Physical Inorganic Chemistry

Physical Inorganic Chemistry
Physica Inorganic
Chemistry
A Coordination Chemistry Approach
S. F. A. KETTLE
Professorial Fellow, University of East Anglia, and
Adjunct Professor, Royal Military College, Kingston, Ontario
Springer-Verlag Berlin Heidelberg GmbH

In memory of Doreen, 1929-1994
British Library Cataloguing in Publication Data
A catalogue record for this book is available from the British Library.
ISBN 978-0-7167-4514-3 ISBN 978-3-662-25191-1 (eBook)
DOI 10.1007/978-3-662-25191-1
Library of Congress Cataloging-in-Publication Data

Kettle, S. F. A. (Sidney Francis Alan)
Physical inorganic chemistry: a coordination chemistry approach I Sidney F. A. Kettle
p. em.
Includes bibliographical references and index
1. Physical inorganic chemistry 2. Coordination compounds.

I. Title.
QD475.K46 1996 541.2'242-dc20 95---44747 CIP
Copyright© 1996 S. F. A. Kettle

Originally published by Spektrum Academic Publishers in 1996
No part of this publication may be reproduced by any
mechanical, photographic, or electronic process, or
in the form of phonographic recording, nor may
it be stored in a retrieval system, transmitted, or
otherwise copied for public or private use without
written permission of the publisher.
Set by KEYWORD Publishing Services, London

Contents
Foreword xiii 3.3 What determines coordination number

and geometry? 42
Preface 'IN
3.4 Isomerism in coordination compounds 43
1 3.4.1 Conformation isomerism 43

Introduction 1 3.4.2 Geometrical isomerism 44
3.4.3 Coordination position isQmerism 44
2 3.4.4 Coordination isomerism 44
Typical ligands, typical complexes 7 3.4.5 Ionization isomerism 45
2.1 Classical ligands, classical complexes 7 3.4.6 Hydrate isomerism 45

3.4.7 Linkage isomerism 45
2.2 Novel ligands, novel complexes 10
3.4.8 Polymerization isomerism 45
2.3 Some final comments 21
3.4.9 Ligand isomerism 46
3 3.4.10 Optical isomerism 46
Nomenclature, geometrical structure 3.4.11 Structural and fluxional isomerism 47
and isomerism of coordination 3.4.12 Spin isomerism 48
compounds 24
3.1 Nomenclature 24
4
Preparation of coordination
3.2 Coordination numbers 31
compounds 51
3.2.1 Complexes with coordination numbers
one, two or three 32
3.2.2 Complexes with coordination number
4.1 Introduction 51
four 33 4.2 Preparative methods 52
five 35 4.2.1 Simple addition reactions 52
3.2.4 Complexes with coordination number 4.2.2 Substitution reactions 54
six 38
4.2.3 Oxidation-reduction reactions 58
seven 38 4.2.4 Thermal dissociation reactions 61
3.2.6 Complexes with coordination number 4.2.5 Preparations in the absence of oxygen 62
eight 39
4.2.6 Reactions of coordinated ligands 65
nine 41 4.2.7 The trans effect 68
3.2.8 Complexes of higher coordination 4.2.8 Other methods of preparing
number 41 coordination compounds 69
viii 1 Contents
····························································································································································································
5 7.8 Tetrahedral complexes 148
Stability of coordination 7.9 Square planar complexes 150
compounds 73 7.10 Other stereochemistries 152

7.11 Ligand field theory 153
5.1 Introduction 73
5.2 Stability constants 74 8

5.3 Determination of stability constants 75 Electronic spectra of transition
5.4 Stability correlations 80 metal complexes 156
5.5 Statistical and chelate effects 84 8.1 Introduction 156
5.6 Solid complexes 89 8.2 The electronic spectra of ym and Ni 11

complexes 157
5.7 Steric effects 90
8.3 Spin-forbidden transitions 163
5.8 Conclusions 92
8.4 Effect of spin-orbit coupling 164
6 8.5 Jahn-Teller effect 166
Molecular orbital theory of transition 8.6 Band contours 170
metal complexes 95 8.7 Band intensities 171

8.8 Tetrahedral complexes 175
6.1 Introduction 95
8.9 Complexes of other geometries 176
6.2 Octahedral complexes 97
8.10 Charge-transfer spectra 178
6.2.1 Metal-ligand ri interactions 97
8.11 Intervalence charge-transfer bands 181
6.2.2 Metal-ligand 7t interactions 103
8.12 Conclusions 182
6.3 Tetrahedral complexes 107
6.4 Complexes of other geometries 110 9

6.5 Formal oxidation states 115 Magnetic properties of transition
6.6 Experimental 117 metal complexes 185
9.1 Introduction 185

7 9.2 Classical magnetism 187
Crystal field theory of transition 9.3 Orbital contribution to a magnetic
metal complexes 121 moment 189
7.1 Introduction 121 9.4 Spin contribution to a magnetic

moment 191
7.2 Symmetry and crystal field theory 122
9.5 Spin-orbit coupling 191
7.3 Crystal field splittings 123
9.6 Low symmetry ligand fields 192
7.4 Weak field complexes 130
9.7 Experimental results 193
7.5 Strong field complexes 136
9.8 Orbital contribution reduction
7.6 Intermediate field complexes 143 factor 195
7.1 Non-octahedral complexes 148 9.9 An example 195

Contents 1 ix
............................................................................................................................................................................................
9.10 Spin-only equation 201 12
9.11 Magnetically non-dilute compounds 203 Other methods of studying coordination
9.12 Spin equilibria 208 compounds 269
10 12.1 Introduction 269
Beyond ligand field theory 211 12.2 Vibrational spectroscopy 270

12.3 Resonance Raman spectroscopy 275
10.1 Bonding in transition metal
organometallic complexes 211 12.4 Spectroscopic methods unique to
optically active molecules 277
10.2 Metal-fullerene complexes 215
10.3 Ab initio and XIX methods 220 12.5 Nuclear spectroscopies 281
12.5.1 Nuclear magnetic resonance
10.4 Semiempirical methods 222 (NMR) 283
10.5 Extended Hiickel method 222 12.5.2 Nuclear quadrupole resonance
10.6 Angular overlap model 226 (NQR) 285
12.5.3 Mossbauer spectroscopy 286
10.7 Three examples: ferrocene,
hexacarbonylchromium and 12.6 Electron paramagnetic (spin)
ethenetetracarbonyliron 227 resonance spectroscopy
227
(EPR, ESR) 288
10.7.1 Ferrocene
Hexacarbonylchromium
10.7.2 229 12.7 Photoelectron spectroscopy (PES) 291
Ethenetetracarbonyliron
10.7.3 232 12.8 Evidence for covalency in transition
10.8 Final comments 235 metal complexes 295
12.9 Molar conductivities
11 296
12.10 Cyclic voltammetry 297
f electron systems: the lanthanides
and actinides 238 12.11 X-ray crystallography 299
12.12 Conclusion 301
11.2 Shapes off orbitals 240 13
11.3 Electronic structure of the lanthanide
Thermodynamic and related aspects
and actinide ions 243
of ligand fields 303
11.4 Spin-orbit coupling 247
11.5 Spin-orbit coupling in pictures 249 13.1 Introduction 303
11.6 Excited states of f electron systems 254 13.2 Ionic radii 303
11.7 Electronic spectra of f electron 13.3 Heats of ligation 305
systems 257
13.4 Lattice energies 307
11.8 Crystal fields and f -+ f intensities 260
13.5 Site preference energies 308
11.9 f-+ d and charge-transfer transitions 262
13.6 Stability constants 311
11.10 Lanthanide luminescence 263
13.7 Lanthanides 312
11.11 Magnetism of lanthanide and
actinide ions 265 13.8 Molecular mechanics 314
11.12 f orbital involvement in bonding 267 13.9 Conclusions 315
x I Contents
............................................................................................................................................................................................
14 16.3 Search for reaction intermediates 391
Reaction kinetics of coordination 16.4 Peroxidases 393

compounds 317 16.5 Blue copper proteins 398
14.1 Introduction 317 16.6 Nitrogen fixation 401

14.2 Electron-transfer reactions 320 16.7 Protonation equilibria in bioinorganic
14.3 Mechanisms of ligand substitution systems 403
reactions: general considerations 325
14.4 Substitution reactions of square 17
planar complexes 328
Introduction to the theory of the
14.5 Substitution reactions of octahedral solid state 407
complexes 331
14.6 Base-catalysed hydrolysis of 17.1 Introduction 407
cobalt(III) ammine complexes 335 17.2 Nodes, nodes and more nodes 408
14.7 Mechanisms of ligand substitution
17.3 Travelling waves and the Brillouin
reactions: postscript 337
zone 413
14.8 Fluxional molecules 338
17.4 Band structure 417
14.9 Photokinetics of inorganic
complexes 339 17.5 Fermi surface 422
17.6 Solid state and coordination
15 compounds 424
Bonding in cluster compounds 345
17.7 Spectra of crystalline materials 428
15.2 Bonding in P4 (and B4 Cl4 ) 346 Appendix 1
15.2.1 'Simple ammonia' model for P4 346 Conformation of chelate
15.2.2 'Twisted ammonia' model for P4 348 rings 432
15.2.3 Atomic orbital model for P4 350
15.2.4 Unity of the three models of P4 Appendix 2
bonding 352
Valence shell electron pair
15.3 Wade's rules 353 repulsion (VSEPR) model 435
15.4 Topological models 359
15.5 Free-electron models 362 Appendix 3
15.6 Detailed calculations 375 Introduction to group theory 440
15.7 Clusters and catalysis, a comment 379

Appendix 4
16 Equivalence of dz2 and dx•-y• in an
octahedral ligand field 445
Some aspects of bioinorganic
chemistry 381
Appendix 5
16.1 Introduction 381 Russell-Saunders coupling
16.2 Myoglobin and hemoglobin 384 scheme 446
Contents 1 xi
Appendix 6 Appendix 11
Ligand tT group orbitals of an High temperature superconductors 472
octahedral complex 449
Appendix 7 Appendix 12
Tanabe-Sugano diagrams and some Combining spin and orbital angular
illustrative spectra 455 momenta 477
Appendix 8
Appendix 13
Group theoretical aspects of band
Bonding between a transition metal
intensities in octahedral complexes
atom and a en Rn ring, n = 4, 5 and 6
459
4 79
Appendix 9
Determination of magnetic Appendix 14
susceptibilities 462 Hole-electron relationship in
spin-orbit coupling 484
Appendix 10
Magnetic susceptibility of a
tetragonally distorted dg ion 466 Index 487
Foreword
GEORGE CHRISTOU
Indiana University, Bloomington
I am no doubt representative of a large number of current inorganic chemists in having

obtained my undergraduate and postgraduate degrees in the 1970s. It was during
this period that I began my continuing love affair with this subject, and the fact that
it happened while I was a student in an organic laboratory is beside the point. I was
always enchanted by the more physical aspects of inorganic chemistry; while being
captivated from an early stage by the synthetic side, and the measure of creation with
a small c that it entails, I nevertheless found the application of various theoretical,
spectroscopic and physicochemical techniques to inorganic compounds to be fascinating,
stimulating, educational and downright exciting. The various bonding theories, for
example, and their use to explain or interpret spectroscopic observations were more or
less universally accepted as belonging within the realm of inorganic chemistry, and
textbooks of the day had whole sections on bonding theories, magnetism, kinetics,
electron-transfer mechanisms and so on. However, things changed, and subsequent
inorganic chemistry teaching texts tended to emphasize the more synthetic and
descriptive side of the field. There are a number of reasons for this, and they no doubt
include the rise of diamagnetic organometallic chemistry as the dominant subdiscipline
within inorganic chemistry and its relative narrowness vis-d-vis physical methods
required for its prosecution.
These days inorganic chemistry is again changing dramatically with the resurgence
of coordination chemistry, fuelled by the increasing importance of metals in biology
and medicine and the new and explosive thrusts into inorganic materials encompassing
a wide variety of types and areas of application, of which high-temperature super-
conductors, molecular ferromagnets and metallomesogens are merely the tip-of-the-
iceberg. Modern-day, nco-coordination chemistry is thus a much broader discipline and
one that now demands greater knowledge and expertise in a much larger range of
theoretical or spectroscopic techniques and physicochemical methods, and to a higher
level of sophistication.
At Indiana University, as at most universities I am sure, we have assigned a high
priority to modifying our inorganic chemistry curriculum to accurately reflect the
changing nature of the field and to better prepare our students for the demands on
them of the new century. The general paucity of suitable texts directed towards the
inorganic chemistry student is a problem. There are, of course, many advanced texts
available for consultation but, on the theoretical/physical side at least, these are
frequently directed at the more quantum mechanically and mathematically competent
reader. In my experience as an instructor, the average student of inorganic chemistry
xiv 1 Foreword
picking up an advanced text on magnetochemistry, for example, will probably not

survive the initial jump into the deep waters of quantum mechanics.
This present work by Sid Kettle represents a wonderful bridge for the student. It is
designed as an intermediate-level text that can serve both as a user-friendly introduction
to a large number of topics and techniques of importance to the student of coordination
and physical inorganic chemistry, and also as a springboard to more advanced texts and
studies. It is written in a style that is appropriate for a teaching text, anticipating
and answering the questions that students will typically have on encountering the topic
for the first time, and introduces a large number of theoretical, spectroscopic and
physicochemical techniques without sacrificing the more classical content of a coordi-
nation chemistry text. In this regard, it is a wonderful hybrid of the classic and modern
aspects of coordination and physical inorganic chemistry and is consequently an
admirable text for the student of this area.
Preface
Some twenty years ago, theoretical aspects of inorganic chemistry formed a major
component of any inorganic textbook. Today, this component is much less evident. No
doubt, this shift in emphasis is a proper response both to the undue weight then given
to theoretical aspects and to the developments that have taken place elsewhere in the
subject. However, in the interval there have been theoretical developments that deserve
a place; further, it has probably become more difficult for the interested student to
access the older work. There seemed to me to be a real need for an easy-to-read, and
so largely non-mathematical, text that would bridge the gap between the relatively
low-level treatments currently available and the research level paper, review, monograph
or text. The present book was written with the object of providing a bridge for this
gap. Although the motivation for writing it is seen in its theoretical content, it was
recognized that there are advantages in placing this in a broader context. So, what has
resulted is a book which contains an overview of the relatively traditional and
elementary along with contemporary research areas, wherever possible viewed from an
integrated theoretical perspective. Because a text on physical inorganic chemistry can
easily become a series of apparently disconnected topics, I have given the subject a
focus, that of coordination chemistry, and have included chapters which should enable
the book to double as a text in that area. To keep the size of the book manageable, to
recognize that it is aimed at the intermediate stage reader, and because the topic is
covered so extensively elsewhere, I have assumed a knowledge of the most elementary
aspects of bonding theory.
In a book such as this it is impossible to avoid cross-references between chapters.
However, it is equally difficult to ensure that such cross-references supply the answers
expected of them. I have therefore attempted to make each chapter as free-standing as
possible and have used the resulting duplications as a mechanism for deepening the
discussion. This strategy can produce its own problems as well as benefits; I hope that
the index will provide direction to sufficient additional material to deal with the
problems!
I am indebted to many institutions which provided the hospitality that enabled most
of the book to be written-Chalmers University and the University of Gothenburg,
Sweden; the Royal Military College, Kingston, Canada; the University of Turin, Italy;
the University of Nairobi, Kenya; Tokyo Institute of Technology, Japan; the University
of Szeged, Hungary and Northwestern University, USA. Of the numerous individuals
who have provided helpful comments on sections of the book, and often offered material
for inclusion, I am grateful to Professor R. Archer and his students at the University
of Massachusetts, Amherst, USA, who made many detailed comments on an early
edition of the text, to Professor K. Burger of the University of Szeged, Hungary-his
contributions were very helpful-and to Dr. S. Cotton, who was a constant fund of
xvi I Preface
comment and information. I am particularly indebted to the Rev. Dr. lain Paul who,
in his own inimitable manner, worked through every sentence and made a multitude
of suggestions for improvement and clarification. Defects, errors and omissions, of
course, are my own responsibility.
S.F.A.K.
Introduction
Textbooks on physical inorganic chemistry can, during their preparation,

easily evolve into compilations of apparently unrelated topics. In writing the
present book, therefore, it was decided to circumvent this problem by
adopting a single, unifying theme: coordination chemistry. The benefit of this
approach is that the theme spans almost all aspects of physical inorganic
chemistry; furthermore, the resulting book also doubles up as a text on
coordination chemistry itself. In achieving this duality, some of the material
present might appear out of place in a book devoted to physical inorganic
chemistry alone. However, it is probably no bad thing that, for example, in
addition to a discussion about the chemical bonding within a particular
exotic species, reference can also be found to its preparation. Since, then, the
theme of this book is that of coordination compounds (or, as they are often
called, coordination complexes), our first task is to define the term coordination
compound. This is not straightforward, for the use of the term is determined
as much by history and tradition as by chemistry. In practice, however,
confusion seldom arises. Let us consider an example.
When boron trifluoride, a gas, is passed into trimethylamine, a liquid,
a highly exothermic reaction occurs and a creamy-white solid separates.
This solid has been shown to be a 1:1 adduct of the two reactants, of which
the molecules have the structure shown in Fig. 1.1, the boron atom of
the boron trifluoride being bonded to the nitrogen atom of the trimethyl-
amine. The adduct, resulting from the combination of two independently
stable molecules, is an example of a coordination compound. An electron
count shows that the boron atom in boron trifluoride possesses an empty
valence shell orbital, whilst the nitrogen of the trimethylamine has two
valence shell electrons in an orbital not involved in bonding. It is believed
that the bond between the boron and nitrogen atoms in the complex results
from the donation of these nitrogen lone pair electrons into the empty
boron orbital, so that they are shared by both atoms. Coordination
Rc. 1.1 The structure of the coordination
compound formed between boron trifluoride compounds in which such electron-transfer appears to be largely respons-
and trimethylamine. ible for the bonding are ·sometimes also called donor- acceptor complexes,
2 I Introduction
although it is to be emphasized that, once formed, there is no difference

in kind between these and ordinary covalent bonds; the difference is in
our approach to them. In the boron trifluoride-trimethylamine adduct,
the nitrogen atom of the trimethylamine molecule is said to be coordinated
to the boron atom. That is, the electron donor is said to be coordinated
to the electron acceptor. A coordinating group (usually called a ligand-it
is ligated, 'tied to', the electron acceptor) need not be a molecule and need
not be uncharged. For example, boron trifluoride reacts with ammonium
fluoride to give the salt NH 4 [BF4 ] which contains the complex anion
[BF4 ] - . Here we adopt the convention of placing the complex species of
interest within square brackets, a convention that will almost invariably be
adopted in this book. In the [BF4 ]- anion the boron atom is tetrahedrally
surrounded by ligands, just as it is to a first approximation in [BF3 • NMe 3]
(Fig. 1.1). Notice that, for non-transition metals and metalloids, complex
formation is associated with a change (usually an increase) in the number
of groups to which the central atom is attached. Boron trifluoride, BF3, is
not normally thought of as a complex, but its adduct with trimethylamine
certainly is.
Most workers regard both trimethylamine and the fluorides as ligands
in the adduct (a pattern that has just been followed). It would be a logical
deduction from the picture just presented to conclude that the maximum
number of ligands which can be added to form a complex is determined
by the number of empty valence shell orbitals on the acceptor atom. Whilst
this is generally true, an indication of the difficulty of rigorously defining
'a complex' is given by the fact that, in practice, the criterion of change in
number of bonded atoms outweighs all others for these elements. Thus,
phosphorus pentachloride exists in the gas phase as discrete PCI 5 mole-
cules. The solid, however, is an ionic lattice, containing [PCJ~+ and
[PCI 6 ] - ions. These two species are usually classed as complex ions,
although the molecule in the gas phase is not.
The detailed geometry of a complex molecule is not simply a combina-
tion of the geometries of its components. In the trimethylamine-boron
trifluoride adduct, for instance, the B-F bond length is 1.39 A and the F-B-F
bond angle 170° compared with 1.30A and 120o in the isolated BF3 molecule.
Similarly, the geometry of the bound trimethylamine fragment differs from
that of the free amine. Information about the bonding within a complex may,
in favourable cases, be obtained by a detailed consideration of these bond
length and angle changes. 1 It is not surprising, then, that a recurrent
theme throughout this book will be the relationship between molecular
1 But there are traps for the unwary. In the simpler compound H 3 'B--NH 3 it was found
that a discrepancy exists between the 'B--N bond length determined by X-ray crystallography
(1.564A) and by microwave spectroscopy (1.672A). Some detailed theoretical calculations
have been carried out on the problem and have shown that the energy difference between
these two bond lengths is rather small for the isolated molecule. Simulation of the molecular
environment showed that the longer bond length in the crystal almost certainly arises from
environmental effects and therefore carries no great bonding significance--<>xcept that over
a short range the total bonding energy is rather insensitive to the precise internuclear distance.
A second trap arises from the observation that the (stabilization) energies of complex
formation increase in the order BBr 3 > BC1 3 > BF3 , an observation that has been related
to " bonding between boron and the halogens (being greatest for the bromide). In fact,
accurate calculations have shown that the difference in stabilities results from variations in
the simple donor-acceptor bonding described in the text.
Introduction 1 3
geometry and electronic structure, the link between the two commonly being
provided by group theory.
Complexes are formed by both transition metal and non-transition
elements. Indeed, at the present time all compounds of transition metal
ions, with very few exceptions, are regarded as complexes. However, despite
the argument given above, the simple donor-acceptor bond approach does
not seem immediately applicable to coordination complexes of the transi-
tion metals, since the molecular geometry does not depend greatly on the
number of valence shell electrons-and, so, on the number of empty
orbitals. As will be seen in Chapter 7, in the simplest model of the bonding
in transition metal complexes electron donation is not even considered to
be involved, a molecule being regarded as held together by electrostatic
attraction between a central transition metal cation and the surrounding
anions or dipolar ligands. However, in more sophisticated discussions of
the bonding (Chapters 6 and 10) the donor-acceptor concept is largely
reinstated for these compounds. So we may conveniently (but not always
correctly) regard a coordination compound as composed of (a) an electron
donor (ligand or Lewis base), an individual atom or molecule which possesses
non-bonding lone-pair electrons but no low-lying empty orbitals; and (b) an
electron acceptor (metal atom, cation or Lewis acid) which possesses a
low-lying empty orbital. As in many other areas of chemistry, we shall often
be particularly concerned with the pair of electrons that occupy the highest
occupied molecular orbital (the HOMO) of the electron donor. This is
matched by an interest in the lowest unoccupied molecular orbital of the
electron acceptor (the LUM0). 2 The donor atom of a ligand is usually of
relatively high electronegativity and the acceptor atom is either a metal or
metalloid element.
Chapters 2-4 are full of examples of ligands and coordination com-
pounds and the reader can gain an impression of the field by quickly
thumbing through them. The field is not as complicated as it may appear,
although it will rapidly become evident that at the present time some rather
unusual organic molecules are increasingly being used as ligands and that
neither the methods of preparation nor the molecular geometries formed
need be quite as simple as for the examples given above. Indeed, part of
the current fascination of the subject lies in the elegance of many of the
complexes which are currently being studied. Complexes in which the metal
atom is totally encapsulated, as in the sepulchrates; those in which it is at
the centre of a crown (crown ether complexes, for instance); those in which
it is surrounded by two ligands which interleave each other (complexes of
catenands); those in which it is at the centre of a stockade-like ligand
(picket-fence complexes) and so on. By such means it is proving possible
to design highly metal-specific ligands, which offer the prospect of selective
ion extraction from, for example, low-grade ores or recycled materials.
The future importance of such possibilities in the face of ever-declining
natural resources can scarcely be overestimated. Similarly, the use of such
complexes in small-molecule activation will surely be of vital importance--
for instance, in the fixation of gaseous nitrogen and the synthetic use of
hydrocarbon species which would otherwise be used as fuels.
2 Because the basics of the subject were developed before use of the HOMO and LUMO
terminology became widespread these labels are scarcely to be found in the relevant literature.
4 1 Introduction
Inevitably, current research tends to focus on the unusual and the exotic
and so, since a book such as this attempts to reflect something of current
work, tends to make the subject appear less straightforward than it really
is. Perhaps it is helpful to recognize that even at a simple level, problems
of definition can occur. Thus, an uncharged compound containing a main
group metal or metalloid element bonded to a methyl group is not usually
viewed as a complex in which CH 3 functions as a ligand, although the
CH 3 group is isoelectronic with ammonia, a molecule which is frequently a
ligand. So, compounds such as Zn(C 2 H 5 h and Si(CH 3 ) 4 would not usually
be considered complexes. However, successes in the synthesis of transition
metal-methyl compounds means that there has been a change in attitude
and that these, too, are now regarded as complexes containing the CH 3
group as ligand. The question of whether they should be considered
as complexes of CH3 is not usually regarded as of great importance. A
similar ambiguity is that although the manganate ion MnOi- would be
considered a coordination complex (of Mn 6 + and 0 2 -) the sulfate anion
SOi- would not. Evidently, we have reached the point at which history
and tradition, as well as utility, colour the definition of a coordination
compound.
The father of modern coordination chemistry was Alfred Werner, who
was born in 1866 and lived most of his life in Zurich. At the time it was
known that the oxidation of cobalt(II) (cobaltous) salts made alkaline with
aqueous ammonia led to the formation of cobalt(III) (cobaltic) salts
containing up to six ammonia molecules per cobalt atom. These ammonia
molecules were evidently strongly bonded because very extreme conditions-
boiling sulfuric acid, for example-were needed to separate them from the
cobalt. There had been considerable speculation about the cobalt-ammonia
bonding and structures such as
/NH 3 -CI
Co--NH 3 · NH 3 · NH 3 · NH 2 -CI
"'-NH 3 -CI
which today look quite ridiculous (although based on the not unreasonable
hypothesis that, like carbon, nitrogen can form linear chains) had been
proposed for the cobalt(III) salt CoN 6 H 18 CI 3 (which we would now write
as [Co(NH 3 ) 6 ]CI 3 ). Werner's greatest contribution to coordination chem-
istry came in a flash of inspiration (in 1893, at two o'clock in the
morning) when he recognized that the number of groups attached to an
atom (something that he referred to as its secondary valency) need not
equal its oxidation number (he called it primary valency). Further, he
speculated that for any element, primary and secondary valencies could
vary independently of each other. The chemistry of the cobalt(III)- ammonia
adducts could be rationalized if in them cobalt had a primary valency of
three, as in CoCI 3 , but a secondary valency of six, as in [Co(NH 3 ) 6 ]CI 3 . The
term secondary valence has now been replaced by coordination number and
primary valency by oxidation state but Werner's ideas otherwise stand largely
unchanged.
Subsequently, Werner and his students obtained a vast body of experi-
mental evidence, all supporting his basic ideas. They further showed that
in the complexes they were studying the six coordinated ligands were
Introduction I5
arranged octahedrally about the central atom (Figs. 1.2 and 1.3). Werner
was awarded the Nobel prize for chemistry for this work in 1913. Some
measure of his stature and work is provided by the fact that in one field
(that of polynuclear cobaltammine complexes) there has, to this day, been
scarcely any addition to the list of compounds he prepared.
Most textbooks discuss transition metal complexes separately from
those of the main group elements. There is, in fact, much in common
between the two classes and, whenever possible, we shall treat them as one.
However, complexes of the transition metal ions may possess an incomplete
shell of d electrons which necessitate separate discussion. This character-
istic makes it particularly useful to determine the magnetic and spectral
properties of members of this class of complexes and the exploration of
these properties will require separate chapters devoted to them. In a similar
way, complexes of the lanthanide and actinide elements, with, typically, an
FJ&. 1.2 An octahedral complex ML., where M incomplete shell off electrons tucked rather well inside the atom and away
is represented by the central white atom and
the ligands L each by a shaded atom. A regular from the ligands- and so behaving rather as if they ·are in an isolated
octahedro~ne is shown in Rg. 1.3--has atom-require their own discussion.
eight faces (each an equilateral triangle) and six The water-soluble ionic species of transition elements such as chromium,
equivalent vertices. In an octahedral complex
the ligands are placed at the vertices. In Rg. manganese, iron and copper seem to exist in aqueous solution as, for
1.2, and in similar diagrams throughout this example, [Cr(H 2 0) 6 ]3+, [Mn(H 2 0) 6 ]2+ and [Fe(H 2 0) 6 ]2+ . That is, it is
book, the perspective is exaggerated (the more accurate to talk of 'the aqueous chemistry of the [Cr(H 2 0) 6 ]3+ ion'
central four ligands lie at the comers of a
square) and all ligand atoms are the same size. than of 'the aqueous chemistry of the Cr3+ ion '. Similarly, in solid FeCI 3 ,
In this example, all six ligands are identicaL the iron atoms are not attached to three chlorines but, octahedrally, to six
Even if they are not, provided the geometrical (each chlorine is bonded to two iron atoms). We have already encountered
arrangement shown in Fig. 1.2 is more-or-less
maintained, the complex is still referred to as the fact that solid PCI 5 is really [PCI 4 ] + [PCI 6 r. The lesson to be
octahedral. As molecular symmetry is important learnt from all this is that coordination compounds are much more
for the arguments to be presented in many of common than one might at first think. The colour of many gemstones and
the following chapters, it will often prove
convenient to emphasize this by including in minerals, the chemistry carried out within an oil refinery, element deficiency
structural diagrams, lines which remind the diseases in animals, the reprocessing of nuclear fuel rods, the manufacture
reader of the molecular symmetry. Commonly, of integrated circuits, the chlorophyll in plants, the colours of a television
such lines will link ligands together and , clearly,
should not be interpreted as bonds between screen-all involve complexes, though we shall not be able to cover all
ligands. of these diverse topics in the present book. Although the first example we
gave in this chapter portrayed complexes as being formed between indepen-
dently stable species, and this is often the case, there are also many
fascinating examples of molecules which are only stable when they exist
as part of a complex; even independently stable species have their chemical
as well as their physical properties drastically changed as a result of
coordination.
In summary, there is no precise and time-constant definition of a
coordination compound- at one extreme methane could be regarded as
one-and the usage of the term is extended to all compounds to which
some of the concepts developed in the following chapters can usefully be
applied. Indeed, one could argue that the value of the concept lies in its
flexibility and adaptability so that the absence of a fixed and agreed
definition is no handicap. We shall find that the study of coordination
compounds excludes few elements-the sodium ion forms complexes- and
overlaps with biochemistry and organic chemistry. Further, it will involve
some fairly detailed theoretical interpretations, although in this book the
FJt. 1.3 An octahedron, a regular figure in
which all vertices are equivalent, as are all powerful but surprisingly simple concepts of symmetry are used to reduce
faces and all edges. theoretical complexities to a minimum.
6 1 Introduction
Further reading nial, published in 1967 by the American Chemical Society as

No. 62 in their Advances in Chemistry series (R. F. Gould,
Most inorganic texts published before the mid-1950s give editor). This book contains over 40 chapters on historical and
historical overviews of the development of coordination chem- current (in 1967) chemistry, including chapters devoted to the
istry. The language used sometimes seems strange in terms of Werner-Jergensen controversy (the nitrogen-chain structure
modern usage and should not be allowed to distract the reader for cobaltammines was originally proposed by Blomstrand and
unduly. A browse through the older books in a good library Jergensen, who was his student at the time), polynuclear
should be adequate; as an indication of the variety available complexes of Co 111 ammines and so on.
the following are worthy of mention: A useful source is Volume 1 of Comprehensive Coordina-
tion Chemistry G. Wilkinson, R. D. Gillard and J. A. McClev-
• Modern Aspects of Inorganic Chemistry H. J. Emeleus and erty (eds.), Pergamon Press, Oxford, 1987. Chapter 1.1 is
J. S. Anderson, Routledge and Kegan Paul, London, 1938/ 'General Historical Survey to 1930' by G. B. Kauffman and
1952. Chapter 1.2 is 'Development of Coordination Chemistry since
• The Chemistry of the Coordination Compounds J. C. Bailar 1930' by J. C. Bailar Jr. Although they are frequently too
(ed.), Reinhold, New York, 1956. specialized to warrant inclusion as further reading in the
• An Introduction to the Chemistry of Complex Compounds following chapters, the contents of the volumes of Comprehen-
A. A. Grinberg, 1951, English translation by J. R. Leach, sive Coordination Chemistry provide a wealth of information
Pergamon, Oxford, 1962. on the details of coordination chemistry. The volumes as-
sume, however, knowledge of the basic language and concepts
For the person interested in exploring the historical aspect of the subject, such as can be gained from the following
in depth, one book is essential reading. This is Werner Centen- chapters.
Question cobaltous oxide. Peculiar compounds are produced in this solu-

tion.'
1.1 In the book Principles of Chemistry published in English
in 1881 Mendeleeff wrote: At about the same time, of course, Werner's work was provid-
'The admixture of ammonium chloride prevents the precipitation of ing an understanding of these peculiar compounds. Write a
cobalt salts by ammonia, and then, if ammonia be added, a brown one-page letter to Mendeleeff on behalf of Werner outlining
solution is obtained from which potassium hydroxide does not separate the key points in Werner's understanding.
Typical ligands, typical
complexes
2.1 Classical ligands, classical complexes

As was seen in the previous chapter, ligands are atoms or molecules which,
at least formally, may be regarded as containing electrons which can be
donated to an atom which functions as an electron acceptor. The presence
of such a pair of electrons is a necessary, but not a sufficient, condition.
Thus, the halide anions are typical simple ligands but halogen compounds
such as CH 3 Cl or C 6 H 5 F are very seldom ligands, although the halogen
atoms in them still possesses electron pairs which could be donated. The
reason lies, at least in part, in the high electronegativity of the halogens.
When the halogen possesses a negative charge there is little energetic cost in
reducing this charge, a cost that can be paid for by the exothermicity of the
bond formed. When the halogen is uncharged, the energetic cost of it
becoming positively charged (as it would if it donated electrons) is too great.
However, electronegativity cannot be the sole reason because, as we shell
see, organic oxides and sulfides form many complexes-and yet the electro-
negativity of oxygen is greater than that of chlorine. An illustration
of the complexing ability of ethers, for example, which is more spectacular
than dangerous-although appropriate precautions should be taken-is
to add a drop of diethylether, Et 2 0, to tin(IV) chloride (stannic chloride, a
liquid). The solid complex [SnC14 (Et 2 0)z] is instantly formed. The heat of
reaction is sufficiently great to boil off some of the diethylether, sending
clouds of the white complex into the air. As evident from the previous
chapter, ammonia and related compounds such as trimethylamine form
many complexes. Organophosphorus ligands are also widely used in synthetic
chemistry and we shall meet them in the contexts of organometallic
chemistry (that of complexes in which a ligand which would be regarded
as part of organic chemistry is bonded through carbon to a metal) and
catalysis, in particular.
8 1 Typical ligands, typical complexes
Table 2.1 Some classical ligands which are common in the complexes of either
transition metal andjor main group elements. The names given follow the rules
to be detailed in Chapter 3. Note that some species are shown twice, when they
can coordinate in more than one way. Note, too, that some ions shown once
can, in fact, coordinate in more than one way; examples are provided by eN-, S 2 o~
and OCW
Ligand Name
Donor atoms from group 17(7) F- flu oro

of the periodic table Cl- chloro
sr- bromo
1- iodo
Donor atoms from group 16(6) o2- oxo

of the periodic table OW hydroxo
-o~- peroxo
co~- carbo nato
CH 3C02 acetate
ONO- nitrite
so~- sulfate
so~- sulfite
s2o~- thiosulfito
s2- thio
CH 3S- methylthio
H20 aqua
CH 30H methanol
(NH 2)2CS thiourea
(C2Hsl20 diethylether
Donor atoms from group 15(5) cw cyano

of the periodic table ocw cyanate
sew thiocyanate (note, bonded through N)
N02 nitro
N-
3 azido
NH 3 amminea
CH 3 NH 2 methylamine
N(CH3)3 trimethylamine
C6 H5 N pyridine (usually abbreviated as py)
a Note the spelling, 'mm', not 'm'.
A list of some simple and common ligands is given in Table 2.1. The
entries in this table are confined to classical ligands, such as could well
have been studied by Werner. There are other ligands, many also simple
and common but non-classical-such as the organophosphines, which will
be covered shortly. Inevitably, the distinction we are making is an arbitrary
one. In Table 2.2 are listed representative examples of complexes formed
by some of the ligands in Table 2.1. The detailed molecular geometries of
the complexes in Table 2.2 will not be discussed because for many of them
there are ambiguities. These problems will be dealt with in Chapter 3,
where many of the examples given in Table 2.2 will reappear.
Complexes of most of the ligands that have so far been mentioned have
been studied for almost a century. Although one might expect the field to
be exhausted, each year there are a few new surprises: the discovery of a
method for the easy preparation of complexes of a metal in a valence state
Classical ligands, classical complexes 1 9
that was previously regarded as difficult, or the preparation and character-

Table 2.2 Some of the complexes ization of a complex which was previously believed not to exist. In this
formed by some of the ligands in
Table 2.1. This table endeavours to field, however, the main work that is still to be done is not that of
demonstrate that any one metal may preparation, but rather work at a deeper level, deeper perhaps than the
well form complexes with different fields covered in this book. For instance, in the solid state, how does a
numbers of ligands and that many complex ion interact with its environment? How are these interactions
complexes are not monomeric (there changed with different counterions; to what extent are the properties that
are sulfur and iodine bridging atoms
in the two cases not explicitly we observe those of complex plus environment and different from those of
detailed). Note that when two different the isolated complex itself? Why, for instance, does the [Ni(CN) 5] 3 - anion
complexes contain the same number have two different geometries in the crystals of one of its salts? There has
of ligands it does not necessarily been some considerable interest generated by the recent discovery that some
mean that the geometrical anions which have always been regarded as having very little tendency to
arrangements of the ligands is the
same in the two cases. Note, too, that coordinate can actually do so-perhaps the best example is provided by the
the attempt to show variety means anion [B(C 6 H 5 ) 4 ]-. The current thrust of preparative work exploits the fact
that this table does not properly that several donor atoms, oxygen and nitrogen in particular, can be strung
reflect the fact that the majority of together with a web of carbon (and sometimes boron or phosphorus) atoms.
complexes contain metal ions bonded An almost infinite variety of exotic ligands becomes possible. This field has
to six ligands
many attractions. By choosing the ligand to be one with a rather rigid
[Co(NCS) 4] 2 - backbone it is possible to impose an unusual coordination geometry on a
[Co(CN) 5] 3 - metal atom. A very popular strategy at the present time is to chose a ligand
[Co(N0, )6] 3 - which has very bulky, and so sterically demanding, substituent groups. In
[
[Co(NH 3 ) 5 N3 ] 2~ ] 3+ the complexes it forms there simply is not enough space to fit very many
ligands around the central atom and so a low coordination number or
(NH3)4< >(NH3)4 unusual geometry results. It is found that metal ions in unusual coordination
geometries often have unusual reactions and/or properties and this makes
NH 2 them of particular interest. Thus, with suitable choice of ligand it is possible
[Cr04 ] 2 -
[Cr(SCN)6]3-
to make volatile compounds of sodium! Alternatively, by careful tuning of
[Mo(CN) 8] 3 - the ligand geometry it may be possible to make it highly specific for a
[Mo2S2(CN)8]6- particular metal. This produces visions of metal recovery from low-grade ore
[CuCI4] 2 - and even gold from sea water (such schemes tend to fail because the cost of
[CuCI5]3- the ligand and its recovery for reuse exceeds the value of the metal obtained).
[Fe(H2 0) 6 ] 3 +
[PtCI 2 (py)2 ]
Next, it may be possible to produce a ligand which closely mimics, in
[Pt(NH3),(0H),] its geometry and composition, that of one found in nature in a complex
[AI(OH)(H 2 0) 5 ]2+ of biological importance. The biological compound is almost certainly only
[TiCI 4 (Et,0) 2 ] available in small quantities, difficult to purify and unstable under most
[SnCisr laboratory conditions. Working with a model compound is much easier than
[AuF.r
[ZrF7 ]3- working with the real thing! This topic will be covered in much more detail
[BeF4]'- in Chapter 16.
[SbBr6 r There is one further advantage to working with ligands containing more
[Ailela] 2 - than one donor atom. This is that the (thermodynamic) stability of a
complex in which two or more donor atoms are part of the same ligand
molecule often appears much greater than if the same atoms were in
separate ligand molecules. There has been much debate on the origin of
this co-called chelate effect. Chemists like to use their imaginations and to
compare a metal ion held between two donor atoms on a ligand with a
crab holding its prey in its claws-hence chelate (Greek chelos-a claw).
It is common to talk of chelating ligands and of chelate complexes.
Complexes are often conveniently divided into two classes, labile and inert.
In the former, consisting of most complexes of main group metals and many
of the more familiar transition metals, ligands are readily replaced. In the
latter, for example complexes of Cr111 and Co 111, ligand replacement is very
slow except under forcing conditions. The chelate effect is a phenomenon
which increases the inertness of complexes (which may mean making a
complex less labile). We shall return to the chelate and related effects in
Section 5.5. At this point all that will be added is a word of caution:
although the occurrence of a chelate effect is common it is not invariably
present. So, organic isocyanides, RNC, form many complexes (bonding to
the metal through the terminal C). The R-N-C sequence remains essentially
linear in complexes and the metal atom also bonds colinearly. The result
is that if a bidentate organic ligand containing two RNC groups is
synthesized then there has to be a sizeable number of carbon atoms (seven
CH 2 units, for instance) between the two -NC groups if they are both to
coordinate to the same metal atom and thus form a chelate. This means
a 12-membered ring system and, as will become evident in Chapter 5,
12-membered rings show no hint of a chelate effect.
In Table 2.3 are listed some of the more common, classical, polydentate
ligands. Again, imagination. The ligand is now pictured as biting, and thus
holding onto, the metal with several teeth (bidentate1 = two donor atoms,
tridentate= three donor atoms). The (minimum) distance between two
donor atoms in a bidentate ligand is sometimes referred to as the bite of
the ligand and the angle subtended at the metal atom the bite angle. In
Table 2.4 are detailed a selection of some of the more exotic ligand species
under current study. The systematic names of these molecules are usually
so horrendous that trivial, often physically descriptive, names are preferred.
Typical examples are picket fence, crown and tripod, some of which are given
in Table 2.4. Table 2.5 shows a selection of the complexes formed by the
ligands contained in the previous two tables.
2.2 Novel ligands, novel complexes

Since the 1950s it has been clear that the simple 'lone electron pair donor'
picture of a ligand and 'lone electron pair acceptor' picture of a metal in a
complex is inadequate. This is nowhere clearer than in the field of organo-
metallic chemistry, where a host of organic molecules, in which all of the
valence electrons are involved in bonding within the organic molecule, form
complexes with metal atoms. Notice the use of the word atom-the metal
is commonly, formally, zero-valent in these compounds and so any simple
electrostatic model for ligand-metal bonding which might be applied to their
classical counterparts seems rather implausible. The same conclusion is also
forced upon us by the nature of the ligands commonly involved-molecules
such as hydrocarbons, carbon monoxide and all sorts of unexpected species,
even, on rare occasions, the H 2 molecule. However, it is also clear that there
are links with the more classical complexes-with ligands such as the halides,
sulfides and organic phosphines, being common to both sets. As is so often
the case, there is a continuous gradation. We have looked at the classical case
1 The current recommendation is that the word bidentate be replaced by didentate but
this recommendation has yet to gain general acceptance.

Novel ligands, novel complexes 1 11
Table 2.3 Some common polydentate ligands" {charges on anions are omitted)
Name Common abbreviation Structure

Bldentate• ligands
H
I
,...c,
Acetylacetonato or 2,4-pentanedionato acac CH3-C ,--, C-CH 3
II \I
o' 'o
"'
2,2'-Bipyridine 2,2'-bpy
often written as bpy
0 0
Oxalato ox
0
,,,--,
~c-cIt·
I I
\
0
\ ';
CH2-CH2
I \
Ethylenediamine or 1,2-ethanediamine en NH2 NH2
\ I
©(As(CHal2
of>henylenenebis(dimethylarsine) or
1,2-phenylenebls(dimethyarsine)
diars
0 ~Hal2
Glycinato gly
8-Hydmxyquinolinato oxinate
w o-
1,10-Phenanthroline phen
©0© \ I
C~a /Ha
1/-c~
Dimethylglyoxlmato or 2,3-butanedione dioximato dmg O-N N-O
(see Table 2.5) /
I
\ I \
I
\ (continued)
Table 2.3 (continued)
Tridentate• ligands
2 ,2' ,6' ,2"-Terpyridyl terpy
Diethylenetriamine dien
CH2-N...._
I H2
CH-N_...,.
I H2
1,2,3-Triaminopropane tap CH2-W.-
H2
Tetradentate• ligands /CH2-CH2

NH "'NH
I "' -'\
CH2 CH2
Triethylenetetraamine trien
\ I
C~ f ~ /CH2
NH2 NH2
CH:i'""""/\-C~
I CH2 CH2
CH2 I I
tren
I;~ c~ / -NH2
Tris(2-aminoethyl)amine NH2 NH2
CH:i'""""/\-C~
1 CH2 c=o
O=C I I
1,...1-c'- / -o
0 0 0
Nltrilotriacetato NTA
~L~
~~ ¢ ~~.
Tris(2-diphenylarsineophenyt)arsine QAS
'AsPh2

p; porphyrins are usually

labelled according to their
Porphyrino substituents -thus. tpp =
tetraphenylporphyrin, see
Table 2.4.
Phthalocyanino pc
Hexadentate• ligand
o=c-o..- --o-c=o
I I
CH2 CH2
I I
Ethylenediaminetetraacetato EDTA --N-CH2-CH2-N--
I I
CH2 CH2
I I
o=c-o-- --o-c=o
a This notation is probably self-explanatory because of the examples given. If not, it is described towards the end of Section 3.1.
in some detail if not depth (that will come in the next few chapters). What
of the complexes formed by non-classicalligands and, in particular, what is
the ligand-metal bonding involved? The answer to this question has become
so important in inorganic chemistry that an outline answer will be given
here, although it will have to be refined later. The pattern is illustrated by
a discussion of the carbon monoxide molecule and the way that it bonds
to a transition metal; with not much modification a general picture emerges.
At the heart of it is the simultaneous-and interdependent- coexistence of
two distinct bonding mechanisms.
At first sight one might well expect CO to bond to a metal through its
oxygen atom-this oxygen has a lone pair of a electrons and oxygen donors
form many complexes; we would expect the oxygen to be more negatively
charged than the carbon because of the difference in their electronegativities.
Although oxygen-bonded CO complexes are known, the almost invariant
mode of bonding of CO to a transition metal is through the carbon atom.
Table 2.4 Some more exotic polydentate ligands
't
o=c\
NH
Oibenzo-l!k:rown-6
A crown ether with 18 atoms in the ring of which 6 are oxygens
A picket-fence porphyrin
(N'D
GI')Nyr;N
N
2,2 ,2-crypt
A cryptand (= Greek hidden); the 2s indicate the number of
liN)
oxygens In each N.. · N chain The sepulchrate ligand, which is hexadentate (the top
and bottom nitrogens do not normally coordinate)
tpp
Tetraphenylporphyrin
Bipyridyl groups grafted onto a cyclic hexamine
A basket-handle porphyrin
When three bipyridine ligands are capped (twice) by a triamine
the cage ligand that results can complex three metal atoms
simultaneously (in this example the 'caps' are, essentially,
the ligand tren, tris(2-aminoethyl)amine)
Table 2.5 Selection of complexes formed by some polydentate ligands
CH3 /O-H···~
'c-:-N, -.r:!::-c-CH3
/,;>f 'N·---/ 'I
cH;;-c,~-?-=-:: '-.......N~c-cH3
I /
0 .... H-0
The square planar complex [Ni(dmg)2 ]: note

The general structure of [M(bidentate) 3 ]'"' complexes where the interligand hydrogen bonds (also
(bidentate) is one of the bidentate ligands of Table 2.3 indicated in Table 2.3)
2+
The K+ complex of dibenzo-18-crown-6
(continued)
Table 2.5 (continued) +
Rb+ 2,2,2-cryptate: oxygen and

nitrogen donor atoms are shown,
black dots indicate CH 2 groups
3+
(Ntl
The con complex of tpp: one of two
equivalent structures (cannonical
GN'fj(N)
l\
forms) of the ligand is shown
Q) N
[Co(sepulchrate)] 3 +
The crystal structure of the M = Ag+ species Cu 2• phthalocyanin: a complex which

with this ligand has been determined is used as a blue pigment
There is also a lone pair of r:r electrons on the carbon which is larger, and
so more available for bonding, than that on the oxygen. This, then, is the
first member of the bonding partnership, a perfectly normal r:r donation from
carbon to the metal. But, given that a carbon atom is involved and that
carbon atoms seldom act as ligands in classical complexes unless there is a
negative charge around (as in CN-), it would be surprising if this r:r donation
above were to lead to a strong bond. It needs reinforcing. The CO molecule
is well able to provide this reinforcement because not only is the lone pair
r:r orbital larger on carbon than on oxygen, so too are the lobes of the orbitals
Fl•. 2.1 (a) The u donaljon OC ~ metal: here,

the transition metal dz2 orbital is shown as the
acceptor but it could be some mixture of s, Pz
and dzl · Note that there is lone-pair u electron
density on both 0 and C. That on C is the larger
and so has the greater overlap with the empty (a)
metal orbital. In this, and all other figures in
this chapter, filled orbitals are shaded; the
phases of orbitals are given expliciijy. (b) The n
back-donation metal ~ CO: the metal orbital is
almost pure d, the CO orbital is an empty
anljbonding n orbital. Note that for a linear
triatomic OCM system there is second,
equNalent, interaction to that shown above
(it is like that shown but rotated 90' about the
OCM axis and so is located above and below
the plane of the paper).
(b)
M
which are the anti bonding counterparts of the C=O rr bonding orbitals. In
CO itself, of course, these orbitals are unoccupied but that is no reason for
ignoring them in complexes. They are LUMOs and a lesson which has
been well learnt in recent years is that LUMOs are seldom disinterested
spectators- they commonly play a key role in determining the outcome
of chemical reactions (and that means that they get involved in the bonding
somewhere along the way). It is therefore not surprising to learn that there
is both experimental and theoretical evidence that electron dona tion from
transition metal d orbitals to the CO rr antibonding orbitals is of vital
importance in the M--CO bonding. That is, the metal to carbon bonding
consists of two parts; 11 electron donation from the carbon to the transition
metal and rr electron back-donation from the transition metal to the carbon
(a) atom. This bonding is pictured in Fig. 2.1. Either bonding mechanism, on
its own, would lead to charge buildup on the metal (11) or the carbon (rr);
by acting together, the resultant charge buildup is small and each charge
transfer can proceed further than wo uld be possible in the absence of the
other. One talks of synergic bonding (synergismus is Greek for 'working
together').
Whenever one learns that a ligand bonds strongly to a transition metal
ion but much more weakly to a main group element, the involvement of
some rr-mediated back-bonding synergic mechanism is likely. Indeed, such
ligands are usually referred to as rr bonding ligands, not only because of
the existence of this mechanism but also because of its constancy, the
bonding always involving ligand empty rr orbitals of one sort or another.
In contrast, there can be more variability in the 11-bonding mecha nism.
Thus, for the latter the electrons involved can be those that are associated
with a chemical bond rather than with a particular atom. An excellent
example is provided by the ethene (ethylene) molecule, C 2 H 4 . Its mecha n-
ism of bonding (the usual pattern of 11 donation a nd rr acceptance) is shown
(b) in Fig. 2.2. The orientation of the bonded molecule is such that it is clear
tha t its orbital involved in 11 bonding is the C=C rr bonding orbital (which,
Fie. 2.2 (a) The u donation C2 H4 n ~ metal
empty orbital. (b) The rr back-donation of course, is occupied). The rr antibonding orbital is also that associated
metal ___. C2 H4 n antibonding orbital. with the C=C rr bond (a nd which, of course, is empty in the isolated ethene
molecule). If ethene seems strange as a ligand it is salutary to learn that the

Table 2.6 Some of the relatively simple
ligands which occur in the organometallic first compound containing ethene n-bonded to a metal was prepared
chemistry of transition metals. Ugands (largely accidentally) by Zeise, a Danish chemist, in 1827. One of the two
such as CO frequently occur along with compounds he prepared, K[Pt(C 2 H 4 )Cl 3 ], bears his name to this day-it
such ligands is almost invariably referred to as Zeise's salt.
In Table 2.6 are listed a few of the simpler n bonding ligands and in
Table 2.7 some of the species in which they are found. The bonding in
these compounds will be covered in more detail in Chapter 10, but, for the
moment, one important point suffices. This arises from the fact that the u
donor-n acceptor model of ligand-metal bonding involves different metal
RC5CR orbitals. It must, because, seen from the ligand, the metal u and n orbitals
t have different nodality. This will be true for each n bonding ligand in the
compound. The maximum number of such ligands which can be attached
Alkenes Alkynes will be limited by the availability of suitable metal orbitals-either empty
orbitals into which to donate or filled orbitals from which to accept.
Eventually, we will run out of metal orbitals. Now, the valence shell orbitals
of the (transition) metal are nd, (n + 1)s and (n + 1)p (for first row
elements 3d, 4s and 4p). This is a total of 5 + 1 + 3 = 9 orbitals so it seems
But-1,3-<liene (cis-butadiene) likely that 18 electrons will be the maximum that will be involved in
bonding, one way or another. It is not clear whether or not this hand-waving
argument is generally valid; what is clear is that the vast majority of
complexes of n bonding ligands obey the so-called '18-electron rule'. A
book-keeping exercise of the number of electrons associated with the metal
in the final complex (be they metal-originating or ligand-originating) usually
leads to the number 18. In Table 2.8 this is demonstrated for some of the
complexes of Table 2.7, but, to maintain a balance, two exceptions are
1,5-Cyclooctadiene (cod) included.
The choice of ligands in Table 2.6 has been made rather selectively.
Although at the present time it would be difficult to find examples similar
to the exotic ligands of Table 2.4, there are many unusual ligands that
could have been included in Table 2.6. Thus, not all of the ligands in
low-valence state complexes are independently stable molecules. For example,
The allyl group, a 3-electron donor; complexes in which cyclobutadiene (C4 H 4 ) is coordinated to a transition
alternatively, it may be helpful to think metal are commercially available, yet organic chemists have sought for
of it as (allylr, a 4electron donor
generations to prepare this molecule. Further, in a sense, it would have been
useful to have included many of the ligands in Table 2.6 several times over.
©
This is because many of the ligands listed can bond in several different ways,
each way giving rise to different complexes. In that a classic ligand such as
Cl- can bond not only to one but also to two metal ions (a bridging chloride
ligand) or to three which lie at the corners of a triangle (a face-bonding
chloride ligand) it is not surprising that a ligand such as CO should do the
Benzene same. How CO differs is that, much more than Cl-, it seems to form an
almost continuous range of intermediate bonding patterns between these
three main types. A ligand such as C 5 H 5 may have one, two, three, four or
five of its carbons bonded to a transition metal; when there are five they
may not all be equally bonded, and so on. Again, the range is enormous.
The cyclopentadienyl group CsH., The existence of such a range strongly suggests that the energy differences
usually written Cp.lt is a 5-electron between the various arrangements may be small. It seems that this, indeed,
donor (some prefer to regard it as is so. There are several possible consequences. First, the thermal vibrations
C5 H5 and as a 6 -electron donor)
- - - - - - - - - - - - - - - o f the molecule may contain sufficient energy to enable such a ligand to hop
Table 2.7 Examples of complexes formed by some of the ligands in Table 2.6 and related ligands
©JI
The most famous complex of (1) A fulvalene complex (6)
"Q
cyclopentadiene, ferrocene
Fe
@
(C0) 2 Ru---Ru(CO),
A diphenylethyne
(7) Q~
w
(diphenylacetylene) complex
#
Dibenzenechromium; the corresponding (2)
cation [Cr(C6H6),j+ was made in 1919 but """c
C~'•
not recognized as such : ~l. . ~ _c--o
Cr ,,/co,
©
_....Co I C-....,
_....c I\ c o
' o c c I
I 1 o
~
0 0
d-
A flyover complex in which a
single hydrocarbon ligand
CH 2 Ph straddles two cobalt atoms
A complex of a cyclopentadiene derivative. (3) which are bonded to each
P(C6H5)3 is a ligand that plays a similar ,::::::- other. Each cobalt is also
H
bonded to an allyl group
role to CD __ I (these bonds are shown large
I dotted) and to a carbon atom.
Fe
CD/ ~~PPh3
A bisallyl palladium complex
(4)
'"~
G)= a
I 1,5.Cyclooctadiene (cod) (10)
~P"';r
Two of the compounds Fe( COb complex of platinum(Il)
LJ/ \r
formed by reaction of Fe(C0) 5
with alkynes
~ Cr
~
Some more exotic complexes
which, despite appearances,
are not significantly different
in kind from those above.
©
Cr
Table 2.8 Electron counts of complexes (1-10) in Table 2. 7. Although it is some-

times convenient to assign a formal charge to the metal when making these
counts, experience is that fewer mistakes are made if metal atoms are treated
as uncharged, if at all possible. This in tum means that ligands must also normally
be regarded as uncharged. So, C5 H5 is a five-electron donor. Note that, although
in the text there is talk of two metal-CO bonding mechanisms, it is only the a
donation from the carbon of the CO that increases the number of electrons
formally associated with the metal. So, CO and ligands such as phosphines which
are believed to behave similarly, are regarded as two-electron donors
Complex Metal Formal Number Ligands Number Total

charge of metal of ligand electrons
valence electrons m+l:l
electrons (/)
(m)
(1) Fe 0 8 2 C5 H5 2x5 18
(2) Cr 0 6 2 C6 H6 2 X 6 18
}
(3) Fe 0 8 C5 H5 R 4
2 co 2 X 2 18
P(C6 H5 ) 3 2
(4) Fe 0 8 C5 H4 0
3 co
4
3x2 } 18
3x3 }
(5) Fe 0 8 C6 Me 4 0 4
18
3 co
(6) Consider each Ru separately: the Ru-Ru bond means that each Ru gains one electron
from the other for electron counting purposes:
(7)
Ru 0 8
n electrons to each Co, or as forming two

C5 H4 R
Ru-Ru
2
c--eo
co LJ 18
Either the diphenylethyne (diphenylacetylene) ligand may be regarded as donating two

bonds (and thus contributing two
electrons also). Again, the metal-metal bond contributes one electron to the electron
count of each metal. Consider each Co separately:
(8)
Co 0 9 C2(CsHs)2
Co-Co
3 co LJ
Here, counting Co as Co 0 means that the allyl group is a three-electron donor. The
18
cobalt-carbon and cobalt-cobalt bonds each contribute one additional electron to each
Ll
cobalt:
Co 0 9 C3 H2RR'R"
Co-Co
18
Co-C
3 co
(9) Again, is it simplest if the pattern adopted in (8) is followed:
Pd 0 10 2 C3 H4 R 2 x 3 16
This is an example of a complex which does not follow the 18-electron rule; square
planar complexes of d8 ions usually deviate, giving 16 instead.
(10) In this final example it is difficult to avoid a formal charge on the metal because to
work with Pt0 would mean working with Br rather than Br-:
Pt 2+ 10 cod 2 X 2 } 16
2 Br- 2x2
Another d8 square planar complex that does not follow the 18-electron rule.
Some final comments 1 21
from one arrangement to another, perhaps equivalent to the first but perhaps
not, at quite a high frequency. Such molecular gymnastics of so-called
fluxional species have been the subject of extensive study, most notably by
NMR. This is a topic which will be covered in Section 14.8.
Second, in a chemical reaction it may happen that the most stable
orientation of a ligand in the starting material is not the most stable over
some parts of the reaction pathway, nor, perhaps, in the reaction product.
The third point stems from the second. When attached to a transition
metal, a ligand may be stable in a range of convoluted geometric orientations
that are not readily available to it as a free ligand. This may well mean that
some new, otherwise impossible, reaction pathways become available. Fur-
ther, their availability and nature are likely to be influenced by the other
ligands present, both for bonding and steric reasons. The consequence is that
the use of transition metal complexes in catalytic reactions has been an
enormous field of study, one in which it is of interest to study large numbers
of closely related compounds. With just the right, carefully tuned, catalytic
molecule one may hit the jackpot and become able to turn an otherwise
useless byproduct into a highly valuable compound! At a more realistic level,
organometallic compounds are the catalysts in several important industrial
processes. They do have a problem, though. If they involve an expensive
metal, platinum for instance, should catalyst recovery be difficult or incom-
plete, the cost of replacement catalyst may make uneconomic an otherwise
viable process. This is just one reason that heterogeneous catalysts-involving
surface complexes-are more important than homogeneous-involving dis-
solved complexes-unless the metal involved in the homogeneous catalysis
is cheap, the products volatile-and so easily separated-or the products
obtained of particular value. There will be more on this topic in Chapter 15.
2.3 Some final comments

The object of the present chapter has been that of introducing the reader to
something of the breadth of complexes which are there to be studied, without
going into great detail at any point. In such a survey it is difficult to avoid
neglect of the majority, but commonplace, and instead to emphasize the
novel. Some final comments will help to complete the picture without unduly
distorting it.
First, so far in this book we have given the impression that coordination
compounds are obtained by reaction between a ligand or ligands and a
metal ion or compound. We shall see in Chapter 4 that, whilst this is
commonly the case, alternatives are possible. In particular, a ligand coor-
dinated to a metal may undergo an organic-chemistry-type reaction with
another (organic) molecule, leading to an extension of the ligand with,
often, an increase in the number of atoms of the-enlarged-ligand coor-
dinated. Alternatively, two ligands independently bonded to the metal
may be linked by a similar reaction. Such reactions, in which ligands which
are held anchored to a metal ion undergo ligand-extending reactions at a
site remote from the metal ion, are called template reactions. Although this
particular discussion is relevant to classical complexes, it probably is even
more important for organometallics. In the early years of organometallic
chemistry it was commonplace to take a transition metal carbonyl such
as Fe(C0) 52 with some likely-looking organic compound-an alkene or

alkyne derivative, for instance--and to heat them together in an autoclave
for hours. Unholy mixtures of fascinating compounds resulted, some of which
have not been properly studied to this day. It is likely that the novel ligands
found in most, if not all, of the products resulted from template-type
building-up reactions involving organic molecules (and often CO, too)
coordinated to the metal. However, their formation is seldom regarded as
providing examples of template reactions. Template syntheses are planned
and usually of high yield; in contrast, the stabilization of an organic molecule
in a speculatively prepared complex is often adventitious and each compound
represents a minor component in a complicated reaction product mixture.
Secondly, we have tried to keep our examples simple, although hinting
at the existence of some points of a greater complexity. Thus, in recent
years there has been enormous study of compounds containing many metal
atoms, linked directly to each other, linked indirectly through ligands, or
both. So wide has this field become that there is real discussion about
whether or not compounds with perhaps 40 or 50 metal atoms or more,
directly bonded one to another, should be treated as fragments of a metal
as much as chemical molecules. Metal clusters will be the subject of Chapter
15 and metals in Chapter 17.
Finally, whilst the ideas about bonding in complexes so far presented
have been rather simple, it has already become clear that there are aspects
of real molecules that they do not even begin to cover. Chemists tend to
start with the usual and common and wonder what happens when new
constraints are added. A popular, and fruitful, line of current enquiry has
already been mentioned. What happens if groups attached to the ligating
atom are made very bulky? Will the increase in steric interactions lead to
a weaker metal-ligand bonding (which could mean interesting properties)
or to a low coordination number (which could also mean interesting
properties)? Generally speaking, the answer to both these questions is 'yes'
and so a need has developed to quantify in some way the steric demands
imposed by individual ligands-and none of the simple bonding theories
so far mentioned addresses this question. This point will be returned to in
Chapter 5 where an angle, the cone angle, will be described, which provides
a measure of steric elfects3 in some complexes, at least. The molecular
modelling to be described in Chapter 13 is one technique which aims to
give an accurate description of steric effects.
2 Note that, although this compound is a complex, the rule that complexes are written
in square brackets is often applied rather sloppily in organometallic chemistry.

3 See C. A. Tolman, Chern. Rev. (1977) 77, 313.
Further reading Much of the further reading suggested for Chapter I is

relevant for Chapter 2 also, although to include transition
The contents of this chapter have been rather broad-brush, metal organometallic chemistry, the browse through the intro-
stretching from the traditional through to the contemporary. ductory coordination chemistry section in general inorganic
Further reading in this vein can be obtained by thumbing texts will need to cover texts published from the 1960s through
through any contemporary text in inorganic chemistry or, to to the early 1980s. More recent texts tend to include material
come really up-to-date, any current issue of any journal relevant to the later chapters of the present book but which
devoted to inorganic chemistry. has only briefly been alluded to in Chapter 2.
Questions 1 23
Behind the content of this chapter, but not discussed within • 'Powerful new metal chelating agents developed' in Chern.
it, is the way that the information was obtained. Sometimes, Eng. News (August 1st 1988) 21.
there is a helpful story to be told. Two that may be mentioned • 'Molecules with Large Cavities in Supramolecular Chem-
are the discovery of the first complex containing coordinated istry' by C. See! and F. Viigtle, Angew. Chern., lilt. Ed. (1992)
N 2 : 'The Discovery of [Ru(NH 3 ) 5 N 2 ] 2 - ' by C. V. Senoff. 31, 528.
J. Chern. Educ. ( 1990) 67, 368 and, rather older. the nature
• 'Calixranes-supramolecular pursuits' by A. McKervey and
of the species giving rise to the intense blood-red colour
V. Bohmer, Chemistry in Britain (1992) 724.
obtained when thiocyanate, seN-, ions are added to iron(Ill)
solutions, 'The nature of iron(III) thiocyanate in solution' by • 'The Specification of Bonding Cavities in Macrocyclic Lig-
ands' by K. Hendrick and P. A. Tasker, Prog. Inorg. Chem.
S. Z. Lewin and R. S. Wagner, J. Chon. Educ. (1953) 30. 445.
Although both make reference to ideas developed later in (1985) 33, 1.
this book, this problem should not unduly inhibit under- • Supramolecular Chemistry, an Introduction by F. Viigtle, 1.
standing. Wiley, Chichester, 1991.
There are many sources of information on the exotic
A Nobel lecture by the major authority in the area, which
ligands discussed in the text. Samples which give easy-to-read
spans a wide range of topics from the history of the subject
insights are:
through to its applications in catalysis, photochemistry and
• 'Coordination Chemistry of Alkali and Alkaline-earth cat- biochemistry is 'Supramolec'.l!ar Chemistry-Scope and Per-
ions with Macrocyclic ligands' by B. Dietrich, J. Chern. spectives, Molecules, Supramoleculcs and Molecular Devices'
Educ. (1985) 62, 854. by J-M. Lehn, Angew. Chern., Int. Ed. (1988) 27, 89.
Questions • 18-crown-6
• 21-crown-7
2.1 Use Table 2.2 to compile a list of atoms (which may be
• dibenzo-12-crown-4.
part of a polyatomic ligand) which coordinate to metal ions
and relate the list to the region of the periodic table where 2.5 Cryptand ligands have been prepared in which up to
the atoms fall. If in doubt about whether an atom can half of the oxygens in 2,2,2-crypt (Table 2.4) have been
be a ligand, the answer is almost certainly 'yes', although even replaced by either S or NCH 3 . Suggest some ligands of this
so the final list will be far from comprehensive. type which might well have been studied.
2.2 Unlike Table 2.2, Table 2.3 contains no examples of 2.6 Select any three of the complexes shown in Table 2.7
ligands in which a halogen atom is the donor atom. Suggest and carry out a valence electron count on them. Compare
a reason why such species are rare. your results with those given in Table 2.8 (if any of the last
three complexes in Table 2.7 were chosen some thought will
2.3 Although Table 2.2 contains examples of ligands in which be needed in using Table 2.8 ).
nitrogen or arsenic are donor atoms there are no examples
which contain phosphorus. Suggest a few possible phosphorus- 2.7 As Table 2.6 hints, complexes are known in which 1,5-
containing ligands (if you wish, some of the ligands in Table cyclooctadiene (cod) bonds through both of its double bonds
2.2 might be appropriately modified). to a single metal atom. It may be that Fig. 2.2 can be
applied to each double bond separately. Equally, it could be
2.4 Using the ligands in Table 2.4 as a guide, suggest probable argued that although this approach may be valid for the rr
structures for: back-bonding, it is inadequate for the (J donation from the
• 15-crown-5 cod. Suggest a reason for this reservation.
Nomenclature, geometrical
structure and isomerism of
coordination compounds
3.1 Nomenclature
In order to facilitate communication between chemists it is desirable that a
convention for naming coordination compounds be followed. This section
contains an outline of the system suggested by a Nomenclature Committee
of the International Union of Pure and Applied Chemistry (IUPAC). 1
Although this convention is commonly adopted-the Russian literature
contains some variants and each language uses its own words, or spelling
(as in 'sulphate' and 'sulfate' for UK and USA, respectively, although the
UK has recently agreed to adopt the American version)-it is often simpler
to give a structural formula, e.g. [Co(NH 3 ) 4 CI(N02 )]+, than to write the
name in full and this will frequently be done in the following chapters. Apart
from this device for side-stepping the problem, in this book we generally
follow the most recent IUPAC recommendations except for those which
have yet to gain general acceptance.
The IUPAC system has both advantages and disadvantages. One major
advantage is that the most recent recommendations are just that, recom-
1 Perhaps the most generally available complete system is the American version because it
is contained in the Handbook of Chemistry and Physics published annually by the CRC
Press and which can be found in most libraries. Inevitably, there is a time lag before the most
recent recommendations appear so it is as well to check on this. The greatest detail and the
most current recommendations have been published in Nomenclature of Inorganic Chemistry,
Recommendations /990 (IUPAC) ed. G. J. Leigh, Blackwell, Oxford and it is these that have
been followed in the present text. At the same time as the IUPAC book appeared so, too, did
another, Inorganic Chemical Nomenclature by W. H. Powell and W. C. Fernelius, published by
the American Chemical Society (the publication is the outcome of the deliberations of this
Society's nomenclature committee), 1990. Whilst there are minor points of disagreement with
the IUPAC book, with its greater emphasis on the notations that will be found in the older
literature, the ACS publication may be seen as an acceptable complement to it.
Nomenclature 1 25
mendations. Previously, they had been 'rules', a situation that led one
group of writers to comment 'such rules often represent a compromise of
conflicting views and may not be completely acceptable (or convenient) to
all'. In what follows, in the spirit in which they are presented, the current
recommendations are not regarded as above criticism.
Having mentioned one advantage, let us get the disadvantages out of
the way. Names are often quite long, and involve numbers, brackets,
subscripts, superscripts and Greek letters. When reading a name all of these
details have to be remembered because it is not until the name is completed
that one can really start building up the complex. The name of the metal
is given at the end, whereas the thinking of most chemists starts with the
metal. It is not surprising that 'nickel tetracarbonyl' is in much more
common usage than the IUPAC name 'tetracarbonylnickel(O)'. So, whilst
the IUPAC notation is that generally encountered in the scientific litera-
ture, when chemists talk to each other they either greatly simplify it or use
one that has grown up in their specialist field. It is therefore only to be
expected that trivial names persist, so, although one should talk of the
hexacyanoferrate(II) and hexacyanoferrate(III) anions, most workers bow
to common usage and continue to speak of them as ferrocyanide and
ferricyanide, respectively. The practice of naming what has at some time
or other proved an important coordination compound after the person who
first prepared it is widely followed, thus: NH 4 [Cr(NH 3 h(NCS) 4 ], Rein-
ecke's salt; [Pt(NH 3 ) 4 ][PtC1 4 ], Magnus's green salt; [IrCOCl(P(C 6 H 5 h) 2 ],
Vaska's compound; [RhCl(P(C 6 H 5 hhJ, Wilkinson's compound and, one
met in Chapter 2 K[Pt(C 2 H 4 )Cl 3 ], Zeise's salt. A system which has
mercifully disappeared, but which will be found in the very old literature,
is that of naming a compound according to the colour of the corresponding
cobalt(III) complex (no matter the colour of the complex itself!). Thus,
'purpureo' salts meant ions with the general formula [M(NH 3 ) 5 Cl]"+.
The following rules summarize the more important recommendations
of the IUPAC committee and contained in their 1990 publication. It should
be recognized that when these differ from the previous rules the earlier
version is the one likely to be met by the reader~the new recommendations
are only just being accepted and used.
Although in writing the formula of a complex the central atom is given
first, in the corresponding name it is given last, thus [Fe(CN) 6 ] 3 -, hexa-
cyanoferrate(III). For anionic complexes the characteristic ending is '-ate'
(as hexacyanoferrate(III)), but for neutral or cationic complexes the name
of the central element (normally a metal) is not modified: [Fe(H 2 0) 6 ] 2 +,
hexaaquairon(II). A distinction is made for anionic complexes so that the
corresponding acids can be systematically named, the characteristic ending
for the acid being '-ic', as for H 4 [Fe(CN) 6 ], hexacyanoferric(II) acid. As
indicated in these examples, the formal oxidation state of the central atom
(Werner's primary valency) is indicated by a Roman numeral in paren-
theses after the name of the complex, but with no space between them. A
formal oxidation state of zero is indicated by (0) and a negative state by
a minus sign, e.g. (-I).
The name of the complex species is written as one word. Ligands which
may be regarded as carrying a negative charge (Cl-, SO;i- etc.) all end in
'-o' (chloro, sulfato etc.). Whereas, previously, the negatively charged
26 1 Nomenclature, geometrical structure and isomerism of coordination compounds
ligands preceded uncharged ligands, now ligands are listed in alphabetical

order, without regard to charge. Prefixes such as di-, tri-, bis-, tris- (see
later) are disregarded in determining the alphabetical order. Previously,
ligands in each class (anionic, neutral) were written in order of increasing
complexity. A problem with an alphabetical order arises when different
languages are used-the name can become language dependent. In prac-
tice, since English has become the international language of science, the
problem is not severe.
A different pattern is recommended for writing the formulae of com-
pounds. The central atom is written first (in contrast to its name, where it
comes last), followed by the anionic ligands in alphabetical order and then
the neutral ligands, again in alphabetical order. Polyatomic ligands are
enclosed in parenthesis (otherwise there could sometimes be confusion
about which atom belongs to which ligand). This latter rule also applies
when abbreviated symbols are used for ligands (e.g. (py) for pyridine). Some
ligands have names dictated by common usage: H 20 aqua, CO carbonyl,
NH 3 ammine, NO nitrosyl. Some names which have been used in the past
are no longer recommended. One is aquo (which has been replaced by
aqua), another is mercapto for (SH)-. In future sulfanido should be used
instead, but it remains to be seen whether this gains common acceptance.
Other names can beg questions. Thus, when an H atom coordinates it is
regarded as H-, and so named hydrido, even if there is experimental
evidence that the ligand behaves chemically more as H + than as H-. When
there are several different possible attachment modes of a ligand, for simple
cases the attached ligand can be indicated by a bond:
(-DNO)-, nitrito-0; (-NO,)-, nitrito-N
(although nitro is also acceptable for this ligand, N-coordinated).

For more complicated ligands a Greek kappa (K) is placed before the
coordinated atoms(s). Thus, although one would normally write thio-
cyanato-N for (-NCS)-, it could be called thiocyanato-KN. Similarly,
(-SCN)- could be called either thiocyanato-S or thiocyanato-KS. Ligands
which commonly coordinate to one coordination position but can do so
in more than one way, and we have just met two simple examples, are
called ambidentate ligands-more examples of the use of the word dentate
will be given shortly; it was first mentioned in Chapter 2. Ligands which
bond to more than one coordination site and can do so in more than one
way are termed flexidentate. When several identical ligands are coordinated
to the same central atom, two cases arise. If the ligand is simple and with
a simple name, the number of ligands is indicated by the appropriate prefix:
di-, tri, tetra-, penta-, or hexa-. Several examples of this usage have already
been given. When the ligand is so complicated that it has a polysyllabic
name-perhaps already including one of the above prefixes-or ambiguity
might arise for some other reason, the name is enclosed in parentheses and
the number of ligands present indicated by the prefix bis-, Iris-, tetrakis-,
pentakis-, or hexakis-. Examples are ethylenediamine,2 NH 2-CH 2-CH 2-
NH2, which gives rise to the complex tris(ethylenediamine)nickel(II),
2 The correct name for this ligand is 1,2-diaminoethane, a name which is being increasingly
used. However, in the literature and in texts ethylenediamine is the name which is usually
encountered and for this reason is used in the present book also.
Nomenclature 1 27
FIC. 3.1 The complex [Rh 4 (C0) 4 (~ 4 -P(C 6 H 11 )

The rhodium atoms are
(~·P(C 6 H 11 ),) 4 ].
indicated in black. The carbon atom of each
cyclohexyl group, C6 H11 , that is attached toP is
the only one shown and is drawn as an open
circle.
[Ni(enh]2+ (en is the usual abbreviation for ethylenediamine), and tri-

phenylphosphine, P(C 6 H 5 lJ, in the complex [NiC1 2 (P(C 6 H 5 hhJ, dichloro-
bis(triphenylphosphine)nickel(II). A name which is increasingly being used
to describe complexes in which all the ligands are identical is to say that they
are isoleptic. So, [Ni(enlJ]2+ is an isoleptic complex but [Ni(P(C 6 H 5 lJhC1 2]
is not
Bridging groups attached to two distinct coordination centres are indicated
by the prefix J1 (mu):
4
NH 2 ] +
[ (NH 3 ) 4 Co~NO.,~Co(NH 3 ) 4
,u-amido-,u-nitrobis(tetraamminecobalt(lll))
When more than two metal atoms are spanned by a single ligand then
the number of centres spanned is indicated by a subscript: J1 3 , J1 4, Jls
(although the reader should be warned that some authors prefer to use
superscripts: J1 3 , 11\ etc.). In such cases, since each metal usually has its
own set of ligands, the formal name can stretch over several lines and, not
surprisingly, its use is avoided; a formula is given instead. So, the molecule
shown in Fig. 3.1 , in which P(C 6 H 11 h groups bridge four Rh(CO) units
which themselves lie at the corners of a square, the whole being capped
by a P(C 6 H 11 ) group, is written [Rh 4 (COMJ1 4 -P(C 6 H 1 d(J1-P(C 6 H 11 ) 2 ) 4 ],
although, written like this, the 1:1 relationship between the Rh and CO is
not explicitly stated. Note that J1S appear in sequence of decreasing suffix
values, although when a ligand is both bridging and non-bridging the
non-bridging is listed first (e.g. Cl - as both bridging and terminal ligand).
Not only can a ligand be attached to more than one metal atom, a given
ligand may be bonded through several atoms to a single metal. This is
described by the prefix '1 (eta), with a superscript to indicate the number
of atoms bonded. One talks of the hapticity (Greek: hapto to fasten) of the
ligand, thus: '7 1 monohapto, '7 2 dihapto, '7 5 pentahapto etc. This nomen-
clature is particularly important in organometallic chemistry, where, as
mentioned in Chapter 2, a given ligand may often coordinate in a variety
of ways and jump from one mode to another. So, the compound in Fig.
3.2 is tricarbonyl('1 4 -cyclobutadienyl)iron(O). Although its name would
lead one to adopt the representation shown in Fig. 3.2(b), for simplicity
that given in Fig. 3.2(a) is usually preferred, unless a specific point has to
be made (for instance, if, in a reaction, one of the four Fe-C bonds breaks).
A detailed discussion of the bonding in such compounds will be given in
Chapter 10. The above paragraphs have been written with the intention
of providing a comprehensive overview, not detail. The latter is contained
in Tables 3.1-3.3 which give specifics and examples.
Names such as those we have given above indicate the ligands attached
to a central metal atom but do not detail the positions of the ligands
relative to one another. If this information is important and has to be
included, an extension of the nomenclature is necessary. It often happens
that the prefixes cis or trans are adequate, as in the complex cis-dichloro-
FIJI. 3.2 Alternative symbolic representations of di(pyridine)platinum(II) in which the platinum and the four atoms co-
the bonding between C4 H4 and Fe(CO)a groups.
In (a) the delocalized nature of the " electron ordinated to it are coplanar:
system of the C4 H4 group is emphasized. In (b)
Cl py
the ~· aspect of the bonding is made evident.
"'pt/ (py = pyridine)
Cl/"' py
The most recent IUPAC recommendations on this point represent a
considerable departure from those previously made. We first outline the
new and then review the old, for as already mentioned, it is the latter which
Table 3.1 Anions and their names will be encountered in all but the most recent literature and texts.
when acting as ligands
Free anion Coordinated anion Table 3.2 Examples of the nomenclature of simple coordination compounds. Some
Amide (NH:2) amido (or azanido) of these examples contain, and adequately define, points not explicitly covered in the
Azide (Nil n~rido (azido will also text
be met)
Compound Nomenclature
Bromide <Bn bromo
Carbonate (Co~-) carbonate K,[ReF8 ] potassium octafluororhenate (note: only 'potassium')
Cyanate (CNO-) cyan ato [Cu(NH 3 ) 4 ]S04 tetraamminecobalt(ll) sulfate (note: 'aa' and 'mm')
Fluoride (F-) fluoro (not fluo) [CuCI2(py),] dichlorobispyridinecopper(ll) (note: bipyridine is the present name for the
Hydroxide (OW) hydroxo (or hydroxide 2,2'-bipyridine ligand-see Table 2.3. More strictly, and as in the text,
or hydroxy) di(pyridine) should be used to give dichlorodi(pyridine)copper(ll). How-
ever, in the spoken language an ambiguey can arise)
Nitrite (N0:2) nitro or nitrito-N
(see text) [Hg(C2Hsl2l diethylmercury(ll)
Oxide (02 -) oxo (or oxido) [Ni(PPh 3 ) 4 ] tetra(triphenylphosphine) nickei(O)
Thiocyanate (SCN)- thiocyanato-N [Ru(NH 3 ) 5 (N 2)] 2 + pentaamminedinitrogenruthenium(ll) (note: similarly, 0., is dioxygen, but
beware confusion w~h 0:2, superoxo and o~-, peroxo)
(N-bonded),
thiocyanato-S K,[FeC1 4 ] potassium tetrachloroferrate(ll)
(S-bonded) (NH 4 ),[SnCI6 ] ammonium hexachlorostannate(IV)
Nomenclature 1 29
Table 3.3 The nomenclature of compounds containing bridging groups
di-p-chloro-bis[diammineplatinum(II)l chloride
p-amino-p-peroxo-bis[tetraamminecobalt(lll)l perchlorate
bis[carbonyl(u-carbonyl)'l5-cyclopentadienyliron(0)1
(one could add trans at the front as the cis form also exists)
,...,cH2 ,...c1 .- c1
CH' -Pd 'AI
''cH2 'c( 'c1
rfl-allylpalladium(II)di-p-chlorodichloroaluminium(III)
Each symmetry-distinguishable geometry about the central metal is

given a so-called polyhedral symbol. The symbol is based on the name for
the geometry and these are discussed later in this chapter: OC for octahedral,
T for tetrahedron, SP for square planar, SPY for square pyramid and TBPY
for trigonal bipyramid are the most important. Some of the others are
contained in Question 3.9. Because there are cases where ambiguity would
otherwise arise, the coordination number is given immediately after the
polyhedral symbol: OC-6, T-4, SP-4, SPY-5, TPBY-5. Then, ligating atoms
are assigned a priority, the highest atomic number having highest priority,
the lowest atomic number the lowest. When the ligating atoms are identical
but with different substituents, the same rule is applied to the substituents.
So, -N02 has a higher priority than -NH 3 , but -DH- has a higher priority
than either. Consider complexes with the two geometries shown in Fig. 3.3.
They are distinguished as follows (the general case is similar). Consider the
ligand of highest priority (add the words 'which lies on the axis of highest
FIJI. 3.3 (a) The complex
(OC-6-22)-triamminotrinitrocobalt(lll)~this is rotational symmetry' if ambiguity persists). Write down the priority number
commonly referred to as the fac (facial) isomer, of the ligand trans to it. Here, -N02 is of highest priority and has priority
i.e. fac-triamminotrinitrocobalt(lll). (b) The
number I and -NH 3 has priority number 2 (had the complex contained
complex (OC-6-21)-triamminotrinitrocobalt(lll)~
this isomer is commonly referred to as the mer -DH- these two numbers would have been 2 and 3). So, both the isomers,
(meridional) isomer. so far, have the symbol (OC-6-2). This is because for that in Fig. 3.3(b), of
the two axes containing -N02, the rules require that we choose that axis
for which the priority number difference is a maximum. We now move to
the plane perpendicular to the axis already chosen and again select the ligand
of highest priority number. We again write down the priority number of the
ligand trans to it. So, the two isomers above become (OC-6-22) and
(OC-6-21), respectively, and they are distinguished. This completes the
process so that the isomer in Fig. 3.3(a) is called:
(OC-6-22)-triamminetrinitrocobalt(lll)
Optical isomers are not distinguished by the notation so far presented but
it can be extended to cover this requirement.
In the past, the system adopted has been to number the coordination
positions. The numbering system adopted for square planar complexes was
1
I
4~M-2
I
3
so that an alternative to cis-dichlorodi(pyridine)platinum(II) is 1,2-di-
chlorodi(pyridine)platinum(II). Notice the brackets around (pyridine). This
practice is recommended and in the present case serves to remove any
confusion with the ligand 2,2'-bipyridine (previously dipyridine). For octa-
hedral complexes the cis and trans nomenclature is often simplest, but
for complicated cases the numbering systems shown in Fig. 3.4 has been
adopted. The ligands which are at the corners of the front face of the
octahedron are cyclically numbered 1 --+ 3 and those on the back face
numbered 4 --+ 6. There are just two ways of arranging two sets of three
identical ligands, [ML 3 L3], in an octahedral complex. When identical
ligands lie at the corners of a face of the octahedron, one has what
traditionally has been called the facial-denoted fac-arrangement (Fig.
3.3(a)). When the three identical ligands lie on a plane that bisects the
octahedron, the meridional-denoted mer-arrangement results (Fig. 3.3(b)).
As has been recognized, two atoms of the same ligand may coordinate
to the same metal, leading to the formation of a ring structure. As
1 mentioned in Chapter 2, the formation of such rings by coordination is
termed chelation and the ligand is called a chelating ligand. Historically,
the language used to describe chelates depended on which side of the
Atlantic you lived. Traditionally, Americans talked of bidentate, tridentate,
tetradentate, pentadentate and hexadentate, the whole class being called
polydentate ligands (ligands attached by a single atom being monodentate).
British textbooks preferred bidentate, terdentate, quadridentate, quique-
dentate and sexadentate, calling the whole set multidentate ligands; single
atom attachment being unidentate. However, the latest IUPAC recom-
mendations are for didentate, tridentate, tetradentate and so on. Whether
didentate will replace the one name common to both American and British
notations-bidentate-remains to be seen. In this book the American
6 usage will be followed, this being the more commonly met. (This notation
Fig. 3.4 Ugand numbering system for an was used in Table 2.3.) A ligand which coordinates four atoms to each of
octahedral complex, indicating how the
numbering is related to the selection of two
two metals is called bisquadridentate, thus combining both notations, and
opposite faces of the octahedron. so on. Note that a polydentate ligand is not necessarily a chelating ligand
Coordination numbers 1 31
(although commonly it is). because the coordinating atoms of the ligand

may be so arranged that they cannot be coordinated to the same metal
atom. So, a polydentate ligand it is not necessarily a chelating one. The
example of multidentate isocyanide ligands, which have to be rather special
if they are to coordinate to a single metal atom, was mentioned in Section
2.1.
Ligands, and polydentate ligands in particular, can be rather com-
plicated. To simplify the description of complexes involving such ligands,
abbreviated forms of the ligands' names are used-we have already used
en as a shorthand for ethylenediamine and py for pyridine. Although there
are IUPAC recommendations, these are not always followed and, in
practice, there is only an approximate system of standard abbreviations.
Fortunately, it is normal for authors of papers to define the abbreviations
they use, so little confusion arises. A representative selection of such ligands
was given in Table 2.3 together with commonly used abbreviations.
There are two other areas covered in this book for which notations are
needed. These are in the fields of cluster compounds and of bioinorganic
chemistry. Both commonly involve very complicated molecules and
attempts to produce systematic names such as those met in this chapter can
defeat the whole purpose--to make communication easy. In both cases it
is difficult to separate the notations currently used from developments in
our understanding of the topic. Discussions of nomenclature will therefore
be given in Chapters 15 and 16 rather than introduced here.
So far our concern has been entirely with a ligand and its attachment
to a metal atom. It has been assumed that we could ignore atoms of the
ligand that are not bonded to the metal. In some cases this is not justified.
The reason is that, for chelates in particular, there is more than one
geometric arrangement possible for the ligand atoms between those bonded
to the metal. The subject of the conformation of chelate rings and the
consequences of the various conformations has been much studied. It is
an interesting and elegant subject area; a brief introduction to it is given
in Appendix I.
3.2 Coordination numbers

Werner was the first to recognize that one characteristic of a coordination
compound is the number of ligands directly bonded to the central atom.
He called this number the secondary valency of the central atom, but this
usage has not persisted and it is now called the coordination number. The
coordination number need not have a unique value for a particular metal
(b)
ion. For example, in pink cobalt(II) chloride it is six and in the blue form
Fig. 3.5 When cobalt(ll) salts dissolve in water
it is four (see Fig. 3.5). In particular, when a coordination compound is
they commonly give pale pink solutions,
normally associated with the octahedral participating in a reaction in which one ligand is being replaced by another,
hexaaqua species (a). Addition of concentrated there is overwhelming evidence that the coordination number in the
hydrochloric acid gives a deeper, beautiful blue
reaction intermediate is different from that in either the initial or product
colour, 'cobalt blue', normally associated with
the tetrahedral tetrachloro species (b). In fact, compound.
the colours are perhaps more associated with Coordination number is more than just a convenient method of classifying
the two different geometries than
coordination compounds-complexes of a given transition metal ion with
ligands-replacing an H2 0 by Cl- in (a), for
example, does not greatly change the observed the same coordination number often also have closely related magnetic
colour of the solution. properties and electronic spectra. In later chapters of this book, these
properties are discussed in some detail. In the present section are described
some of the geometrical arrangements of ligands which have been found for
various coordination numbers and examples of each are given.
First, however, we set the scene by noting that the frequency of occurrence
of coordination numbers for some ions of the first transition metal series is
roughly as follows:
chromium(lll) 6(oct) » 5, others very rare
iron( Ill) 6(oct) > 4(tet) > 5"' 7
cobalt(ll) 6(oct) > 4(tet) > 5 > 4(planar)
cobalt(lll) 6(oct) »> 5 > 4
nickel (II) 6(oct) > 4(planar) > 4(tet) "' 5
copper( II) 6(oct)t > 4(planar) > 5t "' 4(tet)t
t usually distorted
Although not exhaustive, this series illustrates the fact that there i& no fixed
coordination number for any ion. It cannot be emphasized too strongly
that the empirical formula of a compound often has little connection with
either the coordination number or geometry of any complex species it
describes-this recognition was Werner's breakthrough.
3.2.1 Complexes with coordination numbers one, two

or three
Coordination numbers of one, two and three are rare. Although 1t 1s
possible to conceive of complexes with coordination number one-- presum-
ably a very bulky ligand with only one coordinating atom tucked well into
its centre might form such complexes-it is only recently that two have been
reported. The compounds are actually organometallic complexes of copper
and silver with a ligand which has three phenyl groups symmetrically
attached to a central phenyl, which is bonded to the metal through another
carbon of the phenyl ring: 2,4,6-triphenylphenylcopper and -silver.
The best-known example of coordination number two is the complex
ion formed when silver salts dissolve in aqueous ammonia, [Ag(NH 3 h]+.
This, like all other known cases of this coordination number, is linear,
[H 3 N-Ag-NH 3]+ although it should be possible to obtain bent examples,
perhaps when two different but sterically-demanding ligands are involved.
Other complexes of this coordination number, which is almost entirely
confined to copper(!), silver(I), gold(l), and mercury(II), are [CuCI 2 r and
[Hg(CNh]. One way of reducing the coordination below the normal for
a particular metal ion is to choose ligands that are so bulky that they block
the entry of further ligands. So, the ligand P(C 6 H 5 h causes zerovalent
Pt-usually either three- or four-coordinate--to have coordination num-
ber two in Pt(P(C 5 H 5 hh, where the P-Pt-P sequence is linear.
Examples of coordination number three are few, the [Hgi 3] - anion
perhaps being one of the best characterized. In this anion the iodide ions
are arranged at the corners of a slightly distorted equilateral triangle which
has the mercury atom at its centre. In the anion [Sn 2 F 5 ] - two SnF2 units
are bridged by the fifth F, leading to a (distorted) three-coordinated
structure around each tin atom. Examples of three-coordination in transi-
Coordination numbers I 33
tion metal chemistry are the iron(III) complex [Fe(N(SiMe 3 hhJ-the

chromium compound is similar-and the complex formed when copper(!)
halides dissolve in aqueous solutions of thiourea, [Cu(SC(NH 2 hhl Although
these mostly involve a planar coordination around the metal ion, in some
examples the metal ion is slightly out of the plane.
The [AgC1 3 ] 2 - anion has recently been found to have a perfectly planar,
equilateral D3 h structure. It provides an example of how large ligands can
lead to interesting structures. The anion occurs in the complex dibenzo-
18-crown-6-KCl· AgCI (the ligand is that shown in Table 2.4). In the crystal
the ligand complexes the potassium, the resulting complex resembling a
slightly buckled wheel with the potassium at its centre. Perpendicular to
the wheel is a Cl- of the [AgCl 3 ] 2 - anion, coordinated to the potassium.
As a result, three of the wheels surround the anion (Fig. 3.6); presumably
steric interactions between the wheels impose the threefold symmetry in
which the bulk of the wheels prevents any other ligand having access to
the silver ion and thus increasing its coordination number.
FJg. 3.6 The complex formed between

dibenzo-18-crown-6-KCI, a complex of
potassium, and AgCI which contains an
[AgCI3 ] 2 - anion. The potassium atoms are
shown bonded to the oxygens of the crown
ether, the other atoms are carbons. The
ellipsoids give an indication of the vibrations of
individual atoms. If the ellipsoids associated
with the chlorine atoms, for example, were very
large (which they are not) one might have
doubts about whether the equilibrium geometry
of [AgCI3 ] 2 - is equilateral triangular
(reproduced courtesy of Prof. S. Jagner).
3.2.2 Complexes with coordination number four

A tetrahedral arrangement of ligands is commonly exhibited by complexes
with coordination number four. It is found for both transition metal and
non-transition elements; for the latter it is rather common. In Chapter I
the two species [BF3 · (NMe 3 )] and [BF4 ] - were mentioned, in both of
which the boron is tetrahedrally coordinated; other examples amongst
main group elements are the [BeF4 ] 2 -, [ZnC1 4 ] 2 - , and [Cd(CN) 4 ) 2 -
anions. Complexes of transition metals in their higher oxidation states are
often tetrahedral and often also anionic-TiC1 4 , [Cr04 ) 2 - and [Mn04 F-
are examples- but the same geometry is found for other valence states
also. Transition metal chlorides, for instance, quite often give tetrahedral
anionic species when dissolved in concentrated hydrochloric acid: iron(III)
chloride gives the yellow ion [FeC1 4 r and cobalt(II) chloride gives the
well-known blue ion [CoC1 4 ) 2 -.
The four-coordinate arrangement in which the ligands lie at the vertices
of a square (square planar complexes) is almost entirely confined to
transition metal complexes (but XeF4 also has this structure), where it is
common and dominant for ions of the second and third transition series
having dB configurations- rhodium(!), iridium(!), palladium(II), platinum(II)
r
and gold(III). Examples are the [PtC1 4 ] 2 - , [PdC1 4 ] 2 - and [AuF4 anions.
Nickel(II), also a dB ion, is interesting in that it forms both tetrahedral and
square planar complexes. (The red precipitate obtained in the gravimetric
analysis of nickel salts and made by adding dimethylglyoxime to nickel(II)
solution is a planar complex of Ni 11 • Square planar complexes of this ion are
often yellow, orange, brown or red.)
As Fig. 3.7 shows, it is possible in principle, if not in practice, to distort
a tetrahedral arrangement of ligands so that they eventually assume the
square planar structure, and vice versa. This suggests that complexes may
exist with structures which are neither tetrahedral nor square planar, but
intermediate between the two. Indeed, such is the case with the [CuC1 4 F-
anion, made by dissolving CuC1 2 in concentrated hydrochloric acid. How-
ever, caution is needed because symmetry arguments can be invoked
FIJI. 3. 7 At least in principle. by a smooth,

coupled motion of all ligands (that motion
indicated by the arrows) it is possible to change
a square planar complex into a tetrahedral one.
It is therefore not surprising to find that
complexes with intermediate geometries do
indeed exist.
which suggest that tetrahedral:;:::::: square planar interconversions are

unlikely.
A related, but rather different, structure (Fig. 3.83 ) is found for some
main group compounds, for example [SbC1 4 ] - and [AsC1 4 ] - . The co-
ordination geometry here may be regarded as derived from trigonal hi-
pyramidal five-coordination (Fig. 3.9 and Section 3.2.3) by omitting one
of the three equatorial ligands. It has been suggested that the empty
coordination position of the trigonal bipyramid is not really vacant but is
occupied by a non-bonding lone pair of electrons of the central atom
(Sb, As).
Fig. 3.8 The mode of fourfold coordination This does not exhaust the list of possible four-coordinate geometries. A
(C2 v) which seems to be related to the trigonal trigonal monopyramidal, C3 "' arrangement is known when it is imposed
bipyramidal mode of five-coordination (Fig. by the ligand. So, complexes of Mm (M = Ti, V, Cr, Mn and Fe) are known
3.9(a)). The distortion of the ligands away from
colinearity can be understood in tenms of for the ligand R 3 N, where R- is (t-BuMe 2 Si)NCH 2 CH 2-, all four nitrogens
VSEPR theory (Appendix 2). being coordinated. The three bulky (t-BuMe 2 Si) groups block the second
axial position; the formal negative charge carried by the N in R- means
that the complexes, [R 3 NM], are electrically neutral and so offer little
attraction for most ligands. Ligand-induced geometries exist for the other
coordination geometries that now follow but, in general, will not be
included in the discussion.
3.2.3 Complexes with coordination number five

Many examples of five-coordination have been found and it is now clear
that this coordination number is much more common than was once
supposed. Although in practice they are usually found to be distorted, there
are two idealized five-coordinate structures, the trigonal bipyramidal and
the square pyramidal arrangements (Fig. 3.9). These structures are ener-
getically similar and there seems to be no general way of anticipating which
is adopted by a particular complex.4 Indeed, it is possible that the structure
is determined by intermolecular forces within the crystal (almost all struc-
tures have been determined in the solid state). In some five-coordinate
compounds it has been shown that there is a facile interchange of ligands
between the non-equivalent sites in either structure. The most probable
mechanism for this is shown in Fig. 3.10(a, b). Only relatively small angular
displacements are needed to interconvert the square pyramid and the trigonal
bipyramid, and alternation between the two would lead to the observed
interchange of ligand positions. Moreover, geometries between the two
extremes are possible and are those commonly observed, particularly for
complexes containing chelating ligands.
The interconversion shown in Fig. 3.10 is called a Berry pseudorotation-
Berry because this is the name of the person who first suggested the
mechanism and pseudorotation for the reason shown in Fig. 3.10(c,d).
Whereas the top half of the figure shows the motions required to turn a
3 The caption to this figure makes use of group theoretical notation. The reader who is
unfamiliar with this notation should at least read Appendix 3. which includes an outline of
it, at this point..
(b) 4 Two partial exceptions should be noted. First, transition metal complexes containing
Fig. 3.9 (a) Trigonal bipyramidal (03 ") and strongly n-bonding ligands tend to adopt the trigonal bipyramidal configuration. Secondly, it
(b) square pyramidal (C.,) modes of is possible to make approximate predictions for main group complexes, although a delicate
five-coordination. interplay of factors is involved.
r
'
M
~
Fig. 3.10 The synchronous ligand motion
which serves to tum (a) a trigonal bipyramidal
into a square pyramidal complex and (b) a
square pyramidal complex into a trigonal
bipyramidal one. If a complex undergoes the
displacement shown in (a) but does not
become locked into the square pyramidal
geometry (c) then, as the ligands continue their
original motion a trigonal bipyramidal geometry
is reattained (d). Comparison of the geometries
in (a) and (d) might lead an observer to think
that (a) has been rotated to give (d), although
of course it has not. For this reason the
sequence is called pseudorotation.
Pseudorotation serves to interchange axial and
equatorial ligands of a trigonal bipyramidal
rM I
(o)~
complex.
(d)
pentagonal bipyramid into a square pyramid and vice versa, the bottom
half shows what happens if the amplitude of the vibration shown in the
top left hand corner is so great that the atoms carry on beyond the square
pyramid arrangement to give another trigonal bipyramid. A casual observer,
knowing nothing of the square pyramidal intermediate and seeing only the
before (a) and after (d) arrangements, might well conclude that the original
molecule had simply been rotated to give the final one. Of course, no
rotation has occurred, it just looks as if one has. Hence the use of the word
pseudorotation. Other mechanisms, some rather ingenious, have been pro-
posed as involved in equational-axial interconversions in trigonal bipyram-
idal complexes but the Berry mechanism is believed to be the most
important. Indeed, it has become quite common in crystallographic work
in which the structure of a five-coordinate complex is reported, for the
authors to comment on the position on the Berry reaction pathway that
their particular complex occupies.
The small energy difference between the two modes of five-coordination
is demonstrated in the crystal structure of the compound
[Cr(enlJJ[Ni(CN) 5 ]1.5H 2 0, where there are two distinct types of
[Ni(CN) 5 ] 3 - anions, one square pyramidal and the other approximately
trigonal bipyramidal. Were one form to be appreciably more stable than
the other, then that would be the only one present in the crystal. Although
it was not evident from the crystal structure work, it seems that the 1.5H 2 0
play a key role presumably by hydrogen bonding to the anions; on
dehydration of the compound there is spectroscopic evidence that all
the anions become square pyramidal.
Examples of trigonal bipyramidal structures are the [Co(NCCH 3 ) 5 ] +
and [Cu(bpyhl] + cations. In the latter, one nitrogen of each bipyridyl is
in an axial position. Anionic examples are [CuC1 5 ] 3 -, [SnC1 5 r and
FIJI. 3.11 The C2 h geometry of the complex

anion [Cu 2 CI 6 ] 2 -.
[Pt(SnC1 3 ) 5 ] 3 -. The latter, with Pt-Sn bonds, is formed when acidic tin(II)
chloride solution is added to many platinum salts.
Some main group halides have trigonal bipyramidal structures but care
is needed-structures may well differ from gas to solid (with solutions being
different again). PF5 and SbC1 5 retain the structure in gas and solution
but PC1 5 in the solid is better described as [PC1 4 ] + [PC1 6 r;
SbC1 5 remains
a trigonal bipyramid in the solid; NbC1 5 , TaC1 5 and MoCI 5 exist as dimers
in the solid state, the two chlorine bridges making each metal atom
six-coordinate.
Perhaps the best-known example of square pyramidal coordination is
the compound bisacetylacetonatovanadyl, [VO(acac) 2 ], where acac is acetyl-
acetone5 (CH 3-c0-CH 2-cO-cH 3 ) less one of the central (acidic) hydro-
gens, and in which the oxygen atom directly bound to the vanadium
occupies the unique position. 6 In one salt of the [Cu 2 Cl 6 ] 2 - anion, bridges
between adjacent anions lead to a square pyramidal configuration about
each copper atom (Fig. 3.11); compare this example with [CuCI 5 ] 3 -
mentioned above. Among the main group elements, the [SbC1 5 ] 2 - anion
provides an example of square pyramidal coordination.
A feature of square pyramidal structures is that there is the possibility
of an additional ligand occupying the vacant axial site to produce a
six-coordinate complex. Some of the small variations that have been
observed in the electronic spectrum of [VO(acac),] in different solvents
are believed to be caused by a solvent molecule being weakly bound at
the sixth coordination position. There is evidence that good donor solvents
sometimes also introduce a ligating atom cis to the vanadyl oxygen.
5 The correct name for this ligand is pentane-2,4-dionate, a name that
is being increasingly
used. However, in the literature and most texts the name acetylacetone is the one which will
be encountered and for this reason is the one used in this book.
6 Note the use of the ending yl in vanadyl.
The vanadyl cation is V0 2 +; the vanadium
and oxygen are strongly bonded. The fact that this bond remains intact through most
reactions and that we are dealing with a cation containing oxygen bonded to a metal is
indicated by the yl. Another example is uranyl, UO~ +. These two ions are sometimes quoted
as examples of one- and two-coordination, respectively (see Section 3.2).
3.2.4 Complexes with coordination number six

The majority of coordination compounds that one encounters are six·
coordinate, the structure adopted being that of a regular or slightly
distorted octahedron (Fig. 3.12). It is important to recognize that the
octahedral geometry is found for complexes of both transition metal and
main group elements (although much less common for the later lanthanides
and actinides). The emphasis which will be placed on transition metal
complexes in some of the following chapters may tend to obscure this fact.
Examples of octahedral complexes of main group elements are [Al(acaclJ],
r.
[InC1 6 ] 3 - and [PC1 6
An alternative but rare form of six-coordination is the trigonal pyramidal
arrangement which is found in some sulfur ligand complexes such as
[Re(S 2 C 2 Ph 2 h] (Fig. 3.13). The most noteworthy example of this form
of six-coordination is the compound [W(CH 3 ) 6 ]; in that all complexes
[M(NH 3 ) 6 ]"+ are octahedral it seems that hexamethyltungsten must have
Fl~. 3.12 An octahedral complex. Distorted
octahedral complexes are usually described by a trigonal pyramidal, D3 ., arrangement for electronic reasons. These are
an axis along which a compression, elongation not at present understood, although it may be that the discussion below
or other change may be regarded as occurring. on the bonding in molecules with cubic eight-coordination hints at the
Possibilities are a tetragonal distortion (along a
C4 axis), a trigonal distortion (along a C3 ) and, explanation. This form of six-coordination is also the configuration about
more rare, a digonal distortion (along a C2 ). the metal atoms in MoS 2 and WS 2 (the crystals of these compounds
contain layer lattices and not discrete molecules).
Fl~. 3.13 An example of a molecule showing

the trigonal prismatic mode of six-coordination.
Another possible six-coordinate arrangement is that of six ligands at

the corners of a regular hexagon with the metal atom at the centre. It has
been found, but only when the geometry is imposed by the structure of
the ligand as for the K + at the centre of the 2-dibenzo-18-crown-6 ligand
in Fig. 3.6.
3.2.5 Complexes with coordination number seven

There are three main structures adopted by complexes with coordination
number seven; as is commonly the case with the higher coordination
numbers there appears to be no great energy different between them. In
the salt Na 3 [ZrF7], the anion has the structure of a pentagonal bipyramid
(Fig. 3.14), but in (NH 4 h[ZrF7 ], it has the structure shown in Fig. 3.15,
in which a seventh ligand caps one rectangular face of what would
otherwise be an approximately trigonal pyramidal six-coordinate complex.
No doubt the hydrogen bonding in the ammonium salt is a factor contribut-
ing to the difference in geometry,7 just as in the case of the [Ni(CN) 5 ] 3 -

anion discussed above. The anion [Nb0F6 ] 3 -, which is isoelectronic with
[ZrF7 ]3+, adopts the third mode of seven-coordination, shown in Fig. 3.16.
This is derived from an approximately octahedral six-coordinate arrange-
ment by an additional ligand capping one face.
A nomenclature which is sometimes used to distinguish these three forms
of seven-coordination is to proceed down the axes indicated by arrows in
Figs. 3.14-3.16 (these are the axes of highest symmetry) and to list the
number of ligands lying in planes perpendicular to those axes. In this
nomenclature the geometries are called the 1:5:1 (Fig. 3.14), 1:4:2 (Fig.
3.15) and 1:3:3 (Fig. 3.16) modes of seven-coordination.
Fig. 3.14 The pentagonal bipyramidal (1:5:1, Fig. 3.15 The one-face centred trigonal Fig. 3.16 The one-face centred octahedral
05h) mode of seven-coordination. prismatic (1:4:2, C2 ) mode of (1:3:3, C3 J mode of seven-coordination.
seven-coordination.
3.2.6 Complexes with coordination number eight

There are two common arrangements of eight ligands about a central
atom, the square antiprismatic and the dodecahedral arrangements. Con-
sider an array of eight ligands at the corners of a square-based box (not
necessarily a cube). If the top set is rotated by 45° about the four-fold
rotation axis a square antiprism results (Fig. 3.17). Dodecahedral coordina-
tion is more difficult to describe. Consider two pieces of cardboard cut and
marked as shown in Fig. 3.18(a). If these are interleaved as shown in Fig.
3.18(b), the eight points lie at the corners of a dodecahedron (Fig. 3.18(c)).
(A dodecahedron has twelve faces and eight vertices-if you are in doubt,
make a model and count them.) There appears to be little energetic
difference between the two structures so, whilst [Zr(acac) 4 ] is square
anti prismatic, [Zr( ox) 4 ] 4 -, involving the oxalate anion C 2 0~-, is dodeca-
hedral. Similarly, the [Mo(CN) 8 ] 4 - anion may have either arrangement
in the crystal, the shape adopted varying with the cation. Other examples
7 It is interesting that the salt (NH,),(HfF ) is actually (NH,),[HfF ] + NH F; this

7 6 4 is one
Fig. 3.17 The square antiprismatic (04 d) mode of the few clear-cut distinctions between the chemistries of zirconium and hafnium~the
of eight-coordination. zirconium compound contains the [ZrF7 ] - anion.
\
Cut here
(a)
f!C. 3.18 The dodecahedral (02 ") mode of of square anti prismatic coordination are the [TaF8 ] 3 - and [ReF8 ] 2 - anions.
eight-coordination (a dodecahedron is per11aps Typically, bidentate ligands with relatively short separations between the
best thought of as a solid figure with 12 faces,
which can be, but are not required to be,
two coordinating atoms, i.e. with a short bite, form dodecahedral complexes.
equilateral triangular). The arrangement is Examples are [Co(N03 ) 4 ] 2 - , in which two oxygens from each nitrate
shown in (c)-note the approximate pentagon coordinate to give four-membered rings, and [Cr(02 ) 4 ] 3 - in which both
subtended by this projection of the five 'outer'
ligands. The arrangement is best understood by
atoms of the peroxy o~ - anions coordinate to give three-membered
its construction from two pieces of card (a), rings.
interleaved as in (b). The diagram (c) is drawn Both the dodecahedron and square antiprism may be regarded as
such that one of the planes in (b) is
approximately in the plane of the paper.
distortions of a cubic arrangement of ligands (Fig. 3.19). They are favoured
because a cubic configuration would involve greater interligand steric
FIC. 3.19 By an appropriate concerted motion (a)

of the ligands, a cubic arrangement can be
converted into (a) a square antiprism or (b) a
dodecahedron. A cubic intermediate therefore
offers one explanation of the interconversion
between square antiprismatic and dodecahedral
arrangements (this interconversion seems to
occur rather readily).
(b)
interactions. However, a similar argument favours the octahedron as opposed

to a trigonal prism (for six-coordination) and yet the latter arrangement has
been found_ This suggests that the cubic eight-coordination should exist. In
fact, it has been found for the [PaF8 ] 3 - anion in Na 3 PaF8 , as well as in
[UF8 ] 3 - and [NpF8 ] 3 - _Comparison of the symmetries of the antiprismatic
and cubic geometries suggests that the involvement of f orbitals in the
bonding is required for the cubic arrangement to be stabilized 8
Another form of eight-coordination, largely confined to the actinide series,
is the hexagonal bipyramidal arrangement of ligands (Fig_ 3.20). The ideal
geometry has not yet been observed. When the distance between adjacent
pairs of the six equivalent ligands are equal then the hexagon is found to be
Fig. 3.20 The ideal hexagonal bipyramidal puckered_ If the hexagon is planar then these distances are alternately
(06 h) mode of eight-coordination.
long and short. The two axial ligands are usually oxygen atoms which are
strongly bonded to the central metal, as in the trisacetatouranyl, commonly
called uranylacetate, anion, [U02 (acetatehr. A pattern closer to the ideal
is found in some complexes containing crown ethers. For instance, in its
18-crown-6 complex the K + ion is surrounded by a near-regular hexagon of
oxygens; additional ligands can be accommodated above and below this
hexagon to give eight-coordination .
• • 3.2. 7 Complexes with coordination number nine

A spectacular example of nine-coordination is that of the [ReH 9 ] 2 - anion.
This has the structure commonly found for nine-coordination, a trigonal
prismatic arrangement of six ligands, each of the three rectangular faces
of the prism being capped by an additional ligand (Fig. 3.21). Many
hydrated salts of the lanthanide elements (for example [Nd(H 2 0) 9 ]3+)
Fig. 3.21 The three-face centred trigonal adopt this coordination. It is also found for salts such as PbCl 2 and UC1 4
prismatic (D3h) mode of nine-coordination. in their extended lattices.
3.2.8 Complexes of higher coordination number

Although examples exist, coordination numbers of 10 and above are
relatively rare. Further, it seems that the concept of coordination geometry
becomes less applicable_ The reason is that, whilst idealized geometries can
be identified, most real structures show distortions and there may be some
arbitrariness about which of the ideal structures the distorted structure is
derived from. Examples of idealised coordination geometries are given in
Figs_ 3.22 (coordination number 10), 3.23 (coordination number II) and
3.24 (coordination number 12). The captions to these figures describe the
construction of the polyhedra.
8 Because it is the fluoride ligand which is involved we confine our discussion to u bonding.
In the square antiprism (04 ,), the set of eight ligand a orbitals spans the irreducible
representations A 1 + B2 + £ 1 + E2 + E,-a set which matches the s, p and d orbitals on
the central metal: s(A 1 ); p,(B2 ); p., p,(£ 1); dx,, dx'-y'(E 2 ); dm d,,(£ 3 ). In the cube (0,) the
eight ligand a orbitals span A 1, + T2 , + A2 , + T 1,. Only if an f orbital is included is this
set spanned by the orbitals of the metal atom: s(A 1,); dxy• d,,, d"(T2 ,); fx,,(A 2 ,); Pxo p,, p,(T1,).
In neither set is the metal dz2 involved in the a bonding; in the cube, dxl-y2 is not involved
either.
• •
fiC. 3.22 One mode of 10-coordination--the fiC. 3.23 Eleven-coordination is very rare. One
bicapped square antiprism (0..,). The two possible mode of coordination is the all-face
capping ligands are those at the top and capped trigonal prism (03h). This differs from
bottom of the diagram. At the present time no Rg. 3.21 by the addition of ligands at the top
complex is known which contains ten and bottom of the figure.
monodentate ligands.
3.3 What determines coordination number

and geometry?
So far in this chapter the question posed by the title of this section has
largely been ignored. Rather, an attempt has been made to assemble the
available data in a reasonably compact and accurate form. However, what
has been said so far gives us little confidence in our ability to answer it.
The question implicitly separates out the central atom and the ligands
bonded to it from more remote atoms. Yet in our discussion of two- and
three-coordination we met cases where remote steric effects clearly seem
to be of dominant importance. Similarly, in our discussion of five-coordina-
tion we suggested that in one case hydrogen-bonding involving non-
coordinated water molecules plays a determining role. Clearly, what is
involved is a delicate balancing of interactions, some of which may not
fill, 3.24 There are several modes of be immediately obvious. For instance, in forming an aqua complex in
12-coordination, all of high symmetry (although aqueous solution, there is a cost in removing each coordinated water
no complex containing the twelve monodentate molecule from the bulk solvent which has to be included somewhere in
ligands needed to give these symmetries is
known). That shown is the cuboctahedron (0,. the balance sheet.
It is best known as the cubic close packing Part of the difficulty with the question is an assumption about the form
arrangement in metals. The solid figure may be of the answer. We are conditioned to seek an answer in a simple language
regarded as derived from the octahedron (Rg.
3.12). The equilateral triangular face evident in involving orbitals, their overlap, their bonding and repulsion. We look for
the cuboctahedron is derived from a face of the an answer in terms of individual electrons, or at least individual one-
parent octahedron, but is smaller. This is electron orbitals. Unfortunately, these are approximations. In reality the
because a square face has been obtained by
cutting the solid octahedron perpendicular to behaviour of all of the electrons within a molecule is indivisible. We shall,
each fourfold axis to give the cuboctahedron. at several points in this book, arrive at the conclusion that an orbital model
This has the effect of removing the comers of may well not be capable of providing an answer to a question. One has
the octahedron face, so that the comers of the
triangular faces of the cuboctahedron are at the to carry out detailed and accurate calculations for a variety of geometries
mid-points of faces of the parent octahedron. and compare the results. Whilst such calculations are available for lighter
Isomerism in coordination compounds 1 43
(with not too many electrons) main group elements, calculations on transition
metal compounds are less accurate.
In Appendix 2 is outlined the most popular and successful simple model
for predicting molecular geometry of main group compounds, the valence
shell electron pair repulsion (VSEPR) model. However, alongside it are
presented the results of some detailed calculations which prompt the
comment 'the VSEPR model usually makes correct predictions, but there
is no simple reason why'. The problem of the bonding in transition metal
complexes will be the subject of models presented in Chapters 6, 7 and I 0;
this last chapter reviews the current situation. At this point it is sufficient
to comment that the most useful applications of current simple theory are
those that start with the observed structure and work from there. In the
opinion of the author, the general answer to the question posed at the
head of this section is that we really do not know.
3.4 Isomerism in coordination compounds

The evidence used by Werner to conclude that six-coordinate complexes
are almost invariably octahedral was obtained from a study of the isomerism
of these compounds. Although it is a topic that sometimes excites consider-
able attention, as when recently a new form of isomerism was suggested (see
below), there are few studies in inorganic chemistry nowadays of isomerism
per se (except optical isomerism). Nonetheless, it remains an important
aspect of the chemistry of coordination compounds because ligand inter-
change often occurs readily in solution. That is, when a pure complex is
dissolved, the solution may contain a variety of compounds, including
isomers of the original solid-state structure. To work intelligently, one must
be aware of what may have happened so that a scheme can be devised to
check out the possibilities, should it be necessary. Hence, in the following
pages the various forms of isomerism which have been recognized for
octahedral complexes are outlined. It should be remembered that the
categories are not mutually exclusive and that two or more of the classes
we define may have to be invoked to describe fully the isomerism between
two given compounds. The differences between isomers are evident crystallo-
graphically and, usually, spectroscopically. In some cases analytical differ-
ences may also exist.
3.4.1 Conformation isomerism

This is a simple form of isomerism in which the isomers have different
stereochemistries but are otherwise identical, e.g. trans planar and tetra-
hedral NiC1 2 (Ph 2 PCH 2 PPh 2 h (cf. Fig. 3.7). Its occurrence is confined to
relatively few metal ions, usually those with a d 8 configuration.
The flurry of recent interest in isomerism arose because of the suggested
existence of an isomerism related to conformation isomerism, a isomerism
which is variously called distortional isomerism or bond-stretch isomerism.
The suggestion was made that a bond length in a given complex could
have either of two very different values. This could arise, for instance, if
there were two different bonding interactions, each leading to stability at
a different internuclear distance. What was the ground state at one distance
would correspond to a low lying excited state at the other, and vice
versa. Although it was at first found that the suggested example, in the
compound [Mo(O)CI 2(PMe 2Phh], was erroneous (one crystal structure
determination was on an impure crystal and gave misleading results), it
has stimulated great interest in the possible existence of this form of
isomerism. A subsequent reinvestigation has revealed two (pure!) crystal
forms of the compound in which the rather asymmetrical phosphine ligands
adopt rather different conformations. The Mo-O bond length is 1.663 A
in one isomer and 1.682 A in the other. There are different bond lengths
but the name distortional isomerism, the original one, perhaps is the more
appropriate if the phenomenon is regarded as a form of conformational
isomerism. However, there are some clearly established examples of bond
length differences in some dimeric ruthenium complexes such as ['1 5 -
Cp*RuCI(JJ-CI)]2 in which Cp* is the sterically demanding ligand C 5 (CH 3 )s;
apart from its steric effects, which are currently giving rise to considerable
study,9 it behaves like C 5 H 5 • In these, one isomer is diamagnetic, has no
unpaired electrons, and has a Ru-Ru separation of 2.9 A. The other isomer
is paramagnetic, it has unpaired electrons, and a Ru-Ru distance of 3.8 A.
However, because of the magnetic differences between the two isomers,
they are perhaps better regarded as spin isomers, a type which will be
described later.
3.4.2 Geometrical isomerism

This form of isomer has already been met when discussing nomenclature;
cis and trans isomers are examples of geometrical isomers. Interconversion
between two geometric isomers is often an important step in mechanisms
postulated as those by which coordination compounds catalyse reactions,
particularly those involving unsaturated organic molecules.
3.4.3 Coordination position Isomerism

In this form of isomerism the distribution of ligands between two coordina-
tion centres differs; an example is shown below.
OH 1 3+
CI(NH 3 ) 3 Co/ """Co(NH 3 ) 3 CI
"""OH /
Note that each of these two cations exists in a number of isomeric forms.
The reader may find it a useful exercise to draw pictures of all of the forms
and to enquire into the isomeric relationship between pairs.
3.4.4 Coordination isomerism

This may occur only when the cation and anion of a salt are both complex,
9 Similarly, transition metal complexes of the pentaphenyicyclopentadienyl ligand have
been studied. Unfortunately, they tend to be rather insoluble, a clear disadvantage. It is likely
that the addition of alkyl groups to the phenyl rings would increase the solubility of the
complexes formed. However, if such substitution is not symmetrical then further complica-
tions ensue. The recent synthesis of complexes of the penta-p-tolylcyclopentadienyl ligand
suggests that this may become a much studied ligand in the future.
the two isomers differing in the distribution of ligands between the cation
and anion:
and
The same metal may be the coordination centre in both cation and anion:
[Cr(NH 3 ) 6 ][Cr(SCN) 6 ] and
3.4.5 Ionization Isomerism

Two coordination compounds which differ in the distribution of ions
between those directly coordinated and counterions present in the crystal
structure are called ionization isomers:
and
The difference between these isomers is analytically apparent-an aqueous

solution of the first gives an immediate precipitate with barium chloride
solution and the second with silver nitrate.
3.4.6 Hydrate Isomerism

Hydrate isomerism is similar to ionization isomerism except that it really
only applies to crystals. An uncharged ligand changes from being co-
ordinated to being in the crystal but uncoordinated whilst another ligand
moves in the opposite sense. Although the uncharged ligand need not be
a water molecule, in practice it almost always is (and hence the term
hydrate isomerism), for example
and
3.4. 7 Unkage isomerism

In our discussion on nomenclature the problem that some ligands may
coordinate in two or more ways was encountered. As has been mentioned,
such ligands are sometimes called ambidentate ligands. Corresponding to
this is the phenomenon of linkage isomerism, for example
[Cr(H 2 0) 5 (SCN)] 2 + and [Cr(H 2 0) 5 (NCS)] 2 +

[Co(NH 3 ) 5 (N~)] 2 + and [Co(NH 3 ) 5 (ONO)] 2 +
[Co(NH 3 ) 5 (SSDal] + and [Co(NH 3 ) 5 (OS~S)] +
3.4.8 Polymerization Isomerism

Strictly speaking, polymerization isomerism, in which n varies in the
complex [MLmJn (the Ls need not all be identical), is not isomerism. It is
included in this list because it represents an additional way in which an
empirical formula may give incomplete information about the nature of
complex. For example, all members of the following series are polymeriza-
tion isomers of [Co(NH 3 h(N02 h] •.
[Co(NH 3 ) 3 (N0,) 3 ] n= 1
[Co(NH 3 ) 6 ][Co(N0,) 6 ] n= 2
[Co(NH 3 ) 4 (N0,) 2 ][Co(NH 3 ) 2 (N0,) 4 ] n= 2

[Co(NH 3 ) 5 (NO,)][Co(NH 3 ) 2 (N0,) 4 ] , n= 3
[Co(NH 3 ) 6 ][Co(NH 3 ) 2 (N0,) 4 h n= 4

[Co(NH 3 ) 4 (N0,) 2 h[Co(N0,) 6 ] n= 4
[Co(NH 3 ) 5 (N0,) 6 MCo(N0 2 ) 6 ] 2 n= 5
3.4.9 Ugand isomerism

If two ligands are isomers, the corresponding complexes are isomers also;
for example
"' "'
CH 2 -CH-CH 3 and CH 2 -CH 2 -CH 2
NH 2/ NH 2 NH 2 / NH 2
propylenediamine trimethylenediamine
(pn) (tn)
are isomers, both of which form complexes of the type shown in Fig. 3.25
(where a convenient representation has been adopted for the two isomeric
ligands which shows only the coordinated atoms). In this situation, the
two isomers, indistinguishable by elemental analysis, are termed ligand
isomers.
A special form of ligand isomerism arises when two different sites in a
ligand can be protonated. If only one proton is added then two different
~- 3.25 The bidentate ligands are shown
species result, sometimes called protonation isomers. They are important
here very schematically. If in one complex they because both in the case that the proton is replaced by a metal and in the
represent a particular ligand but in another case that the unprotonated site coordinates, different complexes result.
complex an isomeric ligand, then the two Such differences are important in some biochemical systems.
complexes provide an example of ligand
isomerism. A special case of ligand isomerism also arises when the ligands are
optical isomers-enantiomorphs- of each other. One interesting problem
is the extent to which electron absorption bands, which, as a first approxima-
tion, are supposed to be localized on a transition metal ion, acquire optical
activity because of the activity of a coordinated ligand. An example of this
is provided by the ligand mentioned above, propylenediamine (pn) which
exists in optically isomeric forms.
3.4.10 Optical isomerism

A molecule is optically active when it cannot be superimposed on its
mirror image. Although this condition is met by an octahedral complex such
as ML.LbL,LdL,Lr it is rare indeed to be able to resolve such a complex. In
practice, optical activity is largely confined to octahedral complexes of
chelating ligands. Optical activity has also been observed for chela ted
tetrahedral and square planar complexes but only rarely. It is necessary
for the chelated complex to be stable kinetically; to permit resolution,
it must retain its configuration for at least a matter of minutes. This con-
fines attention to complexes of a few ions, of which cobalt(III), chromium( III)
en~ 3+
Fig. 3.28 The complex [Co(en)g] 3 + and

cis-[Co(en) 2 L,]+ (l ~anion-). The two
~
en_)
n+
molecules represented in this figure for each
species are mirror images which cannot be
superimposed.
and rhodium(III) are examples. Although the optical activity of complexes

of many polydentate chelating ligands has been studied, for simplicity our
discussion will be confined to the bidentate case.
Two classes of optically active complexes formed by bidentate chelating
ligands which have been the subject of much work are [M(L 2 h] and
cis-[M(L 2 lzL2J, where (L 2 ) is a bidentate ligand and L' a monodentate.
Figure 3.26 shows the pairs of isomers for M = Co and L2 = ethylene-
diamine (en). At this point the discussion contained in Appendix 1 becomes
relevant, because amongst the molecules considered there are the [M(L 2 h].
If a compound is optically active it cannot crystallize in a centrosymmet-
ric space group (the action of a centre of symmetry serves to convert one
optical isomer into the other). A consequence is that in X-ray diffraction crys-
tallographic studies, pairs of related diffracted beams (h, k, I and -h, -k, -/),
which in a centrosymmetric crystal would have identical intensities, no longer
do so. An analysis of the difference in intensities in such pairs provides
the absolute configuration of the optically active species.
3.4.11 Structural and fluxional isomerism

For simplicity, almost all of the forms of isomerism discussed above
concerned classical octahedral complexes. It was implicitly assumed that
each complex has a single structure and that this structure does not change
with time. Other forms of isomerism are recognized if we remove one
or both of these restrictions. For instance, in Section 3.2.3, we met the
fact that the anion [Ni(CN) 5 ] 3 - can exist in two different geometries,
trigonal bipyramidal and square pyramidal (a phenomenon which is usually
classified under the heading of structural isomerism). A rather more extreme
example is provided by Co 2 (C0) 8 of which at least two forms coexist in
solution:
and (CO) 4 Co--Co(CO) 4

Fig. 3.27 Potential energy profiles in

~ 5 -C5 H 5 Mn(C0) 3 • Consider a microscopic
probe, shown at the right hand side of the
dotted circle surrounding the manganese atom.
If the molecule is held rigid and the probe
rotated around the dotted circle, the top end of Threefold
the probe will experience five bumps per circuit
and the bottom end will experience three. That t
6
is, the ~ 5 ·C5 H 5 and Mn(C0)3 units generate
fivefold and threefold potentials, respectively.
Now forget the probe and consider just one k
caribon atom of the C5 H5 ring as the ring is '
rotated against a rigid Mn(C0)3 unit. The carbon '
atom will experience three bumps per circuit. '' ..... ___ ., ;
So too will each of the other tour carbon atoms

in the C5 H5 ring. But the geometry is such that,
although all of these bumps will be equal, none
Fifteenfold
will coincide. That is, in a complete circuit there
will be a total of 5 x 3 = 15 bumps, a 15·fold
potential.
Indeed, the ability to undergo such molecular gymnastics is a character-

istic of many organometallic compounds, particularly of transition metals.
As another example consider the compound q 5 -C 5 H 5 Mn(CO)J, shown in
Fig. 3.27. Here, the q 5 -C 5 H 5 Mn unit has a local fivefold rotational axis
and the Mn(COh unit a local threefold. Bringing the two together, one
has a 5 x 3 = 15-fold rotational barrier. The logic behind this arithmetic is
contained in the answer to Question 3.1. However, a 15-fold rotational
barrier means 360°/15 = 24° between equivalent potentials-and so just
12° between maxima and minima in the potential surface. This is also
illustrated in Fig. 3.27. Apart from the fact that for such a small angle the
difference between maxima and minima must be small, the atoms involved
are big relative to the corrugations of the potential energy surface. Not
surprisingly, if the molecule is labelled in some way (perhaps by inclusion
of a 13 C in the C 5 H 5 and another in the (CO h) it is found that rotation
of the two halves of the molecule relative to each other is rather free. It is
often convenient to think of the rotation as confined to the C 5 H 5 system,
particularly when the ring is less symmetrically bonded than in our
example, because the rotation then makes the time average of all ring
positions identical. One talks of ring whizzers, evocative of a firework
display-chemists have a sense of humour too. The collective name given
to such interchange phenomena is to talk of fluxionality and of fluxional
molecules. We shall meet them again in Chapter 14.
3.4.12 Spin Isomerism

As will be seen in Chapter 7, octahedral complexes of Fem can exist in one
of two spin states, high spin and low spin. As will be explained in that
chapter, the difference can be attributed to different magnitudes of splitting,
A, between two sets of d orbitals:
Further reading 1 49
high spin low spin

-t--
-1--
.1 .1
-t-- -j--
-i-- - i L-
-j-- -l L-
In some complexes, usually of Fem but also for other ions, most notably
Fe 11 and Co 11, it seems that the magnitude of the splitting .1. is such that
both forms occur; spin isomers coexist in the same sample An example is
provided by the octahedral complex [Fe(S 2 CNMe 2 }J], where the N,N-
dimethyldithiocarbonate ligand is
"'
CH 3 S
/
N-C/ (.
"'"./
""'
CH 3 S
Here, one has to be careful because spin isomerism behaviour in solution
may well be different to that in the solid state. In the solid state the individual
magnetic ions couple weakly together but often sufficiently strongly for the
phenomenon to have a cooperative aspect and to show hysteresis; the
stronger the cooperativity the more abrupt the transition. The spin-crossover
may be induced not only thermally but also by application of pressure, for
small structural changes accompany the spin change.
For both of the last two classes of isomerism we have detailed-fluxional
and spin isomerism-the lifetime of individual isomers may be rather short.
Spin isomers, for instance, typically live for about 10- 7 s (but see Section
3.4.1 and the Further Reading at the end of this chapter). Some would
argue that classical isomerism refers only to species capable of physical
separation, and so of long lifetime. However, with the increasing use of
methods which explore short lifetimes-NMR and EPR, particularly, in
the present context-it seems sensible to ignore this limitation.
Further reading Other discussions exist, for instance, 11-coordination in W. 0.

Milligan, D. F. Mullica, H. 0. Perkins D. A. Grossie and
Progr. Inorg. Chern. has featured a number of articles devoted C. K. C. Lok, Inorg. Chim. Acta (1984) 86, 33.
to specific coordination numbers. They are: Isomerism in general has been reviewed in 'The isomerism
of complex compounds' R. G. Wilkins and M. J. G. Williams
• 4- and 5-coordination M. C. Fa vas and D. L. Kepert 27, in Modern Coordination Chemistry, R. G. Wilkins and J. Lewis
325 (eds.), lnterscience, New York, 1960. Although old, this remains
a useful review.
• 5-coordination R. R. Holmes 32, 119 The fact that spin equilibria (spin isomerism) has been the
• 5-coordination J. S. Wood 16, 227 subject of much recent work is indicated by the almost
• 6-coordination D. L. Kepert 23, I simultaneous appearance of three reviews on the subject:
• ?-coordination M. G. B. Drew 23, 67
'Dynamics of spin equilibria in metal complexes' by J. K.
• ?-coordination D. L. Kepert 25, 41 Beattie in Adv. Inorg. Chern. (1988) 32, I; 'Static and dynamic
• 7-and8-coordination S. J. Lippard 21,91 effects in spin equilibrium systems' by M. Bacci, Coord. Chern.
• 8-coordination D. L. Kepert 24, 179 Rev. (1988) 86, 245; and 'Spin equilibria in iron(II) complexes"
• 8-coordination S. J. Lippard 8, I09 by H. Toftlund, Coord. Chern. Rev. (1989) 94, 67. See also
Section 9.12. Sometimes it is possible to photoexcite a molecule Hendrickson Angew. Chern., Int. Ed. (1994) 33, 425 (discusses
to the less stable spin state, whereupon it may be stable for the isomerization in the context of spin-crossovers and 'show
weeks! An example, which shows how the methods introduced that it is a reality'). A paper that looks back at the origins of
in later chapters of this book may be used to investigate such the controversy, and comes up with the suprising answer
systems, is in a paper by Gutlich and Poganiuch in Angew. contained in its title, is 'Studies of Distortional Isomers. 2.
Chern., Int. Ed. (1991) 30, 975. Evidence That Green [LWOC1 2 ]PF6 is a Ternary Mixture'
A modern general review on isomerism is to be found in by P. J. Desrochers, K. W. Nebesny, M. J. LaBarre, M. A.
an article by J. C. Bailar in 'Coord. Chern. Rev.' (1990) 100, I. Bruck, G. F. Neilson, R. P. Sperline, J. H. Enemark, G. Backes
For those interested in learning more of the bond-stretch and K. Wieghardt Inorg. Chern. (1994) 33, 15; see also the
isomer story there is detailed review by V. C. Gibson and M. footnote on page 2.
McPartlin in J. Chern. Soc., Dalton Trans. (1992) 947 and a A useful source is Volume I of Comprehensive Coordination
shorter one by J. M. Mayer, Angew. Chern., Int. Ed. (1992) 31, Chemistry G. Wilkinson, R. D. Gillard and J. A. McCleverty
286. Both became slightly out of date because of another (eds.), Pergamon Press, Oxford, 1987, and in particular Chap-
article: A. P. Bashall, S. W. A. Bligh, A. J. Edwards, V. C. ters 2 ('Coordination Numbers and Geometries' by D. L. Kepert),
Gibson, M. McPartlin and 0. B. Robinson Angew. Chern., Int. 3 ('Nomenclature of Coordination Compounds' by T. E. Sloan)
Ed. (1992) 31, 1607. Even more recent is 'Bond Stretch Isomers: and 5 ('Isomerism in Coordination Chemistry' by J. MacB.
-
Fact not Fiction' by P. Giitlich, H.A. Goodwin and D. N. Harrowfield and S. B. Wild).
Questions 3.8 The discussion in the text concerned the combination

of a fivefold and threefold rotation. As a simpler example
3.1 Give systematic names for the complexes listed in Table show that in the molecule [(py)Ni(CO),], in which the pyridine
2.2 (in some cases, because full structural details are not given is bonded to the nickel through its nitrogen atom, the twofold
in the table, there will be some ambiguity about the correct rotational symmetry of the py-Ni unit combines with the
name). threefold rotational symmetry of the Ni(CO), to give a sixfold
3.2 Use a piece of paper to cover up the formulae in Table rotational barrier. Hint: draw each of the six equivalent
3.2 leaving just the names visible. Write down the correspond- arrangements; invent a system of labelling which enables
ing formulae and check your answers by removing the paper. equivalent arrangements to be distinguished.
Repeat the exercise, this time covering up the names and
attempting to write them out. 3.9. The author of this book believes that a better (including
a language-independent) polyhedral system than that recom-
3.3 Repeat the exercise of Question 3.2, this time using
mended by IUPAC would be to give a point group (symmetry)
Table 3.3.
symbol followed by the coordination number. There follow
3.4 When VC1 3 is dissolved in acetonitrile (CH 3 CN), there two lists, one in the author's and, in a different sequence, one
is evidence that above 50 oc two non-ionic monomeric octa- of the corresponding IUPAC symbols. None of the latter have
hedral complexes co-exist. Suggest reasonable structures for been used in the present chapter, although all of the poly-
these two complexes. hedra have been mentioned. Pair off corresponding symbols.
3.5 Figure 3.26 shows cis-[Co(en),L 2 ] + (L is anionic) as This problem should give more familiarity with both the
an example of a complex which can be optically active. Could polyhedra and the use of point group notation.
either (a) trans-[Co(en)zL 2 ]+ or (b) [Co(NH 3 ) 4 L 2 ]+ be optic-
Dxh-2 c2v-2 TP-3 PBPY-7
ally active?
D3h-3 C3v-3 TPRS-7 HBPY-8
3.6 If you were given the task of attempting to prepare a
complex with an unusual coordination number, which coor- D5h-7 C3v-7 TPY-3 A-2
dination numbers should you seek to avoid, and why? C,.-7 Oh-8 OCF-7 SAPR-8
3.7 You are asked to attempt to prepare as many isomers D4d-8 D2d-8 DD-8 CU-8
as possible of [Co(NH 3 )BrCl(en)] 2 S04 • As a first step, pre-
pare as complete a list as possible of these isomers. Dsh-8 L-2
Preparation of coordination
compounds
4.1 Introduction
This chapter reviews the most common methods by which coordination
compounds are prepared. However, current research is almost invariably
aimed at producing the unusual and exotic, not the common. So, a
contemporary research journal would describe methods rather less simple
than most of those covered here. A flavour of the current has therefore been
included, although the reader is unlikely to meet some of the compounds
outside the research laboratory. In the reactions described in this chapter,
there are two important variables-coordination number and oxidation
number (the latter is often called the valence state). In principle, either may
increase, decrease or remain unchanged in a reaction, and the reader may
find it helpful to classify the preparative methods described according to
changes in these two numbers. In practice it is not always possible to be
certain of either without more information than that contained within a
chemical equation or chemical formula. A ligand which is potentially triden-
tate may, for example, act as a bidentate ligand and so the coordination number
differs from that expected. Similarly, is the complex ion [Co(NH 3 )sN0]2+
a complex of cobalt(II) or one of cobalt(III)? It depends on whether you
believe that the NO is better represented as NO" (where, in the complex, the
odd electron is paired with a cobalt electron) or as NO- . This problem of
formal valence states will reappear later in this chapter and again in
Chapter 6.
Complications apart, reactions in which the coordination number of an
electron acceptor is increased are called addition reactions, and when it is
unchanged they are called substitution reactions. The coordination number
decreases for dissociation reactions. Reactions involving valence state changes
are called oxidation or reduction reactions, as appropriate.
An important classification of complexes depends on the speed with which
52 1 Preparation of coordination compounds
they undergo substitution reactions. When excess of aqueous ammonia is

added to a solution of copper(II) sulfate in water the change in colour
from pale to deep blue is almost instantaneous, because an ammine complex
is formed very rapidly (in this reaction ammonia replaces some of the water
molecules coordinated to the copper(II) ion). This is an example of the
generalization that copper(II) forms kinetically labile complexes. On the
other hand, it takes hours (or even days at room temperature) to replace
water molecules coordinated to a chromium( III) ion by other ligands. Again
we can generalize: chromium(III) forms kinetically inert complexes. It is
important to recognize the distinction between kinetic and thermodynamic
stability at this point. The thermodynamic stability of a complex (which will
be discussed at length in the next chapter) refers to the concentrations of
complex species and ligands at equilibrium. Kinetic stability refers to the
speed at which equilibrium conditions are reached. As one would expect, the
preparations of kinetically inert and labile complexes present quite different
problems. In general, the ions of the second- and third-row transition
elements usually form kinetically inert complexes. With the exception of
chromium(III) and cobalt(III), the common ions of first-row transition
elements usually form kinetically labile complexes. Metallic main group
elements usually form labile complexes.
Complexes involving low valence states, organometallic complexes for
instance, are usually inert. However, inertness relates to kinetics and kinetics
depend on mechanism. An organometallic compound which normally reacts
slowly may spontaneously catch fire, or, less dramatically, rapidly oxidize,
if exposed to air. Not surprisingly, special inert atmosphere techniques have
to be used in preparing such compounds. Gaseous oxygen, of course, is a
diradical, with two unpaired electrons, and so it is not unexpected that it
should react rather differently to many other potential reactants.
4.2 Preparative methods

It is difficult to present reaction techniques in an order which is obviously
logical and sequential. In the following pages the pattern usually adopted is
to move from the simple to the complicated, although simplicity has its own
complications. So, the first reaction considered is a simple gas-phase reaction
between molecules, but one for which a quite complicated glass vacuum line
would be needed. The reaction between aqueous Cu" and aqueous ammonia,
considered later, can be carried out using a couple of test tubes, but is
chemically quite complex.
4.2.1 Simple addition reactions

The most direct method of preparing [BF3 (NH 3 )] is by gas-phase addition,
in which a carefully controlled flow of each of the gaseous reactants is led
into a large evacuated flask, where the product deposits as a white powder:
BF3 + NH 3 .... [BF3 (NH 3 )]
When one reactant is a liquid and the other a gas at room temperature, a
different technique is usually followed. In the preparation of [BF3 (0Et 2 )],
for instance, the diethylether and boron trifluoride, stored in separate bulbs
Preparative methods 1 53
on a vacuum line, are condensed separately into an evacuated flask cooled

in liquid nitrogen. When the flask is warmed slowly, a controlled reaction
takes place:
Reactions between liquids or solids are best carried out by mixing solutions
of them in a readily removable inert solvent, e.g.
40-60 ·c bp
SnCI 4 + 2 NMe3 --pe-tr-ol-eu_m_e-th_e_r-+
If at all possible, the presence of a solid reactant should be avoided unless

it is one of those reactions in which an otherwise insoluble compound
dissolves in the presence of a complexing agent. Many of these reactions
occur with no change in valence state, as when silver chloride dissolves in
aqueous ammonia:
or when the gelatinous precipitates formed on adding alkali metal cyanides

to aqueous solutions of many metal ions, dissolve in excess cyanide, for
instance
Solids may dissolve in complexing agents with a change in valence state (so
it is debatable whether such reactions should be classified as simple addition).
In the case of the dissolution of metallic silver or gold in water in the
presence of cyanide ion, the oxygen of the air acts as the oxidizing agent:
2M(s) + 4CW(aq) + ~0.., + H2 0 .... [M(CN 2 lr + 20W (M = Ag or Au)
More commonly, however, the oxidizing agent is carefully chosen, as when

the sparingly soluble PbC1 2 dissolves in aqueous hydrochloric acid through
which chlorine is bubbled:
PbCI 2 (s) + 2HCI(aq) + Cl 2 (g) .... H2 [PbCI 6 ]
The product of this reaction is relatively unstable, decomposing by the

reverse of the formation reaction. A trick which is widely used in such cases
is to add a large, poorly polarizing counterion (a cation such as pyridinium,
C 5 H 5 NH+, added as the chloride, in the present case). One then obtains
either a precipitate or crystals in which the unstable cation or anion is less
prone to decompose. Other counterions which are commonly used in this
way are [As(C 6 H 5 ) 4 ]+, (t-C 4 H 9 lJNH+, [Co(NH 3 ) 6 ]3+, [B(C 6 H 5 ) 4 and r
[Cr(SCN) 6 ] 3 -. There have even been attempts to place this on a quantitative
footing by defining size and shape parameters for such species.
An example of a reaction, involving a solid, which would be avoided if
at all possible is the apparently simple reaction
because it is difficult to ensure that reaction is complete. Further, purification

of the product may be difficult; however, a large number of anionic complexes
of formula [Mx.r-, where X is a halogen (usually F or Cl), have been
made in this way, for example

2KCI + TICI 4 --+ K2 [TiCI 6 ]
The important factor is whether the product forms an impenetrable layer
around the crystals of the solid reactant.
As one would expect, good examples of simple addition reactions of
transition metal complexes are confined to those ions which readily change
their coordination number. Copper(II) provides many examples provided
that excess of incoming ligand is used so that a mixture of products is
avoided. An example is provided by the addition of pyridine to [Cu(acach],
a four-coordinate complex becoming five-coordinate:
[Cu(acac) 2 ] + py --+ [Cu(acac) 2 {py)]
As kinetic studies show (Chapter 14), many reactions in solution proceed
through a reaction intermediate in which the solvent is coordinated.
What may, on paper, appear to be an addition reaction may in fact be a
ligand substitution reaction.
4.2.2 Substitution reactions

The majority of complexes, both of transition and non-transition elements,
may be prepared by substitution reactions. The mechanisms of some of these
reactions have been extensively investigated and will be discussed in Chapter
14. Although the coordination number of the atom at the coordination centre
in both reactant and product species is the same in these reactions, it must
be emphasized that only limited information can be inferred about the
reaction mechanism from a study of the products of a reaction. In particular,
phrases such as 'the ligand A displaces ligand B' should be avoided in
detailed discussions unless the reaction has been properly investigated. For
the non-transition elements in particular, where substitution reactions
usually proceed if thermodynamically favourable, a study of these reactions
enables both qualitative and quantitative assessments to be made of the
relative strength of donor-acceptor bonds. Thus, because ammonia displaces
diethylether from BF3 • 0Et 2 , even in ether solution, to give crystals of
BF3 · NH 3 it has been concluded that the B-N bond is stronger than the
B-0 bond, although this argument is open to the objection that if BF3 • NH 3
has a high lattice energy, perhaps because of hydrogen bonding, then it could
be this fact that leads to the formation of BF3 • NH 3 rather than a higher
B-N bond strength.
As noted earlier, there is an experimental distinction between the substitu-
tion reactions of labile and inert complexes. The formation of labile
complexes is virtually instantaneous upon mixing of the reactants, so that
there are few practical difficulties in their preparation, but three points must
be remembered. First, for classical, Werner-type, complexes it is found in
practice that it is difficult to prepare such complexes with several different
non-ionic ligands bonded to the same metal atom, although it is much easier
to prepare complexes in which an anionic species is coordinated together
with a neutral ligand. Secondly, although it may be possible to isolate and
characterize a solid complex, quite a different complex may be the pre-
dominant species in solution. So, the blue complex Cs 2 [CoC14 ] crystallizes
from pink aqueous solutions containing octahedral Co 11 and CsCI. The third
point, that some complex ions display incongruent solubility, as will be seen,
is related to the second.
If an aqueous solution containing iron( II) sulfate and ammonium sulfate
in a 1:1 molar ratio is allowed to crystallize, then a compound which
historically is variously known as Mohr's salt and as ferrous ammon-
ium sulfate, [Fe(H 2 0) 6 ]S04 (NH 4 lzS04 , is obtained. Mohr's salt is said to
show congruent solubility. On the other hand, if an aqueous solution
containing a 2:1 molar ratio of potassium chloride and copper(II) chloride
crystallizes, crystals of potassium chloride are obtained first. Only later does
the complex K 2 [Cu(H 2 0lzCl 4 ] crystallize. Similarly, attempts to recrystal-
lize the salt will lead to the initial deposition of potassium chloride. The
complex is said to display incongruent solubility; it can only be obtained
from aqueous solutions containing excess of copper(II) chloride. A system
which displays incongruent solubility at one temperature may display
congruent solubility at another.
Examples of the formation of complex ions by substitution reactions
of labile complexes are the following.
1. The action of excess of ammonia on aqueous solutions of copper( II) salts:
[Cu(H 2 0) 4 ] 2+ + 4NH 3 (aq)--+ [Cu(NH 3 ) 4 ] 2+ + 4H 2 0
Although this equation 1
shows the complete substitution of coordinated water
by ammonia all such reactions occur in steps and the species [Cu(H 2 0) 4 ]2+,
[Cu(H 2 0JJNH 3 ]2+, [Cu(H 2 0h(NH 3 lz]3+, [Cu(H 2 0)(NH 3 JJ]2+, and
[Cu(NH 3 ) 4 ] 2 + are all present in the solution, although the concentrations of
some are low. By a suitable choice of concentration (using stability-constant
data of the sort discussed in Chapter 5) it is possible to ensure that the
concentration of one particular component, [Cu(H 2 0) 2 (NH 3 ) 2 ] 2 + say, is a
maximum in the solution. However, it does not follow that if crystallization is
induced (for example, by adding ethanol to the solution and so decreasing
the solubility of the complex species) the complex which crystallizes will
contain the [Cu(H 2 0lz(NH 3 lz] 2 + cation. There are many labile complexes
which may be studied readily in solution but which are very difficult to
obtain in the solid state. The converse is also true. Copper(l) bromide reacts
in ethanol with Br- to give solutions in which only the [CuBr 2 anion has r
been identified. From such solutions, crystals of salts containing anions such
as [Cu 2 Br 5 ]3+ and [Cu 4 Br 6 ] 2 -, as well as [CuBr 2 r,
have been obtained.
When a salt such as [N(CH 3 ) 4 ] 3 [Cu 2 Br 5 ] is dissolved in nitromethane, the
dominant species in solution is again [CuBr 2 r.
2. The reaction between aqueous solutions of thiourea and lead nitrate:
[Pb(H 2 0) 6 ] 2+ + 6SC(NH 2 ) 2 --+ [Pb(SC(NH 2 ) 2 ) 6 ] 2+ + 6H 2 0
The lead(II) ion in aqueous solutions exchanges water between its coordina-
tion sphere and the bulk very rapidly but is probably best regarded as six-
coordinate (although some evidence indicates that the coordination number
1 A common coordination geometry for the copper(II) ion is to be surrounded by four ligands
in a plane which, together with two ligands one above and one below this plane but further
from the copper atom, form a tetragonally distorted octahedron. In this discussion these two,
more weakly bonded, ligands have been neglected.
may be as high as eight). However, an aqueous solution of lead nitrate, say,

may also contain polymeric species and the reaction given above is, therefore,
oversimplified both for this reason and because it makes no mention of
species intermediate between [Pb(H 2 0) 6 ]2+ and [Pb(SC(NH 2 h) 6 ]2+.
3. If an uncharged complex is prepared in aqueous solution from ionic
species it is often precipitated from aqueous solution and, unless highly
polymeric, may usually be recrystallized from organic solvents, e.g.
[Fe(H 2 0) 6 ] 3 •; + 3acac--+ [Fe(acac) 3 ] + 6H 2 0
insoluble
in water
Some examples of preparations involving substitution reactions of inert
complexes are given below.
1. The oxidation of (labile) cobalt(II) salts in aqueous solution containing
both ammonia and ammonium carbonate by air bubbled through the
mixture leads to the formation of the (inert) [Co(NH 3 ) 5 (C03 )]+ cation. It
is only on heating with aqueous ammonium hydrogen fluoride solution at
90 oc for 1 h that this is converted into the [Co(NH 3 ) 5 F] 2 + cation:
[Co(NH 3 ) 5 CO:!J+ + 2HF-+ [Co(NH 3 ) 5 F] 2 + + F- +CO;,+ H2 0
The species [Co(NH 3 ) 5 H 2 0]3+ is almost certainly an intermediate in the
reaction. Surprisingly, cobalt(III) complexes containing coordinated carbon-
ate ion lose C02 rather easily-they fizz when dilute acid is poured onto
them-but this reaction does not involve breaking the Co-O bond; isotopic
studies show that the oxygen atom in the final Co-OH 2 bond is the same as
that in the original Co-OC02 .
Reactions of Co 111 complexes played a key role in the development of
coordination chemistry and so mention of a few more of these reactions is
appropriate. In the preparation of Com salts from Co 11, the composition of
the reaction mixture, the choice of oxidant (H 2 0 2 , Pb02 , perhaps charcoal
being added as a catalyst) and temperature are the key variables. A molar
ratio Co 11 :NH 4 Cl:NH 3 :NaN02 of 1:1:2:3 with air as the oxidant at room
temperature gives mer{Co(NH 3 JJ(N02 h] as the major product. Notwith-
standing what has been said above, this is the product most soluble in water
(presumably the ligands strongly hydrogen bond with the water) and is
thus separated. [Co(NH 3 h(N02 h] reacts at room temperature with concen-
trated hydrochloric acid over a day with evolution of brown fumes of
nitrogen oxides to give [Co(NH 3 JJ(H 2 0)Cl 2 ] +. This, in turn, reacts with
ice-cold aqueous ammonia over about 2 h to give a dimeric compound
containing three hydroxyl bridges:
The contrast between these reactions and those of labile complexes, where
reaction is complete almost as soon as the reactants are mixed, is very
evident.
Preparative methods I 57
2. Potassium hexanitritocobalt(III), potassium cobaltinitrite, reacts with an

aqueous solution of ethylenediamine fairly rapidly at ca 70 oc to give
cis-dinitro bis( ethylenediamine )cobalt(III):
[Co(N0,) 6 ] 3- + 2en--> cis-[Co(en) 2 (N0,) 2 ]+ + 4N0:2
In this reaction the complex [Co(en)(N02 ) 4 r

is presumably an inter·
mediate. However, the solid obtained by removing the solvent from a
solution in which the major component is cis-[Co(en)z(N02 Jz]+ will also
consist largely of this complex ion, because it is kinetically inert.
In the preparation of some complexes, particularly organometallic com·

plexes, the presence of water must be avoided. An important example from
classical coordination chemistry is that the action of ammonia (either as a
gas or in solution) on hydrated chromium(III) salts-those commercially
available-leads to the precipitation of insoluble hydroxy complexes and not
to the formation of [Cr(NH 3 ) 6 ]3+. This complex is prepared by reaction
between liquid ammonia and anhydrous chromium(III) chloride.
A variety of methods have been developed for the preparation of such
anhydrous chlorides. Reaction of the heated metal with chlorine is one
obvious procedure, but is difficult to control. Better is removal of water from
the hydrated salt by chemical reaction on heating with thionyl chloride
(unpleasant), dimethyoxypropane or triethylorthoformate:
H2 0 + SOCI 2 -->SO,+ 2HCI
--> 2C2 H5 0H + HC(O)OC 2 H5
Probably most useful is reaction of a metal oxide with a chlorinated

hydrocarbon; the high boiling hexachloropropene, C(Cl) 2 C(Cl)CC1 3 , is
favoured, the terminal --{:Cl 3 becoming --{:QCJ in the reaction.
Important though anhydrous halides are in synthetic coordination chem·
istry, they suffer from one disadvantage. They tend to have low solubilities
and to react slowly. In such cases, an alternative is to form a complex which,
whilst stable, is one from which the ligands are readily displaced. For
instance, if triethylorthoformate is used as a dehydrating agent with ethanol
as a solvent, complexes such as [Mg(C 2 H 5 0H) 6 ]2+, [Co(C 2 H 5 0H) 6 ]2+
and [Ni(C 2 H 5 0H) 6 ]2+, from which the ligands are readily displaced, are
obtained. A ligand which has become increasingly popular for this purpose
since it became commercially available is the trifluoromethansulphonate
anion, CFrSO:J, more usually called triflate (hence triflic acid and, as a
ligand, triflato ). The pure acid itself is very corrosive and care has to be
taken in its use. However, it is probably much less dangerous than the
perchlorate anion which was previously similarly used, for the latter has a
well-known tendency to destroy apparatus and to remove parts of the
anatomy-perchlorates are prone to explode. Although substitution of
CF3-SO:J into Co 111 is slow; the preparation of cis-[Co(en)z(OS02 CF3 Jz]+
from from [Co(en)zC1 2 ]+ requires use of triflic acid at 100 oc for 3 h, the
triflato ligand is readily displaced by reaction with a replacement ligand
in a relatively inert solvent such as acetone. Anhydrous metal triflates may,
usually with advantage, replace anhydrous metal chlorides in organometallic
chemistry. These compounds may be made, for example, by refluxing the

anhydrous chloride with triflic acid.
We have now moved to the area of coordination chemistry in which water
is to be avoided as a solvent. This is a general area with many facets.
Examples of the preparation of complexes by substitution reactions in
non-aqueous media are the following.
1. Potassium thiocyanate melts at 173 oc and may be used as a solvent at
temperatures above this. For example, in this medium, water is readily
displaced from the [Cr(H 2 0) 6]3+ ion:
2. As has been seen, refluxing thionyl chloride reacts with water and may
be used to prepare anhydrous metal chlorides from the hydrates. Addition-
ally, it is a suitable solvent for the preparation of the chloro anions of
metals:
At high temperatures thionyl chloride slowly decomposes to give chlorine

and this reduces its usefulness as a solvent because the chlorine may become
a reactant.
3. Most salts are converted by bromine trifluoride into the highest fluoride
of the element, or, if an alkali metal salt is present, into a fluoro anion. It is
so powerful a fluorinating agent that it will even react with metals and alloys.
For example, with a 1:1 alloy of silver and gold,
BrF3
AgAu(alloy) - Ag[Auf4 ]
4. As will be discussed in more detail in Chapter 15, there are three distinct
bonding mechanisms which contribute to the metal-metal bonding in
[CI 4 Re--ReCI4 ] 2 -, u, n and <5 (the chlorines are all terminal and eclipsed),
and so it is an anion of particular interest. It is readily prepared from the
commercially available ReCI 3 , which consists of molecules containing a
triangle of rhenium atoms, by fusion in molten (220 oq diethylammonium
chloride:
4.2.3 Oxidation-reduction reactions

As has been seen, inert complexes of the transition metals may be inter-
converted by substitution reactions, but such methods cannot generally be
relied upon and it is preferable to prepare inert complexes by a different
method. The chosen method is to take a compound containing the metal in
a different oxidation state and oxidize or reduce it, as appropriate, in the
presence of the selected coordinating ligand. This technique is used extensively
in the preparation of oxalato complexes of chromium(III). Other chromium-
(111) complexes are prepared by the oxidation of chromium(II) salts.
Preparative methods I 59
The success of this general preparative method rests on two factors. First,
although the product is an inert complex, the starting material is one which is
relatively labile. Other things being equal, concentrations used in the
preparation approximate to those which maximize the concentration of a
complex species identical in composition with the desired product but
differing from it in charge. Electron addition or removal (i.e. reduction or
oxidation) then gives the product. Secondly, as has been mentioned earlier,
there will be several labile complexes in equilibria, each of which can undergo
oxidation (or reduction) to give an inert product. In general, the product
actually obtained will be derived from that labile complex which is the most
readily oxidized (or reduced).
Examples of complexes prepared by oxidation-reduction

reactions
1. Some examples of complexes of cobalt(III) prepared by oxidation-

reduction reactions have been given earlier, as part of the description of the
substitution reactions of the inert compounds formed. However, the prepara-
tion of what may be regarded as the parent complex, hexaamminecobalt(III)
chloride, was not described. As Werner found, this is made by hydrogen
peroxide oxidation of an aqueous solution of cobalt(II) chloride made
alkaline with ammonia in the presence of ammonium chloride:
charcoal
2[Co(H 2 0) 6 ]CI 2 + 2NH 4 CI + 10NH3 + H2 0, - 2[Co(NH 3 ) 6 ]CI3 + 14H2 0
This reaction is catalysed by the presence of charcoal; in its absence the
product consists largely of pentaamminecobalt(III) complexes, the sixth
coordination site being occupied by either H 2 0 or Cl-. The function of the
charcoal is not known with certainty but it is believed that it may act by
donating an electron to a pentaamminecobalt(III) ion, converting it momen-
tarily into a labile pentaamminecobalt(II) species into which a further
ammonia molecule substitutes.
2. An aqueous solution of oxalic acid and potassium oxalate reduces
potassium dichromate to the trisoxalatochromium(III) anion:
3. Complexes containing manganese in less-common formal valence states

may be made either by reduction of the permanganate anion, [Mn04 or r,
by oxidation of the hexaquamanganese(II) cation. Sometimes, as in the
preparation of the [MnF5 (H 2 0)] 2 - anion, containing manganese(III), the
two are combined:
In this example is seen another reason for using an oxidation-reduction

reaction for the preparation of a complex, the non-availability of suitable
precursors in which a metal is in the desired valence state.
Systems in which a series of complex ions of identical stoichiometry are

interrelated by a series of one-electron oxidations or reductions have been
extensively studied. The existence of such a related series is conveniently
investigated by electrochemical methods, of which one is polarography,2 the

half-wave potentials obtained suggesting suitable chemical oxidizing or
reducing agents for the bulk preparation of the species. An alternative
technique is to carry out the oxidations or reduction electrolytically,
preferably at a potential which is held constant despite fluctuations in
Fig. 4.1 The square planar complex nickel(ll) current. This technique is called controlled potential electrolysis; some related
bis(stilbenedithiolate). techniques will be considered in more detail in Chapter 12. The method can
throw up some interesting problems. For example, it has been shown
polarographically that the square planar complex of nickel given in Fig.
4.1, undergoes reduction to give the species [Ni(S 2 C 2 (C 6 H 5 ),) 2 and r
[Ni(S 2 C 2 (C 6 H 5 ),),] 2 -. There has been considerable discussion of the
valence state of the metal atom in these and similar complexes. Is it the metal
or is it the ligand which is being reduced, or is it both? It seems that the
electrons added in the reduction are delocalized over the whole complex (the
stilbenedithiolate ligand is planar and has a delocalized n system) and
so it becomes difficult to define the valency state of either the metal or ligand.
Such problems will be considered in more depth in Chapter 6.
Although in the above discussion it has been clear that there can be a
change of coordination number associated with a change of valence state--
tetrahedral chromium and manganese in [Cr 2 0 7 ] 2 - and [Mn04 becom- r
ing octahedral in complexes of the trivalent metals-there was no clear link
between the two changes. However, there exist many reactions where there
is such a link, one that is easy to see. As has been mentioned, square planar
complexes tend to be formed by transition metal ions with the dB configura-
tion. There exists a whole series of reactions of such complexes in which the
metal is, formally, oxidized by two units, and so becomes d 6 , and, simulta-
neously, increases its coordination number by two, becoming octahedral.
The linking of these two changes is signalled by the name given to this
type of reaction-oxidative addition reactions. A typical example involves
Vaska's compound trans-[Ir(PPh 3 lz(CO)Cl] (Fig. 4.2).
FIJI. 4.2 Example of oxidative addition-

to give Vaska's compound,
trans-[lr(PPh3 ) 2 (CO)CI].
Molecules such as H 2 , 0 2 and S02 can replace HBr in this reaction, to

give complexes which readily lose these molecules again. The tendency to
undergo oxidative addition reactions increases, roughly, from top right to
bottom left in the dB series (Fig. 4.3).
2 In polarography the potential between two electrodes in a solution is continuously varied
and the consequent variation in current recorded. At certain potentials electrolytic reduction
of species in solution occurs and the current rises. Because the cathode is very small (either a
flow of mercury drops through a capillary or a thin platinum wire) the increase in current
is diffusion-limited (and so concentration-dependent) and a characteristic step-like plot of
current against voltage results, one step for each reduction process. The voltage corresponding
to the mid-point of the riser part of a step is a characteristic of the reduction process and is
Fig. 4.3 Series of d8 configurations, showing termed the half-wave potential. The height of each step is a measure of the concentration of the
the tendency to undergo oxidative addition. corresponding species in solution.
Inevitably, oxidative addition reactions are not confined to these elements,

these valence states or, indeed, to these geometries but it is for them that
this type of reaction is most important.
Reactions which are, effectively, the reverse of oxidative addition also
occur. Not surprisingly, they arc known as reductive elimination reactions.
The decomposition of complexes of Vaska's compound with molecules such
as H 2 and 0 2 , mentioned above, provide obvious examples. Another is
provided by elimination of CH 4 from some phosphite ester complexes of
cobalt containing H and CH 3 as ligands, in which the cobalt may formally
be regarded as changing from d 7 , Co 11 to d 9 , Co 0 :
4.2.4 Thermal dissociation reactions

By controlled heating, some complexes can be degraded to others, a volatile
compound being expelled. For example, when it is h.eated, ordinary blue
copper(II) sulfate loses water in a stepwise manner until above ca 220 oc
the anhydrous sulfate is left. The first water molecule to be lost comes from
the lattice; subsequent ones come from the coordination sphere of the Cu 11
and are replaced by sulfate oxygens:
96.5 ·c 102 ·c
CuS04 ·5H 2 0 - CuS04 ·4H 2 0 - CuS04 ·3H 2 0
115 ·c 220 ·c
CuS04 · H2 0 - CuS04
Many other complexes behave similarly and heating (usually under vacuum)
to a carefully controlled temperature is a useful preparative method.
Hydrogen halide elimination, for example, is a reaction which occurs readily
for almost any complex which has the electron-donor atom attached to a
hydrogen (e.g. H 2 0, ROH, NH 3 , R 2 NH) and the electron acceptor attached
to a halogen (e.g. BF3 , SnBr4 , FeCI 3 ). So, in the preparation of a complex
such as the first given in this chapter, [BF3 (NH 3 )], the product readily
eliminates hydrogen fluoride to give a series of compounds and ultimately
the polymeric solid BN. Another very common thermal reaction is the
expulsion of one or more neutral ligands (as in the case of copper(II)
sulfate), with a consequent reduction in the apparent coordination number
of the central atom. In fact, quite often, some previously monodentate ligand
(generally an anion) becomes either bidentate or a group that bridges two
metal atoms (the acetate and halide anions, respectively, exemplify these two
cases). Another possibility is that an anion, initially non-coordinated,
becomes attached to the metal. The actual coordination number of the
central atom is seldom reduced. Examples have already been given in Section
3.4.6.
Heating to a relatively high temperature can lead to the complete
dissociation of the complex species. For example
K[BF4 ] -+ BF3 + KF
and
BrF2 [RuF6 ] -+ BrF3 + RuF5
In the absence of some other suitable cationic species, complex fluorides
containing the [BrF2 ] + cation are formed when transition metals are
dissolved in BrF3 • Thermal decomposition of these salts is a convenient way
of making small quantities of many fluorides.
Another example of a complex prepared by a thermal dissociation
reaction is the preparation of cis-[Cr(en)zC1 2 ]+ by heating [Cr(en),]Cl 3 :
210'C
[Cr(en) 3 ]CI 3 - cis-[Cr(en) 2 CI 2 ]CI +en
The temperature of reaction has to be carefully controlled to achieve

maximum yields; the reaction is complete in 2-3 h. In an analogous reaction
trans-[Cr(enlz(SCN)z]SCN is prepared by heating [Cr(en),](SCN), to
130 oc.
Two experimental techniques have been much used to study the prepara-
tion of complexes by the thermal dissociation reactions.
Differential thermal analysis (DTA)

Two cells, one containing the complex under study and the other a similar
amount of a thermally stable material, are slowly heated, each receiving an
identical amount of thermal energy. The temperature difference between the
two cells is measured and remains essentially zero until thermal dissociation
of the complex occurs. Both the temperature (at a particular pressure, usually
atmospheric) and enthalpy of dissociation may be obtained by this technique.
Thermogravimetric analysis (TGA)

The weight of a complex is measured as its temperature is raised. When
thermal dissociation occurs the weight loss due to any volatile ligand expelled
is measured. The empirical formula of the product may usually be deduced
if the identity of the ligand expelled is known. Both dissociation pressure
and temperature are recorded as they are, of course, interrelated.
4.2.5 Preparations in the absence of oxygen

It has been seen earlier that it may be necessary to work in the absence of
water to prepare some complexes. An even wider range becomes accessible
if we work in the absence of air-much of the field of organometallic
chemistry, for example. Not that it is impossible to prepare organometallic
compounds in the presence of air-Zeise's salt K[PtC1 3 (C 2 H 4 )]H 2 0, the
first organometallic compound made, was prepared with no attempt to
exclude air. However, it must be admitted that in this preparation the ethene
itself serves to provide an inert atmosphere. Nowadays, it is prepared by
bubbling ethene through a solution containing [PtC1 4 Y- in strong hydro-
chloric acid (best with a trace of Sn 11 as catalyst) for a few hours. Rather
similar is the cyclooctene, C 8 H 14, complex which serves as a precursor for
many complexes of rhodium(!), [RhCl(C 8 H 14 )z]., made by allowing com-
mercial hydrated RhCI 3 to stand for a week with cyclooctene in 2-propanol
(which is oxidized to acetone in the reaction) in a flask filled with nitrogen.
Such one-step reactions are the exception rather than the rule. More
commonly, it is necessary to carry out a series of successive reactions and
procedures, such as refluxing, distilling, crystallization, filtration and washing
in an inert atmosphere. The most evident way of doing this is to work in an
inert-atmosphere filled glove box using conventional apparatus. This is both
possible and common, but the volume of a typical box is such that it is rather
difficult to reduce and maintain the oxygen concentration at an acceptable
level. There exist sophisticated boxes with recirculation of the inert gas
through oxygen-removing trains and entrance ports which can be thoroughly
evacuated, but they are very expensive. At the other extreme, it has been
pointed out that it is possible to work with a cheap transparent plastic glove,
in which the fingers are used to store and mix reagents, nitrogen being passed
in through the sealed-off wrist.
Many workers make use of so-called Schlenk tube techniques. This is the
name given to a whole series of simple, but versatile, devices that enable
reactions to be carried out in an essentially closed apparatus of low volume.
Examples include the use of septum caps (those used to close the vials
containing materials used for medical injections). Solvents can be taken in
and out of the apparatus using hypodermic needles; nitrogen can be passed
in through such a needle and allowed to escape through another; on removal
of all needles the apparatus is automatically sealed. Filter sticks are
used-glass tubes with a glass sinter sealed halfway down. If one is on top
of the solution to be filtered, inversion of the apparatus (perhaps with gentle
use of nitrogen gas pressure) leads to filtration. Solids can be placed in a
limb of a tube which has a bend of ca. 90° in its middle and which is held
with the angle at the top. When the solid is needed for reaction then, if the
solid is in the left-hand limb, rotation by 180° about the right-hand limb
causes the contents of the two limbs to mix. Tubes are interconnected by
taps (often greaseless) so that alternative routes exist for gas and/or liquid
flow and add to the versatility.
An example of a series of reactions involving Schlenk tubes is the
preparation of (CH 3 )NGaH 3 . GaC1 3 is dissolved in diethylether and slowly
added to a slurry of LiH in diethylether through a greaseless valve. After
reaction, the product is filtered through a filter stick. To the filtrate
(Li[GaH 4 ] in diethylether) is added [(CH 3 )eNH]Cl by the rotating arm
technique. After vacuum removal of the solvent, the product is separated by
vacuum sublimation from the reaction mixture.
Reactions involving metals, either bulk or, more commonly, finely divided,
are an entry point for organometallic complexes of transition metals.
Examples of direct reaction of a metal are
Ni + 4CO --+ Ni(C0) 4
and
Fe + 5CO --+ Fe(C0) 5
It should be noted that classical complexes of low-valence states can also be

prepared from metals; thus, the simplest starting point for the preparation
of pure complexes of chromium(II) is by the action of an oxygen-free acid
on the pure metal in a inert atmosphere. More commonly, however, a
reduction reaction is used in the preparation of organometallic complexes
because in them the metal is usually in a low formal oxidation state.
Examples of reduction reactions include those in which gaseous carbon
monoxide is both reducing agent and reactant. For instance, when CO is
passed over solid RhCl 3 at about 100° C, the chlorine-bridged dimer
[Rh(COhC1] 2 sublimes into the cold parts of the reaction vessel:
2RhCI 3 + 6CO--+ [Rh(C0) 2 CI] 2 + 2COCI 2

This process can be carried further, in which the product reacts with carbon
monoxide in hexane in the presence of a mild alkali:
2[Rh(C0) 2 CI] 2 + 6CO + 2H 2 --+ Rh 4 (C0) 12 + 2CO, + 4HCI
The product is a black crystalline material. The molecules consist of a

tetrahedron of rhodium atoms, each rhodium atom being bonded to the
other three rhodiums and to three CO ligands, [Rh(C0h] 4 (this compound
will be the subject of discussion in Chapter 15).
This same pattern was noted earlier, when describing the preparation of
Cr111 complexes from chromates: that, if at all possible, a ligand is also
used as reducing agent. Other examples are the reaction of rhodium
trichloride with triphenylphosphine in hot ethanol to give Wilkinson's
compound, [RhCl(P(C 6 H 5 hhJ:
This product is important in organic chemistry because it reversibly absorbs

H 2 under normal laboratory conditions and catalyses the hydrogenation of
alkenes and alkynes. Another example is the reduction of the [OsC1 6 ] 2 -
anion, which is commercially available, with hydrazine hydrate, a liquid,
without additional solvent, to give complexes with one and two coordinated
N 2 molecules:
5(NH 4 ),[0sCI 6 ] + 13NH 2 NH 2 H2 0--+

4[0s(NH 3 ) 5 (N 2 )]CI 2 + cis-[Os(NH 3 ) 4 (N 2 ) 2 ]CI 2 + 20HCI + 13H 2 0
A very important compound in Ni 0 chemistry is bis(l,5-cyclooctadiene)-

nickel(O) for the ligands are readily displaced. It is made by the reduction
of [Ni(acaclz] by an aluminium alkyl, usually Al 2 Et 6 , in the presence of
1,5-cyclooctadiene using toluene as a solvent. The reaction occurs at room
temperature or below, the product precipitating. An example of the use of
this compound in synthesis is in the preparation of Ni(PPh 3 ) 4 ; the latter is
formed at room temperature or below by the addition of triphenylphosphine
in hexane.
As this last synthesis illustrates, the majority of reactions of transition
metal organometallic compounds involve a starting material which is already
in a low-valence state. As another example, hexacarbonylchromium reacts
with arenes in a high-boiling solvent to give complexes with a n-bonded
benzene ring:
It is often a characteristic of organometallic reactions that they lead to a

wide variety of products. For example, Co 2 (C0) 8 reacts with CS 2 to
give about 20 identified products; Fe(C0) 5 reacts with diphenylethyne,
(C 6 H 5 )C-=C(C 6 H 5 ), to give about the same number. In such cases chromato-
graphic methods are used to separate the individual component products.
Main group organometallics are usually prepared from the corresponding

Grignard, lithium or mercury organic species,
EtMgBr
[SnCI 4 (Et2 0) 2 ] ~ SnEt4
2AI + 3HgEt2 --+ AI 2 Et;; + 3Hg

and similar methods are used for some transition metal compounds; the
product in the example below is only stable at low temperatures:
3UMe + WCI 6 --+ WMe6
(it is found to be important to use a 3:1 reactant ratio; the yield is only ca.
50%). Closely related are reactions in which a hydrocarbon, such as
cyclopentadiene, C 5 H 6 , which has an acidic hydrogen (loss of a proton gives
the C 5 H5 anion, with an aromatic n system), reacts either with a metal or
strong alkali to give the anion. This is then commonly reacted with a metal
halide or metal halide complex:
THF
2C5 H6 + 2K - 2KC5 H5 + H2
4KC5 C5 + UC1 4 --+ U(C5 H5 ) 4 + 4KCI
Sometimes, the steps are contained within one reaction mixture, as in the
preparation of ferrocene:
4.2.6 Reactions of coordinated ligands

It is almost a tautology to say that substitution reactions of inert complexes
proceed slowly at room temperature; however, there are exceptions. For
example, as has been mentioned, addition of acid to the [Co(NH 3 ) 5 C03 ] +
cation leads to the rapid evolution of carbon dioxide and formation of the
complex ion [Co(NH 3 )s0H 2 ] 3 +. Indeed, a whole series of facile interconver-
sions exists between species containing Co-OH 2 , Co-OH, Co-C03 , Co-S03 ,
Co-N02 , and similar bonds. The explanation for this non-typical behaviour
is that in none of these reactions is a Co-O bond broken (although, because
they are commonly written as above, it is not immediately obvious that all
of them contain Co-O bonds). Note that the N02 anion may also bond
through the nitrogen (see Section 3.4). It was mentioned earlier that in the
reaction of [Co(NH 3 ) 5 0C02 ]+ with acids, it is the 0-C bond that is broken,
not the Co-O. Another reaction of the same type is:
HNO,
[Cr(NH 3 ) 5 (H 2 0)] 3 + - [Cr(NH 3 ) 4 (0N0)] 2 +
and a rather unusual reaction which proceeds in aqueous solution is
H202
[Co(NH 3 ) 5 NCS] 2 + - [Co(NH 3 )6 ] 3 +
(the mechanism of this reaction does not seem to be known).
So far we have considered only reactions of an atom directly bonded to
a transition metal. There has been much work on the more remote
modification of a ligand. One relatively simple complex which has been the
CH3 CH3
I I
0-C 0-C
Fig. 4.4 Bromination at the ligand in
trisacetylacetonatochromium(lll).
cr ~Q~c-H
0-C
10'
Cr \
0-C1
C-Br
I I
CH3 CH3
3 3
subject of extensive study is trisacetylacetonatochromium(III). The action of

bromine in acetic acid leads to bromination of the acetylacetone ring (Fig.
4.4).
It is reported that the N-halosuccinimides are the best agents for
halogenating coordinated acetylacetonate rings. All the available evidence
indicates that it is the coordinated ligand which reacts, and not any free
ligand in equilibrium with it. Examples of other groups which have been
used to replace the active hydrogen atoms in metal acetylacetonates are
-N02 , -NH 2 , -Nj, -CHO, -COCH 3 and -8Cl.
An important group of reactions of coordinated ligands which does
involve the metal-ligand bond are the so-called insertion reactions. They are
often of importance in the catalysis of organic reactions by transition metal
complexes, although insertion reactions have a history dating back to the
mid-19th century in the reaction
SbCI 5 + 2C2H2 -+ CI 3 Sb(CHCHCI) 2
More relevant to the present topic is the reaction
CH 3 Mn(C0) 5 + CO -+ CH 3 C(O)Mn(C0) 5
Isotopic and kinetic evidence have demonstrated unambiguously that the
CO which enters the Mn-CH 3 bond is one of those already attached to the
manganese; there is an equilibrium:
CH 3 Mn(C0) 5 ~ CH 3-GOMn(C0) 4
Another important, but often difficult to prove, example is the insertion of

an alkene into a M-H bond to give an alkyl:
CH 2=CH2
!
M-H -+ M-GH 2-GH3
Such reactions of coordinated ligands are important in the organometallic

chemistry of transition elements; they may proceed in the opposite direction,
an alkene hydride being formed from an alkyl. A trick widely adopted to
prevent this reverse reaction is to replace the f3 hydrogens by alkyl
groups or the f3 carbon by a silicon (with alkyl groups attached).
Another example is provided by the attack of alkoxide ions on coordin-
ated CO to give M-C02 R groups:
[lr(C0) 3 (P(C6 H5 ) 3 b]+ + CH 3 0- -+ [lr(C0) 2 (C~CH 3 )(P(C 6 Hs) 3 ) 2 ]
Often, insertion reactions involve, formally, H + or H- and may be difficult to

distinguish from alkene insertion reactions of the type given above. Examples
Preparative methods I 6 7
are given, firstly, by the reaction of the cobalticinium cation with (boro )·
hydride,
[(~ 5 -C 5 H 5 ) 2 Co]+ + W-+ [(~ 5 -C 5 H 5 )Co(~ 4 -C 5 H 6 )]
and, secondly, the protonation of a-allylic complexes (Fig. 4.5).
Fig. 4.5 Protonation of a a-allylic complex. H+

C5H5Mo-CH2-CH=CH2
(C0)3
A quite different, but very important, class of reaction of coordinated

ligands is in the synthesis of new coordinated ligands. An excellent example
is provided by the synthesis of cobalt sepulchrate (cf. Table 2.5) from
[Co(enh]3+ by the action of formaldehyde and ammonia in a basic medium
(aqueous lithium carbonate), the reaction taking about 2 h to reach com-
pletion at room temperature (Fig. 4.6). The reaction occurs without rupture
of the Co-N bonds. It seems that nucleophiles with the same C 3 , loca!
symmetry as ammonia may replace it in this synthesis-so CH 3-N02 leads
to cages with =C-N02 apices; the N02 group can be reduced, opening the
3+
Fig. 4.8 Reaction of coordinated ligands to

prepare cobalt supulchrate.
~ ;;;; -CH2-CH2-
way to an extensive study of the chemistry of one group of coordinated
ligands. Syntheses of the type just described are often called template
reactions for an obvious reason-the metal together with the ligands already
in place form a template for the ligand to be created.
A rather important group of complexes which may also be made by
template synthesis are those of imines. lmines are formed by condensation
of an amine and a carbonyl (Fig. 4.7). The amine can be coordinated to a
metal and the above reaction still proceeds, the amine (or, more usually,
diamine) remaining coordinated. Typical is the reaction between bisethylene-
diaminenickel(II) chloride and acetylacetone. An aqueous solution of the
mixture, to which a few drops of pyridine have been added, is refluxed
R
Fig. 4. 7 Condensation of an amine and a \
carbonyl to form an imine. C=NR"
I
R'
Me
Py I
C=O
NH2
HrnN I
Ni +2
I
CH2 + 2py +en
H21~ ! NH2
I
C=O
Py I
Me
FI,C. 4.8 Reaction between for 2 h to give the product. The reaction is probably between [Ni(enlz(pylz]2+
bisethylenediaminenickel(ll) chloride and and the acetylacetone (Fig. 4.8). Template syntheses are finding particular
acetylacetone.
applicability to the synthesis of large, sometimes complicated, ligands. The
improvement in yield of ligand, compared with an organic synthesis in the
absence of metal, is sometimes quite spectacular.
Many reactions are known in which the presence of a metal ion influences
the products of a reaction and the explanation for this may well lie in the
different reactions of free and coordinated ligands. Peptide chemistry is one
field in which this may prove to be of great importance.
4.2. 7 The trans effect

The ligand arrangement around an atom after a substitution reaction may
or may not be similar to that of the starting material, even for inert
complexes. An example of such a change is provided by the formation of
trans-[Cr(enlz(NCS)z]+ on heating [Cr(enlJ]3+ with solid ammonium
thiocyanate at 130 oc. Similarly, cis-[Co(NH 3 ) 4 (H 2 0)Cl]S04 is converted
into trans-[Co(NH 3 ) 4 Cl 2 ]HS04 by the action of a mixture concentrated
hydrochloric and sulfuric acids at room temperature.
The chemistry of platinum(II) and, to a lesser extent, those of Pd 11 and
Au 111 (all three form d 8 square planar complexes) is therefore noteworthy in
that the major product of a substitution reaction can be predicted with
confidence. This is because the lability of a ligand bonded to platinum(II)
is largely determined by the group which is trans to it and not by
the nature of the ligand itself. Although this trans effect is not fully
understood its operation is reasonably reliable and renders the synthesis of
platinum(II) complexes, in particular, a class on its own. The stereochemistry
of the products of reactions of platinum( II) complexes can often be varied
by altering the order of reagent addition. An example of this is provided by
the synthesis of cis- and trans-[Pt(NH 3 )(N02 )Cl 2 ] - starting from [PtC1 4 ] 2 -
(Fig. 4.9).
Ligands can be arranged in a series depending on the relative magnitude
of the trans effect which they exert. In view of the above discussion, it would
be expected that different sources would agree on the sequence of ligands
which corresponds to increasing trans effect. In fact, there is little unanimity,
some lists appearing quite aberrant. The following sequence, however, is
largely accepted:
F- ~ow ~ NH 3 ~ py < cl- < sr- < 1- <-sew ~ -N02 <thiourea
< PR3 ~ AsR3 ~ H- < CO ~ CN- ~ C2 H4
One can see from this list the difficulty of proposing a general explanation
CI--NH31- CI--NH31-
I Pt \ I Pt \
Cl Cl Cl N02
cis
~ NH3
CI--CI 12-
Flil. 4.9 The sequence of reagent addition can I Pt \
be altered to selectively provide cis and trans Cl Cl
products from the same starting complex.
1 N02
CI--N02 12- NH3 CI--N02 1-

I Pt \ - I Pt \
Cl Cl H3N Cl
trans
for the trans effect. Almost all those ligands exerting a strong trans effect are
n bonding, but a n bonding explanation does not explain the position of H-
(a polarization model, based on the unique characteristics of the H atom is
usually added to cover this). However, there seems no explanation of why
the effect is largely confined to Pt11 , perhaps along with Pd 11 and Aum-
certainly, neither the n bonding nor the polarization model is metal-specific.
It could be that the notion that the trans effect is largely confined to Pt 11 is
incorrect; it is just that other elements have not been so extensively studied.
Although this point has substance as far as Aum and Pd" are concerned, it
does not seem to have general validity. Above all, it must be remembered
that the trans effect is a kinetic effect, associated with bond breaking and
formation. It could be more a phenomenon of the reaction pathway (an
activated complex pathway, an activated complex or transition state) than
the ground state. In Chapter 14 the kinetics of the reactions of square planar
Pt11 complexes will be the subject of some discussion, a discussion that will
include the trans effect.
4.2.8 Other methods of preparing coordination

compounds
The title of this section rather overstates its contents. One could write
endlessly on the subject. Rather, its purpose is to emphasize the variety of
techniques available but which have not been mentioned elsewhere in this
chapter. Ways of avoiding decomposition or, rather, reaction, in the presence
of oxygen and water have been mentioned. No less important is the ability
to avoid thermal decomposition. There are two related techniques available
here. In both, reactants are cocondensed on a cold surface, cooled to anything
from the temperature of liquid helium upwards. If the spectroscopic prop-
erties of the unstable species are the point of interest then the reactants are
condensed along with an inert diluent, typically a noble gas such as argon,
to give the product in a matrix of the noble gas-the so called matrix-
isolation technique. If the preparation of large quantities is of more interest
then no matrix is used. Typically, metal atoms are evaporated into the high
vacuum chamber that holds the cooled surface, typically cooled with liquid
nitrogen. Many heating techniques-resistive heating, induction heating,
laser ablation are a few-may be used to generate the metal atoms. Along
with the metal, but perhaps in alternate pulses, are condensed the chosen
ligand(s). By such techniques tin carbonyls, compounds unstable at tempera-
tures well below room temperature, have been prepared and characterized.
Paradoxically enough, such metal atom synthesis methods can also provide
a convenient high yield route to compounds which are stable at room
temperature; commercial equipment is available for those wishing to work
on a large scale. A simple version is also available for use in student
laboratories. It is methods such as these which are being used to prepare
fullerene, C 60 , and related species. This soccer-ball-like molecule can encap-
sulate some ions-Sr2+ and LaH are two examples-if suitable sources for
them are present in the carbon rods from which the fullerene is prepared
(they are evaporated by an electric arc struck between them to give a low
yield of fullerenes). The bonding is these so-called endohedral molecules is
discussed in Chapter 10.
Photochemical methods, usually irradiation with ultraviolet light from a
mercury discharge lamp, but sometimes visible light too, have long been used
in the preparation of coordination compounds. They depend on the fact that
the chemistry of an electronically excited molecule is different from that of
the same molecule in its electronic ground state. However, this can only
be exploited synthetically if the excited state lives long enough for chemistry
to be performed on it-and most excited states have very short lifetimes
because excited molecules give up their extra energy to other molecules in
a wide variety of processes. It is likely, therefore, that the success of
photochemical methods depends, in part, upon there being an insulated step
down the ladder of energy changes that lead to deactivation; a level of long
lifetime, from which deactivation is slow. Such levels exist-lasers depend
on them for their action-but it is difficult to predict them and so to predict
whether photochemical methods will lead to new compounds or just to the
destruction of those already present. One compound which is always made
photochemically is the carbonyl Fe 2 (C0) 9 , by the reaction
h,.
2Fe(C0) 5 - Fe2 (C0) 9 + CO
In recent years it has become clear that there exist complexes which
contain the H 2 molecule as a ligand; not two separated H atoms, but H 2 •
One method by which such complexes might be prepared is by the action
of H 2 on suitable coordination compounds. Unfortunately, there is a
problem-the solubility of H 2 in most solvents is rather low, so high
pressures of H 2 are needed. Not surprisingly, most workers would prefer to
work with low pressures of H 2 than with high. It is here that the ingenuity
of experimentalists becomes apparent. Supercritical fluids have properties
which in some ways resemble those of liquids. For example, they can act as
solvents for coordination compounds, and in some ways they resemble gases
in that they are miscible with gases, usually over the entire concentration
range. Here, then, is a solution to the problem-study the reaction of
coordination compounds with H 2 using supercritical fluids as solvents. The
pressures needed to maintain suprecriticality are often modest-a few tens of
atmospheres for carbon dioxide or xenon, for instance. At the time of writing,
this is an area in its infancy but it could lead to important new developments
in the preparation of coordination compounds. 2
Finally, the solid state. The preparation of solid state compounds by high
temperature synthesis is a long established method. Unfortunately, the
available techniques have been rather limited-grind the reactants together
to a fine powder, heat, regrind, reheat and so on until uniformity is
reached-is a typical procedure. But interest in the solid state is growing
rapidly. For instance, some simple inorganic materials show long-range
structural correlations which are difficult to understand or reproduce (for
instance, despite the simple picture presented in most introductory inorganic
textbooks, ZnS has been found to crystallize in several hundred different,
but related, crystal structures, although there is no known method of
controlling which form is produced). Again, the discovery of ceramic
high-temperature superconductors, (a topic which is dealt with in Appendix
10) incorporating Cum, a rather unusual valence state, has sparked off a
search for similar novel properties of solid state materials containing metal
ions in unusual valence states.
There have been developments in synthetic methods. First, related to the
'heat and grind' method, gel/colloid methods of producing reaction pre-
cursors are giving much more control and more reproducibility of the final
product. Secondly, hydrothermal methods are finding utility. In these,
reactions are carried out at high temperature and pressure conditions
in the presence of a solvent (not necessarily water, although this has been
most commonly used). For instance, Fe"Fe~'F8 • H 2 0 was prepared in this
way using liquid HF as a solvent.
2 SeeM. Po!iakoff, S.M. Howdle and S. G. Kazarian, Angew. Chern., Int. Ed. (1995), 34, 1275.
Further reading Specific

• Reactions of Coordinated Ligands (2 volumes), P. S. Brater-
General man (ed.), Plenum, New York, 1987.
Whilst many textbooks of practical inorganic chemistry pro-
• E. C. Constable, Metals and Ligand Reactivity, E. Horwood,
vide useful information, the most general source, used for much
New York, 1990.
of the material in the present chapter, is the series Inorganic
Syntheses that McGraw Hill started in 1939 and is still
continuing, although now published by J. Wiley (over 20 vol- A recent discussion of template reactions is 'Template
Syntheses', a long review by R. Hoss and F. Vogtle in Angew.
umes have been published).
Chern., Int. Ed. (1994), 33, 375.
Another source, almost as valuable, is Handbook of Prepara-
tive Inorganic Chemistry, G. Bauer (ed.) in two volumes,
A useful source is Volume 1 of 'Comprehensive Coordina-
tion Chemistry', G. Wilkinson, R. D. Gillard and J. A. McClev-
Academic Press, London 1963 and 1965.
erty (eds.), Pergamon Press, Oxford, 1987, Chapter 7.4 ('Reac-
A standard reference is D. F. Shriver, The Manipulation of
tions of Coordinated Ligands' by D. St. Black) and 7.5 ('Reac-
Air-Sensitive Compounds, McGraw Hill, New York, 1969.
tions in the Solid State' by H. E. LeMay Jr.).
Other useful texts include:
Other useful texts include:
• D. M. Adams and J.B. Raynor, Advanced Practical Inorganic • F. Basolo and R. G. Pearson, 'The trans effect in Metal Com-
Chemistry, Wiley, London, 1965. plexes', Prog. Inorg. Chern. (1962) 4, 381.
• G. Pass and H. Sutcliffe, Practical Inorganic Chemistry, • J. R. Blackborow and D. Young, Metal Vapour Synthesis in
Chapman and Hall, London, 1974. Organometallic Chemistry, Springer-Verlag, Berlin, 1979.
• R. J. Angelici, Synthesis and Technique in Inorganic Chemistry, • A. Rabenau, 'The Role of Hydrothermal Synthesis in Prepar-
Saunders, Philadelphia, 1977. ative Chemistry', Angew. Chern., Int. Ed. (1985) 24, 1026.
Questions 4.3 In the preparation of the complexes Cr(CO),X (X= sul-

fur donor ligand) the recommended method is to UV-irradiate
4.1 The cation [Mo(H 2 0) 6 ]J+ can be prepared by dissolving a solution of Cr(C0) 6 in THF (tetrahydrofuran) in an oxygen-
K 3 [MoCl 6 ] in 0.5 M p-toluenesulphonic acid (which behaves free apparatus to give the complex [Cr(C0) 5 (THF)]. This is
similarly to the triflic acid discussed in the text) and allowing followed by reaction with the ligand X. Direct reaction between
the solution to stand at room temperature in the absence of 0 2 , Cr(C0) 6 and X generally leads to sulfur-bridged complexes.
for a day. Suggest why this sequence is more successful than Explain the thinking behind this route to Cr(CO),X.
the action of H 2 0 on K 3 [MoC1 6 ].
4.4 One procedure for the preparation oflr4 (C0) 12 is to start
4.2 Heating cis-[Co(en),Cl,]Cl in triflic acid, CF3 S03 H (TH), with commercially available Na 2 [lrCl 6 ] in ethanol and reduce
at 100 'C for 3 h gives cis-[Co(en) 2 T2 ]T. Surprisingly, it was r.
it to the [IrC1 6 ] 3 - ion with Gaseous CO is then passed
found that trans-[Co(en) 2 Cl 2 ]Cl also gives the cis product into the solution, and together with the addition of solid
when heated with TH. Suggest a possible synthetic exploitation K 2 C03 , leads to the product, a black solid. Give a retrospective
of this observation. rationale of this sequence.
Stability of coordination
compounds
5.1 Introduction
The statement that a compound is stable is rather loose, for several different
interpretations may be placed upon it. Used without qualification it means
that the compound exists and, under suitable conditions, may be stored for
a long period of time. However, a statement such as ' a compound is stable
in water' may mean one of two things, either that there is no reaction with
water which would lead to a lower free energy of the system (thermodynamic
stability) or that, although a reaction would lead to a more stable system,
there is no available mechanism by which the reaction can occur (kinetic
stability). For example, there may not be enough energy available to break
a strong bond, although once broken it could be replaced by an even stronger
one. As we have seen, boron trifluoride forms a stable complex with
trimethylamine, [BF3 (N(CH 3 h)]. A similar complex is formed with trisilyl-
amine, [BFiN(SiH 3 lJ)], which is thermodynamically unstable with respect
to the reaction
[BF3 (N(SiH 3 h)J .... (BF2 )N(SiH 3 ) 2 + SiH 3 F

The complex [BF3 (N(SiH 3 h)] can be prepared and stored at low tempera-
tures ( ~ -80 oq since the decomposition then proceeds very slowly-at this
temperature the complex is kinetically fairly stable. At room temperature
the complex is kinetically unstable and the rate of decomposition is much
greater. This is the key distinction made in Chapter 4 between kinetically
inert and kinetically labile complexes. There it was pointed out that the species
which crystallizes from a solution of a mixture of related labile complexes
depends not only on the cation and ligand concentration but also on the
solvent and crystallization temperature. Although it may be a relatively
minor component in the solution, the least soluble complex is probably the
one which crystallizes. In the solution there is a series of equilibria such that,
74 1 Stability of coordination compounds
if one component crystallizes, the concentrations of the others also change.

This chapter is devoted to a discussion of the stability constants which
characterize such equilibria.
5.2 Stability constants

When a complex is formed by the reaction 1
M + L.,tML
the equilibrium constant K 1 for the complex containing a single ligand will
be
[ML]
K1 = [M][L]
where, for the moment, activity coefficients of unity have been assumed. If
ML adds a further molecule of L,
ML + L+t ML,
then the equilibrium constant for the complex containing two ligand
molecules is
In general, the equilibrium constant for the formation of the complex ML.
from ML._ 1 will be
Kn = [Ml.,]
[Mt.,_ 1 )[L]
The equilibrium constants Kt> K 2 , ... , K. are known as stepwise formation

constants. Alternatively, one may consider the equilibrium constant for the
overall reaction
as
[Ml.,]
Pn = [M][L]"
Pn is known as the nth overall formation constant. Pn is related to the stepwise

formation constants Kt> ... , K. by
Pn = K1 X K, X · · • X Kn
That is
1 Throughout this chapter we shall often not specify the charges on the species in reactions
or equilibria. Square brackets are used to indicate both the concentrations of complex species
and the species themselves. It will be clear from the context which is intended.
Determination of stability constants 1 75
Table 5.1 Typical stability constant data for monodentate ligands. All values are logarithmic so, for the Sn 2 + fCI- system,
logK1 = 1.51
Metallon Ugand Stability constants
Sn 2 + c;- K1 = 1.51 K2 = 0.73 K3 = -0.21 K4 = -0.55

Pd2 + c;- K1 = 6.1 K2 = 4.6 K3 = 2.4 K4 = 2.6 K5 = -2.1 K6 = -2.1
Ni2 + NH 3 (30 'C) K1 = 2.67 K2 = 2.12 K3 = 1.61 K4 = 1.07 K5 = 0.63 K6 = -0·09 {36 = 8.01
Cu 2 + NH 3 (30 'C) K1 = 3.99 K2 = 3.34 K3 = 2.73 K4 = 1.97
Mn 2 + F- K1 = 5.52 {J, = 9.04 {J, = 11.64 p. = 13.4 {J5 = 14.7 {36 = 15.5
Pb2 + ;- K1 = 1.98 p, = 3.15 {J, = 3.81 fJ. = 3.75 {J5 = 3.81
Fe 2 + CW {J6 =24
Fe3 + cw {36 = 31
These data refer to 25 'C unless otherwise stated and to zero ionic strength. As a comparison, the data for Ni 2 • /NH 3 (30 'C) in 2M NH4 NO:, are:
K 1 = 2. 78 K, = 2.27 K3 = 1.65 K4 = 1.31 K5 = 0.65 K6 = 0.08
Notice, particularly, the effect of the change on Ks· The entries included in this table have been chosen to illustrate the variety of formats that are
encountered and yet to be internally self-explanatory. As an example, the statement that K1 = 1.51 for the Sn 2 + ;cl- system is to be interpreted as
[SnCj+J = 101·51 = 32.6 L mol- 1

•][cn
[Sn 2
There are the same number of overall formation constants as stepwise

formation constants:
fJ1 = K1; {J2 =K1 xK2 ; P3 = K1 x K2 x K3 etc.
Some typical stability constants are given in Table 5.1. A point to

remember is that when values of K., K.+ 1 etc. are similar then an equilibrium
solution will contain mixtures of the complexes (the Pb 2 +;r case in Table
5.1); when K., K.+ 1 values are very different then it is possible to obtain
solutions containing, essentially, only a single complex. Generally speaking,
K 1 > K 2 > K 3 etc. but, as Table 5.1 illustrates, exceptions occur. So, notice
in this table the expression of the fact that the common anionic chloro
complex of Pd2+ is [PdCI 4 ] 2 -.
5.3 Determination of stability constants

In order to determine the values of n formation constants, n + 2 independent
concentration measurements are needed. These can then be used to obtain
the concentrations of the n species ML, ML 2, ..• , ML. and also those of M
and L. Two pieces of information are at once available; we (should!) know
the quantities of M and L (or alternative starting materials) used in the
measurement. This means that n additional pieces of information are needed.
If it is certain that only one complex, of known empirical formula, is formed,
then a measurement of the concentration of the uncomplexed M or L is
sufficient to determine the formation constant. This measurement can be
made in many ways: by polarographic or emf methods (if a suitable reversible
electrode exists), by pH measurements (if the acid dissociation constant of
HL is known) and by many other techniques, including the whole galaxy of
spectroscopic methods. A recent source book for stability constant data (that
by Connors, see Further Reading) distinguishes over 30 methods.
For the more general case where more than one formation constant is to
be determined, the problem is usually more difficult. For inert complexes it
may be possible to separate, and separately obtain the concentrations of,
the various complex species. In this way Bjerrum was able to determine the
six stability constants within the [Cr(H 2 0) 6 ]3+, [Cr(H 2 0) 5 SCN] 2 +, ... ,
[Cr(SCN) 6 ] 3 - series. However, this is a potentially unreliable method and
has been little used. Some methods of tackling the general problem will now
be indicated. The variants are many for this is a field in which considerable
ingenuity has been used in the design of experiments and in the analysis of
experimental data. As an example consider a ligand for which the species
H 2 L, HL- and L 2 - all exist. The metal ion M and protons may be regarded
as being in competition for L 2 -. If we titrate H 2 L against standard NaOH
solution we obtain a pH-volume curve of the form shown dotted in the
upper part of Fig. 5.1. Now add a known amount of M. Some H+ will be
displaced. If the mixture is again titrated with NaOH the curve shown dotted
in the lower part of Figure 5.1 is obtained. Compare the titres [NaOH] 1
and [NaOH] 2 at the same pH (shown dashed in Fig. 5.1). Now, the H+
liberated by the added M is dependent on the amount of ligand bound to
M, ligand which was previously protonated, the average number of H+ ions
per ligand being nH (nH = 2 for H 2L). We have
[W ion liberated] = [L bound to M] x nH = [NaOH] 2 - [NaOH] ~
So,
[L bound toM]= [NaOH],- [NaOH]~
nH
Define ii as the average number of ligand molecules bound to M, so

11 = [L bound to M] = [NaOH], - [NaOHL
[MJ nH[M]
all the quantities in the right hand side expression are known.
Fig. 6.1 The experimental sequence is shown

by the solid circles on the titration curve. A r
c.
protonated ligand is ~\rated against NaOH to a
suitable pH. Metal ion is added (indicated by
[M]) and the titra~on continued until the same
pH is regained.
NaOH--
Determination of stability constants I 77
By such measurements the average number of ligand molecules complexed

with each metal atom, n, is obtained. Although it is always true that n
indicates the most abundant species, i.e. n = 3 means that the concentration
of ML 3 is a maximum provided that equilibrium has been established, it
does not immediately tell us anything about the other species. However,
measurements on a variety of solutions with different n can. We first note
that n is related to the composition of the mixture of complex ions by the
equation
N
I n[ML,]
n= 1[ML] + 2[ML:z] + · · · + N[Ml,.] n=1
[M] + [ML] + [ML:z] + · · · + [Ml,.] [M],
where the complex species are ML, ML 2 , ..• , MLN and [M], is the total
concentration of metal ion, complexed and uncomplexed, in the solution.
Now,
N N
[M], = I [ML,] = [M] + I [ML,]
n=O n=1
and
[Ml.,] = fln[M][L]"
so, substituting for [M], and [ML.] in the equation above,

N
I nPn[M][L]"
n=O
n= " N
[M] + I Pn[M][L]" 1 + I Pn[L]"
n=O n=O
If a series of solutions, with varying n, is prepared and [L] measured in each,

then values of P., n = 1, 2, ... , N, must be chosen so that the above equality
holds. This may be done, for example, by plotting n against [L] and
determining the ps by an iterative, best fit, method, most readily carried out
using a computer. A large number of programs exist; a book detailing many
of them is referenced at the end of this chapter. One further point: do not be
deceived by the fact that there is no explicit dependence of n on [M] in the
above expression. [L] is the free ligand concentration and its value is
dependent on [M].
If it is easier to measure [M] than [L], a different but related relationship
is used. We know that
[M], = [M] + [ML] + [ML:z] + · · · + [ML,.]

[M] [M]
= [~] ([M] + MMJ[L] + p2[M][L] 2 + ... + PN[M][L]")

= 1 + p1[L] + p2[L] 2 + ... + PN[LJ"
that is,
[M],
[M]
= 1 + I
n=O
/Jn[L]"
If both [M] and [L] can be measured, keeping [M], constant, but changing
the total ligand concentration for each measurement, this relationship leads
to a set of simultaneous equation in the f3s (one equation for each
measurement) which provide a quick way of determining f3 values. Alterna-
tively, if [L] is made large, so that [L] » [M], it is essentially constant. By
varying [M], and measuring [M], the f3s may similarly be found.
A variety of optical methods is used to determine both complex formation
and stability constants. Job's method of continuous variations is the best
known. A wavelength is chosen at which the complex in question absorbs
(usually in the visible or near-ultraviolet region). Several solutions are
examined, for all of which ([MJ. + [L],) is a constant, C, although each has
different values for [M], and [L],. Here, [L], is the total concentration of
ligand, free and complexed. Some measure of the intensity of absorption
(absorbance or optical density) is plotted against composition (usually
against the ratio [LJ./([M], + [L],); that is, against [L],/C, a ratio we shall
call a. As a simple example of the application of this method consider the
case where only a single complex is formed. If a single complex is formed
then we have an equilibrium
M + nL <=> ML,
Initial concentration (1- ~)c ~c o
Final concentration (1- ~- y)C (~- n,)C yC
where, as follows from the definition of a 1, aC is the initial concentration of

the ligand L, yC is the final concentration of the complex species ML. and
n is a fixed (but unknown) number, although we expect it to be an integer.
Explicitly, the equilibrium concentrations are
[M] = (1 - a - y)C
[L] = (~ - ny)C
[MLnJ = yC = /in[M][L]n
The values of [M], [L] and [ML.J all change with a; we wish to find the
relationship that holds when [ML.J is a maximum. To do this we differen-
tiate each of the above equations with respect to a. We obtain
d[M] = -C
d~
d[L] = C
da
d[Ml,] = lin[L]n-l([L] d[M] + [M] n d[L])

d~ da d~
To find the maximum in [ML.] we set the last equation to zero and
substitute the first two. The result is
[L]( -C) + [M](nC) = 0
Introducing the explicit expressions for the concentrations [L] and [M] we
obtain
-(~max- ny) + n(1- ~max- y) = 0
Determination of stability constants 1 79
1l
!6
FJC. &.2 The Job's plot obtained when only one €
complex species, ML,, is formed between M <~
and L
0.0 0.2 0.4 0.6 0.8 1.0 [M]

1.0 0.8 0.6 0.4 0.2 0.0 [L]
that is,
n
"'max=l+n
It follows that since ac is known (from the composition with which the
solution was made up) then, if the visible or ultraviolet wavelength chosen
corresponds to absorption by the complex, then n can be determined from
the maximum in plot of absorbance against cc. Conversely, if the absorption
at the chosen value is primarily due to either M or L, the minimum of the
absorbance versus ac plot gives n. Figure 5.2 shows a typical Job's plot for
a ML2 complex (ccmax = i). The rounding at the peak is due to the fact that
we have chosen to show the case of a not-very stable complex. The more
stable the complex the less rounded the peak; indeed, in favourable cases
the peak shape may be used to determine P•. When more than one complex
is formed there will usually be a corresponding number of peaks in a Job's
plot, but stability constants may only be determined from such a plot under
very special circumstances.
There is another method for the determination of stability constants
which, although not much used for this purpose, is of some value in reaction
kinetics as a method of estimating rate constants which cannot be measured
readily (see Chapter 14). Suppose the reaction
ML, + L-+ Ml.,+ 1

proceeds by a one-step process. The rate of the formation of ML.+ 1 is
kr[ML.][L], where kr is the rate constant of this forward reaction. The
corresponding backward reaction is
and the rate of disappeance of ML.+ 1 is kb[ML.+ 1], where kb is the rate
constant of the backward reaction. At equilibrium,
SO 1 Stability of coordination compounds
Therefore,
[Ml,+1] - K - kt
[Ml,][L]- n + l - if;;
That is, if the rate constants kc and kb can be independently determined, the
corresponding stepwise stability constant is given by their quotient.
So far, activity coefficients of unity have been assumed, but they seldom
have this value. Consequently, the stability constants determined above
are not constants at all, but are a function of concentration. The simplest
way out of this difficulty is to determine the stability constants over a range
ofreactant concentrations and then extrapolate to zero concentration (where
activity coefficients are unity). K. and fJ. at zero concentration are true
thermodynamic equilibrium constants and are distinguished by the super-
script T, thus TK. and TfJ. (although, to avoid complexity early in the
chapter, they were not designated as such in Table 5.1). More commonly,
activity coefficients are kept essentially constant by carrying out measure-
ments in the presence of a backing electrolyte which keeps the ionic strength
of the medium constant. Stability constants obtained in this way are
proportional to the thermodynamic stability constants, the constant of
proportionality being a function of the activity coefficients of the species
involved in the complex. Such stability constants are referred to as stoichio-
metric stability constants. Strictly, it is the determination of these which we
have discussed-they are given the symbols K. and fJ. which we have been
using.
5.4 Stability correlations

Two schemes have been proposed which systematize the available stability
constant data. In addition to data contained in the compilations given at
the end of this chapter, more qualitative evidence, based for example on the
results of displacement reactions, has been included in arriving at the
generalizations. Historically, the first scheme is that due to Chatt and
Ahrland who pointed out that electron acceptors may be placed in one of
three classes. Class-a metals, the most numerous, form more stable complexes
with ligands in which the coordinating atom is a first-row element (N, 0,
F) than with those of an analogous ligand in which the donor is a second-row
element (P, S, Cl). Class-b has the relative stabilities reversed. It is not difficult
to extend the stability relationships to include heavier donor atoms. Class-a
behaviour is, then, typified by a stability order
F- » Cl- > Br- > 1-
0»S > Se > Te
and
N»P>As>Sb>Bi
Class-b behaviour is rather more complicated and is typified by relative
stability constants in the order
F- «CI- < Br- <1-
0«S ~ Se ~Te
Stability correlations 1 81
and
Table 5.2 Classification of acceptor
species {after Ahrland, Chatt and Davies) N«P>As>Sb>Bi
In addition, there is a third class of electron acceptor for which the stability
Class-a behaviour
H, the alkali and alkaline earth metals, the constants do not display either class-a or -b behaviour uniquely. The
elements Sc -+ Cr, AI -+ Cl, Zn -+ Br, In, Sb class-ajclass-b classification of some metal ions is given in Table 5.2 (normal
and I, the lanthanides and actinides valence states are assumed).
Class·b behaviour Although not included within the above classification, there are some
Rh, Pd, Ag, lr, Au and Hg other useful gradations which have been noted and are conveniently included
Borderline behaviour
at this point. For a given ligand, corresponding stability constants of
The elements Mn -+ Cu, Tl -+ Po, Mo, Te, complexes of bivalent ions of the first transition series are usually in the
Ru, W, Re, Os and Cd natural order (sometimes called the Irving-Williams order):
Mn 11 < Fe 11 < Co 11 < Ni 11 < Cu 11 > Zn 11

Copper(II) does not coordinate a fifth and sixth ligand particularly strongly,
and this order is incorrect for stability constants relating to CuL 5 and CuL 6
complexes. Complexes of chelating ligands also tend not to follow this order.
For non-transition metal ions complex stability decreases roughly in the
order of ionic potential (or polarizing power), which is defined as (formal
charge)/(ionic radius). Thus, corresponding stability constants decrease in
the order
Li+ > Na+ > K+ > Rb+ > cs+
Mg2 + > Ca 2 + >sf!+> Ba 2 + > Ra 2 +
and
Al 3 + > Sc3 + > ya+ > La 3 +
provided that the ligand is not changed from one ion to the next and subject
to the absence of size-matching conditions that will be discussed later in this
chapter. Similarly, for approximately constant ionic radius, the stability
constants are in the order of decreasing charge, thus,
Th 4 + > Y3 + > Ca 2 + > Na+
and
La 3 + > s,.:<+ > K+
Although the formal charge on an ion is a well-defined quantity, the actual

charge (which is not so well defined) may differ significantly from it-we
shall meet this point again in Section 7.5. Further, the concept of ionic radius
is also somewhat elusive. If determined from crystal structures-so that, for
example, the radius of Cl- is given by the point of minimum electron density
along the Na +--ci- axis in NaCI-then it is not a quantity which is
constant because it varies somewhat with the crystal structure chosen to
measure it. In retrospect, it is perhaps surprising that the above inequalities
hold as well as they do!
The second and more recent approach which has been used to classify
metal ion-ligand interactions is based on the concept of hard and soft acids
and bases, often denoted HSAB. Hard metal ions are those which parallel
the proton in their attachment to ligands, are small, often of high charge
and have no valence shell electrons that are easily distorted or removed. Soft
metal ions are large, of low charge or have valence shell electrons which are
easily distorted or removed. They bond strongly to highly polarizable

ligands-which often have a very small proton affinity. Similarly, ligands are
divided into those that are non-polarizable (hard) and those that are
polarizable (soft). Remembering the Lewis definition of acids and bases as
electron acceptors and electron donors respectively, the cations are classified
as either soft or hard acids, whilst the ligands are classified as either soft or
hard bases-hence, HSAB. An important empirical generalization is that the
most stable complexes are those of soft acids with soft bases and of hard
acids with hard bases.
This approach has the advantage that it is not restricted to complexes of
transition metal ions. Indeed, it may be applied to a wide range of chemical
equilibria and systemizes a great deal of chemical intuition. A particularly
useful concept is that of symbiosis. A cation which is classified as a relatively
hard acid (or, indeed, which is regarded as borderline) is made softer by the
coordination of a soft ligand (or harder by the coordination of a hard ligand)
and so it is more likely to add further soft (or hard) ligands. For the
non-transition elements in particular, the principle of hard or soft acids and
bases systematizes stability data in a useful way. In Table 5.3 some metal
ions and ligands are classified as hard or soft. Generally, ligands in which
the coordinating atom has a high electronegativity are hard bases; those in
which it has a low electronegativity are soft.
Recent work2 has made the connection between electronegativity and
the hard-soft/acid-base concept clearer and has served to give the latter
Table 5.3 Classification of some (formally) ionic species as hard and soft acids and
bases (after Pearson)
Hard acids H+ u+ Na+ K+

Be"+ Mg"+ Ca2 + sr>+ Mn2 +
Al3 + Sc3 + Ga3 + ln 3 + la3 +
cr"+ eo3+ Fe3 + ee3+
Si4 + Ti4+ zr"+ lh4+
vo•+ vo~+ Ma03+
Soft acids cu+ Ag+ Au+ 11+ Hg+

Cd 2 + Hg2+ Pd2 + pf+
Tl 3 +
Pt4+
Borderline Zn 2 + Sn 2 + Pb2 +
Fe2 + Co2 + Ni2 + Cu 2 + Ru 2 + Os2 + Sb3 + Bi3 + Rh3 + lr"+
Hard bases H20 R20 ROH NH3 RNH2

OW OR- C104 NO:l CH:l CO:l
so~- co~-
Po~-
Soft bases R2S RSH P3 P R3 As

Rs- 1- sew eN- w R-
S 3 0~-
Borderline Cl- Br- N:l NOi so~-
2 R. G. Pearson, J. Chern. Ed. (1987) 64, 561 and Inorg. Chern. (1988) 27, 734.
Stability correlations I 83
Fig. 5.3 The plot used to define the

electronegativity and hardness of the species
M" + in terms of I and A. See the text for a full
discussion.
A
___ j ___ _
----------2-
Number of electrons ------
concept a quantitative basis. This work starts from a rather different

direction to that given above but, in the end, arrives at the same point.
Consider Fig. 5.3. This figure shows a plot of the total energy of an atom
(or molecule). If we start with M"+ (the central point on the graph) then
the energy change to give M(n+ I)+ is the ionization potential of M"+; this
we will call I. Similarly, the energy change to give Mrn-l)+ is the electron
affinity of M"+ (it is the negative of the ionization potential of Mln-l)+).
Call this A. Following Mulliken's definition of electronegativity, we equate
(I + A)/2 with the electronegativity of M"+. From Fig. 5.3 we see that if the
curve joining Ml•+ I)+ and Ml•-ll+ were a straight line then (I+ A)/2 would
be the slope of that straight line. To a first approximation, (I+ A)/2, the
electronegativity of M"+, is the slope of the curve at M"+ However, the
curvature of the plot, the deviation from a straight line, is measured by the
difference between I and A.
The approximate curvature at M"+, (I- A)/2, may be taken as a measure
of the hardness of M"+. Physically, this definition relates hardness to a
resistance to deformation of charge. Hard metal ions are those for which it
costs a great deal to change electron densities. Unfortunately, this model
cannot immediately be extended to all ligands; for anions, in particular,
electron affinities are unknown. In such cases it seems that use of data for
corresponding neutral species gives reasonable results. In Table 5.4 are given
values of absolute hardness for typical metal and ligands.
At several points in this book there are mentioned the consequences
of particular HOMO-LUMO (highest occupied molecular orbital-lowest
unoccupied molecular orbital) energy separation patterns. The present
discussion provides another. Hard ions or molecules have a large HOMO-
Table 5.4 Absolute hardness values for common metals and ligands
Species Ionization Electron Absolute hardness,

potential, I affinity, A i<I-A)
Cations
Na+ 47.29 5.14 21.08
K+ 31.63 4.34 13.64
cu+ 20.29 7.73 6.28
Cu 2 + 36.83 20.29 8.27
Ag+ 21.49 7.58 6.96
Mg2+ 81.14 15.04 32.55
Ca 2 + 50.91 11.87 19.52
cr"+ 30.96 16.50 7.23
CrH 49.1 30.96 9.1
Mn2 + 33.67 15.64 9.02
Mn3 + 51.2 33.67 8.8
Co 2 + 51.3 17.06 8.22
Co3 + 51.3 33.50 8.9
Pd 2+ 32.93 19.43 6.75
Pd2 + 31.94 18.76 8.46
Ugands
co 14.0 -1.8 7.9
PF3 12.3 -1.0 6.7
C2H4 10.5 -1.8 6.7
C5 H5 N 9.3 -0.6 5.0
c2c2 11.4 -2.6 7.0
H2 0 12.6 -6.4 9.5
(CH3),S 8.7 -3.3 6.0
NH 3 10.7 -5.6 8.2
F' 17.42 3.40 7.01
Cl 8 13.01 3.62 4.70
Br' 11.84 3.36 4.24
a Data for atoms; taken as models for the corresponding anions.
LUMO separation; soft ions or molecules have a small HOMO-LUMO

difference-this is just another, orbital, interpretation of the quantity (I - A).
With these recent developments, it seems that the concept of hardness and
softness will increasingly replace the older class-ajb distinctions. However,
the latter nomenclature is widespread in the chemical literature and likely
to remain so for some time. Indeed, the a/b distinction separates species that,
from Table 5.3, show relatively small differences in hardness and so, perhaps,
the notation will remain useful.
5.5 Statistical and chelate effects

The variations in stability constants which have been discussed so far in this
chapter are typically rather large. Superimposed on them are some smaller
but systematic variations which also merit attention.
Most stability constant measurements are made in aqueous solutions
at a constant ionic strength. The species we have called M must, in reality,
be a complex mixture of species. The majority will probably be [M(H 2 0).]'+,
where n may well be four or six although larger numbers also occur. Other
Statistical and chelate effects 1 85
species present will include foreign anions-ones not involved in the reaction,
but arising from the salt used to maintain a constant ionic strength-and
some of these might well be coordinated to M. The addition reaction
Ml., + L-+ Ml.,+1
should more properly be written as a substitution reaction (L' is usually
H 2 0 but in this section is to be generally taken to mean 'solvent molecule'):
ML;,Ln + L-+ ML;,_1Ln+ 1 + L'
In this reaction it has been assumed that the number of ligands around M
is constant, and equal to (m + n). If we let this number (usually four or six)
be N, then the reaction may be written as
Ml.N-nl., + L-+ Ml.N-n-1Ln+1 + L'
As has been seen in Table 5.1, it is commonly found that for a given M and
L there is a decrease in successive formation constants, K 1, K 2 .•• or /3 1,/32
... This is largely a statistical effect. It is easier to attach L to ML~ than to
ML~_ 1 L because there are N reaction sites on the former but only
N- 1 on the latter (replacement of L in ML~_ 1 L by L gives an identical
molecule).
Consider the equilibrium:
Ml.N-nl., + L <" Ml.N-n-1l.,+1 + L'
The rate of formation of ML~-.- 1 L.+ 1 is, for a simple one-step reaction,
k,[ML~-.L.][L]. If the statistical effect is the only one which varies as n
changes, the forward rate constant will vary with n in proportion to the
number of reaction sites, that is, the number of L' ligands in ML~-.L•. We
therefore rewrite the above rate as k((N- n)[ML~_.J[L'], where k( is
expected to be constant for all values of (N- n). Similarly, for the rate of
disappearance ofML~-.- 1 Ln+ 1 , kb[ML~-.- 1 L.+ 1 ][L], the backward reac-
tion rate constant, kb will vary with n in proportion to the number of L
ligands in ML~-.- 1 L.+ 1 • Hence, in the above cases kb = k~(n + 1), where
k~ is constant for all values of (n + 1). It follows that
k, ki(N -n)
Kn+1 = !(.; = kt,(n + 1)
In a similar way it is found that
K =kf(N-n+1)
n kt,n
The ratio between successive formation constants

Kn+ 1 = n(N- n)
Kn (n + 1)(N - n + 1)
The value of this expression for N = 6 and n = I -+ 6 are compared with

the experimental data for the Ni2+ /NH 3 system (where L' is H 2 0) in Table
5.5. The agreement is far from exact, but the predictions are of the correct
order of magnitude. The experimental ratios are, as is commonly the
case, rather smaller than those predicted statistically.
Table 5.5 Statistical predictions and experimental ratios for the equilibrium con-
stants of the system
[Ni(H20l6-n(NH3ln] 2 + + NH3 :;:=:: [Ni(H20ls-n(NH3)n_ 1 ] 2 + + H20
Ratio n Experimental ratio Statistical
(data from Table 5.1) prediction a
K2/K1 1 0.28 0.42

K3/K2 2 0.31 0.53
K4/K3 3 0.29 0.56
Ks/K4 4 0.36 0.53
KB/Ks 5 0.19 0.42
' Calculated using KM 1 ~ " 16 - n)
K, (n + 1)(7 - n)
Table 5.6 Comparative stability of the Ni 2 +/NH 3 and Ni 2 +;en

systems
Equilibrium• logK
Non-chelated complex
Ni 2 + + 2NH 3 :;:=:: [Ni(NH 3 )2]h 5.05
[Ni(NH3) 2 ] 2 + + 2NH3 :;:=:: [Ni(NH 3 ) 4 ] 2 + 2.96
[Ni(NH 3) 4 ] 2 + + 2NH 3 :;:=:: [Ni(NH 3 )6 ] 2 + 0.73
Chelated complex
Ni 2 + +en:;:=:: [Ni(en)] 2 + 7.51
[Ni(en)] 2 + +en:;:=:: [Ni(en) 2 f+ 6.35
[Ni(en) 2] 2 + +en""' [Ni(en) 3 ] 2 + 4.32
a Coordinated water is omitted for simplicity.
When a N-H bond in a ligand such as ammonia is replaced by a N-alkyl

bond, the corresponding stability constants of complexes are usually lowered.
This may be due partly to increased steric interaction in complexes of the
substituted ligand, but the observation that the stability constants of
complexes with sulfur-containing ligands are usually H 2 S < RSH < R 2 S-
that is, in the opposite order-suggests that other factors also operate
(for example, both the energy and the shape of the lone-pair-containing
orbital(s) of the coordinating atom will change slightly on substitution). Even
so, it might reasonably be expected that complexes of a bidentate ligand
such as ethylenediamine (NH 2...CH 2...CH 2-NH 2 ) are less stable than the
corresponding complex with two ammonia molecules. Quite the opposite is
true. Complexes containing chelate rings are usually more stable than similar
complexes without rings. This is termed the chelate effect, and is illustrated
in Table 5.6; it was first met in Section 2.1. The origin of this effect has been
the subject of some controversy. There have been those who have argued
that the effect does not exist (despite the experimental observation that a
ligand such as en seems invariably to displace NH 3 ; that bpy (2,2' -bipyridine)
replaces py; that oxalate displaces acetate and so on). The problem is a
fascinating one; it relates to the fact that quantities such as K 1 and K 2 etc.
have units (mol- 1 L) so that in comparing an equilibrium involving 2NH 3
with one involving en, the corresponding fJs have different units (mol- 2 U
Statistical and chelate effects I 87
and mol- 1 L, respectively). How does one compare quantities with different
units? This is not the end. When we write RT InK (as we shall do shortly)
we have to recognize that we can only take the logarithm of a number.
Where have the units of K gone? The answer here is that we really evaluate
RT ln(K/1), where the I refers to a standard state which has the same units
as K. This leads to a discussion of the characteristics of the standard
state and this, in turn, to the recognition that there are three different
concentration scales in use: molarity, molality and mole fraction. These
concentration scales are not directly proportional to each other so that, for
example, it is theoretically possible for a given solution to have activity
coefficients of unity on one scale but not on the others. The chelate effect is
not enormous and so it is not surprising that such fine points can assume
undue prominence and that controversy and apparent misunderstanding
exist! A very readable account of the general problem can be found in
Chapter 2 of the book by Connors and in the articles given at the end of
the chapter. Fortunately, we can side-step most of the pit-falls and learn
more about the chelate effect from a more detailed analysis of stability
constant data. To do this, we note the relationships
which show that the chelate effect could originate in either the heat term,
!.lH 0 , the entropy term, !.lS0 , or both.
To proceed further, we could analyse the equilibrium constants in Table
5.6, in pairs, using these thermodynamic relationships. We can simplify
matters by considering, instead, the equilibrium
[M(NH 3 ) 4 ] 2+ +en:;;=: [M(NH 3 ) 2 (en)] 2 + + 2NH 3
Available data at 298 K for M = Ni 11 and Zn 11 (remembering that !.lS0 is

quoted in terms of J mol- 1 K - 1 , whilst /.lH 0 and !.lG0 are given in kJ mol- 1 )
are
Ion Mf' - T~S 0 (kJ mol- 1 )

-3.4 -2.0 4.8 -1.4
-1.5 0.1 5.3 -1.6
These results are qualitatively general: the chelate effect is largely an entropy
effect; for non-transition metal ions the heat term is particularly small.
There are two important contributions to the !.lH 0 term when it makes
a significant contribution. First, when two monodentate anionic ligands
are brought together to occupy adjacent coordination sites in a complex,
there will be an electrostatic repulsion between them against which work
has to be done. The same is true for uncharged monodentate ligands because
such ligands are always dipolar. For chelating ligands the coordinating
centres do not have to be brought together and most of this repulsive
energy is 'built in' (it makes a contribution to the enthalpy of formation of
the ligand). That is, in the above equilibrium this contribution to the heat
term would be expected to favour chelated species because the oriented
ammonia molecules repel each other. The second, more variable, contribu-
tion to the 11H0 term comes from solvation energies. Each of the species in
the equilibrium will be solvated (by hydrogen bonding, for example). Further,
there will be anions closely associated with the cationic species. We shall see
in Chapter 14 that there is kinetic evidence that counterions, although not
directly bonded to the coordination centre--they are not in the first
coordination sphere--are often present in an outer sphere (the so-called
second coordination sphere). It does not seem possible to predict, in general,
whether this term makes a positive or negative contribution to 11H0 , but it
appears to make a positive contribution when the coordination centre is a
main group element.
The entropy effect is readily understood. An increase of randomness
is associated with a concomitant increase in entropy. A complex molecule
has a lower entropy than its separated and therefore independent com-
ponents. In the equilibrium above there is a 50% increase in the number of
independent molecules in the right-hand side compared with the left and,
so, the right-hand side is favoured. Another aspect of the entropy effect is
the following. When one end, and only one end, of an ethylenediamine
molecule is coordinated, the effective concentration of the other end in the
system, and the probability that it will coordinate, is high, because it is
constrained to stay close to the cation. This means that it is easier to form
a chelate ring than to coordinate two independent molecules because the
two acts of coordination are related for the former, whilst for the latter they
are entirely independent of each other. All these, then, are potential
contributors to the chelate effect. They are difficult to relate to the kinetic
data on the effect. These show that it occurs because the ring opening
is slower than expected for the dissociation of the first of two independent
ligands, not because the formation of the ring is faster (it is the ratio of these
two that is the equilibrium constant K ). It is understandable that the topic
continues to be the subject of debate.
The chelate effect varies with the size of the ring formed on coordination. It
is usually a maximum for five-membered rings and only slightly smaller for
six-membered rings. It is not difficult to see one reason why this should be.
For an octahedral complex, the angle subtended by a chelate ring at the
metal, no matter the size of the ring, will be close to 90°. For simplicity, we
fix it at this value. What will be the bond angles at the other atoms in the
ring? Assuming planar rings, the average values will be:
Ring size Average angle
4 90'
5 112'
6 126'
7 135'
So, if in a saturated ring system, such as that formed by en, the five-membered
ring system will have angles that are closest to the natural, tetrahedral, value
(109.SO). Similarly, for a ring that is conjugated, as in acac-, which may be
drawn as shown in Fig. 5.4, the six-membered ring has an angle closest to
the natural trigonal value of 120°. Of course, ring puckering can and does
occur, but can only be of a large magnitude at a cost of either steric
Solid complexes I 89
Me-..... ....,;CH /Me Me, ..-CH ..-Me

c:;;-- 'c 'c..- .;::.c..-
I II 11 I
0~ ,......o 0, ....,;0
(a) '-"M 'M:;;--
Fig. 5.4 (a) Two equivalent resonance hybrids
(canonical forms) of coordinated acac.
However, the conjugated nature of the system
is more commonly represented as shown in
(b).
interactions between rings or loss of conjugation within a ring. Further, the

stabilization of six-membered chelate rings is not confined to conjugated
systems, although the chelate stabilization of non-conjugated systems is
usually greater for five-membered rings, where comparison is possible.
Clearly, this discussion of the chelate effect could be elaborated considerably.
What of polydentate ligands? This is a topic very relevant to the ligands
encountered in the complexes of bioinorganic chemistry and discussion of it
will be deferred until Chapter 16.
Although complexes with rings of size other than five or six members
have been synthesized they show little sign of the chelate effect. This is partly
because of the reduced effective concentration of the other end of the ligand
for larger chelating molecules and because of the increase in work against
electrostatic forces needed to bring the coordinating atoms together. Further,
as has been mentioned, another potentially destabilizing influence for
larger-membered rings is the relative difficulty of finding a sterically non-
crowded ring configuration. The relationship between the geometry of free
and coordinated multidentate ligands is currently the subject of research,
particularly for the large ligands that occur in bioinorganic complexes
(Chapter 16); it also finds a faint echo in Appendix I. It is clear that ligand
geometry is an important factor in determining the relative stabilities of
complexes formed by such ligands. For instance, a coordinated ligand will
normally be less flexible than the free ligand. This loss of motional freedom
will be reflected in llS values and, thus, in the stability of the coordination
compound. We shall return to this point in Chapter 16.
5.6 Solid complexes

In this chapter our concern has been with equilibria in solution. As has
already been noted, it is sometimes possible to study a complex in solution
but not to obtain it in the solid state or vice versa. Why should stability be
related to phase in this way? It is because an additional energy factor is
involved, the relative lattice energies of the crystals which might be formed
on crystallization. Usually, the crystal form with the highest lattice energy
will be obtained (this will be the least soluble complex). For an ionic complex
species the lattice energy of a crystal will depend on the counterion
with which it crystallizes. Lattice energies are a maximum when the cation
and anion are of similar charge and size (and, usually, of the same hardness
or softness). This suggests a method by which crystals containing an elusive
complex may be obtained and, as we saw in Chapter 4, accounts for the

widespread use in preparative complex chemistry of such species as [PF6 ]-,
r,
[B(C 6 H 5 ) 4 [NEt4 ]+, and [As(C 6 H 5 ) 4 ]+. Such symmetrical cations and
anions have, however, one disadvantage. Because of their potentially high
symmetry (Oh or Th), they are somewhat spherical and have an unpleasant
tendency to adopt a disordered arrangement in a crystal, making structure
determination more difficult. This is one reason for the apparently perverse
habit of preparing compounds in which the symmetry of such counterions
is reduced by, for example, replacing one of the four phenyl groups in
[As(C 6 H 5 ) 4 ]+ with an ethyl to give [As(C 6 H 5 )J(Et)]+.
5. 7 Sterle effects
As has been indicated above, steric effects can play a part in determining
the stability of complexes. Indeed, this has become an area of much research.
If a very bulky ligand coordinates it may well form a weak bond, in which
case molecules containing it may well be rather reactive. If, on the other
hand, despite its bulk it coordinates strongly, then it may well cause the
bonding of other ligands to be weaker than normal. Again, an enhanced
reactivity is likely. Enhanced reactivity, particularly in the field of catalysis,
in which, often, a ligand dissociates from a metal ion at one point in the
mechanistic sequences and recoordinates at another, is always interesting.
A question at once arises: can we classify steric effect? The answer is that we
can, to some extent. Before attempting this, however, there is an important
point which has to be made. Consider a ligand such as ammonia but consider
it as a free molecule. The energy required to make it rotate, to spin, about
its threefold axis is low-it falls in the microwave region of the spectrum.
When the ammonia molecule is coordinated to a metal this rotation becomes
less free--it may well become a libration, in which it oscillates about its
threefold axis as a solid unit but does not overcome the barrier preventing
it reaching an equivalent position-but it is still of low energy. Everything
depends on the thermal energy available compared to the barrier height. At
room temperature the NH 3 groups in ammine complexes are commonly in
a state of large-amplitude oscillation, if not almost free rotation. This is
important because it means that the 'bumps' on the ammonia molecule--the
hydrogen atoms-may well behave as rather smeared out as far as steric
effects are concerned. It may not be too bad an approximation to regard
each ligand as effectively acting as a circular cone, with its apex at the metal
atom. This immediately leads us to some qualitative predictions. As far as
steric effects are concerned, we would expect the linear M-N=<:~S and
M-N=<:~e systems to be more stable than their bent isomeric counterparts
M-S M-Se
"'-c "'-c
~N ~N
A case where this may well be evident is in the square planar complex formed
between the tridentate ligand diethylenetriamine, NH 2-CH 2-CH 2-NH-
CH2-CH2-NH2, dien, and palladium( II), [Pd(dien)(SeCN)]. In this complex
the SeeN- anion is coordinated through Se and is bent. When the terminal
Steric effects 1 91
cone angle
Table 5.7 Cone angles for some simple systems, see Fig. 5.5
(ligand charges are not shown
Ligand Cone angle Ligand Cone angle

(degrees) (degrees)
H 75 PH 3 87
Me 90 PF3 104
2.2aA F 92 P(0Me) 3 107"
CO,CN,N 2 ,NO ~95 P(0Et) 3 108"
Cl, Et 102 PMe 3 118
Br, C6 H5 105 PCI 3 124
FIJI. &.& The definition of the cone angle tor a
symmetrical phosphine. PR3 • It is necessary, of I 107 PEt3 132
course, to define the cone angle with respect to i-Pr 114 PPh3 145
the metal atom to which the phosphine is t-Bu 126 P(i-Pr) 3 160
coordinated. It is assumed in the definition that CsHs 136 P(t-Bu) 3 182
this atom is 2.28A from the phosphorus. For a
a In obtaining these values, Tolman used a conformation which has yet to be
different coordinated atom a different
observed.
metal-ligand separation would have to be
assumed.
NH 2 groups of the dien ligand are replaced by NEt 2 groups, the SeeN-
anion is coordinated through nitrogen and is linear. Steric effects seem to
be the factor causing the change. Can the effect be treated in a reasonably
quantitative manner? The answer is yes. By a study both of crystal structures
and of molecular models, Tolman3 has compiled data for steric effects,
described in terms of cone angles, defined as in Fig. 5.5, for a large number
of ligands, particularly for those in which phosphorus is the donor atom. A
selection of his data is reproduced in Table 5.7. There have been many
variants in this field-the use of X-ray data to determine cone angles, the
use of detailed theoretical calculations, methods of averaging when very
different substituents are present, particularly on a phosphorus atom, and
so on. Generally, these other approaches lead to cone angles which are
somewhat smaller than those given by Tolman.
A rather different form of size effect occurs for those ligands which are
cage-like and which largely envelop a cation at their centre. For these, it is
the matching of the size of the central cavity of the ligand to the size of the
cation which determines stability. In Chapter 3 the cryptand ligands were
mentioned; these are shown again in Fig. 5.6. They are given shorthand
names which simply list the number of oxygen atoms in each bridge between
the two nitrogen atoms. Also in Fig. 5.6 are given the approximate radius
of a hypothetical spherical cavity at the centre of each ligand. These values
are to be compared with the approximate ionic radii of the alkali metal
cations, as determined experimentally by X-ray data diffraction,4 of Li +
0.76A, Na+ 1.02A, K+ 1.38A, Rb+ 1.52A and cs+ 1.67A. The relative
stabilities of the alkali metal cryptand complexes are
[211] u+ > Na+

[221] u+ < Na+ > K+ > Rb+
[222] Na+ < K+ > Rb+
[322] Rb+ < cs+
3 C. A. Tolman, Chern. Rev. (1977) 77. 313.
4 Data from R. D. Shannon, Acta Cryst. (1976) A32, 751.
[211]
o.sA [221]
Fig. 5.6 Four cryptand ligands. The three 1.1A
numbers within the square brackets correspond
to the three chains linking the two nitrogen
atoms and give, in descending numerical order,
the number of oxygen atoms in these chains.
Below each ligand designation is given the
approximate size of the cavity within that ligand.
[222]
[322]
1.4 A
1.8A
The correspondence between the best cavity-size to cation-size match and

maximum stability is perfect. Not surprisingly, this interpretation of stability
constant data for encapsulating ligands is being widely applied, not least to
some antibiotics and biological molecules which are involved in the transport
of Na + and K + in living systems. It is also being applied to ligands which
surround a molecule without totally encapsulating it-for instance, ligands
that surround a metal, holding it in a square planar environment. Again, such
species have biological analogues, to which we shall return in Chapter 16.
5.8 Conclusions
Although in this chapter many aspects of stability have been discussed, there
are many others that have not. It is now accepted that for some ligands,
metal-ligand double bonding occurs. 5 Just how are bond order and stability
related? Is the distinction between inert and labile transition metal complexes
related in any way to the d-electron configuration of the transition metal
(we shall have more to say about this in Chapter 14)? In addition to the
kinetic trans effect, discussed at the end of Chapter 4, there is a static trans
effect, sometimes called the trans influence. In this, a metal-ligand bond
5 See W. A. Nugent and J. M. Mayer, Metal-Ligand multiple honds. Wiley, New York, 1988.
Further reading I 93
length, vibrational frequency or some other characteristic of the metal-ligand

bond, is influenced by the ligand trans to it. So, in [OsNCI 5] 2 -, in which
the Os-N bond is short and usually regarded as a multiple bond, the Os--Cl
bond trans to the nitrogen is over I 0% longer than that cis. What is the
consequence of the trans influence on stability? In the case quoted, it seems
that the chlorine trans to nitrogen is kinetically the more labile. The
observations and consequent questions are almost endless; for some, incom-
plete answers exist; for others, nothing. We note that although several aspects
of the stability of coordination compounds have been discussed in this
chapter we have avoided the most fundamental question of all: why is one
particular complex more stable than another? What is the fundamental
explanation for class-a and -b behaviour (or for the hard and soft distinc-
tion)? As has been seen earlier, an answer to this last question is emerging,
but it is still incomplete. It is now generally accepted that the interaction of
the orbitals of the two atoms to be bonded is the most important factor.
This interaction depends on the matching of the orbitals-orbitals of similar
energies often also have similar sizes-but it is not a simple matter because
several orbitals or sets of orbitals of different energies and sizes on each atom
are involved. Our discussion of LUMO-HOMO effects in the context of
hard and soft ligands probably points the way ahead. Some aspects of this
will be considered in more detail in Chapter 10. However, at the present
time it does not seem possible to give more than a semiquantitative general
answer to such questions.
Further reading Stability Constants Database, NIST, Gaithersburg, MD, USA,

1993.
Stability constant data are to be found in some early, pioneer- Very useful is Determination and Use of Stability Constants,
ing, compilations: 2nd edn, VCH, New York, 1992.
Computer methods for the interpretation of experimental
• Stability Constants of metal-ion complexes, Special Publica- data are described in Computational Methods for the Deter-
tion No. 17, The Chemical Society, Burlington House, Lon- mination of Formation Constants, D. J. Leggett (ed.), Plenum,
don, 1964. New York, 1985.
• Stability Constants of metal-ion complexes, Supplement, Spec- With a content covering a much wider area than the
ial Publication No. 25, The Chemical Society, Burlington subject of this chapter, but with much relevant material,
House, London, 1971. particularly on measurement methods, is K. A. Connors, Bind-
• Stability Constants of metal-ion complexes Part A (Inorganic ing Constants, Wiley, New York, 1987.
Ligands), Pergamon Press, Oxford, 1982. Graphical representations of data related to those discussed
in this chapter will be found in J. Kragten, Atlas of Metal-
• Stability Constants of metal-ion complexes Part B (Organic
Ligand Equilibria in Aqueous Solution, E. Horwood, Chichester,
Ligands), Pergamon Press, Oxford, 1979.
1978.
All four of the above volumes are needed as data are not An easy-to-read and enlightening article is 'A comparison
repeated unless new measurements are reported. The first two of different experimental techniques for the determination of
volumes give the more detailed and helpful description of the the stabilities of polyether, crown ether and cryptand com-
use of the tables. plexes in solution', H.-J. Buschmann, lnorg. Chim. Acta (1992)
A continuing compilation is that of A. E. Martell and R. M. 195, 51.
Smith, Critical Stability Constants, Plenum, New York, Vols. Other relevant articles:
1-6 (1974, 1975, 1976, 1977, 1982 and 1989).
Related, and similarly arranged, thermodynamic data are • 'Potentiometry revisited: The Determination of Thermody-
to be found in J. J. Christensen and R. M. Izatt, Handbook of namic Equilibria in Complex Multicomponent Systems',
Metal-Ligand Heats, Marcel Dekker, New York, 1979. A. E. Martell and R. J. Motekaitis, Coord. Chern. Rev. (1990),
A database is A. E. Martell and R. M. Smith, Critical 100, 323.
94 I Stability of coordination compounds
• 'Steric Effects of Phosphorus Ligands', C. A. Tolman, Chem. • 'Hard and Soft Acids and Bases-the Evolution of a Chem-
Rev. (1977), 77. 313. ical Concept', R. G. Pearson, Coord. Chem. Rev. (1990), 100,
• 'Ligand Interactions in Crowded Molecules', H. C. Clark and 403.
M. J. Hampden-Smith, Coord. Chem. Rev. (1987) 79, 229. • 'Absolute Electronegativity and Hardness', R. G. Pearson,
• 'Ligand Steric Properties', T. L. Brown and K. J. Lee, Coord. Chem. in Britain (1991) 444.
Chem. Rev. (1993) 128. 89. • 'Molecular Organization, Portal to Supramolecular Chem-
• 'The Specification of Bonding Cavities in Macrocylic Lig- istry. Structural Analysis of the Factors Associated with
ands', K. Hendrick, P. A. Tasker and C. F. Linday, Prog. Molecular Organization in Coordination and Inclusion
Inorg. Chem. (1985) 33, I. Chemistry, including the Coordination Template Effect',
• 'The Chelate Effect Redefined', J. J. R. Frausto da Silva, D. H. Busch and N. A. Stephenson, Coord. Chem. Rev. (1990)
J. Chem. Educ. (1983) 60, 390. 100, 119.
• 'Misunderstandings over the Chelate Effect', D. Munro,
Chem. in Britain (1977) 100.
Questions 5.3 Use the data in Fig. 5.2 to show that, as claimed in the
caption, it represents the formation of the ML2 complex.
5.1 In Fig. 3.27 was shown an example of a compound with
a low barrier to internal rotation. There are many similar
compounds, some of which spontaneously ignite when exposed 5.4 The experimental data used in Table 5.5 are those for
to air. Comment on the applicability of the terms kinetic zero ionic strength in Table 5.1. Calculate the corresponding
stability and thermodynamic stability to such species. experimental data for 2 M NH4 N03 (given in the caption to
Table 5.1 ). Comment on the relative agreement of the two sets
5.2 (Note that in this question square brackets indicate con- of experimental data with experiment.
centrations.) The consequences of the addition of base to
aqueous Cr 111 was followed for four years. [Cr 2 (0H),] 4 +
maximized after a few days whilst [Cr 3 (0H) 4 ]<+ increased 5.5 Compare Tables 5.2 and 5.3 and thus comment on the
throughout the period. [Cr4 (0H) 6 ]6+ was constant after a few question of whether the class-a and -b, and hard and soft
months. Comment on these observations in the light of the classifications are simply restatement of the same experimental
reported K •• values: observations.
[Cr0 (0H)•]"+
5.6 Use Table 5.4 to associate numbers with the broad
Kab = [Cr(OH),]+[cr•• 1 (0H)•-2]<n il+
divisions given in Table 5.3 (for instance, what is the range of
logK22 = 5.1; logK34 = 6.9; logK46 = 5.2 absolute hardness values that corresponds to hard?).
Molecular orbital theory of
transition metal complexes
6.1 Introduction
The traditional approach to the electronic structure of transition metal
complexes (which is the subject of the next chapter) is to assume that the
only effect of the ligands is to produce an electrostatic field which relieves
the degeneracy of the d orbitals of the central metal ion. The most serious
defect of this model is that it does not recognize the existence of overlap,
and hence the existence of specific bonding interactions, between the ligands
and the metal orbitals. Yet calculations which assume reasonable sizes for
the orbitals (together with a considerable body of physical evidence which
will be reviewed in Chapter 12) point to the existence of overlap. How should
this be taken account of? The simplest answer is to be found in the
L•
application of symmetry ideas to the problem, and this is the subject of this
chapter. In it the readet' will be assumed to have some familiarity with the
basics of group theory. Appendix 3 gives an introduction to the subject; the
following lines are intended to provide a brief overview of aspects needed
to make a start on the present chapter.
It is convenient to focus the discussion on octahedral complexes (Fig. 6.1 ).
These are molecules with high symmetry, possessing, for example, fourfold
------·
"1M rotation axes, threefold rotation axes, twofold rotation axes, mirror planes,
and a centre of symmetry (Fig. 6.2). Symbols such as a 19 ('aye one gee '), e9
and t 19 are used to distinguish the behaviour of different orbitals or sets of
(b) orbitals under the various operations associated with the rotation axes, mirror
planes etc. of the octahedron. Note the shift from 'symmetry elements' (such
fiC. 6.1 (a) An octahedron. (b) An octahedral
complex. The perspective adopted here is the as rotation axes, mirror planes) Fig. 6.2, to 'symmetry operations'- the act
same as that for (a), and was chosen to make of rotating, reHecting etc. In group theory one is concerned with symmetry
(a) as easy as possible to visualize. The way operations, rather than with symmetry elements. Lower case symbols are
that an octahedral complex is drawn will vary,
the perspective adopted depending on the point used to denote orbitals or, more generally, wavefunctions. So, a point to
under discussion. which we shall return, an s orbital of an atom at a centre of an octahedral
96 1 Molecular orbital theory of transition metal complexes
Fig. 6.2 Some of the symmetry elements of an

octahedron. The symbols used are those
standard in chemical applications of group
theory (see below). Where there is a number in
brackets it indicates the total number of
elements of that type.
C4 : fourfold rotation axis
C3 : threefold rotation axis
C2 : twofold rotation axis
i: centre of inversion
a": (horizontal) mirror plane. Each of these
planes contains two C4 axes
a": (dihedral) mirror plane. In the octahedron
each of these planes contains two C2 and
two C3 axes
Each C4 and C3 axis has two symmetry
operations associated with them (the C4 axis
also has a coincident C2 ). Not shown in the
diagram is the 'leave alone, do nothing'
operation. This is called the identity operation
and is denoted by E. Although there is only one
E (C1 ) operation there is an infinite number of
C1 axes.
complex is referred to as an a 19 orbital. The symbol a, (and b, when it is
met), with any suffixes or primes, indicates an orbital which is singly
degenerate. Similarly e, with or without suffixes or primes, indicates a pair
of degenerate orbitals. 1 Finally, t indicates a set of three degenerate orbitals.
The subscripts g and u indicate behaviour under the operation of inversion
in the centre of symmetry. An orbital of A 19 symmetry (an a 19 orbital) is
turned into itself under inversion (g = gerade, German for even); something
of A 1 • symmetry (an a 1• orbital) is turned into the negative of itself, hence u
(ungerade, odd). Other subscripts (and/or primes where these appear) serve
only to distinguish general symmetry properties. So the symbols t 19 and t 29
represent two sets of triply degenerate orbitals, all centrosymmetric, but the
members of one set behave differently from the members of the other under
some symmetry operations. Although the same symbols may be applied to
molecules of different symmetries there need be no obvious logical connec-
tion between them. Symmetry symbols and symmetry, reasonably, go
together. A change of symmetry probably means a change of symbols. For
example, although the s and three p orbitals of a metal atom at the centre
of an octahedral complex are labelled a 19 and t '"' in a tetrahedral complex
the labels are a 1 and t 2 .
In a non-linear polyatomic system one cannot, strictly, talk of a, n and
b interactions. (Draw a a bond involving the s and p orbitals of two atoms
and then allow a third, non-colinear, atom to participate. The interactions
between an orbital of this third atom with those forming the a bond will
not be purely a unless the orbital on the third atom is a pure s orbital). For
simplicity, in the discussion that follows, ligand-ligand overlap will be
neglected (although some workers hold that this is of importance) and only
ligand-metal interactions considered. This means that problems associated
with the presence of third atoms can be neglected and a molecule is regarded
1 The fact that some orbitals are members of degenerate sets will be taken for granted in
the present chapter. In the following chapter it will be necessary to explore this degeneracy in
much more detail and so a detailed discussion is deferred until then.
Octahedral complexes I 97
as held together by a network of diatomic interactions. From the point of

view of the metal in a complex, each individual interaction may then be
classified as either a or n (when bonding is to another metal atom b bonding
may have to be included, as will be seen in Chapter 15). We start by
considering octahedral complexes in some detail and, first of all, the
metal-ligand a bonding in them.
6.2 Octahedral complexes

6.2.1 Metal-ligand cr interactions
The interactions between the six ligand a orbitals, one orbital on each ligand
and pointing towards the metal, and the valence shell orbitals of the metal
atom in an octahedral complex are the subject of this section. The case of
a complex formed by a metal ion of the first transition series will be the one
considered, so that the relevant valence shell orbitals of the metal are the
3d, the 4s, and the 4p. The ligand a orbitals are all formally occupied by
two electrons, which, in the simple picture of Chapter 1, one might regard
as being donated to empty orbitals on the transition metal ion. Such
a-valence bond-picture of the bonding is that presented by Pauling in his
classic text The Nature of the Chemical Bond where the empty orbitals
considered were d 2 sp 3 metal orbital hybrids. Although there is increasing
attention being paid to the valence bond model, we use a molecular orbital
approach because this enables us more readily to exploit the octahedral
symmetry of the molecule. Indeed, the context of the present chapter is largely
symmetry-determined. The reader may easily see this by briefly comparing
the content of the present chapter with that of Chapter 10. In Chapter 10
symmetry will be exploited as far as possible but that chapter also covers
situations in which the symmetry is too low to be of any real assistance. The
application of symmetry in the present chapter is at once a strength and a
weakness; it enables an elegant and enlightening discussion but the study of
real-life molecules often requires the content of Chapter 10.
The major impact of symmetry on the present discussion arises from the
fact that only orbitals of the same symmetry type have non-zero overlap
integrals (this is demonstrated in outline in Appendix 3). That is, if there are
two sets of orbitals of E. symmetry, (one on the ligands, the other on the
metal), they will in general be non-orthogonal (i.e.have a non-zero overlap
integral) and therefore interact. On the other hand both will have a zero
overlap integral with all orbitals of A 19 , T 1 ., T 2 • and all other symmetry
species, wherever these are located in the molecule, and will not interact with
them. Not surprisingly, this fact enormously simplifies the problem. First
then, we classify all the orbitals under consideration according to their
symmetries. It is convenient to start by classifying the valence shell atomic
orbitals of the central metal atom. It is an excellent exercise to demonstrate
that the symmetry labels that follow correctly describe the transformation
properties of the metal orbitals. This is an easy task for those with some
experience of group theory; the inexperienced should immediately read
Appendix 3.2 Key is the character table of the octahedral group 0, given in
2 A detailed derivation is given in the literature in a paper that contains much other material
relevant to the present chapter; see S.F.A. Kettle, J. Chern. Educ. (1966) 43, 21.
Fig. 6.3 The metal s orbital.
X~
abbreviated form in Table 6.1 and also in Appendix 3 (Table A3.3). Because
the 4s orbital (Fig. 6.3) of the central metal is turned into itself by all the
operations of the group--multiplied by 1-its behaviour is described by the
A 1 irreducible representation of Table 6.1. In the full octahedral group
Oh (which, unlike the group 0, contains the operation of inversion in a centre
of symmetry), the label describing the behaviour of the 4s orbital is A 1 • (note
that an equivalent way of talking about an orbital labelled a 1 • is to say that
it 'is of A 1 • symmetry' or that 'it transforms as A 1 .'). In similar fashion, the
set of three 4p orbitals (Fig. 6.4) have T 1" symmetry (T 1 in Table 6.1). Finally,
the five 3d orbitals split into two sets, of E. (Fig. 6.5) and T 29 (Fig. 6.6)
symmetries (E and T 2 in Table 6.1). This latter splitting, which is denoted
11, is of crucial importance to the understanding of the properties of transition
metal complexes. One considerable attraction of crystal field theory, the
subject of the next chapter, is that it gives a very simple physical explanation
of this splitting. In the present context it is perhaps most easily regarded as
a splitting which is group-theoretically allowed and so, presumably, may
well exist. As the chapter develops we shall find good reasons for the
splitting.
Evidently, the next step is to classify the ligand orbitals according to their
symmetry types. Two problems arise. First, for a regular octahedral complex,
all six ligand orbitals look alike-how then can they be classified differently?
Second, a characteristic of a symmetry-classified set is that all of those
symmetry operations which send an octahedron into itself send one member
Table 6.1 The 0 character table of the set into itself, another member or a mixture of members of the set.
The reader who is not familiar with this pattern is asked to take it on trust
0 E 6C4 3C2 6C2 8C3 for the moment. It will be demonstrated in the next chapter in showing the
degeneracy of the metal dx'- Y' and dz, orbitals in an octahedral crystal field
A, 1 1 1 1 1 (Section 7.3); alternatively, Appendix 4 is dedicated to the problem. As these
symmetry operations send one ligand a orbital into another, why do these
A2 1 -1 1 -1 1 orbitals not already constitute a set? The answer is, they do. However, this
set can be broken down into smaller sets and it is to these latter that
E 2 0 2 0 -1
symmetry labels are attached. 3 In these sets the individuality of the ligand
Tl 3 1 -1 -1 0 a orbitals is lost. We talk of the wavefunctions of various sets of ligand
orbitals rather than of the wavefunctions corresponding to individual ligand
T2 3 -1 -1 1 0
3 Appendix 6 shows how these subsets may be obtained.
Octahedral complexes 1 99
z z z
t t t
,/~-, .-/L~j~\-, ,/¢.,

Px Py Pz
Fig. 6.4 The metal p,, p, and p, orbitals.
z
z z
I
dxy dyz dzx
Fig. 6.5 The metal d,, d"' and d~ orbitals.
z z
dx2- Y' d,2

Fig. 6.6 The metal dx2 -y2 and dz2 orbitals. orbitals. The members of the symmetry-classifiable sets are linear combina-
tions of the ligand rr orbitals and are often referred to either as 'ligand group
orbitals' or 'symmetry-adapted combinations'. The step of moving from the
individual ligand rr orbitals to their symmetry-adapted combinations in-
volves the use of group theoretical procedures. The derivation is not difficult
but it is lengthy and so would be out of place in the present section. It is
given in Appendix 6, where several different ways of tackling the problem
will be found. The explicit form of these ligand group orbitals is given in
Table 6.2. The individual rr orbitals are identified by the labels rr 1 --> rr6 as
Table 6.2 Ligand u group orbitals of six octahedrally-orientated

ligands
Symmetry Ligand group orbitals
J61 (u1 + <12 + <Ja + "'• + "'s +Us)
Fig. 6. 7 The numbering system adopted for

the ligand u orbitals in an octahedral complex
(cf. Rg. 3.4).
shown in Fig. 6.7, the labels being chosen in conformity with the convention
given in Fig. 3.4. As mentioned earlier, in the explicit forms given in Table
6.2 it has been assumed that the ligand a orbitals do not overlap each other,
although non-zero ligand-ligand overlap integrals would only affect the
normalization factors (the numerical coefficients at the front of each
expression) given in the table, not the general form of the combinations.
The ligand a group orbitals listed in Table 6.2 are of A 19 , E 9 and T 1u
symmetries and each set will overlap with metal orbitals of the same
symmetries, so that the 4s(a 19 ), 4p(t 1.) and 3d(e9 ) orbitals of the metal will
be involved in a bonding. Of the valence shell orbitals of the metal atom
only the 3d(t 29 ) orbitals are not involved in this bonding. The interactions
of A,., T 1• and E. symmetries are shown pictorially in Figs. 6.8-6.10, where
it can be seen that there is the expected close matching between ligand group
orbitals and the corresponding metal orbitals. If symmetry had not been
applied to the u bonding problem then we would have had to consider the
interaction between the six ligand u orbitals and nine metal orbitals, with
no means of knowing in advance that three of the 3d are not involved in
this u bonding. In contrast, not only does the symmetry break the problem
up into subproblems-the A 19 , E9 and T,.- but for each degenerate case,
E9 and T1., only one member of each set need be considered (by symmetry,
consideration of any other member must lead to the same result). That is,
for each of the three subproblems only the interaction between a single metal
orbital and a (delocalized) ligand orbital has to be treated. For simple models
the calculations really can be done on the back of an envelope! Fig. 6.11
shows, schematically, the energy level pattern obtained as a result of these
u interactions. It includes metal and ligand electrons; the detailed problem
of how to include those originating on the metal will be a major concern of
Chapters 7 and 8. Note, particularly, the splitting d in Fig. 6.11. Many
FJC. 6.8 The metal-ligand a bonding in an
octahedral complex involving orbitals of A"'
symmetry. The metal s orbital is shown at the
centre.
FJC. 6.9 The metal-ligand a bonding in an

octahedral complex involving orbitals ofT:w
symmetry. In this figure the interaction involving ..,...-- a:::-----::,..:---....------e
the metal p, orbital is shown; there are similar x
interactions involving Py and Pz·
z z
X
......-- --........ y --........ y
(a) eg (d.,_ y.z) (b)
Fill. 6.10 The interaction (a) between metal

d,.,_, and (b) dz' orbitals and the
corresponding ligand group orbitals.
Blg(2)
4p====:
4s------(
eg(2)
eg(1)
tlu (1)
a1g(l)
Metal orbitals Molecular orbitals Ligand orbitals
Fig. 6.11 A schematic molecular orbital energy research groups have tried to carry out reasonably accurate calculations
level scheme for an octahedral complex of a aimed, in part, at obtaining molecular orbital energy level schemes (such as
first row transition metal ion (only a interactions
are included). The ligand orbitals are shown that in Fig. 6.11) accurately, although this is a difficult task. It in this
occupied; the metal d orbitals are partially filled. endeavour that many of the methods to be discussed in Section 10.3.2 were
The example shown is appropriate to a d7 ion in developed.
a strong-field (low-spin) complex. The full
meaning of these latter terms will only become Figure 6.11 indicates that the idea that the ligands function as electron
evident in Chapter 7. The numbers in donors is correct. Electrons which, before the interactions were 'switched
parenthesis follow the established convention of on', were in pure ligand orbitals are really in delocalized molecular orbitals
counting from the bottom up. So, the bottom
a'<l is a,..(1), the next a,..(2). which take them onto the metal atom-the electron density on the metal
atom is increased as a result of the covalency. The consequences of this for
the concept of formal valence states will be discussed later. The most
important feature of Fig. 6.11 is the fact that the two lowest unoccupied
orbital sets are, in order, t 29 and e.(2) (this latter being weakly u antibonding).
So far, no metal electrons have formally been included and so these
unoccupied orbitals have to accommodate them. Of course, the number of
electrons occupying these orbitals is the same as the number present in the
valence shell orbitals of the metal before the metal-ligand interaction was
switched on; the d electrons of the uncomplexed metal ion may be regarded
as being distributed between the t 2 • and e.(2) molecular orbitals. This is a
situation which will be explored in the next chapter where care will be taken
to state that .1., the label given to the energy separation between a lower t 2 •
and an upper e. set, is an experimental quantity. What the next chapter will
not show is the way that the e9 set is antibonding. The present discussion
also leads us to recognize that metal electrons in the eg(2) orbitals are to
some extent delocalized over the ligands. In Chapter 12 experimental
evidence will be adduced in support of this conclusion.
6.2.2 Metal-ligand 1t interactions

So far we have only considered rJ bonding in an octahedral complex. What
of n bonding? It appears to be generally true in chemistry that n is rather
weaker than rJ bonding. Thus, there is no compound known in which it has
been established that, in the ground state, there is a n bond but no rJ bond.
On the other hand, the vast majority of molecules have at least one rJ bond
with, apparently, no associated n bond. Consequently, for our purposes, it
can reasonably be assumed that the effects of n bonding will be to modify,
but probably not drastically alter, Fig. 6.11.
Our basic approach to the treatment of n bonding is similar to that of rJ
bonding. It is convenient to assume that the ligands involved are simple,
something like a halide or cyanide anion (cases which will be explicitly
explored later). It follows that each ligand has two orbitals available for n
bonding on each ligand-ligand Px orbitals or (ligand) molecular orbitals
with the same symmetry characteristics as the corresponding Px orbitals.
That is, they have their maximum amplitude perpendicular to the metal-
ligand axis and this axis lies in a nodal plane of the ligand n orbitals (Fig.
6.12).
As there are two such orbitals on each ligand, there is a total of 12 in an
octahedral ML 6 complex. Again, group theory can be applied to obtain the
Fig. 6.12 Ligand n orbitals in an octahedral 1t3y

complex. The arrow heads represent the lobe of
each orbital which has positive phase. Although,
for generality, these orbitals are referred to as
rr. those actually drawn are pure p orbitals. All p
orbitals are labelled according to octahedral
axes using the outlining notation given at the
foot of the diagram.
'•,,,.•
:e:····.;:: Px
Fig. 8.13 (a) One of the three ligand " group

orbitals of r,. symmetry. Notice that of the 12
" ligand orbitals only four are involved in this
particular combination. The remaining eight
similarly divide into two sets of four, with no (a)
member common to any two sets. A similar
. . --.~ r (_-. ~-
pattern holds for all the other " combinations
shown in the following figures. (b) One of the
three ligand "group orbitals of T2 u symmetry. As
for all other" combinations in adjacent figures,
-(~F·-----i\:)
the combination shown is the first listed in the
·~
appropriate section of Table 6.3.
• l •II
\
\I
I
II
\_J
\I
(b)
symmetry-adapted combinations. The symmetry-adapted combinations con-

sist of four sets with three ligand group orbitals in each set. The sets have
T 1., T 1 u, Tz. and Tzu symmetries; explicit expressions for them are given in
Table 6.3 using the ligand n orbitals labelled as shown in Fig. 6.12. The
Table 6.3 " ligand group orbitals for derivation of the explicit forms of these functions is somewhat lengthy and
the six ligands in octahedral complexes is not given in this book; the procedure follows that given in Appendix 6; a
(a definition of the labels is given in detailed but simple treatment is given elsewhere.4 There are no metal orbitals
Fig. 6.12)
of T 1• and Tzu symmetries within our chosen valence set (although, just for
Symmetry Ugand group orbital the record, one set of metal f orbitals transforms as Tzu and one set of metal
1
g orbitals as T 1.). The ligand n group orbitals ofT1 • and Tzu symmetries are
~(x2z + 1C3z + x 42 + n5z) therefore carried over, unmodified, into the full molecular orbital description.
t1u ~(ltlx + """ + """ + lt4x) A representative example of each is shown in Fig. 6.13. We are left with Tz.
~(xly + n3y + "5y + ns,> and T~u sets. The T'" set will interact with the metal p orbitals (also of T 1u
symmetry), but it is simpler to think of their effect on the occupied T~u u
tog 1 ;(n>x - "ay - """

~(ltlx- "2z- "6y
+ "5Y)
+ 1t4z)
~(n 1y- lt3z- """ + n5z)
molecular orbitals in Fig. 6.11 and to consider two cases.
Case 1: The ligand 1t orbitals are occupied
1 ~(n2x + nay- 1t4x - 1tsy) In this case, illustrated in Figs. 6.14 and 6.15(c), it follows that the T 1u
symmetry ligand n orbitals are also occupied. Interaction with the t '"(1)
t2g ;(nly + lt3z- " " ' - n.,.)
molecular orbitals of Fig. 6.11 will raise or lower this latter set depending
~("'"' + 1C2z- " " ' - 1C5z) on whether its energy is higher or lower than that of the ligand n(T~u) set.
1 ~(lt2z- "3z+ lt4z- lt5z) As long as we retain two occupied t'" sets the orbital occupancy is unaffected
t2u + """- n4x)
'(ltlx- lt2x 4 Appendix 4 in Symmetry and Structure, 2nd edn., by S. F. A. Kettle, Wiley (Chichester and
'("1y- "ay + "5y- "sy) New York) 1995. Note that the atom labelling pattern used in this text differs from that in the
present.
by this interaction and, because the t 29 - eg(2) separation is our concern, for
this case we may forget the t lu orbitals and their interactions.
Case 2: The ligand n orbitals are unoccupied
,-, ,-, In this case it follows that the ligand n combinations of T 1• symmetry are
also unoccupied. Interaction between the ligand nand (occupied) t~u(l) sets
I+ +I
of Fig. 6.11 will result in a repulsion between the two and the t~u(1) set will
' '
··~----+-+-------·
'1- '
,_,I '
,_,
I -I \
be stabilized somewhat but the orbital occupancy will remain unchanged.
This is illustrated in Figure 6.15(b). We could, as a separate case, consider
the interaction between the t 1.(2) molecular orbital set of Fig. 6.11 and the
ligand n(t~u) set. Again, this would only be necessary if the interaction leads
Fig. 8.14 A ligand " group orbital of Tw to one of the orbital sets becoming occupied. In fact, there is no recognized
symmetry and the molecular orbital of Fig. 6.9
with which it interacts (that in110lving the metal case in which the ligand n(t~u) set, whether occupied or not, needs to be
Pz orbital). considered.
We are left with the ligand n set of T 29 symmetry. This set may interact
with the metal t 29 orbitals (dxy• d,., d,x) which, so far, have been non-bonding
because there is no ligand a combination ofT29 symmetry. The consequences
eg(2)
~
t1u
eg(1)
t1u (1)
t1u
81g(1)
(a)
Occupied
ligand 1t t1u
Unoccupied
ligand 1t t1u (c)
(b)
Fig. 8.15 The interaction between the lowest

t1u orbitals of Fig. 6.11(a) with (b) an
unoccupied ligand x(!w) set and (c) an
occupied ligand x(!wl set. Largely ignored in
the text is the !1u(2) set of Fig. 6.11;
corresponding arguments can be developed,
with similar conclusions, for this set.
t2g
Ligand
eg Metal
Metal t2g
cr bonding only cr bonding only

cr + 1t bonding
Ligand (c)
cr + 1t bonding
(b)
~- 8.18 (a) The interaction between a metal
t, orbital (d.,) and the corresponding ligand n of interaction between the ligand n(t 2,) orbitals and the metal t 2 • orbitals
combinations. Overlapping lobes are joined by shown in Fig. 6.11 depend upon which of the two t 2, sets is higher in energy.
lines. (b) The consequences when the ligand
n(t211 ) are low lying (and occupied); tJ. is The alternative situations are shown in Fig. 6.16. If the ligand 1t set is the
decreased. (c) The consequences when the higher in energy then the metal t 2 , set is pushed down; if the ligand n set is
ligand n(t,..> are high lying (and empty); tJ. is the lower, the metal set is raised in energy. As Fig. 6.16 shows, these
increased.
movements have a direct effect on the t 2 , - e,(2) splitting. That is, the
magnitude of~ depends, in part, on 1t bonding. It is believed that the halide
anions provide examples of the situation shown in Fig. 6.16(b). For these
ligands, two of the filled p orbitals in their valence shell will be the orbitals
involved in 1t bonding and interaction with them will raise the energy of the
metal t 2 • orbitals making the latter weakly antibonding. This explains,
partially at least, why complexes containing halide anions as ligands have
relatively small values of~ (see Table 7.1 ). In contrast, the cyanide anion is
associated with large values of~- The cyanide anion possesses two sets of 1t
orbitals both of which will interact with the metal orbitals. The effect of the
occupied C=N- 1t bonding orbitals will be similar to that of the occupied 1t
orbitals of the halides. However, these 1t orbitals have energies well removed
from those of the metal d orbitals and so their effect is small. Much more
important are the empty C=N- n antibonding orbitals, which behave as
shown in Fig. 6.16(c). Interaction with the metal t 2 , orbitals stabilizes the
latter (which become weakly n bonding) and so increases the metal
Tetrahedral complexes 1 107
t 2" - e.(2) separation of Fig. 6.11. This behaviour is clearly consistent with
z the observation that the cyanide anion gives rise to large values of !J.; it exerts
a very large ligand field. In Chapter 10 we will see that the carbon monoxide
ligand behaves similarly.
Finally, a cautionary note. There is theoretical evidence, despite all that
has been said in this chapter, that in a complex such as [Cr(CN) 6 ] 3 - the d
electrons are not in the highest occupied orbitals. Rather, that all the CN-
n occupied and, indeed, the CN- (J bonding orbitals, all occupied, have
higher energiess The reason for this apparently ridiculous behaviour is to be
found in a phenomenon which so often wrecks our simple pictures-electron
repulsion within the molecule. How can there be filled orbitals above
partially occupied ones? The increase in electron repulsion energy in taking
an electron from a (delocalized and therefore diffuse) n occupied or (J bonding
orbital and putting it into a (largely localized and a therefore concentrated)
metal d orbital •costs' more than the energy gained from moving the electron
into what is a more stable orbital. The holes in the d orbitals are protected!
z Fortunately, this complication turns out to be unimportant for the discussion
in this chapter. It doesn't really matter too much where the d orbitals are
relative to the ligand orbitals. As has been commented 'if one is only
interested in the energy pattern, ligand field theory remains a reliable guide'.
Since virtually all of the measurements made on transition metal complexes
are concerned with the detailed energy pattern, all is well. The reader who
is either unhappy with this situation or is so interested in it that they wish
to learn more, should turn to Section 10.7 where it is encountered again, in
the context of ferrocene, and discussed in more detail. In Section 12.7
measurements that go beyond energy patterns will be described and com-
pared with theory, whereupon the relative energies of the ligand orbitals will
become of importance.
FIC. 8.17 Metal d orbitals in a tetrahedral

complex. 6.3 Tetrahedral complexes
In this and the next section the bonding in complexes with other-than-
octahedral stereochemistries are considered. That the d orbitals split into t 2
and e sets in tetrahedral complexes may be seen from Fig. 6.17. The key is
to recognize that a tetrahedron is closely related to a cube. If, starting from
one vertex, lines are drawn across the face diagonals of a cube to connect
with the opposite vertices and the process continued, then a tetrahedron
results (Fig. 6.18). If Cartesian axes are drawn as in Fig. 6.17 then
x ""y"" z so that dxy• dyx and dzx must be degenerate. The degeneracy of
dx'- y' and dz, follows just as it did for octahedral complexes (for a detailed
justification, see Section 7.3 and/or Appendix 4). The other metal orbitals
in the valence shell are 4s(a 1 ) and 4p" 4pY, 4pz(t 2 ), again considering a
first-row transition element. It is a simple exercise in group theory to show
that the four ligand (J orbitals give A, and T 2 symmetry-adapted combina-
tions. The ligand group orbitals may be obtained by the methods of
Appendix 6 (for the third method described there, also use the operations
5 There are many papers in the literature which indicate this. See, for example: Vanquicken-
FJg. 8.18 The relationship between a cube and borne et a!., lnorg. Chern. (1984) 23, 1677; Excited States and Reactive Intermediates, ACS
a tetrahedron. Symposium Series 307 (1986) 2; and Inorg. Chern. (1991) 30,2978.
z
Table 6.4 Ligand group orbitals of (J symmetry in a tetrahedral
complex; the labels used are those of Fig. 6.19
Symmetry Ligand group orbital
~(<J1 + <J2 + G3 + a4)
E, 8C 3 and 3C2 ). Using the orbital labels shown in Fig. 6.19, the A 1 and T 2
ligand group orbitals are given in Table 6.4 and pictured in Fig. 6.20, a figure
04 which also shows the metal orbitals of the same symmetries.
As before, only orbitals of the same symmetry species may interact with
Ftg. 6.19 The labels used in the text for the
ligand u orbitals of a tetrahedral complex. each other. The A 1 interactions are straightforward but the T 2 more
complicated than the corresponding octahedral case because both metal sets
4pxo 4pY, 4pz and 3dxy• 3dy, 3dzx have T 2 symmetry and so interact with
the ligand orbitals of this symmetry. It does not appear possible to make a
general statement about the details of the outcome of the T 2 interactions but
that given in Fig. 6.21 is about as close as one can get. Figure 6.21 gives a
schematic a bonding-only molecular orbital diagram for a tetrahedral
complex. As we shall see in the next chapter, just as in octahedral complexes,
the more stable d orbital set according to crystal-field theory, that of e
symmetry, is not involved in a bonding but the less stable, the tz, is involved.
Many tetrahedral complexes involve the oxide anion as a ligand (e.g. the
Mo~- anions commonly known as permanganate, chromate, and ferrate)
and contain a metal atom in a high formal valence state. Since oxygen
is usually regarded as forming two covalent bonds and because a high metal
charge will favour ligand-to-metal charge migration, it can be anticipated
that n bonding may well be of potential importance for such tetrahedral
complexes. Unfortunately, the consequences of n bonding are not as clear-cut
as for octahedral complexes. An important difference between octahedral
(a)
and tetrahedral complexes is that the latter do not have a centre of symmetry.
Such a centre of symmetry separates the p orbitals from the d orbitals on
the metal (the ps are ungerade and the ds are gerade) and no centrosymmetric
ligand field can mix them. A tetrahedral ligand field can mix ps and ds on
the metal and there is good evidence that such mixing occurs~it will be met
in Chapter 8.
The n bonding problem in tetrahedral molecules starts difficult and
remains so throughout. It is not a trivial task to demonstrate that the ligand
n orbital symmetry-adapted combinations are of T 1 + T2 + E symmetries.
To show this it is important to chose the orientation of the ligand n orbitals
carefully if the task is to be made (relatively) easy. Guidance is given in Fig.
(b) 6.22. Appendix 4, and the character of -1 described at the end, may well
Ftg. 6.20 (a) Interaction of the ligand a1 group be needed also. The next step, that of the generation of the ligand group
orbital of Table 6.3 with the metal s orbital. orbitals, is not trivial. 6 The explicit expressions for the ligand group orbitals
(b) Interaction of the first ligand t 2 u group
orbital listed in Table 6.4 with the 6 A fairly simple derivation is given in detail in S. F. A. Kettle, Symmetry and Structure, 2nd
corresponding metal p orbital. edn., Wiley, Chichester, 1995, Appendix 4.

s ,;;;Bl;,__ _ _..(
~+e ~
Fig. 8.21 A schematic molecular orbital energy d
level diagram for tetrahedral complexes with
~~~~~~K~~-----c~=:::;:J=~~)
only u interactions included. The ligand u
electrons are stabilized by interaction with the
corresponding metal orbitals. The (three)
electrons originally in the metal d orbitals
correspond with those distributed between the
e(l) and t 2 (2) molecular orbitals.
81 (1)
Metal obitals Ligand cr obitals
1t2v
Fig. 8.22 A symbolic representation of the

ligand n orbitals in a tetrahedral complex. On
each ligand there are two n orbitals and these
may be regarded as lying in a plane
perpendicular to the local ligand-metal axis (at
each ligand this plane is represented by a circle
which. however, may appear as an ellipse
because of the perspective). At each ligand an
orbital is labelled nv if it lies in a plane
containing the z axis and nh if it is
perpendicular to this plane.
1t3v
are given in Table 6.5 using the notation of Fig. 6.22 and illustrated in Figs.
6.23-6.25, the relevant metal orbitals being included in the e and t 2 cases.
For tetrahedral complexes 1t bonding involves all of the metal d orbitals, not
just one set (as in the octahedral case), and it is not possible to give a simple
diagram analogous to Fig. 6.16. Both the t 2 (2) and e levels of Fig. 6.21 will
Fig. 6.23 Ugand n group orbitals of E

symmetry: (a) that which overlaps with the
metal dz2; (b) that which overlaps with the
metal dx2~y2· The pattern of overlaps in this
diagram is worthy of careful study.
Table 6.5 n ligand group orbitals of a tetrahedral complex.

Figure 6.22 gives a definition of the orbitals
change with n bonding and it is not possible to discuss the behaviour of

their separation, ~"'' in general terms although it is quite clear that ~'"
will be sensitive to n bonding. However, both e and t 2 n interactions are
likely to be either both bonding or both antibonding so that the difference
between the e and t 2 (2) orbital energy changes may be small and this may
mean that n bonding has a relatively small effect on ~tot·
A schematic molecular orbital pattern for a tetrahedral complex with a
significant n bonding contribution is shown in Fig. 6.26. It is emphasized
Fig. 6.24 A ligand n group orbital of T,
that the relative energies shown for the molecular orbitals in this figure are
symmetry. This orbital has the same symmetry to be regarded as highly flexible.
properties as a rotation about the z axis. It is
the fact that the ligand n orbitals have been
chosen to be oriented with respect to the z 6.4 Complexes of other geometries
axes (see caption to Fig. 6.22) that makes this
orbital, mathematically and pictorially, Apart from one example, other geometries will not be discussed in detail
particularly simple. Its two partners in Table
6.5 have the same symmetry properties as
because there are none for which we could arrive at firm general conclusions.
rotations about the x andy axes, respectively. Instead, in Table 6.6 are listed the symmetries of the <J and n ligand group
Complexes of other geometries 1 111
orbitals for common geometries and compared with the symmetries of the
orbitals comprising the valence set of the transition metal ion. This table is
group theoretical in origin and is designed to enable the reader to develop
a qualitative bonding argument for almost any complex with up to eight
ligands with any significant symmetry (if a molecule is only slightly distorted
from a high symmetry it often pays to pretend that there is no distortion at
all, as a first approximation). The reader who is unhappy about the plethora
oflabels in Table 6.6 need not be too concerned-these labels can be regarded
as simply indicating what interactions can occur-labels have to be identical
for an interaction to be possible; there is no vital need to enquire into their
deeper (group theoretical) significance.
As an example of the use of Table 6.6 its application to a trigonal
bipyramidal complex of D3h symmetry (Fig. 6.27) will be outlined. First, it
has to be recognized that our discussion of octahedral (Oh) and tetrahedral
(Td) complexes will be oflittle direct use in discussing ML 5 (D3 h) complexes.
Fig. 6.25 A ligand " group orbital of T2 Had we been concerned with complexes with either six or four ligands,
symmetry together with the metal p, with which
it ove~aps. The dominant orbital overlaps are however, it might well have been a good idea to start from the appropriate
indicated. The form of this ligand group orbital high symmetry case, as mentioned above. As Table 6.6 shows, in a ML 5
is particularly simple for the reason given in the complex the metal d orbitals split up into three sets, d., has A; symmetry,
caption to Fig. 6.24. Note carefully the phases
of the ove~ap--they are not incorrect! dx'-y' and dxy, together, are of E' symmetry whilst dyz and d.x, together,
have E" symmetry. The somewhat surprising fact that dx'-y' pairs with dxy
FJC. 6.26 A schematic molecular orbital energy

level diagram for a tetrahedral transition metal
complex in which both G and n: interactions are
4p =======/
;;;;;
important.
4s - - - - - . . . - -
t<1tet
Ligand 1t
(e+ t1 + t2)
Fig. 6.27 A trigonal bipyramidal ML,; complex.

····························································································································································································
Table 6.6 The symmetries of ligand u, " and metal orbitals for common geometries
Molecular geometry Symmetry Symmetries of metal orbitals

symbol
s Pz Px Py d.. d,.._,.. d'G'
\.__
'---y---1
Octahedral, Mls 0, al&' tlu eg \_____
'--y-J
Tetrahedral, ML4 1d a, r, e
'---y--/
trans-Octahedral, ML4 M2 o.. al&' a,u eu al&' blg bos
'---y---J
Square planar, ML. o.. al&' a,u eu al&' blg bos
Octahedral with a bidentate '---y--/ '---y--/
chelating ligand, M(L,)3 o_, a, a, e a, e
Trigonal bipyramidal, ML,; o_, a;_ a; ~ a;_ ~

'---y--/
All cis-octahedral, MLaL3 Cav a, a, e a, a, a,
'---y---J
Tetrahedral ML,L' Cav a, a, e a, a, a,
One face centred octahedral, '---y---J

ML,L3L" Cav a, a, e a, a, a,
Square based pyramidal, '---y--/

ML4 L' c.v a, a, e a, b, b,
Octahedral, ML5 L' or '---y--/
trans-ML.L'L" c.v a, a, e a, b, b,
cis-Octahedral c2v a, a, b, b, a, a, a,
Tetrahedral ML,L:, C2v a, a, b, b, a, a, a,
Square face-centred trigonal

prism, ML.L:,L'' c,v a, a, b, b, a, a, a,
'---y--/
Dodecahedral, MLa o,. a, b, e, a, b, b,
'---y--/ '---y--/
Square antiprism, MLa o.. a, b, e, a, e,
and not d.2 is easily explained-it is a general rule that the axis of highest
symmetry is (almost7) always chosen as the z axis, so here we choose the
threefold rotation axis; for an octahedron the z axis was a C4 and for a
tetrahedron an S4 .
The next step is to include, qualitatively, u bonding in the picture. The
five ligand group u orbitals are of2A~ + A2 + E' symmetries. They can easily
be obtained by the methods of Appendix 6 (treat non-equivalent ligands
separately) and are pictured in Fig. 6.28. From this figure it is evident that
7 When working with an icosahedral molecule such as [B 12 H 12 ] 2 - or C 60 life is much easier if
a C2 (rather than a C,) axis is chosen as z because x andy can then also be orientated along
C2 s and the three coordinate axes are symmetry-related. With z chosen to lie along a C, axis
the three coordinate axes are not symmetry-related.
Symmetries of the ligand orbitals Comments
dyz dzx t1 1t
t2g a.1g+eg+t1u t1.g + tlu + t2g + t2u
t2 al + t2 e+ t 1 + t2
~
eg 2alg + a2u +big+ eu a2g + b2g + a2u + b2u + 2eg + 2eu
~
eg alg + b2g + eu a2g + b2g + a2u + b2u + eg + eu
~
e a 1 + b2 + 2e 2a 1 + 2a 2 + 4e
~
e'" 2a1 +a;+ e' a:, +a; + 2e' + 2e"
~ the threefold axis is the z
e 2a 1 + 2e 2a 1 + 2a 2 + 4e
axis
~
e 2a 1 + e a1 + a2 + 3e
~
e 3a 1 + 2e 2a 1 + 2a 2 + 5e
~
e 2a 1 + b 1 + e a 1 + b 1 + a 2 + b2 + 3e
~
e 3a 1 + b 1 + e al + bl + a2 + b2 + 4e
b2 bl 3a 1 + b 1 + a 2 + b2 2a 1 + 4b 1 + 2a 2 + 4b 2
b2 bl 2a 1 + b1 + b2 2a 1 + 2b 1 + 2a 2 + 2b 2 b 1 and b 2 may be interchanged by some
authors
b2 bl 3a 1 + 2b 1 + a2 + b2 3a 1 + 4b 1 + 3a 2 + 4b 2
~
e 2a 1 + 2b 2 + 2e 2a 1 + 2b 1 + 2a 2 + 2b 2 + 4e
~
e3 a 1 + b 2 + e 1 + e 2 + e3 a 1 + b 1 + a 2 + b2 + 2e 1 + 2e2 + 2e3
both of the a~ ligand orbitals may interact with the same metal a~ orbital.
The relative importance of the two interactions will depend on the geometry
of the complex-which of the (two) axial and the (three) equatorial ligands
are the closer to the metal? Usually, the equatorial. The problem is made
more complicated, of course, by the presence of two metal orbitals which
are A'1 , because, in addition to d,,, the metal s orbital has this symmetry.
Although by no means always justified by detailed calculations, it is often
convenient to take sum and difference of the d,, and s orbitals. This leads
to one metal A'1 s-d mixed orbital largely, if not exclusively, interacting with
the axial ligands and one with the equatorial, thereby simplifying the
problem. The metal p, orbital has A2 symmetry and so interacts uniquely
with the ligand group orbital of this symmetry.
The two ligand cr orbitals of A1 ' symmetry:
may each interact with either or both of the metal orbitals of A1 ' symmetry:
sand d}
Similarly, the ligand A2 " cr orbital interacts with the metal Pz orbital
and the ligand e' orbitals interact with metal Px and Py orbitals:
with
Fig. 6.28 ~ bonding in a O,h complex.

Formal oxidation states 1115
and
Fig. 6.28 continued.

with
Finally, there are both d orbital and p orbital sets of E' symmetry (the d
we have discussed above, the p are Px and p,). Again, the general outcome
is not clear. Nonetheless, we can make an educated guess. Although there
is no required relationship, we can actually relate the trigonal bipyramidal
problem to that of the octahedron, discussed earlier. If, for each, we separate
out two axial ligands, then in the trigonal bipyramidal they are partnered
by three equivalent coplanar ligands, in the octahedron by four. Perhaps the
outcome in the two cases is not too dissimilar. If this is so, it would be
concluded that the a'1 (dz,) and e' (dx'-y' and dx,) orbitals are actually <J
antibonding. The metal d orbitals of E" symmetry, d,z and d"" are
non-bonding. When n bonding is included, the ligand orbitals a2, a~, 2e' and
2e" are added to the problem (Table 6.6). They will not be discussed in detail;
we merely note that the a~ orbitals are only involved in <J bonding, the e"
only in n bonding and the e' in both <J and n bonding. The appearance of
two orbitals of a symmetry species (2a;, 2e;, 2e") in the above discussion
need cause no concern. They arise because the axial and equatorial ligands
are not symmetry related and so contribute additively. The way they were
handled for the 2a; case illustrates the general approach-include them all
together, the actual number of them is not important.
6.5 Formal oxidation states

In crystal field theory a complex ion is assumed to be composed of a cation,
M"+, surrounded by, but not overlapping with, a number of ligands. What
is the effect of covalency on the electronic nature of the cation? Let us
consider two octahedral complex ions, one of Fe3+ (d 5 ), the other of Fe2+
(d 6 ), both assumed to have as many unpaired electrons as is reasonably
possible, and with identical ligands. Figure 6.29 gives schematic molecular
orbital energy level diagrams for the two complexes. The electrons in the
a,.(l), t 1 .(1), and lower eg(l) orbitals originated on the ligands but in the
complex ions, occupy molecular orbitals and are therefore delocalized onto
the cation to some extent.
Because of its higher charge and smaller size, the polarizing power of the
Fe3+ cation would be expected to be greater than that of Fe2+ so that the
transfer of electron density from the ligands to the cation would be expected
to be greater for Fe3+ than for Fe 2 + Let us suppose that an effective transfer
==~=±=== eg (2) eg(2)
==~±::::~= t2g (1) t2g(1)

H
FJg. 6.29 Schematic molecular orbital energy
level diagrams for high spin Fe 3 + (d 5 ) and Fe2 +
(d 6 ) octahedral complexes.
~++t==±+:t:+ = eg(1) H H eg(1)
HH t1u (1)
---++-- a1g (1) ---++----- a1g (1)
Fe3+ complex (high spin) Fe 2+ complex (high spin)
of two electrons occurs for the former and one for the latter (we choose
integers for simplicity and ignore the rather difficult problem of how to
calculate the number of electrons transferred). The resultant charges on the
two cations are therefore both + 1 (( +3- 2) and ( +2- 1)) although
we started with one Fe3+ and one Fe 2 + complex. The charges we assumed,
3+ and 2+, are free-ion charges. One expects that the magnitude of an
actual charge will always be less than the absolute magnitude of a free-ion
charge, for both ligands and cations. This, of course, is a restatement of
Pauling's electroneutrality principle. Contemporary calculations indicate
that whilst the actual charge of a Fe 3 + ion is likely to be rather greater than
that on a FeZ+ ion if the ligands are identical, the difference between them
is only of the order of one-third of an electron. The actual charges themselves
would be of the order of unity (positive). It is not surprising, therefore, that
the use of free-ion charges and valence states, will sometimes prove difficult.
In a transition metal complex containing H as a ligand, should this
be regarded asH+ or H-? The charge on the metal depends on which we
choose. In practice, the problem could either be sidestepped or worked the
other way round-a formal charge first assigned to the metal from which
the, equally formal, charge to be allocated to the H ligand would be deduced.
Alternatively, the problem could be resolved by appeal to chemistry-does
the H ligand behave more like H+ or H- in its reactions? In transition metal
chemistry the answer is quite often H- because transition metal hydrides
are more characteristically reducing agents than acids. Although the charges
indicated by the symbols FeZ+ and Fe3+ are misleading, this representation
is far from valueless. In particular, these charges lead to a correct count of
the number of electrons in the t 29 and upper e9 orbitals such as in Fig. 6.11.
It is essential to get this number right if we are to be able to correctly interpret
the physical and chemical properties of a complex. The usual compromise
adopted is to refer to iron(III) and iron(II), as has usually been done
throughout this book, rather than Fe3+ and Fe 2 +, thereby avoiding the
difficult problem of the actual charge distribution within the molecule. At
some points in the text, usually for emphasis or to facilitate electron counting,
there has been a reversion to the Fe3+ and Fe 2 + convention.
It is common practice to assign a charge to species such as [Fe(H 2 0) 6 ] 2 +
This assumes that there is no covalent interaction between the complex ion
Experimental 1117
and surrounding molecules. For example, it is assumed that in aqueous

solution [Fe(H 2 0) 6 ] 2 + either does not hydrogen bond with the solvent or
that any hydrogen bonding does not affect the electron distribution within
the complex ion. When attention is focused on the metal ion in these species,
this approximation introduces no difficulties. For quantitative work on the
interaction of a complex ion with its environment-and such an interaction
can be of vital importance in bioinorganic chemistry-it might become sen-
sible to further modify the nomenclature and to call this species [Fe(H 2 0) 6 ] 11
or something similar.
6.6 Experimental
The content of this chapter has been theoretical and so an experimental
section seems somewhat out of place. Yet it is not, for experimental evidence
from a variety of sources is becoming available which helps pin-point the
strengths and weaknesses of the discussion. The focus is on X-ray and
neutron diffraction data. Although in X-ray crystallography it is usual to
assume that atoms are spherical (so that it is assumed that on each atom in
a crystal there is a spherical distribution of electron density and it is this
which is responsible for diffracting X-rays), the precision of current apparatus
and techniques, at their best, is such that deviations from spherical can be
measured. In Fig. 6.30 are shown these deviations for the cobalt ion and
four coplanar N02 ligands in the complex ion [Co(N02 ) 6 ] 3 -. In this
complex the cobalt(III) has a t~.e~ configuration and the consequences
of this are evident in Fig. 6.30--there is a depletion of electron density in
the 'e• region', i.e. along metal-ligand bonds, and buildup in the 't 2 g regions',
i.e. between metal-ligand bonds, just as expected. The reader may also note
a buildup of electron density near the N atom of the Co-N bonds; even the
N-0 bonding electrons are visible. This result is typical of X-ray electron
density difference measurements. They support the general picture presented
both in this chapter and in Chapter 7.
FIC. 6.30 Electron density difference map for

the [Co(N0,) 6 ] 3 - ion in the CoN4 plane.
Within a solid contour the electron density has
been depleted and within a dotted contour it
has increased. Adapted and reproduced with
permission from s. Ohba, K. Toriumi, S. Sato
and Y. Sa~o. Acta CI)'St. (1978) 834, 3535.
',, '
: ___'',
' -
N
In most neutron diffraction experiments, it is (loosely put) the conse-

quences of neutrons bouncing off nuclei that are measured. Electrons are
not involved. However, this last statement is not always true. Neutrons,
like electrons, have a spin of! and so, like electrons, behave a bit like tiny
bar magnets, an analogy that will be pursued in Chapters 8 and II. This
means that, experimentally, one can sort, and work with, neutrons either
with spin up or with spin down. If such a beam of polarized neutrons
impinges on a crystal containing unpaired electrons then the associated
electron magnets will interact with the neutron magnets. If the electron spins,
the electron magnets, are at least partially orientated by application of a
magnetic field the electron-neutron interaction will depend on whether the
neutron magnets are up or down. The crystals of many transition metal
complexes contain unpaired electrons and diffraction measurements made
with polarized neutrons enable unpaired electron density distributions to be
Fig. 8.31 Spin density distribution in the measured. The [Cr(CN) 6 ] 3 - anion has a 4 A 2 .(t~.) ground state, is of high
(Cr(CN) 6 J"- anion in the CrC4 plane. The solid
contour defines a region of spin-up electron symmetry (the detailed geometrical structure of many ligands is such that
density and the dotted contour of spin-down. their complexes can never be truly octahedral), and is of particular interest
This diagram is from a model of the because ofthe expected involvement of chromium to CN- n bonding. Figure
experimental data which over-sharpens some
features so that fine detail is not reliable. 6.31 shows the experimental spin density distribution in this anion. Most
Adapted and reproduced with permission from obviously and surprisingly, it contains both spin up and down densities.
B. B. Rggls, s. B. Forsyth and P. A. Reynolds, Ignoring this problem for the moment it is noticeable that the spin up density
lnorg. Chem. (1987) 26, 101.
on the chromium is essentially where we expect it to be--in the t 2 • orbitals,
between the metal-ligand axes. Further, some of this electron density has
been transferred to a n orbital on the nitrogen, in accord with our n bonding
model. So far, so good, but what of the negative spin density? It is
energetically favoured for the three electrons in the t 2 • shell to have
their spins parallel-as will be seen in the next chapter, we have a 4 A 2 •
ground state. One talks of 'the (electron) exchange stabilization'. Why
should this privilege of exchange between parallel spins be restricted to the
chromium t 2 • electrons? In fact, other electrons can participate. In particular,
in the lone pair on the carbon of CN-, the two electrons are not uniformly
distributed in space. The one with spin up tends to be closer to the chromium
t 2 • electrons with spin up--this is exchange-preferred. There are two conse-
quences. First, these ligand electrons appear somewhat merged with the
chromium t 2 • electrons in Fig. 6.31. Second, they leave behind a balancing
down spin density in a u orbital on the carbon atom, thus accounting for
the experimental result.
Where does all this leave the model developed in this chapter? Clearly,
basically correct (if one is prepared to accept that the wrong placement of
ligand orbital energy levels is unimportant) but unable to account for fine
details such as unpaired spin distributions (the total uneven spin density on
each carbon in [Cr(CN) 6 ] 3 - amounts to about 0.1 electron). How can the
model be improved? The explanation given for the observation of an uneven
spin density depended on, loosely, how one electron behaved consequent on
the behaviour of another. The model presented in this chapter is a one-
electron model; electrons were talked of as individuals. In order to explain
fine details this one-electron model is inadequate, two-electron correlations
have to be included in our treatment, the second time in this chapter that
this conclusion has been deduced. This is not the last time that we shall find
a need to explicitly include electron correlation. At this point all that needs
Questions 1 119
to be added is that theoretical models which include electron correlation

not only predict, qualitatively, the presence of uneven spin densities in
[Cr(CN) 6 ] 3 - but also give quantitative data that are not too far from the
experimental observations. However, the routine and accurate inclusion
of correlation effects in calculations on transition metal compounds remains
an elusive goal. It is the subject of much current theoretical research. It
highlights the difference in difficulty presented by accurate calculations on
transition metal compounds and organic molecules-for the latter, the
inclusion of correlation is essentially a solved problem. As seen earlier, at
the end of Section 6.2, inclusion of electron correlation in calculations on
transition metal complexes may mean orbital energy level patterns rather
different from those given by calculations which exclude it.
Further reading (c) Modern Coordination Chemistry, J. Lewis and R. G. Wil-

kins, Interscience, New York, 1960.
Most contemporary texts in inorganic chemistry include a Recent theoretical work is, largely readably, illustrated in
treatment of the material in this chapter, although in less detail The Challenge of d and f Electrons, D. R. Salahub and M. C.
and depth. They may be useful, however, in painting a broad- Zerner (eds.), Symposium Series 394, American Chemical Soci-
brush picture. Other treatments tend to be rather mathematical ety, 1989.
and to give at least as much emphasis to crystal field theory Electron density distributions in inorganic compounds are
as to molecular orbital (or to their combination, ligand field reviewed by K. Toriumi and Y. Saito in Adv. Inorg. Chern.
theory); some older works remain the easiest to follow. Exam- Radiochem. ( 1984) 27, 27.
ples are: A good introduction to spin density distributions-which
1. A classic text, Introduction to Ligand Field Theory C. J. Ball- concludes that long range effects may well be important-
hausen, McGraw Hill, New York, 1962; the reader of the is 'The Magnetization Density of Hexacyanoferrate(III) Ion
present book may well find it easiest to skip the first few Measured by Polarized Neutron Diffraction in Cs 2 KFe(CN) 6 '
chapters of Ballhausen at a first reading. by C. A. Daul, P. Day, B. N. Figgis, H. U. Glide!, F. Herrin,
A. Ludi and P. A. Reynolds, Proc. Roy. Soc. Lond. (19-
2. Other texts or compilations-again, some selectivity will be 88) A419, 205.
needed. A very worthwhile article which emphasizes that main
(a) Introduction to Ligand Fields, B. J. Figgis, Wiley, New group and transition metal ions have much in common is 'The
York, 1966. Roles of d Electrons in Transition Metal Chemistry: a New
(b) Some Aspects of Crystal Field Theory, T. M. Dunn, R. S. Emphasis' by M. Gerloch, Coord. Chern. Rev. (1990) 99, 117.
McClure and R. G. Pearson, Harper and Row, New York,
1966.
Questions 6.2 Rehearse the arguments which lead to the conclusion

that only interactions of T 2 • symmetry need to be considered
6.1 Use Table 6.1 and Fig. 6.2 to show that the following when the problem of 1t bonding in octahedral complexes is
metal orbitals have the symmetries indicated studied.
6.3 Although both octahedral and tetrahedral complexes are

==
cubic (x y z), it is possible to be much more specific about
the effects of metal-ligand bonding in the former. Why?
Table 6.1 does not enable the g and u suffixes to be determined.

The answer to this problem is worked out in detail in the 6.4 Show that the five M-L a bonding orbitals in a ML 5
reference at the foot of page 97; the solution in this reference trigonal bipyramidal complex have 2A'{ + A; + E' symmetries.
includes the suffixes. For this problem the character table of the D3 • point group is
needed: 6.5 As indicated in the text, Fig. 6.30 shows a depletion

of electron density along the Co-N axis; it is to be noted that
Dah E 2C3 3C2 "h 2S3 3ad
this depletion is close to the Co (as expected). However,
A1 1 1 1 1 1 1 close to the N there is a buildup of electron density.
Suggest reasons for this buildup.
A2 1 1 -1 1 1 -1
E' 2 -1 0 2 -1 0
A1 1 1 1 -1 -1 -1
A2 1 1 -1 -1 -1 1
E" 2 -1 0 -2 1 0
Crystal field theory of
7.1 Introduction
Although little use is made now of the theory presented in this chapter,
it contains the basis of all of those that are used. It provides the foundation ,
particularly for the understanding of spectral and magnetic properties; all
else is elaboration and refinement. A knowledge of simple crystal field theory
is therefore essential to an understanding of the key properties of transition
metal complexes and particularly those covered in Chapters 8 and 9. This
chapter deals exclusively with transition metal complexes. In one or more
of their valence states, the ions of transition metals have their d orbitals
incompletely filled with electrons. As a result, their complexes have character-
istics not shared by complexes of the main group elements. It is the details
of the description of these incompletely filled shells which is our present
concern; this is in contrast to the discussion of the previous chapter where
the topic was scarcely addressed. Ions of the lanthanides and actinides
elements have incompletely filled f orbitals and so necessitate a separate
discussion which will be given in Chapter 11.
In 1929 Bethe published a paper in which he considered the effect of
taking an isolated cation, such as Na +, and placing it in the lattice
of an ionic crystal, such as NaCI. In particular, he was interested in what
happens to the energy levels of the free ion when it is placed in the
electrostatic field existing within the crystal, the so-called crystal field. The
energy levels of a free ion show a considerable degeneracy, particularly if
one is prepared to ignore effects which cause only small splittings. That is,
in the free ion there exist sets of wavefunctions, each member of any set being
quite independent of all other wavefunctions (i.e. orthogonal to them), yet
all members of any one set correspond to the same energy. What happens
to these ions when placed in an ionic crystal? Do the wavefunctions which
in the free ion had the same energy still all have the same energy in the
122 1 Crystal field theory of transition metal complexes
crystal? Bethe showed that in some cases the free ion degeneracy is retained
and in others it is lost, the crucial factors being the geometry of the crystalline
environment and the term ('S, 3 P, 2 D, 1F etc.) of the wavefunctions of
the free ion.
Two years later, in 1931, Garrick demonstrated that a simple ionic model
gives heats of formation for transition metal complexes which are in
remarkably good agreement with the experimental values. That is, these
complexes behave as if the bonding between the central metal ion and the
surrounding ligands is purely electrostatic, just as in a simple picture of the
bonding in NaCI. If this is so, then Bethe's work may be applied to complexes
as well as to ionic crystals and the energy levels of the central metal ion
related to those of the same ion in the gaseous state. All that is needed is a
suitable quantitative calculation to obtain the energy level splittings due to
the crystal field. This approach to the electronic structure of transition metal
complexes is known as crystal field theory and it is the subject of the present
chapter.
7.2 Symmetry and crystal field theory

Almost all the material of the present chapter arises from the symmetry
of the molecules considered. This symmetry finds expression in the subject
of group theory and, as in the previous chapter, to follow the arguments
we use the reader will need to be reasonably conversant with group-
theoretical jargon. The reader who needs a refresher course on the subject,
or perhaps somewhere in the depths of this chapter becomes uncertain of
their command, is reminded that there is a brief review of the essentials in
Appendix 3, although this is not to be regarded as a substitute for a proper
study.1 Some of the language that will be used is reviewed in the next few
paragraphs and, as in the previous chapter, the discussion will be confined
to octahedral complexes.
In the present chapter the symbols A 19 , A 29 , E9 , T 19 and T 29 will be
encountered with a spin state designated, for example, 2 E 9 ('doublet ee gee')
and 3 T 19 • The lower-case symbols used in Chapter 6 are also of frequent use,
so, a 19 , e., t 1• etc. These symbols may also have superscripts. We shall refer
to t~ 9 , ('tee two gee two'), for example when talking about two electrons in
a set of t 29 orbitals (other books may refer to this as (t 2.) 2 ). Symbols such
as d. or (t 29 ) 3 indicate that the set of three orbitals labelled t 29 are occupied
by three electrons. This usage follows the similar use of s, p and, very
important for this chapter, d to indicate an orbital occupancy. One talks of
sp(=s 1 p 1 ), d 2 , d\ d 3 p 2 and so on as configurations (appropriate to an
isolated atom). In just the same way, symbols such as t~ 9 , t~ 9 , and t~ 9 e: refer
to electron configurations in (for the examples given) octahedral (Oh)
symmetry.
It often happens that, for isolated atoms, one wishes to discuss all the
electrons-or, certainly, the outermost ones-collectively rather than as
individuals. One then refers to states or terms and uses uppercase symbols.
1 The present author has written a, hopefully, easy to read and follow, non-mathematical
text on group theory called Symmetry and Structure, 2nd edn., published by Wiley, Chichester,
1995.
Crystal field splittings 1 123
The difference between these names is rather subtle and different usages are
met. The most common is that in which the detailed mathematical specifi-
cation decreases in the order state > level > term. So, in most of our
discussion we will talk of terms, although the repeated use of this word may
give rise to some ugly language and although in some papers and texts the
word state is used almost interchangeably with it. After a phenomenon
known as spin-orbit coupling has been included we will talk of levels and
if we wish to be even more detailed and talk about an individual wave-
function, we will use the name state. Care must be taken to avoid confusing
this use of state with phrases such as 'the ground state' and 'an excited state'.
For our present purposes the important thing to note is that under the
rotational operations of an octahedron2 any S term has the same symmetry
as an s orbital, a P term the same symmetry properties as a set of three p
orbitals and so on. Key, however, is the fact that whilst a solitary electron
in a set of p orbitals gives a P term, it transpires that two electrons in the
d orbitals can give one also, as can three d electrons. We talk of 'the P term
arising from the p 1 configuration', or 'the P term arising from the d 2
configuration' and so on. A crucial factor turns out to be the number of
unpaired electrons, n, associated with each term and this is indicated by the
number (n + 1) as a superscript thus: 2 P, 3 D and so on. If the choice of
(n +I) seems odd-perhaps n seems more sensible-reflect on the fact when
n = 1, that is, when there is just one unpaired electron, this electron can
have spin up or down---there are two, (n + 1), different spin possibilities.
Although it is usual to arrive at symbols such as 3 P by feeding electrons into
orbitals, the absence of any specific reference to these orbitals in the final
symbol suggests that they are merely a convenient vehicle, and not essential.
This is the case. Symbols such asP and 3 P, like their lower case counterparts,
are a consequence of the (rotational) symmetry of a sphere. Indeed, the
symbols S, P, D, etc. are the labels of irreducible representations of the
relevant spherical group.
Although an octahedron has a much lower symmetry than a sphere it
would be reasonable to expect that many-electron wavefunctions would be
handled similarly. This is so-symbols such as 3 T 19 , 2 E 9 and 1 A 19 , like t 1 g,
e9 and a 19 orbitals, imply, respectively, triple, double and single orbital
degeneracy. In each case they are associated with a spin degeneracy which,
in each of these three examples, is identical to the spatial degeneracy.
However the two vary independently and so symbols such as 2 T 1 ", 3 £ 9 and
19 are perfectly reasonable.
6A
7.3 Crystal field splittings

In crystal field theory a complex is regarded as consJstlng of a central
metal cation surrounded by ionic or dipolar ligands which are electrostatically
attracted to the cation. The bonding within the complex arises from the
electrostatic attraction between the nucleus of the metal cation and the
electrons of the ligands. The interaction between the electrons of the cation
and those of the ligands is entirely repulsive. These repulsions will be central
2 In the previous chapter, the extension to include operations such as reflection and inversion
was indicated; a similar discussion will appear later in this chapter.
to the content of this chapter, for they are largely responsible for the energy
level splittings which were the subject of Bethe's paper. As we will see, group
theory tells us to expect these splittings (this is what Bethe showed). However,
group theory tells us nothing about the magnitude or even the sign of the
splittings. We need a specific model to do this and the simplest is the crystal
• field model. Actually, if we try to get numbers out of the model which can
then be compared with experiment we find that the model is not a very good
one. This is one aspect which has led to the development of the more
realistic model of Chapter 6. However, historically, virtually the entire detail
of the theory of transition metal ions was developed using the crystal field
model. The trick is never to use the model to get numbers. Rather, it is used
Fig. 7.1 An octahedral complex. In this chapter to focus our attention on energy differences and the relationships between
the way that an octahedral complex is drawn them. Experimental data are then used to obtain the numbers! It is this,
will vary, the perspective adopted depending on together with the fact that (as we shall see) the results are largely symmetry-
the point under discussion. Frequently, lines will
be drawn which represent the edges of the determined that lead to the utilization of the method. Consider the octahedral
octahedron rather than chemical bonds. complex shown in Fig. 7.1. What will be the effect of the crystal field on a
singles electron of the central metal ion (Fig. 7.2)? The ligand-metal electron
z repulsion which we have associated with the crystal field will raise the energy
of the s electron (or an S term), but as there is no orbital degeneracy, no
orbital splitting can result. Next, what will be the effect of the crystal field
on a single p electron (or a P term) of the metal ion? As is evident from
Fig. 7.3 all the p orbitals are equally affected by the crystal field and so, no
matter which of them the p electron occupies, the repulsion is the same. That
is, the p orbitals (or the components of a P term) remain triply degenerate
in an octahedral crystalline field.
The case of a single d electron (or a D term) is both more difficult and
more interesting. All five d orbitals are not spatially equivalent. Three, dxy•
~ dyz and dzx• are evidently equivalent for they are equivalently situated with
Y respect to the ligands (Fig. 7.4) and may be interchanged by simply
interchanging the labelling of the Cartesian coordinate axes. The other two
Fig. 7.2 A metal s orbital in an octahedral d orbitals, dx'-y' and dz'• are not equivalent, although they both have their
complex.
maximum amplitudes along the Cartesian coordinates axes (Fig. 7.5). Inter-
change of the labels associated with each of these axes has the effect not of
interchanging the orbitals but of generating new orbitals. So, starting with
the coordinate system of Fig. 7.5 the interchange x -> z -> y -> x gives us
z z z
t t
.-//0~0\ -, ,//c~0!\ . .~, ,-/~~,

Px Pr Pz
Fig. 7.3 A set of metal p orbitals in an octahedral crystal field. Because the coordinate axes of the octahedron are equivalent so too are the p orbitals.
They remain triply degenerate.
Crystal field splittings I 125
z
z z
J.
/ ~/
X X
dxy dyz
Fig. 7.4 The d,, d, and d~ metal orbitals in an octahedral crystal field. Because they are all equivalent (they can be interconverted by rotating
around the threefold axis approximately perpendicular to the plane of the paper) they remain triply degenerate.
z z
J.
Fig. 7.5 The dx2 -r and dz" orbitals in an

octahedral crystal field. Although they look very
different, the fact that they can be mixed by an
axis relabelling shows that they are a
degenerate pair (see the text and Fig. 7.6).
X X
Fig. 7.6 When the octahedral axes of Fig. 7.5

are relabelled
z~x
'\ <'
y
then the orbitals dx" and dy"-z" are obtained. It
may help to see this if both of the octahedra
drawn are mentally rotated by 120= anticlockwise
so that the axes return to the positions they
have in Fig. 7.5. Since nothing else has been
added or removed, the orbitals dx2 and dy2-z2
must be mixtures of the original dxz and dy2-x2· y ,;:/
dY' ~z2 d..,
Fig. 7.6 in which these same d orbitals are now labelled dx' and dy'-z" We
have not invented something new; the 'new' orbitals are simply mixtures of
the 'old'. The fact that it is possible to mix two orbitals by such a trivial
operation as relabelling the axes shows that the two orbitals arc degenerate.
If they were not, the mixed orbitals would have different energies from those
of the starting pair, and it is obviously ridiculous that the energies should
be a function of the labelling of the axis system. In Appendix 4 it is shown
in a more formal way that the new d orbitals are simply mixtures of the old
ones, dz' and dx'-y'· We conclude that the d orbitals (or aD term) split into
z z
X___-
Fig. 7. 7 In an octahedral crystal field the f

fz,; (fxs and fys are similar) fz (x:>- ys); (fx (Y'- z•l and fy (z2-x2) are similar)
orbitals of a metal atom split into two sets
which are triply degenerate and one orbital
which is singly degenerate.
z
X___-
fxyz
two sets, a set of three degenerate orbitals and a set of two degenerate
orbitals. The relative energies of these two sets will be discussed shortly.
Finally, we consider the effect of an octahedral crystal field on a single f
electron (or an F term). Fig. 7.7 pictures the seven f orbitals in an octahedral
environment. The lobes of three, fx'' fy, and f., point along axes. Three,
fz(x'-y'J' fx(y'-z'J and fy(z'-x'J have lobes located in coordinate planes (for the
fz<x'-y'J orbital shown in Fig. 7.7 these are the zx and yz planes). The last,
fxyzo has lobes pointing between all coordinate axes. Clearly in an octahedral
crystal field the f orbital sevenfold degeneracy is lost to give two sets of triply
degenerate orbitals and one singly degenerate orbital.
So far it has been shown that sets of d and f orbitals (and therefore D
and F terms) split into subsets in an octahedral crystal field but nothing has
been said about the relative energies of these subsets. For the moment, the
discussion will be restricted to orbitals because it is easy to give pictures
of them. In subsequent sections the discussion will be extended to the
corresponding terms. In preparation for this extension, it would be helpful
if the reader has some idea of their derivation so that he or she is fully aware,
for example, that an F state means seven spatial (as opposed to spin)
functions, just like a set off orbitals. One of the simplest ways of appreciating
this is through the Russell-Saunders coupling scheme, that which is adopted
to obtain the explicit functions themselves. This scheme is outlined in
F=== dx2-y.o. ct 22
;-----fxyz
(a)
(b)
Fig. 7.8 The combined repulsive and splitting

effect of an octahedral crystal field on the Appendix 5; part of it is also given in Section 11.3 which contains an example
energies of (a) a set of metal d orbitals and worked through in outline. If difficulty is encountered as the argument
(b) a set off orbitals.
develops over the following pages, these resources should provide help.
Consider the d orbitals shown in Figs. 7.4 and 7.5. The more stable set
is that in which an electron experiences least repulsion- .. destabilization-from
the electrons on the ligands. Evidently, this set is that composed of dx,, d,,
and d'"" because this set keeps the d electrons away from the ligand electrons,
at least when the ligands are represented as point charges. This set has T 2 ,
symmetry and we shall refer to these orbitals as 'the t 29 orbitals' or 'the t 2 ,
set'. The less stable set, the' e, orbitals' or' e, set', consists of d,, and dx'-y'·
In a similar way it can be seen that the relative stabilities of the f orbitals
(Fig. 7.7) is: fxyz (a 2 ,), most stable; fx(y'-z')• fy(z'-x') and fz(x'-y') (t 2 ,), inter-
mediate stability; fx,, f,, and f,, (t 1 "), least stable. The splitting patterns ford
and f orbitals are shown in Fig. 7.8 3 We shall return to the f orbital splitting
but for the moment confine ourselves to the d orbital case. The vast majority
of experimental data on transition metal complexes gives information on the
splitting between the d orbitals but not on their absolute displacements from
the free ion energy. It is therefore convenient to delete this unknown quantity
from the diagrams and to regard the free ion energy as lying at the centre
of gravity of the energies of the split orbitals (Fig. 7.9). The splitting between
the t 2 , and e9 sets of d electrons we shall call tJ. (some authors prefer to call
it I 0 Dq). For elements of the first transition series tJ. has a value of around
IOOOOcm- 1 ; for dispositive transition metal ions its value is usually
5000-15000cm- 1 ; and for tripositive ions I0000-30000cm-l Its value
increases roughly in proportion to the cation charge, depends markedly on
the ligands and, to a smaller extent on the metal (within any one transition
series). A complex of the second or third transition series has a value of tJ.
which is up to twice that of the corresponding first row complex.
The modified d orbital splitting pattern showing only the crystal field
effect given in Fig. 7.9 defines the splitting which is crucial to our discussion.
However, the argument we used to derive the splitting used plausibility in
place of mathematics so we cannot be absolutely confident of the results.
Ultimately, the justification for the splitting in Fig. 7.9 is experimental so
3 Beware the assumption that this diagram, although correct, may be used to explain the
properties of ions containing unpaired f electrons. The situation is more complicated, as will
become evident in Chapter 11.
dx,- !"'• d 22 (eg)
RJI. 7.9 The d orbital splitting in an octahedral

crystal field after the repulsion term common to
t<l
both sets has been deleted. The unsplit d
orbitals are at the centre of gravity of the split
sets.
~Ll
dXJ" dyz• dzx (t2g)
+I ++ ++
} Smallll.
I +
+
II
d.
+I
+
ds
++I
d6
N
d7
High spin
RJI. 7.10 The high spin (small Ll.) and low spin
(large Ll.) possibilities for d4 -d 7 octahedral
complexes.
Low spin
that ~is to be regarded as an experimental quantity. This is why we prefer

~to 10 Dq to describe the overall splitting-the D and q in 10 Dq relate to
specific mathematical functions in crystal field theory.
One immediate consequence of the splitting of d orbitals into t 29 and e9
sets must be recognized. When there are between four and seven d electrons
present there exist two quite different low-energy ways of allocating these
electrons to the t 29 and e9 orbitals. These are shown in Fig. 7.10. There is a
competition between the tendency of the electrons to stay as far apart
as possible-they repel each other-and the preference to occupy the
lowest-energy empty orbitals (remember that there is no covalency in
our model-pairing electrons brings no bonding stabilization). The high spin
arrangement is the one in which the interelectron repulsion between the d
electrons is smallest-the electrons are spatially less concentrated--and the
interelectron exchange stabilization is greatest (which is also why the

'maximum number of spins parallel' arrangement is of lowest energy). There
also exists the low spin arrangement which differs from the high spin in that
one or more electrons have been transferred from the less stable e9 to the
more stable t 29 orbitals. Each electron so transferred contributes a crystal
field stabilization of !l. to the system, but only at the cost of an increase in
the electron repulsion destabilization and a decrease in the exchange
stabilization (the latter, together, are often called the pairing energy). For
any one transition metal ion with from four to seven d electrons the vital
factor determining whether a particular complex is of the high or low spin
type is the magnitude of !l.. The change from one type to the other is
discontinuous; for other d" configurations there may be related, but less
dramatic, continuous changes associated with the transition from high
to low type. We may note at this point that the two types of complex
display quite different spectral and magnetic properties. For example, the
[Fe(CN) 6 ] 4 - ion, which is a low spin complex of Fe", a d 6 ion, is yellow
and has no unpaired electrons. The [Fe(H 2 0) 6 ]2+ ion, a high spin complex
of Fe" is pale blue4 and is paramagnetic, with four unpaired electrons. It is
easy to show the difference in the number of unpaired electrons. If finely
ground crystals of K 4 Fe(CN) 6 and then of FeS04 · 7H 2 0 are separately
dropped onto the poles of a strong magnet the latter show a definite tendency
to stick, whereas the former do not (neither stick to unmagnetized steel). This
different behaviour leads to an interest in the magnetic properties of
transition metal complexes and this forms the subject of Chapter 9.
In the crystal field model, the ligands are approximated by point charges
or dipoles, the value of !l. for a particular complex depending on the
magnitude of both this charge and that on the metal. This suggests that it
should be possible to place ligands in order of increasing effective charge
and, therefore, of increasing !l.. Further, this order should be the same for
all metals. Such an order of ligands was discovered by Tsuchida before the
advent of crystal field theory, and is called the spectrochemical series. As its
name implies the series was discovered as a result of a study of the (visible
region) spectra of transition metal complexes. It is therefore particularly
relevant to the content of Chapter 8, which deals with these spectra. An
abbreviated spectrochemical series, in order of increasing !l., is:
1- < Br- < SCN- (S-bonded) < Cl- < F- <OW < H2 0 < SCN- (N-bonded)
< NH 3 "'py <So~- < bpy < N02 (N-bonded) < CW
A similar series exists for the variation with metal ion in which it is seen,
as already mentioned, that !l. increases with the formal charge on the ion
and down the periodic table:
Mn 11 < Ni 11 < Co" < Fe 11 < V11 < Fe 111 < Cr 11 < V111 < Co"' < Mn 1v < Rh 111
< Pd 1v < lr111 < Pt 1v
Despite the above argument, the former series should not be regarded as a
series in which the charge on the ligand increases from left to right. Other
factors are involved, as was evident from the discussion in Chapter 6.
4 It commonly appears green, but this is caused by contammation with a small amount of
the yellow [Fe(H 2 0) 6 ] ' " .

These two series can be brought together for octahedral complexes,

Table 7.1 M and L1 values for
octahedral complexes including those which contain a mixture of ligands (here we anticipate
the 'rule of average environment', to be described in Section 8.9). Using Table
Metal M Ugand L, 7.1, one simply forms the product
Mnln 0.80 Br- 1.27
Ni 11 0.89 Cl- 1.33
Co" 0.93 F- 1.50
Fe 11 1.00 ow 1.57 to predict a value for Ll (em - t ). Here M and L 1 are taken from Table 7.1
cu" 1.20 H2 0 1.67
and n1 is the number of ligands of type I associated with L 1 in the
V" 1.23 seN- 1.72
Fe 11 1.40 NH3 2.08 complex (for genuine octahedral complexes n1 = 6). This procedure has been
C~" 1.74 bpy 2.38' the subject of some controversy in the literature, but the fact is that it works
VIII 1.86 NO:I 2.50 remarkably well.
Colli 1.90 CW 2.83 We now return to the problem we met at the beginning of the chapter.
Til II 2.03 What are the energy levels of a transition metal ion in an octahedral crystal
Mn 111 2.10
Mnlv 2.30 field? There are alternative approaches to this problem depending on whether
Rh 111 2.70 one is interested in a high or a low spin complex. Before embarking on this
I~" 3.20 discussion, a limitation in our definition must be recognized. We have talked
Pt'v 3.60 of high and low spin complexes, but recognized that this distinction is only
a For each donor atom; muttiply by 2 for the bidentate applicable to ions with between four and seven d electrons. Clearly, the
ligand.
magnitude of Ll can vary in a similar way for the other d electron
configurations, it is just that its variation does not have such evident
consequences. When we wish to talk of the entire set of d electron cases,
d 1-d 9 , we shall talk, instead, of weak field (small Ll) and strong field (large
M complexes although, in fact, these terms are not even synonymous with
high spin and low spin for the d4 -d 7 cases (the relationship will be made
clear later). The two cases will be considered separately and followed by a
discussion of the real-life case, in which the crystal field is of a strength
intermediate between those appropriate to the two extremes.
7.4 Weak field complexes

Weak field complexes are those for which the crystal splitting Ll, is smaller
than the electron-repulsion and -exchange energies. This at once indicates a
suitable theoretical approach to a discussion of their electronic structure. In
a general discussion of the energy levels of a free atom or ion one usually
considers the various interactions in order of importance. The most important
is the attraction between an electron and the nucleus. Next, the effects of
interelectron repulsion (this includes the exchange energy) are considered,
then the coupling between the spin and orbital motion of the electron
(spin-orbit coupling) and so on.
There would be no need to preserve this pecking order if each step in the
calculation were carried out exactly for both the ground and all excited states
of the atom-and this is what one would hope to achieve if a computer were
used to attack the problem. However, a deeper understanding is obtained
by following an algebraic approach. Here, each step usually involves
approximations and is only carried out for the ground and a few low-lying
excited states. Consequently, the step-wise procedure becomes necessary to
ensure that the properties of the ground state and terms immediately above
it in energy are described with fair accuracy.
Weak field complexes 1 131
In crystal field theory the central atom of a complex is regarded as a free

ion subject to an additional perturbation due to its environment. Evidently,
this additional perturbation must be introduced at the correct point in a
calculation and, for weak field complexes, this is after the effects of electron
repulsion have been dealt with. That is, when the calculation has reached
the stage of the classification of terms, such as 2 D, 3 F, 1 S, and so on. At this
point a very important characteristic has emerged. The most stable electron
arrangement in the free ion, which will be preserved in weak field complexes,
is one in which electron-electron repulsion is a minimum. The electrons are
as far apart as is possible. This will be when they occupy different orbitals
as much as is possible and this, in turn, means that the term of highest spatial
(orbital) and spin multiplicity will be the ground state. In this section we
shall therefore principally be interested in such terms. However, the way also
has to be prepared for the strong field case and this will necessitate
consideration of other, excited state, terms. Most important will be those of
the same spin degeneracy as the ground state, for spme of the physical
observables associated with transition metal complexes-colour, for instance
-can only be understood if they are included.
As has been pointed out, the spatial degeneracy implied by labels such
as P, D, F, G, ... (3, 5, 7 and 9, respectively) arise from the high rotational
symmetry of a sphere. In the Oh symmetry of an octahedral complex, the
central metal atom, instead of an environment with spherical symmetry, is
in one with only 24 rotational operations (cf. Fig. 6.2). This reduction in
symmetry means that there will also be a reduction in spatial degeneracy.
What we must now do, then, is determine how, for example, the 21
wavefunctions of the term which is the ground state of the free ion d 2
configuration, 3 F, (three spin wavefunctions each combined with any one of
seven spatial wave functions) split under the influence of the crystal field.
This, of course, is where we came in at the beginning of the chapter and is
the topic which was the subject of Bethe's paper. We ignore the spin
degeneracy (because the effects of a crystal field are the same whether an
electron spin is up or down; the crystal field may only have a small indirect
efTect on the spin degeneracy through the coupling which exists between the
spin and orbital motions of an electron). We arc left with the problem of
the F term and, as has already been pointed out, the splitting of an F term
parallels the splittings of f orbitals so that we conclude that the F term is
split into three substates, two of which are triply degenerate. It has already
been seen that a set off orbitals split up into t 1 "' t 2 , and a 2 u subsets. Similarly,
an F term splits up into either T,"' T2 "' and A2 " or T 19 , T 29 and A 29 subsets.
Which of these is correct is determined by the g or u nature of the
configuration from which the F term is derived. Because f orbitals are u in
character the 2 F term corresponding to an f 1 configuration splits up into
2 T"" 2 T 2 ,, and 2 A " components. Similarly, the 3 F term derived from the
2
d 2 configuration splits into 3 T 19 , 3T 29 and 2 A 2 " components because the d
orbitals are g in character. Note a clever, but potentially misleading, trick
here: the group of all rotations of a sphere docs not include the operation
of inversion in a centre of symmetry, i, and so the labels, S, P, D, ... tell us
nothing about behaviour under this operation. As a result, we are able to
apply them equally tog and u functions but as soon as a centre of symmetry
is encountered (as in Oh) we have to check whether we arc dealing with g
Table 7.2 Crystal field components of the ground and some excited terms of dn
(n = 1-9) configurations
Configuration Free ion Crystal field subterms Important Crystal

ground term excited term field term
d' 'O 2T:zg + 2Eg
d' 3F 3T'<I + 3T:zg + 3A2g 3p 3T'<I
d3 •F •r'<~ + •r"" + 4A:zg 4p •r'<~
d4 so sr:zg + sEg
ds •s "A:~g
d" so sr,g + sEg
d7 •F •r'<~ + •r,g + 4A:zg •p 3r.
3p 31<1
d" 3F 3T1g + 3T2g + 3A:zg Tlg
dg 20 2T2g + 2Eg
or u functions. In this chapter our concern is only with the crystal field
splitting of terms derived from d" configurations and because all d orbitals
are centrosymmetric, we shall encounter only g suffixes.
Table 7.2 gives the behaviour for all of the transition metal ions (d 1-d 9
configurations); it lists ground terms and, where the information is needed
later, the behaviour of the lowest excited state. Two points should be noted
in connection with this table. First, that only S, P, D, and F terms occur;
for each the splitting is similar to that of the corresponding orbitals. Secondly,
the table has some symmetry. Apart from the first column, the bottom half
is the mirror image of the top half. These two features combine to simplify
the remainder of our discussion of weak field complexes. The ground and
excited terms of transition metal ions have been introduced in Table 7.2
without any justification apart from that in Appendix 5. Such a justification
will be needed for the f electron systems in Chapter 11. The reader who
is unhappy with the present ex cathedra presentation should read Section
11.6 where the procedure is detailed; they may be fortified by the knowledge
that d electron systems are easier than their f electron counterparts! The
crystal field splittings given in Table 7.2 follow from the discussion earlier
in this chapter.
We turn now to the problem of the relative energies of the crystal-field
components listed in Table 7.2. Consider the D terms, which give T 2 • and
E 9 components. Which component is the more stable and by how much? It
will be taken for granted that it is an experimental fact that the five d orbitals
split as shown in Fig. 7.9 the splitting being denoted by .1.. Consider the
d 1 case, Fig. 7.11. Here, the ground state will be that in which the electron
occupies the lowest, t 2 ., orbitals. Just as a d 1 configuration gives rise to a
2 D term so, too,~~. configuration (which is what we have here) gives rise to
e:
a 2 T 29 term. Similarly, the (excited) configuration gives rise to a 'E• term.
We conclude that, for the d 1 case, the 2T 29 term is the more stable because
it means that the solitary d electron is in the t 29 orbitals. The (excited)
t2g
2 E. term is generated from it by excitation of an electron from the t 2 • orbitals
(a) (b)
to the e• orbitals. This, by definition, requires an energy .1., so we conclude
that the 2 T 2 • and 2 E. terms are also separated by the energy .1..
Fig. 7.11 (a) Ground and (b) excited states
derived from the d1 configuration in an Two points should be noted. First, a detailed argument is required to
octahedral crystal field. relate the splitting of orbital energies to the splitting of term energies.
Weak field complexes 1 133
This caution is necessary because, as will be seen in the following paragraph,

the fact that t 29 orbitals are more stable then e9 does not imply that
T 29 terms are necessarily more stable than £ 9 . Second, nine more electrons
could be accommodated in the orbitals of Fig. 7.11. The other positions are
vacant or, as it is more usually put, occupied by holes. A filled shell of
electrons has spherical symmetry; so too does a half-filled shell, provided
that no orbital of the shell is doubly occupied. The complement of a shell
(a) (b) half-filled with electrons is a half shell of holes which, therefore, also has
Fig. 7.12 (a) Ground and (b) an excited state spherical symmetry. As has been indicated, our discussion is symmetry-
with the same spin multiplicity, derived from the derived and something of spherical symmetry (more precisely, something
d4 configuration in an octahedral crystal field. In which is totally symmetric) never changes symmetry arguments and makes
the ground state there is a hole 1n the eg
orbitals; in the excited state the hole is 1n the a constant contribution to energies (this is the reason that inner-shell
t2 g orbitals. electrons can be ignored). It follows that when we have a half shell of
electrons or, equivalently, holes, this half shell may be neglected. This is
illustrated in the next paragraph, which develops the electron-hole relation-
ship.
We consider the d 1 case in which the t 29 orbital is the one occupied. There
are two ways of describing the situation in which the t 29 orbitals are occupied
by a single electron. We may say that the situation differs from spherical
symmetry either by the presence of a t 29 electron or by the presence of two
holes in the e9 set and two holes in the t 29 set-that is, a hole is missing in
eg 4+++::1=== the t 29 set. Obviously, it makes no sense to use the hole description ford 1 (ti,)
case but sometimes it is useful to talk in terms of holes. Such a case is
provided by the ground state 5 D term of the d 4 configuration. The quintet
spin state means that all four of the d electrons have parallel spin, as shown
in Fig. 7.12. The electron distribution differs from spherical symmetry by the
H+ presence of four electrons, or, what is equivalent, by the presence of one
(a) (b)
hole. Just as it was sensible to discuss the d 1 case in terms of one electron
rather that four holes, so it is sensible to discuss the d 4 case in terms of one
Fig. 7.13 (a) Ground and (b) lowest excited
state of the same spin multiplicity, derived from
hole rather than four electrons. The most stable situation is that in which
the d6 configuration in an octahedral crystal the hole is in the e, orbitals. The ground term is therefore 5 £ 9 . At an energy
field. Ll. above the ground state is the 5 T 29 term, in which the hole is in the t 2 ,
orbitals. In the d 4 case, therefore, the splitting of the £ 9 and T 29 levels is the
inverse of that in the d 1 case. It is worthwhile emphasizing again that, as
has just been seen, the fact that t 2 , orbitals are more stable than e9 does not
mean that T 29 terms are automatically more stable that £ 9 •
In the 5 D term arising from the d 6 configuration the orbital occupation
is as shown in Fig. 7.13. It differs from spherical symmetry by the presence
of a single electron, which is most stable when in the t 29 orbitals. The splitting
therefore follows the d 1 case, the 5 T 2 , term being more stable than the 5 E,
H H H term by an energy Ll.. Similarly, the 2 D term arising from the d 9 configuration
differs from spherical symmetry by the presence of a hole in the e9 orbitals
in the ground state, as shown in Fig. 7.14. The ground state is therefore
of 2 E9 symmetry and the excited state, at an energy of Ll. above. is of 2 T 2 ,1
HH HH
symmetry.
H Why is the hole formalism so useful? When a set oft 29 orbitals contains
(a) (b)
a single electron a T 29 term results. What if it contains two electrons; does
this also result in a T 29 term'' The answer, as we shall see later. is that it
Fig. 7.14 (a) Ground and (b) excited state
derived from the d9 configuration in an docs, but other possibilities also exist (in the same way that a d 2 configuration
octahedral crystal field. gives rise to other than D terms). Similarly. if we evaluate the terms arising
from the t~.e; configuration (using methods to be described later) we find

a multitude of them. Some terms correspond to singlet spin terms, others to
triplets and one to a quintuplet. If we are interested in the splitting of the
5 D term (the ground state of the free ion d 4 configuration) we are only
interested in the spin quintuplet, and work is obviously involved in sorting

it out from the others. The hole formalism does this work for us. Similarly,
if we wish to write down an explicit expression for the wavefunctions of the
5 D term arising from the d 4 configuration it is much easier to write down a
wavefunction appropriate to the single hole than to write down a (product)

wave function appropriate to the four electrons. However, care is needed in
the use of these pseudowavefunctions because holes behave differently from
electrons. For example, it was argued earlier that metal electrons are repelled
by the ligand field. It follows that, in contrast, holes are attracted by a ligand
field.
The lowest term arising from the d 2 configuration of a free metal ion is
3 F.lt has already been shown that this gives 3 T 1., 3 T2 • and 3 A 2 • components
in a ligand field. What is their relative ordering? Following our discussion

of the splitting off orbitals in a crystalline field we might anticipate that the
T 2 • term would be of intermediate stability and, following our discussion of
the splitting of D terms, expect that T 1• and A 2 • would alternate as the ground
state for the d 4 , d 7 and d 8 weak field cases. Detailed calculations confirm
both of these predictions. Although it is beyond the scope of this book to
give the detailed calculations,5 this ordering of terms will be justified, and
values for the relative energies of the components obtained, in Section 7.5.
Having done this, the subject becomes easier because the splittings of all
other F terms follow from the fact that for the 3 F term derived from the d 2
configuration the 3 T 1• term is lowest and 3 A 2 • the highest. This will now be
done fairly briefly; the task will be made easier by anticipating the results
we shall later derive for the orbital occupancies associated with each term.
This derivation will also help to deepen the understanding of the following
results. At this point, remember that all starts from the d 2 case and follows
either directly (d 7 ), or by the hole-electron analogy (d 3 and d 8 ). For d 7 and
d 8 , the spherical symmetry of a half-filled shell is also involved.
In theer,.) ground term derived from the 3 F term of the d 2 configuration
we will find that there are two electrons in the t 2 • orbitals. 6 For the 4 F term
arising from the d 3 case, the~~. term derived from it in an octahedral crystal
field corresponds to the presence of two holes in the t 2 • orbitals and so is
an excited state (Figs. 7.15 and 7.16); the 4 A2 • state is the ground state in
this case. Note the switch from electrons to holes in the last two sentences
and the consequent change of ground state into excited state when referring
to the~~. term. For the 4 F term of the d 7 configuration the ground state
is 4 T 1.; the configuration differs from spherical symmetry by the presence of
two electrons in the t 2 • orbitals (Fig. 7.17) and so the splitting parallels the
d 2 case (the spin states differ, of course, but this is irrelevant). The 3 F term
5 A reasonably simple treatment is given in Valence Theory by J. N. Murrell, S. F. A. Kettle

and 1. Tedder, Wiley, London (1965) Chapter 13.
6 This statement is not quite correct, as we shall see at the end of Section 7.5. Similarly,
errors are contained in the statements on the d 3 , d 7 and d 8 configurations. These errors are
made in the interest of linguistic and conceptual simplicity and in no way invalidate the general
argument.
Weak field complexes I 135
····························································································································································································
eg
FJC. 7.15 (a) Ground and (b, c) two excited

state configurations of a d2 ion in an octahedral
crystal field. The triple orbital degeneracy of the
3 T,_, ground state may be associated with the t2g
three possible orbital sites tor the hole in (a).
(a) (b) (c)
eg
Fig. 7.18 (a) Ground and (b, c) two excited
state configurations of a d3
ion in an octahedral
crystal field. Note that there is only one
distinguishable way of arranging the electrons in
t2g
the three orbitals in the ground state-and so ~-
the ground state is orbitally non-degenerate.
(a) (b) (c)
Fig. 7.17 (a) Ground and (b, c) two excited

eg
H+ +I +I
crystal field.
+
H H H+
t2g
(a) (b) (c)
FJC. 7.18 (a) Ground and (b, c) two excited

eg H+ HH
crystal field.
H +
H H HH H+
t2g
(a) (b) (c)
derived from the dB configuration gives rise to an excited 3T 1• term (Fig.

7.18), two holes occupying the t 29 orbitals, and a 3 A 2 • ground state,
paralleling the d 3 case-and so the opposite splitting pattern to d 2 -
remember that holes and electrons behave in opposite ways in crystal fields.
The only other weak-field case which remains to be discussed is that of
the 6 S ground state of the d 5 configuration. This ground state is orbitally
non-degenerate and so no crystal field splitting can occur. The behaviour of
this 6 S term parallels that of an s orbital and becomes 6 A 19 in an octahedral
crystal field.
So far the P excited states of the d 2 , d 3 , d 7 , and dB configuration which
were given in Table 7.2 have not been mentioned. These excited states give
rise to a T 19 term which is of the same spin multiplicity as the T 1• term
derived from the corresponding free-ion F ground term, the energy of which
has just been discussed. These two T 1• terms, being of the same spin
multiplicity and orbital symmetry, may interact under the influence of the
crystal field. In the limit of a very weak crystal field-and such a field is our
present concern-this interaction is correspondingly weak and may be
ignored. Attention is drawn to it at this point as it is referred to in Sections
7.5 and 7.6.
sfg
•r,g
6s 6A1g so 4F 3F
3T2g
sr,g
4hg
d5 d6 Ll d7 Ll
3A2g
The arguments of this section are summarized by showing in Fig. 7.19

the splittings, in the weak field limit, of the ground state free-ion terms listed
in Table 7.2. The behaviour shown is simple-the energies are a linear
function of Ll. (introduction of the additional T 1 , levels mentioned above,
will, when we include them, introduce a curvature into the T,, behaviour).
So far, the actual energies for the components of split D terms are the only
ones to have been discussed; although F terms have been considered in
outline, the energies of their components in an octahedral crystal field have
not been obtained-this will be done later.
7.5 Strong field complexes

Fig. 7.19 The splitting of the weak field d"
Strong field complexes are distinguished from weak field complexes because
(n ~ 1-9) ground state terms in an octahedral their crystal field splitting energy, Ll., is greater than the energies associated
crystal field. with electron pairing. In an actual calculation this means that instead of
finding the terms arising from a d" configuration-otherwise the first step
in a step-wise, pecking order, calculation-one first applies the crystal field
perturbation. This divides the d orbitals into e, and t 2 , sets. Ad" configuration
splits up into sets which differ in the occupancy of the t 2 , and e, orbitals. So,
for example, the d 2 configuration splits into three sets t~,, tLe~ and e~. A
Strong field complexes 1 137
complete list of these substates is given in Table 7.3 which possesses

Table 7.3 Strong-field substates for the same sort of symmetry as Table 7.2. The number of substates listed for
d" configurations
a d" configuration is the same as that for a d 10 -• configuration although
d" Strong field configurations inspection shows clear differences in orbital occupancies. These differences
are only apparent. The reader should easily be able to show that, if the hole
dl t~, ei formalism is used for one of them, the difference between them disappears.
d2 f,g, t~gei, e~
For the d 8 configuration, for example, the hole formalism leads to t~ 9 , ti.e~
d3
d4
~g. t~ei, t~e~, e~
f,., ~.ei, f',.e~, t~e~, e; and e; hole configurations, the same as the electron configurations listed
d5 t~, tt,ei, ~e~, f,ge~, t~e; for the d 2 configuration in Table 7.3. The relative energies of the terms given
d6 F,g, ~ei, t<,.e~, ~e~, f',.e; in Table 7.3 are readily evaluated; in evaluating them the advantage of
d7 F,gei, t~e~, f,gei, ~e; arbitrarily placing the energy of the d orbitals of the free ion at the centre
d" t~e~, t~e~, t~e; of gravity of those of the complexed ion will be seen. With this convention,
dg t~ge:. t~e; it is obvious (Fig. 7.9) that the t 29 orbitals are stabilized by %.1. and the e9
orbitals destabilized by~ .1.. Following the usual sign convention, the energy
of the t 29 orbitals is-% .1. and that of the e9 orbitals~ .1.. That is, each electron
in a t 29 orbital contributes -% .1., and each electron in an e9 orbital ~ .1.,
to the total d orbital energy. As an example of the application of this,
consider the ti 9 , ti 9 ei and e; configurations derived from d 2 These have
energies, respectively, of 2( -% .1.) = -% .1., ( -% .1. + ~ .1.) =! .1., and 2(~ .1.) =
~ .1.; three levels, each .1. away from its neighbour(s).
This is a convenient point at which to begin to introduce some diagrams
to which our discussion is leading. The beginnings of the diagram relevant
to the d 2 configuration is given in Fig. 7.20. The weak field limit (.1. = 0) is
represented by the left hand vertical line on which will be seen two of the
labels which featured in the discussion of the weak field case, F and P. At
the right hand edge of the diagram (.1. infinite) will be seen-a bit buried-
the energies of the three strong-field configurations arising from the d 2
configuration in the octahedral case (d 9 at -% .1., ti 9 e: at! .1. and e; at~ .1.).
The second of these configurations differs from the first in that it has one
fewer t 29 electrons and one more e9 • The third differs from the second in the
same way. Now the energy required to move an electron from a t 29 orbital
to an e9 is .1. (this is the definition of .1.). It follows that in Fig. 7.20 the three
configurations are equally spaced (by .1.) along the right-hand vertical axis.
Along the horizontal axis .1. runs from 0 to oo, as in Fig. 7.20, and this entire
range is included in the figure. It follows that the .1. scale must be nonlinear.
Nonetheless, it will be assumed to be linear near the weak field and strong
field axes and that the non-linearity is somehow accommodated in-between.
As Fig. 7.20 shows, the discussion so far has left us in a rather difficult
position. At the weak field limit we know the symmetries of the crystal field
terms but only for one, the 3T 19 arising from the free ion 3 P term (usually
denoted 3T 1.(P)) do we know an energy. Because the 3 P term does not split
in a crystal field, the energy of the 3 T 1.(P) term is independent of .1.. The
energy is therefore shown as a horizontal line at the weak field limit. In the
strong field limit we know energies; all terms arising from thee; configuration
have an energy of~ .1., all arising from ti.e: have an energy of! .1. and all
arising from tL have an energy of -% L\. In Fig. 7.20 lines of the appropriate
slope have been drawn at the strong field limit but we neither know the
symmetry labels to be attached to these lines nor, indeed, whether each line
represents more than one energy level. Our next task is to tackle this latter
problem; once this is done we shall be able to complete Fig. 7.20.
Fig. 7.20 The weak and strong field limits for

the d2 configuration in an octahedral crystal field.
Weak Strong
field field
limit limit
ll=O Ll==
By the end of the discussion of weak field complexes the effects of both
the (weak) crystal field and electron repulsion had been included. To be
consistent we must consider the effects of electron repulsion in strong field
complexes, and this we now do. Electron repulsion causes some arrangements
of electrons within an unfilled shell to be more stable than others-those
arrangements in which unpaired electrons are kept farthest apart and have
parallel spins will be the most stable (Hund's rules). That is, in a free
atom or ion, electron repulsion causes the terms arising from a configuration
to have different energies, as we have seen. Similarly, in a crystal field, the
terms arising from a configuration like t~" will, in general, have different
energies because of electron repulsion. How does one determine the terms
arising from such a configuration? What are the relative energies of these
terms? We shall answer these questions by looking at the group theory of the
problem. Results that are qualitatively correct will be obtained, with no need
to evaluate a single integral.
Table 7.4 is a table of direct products; its derivation is included in Appendix
3. This table is important, for it is used whenever one is simultaneously
interested in two similar quantities associated with an octahedral molecule;
for example, if we are interested in the symmetry properties of two electrons
as a pair rather than as individuals. Similarly, this table will be used to
discuss spectra, for which the ground and excited states of a molecule have
to be considered simultaneously. Table 7.4 does not give g and u suffixes;
they could have been included, but this would have made the table four
times larger with no increase in real content. The way that these suffixes may
Strong field complexes 1 139
Table 7.4 The oh direct product table (g X g= u X u = g; g X u = u)
oh Al A2 E T1 Tl
Al Al A2 E Tl Tl
A2 A2 Al E Tl Tl
E E E A1 +A2 + E Tl + Tl T1 + Tl
Tl Tl Tl rl + r2 A1 + E + T1 + T2 A2 +E+T1 +T2
Tl Tl Tl Tl + Tl A2 + E + T1 + T2 ~+E+T1+T2
be added will be shown shortly but first the meaning of the entries in Table
7.4 must be explained.
As our immediate aim is to complete Fig. 7.20 our discussion will be
confined to the d 2 case. Consider first the ti.e~ configuration.The first
electron may be fed into any one of three r2 " orbitals and the second into
any one of two e,. That is, there are 3 x 2 = 6 ways of feeding the two
electrons in; there are six orbitally-different wavefunctions. Table 7.4 shows
that the direct product of t 2 (extreme left-hand column) with e (top row)
written T 2 x E, is equal to T 1 + T 2 or, including g suffixes in an obviously
sensible way, T 29 x E, = T 19 + T 29 (note that T 2 , x Eu or T 2 u x E9 would
have given T 1u + T 2 uJ. The other possibility, T 2 u x E,, would also have
given T,. + T 2 , (g x g = u x u = g; u x g = g x u = u). The sum of the
degeneracies implied in T 19 and T 29 (3 + 3 = 6) is the same as the number
of orbital wavefunctions arising from the ti,e~ configuration (3 x 2 = 6). It
will not surprise the reader to learn that these six ti,e~ two-electron
wavefunctions divide into two sets of three each, one set of T 19 symmetry
and one set of T 29 symmetry. That is, the configuration tie; gives rise to T 19
and T 2 , terms. This is no accident-it is a group theoretical requirement and
would be just as valid in a different context (molecular vibrations, for
instance, if we simultaneously excite T 29 and E, molecular vibrations of an
octahedral complex the molecule could end up in either a-vibrational-T 1 "
or T 2 " state). So far, the spin of the electrons has not been mentioned. Because
in the tLe; configuration the two electrons always occupy different orbitals,
there are no constraints and all paired (singlet) and parallel (triplet) spin
arrangements are compatible with all the orbital symmetries that arise. We
conclude that the tLe; configuration gives rise to 3 T 1,, 3 T 2 ,, 1 T 1" and 1 T2 ,
terms. This means that the line at the strong field limit in Fig. 7.20 with
slope ! D. is a superposition of lines corresponding to these four terms.
However, because Fig. 7.20 is only concerned with triplet spin terms, only
the labels 3T 19 and 3 T 29 have to be added to this line in Fig. 7.20.
As a check on the answer that has been obtained it is helpful to count
wavefunctions. Previously, we only counted orbital functions; what if we
include spin° Now, there are six different ways of putting an electron into
the t 2 , orbitals (three orbitals, two spins) and four ways of putting one
into the e., a total of 24 different ways of putting in the two electrons;
that is, there are 24 different wavefunctions. This is just the number implied
by 'T,, + 3T 2 " + 'T," + 1T 2 , (9 + 9 + 3 + 3). The counts agree, as they
have to.
We now consider the e~ configuration, associated with the strong field
line of slope %D. in Fig. 7.20. The first electron can be fed into the
e, orbitals in any one of four ways (two orbitals, two spins). The second
cannot be in the same orbital with the same spin as the first and so the
number of distinct two-electron wavefunctions is 1(4 x 3) = 6. The 1 arises
from the word distinct: for instance, the 4 x 3 includes both (j l) and (L j)
yet these are not distinct; we have counted everything twice. As Table 7.4
shows, the direct product E, x E 9 = A 19 + A 29 + E,. We know that the e;
configuration must give rise to at least one spin triplet (because we can put
one electron into each e, orbital and they can be put in with parallel spin).
But, as Fig. 7.20 shows, in the weak field limit there is no 3 A 1 , or 3 E, term,
only 3 A 29 • We conclude that the e; configuration gives rise to the terms
19 + A 2 , + E 9 • This conclusion can be checked by counting the number
1A 3 1
of two-electron wavefunctions implicit in these terms. It is I + 3 + 2 = 6, in

agreement with the number obtained at the beginning of this paragraph.
Since we are only interested in triplet spin terms, the label 3 A 2 , may now
be added to the line of slope %L1 in the strong field limit of Fig. 7.20.
Notice an important distinction between the treatment of the ti.e; and the
e~ (and, in the next paragraph, the t;) configurations. For the ti,e~ case the
spin and orbital motions of the electrons were treated independently-a term
of given orbital symmetry appeared both as spin singlet and as spin triplet.
When the two electrons are in the same orbital set this independence
disappears; a term of given orbital symmetry appears as either a spin singlet
or as a spin triplet, never both. This apparently arbitrary distinction actually
arises from the fact that the (spin plus orbital) wavefunctions have to be
antisymmetric (go into themselves multiplied by a factor of -I) on
interchange of two electrons in the same orbital set. The connection between
the requirement and the consequence is not immediately obvious but can be
expressed group theoretically by an extension of the concept of the direct
product beyond that used in this chapter.
We now turn to the d, configuration but we already know what we expect
to find-those terms which appear at the weak field side of Fig. 7.20 and
which have not yet been obtained. That is, we are looking for a 3 T 19 term,
the only spin triplet unaccounted for. As Table 7.4 shows, the direct product
T 19 x T 1 , = A 1 , + E9 + T 19 + T 29 • We expect that the addition of spin labels
will lead to 1 A 19 + 1E, + 3 T 19 + 1T 29 . Is this correct? A partial check can
be made by counting two-electron wavefunctions. The configuration t~,
implies 1(6 x 5) = 15 two-electron wavefunctions. This is the number implicit
in 1 A 19 + 1 E, + 3 T 19 + 1T 29 (1 + 2 + 9 + 3 = 15). It is reasonable then, to
add the label 3T 19 to the line of slope -4 L1 at the strong field limit
of Fig. 7.20.
All that we need to complete Fig. 7.20 is a knowledge of the relative
energies (i.e. the L1 dependence) of the 3 A 2 , + 3T 19 + 3 T 29 components
of the 3 F term in the weak field limit. This is easy. We know the energies of
the 3 A 29 and 3 T 29 states in the strong field limit; they are %L1 and ! L1,
respectively. There is no reason why these L1 dependencies should change as
L1 decreases and so we conclude that these values are also their energies in
the weak field limit. All that remains is to obtain the energy of the 3 T 19 term
arising from the 3 F term (usually denoted 3T 1,(F)). We might be tempted
to simply look at the 3 T 19 term arising from the t~, strong field configuration
and conclude that the answer is -4 L1. However, a little thought will indicate
that caution is needed. The 3T 19(P) term does not depend on L1 in the weak
Strong field complexes I 141
field limit but has to correlate with the 3 T 19 term arising from the t~ 9 e:
configuration, a term which has an energy of! .1. (this has to correlate because
if it did not the non-crossing rule-the rule that the energy levels of terms
of the same symmetry and spin degeneracy do not cross-would be violated).
So, the term 3 T 19 (P) changes its dependence on .1. as .1. itself changes-and
this is caused by the presence of the second 3 T 1• term, with which it
interacts-and it is this latter that we are interested in. The moral is: be
careful when there is more than one term of a given symmetry and spin
multiplicity. Actually, it is not difficult to calculate the energy of the 3T 19(F)
term. Note that it is the splitting of the 3 F weak field term that gives
3 A 29 + 3 T 1 • + 3 T2 • terms. The word splitting implies that the energies of the
components sum to that of the 3 F term; their .1. dependencies sum to zero.
Denoting the energy of the 3 T,.(F) term by S(T,.), taking care to weight each
energy by the number of wavefunctions with this energy, we have
(3 X ~ .1.) + (9 X 8(T1g) + (9 X g.1.) = 0
That is, S(T1•) = -~ .1.; our caution was justified-in the weak field limit the
3 T 1.(F) term does not have the same energy as in the strong field limit. The
two energies differ by -! .1., equal and magnitude but opposite in sign to
the difference between the energies of the 3 T,.(P) term in the same limits.
The crystal field has caused the two 3 T 1• terms to interact; they 'push' each
other apart by equal and opposite amounts. Figure 7.20 can now be
completed. The final version is shown in Fig. 7.21 which embodies all the
content of the above discussion. One final point. As befits the above
discussion, in this figure straight lines are drawn representing 3 A 2 • and 3 T 29
terms connecting the weak and strong field limits. Given what was said
earlier about the non-linear scale of the .1. axis in this figure, such straight
lines cannot be justified. The moral is clear: Fig. 7.21, and related ones which
will be presented shortly, are of pedagogical use only. However, they serve as
an excellent introduction to related diagrams (Tanabe-Sugano diagrams)
which have well-defined energy scales.
The discussion so far almost, but not quite, enables an extension to all
d" systems either by the electron-hole parallel, by neglect of half-filled shells,
or both. The omissions can be dealt with by arguments paralleling that which
follows for the d. configuration, a configuration which poses a problem.
The most obvious way of obtaining the orbital terms arising from this
configuration is to simply take each of the orbital terms arising from the d.
configuration and combine each with a further t 29 orbital function. That is,
to form the triple direct product T 2 • x T 2 • x T 2 • or, equivalently, consider
the sum of direct products
(Alg X T2g) + (Eg X T2g) + (Tlg X T2g) + (T2g X T2g)
This would be wrong. Some of the spin singlet wavefunctions of the T~.
configuration represent two electrons occupying the same orbital. To form
direct products blindly would, for some of the three-electron wavefunctions,
be to allocate all three electrons to one orbital! The simplest way to avoid
this problem and obtain the correct answer is as follows. In the 3 T 19 term
of the tL
configuration we know that the electrons have parallel spins and
so must occupy different orbitals (because of the Pauli exclusion principle).
Adding a third t 2 • electron can never give us three electrons in one orbital.
Fig. 7.21 The correlation between

weak and strong field limits for the 3F
d2 configuration in an octahedral
crystal field.
Weak Strong
field field
limit limit
Figure 7.22 gives diagrammatic representations of the 3 T 1, orbital wave-
functions (spin is not specified). Also in Fig. 7.22 are shown all the possible
ways of orbitally allocating electrons in tL configuration, again without
specifying spin. It is easy to see that all of the tL arrangements may be
obtained from those of the orbital components of the 3 T 19 term. This suggests
that the direct product T 19 x T 2 , will give us the symmetries of all the sets
of d, three-electron wavefunctions. This is so; from Table 7.4 it is seen that
they are A 29 , E,, T 19 and T 29 • The spin multiplicities of these terms now have
to be added. It is easy to sec that only one spin quartet term can exist and that
this is orbitally singly degenerate~ there is only one way of allocating three
electrons with a spin to the three t 29 orbitals. The other terms must therefore
be doublets; that is, we have 4 A2 ,, 2 E,, 2T 1 , and 2T 29 . Again, the total
degeneracy (4 + 4 + 6 + 6 = 20) equals the number of distinguishable and
allowed ways of feeding three electrons into the t 29 orbitals:
(b)
Fig. 7.22 (a) The three orbital arrangements _.._(6_x---,-5_x_4c:-c) = 20

associated with the triple orbital degeneracy of (1 X 2 X 3)
the 3 T,, ground state of the d2 (t~)
configuration (cf. Fig. 7.15). (b) The seven
different orbital arrangements corresponding to Similar arguments may be applied to all the other configurations which arise,
the ~g configuration may be derived from those although when either or both of the t 2 • and e" shells are more than half full
of the t~g as indicated. It will be noticed that it is convenient to work in terms of holes. The results, of course, are the
there are three arrows connecting with the top
line of (b). One spin arrangement associated same as those for the similar electron configurations, both for space and
with this top line is 'all spins parallel'-4 A2 g. The spin. Results are collected together in Table 7.5 which may appear frightening
other two must be spin doublets, which, in its complexity. Fortunately, Hund's rule applies to each configuration and
together with the six below, give a total of
eight spin doublet functions, corresponding to we shall usually be interested only in the most stable terms, those of highest
the 'Eg + 'r,.+ 'T,g terms (cf. Table 7.5). spin multiplicity, of each configuration.
Intermediate field complexes 1 143
Table 7.5 Terms arising from t~eg configurations
Number of Configuration Terms arising

d electrons
1, 9 t2g t~ge; 'T"'

e, t~ge~ 'Eg
2,8 t~g tige; 3Tlg, lAtg, lEg, 1T2g
t,ge, t~ge~ 3Ttg, 3T2g• 1Ttg, 1T2g
e'g t~ge~ 3A2g, 1A4i, lEg
3, 7 ~. ~.e; 4 A2g, 2 Eg, 2T41 , 2 T2g
t~geg tige: 4T
41 , 4 T2 g, 2 A41 , 2A2g, 2 2 E,, 2 2 T41 , 2 2 T2 g
t 2 ge~ t~ge~ 4T 2 2 T2 g
41 , 2 2 T1,,
e3g t~geg 'Eg
3T
4, 6 tig t~ge; 41 , A41 , 'E,, T41
1 1
~geg ~.ei 5Eg, 3A 41 , 3A2,, 2 3 Eg, 2 3 T;;g. 2 3 T41 , 1A 41 , 1A2g.

1 Eg, 2 1 T , 2 1 T g
41 2
t~ge~ tjge~ 5T2g, 3A 2g, 3 Eg. 3 3 T41 , 2 3 T2 g, 2 1A 41 , 1A2,,

3 1 Eg. 1 T1 g, 3 1 T2 g
t 2 ge~ t~geg 3T
41 , 3 r,•. 'r41 • 'r,,
e4g F,g 'A"'
5 t~g t2ge; 2T2g
tigeg t~ge~ 4T
41 , 4 T2 g, 2A41 , 2A2 g, 2 2 Eg. 2 2 T41 , 2 2 T2 g
~ge~ 6A 41 , 4A41 , 4 T2g, 2 4E,, 4 T41 , 4 T2g, 22A 41 , 2A2g,
3 2 Eg, 4 2 T41 , 4 2 T2g
In the preceding paragraphs, our concern has largely been with the
d 2 case. The results obtained for this configuration may readily be extended
to other cases; this extension is given in the next section. The present section
is concluded by enquiring into the d orbital occupancy when a weak field
d 2 ion is in its ground state (the d 7 case is similar). The energy of the 3 T 19
ground term, -~ L\, must correspond to an electron distribution of ~
electrons in the t 29 orbitals and 1 in the e, orbitals:
(g X -~ M + (t X g!l) ~ -~"'
If talking in terms of fractions of electrons seems strange, remember that

these fractions refer to the probability distribution of the electron density.
Physically, because electron repulsion is larger than the crystal field in weak
field complexes, this repulsion forces some electron density into the e,
orbitals.
7.6 Intermediate field complexes

In the vast majority of transition metal complexes the energies associated
with electron repulsion and the crystal field arc of the same order of
magnitude. This means that neither the weak field limit (electron repulsion»
crystal field) nor the strong field limit (crystal field» electron repulsion) are
met with in practice. There is no separate theory for the intermediate, real
life, region. It is approached from either the strong or the weak field end,
whichever seems the more appropriate. For d 4 -d 7 complexes the choice is

dictated by the distinction used when introducing high and low spin
complexes-by the number of unpaired electrons present on the metal ion.
The qualitative behaviour of the energy levels in the intermediate-field
region can readily be obtained from a knowledge of the energy levels in the
weak and strong field limits. When discussing strong field complexes every
possible term arising from every possible configuration was considered.
However, for weak field complexes only the terms arising from the lowest
free ion term were included in the discussion. To give a complete treatment
of the intermediate field region the weak field treatment must be extended
to include the terms arising from all of the other free ion terms. This
information is given in Table 7.6, which also includes the splittings of those
terms which we have already considered in detail. The terms arising from
d" configurations are listed in Table 7.7.
The relative energies of the components of free-ion terms in a crystal field
may be obtained by the methods which have already been described or by
others which are related to them. Two problems arise. First, as has already
been noted, for a given d" configuration there is usually more than one term
of a given spin multiplicity and symmetry. These different terms interact, the
interaction between them becoming increasingly important as the field
increases. Second, if we wish to show the transition from weak to strong
fields diagrammatically, the diagrams will be rather complicated (and so,
too, would the calculations-this is where a computer comes in handy). For
the d 4 and d 6 configurations there are 43 energy levels the behaviour of
Table 7.6 Crystal field splitting of Russell-Saunders terms

arising from d" configurations in an octahedral crystal field. The
spin multiplicity, not included in this table, is the same for the
crystal field terms as for the parent Russell-Saunders term
Russell-Saunders tenn Crystal field components

5 A,..
p r,..
D Eg + T2g
F A2g + r,.. + T2g
G A,..+ Eg + T:~g + T2g
H Eg + 2T:~g + T2g
I A,..+ A2g + Eg + r,.. + 2T2g
Table 7.7 Terms arising from d" configurations
Configuration Terms•
d1 , d 9 20
d2, da 3F, 3p, 1G, 'D, 15
d3, d7 •F, •p, 2H, 2G, 2p, 220, 2p
d4 , d6 5 0, 3 H, 3 G, 2 3 F, 3 D, 2 3 P, 1 /, 2 1 G, 1 F, 2 1 0, 2 1 5
d5 65, •a, 4F, •o, 4P, 2/, 2H, 2 2G, 2 2F, 3 2D, 2P, 25
a 2 20 means that there are two distinct 2D terms.
which would have to shown in a diagram (combine the relevant data in

Tables 7.6 and 7.7).
In order to use diagrammatic representations of the behaviour of these
energy levels it is necessary to make simplifications. We know which are the
more stable terms in the weak and strong field limits and we will show only
these. If we do not wish to solve the problem of the interactions between
terms exactly, the non-crossing rule gives a qualitative idea of what happens.
It is convenient first to consider the behaviour of the ground and low-lying
terms obtained in the weak-field limit as the crystal field increases, and then
to extend the discussion to include the strong field limit. This is also a
convenient point at which to broaden our discussion to include all d"
configurations.
The behaviour of the T 2 , and E, states derived from the D term of highest
spin multiplicity for the d 1 , d4 , d 6 and d 9 configurations is given in Fig. 7.23.
This figure summarizes the relevant discussion of the two previous sections
and, in particular, the relevant parts of Fig. 7.19. Figure 7.23 is a modified
form of a type of diagram first introduced by Orgel and usually called Orgel
diagrams. It will be noted that some of the g suffixes that have been used in the
discussion so far have been dropped in Fig. 7.23. The reason for this is that,
as will be seen in the next section, by so doing, tetrahedral complexes can
be included in Fig. 7.23. As expected, in the weak field (and, in this case,
also in the strong field), limit the E1,l terms in Fig. 7.23 have a ±~ ~
dependence on~ ( + when we work in terms of holes, and this means the d 4
and d 9 cases, and -when we work with electrons, the d 1 and d 6 cases). The
T21 ,l terms have a ±~ ~ dependence (- for holes, + for electrons).
Because d 4 and d 6 configurations give rise to quintet spin terms but d 1 and
d 9 to doublet, there is no mention of spin terms in Fig. 7.23 so that all of
these cases can be included in the same diagram.
Octahedral Tetrahedral
Fig. 7.23 A modified Orgel diagram for weak
field complexes with free ion 0 terms as ground
state (d\ d4 , d6 and d9 ), I
Tetrahedral Octahedral
I I
e1 (d 1 ) } t2} eg' (d 4 )}
e3 ti(d 6 ) t2~ el (d 9 )
Stro'-n-g----,..--Z-e-ro--,..-----S_.Jtrong
field field field
limit limit
146 I Crystal field theory of transition metal complexes
Tetrahedral Octahedral
Fig. 7.24 A modified Orgel diagram for weak

field complexes with free ion F tenns as ground
state (d 2 , d3 , d 7 and d8 ).
e4 ti (d 7 )} tJ (d 3 ) }
e2 (d 2 ) t2~ ef (d8 ) Octahedral Tetrahedral
L-------~--------~
Strong Zero t.- Strong
field field field
limit limit
The ground and low-lying weak field terms of the d 2 , d 3, d 7 and

d 8 configurations are given in Fig. 7.24. Again, this figure summarizes the
relevant discussion in the previous sections. In particular, the interaction
between the T 1 <•l(P) and T 1 <•l(F) terms has been included and is responsible
for the appearance of curvature on the lines associated with these terms in
the intermediate field region. Just as for the splitting of D terms, so the
splitting ofF terms is inverted for holes (d 3 and d 8 ) compared with electrons
(d 2 and d 7 ). In the former cases this brings the T1 <•l(F) and T1 <ul(P) terms
close together and, although the magnitude of the interaction between these
two terms is the same as for the d 2 and d 7 cases, the effect of the interaction
is much greater and leads to considerable curvature in the Orgel diagram
in the intermediate field region.
The only case not so far included, the d 5 case, is trivial and a diagram is
not given for it (a more complicated one will be given later). A diagram
given at this point would consist of a single horizontal straight line joining
the 6 S free ion term to the t~e 2 strong field configuration (again, the g suffixes
are omitted).
The next step is to include in these diagrams additional terms appropriate
to strong field complexes for those cases where these have different ground-
state spin multiplicities from the corresponding weak field complexes.
Figures 7.25-7.27 include the additions, the last named showing the d 5 case.
It will be noticed that the g suffixes, appropriate to octahedral complexes,
have been reinstated in these figures-and this is because they cannot be
extended to tetrahedral complexes. Figure 7.26 is particularly interesting
because, whereas in the d 3 case a spin quartet (4 A 29 ) remains the ground
state for all values of .1., for the d 7 case the 4 T 1• weak field ground state is
supplanted by a 2 E9 as the crystal field increases. The reasons for this
difference are either contained in or implicit in our earlier discussion and the
Fiji. 7.26 A modified Tanabe-Sugano diagram

for d4 and d6 octahedral complexes.
tilt
Strong Zero Strong
field field field
limit limit
3
eg
Fig. 7.26 A modified Tanabe-Sugano diagram

for d3 and d7 octahedral complexes.
Strong -Ll Zero Ll- Strong

field field field
limit limit
reader will find that explaining it provides a useful check on his or her
understanding.
Figures 7.25-7.27 show in a qualitative fashion the crystal field energy
levels changes which lead to the experimental distinctions between high
and low spin complexes; the close relationship between d" and d 10 -•
configurations being brought out by including them in the same diagram.
For the diagrams relevant to the d4 -d 7 configurations at some point the

lowest level at the weak field limit is replaced as ground state by a term of
lower spin multiplicity with increase in crystal field. This is the point of
transition from a high spin to a low spin complex, although it would not be
correct to regard it also as a point of transition from weak to strong field
't (although it is often convenient to loosely relate the two transitions). As has
been seen, the high spin-low spin transition does not occur at a single point
for a given d" configuration-it is a function of the ligands in a complex.
For a given complex, the crossover point will be determined by an
appropriate balance of crystal field and electron repulsion energies. Mathe·
s l--.r----~~~r--lt]6 e;
1
6
matically, in the crossover region both high and low spin terms would be
treated using an intermediate field approach (which is why the transition is
6A,g
not also between strong and weak fields). The intermediate field approach
is applicable whenever curvature is present in the lines shown in Figs.
7.24-7.27. The fact that curvature, always in one sense, is present on the
lines corresponding to low spin terms in these figures indicates that
they have been drawn in a selective way-other lines, at higher energies,
with opposite curvature have been omitted. Further, as has been seen,
Zero .1- Strong
field field Figs. 7.23-7.27 have a limitation. The scale along the horizontal axis cannot
limit be properly defined. To interpret spectra, in particular, it is convenient to
Fig. 7.27 A modified Tanabe-Sugano diagram have diagrams in which the A scale is accurately linear. In Chapter 8 such
for d5 octahedral complexes.
plots will be introduced-they are called Tanabe-Sugano diagrams. Those
lines which are shown meeting at A = oo in Figs. 7.23-7.27 appear as parallel
lines in the accurate Tanabe-Sugano diagrams because there is no attempt
to include infinity in them!
7. 7 Non-octahedral complexes
So far, this chapter has been concerned entirely with octahedral complexes.
Whilst the majority of complexes are octahedral, almost all of them display
some slight deviation from the ideal geometry. Other complexes have quite
different geometries, as was seen in Chapter 3. For the moment the discussion
will be extended in a more limited way, so as to include only tetrahedral
and square planar complexes.
7.8 Tetrahedral complexes

In tetrahedral complexes the five d orbitals of a transition metal ion are
again split into a set of three (dxy• dyzo and dzxJ, denoted t 2 , and a
set of two (dx' _ Y' and dz,) denoted e. These labels are those obtained by
dropping the g suffix from the labels given to the same orbitals in
the octahedral group, but the reader must be wary, for this pattern does not
always hold. As was seen in Chapter 6 the label t 1 u is used when discussing,
for example, the metal p orbitals in octahedral complexes. This label becomes
t 2 , not t, when carried over to tetrahedral complexes.
That the d orbitals split into t 2 and e sets in tetrahedral complexes may
be seen from Fig. 7.28. As in Chapter 6, the key is to recognize that a
tetrahedron is closely related to a cube (Fig. 7.29). If Cartesian axes are
............................................................................................................................................................................................
Fig. 7.28 Metal d orbitals in a tetrahedral

ligand field.
==
drawn as in Fig. 7.28 then x y z so that dxy• d,, and dzx must be
degenerate (Section 6.3 and Appendix 4). The relative ordering of these two
sets energetically may also be seen from Fig. 7.28. For example, one may
say, loosely, that each lobe of the dx, orbital is half a cube edge away from
each ligand (regarded as a point), but each lobe of the dx'-y' orbital is half
a cube diagonal away. As for the octahedral case, this argument must be
regarded as indicative only. The conclusion that the e set is less destabilized
by the ligand field than the t 2 set (Fig. 7.30)-the opposite sense of
splitting to that which occurs in an octahedral field-is confirmed by
experiment and by detailed calculations. Although the agreement between
experiment and the crystal field calculations is no better than for the
octahedral case, these calculations also show that, for a given metal and
ligand and constant metal-ligand distance, the magnitude of the splitting in
a tetrahedral complex is ~ of that in an octahedral. That is,
A,., = - ~ Aoc,
Experimentally, this relationship has been found to hold remarkably well, a

fact which is perhaps not too surprising in crystal field theory because the
~ factor arises from the geometric relationship of a tetrahedral complex to
an octahedral-it is the squared ratio of the number of ligands: (~) 2 . It
becomes more surprising when, as in the previous chapter, the n bonding
in octahedral and tetrahedral complexes is introduced and compared.
The next step is to obtain crystal field splitting diagrams of the type that
were given in the last section. Surprisingly, we already have them! Remember
that when discussing, for example, the d 8 weak field octahedral case the
energy levels were obtained from those of the d 2 case by talking in terms of
holes rather than electrons. Because holes are attracted by the ligands,
whereas electrons are repelled, for holes the crystal field splitting of the d
orbitals is, effectively, -fl? That is, the d 8 case is obtained from the d 2 by
replacing fl by -fl. Now, the crystal field splitting (for electrons) in
tetrahedral complexes is -~fl ••,. That is, the splitting diagram for a d 2 ion
in a tetrahedral crystal field is qualitatively the same as that for a d 8 ion in
Fig. 7.29 The relationship between a cube and 7 As in previous sections we shall use A without any suffix to denote the octahedral case.
a tetrahedron. Other cases (and the octahedral also, when in the interest of clarity) carry suffixes.
Fig. 7.30 d orbital splitting in a tetrahedral

ligand field. As in Fig. 7.9, a common repulsion
term has been omitted. L'ltet
an octahedral field and accurately so if we include a scale factor of%. It was

for this reason that we dropped some of the g suffixes in Figs. 7.23 and 7.24.
Fig. 7.23 shows the energy levels of d I, d 4 , d 6 and d 9 weak field complexes
in both octahedral and tetrahedral ligand fields. Similarly, Fig. 7.24 shows,
qualitatively, the energy levels of d 2 , d 3 , d 7 and d 8 weak field complexes in
both of these fields. The reader should easily be able to show that d 3-d 6
ions would be expected to form weak and strong field tetrahedral complexes,
differing in the spin multiplicity of their ground states. Strong field tetra-
hedral complexes are very, very rare so they will not be discussed further.
In particular, no attempt has been made to include tetrahedral complexes
in Figs. 7.24-7.27. A distinction between Orgel and Tanabe-Sugano diagrams
is that the former include both tetrahedral and octahedral cases whilst the
latter refer to octahedral complexes only.
7.9 Square planar complexes

There are two conceptually different approaches to square planar complexes.
One may regard the d orbital splitting as being that obtained when two trans
ligands are simultaneously removed from an octahedral complex (conven-
tionally, along the z axis). Alternatively, one may use the same approach as
z that used for octahedral and tetrahedral complexes and consider the splitting
of the free ion d orbitals by a square planar ligand field. Since, in a square
planar complex, x = y (¥;z) (Fig. 7.31) with the ligands in the xy plane, the
dzx and dyz orbitals will be more stable than dxy• leading to the qualitative
d orbital splitting pattern shown in Fig. 7.32. If, as shown in Fig. 7.32, one
regards a square planar complex as derived from an octahedral, then there
is no evident reason to expect any significant change in energy levels of d
orbitals largely localized in the xy plane. That is, dxy and dx' _ y' are essentially
unchanged in energy. This is what is shown in Fig. 7.32 but a disadvantage
must be recognized. This is that the centre of gravity of the split energy levels
no longer coincides with the free-ion level. This defect is corrected in
Fig. 7.31 The usual choice of coordinate axes
in a square planar complex.
Fig. 7.33, but Fig. 7.33 has its own problems: it might easily lead one to
Square planar complexes I 151
Fig. 7.32 Correlation of the d orbital splitting

in a square planar ligand field with those in an
octahedral ligand field.
Free ion
eg (dyz• dzx)
Octahedral crystal field Square planar crystal field
Fig. 7.33 Modified correlation between

octahedral and square planar crystal field
splittings.
Free ion
eg
Octahedral crystal field Square planar crystal field
conclude that in square planar complexes dxy and dx'- ,., suffer an extra
repulsion. This is but a minor disadvantage and Fig. 7.33 is the one generally
encountered. Note that whilst b. may be regarded as the energy difference
between the centres of gravity of the upper pair and lower trio of levels, two
additional splitting energies are needed to give a complete specification of
the energy level diagram for square planar complexes. These arc the splittings
of the b 1 ,-a 1 " orbitals and of the b 29 -e 9 . There is no evident connection
between these energies because the consequence of removing the ligands on
the z axis must be expected to be different for the d=' and the dxz and dyz
Table 7.8 Correlation between symmetry labels used for octa-

hedral and square planar geometries. Although g and u suffixes
are not given and have to be added, the expressions in brackets
indicate how the various d orbitals transform. Thus, dz2 has A:~,g
symmetry in a square planar complex
Octahedral Square planar
Al Al
A2 81
E A1<z"> + a1<x>- >">
rl A2 +E
T2 B2 (xy) + E(zx, yz)
orbitals. Not surprisingly, theoretical efforts have been made to relate the
two splittings and thus make the problem of interpreting the d-d electronic
spectra of square planar complexes more tractable.
Note that the symmetry labels used for square planar complexes are
different from those used for octahedral complexes. The correlation between
the two sets is shown in Table 7.8. In this table are included some symmetry
labels which have not yet been used in the discussion, anticipating a need
for them in later chapters. The g and u suffixes have been omitted in both
geometries in order to keep the table compact. (The rule, of course, is g ..... g
and u --> u.) There is some freedom about the choice of the B1 and B2 labels
in square planar complexes in that they may be interchanged. What one
author called B 1 another may call B2 • In this book the choice shown in Table
7.8 is used. Table 7.8 may be used for either orbitals or terms. For example,
the 2T 2 • term of a d 1 configuration in an octahedral ligand field splits into
two, 2 B2 • and 2 E., if the ligand field is reduced to square planar. Of these,
the 2 E• is more stable because it corresponds to the single d electron
occupying the d.. and d,, orbitals (which are more stable than dx,, cf.
Fig. 7.33).
7.10 Other stereochemistries

The d orbital splitting patterns in low-symmetry geometries other than those
which have been discussed are most readily obtained in the way outlined for
square planar complexes. The results for two important geometries are given
in Table 7.9. However, care must be exercised in using this table. First, as
for square planar complexes, some choice exists in the allocation of symmetry
labels for digonally distorted octahedral complexes (Fig. 7.34). Secondly, the
choice of z axis for an octahedral complex (along a fourfold rotation axis)
is not consistent with the usual choice for trigonal distorted octahedral
complexes, for which the threefold axis is chosen as z axis (Fig. 7.35). This
means that, for example, the d,, orbital in the latter geometry is not the same
d,, as that in the corresponding octahedral complex, but turns out to be a
mixture of the octahedral dx,, d,, and dzx orbitals. This perhaps seems rather
strange but, in fact, illustrates a rather general point which will be more fully
dealt with in Chapter 11. There it will be found that two complete sets off
orbitals exist, the shapes of the orbitals in one set not all being matched in
Ligand field theory 1 153
Table 7.9 Correlation between symmetry labels used for true octahedral, trigonally
distorted octahedral and digonally distorted octahedral geometries, As for Table 7.8
g and u suffixes are omitted but must be added. So, A1 (xy) and A(i') mean,
respectively, that dxy transforms as A:~g (trigonally distorted octahedron) and dz'
transforms as Ag (digonal distortion). For digonally distorted complexes the labels
Bv 8 2 and 8 3 (with or without g or u suffixes) may be used differently by different
workers, although those with g suffixes are never interchanged with those with u
suffixes
Octahedral Trigonally distorted octahedral Digonally distorted octahedral

field field
Al Al A
A2 A2 A
E E A(z 2 ) + A(x' - y")
T1 A2 + E B 1 (xy) + B2(yz) + 83(zx)
the other. Which set is used depends on the molecular geometry, just as
different d orbital sets are used for octahedral and trigonally distorted
octahedral geometries.
One final point: just as we found that the number of unpaired electrons
in an octahedral complex can vary with the crystal field, so, too, in other
geometries. However the effects tend to be more subtle and difficult to
disentangle. Thus, for five-coordinate complexes they are interwoven with
the trigonal bipyramidal and square pyramidal structural possibilities. 8
7.11 Ligand field theory

This chapter is concluded by returning again to octahedral transition metal
complexes. In Chapter 6 the molecular orbital method was found to give an
account of the bonding in these complexes which gave good general
agreement with the observations except at points where the independent
electron model is deficient. In the present chapter crystal field theory had
rather little to say about bonding but showed considerable value in modelling
the behaviour of d electrons in complexes. In that they focus on rather
different aspects, could they not be combined into a single, unified, theory?
They can and have been. This theory, called ligand field theory, differs from
simple crystal field theory in that allowance is made for the existence of
covalency. This is done by dropping a requirement inherent in simple crystal
c:+7
field theory: that when a quantity is available from spectral data on free
ions-an electron repulsion energy, for instance-it should used in crystal
field calculations. Covalency means that the orbitals occupied by the d
r---'
electrons in a complex must be expected to differ from those of the free ion
(b~ -- -
and so, too, therefore should the value of electron repulsion and other
energies. In practice, this means that such quantities become parameters in
the theory, given those values which produce the best agreement with
FJg. 7.34 A digonally distorted octahedral
complex: (a) an octahedral complex and a experiment. These additional parameters will be met in the next two chapters.
distortion which is symmetric with respect to The magnitudes of the additional parameters are determined experimentally,
the c2 axis indicated; (d) the distorted molecule
with the original octahedron indicated dotted. 8 SeeR. R. Holmes, J. Amer. Chern. Soc. (1984) 106, 3745, for a detailed discussion.
Fig. 7.35 A trigonally distorted octahedral

complex: (a) an octahedral complex and a
distortion which is symmetric with respect to
the c3 axis indicated; (b) the distorted molecule
with the original octahedron shown dotted; (c)
an alternative view of the trigonally distorted
octahedron.
(a) (c)
but the number of independent observables is usually less than the number
of parameters. It cannot. therefore, be asserted that ligand field theory, is in
general, proven. One may say only that it provides a consistent explanation
for experimental data using parameters which, almost invariably, have
physically reasonable values. Not surprisingly, there have been attempts to
devise cunning experiments which provide additional information so that we
end up with more data than parameters. For instance, instead of studying
complexes randomly orientated in solution one might look at molecules fixed
in a single crystal of known structure. The orientation of the crystal, and
with it any inherent molecular anisotropy (such as a tetragonal distortion),
could be varied relative to the direction of polarization of polarized incident
light, for example, and so give additional data. However, octahedral and
tetrahedral complexes are isotropic (x = y = z) so this method is only
applicable to lower symmetry molecules requiring even more parameters to
start with, as was seen in Section 6.9. Suitable cases for study are already
bad ones! Again, instead of working at one temperature, we could study a
property- the magnetic properties of a complex for instance-as a function
of temperature. The varying thermal population of low-lying energy levels (a
magnetic field frequently induces suitable small splittings, as we shall see in
Chapter 9) then gives information on the energy separation between these
levels. Unfortunately, at the very low temperatures that really lead to the
depopulation of levels that are only slightly split apart, and so offer the
greatest potential for additional useful information, other weak phenomena-
such as the freezing out of lattice vibrations-have to be considered and this
requires yet more parameters! We seem to be hitting a brick wall at every
turn. This situation has inspired a search for theoretical models which might
reveal interrelationships between the various parameters of ligand field
theory whilst themselves involving relatively few, if any, additional parameters.
Many such models have been investigated, including those that we will
meet in Chapter 10. Entire books have been written on the subject. 9 Suffice
to say that the search continues. Although clear progress has been made, no
single model is currently accepted as providing a generally applicable
method.
We conclude by reconsidering what we must now call the ligand field
splitting parameter of an octahedral transition metal complex, Ll, since the
factors affecting its magnitude are evidently more complicated than we at
first supposed. We have encountered three such factors:
9 See, for example, M. Gerlach and R. S. Slade Ligand-Field Parameters, Cambridge
University Press, Cambridge, 1973.
Questions 1 155
• First, the magnitude of L1 depends on the electrostatic field generated by the

ligands-the crystal field.
• Secondly, it depends on ligand-metal rr bonding because the energy of the
metal e, orbitals depends on this.
• Thirdly, it depends on the ligand-metal n bonding, since this affects energy
of the metal t 2 , set.
This does not exhaust the list of factors influencing the magnitude of t1, but
there are believed to be no others of comparable importance. A recitation
of most of the evidence that supports the ligand field model in preference to
the crystal field model is deferred until Section 12.1.
Further reading 1970 and, more simply, in Chapter 12 of The Chemical Bond
by the same authors, J. Wiley, Chichester, 1985.
In practice it is not possible to separate the further reading An excellent introduction is given in 'Ligand Field Theory'
relevant to this chapter from that appropriate to Chapter 6. by J. S. Griffiths and L. E. Orgel, Quarterly Rev., Chern. Soc.
The contents of the two chapters go together. In addition to (London) (1958) 11, 381. This particular article was perhaps
the references given at the end of Chapter 6, two others which more responsible than any other for the introduction of the
follow the pattern of the present chapter but with a more ideas of crystal and ligand field theories into inorganic chem-
mathematical approach are Chapter 13 of Valence Theory by istry.
J. N. Murrell, S. F. A. Kettle and J. Tedder, Wiley, London,
Questions differences will be reflected in diagrams such as that in Fig.

7.27).
7.1 Outline the difference(s) between members of the follow-
ing pairs: 7.5 With as little reference to the text as possible, rehearse the
arguments that lead to Fig. 7.20.
1. symmetry elements and symmetry operations;
7.6 With as little reference to the text as possible, rehearse
2. tL and d 9 ;
the arguments that lead from Fig. 7.20 to Fig. 7.21.
3. a p orbital and a P term;
4. T 1 , and 3 T1 ,. 7.7 A tetrahedron and a cube are closely related. Indeed, as
Fig. 7.29 shows, a cube may be regarded as composed of two
tetrahedra. Suggest, qualitatively, the form of the relationship
7.2 Draw diagrams showing the splitting of the following that exists between .1.,,. and .1-,.b;, for complexes involving the
orbital sets in an octahedral crystal field: s, p, d and f. Make same metal and ligands. The present chapter contains data
an educated guess at the likely pattern of splitting of a set of which suggest a probable quantitative relationship. What is
g orbitals (the answers to all of these may be deduced from it?
Table 7.6).
7.8 You have just astonished the scientific world by your
7.3 Write brief notes on: discovery of a class of monodentate ligands that form cubic
1. the d 5 high and low spin configurations in an octahedral complexes, [ML 8] 2 +,with all of the ions of the first transition
ligand field; series. It is clear that, depending on the choice of L, some metals
2. pairing energy; may form either high spin or low spin complexes. For which
3. the spectrochemical series. ions of the first transition series does the possibility of high and
low spin cubic complexes occur?
7.4 Starting with (a) the weak field and then (b) the strong
field octahedral d 5 configurations, write down the excited state 7.9 Enumerate and briefly explain those factors which are
configurations which may be derived from them. Comment on believed to make a significant contribution to .1., the ligand
the consequences of the differences between the two lists (these field splitting parameter of an octahedral complex.
Electronic spectra of
8.1 Introduction
In this chapter our concern will be with the electronic spectra of transition
metal complexes, particularly those of the first transition series. Although
nowadays there are fewer studies of these spectra, per se, than previously,
their study is essential if the electronic excited states of complexes are to be
understood. The energy required for the promotion of an electron from one
orbital to another or, more precisely, the excitation of a molecule from its
electronic ground state to an electronic excited state, corresponds to
absorption of light in the near-infrared, visible or ultraviolet regions of the
spectrum. For transition metal complexes the absorption bands in the first
two of these regions are relatively weak and are associated with transitions
largely localized on the metal atom. The ultraviolet bands are intense. They
are associated with the transfer of an electron from one atom to another and
so are called charge-transfer bands. These bands are responsible for the
colour changes associated with indicators for inorganic cations, such as the
thiocyanate test for Fe 111 or the indicators used in EDTA (compleximetric)
titrations, but in these cases the charge-transfer bands fall in the visible region
of the spectrum. The intense bands will be dealt with in the latter part of
this chapter; for the moment our concern will be with the weaker bands.
Whilst these are most simply explained by crystal field theory, a detailed
comparison of the data with the theoretical predictions will show that ligand
field theory provides a more appropriate explanation.
When spectral bands are weak there is a reason. Often it means that they
are bands which are forbidden but- obviously- not totally forbidden. The
weak bands with which we are concerned are of this type. Electronic
transitions which correspond to strong bands are electric dipole in type.
Classically, the electric vector associated with the incident light beam behaves
like a pair of alternating + and - charges across the molecule. These
oscillating charges induce an oscillating dipole in the molecule; when the
Electronic spectra of V 111 and Ni 11 complexes 1 157
frequency of the molecular oscillating dipole corresponds to a natural

frequency ofthe molecule, resonance occurs and the molecule acquires energy
at the expense of the light wave. Clearly, at the heart of this explanation is
"uc: an electric dipole-hence the phrase electric dipole allowed. In an atom, the
e"' electric dipole-allowed transitions are those associated with transitions
g between the orbitals s .-. p, p .-. d and d.-. f (where the double-headed
~ arrows indicate that the transition is allowed in either direction). 1 Forbidden
ares.-. d, p .-. f, s .-. s, p .-. p, d <-+ d etc. The weak bands observed in the
visible region for transition metal complexes are d <-> d in type, and so,
formally forbidden but made slightly allowed by mechanisms which will be
discussed later. First, we need an explanation for the number of bands and
Fig. 8.1 The d~d spectrum of the [V(H 2 0).,] 3 + their relative energies. We start with the crystal field model and later
cation (d2 ).
consider the changes needed to accommodate the ligand field approach.
As an example, consider the pale violet coloured ion [V(H 2 0) 6 ]3+ which
shows two weak absorptions in the visible. One is in the blue at ca. 3900 A
and the other in the yellow at ca. 5800 A. The spectrum of this d 2 ion is
shown in Fig. 8.1. The explanation provided by crystal field theory for the
weak transitions is presented in diagrams such as Fig. 8.2, which shows the
effect of an octahedral crystal field on the energy levels of a d 2 ion such as
3p ~-~---r-r vm. This energy level diagram was developed in a qualitative fashion in
Chapter 7 (see Fig. 7.21) but for the purposes of the present chapter, a form
'F
in which the energy scale is well defined is needed and obtaining this
is our first task. The ground state is 3 T 19(F) and spin-allowed transitions-
those in which the spin is the same in both ground and excited states-may
occur to the 3 T29 , 3 T 1 g(P) and 3 A 29 excited states, this being the usual order
~- of increasing energy. The origin of these predictions is shown by the arrows
FJg. 8.2 The part of the d2 Tanabe~Sugano in Fig. 8.2 (the triple-headed arrow is that relevant; those to the right explain
diagram needed for the interpretation of the
spectrum in Fig. 8.1.
it). As is evident from Fig. 8.1, experimentally two transitions are observed.
Calculations (see below) indicate that the third, high energy, band is usually
obscured by an intense charge-transfer band in this region. Later, it will
transpire that this third band should, in any case, be particularly weak and
broad. It thus appears that we can account, qualitatively, for the spectrum.
Whether we can do so quantitatively is discussed in the next section.
8.2 Electronic spectra of V111 and Ni 11 complexes

Before a quantitative assessment of the applicability of the crystal field model
is possible, explicit expressions for the energies of the excited states of the
V111 ion relative to that of the ground state are needed. These energies depend
on three quantities: the crystal field splitting parameter ~. the magnitude of
the 3 F- 3 P separation in the free ion, and the magnitude of interaction
between the two 3T 19 terms. The second of these quantities is usually denoted
15B where B is a sum of electron repulsion integrals which it is convenient
not to evaluate explicitly and which is referred to as a Racah parameter.
Other Racah parameters, also composed of electron repulsion integrals,
which the reader may encounter are denoted A and C. The third of the
quantities mentioned above, the energy of interaction of the two 3T 19 terms,
is, fortunately, a function of ~ and B only, so that these two quantities are
1 Strictly, of course, we should work with terms rather than orbitals.
158 1 Electronic spectra of transition metal complexes
all that is needed to interpret the spectra of octahedral V111 complexes. 15B
is the separation of the 3 F and 3 P free ion terms, so it should be possible
to obtain the value of B from atomic spectral data. This corresponds to the
pure crystal field approach, but, significantly, agreement between experiment
and theory for complexes is only generally obtained if B is treated as a
parameter, the value of which may be varied. It turns out that B has to be
given a value which is rather smaller than that obtained from atomic spectra.
This parameterization of B is part of the ligand field method and is consistent
with our ideas of the consequences of covalency in transition metal complexes.
Covalency implies
1. that the metal electrons will be partially delocalized onto the ligands;
2. that the effective positive charge on the transition metal will be smaller
than in the free ion.
Both of these effects mean that the metal electron cloud will be more diffuse
in the complex than in the free ion, and repulsion between the electrons
making up this cloud is therefore reduced. This conclusion is confirmed by
a more detailed analysis.
The energies of the crystal field spin triplet terms of the d 2 configuration,
relative to the 3 F free-ion term as zero, are given explicitly below. It is beyond
the scope of the present text to derive them so we give these expressions
without proof;2 however, the reader may readily check that they reduce to
those given in Section 6.5 for the weak field limit 3 !1--+ 0 (expand the square
root, using the binomial theorem in the form (y 2 + xy) 112 ::::: y + x/2). These
expressions also give the strong field energies; note that the 15B term was
not included in the energies given in Section 6.5, not that this is important
because in the strong field limit the effects of electron repulsion are negligible
in comparison with the ligand field. In Chapter 7 this situation was
accommodated by making the ligand field infinite but in the present context
it is simpler to set B = 0. The square root expressions are those which allow
for the mixing of the 3 T 1.(F) and 3 T 1,(P) terms by the crystal field. Indeed,
it should be evident from the form of these two energy level expressions that
they arc obtained as the roots of a quadratic equation. The energies of the
3T
19 and A 29 terms are those which were obtained in Chapter 7.
3
Term Energy
~[158- ~ Ll- (2258 2 + 188!1 + !1 2 ) 112 ]

g!l
~[158 - ; Ll + (2258 2 + 188!1 + !12 ) 112 ]
~Ll
It follows that the two observed bands, assigned to the 3 T 29 +- 3 T 19 (F) 4
2 A derivation is given in J. N. Murrell, S. F. A. Kettle and J. M. Tedder, V a/ence Theory,

Wiley, London and New York, 1956, Chapter 13. A simpler derivation is given by the same
authors in The Chemical Bond, Wiley, Chichester and New York, 1985, Chapter 12.
3 Strictly, of course, the weak field limit is ~ = 0. However, substituting this value simply
gives the 3 F~ 3 P separation in the free ion, 15B.
4 Here we follow the usual spectroscopic practice of giving the exc1ted state first.
Electronic spectra of V 111 and Ni 11 complexes 1 159
and 3 T 1.(P) <-- 3 T 1.(F) transitions, have energies given by the differences
between the two relevant expressions above:
70
Transition Energy
60
3 T2g +-- 3T,fF) ~[!'. - 158 + (2258 2 + 1888 + 82)112]
15 3T11(P) +-- 3T11(F) (2258 2 + 188 8 + 8 2)112
50
3T2g
We have experimental transition energies to fit these expressions, so we can
L..--""7- 1A1B
now obtain values for A and B. This task is not difficult. Divide the two
expressions above by B. We then have
30
E( 3 T2B +-- 3 T:,fF))
1G -1 [A
- - 15 + ( 225 + 18-
A + -A2)112] (8.1)
20 8 2 8 8 82
E( 3 T11(P) +-- 3 T1g(F)) A A2)112
( 225 + 188+ 82 (8.2)
8
The right-hand side expressions are functions of A/B. Similarly, the energies
of the terms themselves, divided by B, may be expressed as functions of
10 20 30
!'./8
A/B. Tanabe and Sugano have published diagrams which show these
relationships; they take as their energy zero the energy of the ground
FJC. 8.3 A complete Tanabe-Sugano diagram
for octahedral d2 complexes. state, so that, for the d 2 case, they effectively plot the functions
and
against A/B. It is to be emphasized that the ground state is coincident with

the abscissa in their diagrams. If a change in ground state occurs-and this
can happen for d 4 -d 7 ions, where a high spin/low spin distinction exists-
there is a discontinuity in the diagram. Figure 8.3 is a Tanabe-Sugano
diagram for the octahedral d 2 case. A complete set of Tanabe-Sugano
diagrams is given in Appendix 7; similar sets appear in many textbooks
of inorganic chemistry. It may well be found helpful to compare these, real,
Tanabe-Sugano diagrams with the modified versions introduced in Chapter
7. We shall return to the Tanabe-Sugano diagrams later in this chapter and
the discussion there should explain any difficulties encountered when making
the comparison just suggested.
Consider as a specific example the [V(H 2 0) 6 ]3+ ion, which has the
spectrum shown in Fig. 8.1; it has weak peaks at ca. 17 200 em - 1 (ca. 5800 A)
and 25 600 em - 1 (ca. 3900 A). We conclude that E('T 1.(P) <-- 3 T 1.(F)) =
17 200 em- 1 and that E('T1.(P) <-- 3 T 1.(F)) = 25 600 em - 1. Equations 8.1
and 8.2-and therefore Fig. 8.1-can be used provided that the Bs appearing
on the left-hand side of each equation are eliminated. This can be done if
0.8
:=:i~
~~~ 0.6
~it
Fig. 8.4 A plot of eqn 8.3, useful in the
interpretation of the spectra of octahedral d2 ~;.:1:. 0.4
~~~
complexes.
""'"
Cui~ 0.2
ar
0
eqn 8.1 is divided by eqn 8.2 to give
L'> ( L'> t,2)1/2

E!'T2g <- 3T1g(F))
8 - 15 + 225 + 18 8 + 82
(8.3)
E(3 T1g(P) <- 3 T1g(F))
2(225 + 18~+f,t
2
In the case of [V(H 2 0) 6 ] 3 + the value of this quotient is 17 200/25 600 = 0.67.
We can now proceed in one of two ways-and a third will immediately occur
to the computer enthusiast; it is not difficult to write a suitable program to
do the work. We could apply a trial and error process to Fig. 8.3 until we
find the correct value of !3./B (in our case tJ.jB = 28, shown dotted in Fig.
8.3). Alternatively, we could plot the right-hand side of eqn 8.3 against tJ.jB.
Such plots are useful when a large number of data have to be analysed. That
appropriate to the d 2 configuration is shown in Fig. 8.4 where, again, the
[V(H 2 0) 6 ]3+ case is indicated by dotted lines. Having determined tJ.jB from
Fig. 8.4 (again, of course, obtaining a value of 28), we return to Fig. 8.3 and
use this ratio to obtain E{'T1.(P) <-- 3 T 19(F))/B and E('T 1.(P) <-- 3 T 19 (F))/B.
Alternatively, and more accurately, we could substitute for !3./B in eqns 8.1
and 8.2. The right-hand sides of these equations are then found to have
values of 25.9 and 38.6, respectively so, using the experimental transition
energies, both give B=665cm- 1 • Since tJ./B=28 it follows that 11=
18 600 em - 1 . This analysis follows the ligand field approach, with B regarded
as a parameter.
In the crystal field approach we use the free-ion value of B, 860 em - 1 . It
follows that E('T 19 (P) <-- 3T 19 (F))/B = 17 200/860 = 20 and E('T 19 (P) <-
3T10(F))/B = 25 600/860 = 29.8. Using Fig. 8.3 these lead to values of tJ.jB
of 22.5 and 18.0 respectively; that is, tJ. values of 19400 and 15 500 em - 1 •
This internal inconsistency does not occur in ligand field theory. However,
it is only removed by introducing an additional parameter, and, consequently,
there are no additional data with which to test the theory, for we have used
two experimental quantities (transition energies) to define two parameters
tJ. and B. Only if the 3 T 29 <-- 3 T 19(F) were to be observed could we test the
theory. Using the ligand field values for tJ. and B obtained above, together
with the expression for the transition energy obtained from the term energies
given earlier, this transition is predicted to be at ca. 36 000 em- 1 , where it
is obscured by strong charge-transfer bands. However, a weak band has been
observed in this region in closely related complexes.
Electronic spectra of v"' and Ni 11 complexes 1 161
1lc
Fig. 8.5 The d-d spectrum of the -e"'
0
[Ni(H 2 0) 6 j>+ cation (d 8 ). <J)
.c
<(
30000 20000 10000

cm-1
60
50
40
Fig. 8.6 The part of the d8 Tanabe-Sugano
diagram believed to be relevant for the en
interpretation of the spectrum in Fig. 8.5.
"' 30
'
~--~~----~----~------''A2g
10 20 30
fl/B-
As a further example of the interpretation of d-d spectra consider the case

of the d-d ion Ni" in the complex [Ni(H 2 0) 6 ] 2 +, the spectrum of which is
shown in Fig. 8.5. The three bands, at 8500 em- 1, ca. 14 100 em- 1 and
25 300 em - 1 are assigned, respectively, to the 3 T 29 +- 3 T 2 ., 3 T 19(F) +- 3 A 29
and 3T 1.(P) +- 3 A 29 transitions (see the simplified Tanabe-Sugano diagram
in Fig. 8.6). Of these, the central band shows some structure which will be
discussed later in this chapter. The energy level expressions are derived
immediately from those given earlier for d 2 , except that now the 3 A 29 level
is the ground state (which means that we simply have to change the sign of
11). It follows that the lowest energy transition has an energy 11. So, we take
11 = 8500 em-'- The next transition, to the 3 T 19(F) term, has an energy of
H158 + 31'. - (2258 2 - 1881'. + t.2) 112]

(remember, the sign of !1 has to be changed, here we are dealing with two
holes). In the crystal field model the free-ion value of B, 1082 em - 1 is
inserted, to obtain a second value of !1. This can be done by dividing
throughout by B, to obtain
14 100
- - =-
1[ t. ( t. t.
15 + 3- - 225 - 18- +- 2)1/2]
1082 2 8 8 82
and solving for !1/B, thus obtaining !1/B = 7.71. With the free-ion value of
B (the crystal field model) this gives !1 = 8340, a bit different from the
previous value. In the ligand field model we insert !1 = 8500 and solve for B
(the B 2 terms disappear) to obtain B = 900 em - 1 .
In this particular example there was a third band to consider. Really, this
should be set on a par with the other two to obtain values for !1 and B,
particularly on the ligand field model. Instead, we will use our calculated
ligand field values of !1 and B to predict its position-for this will provide
a test of the ligand field model. The energy ofthe 3 T 19 (P) +- 3 A 29 transition is
H158 + 3t. + (22582- 188t. + t. 2J112J

which, inserting B = 900 em- 1 and !1 = 8500 em- 1 predicts that the third band
will be at 24 900 em - 1 . In contrast, the crystal field model (B = 1082 em - 1 ,
!1 = 8500 em- 1 ) predicts 27 400 em- 1 or, with !1 = 8340 em- 1 , the transition
is predicted to be at 27 I 00 em- 1 • Clearly, the ligand field theory is to be
preferred.
As this example shows, those cases in which there are sufficient data to
provide a test of ligand field theory show that it leads to reasonable
agreement between observed and calculated band positions. It will be
noted that the discussion has been confined to spin-allowed bands (both
ground and excited states were spin triplets). Other, extremely weak bands
occur and have been assigned to spin-forbidden transitions (for example
1T2 • +- 3 T .(F)).
1
The spin-allowed spectral bands of all other cases, weak field and strong
field alike, may be analysed in a similar way to that given above. Tanabe
and Sugano have given diagrams for all other d" cases (except d 1 and d 9
which are trivial, having only one excited state at an energy !1 above the
ground state) and a complete set is given in Appendix 7. Alongside each
diagram in this Appendix is an example of a spectrum to which it refers. The
reader who glances forward to this Appendix at this point will find that there
are quite often complicating features which have not so far been discussed.
Most will be covered in later in this chapter. As has been demonstrated,
both !1 and B may be obtained from Tanabe-Sugano diagrams with
sufficient accuracy for most purposes. The physical reasonableness of the
results obtained may be assessed by reference to the spectrochemical series
(for !1) and the free-ion value of B. The latter are given in Table 8.1 for
ions of the first-row transition elements. It is assumed that an incorrect
assignment of the spectral bands will lead either to an incompatibility
with the Tanabe-Sugano diagram or to physically unreasonable values of
!1 and B5
5 Of particular interest to the computer enthusiast are general expressions for A and B.
See Y. Do, J. Chern. Ed. (1990) 67 134.

Spin-forbidden transitions 1 163
Table 8.1 Free-ion values of the electron repulsion parameter B for first row
transition metal elements with d" configurations, measured in em~ 1 . A blank
indicates that the value is not known; a dash indicates that the configuration has
either one or no electrons outside a closed shell
Metal atom charge Metal atom
Ti v Cr Mn Fe Co Ni Cu
0 560 579 790 720 805 789 1025

+1 681 660 710 872 870 879 1038 1218
+2 719 765 830 960 1059 1117 1082 1239
+3 860 1030 1140
+4 1040 1144
It is convenient at this point to complete our discussion of the reduction

in the free-ion value of the Racah parameter B which occurs on complex
formation. This, it will be recalled, appears to be a consequence of the
metal electrons being delocalized over a larger volume of space in the
complex than in the free ion. It has proved possible to arrange ligands in a
series such that, for a given metal ion, the B value required to fit the spectra of
the [ML 6 ]"+ ions decrease down the series
F~ > H2 0 > NH 3 >en> -NCS~ > Cl~ :>: CW > Br~ > S 2 ~ :>:I~
This series has been called the nephelauxetic series (nephelauxetic = cloud
expanding). There are exceptions, but this order holds quite well. It should
be noted that what are probably the most polarizable ligands give the
lowest B values and vice versa. Another way of looking at the list is to
recognize that hard bases are to the left and that soft are to the right (see
Section 5.4). A similar list exists for metal ions but it is difficult to attach
any clear interpretation to it, although some regularities may be teased out.
Such a list, with B decreasing left to right relative to the free ion values, is
Mn 11 "' V" > Ni 11 :>: Co" > Mo111 > Cr'" > Fe111 > Rh 111 :>: lr111 > Co 111
> Mn'v > Pt'v > Ptv'
8.3 Spin-forbidden transitions

As mentioned above, extremely weak bands, assigned to spin-forbidden
transitions, may sometimes be observed in the spectra of transition metal
ions. Indeed, for the d 5 high spin case (Mn11 and some Fe111 complexes),
the ground state is the only sextet spin term (6 A 1.) and so all the observed
d-d transitions are spin-forbidden. Evidently, the spin selection rule, like so
many others, is not absolute. Similarly, most of this chapter is concerned
with transitions which, in large measure, are d-d transitions and therefore
forbidden (they are often said to be Laporte forbidden). Later we shall return
to the problem of why these forbidden transitions occur.
The problem of accounting quantitatively for the spin-forbidden bands is
rather more difficult than carrying out calculations for their spin-allowed
counterparts. Not surprisingly, a new parameter has to be introduced. This is
the Racah parameter C, another electron repulsion parameter, which like B,

is expected to be numerically smaller than its free-ion value. Jf the treatment
given above for the d 2 case is followed, C appears in the final expressions for
(transition energy)/B as the quotient C/B. Before a Tanabe-Sugano diagram
which includes terms of all multiplicities can be constructed it is necessary
to give a numerical value to C/B (or, as it is sometimes called, y). For different
values of C/B the Tanabe-Sugano diagram will be different. However, this
parameter only affects the interscaling between terms of different spin
multiplicities-so that, for instance, the relative energies of the 3 A 2 ., 3 T 1.(P),
3T 2 • and 3 T 1.(F) terms of the d 2 configuration are independent of C/B. 6 The
Tanabe-Sugano diagrams which are commonly encountered are plotted

for C/B ~ 4.5. This value is rather larger than the free-ion ratio which, for
the first row transition elements, has an average value of ca 4.0 (but
which varies from 3.2 to 4.8). Before the availability of suitable computer
programs it was not an uncommon practice to assign spin-forbidden bands
using whatever Tanabe-Sugano diagram was to hand. When the job
has been done properly it has been found that it is possible to obtain excellent
agreement between theory and experiment. In the computer-based approach
it is not generally possible to work with relatively simple equations of the
type that were used above for the d 2 and d 8 cases. It is necessary to go back
to the beginning and to set up the problem in a way that is rather different
from the presentation given in the present text.
8.4 Effect of spin-orbit coupling

The spin-allowed d-d bands which dominate the visible spectrum of
complexes of many transition metal ions are rather broad, with half-widths
of ca. 3000 em- 1 . This means that if there is some interaction within the
system which has been overlooked, its presence will not be apparent from
the electronic spectrum unless it causes splittings in the energy levels of
perhaps 1000 em - I In fact, two such effects have been omitted-spin-orbit
coupling and the Jahn-Teller effect.
When developing crystal and ligand field theories, it was assumed that
the orbital properties of the electrons are something quite separate from their
spin. This separation is reflected in the symbolism; in the symbol 3 F for
example, the 3 refers to the spin multiplicity and the F to the orbital
multiplicity. Now, the electron has an intrinsic magnetic moment-that is,
it behaves like a tiny bar magnet-the spin of the electron describes the same
phenomenon. These tiny bar magnets, spins, orientate themselves in a
magnetic field, a phenomenon which makes relevant the magnetic measure-
ments described in the next chapter. Similarly, there is an orbital-derived
magnetic moment for all but S terms; this orbital moment may be likened
to the magnetic moment of a solenoid, the circulation of the electron around
the nucleus being akin to the circulation of a current through the turns of a
6 The general development in this and the preceding chapter follow the traditional pattern
and are based on the so-called Slater-Condon-Shortley model for the description of the energy
differences between terms arising from the same configuration. Fundamental objections have
recently been raised to certain features of this model but the general picture that it presents
survives, although, for example, some detailed features that it attributes to electron repulsion
may actually arise from nuclear attraction!
Effect of spin-orbit coupling 1 165
solenoid. We see that the separation of the spin and orbital properties of the
electrons is equivalent to assuming that two magnets, spin-derived and
orbital-derived, do not interact. Of course, they interact, i.e. couple, to some
extent and this is reflected in the phenomenon of spin-orbit coupling. The
formal theory of spin-orbit coupling treats the phenomenon as the coupling
of two angular momenta rather than bar magnets. We shall have to move
in this more formal direction in our discussion of the properties off electron
systems (Chapter 11), although we shall there attempt to compensate by
introducing pictorial representations of spin-orbit functions. The conse-
quences of spin-orbit coupling for d electron systems are discussed in more
detail in the next chapter. At this point it is sufficient to simply state that
spin-orbit coupling causes splitting of some of the degeneracies implicit in
the orbital energy level diagrams encountered so far. For example, a 4>f19
term is 12-fold degenerate (4 x 3). This splits into three sublevels as a
consequence of spin-orbit coupling because, loosely, some arrangements of
spin and orbital magnets are energetically more stable than others. The
effective number of spin-orbit components of crystal field terms depends
only on their spin and orbital multiplicities and these are shown in Table
8.2. The actual magnitude of the splitting caused by spin-orbit coupling may
be determined from the electronic spectra of gaseous transition metal atoms
and ions (excited, for instance, in an electric discharge). For these it depends
on the metal. Although, in theory, the dependence should vary with the
atomic number Z as Z\ the actual experimental dependence on Z is not so
simple, as Table 8.3, which gives spin-orbit splitting constants for some
typical ions, shows.
For Tim the free-ion value of the spin-orbit splitting constant, denoted
by C (zeta), is 158 em -I. This is the quantity which determines the actual
spectral splittings observed and which varies from one ion to another.
So, given band half-widths of ca. 3000 em- 1 , fine structure caused by
spin-orbit coupling is most unlikely to be seen in the spectra of Tim
complexes. For Ni 11 the value is 703 cm- 1, which means that small splittings
might be caused by this effect, although none have been unambiguously
found. Spin-orbit coupling is more likely to be of importance in the
assignment of the very weak bands associated with spin-forbidden transitions
(which, as we will see, largely owe their intensity to spin-orbit coupling).
These are frequently much narrower than those corresponding to spin-
allowed transitions and, consequently, much smaller splittings will be visible.
For Rhm and Irm, both of which form stable complexes, Chas values of ca.
1400 and 4400 em -1, respectively so, for complexes of these ions spin-orbit
splitting of d-d bands is to be expected. Indeed, spin-orbit coupling is of
potential importance for all ions of the second and third transition series.
Table 8.2 Spin-orbit levels arising from crystal field terms
Crystal field term 1Al;l "Al;l 2A:>g 3A2g •A:>g 2Eg 5Eg
Spin-orbit levels 1 1 1 1 1 1 1
Crystal field term 3Tl;l •rlg 5r1g 2r2g 3T2g •r2g 5r2g
Spin-orbit levels 3 3 3 2 3 3 3
Table 8.3 Atomic spin-orbit coupling parameters (em - 1 )

Spin orbit coupling is apparent as splittings in the spectra of many-electron atoms.
However, it is not difficult to derive one-electron quantities from these splittings and
these one-electron quantities enable comparisons to be made between different
electron configurations. It is these one-electron parameters that are listed in this
table. All the configurations covered are of the form d" with the exception of the
uncharged metal atoms, all of which are d"s2 , and some highly charged ions of Ti
and V, all which are p". At the foot of each column of data for the first row transition
metals is given a number, m, which indicates the m dependence on Z, the atomic
number, of the data above in the form zm. An asterisk indicates a closed shell species
so that no spin-orbit splitting is seen in the spectrum.
Metal Charge on metal atom
0 1+ 2+ 3+ 4+ 5+ 6+
Ti (123) 104 131 158 (3716) (3973)
v (179) 154 187 220 253 (4900)
Cr (248) 219 256 296 337 378
Mn (334) 294 343 338 436 486 536
Fe (431) 388 441 499 554 612 673
Co (550) 500 561 625 695 760 830
Ni (691) 634 703 775 851 934 1012
Cu (857) 870 931 1037 1127 1224
m 7 7.5 7 6.5 6 6 5.5
Ru (1042) 968 1082 1201 1319 1441 1564
Rh (1259) 1177 1299 1426 1567 1689 1825
Os (3381) 3174 3531 3898 4259
lr (3909) 3690 4056 4430 4814
Data are adapted from Handbook of Atomic Data by S. Fraga, J. Karawowski and K. M. S. Saxena, Elsevier,
Amsterdam, 1976.
A more recent compilation is given by J. Bendix, M. Brorson and C. E. Schaffer, /norg. Chem. (1993) 32,
2838 but the definitions that they adopt, although with merit, are not those generally encountered in the
literature.
This is why our discussion of crystal and ligand theories has been exemplified
by complexes formed by elements of the first transition series. Had we
included spin-orbit coupling within the crystal field model (and in a more
complete treatment this would have been done), within the crystal field model
it would have appeared with its free-ion value. Not surprisingly, in ligand
field theory it becomes a parameter which, characteristically, is found to have
a value somewhat lower than that found for the free ion.
8.5 Jahn-Teller effect

The second effect which has been neglected is that due to low-symmetry
crystal fields. As the discussion in this chapter has been confined to
octahedral ML 6 complexes one might assume that low-symmetry ligand
fields could safely be ignored, but this is not so. A theorem due to
Jahn and Teller states that any non-linear ion or molecule which is in an
orbitally degenerate term will distort to relieve this degeneracy. This means
that all E., T 1• and T 2 • terms of d" configurations, in principle, are unstable
with respect to some distortion which reduces the symmetry. Of course, as
Jahn-Teller effect 1 167
has been discussed above, some of the orbital degeneracy may be lost because
of spin-orbit coupling. If this mechanism were to totally relieve the orbital
degeneracy then there would be no need to consider the Jahn-Teller theorem.
The Jahn-Teller theorem (which is group-theoretical in origin) tells us
that a regular octahedral complex will often be unstable with respect to a
distortion, but it says nothing at all about the magnitude of the distortion. An
exceedingly small distortion, small enough to escape detection by the most
sensitive technique, could in principle satisfy the Jahn-Teller requirement.
Indeed, there is often argument over the crystallographic evidence cited in
FJg. 8. 7 An octahedral complex, compressed support of the phenomenon, although the importance of the Jahn-Teller
along one fourfold axis (taken to be the z axis). effect is generally accepted and it is frequently invoked in explanations. The
most convincing evidence for the operation of the Jahn-Teller effect in
transition metal complexes is found in studies on Cu11 complexes, although
its importance is rather more clearly established in other areas, notably some
parts of solid-state physics.
Physically, the Jahn-Teller effect may be regarded as operating as follows.
Suppose that an octahedral complex is momentarily distorted as a result of
a molecular vibration and has the shape shown in Fig. 8.7. This vibration
will cause d., and dx'-y' to lose their degeneracy (the distorted molecule has
the symmetry of the trans octahedral molecule in Table 6.6). Suppose too
that in the undistorted complex the e9 orbitals were occupied by a single
electron. In the distorted complex this electron could be in either d., or
dx'-y'· Consider both possibilities. The distortion shown in Fig. 8.7 results
from a metal-ligand stretching mode which is of the form shown in Fig.
8.8(a); from this figure two things should be evident. First, that the caption
to Fig. 8.7 is incomplete-not only do the axial ligands move in, but the
equatorial ligands move out, although by a smaller amount, actually one half.
Secondly, the general nodal pattern and amplitudes in Fig. 8.8(a) follow
those characteristic of a d., orbital. We know that, in an octahedron, d., is
partnered by dx'-y'· So, the vibration in Fig. 8.8(a) is partnered by that in
Fig. 8.8(b). The two, as a pair, have e9 symmetry. We could discuss either
member but the discussions would differ in detail. We shall consider only
the d.,.Jike vibration, that shown in Fig. 8.8(a.) This is the one for
which, if it led to a static distortion, simple crystal-field arguments would
lead us to expect that the solitary e9 electron would be more stable in the
dx'-y' orbital. If the arrows in this diagram are reversed, so that the
distorted complex is one in which the four in-plane ligands move in and the
axial ligands move out compared with their positions in the undistorted
molecule, then we would expect the e9 electron to be more stable in d.,.
Consider now Fig. 8.9, which shows a diatomic potential energy diagram
which we apply, separately, to each bond in our complex. We compare the
cost in energy terms of the (two short, four long) and (two long, four short)
alternatives. In doing this we are just talking about 'cost' and so ignoring
the Jahn-Teller stabilization which is the driving force for the distortion-
but, as said earlier, the argument is indicative only, it certainly is not
complete.
Fig. 8.8 Two metal-ligand bond-length change Remembering that the amplitude associated with the two ligands is twice
vibra~ons of an octahedral complex. Those that associated with the four, it can be seen that the inherent asymmetry in
shown, together, have Eg symmetry; (a) has a
displacement pattern of a d, orbital, (b) the potential energy surface indicates that the cost of (two short, four long)
resembles dx2-y2· will be greater than that of the (two long, four short). This conclusion is in
Undistorted M-L bond length
~
Fig. 8.9 A M--l (diatomic) potential energy
diagram which enables an assessment of
energy changes consequent of the alternative
distortions shown in Fig. 8.8.
4 short M-L bonds

2 long M-L bonds
4 long M-L bonds

2 short M-L bonds
accord with the distortions that have been observed in crystal structure
determinations and other techniques. Octahedral complexes of Cu11 are
almost invariably distorted, commonly in the way shown in Fig. 8.10. This
is entirely consistent with the operation of the Jahn-Teller effect because the
ground-state electron configuration of a Cu11 ion in an octahedral field is
~~.e~, 2 E 9 • It has an odd number of electrons in the e9 orbitals.
Fig. 8.10 The distortion commonly found in There is one result of the general theory which should be mentioned. It
octahedral complexes of Cu". In real life the can be shown that in a regular octahedron the Jahn-Teller effect only
picture is often more complicated because operates by way of vibrations of e9 symmetry when the electronic state is
the two long-bonded ligands differ from the
other four. "E 9 (the value of n is irrelevant because the Jahn-Teller theorem applies only
to space functions, not spin). For electronic states of either "T19 or "T29
symmetries then the Jahn-Teller effect operates through vibrations of either
e9 or t 29 symmetries; a vibration of the latter symmetry is shown in Fig. 8.11.
Of course, the orbital degeneracy in an octahedral complex may be relieved
by distortions other than those shown in Figs. 8.8 and 8.11. However,
in such cases we may conclude that whatever is responsible for the distortion
it is not the Jahn-Teller effect. In particular, all of the Jahn-Teller-active
vibrations of an octahedron carry the g suffix and this means that they
cannot give rise to a distortion which destroys the centre of symmetry of an
octahedron. Distorted octahedral complexes which lack a centre of symmetry
cannot owe their distortion to the operation of the Jahn-Teller effect.
Fig. 8.11 A vibrational mode of T211 symmetry
of an octahedral complex. These vibrations This account of the Jahn-Teller effect indicates why it is of little importance
change bond angles, not bond lengths. when the t 29 orbitals are unequally occupied. Occupation of these orbitals
Jahn-Teller effect 1 169
2 long 2 short
4 short 4 long
more stable less stable
Fig. 8.12 The effect of the distortion shown in
Fig. 8.8(a) on an excited state in which the involves less destabilization than does occupation of the e9 orbitals-they
Jahn-Teller effect is operative. It follows that
the excited state must be orbitally degenerate; are, to a first approximation, non-bonding.
this figure demonstrates the loss of that The Jahn-Teller effect is of importance in the spectra of octahedral
degeneracy consequent on the distortion. transition metal complexes because the transitions observed usually involve
the excitation of a single electron from a t 29 to an e9 orbital. It follows that
there must be an odd number of e9 electrons either before or after the
transition. That is, if the ground state is not subject to a Jahn-Teller
distortion, the excited state is, and vice versa. To explore the effect of a
Jahn-Teller distortion on the electronic spectrum of a complex we shall
consider a simplified model. Suppose that the Jahn-Teller effect is operative
in the excited state and that molecular distortions have transiently distorted
the molecule in its electronic ground state so that it has the shape shown in
Fig. 8.1 0. Suppose too that the stable molecular geometry in the excited state
is also that shown in Fig. 8.1 0-that is, it is that briefly assumed by the
ground state. Then there also exists a less stable excited state for which the
molecular geometry is as shown in Fig. 8.12. The two excited states differ in
that in the one the e9 electron is in dx'-y' and in the other it is in dz'· If
whilst the ground state has the molecular geometry of Fig. 8.10 it absorbs
light and assumes an electronically excited state then we have to consider
the two possibilities shown in Fig. 8.12. First, the excited state could be the
more stable one, so that the molecule finds itself in its vibrational ground
state. Secondly, the molecule could assume the less stable excited state but
would then find itself well away from its equilibrium geometry-it is
Q) vibrationally excited, possibly by several quanta. These quanta may each
"t1l<= have an energy of ca. 300 em- 1 (assuming no great change in frequency
-e
0 between ground and excited state). Remembering the intrinsic difference in
<J)
.0
<( energy between the two excited levels in their vibrational ground states, we
conclude that such excited state Jahn-Teller distortions may well be
spectrally apparent. For example, this is believed to be the explanation of
the asymmetry in the 2 E9 <-- 2 T29 d-d d transition of the [Ti(H 2 0) 6 ]3+ ion
shown in Fig. 8.13.
Fig. 8.13 The asymmetry in the d-d spectrum
of the [Ti(H 2 0) 6 ] 3 + ion (d 1 ), believed to result
If the Jahn-Teller effect operates in the electronic ground state of a
from the operation of the Jahn-Teller effect in complex it is unlikely to be apparent in the electronic spectrum. If the two
the excited state. ground states are split sufficiently far apart for transitions to the excited state
to be resolvable, the upper of the two ground states will not be sufficiently
thermally populated for transitions from it to be seen. On the other hand,
in this situation there will be an additional transition in the infrared, although
probably difficult to locate amongst the forest of vibrational bands in this
spectral region. This additional band corresponds to a simultaneous electronic
and vibrational excitation from the lower of the ground states to the upper.
A Jahn-Teller distortion may be either static or dynamic and so far only
the former has been considered. There are, in fact, theoretical reasons for
expecting static Jahn-Teller distortions to be atypical, which is perhaps why
it is so difficult to point to firm evidence for the effect in ground states. The
dynamic case occurs when the potential barrier separating, for example, the
three equivalent distortions of the type shown in Fig. 8.1 0, an elongation
along x, y or z axes, is of the order of kT. The distortion then rapidly
alternates between the possibilities. In this way a small Jahn-Teller effect
may be unobserved--the time average is an undistorted structure (as will
become apparent towards the end of Chapter 9, small distortions may be
detected by magnetic measurements). Because the time taken to jump from
one configuration to another is of the same order as the time taken for a
simple vibration, a measurement taking much longer than this to perform
will give only an average. A measurement of 10- 10 s duration takes too long!
The dynamic Jahn-Teller effect does not affect the visible appearance
of the electronic spectra of transition metal complexes but this is because of
the energetics of the effect, not because of the timescale.
8.6 Band contours

The d-d bands of transition metal complexes are weak, sometimes very weak.
Some of the bands are sharp, some broad and some so broad that it is
difficult to be certain that they exist. In the next section we shall consider
the problem of intensities; in this section, the problem of band shapes. The
two are related. The very weak peaks are usually sharp. If they were not,
they would escape detection and no doubt many broad, very weak, peaks do
pass undetected.
When discussing the Jahn-Teller effect a refinement was introduced into
our model of a transition metal complex: it was allowed to vibrate. This is
an important refinement for it enables us to understand both band intensities
and shapes. Consider an octahedral metal complex in which the totally
symmetric metal-ligand stretching mode (the breathing mode) is excited.
That is, the complex retains its octahedral geometry because all of the M-L
bonds are contracting and elongating in phase. The essential point is that
the repulsion between the ligands and metal d electrons and, therefore, the
crystal field splitting parameter, .1., will vary with the metal-ligand distance.
On the molecular orbital model, the overlap between the ligand and metal
orbitals will be modulated by the vibration. The conclusion is that .1. is not
a fixed quantity but rather one that varies with metal-ligand separation, so
that at any instant it varies from one molecule to another. A convenient way
of thinking about this is to regard a collection of molecules as being
represented, not by a line in a Tanabe-Sugano diagram (as in Fig. 8.3) but
by a band, covering a range of values of .1-jB. An absorption peak will
therefore not consist of a narrow line but, usually, a broad one, a superposition
Band intensities I 171
of a multitude of sharp ones. The relative breadths of the spectral lines will
be determined by the relative slopes of the lines which, in a Tanabe-Sugano
diagram, represent the excited states. For example, Fig. 8.3leads us to predict
that transitions to the 3T 1.(P) and 3T 29 terms of the d 2 configuration will
give bands of similar widths. As Fig. 8.1 shows, this prediction is roughly
confirmed by experiment. We further predict that the transition to the 3 A 29
excited state will give a band which is about twice as broad as the other
two. This, together with its inherent low intensity (see below), is no doubt
why it is difficult to observe. However, as the Ni 11 example given earlier in
this chapter shows, it is possible to see such transitions in favourable cases.
On the other hand, some of the lines in a Tanabe-Sugano diagram are
almost parallel to that representing the ground state. These lines are
invariably amongst those derived from the same strong field configuration
as the ground state. For example, as Fig. 8.3 shows, in the d 2 configuration
the .1. dependence of the 1 A 19 , 1 E 9 and 1T 29 terms is roughly parallel to that
of the 3 T 19(F) ground state. These four terms are those which arise from the
t~ 9 strong field configuration. That is, a transition from the ground to another
of these terms corresponds to a rearrangement of electrons within the t 29
orbitals and so does not depend on the t 29-e9 separation, .1.. This statement
is rigorously true in the strong field limit but is only approximately true in
weak fields. In the latter case, mixing of the 1 A 19 , 1 E9 and 1T 29 excited terms
e;
derived from the t~ 9 e; and configurations, into those derived from the t~ 9
configuration, makes the lowest 1 A 19 , 1 E9 and 1T 29 term each have a different
dependence on .1.. However, in intermediate and strong ligand fields it is true
that the energies of transitions to these terms are scarcely modulated by the
.1. variations caused by ligand vibrations. It follows that transitions to these
terms will appear as relatively sharp lines in the spectrum. However, they
are very weak because they are spin forbidden (a spin triplet to a spin singlet)
and for this reason are not easy to detect. The existence of these transitions is
exploited in those lasers which involve transition metal ions, for instance in
the ruby laser. Ruby is an Al 2 0 3 crystal containing a small amount of Crill
as impurity, the chromium being approximately octahedrally surrounded by
oxygen atoms. It is on this Crill, a d 3 ion, that we focus our attention. As
Fig. 8.14 shows, the energies of the transitions from the ground term to the
lowest 2 E 9 , 2 T 19 and 2 T 29 terms are essentially independent of .1. and so the
bands are sharp. Emission of light from Crill ions in the 2 E9 excited term as
they fall back to the ground state gives rise to the essentially monochromatic
red light emitted by the ruby laser.
In the above section only the effect of the totally symmetric breathing
vibration on the value of .1. was considered. Similar conclusions follow if
vibrations of other symmetries are considered in detail. As will be seen
in the next section, some of these other vibrations are responsible for the
appreciable intensity in formally forbidden d-d transitions.
8. 7 Band intensities
The fact that very weak d-d transitiOns may be observed in atomic
spectroscopy indicates the approximate nature of the rule that only s ..... p,
p <--> d, d <--> f etc. orbital transitions are allowed. The spin-allowed d-d
transitions in an octahedral metal complex are of much higher intensity than
4 T1g IPJ
- - - - - - - , - - - - - - - - . - - - - - - - - "T-----
''
70
'
''
'
60 ---------,--------
' r---
' ---
'
'''
'
'
50 '
- - - - - - -'1 - - - - - - - - .J---
Fig. 8.14 A simplified Tanabe-Sugano diagram ' '
for the d3 configuration (appropriate to Cr"), '
' '''
showing the excited terms which have a
dependence on !1/B which is almost the same
as that of the ground state. Such parallel co 40 4T2g
LiJ
I
behaviour only occurs when the ground and
excited terms correlate with the same strong
field configuration. In the present case this is
~. (see Table 7.5). 30
~~--~~~r:~~~---:· 2Gg
'
;__;_;:_;:_: : :;_
'
'2r1g
20 d.~.-..,b.=~~~-t-:;;_:;:i:;;_;_ i 2Eg
2G
4p ''' '
10 ''
----~--------
4F
0 10 20 30
!J.fB-
their atomic counterparts so that some new intensity-generating mechanism

exists in the complex which is not present in the free ion. Not surprisingly,
it involves metal-ligand vibrations. Consider, for example, an octahedral
complex which has momentarily been distorted by the vibration of T 1 u
symmetry shown in Fig. 8.15. This vibration destroys all of the threefold and
all but one of the fourfold axes (that along which the vibration takes place).
The molecular symmetry is, in fact, the same as that of a trans-ML 4 L'L"
complex, Ck As Table 6.6 shows, in this symmetry, p, s, and d,, carry the
same symmetry label (a 1 ) and so may mix. Similarly, Px and pY together have
the same symmetry (e) as d,x and dy, and so these pairs of orbitals may also
mix. In the distorted molecule of Fig. 8.15 what we have called d,, becomes
a mixture of d,, with a little of p, and of s. The contamination by these other
orbitals disappears when the molecule becomes accurately octahedral again.
Similarly, what we have called d" is contaminated by Px and dyz by Pr This
means that if an electronic excitation occurs whilst the molecule is distorted,
a transition in which an electron jumps formally from d,x to d,, also contains
a component in which the electron jumps from d" to p, and from Px to d,,,
both of which are allowed by the selection rules. The d-d transition gains
intensity, therefore, because it is not entirely d-d. The intensity depends on
FJg. 8.15 An M-L bond stretch vibration of T1 u the extent of d-p mixing and, in turn, this depends on the nature of the
symmetry of an octahedral complex. distortion at the moment of excitation. Any distortion which fails to remove
Band intensities 1 173
the centre of symmetry of the molecule will not introduce any intensity, for
the essential requirement is that the vibration mixes atomic orbitals of q and
u types (and, so, in our example, d and p orbitals, respectively). This topic
is explored in some detail in Appendix 8, an appendix that shows that there
will be perhaps ten vibrational subpeaks, each with a L1 modulation and
Jahn-Teller broadening, leading to the relatively weak, broad peaks observed.
The above discussion is associated with a breakdown in the Born-
Oppenheimer approximation. This states that electronic and vibrational
energy levels may be treated separately, so that under this approximation
an electronic transition is something quite separate from a vibrational
transition. However, in our discussion we have found that the one depended
on the other; that there is a coupling between vibrational and electronic
states. This is called vibronic coupling. The intensity introduced into a d-d
transition by vibronic coupling cannot come from nowhere. As is evident
from our discussion, it is in fact 'stolen' from what in a regular octahedron
is an allowed transition-for example, the equivalent in ·the complex of the
free ion d --> p transition, a transition which will lie in the ultraviolet. A more
detailed analysis shows that the magnitude of the stolen intensity is expected
to be an inverse function of the energy separation between the d-d band
and the allowed band. This means that the highest energy d-d band, that
with the smallest separation from the allowed band, will generally be the
most intense, other things being equal. This perhaps explains why the
29 transition in [Ni(H 2 0) 6 ] + -a two-electron promotion (see
3 T (P) <-- 3 A 2
19
Fig. 7.18)-is seen at all.
We now turn to the origin of the intensities of the spin-forbidden d-d
transitions. One possible source is a magnetic dipole mechanism. In electron
paramagnetic (spin) and nuclear magnetic resonance spectroscopies (EPR
and NMR), electromagnetic radiation is used to 'turn over' the spin of either
an electron or nucleus in a molecule (this simple view of the process is
adequate for our present purposes). The electromagnetic radiation does this
by virtue of its associated magnetic vector and the process is said to be
magnetic dipole allowed. Classically, the alternating N-S magnetic dipole
associated with the light wave causes similar magnetic dipoles in the molecule
to alternate in direction in phase with it. When the alternation coincides
with a resonant frequency, a transfer of energy from the light wave to the
molecule occurs. Although a similar process undoubtedly contributes to the
intensity of the spin-forbidden d-d bands-such transitions are magnetic
dipole allowed-detailed calculations indicate that its contribution is small,
although, as we shall see in Chapter II, this mechanism is of importance for
f electron systems. Much more important is spin-orbit coupling. Our
discussion in Section 8.4 demonstrated that the spin and orbital properties
of an electron are not completely independent of each other. Earlier in this
section we described a vibrational mechanism by which an orbital transition
becomes weakly allowed, so, in principle at least, spin-orbit coupling
may transfer some of this allowedness into what is, formally, a spin-forbidden
transition. Let us consider a specific case.
As we have seen, the ground state of an octahedral d 8 complex is 3 A 29 .
A low-lying spin-forbidden transition is to a 1E 9 term and a spin-allowed
transition is to the 3 T 19 term (see the modified Tanabe-Sugano diagram
shown in Fig. 8.16). Now, spin-orbit coupling has the effect of contaminating
174 I Electronic spectra of transition metal complexes
------ T--- -----:---- ----~-- ------

, '
' '
''
60 '' ----- i'' -----
----- -i---
' '
' '
: :
'
'
50
40
Fig. 8.16 Spin-orbit coupling causes an "'

interaction between the ~Eg low-lying excited LJ
state of an octahedral d8 complex with two of 30
the wavefunctions of a 3 T14 . This interaction
leads to mutual repulsion and a non-crossing of
the energy levels.
'
'
'
'
'
'
- - - - - - - T' - - - - - - - - , - - - - - - - - l'
'
'
'
'
'
'
10 20
6./8---
two of the nine wavefunctions of the 3T 1 • term with those of the 1Eg term
(of which, of course, there are also two) and vice versa. When the energy
separation between these two terms is small (and this occurs at C./ B :::o II in
Fig. 8.1 6) the contamination becomes gross pollution! The result is that for
these four coupled wavefunctions the two lines on a Tanabe-Sugano diagram
do not cross but repel each other and behave as shown in Fig. 8.16 so that
those two wavefunctions which in a weak field belonged to 3 T 1 • become 1 E.
in a strong field and vice versa. This means that some of the intensity of
what was the 3 T 1• +- 3 A 2 • transition (before we included spin-orbit coupling)
is transferred to the 1 E 9 +- 3 A 2 • transition once spin-orbit coupling is
included. All other spin-forbidden transitions are believed to gain intensity
by similar mechanisms. One explanation which has been suggested for the
double-headed peak in the [Ni(H 2 0 6 )]2+ spectrum of Fig. 8.5 is that it
originates in this mixing, the d-d intensity being shared by the two
components. Against this, it has been argued that a detailed analysis,
including all interactions, leads to a diagram sufficiently different from Fig.
8.16 for it to be unlikely that this explanation holds for this particular
example-it is not true that the complex falls at the crossover point. An
alternative explanation for the double-headed band is that the splitting
originates in the Jahn-Teller effect associated with the r~.e~ strong field
configuration with which the 3 T 19(F) excited term correlates. But in this case
why is the lower band, to 3T 29 , which also correlates with r~.e; not also
split-and, if this is the explanation why is the band-splitting not found in
most, if not all, octahedral Ni 11 complexes? At the present time there is no
agreed explanation for the splitting.
Two points should be mentioned. First, although the 1E 9 <-- 3 A 29 transition

of the above example would, by the arguments of the previous section, be
expected to be sharp, mixing with the 3 T 19 <-- 3 A 29 transition will broaden
it. In general, the lower the intensity of a spin-forbidden band the narrower
it is likely to be. Secondly, the greater the spin-orbit coupling, the more
important the spin-forbidden transitions are likely to be in the spectra. We
shall have more to say about this when we discuss f electron systems in
Chapter II. The relative magnitudes of spin-orbit coupling constants has
been discussed in Section 8.4. We conclude this section by considering
transitions such as 3 A 29 <-- 3 T 1 g(F) of the d 2 case. In the strong field limit
this corresponds to a transition from the ~~. to the e; configuration. Now,
except in a very special circumstances, the effect of electromagnetic radiation
is to excite one electron at a time, not two. In the strong field limit the above
transition is therefore forbidden (and rigorously so, within the general
approximation scheme that we are using). In the intermediate and weak field
regions the 3 T,.(F) level is a mixture of the d. and t~ 9 e: configurations (this
was discussed in Section 6.5) and so the 3 A 2 • <-- 3 T 19(F) transition has a
e;
t~.e: <-- component which, of course, corresponds to a one-electron jump.
This component can acquire intensity by a vibronic mechanism. It follows
that the most favourable situation for observing the 3 A 29 <-- 3 T 19(F) transition
in d 2 complexes will be with very weak field complexes. In this situation it
will lie at a relatively long wavelength and, hopefully, will not be obscured
by the charge-transfer bands which, as we will see later, commonly occur in
the near ultraviolet. In contrast to the conclusion that we have just reached,
we note that, of all the transitions in the [Ni(H 2 0) 6 ] 2 + spectrum, the highest
is the most, not the least, intense. This clearly indicates a weakness in our
model; most probably the fact that we have ignored any mixing between
d-d and the intense charge transfer transitions that will be discussed in
Section 8.10; a mixing that was also mentioned earlier.
8.8 Tetrahedral complexes

So far in this chapter we have only discussed octahedral transition metal
complexes. However, all of what we have said may be carried over to the
tetrahedral case, provided that we remember that .1.,., is of opposite sign to
.1. (in a tetrahedral complex d,,., d,.z and d" are above dx'-r' and dz,, a
0 "
point developed in some detail in Section 7.7). This means, for example, that
the spectrum of a tetrahedral d 2 complex can be interpreted using a Tanabe-
Sugano diagram for the octahedral d 8 case because, as was discussed in
Section 7.4, the splitting pattern for d 8 is the opposite to that for d 2 . It is to
facilitate this application that Tanabe-Sugano diagrams quite often do not
contain any g or u suffixes in their term labels. There is only one major point
of difference that has to be discussed and this concerns the intensities of d-d
bands of tetrahedral complexes. As is evident from Fig. 8.17, where the
spectrum of the octahedral [Co(H 2 0) 6 ]2+ ion is compared with that of the
tetrahedral [CoC1 4 f- anion, these intensities are at least an order of
magnitude greater in tetrahedral complexes than in their octahedral counter-
parts. Evidently, some new intensity-generating mechanism is available in
tetrahedral complexes.
E 2
Q)
·;:;
~0
"
c:
0 600
Fig. 8.17 The spectra of the d7 species :g
[Co(H 2 0) 6 ] 2 + (which is weak field) and c:
[CoCI 4 ] 2 ~. The latter is two orders of 'iii
UJ
magnitude more intense.
400
200
25 000 20 000 15 000 10 000 5000

cm-1
One important aspect of the electronic structure of tetrahedral complex

ions which was not discussed in detail in Chapter 7-although it was
mentioned in Chapter 6 and is contained in Table 6.6-is that, unlike the
octahedral case, in a tetrahedral complex the metal p orbitals have the same
symmetry (t 2 ) as do the metal dx,, d,, and d,. orbitals. Because they are of
the same symmetry, the t 2 set of d orbitals will be mixed by the crystal field
with the metal p orbitals. This in turn means that an e -+ t 2 transition
contains some d -+ p component and this component is an allowed transition.
Evidently, the intensity of the e -+ t 2 transition is related to the extent of
d~p mixing so that this may be worked backwards and the extent of mixing
assessed, at least qualitatively, from the intensities of d~d bands in the
spectra. Notice both the similarity and the difference between the intensity-
generating mechanisms for d~d transitions in octahedral and tetrahedral
complexes. Both depend on d~p mixing but only for the tetrahedral case
does this mixing occur for the non-distorted molecule.
8.9 Complexes of other geometries

As the ligand-field splitting of the d orbitals in lower symmetry complexes
lead to lower degeneracies and so a need for more parameters to describe
them (see Table 6.6), so the d~d transitions of complexes of other than
octahedral and tetrahedral geometries can only be discussed by introducing
these parameters. For square planar complexes, for instance, two additional
parameters must be introduced.
''
''
'
''
''
''
'
FIJI. 8.18 The spectrum of [Co(NH3 ) 4 (ox)]+
(second down, indicated by vertical bars) to a
first rather good approximation is a weighted
interpolation of those of [Co(NH3 )s] 3 + and
[Co(ox) 3 ] 3 -. This is an example of an
application of the rule of average environment.
30000 15000
The number of observed spectral bands is seldom sufficient to allow all

of the parameters to be simultaneously determined. Commonly, therefore,
one has to be content either with tentative band assignments or to extract
data appropriate to an oversimplified model-assuming that deviations from
an ideal geometry may be ignored, for example. A useful first approximation
to the assignment of the spectra of coordination compounds in which the
symmetry is low by virtue of the coordination of several different ligands is
provided by the rule of average environment. This states that, for example,
the value of 11 for the complex [ML 3 L~] will be approximately the
mean of the 11 values for [ML 6 ] and [ML6]; similarly, for [ML 4 L;] the
observed 11 will be approximately
~(211[MleJ + L1[M6J)
As mentioned in Section 7.3, the rule of average environment may be used

with the data in Table 7.1. The ligand non-equivalence may manifest itself
by a splitting of the bands expected for an octahedral complex with the value
of 11 predicted by the rule of average environment. An example of the
application of this rule to spectra is contained in Fig. 8.18, where it is shown
that the spectrum of [Co(NH 3 ) 4 (ox)] + may be predicted rather well from
those of [Co(NH 3 ) 6 ]3+ and [Co(ox).] 3 -.
8.10 Charge-transfer spectra

As is evident from the simple description given in Section 8.1, if the
absorption of light is to cause an electronic transition within an atom or
molecule, it is essential that the absorption results in a movement of charge
density. This displacement may be localized on one atom-as it is in the
d-d spectra which have been the subject of this chapter up to this point-or
it may be the displacement of charge from one atom to another. Electronic
transitions which can be ascribed to the latter process are termed charge-
transfer transitions. These are not always simple processes, as shown by
the fact that change of solvent or of ions present in solution can influence
the position of a charge-transfer band, particularly when the complex is
asymmetric (and so has some inherent polarity). For the moment, such
complications will be ignored. Three main categories of charge-transfer
processes may then be distinguished in coordination compounds.
Class 1-lntraligand transitions
A ligand such as SCN- has internal charge-transfer transitions, usually
located in the ultraviolet region of the spectrum. Corresponding transitions
occur in the coordinated ligand but can usually be identified by comparison
with the spectrum of the free ligand. Any metal ion to which the ligand is
complexed is not significantly involved in the transition. This class will
therefore not be discussed further.
Class 2-Ugand to metal (reduction) charge-transfer
transitions (LMCT)
These are a common type of transition, in which a ligand electron is
transferred to a metal orbital and the charge separation within the complex
thereby reduced. The process can lead to a permanent transfer of charge so
that the compound involved is photolytically changed. This offers not only
a method of measuring the intensity of a light beam (actinometry) but in
recent years has been the subject of much research in an attempt to couple
such a process into the reaction
2H 2 0 -+ 2H 2 + 0,
If this could be cheaply achieved it would have considerable economic
consequences. In Chapter 14 the topic will be briefly explored. In the present
section we are not concerned with such permanent charge transfer but,
rather, the process in which electron transfer from one orbital to another is
caused by the absorpion of light (the return of the molecule to its ground
state subsequently occurring by some energy degradation mechanism which
is not our present concern). One problem associated with this class of
charge-transfer transitions is the lack of certainty about which orbital the
electron comes from. Thus, one is tempted to discuss such a transition in
terms of molecular orbital energy level diagrams such as those of Figs. 6.11
or Fig. 6.26. Figure 6.26 is more complete than is Fig. 6.11 because the former
covers both ligand a and n electrons, a total of six on each ligand. But a
monatomic ligand such as F- or 0 2 - has eight valence electrons. The two
which have not been included are in a a orbital which points away from
the metal atom. These a orbitals give rise to ligand group orbitals which
are probably higher in energy than most of the occupied ligand orbitals of
Charge-transfer spectra 1 179
Figs. 6.11 and 6.26 because they are not stabilized by interactions with the
metal orbitals. They might, then, be expected to give the lower energy
charge transfer bands (as, indeed might some of the other ligand orbitals
mentioned above, for the reason indicated at the end of Section 6.2). This is
probably the case but, as the intensity of charge-transfer bands is a function
of the overlap of the orbitals between which the overlap occurs, the
corresponding intensities are small. Clearly, there is a danger that these bands
may be confused with d-d transitions. This is in contrast to most charge-
transfer bands, which are intense, and so easy to distinguish.
Class 3-Metal to ligand (oxidation) charge-transfer
transitions (MLCT)
In this type of transition a metal electron is transferred to an orbital largely
located on the ligands. It is therefore, in a sense, the opposite to type of
charge-transfer transition described in the prevous section. So, it corresponds
to an oxidation of the metal and reduction of the ligand. An example is
provided by the [M(H 2 0) 6 ] 2 + ions of the first transition series, for which
Dainton has shown that the energy ofthe first charge-transfer band is linearly
related to the redox potential of the system
M2 +(aq) ..... M3 +(aq) + e-(aq)
Much of what has been said about ligand metal charge-transfer bands applies
to this class also-they, too, can be used as a basis for actinometry; it would
be attractive to combine both types in a catalysed photolytic decomposition
of water.
It is not always easy to decide whether category 2 or 3 is involved in a
particular transition. If it is 2, then, for an allowed transition in an octahedral
complex, the excitation must be from a ground-state orbital of u symmetry
(and, therefore, t 1• or t 2 • as there are no other occupied u orbitals) and the
electron must be excited into either a t 2 • or e. orbital. If both t 2 • and e. sets
are incompletely filled then both excitations may occur and two bands (or
families of bands) separated by Ll are to be expected. If the t 2 • set is full then
only excitation to the e• set will occur. This probably explains why there is
a long wavelength charge-transfer band in [Fe(CN) 6 ] 3 - (d 5 ) which has no
counterpart in [Fe(CN) 6 ] 4 - (d 6 ). A similar pattern is observed in the anions
[IrBr6 ] 2 - (d 5 ) and [IrBr6 J'- (d 6 )-the latter pair are illustrated in Fig.
8.19. In both cases the band in the d 5 species which disappears is about Ll
away from the one which remains (the Ll being that of the higher valence
ion, of course), as the above explanation requires. If we regard charge-
transfer spectra as involving an electron moving towards or away from a
cation then the fine details of the cation's electronic structure assume a
lesser importance, although it may influence the number of bands, as we
have seen. Adopting this viewpoint, then in cases such as those just dis-
cussed in which there are similar complexes differing by a single formal
charge, we would expect ligand -> metal charge-transfer to become easier-
and so of lower energy-as the formal charge on the metal ion increases.
Conversely, metal -> ligand charge-transfer would be expected to become
more difficult and so of higher energy. These, then, are criteria by which the
type of charge transfer transition involved in a particular band can tentatively
be identified.
[lr Br6 ]3- t~g
Rg. 8.19 The visible and near-ultraviolet

spectra of the related octahedral complexes
[lrBr6 ] 2 - and [lrBr6 ] 3 -, d5 and d6 , respectively.
There is a complicated long-wavelength band
pattern in the d5 case which is absent in the d6 •
The complexity suggests that orbitally
degenerate states are involved, consistent with
a charge-transfer transition in which F,g (ground
state) becomes a ~ in the excited state.
m~
Because the charge-transfer process involves electron transfer the spectra

are sometimes called redox spectra and, certainly, band positions can
correlate with the ease of oxidation or reduction of the metal ion or
ligand-we have already met one example of this. A second, rather different,
example is given in Fig. 8.20. This figure shows the charge-transfer spectra of
[Co(NH 3 ) 5 X]2+,X = C1-,Br- andi-.Asourdiscussionleadsustoexpect,
the transition energies decrease in the order Cl- > Br- > r. Usually, in
such series the charge-transfer process does not lead to a permanent
oxidation or reduction. However, along the series [CuCI 4 ] 2 -, [CuBr4 ] 2 -
and [Cul4 ] 2 - charge-transfer absorption occurs at longer and longer
wavelength until, for the latter, the process becomes irreversible and
decomposition into Cu' and 12 becomes a spontaneous process. Similarly,
copper(II) chloride is green, copper(II) bromide is black and, of course,
copper(II) iodide is not stable. These sequences provide an excellent
demonstration that the particular charge-transfer process involved is ligand-
to-metal in nature. Similarly, [FeF6] 3 - is virtually colourless, but replacement
of the fluoride ion with the heavier halogens leads first to yellow solutions
(CI-), then brown (Br-) and finally to the spontaneous reduction of Fe"'
by I-, at least in aqueous media.
It has been found possible to calculate the positions of charge-transfer
bands by assuming that they depend on an electronegativity difference
between the ligand and the metal ion. The values of these so-called optical
.,g electronegativities indicate, as observed, that the first charge-transfer band
-e0 in the spectrum moves to lower energies in the order:
~ Pt'v > Rh 111 > Cu 11 > Co 111 > lr111 > Ni 11
Finally, as already mentioned, it seems clear that the solvent molecules can
be involved in the charge-transfer process. Detailed discussions of charge-
Wavelength - transfer spectra in transition metal complexes quite often label the corre-
Rg. 8.20 The lowest energy charge-transfer sponding bands quite separately, giving them the label CTTS-charge
bands of the complexes [Co(NH3 )sl(] 2 +, transfer to solvent. So, the fact that Fel 3 has recently been prepared in
X ~ Cl-, Br-, 1-. Qualitatively, the less
electronegative the X, the longer the wavelength non-aqueous media suggests that the solvent-water-is not always the
(the easier the transition). mere spectator that it was implicitly assumed to be above.
lntervalence charge-transfer bands I 181
8.11 lntervalence charge-transfer bands

So far in this chapter we have followed the assumption implicit in the
previous two chapters-that we could treat transition metal ions as indi-
viduals, isolated along with their ligands. It will become apparent in the next
chapter that there is magnetic evidence that this assumption is not always
valid and so it is appropriate to recognize that similar spectroscopic evidence
also exists. The evidence is not confined to transition metal ions and so we
shall broaden this section to cover the entire periodic table (or almost!),
although referring solely to 'metals' for linguistic simplicity. It quite often
happens that when a metal appears in two different valence states in a
compound that the colour of the compound is not that corresponding to
the sum of the colours of its components. Examples are I3, which may
be regarded as derived from I 0 -I 2 has a purple vapour-and I-, the I-
anion is colourless; soluble Prussian blue, the deep blue almost black
KFe[Fe(CN) 6 ] · H 2 0, prepared from Fe", pale blue, and [Fe(CN) 6 ] 3 -,
yellow.
These colours arise because there is an overlap between the orbitals on
the two or more metal atoms with different valence states. This overlap can
be direct, as in 13, or indirect, through mutual overlap with an intervening
ligand, as in KFe[Fe(CN) 6 ] · H 2 0, where the CN- anion is involved. This
overlap involves both ground and excited electronic states. The intense
colours arise because if the electron density (or just 'electron') that is excited
is in a ground state orbital largely located on one atom, then in the excited
state it is largely located on the other. This pattern is easy to see for the case
of just two interacting atoms. The effect of the overlap between two orbitals
is to give sum and difference combinations. If the sum combination is, in
fact, largely located on one atom the difference combination has no choice
but to be largely located on the other. Movement of electron density from
ground to the excited state will be associated with a large electric dipole
change and, hence, absorption. These are called intervalence-transfer bands,
to distinguish them from the charge-transfer absorptions discussed in the
previous section. Commonly, but not always, they involve identical atoms,
differing only in their formal valence states-iodine and iron in the two
examples that have been cited. The ground and excited states of such atoms
are likely to have a compatibility seldom shared by dissimilar atoms.
Not all mixed valence compounds show intervalence bands. Those that
show not the slightest evidence for them are often called class 1. In class 2
intervalence compounds the intervalence-transfer band may well dominate
the visible spectrum, swamping the spectra of the individual ions. However,
these may be seen; the important point is that the different ions retain their
chemical individuality. Soluble Prussian blue is a case in point. This contains
Fe" and Fe111 bridged by CN- ligands. Isotopic tracer measurements show
that there is no doubt-the Fe111 is bonded to theN atom of the CN- ligand
and the Fe" to the C. If radioactive Fe is used in the preparation and the
compound subsequently decomposed, the activity remains in the iron of the
same valence state in which it was incorporated. A class 3 also exists,
exemplified by the I3 anion, in which it is not possible to associate a unique
valence state to individual metal ions; typically, but as our example shows,
not always, they are structurally indistinguishable. When they form an
Table 8.4 Examples of intervalence-transfer compounds (after

Robins and Day)
Class 1-no detectable interaction

Cr 1Cr 11 F5 Cr2 F5
Fe 11 Feg'F8 ·10H 20 Fe 3F8 · 10H20
PbgPb 1v04 Pb 304 (red lead)
Ti 1TI 111 (CN) 4 TI 2 (CN) 4
Class 2~1early interaction, but chemical identity retained

KFe 11 [Fe 111 (CN) 6] soluble Prussian blue
Fe~1 [Fe 11 (CN) 6 ] 3 insoluble Prussian blue
[Cu'Cu"CI 6] 3 - [Cu 2CI 6] 3 -
[Au 1Au 111 CI 6] 2- [Au 2CI 6f-
[Sb111Sb1vBr12]4- [Sb 2Br12 ] 4 -
[Pb 11 Pb 1vCI 12 ] 6 - [Pb 2CI 12] 6 -
Ciass 3~1early interaction, chemically distinct species

cannot be identified
[Cu4s3r
[Cu 4 s 4r
[Nb6CI12]2 +
[Nb6CI14]4+
[Tc 2 CI 8] 3 +
AgF2
extended lattice such compounds typically have a metallic appearance and

even metal-like electrical conductivity. Examples of all three classes are given
in Table 8.4.
Finally, we return to transition metal ions, but in a slightly different
context. The class 1, 2 and 3 behaviours which have just been detailed refer
to interactions between two or more ions in different valence states. In the
next chapter we will meet interactions which can be between ions in the same
valence state, interactions which are evident in the magnetic properties of
the ions. Magnetically coupled (or exchange coupled, as it is usually put)
ions also show spectroscopic evidence for the coupling-what in an isolated
ion are spin-forbidden transitions appear with enhanced intensity (often
temperature dependent, the intensity rising with temperature) and additional
fine structure. In addition, new absorptions appear with energies approxi-
mating the sum of energies of two single excitations. Overall, the colour may
appear somewhat different. Thus, copper(II) acetate, an exchange-coupled
dimer, is green, whereas the common colour for copper(II) salts is blue.
8.12 Conclusions
The discussion in this Chapter has, at times, been somewhat complicated,
the complications centring around such topics as the extraction of quantitative
data from spectra or the details of how a forbidden band obtains intensity.
The assignment of the band pattern has, tacitly, been assumed to be
straightforward. This is by no means always so. For first row transition metal
complexes there is seldom a problem but for the others the larger values of
~ mean that there is more overlap between d-d and charge-transfer bands.
Further, the larger values of the spin-orbit coupling constants for these
elements make intensity and band-width criteria for distinguishing between
spin-allowed and spin-forbidden transitions less reliable. Even more difficult
is the case of lower-than-octahedral symmetry. It is probably true that the
electronic spectrum of no transition metal complex has been studied in as
much detail as has that of the square-planar [PtC1 4 ] 2 - ion. Despite all of
this work, it is only relatively recently that some general agreement on the
assignment of the spectrum has been achieved and few would regard as
completely improbable a revision of some aspect of the current interpretation.
Finally, apart from a brief reference to the ruby laser, the content of this
chapter has been concerned with the absorption of light. Some complexes
emit light also, giving rise to the phenomenon of fluorescence (the excited,
emitting state has the same spin multiplicity as has the ground state) and
phosphorescence (the excited state has a different spin multiplicity from the
ground state). Such processes arc currently exciting much interest but as the
identity of the actual excited states involved is seldom agreed they will not
be discussed in this chapter. They will be referred to in Chapter 14 because
-
a related topic, that of the reactions of electronically excited states, is better
defined.
Further reading Compounds', Chapter 4 in Modern Coordination Chemistry

J. Lewis and R. Wilkins (eds.), Interscience, New York,
An old but still very useful article that not only covers the 1960.
material indicated in the title but which also explains very
• C. K. 1mgensen 'The Nephalauxetic Series' Prog. Inorg.
simply how spectral data may be interpreted and also lists
Chern. (1962) 4, 73
spectral data for many compounds is 'The Nephalauxetic
Effect. The Calculation and Accuracy of the Interelectronic • M. B. Robin and P. Day 'Mixed Valence Chemistry-a
Repulsion parameters for cubic high spin d 3 , d 3 , d 7 and d 8 Survey and Classification' Adv. Inorg. Chern. Radiochem.
systems' by E. Konig inStruct. Bonding (1971) 2, 175. (1967) 10, 247.
Much of the basic work in this subject was done in the
1960s and so many of the more useful references tend to date • N. S. Hush 'lntervalence-Transfer Absorption' part 1 (with
from this period. Examples are: G. C. Allen) and part 2, both in Prog. Inorg. Chem. (1967)
8, 357 and 391, respectively.
• C. 1. Ballhausen 'Intensities of Spectral Bands in Transition • A useful reference, which although devoted to crystal spectra
Metal Complexes' Prog. Inorg. Chern. (1960) 2, 251. is also of value for solution work, is N. S. Hush and R. 1. M.
Hobbs, Prog. Inorg. Chern. (1968) 10, 259.
• 1. Ferguson 'Spectroscopy of 3d Complexes' Prog. Inorg.
Chern. (1979) 12, 158. • A more recent, more mathematicaL discussion is by K. Y.
Wong and P. N. Schaty 'A Dynamic Model for Mixed-
• C. 1. 10rgensen Absorption Spectra and Chemical Bonding in
Valence Compounds' Prog. Inorg. Chem. (1981) 28. 369.
Complexes Pergamon, Oxford, 1962. This book is a useful
source of data on individual species. • P. 1. McCarthy and H. U. Gi.idel ·optical Spectroscopy of
Exchange-Coupled Transition Metal Complexes' Coord. Chem.
• C. 1. Ballhausen Introduction to Ligand Field Theory McGraw- Rev. (1988) 88, 69.
Hill, New York, 1962. This book is something of a bible in
the field: Chapter 10 is relevant to the present chapter.
Probably the most useful current reference, one that contains
• A useful source is Volume 1 of Comprehensive Coordination not only theory but also data on a large number of species
Chemistry G. Wilkinson, R. D. Gillard and 1. A. McCleverty is A. B. P. Lever 'Inorganic Electronic Spectroscopy' Elsevier,
(eds.), Pergamon Press, Oxford, 1987, Chapter 6, 'Ligand Amsterdam, 1984.
Field Theory' by B. N. Figgis. A much earlier article by the same author gives a short.
simple but useful review of 'Charge Transfer Spectra of Tran-
• T. M. Dunn 'The Visible and Ultraviolet Spectra of Complex sition Metal Complexes' in.!. Chem. Educ. (1974) 51, 612.
Questions [Ni(H 2 0) 6 ] 2 + (Fig. 8.5). To what extent do these data conform

to those predicted by the rule of average environment" Data
8.1 An octahedral vanadium(III) complex has d-d bands at are in 10 3 em- 1
20000 and 30000 em - 1 Interpret these data in terms of (a) the
crystal field and (b) the ligand field models. [Ni(en) 3)f • 11.2 18.4 29.0
[Ni(en),(gly)] + 10.8 17.9 28.6
8.2 Whereas the central d-d band in the spectrum of
[Ni(en)(gly) 2 ] 10.5 17.3 28.1
[Ni(H 2 0) 6 ] ' + (Fig. 8.5) shows a splitting, the corresponding
band of [Ni(NH 3 ) 6 ]2+, at ca. 17 300 em_,, shows, at most, a
[Ni(gly>3r 10.1 16.6 27.6
slight asymmetry. That in [Ni(en),] 2 +, at 18400 cm- 1 , is quite
symmetrical. Do these two latter observations enable a dis- 8.5 Outline the spectral consequences of (a) the Jahn-Teller
tinction to be made between the two explanations offered in effect and (b) spin-orbit coupling for the first row transition
the text for the splitting observed for [Ni(H 2 0) 6 ] 2 •" metal ions. How would corresponding second and third row
ions be expected to differ"
8.6 Use the Tanabe-Sugano diagrams given in Appendix

7 to suggest transitions which might be candidates to provide
monochromatic emitted (laser) light. What is the feature
common to all such transitions?
.,
"c:
-e"'
0
U)
Ql
"c:
-e"'
0
U)
17400 24500 33 300
~
cm-1
Fig. 8.21 The d-d spectrum of [Cr(H 2 0) 6 ] 3 ".
8.3 Figure 8.21 shows the d-d spectrum of [Cr(H 2 0) 6 ] 3 +. Wavelength -

Provide an explanation for the bands observed. Note: this
Fig. 8.22 The charge-transfer spectra of (a) [CoCI 4 f-, (b) [CoBr4 ] 2 -
question can be answered at several levels. Simplest is to use and (c) [Col 4 ] 2 -.
the appropriate Tanabe-Sugano diagram in a qualitative
fashion. Deeper, is to use the d 2 discussion of the text together
with the hole-electron analogy. 8. 7 Figure 8.22 shows the charge transfer spectra of the ions
[CoC1 4 ] 2 - , [CoBr4 ] 2 - and [Col 4 ] 2 -. What is the corre-
8.4 The following are the positions of the d-d bands of spondence between spectra and species" Give reasons for your
some Ni 11 complexes. The assignments correspond to those of choice.
Magnetic properties of
9.1 Introduction
In the last chapter, in Section 8.4, it was suggested that the spin of an electron
is best thought of as meaning that the electron behaves like a tiny bar magnet.
When there are several unpaired electrons the spin degeneracies can be
thought of as resulting from the variety of ways of arranging bar magnets
side-by-side. Similarly, the orbital motion of the electron, the circulation of
charge around the nucleus, can be thought of as leading to a solenoid-like
magnet. These spin and orbital magnets will interact with an applied
magnetic field , so that when an atom or molecule is placed in a magnetic
field any spin degeneracy may be removed- the different resultant magnets
behave differently. Thus, a level which is orbitally non-degenerate but is a
spin doublet has this spin degeneracy split into two levels with slightly
different energies in a magnetic field (corresponding to the N~S and S ~N
arrangements). If there is orbital degeneracy this too may be removed by a
magnetic field. As we shall see, it is such splittings which determine the
magnetic properties of a complex. Note particularly that either or both of
spin and orbital degeneracies give rise to the magnetic effects. Sometimes
statements such as 'the number of unpaired electrons in a complex may be
determined from magnetic susceptibility measurements ' are encountered.
Whilst true within their own context, they should not be read as meaning
that in electron spin lies the sole source of magnetic effects.
The splittings produced by magnetic fields are very small, about I em- 1
for a field of 0.5 T, and, for the majority of cases, are proportional
to the magnetic field. Because the splittings are so small, any particular atom
or molecule may be in any one of the several closely spaced states resulting
from the splitting. For a macroscopic sample, however, there will be a
Boltzmann distribution between the levels; at room temperature kT is about
200 em - I so ample energy is available. Clearly if we are to be able to interpret
186 I Magnetic properties of transition metal complexes
experimental results, we must consider both the splittings and the Boltzmann
distribution over them. However, because the effect of the magnetic field is
so small we must consider any other interaction which involves energies
corresponding to more than about I em- 1 • This is because such interactions
will play a part in determining the ground state of the complex before the
magnetic field is switched on-this field will only cause the subsequent
splitting. Two such interactions-spin-orbit coupling and the presence of
low-symmetry components in what is otherwise an octahedral crystal
field-have already been discussed in outline. The latter effect can seldom
be neglected. One might think, for example, that an isolated [Co(NH 3 ) 6 ]3+
ion would be accurately octahedral. This is not so, for there is an incompati-
bility between the threefold axis of each NH 3 and the coincident fourfold
axis of the CoN 6 octahedron. Add to this the effect of the environment (and
this means 'do not consider an isolated molecule; in solution it will be
surrounded by a jumble of solvent molecules and in a crystal by anions-and
these will seldom respect its octahedral symmetry'), recall the Jahn-Teller
effect, remember that some metal-ligand vibrations will be thermally excited
down to quite low temperatures and that lattice vibrations will persist to an
even lower temperature, and it becomes evident that from the point of view
of magnetism, a regular octahedral environment is a rare, if not extinct,
species. For basically octahedral molecules, however, an octahedral model
is a good first approximation and may be refined to take account of the
above effects. For the majority of this chapter we shall confine ourselves
largely to such octahedral complexes.
Before we can proceed, some of the vocabulary of magnetism has to be
introduced. Closed shells of electrons have neither spin nor orbital degeneracy
and are represented by a single wavefunction. A magnetic field therefore
produces no splitting. It does, however, distort the electron density slightly,
in a manner akin to that predicted by Lenz's law in classical electrodynamics.
That is, effectively, a small circulating current is produced, the magnetic effect
of which opposes the applied magnetic field. Because there is no resistive
damping, the current remains until the magnetic field is removed. Molecules
with closed shells are therefore repelled by a magnetic field and are said to
exhibit diamagnetism or to be diamagnetic.
Suitably oriented magnets are attracted towards the magnetic field of a
stronger magnet and the same is true for any orbital magnet and the intrinsic
(spin) magnet associated with an unpaired electron. Molecules with unpaired
electrons are therefore attracted into a magnetic field 1 and are said to to
exhibit paramagnetism or to be paramagnetic. For any transition metal ion
which has both closed shells and unpaired electrons the diamagnetism of the
former and the paramagnetism of the latter are opposed. All that can be
measured is their resultant. Fortunately, the effect of paramagnetism is about
100 times as great as that of diamagnetism so that it takes a great deal of
the latter to swamp the former. Fortunately, too, it is found that the effects
of diamagnetism are approximately additive-each atom makes a known,
and approximately constant, contribution to the diamagnetism of a molecule
and so diamagnetism may be fairly accurately allowed for once the empirical
formula of a complex is known. This means that it is possible to deduce the
1 Exceptions to this statement exist at very low temperatures.

Classical magnetism I 187
intrinsic paramagnetism of a complex to an acceptable level of precision and

to compare it with theoretical predictions. Finally, a discussion in terms of
an isolated molecule requires that the (para)magnetic units do not interact;
that is, these units must be spatially well separated. The jargon is to say that
the sample must be magnetically dilute. The detailed theory of magnetically
non-dilute substances is beyond the scope of this book, although some
general comments on the topic are contained in Section 9.12. Actually,
measurements at very low temperatures have shown that some of the most
classic of coordination compounds exhibit a tiny amount of magnetic
coupling between the transition metal ions and so are slightly non-dilute.
Examples include K 3 [Fe(CN) 6 ], [Ni(enlJ](N03 lz, K 2 [Cu(H 2 0) 6 ](S04 h
and [Ni(H 2 0) 6 ]Cl 2 . At temperatures much above those of liquid helium
such compounds behave as magnetically dilute.
9.2 Classical magnetism

A theory of magnetism was developed long before the advent of quantum
mechanics. This section contains an outline of this theory and defines some
of the experimental quantities which may well be encountered.
When a substance is placed in a magnetic field, H, the total magnetic
induction within the substance, B, is proportional to the sum of H and M,
where M is the magnetization of the substance:
8 =f.1 0 (H + M)
where Jlo is the vacuum permeability (in the CGS system of units Jlo is unity,
in SI it is defined to be exactly 4n x 10- 7 N A- 2 ; note that N A- 2 is the
same as H m - 1-either may be encountered). This equation may be
written as
~ = f.lo( 1 + ~) = f.10 (1 + K)
where K is the volume susceptibility (the ratio of the induced magnetization

to the applied magnetic field). A more useful quantity is the susceptibility
per gram, X (chi), called the specific susceptibility, which is given by
K
xx-p
where p is the density of the substance. Alternative forms of X which are
encountered are XA and XM, the atomic and molar susceptibilities, respectively.
These latter are obtained by multiplying X by the atomic and molecular
weights, respectively, of the magnetically active atom or molecule. They are
the susceptibilities per (gram) atom and mole, respectively. The former is
no longer used but will be met in the older literature. Because our interest
is in paramagnetism, some correction has to be made for the underlying
diamagnetism, a topic dealt with in more length in Appendix 9. If such a
correction has been made, it is indicated by a prime, thus x~ and x;,. Now,
MV
pX (MW) X H
K
XM =
188 1 Magnetic properties of transition metal complexes
where (MW) is the molecular weight and Vis the molar volume. Therefore,
total magnetization per mole

XM = H
average magnetization per mole x NA

H
that is
(9.1)
where rn is the average moment per molecule and NA is Avogadro's number.

If we have a collection of identical molecules, each of magnetic moment
Jl and free to orient itself in a magnetic field, then in such a field
there will be some alignment but this will be opposed by the thermal motion
of the molecules. That is, the measured moment decreases with increasing
temperature although Jl itself is a constant. Langevin showed that in this
situation the average (measured) magnetic moment rn, and the actual
moment Jl, are related by:
- J1.2H
m=- (9.2)
3kT
a derivation of which will be found in almost every text on magnetism.

His derivation contains the assumption that B = JloH, so the theory can
only be expected to hold for gaseous molecules, although, as is commonly
done, we shall ignore this limitation. Combining the last two equations
we have
(9.3)
where C, the Curie constant, is equal to NJ..J1 2 f3R. The equation x = C/T is
known as the Curie law-susceptibility is inversely proportional to the
absolute temperature. Surprisingly, this law is obeyed rather well by many
liquids and solids, and in particular by complexes of the first row transition
elements. The origin of this general agreement is not clear, except under
rather special limiting conditions. The detailed quantum mechanical treat-
ment does not at all readily lead to a prediction of a CfT type of behaviour.
For complexes of the first row transition elements it turns out that the
agreement is because the spin-orbit coupling constants are comparable in
magnitude to kT; at low temperatures they do not follow the Curie law so
well. Rearranging the above equation we find:
(9.4)
Orbital contribution to a magnetic moment 1 189
It is convenient to express J1 in units of Bohr magnetons,2 jJ, and so we write

J1 = Jl,rrf1· Combining the last two equations we find that
(3RTzM)1;2
(9.5)
l'eff = NAfJ
Notice that, defined in this way, llorr is a pure number, the Bohr magneton
number, which refers to a single molecule. Although often met, it is not
strictly correct to call it the effective magnetic moment, and it certainly is
incorrect to express it in units of Bohr magnetons, although this is a usage
met all too often.
9.3 Orbital contribution to a magnetic moment

We now turn to the quantum mechanical approach to the phenomenon of
paramagnetism and discuss first the orbital contribution to a magnetic
moment. it is convenient at this point to make a few general statements which
will be explained by the subsequent discussion.
The phrase orbital degeneracy was used in Section 9.1, implying that for
individual cases one may or may not exist. If in the free ion-in the absence
of a magnetic field-there is orbital degeneracy (that is, if we have anything
other than an S state ion) the free ion has an orbital magnet. If the orbital
degeneracy is lost in a real environment-by chemical bonding or crystal
field effects-the orbital contribution to the total magnetic moment is said
to be quenched. If the orbital degeneracy is merely reduced, the orbital
contribution is partially or incompletely quenched. Orbital degeneracy,
however, although a necessary condition for an orbital moment, is not a
sufficient condition.
It may be recalled that in Section 8.4 the orbital moment was likened to
the magnetic effect produced by a current in a solenoid. An even better
analogy would be with the magnetic moment associated with a current
flowing around a circular ring of superconducting material. If the super-
conducting ring is rotated by 90"-or any other angle-about its unique
axis, one is left with a physically identical situation (for a solenoid the ends
of the wire coil would be in a different position after the rotation). The
requirement for a non-zero orbital-derived moment is similar. An orbital
2 The Bohr magncton is a fundamental quantity in the quantum theory of magnetism. Bohr's
explanation of atomic structure \Vas based on the assumption that the angular momentum of
an electron circulating about the nucleus of an atom is quantized and equal to nh 12n:, where n
is an integer and h is Planck's constant. That IS, nh/2n = ma 2 w, \Vhere m is the mass of the
electron. a the radius of its orbit and w its angular velocity in radians. The area of the circular
orbit is Ra 2 and the current to which the electron circulation JS equivalent is e x (c•J/27r). From
the theory of a current flowing through a circular loop of wire, the magnetic moment assoc1ated
w1th the circulating electron is equal to the product
er:.J ea 2 ro
current x area = - x l':a 2 = - -
2" 2
From Bohr's postulate, this equals
ea 2 nh he he
- x - -2 = n - = nji. where fi = -
2 2rcma 4rrm 4nm
That is, the magnetic moment is an integer times {J. \vhere fJ is the Bohr magneton. It transpires
that f3 is also the fundamental quantity in the modern qu<tntum mechanical treatment.
degeneracy is needed-and this only occurs when orbitals are unequally

occupied-but this orbital degeneracy must be such that there exist two or
more degenerate orbitals which can be intcrconvertcd by rotation about a
suitable axis. Consider the d,x and d,, orbitals in an octahedral complex,
shown in Fig. 9.1. Rotation by 90° about the z axis interconverts these two
orbitals, so that if an electron were initially in the d"' orbital it could circulate
about the z axis by jumping between the dzx and d,., orbitals alternately. This
circulation is equivalent to a current flowing and so it produces a magnetic
effect. The circulation may take place in either direction, clockwise or
anticlockwise, and in the absence of a magnetic field the two possibilities
are degenerate (this is implicit in the orbital degeneracy). On application of
a magnetic field the two directions of circulation have different energies, and
this is associated with the loss of orbital degeneracy in a magnetic field.
Four comments are relevant at this point. First, the dx,. and d,x orbitals
arc interconverted by a rotation about the x axis and the dx,. and the d,.,
by a rotation about the y axis, and so the discussion above is relevant to
these pairs also. Second, a more detailed analysis shows that an electron in
the d,, orbital does not have to jump 90° to get into the d,, orbital because
these two orbitals overlap each other. Note that this statement is not
incompatible with the fact that these two orbitals have a zero overlap
integral. A continuous range of rotations is allowed, so that the electron
cloud experiences no barrier to free rotation about the z axis. Third, the
Jahn-Teller theorem requires that the orbital degeneracy which has just been
invoked, actually, never exists! However, because they are largely non-
bonding, it is usually assumed that any Jahn-Teller splitting of the dx,, d,,
and d,x degeneracy is small and that the orbital contribution to the moment
is only slightly quenched. Physically, this means that an electron circulating
about the z axis in the d,x and d,, orbitals does experience a barrier to free
rotation but that this barrier is usually small relative to thermal energies.
Finally, the dx, and d,,_,, orbitals are interconverted by a 45o rotation
about the z axis. However, in an octahedral complex they are not degenerate
and so give rise to no magnetic effect. In a free atom, on the other hand,
they are degenerate and so contribute to the orbital magnetic moment.
Summarizing, it turns out that for octahedral transition metal ions, only
ground states of T 1 " and T 2 " orbital symmetries give rise to orbital-derived
z magnetic moments. E" ground states have no such moment.
A question which at once arises is 'can a complex which has no unpaired
electrons be paramagnetic just because of its orbital magnetism'? The answer
is yes; two examples are provided by the permanganate ion Mn04 and by
the ion [Co(NH 3 ) 6 ] 3 + This seems strange. There is no case in which the
ground state of a transition metal ion has orbital degeneracy without also
having a spin degeneracy. So, both of the two examples just given have
orbitally non-degenerate ground states and, of course, no spin degeneracy.
We would expect them to be diamagnetic. Without orbital degeneracy, how
can an orbital magnetism exist? The answer lies in the existence of an excited
Fig, 9.1 The circulation of electron density
about the z coordinate axis in an incompletely
state in each case which does have the required orbital degeneracy and yet
filled t2g shell of an octahedrally coordinated is a spin singlet. The magnetic field mixes some of the excited state into the
transition metal ion. The electron density may ground stale (or, equivalently, pushes a tiny bit of electron density into
be thought of as jumping from dzx to dyz to dzx
and so on, but see the text for a more
the excited state) and so the ground state assumes some of the properties of
accurate description. the excited. In particular, measurement shows that such a complex has
Spin-orbit coupling 1 191
a-small-orbital paramagnetism, which is sometimes large enough to

cancel out the inherent diamagnetism. Because this paramagnetism does not
depend on the thermal population of levels-unlike simple paramagnetism-
it is called temperature independent paramagnetism, or TIP for short. 3 TIP
is no mere academic curiosity. For diamagnetic transition metal ions it can
be very important in determining the position of NMR resonances (the Co 111
case is particularly well studied and is cited in the 'Further reading' at the
end of Chapter 12). In Co 11 octahedral complexes the magnitude of the TIP
is often very comparable to the correction that has to be made for the
underlying diamagnetism. In such cases, it is better to make no correction,
and rely on the approximate cancellation of the two effects, than to correct
for one but not the other.
9.4 Spin contribution to a magnetic moment

As has been mentioned earlier in this chapter, the phenomenon of electron
spin is best thought of as the electron behaving like a tiny bar magnet. This
bar magnet is essentially insensitive to its environment because the latter
imposes an electrical, rather than a magnetic, field. A single bar magnet can
be thought of as having two orientations with respect to an applied magnetic
field, parallel and antiparallel, and these have different energies; with a pair
of bar magnets the possibilities increase to three (both parallel, one parallel
one antiparallel, both antiparallel), again each of the three arrangements
have different energies. Higher spin multiplicities can similarly be given
simple physical descriptions. It is possible to derive a simple equation which
holds when the spin is the sole cause of the magnetic properties of a complex,
and this will be done in Section 9.10. Despite its simplicity, we cannot
immediately move to it because it is important to be aware of the
circumstances in which it is applicable. These are when other, complicating,
effects are absent. First, these complications must be explored. In the general
case, both orbital and spin motions will contribute to the magnetism
displayed by a complex; their effects may be additive or opposed. Further,
the spin and orbital magnets may interact with each other, the phenomenon
of spin-orbit coupling. This will be covered in the next section.
9.5 Spin-orbit coupling

Spin-orbit coupling has to be included in a discussion of the magnetic
properties of transition metal complexes whenever a spin-derived magnetic
moment and an orbitally-derived magnetic moment coexist. Spin-orbit
3 It is instructive to look at the Com case in more detail. The d electron configuration of
Com is t~,e~ and so the ground state is 1 A 111 • Earlier in the section it was found helpful to talk
in terms of the magnetic properties of a current flowing in a ring of supcrconducting material.
This analogy is helpful in the present context. In the point group Oh the set of three rotations
Rx, K,. and Rz transform as T 11r The analogy suggests that the components of the magnetic
field also transform as T 1 ,. This is, mdeed, the case. Now, if an excited state-call its symmetry
species r -is to be mixed with the ground state, 1 A 1 q, by the magnetic field, then the direct
product T 1 q x r must contain 1 A 11, the symmetry of the ground state. This only occurs when
r = T 1 q and so it has to be an excited 1T 111 state that gives rise to the TIP of Co 1ll species.
Reference to the d 6 Tanabe-Sugano diagram shows that such a low-lymg excited state does,
indeed, exist.
coupling splits terms with both spin and orbital degeneracy into a number
of sublevels. This phenomenon is relevant to the spectroscopic properties
of transition metal complexes and so in Table 8.2 were listed the number of
components obtained as a result of spin~orbit coupling.
There is a simple method by which one may discover whether spin--orbit
coupling mixes two d orbitals. For each d orbital, note the number of its
nodal planes which also contain the z axis (i.e.those in which the z axis lies
on a nodal plane). If two d orbitals differ by either one or zero in these
numbers then the two orbitals may be mixed by spin~orbit coupling. Using
this approach it follows that spin~orbit coupling mixes d orbitals as indicated
below. Numbers in brackets indicate the number of nodal planes which, for
that orbital, contain the z axis:
d,,(O)
/"""'-
dyz(l)---<lzx(l)
1X1
dxy(2)-dx'-y,(2)
The origin of this pattern will become more evident from the detailed
discussion that will be given Sections 11.4 and 11.5.
The magnitude of the spin~orbit coupling for a particular ion is usually
given in terms of one of two different so-called spin -orbit coupling constants,
( (zeta) and ).. The former is the one-electron spin~orbit coupling constant
and is useful when comparing the relative magnitudes of spin~orbit coupling
for different ions. In practice, for many-electron ions, what is measured is a
resultant spin~orbit coupling between the resultant spin magnetic moment
and the resultant orbital moment. It is this latter spin~orbit coupling
constant which is called ).. For ground states the two constants are simply
related:
where Sis the spin multiplicity of the ion and the plus sign refers to d 1 ~d 4
ions and the minus sign to d 6 ~d 9 For these latter, it is simplest to think of
holes circulating, the holes having the opposite charge to electrons. This
point is detailed in appendix 12, which contains a specific example, although
this particular appendix is specifically addressed to a problem that will be
encountered in Chapter II. Values for), (and thence 0 are obtained for free
ions from atomic spectral data and for complex ions from magnetic
measurements of the type discussed in this chapter.
9.6 Low-symmetry ligand fields

Because they can, and usually do, remove orbital degeneracies-and thus
reduce the orbital contribution to the magnetism-low-symmetry ligand fields
cannot be ignored in a study of the magnetic properties of transition metal
complexes. Unfortunately, it is no easy matter to determine the splitting
effects of low symmetry fields; usually it is necessary to work with two
Experimental results I 193
splitting parameters. Because the effects of low symmetry are so much more
important in magnetism than in, say, spectroscopy, the magnitude of the
parameters is best determined magnetically. However, the values of the
distortions may be small and so impossible to detect by any other method;
the distortion may not even be evident in a structure determination, for
instance. In this situation the magnitude of the parameter becomes something
of a 'fudge factor'-it is given the value that produces best agreement
between experiment and theory.
There is an important theorem due to Kramer, which states that when
the ground-state configuration has an odd number of unpaired electrons
there exists a degeneracy which a low symmetry ligand field cannot remove.
This degeneracy, usually known as Kramer's degeneracy, arises from the fact
that an orbital may be occupied by an electron in two ways-the spin may
be up or it may be down. In the absence of a magnetic field these two
orientations have the same energy. The application of a magnetic field causes
the two to differ in energy-by about 1 em - 1 -and it is the greater
occupation of the more stable in a macroscopic sample which causes the
sample to be attracted into a magnetic field. When there is an even number
of unpaired electrons a low-symmetry ligand field can relieve degeneracies,
but the application of a magnetic field then causes the lowest state to become
even more stable.
9. 7 Experimental results
If there were no orbital contribution to the magnetic moment of a complex
ion, one would have to worry a great deal less about the effects of spin-orbit
coupling and low-symmetry fields. Table 9.1 lists the moments that would
be expected for ions of the first transition series if there were no orbital
contribution (spin-only moments) and compares these with room temperature
experimental data. Also indicated in Table 9.1 is whether a significant orbital
contribution is to be expected (that is, whether there is a ground state T 19
or T 29 term). In Section 9.10 the spin-only equation used to predict the
moments given in Table 9.1 will be derived. On the whole, the agreement in
this table is not at all bad and if all one is interested in is whether a complex
is high or low spin then, if the oxidation state is known, a simple room
temperature magnetic susceptibility measurement4 on a tetrahedral or
octahedral complex of a first-row element may readily be interpreted using
the data in this table. A detailed analysis shows that the reasonable
agreement between spin-only and experimental moments shown in Table 9.1
is somewhat fortuitous. For d 1-d 4 ions, spin-orbit coupling has the effect
of reducing the observed moment and, roughly, cancels any orbital contri-
bution. For d 6 -d 9 ions, spin-orbit coupling increases the observed moment
and adds to the orbital contribution. This, together with the increased
magnitude of the spin--orbit coupling constants for these ions explains the
few gross disagreements between spin-only and experimental values. Because
of the very large spin-orbit coupling constants of elements of the second
and third transition series, their complexes usually have moments much
4 Appendix 8 contains both a description of how such mcasur~ments are made and the
treatment of the data obtained; how the diamagnetic corrections are made, for instance.
Table 9.1 Comparison of calculated spin-only moments and experimental data for
magnetic moments of ions of the first transition series
Ion Configuration• Orbital Theoetical Range of experimental

contribution spin-only values found at
expected? value room temperature
Octahedral complexes
Ti 3 + dl Yes 1.73 1.6-1.75
v•• dl Yes 1.73 1.7-1.8
v3+ d2 Yes 2.83 2.7-2.9
cr"• d2 Yes 2.83 ca. 2.8
v2• d3 No 3.88 3.8-3.9
cr"• d3 No 3.88 3.7-3.9
Mn 4 + d3 No 3.88 3.8-4.0
cr"• d4 hs No 4.90 4.7-4.9
cr"• d4 Is Yes 2.83 3.2-3.3
Mn 3 + d4 hs No 4.90 4.9-5.0
Mn 3 + d4 Is Yes 2.83 ca. 3.2
Mn 2 + d 5 hs No 5.92 5.6-6.1
Mn 2 + d 5 Is Yes 1.73 1.8-2.1
Fe3 + d5 hs No 5.92 5.7-6.0
Fe3 + d5 Is Yes 1.73 2.0-2.5
Fe2 + d6 hs No 4.90 5.1-5.7
Co2 + d 7 hs Yes 3.88 4.3-5.2
Co2 + d 7 Is No 1.73 1.8
Ni 3 + d7 Is No 1.73 1.8-2.0
Ni 2 + da No 2.83 2.8-3.5
Cu 2 + dg No 1.73 1.7-2.2
Tetrahedral complexesb
cr>• dl No 1.73 1.7-1.8
Mn6 + dl No 1.73 1.7-1.8
cr"• d2 No 2.83 2.8
Mn 5 + d2 No 2.83 2.6-2.8
Fe 5 + d3 hs Yes 3.88 3.6-3.7
unknown d4 hs Yes 4.90
Mn 2 + d5 hs No 5.92 5.9-6.2
Fe 2 + d6 hs No 4.90 5.3-5.5
eo2• d7 No 3.88 4.2-4.8
Ni 2 + d" Yes 2.83 3.5-4.0
a hs = high spin, ls = low spin.
b Note that low-spin tetrahedral complexes are very rare-if any exist at all-and are not included in this
table.
smaller than the spin-only values. We shall see why this is so in Section 9.9.
Lest it be thought that the sole effect of spin-orbit coupling is that of making
life more complicated, it should be mentioned that a large spin-orbit
coupling tends to reduce the sensitivity of the magnetic moment to low-
symmetry fields, leading to a more octahedral-like behaviour-provided that
the complex is approximately octahedral to start with.
An example 1 195
9.8 Orbital contribution reduction factor

So far in this chapter an essentially crystal field approach has been followed.
How has it to be modified to take account of covalency? One way is to let
the spin-orbit coupling constant vary, to become a parameter, rather than
giving it its free-ion value. Because a molecular orbital differs from an atomic
orbital, we would expect them to have different magnetic moments and also
different spin-orbit coupling constants. So, the t 2 , set of d orbitals can give
rise to an orbital contribution. What happens when it contains a ligand
component-which, in this case, has to ben? The arguments used to justify
an orbital contribution from the metal orbitals apply equally to the ligand n
orbitals-the appropriate ligand combinations can be interchanged by
suitable rotations, although it is unlikely that they will overlap with
each other in the same way. The real question, therefore, is whether there is
any reason to expect the ligand contribution to the moment to be equivalent
to that part of the metal contribution which it has repla~ed. There is no such
reason and so we must allow for the difference somewhere within our model.
The usual way is to multiply the orbital contribution, calculated for the free
metal ion, by a parameter, choosing that value of the parameter which gives
the best agreement with experiment. This parameter is usually denoted k and
is called the orbital reduction factor because it generally turns out to have
a value of less than unity.
9.9 An example
This Section gives an outline of an algebraic calculation of the magnetic
properties of a complex ion. The treatment is repeated in more detail in
Appendix 10. In real life, the calculations would be carried out by a computer
and some of the approximations that will be made here would not be
necessary. However, the language of the subject is so wedded to the algebraic
treatment that it is essential to study it, at least superficially.
The problem that will be considered is very simple, at least in principle.
It is that of an octahedral ti, complex. So, the theory is that of a strong field
complex of Ti 111 -strong field because the e9 orbitals are considered so high
in energy that they can be ignored. It would be good at the end of our
development to be able to say that the final equation gives an excellent
account of the magnetic properties of Ti 111 complexes. Unfortunately, it is
not possible to be so enthusiastic-indeed, it is possible that the theory is
fundamentally flawed, although it will only be possible to appreciate why
this is so after the theory has itself been developed.
The d, configuration gives rise to a 2 T 29 ground term-- three orbital
functions, each of which may be combined with either of two spin functions,
giving a total of six functions to be considered. Eventually, we will arrive at
a picture in which these six levels, although clustered together, are split apart.
We will be concerned with the Boltzmann distribution over these separated
levels and, in particular, the sensitivity of the distribution to the application
of a magnetic field. In this situation, it would clearly be ridiculous to consider
just the effects of the magnetic field and to ignore the larger effects of any
distortion and of spin-orbit coupling. For simplicity, a distortion from
octahedral which retains some orbital degeneracy will be considered. This
means a distortion which retains either a fourfold or a threefold axis. Both

are mathematically similar, only one additional low-symmetry field splitting
parameter being involved in either case. Of the two it is simpler to consider
the tetragonal case because we can then retain the coordinate axes of the
octahedron. Such a tetragonal distortion leads to the orbital triplet (t 2.)
splitting into an orbital singlet (b 2 •) and an orbital doublet (e.), the latter
two labels being appropriate to D4 h symmetry. We will assume that the
doublet is the lower in energy. Formally, we have two wavefunctions, a 2 B2 •
term, above a set of four wavefunctions, a 2 E 9 • Spin-orbit coupling splits the
2 E into two doublets. The final pattern is one with three distinct levels, each
9
doubly degenerate, each a Kramer's doublet. These double degeneracies can
only be split by a magnetic field.
All that remains is to remove the Kramer's degeneracies by the application
of a magnetic field. Here, there is a clear divergence between the computer
and algebraic solutions to the problem. In the algebraic approach, a
distinction is made between the effect of a magnetic field on a particular
level itself (the so-called first-order Zeeman effect) and its effect in mixing
different levels together (the second-order Zeeman effect). The first-order
Zeeman effect is proportional to the magnetic field strength, H, and the
second order to the square of the magnetic field. Computer calculations show
that this separation into terms proportional to H and to H 2 is only
approximate, although for the strength of magnetic fields that are used in
experimental work, the approximation is usually a good one.
The pattern of splittings that has just been described is shown in Fig. 9.2,
where it has been assumed that the tetragonal field and the spin-orbit
coupling produce effects of similar magnitudes. For clarity, in this figure, the
effects of the magnetic field have been greatly exaggerated. In order to relate
the energy level sequence shown in Fig. 9.2 to experiment, a value for the
average magnetic moment per molecule XM is needed {see Section 9.2). Now,
XM is
2:
all molecules
(magnetic moment of a molecule) x
(number of molecules with this moment)
XM = ------'-------------
(total number of molecules)
There is a Boltzmann distribution of molecules over the levels shown on the

right-hand side of Fig. 9.2, so that the relative number of molecules occupying
a level with energy E is exp(E/kT ), a quantity which can be used in the
expression above provided that we have explicit expressions for the energy
levels-and Appendix 10 provides them. The Zeeman part of the calculation
provides the value of the magnetic moment for the molecules in a particular
level. Having thus obtained a complete expression for XM, this can be related
to the quantity usually discussed, the Bohr magneton number llerr because,
as was shown in Section 9.2 (eqns 9.4 and 9.5)
(9.6)
Rather than work with the rather unwieldy equations resulting from the
expressions given in Appendix 10 we shall simplify. We ignore the tetragonal
An example 1 197
Fig. 9.2 Schematic energy level diagram

illustrating the various splittings determining the
ground state derived from a t~g configuration.
---<_______
===
Crystal
field
Tetragonal
splitting
Spin-orbit<
coupling First-order S d d
Z econ -or er
e~m~n Zeeman
e ec effect
Magnetic field
distortion. With this step we obtain the Kotani model, in which just the
effects of spin-orbit coupling and the magnetic field on a t}. configuration
are considered. We thus obtain an important expression, derived in reasonable
detail in Appendix 10:
3\ (-3\)
--8exp - - +8
~~ff = k~- e[ x p( -:;;

-+2
J
kT 2kT
or, by putting (/kT = x (a number) the equation simplifies to
2
(3x- 8) exp( -=F) +8
f.leff =
another equation first derived by Kotani.

We can now plot the value of Jl,rr (given by the positive square root of
the above expression) against x, or, more conveniently, against 1/x, to give
a so-called Kotani plot. This has been done in Fig. 9.3. Also on this figure are
indicated the values of (/kT for some d 1 ions at 300 K, ( being given the
Spin-only y4+
--------- -~--------------- --- _,_...---;.__...............__

Fig. 9.3 A Kotani plot for the t~g configuration
together with 300 K values for some th ions.
The spin-only formula gives ~.ff = 1. 73; 1
whereas first row transition metal ions roughly
approximate to this value at room temperature,
values for second and third row ions can be
very different.
0.5 1.0
kT 1
X
free-ion value. It is seen that /l.rr is almost independent of temperature at

high enough temperatures (the larger (, the higher the temperature needed);
room temperature measurements on first row transition metal ions give,
essentially, their 'plateau' values. It is under these limiting conditions that
simple formulae such as
!lett= Jn!n + 2)
the spin-only formula (which will be derived in the next section), in which
n is the number of unpaired electrons, become appropriate. For complexes
of the second and third row transition series the spin-only formula is not
applicable and magnetic measurements over a temperature range are
absolutely essential, even if it is only the number of unpaired electrons which
is to be determined. For measurements on such compounds at room
temperature, the plateau has not been reached.
For electronic configurations other than the one discussed above, anal-
ogous calculations lead to relationships which are roughly similar to that
shown in Fig. 9.3. /l.rr does not generally drop to zero at 0 K and, down to
i.jkT;:::; 1.5 it may increase slightly with decreasing temperature, decreasing
as the temperature is lowered further. If the model is not simplified and k
(the orbital reduction factor) and t (the factor describing the distortion to a
tetragonal field), or related functions, are included in the final energy-level
expressions, the temperature dependence of /l,H is less than in the corre-
sponding case in which they are omitted. As an illustration of this, in Fig.
9.4 is shown a comparison between the predicted and experimental results
for [VC1 6 ] 2 -, the cation being the pyridinium ion, C 5 H 5 NH+. Simple
Kotani theory is roughly followed if the spin-orbit coupling constant is
reduced to 190 em - 1 from the free-ion value of 250 em - 1 (Fig. 9.4(a)). Much
better agreement is obtained if distortion and covalency are allowed for using
the full energy level pattern of Fig. 9.2. This better agreement is shown in
Fig. 9.4(b). In that figure, the theoretical curve is that calculated fork= 0.75,
( = 150 em - 1 and t = 150 em - 1 in the expression given in Appendix 10.
The negative sign on t implies that the orbital singlet lies lowest, not the
orbital doublet-the distortion is the opposite to that assumed in Appendix
I 0 and, indeed, earlier in this chapter. The job of fitting a theoretical curve
An example 1 199
1.8
l
•
1.2
100 200
(a) Temperature (K) -
Fig. 9.4 (a) A simple Kotani plot for [VCI 6 ] 2 -

(( ~ 190 cm- 1 ). (b) A Kotani plot including
best-fit tetragonal distortions and orbital
1.8
reduction factors. In both cases experimental
data are given as dots.
"'
"-
1.2
100 200 300
(b) Temperature (K) -
to the experimental results is best done by computer; the need for high
experimental accuracy is evident.
This, then, is the algebraic model. What, if anything, is wrong with it?
Look again at Fig. 9.4(b ), where the best fit between the modified Kotani
theory and the experimental data is shown. The lower temperature points
seem to show a systematic divergence between experiment and theory. This
is not surprising, for as either a qualitative consideration of the effect of
decreasing temperature on the population of the levels in Fig. 9.2 or,
equivalently, a study of Fig. 9.4(a) (the Kotani plot) shows, the most severe
test of the theory is to be expected at the lowest temperatures, for here the
average magnetic moment has its greatest temperature dependence. The
low-temperature cut-off point in Fig. 9.4 is determined by the boiling point
of liquid nitrogen. What if the temperature is taken down to that of liquid
helium (3 K), an extension which is now normal. Inevitably, complications
set in! These complications can be many-faceted. Small distortions away from
true trigonal or tetragonal may become evident, as may long-range magnetic
coupling between what would otherwise be regarded as isolated magnetic
centres (some examples of this were given at the end of Section 9.1 ).
Fortunately, at such low temperatures it is possible to measure the changes
in thermal population of the magnetic-field split levels by another method,
that of specific heat measurements. Such additional data are not without
additional complications. In particular, the varying thermal population of
the multitude of lattice vibrations has to be taken into account. Fortunately,
at low temperatures this part of the specific heat of most solids has a T 3
dependence and may usually be allowed for with adequate accuracy. As this
example and as the data available in the literature make clear, high
temperature magnetic susceptibility data are of limited value-and some
would say that 100 K is high-and that much more is revealed at low
temperatures. An additional bonus is that at low temperatures it may be
possible to use electron paramagnetic (spin) resonance spectroscopy (EPR,
discussed in Section 12.6) to study transitions between levels such as those
of Fig. 9.2 and so to have independent measures of them. A limited number
of ions give room temperature EPR spectra which are well resolved but some
which do not are found to give quite sharp lines at low temperatures.
It is here that our circle closes. Such measurements have been made on
some Ti 111 salts and have shown the presence of a major energy contribution
which is not present in Fig. 9.2 and is also absent from the associated text
and Appendix 10. This contribution arises from a phenomenon that was
discussed in Section 8.5, the dynamic Jahn-Teller effect. Throughout our
development of the ti_ case, we assumed that the molecule under study was
rigid. The dynamic Jahn-Teller effect introduces (some, but not all) metal-
ligand vibrations into the picture. How important is all this? Well, it has
been emphasized earlier that the magnetic properties of a complex are very
sensitive to small distortions and therefore to vibrations. The lattice vibrations
mentioned above and in Section 9.1 do not distort molecules-in them the
molecules move as rigid bodies-and metal-ligand vibrations, also mentioned
in Section 9.1, will normally be frozen out at very low temperatures. The
dynamic Jahn-Teller effect is much less readily frozen out and so, indeed,
it is a potential complication. When it is included in the calculation of the
energy levels in the ti 9 case, patterns such as that in Fig. 9.5 are obtained.
The difference from Fig. 9.2, although small, is significant. Just as a
reasonably complete understanding of the electronic spectra of transition
metal complexes requires the inclusion of vibronic (vibrational-electronic)
effects so it seems likely that an understanding of magnetism will too. This
aspect of the subject is still in its infancy, although it is clear that there are
other aspects of the magnetism of coordination compounds in which vibronic
effects are implicated (the dynamic Jahn-Teller effect is just one form of
vibronic coupling).
The developments outlined in the above paragraph may well help to refute
criticisms of that theory of magnetism which has been developed in the last
few pages. When data from a large number of complexes were reviewed it
was concluded that
1. the theory produces too many ligand field parameters which, in the event,
often seem to have no obvious chemical relevance;
2. the models and parameters used for molecules with different geometries
seem to bear little relationship with each other; they do not vary with
geometry in a way that is obviously sensible;
3. molecules with very low symmetry could not be handled.
Since these criticisms were made the situation has improved, not only
because the need to include the dynamic Jahn-Teller effect has been
recognized, but also by theoretical developments. Rather than start with an
octahedron and distort it, the present tendency is to, theoretically, build up
Spin-only equation 1 201
,-
''
'''
'
''
'''
'
''
'''
'
' ''
:
==i
==\
''
''
''
''
''
''
'
\---
- _,/ ::
----
___
'
==',\
/
I
Crystal , /
field ', , ,/
\ ',,,
D , --------
~~~~c ',Tetragonal:,~~-'--<:,
effect
~~-======== ------
splitting
Spin-orbit ----
coupling First-order
Second-order
Zeeman
Zeeman
effect
effect
Magnetic field
Fig. 9.5 A schematic energy level diagram corresponding to Fig. 9.2 but with inclusion of a dynamic
Jahn-Teller effect.
a complex using geometry-sensitive but transferable parameters for each

ligand. Surprisingly, an economy of parameters can result. This approach
will not be developed here (although the Further Reading at the end of the
chapter contains a reference to it) because it tends to be molecule-specific
but the general aspects of it will be covered in the next chapter when the
angular overlap model is outlined. The reference in the Further Reading at
the end of Chapter 10 on modern developments of the angular overlap
model provides an entry into the literature on the subject.
9.10 Spin-only equation

The previous section has shown that even if there is an orbital contribution
to the magnetic moment it may be reduced by covalency and by distortions.
What if we assume that there is no orbital contribution at all-that it is
completely quenched? In this section this problem is considered and we will
be led to the spin-only equation mentioned several times earlier in this

chapter. The problem will be worked through in some detail to give the
reader who has decided not to tackle Appendix 10 some idea of how the
calculations are performed.
We are interested in the case in which there are n unpaired electrons, each
with spin -!, in an isolated ion which couple to give a resultant, denoted S,
where S = n/2. The allowed components of S along the direction (z) of a
magnetic field, when such a field is applied, are
Sz = S, (S- 1), (S- 2), ... , (S- (2S- 1)), (S- 25)
That is, the allowed components run from S through to-Sand are (2S +I)
in number. Their energies in a magnetic field will be proportional to the
magnetic field strength and to the magnitude of the z component of S, the
value of S,. So, their energies will be of the form -2S,f3H, the negative sign
indicating that the higher S, values are stabilized and the 2 being the Lande
g factor for the electron-the splittings in a magnetic field are twice as great
as one would expect from the magnitude of the spin angular momentum.
So, their energies are
-2Sf3H, -2(5- 1)[3H, -2(S- 2)[3H, ... , 2Sf3H
The average magnetization per molecule will be given by an expression of

the form
L (magnetic moment of a molecule) x (number with this moment)
Assuming that in a macroscopic sample there is a Boltzmann distribution

of ions over the various S, levels, we have that the number of ions
with moment 2S,f3H will be exp(2S,f3H/kT ). Using this and eqn 9.6 together
with eqn 9.1, we obtain
3kTNA
-s
2:2Sz X f3 X exp (25
_z_
[JH)
2 s,"s kT
l'ett =
NAf3 2
s
2: (25
exp _ z _
[JH)
s,"s kT
Simplifying and expanding the exponentials using the equation exp(x):::::; I + x,
L Sz ( 1
6kT -s 25zf3H)
+_ _
2 s,"s kT
s ( 2S f3H)
l'ett =
HfJ L 1+-z-
S,"S kT
This equation contains three standard summations:

-a -a a -s
LX= 0; 2: i' =-(a + 1)(2a + 1); 2: 1 = (25 + 1)
x=a 3 S,"S
so,
6kT 2f3H S
2
/lett= - X - X - X (S + 1) = 45(5 + 1)
Hf3 kT 3
Magnetically non-dilute compounds 1 203
That is,
llet1 = 2jS(S + 1) = Jn(n + 2)

As we have seen, this relationship holds quite well for the first row transition
series. For the second and third row elements it gives totally misleading
results because the spin-orbit coupling of these ions is much greater than
kT -so that, usually, the plateau of the Kotani plot has not been reached.
9.11 Magnetically non-dilute compounds

When a simple Curie law plot of XM against T _, (this pattern was discussed
in Section 9.2) does not give a straight line it is often found that a modified
form, the Curie-Weiss law, does:
c
ZM = T + 8
Here, 0 is a constant, the Weiss constant, which has the dimensions of

temperature, and is measured in degrees. Again, despite its widespread
applicability, it is difficult to give a quantum-mechanical interpretation of
this relationship. In the limiting cases where this is possible, 0 is associated
with a breakdown of the Langevin assumption that B = H, and values of
0 "# 0 are generally regarded as indicative of magnetic interaction between
discrete molecules in condensed phases.
This brings us back to an assumption made at the beginning of this
chapter, that we could restrict our discussion to magnetically dilute materials.
Magnetically non-dilute materials fall into one of three main classes. These
are ferromagnetic, antiferromagnetic and ferrimagnetic-terms that, strictly,
apply to extended lattices. In addition we will discuss the phenomenon of
ferromagnetic and antiferromagnetic coupling between individual ions,
leading to an understanding of the consequences of local magnetic coupling.
In Fig. 9.6 are shown diagrams representing paramagnetism, ferro-
magnetism, antiferromagnetism and ferrimagnetism, both with and without
a (very large!) magnetic field applied. In Fig. 9.7 is shown the temperature
behaviour of the magnetism associated with each type. Of the four classes
of magnetically non-dilute behaviour shown in Fig. 9.6, that of ferromagnetic
behaviour is well known-everyday magnets are made of ferromagnetic
materials. As with the other three classes, the individual magnetic centres
must be considered together, not separately, the correct building block being
the unit cell of the crystalline solid. In ferromagnetic materials, such as
metallic iron, the electron spins of each of the atoms couple together to form
a resultant unit cell magnetic moment. Iron has a cubic unit cell so let us
suppose that the unit cell moment is perpendicular to one of the faces of the
unit cell. There are three pairs of such faces and in a perfect crystal of
non-magnetized iron there are domains-within each of which there is
magnetic alignment between adjacent unit cells-but in the whole crystal
there is a random distribution over all of the possible orientations. The
process of magnetization brings these moments into alignment and a
resultant permanent moment results. These features make ferromagnetic
materials quite different from most of those which concern a chemist and
we shall not consider them further here, although we shall have something
to say about molecular ferromagnetism.
The second-and most important case for the chemist because of its
molecular counterpart, to which we shall turn shortly-is that of antiferro-
magnetism (although we will find that our discussion on this topic takes us
to molecular ferromagnetism). In antiferromagnetic materials, transition
metal ions are separated by (usually small) ligands, so that many transition
metal oxides and halides show antiferromagnetic behaviour. In such com-
pounds, adjacent metal ions couple with their spins antiparallel; there are
always equal numbers with the two arrangements so that there is no resultant
magnetization in the absence of a magnetic field. As an example of one
complicating feature of antiferromagnetism consider the sets of four two-
dimensional crystallographic unit cells, as determined by X-ray diffraction,
Magnetically non-dilute compounds 1 205
shown in Fig. 9.8. X-rays are blind to magnetism5 and so the magnetic centres
are represented by circles in the first diagram (ligands are omitted). We now
admit the existence of antiferromagnetic coupling between some pairs of
magnetic centres. In the other three sets of cells in this figure are given three
possible antiferromagnetic arrangements of spin orientations within the
block of four cells. It can be seen that these are all different; the bottom pair
of arrangements contain only two magnetic unit cells. Compared with the
32 crystallographic point groups and 230 space groups of classical crystal-
lography there are 90 crystallographic magnetic point groups and 1651
magnetic space groups. By using polarized neutrons (neutrons have an
intrinsic spin and so behave like tiny bar magnets; in a beam of polarized
FIJI. 9.7 Schematic temperature-susceptibility neutrons these magnets are all essentially parallel), neutron diffraction data
plots for various types of magnetic behaviour:
(a) antiferromagnetism (the arrow indicates the are capable of allocating an antiferromagnetic material to its correct
maximum which has sometimes been called the magnetic space group. If this is not known, it is not possible to give a
Curie temperature, Tc, or the NSel temperature, complete theoretical discussion of the magnetic properties of an antiferro-
TN); (b) paramagnetism; (c) ferrimagnetism;
(d) ferromagnetism. Note that this diagram does magnetic material.
not extend to very low temperatures (see Antiferromagnetic materials show a maximum in a plot of XM against
Question 9.5, for instance) or high magnetic temperature, at the so-called Nee! point-this is shown in Fig. 9.7. Below
fields (when saturation effects occur).
the Nee! point the susceptibility is to some extent field-dependent. These
properties provide an indication of the existence, or absence, of antiferro-
magnetism. Other tests include a comparison of the results of solution and
solid-state measurements, where this is possible, and, where it is not, by
the technique of the dilution of the magnetic ions in the lattice by their
partial substitution with an isomorphous non-magnetic ion. At temperatures
sufficiently above the Nee! point, antiferromagnetic materials follow a
Curie-Weiss law. Even if the Nee! point is at too low a temperature to
be measured conveniently, the observation that a material follows the
Curie-Weiss law usually implies a residual antiferromagnetic coupling.
The third class is that of ferrimagnetic materials. Just as for materials
0 0
t t
0 0 t t
0 0
t t
fill. 9.8 Crystallographic and magnetic unit
0 0 t t
cells. X-ray diffraction would give for all of these
systems the unit cell pattern of the top
left-hand diagram. Polarized neutron diffraction
would show doubled, magnetic, unit cells for
two of them.
5 This is a marginal overstatement. If the very intense X-ray beam from a synchrotron source
is used, weak X-ray magnetic scattering-which is relativistic in origin-can be observed. The
effect is rendered more observable by tuning the X-ray energy to coincide with an energy
difference in the material under study, when a resonance enhancement occurs.
showing antiferromagnetic behaviour, ferrimagnetic materials have ions on

two sets of lattice sites. These have opposed spin arrangements but as they
do not cancel each other out there is a resultant permanent moment. The
best known example is Fe 3 0 4 , the black mineral which used to be called
magnetic oxide of iron, but which is a complicated case because one site
contains (formally) Fe 11 and Fe"' in equal amounts, whilst the other contains
only Fem
A topic which has gained considerable attention is that of molecular
antiferromagnetism (because it potentially leads to the possibility of infor-
mation storage at the molecular level). This occurs when two transition metal
ions both have unpaired electrons and are bonded together, most usually by
bridging ligands. In such a case it is common to find that the magnetic
properties of the complex are not simply the sum of those of the two
individual metal ions. There is a magnetic interaction between them. This
situation has already been anticipated at two points in the text. At the end
of Section 8.11 it was commented that such magnetic interactions can lead
to spectral changes; at the end of Chapter 6 it was recognized that polarized
neutron diffraction measurements indicate that the distribution of electrons
in a ligand is affected by the unpaired electrons on the metal atom to which
it is bonded. The following discussion could be based on the details of Fig.
6.31 but it is simpler to consider a case in which there is bonding between
the metal orbital containing the unpaired electron and a ligand pure p
orbital. This is shown in Fig. 9.9(a). Because we are interested in a ligand
that bridges two transition metal atoms this is the situation pictured in Fig.
9.9(a). Start at the left-hand side of Fig. 9.9(a). The metal ion electron has
spin up. Although the ligand p orbital formally contains two electrons, we
have seen in Section 6.2.1 that it can nonetheless be involved in bonding
with the metal orbital. Such bonding means a pairing between the metal
unpaired electron and that of opposite spin in the p orbital. This leaves an
election in the p orbital which is of the same spin as that in the left-hand
metal d orbital. Pairing between this electron and the metal d electron in
the orbital at the right-hand side of the diagram requires that this latter d
electron has its spin in the opposite direction to that of the d electron in
the orbital at the left-hand side of the diagram. That is, the two metal ions
are antiferromagnetically linked by the ligand. In this example the two
metal ions were implicitly taken to be identical. What if they are different?
Suppose that one-the left-hand one-has its unpaired electron in an e9
orbital (we assume octahedral symmetry) and the other, the right-hand, has
its unpaired electron in a t 29 . Starting with the left-hand d orbital, the
argument follows that given above until we reach the right-hand d orbital,
an orbital which is orthogonal to (that is, has a zero overlap integral with)
the ligand p orbital. Unpaired electrons in orthogonal orbitals tend to align
themselves with parallel spins (examples of this can be seen by looking back
at Fig. 7.10)-this arrangement is exchange-stabilized. In Fig. 9.9(b), therefore,
the electron in the right-hand d orbital is of the same spin as that in the
left-hand. The two metal ions are ferromagnetica/ly linked.
Key to the development of ferromagnetic coupling in the argument above
was a step involving the orthogonality of two orbitals. The orthogonality
need not be between metal and ligand orbitals. In Fig. 9.9(c) is shown an
interaction pathway which includes two p orbitals on the same atom; these
Magnetically non·dilute compounds 1 207
Fig. 9.9 (a) Ligand a-orbital-mediated

antiferromagnetic coupling; (b) and (c) show
two ways in which orbital orthogonality can lead
to ferromagnetic coupling.
(c)
p orbitals, of course, are mutually orthogonal. Again, ferromagnetic coupling

results; in Figs. 9.9(b) and 9.9(c) the point at which exchange stabilization
leads to parallel spins is indicated by double-headed arrows. It would be
sensible for the reader to stop reading at this point and to trace through the
pathway of Fig. 9.9(c) and check that a ferromagnetic interaction, indeed,
results.
Clearly, by careful choice of metal ions and bridging ligands (which have
to be sufficiently rigid to provide rather fixed geometrical relationships
between the interacting orbitals) an orthogonal step can be introduced so
Fig. 9.10 (a) A Cu 11-VO complex showing

ferromagnetic interaction between the copper
and vanadyl ions. The ligand skeleton which
holds the two ions together is shown dotted.
(b) On the d9 ion Cu 2 • an unpaired electron is
in an eg type of orbital; on the d1 ion V0 2 + an
unpaired electron is in a t>t type of orbital.
These orbitals are, respectively, antisymmetric
and symmetric with respect to the vertical
mirror plane shown as a solid line, and are
therefore orthogonal.
(b)
that ferromagnetic coupling occurs between two metal atoms, leading them
to have parallel spins. An example of such a ligand is shown in Fig. 9.10.
Although with cunning choice ofligand and metal ion it is possible to obtain
ferromagnetic complexes such as that shown in Fig. 9.10, in most cases the
coupling between two linked metal atoms, each with unpaired electrons,
will contain at least one antiferromagnetic coupling pathway-a pathway
involving non-orthogonal orbitals-and, perhaps, a ferromagnetic coupling
through another orbital sequence. In such cases, the antiferromagnetic
coupling almost invariably dominates. Lastly, an important word, none the
less important for having been left until the end. The discussion of the last
few paragraphs hinged on the effects of exchange between electrons on
different atoms. Irrespective of whether the outcome is ferromagnetic or
antiferromagnetic coupling, the general mechanism involving ligand orbitals
to mediate coupling between metal electrons is referred to as superexchange.
Finally, although strictly out of place in this chapter, we mention
high-temperature superconductors. They are introduced at this point because,
as in superexchange, superconductivity involves the properties of electrons
on different atoms being correlated with each other. At the time of writing
this book the detailed mechanism by which high-temperature superconduc-
tivity occurs is not known. The interested reader will find the problem
explored in Appendix 11; they should be warned, however, that this appendix
is likely to date rather rapidly!
9.12 Spin equilibria

In the final section of this chapter we briefly review some phenomena which,
although involving no new concepts, nonetheless lead to some unexpected
magnetic results. In Chapter 3 we met two forms of isomerism which have
magnetic implications. Because the splitting of d orbitals is a function of

coordination geometry, a change in geometry may well mean a change in
the number of unpaired electrons. So, the existence of geometrical isomerism
(discussed in Section 3.4.2) may well be magnetically evident (it will be
spectrally apparent also). Many examples are provided by Ni" complexes;
this ion forms many square planar complexes-which are diamagnetic-
which may isomerize to give tetrahedral species. The latter, being d 8 , have
two unpaired electrons; the reverse isomerization may also occur, of course.
This pattern of behaviour is particularly common for Ni" complexes of
general formula [NiP2 X 2], where X is a halogen and P is an organophosphine
ligand. Typically, the square planar form is red or brown and the tetrahedral
green, the Ni" in the latter having a magnetic moment of ca. 3.5 (see Table
9.1).
The second form of isomerism which has magnetic implications is spin
isomerism (see Section 3.4.12). This occurs because with the correct choice
of ligand (and much effort has been spent on ligand design and variation)
the value for 11, for an octahedral complex can be made to be close to that
appropriate to the change-over point from high to low spin complexes in
d 4 , d 5 , d 6 and d 7 systems (see the Tanabe-Sugano diagrams in Appendix 7
for an idea of the magnitude of ligand field required). Most known examples
come from the 3d series-Mn", Mn 111, Co", Co 111 and, particularly, Fe" and
Fe111 • Because the two spin states in equilibrium involve different t 2
populations, the metal-ligand bond lengths would be expected to vary
.-e.
between the two spin isomers. X-ray measurements show that the M-L bond
lengths tend to be roughly 0.2 A longer in the high spin form because the
e. orbitals are weakly antibonding and have a higher occupancy in the high
spin isomer. Not surprisingly, therefore, the spin equilibria are pressure
sensitive, as well as sensitive to the choice of counterion, any solvent of
crystallization as well as steric effects originating in the ligands. Many studies
have been made of the lifetimes of each of the species in equilibrium. These
lifetimes are about 10- 7 s, but vary by at least one order of magnitude in
either direction. Often, the magnetic susceptibility changes on warming a
sample are not opposite of those on cooling, this hysteresis indicating some
cooperative, perhaps domain, behaviour. Although, magnetically, both spin
states are unexceptional, the study of the equilibria between them has been
energetically pursued by many workers and has involved the use of almost
all of the techniques described in this book (especially those described in
Chapter 12), and more.
Further reading Chemistry J. Lewis and R. G. Wilkins (eds.), Interscience,

New York, 1960.
The literature on this subject tends to divide between the old,
which provides the best available treatment of the classical • 'The Magnetic Properties of Transition Metal Complexes'
(including the classical quantum mechanical), and the recent, B. N. Figgis and J. Lewis, Prog. Inorg. Chern. (1964) 6, 37
which provides an up-dating of the earlier. Older are:
• A useful review which also includes an extensive compilation
• Magnetism and Transition Metal Complexes F. E. Mabbs and of diamagnetic corrections is 'Magnetochemistry-Advances
D. J. Machin, Chapman & Hall, London, 1973. in Theory and Experimentation' C. J. O'Connor, Prog. Inorg.
Chern. (1982) 26, 203
• 'The Magnetochemistry of Complex Compounds' B. N.
Figgis and J. Lewis, Chapter 6 in Modern Coordination • A still-current view of theory, experiment and problems is
• 'A Local View of Magnetochemistry' M. Gerloch, Inorg. temperatures' J. E. Rives, Transition Met. Chern. (1972) 7, I.
Prog. Chern. (1979) 26, 1. A more general account is 'Magnetic Symmetry' by W. Ope-
chowski and R. Guccione, Chapter 3 in Magnetism, Volume IIA,
Problems in the effects of magnetic coupling are reviewed in G. T Rado and H. Suhl (eds.),Academic Press, New York, 1965.
'Magnetochemistry: a research proposal' R. L. Carlin, Coord. More recent, and concentrating on systems of two transition
Chern. Rev. (1987) 79, 215. metal ions, is 'Magnetism of the Heteropolymetallic Systems'
An excellent current book is Molecular Magnetism 0. Kahn, 0. Kahn, Struct. Bonding (1987) 68, 89.
VCH, Weinheim, 1993; a rather different account is 'Organic Spin equilibria are covered in
and Organometallic Molecular Magnetic Materials-Designer
Magnets', a very long review by J. S. Miller and A. J. Epstein in • 'Dynamics of Spin Equilibria in Metal Complexes' J. K.
Angew. Chern., Int. Ed. (1994) 33, 385. Beattie, Adv. Inorg. Chern. (1988) 32, I.
A detailed account of magnetic phases and phase transitions • 'Static and Dynamic Effects in Spin Equilibrium Systems'
at low temperatures is 'Magnetic phase transitions at low Coord. Chern. Rev. (1988) 86, 245.
Questions 9.5. In Fig. 9.9 and the associated discussion within the text,
the ligand orbital involved in superexchange was a <I orbital.
9.1. Show that the metal d orbital and the ligand orbitals
<I Ligand n orbitals may also be involved. Figure 9.12 is the ligand
that contain unpaired spin density in Fig. 6.31 are orthogonal n orbital equivalent of Fig. 9.9(a). Give an account of Fig. 9.12
to each other. (Hint: consider their symmetries.) which parallels that given in the text for Fig. 9.9(a).
9.2. Show that the interactions shown in Fig. 9.9(c) lead to a

ferromagnetic (parallel spin) coupling of the two metal ions.
9.3. If one step involving orthogonal interactions leads to a
ferromagnetic interaction, then two such steps should lead to an
antiferromagnetic. Check this conclusion by replacing one of
the d orbitals shown in Fig. 9.9(c) by a t 2 • d orbital and tracing
through the coupling pathway.
9.4. Figure 9.8 does not exhaust the possible two-dimensional
antiferromagnetic arrangements of two magnetic ions; three
more are shown in Fig. 9.11. However, unit cells similar to M L M
those of Fig. 9.8 are not given. Insert them. To complete this
task for the third diagram will require the recognition of a Fig. 9.12 Question 9.5.
feature not discussed in the text.
9.6. There is a, perhaps, unexpected result contained in Fig.
9.4. At 0 K a system with a single unpaired electron is predicted
to be non-magnetic. Give a qualitative explanation for this
behaviour (remember-there are two sources of magnetism,
orbital and spin).
Fig. 9.11 Question 9.4.

Beyond ligand field theory
In this chapter an overview will be presented of many of the theoretical

methods currently being used to obtain insights into the electronic structure
of coordination compounds. The discussion is rather general but it is helpful
to focus on one area. Some attention is therefore given to transition metal
organometallic and related complexes, this class being selected because
nowhere else in this book are they explicitly treated. The first part of the
chapter therefore provides a background by outlining the simplest, mostly
symmetry-based, treatment of their structure.
10.1 Bonding In transHion metal organometallic

complexes
The present-day qualitative approach to the bonding in transition metal
organometallic complexes is based on a combination of two models.
The first, the IS-electron rule, has already been met in Chapter 2 (see Table
2.8)- ' ligands that bond to a transition metal in a low valence state normally
do so in such a way that the metal atom is, formally, surrounded by
IS-electrons'. Of course with a few exceptions, the classical complexes of the
Werner type and which are the subject of most of this book do not obey
this rule, so that it is clear that much hinges on the low valence state
requirement. Equally, the fact that most transition elements form neutral
bis-17 5 -cyclopentadiene complexes, M(C 5 H 5 h, shows that even for organo-
metallic complexes of transition metals it is only an approximation. Many
examples of molecules with more, and many examples with fewer, than 18
valence-shell electrons are known. Those with more tend to be readily
oxidized, those with fewer tend either to be sterically hindered or readily
add further ligands.
The second model, which is widely applied to organometallic complexes
of transition metal ions, is the requirement that the bonding interactions are
symmetry-determined. A very common and important case is provided by
the bonding of cyclic conjugated C.R. systems-of which C 5 H 5 is perhaps
the best known example. These are discussed in more detail in Appendix 13
212 1 Beyond ligand field theory
for the cases in which n = 4, 5 or 6, where two apparently rather different

Table 10.1 The C-C bond distance in
C2 H4 and its complexes. It is salutary to approaches to the problem are described. The approaches are actually
note that Zeise's salt, for which the extensions of the Chatt-Duncanson model for the bonding in Zeise's salt,
Chatt-Duncanson bonding model was described in Chapter 2-see Fig. 2.2 and the associated text. In general, the
first suggested, shows the smallest Chatt-Duncanson model is supported by the experimental data-see, for
effect. The last complex is one in which instance, Table 10.1. It therefore seems reasonable to extend it to other,
C2 H4 acts as a bridge between two Zr
and so is bonded to both related organic ligands. In this extension of the model, the lower nodality,
occupied, n orbitals of the organic ligand donate electron density to the
Molecule C-C distance (A) metal atom whilst the higher nodality, unoccupied, n orbitals accept electron
density from the metal orbitals (which, because of the higher-nodality
C2H4 1.34
[Pt(C2H4)CI,r 1.35 requirement, will be d orbitals). It would be expected that of particular
V(C2H4)(PMe3)2(C5H5) 1.36 importance in this bonding will be the ligand HOMO as electron donor and
Fe(C2H4)(PEt3 ), 1.38 the ligand LUMO as electron acceptor. Because of the high symmetry of
Ni(C2H4)(PPh3 ), 1.43 the isolated, planar, ligand, many of the n orbitals occur as degenerate pairs.
Pt(C2H4)(PPh 3 ) 2 1.43
Ru(C2H4 )(PMe3)4 1.44
When two such ligands are simultaneously coordinated then the normal
Fe(C2H4)(C0) 4 1.46 pattern is that the ligand n orbitals again occur in pairs, the sum and
Zr2(C2H4)CI6 (PEt3 ) 4 1.69 difference of the corresponding n orbitals of the individual ligands (the
sum and difference would normally have different symmetry labels). Overall,
then, we have pairs of pairs. If one of the sum and difference pairs were more
extensively involved in the bonding than the other, the molecular orbitals
would be separated in the energy level diagram-although it is usually not
difficult to recognize such pairings. Figure 10.1 shows such a qualitative
energy level diagram for ferrocene, Fe(C 5 H 5 )z. Care will have to be taken
using this diagram because, to avoid unduly complicating it, not all of the
energy-level tie lines have been included. 1 The pairs of pairs just discussed
appear on the left-hand side of this figure. In Section 10.7 this diagram will
be compared with the results of some detailed calculations. Figure 10.2 shows
in more detail, the general bonding pattern for metal-Cn(ring) complexes,
n = 4, 5 and 6, in which the matching of ligand and metal orbital nodal
patterns is emphasized (this is a topic discussed in some detail in Appendix
13 and is summarized in Fig. 10.2). The fact that a ligand can bond
symmetrically does not mean that it does bond symmetrically: Fig. 10.3 gives
some of the ways that benzene, C 6 H 6 , has been found to bond to transition
metals-it can be seen that it can function as a two-electron, a four-electron
or as a six-electron donor; in such cases the electron count would normally
be that which satisfies the 18-electron rule. In addition, many complexes are
known in which the benzene molecule bridges two or more transition metal
atoms; in these the benzene is seldom symmetrically bridged. Such asymmetric
attachments of hydrocarbon ligands is even more probable for chain-
conjugated hydrocarbons and related species, two examples of which are
given in Fig. 10.4. However, as this figure shows, there is a persistence of a
pattern of n orbitals which can be ordered according to the number of nodal
planes, even when, as in the examples shown, the introduction of a
hetero-atom (here, an 0 in place of a CH 2 group) produces considerable
changes in the detailed forms of the ligand n orbitals.
1 In the gas phase ferrocene has an eclpsed (D,.) geometry; in the crystal it is staggered (Ds;).
The energy difference between the two is small (ca. 4 kJ mol- 1 , 0.8 kcal mol- 1 ). Although the
symmetry labels change between the two, the various interactions involved in the bonding are
little different. In this chapter the staggered configuration will be the one discussed because most
of the theoretical and experimental work invoked assume this geometry.
Bonding in transition metal organometallic complexes I 213
(CP)2
FIJt. 10.1 Schematic molecular orbital energy An example in which the IS-electron rule and the symmetry-determination
level diagram for ferrocene. A more accurate
diagram is given in Fig. 10.9. (Adapted and
of interaction patterns come together rather strongly is in the molecule
reproduced with permission from Shriver, hexacarbonylchromium, Cr(COk This is an octahedral complex, to which,
Atkins and Langford, Inorganic Chemistry). in principle at least, all of the relevant arguments of Chapters 6 and 7 may
be applied. That the IS-electron rule is satisfied is easily seen; a Cr 0 species
has six valence-shell electrons-which will fill the t 2 " set of d orbitals--and
each CO ligand contributes two (J electrons from each carbon (formally,
these electrons are (J-donated to the metal), giving a total of 18. It is perhaps
not surprising that Cr(C0) 6 is colourless; CO, like CN-, is a strong field
ligand (we met this in Section 6.2.2) and so the t 2 , ---+ e, transitions fall in
cb I
M
Metal orbitals with 0 nodes
FJC. 10.2 A classification of orbitals according

to the number of their nodal planes which Metal orbitals with 1 node
contain the z axis for metal and planar cyclic
Cn Rn systems. The z axis is the axis of highest
rotational symmetry (top).
(a) Different metal orbital sets differ in the
nodality patterns they span: u orbitals only have
zero, n orbitals zero and one, {J orbitals, zero,
one and two. (continued)
Px
Metal orbitals with 2 nodes
{a)
~~ dx2- y2 dxy
the near-ultraviolet region of the spectrum. This, then, is a molecule for which
Figs. 6.11 and 6.16(c) must be brought together, and this is done in Fig. 10.5.
It would be worthwhile for the reader to stop at this point and examine in
some detail the connection between these three figures. One hidden problem
relates to the simple picture of COn bonding given in Section 2.2 and shown
in Fig. 2.1. The simple picture has, effectively, two electron pairs back-
donated from metal d orbitals into the n antibonding orbitals of the CO.
The group theory associated with Fig. 6.16, in which it was shown that the
12 ligand n orbitals transform as t 19 + t 1• + t 29 + t 2 ., is to be compared
with the symmetries of the available metal orbitals; of these four sets, only
two, t 1• + t 29 , find matching partners in the metal set. So, the simple picture
Metal-fullerene complexes I 215
............................................................................................................................................................................................
n= 3 4 5 6
~+ ~~+ +- + + __ +
+ + + + + + 0 nodes
- - - -
- - - - - -
(2)
fl- ~+
-
+
- (2)
-
+
+
-
+
+
- (;)
-
1 node
~~$-b 2 nodes
~~~ (2) (2)
(b)
FIJI. 10.2 (continued) (b) Different planar overestimates the extent of n back-bonding. In Section 10.7 the qualitative
cyclic CnRn ring systems differ in their p, orbital picture of bonding in Cr(C0) 6 will be compared with the results of detailed
nodality patterns: n = 2, zero and one; n = 3,
also zero and one. However, there are two calculations.
different orbitals with one nodal plane
containing the z axis (indicated by the (2) in the
figure). The nodal plane shown is that parallel 10.2 Metal-fullerene complexes
to the plane of the paper; the other orbital has
its nodal plane perpendicular to the plane of In recent years some considerable excitement has been generated by the
the paper. The case n = 4 has zero, one (two
of) and two nodal planes; the case n = 5 has discovery of a series of carbon cage compounds, generated in low yields
zero, one (two of) and two (two of). The case when an electric arc is struck between carbon rods in a low pressure of a
n = 6 has zero, one (two of), two (two of) and gas such as argon. Although by no means the only species formed, there has
three. The case n = 7 (not shown) has zero,
one (two of), two (two of), three (two of) etc. been particular study of C 60 and C 70 , the major products of the preparation.
Of these, C 60 has an icosahedral structure and, because of the general
resemblance to the geodesic domes designed by Buckminster Fuller, has
come to be known as buckminsterfullerene; it is shown in Fig. 10.6. The
whole family has come to be called fullerenes. In contrast to C 70 , which has
a D 5 h structure and five structurally different types of carbon atom, all of
the carbon atoms of C 60 are structurally equivalent. Our discussion will
therefore concern this molecule alone. The molecule is just over 7 A in
diameter, the central hole has a diameter of half of this. It is therefore large
enough to accommodate an atom and such metal complexes have been
prepared. They have been called endohedral complexes (as opposed to
Modes of six-electron donation
~ ~
~
I M--M
~
M
Modes of four-electron donation
~
M
M
I Q M
Q M--M
Mode of two-electron donation
9 M
Fig. 10.3 The benzene molecule can function as a six-electron, a four-electron and a two-electron
donor. It does not necessarily remain planar and may be involved in donation to more than one metal.
The model of donation is usually clear from a crystal structure determination and, in all probability,
also by application of the 18-electron rule. Examples of relevant molecules are, left to right:
six-electron donation-[Mo(~ 6 -C 6 H 6 )(C0) 3 ], [V2 (~ 3 .~ 3 -C 0 H 0 )(~ 5 -C 5 H 5 },H 2 ],
[Os 3 (~ 2 , ~ 2 , ~ 2 -C 0 H 6 )(C0) 9 ];
four-electron donatlon-[Os(~ 4 -C0 H0 ) ~°C 0 H 0 )], [ (Re(~ 5 -C5 Me 5 ) (C0) 2 ) 2 (~ 2 , ~ 2 -C 0 H 0 )],
[Pd 2 (~ 2 , ~ 2 -C 0 H 0 ) 2 (AIC1 4 ) 2 ];
two-electron donatlon--[(~ 2 -C 0 H 0 )Re(C0) 2 (~ 5 -C 5 Me 5 )].
exohedral complexes in which a metal is attached externally to the carbon

cage. The bonding in these latter is similar to that discussed above and in
Chapter 2).
A symbol which has been suggested to denote endohedral complexes is
@; so, La@C 60 . The method of preparation is scarcely sophisticated;
the carbon rods used in the preparation of C 60 are impregnated with a salt
of the metal and a small amount of the metal complex is formed in parallel
with the preparation of C 60 . The method is reminiscent of the use of similar
impregnated carbon rods to produce luminous arcs in the early days of
cinema projection-it may be that this was when the complexes were first
produced! Endohedral complexes have been prepared for carbon cages C 28
(U @C 28 ) and upwards; for the larger cages it has proved possible to
incorporate up to three metal atoms endohedrally, as in Sc 3 @C 82 . However,
it is found that such complexes are more difficult to prepare the more d
electrons the endohedrally incorporated element possesses; most of the
compounds that have been prepared are of the lanthanides, although barium
and strontium compounds have also been reported. It is not difficult to see
why this should be. The metal atom in such a complex must be stabilized by
interactions between its orbitals and the orbitals of the carbon atoms which
point towards the centre of the icosahedron. The latter will be of a a type
(viewed from the metal at the centre of the icosahedron) and, for C 60 , 60 in
Metal-fullerena complexes I 217
CH2=CH-CH=CH-CH=CH2
3 3
2 2
1 1
'
'
~0
(a)
~0
(b)
FIC. 10.4 P. molecular orbitals of some linear number. However, they are different when viewed from the perspective of
conjugated molecules in the HOckel
approximation (this is a crude, but adequate,
the rings of carbon atoms that make up the C60 cage, each ring being
model). Nodal planes are shown dotted and approximately planar. From this point of view, it is a reasonable approxi-
the total indicated. The more nodes, the higher mation to regard them as the counterparts of the n orbitals of the organic
the energy. (a) Butadiene (in the Huckel
approximation the cis and trans species are the
ring molecules discussed earlier in this chapter. They can be treated similarly
same). (b) Acrolein-this molecule may be and the MOs to which this gives rise ordered in energy sequence according
regarded as butadiene in which a terminal CH2 to the number of inherent nodal planes that they possess. This has been done
group has been replaced by 0. The detailed
consequences for the p. molecular orbitals are
in Fig. 10.7.
considerable but the general pattern is Because of the high symmetry of the icosahedron, higher even than that
unchanged. (c) Hexatriene-comparison of this of the octahedron, orbital degeneracies of four and five occur in icosahedral
pattern with (a) indicates the nature of the
general pattern.
molecules and this will be seen in Fig. 10.7, along with their symmetry labels
in the icosahedral group (although these labels add nothing to a discussion
except at a rather detailed level). Paralleling the organic rings discussed
earlier and in Appendix 13, the nodal patterns of the C 60 cage orbitals have
to match those of the orbitals of the endohedral atom. Fortunately, again
because of the near-spherical symmetry of the icosahedron, this is rather
simple (and can be shown to be so by looking at the detailed significance
of the symmetry labels in Fig. 10.7). The C 60 no-node combination interacts
4p===~
4s----~
eg(2)
3d
eg eg(1)
81g
Metal cr-only cr+n Ligand cr,
orbitals molecular molecular nand
orbitals orbitals somerr'
group
orbitals
Fig. 10.5 A qualitative a + n bonding exclusively with the s orbital of the endohedral atom, the three C 60 one-node
molecular orbital energy level diagram for
Cr(C0) 6 based on Figs. 6.11 and 6.16(c)
combinations interact with the endohedral atom's p orbitals, the five C 60
together with the associated discussion. This two-node combinations interact solely with the endohedral atom's d orbitals,
schematic diagram may be compared with Fig. the seven C 60 three-node combinations (which split into two little-separated
10.10 which is a more accurate molecular
energy level diagram for Cr(C0) 6 •
sets) interact with the endohedral atom's f orbitals, and so on. Now, just as
in the neutral hydrocarbon rings discussed earlier, the total number of
electrons to be accommodated in the C 60 orbitals is equal to the number of
atoms, 60. So, the bottom 30 orbitals are full, and this is indicated in Fig.
10.7. This means that the no-node, the one-node, the two-node and the
three-node orbitals (and 14 others, too) are all full. When they interact with
an endohedral metal atom the only interactions that can contribute a
stabilization are those with empty endohedral atom orbitals. When an
endohedral atom has occupied d orbitals in its valence shell then those d
electrons are forced to occupy antibonding orbitals if interactions involving
the d orbitals contribute significantly to the molecular stability. Similar
arguments apply to the s and p orbitals, of course.
Metal-fullerene complexes 1 219
The experimental observation that the only atoms which form stable
endohedral complexes are those with empty orbitals is at once explained.
The content of Chapter II suggests that the f electrons of the lanthanides
are expected to be too well tucked away inside the atom to be much involved
in bonding and so, for this reason, the above discussion does not apply very
strongly to the C 60 three-node orbitals; Fig. 10.8 illustrates this argument.
Although this seems reasonable, some recent rather detailed calculations on
Ce@C 28 suggest that at least some f orbital covalency may exist, although
not enough to alter the above general conclusions. Hints that f orbital
covalency is not unreasonable will be found in Chapter 11, at the end of
Section 11.2. Finally, it should be noted that there is a similarity between the
fullerenes and the metal clusters which form the subject of Chapter 15. The
methods which have been used to describe the bonding in these clusters are
also applicable to the fullerenes although in the case of the spherical tensor
FJg. 10.6 The cage molecule C60 , method (Section 15.5-the method is to some extent anticipated by the above
buckminsterfullerene. discusssion) the complication of d orbital involvement ,on the cage atoms is
absent.
t2g
gu -- -- 8
7
gg ---- 6
hu ----- 7
-----
t2u 5
hg 6
t1g 6
5
Fig. 10.7 The molecular orbital energy level

pattern for those orbitals of C60 involved in
bonding an endohedral atom.
***** 5
gg+ hg
********* 4
gu
**** } 3
t2u
***
*****
hg 2
t1u
*** 1
Bg
* 0
Nodal
planes
---- 1
-----
'
-----.. ' ',,_
*****
I
·.,....---- d
* * * * * * * * */
J2
'
''
'
Fig. 10.8 Schematic diagram of the interaction
between the orbitals shown in Fig. 10.7 and the
**** ''
atomic orbitals of an endohedral atom. ***
* * * * *'~-- 2
1
0
Nodes
Ceo Metal atom
10.3 Ab initio and Xa methods

At the present time it is a routine matter to carry out ab initio calculations
on small organic molecules. Such calculations attempt to solve the Schrodinger
equation for the molecule to a high level of approximation. Essentially, one
needs rather accurate mathematical descriptions of the orbitals involved-
these are derived from work on atoms and approximated by a series of
functions. This is followed by a proper mathematical treatment which
transforms the Schrodinger equation into a form both adapted to the
mathematical description of the orbitals used and to solution by computer.
The commercial availability of programs which run on relatively small
computers, coupled with molecular graphics programs to display the results,
is changing organic chemistry. Such calculations can be expected to yield
bond lengths and bond angles in good accord with experiment-bond
lengths accurate to ca. 0.01 A-but usually slightly short in the best
calculations-and angles accurate to within 2-3°. What is not yet available
is anything comparable for molecules containing transition metal atoms. The
current situation is made evident when it is recognized that it is difficult
theoretically to reproduce experimental data for the isolated nickel atom
or the diatomic Cr2 molecule-although this may be too pessimistic, for
these two examples have particular problems, such as a very low-lying excited
state which, because of the approximations made, could be calculated to be
the ground state.
The problem for transition metals seems largely to arise from the difficulty
of fully allowing for electron repulsion; the particular aspect of electron
repulsion which is relevant is called electron correlation and it has been briefly
mentioned several times in earlier chapters. The simplest method of appreci-
ating the problem is to place the nuclei at the experimental positions
in a molecule. All electrons except one are placed in the lowest orbitals,
Ab initio and Xoc methods 1 221
although one of these will have a hole because of the selected electron, which
is ignored for the moment. The repulsive field generated by all of the included
electrons is averaged and taken as that experienced by the selected electron,
which modifies its orbital accordingly. The selected electron is then placed in
this modified orbital. This procedure is repeated for all of the electrons, each
being selected in turn, over and over again, until the input and output
arrangements are essentially the same and a self-consistent field is obtained.
A weakness of this model is that it effectively smears the density of the
selected electron over the whole molecule; in reality, the other electrons will
tend to stay away from where the selected electron is, not from a smeared-out
distribution-the electron positions are correlated with one another. In
calculations this problem can be dealt with, in some measure, by adding to
the ground state wavefunction additional functions which change it so as to
compensate for the error. The functions most evidently available are those
involving what would formally be regarded as excited states and so different
allocations of electrons to orbitals-that is, different electron configurations.
In this way the interactions between these configurations is, has to be,
incorporated. So, the process of improving the ground state wavefunction
involves so-called configuration interaction. Other things being equal, the
lower-lying the excited state, the more extensively it is likely to be involved
in the ground state correction.
For transition metals there is a large number of low-lying excited states
to be considered; many of these-and the ground state also-may not be spin
singlets, a situation of particular difficulty (the theory is different). Further,
the d electrons are confined to a small volume and it can be difficult to
maintain a balance between these and the more diffuse ligand densities.
Unless a proper balance is maintained, the more diffuse ligand orbitals may
distort so as to compensate for the deficiencies in the less adaptable set. Add
to this the fact that the correlation energies are likely to be comparable in
magnitude to metal-ligand binding energies. Further, relativistic corrections,
needed where spin-orbit coupling or heavy elements are involved, can alter
bond lengths and total energies and you have a very difficult situation.
Indeed, several examples are known in which apparent excellent agreement
with experiment has disappeared when the calculation was improved,
showing that the initial good agreement was, in fact, illusory. Lest this appear
too pessimistic, it is to be emphasized that good calculations are beginning to
appear, but they are far from routine. Ferrocene2 provides an excellent
example. After a calculation including configuration interaction with over a
million excited state configurations and a correction for relativistic effects,
the Fe-C bond length was still 0.07 A longer than found experimentally, a
discrepancy which should be compared both in magnitude and direction
with that quoted above as resulting from far less sophisticated calculations
on organic molecules. This error of0.07 A seems to be common-it has been
found in the c-cr bond length in di-1] 6 -benzenechromium and also in
1] 6 -benzenetricarbonylchromium. The error probably reflects a common
deficiency in the functions used to describe the orbitals of the metal atom,
perhaps a need to include f orbitals. It is surely significant that it is only for
2 Sec H. P. Liithi. P. E. M. Siegbahn, 1. Almlof, K. Faegir and A. Heiberg, Chem. Phys. Lerr.
(1984) l 11, 1.
a main group cyclopentadienyl, Mg(C 5 H 5 h, that acceptable agreement with

experiment has been obtained. In the case of ferrocene, the inclusion of
extensive configuration interaction had the effect of showing the Fe-C
bonding to be significantly more covalent than would otherwise have been
concluded.
Although the situation is improving, the computer demands of ab initio
calculations of the type just considered impose severe limitations. Alternative
approaches which are less computer demanding but, hopefully, comparable
in accuracy are attractive! One such method is very different in approach.
The molecule is, pictorially and mathematically, divided up, a spherical shell
being placed around each atom (the shell is usually called, pictorially, a
'muffin tin'-although 'cake tin' might be more appropriate in the UK).
Each sphere is treated separately. What this means is that when, as before
in the ab initio method, an electron is selected out, it is selected only from
within the sphere, not some more distant part ofthe molecule. If the potential
between the spheres is taken to be constant, it is not difficult to ensure that
the resultant wavefunctions vary smoothly crossing the surface of the shell.
Although this model sounds artificial, it is not, for within each sphere
one is dealing, essentially, with an atom-and it is much easier to do
calculations on isolated atoms than on the same atoms embedded within a
large molecule. Of course, the electron density within any one sphere is
dependent on that in the others, as we would expect. This so-called Xoc
method has been widely used in calculations on transition metal complexes
and is almost certainly the best method presently available for treating large
molecules such as the metal clusters which will be discussed in Chapter 15
and the solids which will be the subject of Chapter 17. In the name Xrx, the
rx is a number, which some treat as a variable and others as a fixed quantity,
which has a value of between i and I. It occurs in the formalism by which
the method treats the electron repulsion involving the selected electron.
However, the results resulting from the use of the Xrx method have been
somewhat erratic and contemporary work is turning increasingly towards
more sophisticated, but related, approaches known as density functional
methods. In these, the (local) density serves a role akin to that of the
shells, the muffin tins, of the Xoc method. As this comment suggests, current
research is aimed as much at producing reliable computational methods as
producing results, so that it is scarcely possible to provide a survey of these
results, although later in this chapter we shall look at ferrocene again.
10.4 Semiemplrlcal methods

In the previous section, the ab initio method was presented as one that makes
a serious attempt to obtain very close approximations to an exact solution
of the Schrodinger equation. The best of such calculations are enormously
demanding of the time of the fastest and most modern computers-many
millions of integrals over three-dimensional space have to be calculated to
high precision. If progress depended on such calculations being available for
all molecules, progress would be slow indeed. Faced with a need for relatively
quick, hopefully reasonably reliable, calculations, many have asked why
reinvent the wheel?, or, in context, why try to calculate that which is already
known? Specifically, rather than spend a lot of computer resources trying to
Extended HOckel method 1 223
include accurately, for example, all of the electron-electron repulsion

and nucleus-electron attraction contributions to the atomic orbital energies
involved in the calculation, why not simply take them from experiment? You
may not be able to say much about how the energies arise, but at least you
get the answer right! Of course, there are alternatives. One could set out to
do the massive sums but simply not include those integrals for which there
is good reason to believe that the answer would be very small. Or, a relatively
small calculation could be undertaken but the results scaled to those of a
more accurate calculation. Or, an approximate theory could be developed
which could be applied to a wide range of experimental data but which
contains parameters. Variations in the values of these parameters could then
be systematically explored until a good fit with the data is obtained.
One then asks whether the parameter values are reasonable and vary sensibly
between related compounds. All of these various possibilities will not be
explored in this chapter. Rather, two somewhat different approaches will be
considered in outline. Others which the reader may well encounter in the
literature and which make serious attempts to model the exact solutions are
complete neglect of differential overlap (CNDO), mentioned in Table 10.2,
intermediate neglect of differential overlap (INDO), and a method due to
Fenske and Hall, mentioned in Table I 0.3. They are excluded simply because
a non-mathematical discussion is scarcely feasible for them; they differ in the
ways they handle various integrals that arise in the detailed theory.
10.5 Extended Hiickel method

The original HUcke! method applied solely to unsaturated organic molecules,
particularly hydrocarbons. It was concerned with the rr electrons and the
interactions between them which led to delocalized rr molecular orbitals.
For hydrocarbons just two quantities were involved, the energy of the
isolated rr electron and the energy of interaction between pairs of them. Of
these, only the latter, usually denoted {3, affected the results. It was a crude
and simple model-but it worked! The results could be used to explain
stabilities, spectra, reactivities and dipole moments; Fig. 10.4 is based on
them. It is not surprising that chemists should seek to adapt the model to
the more complicated molecules of inorganic chemistry, for the HUcke!
approach is perfectly applicable to molecules of low symmetry (it actually
increases their symmetry in a subtle way-chemically distinct atoms are
given the same parameters). The model which has resulted is called the
extended HUcke! method, although inevitably many variants exist (all, like
their parent, given the opportunity also subtly increase symmetry).
In the most common species of the method encountered, each orbital to
be included is given an energy, the value of which is taken from atomic
spectral data. The energy of interaction between two orbitals is assumed to
be related to their overlap integral (to evaluate this, of course, explicit
algebraic forms for the orbitals must be assumed). An overlap integral is a
number and an energy of interaction is needed. It is therefore necessary to
multiply the overlap by an energy and many suggestions for this step have
been made-the outcome of the calculation is dependent on the choice made.
Simplest, and most widely used, is to multiply by the average of the energies
taken as those of the interacting orbitals. The product (overlap integral x
energy term) is taken to be proportional to the interaction energy. The

two are related, then, by a proportionality constant, the practice being to
use that value which gives best agreement with experiment (this constant
has sometimes been denoted F, a notation which has become less common-
perhaps because cynics have been heard to comment that F stands for fudge
factor). This comment is unduly uncharitable not least because once the
constant is given a value it is fixed, at least for a series of related compounds.
Commonly, the value of 1.75 is used. It is important in the method that the
proportionality constant is given the same value for each and every
interaction within the molecule. In this way the method is made rotationally
invariant. That is, the answer obtained does not depend on the direction
taken for the molecular z axis nor does it change if, say, s and p orbitals are
mixed to form sp hybrids and these used instead of separate s and p orbitals
in the calculations.
The theory so far described suffers from one major defect. The maximum
stabilization resulting from a particular interaction is a maximum when the
corresponding overlap is a maximum. So, bond distances (and the total
bonding) are determined by some sort of weighted average over all of the
bonding interactions included. In reality, internuclear distances are deter-
mined by a balance between attractive and repulsive forces; so far the model
lacks repulsive forces. There have been several suggestions for introducing
them into the calculation. One of the most useful uses the same orbitals as
used to obtain overlap integrals to calculate repulsion integrals. This model
is called the atom-superposition electron-delocalization molecular orbital
(ASED-MO) method, has given good results and is incorporated in some
(but by no means all) of the computer packages available for extended
Hiickel calculations. Other modifications to the simple model write the
proportionality constant as a product of factors which, for instance, introduce
an explicit internuclear separation dependence to improve the predictions at
larger internuclear separations or to reduce the extent to which low-lying,
compact, filled orbitals are mixed with higher-lying diffuse orbitals (so-called
counterintuitive orbital mixing, a problem akin to the need for balanced
basis sets, mentioned above).
The value of extended Hiickel theory is that it enables one to apply simple
and acceptable ideas of energy differences and overlaps to low symmetry
molecules and so is applicable to systems for which symmetry alone is of
little help. In so doing it can both highlight major interactions and
competitions between interactions, a concept which can then be separated
from the model itself and discussed separately, and, perhaps, largely quali-
tatively. The danger of extended Hiickel theory is that more meaning can
easily be attached to the energy levels obtained than is warranted. That this
is so can best be seen if some hidden assumptions of the model are
highlighted. First, in carrying out the calculations, it is necessary to assume
a molecular geometry. If the geometry were to be systematically changed,
so as to locate the total energy minimum, this would probably be found to
be well away from the assumed geometry-and quite possibly at a geometry
which would be felt to be quite ridiculous (because of the absence of any
repulsive terms, for instance). Unless specific steps are taken to remedy this
situation, the calculations are made at some high point on a potential energy
surface and, without some detailed justification, cannot be expected to give
Extended Hilckel method 1 225
a reliable account of the energetic consequences of bond length changes. The

remedy is to systematically change the input parameters until the input
internuclear separations are minima; unfortunately, it is not always clear
from published results whether this has been done. Strictly, the geometry
should be one in which any small movement of any atom leads to a less
stable molecule-and it is likely that few of the published calculations satisfy
such a strict requirement. Second, in carrying out the calculations it is
necessary to assume formal charges on the atoms in order to be able to make
use of energies from atomic spectroscopy and to determine the precise form
of the orbitals to be used to calculate overlap integrals. These charges are
not unambiguous. For ferrocene, for example, do we take Fe 0 or Fe11?
Probably the former, because the latter would force us to work with C 5 H5,
and C 5 H 5 is easier. However, this way of determining the choice begs the
Fe 0 -Fe 11 question, it does not solve it. Unfortunately, the results of an
individual calculation are dependent on the choice made.
This highlights a third point. If, after completing the calculation, the
formal charges on the atoms were calculated (and approximate simple
ways are available for this) they would be found to differ from the starting
charges. The new charges would probably be non-integers but this is no real
problem since it is not difficult to interpolate reliably between free-ion
energies and, perhaps, orbital parameters. We could then rerun the calculation
and keep on repeating until the input and output atomic charges became
essentially identical. It seems that such self-consistent charge and con-
figuration (SCCC) calculations do not give better agreement with experiment
than calculations which are not self-consistent. As a result, in most, if not
all, current applications of the extended Hiickel method there is no iteration;
the calculations are stopped once the first output has been obtained (just as
in the Hiickel theory of organic chemistry). It follows that formal atomic
charges calculated by this method are not self-consistent and that arguments
based on such charges should only be accepted cum grana salis. The
final charges depend on the initially chosen charges.
The extended Hiickel method is of value because it addresses the problem
of energy levels and energy level changes when symmetry has either been
exhausted or symmetry does not exist. It is simple, so that it can be applied
to molecular fragments and to crystalline solids alike, and has been applied
extensively to both. It is neither computationally nor mathematically
demanding; computer programs are readily available so that it can be-and
is-used by inorganic chemists as an everyday tool. It is also very flexible,
a feature which is at once both a weakness and an advantage. So, if the
molecular orbital energy level sequence for a particular molecule were
determined beyond all reasonable doubt, perhaps by a very detailed ab initio
study, then there is little doubt that the results could, in large measure, be
duplicated by an extended Hiickel calculation. The advantage of the method
is that it could then be applied to related molecules with considerable
reliability. The virtues and weaknesses of the method are perhaps best
summed up in the words of one of its leading practitioners: 'since all
other methods are superior to it, it inculcates in its user a feeling of humility
and forces him or her to think about why the calculations come out the way
that they do'. However, as reference to any issue of a current journal dealing
with inorganic chemistry will testify, the extended Hiickel method is widely
used both for molecules and solids. Its status is evident in the fact that when
detailed ab initio or similar calculations are published it is normal to compare
the results obtained with those given by the extended Hiickel method. Later
in this chapter there is an example which illustrates how, despite all its
shortcomings, the method can provide insights which could scarcely have
been obtained in its absence.
10.6 Angular overlap model

The angular overlap model was originally introduced as a way to extend
the ligand field model to complexes of low symmetry without dispropor-
tionately increasing the number of unknowns, the number of parameters, in
the problem. In its original form it did this by postulating that all bonding
interactions are proportional to the square of the corresponding overlap
integral (an overlap integral can be either positive or negative, the square
can only be positive). As befits a development of the ligand field model, the
angular overlap model usually concerns itself exclusively with the d orbitals
on the transition metal. If the form of the atomic orbitals is known along
with the molecular geometry~or reasonable approximations can be made~
then the method contains only two unknowns, the proportionality constants
relating overlap to interaction energy for (J and for n interactions. Commonly,
these proportionality constants do not have to be evaluated explicitly. Often,
the problem becomes trigonometric, dependent only on the angular com-
ponent of the orbital wavefunctions, so that it is not necessary to evaluate
overlap integrals either. A further advantage of the method is that, although
normally applied to low-symmetry molecules, when applied to high-symmetry
species it gives results equivalent to those coming from crystal or ligand field
models. For instance, it gives the result that tl.o" = -9/4 tl."'' a relationship
that we met in Chapter 7. So, it is a method with a minimum of parameters
and has been found to give good agreement with experimental results, at
least for transition metal complexes. Its development with time has been in
the opposite direction to that of extended Hiickel theory; its practitioners
have sought to develop the model by its application in increasing detail to
a limited range of molecules. So, it has been refined to the point to which
it can predict with accuracy such diverse quantities as anisotropies in the
magnetic properties and the intensities of electronic spectral bands. These
developments have been accompanied by subtle changes in the model itself.
For example, it has been found possible to incorporate within the, formally,
d electron-only model, some of the effects of s and p orbital involvement in
the bonding~and to do so in a reasonable way. One way in which this has
been achieved is to distinguish between a free ligand and a ligand in a
complex; to allow for a difference between the orbitals and orbital energies
of the free and complexed ligands. Any attempt to deal with the ligands in
any global sense has been abandoned by some workers; each ligand is treated
individually (the most developed current model is a cellular angular overlap
model). This progress has been made by a painstaking detailed study of
individual molecules; the theory is only just reaching the point at which it
can be generally applied without need for molecule-specific development
work.
Three examples: ferrocene, hexacarbonylchromium and ethenetetracarbonyliron 1 227
10.7 Three examples: ferrocene,

hexacarbonylchromium and
ethenetetracarbonyliron
10.7.1 Ferrocene
There have been numerous calculations reported on ferrocene, at a variety
of levels of sophistication. Although there is agreement on the general
bonding pattern in the molecule, this agreement is far from obvious. The
problem is made evident when the results of ab initio calculations are
compared with all of the others. This comparison is given in Table 10.2.
Whereas all non-ab initio methods give the highest orbital as an a 19 , an
orbital which largely consists of the metal 3dz'• this same orbital is at the
bottom of the ab initio list. The difference is deceptive; it arises from a
different definition of what is meant by an orbital energy in the two cases.
The origin of the difference is important because i.t. .has always to be
remembered when ab initio and other calculations are compared.
When an electron is removed from a molecule-and such ionizations are
implicit in sequences such as those of Table 10.2-then the remaining
electrons must be expected to change their distributions and orbitals slightly.
For the ab initio calculations in Table 10.2 this adjustment, the reorganization
energy, has not been included because the readjustment has not taken
place-the energy levels are those of the neutral ferrocene molecule. The
other methods give sequences which answer the question 'what is the order
of ease of ionization of the electrons?' -and so include the adjustments. The
way in which this adjustment energy appears in the ab initio calculations is
as follows. Ab initio calculations on the ferrocene cation, so that there is one
electron less than for the neutral molecule, indicate that the ground state of
the cation is one in which an a 1.(3d) electron is removed from neutral
Table 10.2 Comparison of orbital energy level sequences resulting

from ab initio and other calculations on ferrocene. The most
readily ionized electrons are at the top (but with exceptions, see
the text). Orbitals printed together are close in energy. The
results of two different ab initio calculations are included
Ab initio Ab initio Xot Extended CNDO"

Hiickel"
e1 u(nCp) e1 )nCp) a,., a,., a,.,

e,.,(nCp) e,.,(nCp) e>g e>g e>g
e2g(3d) e2g(3d) elu elu e,.,

e,., e,.,
Energy e2u(uCp) e2u elu
a2u(nCp) a2u(4p) e,.
e2g(uCp) a,u e2g
a,.,(3d) a,.,(nCp)
a,.,(3d)
a2u
a,.,
a The G C5 H5 were not included m the calculation.
ferrocene. That is, it comes from the orbital which is at the bottom
of the ab initio list in Table 10.2. The reorganization energy compensates for
the fact that the a 19(3d) orbital in ferrocene is relatively low in energy. The
reason that e29 is the second orbital in the non-ab initio list, but not in the
ab initio list is similar. There is a basic and important reason for all of this.
Both of the more readily ionized electrons, a 19 and e29 , are largely metal in
character. In the neutral molecule they are confined to a small volume of
space and it is not surprising that the molecular electron density distribution
should change significantly when one of these electrons is removed. It is the
consequential change in energies that result in the apparently anomalous
position of the a 19 and e29 orbitals in the neutral molecule energy level
sequence. In contrast, orbitals which are higher lying in the ferrocene ab
initio energy level sequence are largely ligand in nature. This means that
their electron density is spread over at least 10 atoms and so is relatively
diffuse. As a result, the change in the molecular electron density distribution
following the removal of such an electron is relatively small and so the
rearrangement energy is small also. This is not just some subtle difference
between ab initio and other calculations. If, as well may be the case, we are
interested in the bonding in the ferrocene molecule, then the energy level
pattern given by the ab initio method is that which is appropriate-it more
accurately shows the stabilizations resulting from the bonding interactions.
If, on the other hand, we are more interested in the chemical reactions of
ferrocene-that is, in situations in which electron density is displaced, then
the orbital energy level diagrams of the other methods become more relevant.
Because of its relative ease of ionization, a 19 and e29 behave as if they are
the HOMOs, even if, strictly, they are not. This, of course, is a situation
which is very similar to that met at the end of Section 6.2, where for classical
coordination complexes it was found that the (incompletely filled) d orbitals
are not, in fact, the highest lying occupied orbitals.
At the present time it does not seem to be unambiguously determined
whether the lowest ionization potential of ferrocene corresponds to the loss
of an electron from the a 19 orbital or whether it comes from the e 29 • As
Table 10.2 shows, most theoretical work favours a 19-and there are electronic
spectroscopic arguments in favour of this assignment-but the ab initio
calculations point to the alternative, e 29 assignment and find support in both
EPR and photoelectron spectroscopic data (the latter will be given and
discussed in Section 12.7).1t is likely that the ab initio assignment is correct.
The results of an ab initio calculation on ferrocene are also shown in Fig.
10.9.1t is perhaps more useful than the same data in Table 10.2 in addressing
the question of whether these calculations support the simple picture of the
bonding in ferrocene given in Section 10.1. The answer, comfortingly, is yes.
This conclusion is perhaps most easily seen by looking at the empty,
antibonding orbitals in Fig. 10.9. As we have seen, occupied n orbitals in a
hypothetical (C 5 H 5 )z unit which could act as electron donors to the iron
atom are of A 19 + A 2 u + E 1• + E 1• symmetries. If these interactions do
indeed contribute to the metal-ligand bonding then we would expect this
to be signalled by destabilized, antibonding, orbitals of these same symmetries.
As Fig. 10.9 shows, such orbitals do, indeed, exist. Looking at antibonding
orbitals in this way is a simple, approximate, way of avoiding the problems
posed by the plethora of c--c and C-H bonding orbitals in the bonding set.
Three examples: ferrocene, hexacarbonylchromium and ethenetetracarbonyliron I 229
Metal orbitals
Alg p, ring molecular orbitals
Fe (cp), molecular orbitals
FJC. 10.9 Schematic ab initio molecular orbital Rather similar arguments hold for the potentially 1t acceptor orbitals on the
energy level diagram for ferrocene (data
from T. E. Taylor and M. B. Hall, Chem. Phys.
hypothetical (C 5 H 5 ), unit, E 2 • + E 2 •• There is no metal orbital of E 2 •
Let!. (1985) 114, 338). This figure differs from symmetry but there is one of E 2 •• Again, there is an empty E 2 • orbital,
Table 10.2 in that the table, but not this figure, corresponding to an occupied one, just as required by the simple picture of
includes C5 H5 ring orbitals. This diagram does
not demonstrate the fact that because of
Section 10.1.
configuration interaction the average occupancy
of the doubly occupied bonding orbitals is
sligh~y below 2 (ca. 1.98) and that of the
o
empty antibonding slighUy above <ca. 0.02). 10.7.2 Hexacarbonylchromium
Section 10.1 contains a schematic molecular orbital energy level diagram
for Cr(C0) 6 , Fig. 10.5, which was derived from a model in which it was
regarded as a complex in which extensive 1t back-bonding occurs from metal
t 2 • to 1t*t 2 •• There is abundant evidence that this back-bonding is a real
phenomenon. Theoretically, all detailed calculations concur on its reality;
experimentally, evidence for its presence is revealed by a detailed analysis of
the electron distribution about the Cr--c--o axis, obtained from accurate,
low-temperature X-ray diffraction measurements. Perhaps the simplest
evidence is provided by the changes in the average v(CO) stretching
frequencies (em_,) along the isoelectronic series

TI(CO)~- V(C0)6 Cr(C0) 6 Mn(C0)6
1780 1897 2017 2113
The argument is that the different frequencies reflect different electron
populations in the CO n*, antibonding, orbitals, electron density which
originates in donation from the metal atoms to the CO groups. The greater
the antibonding electron population, the lower the frequency. If, as seems
entirely reasonable, the formal charges on, for instance, Mn+ and v-,
respectively, inhibit and enhance the metal -+ ligand n electron donation
then an entirely self-consistent picture is obtained. The picture makes the
assumption that the u electron density does not share the sensitivity to formal
metal charge shown by the n, but there is ample evidence in support of this
assumption.
Table 10.3 summarizes the results of some of the available calculations
on Cr(C0) 6 and Fig. 10.10 is a molecular orbital energy level diagram based
on the ab initio data in Table 10.3. All of the calculations in Table 10.3 agree
about the HOMO-it is a t 29 molecular orbital which is extensively involved
in the n bonding that was discussed at the end of the previous paragraph.
An electron in one of these t 29 orbitals is approximately one-third located
in ligand n* orbitals. Apart from these t 29 orbitals, all of the other orbitals
listed are largely located on the ligands. Because they are distributed over
six ligands they are relatively diffuse and Koopmans' theorem holds (this
theorem will be discussed in Section 12.7; it equates an ionization energy
with an orbital energy). It follows that there are no complications of the sort
encountered with ferrocene, where some of the ab initio results could not
immediately be compared with the others. However, this is not to say that
electron correlation is unimportant in hexacarbonylchromium. For the ab
initio calculations it has the effect of decreasing the electron density in the
metal t 29 orbitals and to decrease it in the e9 , a step that serves to bring the
ab initio and Xa results closer together.
Table 10.3 Orbital energy level sequences resulting from

various calculations on Cr(C0) 6 • Orbitals printed close
together are close in energy
Ab Initio X.. Extended ca. Fenske-

Hackel Hall
t:zg t:zg t:zg t:zg

t1u t1u eg t1u
eg t~g t1u t~g
t2u t2u
Energy t~g a~g
t2u t:zg t1u
t~g
t:zg t1u a~g
t1u t2u
eg eg
a~g t1u
a~g t:zg
t:zg
Three examples: ferrocene, hexacarbonylchromium and ethenetetracarbonyliron 1 231
'
s, 81g
t2u
tlg
t2u
'"' 12
co It'
t1u t1u
d, eg
t2g
t?g
eg
Metal orbitals
6
t1u co (J
81g
tlu
tl.g tlg
12
t2u t2u
C01t
t?g t1u
co
t1u t?g orbitals
eg (CO)s
orbitals
81g
Cr(CO)s
molecular
orbitals
FJC. 10.10 Schematic X,. molecular orbital As was discussed in Chapters 6 and 7, the occupied ligand orbitals of an
energy level diagram for Cr(C0) 6 (data from
R. Arratia-Perez and C. Y. Yang, J. Chern. Phys.
octahedral complex can be classified as either a or lt. The latter span the
(1985) 83, 4005). Electrons which would be irreducible representations T 1• + T'" + T 2 • + T 2 •• Of these, the T,. and T 2•
counted in an application of the 18-electron are non-bonding because there are no orbitals of these symmetries on the
rule are shown as lines, those that would be
excluded are shown as dots.
metal. It is therefore not surprising to see in Table 10.3 that they are found
to have almost identical energies-any differences are due to ligand-ligand
interactions. Three of the entries in Table 10.3 show a cluster of orbitals of
T 1• + T 1• + T 2 • + T 2 • symmetries which it is tempting to equate with the
entire n set. This identification would not be entirely valid for the T 1 • orbitals.
There are two sets of orbitals of this symmetry and both are mixtures of
ligand a and ligand n. The ligand a contribution is the greater for the higher
lying, not for the lower. For the lower, the n contribution dominates. Both
of the approximate methods place an orbital of T 29 symmetry as the lowest
listed; the actual wavefunctions show a significant metal t 2 • component. The
ab initio and XIX calculations do not give this result, and we must conclude
that it is an artefact introduced by the approximations of the less rigorous
methods.
Fig. 10.11 Simultaneous donation from CO a

and n: occupied orbitals into a chromium 4p
orbital in Cr(C0) 6 . Only orbitals of T1.u symmetry
are involved in this simultaneous donation but it
occurs for both of the bonding T1" orbital sets
of Fig. 10.10.
The ligand a orbitals span the A 1• + E. + T 1 • irreducible representations.

Both the ab initio and Xcx calculations agree that the a 1 • is the lowest but
differ in their placing of the e. orbitals. The origin of this disagreement is
not clear, although it is associated with different metal orbital contributions
(about 20% for the ab initio and 33% for the Xcx); it would probably be
expected that the ab initio result is the more reliable. If the ab initio results
are taken as a criterion, the approximate methods do not fare well, although
as has been pointed out, some inherent flexibility may well enable them to
duplicate the ab initio results more closely, if this were to be taken as the
goal. Details apart, however, all the methods agree in that the qualitative
picture presented in Section I 0.1 is supported. However, there is one
unexpected result that merits mention. In Section 7.2.2, we recognized the
possibility of interaction between a set of a bonding and a set of n bonding
t 1• orbitals but did not explore the possible outcomes in detail. As we have
indicated above, the t tu molecular orbitals actually contain contributions
from ligand a, ligand n and metal p orbitals. The detailed calculations reveal
an unexpected picture, one in which there is simultaneous electron donation
from adjacent CO groups, the donation from one CO group being a whilst
that from its neighbours is n. This interaction is shown in Fig. I0.11; it occurs
for all of the bonding t~u orbitals of Table 10.3 and Fig. I0.1 0.
10.7.3 Ethenetetracarbonyliron
This compound, Fe(C0) 4 C 2 H 4 , is prepared by the reaction between a high
c pressure of C 2 H 4 and Fe(C0) 5 and has the structure shown in Fig. 10.12.
0 It is a yellow oil at room temperature and is relatively unstable, decomposing
Fig. 10.12 The molecule Fe(C2 H4 )(C0) 4 • The into C 2 H 4 and Fe 3 (C0) 12 , the latter presumably being formed from three
arrangement of atoms around the Fe is
approximately octahedral (the edges of the Fe(C0) 4 units. This section will be concerned with the answers to two
octahedron are shown dotted). questions, to which we will find that the insights provided by extended
Three examples: ferrocene, hexacarbonylchromlum and ethenetetracarbonyliron 1 233
Hiickel theory provide answers:

• Why is this compound stable at all?
• Why is it not more stable?
For simplicity, it will be assumed that decomposition occurs by Fe-C 2 H 4
bond fission, an assumption which means that we can restrict our discussion
to the interaction between Fe(C0) 4 and C 2 H4 units. As Fig. 10.13(a) shows,
the Fe(C0) 4 fragment is rather like part of an octahedral molecule and we
could well be tempted (incorrectly) to assume that the metal has two
additional orbitals available to complete an octahedral bonding pattern. If
this were so, then, using the molecular C 2 v point group, it is a simple matter
to show that these two orbitals give rise to combinations of A 1 + B1
symmetries, combinations that are shown in Fig. 10.13(b). It would have
been logical to have concluded that the A 1 combination is the more stable
because it is the no-node combination, whereas the B 1 has a planar node.
However, we would have been wrong in a way that is readily revealed by
the extended Hiickel calculations-the A 1 combination is Fe-CO antibonding,
0
c
I
oc--u6-
co
c
(a) 0
FIJI. 10.13 (a) The Fe(C0) 4 unit regarded as a

fragment of an octahedron (drawn from the
same viewpoint as Fig. 10.12). The Fe orbitals
required to complete the octahedron are
shown. (b) The A1 and 8 1 combinations of the
octahedron-completing Fe orbitals. The
symmetry elements of the c, point group are
shown inset. The orbital combinations are
drawn from a viewpoint lying in the u~ mirror
plane.
c
(b) 0
whereas the B 1 combination is Fe-CO bonding. The extended Hiickel

calculations reveal this fact and thus that the orbital ordering is the
opposite of that given by the simple model; in fact, the A 1 orbital is above
the B 1 . These orbitals are shown schematically in Fig. 10.14.
The picture is completed by noting that when it spans the two empty
octahedral sites of the Fe(C0) 4 unit, the ethene n bonding orbital has A 1
symmetry and then antibonding orbital 8 1 symmetry. The partial molecular
orbital energy level diagram shown in Fig. 10.15 is thus obtained, and, with
it the existence of the molecule explained-there are appropriate bonding
orbitals occupied. But why is it not more stable, our second question? Most
probably because the bonding within the Fe(CO)z unit shown in Fig. 10.14
is weakened by the addition ofC 2 H 4 . The occupied B 1 orbital of the fragment
can only become involved in C 2 H 4 bonding at the expense of the Fe-(CO)z
o[XJ:~~o
bonding. Put another way, electron density in the A 1 fragment orbital means
population of an Fe-(C0) 2 anti bonding orbital. In retrospect, it can be seen
that the existence of Fe(C0) 5 , a trigonal bipyramidal molecule, provides an
Fe 81 argument in favour of the 'A 1 high energy, B 1 low energy' pattern.
This molecule differs from Fe(C0) 4 C 2 H 4 by a CO group taking the place
+ -
of the C 2 H 4 . Following the usual model of CO bonding to a metal, a
donation from the carbon of the CO group to the metal requires an empty
Ftg. 10.14 The A1 and 8 1 combinations of A 1 orbital on the iron. Similarly, back bonding into an empty n antibonding
Rg. 10.13(b) as given by the extended HUcke! orbital of the CO requires a filled 8 1 orbital on the iron. This A 1 high, 8 1
method, viewed along the linear OC-Fe-CO axes low, pattern is just that revealed by the extended Hiickel calculations.
of Fig. 10.13(b). The nodal plane indicating
Fe-C antibonding in the A1 orbital is dotted. Before concluding our discussion of the Fe(C0) 4 unit it is worthy of
The Fe-<: overlaps responsible for the bonding comment that it is often thought of as a member of an isolobal series
interactions in the 8 1 orbital are shown as such as that given in Fig. 10.16. Members of an isolobal series are always
double-headed arrows.
shown as related to each other by the symbol --..->. Membership of such a
series is based on experimental evidence as much as theoretical, but members
of an isolobal series usually have similar orbital patterns, both in terms of
orientation and energy. Experimentally, they form similar compounds-so,
each member of the middle series shown in Fig. 10.16 forms an H 2 com-
pound. However, of the members of this series only Fe(C0) 4 has the A 1 high,
B1 low energy level pattern. Theoretically, this species is the only member of
1t antibonding
I
bi
~\ I
----~\
b1 \ \ 1I
FJg. 10.15 Schematic molecular orbital
diagram showing the interaction between the n
orbitals of C2 H4 and the Fe(C0) 4 fragment
Orbitals of
Fe(C0)4
\' ,'/.I/
.· . 1t bonding
~~'-----b1...
orbitals.
"''""' /
Final comments 1 235
One orbital, one-electron isolobal species
---u-- -u-
r
-CH, -Mn(C0)5 -Co(C0)4
-
0Fe(C0) 2 Cp
-~C>
----o- -H
----o- -Br
Two orbital, two-electron isolobal species
Fig. 10.16 Examples of isolobal series.
'-....C 0 (CO)C ~
>=&
~0 ~ ~S
~ p u ~ u ~
Three orbital, three-electron isolobal species
the series for which the two electrons accommodated in the pair of orbitals
shown would normally be regarded as paired. This might be thought of as
a weakness in the isolo bal concept, but, in fact, the value of the concept lies
in its flexibility. It has come to be widely used in inorganic chemistry.
10.8 Final comments

There is every indication that there will be many changes in the area of this
chapter over the next few years. The increasing availability of ever-increasing
computer power will surely mean that higher-level calculations become more
routine. If this is true, it could well be that the present time (1994) is close
to the time at which the extended Hiickel method is at its zenith. Current
results coming from methods as diverse as the cellular angular overlap and
the density functional models are impressive. Then there are developments
on the horizon which presage change. The first is the wider incorporation
of relativistic phenomena. They are not trivial; the spin-orbit coupling which
has featured so often in Chapters 8 and 9 is relativistic in origin. It has
recently been shown that the reason that Bi(C 5 H 5 ) 5 , a molecule with C4 "
symmetry, is an intense violet colour is because of relativistic effects. In C4 "
symmetry, but not D,., relativistic effects cause the LUMO to drop in
energy, moving an allowed transition into the visible (in the former
symmetry, but not the latter, the bismuth 6s orbital can participate in
the LUMO; for a discussion on the relevance of this see the footnote at the
beginning of Section 11.3 ). One must anticipate similar, if often less dramatic,
phenomena for many compounds of the heavier elements. An example of
one of the more dramatic phenomena is the prediction that relativistic effects
may well lead to the molecule HgF4 having some stability (to date, it has
not been prepared). Finally, there is the development of spin-coupled valence
bond theory. This book, like most of its generation, has concentrated on
molecular orbital theory. Valence bond theory was mentioned at the
beginning of Section 6.2.1 but then forgotten. Subsequently, problems were
found with the molecular orbital theory model arising from electron
repulsion. Such problems are much less severe for valence bond theory, but
it has problems of its own. These are in large measure overcome in
spin-coupled valence bond theory and so this method offers the prospect of a
quite new approach to the electronic structure of many of the molecules of
this chapter. This development is for the future because, at the moment, the
method is only being applied to diatomic transition metal species.
Further reading Ziegler, V. Tshinke, L. Fan and A. D. Becke, J. Amer. Chern.

Soc. (1989) 111, 9177. The work reported used density function-
This Chapter has deliberately avoided the mathematical com- al methods, which represent a development and improvement
plexities associated with the methods it describes. In practice, on Xot calculations. The Xot method is readably described in
for most readers who wish to explore the topics at greater the first few pages of 'The Self-Consistent Field for Molecules
depth, the problem will rapidly translate into one of working and Solids', which is volume 4 of the series Quantum Theory
intelligently with a suitable computer program. A helpful book, of Molecules and Solids, J. C. Slater, McGraw-Hill, New York,
with seven chapters related to the topics of this chapter (and 1974. Perhaps even more readable (the only, two, equations are
part of Chapter 13) is Computational Chemistry Using the PC in the first paragraph) is 'Electronic Structure Calculations
by D. W. Rogers, VCH, Weinheim, 1990. using the Xot Method', D. A. Case, Ann. Rev. Phys. Chern. (1982)
General accounts of the chemical bonding concepts discus- 33, 151. An overview of the density functional method can be
sed in this chapter are to be found in the following two books, gained by a browse through Density Functional Methods in
although only the second covers isolobality. Chemistry, J. K. Labanowski and J. W. Andzelm (eds.) Springer-
• J. N. Murrell, S. F. A. Kettle and J. M. Tedder The Chemical Verlag, New York, 1991. In general easier to read (in the second
Bond, Wiley, Chichester, 1985. half, at least) is 'Approximate Density Functional Theory as a
• T. A. Albright, J. K. Burdett and M-H Wangko Orbital Practical Tool in Molecular Energetics and Dynamics'. by T.
Interactions in Chemistry, Wiley, New York, 1985. Ziegler, Chern. Rev. (1991) 91,651. A paper which includes not
only ab initio but also density functional results is 'Selected
An excellent source of ab initio results relevant to the present Topics in ab initio Computational Chemistry in Both Very
chapter is 'Ab initio Calculations of Transition-metal Organo- Small and Very Large Chemical Systems', by E. Clementi, G.
metallics: Structure and Molecular Properties', A. Veillard, Corongiu, D. Bahattacharya, B. Feuston, D. Frye, A. Preiskorn,
Chern. Rev. (1991) 91, 743. The paper contains not a single A. Rizzo and W. Xue in Chern. Rev. (1991) 91, 679. It includes
equation and so is very easy for the general reader to read; a detailed discussion of buckminsterfullerene and also a topic
unfortunately, it contains not a single diagram either. not included in the present chapter-the use of theoretical
A very readable account of the bonding in ferrocene is given methods to calculate the way that molecular systems evolve
by M. M. Rohmer and A. Veillard in Chern. Phys. (1975) 11, with time.
349. Also of interest is 'The Transition Metal-Carbon Bond', The first paper to describe the extended Hiickel method in
by E. R. Davidson, K. L. Kunze, F. B. C. Machado and S. J. its present form is by R. Hoffmann, J. Chern. Phys. (1963) 39,
Chakravorty, in Ace. Chern. Res. (1993) 26, 628. This contains 1397, but this covers its application to organic molecules.
an account of the bonding in Cr( CO )6 which is both a bit more A closely related, very readable blow by blow account for
advanced than and different from that in the text (whilst being inorganic molecules is by H. D. Bedon, S. M. Horner and S. Y.
readable and consistent with that in the text). Tyree Inorg. Chern. (1964) 3, 647.
A paper that uses the general approach developed in this The angular overlap model in its simplest form is described
chapter but which extends both it and its applicability is by T. readably and in detail in Appendix 3 of Inorganic Electronic
Questions 1 23 7
Spectroscopy, A. B. P. Lever, Elsevier, Amsterdam, 1984. A S. J. Essex, M. Gerloch and K. M. Jupp Mol. Phys. (1993) 79,
more recent overview is provided by 'The Angular Overlap 1147, an article which is useful for the non-mathematically
Model as a Unified Bonding Model for Main Group and minded because towards the end it contains a review of the
Transition Metal Compounds' by D. E. Richardson, J. Chern. physical significance of the parameters contained within the
Educ. (1993) 70, 372. More developed forms of the model are developed angular overlap model.
the subject of a review by M. Gerloch and R. G. Wooley in Finally, the spin-coupled valence bond theory is described
Frog. Inorg. Chern. (1984) 31, 371. Most recent developments in 'Applications of Spin-Coupled Valence Bond Theory', D. L.
are covered by C. A. Brown, M. J. Duer, M. Gerloch and R. Cooper, J. Gerratt and M. Raimondi, Chern. Rev. (1991) 9/,
F. McMeeking in Mol. Phys. (1988) 64, 825 and M. J. Duer, 929.
Questions 10.5 In the last example of Table 10.1 an ethene molecule

bridges two zirconium atoms. Assume that the Zr 2 C 2 unit is
10.1 The ring system used at the top of Fig. 10.2 to define planar with D2 • symmetry and that the ethene hydrogens lie
the z direction is a planar eight-membered ring such as in a plane perpendicular to the Zr 2 C 2 • Extend the Chait-
cyclooctatetraene, C 8 H 8 • Extend Fig. 10.2(b) to include both Duncanson model to cover this case and thus suggest why the
it and the seven-membered ring mentioned in the caption. C-C bond length is the longest in Table 10.1.
10.2 It has been suggested that there is a similarity between 10.6 In the text it was shown that the simple picture of n
the bonding of C 2 H4 to CH 2 in cyclopropane, C 3 H 6 , and the back-bonding in metal-Cr bonding given by pictures such as
Chatt-Duncanson model of the bonding of the C 2 H 4 to Pt in Fig. 2.1 overestimates the extent of this bonding in Cr(C0) 6 .
Zeise's salt. Critically assess this suggestion. Show that it similarly overestimates it in the tetrahedral
molecule Ni(C0) 4 . For this, the character table of the T, point
10.3 On a relatively small modern computer it is now possible group will be needed.
to carry out approximate ab initio calculations on quite large
organic molecules. However, with none of the programs avail- E 8C3 3C2 6S 4
Td 6ad
able for this, is it possible to include a transition metal atom?
Outline the reasons for this and explain how approximate A1 1 1 1 1 1
methods attempt to circumvent the problem.
A2 1 1 1 -1 -1
10.4 Crystal and ligand field theories predict that the d 0 ion
E 2 -1 2 0 0
[TiH 6 ] 2 - will be octahedral. Extended Hiickel, however, pre-
dicts a C2 " bicapped tetrahedron, structure. Initially, ab initio T1 3 0 -1 1 -1
methods predicted an octahedral structure but with the inclu-
sion of configuration interaction a trigonal prismatic structure T2 3 0 -1 -1 1
is indicated (see Inorg. Chern. (1989) 28, 2893). Using this
discordance of results as a basis. suggest those situations in
which each method may be expected to make reasonably
reliable geometry predictions (if ever!).
f electron systems: the
lanthanides and actinides
11.1 Introduction
In recent years there has been an increasing study of compounds of the
lanthanides and, to a lesser extent, of the actinides. These two groups
of elements have varying numbers of electrons in their f orbitals (4f for the
lanthanides and 5f for the actinides), thus inviting a comparison with the
transition metal elements, discussed in Chapters 6- 9, with their varying
number of d electrons. However, as we shall see, such a comparison is not
particularly helpful, a situation which has contributed to an attitude
commonly encountered-that f electron systems are difficult to understand,
that the theory is difficult. It is hoped that it will be possible to demonstrate
in this chapter that this is not the case. Indeed, it is hoped to convince the
reader that a study off electron systems is not only of value in its own right
but that such a study helps in the understanding of d electron systems. It
does this by its concern with phenomena which also exist in d electron
systems but which are currently largely ignored when discussing them.
What are the difficulties with f electron systems? First, they involve f
orbitals and these are unfamiliar. Fortunately, they have already been met
in this book, in Section 7.3, where it was found useful to assess their relative
energies, the f orbital splittings, in an octahedral crystal field. Secondly,
spin- orbit coupling is important. Again, this phenomenon has already been
met (in Sections 8.4 and 9.5) but because we shall have need of a deeper
understanding, pictures of spin-orbit functions will be introduced. Thirdly,
crystal and ligand field effects are small, something which poses problems
when one is more familiar with molecules in which they are large. Finally,
coordination number and geometry are much more varied (and more
uncertain) than for d electron systems. Fortunately, the last two problems
tend to cancel each other out. When crystal field effects are small, knowledge
of the detailed ligand arrangement becomes less important.
Introduction 1 239
Although the discussion in this chapter will be largely concerned with

theoretical matters, it is helpful to first take a brief look at the chemistry of
the lanthanides and actinides. The solution and solid state chemistry of the
lanthanides is dominated by the trivalent state. Other valence states exist-
cerium(IV) is widely used as an oxidizing agent, for instance-but although
their chemistry has been much studied, the knowledge of their spectroscopic
and magnetic properties is still relatively primitive. It seems that these
properties resemble those of the corresponding trivalent ions. Just as the
spectroscopic and magnetic properties of iron(III) resemble those of man-
ganese(II), because both are d 5 (although high spin-low spin complications
occur), so one would expect europium(II) to resemble gadolinium(III)
because both are f7 (and without spin complications). Our discussion will
therefore be confined to the trivalent lanthanides and actinides. This
simplification is much more justified for the lanthanides than the actinides.
All the lanthanides, but not all the actinides, have stable trivalent species.
Although this valence state is stable for the simple actinides towards the end
of the series, americium(III) onwards, for the early members it is usually
strongly reducing and difficult to characterize. Thorium(III) probably does
not exist except in the gas phase. Nonetheless, we shall confine our discussion
of the actinides to the trivalent state, for then we can treat them and the
lanthanides together.
Apart from oxidation-reduction behaviour, the chemistry of the earlier
lanthanides resembles that of calcium( II)-the oxides absorb water to give
hydroxides and they absorb carbon dioxide to give carbonates. The hydroxides
tend to be slightly soluble in water giving alkaline solutions, the carbonates
tend to be insoluble and so on. The chemistries of the later lanthanides tend
to be more like that of aluminium, although the hydroxides are not
amphoteric. These similarities with the properties of more common ions is
not just an aide-memoire; it can be exploited. Thus, calcium(II) ions are
biologically very important-their movement is involved in nerve action, for
example. Unfortunately, the study of such calcium ions is very difficult
because calcium(II) lacks any convenient spectroscopic property-a spectro-
scopic 'handle'-by which it can be studied. Because of their similar chemistries,
it is possible to replace the calcium(II) with an ion such as terbium(III), which
does have convenient handles, as we shall see. Study of the terbium(III) then
gives the information we could not get directly for calcium(II). Such
replacements are a popular trick in bioinorganic chemistry (see Chapter 16).
The smooth change in chemical properties of the trivalent ions of the
lanthanides across the series is associated with the so-called lanthanide
contraction-the size of the trivalent ions becomes progressively smaller with
increasing atomic number. Although the ionic radii often quoted in support
of this statement are not free from objection, 1 the phenomenon seems real
enough. Because the 4f orbitals are highly nodal and all of the nodes pass
through the nucleus, the probability of finding an f electron close to the
nucleus is low. The 5d and 6s orbitals-which are formally empty orbitals-
may well accept (paired) electrons donated by ligands and in so doing help
to determine the ionic radius of the lanthanide. But electrons in these orbitals
have a much higher probability of being very close to the nucleus than do
1 See footnote on page 303.

240 1 f electron systems: the lanthanides and actinides
4f
Fig. 11.1 Radial probability functions for

hydrogen-like orbitals. Although the 4f (solid) 5d
has its maximum closer to the nucleus than 5d ~
,~,
(dashed) and 6s (dotted), close to the nucleus :0 I \ 6s

I \
e"'
the density of the latter two both exceed that of .0
I \
4f. I \
0..
I \ •
\•
',
~
r-
_:
' .
the 4f (see Fig. 11.1). Moving across the lanthanide series, the addition of a
4f electron to balance each increase in positive charge on the nucleus is not
sufficient to prevent electrons in the 5d and 6s (and, for that matter, the 5s
and 5p) orbitals feeling the increased nuclear charge.2 The effect on the 5p,
5d and all other outer electrons is a progressive orbital contraction and, with
it, a decrease in the ionic radius. Although the data are incomplete, it is clear
that the actinides exhibit an actinide contraction which closely parallels the
lanthanide contraction.
A study of the electric spark discharge spectra of the elements (and, so,
of the electronic energy levels) shows that the ground states of most atoms
of the lanthanide elements have an outer-shell electronic configuration of
the form ... 4f"6s 2 , although some have ... 4f"- 1 5d 1 6s 2 In the trivalent ions,
also seen in the arc spectra, the two s electrons and one other are lost; the
other one being an outer d electron if one were present in the atom. The
electronic ground states of the trivalent lanthanides therefore have filled 5s
and 5p shells, empty 5d and 6s and a number of 4f electrons which varies
from none in lanthanum(III) through to 14 in lutecium(III). These configura-
tions are detailed in Table 11.1. Following the lanthanides, the trivalent
actinide ions contain a variable number of 5f electrons, no 6d and no 7s.
The actual electronic configurations of the ground states of the atoms
and trivalent ions are given in Table 11.2.
11.2 Shapes of f orbitals

Because of their importance to us, the object of the present section is to give
the reader more familiarity with the shapes off orbitals than that provided
in Section 7.3. In that section the f orbitals were used to indicate the way
that F terms (arising from d" configurations, n = 2, 3, 7 or 8) split in an
octahedral ligand field. The labels that were used to describe these f orbitals
were abbreviated, just as the label dz, is an abbreviation for d 2 z'-x'-y'· The
complete labels for the f orbitals accurately describe the lobes of the orbitals,
their relative phases and their positions.3 Drawings of the cubic set of
f orbitals are repeated in Fig. 11.2, together with their abbreviated and
2 Some textbooks treat the lanthanide contraction as involving only 4f electrons. This cannot
be correct because these electrons do not determine the ionic radius~they are too well tucked
away inside the ion.
3 Here, as usual, we consider only the angular part of the complete orbital.
Shapes off orbitals 1 241
Table 11.1 The electronic configurations of the neutral atoms and trivalent lanthanide
ions. Core electrons have been omitted
Atomic Name Symbol Isolated atom M 3 + f electron

number electron configuration configuration
57 Lanthanum La 5d 16s2 4f0

58 Cerium Ce 4f15d 16s2 4f 1
59 Praseodymium Pr 4f3 6s2 4f 2
60 Neodymium Nd 4f 4 6s2 4f 3
61 Promethium Pm 4f5 6s2 4f 4
62 Samarium Sm 4f6 6s2 4f 5
63 Europium Eu 4f7 6s2 4f6
64 Gadolinium Gd 4f 7 5d 16s2 4f 7
65 Terbium Tb 4f9 6s2 4f8
66 Dysprosium Dy 4f1o6s2 4f 9
67 Holmium Ho 4f116s2 4f10
68 Erbium Er 4f126s2 4f 11
69 Thulium Tb 4f13 6s2 4f12
70 Ytterbium Yb 4f146s2 4f 13
71 Lutecium Lu 4f145d 16s2 4f 14
Note: Promethium, effectively, does not occur in nature. It is a fission product of uranium and may be made,
for example, by neutron bombardment of neodYmium to give an isotope witl1 a half·life of just under 4 years.
Table 11.2 The electron configuration of the neutral atoms and trivalent actinide
ions. Core electrons have been omitted
Atomic Name Symbol Isolated atom M3 + f electron

number electron configuration configuration
89 Actinium Ac 6d 17s2 5f 0
90 Thorium Th 6d27s2 5f 1
91 Protactinium Pa 5f 26d 17s2 5f 2
92 Uranium u 5f3 6d 17s2 5f3
93 Neptunium Np 5f 4 6d 17s2 5f4
94 Plutonium Pu 5f6 7s2 5f 5
95 Americium Am 5f 7 7s2 5f 6
96 Curium em 5f 76d 17s2 5f 7
97 Berkelium Bk 5f8 6d 17s2 or 5f 9 7s2 5f8
98 Californium Cf 5f1o7s2 5f 9
99 Einsteinium Es 5f11 7s2 5f10
100 Fermium Fm 5f12 7s2 5f 11
101 Mendelevium Md 5f13 7s2 5f 12
102 Nobelium No 5f14 7s2 5f 13
103 Lawrencium Lr 5f14 6d 17s2 5f 14
complete labels. The reader may have noted the use of the phrase cubic set
in the preceding sentence. This is because the f orbitals of Fig. 11.2 are only
appropriate for cubic molecules (in practice, this means molecules with Oh
or Td symmetries or related point groups-molecules with symmetries such
that the x, y and z axes are symmetry-related). For non-cubic geometries
one uses a different set of f orbitals. Actually, the situation is not all that
unfamiliar. Consider a d., orbital in an octahedron. It is so normal to choose
the z axis to coincide with a fourfold axis that we seldom consider any
~
X ~y
f" = f, (2z' _ 3 x, _ 3y,1; similar are fx3 = fx (2x2 _ 3Y' _ 3,,1and fy 3 = fy (2Y' _ 3z2 _ 3,121
Fig. 11.2 The cubic set off orbitals together

with their shortened and detailed Cartesian
angular forms.
~y
f, (x'- Y'l; fx (Y'- z2) and fy(z2- x>) are similar
X~ ~y
alternative. However, in real life an octahedral complex is often slightly

distorted and this distortion is frequently a distortion along a threefold axis
(this case was discussed in Section 7.10). The threefold axis is then the axis
of highest symmetry and so becomes the z axis and the dz, orbital has to be
oriented along it. This dz, orbital cannot be the same as that which is oriented
Electronic structure of the lanthanide and actinide ions 1 243
along a fourfold axis in the undistorted octahedron, although they look the
same. In fact, the dz, (threefold axis) is a mixture of the t 2 • set, dxy>
d,z and dzx (fourfold axis). For dz, the existence of two choices of axes does
not mean a change in shape; for the f orbitals a change in axis set does mean
a change in shape.
There is another way of looking at this. In Oh symmetry the f orbitals
transform as A 2 • + T 1• + T2 •. Reduce the symmetry by a distortion (com-
pression or elongation, for example) along a fourfold axis so that the
symmetry is now D4 h. The f orbitals now transform as (B 1.) + (A 2 • +E.)+
(B 2 • +E.), where the brackets correspond, in order, to the symmetries in Oh
given above. It can be seen that two different sets of f orbitals have E.
symmetry. Because they have the same symmetry they can mix and it proves
convenient to let them do this and to work with combinations of them. As
a result, some differently shaped f orbitals arise.4 The members of this
so-called general set of f orbitals are shown in Fig. 11.3, where both their
abbreviated and complete labels are given.
The lanthanide and actinide trivalent ions commonly occur with high
coordination numbers, typically 7, 8, or 9 but up to 12, in low-symmetry
geometries. Simple theoretical models, such as the extended Hiickel and
angular overlap models (Sections 10.5 and 10.6) have therefore been applied
to them since symmetry alone does not give much insight. In such calculations
the general set of f orbitals would be used. It is perhaps appropriate to
comment that if the bonding in lanthanide and actinide complexes were
entirely ionic then it would be nonsense to apply such models to them-they
depend on the existence of a (covalent) overlap between the metal and ligand
orbitals. The fact that the f electrons in these compounds have properties
which are very close to those of the isolated M 3 + ions should not lead one
to conclude that the same is true of all other electrons.
11.3 Electronic structure of the lanthanide and

actinide Ions
Although the idea can be over-emphasized,5 there is little doubt that f
electrons are tucked away well inside the atoms of the lanthanides and
4 The argument used here is correct for all but one of the geometries of interest. It does not
hold for the group Dx> (which is appropriate for the important ion UO~+), where the f orbitals
split into A 2" + E~u + E2 ., + £ 3 ,., and no symmetry species appears twice.
5 A note of caution is appropriate because we have ignored relativistic effects. These effects
most directly concern ls electrons because s orbitals do not contain any node at the nucleus.
The Is electrons are distinguished because they may be very close to the nucleus and only avoid
capture because of their high speed, which for the heavier elements approaches the speed of
light. This, relativistically, increases the mass of the Is electrons which in turn means a smaller
orbital (in the Schrodinger model of H-like atoms the most probable distance of a Is electron
away from the nucleus is inversely proportional to the electron mass; the essentials carry over
into more complicated atoms). So, the ls electrons screen the nucleus a bit more effectively
than expected on a non-relativistic model. Seeing a smaller positive charge on the nucleus, f
and d electrons occupy orbitals which are both larger and have lower ionization potentials
than we might have expected, thus increasing their availability for bonding. In the context of
the present chapter this is particularly important in the context of the f orbitals. Although the
primary relativistic effect concerns the ls electrons, all s orbitals have to remain orthogonal to
each other. So, if the ls contracts, so too must all others orbitals, to maintain orthogonality.
There therefore is an effect on outer s electrons also, an effect which seems particularly
pronounced for the elements Pt, Au and Hg.
fz3 = fz(2z2 -3x2 -3y2)
z
fxz2 = fx (4z2- x2- y2)•
FIJI. 11.3 The general set off orbitals together fyz, = fy 14,, -x' _Y'l is similar
with their shortened and detailed cartesian
angular forms.
y
z
X
fx(X2 -3.)<1)• similar is fy(Jx' _ Y'l
fxyz
actinides. This means that they are rather insensitive to their molecular
environment. So, the crystal field splittings of f orbitals are about one-
hundredth of those of d orbitals. That is, in the language of crystal field
theory, all lanthanide and actinide complexes are weak field, high spin.
Similarly, the f-f electronic spectra of the complexes are very similar to those
of the free ions as seen in arc spectra. The fact that the spectra of, nearly,
isolated ions can be seen in crystalline materials has led to extensive studies
aimed at their understanding. We will return to these spectra later in this
chapter; it is the purpose of the present section to begin to assemble the
background which will make such a discussion possible.
Electronic structure of the lanthanide and actinide ions I 245
Clearly, our discussion of the f electrons of lanthanides and actinides must

be appropriate to the free ion with, initially, total neglect of crystal fields. If
we insisted, we could work with the f orbitals of the previous section-and
they have the clear advantage of representing static charge densities.
However, we will prefer to use rotating charge densities, for they are then
associated with an angular momentum and it turns out that this angular
momentum is quantized. The connection between static and rotating6 charge
densities and with angular momentum is simple and connected with
the number of planar nodes of the static orbitals seen down the z axis. (In
the case of the free ion it is helpful to think of it as being subjected to a
weak electric field which thus enables a z axis to be defined; in a molecule
it is even easier to think of such electric fields defining a z axis.) Figure 11.4
indicates the answer. It shows that one pair of general f orbitals, f*' _ 3 ,,>
and f,( 3 x'-y')' have three planar nodes containing the z axis. Static charge
distributions are obtained from a pair of rotating charges, members of the
pair being equivalent except that they are rotating in Qpposite senses. The
two rotating charge densities are combined in phase and out of phase to
give the two static distributions. So, the two static f orbitals of Fig. 11.4 are
combinations of rotating distributions, rotating distributions which have
angular momentum quantum numbers of 3 and - 3. As Fig. 11.4 shows, the
number 3 reappears in the two static orbitals as the number of planar nodes
containing the z axis. Similarly, the static orbitals fx,z and fz<x'-y'>' each with
two nodal panes containing the z axis, are combinations of those rotating
orbitals with angular momentum quantum numbers 2 and -2. The orbitals
fxy' and f,z,, with one relevant nodal plane, are derived from the rotating
orbitals with angular momentum quantum numbers 1 and -1. Finally, the
static orbital fz, is also the function with angular momentum 0; as it has 0
angular momentum it is a static function already.
We can now move towards finding the electronic ground states of all the
trivalent lanthanides and actinides. We already know that these are weak
field, high spin, species, with the maximum number of unpaired f electrons
possible. This means that as far as possible the f electrons occupy different
orbitals. It is at this point that the use of rotating charge densities, orbitals
classified according to their angular momentum, becomes convenient. An
example shows how this works. Consider the f 2 configuration. Electron
repulsion means that the electrons will stay as far apart as possible and this
means that they will occupy the maximum number of different orbitals. So,
we allocate each electron to the orbital with the highest angular momentum
available. For the f 2 configuration this means that the first electron is
allocated to the orbital with angular momentum 3 and the second to the
orbital with angular momentum 2. The total angular momentum is the sum,
3 + 2 = 5. It is this simplicity of addition which makes rotating charge
densities so much simpler to work with. The relationship between the total
angular momentum and the term of the ground state is the usual one:
Total angular momentum 0 1 2 3 4 5 6 7 8

Term symbol S P D F G H I J K
6 Note that we use the terms static and rotating charge densities (or orbitals) in preference
to the more conventional names real and complex because they are felt to be easier to understand
by most readers.
fx (x2 - 3y2) } 3 nodes

fy(3x2- y2)
The diagram shows fx(x2 -3y2)·

X
The fy( 3x2 _ y2) orbital is obtained
by an anti clockwise rotation of 30•.
Rg. 11.4 The nodal patterns of the general

set off orbitals, viewed down the z axis (cf. f, (x2 - y2) } 2 nodes
Rg. 10.2(a)). fxyz
fz(x2-y2) is shown; fxyzis obtained by

X
an anticlockwise rotation of 45•.
~:} 1 node
fxz2 is shown; fyz2 is obtained by
an anticlockwise rotation of go•.
X
fz3. 0 node
It follows that the ground state term of an f 2 ion is H. Because we have

two unpaired electrons it is 3 H. Moving to the f3 configuration, the extra
electron is allocated to an orbital with angular momentum 1, giving a total
angular momentum of 6 and a 4 / term. Continuing this pattern gives the
data in Table 11.3, which also shows the details of each calculation.
Had the f electrons in complexes of the lanthanides and actinides
experienced crystal fields comparable to those in complexes containing
d electrons, we would next have to include these fields, just as was done in
Chapters 7 and 8. But these fields are small-much more important
Spin-orbit coupling 1 24 7
Table 11.3 Trivalent lanthanide ion ground state characteristics. For all of these ions
the f electrons are the ones that determine the (many electron) term of the ground
state of the ion
lon f electron Angular momentum Total orbital Term associated

configuration of relevant orbitals angular with the
momentum ground state
La"' fO none 0 ls
eelu fl 3 3 2F
Pr111 f2 3,2 5 3H
Nd 111 f3 3,2,1 6 4/
Pmul f4 3,2, 1,0 6 5/
Sm111 f5 3, 2, 1, 0, -1 5 sH
Eu 111 t• 3, 2, 1, 0, -1, -2 3 7F
Gd 111 f1 3, 2, 1, 0, -1, -2, -3 0 •s
Thill t• 3. 3 7F
Dy"' f9 3,2 5 sH
Ho111 flO 3,2,1 6 5/
E~" f11 3,2, 1,0 6 4/
Tmul f12 3, 2, 1, 0, -1 5 3H
Yblll f13 3, 2, 1, 0, -1, -2 3 2F
Lu 111 f14 3, 2, 1, 0, -1, -2, -3 0 ls
a From Tb111 onwards, for simplicity, the half.filled shell that is also present is not detailed.
are the effects of spin-orbit coupling and it is these that we now consider.
As will be seen, the effects of spin-orbit coupling are most simply covered
using rotating orbitals, although later an attempt will be made to give their
static equivalents.
11.4 Spin-orbit coupling

The phenomenon of spin-orbit coupling was first met in Section 8.4, where
we were interested in its spectroscopic consequences and, again, in Section
9.5 when it was relevant to the magnetic properties of transition metal
complexes because its effects are much greater than those of the magnetic
field. In this latter section it was found convenient to regard spin-orbit
coupling as resulting from the coupling of orbital, solenoid-like, magnets and
spin, bar-like, magnets. There is a fairly evident connection between rotating
orbitals, solenoid-type magnets and angular momenta. Now, the phrase 'the
electron has an intrinsic spin' is actually equivalent to the statement that
'the electron has an intrinsic angular momentum'. Angular momentum is a
characteristic of both orbital and spin magnets. So, equivalent to the
statement that 'the orbital and spin magnets couple together in the phe-
nomenon of spin-orbit coupling' is the statement that 'the coupling of spin
and orbital angular momenta gives a resultant angular momentum'. This is
a more useful statement because it means that we can treat spin-orbit
coupling with no more difficulty than we had in compiling Table 11.3. Let
us look at an example, that of a single f electron. An individual electron can
have spin, that is angular momentum, of either +!or -!. In Table 11.3 the
f 1 case seems almost trivial; a single f electron, one electron with spin ±!
in an orbital set which has a maximum angular momentum of 3, gives rise
to a 2 F term. Bringing these angular momenta together, there are only two
possible resultants, 3 ± f, i.e. i and 1. So, spin-orbit coupling splits the 2 F
ground state term of cerium(lll), an f 1 ion, into two levels, 2 F112 'doublet
F seven halves' and 2 F512 'doublet F five halves'. The subscripts are the
two possible values of j, where j denotes a sum of spin and orbital angular
momenta, j = I + s. It is perhaps helpful to think of this splitting resulting
from the interaction of two magnets (which, for simplicity can be thought
of as bar magnets), side by side. The lowest energy arrangement is that in
which the north pole of one magnet is next to the south pole of the other.
So, here, the 2 F 512 is the more stable level. It is a general pattern for atoms
that an angular momentum ofm implies a (2m+ I) degeneracy. For instance,
spin functions with s = f are doubly degenerate; s functions with angular
momentum of 0 are singly degenerate; p functions with an angular momentum
of I are threefold degenerate and so on. So, here the j = i functions are
eightfold degenerate and the j = t are sixfold, a total of I4 functions. This
is just the number implied by the parent symbol, 2 F; a 2 x 7 fold degeneracy.
The reader may reasonably complain that we have treated the two angular
momenta in different fashion in the above development. We have taken the
maximum orbital angular momentum and added to it, in turn, each component
of the spin angular momentum. It is by no means self-evident that this
procedure is valid; unfortunately it would break the continuity of the
discussion too much to present a justification for it here but one is given in
Appendix I2. Using this procedure we arrive at the spin-orbit split levels
given in Table Il.4.
The final question to be answered for each configuration is which of the
spin-orbit levels becomes the ground state? An answer has already been
given for the f 1 case, where it was concluded that the smallest value of j
becomes the ground state (an analogy was drawn with a pair of magnets
being most stable when the north pole of one is next to the south of the
other). This result is general for the first half of the lanthanide series, the
lowest value of j is the most stable. Just as was found for d electron
systems, so too for f electron systems; for more than half-filled shells it is
simplest to work in terms of holes rather than electrons. Now, if in an orbital
there is a single electron with spin up, then if we describe this situation using
the hole formalism then we have to talk of a hole with spin down (this is
the spin of the electron which is absent). It follows that the spin-orbit state
which is most stable when talking about electrons will be the least stable
Table 11.4 Spin-orbit levels arising from f electron ground state Russell-Saunders
terms
f electron configurations Term Spin-orbit levels

f\ f13 2p 2p5/2• 2p1/2
f2, t12 3H 3H4, 3Hs, 3Hs
f 3 , f 11 •t 4/9/2• 4 /11/2• 4,13/2
f4, flO 5/ s/4, sis, sis, s/7> sla
f5, f9 BH 8H5/2• 6 H7!2• 8H9/2• "H11/2• 8H13/2• 8H15/2
f 8, f 8 'F 7Fo, 7p1• 1p2• 7F3, 7p4• 7Fs. 7ps
t' •s •s,/2
Spin-orbit coupling in pictures 1 249
Table 11.5 Ground and low-lying electronic levels of the lanthanides and actinides
f electron Relevant ions Ground state Low-lying excited levels

configuration
fO Lalli, Aclll lSo
fl eem. Thill 2Fsp 2F7!2
f2 p~n. Pam 3H• 3Hs, 3He
f3 Nd 111 ,U 111 .,9/2 4111/2• 4 ,13/2• 4,15/2
t• Pm 111 , Np111 5'· sis, sla, s/7, sla
f5 Sm 111 , Pum "Hsp aH1f2• eH9/2• sHll/2• eH13/2• eHlS/2
t• Eu 111 , Am 111 7Fo 7Fl, 7F2, 7Fa, 7p4• 7fs, 1fe
t7 Gd 111 , Cm 111 "s112
t• Tb111 , Bk111 7Fe 7fs, 7f4, 7fg, 7p2• 7F1, 7Fo
f9 Dy'". Ct'" 6H1s12 sH13/2• sH11!2• eH9/2• eH7/2• eHs/2
flO Ho111 , Es111 s,. 5'7· 5'6• 5'~·· ~14
fll E~ 11 , Fm 111 4 /15/2 .,13/2• .,1,;, ••,9/2
t12 Tmm, Md'n 3H• 3Hs. 3H•
t13 Ybm, Naill 2F7!2 2Fsp
f14 lulll, uJII 1so
Note the symmetry in this table (compare the first level listed for an f" ion with the last listed for the 114 -").
This symmetry is detailed in the text.
when talking about holes. So, for the second half of the lanthanide series
it is the spin-orbit state with the highest j value which becomes the ground
state. This is not the easiest of arguments to follow and so an example is
given in some detail in Appendix 14. For the f1, the half-filled shell case,
where the orbital angular momentum contributions sum to zero (although
the spins most certainly do not) there is no spin-orbit splitting.
We are now in a position not only to detail the ground states of all the
lanthanide and actinide trivalent ions but also to give some of the low-lying
excited states when these result from spin-orbit splitting. Excitation to these
low-lying levels corresponds to energies in the infrared or near infrared
regions of the spectrum for the lanthanides, the splittings resulting from
spin-orbit coupling being of the order of 1000 em -I. These states are detailed
in Table 11.5. This Table has an underlying symmetry, perhaps most readily
revealed if the 8 S, f 7 , entry is moved sightly to the right and then regarded
as an approximate centre of symmetry.
11.5 Spin-orbit coupling In plctures7

As has been pointed out, for the lanthanides and actinides spin-orbit
coupling has a greater effect on f electron energy levels than does the crystal
field generated by the surrounding ligands in their complexes. In the previous
section it has been seen that it is not difficult to introduce spin-orbit
coupling mathematically if rotating orbitals are used. The question inevitably
arises of whether it is possible to give any sort of static orbital picture of the
7 This section is an attempt to make understandable the subject of so-called double groups.
These are important for an understanding of systems with an odd number of electrons, not
only for f electron systems but for d also.
Fig. 11.5 View down the z axis of orbitals with

4, 2 and 1 nodes containing the z axis.
dxy
Py
phenomenon. One would expect the answer to be in the affirmative. After

all, the (static) p and d orbitals with which the chemist is so familiar actually
first appear as rotating orbitals, when the Schrodinger equation for the
hydrogen atom is solved. In this section is described the static orbital pictures
of spin-orbit coupled functions. Unfortunately, the pictures contain a feature
which makes them less immediately useful than those of p, d and f orbitals.
We start with an unexpected step. We introduce the orbital gxy(x'-y')· The
angular part of this orbital is shown in Fig. 11.5 along with dxy and Py· This
series is chosen because the g orbital is a static orbital corresponding to
angular momentum 4 (it has four planar nodes containing the z axis), the d
orbital corresponds to angular momentum 2 and the p orbital to angular
momentum I. Each corresponds to one-half of the angular momentum of
the orbital which precedes it. Figure 11.5 has been used to compile Table
11.6, a Table in which we have included the next member in the geometric
series 4, 2, I,!. We can learn something about the static orbital picture of
the angular momentum = ! case by extrapolation from the pattern revealed
by the other three. The reader should now study Table 11.6 and complete
it by filling in the missing entries, by replacing the question marks by the
appropriate number and angle.
Table11.6
Property Function
-- --
g,(x'-r> dxy Spin!
--
Py
I angular I -c2 -c2 -c2
-- -- --
1
4 2 ~
momentum
-c2 -c2
Number of angular 4 2 ?
-- -- --
nodes (Rg. 11.5)
x2 x2
Smallest angle between 45' 90' 180' ?
nodes (Fig. 11.5)
The number, of course, is one-half and the angle three-hundred and sixty
degrees. This seems nonsense. What meaning can be given to half a node
and what does it mean to have three-hundred and sixty degrees between
complete nodes? Actually, both are entirely sensible. The essential step is to
change the identification of the identity operation (the E at the head of
character tables, the 'leave alone' operation). This operation can be equated
with a rotation of 360° (although this equation is not often explicitly stated).
Instead, we now equate it with a rotation of 720°. It is easy enough to make
this change for the functions shown in Fig. 11.5. This is done in the top row
of Fig. 11.6 which shows the angles oo through to 360°, compressed into the
(actual) region oo to 180°. The diagrams are completed in the middle row,
where the angular pattern from 360° through to 720° simply duplicates that
for oo to 360°. In this row the static counterpart of the angular momentum
! case has been included, which can be seen to comply with the requirements
imposed on it by the completed Table 11.6. It is evident that these diagrams
are becoming rather congested and so at the bottom of Fig. 11.6 are given
simplified pictures, drawn in a pattern that will be followed for the remainder
of this section.
Although a picture of an angular momentum = ! function (more accurately,
a 1!1 function) has been obtained, it still seems somewhat artificial. It can be
given more physical reality using a Mobius strip. The top of Fig. 11.7 shows
a long strip of paper, creased so that it can easily be bent (away from the
viewer) along its central axis (indicated by the arrows). Lobes are drawn
with phases as shown, the lobes being terminated by nodes at the edge of
the paper. Suppose that the back of the strip of paper is coated with a contact
adhesive, so that as soon as the strip is folded back it sticks to itself. Join
the ends of the strip together but in doing so twist the strip so that the back
comes to the front. The result is shown at the bottom of Fig. 11.7. Starting
at any point on this Mobius strip it is necessary to go round twice,
akin to rotating by 720°, before regaining the starting point. The nodal
pattern encountered in traversing the Mobius strip is just that shown
in Fig. 11.6 for the angular momentum =!function. Having pictured a static
angular momentum ! function it is not difficult to extend the approach to
other half integer functions and this is done in Fig. 11.8, which also shows
the corresponding unfolded, unstuck, Mobius strips. A few points remain
to complete this section. Just as in all other cases except the angular
momentum= 0 case, all functions appear in pairs (only one member of each
i')
U1
i')
..
~
~
540
a
:::1
540 540 ..
~CD
3
!'!
720 :Et
:CD
360 I === :::w:::: .......:: • X 360 I :::== 7t\: ::::::::::::::: • X 360 X
0 :fir
·a
::r
..
:::1
a:
CD
180 180 180 :
. :::1
...
Q.
..
~
:;·
CD
.a:
720
360 I 3> <: I • X 360 ::;:;. )I(< • X 360 >X< I X 360 X
=-= • 0
+ 720
I T::-:=>Y<C .· I~-;- X -+----:;;,I(E----+-- -x -+-------3~---+--x X
0
+ -
Fig. 11.6 The identity operation~ 720' representation of the orbitals of Fig. 11.5, together with the spin~~ function.
............................................................................................................................................................................................
FIC. 11.7 The Mobius strip representation of a

spin = ~ function. There is a node on each side
of the strip at the point indicated by an arrow
(this is the point of contact of the two ends of
the top diagram). In the bottom diagram only
the signs of the function are given. There is no
attempt to indicate amplitude. Note that two Stick, twist t and join
different MObius strip representations exist-the
twist can be made in either of two senses.
Angular momentum = I~I
Fig. 11.8 The MObius strip constructions for

angular momentum = ~ and ~functions.
Angular momentum = I~I

CD 8
720 720
360 0 360 0
-
The two static functions corresponding
® ®*.
to angular momentum ~ ± ~
~. 11.9 The pairs of static (real) functions
corresponding to angular momentum~ ±~
and±~.
720 720
'"' 0 ""' - 0
The two static functions corresponding

to angular momentum = ± ~
pair is shown in Fig. 11.5). So, too, for the angular momentum = f, ~, ... ,
and so on cases. Figure 11.9 shows both members of the f and ~ pairs.
Finally, this section has been entirely concerned with half-integer values of
the angular momentum. Spin-orbit functions may also have integer values
of angular momentum; for these the pictures look the same as for the familiar
orbitals with the same angular momentum, although their meaning, of
course, is somewhat different.
11.6 Excited states of f electron systems

When d electron systems were considered in Chapter 7 it was found that a
detailed study of the d 1 and d 2 configurations could easily be extended to
cover all the low-lying excited states of the same spin multiplicity as the
ground state for all d" configurations. Even if spin-orbit coupling could be
ignored, the f electron case is more complicated. As Table 11.5 shows, not
only f 1 and f 2 but also f 3 configurations would have to be included before
all the possible ground terms F, H and J, had been covered. But spin-orbit
coupling cannot be ignored and this means that we cannot even talk about
spin allowed and spin forbidden transitions and thus restrict the discussion
to terms with the same spin multiplicity. Faced with this complexity, the f 2
problem and its solution will be outlined. This will enable us to see how to
tackle the other configurations-the extension of our arguments to include
them is more tedious than difficult. Despite what has just been said, we
shall at first appear to ignore spin-orbit coupling. However, we shall also
ignore any restriction on spin multiplicity (in Chapter 7 the discussion was
restricted to terms with the same spin multiplicity as the ground state) and
this will enable us, later on, to mix the different spin multiplicities by the
mechanism of spin-orbit coupling.
In Section 11.3 a 3 H ground term for the f 2 configuration was obtained
by feeding electrons with spins parallel into rotating orbitals characterized
by angular momenta of 3 and 2. Had we been prepared to pair these electrons
they could both have been placed in the orbital with angular momentum 3
Excited states of f electron systems 1 255
Table 11.7 Terms arising from the f 2 configuration (left-

hand column) and states consequently resulting from
spin-orbit coupling
Russell-Saunders term Spin-orbit states
's 'So
3p lpo 1p1 tp2
'o 'o,
3F 3p2 3F3 3F4
'G 'G•
,,
3H 3H4 3Hs 3H6
'16
to give a 1 I term. This 1 I term and the 3 H are the start of a neat pattern
based, ultimately, on the pattern that the angular momenta of the two
electrons can either add or subtract, 8 but never go negative. The greatest
orbital angular momentum they can have as a pair is 6 and the smallest 0,
seven values in all. Write down, in order, all the term symbols corresponding
to all values of the orbital angular momenta from 0 to 6. We obtain
S+P+O+F+G+H+I
Now add the only two possible spin labels, I and 3, alternately, in such a
way as to include 3 H and 1I. We obtain:
15 + 3p + 1 0 + 3F + 1G + 3H + 1/
This is a complete list of the terms that arise from the f 2 configuration. A
check on the correctness of the result can be obtained by counting
wavefunctions. For the f2 configuration the first electron can be inserted in
any one of 14 ways (we have seven orbitals, the spin can be either up or
down); the second electron can be fed in in any one of 13 ways (it cannot
be in the same orbital as the first and also have the same spin). Because the
electrons are indistinguishable we have counted every possibility twice-so,
it and t i have been counted as different. It follows that the total number
of wavefunctions is 14 x 13/2 = 91. This number, 91, is the same as the
number of functions contained in the terms that we generated above:
15 + 3p + 1 0 + 3F + 1G + 3H + 1/
1 + 9 + 5 + 21 + 9 + 33 + 13 ~ 91
It is now time to introduce spin-orbit coupling and this is done using the
procedure that has already been described in Section 11.4 and Appendix 12.
We simply add each component of the spin angular momentum to the orbital
angular momentum. The result is given in Table 11.7, where the final levels
are stacked according to their total angular momenta, according to their j
values, this being given by the subscripts. Now comes an important point;
8 The reader may reasonably object-what about cases such as -3 combined with -3, this
gives us - 6? In fact such cases are included in the discussion in the text. So, for example, the
I term, from the 1 /, comprises 13 (2L + 1; here L ~ 6) different functions, differing from each
other because the z component of the orbital angular momentum spans the 13 values from + 6
through to -6. We do not have freedom to use the -6 function a second time; the angular
momenta which are reflected in term symbols such as S. P, D, F, , are never negative.
~~~ 1 . "
"
/
----1so
45000 s //
I
Fig. 11.10 The electronic energy level diagram I
for Pr111 (4t2). A typical crystal field splitting is I
shown at the right; its effects are much smaller
than those of spin-orbit coupling. As with 20000
Tanabe-Sugano diagrams, the horizontal axis is
taken as the ground state (3 H4 ). This enables ~
'E
__....---10:!
the effects of interactions between levels with
the same total angular momentum to be more " 15000
_1._
clearly seen. Note, for instance, the relative
upward displacements of levels with the j
quantum number 4 (3 H4 , 3 F4 , 1 G4 ).
10000
T
Crystal
field
....-----3H5
0 L-3..:H.:....""""":;::":::/_"_ _ _ _ _ _ _3:_:H4i.__ _ __
Electron repulsion Spin-orbit coupling
levels which are in the same column can interact together because of
spin-orbit coupling; they have the same values of j. In terms of the pictorial
representation of the previous section, the spin orbitals that interact have
the same number of nodal planes containing the z axis; they have the same
symmetry. As is evident from Table 11.7, some levels stand in isolation and
so remain unchanged by spin-orbit coupling (' P1 for instance); for others,
mixing is between functions with the same spin ('S0 and 1 P0 ). However, the
fact that the ground state, 3 H 4 , mixes with 1 G4 ensures that all of the f -> f
electronic transitions of an f 2 ion such as praseodymium(III) are of mixed
spin character (a mixture of spin triplet and singlet), emphasizing the fact
that because of spin-orbit coupling it is not useful to attempt to talk of spin
on its own.
The above discussion has been relatively superficial but has covered all
of the important points. In Fig. 11.10 is given an actual energy level diagram,
which is not relatively superficial, and which shows how the primary
separation between levels is due to electron repulsion, a topic which was
considered in some detail in Chapter 8, in particular. Note the way that
spin-orbit coupling changes the levels. So, although the 3 H 4 is the most
stable level originating in the 3 H, 3 F4 is the least stable of those coming from
3 F; the splitting between 3 F and 1 G is much greater than that between 3 F
4 4
and 1 G. Several other similar patterns can be discerned in Fig. 11.1 0.
This completes our discussion of the excited states off electron systems.
For completeness, however, Table 11.8 gives a more complete list of the
Russell-Saunders terms arising from all f" configurations. It does not include
the effects of spin-orbit coupling, which have to be worked out for each case
individually. The reader may find it a helpful exercise to select a configuration
from Table I 1.8 and use it to construct a table similar to Table 11.7.
Electronic spectra off electron systems 1 257
Table 11.8 Terms arising from fn configurations
f electron configuration Number of Total number Terms arising

functions listed of functions
fl, f13 14 14 2p
t2' f12 91 91 3H + 3p + 1G + 10 + 3p + 1/ + 15
f3, f11 294 364 ~+~+~+~+~+~+~+~+~+~+~+~
~·, 11o 394 1001 ~+¥+~+~+~+~+~+~+~+~+~+~
15, 19 238 2002 6H + 6p + •H + •o
+ 6p + ., + •s
16 , 18 49 3003 7p
I" 478 3432 ~+~+~+~+~+~+~+~+~+~+¥+~
For tn, 3 < n < 11, only a selection of terms is given, listed in approximate energy sequence (lowest first). A limited number is given because mixing of the type
shown in Fig. 8.16 occurs and particularly complicates the more highly excited states of those configurations for which the total number of functions is large.
11.7 Electronic spectra of f electron systems

As a class, the complexes of the lanthanides are both beautifully and
delicately coloured. There is a symmetry in the colours of the ions in aqueous
solution, a symmetry which seems to be accidental. These colours are given
in Table 11.9, listed in a way which shows the colour parallel between f"
and f 14 -" ions. Also shown are the colours of some organo-lanthanide
compounds, which do not show anything like the same pattern, colours
which support the idea that the colour pattern of the aqua ions is accidental.
The actual absorption bands that occur in the spectra of the lanthanides
and actinides and which are associated with electronic transitions have been
divided into three types.
1. f-> f transitions These transitions are localized entirely within the f shell
and so, like d -> d transitions, are formally forbidden. However, like
d -> d, they actually occur and give rise to a large number of weak, sharp
bands from the infrared through to the visible region. The absorption
band patterns obtained from species in solution are closely related to the
emission spectrum of the corresponding ion, obtained from the arc spectra
of the elements.
2. nf-> (n + l)d transitions Here, then and (n + I) arc principal quantum
numbers so these are allowed bands in which a 4f (lanthanides) or 5f
(actinides) electron is promoted to a 5d and 6d orbital, respectively. They
give rise to quite intense and broad bands, lowered by ca. 15 000 em - 1
in solution compared to the gaseous ion.
3. Ligand -> metal f (electron-transfer bands) These are usually intense,
broad bands which lie in the ultraviolet region. The charge transfer
means that in the excited state there is one more f electron than in the
ground state.
Of these three types of transition, it is the first which has received, by far,
the greatest study, even though they are the weakest bands. To put this in
perspective, Fig. 11.11 shows the electronic spectra of aqueous Ce 111 , 4f 1, and
Pr 111 , 4f2. Only in the latter are f-> f transitions evident; this is because for
the Ce 111 the only transition is between the two spin-orbit components of
Table 11.9 Typical colours of lanthanide complexes
Configuration I on Colour Configuration Ion Colour

Aqueous ions
t' Celli colourless t13 Ybtn colourless
f2 Pr'" yellow-green f12 Tmlll light green
13 Ndtn red-violet f11 Er'" lilac
14 Pm111 pink flO Hoi II yellow
f5 Smlll yellow fg Dy'" yellow-green
16 Eulll pale pink f" Tb 111 pale pink
17 Gd 111 colourless
Lanthanide tricyclopentadiene complexes•

tl Celli orange f13 Yb 111 dark green
f2 Prill greenish f12 Tmlll yellow-green
13 Nd 111 blue f11 Er'" pink
14 Pmlll yellow-orange flO Ho' 11 yellow
Is Smlll orange fg Dy"' yellow
16 Eulll brown t• Tb 111 colourless
f7 Gdlll yellowish
a The crystal structures of these complexes show three cyclopentadiene rings q5 -coordinated to each
lanthanide, one of these rings additionally being bonded in q1 or q2 fashion to an adjacent lanthanide.
Fig. 11~11 (a) The electronic spectrum of Ce 111 ,

11 is dominated by charge-transfer bands. The I I
f - f spectrum is expected to be very simple, 50 40 30 20 10 0
consisting of a single transition between the
two spin-orbit levels given in Table 11.4. (a) cern f 1
This transition will be both weak and in the
infrared-towards the right-hand side of
Rg. 11.11(a)-where it is lost somewhere in
the forest of vibrational bands that occur in this
spectral region. (b) The electronic spectrum
~-L--~
of Pr111 , an f 2 ion, is more characteristic of
lanthanides, showing both charge transfer and
f - f transitions (sharp). The involvement of
f-f electron repulsion moves some of these
transitions into the visible region of the
spectrum. The relevant energy level diagram
is given in Fig. 11.10.
40 24 20 16 10 5 0
cm-1 x 103
(b) Prm f 2
the 2 F term. It falls in the infrared region, where it is difficult to distinguish

it from the forest of vibrational bands; not all of the electronic transitions
of an ion may be available for study. As a simplification our discussion will
be restricted to the lanthanides. For the actinides the spin-orbit coupling
is so great that the spin-orbit components of different Russell-Saunders
terms overlap (only the 1J overlaps with the 3 Pin Fig.ll.IO; a corresponding
diagram for an actinide would be much more complicated). Secondly, crystal
Electronic spectra of f electron systems 1 259
fields are about twice as great for the actinides as for the lanthanides so that
it is less acceptable to ignore them in the way that we shall do.
When a lanthanide ion is at a centre of symmetry, in the hydrated
pcrchlorates where the lanthanide is octahedrally surrounded by water
molecules, for instance, the f--+ f transitions are an order of magnitude
weaker than in low-symmetry complexes, indicating that one intensity-
generating mechanism is the mixing off with other orbitals (presumably d,
because this would lead to an allowed f--+ d component in the transition)
by the low-symmetry crystal field, a mechanism analogous to that discussed
for low-symmetry transition metal complexes in Chapter 8 (where d-p
mixing was invoked). Just as for transition metal complexes, too, it seems
that the mechanism by which the transitions in centrosymmetric complexes
gain intensity is through the dynamic distortions caused by molecular
vibrations (for transition metal complexes this mechanism was discussed in
Section 8.7). Support for this explanation comes from experiments such as
those in which a low concentration of a lanthanide ion is doped into
crystalline Cs 2 NaYC1 6 . The yttrium is an f 0 ion and octahedrally surrounded
by CJ- ions. Particularly with low-temperature samples, it is possible to
observe mixed electronic+ vibrational (vibronic) transitions for the doped
lanthanide ion, consistent with a vibrational mechanism for generating the
intensity.
The point in our discussion has now been reached at which the parallel
with d --+ d transitions is little help; new mechanisms for generating the band
intensities have to be introduced. The reason for introducing such new
mechanisms is that those given above prove inadequate when really tested.
For instance, detailed evidence that a vibrational mechanism is involved in
making f --+ f transitions weakly allowed in centrosymmetric complexes has
just been presented. But what of the band origin which is seen under
high-resolution conditions? It has no vibrational component and so no
vibrational explanation can be given for its appearance. The explanation
which is usually given for it is that it is magnetic dipole allowed, an
explanation which itself calls for an explanation! In the Maxwell model, a
monochromatic polarized beam of light consists of two vectors, an electric
vector and a magnetic vector. These are mutually perpendicular to each
other and both are perpendicular to the direction of propagation of the light
wave. Normally, attention is confined to the electric vector because calcula-
tions indicate that the interaction between the magnetic vector and a
molecule is much weaker than that involving the electric vector. 9 Magnetic
vectors, and magnetic fields, cause circular displacements of charge (think of
the path followed by a charged particle in a conventional mass spectrometer).
Such rotations of charge, like all rotations, are symmetric with respect to
inversion in a centre of symmetry. This is in contrast to an electric vector,
and the linear charge displacements to which it gives rise; they are anti-
symmetric with respect to inversion in a centre of symmetry (Fig. 11.12). So,
magnetic-dipole-allowed transitions are g in nature whereas electric-dipole-
allowed transitions are u. Now, f--+ f transitions are u --+ u and u x u = g.
So, although f --+ f transitions are electric-dipole-forbidden they are magnetic-
9 Not that it is zero-all magnetic resonance measurements rely on the fact that the
interaction between an oscillating magnetic field and a molecular system is non-zero.

Invert in a centre
of symmetry
Fig. 1L12 Inversion in a centre of symmetry Translations have u symmetry

changes a translation into a translation in the (a)
opposite direction (a), whereas the sense of a
rotation is left unchanged (b).
Q Invert in a centre
of symmetry
0 Rotations have g symmetry (the rotation is

in the same sense after the inversion)
(b)
dipole-allowed. This is why it is believed that a magnetic-dipole mechanism

is responsible for the appearance of the band origin, with no vibrational
component, in high resolution studies of the spectra of lanthanide ions doped
in host lattices. Note that a similar explanation could apply to the corre·
sponding d -+ d transitions and has, indeed, been suggested.
The fact that a beam of light travels more slowly through a totally
non-absorbing block of glass more slowly than through a vacuum shows
that light interacts with matter, whether or not absorption takes place. That
such an interaction should take place is not surprising-as has just been
pointed out, light waves behave, in part, like oscillating electric dipoles. They
therefore cause electrons to move, to oscillate in sympathy with the electric
vector of the light wave. A fascinating f -+ f band intensity mechanism has
been proposed based on this phenomenon and is currently attracting much
interest. It has been suggested that the light wave used in absorption studies
polarizes the ligands and that the lanthanide ion feels the dipoles created.
That is, it is suggested that the ligands provide a non-centric field at the
lanthanide ion not by virtue of any vibrational movement of their nuclei but
because of the instantaneous polarization of their electron density caused by
the light wave itself.
11.8 Crystal fields and f--+ f intensities

As has been made clear, crystal field splittings are not an important aspect
off electron systems. In solution and the solid the intrinsic half-widths of
the bands observed in the f -+ f spectra are typically 50 em- 1 and the crystal
field splittings seem much the same, so that there is little information on
crystal fields to be gained from f-+ f spectra. Nonetheless, there is clear
evidence that crystal field effects do play a role, albeit not a dominant
role-and not a well-understood one either.
The first evidence of crystal field effects is the recognition that lanthanide
and actinide ions in solution and the solid state show a red shift of their
spectral bands. This has been interpreted in terms of a nephelauxetic effect
Crystal fields and f ..... f intensities 1 261
Table 11.10 Some f --+ f hypersensitive transitions. In this table it has been recognized
that the Eu 111 7 F1 state is low lying and is thermally populated at room temperature so
that the hypersensitive transition could involve it as the ground state
Ion Ground state term Excited state levels of hypersensitive transitions

Lalli
'So none
cenl 2F5/2 none
P~" 3H4 3Hs, 3F2
Nd 111 4/9/2 4 Gs!2• 4 G112• 2 G112• 2 K13/2
Pm'n 5/4 5G2, 5G3

Smlll sH5/2 Bfl/2• 6F3/2• 4H7j2
Eu 111 7Fo(F,) 7p2
Gd 111 as7/2 none
Tb 111 7p6 7p5
Dy"' 6 H1s;2 sH13/2• eHH/2• Bfup

He'll 5/s sGs, 3Hs
E~" 4/15/2 2 H1112• 4 G11;2
Tm 111 3He 3H5, 3H4, 3p4

Yb111 2p7/2 none
(a cloud expanding effect; the occurrence of this phenomenon for transition

metal ions was outlined at the end of Section 8.3), and this, in turn, has been
explained by invoking metal-ligand covalent bonding. Whether the explan-
ation is valid is unclear-one author has commented that 'the nephelauxetic
parameters resulting seem not to have much meaning'-but the existence of
the phenomenon is real enough and its explanation clearly has to involve
the interaction of the f electron system with its environment. Secondly, no
less clear and no better explained, are the so-called hypersensitive transitions.
These are f--+ f transitions which have environment-sensitive intensities.
Their intensities may vary by over three orders of magnitude with change
of environment, with change of ligand for instance. The effect of the ligand
environment is made evident by a band intensity change rather than the
band splitting which experience with d --+ d systems would lead us to expect.
A list of some hypersensitive transitions is given in Table 11.1 0; as this table
shows, there is no evident pattern to them. Not surprisingly, the problem
posed by the fact that some, but not all, f --+ f transitions are hypersensitive
has been the subject of much study. Whilst the answer is still not clear, a
common feature seems to be that the hypersensitive transitions are all
quadrupole-allowed. A quadrupole is not a particularly familiar animal for
most chemists, although the increasing use of quadrupole mass spectrometers
is changing the situation. There are several ways of picturing a quadrupolar
arrangement of charges of which the simplest is to think of it as a side-by-side
arrangement of two opposite dipoles; two dipoles head-to-tail. So, in seeking
an explanation for hypersensitive transitions it is natural to look for a
mechanism involving two dipoles. Several such mechanisms have been
suggested. The first is one which builds on the 'polarization induced by the
electric vector of the light wave' model which was introduced at the end of
the previous section. In this explanation, this induced polarization provides
one dipole and the electric vector of the light wave itself provides the second.
262 I f electron systems: the lanthanides and actinides
Of course, the only light wave which is relevant here is that which has an
energy corresponding to the energy of the hypersensitive transition observed.
On this model, the sensitivity of hypersensitive transitions to different
environments arises from the different polarizabilities of the different ligand
environments.
A second mechanism which has been suggested to explain hypersensitive
transitions applies only to low-symmetry molecules, and so, low-symmetry
crystal fields. The mechanism postulates that one of the dipoles is inherent
in the ligand arrangement; the overall ligand environment is so distorted
that it is dipolar. The second dipole required for a quadrupole again comes
from the incident light wave. On this model, different hypersensitivities arise
because of different intrinsic dipoles in the ligand arrangements. This second
model can be modified to cover the case of high-symmetry arrangements,
arrangements which are centrosymmetric and so cannot be intrinsically
dipolar, for example. This modification supposes that the high-symmetry
molecule is distorted by a vibration, such as those vibrations used to explain
the intensities of d -+ d transitions in Section 8.7. These, for an octahedral
complex, are vibrations which destroy the centre of symmetry. Examples,
again for an octahedral complex, are vibrations of T 1 u or T 2 u symmetries.
The vibrationally distorted molecule is then non-centric, at least whilst the
distortion persists (although only for a T 1 " vibration would it have a-
transient-dipole). This, then, is the source of one dipole; again, the second
is that originating in the light wave itself. The hypersensitivity differences
would, on this model, originate in the different vibrational properties of the
different species.
We have then three different explanations for the phenomena of hyper-
sensitive transitions. In general, they are not necessarily mutually exclusive
although the relative importance of each would surely vary from complex
to complex (one could not invoke the second for a centrosymmetric species,
for instance). Unfortunately, although hypersensitive transitions show that
a lanthanide ion is sensitive to its surroundings, until more is known about
the mechanism of hypersensitivity it is not possible to use the phenomenon
to learn more about this sensitivity. One final point, also puzzling: the
actinide ions, despite the greater importance of ligand fields for these ions,
do not seem to exhibit hypersensitive transitions. Even so, some of their
transitions show small, but undoubtedly real, band intensity changes with
change in ligand environment.
11.9 f--+ d and charge-transfer transitions

In Section 11.7 were listed two categories of electronic transitions which
occur in the lanthanides and actinides in addition to the f -+ f transitions
discussed in the preceding section. It is convenient to treat these two classes,
f-+ d and charge-transfer transitions, together since, spectrally, they occur
together, although generally at lower energies in the actinides than in the
lanthanides. Much of the study of these bands has taken the form of seeking
criteria which enable the two different sorts of transition to be distinguished.
One such distinction is that in the lanthanides, the 4f-+ 5d transitions
tend to be narrower than the electron-transfer bands (their half-widths are
Lanthanide luminescence I 263
ca. 1000 and 2000 em-\ respectively). Although either type of band may be
the lower lying, in the lanthanides the charge-transfer tends to occur at the
lower energy when addition of an electron to an f-shellleads to a half or full
electron shell. The f 6 ion europium(III) and the f 13 ion ytterbium(III) are
species for which this pattern holds. The converse pattern is also general.
That is, when an f-. d transition (a transition which means that the number
off electrons is reduced by one) leads to an empty or half-filled shell then
these f -. d transitions are the lower. The f 1 ion cerium(II) and the f 8
terbium(III) provide examples of this pattern. It seems that the f 2 ion
praseodymium(III) is another example, although this could not have been
predicted. When the lanthanide or actinide is one for which more than one
valence state may be studied, then it is relatively easy to distinguish between
electron-transfer and f-. d bands. As the valence state increases, f-. d bands
move to higher energy with increase in valence state whereas the-ligand to
metal-electron-transfer move to lower. Of course, the spectra of two
differently charged ions will be far from identical but, nonetheless, bands
which appear similar in the two spectra show these relationships. Another
distinction, within a given valence state, is based on the fact that, as one
would expect, the d orbitals are much more sensitive to the ligand environ-
ment than are the f, so that the energies of the f -. d bands are dependent
on coordination number whilst the charge-transfer are not. Conversely, for
a fixed coordination number, the energies of the f -. d bands are less sensitive
to change in ligand than are the corresponding charge-transfer bands.
All of these criteria more-or-less follow simple common sense ideas about
the characteristics of the two types of transition. Together, they enable
distinctions to be drawn in most cases. Although the detailed study of the
bands is in its infancy, it is interesting to note that f-. d transition energies
seem to follow a spectrochemical series, just as do d -. d in the transition
metal ions. So, the following 5f-. 6d transition energies, all x 10 3 em -I,
have been reported for complexes of um and other similar ions:
Ligand: 1- < sr- < cl- < so~- < H2 0 < F-
v(f-. d): 13 ca. 17 ca. 19 ca. 22 25 ca. 25
11.10 Lanthanide luminescence

It has long been known that many lanthanide ions fluoresce under ultraviolet
light, the fluorescence coming from f-+ f transitions; some ions which do
not normally fluoresce at room temperature do so when they are cooled.
This fluorescence property has led to lanthanide ions being incorporated in
the phosphor of domestic fluorescent tubes and in the screens of colour
televisions. 10 When an ion is in an electronically excited state there is a
competition between deactivation by radiative and non-radiative processes.
For an ion to be a good emitter, any non-radiative process must be a poor
second in the competition. If studies are carried out using aqueous solutions,
it is found that the lanthanide ions at the centre of the lanthanide series are
10 These screens contain a multitude of tiny clusters of carefully placed patterns of red, green
and blue light-emitting phosphors. A metallic mask immediately behind the screen has matching
tiny holes or slots which allow only electrons from the green cathode to hit the green
light-emitting phosphor in each cluster, the electrons from the blue cathode to hit the blue
phosphor and so on.
the most efficient emitters; not surprisingly, their observed radiative lifetimes
are found to be anything up to a factor of I 0 3 greater than those of the ions
at either end of the series. Much more surprisingly, the energies of the excited
states of the ions in the centre of the series are higher than those for the end
members by a factor of two or three-as has been seen from Table 11.4.
With the exception of gadolinium( III), the ions at the centre of the series have
more f electron excited states and so the energy levels are more widely spread.
One might have expected that the more energy there is to lose, the more
readily it would be lost-but, clearly, this is not the case. It seems clear that
a key part in this phenomenon is played by the solvent (water). So, if D 2 0
is used as the solvent in place of H 2 0, the observed excited state lifetimes
are all increased by an order of magnitude. The explanation seems clear; an
important non-radiative deactivation process is one in which the energy in
the electronic excited state is transferred to an overtone of the v(Q-H)
stretching vibration of the solvent water. A lower overtone level is needed
for the deactivation of the end members of the lanthanide series than for the
central members. Because the frequency of the v(Q-D) vibration is only
about three-quarters of that of the v(O-H), a much higher overtone is needed
in D 2 0 than H 2 0. Remembering that the transitions involved are f--> f, it
is evident that this deactivation mechanism indicates something that we have
met several times before, that f--> f transitions are not totally insulated from
their environment.
Amongst the most efficient of current commercial phosphors are those
based on the red emission of europium(III), the green emission of terbium(III)
and the blue emission of europium(III). These ions are at the centre of the
lanthanide series so that it seems that the phenomena found in aqueous
solution extend to the solid state. In addition, gadolinium(III) may well be
a component of a phosphor but this is because of its excited state lifetime,
not because of any emission. The actual matrix into which the lanthanide
ions are incorporated in a phosphor are oxide or glass ceramic-forming
oxyanions-borates, silicates, aluminates and tungstates. The choice is
important because one step in the emission process is transfer of the absorbed
ultraviolet radiation (absorbed by a carefully chosen impurity species,
commercially called a sensitizer) through the crystalline lattice to the emitter.
An ordered lattice facilitates this energy transfer but may simply serve to
allow transfer to an (inadvertent) impurity ion which provides a mechanism
for deactivation without emission. The degree of order of the host lattice has
to be optimized, as too does the concentration of the chosen emitter.
Incorporation of an ion such as gadolinium(III) at quite a high concentration
(up to 20% of total cations) in the lattice is found to facilitate long-range
energy transfer. The current understanding of the processes involved may
best be illustrated by an example. Consider a borate lattice containing a high
concentration of gadolinium, GdB 3 0 6 • A typical sensitizer is bismuth(III)
and europium(III) will be chosen as emitter. Current notation would write
this composition as GdB 3 0 6 :BiEu. The sequence of events seems to be:
Ultraviolet
light Gd(lli) __, energy transfer Eu(ill) --> emission
L T L T
Bi (Ill) __, excitation Gd(lll) --> energy transfer
Magnetism of lanthanide and actinide ions 1 265
11.11 Magnetism of lanthanide and actinide ions

In Chapter 9 a review was given of the approach adopted in the calculation
of the magnetic properties of transition metal ions and the same general
philosophy applies to the lanthanides and actinides. First, one has to identify
the ground state and then to make allowance for all those interactions in
which it is involved and which, energetically, are larger than the effect of the
magnetic field. Then, this corrected ground state is subjected to the magnetic
field perturbation, the effect of which is usually to split it into a family of
sublevels. These split levels are thermally populated with a Boltzmann
distribution.
As has been seen in Section 11.4, for the lanthanides the most important
factor determining the ground state energy level is the value of J ( = (L + S))
(here, J is the sum of the individual one-electronj values,j =(I+ s)). Indeed,
in practice it is found that a J-only model, with no allowance at all for minor
things such as crystal fields, gives a good agreement between experiment and
theory. What this means is not that crystal fields are smaller than the effects
of the magnetic field, it has been seen above that they are not, but that their
inclusion does not significantly modify the magnetic properties of the ground
state or its relationship with other thermally populated states. For the
actinides the J-only model does not give such good agreement with
experiment, in keeping with the fact that crystal field effects are larger and
so do more than produce the simple additive correction found for the
lanthanides.
It will be recalled that in Section 9.1 it was shown that the spin-only
model for molecular paramagnetism gave rise to the equation
iieff = J4S(S + 1)
It will therefore not be surprising that, essentially by replacing S by J in the
derivation, the J-only model gives rise to the equation
1-'eff OC JJ(J + 1)
The equality sign in the spin-only equation has been replaced by a
proportionality because, whereas there can be no ambiguity about how
a given value of S arises-one simply has to count the number of unpaired
electrons and divide by two-as has been seen, a given J value can arise
from a variety of different spin and orbital components. Spin magnets and
orbital magnets are not immediately interchangeable and so the constant of
proportionality in the J-only equation, denoted g, has to reflect the particular
mix involved. The equation for g is
3 S(S + 1) - L(L + 1)
g= 2+ 2J(J + 1)
The quantity corresponding to g in the spin-only formula is the factor 2

(which appears as 4 when placed inside the square root). In Table 11.11 is
given a comparison of observed and calculated Jlerr values for the lanthanides.
As has already been commented, the general agreement is very good. Two
additional points should be made. First, a small contribution from tem-
perature independent paramagnetism (TIP, the mixing of excited states into
Table 11.11 f.lett values for lanthanide ions
lon g /lett(calculated) Approximate /lett

(observed)
Lalli 1 0.0 0.0
Celli 6/7 2.54 2.5
Pr" 4/5 3.58 3.5
Nd 111 8/11 3.62 3.6
Pm 111 3/5 2.68
Sm 111 2/7 0.84 1.5
Eu 111 1 0.0 3.4
Gd 111 2 7.94 8.0
Tb111 3/2 9.72 9.3
Dy'" 4/3 10.63 10.6
Hou' 5/4 10.60 10.4
Er" 6/5 9.59 9.5
Tm 111 7/6 7.57 7.4
Yblll
8/7 4.54 4.5
Lu 111 1 0.0 0.0
the ground state by the magnetic field, discussed in Section 9.6) is expected
and has, in fact, been included in the calculated values in Table 11.11.
Secondly, the ions europium(III) and samarium(III) do not give good
agreement. The reason for this is known. These two ions possess very
low-lying excited states which are so low in energy that they are thermally
populated. The J values for these low-lying states differ from those of the
ground states and so the J-only equation above has to be modified to take
account of this. When this correction is made, good agreement with
experiment is obtained.
It is evident from Table 11.11 that several of the lanthanides have rather
high values of Jl.ecc· Physically, this means that when salts of these ions are
placed in a strong magnetic field, they become slightly warm. The system is
stabilized, the salts are attracted into the magnetic field, and the energy of
stabilization has to go somewhere; it appears as heat. At room temperature
this effect is small but at low temperatures it becomes rather significant. This
has led to the use of gadolinium salts, especially Gd 2 (S0 4 h ·8H 2 0, and salts
of dysprosium for obtaining low temperatures. The salts are placed in a
container, cooled with liquid helium and a strong magnetic field applied. As
has just been seen, heat is evolved. When the salts have cooled down to
liquid helium temperature again, the magnetic field is switched off. Cooling-
adiabatic demagnetization-occurs as the salts lose their magnetic orientation.
There is an alternation of natural abundances of the lanthanides, starting
with lanthanum (low), 11 cerium (high), so that both gadolinium and
dysprosium are highs, explaining their selection for this application-
gadolinium is not otherwise, from Table 11.11, the most obvious choice.
Clearly, the total cooling effect depends on having a relatively large amount
of the lanthanide salt available.
11 Low and high here are relative to adjacent elements only. So, lanthanum is four or five
times more abundant than either gadolinium or dysprosium.
f orbital involvement in bonding I 267
11.12 f orbital Involvement in bonding

The question of the extent to which f orbitals are involved in the bonding
of complexes of the lanthanides and actinides is one that has long interested
chemists. The final answer has yet to be given. On the one hand there is the
experimental evidence provided by anions such as [UF8] ' - . mentioned in
Section 3.2.6, which have the fluoride ligands arranged at the corners of a
cube, which is not the arrangement sterically favoured. The geometry
becomes understandable iff orbitals are involved in the bonding because in
an Oh molecule the fxyz orbital can participate. This it cannot do in the most
probable alternative geometry, the D4 d antiprism, a point which is explored
in more detail in a footnote towards the end of Section 3.2.6. This apparently
unambiguous evidence is in contrast to the results of theoretical calculations.
The fact that spin-orbit coupling is of key importance for the lanthanides
and, even more so, for the actinides, is a clear indication that it is
important to include relativistic effects in an accurate calculation. And this
is in what is already a difficult problem, because of the large number of
electrons involved. Not surprisingly, a large number of different methods
have been investigated, ranging from relativistic corrections to the extended
Hiickel model of Chapter 10 through to similarly modified ab initio, Xo: and
density functional methods. Although much of the work has, in reality, been
as much concerned with the exploration of methods of tackling such a
difficult problem, some of the results have shown impressive agreement with
such measurables as bond lengths and spectroscopic transitions. Unfor-
tunately, as was pointed out in Chapter 10, it is far from unusual for such
agreement to disappear when a calculation is improved and this could be
the case here. Nonetheless, some reasonably well accepted generalities seem
to have emerged. First, f orbital involvement is greater for the actinides than
for the lanthanides. This conclusion seems clear, despite the greater difficulty
inherent in calculations on the actinides. Secondly, whilst most calculations
on the actinides do indicate at least a limited f orbital involvement, what is
most unclear is whether this involvement has any chemical consequences-
whether it contributes significantly to bond energies, for instance. The
Table 11.12 A comparison of the results of relativistic

(rei) and non-relativistic (non-rei) calculations on the
uranium atom
Orbital Non-rei Rei ReljNon-rel

Energies (eV)
7s 4.54 5.51 1.21
6d 7.25 5.69 0.78
5f 17.26 9.01 0.52
Radii (fi.)
7s 2.67 2.30 0.86
6d 1.52 1.71 1.12
5f 0.67 0.76 1.13
In this example the inclusion of relativistic effects does not change the
sequence of energy levels. In the case of thorium, in contrast, it seems
that the 6d and Sf energies interchange when relativistic effects are
included.
evidence of [UF 8 ] 3 - is that it does-and it is to be remembered that, as

the footnote at the beginning of Section I 0.3 indicated, the consequence of
the inclusion of relativistic effects on f orbitals is that of making them more
available for chemical bonding. This is illustrated in Table 11.12, which
shows the results obtained on some rather good calculations on the uranium
atom. The calculations were carried out twice, with and without allowance
for relativistic effects. The profound consequences of these effects is evident,
the 7s orbitals becoming smaller and electrons in them more difficult to
ionize. On the other hand, the 6d and the 5f become larger and electrons
occupying them much easier to ionize. Even so, the radius of the 5f
electrons remains small, although changes in the effective nuclear charge of
the uranium, perhaps caused by chemical bonding, could change it yet again.
It is not surprising that opinions about the importance of the f orbital
contribution to that bonding remain mixed.
Further reading and W. T. Carnal! in the American Chemical Society Symposium

Series, 1980, Vol 131, 349. The interpretation goes well beyond
A good, if at some points rather dated, survey is Complexes the content of the present chapter but the article contains
of the Rare Earths, S. P. Sinha, Pergamon, Oxford, 1966. An energy level diagrams and other very useful and accessible data.
excellent recent account of the chemistry of the lanthanides and Many of the topics touched on in this chapter are treated
actinides is to be found in Lanthanides and Actinides, S. Cotton, in more detail and depth in some issues of Structure and
Macmillan, London, 1991. An excellent series of brief review Bonding, notably 13, 22, 25,30 and 59/60(actinides). A reference
articles is to be found in Radiochim. Acta (1993) 61. Three that contains more than its title indicates is 'Lanthanide Ion
examples are 'Overview of the Actinide and Lanthanide (the f) Luminescence in Coordination Chemistry and Biochemistry',
Elements' by G. T. Seaborg (p 115), 'Systematics of Lanthanide W. D. Horrocks and M. Albin, Prog. Inorg. Chern. (1984) 31, 1.
Coordination' by E. N. Rizkalla (p 118) and, for those who are Two articles that provide an overview of the current
interested in learning how to study the chemistry of a couple situation on f orbital involvement in bonding in the actinides
of dozen atoms, 'Atom-at-a-Time Chemistry' by D. C. Hoffman are: 'Cyclopentadienyl-Actinide Complexes: Bonding and Elec-
(p 123). tronic Structure' by B. E. Bursten and R. J. Strittmatter, Angew.
The shapes off orbitals is discussed in readable fashion in Chern. Int. Ed. (1991) 30, 1069 (in its last couple of pages this
two adjacent papers: H. G. Friedman, G. R. Choppin and D. G. provides a brief overview of all the theoretical methods being
Feuerbacher, J. Chern. Educ. (1964) 41, 354, and C. Becker, J. employed) and 'The Electronic Structure of Actinide-Containing
Chern. Educ. (1964) 41, 358. See also C. A. L. Becker, J. Chern. Molecules: A Challenge to Applied Quantum Chemistry' by
Educ. (1979) 56, 511, and 0. Kikuchi and K. Suzuki, J. Chern. M. Pepper and B. E. Bursten, Chern. Rev. (1991) 91, 719. A
Educ. (1985) 62, 206. reference given at the end of Chapter 6 is also relevant, although
For spectra, a useful compilation is Spectra and Energy some of the relevant articles are more readable than others-
Levels of Rare Earth Ions in Crystals, G. H. Dieeke, Interscience, The Challenge of d and f Electrons, D. R. Salahub and M. C.
New York, 1968. Particularly valuable is an article 'Optical Zerner (eds.) American Chemical Society Symposium Series,
Properties of Actinide and Lanthanide Ions' by J. P. Hessler 1989, Vol. 394.
Questions 11.3 The lanthanides and actinides form complexes with

planar cyclic C, R, organic species. Use Fig. 11.4 together with
11.1 Select any non-trivial row from Table 11.5 and show Fig. 10.3 to detail how the bonding in such complexes might
that the degeneracy implicit in the simple Russell-Saunders be expected to differ from those of the corresponding transition
term (e.g. 6 H = 6 x II = 66) is equal to the sum of degeneracies, metal complexes.
L; (2j + I), of the corresponding spin-orbit levels. (Hint: the 11.4 Table 11.9 shows that both Ce 111 , f 1 , and Yb 111 , f 13 ,
2 F case has been worked through in Section 11.4.)
are colourless in aqueous solution. However, the spin-orbit
splitting of these two ions are ca. 650 em - I and ca. 3000 em -I,
11.2 The f 1 ion Ce 111 is colourless in aqueous solution, whereas respectively. Explain why these observations are mutually
the d 1 ion Ti 111 is purple. Outline the reason for this difference. compatible.
Other methods of studying
12.1 Introduction
In preceding chapters the theories underlying some of the methods which
have been used to characterize coordination compounds have been discussed
in some detail. The present chapter provides a less detailed survey of other
methods, stretching from indicators of complex formation through to some
which provide insights into physical structure, others into electronic structure,
yet others into the forces between atoms and molecules.
If one is solely concerned with the question of whether or not a complex
is formed in a particular system, relatively crude methods often suffice-
although, historically, such methods have played an important part in the
development of the subject. The evolution of heat, the crystallization of a
product, a change in chemical properties- the failure to undergo a charac-
teristic reaction, for instance- are all simple but useful indicators. More
sensitive methods often involve the study of some physical property of the
system as a function of its composition and some of these were outlined in
Chapter 5. Other examples of methods used for non-transition metal
complexes are the measurement of colligative properties such as vapour
pressure, boiling point or freezing point. Less commonly, measurements of
quantities such as viscosity and electrical conductivity have been employed.
Another electrical measurement which has been much used is that of pH
change. Most ligands may be protonated at their coordination site so that,
for example, corresponding to the NH 3 ligand there is the protonated species,
NH.(, the ammonium ion. This means that, if in the same solution there is
free ligand L, an acceptor species A (which could well be a metal ion), and
acidic protons- which for simplicity will be written as H +, there will be two
important equilibria involving L:
270 1 Other methods of studying coordination compounds
and
A+L~[AL]
If the second equilibrium exists, that is, if a complex is formed, addition of

more A will cause the first equilibrium to be displaced to the left and the
pH of the solution will decrease. Measurement of pH not only allows the
formation of complexes to be detected but may also be used to determine
the stability constants of the species formed (see Chapter 5). As will be seen,
half-cell potentials and polarographic measurements may be used both to
detect complex formation and to determine stability constants but, before
turning to these a variety of techniques will be outlined which not only
indicate the formation of coordination complexes but also provide infor-
mation on their structure.
12.2 Vibrational spectroscopy

For most workers, vibrational spectroscopy means infrared and Raman, 1
although the availability of a third method should not be overlooked. This
is inelastic neutron scattering; in the simplest form, a beam of monochromatic
neutrons is bounced off a solid sample. The energies of the rebounding
neutrons are measured, the differences from the energy of the incident
neutrons give vibrational frequencies. The measurements are of relatively low
resolution and require national or international facilities but have the
advantage that they may reveal bands which are both infrared- and
Raman-forbidden; they are particularly good at revealing modes involving
the motion of hydrogen atoms. However, the improvements provided by
modern Fourier transform infrared spectrometers have made weak bands
much more accessible; the extension of these methods to Raman spectroscopy
has made this technique both easier to use and applicable to more samples
(laser-induced fluorescence, which can readily swamp the weak Raman
scattering, has been largely overcome by the use of infrared lasers, which
also enable an infrared spectrometer to be used to measure Raman spectra).
The infrared and Raman frequencies of isolated molecules are determined
by the arrangement of the atoms in space and by the forces between the
atoms. When a ligand coordinates, all of these change. So, changes are to
be expected both in the vibrational spectral features associated with the free
ligand and in those of the system to which it becomes attached. An example
is provided by the thiocyanate anion, SCN-. In simple ionic thiocyanates,
such as KCNS, there is a v(C-N) stretching vibration at ca. 2060 ern- 1 , a
v(C-S) stretch at ca. 746 em - 1 and a b(NCS) bend at ca. 480 em - l In the
anion [Co(NCS) 4 ] 2 -, where the SCN- is N-bonded, the v(C-N) mode
rises to ca. 2070 em- 1 , the v(C-S) increases dramatically to ca. 815 em- 1
1 Infrared-allowed transitions are dipole-allowed. That is, the (electric) dipole of the incident
radiation couples in resonance with a vibration which changes the (electric) dipole of the
molecule. So, an infrared-allowed vibration behaves like a dipole and that means like a
coordinate axis. In contrast, Raman involves two light waves, one in, the other out. So, Raman
selection rules reflect this difference and a vibration which behaves like a product of t1vn dipoles,
like a product of coordinate axes, are Raman-allowed. Clearly, infrared and Raman have
different selection rules. In the statements above, the word behaves means 'has the same
symmetry properties as'.
Vibrational spectroscopy 1 271
but the bend only drops by ca. Scm- 1 In contrast, in the [Hg(SeN) 4 ] 2 -
anion, where the seN- anion is S-bonded, the v(e-N) mode rises to ca.
2100 em -•, the v(e-S) drops to ca. 710 em -• whilst the bend drops to ca.
450 em- 1 • It is obvious that these different patterns can be used to determine
the way that a seN- ligand is coordinated. Even more obvious would be
to look for the appearance of v(M-N) or v(M-S) features, but this is not
always easy if there are other ligands present with metal-ligand stretching
modes in the same spectral regions. This explains why there is a particular
interest in ligand-specific modes falling in otherwise relatively clear spectral
regions.
In favourable cases vibrational spectra can be used to determine the ligand
geometry around a central metal atom.2 Best known is the fact that a centre
of symmetry means that bands active in the infrared are not Raman active
and vice versa, although this is just one of many symmetry-spectra
relationships. Such relationships are not infallible. So, a band may exist but
be too weak to be observed separately from other bands ~n the same spectral
region. Alternatively, vibrational couplings may be small. So, if two vibrators
are uncoupled it is irrelevant that they happen to be related by a centre of
symmetry-a single band will appear, coincident in infrared and Raman.
Because of this element of uncertainty, vibrational methods of structure
determination are sometimes referred to as sporting methods-they are not
infallible-although for compounds for which crystals, and thus X-ray
crystallography, cannot readily be made available, they may be the only
methods which it is feasible to use.
As a typical example of the application of vibrational methods, consider
square planar complexes of Pt 11 of general formula PtL 2 X2 , where X is a
halogen and L is a polyatomic ligand. A study of the spectral activity of the
Pt-X stretching modes usually shows whether a particular complex is cis or
trans. If it is cis then the molecular geometry is approximately C 2 v and the
v(Pt-X) vibrations are of A 1 + B 1 symmetries so that both modes are both
infrared and Raman active. It follows that two v(Pt-X) bands are expected
in each spectrum, the bands in one spectrum coincident with those in the
other, to within experimental error. The trans isomer has approximate C 2 •
symmetry and the v(Pt-X) vibrations have A• + A. symmetries. Of these,
the former is Raman active and the latter is infrared. One band is predicted
in each spectrum, at a different frequency in each. The difference in the
spectral predictions between the two isomers is so great that it should be
possible to use the method even if the spectra are incomplete or less than
ideal. An application of these results is illustrated in Fig. 12.1, which shows
the Raman spectra of cis and trans [PtC1 2 (NH 3 h] in the v(Pt-Cl) and
v(Pt-N) stretching regions. The presence of two peaks in the v(Pt-N) region
is sufficient to establish the lower spectrum as that of the cis isomer, without
aid of the infrared spectrum, even though the v(Pt-Cl) only shows one (the
expert might well argue that the greater breadth of the v(Pt-Cl) peak in the
2 This has been exploited in transition metal carbonyl chemistry. Simple metal carbonyl
derivatives often have quite high symmetries and, equally important, a variety of possible
geometric arrangements. The vibrational coupling of v(C-0) groups is so great-and the
resulting bands so well separated-that simple group theoretical methods may often be used
to distinguish between them. For instance, an M(C0) 3 unit could have D31, C3 l, or Cs
symmetries. These could be distinguished because they give rise to one, two and three infrared
bands in the v(C-0) region, respectively.
FIJI. 12.1 The Raman spectra of (a)

trans-[PtCI 2 (NH3 ) 2 ] and (b) cis-[PtCI 2 (NH3 ),].
The v(Pl-N) mode is at ca. 500 cm- 1 and the
v(P\-CI) at ca. 300cm- 1 (adapted and
(a)
reproduced with permission from M. J. Almond,
C. A. Yates, R. H. Orrin and D. A. Rice.
Spectrochim. Acta A (1990) 46A, 177).
(b)
600 500 400 300
cis isomer is indicative of a band splitting which is not resolved in the

spectrum). Another origin of band splitting which may conveniently be
mentioned at this point is that which occurs in a ligand mode when there
is a decrease in ligand symmetry on coordination. For instance, the infrared
active v(S-D) stretches of the isolated (tetrahedral, T4 ) sulfate anion appear
as a single band (the mode is ofT2 symmetry), which splits, loses degeneracy,
on coordination.
Two other examples of the application of the general approach are shown
in Figs. 12.2 and 12.3. Figure 12.2 provides a further illustration of the way
that vibrational spectra can indicate a mode of coordination. So, in
complexes of the N02 anion, a v(N-D) mode appears at ca. 1300 em - I
when the ligand is N-bonded and at ca. 1150 em- 1 when it is 0-bonded.
This is a much-studied and much-exploited pattern which is illustrated for
two cobalt(III) complexes in Fig. 12.2. This particular example was chosen
because an experiment in many undergraduate inorganic chemistry courses
is a study ofthe way that the two species in Fig. 12.2 isomerize, using infrared
spectroscopy to follow the reaction.3
Figure 12.3 shows two Raman spectra of the anion [Fe(CN) 6 ] 4 -.
Although a relationship between them exists, they are surprisingly different.
The first thing to note is that some of the bands appear above 2080 em- 1,
the position of the v(C-N) band in KCN; bands do not invariably drop in
frequency when a ligand coordinates, although this is the common pattern.
The differences between the two spectra in Fig. 12.3 show that vibrational
interactions can occur between different complex molecules in the solid state.
In the present example, between different [Fe(CN) 6] 4 - anions. It is an
unfortunate fact of life that such splittings can easily be confused with
splittings resulting from molecular geometry. The general way that solid-state
(b) splittings originate is shown in Fig. 12.4. This figure shows four single
Fig. 12.2 The infrared spectra of
3 Visible spectroscopy and NMR can also be used; the reaction in solution shows base
(a) [Co(NH 3 ),-<lN0,] 2 + and
(b) [Co(NH3 ),-N0,] 2 +. The v(N-<l) peak in catalysis and provides a useful illustration of this phenomenon. A word of caution, however,
the latter at ca. 1300 cm- 1 overlaps with a the back reaction is photocatalysed so that working in bright sunlight can produce unexpected
peak common to both spectra. results!
Vibrational spectroscopy I 273
Fig. 12.3 The Raman spectra of crystalline

samples of (a) fi.[Fe(CN)eJ and
(b) Na4 [Fe(CN)e]. This figure illustrates several
of the points made in the text. In the [Fe(CN) 6 ] (a)
anion the v(O.N) vibrations are coupled.
The higher frequency bands in both spectra are
the Al<r (totally symmetric, 'breathing' mode)
and the lower Eg. the components of which are
split apart. This latter splitting-and all the
frequency differences between the two
spectra-show the importance of solid-state
effects, in which they originate. (Adapted and
reproduced with permission from T. N. Day,
P. J. Hendra, A. J. Rest and A. J. Rowlands,
Spectrochim. Acta A (1991) 47A, 1251).
2100 2050
Fig. 12.4 A unit cell of a crystal in which the

unit cell contains four molecules. Corresponding ~--------------7( ~--------------/(
vibrations of each molecule (top) couple
together to give four unit cell vibrations (lower). '' '' '' ''
The coupling between the modes of the \ ! :' :'
individual vibrations is represented by the dotted '' '' '' ''
lines drawn between them. Of the four unit cell
vibrators the upper two are Raman active and ~--------------~ ~--------------~
the lower two infrared active. The diagrams
show two·dimensional unit cells but the
principle carries over unchanged into
three-dimensional unit cells.
'K--------------/( 'K--------------7(
'' '' '' ''
'' '' '' ''
''' ''
' '' ''
''
''
~--------------~ /--------------~
vibrators in a unit cell coupling together to give four coupled vibrations; if

the coupling is large enough then the four coupled vibrations have different
frequencies and, selection rules willing, appear as separate bands. Fortunately,
the vibrational analysis of a crystalline solid is no more difficult than that
of an isolated molecule-one treats the contents of the unit cell as a giant
molecule-although the number of cases in which such spectra have been
used to make predictions about crystal structure is rather small! The
theory of these solid state effects is covered more fully in Chapter 17
(Section 17.6).
It is interesting that it is possible to use vibrational spectroscopy as a
probe of more than molecular symmetry; in favourable cases the detailed
molecular structure can be explored. Most of this work has been done on
the infrared spectra of transition metal carbonyls-their v(C--0) stretching
modes occur in a spectral region (ca. 2000 em- 1 ) almost free from other
features. Further, the individual CO vibrators couple together strongly giving
well-separated bands which are associated with intense infrared absorptions.
Finally, the dipole moment change associated with each CO stretch seems
essentially colinear with the bond axis. Suppose we have an M(COh unit,
each CO group being linear, as is usually the case to quite a good
approximation, and at an angle() to each other, as shown in Fig. 12.5. The
in-phase vibration of the two CO groups will have a resultant dipole moment
that bisects (); its magnitude will be 2d cos((J/2), where d is the bond dipole
moment change. The dipole associated with the out-of-phase combination
will be perpendicular to that of the in-phase and will have a resultant dipole
moment change of 2d sin((J/2). Infrared band intensities are proportional to
the square of the corresponding dipole moment changes (dipole moment
changes can be either positive or negative but band intensities can only be
positive). If the intensity of the band associated with the in-phase vibration
(a)
Fig. 12.6 (a) The in-phase vibrations of two
CO groups in an M(C0) 2 unit subtend a
resultant vector (shown dotted) of 2d cos(0/2).
(b) The out-of-phase vibrations of two CO
groups in an M(C0) 2 unit subtend a resultant
vector (shown dotted) of 2d sin(0/2). In both
cases the vibrations are shown and, to the light,
the corresponding vector sums.
. e
(b)
dSin 2
Resonance Raman spectroscopy 1 275
is I, and that associated with the out-of-phase is Jo then it follows that

2
( 2d sin ~)
-;---2~ t 28
( 2d cos ~y ~ an 2
So, measurement of relative band intensities leads to a C-M-C bond angle,

e. This procedure can be generalized to other metal carbonyl systems;
for simple systems, rather good agreement with crystallographically deter-
mined angles is obtained. Unfortunately, few ligands other than CO prove
amenable to such an analysis.
12.3 Resonance Raman spectroscopy

In normal Raman spectroscopy a sample is placed in a (monochromatic)
laser beam and the very weak scattered light of lower frequency is studied.
In such a study the colour of the laser light is usually chosen to be away
from any absorption band of the sample because such a choice reduces the
risk that the focused laser beam will destroy the sample by heating it. In the
resonance Raman effect the laser beam colour is deliberately chosen to
coincide with an absorption band-an electronic transition-of the sample.
Whilst this may lead to the destruction of the sample, for favourable cases
it leads to Raman scattering which is much stronger than normal. This, in
turn, means that the laser power can be reduced, improving the chances of
sample survival. The spectra obtained from compounds showing such a
resonance Raman effect are both simpler and more complicated than normal
Raman spectra. They are simpler because, often, only totally symmetric
vibrational modes are seen. The reason for this is that if the electronic
X 10
6v 1 2v1
?v1
Fig. 12.6 Resonance Raman spectra. (a)

Crystalline Til 4 ; v• 1 is the v(Ti-1) totally symmetric 3v1
breathing mode, reproduced with permission
from R. J. H. Clark and P. D. Mitchell,
J. Amer. Chern. Soc. (1973) 95, 8300.
4v 1
2000 1600 1200 800 400 0

(a)
1\)
01
~
!!l
i
i
~
Jv3+v1 2v3+V1 v3+V1 v1 lOv3 ev, 8v3 5v3 + v2 7v3 4v3 + v2, 6v3 V3+V2,JV3
. .T2vTr·· ••
·r Ii
I
I
~
g
I
~ cm-1
(b)
Fig. 12.6 (continued) (b) Crystalline [NBu 4 ][0s(C0) 2 Br.:J. The anion has D4h symmetry, the totally symmetric modes are: v1 , v(C=O); v2 , v(Os-C); ''••
v(Os-Br); reproduced with permission from F. H. Johannsen and W. Preetz, Z. Naturforsch (1977) 32b, 625. In both cases, although other modes can be
seen, the totally symmetric modes dominate the spectra.
Spectroscopic methods unique to optically active molecules 1 277
excitation involved is one in which the number of bonding electron changes

then the excited molecule will find itself, be produced, in a geometry well
away from its equilibrium. Typically, if the electronic transition which is the
source of the absorption band is located on a metal atom, then a breathing
mode of the surrounding ligands is the only vibrational mode excited-
totally symmetric modes are the only ones that produce volume changes of
the type needed to reach the equilibrium geometry. The spectra are more
complicated because it is not just the fundamental frequency of the breathing
mode which is excited but also a host of overtones-frequently up to the
lOth but cases are known in which overtones approaching the 20th .are
observed. Two examples of resonance Raman spectra are shown in Fig. 12.6.
In the literature there are examples of spectra labelled as 'resonance Raman
spectra' which simply show generally enhanced intensities rather than the
specific characteristics highlighted in Fig. 12.6, perhaps because the electronic
excitation involved is one in which no great change in equilibrium geometry
occurs. It may be that, de facto, a more relaxed definition of the name
'resonance Raman' is emerging but in this section the restricted definition
is used.
Resonance Raman spectra are important for two main reasons. First, they
enable the ready determination of potential energy curves (one associated
with each of the modes excited) to unusually high levels. Secondly, they can
be used to target small parts of a large molecule. So, if in a bioinorganic
molecule containing an Fe111 ion the laser can be tuned to an absorption
band associated with this iron atom (perhaps a d --+ d band, Chapter 8),
then just vibrations originating in the atoms coordinated to the Fe111 are to
be expected. This means that it is possible to probe these atoms-just what
is their valence state and in what geometry they are coordinated? Such data
can be most valuable when dealing with a complicated molecule. Indeed, it
is sometimes possible to probe more than one site in a biomolecule by
judicious choice of exciting laser frequencies and obtaining a different
spectrum for each frequency. Another measurement possible by resonance
Raman but not by conventional Raman is the measurement of the vibrational
spectra of electronically excited states. Typically, a high-power pulse laser
of a frequency corresponding to an electronic transition of the target
molecule is used to excite a large proportion of the molecules into the excited
state. This pulse is immediately followed by another which excites the
resonance Raman spectrum. So, the anion [Re 2 Cl 8 ] 2 - has a o bond
between the two metal atoms (this will be dealt with more fully in Chapter
15). Excitation to an excited state in which an electron is promoted to the
o (antibonding) orbital leads to the Re--Re stretching vibration dropping
from 274 to 262 em - l and the Re--Cl increasing from 359 to 365 em- 1. The
o
changes are small, indicating that the bonding is not strong, but they are
real and in accord with what would be expected.
12.4 Spectroscopic methods unique to optically

active molecules
If a weight at the end of a piece of string is set in motion so that it traverses
a circular path, this motion may be regarded as composed of two mutually
Fig. 12.7 A simple harmonic motion (centre)

may be regarded as a sum of two circular
motions. The motion represented takes place in
the plane of the paper. The thickness of the
lines represents time, rather like an oscilloscope
trace on a tube with a persistent phosphor.
perpendicular simple harmonic motions combined out of phase with each

other by 90°. In a similar way, a single back-and-forth harmonic vibration
of the weight may be regarded as an in-phase combination of two circular
motions in opposite directions. An attempt to show these alternatives in
pictorial fashion is given in Fig. 12.7. Thus, there is always a choice. A circular
motion may be described as a combination of two simple harmonic motions
and, similarly, a simple harmonic motion may be expressed in terms of two
circular motions. This choice is analogous to those met elsewhere: between
real and complex atomic and molecular orbitals- we preferred to call them
static and rotating in Chapter II ; between linearly and circularly polarized
light (and this will be our concern in this section) and a similar choice has
to be made in areas remote from chemistry- in some aspects of electrical
circuit theory, for instance. Which of the alternative descriptions is adopted
is usually determined by that which leads to the simpler mathematics, even
if it involves some initial conceptual difficulties. So, whilst it is true that
optically active compounds rotate the plane of linearly polarized light, it
turns out that it is more convenient to discuss the phenomenon in detail in
terms of circularly polarized light. Figure 12.8 shows how linearly polarized
light may be regarded as made up of two circularly polarized components.
Consider what happens when each of the circularly polarized components
is propagated along the threefold axis of the optically active ion [Co(en),] 3+ .
From Fig. 12.9 it is evident that one of the circularly polarized components
tends to pass along the backbone of the ethylenediamine ligands whilst
the other tends to cut the backbone perpendicularly. This description is
inaccurate in that the wavelength of visible light (and it is in this that
we shall be interested) is of the order of a thousand times greater than
the molecular dimensions. Nevertheless, the vital point remains- the two
Fll(. 12.8 Linearly polarized light (centre)

regarded as a sum of two circularly polarized
components. Figure 12.7 may be regarded as a
view of Fig. 12.8 along the axis of propagation,
the thickness of the lines then indicating
distance.
Spectroscopic methods unique to optically active molecules 1 279
circularly polarized components will encounter slightly different electron

density distributions in passing through the [Co(enlJ] J+ ion. As is well
known, the interaction between a light wave and electron density leads to a
reduction in the velocity of propagation of the light wave- refractive indices
are greater than unity- and the greater the electron density (roughly), the
higher the refractive index and the slower the velocity of propagation. This
means that one circularly polarized component must be expected to pass
more slowly through the [Co(enlJ] J+ ion than the other; the faster might
be that which tends to cut the ligand backbone. Although this last statement
is, in fact, not generally true, it correctly suggests that polarized light may
be used to determine the chirality of an optically active species. Because one
circularly polarized component travels faster than the other, one of the helical
'springs ' shown in Fig. 12.10 will be more extended than the other. This
means that after traversing the molecule the two components will be slightly
out-of-phase. Combining them, as in Fig. 12.10, leads to a rotation in the
FIC. 12.9 Two circularly polarized beams of plane of polarization compared with the incident light, as observed. If a
light passing through a molecule of
[Co(en) 3 ] ' +, the threefold axis of the latter graph is plotted of the angle of rotation against wavelength, the angle of
being coincident with the axis of propagation of rotation being suitably corrected for path length and concentration of the
the light. One circularly polarized beam (the top optically active species, a curve of optical rotatory dispersion, an ORD
one) tends to cut the backbone of an
ethylenediamine molecule, the other to pass spectrum, is obtained. At wavelengths at which the species is transparent,
along it. where there is no electronic transition, the ORD spectrum is unexciting,
varying little with wavelength. It is only when absorption occurs that things
begin to happen!
It is well known that the absorption of light by a molecule is frequently
anisotropic- and this anisotropic absorption occurs simply because molecules
are frequently anisotropic. Thus, a light wave incident on a molecule in one
direction may excite a transition but, light of the same wavelength, incident
r
~
r
~
/ ~ '
~ ~ ~
>c::::> /
FJc. 12.10 The passage of a beam of linearly
polarized light through an optically active
molecule. Note the rotation of the plane of
x::=::>
c--===='
x::=::>
>
......______
~
~
c::=><
c ><
c::=><
'-.....,
~ ~
polarization and correlate this with the different
behaviour of the two circularly polarized
components INithin the molecule.
(
><==:: eX Optically
active
molecule
~ ex
c:
0
%0
<I)
.c
<(
(a)
FIC. 12.11 Idealized diagrams of (a) an

electronic absorption band of an optically active +
species and (b, c) the corresponding ORD (solid
line) and CD (broken line) plots for the two
different hands of the species.
Positive Cotton effect Negative Cotton effect

(b) (c)
in another direction, may not be absorbed. In a region of absorption of light

by an optically active species, plane-polarized light must be expected to be
attenuated, but what of its circularly polarized components; are they equally
attenuated? Since these follow slightly different paths through a molecule,
as seen earlier for [Co(enhJ3+, it would seem reasonable that they should
be absorbed to different extents. This prediction is confirmed by measure-
ments made with circularly polarized light-it is possible to pass right
(clockwise, viewed in the direction of propagation) and left (anticlockwise)
circularly polarized light alternately through a solution and to compare their
relative absorptions. The difference in extinction coefficients between left and
right circularly polarized light, e1 - e, is small but is measurable. It is called
the circular dichroism, CD. Together, ORD and C D are collectively called
the Cotton effect, after the Frenchman who did the fundamental work. A
CD spectrum shows a maximum at the position of maximum absorption of
an ordinary absorption curve (so that if a CD spectrum shows several
maxima it indicates that the ordinary absorption curve consists of several
overlapping bands). The ORD spectrum behaves as the derivative of the CD
spectrum, passing through a point of inflection at the absorption maximum.
These relationships are illustrated in Fig. 12.11; as this figure shows, the
curves for one species are the negative of those of its enantiomorph. A
complication arises, however. Although the [Co(enh]3+ cation is shown
behaving like a left-handed screw in Fig. 12.9, the same molecule, when
viewed in a perpendicular direction, behaves as a right-handed screw (Fig.
12.12)! This duality is not unique to this example but occurs for all optically
active molecules.
FIC. 12.12 The [Co(en)3 ] 3 + ion viewed When a molecule absorbs light and undergoes an electronic excitation
perpendicular to the threefold axis (cf. Rg. 12.9). this occurs through a displacement of electron density within the molecule.
Nuclear spectroscopies 1 281
It turns out that the absolute configuration of a molecule can be related to

the Cotton effect observations only if the direction of charge displacement
is known. For an optically active molecule this displacement occurs along
a helical path 4 and, as has just been seen, within a given molecule there are
always both right- and left-handed helical paths available. This is made yet
more complicated when we recognize that if the displacement is perpendicular
to, say, a left-handed helical backbone, then the displacement itself has the
character of a right-handed helix (Fig. 12.13)1 Yet only if we know something
about the nature of the electronic transition can we hope to deduce a
molecular configuration from Cotton effect measurements. Evidently, it is
advisable to study a species with as few electronic transitions as possible, if
one has choice, for this reduces the probability of error. It is for this reason
that much more detailed work has been done with complexes like [Co(en)J] H
than with others such as [Co(enhCI 2] +.The former has the higher symmetry
and, therefore, fewer absorption bands. If the same d -+ d transition can be
observed in a series of closely related compounds then, remembering that in
the crystal field model, at least, the ligands generate an electrostatic crystal
field but are not otherwise involved, if the absolute configuration of one of
the members can be established, so too can that of each of the others. In
practice, it is usually possible to identify transitions by band intensity criteria
and, fortunately, the absolute configuration of ( + )-[Co(en)J] H and of
several other key species have been determined by X-ray methods. An
alternative would be to be able to have a detailed and reasonably accurate
quantum mechanical treatment of the phenomenon, for these would also
provide the structure- spectra relationship. Such, ab initio, calculations are
beginning to appear but have yet to reach the point at which they fulfil the
above objective.
Finally, it is to be emphasized that the sign of an ORD curve at any
arbitrary point-like the sodium D-line wavelength- is composed of a
Fig. 12.13 When the grey and black helices superposition of tail absorptions from all the ORD bands in the spectrum.
intersect they are at right angles to each other. The sign of the specific rotation at such a point is therefore an uncertain
The two helices, one corresponding to a helical
molecule and the other to a transition occurring guide to molecular structure, although, historically, many have sought to
perpendicular to the molecular helix, are of use them to find reliable specific rotation- structure relationships.
opposite hand.
12.5 Nuclear spectroscopies

In this section three methods will be outlined which have been used to study
the properties of nuclei in coordination compounds. None of the techniques
is of universal application and they all suffer from the disadvantage that the
connection between the spectra obtained and the molecular bonding is
seldom simple so that in this application they are generally best used to
compare two compounds rather than discuss either in isolation. However,
the spectra can be outstandingly useful in obtaining details of molecular
geometry and reactions.
4 The requirement for optical activity is that electronic charge displacement occurs along a
helical path, for then optical isomers will be associated wit h bands of opposite helical charge
displacement. Group theore ti cally, this means that transla tion along and rotation about an axis
transform isomorphically - as the same irreducible representation (and, if the irreducible
representation is degenerate, as the same component).
12.5.1 Nuclear magnetic resonance (NMR)

The use of NMR in kinetic studies will be covered in Section 14.5 and in
particular the way that it is used to study intramolecular gymnastics
(fluxionality). At this point, it has to be emphasized that spectra obtained
at room temperature may be misleadingly simple because of dynamic
exchange processes in which two (or more) nuclei interconvert their chemical
environments so rapidly that they give an averaged spectrum.
NMR has developed to the point at which it is the second most important
technique in chemistry for determining molecular structure (the most
important being X-ray diffraction). As a technique, it is not applicable to all
elements, although it is increasingly rare to find a compound in which at
least one element cannot be studied. When there are two or more such
elements, the range of experiments available, and so the information
available, increases considerably. In Table 12.1 are listed the elements which,
with a suitable spectrometer-or array of spectrometers-can be studied.
Just as infrared spectroscopy has been revolutionized by Fourier Transform
techniques, so too has NMR. In practice, a pulse of high-frequency (in the
MHz range) radio waves is applied to the sample, which is contained in a
magnetic field. The way that the nuclei respond after the pulse is completed
is monitored. In more subtle experiments, other pulses are applied after the
first. The time interval between the pulses, the length of the pulses, the
frequency content of the pulses, the power level of the pulses; all can be
varied, giving rise to a plethora of different measurements. In addition, it is
possible to irradiate at more than one frequency. So, for instance, by using
high power at the correct radio frequency it is possible to excite one set of
nuclei so that the relevant ground and excited state populations become
equal. Because the observation of an NMR spectrum depends on the
Table 12.1 Some of the nuclei which have been studied by NMR
spectroscopy; spin = ~ unless otherwise stated
Nucleus (spin) Natural abundance ("/o) Relative sensitivity"

1H 99.9 1.0
2H(1) 0.02 1.5 X 10- 6
118(~) 80.4 0.13
13c 1.1 1.8 X 10- 4
14N(1) 99.6 1.0 X 10- 3
15N 0.4 3.9 X 10- 6
120(~) 0.04 1.1 X 10- 5
19F 100.0 0.83
2sSi 4.7 3.7 X 10- 4
31p 100.0 0.066
335(~) 0.76 1.7 X 10- 5
35 CI(~) 75.5 3.6 X 10- 3
37 CI(~) 24.5 6.6 X 10- 4
a18r(~) 49.5 4.9 X 10- 2
llssn 8.6 4.4 X 10- 3
1271(~) 100.0 9.3 X 10- 2
195Pt 33.8 3.4 X 10- 3
a Relative sensitivity includes a number of factors of which percentage abundance is one.
difference between such populations, high powers lead to spectral changes,

usually simplifications.
Although measurements such as the variety indicated above are available,
they require a considerable amount of time and expertise~ in order to arrive
at the optimum experimental pattern for a given measurement a fair amount
of trial and error may be involved. Not surprisingly, the vast majority of
NMR measurements remain of the relatively simple variety. Fortunately,
too, much of the sophisticated work is aimed at taking complicated spectra
and making them simple. In this section, therefore, the simpler aspects of the
subject will be briefly reviewed. In the more general case, where such
simplifications cannot be achieved by use of experimental ingenuity, use of
computer programs to match observed and calculated spectra is normal. It
is usually possible to be confident that such a match is unique and so the
relevant molecular quantities~number of nuclei of a particular type,
chemical shifts and coupling constants~are uniquely determined. Finally,
mention should be made of the fact that it is becoming .jncreasingly possible
to study solids by NMR. The method depends on spinning the sample
inclined at the correct angle to the applied magnetic field. In favourable
cases, narrow lines are obtained~normally solids give broad lines~but in
some cases it is not physically possible to spin fast enough to achieve
liquid-like spectra. The method works well for amorphous solids and so can
generally be expected to address questions such as whether the structure in
the solid is the same as in solution. The following paragraphs provide a brief
indication of the ways in which NMR finds application in coordination
chemistry by a study of selected examples.
Complexes of WF6 , [WF6 L] where L is a ligand, exist and in which the
tungsten atom might be seven coordinate. A low-resolution 19 F spectrum
shows three lines of relative intensity 1:1:4, which does not appear to be
consistent with a seven-coordinate structure. The most likely structure is
[WF5 L]+ F-, in which the tungsten atom is octahedrally coordinated by
six ligands, five of which are fluorines. The four coplanar fluorines give rise
to the largest peak and the axial fluorine to one of the others. The fluoride
anion gives the final peak. The fine structure of the peaks confirms this
assignment as does the conductivity of the compounds in liquid sulfur
dioxide (Section 12.7) and, finally, X-ray crystallography. This example, as
presented, depends on an interpretation of the number of observed peaks.
In practice, empirical relationships between structure, chemical shifts and
coupling constants would also be used in such structure determinations.
It is not usually possible to observe the NMR spectra of paramagnetic
species because the lines are extremely broad. In particular cases, however,
the spectra can be observed but cover a frequency range perhaps 50 times
as great as that encountered with similar diamagnetic compounds. As was
seen in Chapter 6, unpaired electrons which, from crystal field theory, would
be expected to be localized on a transition metal, are in fact delocalized onto
the ligands (main group elements rarely form paramagnetic coordination
compounds). If, in particular, the ligands are conjugated systems, the
unpaired electron density will also be delocalized over the entire ligand. The
additional shifts observed in the NMR spectra of paramagnetic compounds
are directly proportional to the unpaired electron densities on the nuclei
giving rise to the various peaks. Unpaired electron densities may be either
Et
•omn-@-o-@:> '
I
0.000~ ~020 1 t ~t
Fig. 12.14 Nickel(ll) aminotroponeimineato
complex.
_ +
+ 0.040
-0.022 2
positive or negative depending on whether the unpaired electron density is

of a or fJ spin (spin densities usually alternate in sign along a conjugated
system in a manner similar to the superexchange mechanism discussed at
the end of Chapter 9; at the end of Chapter 6 an alternation of spin densities
was also encountered). So, in a nickel(II) aminotroponeimineato complex
the unpaired spin densities shown in Fig. 12.14 have been calculated from
the observed spectra. In this compound the shifts of the proton resonances
were studied. The unpaired spin densities given in Fig. 12.14, however, are
those at the corresponding carbon atoms. This is because there is a direct
proportionality between these two quantities. In principle, the spin density
at the carbons could be measured directly, by use of 13 C NMR. However,
this does not seem to have been done.
The opening-out of spectra due to paramagnetism is exploited by the
use of so-called shift reagents. These are paramagnetic compounds, soluble
in organic solvents, which have a through-space effect on the species in
solution. By careful choice of shift reagent, chemical shift differences in
the solute molecules are enormously increased, thus simplifying spectra,
without any undue broadening of the peaks. The detailed mechanism of the
effect is not fully known but it is likely that a small proportion of the solute
molecules form a transient, weak, complex with the shift reagent, the effect
perhaps affecting all solute molecules by a process akin to those described
in Section 9.11. For 1 H spectra, a lanthanide complex of the ligand
Me 3 CCOCHCOCMe 3 with an ion such as Eum or Prm is frequently used.
Earlier in this section the use of empirical correlations involving chemical
shifts were mentioned as an aid to structure determination. There has long
been known a particularly good illustration of the origin of one contribution
to such correlations in coordination compounds. This example concerns
59 Co resonances, for which it has been predicted that the chemical shift
shown by octahedral cobalt(III) complexes should be inversely proportional

to the energy separation between the 1 A '• ground state and the lowest
excited 'T,. term, a separation which may, of course, be measured from the
d--> d spectra. In Fig. 12.15 is given a plot of chemical shift against the
wavelength of this transition (the wavelength is proportional to the inverse
of the energy separation) for some octahedral cobalt(III) complexes. The
amount of the (small) mixing of the 1 T 1 • term into the 1 A 1 • term by the
- Wavelength
magnetic field is inversely dependent on their energy separation, thus
Fig. 12.15 The 59Co chemical shifts shown by explaining the observation. This same mixing was met at the end of Section
cobalt(lll) complexes plotted against the
wavelength of the corresponding 1 T1g ~ 1A1g 9.3, where it was seen that it gives rise to temperature independent
electronic transition. paramagnetism (TIP).
12.5.2 Nuclear quadrupole resonance (NQR)

·~ '"~""'"" Atoms which have nuclear spins greater than -! behave as if the distribution
of charge within the nucleus is non-spherical. The nucleus does not behave
~a+
- - - - - - - - - a-
as a dipole because the nuclear charge distribution remains centrosymmetric.
It does, however, possess an electrical quadrupole moment (C02 , which is
(a) centrosymmetric and linear, is an example of a molecule which has an electric
quadrupole moment). In an applied non-uniform electrostatic field, the
non-uniformly charged nucleus can take up at least two orientations, one of
+
which is more stable than the others (the number of orientations depends
~ Figradient on the magnitude of the nuclear quadrupole moment). A diagram which gives
some idea of the origin of this splitting is shown in Fig. 12.16. It is possible
~a+
- - - - - - - - - a-
to excite the nucleus from a lower to an upper state by application of suitable
radio-frequency radiation. This is a classical description of the phenomenon,
(b) but the essentials are carried over into a quantum mechanical treatment. In
Fig. 12.11 A nucleus (represented by an
practice, the non-uniform electrostatic field is generated by the charge
ellipse) with a quadrupole moment (represented distribution around, but very close to, the nucleus. A more detailed
by the charges within the nucleus) in (a) a more discussion shows that any variation in field gradient is almost entirely due
stable and (b) a less stable orientation within
an applied electrical field gradient. to unequal occupancy of the p orbitals of the atom-so that if there is to
be a field gradient, the p orbitals cannot be symmetry-related. Clearly, this
is a phenomenon which is only applicable to atoms in a low-symmetry
environment (but, note carefully, this does not automatically mean a
low-symmetry complex). The most-studied of the nuclei which exhibit
quadrupole resonance spectra are 35 Cl and 37 Cl. The method is inherently
insensitive and the high concentrations of these isotopes in samples such as
solid K 2 [PtC1 6 ] is a great advantage. Only solids can be studied, anyhow,
because the molecular tumbling in a liquid or gas averages the effect to zero.
The sensitivity of the method has increased in recent years by the advent of
pulse techniques (such as those used in NMR) and also, in suitable cases,
by not looking at the NQR nucleus itself but, rather, at one which is
energetically coupled to it and which is NMR-active.
Other, currently more specialist but of potential wide applicability,
methods include the optical detection of quadrupole resonances-a sample
is laser-excited to an electronically excited state, the return to the ground
state is by phosphorescence; the intensity of the phosphorescence is sensitive
to whether or not concurrent microwave radiation matches an energy
separation in some quadrupole-split intermediate state. Yet another method
depends on correlations between successive f3 or y emissions from excited
quadrupolar nuclei (where the excitation can be achieved by suitable nuclear
bombardment). These do not exhaust the list of current developments-they
have been chosen to illustrate the wide front on which new techniques are
emerging. It is likely that because of these developments the future will see
a wider use of NQR spectroscopy. It is also likely that the interpretation of
the data will become more sophisticated. Traditionally, the experimental data
have been interpreted to give the percentage ionic character of a bond. This
is because, for example, in the Cl- ion all of the p orbitals are equally
occupied whilst in Cl 2 the u bond, if composed of p orbitals only, corresponds
to one electron in the p6 orbital of each chlorine atom, and so Cl- and Cl 2
differ in their resonant frequencies. Interpolation allows a value for the ionic
character of a Cl-M bond to be determined from the chlorine resonance
frequencies in Cl- and Cl 2. Some correction may be applied to allow for the
Table 12.2 Ionic character of the M-el
bond in [MCI 6 ] 2 -. Although these ions
fact that a pure chlorine p orbital may not be involved in the M-el bond.
have Oh symmetry, the symmetry at In this way, Table 12.2 was compiled. When there are two non-equivalent
the chloride is, at most, C4 v (adjacent NQR nuclei in the unit cell of a solid these give rise to separate resonances
cations may reduce this symmetry) which may be resolvable. In this way NQR spectroscopy gives structural
information. Both bromine, 79 Br and 81 Br, and iodine, 127 I, but not fluorine,
Species Ionic character of the
give NQR spectra, as too may 14 N, 55 Mn, 59 Co, 63 Cu, 65 Cu, 75 As, 121 Sb,
M-CI bond (%)
123Sb, 201 Hg, and 2o9Bi.
[PtCI 6 ]"- 44
[PdC16]2- 43
[lrCI 6 ] 2 - 47 12.5.3 Mossbauer spectroscopy
[0sCI6]2- 46
[ReCI 6] 2 - 45 Just as there are ground and excited states of atoms and molecules
[WCiel2- 43 (electronic, vibrational and the like), so too there exist both ground and
[SnCI 6] 2 - 66 excited states of nuclei-they have just been mentioned as sometimes
[TeCI 6 ] 2 - 68 providing a method of measuring NQR spectra. In decaying from an excited
[SeCI 6j"- 56
state a nucleus may emit light, just as an atom or molecule may. In the case
of nuclei, this light is of very short wavelength, it is y radiation. If this y
radiation falls on another, identical, nucleus it may be absorbed, leaving the
second nucleus in an excited state. Nuclei of any one element which are in
different chemical environments will have slightly different energy levels,
but the environment-induced changes are so small that it is possible to
compensate for them with a Doppler shift of the y radiation, achieved by
moving the emitting nucleus either towards or away from the absorber. In
Miissbauer spectroscopy the absorption of·; rays by the sample is recorded
as a function of the velocity of the source. Solid samples are used; the source
may be moved by attaching it to the diaphragm of a loudspeaker driven by
a suitable signal generator. The effect has been observed for relatively few
nuclei at the concentrations at which they occur in most coordination
compounds, of which 57 Fe and 119 Sn have been the most widely studied. A
more complete list is given in Table 12.3. The difference in absorption velocity
and that of a suitable reference standard is called the isomer (or chemical)
shift; it is denoted il and is usually expressed in units of mm s- 1 or em s- 1.
The chemical environment affects the nuclear energy levels through those
electrons which are in orbitals which allow them to make contact with the
nucleus. This means that only s electrons can directly affect isomer shifts
since for all other orbitals the nucleus is contained in a nodal plane (electrons
in p, d, or f orbitals can only influence isomer shifts through their incomplete
shielding of the nucleus, leading to a change in effective nuclear charge which
is felt by the s electrons). Examples of isomer shifts, relative to an iron foil
Table 12.3 Some nuclei which have standard, of 57 Fe in some iron compounds are given in Table 12.4. As this
been studied by Mossbauer spectroscopy
table illustrates, the isomer shift decreases with increase of negative charge
Isotope Natural abundance (%) and increases with increase of coordination number. Both of these general-
izations find application in the observation that the isomer shift of Fe(C0) 5
57 Fe
is greater than that of Fe( CO Ji-. If the same nucleus has two different
2.2
99 Ru 12.0
u9sn 8.6 chemical environments in a compound, these will normally give rise to
121Sb 57.3 separate resonances. In this way it has been shown that the iron atoms in
12sTe 7.0 insoluble Berlin blue, Fe 4[Fe(CN) 6h which is formed by the reaction of
Fem salts and [Fe(CN) 6] 4 - ions in aqueous solution, are not all equivalent
1271 100.0
12gXe 26.2
197Au 100.0
but retain their distinct oxidation states. Indeed, a particular use of Miissbauer
data has been to indicate the valence state of an atom-empirical parameters
are available which compensate for the effects of change of substitu~nt and
Table 12.4 Isomer shifts for some iron
coordination number. So, isomer shift data have been used to conclude that
compounds•
the n bonding ability of ligands decreases in the order
Compound Isomer shift
(mm s- 1 ) NO+ >CO> CW >SO~- > PPh 3 > N0:2 > NH 3
High spin Fe"' ca. 0.3-0.5 A complication arises in that if the resonant nucleus experiences a non-zero
FeF3 (00 ) 0.49 electrostatic field gradient, a quadrupole splitting of the resonance may be
FeCI 3 (00 ) 0.46 observed (nuclear quadrupoles were discussed in the previous section). In
[FeF4r (Td) 0.30
an octahedral environment, such as that in [Fe(CN) 6 ] 4 -, there is no field
Low spin Fe 111 gradient (the chlorine atoms considered in [PtC1 6 ] 2 - in the previous
[Fe(CN) 6 ] 3 - (00 ) -0.12 section were not in an octahedral environment although in an octahedral
High spin Fe 11 ca. 0.9-1.5 complex), but in [Fe(CN) 6 ] 3 - the extra hole in a t 29 orbital has the
FeF2 (0 0 ) 1.48 effect of introducing a very temperature-dependent quadrupole splitting,
FeCI 2 (00 ) 1.16 which presumably occurs through vibrational distortions of the octahedron.
FeBr2 (0 0 ) 1.12 For [Fe(CN) 5 N0] 2 -, where there is a built-in asymmetry, the quad-
[FeCI 2 (H 2 0)4] COo') 1.36 rupole splitting is almost temperature-independent. The [Fe(CN) 6 ] 4 - and
[FeC1 4] 2 - (Td) 0.90
[Fe(CN) 5 N0] 2 - ions are compared in Fig. 12.17. There is evidence that
Low spin Fe 11 quadrupole splittings can be influenced by relatively distant ions in the
[Fe(CN) 6 ] 4 - (00 ) -0.04 crystal lattice-change a counterion, for instance, and there is a small
a Isomer shifts show a slight temperature dependence change in quadrupole splitting.
which has been ignored in thiS compilation. Data
An excellent example of the use of Mi:issbauer spectroscopy in structure
from N. N. Greenwood and T. C. Gibb M6ssbauer
Spectroscopy Chapman & Hall, London, 1971. determination is provided by Fe 3 (C0) 12 . Although the dark green crystals
of this compound are easy to prepare-dissolve Fe(C0) 5 in aqueous alkali
to give the Fe(CO)~- anion and oxidize this with solid Mn02 to give
Fe 3 (C0) 12 -its structure was uncertain for over 30 yearss Although eight
structures, all incorrect, had been proposed for Fe 3 (C0) 12 , it was the ninth,
suggested on the basis of its Mi:issbauer spectrum, which eventually proved
to be correct. The spectrum is reproduced in Fig. 12.18 (peaks are downwards).
The most evident thing is that it corresponds, approximately, to three peaks
of equal intensity. It would be wrong to conclude that each corresponds to
a different type of iron atom. If the iron atoms were all different then they
could not all be in high symmetry environments and so quadrupole splittings
would be expected on most of the peaks. Given that the chemical nature of
the three iron atoms is so similar-and so similar isomer shifts arc to be
expected-the only reasonable interpretation of the spectrum is that the
outer two lines are the quadrupole-split components arising from a peak of
intensity two, centred at about the same position as the central peak (which
itself has but a small quadrupole splitting and is of intensity one). So, it
seems that there are two equivalent, low-symmetry, iron atoms and one of
high symmetry. The correct structure, since confirmed by X-ray studies-
which were made difficult by disorder in the crystal-is shown in Fig. 12.19.
The high-symmetry iron atom is on the right (it has but two types of bond,
I one to terminal CO ligands, the other to Fe atoms). The low-symmetry pair
I
are on the left. The X-ray crystallographic work showed the bridging CO
' - - - [Fe(CN)6l 4- ligands to be asymmetric, off-centre. This means that the low-symmetry iron
atoms are each involved in four different bonds, two to bridging CO
Shift ligands-one long, one short-one to the terminal CO ligands and one to
Fig. 12.17 The Milssbauer spectra of
[Fe(CN) 5 NO]'- (d 5 • ligand assymetry) and 5 The story makes fascinating reading-see 'The Tortuous Trail Towards the Truth' by
[Fe(CN\ 6 ] 4 - (d 6 , symmetric ligand 1ield). R. Desiderata and G. R. Dobson, J. Chern. Educ. (1982) 59, 752.
c:
0
~
0
<II
Fig. 12.18 The Mossbauer spectrum of ~
Fe3 (C0) 12 .
1.0 2.0
Velocity (mm s-1 )
the unique iron atom. A more recent study of the Miissbauer spectrum over
a temperature range has provided explanations for the asymmetries in the
peak-intensity patterns of Fig. 12.18 which are entirely in accord with the
accepted structure.
Finally, mention should be made of the fact that additional information
can be obtained from measurements made with samples in a magnetic field,
a magnetic field that can even be self-generated because of an inherent
magnetism within the sample-a point of particular relevance to Fe 111, d 5
complexes. In Fig. 12.20 is shown the quadrupolar splitting that occurs when
a 57 Fe nucleus is in a magnetic field, together with an energy level diagram
that both explains the observed spectrum and also allows the relevant
selection rules to be deduced.
12.6 Electron paramagnetic (spin} resonance

spectroscopy (EPR, ESR}
As was emphasized in Chapter 9, the energy level splittings caused by placing
a paramagnetic ion in a magnetic field are small, and in order to give a
complete account of them it is necessary to include all of the effects which
give rise to comparable or larger splittings. This is particularly important in
electron spin resonance spectroscopy (or, perhaps a wider and therefore
better, name, electron paramagnetic resonance spectroscopy6 ). In this,
transitions are induced and observed between pairs of levels, split by a
magnetic field.
Suppose that we have a single crystal containing paramagnetic ions, held
co in a fixed orientation in the lattice. At any orientation of the crystal within
an applied magnetic field there will, in general, be a splitting of the energy
levels of the paramagnetic ions. For the moment, let us select just one of
oc these ions. If we now rotate the crystal, keeping the magnetic field constant,
then the magnetic splittings of the ground state will vary with the orientation
c of the crystal, unless both the ion and crystal have perfectly octahedral or
0
Fig. 12.19 The molecular structure of 6 The name electron spin resonance is that which is the more appropriate to the particular
Fe3 (C0) 12 • discussion which follows.

Electron paramagnetic (spin) resonance spectroscopy (EPR, ESR) 1 289
3
2
./- + j.
t
t t I I
at
I
bl c dI e[ f
Fig. 12.20 The Mossbauer spectrum of 57 Fe in I I I I
a magnetic field (reproduced courtesy of
Prof. K. Burger). 1
2 -::. /
',
I I
-6 -2 -4 4 6 8
tetrahedral symmetry, a highly improbable situation. Consequently, the

frequency at which resonance will occur varies with crystal orientation. This
splitting may be thought of as represented by an ellipsoid, with the
paramagnetic ion at its centre. As the ion is rotated, so the ellipsoid rotates
with it, but at all points faithfully represents the relative splittings of the
energy levels between which transitions are being observed. The ellipsoid is
described, mathematically, as a tensor and is referred to as a g tensor, the
splitting of the energy levels at any particular orientation being of the form
g[JH, f3 being the Bohr magneton and H the applied magnetic field. It is the
job of the theoretician to calculate and interpret the g tensor in terms of a
suitable model. In a real crystal, the ions or molecules seldom have their
corresponding axes parallel to each other. In such a real crystal there will
be a family of molecules within a unit cell and a corresponding family of
orientations of molecular g tensors. The result will be distinct sets of lines
that will smoothly move and interchange as the crystal is rotated.
But this is too simple. As has been commented before in this book, it is
helpful to interpret electron spin as the electron behaving like a tiny bar
magnet. The transitions that we have been talking about are, then, associated
with turning the bar magnet end-over-end in the applied magnetic field. But
this is to neglect the presence of other bar magnets, the presence of nuclei
each with their own nuclear spin. These nuclear magnets must also adopt a
particular orientation within the magnetic field and the variety of possible
orientations will be felt by the electron magnet. The result is that the election
spin transitions have an additional structure, the so-called hyperfine structure
N N N N
FIC. 12.21 The EPR spectrum of originating in interactions between the unpaired electron(s) and nuclei with
5-chlorosalicylaldoxinecopper(ll). In this complex non-zero spin. So, copper has two naturally occurring isotopes, each of spin
there are two N atoms, each with spin 1,
coordinated to the copper, so that the resultant
l For each, we expect (2S + 1) = 4 nuclear levels. The EPR spectra of
N spin can be 2, 1, 0, -1 or -2. It follows copper(II), d 9 , complexes therefore show four peaks rather than one.
that a five-line peak structure is expected. The Actually, the two copper isotopes do not give precisely superimposed lines;
Cu spin of ~ gives rise to a four-line structure
(~. ~. -~. -~) and so the spectrum is
rather, two inter-related sets of four lines each, with intensities determined
interpreted as indicated. Clearly, the unpaired by the relative isotopic abundances, 70:30. The (single) four-line pattern will
electron interacts with both copper and nitrogen be seen at the very top of Fig. 12.21.
nuclei. This spectrum is presented in derivative
form, the normal presentation for EPR spectra.
The interpretation of an electron paramagnetic resonance spectrum may
In this presentation the fine structure is made involve all of the parameters introduced in our discussion of magnetic
most evident. For comparison, Fig. 12.22 is susceptibilities in Chapter 9 and Appendix 10-so that EPR measurements
presented in integrated form {reproduced
courtesy of Prof. K. Burger).
can give information on these parameters. Such measurements have the
advantage that they give the individual components of the g tensor, whereas
magnetic susceptibility measurements normally lead only to an average g
value. In suitable cases, individual g values can also be obtained from EPR
measurements on samples in solution. There are problems, however. EPR
spectra can usually only be observed for ions which have spin degenerate
ground states in the absence of a magnetic field-and that means only for
molecules with an odd number of electrons. The spectra observed consist of
transitions between members of this so-called Kramer's doublet (there is more
discussion of Kramer's doublets in Section 9.10 and in Appendix 10). Further,
there are often peak-broadening phenomena which can only be overcome
by working at very low temperatures, perhaps as low as that of liquid helium.
These peak-broadening phenomena are particularly severe when the ground
state has another orbital level not far above it in energy (not far means
comparable with kT ). Thus, it is very difficult to observe the spectra of
octahedral complexes of titanium(III), d 1, for which the ground state is
derived from 2 T 29 by application of spin-orbit coupling and low-symmetry
Photoelectron spectroscopy (PES) 1 291
crystal field perturbations. Here, the ground state is usually such that it has
two orbital (four spin-orbital) terms not far removed from it because of their
common 2 T 29 parentage. Manganese(Il), ds, however, with its high spin
19 ground state, gives room-temperature spectra. Another broadening
6A
mechanism may occur if the paramagnetic species is at all concentrated. For

this reason, single crystal data are usually obtained on a crystal of a
diamagnetic compound, isomorphous with that under study, in which a
very low concentration of the paramagnetic species is incorporated as an
impurity.
It was mentioned earlier that EPR spectra may show fine structure due
to interaction between the unpaired electron(s) and nuclei with non-zero
spin. These nuclei may be those of the paramagnetic ion (for example, 55 Mn,
with spin~ and 100% natural abundance) or those of a ligand. EPR spectra
Magnetic field of copper( II) complexes with nitrogen ligands show evidence for interactions
Rg. 12.22 The EPR spectra of a crystal of
of this sort and have been extensively studied-an example is shown in Fig.
Na2 [PtCI 6 ] · 6H 2 0 containing a small amount of 12.21, where the nitrogen fine structure is additional to the quartet pattern
the isostructural Na 2 [1rCI 6 ]'6H 2 0 as an mentioned above and is indicated by N. The best know·n example, however,
impurity. This spectrum is presented in
integrated form. More common is for the
is that of the EPR spectrum of the [IrC1 6 ] 2 - ion, involving a strong-field
spectrum to be differentiated and the first complex ofiridium(IV), with a d 5 configuration, incorporated as an impurity
derivative plotted, as in Fig. 12.21. in a crystal of Na 2 [PtC1 6 ] · 6H 2 0. The fine structure observed, shown in Fig.
12.22, can be interpreted in terms of interaction of the unpaired electron with
the 35 Cl and 37 Cl nuclei of the ligands. Such observations provide excellent
evidence for covalency in transition metal complexes, emphasizing the
superiority of the ligand field model over the crystal field approach-it
appears that the unpaired electron density in [IrCl 6 ] 2 - is about 5% on each
ligand. As for several other of the techniques described above, EPR is
currently undergoing a rapid development which can only increase its future
impact. Perhaps most important has been the introduction of pulse (and
therefore Fourier transform) techniques, which in turn has led to the same
sort of variety of pulse sequences that has transformed NMR. In addition,
the method is no longer confined to the microwave region but has moved
into the far infrared.
12.7 Photoelectron spectroscopy (PES)

The consequences of the interaction of a beam of light with a molecule
depend on the energy of the light beam. If it is low it can only lead to the
molecule rotating or vibrating. At higher energies, electronic excitations
occur, these becoming more violent with increase in energy of the light beam.
Ultimately, in the vacuum ultraviolet, the interaction leads to the ejection
of electrons. If a monochromatic ultraviolet source is used, the energy of the
incident photons is known. As measurements have to be made under
high-vacuum conditions-this is the vacuum ultraviolet-the energy of the
ejected photoelectrons can be measured by passing them through a magnetic
or electrostatic field and measuring their change in path direction. The
difference in energy between that of the incident photons and of the emergent
electrons is the energy required to ionize the electrons from the molecule.
Such measurements are the basis for photoelectron spectroscopy, a technique
which has provided great insights into molecular structure, particularly for
simple molecules.
It would be ideal to use a vacuum ultraviolet laser as the source of

monochromatic light but, for all practical purposes, such lasers do not
exist 7 Instead, a helium discharge is used. One would expect a multitude of
lines, and indeed they do occur, but with careful control of conditions it is
found possible to have over 99% of the emission in the He' 584 A (21.22 eV)
line. Under different conditions, it is possible, alternatively, to obtain a high
intensity in the He" 304 A (40.81 eV) line. The use of helium gas as a
discharge medium in a ultrahigh vacuum setup is not too much of a
problem-the light from the discharge is admitted into the spectrometer
through a length of capillary tube. Under high vacuum conditions, atoms
and molecules have mean free paths of several centimetres so that a length
of 0.5 mm diameter capillary represents a real obstacle. More of a problem
is removal of the gaseous sample under study after it has passed the narrow
shaft of ultraviolet light. Cunning design and good vacuum pumps provide
the solution!
A PES spectrum consists of many lines. Ignoring fine structure for the
moment, each line in the spectrum may be associated with photoionization
of an electron of different energy. So, if there are four occupied orbitals with
energies above 21.2 eV but six above 40.8 eV, then the He' PES will normally
show four lines and the He11 six. Note the use of the word orbitals in the
preceding sentence. The equation of an ionization energy with an orbital
energy is usually, but not always, valid. It is usually referred to as Koopmans'
theorem, although Koopmans' approximation would be more accurate.
Koopmans' theorem is of greatest validity when an electron is photoionized
from a very diffuse, delocalized, molecular orbital. The adjustment of the
remaining electron density to take account of the loss of an electron is
minimal. In contrast, when an electron is photoionized from a very compact,
localized orbital the local readjustment of electron density may well have
considerable energetic consequences. In such cases it is no use to compare
the numerical results of PES with those from a very accurate ab initio
calculation on the molecule. Rather, the PES data have to be compared with
the differences in total energy between calculations on M and on M + in
the appropriate state. As has been seen in Section 10.7-and will be seen
again, below-this point is of particular importance in molecules such as
ferrocene.
So far, fine structure has been neglected. In fact, fine structure is of key
importance in PES. Suppose that the photoionization observed is from a
bonding orbital. Then the ionized molecule will be more weakly bonded than
the parent and will be produced in what, for it, is a compressed structure.
That is, not only is it produced vibrationally excited-and so there will
be vibrational fine structure in the PES spectrum-but the vibrational
frequencies will be lower than their counterparts in the parent. Conversely,
photoionization from an antibonding orbital gives an ionized molecule with
higher vibrational frequencies than the parent. Photoionization from a
non-bonding orbital gives rise to small or zero vibrational frequency changes.
7 The nearest that one can get in practice is to use monochromatic ultraviolet light
from a synchrotron source. Such sources have the advantage of being tunable-so that the
wavelength of the radiation can be chosen to be the optimum for the measurement-but the
disadvantage that they are only available at a relatively small number of national or
international facilities.
Photoelectron spectroscopy (PES) 1 293
Careful analysis of such vibrational frequencies can even indicate the

symmetry of the orbital from which the electron was ionized.
Other causes of fine structure in the spectrum are spin-orbit coupling and
the Jahn-Teller effect operating in the ionized molecule. These complications
will not be further discussed at this point because they are adequately covered
in Sections 8.4, 11.5 and 8.5, respectively. Two other general points should
be noted: first, that PES is of limited resolution-at best, about 20 meV,
160 em -t, peak half widths (at this resolution a good peak gives about I 00
counts per second for a typical organometallic). This means that the effect
of the so-called fine structure is often to give broad bands and thus band
overlap. Secondly, band intensity changes occur going from He' to He11
which make more difficult attempts to relate band intensity to the degeneracy
of the orbital set from which an electron is photoejected (although the
intensity changes may be related to changes in cross-section; so the metal
3d cross-section increases and the carbon 2p decreases from He' to He 11 ).
In Fig. 12.23 are shown the He 11 photoelectron spectra of Cr(C0) 6 (a)
and W(C0) 6 (b). Their general similarity is obvious. The interpretation
of these spectra is aided by the data given in Chapter 10 in Table 10.4. At
12 16 20 24
(a)
ev-
FJg. 12.23 The He" PES of (a) Cr(C0) 6 and
(b) W(C0) 6 . Adapted and reproduced with
permission from B. R. Higginson, D. P. Lloyd,
P. Burroughs, D. M. Gibson and
A. F. Orchard, J. Chern. Soc., Faraday Trans. 2
(1973) 69, 1659.
8 12 16 20 24
(b)
ev-
(a)
ev-
Fig. 12.24 The He' PES of (a) Mg(C5 H5 ) 2 and
(b) Fe(C5 H5 ) 2 • Adapted and reproduced with
permission from S. Evans, M. C. H. Green,
B. Jewitt, A. F. Orchard and C. F. Pygall,
J. Chern. Soc. Faraday Trans. 2 (1972) 68,
1847.
18
(b)
ev-
about 8.5 eV in the PES, ionization from the metal t 29 d orbitals is seen. For
W(C0) 6 this band has a shoulder, clearly resolved in the He' PES. This
splitting is attributed to spin-orbit splitting in the ti. orbitals of [W(C0) 5 ]+.
Between 13.3 and 14.4 eV lie the CO n bonding combinations t 1., t 19 and
t 2 •• The t 29 member of the n bonding set appears within the span of the CO
a bonding combinations e., t .. and a 1 • at lower energies (they vary in going
from Cr(C0) 6 to W(C0) 6 but are centred at ca. 15.6 eV). In the region
17-22 eV lie the oxygen a lone-pair combinations a 19 , t 1• and e9 (these are
not included in Table 10.4). It is to be noted that although the oxygen a
(outward-pointing) orbitals are physically well separated in the molecule,
the involvement of, and mixing with, what are more inner orbitals-of the
same symmetry-is sufficient to cause large splittings.
As a final example consider the case of ferrocene, Fe(C 5 H 5 ),, for which
the He' PES is given in Fig. 12.24 (b) together with that of Mg(C 5 H 5 ), (a).
Although there are marked similarities between the spectra there are also
clear differences. A pair of bands at ca. 7 eV in ferrocene have no counterpart
in magnesocene and are to be associated with ionization from the d orbitals
of the transition metal. The pair of bands between 8 and I 0 eV are closer
together in ferrocene whilst for this molecule the broad band between 12
and 14 eV is more structured. The pair of bands at ca. 7 eV have been
assigned to e2 g(3d) and an a 1 g(3d)-yet in Section 10.7, and in particular
Table I 0.3, we saw that in ab initio calculations on ferrocene these are rather
Evidence for covalency in transition metal complexes I 295
low-lying orbitals and not those that would be expected to be the most readily
ionized. We have a breakdown of Koopmans' theorem. The reason for this
pattern-in terms of the high degree of localization of the electrons in these
orbitals and so the large readjustment energy of the remaining electrons-
has been discussed in Section 10.4. Returning to the photoelectron spectrum
of ferrocene, the 8-10 eV bands have been assigned to e 1u(n-Cp) and
e 1g(n-Cp) whilst the 12-14 eV region is assigned, qualitatively, to ionization
from ligand CT and n orbitals. It is illuminating to compare the experimental
patterns with the alternative sequences given in Table 10.3.8
One final comment: in this section we have been concerned with vacuum
photoelectron spectroscopy. A similar X-ray photoelectron spectroscopy
exists in which the ejected electrons come from inner electron shells. The
ejected electron energy is sensitive to the chemical environment from which
it originates and so gives information on this.
12.8 Evidence for covalency in transition metal

complexes
This is a convenient point at which to bring together the various pieces
of evidence indicating covalency in transition metal complexes although it
must be remembered that not all are applicable to every species. The pieces
of evidence are listed below.
1. Ligand hyperfine splittings in EPR spectra, discussed in Section 12.6.
2. The percentage of ionic character in metal-chlorine bonds as measured
by NQR spectroscopy, discussed in Section 12.5.2.
3. Unpaired electron spin densities on ligand atoms as measured by NMR
spectroscopy, discussed in Section 12.5.1.
4. The superexchange mechanism of antiferromagnetic coupling which
enables the electrons on one atom to see the spin of those on another
and which operates through their mutual overlap with a ligand atom,
discussed in Section 9.11.
5. The orbital reduction factor k, which is introduced to allow for
the effects of covalency on the magnetic properties of paramagnetic
species, mentioned in Chapter 9 and discussed in some detail in
Appendix 10.
6. The reduction in spin-orbit coupling constant in a complex compared
with the free ion value, discussed in Section 8.4.
7. The reduction in electron repulsion energies (the nephelauxetic effect)
in a complex compared with their values for a free ion, covered in
Section 8.3.
8. Accurate electron density maps obtained from X-ray diffraction data,
which show that there is no point of zero electron density on metal-ligand
axes. It is therefore not possible to regard metal and ligand electrons
as in any way separate.
8 For a more detailed discussion see G. Cooper, J. C. Green and M. P. Payne, Mol. Phys.
(1988) 63, 1031.
9. Polarized neutron diffraction measurements, which show electron spin

densities on the ligands, albeit not always of the spin expected, discussed
in Section 7.6. These data not only show the need for a model which
includes covalency but also indicate the limitations of ligand field and
other one-electron models.
10. The use of shift reagents in NMR, outlined in Section 12.5.1, which
show that unpaired electron density in both transition metal and
lanthanide complexes can be felt by molecules outside the coordination
sphere.
11. The most precise theoretical calculations, described in Chapter 10,
supported by the results of photoelectron spectroscopy-see Section
12.7-unambiguously support the general conclusions of ligand field
theory. However, a general tendency on the part of inorganic chemists
to qualitatively over-estimate the involvement of d orbitals in bonding,
particularly for main group elements9 but also for transition metals, is
to be noted.
12. Band intensities: some spectral bands involving electronic transitions
have intensities other than those which might be expected in terms of
a purely ionic model. This evidence tends to be more speculative than
the others listed above and so it has not been discussed earlier in the
text. There are four intensity anomalies which may be taken to indicate
covalency.
(a) d - d electronic transitions which are more intense when close to
charge-transfer bands. This is indicative of a mixing between the
two classes of transition. which, in turn, implies covalency within
the M-L bonds; see Section 8.10.
(b) Some metal-ligand stretching vibrations have much lower infrared
intensities than would be expected for vibrating, non-overlapping,
ions; see Section 12.2.2.
(c) The optical activity introduced into formally d - d electronic
transitions by an optically-active ligand may be much greater than
expected; see Section 12.3.
(d) In Chapter 11 the mathematical treatment of the origin of f- f
band intensities in lanthanides and actinides was not discussed. The
basic theory, the so-called Judd-Ofelt model, contains a number of
parameters. Of these, at least two are generally regarded as associated
with covalency.
12.9 Molar conductivities

Electrochemical methods have been part of coordination chemistry from its
earliest days. In recent times there has been a revival of interest in such
techniques, so we start with a brief overview of the traditional as a prelude
9 As an example the case of SF6 may be cited; the bonds are rather polar, a feature which
explains its properties without the need to invoke significant d orbital participation on the
sulfur. See J. Cioslowski and S. T. Nixon, Inorg. Chern. (1993) 32, 3209.
Cyclic voltammetry 1 297
to considering the more modern in the next section. In water, equivalent

conductivities-the contribution of an ion to the equivalent conductance of
a salt at infinite dilution-are about 60 ohm~ 1 at 20 "C for each species
(large inorganic ions give lower values whilst H+ and OH ~,for which chain
conduction mechanisms are available, give higher values). As a first approxi-
mation, then, one would expect the contribution to the molar conductivity
of an ion y+ to be about 60r ohm~ 1 . For ions Mm+ and X"~ in the salt
M.Nm, the contributions will be 60m ohm~ 1 (from Mm+) and 60n ohm~ 1
(from X"~). Multiplying by the number of ions of each type and adding leads
to the conclusion that a salt M. Nm will have a molar conductivity of about
120 ohm~ 1 at 20 "C. So, a compound analysing as CoCl 3 5NH 3 is found to
have a molar conductivity of 261 ohm~ 1 and it is concluded that it is
[Co(NH 3 ) 5 Cl]Cl 2 with nm = 2. Similarly, CoC1 3 5NH 3 · H 2 0 has a molar
conductance of 390 ohm~ 1 and so is [Co(NH 3 ) 5 (H 2 0)]Cl 3 , with nm = 3.
Water is a coordinating solvent and, particularly for labile complexes, is best
avoided for conductivity measurements. Nitrobenzene and nitromethane are
commonly used alternatives, although solubility problems can arise. The
usual procedure is to measure the conductivity at a concentration of ca. 10~ 4
molar and, assuming a molecular weight, to compare the value with that
obtained for similar complexes of known ion type (I :1, 1:2, etc.). A rather
better method is to make measurements over a range of concentrations and
to plot i,o and i.,(c), where i.,(c) is the equivalent conductivity at infinite
dilution, against Jc. Again, this plot is compared with those obtained for
similar species in the same solvent. The equivalent weight of the complex is
needed for this method and this may be obtained by choosing a suitable
counterion for the complex species. The potassium ion is a suitable cation
and the perchlorate has been widely used as a suitable anion, although,
following the discussion of Section 4.2.2, the triflate anion CF3 S03 may well
be preferred. None of these species is involved in strong coordination and
so, assuming that one can ignore their complex-forming ability for the
compound under study, the equivalent weight may be obtained from its
empirical formula.
12.10 Cyclic voltammetry

As is well known, oxidation and/or reduction of ionic species in solution
can occur on electrolysis. This is the basis of electroplating and, in reverse,
the mechanism of action of many batteries. Controlled potential electrolysis
as a preparative method was met in Section 4.2.3. Cyclic voltammetry is a
method of studying such oxidation-reduction processes in detail. It is a
method that has gained much in popularity in recent years since among
other things it enables the measurement of thermodynamic redox potentials.
The word cyclic in the name refers to the fact that if in the measurement
A --> B is an electrolytic oxidation, in the measurement it is immediately
afterwards followed by B --> A in an electrolytic reduction. This might sound
rather like a pointless exercise but it, in fact, enables the fate of B to be
probed immediately after it is formed. If it is found that there is less B to be
reduced than there was B formed, experiments can perhaps be designed to
investigate its fate. The method also provides information on the reversibility
of electron gain/loss processes, their kinetics and on the strategies that might
be successful in the bulk preparation of species for which there is cyclic
voltammetric evidence.
At its simplest, cyclic voltammetry (CV) consists of applying a saw-tooth
voltage across two electrodes (often of platinum but carbon or mercury are
also used), usually in an aqueous solution containing a supporting electrolyte
as well as the ion under study. The supporting electrolyte is involved in
electrical current transfer across the bulk of the cell but is not involved in
the electrode reactions. In order to ensure that this is so, the voltage range
of the saw-tooth is limited. To avoid complications caused by the electrolytic
reduction of dissolved oxygen, the solutions are degassed; when non-aqueous
solvents are used great care must be taken to remove traces of water. The
concentration of the ion under study is usually less than I% of that of the
supporting electrolyte. This means that as the electroreduction (say) of
A--> B progresses, the layer of the solution adjacent to the electrode becomes
depleted of A and so the electroreduction, ultimately, becomes diffusion-
controlled. If the potential across the cell is now reversed, as it will be if
the applied saw-tooth spans both positive and negative potentials, then,
ultimately, the electrooxidation B --> A will occur. If B has not been reduced
in concentration by some reaction, a similar pattern will be followed~a
maximum current followed by a drop to a diffusion-controlled limit. The
sequence is illustrated in Fig. 12.25.
Although a sequence A --. B; B --. A has just been followed, it is clear that
if some B has reacted en route then the final condition is not the same as
the starting. That is, successive voltammagrams will not be identical. A
further reason for a difference would be if the products of the side-reactions
of B were themselves electroactive. An important variable in studies aimed
at investigating such phenomena is the slope of the saw-tooth waveform.
Rates of up to ca. 100 V s- 1 are common~ which means that a complete
scan takes only about 10- 2 s~although they can be very much slower.
The method is a very useful one for the qualitative, rather than quantitative,
characterization of compounds. It has found common application in bio-
inorganic chemistry, where species often show a family of oxidation-
reduction steps. If one were trying to synthesize a model complex analogous
to a naturally occurring bioinorganic, one question that it would be natural
to ask is, does it have a similar CV behaviour?
Two final comments: first, although the above discussion has been about
a two-electrode process, the real-life experiment involves at least three, the
third being a reference electrode (usually calomel or silver/silver chloride).
If the supporting electrolyte contributes significantly to the current~as may
well happen if, for instance, the pH is well away from 7, then a four-electrode
cell can be arranged in such a way that the contribution from the supporting
electrolyte is cancelled. Finally, as hinted above, not all electrode reactions
are completely or immediately reversible. Such irreversibility is indicated by
the separation between the cathodic maximum current potential (E,) and
the anodic maximum current potential (E.) being separated by a potential
which is greater than expected (which is ca. 0.059/n, where n is the number
of electrons involved in the electrode reaction).
X-ray crystallography I 299
Ec
20
'-'
'6
_g 10
"
'-'
op.-----
Fig. 12.25 A cyclic voltammagram of 6 mM

K3 [Fe(CN) 6 ] in 1M KN0 3 . Adapted and
'-'
reproduced with permission from P. T. Kissinger '6
0
and W. R. Heineman, J. Chem. Educ.
(1983) 60, 702. ffi -10
-20 ~--~----~--~--~--~
0.8 0.6 0.4 0.2 0 -0.2
Potential, V versus SCE
54 3 210-1-2-3-4-5
La Fem(CN)~-
g Fe"(CNJt'
12.11 X-ray crystallography

Undoubtedly the most important method of structure determination at
the present time is that of X-ray crystallography. It is a large subject, of
which only a few points can be highlighted in the present section. In section
11.7 the classical model of the way that a light beam interacts with matter
was detailed; in particular, the light beam is slowed down. This happens
because the light beam induces a dipole within each atom or molecule, which
then re-emits. The same happens with X-rays but with the important
difference that the re-emissions occur from atoms separated by distances
comparable to or greater than the X-ray wavelength. The result is that both
constructive and destructive interference can occur between the re-emitted
beams. For crystalline samples, destructive interference is the norm, con-
structive only occurring at very special angles between the incoming
(monochromatic) X-rays and the crystal. These angles are determined by the
details of the regular, repeated, geometrical arrangement of the atoms in the
crystal and also by the X-ray wavelength. The intensities of the emitted,
diffracted, beams depends on the position of each atom in the regular array
and, most important, on the chemical identity of the atoms, for this
determines the strength with which each atom re-emits (basically, the more
electrons the stronger the re-emission). So, measurement of the angles of the
diffracted beams gives details of the regular array, and thus to some extent
structure, and measurement of their intensities give the information of
greatest chemical interest, the identities of the atoms and of their positions
relative to other atoms. Clearly, measurement of both angles and of
intensities is important. In most laboratories the process is now highly
automated. A tiny single crystal is selected and mounted on a fibre oflithium
borate glass (low atomic weights, low X-ray scattering). Some preliminary
diffraction measurements enable the computer to orient the crystal and to
orchestrate the collection of data. After some data manipulation, the
intervention of the researcher is generally needed to choose the most
probably correct, but approximate, crystal structure, which is then refined
by the computer and the full data set deconvoluted. It sounds too easy!
Whilst, in favourable cases, little knowledge is needed to solve a crystal
structure, learning how to control the computer can be the major hurdle-
there are traps for the unwary. First, the theory assumes that the crystal is
uniformly bathed in X-rays. No scattering, no diffraction, no absorption.
This is why the crystal has to be tiny, perhaps almost too small to see. This
points to a major problem but one that is not always explicitly addressed,
that of whether the crystal studied is representative of the sample from which
it is selected. Next, heavy atoms dominate the scattering process, so that in
structures containing them it is difficult to accurately place light atoms
nearby and sometimes even to identify these light atoms. In many structures,
hydrogen atoms are not detected (neutron diffraction, best on deutero
materials, solves this problem). Again, at room temperature the atoms in a
crystal will be, thermally, vibrationally excited. This motion 'blurs out' the
structure, reducing precision. However, for some samples, cooling-which
would normally give better precision-shatters the crystal (this was a
problem for ferrocene). Particularly if rather rounded molecules are involved,
disorder can occur in a structure, a disorder which has to be modelled in
some way before the crystal structure can successfully be refined. Rounded
molecules can often be accommodated in several different ways in the
structure, all arrangements being of comparable energy. In such a case
the model would be one in which all of the possible arrangements are
superimposed, each weighted according to its occurrence (the weightings
would be varied to obtain the best fit). A different type of disorder occurs
when solvent molecules occupy some, but not all, equivalent sites. Finally, at
several points in this chapter, as well as elsewhere in the book, mention has
been made of spectroscopic measurement on crystals. Apart from the far
from trivial fact that these often demand crystals which are very much larger
than those for X-ray structure determinations (so that it is not a trivial
problem to check that the large crystal has the same structure as that on
which X-ray work was carried out), it is important to recognize that all
spectral interpretations demand the use of a primitive unit cell. This is because
almost all spectral measurements use wavelengths which are large compared
to typical interatomic separations. Molecules or ions related by pure
translations experience the same electric vector originating in the incident
light wave, for example. To deal with this, the pure translations must be
correctly chosen-and the correct choice is that set which interrelates
primitive unit cells. If, as may well happen, crystallographers find it
convenient to work with a centred unit cell (body or face-centred), the

spectroscopist-and theoretician-has to work with the corresponding
primitive unit cell (there always is one). This particular point will be met
again in Chapter 17.
12.12 Conclusion
In this chapter some of the more important methods of studying coordination
compounds have been reviewed. The detailed interpretation of the experi-
mental results is often rather difficult, and, for the phenomena of optical
activity and NQR, for example, the theory of the method has only
been worked out incompletely. Discussion has therefore been confined to
the qualitative level, it being considered important that the student should
have a pictorial idea of the phenomena considered. In this way he or she
should both have been made aware of which technique is likely to be of use
in tackling a particular problem and also of some of the difficulties associated
with its application. Inevitably, it has been necessary to be selective and this
has been done on the basis either of techniques which the student may well
meet in the laboratory or of techniques which are of particular importance.
Finally, Chapter 16 will provide examples of the application of some of the
techniques described in the present chapter, as well as a few more which
have proved to be of particular value in the study of bioinorganic molecules.
Further reading • 'An Introduction to Cyclic Voltammetry' by G. A. Mabbott,

J. Chern. Educ. 60 (1983) 697.
Two books which cover the majority of the methods described • 'Cyclic Voltammetry-Electrochemical Spectroscopy' by J.
in this Chapter are Structural Methods in Inorganic Chemistry Heinze, Angew. Chern. Int. Ed. (1984) 23, 831.
by E. A. V. Ebsworth, D. W. H. Rankin and S. Cradock,
Blackwell, Oxford, 1987 and NMR, NQR, EPR and Mossbauer • 'Cyclic Voltammetry' by P. T. Kissinger and W. R. Heineman,
Spectroscopy in Inorganic Chemistry, R. V. Parish. Ellis Hor- J. Chern. Educ. (1983) 60, 702.
wood, Heme! Hempstead, 1990. An older book, but one that • 'Molecular Electrochemistry' by R. G. Compton and A. R.
covers a wider subject area, with some good chapters is Physical Hillman, Chemistry in Britain (1986) 1088.
Methods in Advanced Inorganic Chemistry H. A. 0. Hill and P.
Day, eds., Interscience, London, 1968. Readable insights into a • 'Nuclear Quadrupole Interactions in Solids' by J. A. S. Smith,
limited number of particular areas can be found in Spectroscopy Chern. Soc. Rev. (1986) 15, 225.
of Inorganic-Based Materials (Advances in Spectroscopy Vol • 'Modern Techniques in Electron Paramagnetic Resonance
14). R. J. H. Clark and R. E. Hester, eds. John Wiley, Spectroscopy by J. H. Freed, J. Chern. Soc., Faraday Trans.
Chichester, 1987. (1990) 86, 3173.
Other, complementary, useful references are:
• Electron Paramagnetic Resonance of d Transition Metal Com-
• 'A Primer on Fourier Transform NMR' by R. S. Macomber pounds (Studies in Inorganic Chemistry, Voll6). F. E. Mabbs
in J. Chern. Educ. (1985) 62, 213. and D. Collison. Elsevier, Amsterdam, 1992. A large reference
book that at first sight appears mathematical. In fact, there
• Fourier Transform Raman Spectroscopy. Instrumentation and are some easy-to-read chapters; the derivations are presented
Chemical Applications, P. Hendra, C. Jones and G. Warnes, relatively simply and certainly completely.
Ellis Horwood, Chichester, 1991. A general introduction, not
confined to coordination compounds. • 'Pulsed Electron Spin Resonance Spectroscopy: Basic Prin-
ciples, Techniques and Examples of Application' by A.
• 'Polarographic Behaviour of Coordination Compounds' by Schweiger, Angew. Chern., Int. Ed. (1991) 30, 265.
A. A. Vlcek, Frog. Inorg. Chern (1963) 5, 211.
• 'Some Aspects of the Electron Paramagnetic Resonance
• 'Circular Dichroism of Transition Metal Complexes' by R. Spectroscopy of d-Transition Metal Compounds' by F. E.
D. Peacock and B. Stewart, Coord. Chern. Rev (1982) 46, 129.
• NMR, NQR, EPR and Miissbauer Spectroscopy in Inorganic There are two key books on the applications of vibrational
Chemistry by R.V. Parish, Ellis Horwood, Chichester, 1990. spectroscopy: Metal-Ligand and Related Vibrations by D. M.
A book with a minimum of theory which provides an Adams, Edward Arnold, London, 1984; and Infrared and
excellent follow-up for the material in the present chapter Raman Spectra of Inorganic and Coordination Compounds, by
and a bridge to more theoretical treatments. K. Nakamoto, J. Wiley, New York, 1986.
• 'NMR and the Periodic Table' R. K. Harris and B. E. Mann A useful source is Volume I of Comprehensive Coordination
(eds.), Academic Press, New York, 1978; provides an older, Chemistry G. Wilkinson, R. D. Gillard and J. A. McCleverty
but excellent and reasonably readable, specialist treatment. (eds.), Pergamon Press, Oxford, 1987, Chapter 8.1 'Electro-
A very readable review of modern aspects and applications chemistry and Coordination Chemistry' by C. J. Pickett.
of NMR is to be found in the July 1993 (page 589 on) issue
of Chemistry in Britain.
Questions 12.3 In the footnote on page 271 it was claimed that infrared
spectroscopy could distinguish three different M(CO), geom-
12.1 Figure 12.26 shows the infrared spectra of cis and trans etries. Substantiate this claim by working out the predicted
[Pd(NH 3 ),Cl 2 ] in the v(Pd-X) region. Which spectrum corre- allowed bands for each geometry (it will be necessary to use
sponds with which isomer? (Figure adapted from R. Layton, group theory). Could a similar claim be made for Raman
D. W. Sink and J. R. Durig, J. Inorg. Nucl. Chern (1966) 28, spectroscopy?
1965).
12.4 It was found that a complex containing an Fe(CO),
unit, when in solution, gives an infrared spectrum with two
bands of equal intensity (to within experimental error) in the
v(C-Q) region. What is the C-Fe--C bond angle?
12.5 There are reports in the literature-including a (very

early) crystal structure-of a compound [Pt(CH 3 ) 4 ] 4 , con-
taining a tetrahedron of platinum atoms with a CH 3 spanning
each face. More recent work suggests that the compound is
[Pt(CH 3 ) 4 0H] 4 , with faces centred by OH groups. Which of
the methods described in the present chapter might be expected
600 400 600 400 to distinguish between the two formulations?
Fig. 12.28 Infrared spectra, Question 12.1.
12.6 You have been given the task of equipping a general
inorganic research laboratory with a limited but ill-defined
12.2 Based on a study of Fig. 12.3 and using a group budget. Place the instrumentation associated with the tech-
theoretical approach, suggest, qualitatively, what band pat- niques described in this chapter in order of priority on your
terns might be observed in the infrared spectra of crystalline shopping list. For each instrument write a short paragraph
K4 [Fe(CN) 6 ] and Na 4 [Fe(CN) 6 ]. explaining the reasons for its position on your list.
Thermodynamic and related
aspects of ligand fields
13.1 Introduction
One of the earliest applications of crystal field theory was its use to explain
irregularities in thermodynamic and related properties of a series of transition
metal complexes as the transition metal was varied. These applications are
usually dealt with in one of the early chapters of a book such as this; their
consideration has been deferred in order to be able to draw on the
background and additional insights provided by Chapters 6, 7 and II.
Although the notation that will be used implies the use of the crystal field
model, the discussion is not limited to this, as the title of the chapter
indicates. Rather, the concern will be with energy level separations, con-
veniently but not always accurately, thought of as resulting from orbital
energy level differences- hence the convenience of using the language of
simple crystal field theory, even if we do not really mean it! Thermodynamic
aspects of coordination chemistry have already been encountered in this
book- they represent the major theme of Chapter 5. The present chapter
is distinguished from Chapter 5 because here we deal with the fine detail;
the earlier chapter was more concerned with gross effects.
13.2 Ionic radii

The crystals of simple salts such as NaCI and ZnS have lattices which are
often regarded as ionic. That this is a plausible approximation is indicated
by, amongst other things, the fact that it is possible to assign ionic radii and
to use the approximate additivity of these radii to predict interatomic
distances.1 Crystal field theory similarly assumes that the forces between a
1 It is perhaps not sufficiently well recognized that although the ionic radius approach is
reasonably self-consistent, the radii discussed may not correspond to physical reality. For
example, in the KCI crystal the minimum of electron density along the K-CI axis is ca. 1.45 A
304 1 Thermodynamic and related aspects of ligand fields
X=l 3.25
1.1
+~+ 3.00
<llN
<l>
UN
+
c
I 1.Q "";"~+ § i
.!!? +
Fig. 13.1 The lanthanide contraction and <l>"'
'0 :J
·c :c ' ... -~ +_,"'
'+..... 'ON
X~
metal-halogen distances for divalent first-row
transition and related metal ions. The thin solid "'"'
.£
~-~
~ Lanthanides 2.75 I
II
::;;::;;
lines indicate the probable values for the M-X ro r:: 0.9
_J .Q
distances corrected for crystal field effects.
2.50
0.8
Atomic number ____...
transition metal cation and the surrounding ligands are purely electrostatic.
The theory suggests, therefore, that it should be both valid and possible to
obtain values for the ionic radii of transition metal ions from crystallographic
data. How might these radii be expected to vary from one ion to the next?
If crystal field effects are neglected, the essential difference between adjacent
members of a series of ions such as Ti3+, yH, Cr3+ and so on, is that each
successive member has an extra positive charge on its nucleus and an extra,
compensating, d electron. As a first approximation, therefore, no change in
ionic radii along the series might be expected. Recognizing the incomplete
screening of the additional positive charge by the additional electron,
however, a small decrease in ionic radius seems more probable. These
qualitative ideas find support in the gradual decrease in ionic radius exhibited
by successive trivalent ions in the lanthanide series-the so-called lanthanide
contraction, although here, of course, it is f electrons which are involved.
This particular ionic radius decrease is shown in Fig. 13.1 together with the
metal-ligand separation in halides of the first transition series (in these the
metal is octahedrally surrounded by halide ions). By plotting metal-ligand
separations an assumption about the ionic radius of the halide ion is avoided.
Values for copper(II) compounds have been omitted from this diagram.
These complexes are usually highly distorted, so that Cu-X distances both
greater and smaller than the values predicted by interpolation of the data
for adjacent ions have been reported. The smooth curves which were
anticipated do not appear in Fig. 13.1. There is a simple explanation for this
which becomes evident when crystal field effects are included. In the absence
of a crystal field all of the metal d orbitals are degenerate and so, effectively,
equally occupied. Application of an octahedral crystal field removes the
degeneracy, so that the t 2 " orbitals are preferentially filled. This means that
the d electrons are preferentially placed in orbitals which do not screen the
ligands from the increased (attractive) nuclear charge as we go from one
metal to the next. When an electron is placed in an e" orbital, however, it
has an enhanced screening effect, because these orbitals are concentrated
along the metal-ligand axes. For a d 5 (high spin) configuration all of the d
from the potassium and 1.70 A from the chlorine, compared with the usually quoted
ionic radii of 1.33 A and 1.81 A respectively. However, the (presumably) more accurate radii,
such as that of 1.70 for Cl-, arc not constant but vary from compound to compound and so
the additivity is lost.
Heats of ligation 1 305
orbitals are equally occupied and the screening is the same as it would have
been in the absence of a crystal field. It follows that the additional screening
of an electron in an e, orbital compensates for the deficiencies in the screening
of one and a half electrons in the r2 , orbitals. The deviation of ionic radii
from that given by the simple picture is therefore expected to be in the order:
t~ > t~ > t~eg > ttt > t~ "' t~ei > t~g > t~gei
"' t~eg > t2g "' ttte~ > t~e~ "' t~e; "' 0
Remembering that all of the examples given in Fig. 13.1 arc high spin, it will
be seen that this series is followed quite well. The only exception appears to
be d 7 (high spin) cobalt( II) for X= 1-. However, it has already been
recognized in Section 7.5 that cobalt(II) has the configuration ~~~1 5 e: 115 in
the weak field limit--and I- is a weak field ligand. If the weak-field limit
values are taken, then this configuration for cobalt(II), along with d 2 which
has a configuration ~~~5 e:l', should be placed with the configurations
rt,e~ and t~ 9 e 9 in the above series. In practice, high spin cobalt( II) complexes
have configurations intermediate between ~~:1 5 e: 1/ 5 and ti.e; and so occupy
variable positions in the series. A consequence of this general discussion is
that if the distribution of electrons between t 2 , and e, orbitals changes, then
there will be a synchronous change in ionic radius. So, the ionic radii of
transition metal ions in electronically excited states are expected to differ
from their ground state values (to be greater, because the e, population
increases for spin-allowed d-d transitions). Conversely, when the population
of the r2 , orbitals increases at the expense of thee, a decrease in ionic radius
must be expected. It will be seen in Section 16.2 that just such a change
seems to be involved as a trigger in a mechanism which causes iron(II) ions
some 30 A or more apart in hemoglobin to be sensitive to whether or not
their partners in the molecule are coordinated to 0 2 .
13.3 Heats of ligation

Crystal field theory was introduced in Chapter 7 by considering a free,
gaseous, ion and then placing it in an octahedral crystal field. The heat
released in this process is known as the heat of ligation (the ligands are also
assumed to originate as free, gaseous, molecules). Values of the heat of
ligation may be obtained from experimental data on reactions in solution
by the use of suitable energy cycles. The heat of ligation is conveniently
broken down into several parts. Thus, within a purely electrostatic model,
the most important term is the stabilization resulting from the attraction
between the ligand and the transition metal ion. A destabilization results
from the electrostatic repulsion between the ligands themselves--the effect
of this term is to cause a decrease in the increments of energy liberated as
additional ligands are added to the central atom. Because the energy released
when two point charges of opposite sign are brought together from infinity
depends inversely on their final separation, it would be expected that the
major, stabilizing, term would be modulated by the changes in ionic radius
discussed in the previous section. However, there is another term which
changes from one transition metal to another, as the following hypothetical
sequence shows. Form an octahedral complex ion from its infinitely separated
Table 13.1 Crystal field stabilization energies for weak field and intermediate field
octahedral complexes. Where alternative configurations are given, the fractional
values are the weak-field limit and the integer values are the strong-field limit
(between them they give the intermediate field range)
dn Crystal field Crystal field
configuration configuration stabilization energy
do 0
dl t~g -2/51!.
d2 t§.. -4/51!.
t~sel;s -3/51!.
d3 f,g -6/51!.
d4 f,gei -3/51!.
d5 ~.e~ 0
d6
~i -2/51!.
d7 t~ge~ -4/51!.
~;;seil/5 -3/51!.
d" F,ge~ -6/51!.
dg q.e~ -3/51!.
dlO F,ge; 0
components, so that the final metal-ligand distance is that observed in

the complex. Energy will be liberated, most of which will be lost from the
molecule. Suppose that sufficient is retained for the split d orbitals all to be
equally populated, so that the electron distribution is identical to that in the
free ion. At this state in the process the energy lost will differ from one
transition metal ion to another only by virtue of
• their different effective nuclear charges;
• the different metal-ligand distances.
Of these, the first would be expected to vary smoothly along the series.
Now, let the d electrons assume their ground-state configurations. An
additional increment of energy will be liberated if, in the ground state, the
d orbitals are not equally occupied. This energy, the so-called crystal field
stabilization energy, will vary with the metal ion. It is a simple matter to
calculate crystal field stabilization energies for they depend only on the
electron configuration and the crystal field splitting 11. An electron in the
t 29 set contributes a stabilization, of -2/5 11, one confined to the e9 set
contributes a destabilization of 3/5 ,1, (see Fig. 7.9). Crystal field stabilization
energies for weak field octahedral complexes are listed in Table 13.1. A
similar table can be constructed for strong field complexes but an additional
destabilizing term, allowing for the pairing energy of two d electrons, would
have to be added. The presence of this term makes the interpretation of the
experimental data more difficult and so our discussion is confined to weak
field complexes. In Fig. 13.2 are shown the variation of heats of ligation of
transition metal ions with water as ligand. The data are most complete for
this ligand but the available data for other ligands indicate a similar
behaviour. The deviations from a curve drawn through the data points for
calcium(II), manganese(II) and zinc(II) are in good agreement with the
Lattice energies 1 307
-2000
,
Fig. 13.2 Heats of hydration of first row 0
transition metal ions. Experimental values E
(which are subject to varying errors) are shown
by circles. When corrected for the contribution
made by crystal field stabilization energy
(crosses) they conform to a smooth behaviour
(solid line).
-1500~----~----~--~--~~--~--~--~-----
crystal field stabilization energies given in Table 13.1. Indeed, if one corrects
for the crystal field stabilization energies, using this table and the spectro-
scopic values oft., then the resulting points fall on an almost straight line.
For cobalt(II), d 7 , two points are shown, corresponding to stabilizations of
-3/5 t. and -4/5 t., for the reason discussed above. The curve passes
between these limits. This relationship between experimental and corrected
values of the heat of ligation has led to the suggestion that it may be used
as a method of obtaining approximate values oft.. However, our discussion
suggests that the agreement that is found with the spectroscopic values oft.
is somewhat fortuitous, for the ionic radius effect must be superimposed on
the crystal field stabilizations in Fig. 13.2. Further, several assumptions have
been made in the above discussion. In particular, one fact has been
overlooked- that when the randomized arrangement of d electrons changes
to the ground state arrangement there is an increase in the effective
charge seen on the cation by the ligands-the number of e" electrons
screening the nucleus decreases. Accordingly, there is a change in the
electrostatic energy. This emphasizes the interrelationship between effects
which have been discussed separately. However, since the contributions to
the total energy arising from the variation in metal-ligand distances and the
crystal field stabilizations act in the same direction-compare the inequalities
given in Section 13.1 with the relative magnitude of stabilization energies
seen in Table 13.1--the essential point is that the seemingly rather erratic
nature of the experimental data in Fig. 13.1 can be understood.
13.4 Lattice energies

In the previous section an isolated complex ion was discussed. Ionic
lattices, in which the structure consists of interconnected octahedra, tetrahedra
or other geometrical arrangements of ligands, when regarded as coordination
compounds, may be similarly treated. Here, the lattice energy replaces the
heat of ligation as the experimental quantity under discussion. A complication
arises when the members of a series of compounds which one wishes to study
are not isomorphous; they crystallize with different crystal structures,
possibly due to the irregularities in ionic radii that were discussed in the
previous section. A detailed consideration of this problem for the first row
308 I Thermodynamic and related aspects of ligand fields
2800
X~ Cl
~
0'
E X~ Br
£
x~1
Fig. 13.3 Lattice energies of the dihalides,
MX2 , of the first row transition series and
related elements.
2000~----~~--~~----~------L--L--------
dihalides indicates that here, at least, the differences in lattice energies

resulting from these complications are small and may be neglected as a first
approximation. In this way the data shown in Fig. 13.3 were compiled. The
now-familiar pattern is repeated; its interpretation closely follows that of the
previous section.
13.5 Site preference energies

It has been seen that the crystal field stabilization energy is a relatively small
component of the total energy involved in the formation of a complex in the
gas phase-compare the corrections in Fig. 13.2 with the total energy
involved, as indicated by the scale of the ordinate. In a more realistic
situation, when, say, the hexaaquanickel(II) cation is converted into tris-
ethlylenediaminenickel(II), there will be a change in crystal field stabilization
energy because water and ethylenediamine exert rather different crystal fields
(the complexes are green and purple, respectively). However, the major
change is due to the difference in electrostatic attraction between Ni2+ and
a water and an ethylenediamine molecule (there are also entropy and heat
terms which favour the formation of the ethylenediamine complex and which
were discussed in Section 5.5). Unless it is certain that these changes are
zero, or at least very small, any argument based on crystal field stabilization
energies alone must be regarded with suspicion. If they lead to predictions
which agree with experiment, this may be because the changes of crystal field
stabilization energy parallel the major energy changes rather than because
the crystal field stabilization energy is the determining factor. However, as
has been seen, this reservation is less applicable when one compares a series
of related ions for which all energy factors other than crystal field stabilization
energies would be expected to vary smoothly along the series. In this case,
the deviations of individual members of the series from the pattern may
largely be determined by crystal field stabilization energies.
An example of this use of crystal field stabilization energies is the
following. Just as a table of stabilization energies for weak field octahedral
complexes was obtained above, so one can be obtained for tetrahedral
Site preference energies 1 309
Table 13.2 Crystal field stabilization energies for tetrahedral complexes and comparison with high spin octahedral complexes.
As indicated in the text, all quantities involving Ll. are moduli, absolute values without regard to sign (this means that we
do not need explicitly to take account of the fact that t.,.,
and i\. 0 ,, are of opposite sign, although it is convenient to include
negative signs in this table to correctly indicate the fact that it is stabilization energies which are listed)
d electron Tetrahedral field Crystal field stabilization energies (Octahedral) site

configuration configuration preference energies
Tetrahedral field Octahedral field
do 0 0 0
dl el -3/5 L'itet ~ -4/15 L'i 000 -6/15 L'i 000 2/15 8oct
d' e' -6/5 litot ~ -8/15 L'i 000 -12/15 L'i 000 4/15 L'i000
d3 e 2 ti -4/5 L'itet ~ -16/15 L'i000 -54/45 L'i000 38/45 L'i000
d4 e 2 t~ -2/5 L'itot ~ -8/45 L'i 000 -27,/45 ~oct 19/45 L'i000
d5 e 2 t~ 0 0 0
d6 e 3 t~ -3/5 L'itet ~ -4/15 L'i 000 -6/15 8oct 2/15 L'i 000
d7 e 4 t~ -6/5 L'itet ~ -8/15 L'ioct -12/15 L'ioct 4/15 L'i 000
dg e4 ti -4/5 litet ~ -16/45 L'i 000 -54/45 L'ioct 38/45 L'ioct
dg e 4 t~ -2/5 L'itet ~ -8/45 L'i 000 -27/45 L'i 000 19/45 L'ioct
dlO e 4 t3 0 0 0
complexes. A simplified version is given in Table 13.2 in terms of Lltet and,

remembering that
ILI.tetl "' 4/9 lil.octl
also in terms of IL1octl· The corresponding values for weak field complexes
are also included. It can be seen that the octahedral stabilization energies
are invariably greater than the corresponding values for tetrahedral com-
plexes, unless both are zero. The difference-the (octahedral) site preference
site energy-is given in the final column of Table 13.2. For d 3 and d 8 ions
it is particularly large. This has led to the suggestion that it is the cause of
the relative rarity of tetrahedral complexes of chromium(III), d 3 , and
nickel(II), d 8 . Since this suggestion was first made, many tetrahedral
complexes of nickel(Il) have been prepared (the simplest to make is perhaps
[NiC1 4 ] 2 - which is formed when NiC1 2 is added to molten NaCl) and
examples of tetrahedral chromium(III) complexes have also been claimed,
so that the evidence on which the original suggestion was based has
weakened with time.
To put the problem into perspective, consider the complexes [Cr(H 2 0) 6 ]3+
and [Cr(H 2 0) 4 ]3+. The heat of hydration for the former is approximately
4400 kJ mol- 1 and its crystal field stabilization energy 250 kJ mol- 1 , cal-
culated using the spectroscopic value of L1 (17 400 em - 1 = 208 kJ mol- 1 ).
Assuming that 4400/6 kJ mol- 1 is the energy of a Cr-H 2 0 bond, the
difference in heats of hydration of the two species is 1467 kJ mol- 1 • To this
must be added the crystal field stabilization contribution of 38/45 x 208 =
176kJmol- 1, to give a total heat of hydration of [Cr(H 2 0) 4 ]3+ to
[Cr(H 2 0) 6 ] 3 + of 1643 kJ mol- 1. Of this, about 11% is the contribution from
the crystal field stabilization term. This is a crude, order-of-magnitude
calculation, but it indicates that, whilst the crystal field stabilization term is
not negligible, it is by no means the dominant factor in determining the
relative stabilities of the two species (note that the entropy term, a term
which is also of importance, has been neglected in this calculation). However,

such data may be used to explain why, for example, cobalt(II) forms
tetrahedral complexes more readily than does nickel( II)-the octahedral site
preference energy is much smaller for cobalt(II) than nickei(II).
Another use of site preference energies is their application to the spinels.
Spinel itself is a mineral of composition MgA1 2 0 4 and may be considered
the progenitor of a class of mixed oxides of composition M 2 +M~+o~
which are often referred to as the spinels. The oxide anions are arranged
in a cubic close-packed arrangement. As in all close-packed arrangements,
there are two sorts of hole between the close-packed atoms. These are
tetrahedral and octahedral holes, so-called because they are at the centres
of the contact between four and six anions, sets of contacts which have these
geometrical arrangements. In the close-packed oxygens there are twice as
many tetrahedral holes as there are octahedral holes, and there is one of the
latter for each anion. The holes are large enough to be filled by cations. In
MgO, for example, each octahedral hole in a cubic close-packed arrangement
of oxide anions is filled by a magnesium cation to give a sodium chloride-like
structure. One might reasonably expect a similar arrangement in the spinels,
with the divalent cations occupying octahedral holes, for these are larger
than the tetrahedral holes. In fact, the structures are not as simple as this.
There are two main classes of spinels. In normal spinels, the divalent cations
are in tetrahedral holes-the opposite of that which we expected. Because
filling all of the tetrahedral holes would lead to a structure of general formula
M 2 0 and the divalent ions in a spinel are only M 11 0 4 , the divalent cations
occupy one-eighth of the available tetrahedral holes. In normal spinels the
trivalent cations occupy one-half of the available octahedral holes. The
second class of spinels are the so-called inverted spinels. In the inverted
spinels the divalent cations are where we expected them, in octahedral holes,
having changed places with one half of the trivalent cations. It is found that
spinels with M3+ = Cr3+ are normal and most of those with M2+ = Ni2+
are inverted. Similarly, Fe 3 0 4 ( =Fe 2 +Fe~+o~-) is inverted, but Mn 3 0 4 and
Cr 3 0 4 are normal. The problem is to decide what determines the structure
that is adopted.
First, we must consider the large cation-anion attraction term. This is
more difficult than for an isolated complex ion, for this term, together with
the cation-cation and anion-anion repulsion terms, must be summed over
the whole structure to give the Madelung constants for normal and inverted
spinels. Such calculations have been made and indicate that the two spinel
structures have very similar energies. It is therefore reasonable that site
preference energies should play an important part in determining which
structure is adopted. As Table 13.2 demonstrates, it is at once understandable
that both Cr3+ (d 3 ) and Ni2+ (d 8 ) should be located in octahedral holes,
and their spinels be normal and inverted, respectively, for d 3 and d 8 ions are
those that have the largest octahedral site stabilization energies.
The essential difference between normal and inverted spinels is the
interchange of M2+ and half of the M3+ between octahedral and tetrahedral
sites. If site preference energies are the only important factor, the spinels will
be inverted if the octahedral site preference energy of M2+ is greater than
that ofMH. Consider spinels such as Fe 3 0 4 , where M2+ and M3+ are of the
same element, differing only in charge. As a (crude) approximation, set
Stability constants 1 311
Table 13.3 Site preference energies, used to predict spinel type. In this table it has been assumed that ~(M 3 +) " 3/2 ~(M 2 +)
M>+ M3+
High spin Octahedral site High spin Octahedral site Difference in site Predicted
configuration preference energies configuration preference energies preference energies spinel type
dl, d 6 2/15 ll(M 2 +) do, ds 0 2/15 il(M 2 +) inverted

d2, d7 4;15 il(M 2 +) d 1, d6 2115 ll(M 3 ~) ~ 1/5 ll(M 2 +) 1/15 ll(M 2 +) inverted
d3, d8 38/45 ll(M 2 +) d2, d7 4j15 ll(M 3 +) ~ 2/5 ll(M 2 +) 4/9 ll(M 2 +) inverted
d4 , d9 19/45 ll(M 2 ~) d3, d8 38!45 Ll(M 3 +) ~ 19/15 Ll(M 2 +) -38/45 ll(M 2 +) normal
ds, do 0 d4 , d9 19;45 ll(M 3 +)" 19j30 ll(M 2 +) -19/30 il(M 2 +) normal
Ll.oc~(MJ+) = 3/2 Ll.oc~(M2+ ). Using Table 13.2, the data in Table 13.3 are
derived, in which the type of spinel lattice adopted is also predicted. The
prediction is that Mn 3 0 4 (Mn2+ is d 5 ) is normal and _that Fe 3 0 4 (Fe2+ is
d 6 ) is inverted, both as found. Co 3 0 4 (Co2+ is d 7 ) is normal but is predicted
to be inverted. However, the energy difference between the two forms for d 2
and d 7 is only 1/15 Ll. 0 "(M 2 +), so that the prediction is hardly to be regarded
as reliable, particularly when the variability of t 2 , and eg occupancy for d 7
ion is recalled. For Co 3 0 4 and other cases which do not follow the simple
predictions, more detailed analyses have led to agreement with experiment.
It will be noted that high spin configurations have been used throughout
this discussion. The presence of low spin configurations would introduce the
complication of pairing energies.
13.6 Stability constants

The dependence of the heat terms on the crystal field stabilization energies
discussed in the previous sections will presumably be reflected in the
corresponding free~energy values. Now, free~energy differences are related to
equilibrium constants by equations of the form
~G = -Rtlnk
and so differences in crystal field stabilization energies might well be reflected

in equilibrium constants. Figure 13.4 shows the variation of K 1 for [M(enh]2+
across the first transition series. Here, it is the differences between the crystal
field stabilization energies of [M(cn!JJ2+ and [M(H 2 0) 6 ]2+ which are
relevant. The data seem to show a clear effect, except that the Cu 2 + value
is greater than expected. However, as is well known, copper(Il) complexes
are seldom octahedral--the static Jahn-Teller effect (Section 8.5) is usually
invoked to explain the considerable distortions observed-and so the fact
that copper(II) is out~of~line is not too surprising. It also was omitted from
the discussion in Section 13.2. It is to be noted in passing that Fig. 13.4
provides a partial illustration of the Irving-Williams series (Section 5.4)-the
generalization that stability constants vary in the order
an order that is relatively insensitive to the ligand involved.

'2
Fig. 13.4 Variation of log10 K1 across the first ~
transition series for the reaction ~
[M(H 2 0) 6 ] 2 + +en~ [M(H 2 0) 4 (en)]2 + + 2H 2 0 ~
13.7 Lanthanides
The recognition that crystal field stabilization energies play a role in the
chemistry of the first row transition elements has prompted a search for
similar effects elsewhere in the periodic table. In particular, the question has
been asked 'do similar effects occur for the lanthanides'? It was recognized
in Chapter 11 that crystal field effects are small for the lanthanides and so,
too, therefore must be any consequent stabilization. In Chapter 7 it was seen
that the seven f orbitals split into a 2 • + t 1• + t 2 • sets in an octahedral ligand
field. If we make the-significant-assumption that this is the relevant
symmetry for lanthanide complexes, then we need the relative energies of
these three sets. We actually have them. In Section 7.5 we found that in the
octahedral weak field limit-surely the limit applicable to the lanthanides-
the 3 F term arising from the d 2 configuration splits into 3 A2 • (6/5 ~), 3 T2 •
(1/5 ~)and 3 T 1• ( - 3/5 ~)components. In that so much of ligand field theory
is symmetry-determined, it is not surprising to learn that the energies of these
components are directly proportional to those of the split f orbitals in an
octahedral crystal field. The differences arise from the fact that we are dealing
with seven f orbitals, not five d. This means that g suffixes have to be replaced
by u and that ~ has to be replaced by -~. This latter point is most
readily seen by considering, the fxyz orbital. This orbital, of a2 • symmetry, is
the most stable f orbital in an octahedral crystal field because it points away
from all the ligands. In contrast, for the d 2 configuration it was shown in
Chapter 7 that the 3 A 2 • is the least stable 3 F component. So, the splitting
of the f orbitals in an octahedral ligand field is as given in Fig. 13.5. It follows
that the lanthanide crystal field stabilization energies will be
[ -~n(a2u) - tn(t2ul + ~n(t1ulJ ~

where n(a 2 .) is the number of electrons in the a 2 • orbital, and so on. Feeding
electrons into the orbitals of Fig. 13.5 in a high-spin manner, using the above
equation together with a typical value of 8, leads to the stabilizations of Fig.
13.6. Also in this figure are plotted data for some complexes of the
lanthanides. The 2,2'-bipyridine-dicarboxylate anion shown in Fig. 13.7
forms complexes with the trivalent lanthanides in which three of the anions
Lanthanides 1 313
Free-atom
f orbitals
Fig. 13.5 The relative energies of f orbitals in
an octahedral crystal field.
10
....
X 5
<I
LU
~
u 0
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
fiC. 13.6 (a) Octahedral crystal field
(a) Lanthanide (+3)
stabilization energies of the tnpositive
lanthanide ions. (b) Enthalpy data related to
the formation of (octahedral) complexes of the
ligand shown in Fig. 13.7 (see also the caption
to Rg. 13.2). 25
~
L
0
E
:>1
0
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
(b) Lanthanide (+3)
coordinate to give species such as [Gd(bpydicarbh]H, in which there is,

essentially, octahedral coordination.
The argument just developed is weak in that in the calculation of crystal
field stabilization energies individual f electrons have been allocated to
individual f orbitals. It was seen in Chapter 7 that this allocation could not
always be made for d electrons and d orbitals, and this fact was used in the
discussion in Section 13.2. In Chapter II it was seen how much less it is a
valid step for f electrons, where the effects of electron repulsion and
spin-orbit coupling are much more important than those of the crystal field.
Nonetheless, it seems from Fig. 13.6, and certainly has been argued, that
some very small crystal field stabilization occurs for the lanthanide bipyridine-
dicarboxylates. Given the magnitude of any such stabilization, it is not
surprising that no crystal field modulation of the lanthanide ionic radii was
observed in Fig. 13.1.
Fig. 13.7 The 2,2'-bipyridine-6,6'-
dicarboxylate anion.
13.8 Molecular mechanics

So far in this chapter the concern has been to attempt to understand and
predict relatively small structural or related differences. Although it was
found that crystal field stabilization energies are often of vital importance it
was recognized that this can only be so when the much larger contributions
to the total energy are relatively constant across the. 'Variations under
discussion. In this situation, it is inevitable that attempts should be made to
assess these larger contributions and their variations. In principle this is an
area to which quantum mechanics would be expected to make a major
contribution but, as has become evident in earlier chapters, the application
of quantum methods without major simplifications is a task to be carried
out molecule by molecule and is not likely to be able to address the problems
raised in this chapter for many years. On the other hand, as was asked in
Chapter 10 'why re-invent the wheel'? Specifically, the questions of immediate
concern are those about the way atoms and molecules interact with each
other, how geometries are determined and so on. The problems of core
electrons, often an area in quantum mechanics in which approximations are
made, seem scarcely of relevance. After all, most chemists feel that they
understand why there is a barrier to free rotation about the carbon-carbon
bond in ethane, and do so with no need of detailed quantum mechanics. It
is this philosophy which lies behind the subject of molecular mechanics. Is
it possible to find relatively simple expressions for each of the forces acting
between the atoms of a molecule and, by including all that are believed to
be relevant, to balance them against one another and so make structural
predictions? For instance, most of vibrational spectroscopy is built around
the tenet that bond stretching is a simple-harmonic process. Of course, this
does not cover very large bond length changes, but there one can turn to a
Morse function or something similar-more complicated perhaps, but not
much. Similarly, there are simple expressions for such terms as van der Waals
interactions, repulsive forces between non-bonded atoms, bond angle changes,
rotational barriers (such as that in ethane) and so on. It is not difficult to
see that a computer program can be devised in which the energetic
consequences of a molecule twisting, turning and stretching against itself are
explored. When a reasonably stable arrangement has been found, it can then
itself be varied until the optimum arrangement is obtained. Subsequently,
alternative minima can be explored until there is confidence that all feasible
minima have been investigated and the best then selected. Such methods are
well developed in organic chemistry, where the parameters that arise
have been fine-tuned. As a result, it is now possible to apply molecular
mechanics methods to calculate most organic structures and obtain bond
FUrther reading 1 315
lengths that are within about 0.01 A, and bond angles that are within about
1°, of those observed. It is not surprising that considerable efforts have been
made to extend these calculations to inorganic chemistry, particularly at a
time when, as became evident in Chapter 2, there is heightened interest in
large organic ligands and ligands with interesting steric properties, both of
which are properties which can already be treated rather well. Applied to
inorganic species, molecular mechanics is not yet able to yield the accuracy
found in organic chemistry. Not that it is without success-an error in a
crystal structure determination was discovered when molecular mechanics
predicted a different ligand conformation. Indeed, progress is encouraging
but the transition metals exhibit a variability of geometry not matched in
organic molecules and the prediction ofthis is a major problem. Nonetheless,
in well-defined areas such as the subject matter of Appendix 1, predictions
are usually in good accord with experiment. Similarly, when the outcome is
dominated by the properties of the organic ligand, good results are achieved.
A common problem, however is the occurrence of different predicted
molecular geometries with energy separations which are within the errors
inherent in the calculations. One criticism of many of the models adopted
is that they make no specific allowance for those factors which, earlier in
this chapter, were highlighted as holding the key to the particular geometry
adopted. Attempts to circumvent such problems are beginning to be made,
for example by the union of molecular mechanics and angular overlap
calculations (see Section 10.6).
13.9 Conclusions
In this chapter some rather crude approximations based on crystal/ligand
field theory were used to make a variety of predictions or correlations.
Although no one piece of evidence is really definitive, taken together they
provide support for the general crystal field approach. However, it must be
emphasized that arguments based on crystal field stabilization energies must
always be used with great care and with due regard for other energetic and
en tropic factors involved in the processes considered. Such factors are at the
heart of the application of molecular mechanics calculations to inorganic
systems. These calculations have minimal computer demands compared with
those of their detailed quantum mechanical counterparts and represent a
rapidly growing field in which significant advances are to be expected.
Further reading energies. It is arguable that it is more correct to include them;

however, to do so considerably complicates the discussion. See
An excellent source of much thermochemical data is S. J. S. S. Parmar, J. Chern. Educ. (1981) 58, 1035.
Ashcroft and C. T. Mortimer, Thermochemistry of transition Molecular mechanics is covered in: 'Molecular Mechanics
metal complexes, Academic Press, London and New York 1979, Calculations as a Tool in Coordination Chemistry' R. D.
although, today, computer databases are better if access can be Hancock, Prog. Inorg. Chern. (1989) 37, 187; 'Methods for
gained to one. molecular mechanics modelling of coordination compounds'
The classic reference for the material in all but the last two B. P. Hay, Coord. Chern. Rev. (1993) 126, 177; and a review with
sections of this chapter is P. George and D. S. McClure in Prog. a wide variety of examples 'The relation between ligand
Inorg. Chern. (1959) 1, 381. structures, coordination stereochemistry, and electronic and
The approach to crystal field stabilization energies adopted thermodynamic properties' by P. Comba, Coord. Chern. Rev.
in this chapter is one that bypasses the problem of pairing (1993) 123, I.
Questions 13.3 Give a detailed explanation of the lattice energy data

shown in Fig. 13.3.
13.1 Explain carefully what is meant by 'crystal field stabiliz-
ation energy' and compare this term with the alternative, 13.4 Give an account of the factors influencing the stereo-
'ligand field stabilization energy' used by some textbooks. chemistry of complexes of (a) main group and (b) transition
group elements.
13.2 Write an essay on the structure of Fe 3 0 4 (Section 9.11
should also be consulted).
Reaction kinetics of
14.1 Introduction
In the majority of chapters of this book the concern has been with an
understanding of the properties of individual molecules, properties which are
regarded as essentially time-independent. However, no less important are
the chemical reactions of these molecules and, here, changes as a function
of time are of the essence. This chapter is devoted to a review of our present
understanding of some of the reaction types which are characteristic of
coordination compounds. It is as well to recognize the complexity of the
problem. Suppose we are interested in a reaction such as the aquation of an
ion such as [Co(NH 3 ) 5 Cl]2+, a much studied system:
H2 0 + [Co(NH 3 ) 5 CI]2+ -+ [Co(NH 3 ) 5 H2 0] 3 + + Cl -

Because [Co(NH 3 ) 5 Cl]2+ salts are soluble in few other solvents, the study
will probably be carried out with water as a solvent, as well as reactant. In
this case, each [Co(NH 3 ) 5 Cl]2+ ion in solution will undergo a constant
battering from water molecules incident from all directions. No doubt, the
coordinated ammonia molecules will move in response, although perhaps
restrained somewhat by hydrogen bonding to the solvent. For the reaction
to occur, a Co-Cl bond has to be broken and this could be because
vibrations of the coordinated ammonia ligands temporarily expose the
Co-Cl bonding electrons to attack by a water molecule. Equally, it could be
that a transient, strong hydrogen bond is formed between the Clligand and
a water molecule, so facilitating breaking of the Co-Cl bond. Again, Co-Cl
bond breaking could be dependent on much of the vibrational energy within
the molecule localizing itself briefly in the Co-Cl bond stretching mode.
Alternatively, some combination of all three factors could be involved. It
seems likely that a multitude of slightly different reaction pathways exist
and, since we study a large number of molecules simultaneously, all we can
318 1 Reaction kinetics of coordination compounds
measure is some sort of average. Hopefully, there will be only one, clearly
defined, lowest energy reaction route and the average will be dominated by
this and its minor deviants. However, the three contributions listed above
(and others could be added) will presumably have different temperature
characteristics and so the (average) reaction route will also be slightly
temperature dependent. Fortunately, this complication can usually be ignored
-it seems to be of lesser importance than experimental error. Attention
therefore focuses on a single reaction pathway and, in particular, the different
potential energy profiles associated with alternative pathways. The task of
the worker in the area is to use the experimental data, and, in particular,
rate laws, to deduce the most probable reaction pathway. It is not an
easy task. As will be seen, the existence of pre-equilibria (which mean that
the real reactant species is present in much lower concentration than
expected), the involvement of the solvent (which, because its concentration
scarcely changes, will not be evident from the rate law), the ability to work
over limited temperature and concentration ranges, all impose problems.
Fortunately, more techniques are becoming available; thus, the ability to
study reaction rates in solution as a function of the external pressure applied
to the solution offers information as to whether the reaction pathway
involves compression or extension of the reactant species and, thus, insights
into its molecularity.
Only in one area can theory be said to have led experiment. This is in one
aspect of oxidation-reduction, electron-transfer, reactions, a topic to which
we shall return. First, however, we shall give an overview of the very
important topic of reaction rates. Perhaps the most evident thing about them
is their enormous range. They (or, perhaps easier to think of, the time to
half-completion of a reaction) span a range of at least 10 15 • Clearly, a
theory is to be regarded as acceptable if it succeeds in correctly predicting-
or interpreting-an order of magnitude, exact agreement is too much to
hope for. The range of values emphasizes one thing. For a reaction to proceed
it must be thermodynamically feasible. The range of rate and t 112 values
can arise from variations in mechanism-the availability of a facile mechanism
or the absence of one (thermodynamically, CCI 4 should react violently with
water)-but a given mechanism can be associated with very different rates
also. 1 Ultimately, the hope is that the mechanism can be related to the
electronic properties of the molecule(s) under study and relative rates thus
explained. As will be seen in this chapter, considerable progress has been
made, but much remains to be done.
In Section 2.1 the distinction between inert and labile complexes was
encountered. Several attempts have been made to formalize this distinction,
of which the most popular seems to be Taube's definition: 'if no delay is
noted in the substitution reaction under ordinary conditions (i.e. room
temperature, ca. 0.1 M solutions) the system will be described as labile'.
However, for most chemists inert complexes are effectively those for which
their reactions may be studied by classical techniques, such as monitoring
the change in intensity of a visible or ultraviolet spectral peak with time.
Such reactions are half complete in about one minute or longer at 25 oc for
1 So, the rates of substitution of Ni 11 and Co111 are very different (fast and slow, respectively)
but the mechanisms involved are probably rather similar.
Introduction 1 319
concentration of ca. 0.1 M. In the past, labile complexes were not so readily
studied but current techniques have provided a wealth of data on them.
Reactions of inert complexes are usually studied by mixing solutions of the
reactants and monitoring either the appearance of a product species or
disappearance of a reactant. It is difficult to mix two solutions completely
in less than a millisecond (and only then in small volumes and in some sort
of flow system) and this limits the extent to which technical ingenuity may
be used to apply classical techniques to fast reactions. Some further
extension, enabling the study of some not-too-labile complexes, is possible
by working at low temperatures with very dilute solutions, but for very fast
reactions quite different methods must be used. The simplest has already
been mentioned (Section 5.3): if a stepwise equilibrium constant is known,
together with the rate of either the forward or backward reaction, then
enough data are available to enable the unknown rate to be determined.
Other methods study an equilibrium as the following two examples illustrate.
Suppose that a diamagnetic complex containing coordinated trimethyl-
amine is dissolved in trimethylamine and the proton magnetic resonance
spectrum of the solution studied. If the coordinated trimethylamine exchanges
with the solvent slowly, then two resonances will be observed (provided
that complicating features are absent), one due to coordinated and the other
due to free trimethylamine. If the exchange is rapid, only one resonance will
be observed, at some sort of average position. In favourable cases, if the
temperature of the sample is varied, at one temperature slow, and at another,
fast exchange will be observed (when it will be found that the position of
Fill. 14.1 Dynamic exchange in NMR. the single resonance is a concentration, i.e. peak-area-weighted, average of
The top spectrum is the low-temperature, those of the two) as will be the intermediate region, in which broad peaks
no exchange, limit The bottom is the
high-temperature, rapid exchange, limit The occur (Fig. 14.1). From such measurements the rate of exchange at the
central spectrum was recorded at ca. 200 'C. temperature at which intermediate behaviour is observed may be obtained.
Those immediately below and above were Strictly speaking, the terms slow and rapid in this example are relative to
recorded at ca. 190 'C and 210 'C,
respectively. This particular example arises from the NMR timescale and this depends on the separation between the peaks
two proton environments within a molecule due to coordinated and solvent trimethylamine. If the complex is para-
becoming rotationally equivalent with increase magnetic, additional line-broadening and shifts occur. These may be analysed
in temperature. For the case described in the
text the two low-temperature peaks would be to give rate data but the results tend to be somewhat ambiguous, since it is
of different areas and the high temperature at the slower of two processes which is measured. These are the exchange
the area-weighted average position. processes of interest and that of a paramagnetically induced change in the
nuclear spin state of the protons (for proton magnetic resonance) of the
coordinated ligand. Measurements with a different but similar ligand may
establish which process is the slower, but in such cases one often has to be
content with only being able to put a limit on the rate of the exchange process.
A most important approach to the study of the fast reactions of labile
complexes involves relaxation phenomena. The position of dynamic equi-
librium in a system depends not only on reactant concentrations but also
on such quantities as temperature, pressure and an even electric field gradient
(if present). If one of these is suddenly changed the position of chemical
equilibrium will also change slightly and, with rapid response and sensitive
instruments, this change may be detected. The speed with which the new
equilibrium is reached depends on reaction rates which may thus be
measured. In this way Eigen and his co-workers have determined unimolecular
rate constants greater than I 09 s- 1 , which is approaching the limit for
diffusion-controlled reactions.
14.2 Electron-transfer reactions

The simplest class of reaction of coordination compounds which has been
studied is that of electron-transfer reactions. Suppose that a solution of
potassium ferrocyanide (hexacyanoferrate(II)) is mixed with one of potassium
ferricyanide (hexacyanoferrate(III)), then if an [Fe(CN) 6 ] 4 - anion loses an
electron and an [Fe(CN) 6 ]3+ anion gains one, a chemical reaction has
occurred, although there is no change in the composition of the mixture. If
one of the atoms in just one of the complex ions is labelled in some
way-with 14C for example-then the reaction may be studied by seeing
how quickly it appears in the other complex ion. In this particular case the
reaction is fast, the second-order rate constant being ca. I 0 3 M- 1 s- 1 at
25 oc. This rate is much greater than those of reactions involving ligand
exchange of either species, so a simple electron-transfer mechanism is
indicated. Because there is no net chemical reaction there is no heat change
associated with the electron transfer. The Fe-C bond length in [Fe(CN) 6] 3 -
is very slightly shorter than that in [Fe(CN) 6 ] 4 -, so, if an electron were to be
transferred between the anions in their ground-state equilibrium configur-
ations, by the Franck-Condon principle, the product [Fe(CN) 6 ] 3 -, would
be expanded and the [Fe(CN) 6 ] 4 - compressed relative to these configur-
ations. That is, the products would be of higher energy than the reactants,
contradicting the requirement of zero heat change. It follows that an electron-
exchange reaction will only occur between precisely matched molecules
which are vibrationally excited. It should be noted, however, that transfer
would not occur between [Fe(CN) 6 ] 4 - and [Fe(CN) 6 ] 3 - anions, which
are, respectively, compressed and expanded into each other's equilibrium
geometry because this would mean that two vibrationally excited molecules
become two vibrational ground-state molecules. It is only totally symmetric
vibrations-breathing modes-which lead to the synchronous contraction
or expansion of all symmetry-related bonds in a molecule and these are the
modes which have to be excited to achieve the required matching (Fig. 14.2).
It is not essential that two matched anions are in contact at the instant of
electron transfer. The question which then arises is just how far apart can
two ions be and yet participate in an electron-transfer reaction? The answer
to this question is particularly important in bioinorganic chemistry (Chapter
16), for there two potential participants may be unable to approach each
other very closely because ofthe constraints imposed by the large molecule(s)
within which they are found. For most systems the answer to the above
question seems to be less than 10 A but it is becoming clear that much
Fig. 14.2 For simplicity, this figure shows 1- L

L--Mn- ---------M'~'--------
two centrosymmetric ML, molecules. The
M111-L (top left) is drawn much shorter than the
M 11- (top right). By appropriate totally symmetric
(breathing) M-L extensions or contractions the
two molecules become of identical size - L - - - M1-1 - - - L-
(centre). The totally symmetric mode can
involve just one molecule but then the
amplitude is greater (bottom). Evidently, a
continuous range of matching bond lengths
exists. L----.--Mn_I__ .___ L - - - - - - M 1!...
1 -----L
Electron-transfer reactions 1 321
larger separations are possible. At the time of writing the record seems to
be ca. 40 A. Such long-distance electron-exchange reactions are often referred
to as occurring by an outer-sphere mechanism, that is, as not involving the
immediate coordination sphere of either metal. However, it is reasonable to
expect that electron transfer will occur most readily when the two reacting
species are relatively close together. That this is so is indicated by the
observation that outer-sphere electron-transfer reactions are more rapid for
complexes containing ligands such as o-phenanthroline and the cyanide
anion than for corresponding complexes with ligands such as H 2 0 or NH 3 .
That is, a ligand over which a metal electron may be extensively delocalized
(cf. Section 12.8) significantly reduces the magnitude of the barrier to electron
transfer (one may draw an analogy with a current flowing through a piece
of resistance wire: replacing part of the resistance wire by a piece of copper
wire increases the current).
In the example discussed above, the electron configuration of the iron
atom in the [Fe(CN) 6 ] 4 - anion is tL
and that in the.[Fe(CN) 6 ] 3 - anion
is~~ •. Removal of an iron electron from [Fe(CN) 6 ] 4 - leaves a~~. configur-
ation and addition of one to [Fe(CN) 6] 3 - gives a~~. configuration. It is not
always as simple as this. Consider an electron-transfer reaction between a
molecule of a cobalt(III) complex and one of cobalt(II), both octahedrally
coordinated. Cobalt(III) complexes are usually low spin and cobalt(II) are
high spin, so the electron configuration of the cobalt ions will be cobalt(III)
~~ •• and cobalt(II) t~ 9 e;. After transfer of an electron from cobalt(II) to
cobalt(III) these configurations will presumably become t~ 9 e 9 and t~ 9 e 9
(cobalt(II) and cobalt(III), respectively). However, these are not the ground-
state configurations of the ions. That is, after the electron-transfer reaction
both complexes will be electronically excited (this excess of energy will
rapidly be lost either by radiation or, more probably, it will be converted
into thermal energy). Because this electronic energy contributes to the
activation energy of the process, the rate of electron-transfer reactions
between cobalt(II) and cobalt(III) complexes is much slower than that
between the [Fe(CN) 6] 3 - and [Fe(CN) 6] 4 - anions.
There is yet another reason for this. The original cobalt(III) ion in its
normal octahedral ground state would be 1 A 19 • It becomes an octahedral
cobalt(II) ion with a t~ 9 e 9 configuration. It therefore has a 2 E• term (Table
7.5), but the ground state of octahedral cobalt(II) is 4 T 19 (derived from the
~~.e; configuration). That is, in passing from the as-formed cobalt(II) to
ground-state cobalt(II) the spin multiplicity must change. The only available
mechanism is spin-orbit coupling, operating in a way similar to that shown
in Fig. 8.16. Reference back to the text associated with that figure will show
that such spin-multiplicity changes require special, and rare, conditions if
they are to be facile. It is not surprising that an exchange such as
[Co(NH 3 ) 6 ] 3 + + [Co(NH 3 )a] 2 + -+ [Co(NH 3 ) 6 ] 2 + + [Co(NH 3 ) 6 ] 3 +
should be extremely slow, with time for half-reaction measured in hours. The
reader may find it helpful to reconsider the use of charcoal in the preparation
of cobalt( III) complexes (Section 4.2.2) in the light of the above discussion.
At first sight, an attempt to calculate rate constants for electron transfers
of the sort we have been discussing would seem an impossible task. Charged
species, separated by varying amounts of solvent, a solvent which will react
to the sudden change of charge on each of the two ions-and a proper

description of this solvent polarization change proves to be a key component
to a successful theory-the need for there to be a matching of vibrational
energy levels of the two complex ions, all have to be considered. In fact, the
problem is approximately soluble. This is, in part, because so many factors
are involved and, so, fairly crude approximations for each will work
reasonably well. Getting the overall picture approximately correct is more
important than individual accuracy. Further, more than one set of approxi-
mations lead to the same general result. It is clear that a potential energy
surface has to be modelled because vibrations are important. One model,
then, describes the electron transfer as resulting from a coupling of two
potential energy surfaces. If there is no coupling then there is no electron
transfer. The problem of a mathematical description of the process thus
becomes one of a description of a coupling between two potential energy
surfaces. Such a theory contains many parameters but, fortunately, these
largely cancel out when a series of closely related reactions is studied, so
that much of the literature is concerned with reaction series. Inevitably, the
resulting theory is somewhat mathematical but it seems that the Marcus-
Hush theory (the name coming from the two workers who arrived at
essentially the same result following somewhat different routes) correctly
predicts to within an order of magnitude electron-transfer rates of the type
so far considered. This is not at all bad when the rates studied vary by 10 14 .
But there is more. The type of electron-transfer reaction which has so far
been covered is that in which two species differ by one electron-[Fe(CNJ 6 r-
and [Fe(CN) 6 ] 4 - and the pair [Co(NH 3 ) 6 ]3+ and [Co(NH 3 ) 6 ]2+, for
example. In such reactions the concentration of neither species changes with
time (so that some care has to be taken in defining the rate of reaction; in
practice the first-order rate for oxidized -> reduced, or its reverse, is commonly
reported). However, if the reaction is between two such pairs and the oxidized
form of one pair is reacted with the reduced of another, then concentrations
will change with time, until equilibrium is reached. It turns out that
Marcus-Hush theory can make quite good predictions about the rate of such
so-called cross-reactions. Consider the two reactions, each having rate
constants kaa M - 1 s - 1 (a = 1 or 2) as defined above:
and
Then, one of the cross-reactions is
The equilibrium constant for this last reaction, obtained from emf measure-
ments, is K 12 . For reactants and products of the same size and charge type
the simplest form of the Marcus cross-relationship is
Electron-transfer reactions 1 323
(If the size and charge requirements are not met, a further factor, / 12, appears
within the bracket on the right hand side.)2 For the above reactions,
k 11 = 4 x 103, k 22 = 1.2 x 10- 2 and K 12 = 1.07 x 10 6 . It follows that the
calculated value of k 12 is 7.2 x 10 3 . This is to be compared with the
Fig. 14.3 The ligand pyrazine. experimental value of 4.2 x 103 • In general, the equation gives values correct
to within one or two orders of magnitude. When a self-exchange or
cross-reaction does not obey the Marcus-Hush predictions, at least approx-
imately, it is a reasonably reliable indication that some complicating
feature is involved-a stabilization caused by hydrogen bonding between
ligands in a cross-reaction, for instance.
So far, the electron-transfer reactions which have been discussed are those
in which only the formal valence states of the metal ions involved changes,
reactions which can occur when there is no intimate contact between the
two reactants (although this absence of contact is a point which it may be
difficult to prove). There is another class of electron-transfer reaction, in
which intimate contact between the two reacting molecules leads to reaction
by a bridge or inner-sphere mechanism. A class of compound which has been
much studied in this connection is one based on a species first studied by
r
Creutz and Taube, [(NH 3 ) 5 Ru(pyz)Ru(NH 3 ) 5 +,in which the two reacting
centres (one is Ru11 and one is Ru 111 are linked by a bridging ligand, in this case
pyrazine (Fig. 14.3) although variation in choice of bridging ligand is
common. Inner-sphere mechanisms often have indicators which reveal their
presence. Reactions of the type
chromium(ll) + cobalt(lll) .... chromium(lll) + cobalt(ll)
have been extensively studied, again, notably by Taube and his co-workers.
Cobalt(III) and chromium(III) form inert complexes whilst the corresponding
divalent ions give labile complexes. This means that if a ligand is transferred
from cobalt(III) to chromium(III) in the reaction it will be possible to show
that this transfer has occurred.
Consider the reaction:
[Co(NH 3 ) 5 CI] 2 + + [Cr(H 2 0) 6 ] 2 +-+ [Co(NH 3 ) 5 (H 2 0)] 2 + + [Cr(H 2 0) 5 Cif+
The reaction is carried out in water and the final cobalt(II) product is actually
[Co(H 2 0) 6 ] 2 +, but this is immaterial to our discussion. It was found that if
e
the solution contained labelled chloride ion 6 Cl-) none of the activity
appeared in the [Cr(H 2 0) 5 Cl]2+ product; the reaction does not involve the
free chloride ions present in the solution. This observation indicates that
there must be an intimate contact between the reacting species, a Cl- ion
being transferred from cobalt(III) to chromium(II) and an electron migrating
in the opposite direction. It therefore seems likely that a species something
similar to that shown in Fig. 14.4 must be involved. Similar transfer of the
ligand X from [Co(NH 3 ) 5 X]"+ to chromium(II) occurs for X= Cl-, Br- ,
N 3, acetate, SO~- and PO!-. That there is a transient intermediate
something like that in Fig. 14.4 is supported by the observation that for
X= Ncs- (the complex having a Co-N bond) the initial product is
2 / 12 is a function of k 11 , k22 , K 12 and a collision frequency which itself is usually taken as
a constant. In practice, therefore, the presence of J; 2 causes no difficulty but it is simplest to

omit it in the present discussion.
4+
FJg. 14.4 The essentials of the intermediate

species formed during the transfer of c;- ion
from cobalt(lll) to chromium(ll).
[Cr(NH 3 ) 5 SCN]2+, with a Cr-S bond, which subsequently rearranges to

[Cr(NH 3 ) 5 NCS] 2 + (with a Cr-N bond).
Further supporting evidence for the suggested mechanism comes from the
observation of fairly stable intermediate species in some inner sphere
electron-transfer reactions. For example, reaction between vanadium(II) and
vanadyl (VQ2+) complexes, to form vanadium(III) species, rapidly give a
brown intermediate, believed to contain a V-0-V bridge which, relatively
slowly, gives the final product. In some cases there is evidence that there
are two or even three bridges in the intermediate (for example, in some
inner-sphere electron-transfer reactions between chromium(II) and chro-
mium(III) complexes containing the azide anion as ligand). An example of
the ingenious methods that have been used to probe such reactions is
provided by the used of an optically active inert species, such as a resolved
isomer of [Co(en)(oxlzr, reacting with a labile species which consists of
two enantiomers, such as [Co(en)J]2+; and determining to what extent the
inert product [Co(enJJP+ is optically active-an activity which must have
been acquired along with the electron transfer.
In the majority of cases there is a transfer of a (bridging) ligand
concomitant with electron transfer. Examples are known, however, where
although there is strong evidence for a bridge intermediate in the electron-
transfer process, no ligand transfer occurs. Although not developed for it
nor obviously adapted to it, there is evidence that the Marcus-Hush theory
may be applicable to inner-sphere electron transfer. Perhaps this serves to
emphasize the fact that a continuum of electron transfer processes exists,
although it is convenient to divide them into two main categories. The same
point is evident in another class of compounds which has been studied,
those in which both metal atoms participating in an electron transfer reaction
are contained within a single complex ion, being linked by a bridging ligand.
Undoubtedly the most important series of compounds which has been
studied are species derived from the Creutz-Taube ion, mentioned above,
by choosing bridging ligands related to, but different from, the pyrazine
of the parent species. Another example is provided by the molecule shown
in Fig. 14.5 which contains one iron(II) and one iron(III). Such complexes
can be made by partial oxidation or reduction of a symmetrical species. They
can be studied by a variety of methods, most notably by electronic
spectroscopy where an additional metal-to-metal charge transfer band is
observed, the frequency and intensity of which turn out to be related to the
5-
FJg. 14.5 Complex anion containing both
iron(ll) and iron (Ill).
(CN)5Fe-NQ>--<Q>-Fe(CN)5
Mechanisms of ligand substitution reactions: general considerations I 325
rate constant of electron transfer. Clearly. the Robins-Day classification

(Section 8.11) is relevant here. When both metal ions are not identical it is
sometimes possible to prepare a molecule with one metal ion in the oxidized
form and then measure the slow transfer of an electron from what initially
is the reduced metal to that which initially is the oxidized.
14.3 Mechanisms of ligand substitution

reactions: general considerations
In this section reactions of coordination compounds in which one ligand
is replaced by another will be discussed, i.e. ligand substitution reactions. It is
convenient to start by a consideration of a general potential energy profile,
a topic which, whilst relevant to the electron transfer reactions of the previous
section, has been deferred until the present because of its importance to it.
In the introduction it was seen that it is convenient to make the approximation
that all molecules undergoing a particular reaction follow identical energy
pathways. As shown pictorially in Fig. 14.6, in the simplest case a potential
energy barrier separates reactants from products, along the pathway of
atomic movements that leads the reactants to become products (the so-called
reaction coordinate). The rate of reaction is determined in part by the height
of this barrier (another factor is the accessibility of the reaction coordinate-
this is where energy-matching requirements, such as those met in the last
section, come in). The equilibrium position is in part determined by the
relative heights of the reactant and product potential energy minima. The
'in part' in the last sentence arises because the equilibrium position is
determined by the relative free energies, and so there is also an entropic
contribution. It is often found, both in organic and in inorganic chemistry,
that there is a proportionality between the free energy 'hill' and the free
energy difference between products and reactants. The existence of such
proportionalities gives rise to the study of linear free-energy relationships.
Potential
energy
barrier
I
Fig. 14.6 A simple potential energy diagram

relating reactants and products. The reaction ~Q)
c:
coordinate can involve quite a complicated set Q)
of atomic motions, of which minor van ants

are admissible. It follows that this diagram is "'"'c:
Q)
just one of a large family, of which all the 0
0..
acceptable variants are members.
Products
Reaction coordinate -----....

Transition
states
RC. 14.7 A more realistic potential energy Successor

diagram than that shown in Fig. 14.6. complex
An additional complication may be that the
reaction coordinate may well not be the
/__
smooth, continuous, motion implied by this
diagram. So, for example, the motions that lead
from reactants to reaction intermediate may be
rather different from those involved in the
passage from reaction intermediate to
products.
Products
Reaction coordinate -
A little consideration soon shows that Fig. 14.6 is likely to be somewhat

idealized. It implies that the path from products to reactants is smooth and
uneventful. There are no sticky spots. But consider the reaction between a
cationic complex, [Ni(H 2 0) 6 ] 2 + for example, and an anion, Cl- say, to
give the complex ion [Ni(H 2 0)sCl] +. Simple electrostatic considerations
suggest that the ion pair [Ni(H 2 0) 6 ] 2 + ·Cl- may well have some stability,
particularly in solvents with a relatively low dielectric constant. The reactants
are 'sticky'; the ion pair is likely to exist until atomic positions and momenta
are either such as to allow the reaction to proceed or the ion pair to
dissociate. Such an intermediate, formed between the reactants but in
advance of reaction between them, is called a precursor complex. Similarly,
when the reaction is one which involves loss of an anionic species by the
complex, a so-called successor complex may be an intermediate on the way
to the final product. This pattern is shown in Fig. 14.7 which also includes
the possibility of a reaction intermediate of some stability. Fig. 14.7 shows
a situation which is complicated and would therefore have a complicated
rate law. Most systems studied either are, or are assumed to be, rather
simpler.
Despite its complexity, Fig. 14.7 begs one important question. Consider
either the reaction intermediate in that figure or, when no such intermediate
is formed, the corresponding transition state (shown dotted in Fig. 14.7).
Compared to the coordination number of the metal in the reactant complex,
at this point in the potential energy surface the coordination number of the
metal could have increased by one, decreased by one or stayed the same.
The rate-determining step in the reaction mechanism involves association,
dissociation or an interchange, respectively, and these labels, A, D and I,
are used to describe the reaction type. The job of classifying a particular
Mechanisms of ligand substitution reactions: general considerations 1 327
reaction is an experimental one. While it is easy to provide experimental

criteria which enable such a distinction they refer to ideal cases and real life
examples can be more complicated. Thus, there are cases where it seems that
A and D mechanisms operate in parallel.
The distinction between an I mechanism and either A or D is that in the
latter there should be proof of a reaction intermediate. In a D mechanism
there is no dependence of the reaction rate on the incoming group. Similarly,
in an A mechanism there is no dependence of the rate on the leaving group.
These criteria require the ability to study a closely related series of reactions
which follow similar reaction pathways. Of course, one cannot know in
advance that all members of the series will behave as expected and so if one
or two behave differently, a problem arises that can only be resolved by
appeal to other data. Fortunately, the most readily available additional
data refer to individual reactions and to the properties of the transition
state. These are the entropy of activation, f).S*, obtained from the intercept
of a log k against 1/T plot (T = absolute temperature), and the volume
of activation, f).V *, obtained from the slope of a log k against P plot
(a In k/8P = -f).V* /RT, where P =pressure). Of these, f).V*, although the
more difficult to obtain, is perhaps the more reliable. Negative values of
f).V *-the rate increases with increasing pressure-implies an A mechanism,
positive values aD mechanism. In the case of solvent exchange, for a limiting
A mechanism the f).V * value equals the partial molar volume of the solvent
and for a limiting D, it is negative. Smaller, positive, values indicate an Id
(d =dissociation) mechanism of solvent exchange; smaller and negative
values an I. (a = association) mechanism.
Type D mechanisms imply an increase in the effective number of particles
in the system, an increase in the number of possible arrangements, and so
f).S* is positive. Conversely, negative f).S* values imply A mechanisms. Type
I mechanisms are associated with very small magnitudes of f).S*. However,
the view just adopted is somewhat simplistic in that there are many
contributions to the f).S* -translational, rotational, vibrational-of the
reactants, together with a contribution from solvent ordering, and it is their
sum which is determined. For this reason, and because a long graphical
extrapolation is needed to obtain them, f).S* values are perhaps less useful
than f).V *. However, they are easier to obtain.
This far from exhausts the list of ways of gaining insight into reaction
mechanism by kinetic studies-changes in the nature of the solvent and
solution composition are obvious additional variables. Less evident are
techniques such as carrying out measurements in the intense magnetic field
available from a superconducting coil. Such experiments have revealed that
a reaction intermediate can have quite different magnetic properties from
either reactants or products. As a result, a magnetic field, which interacts
with any unpaired electron present, can affect a reaction rate. This is
particularly relevant to electron-transfer reactions-for at least one reactant
and one product species must be paramagnetic.
It is perhaps not surprising that f).S* (and f).V * where available) values
quite often indicate that there are I mechanisms with a rate dependence on
either entering or leaving ligands. That is, I mechanisms which have some A
character and others which have some D. The I class is therefore usually
subdivided into I. and Id, labels we have already met. Although a constant
research interest is where a particular reaction falls in the sequence:

A -+ 10 -+ ld -+ D
and we shall certainly need to refer to the topic, 3 it will not be a major
concern in this chapter. Rather, we shall look at two particular, well
researched, areas in some detail. The first, the reactions of square-planar
platinum(II) complexes, provides an excellent example of a set of reactions
which are essentially A-type. The second, the reactions of octahedral
cobalt(III) complexes, provide examples of reactions which are in the Id to
D region. Two words of caution. First, it should not be thought that in all
respects they are representative of their type-that all square planar
complexes behave like platinum(II) and octahedral like cobalt(III). For
instance, gold(lll) forms square planar complexes but can also be reduced
to gold(!) by the iodide anion, for example. Not surprisingly, this opens up
reaction pathways and some of the reactions of gold( III) have no parallel
in the chemistry of platinum(ll). This apart, there seems to be a similarity
between the kinetic patterns of square planar complexes of Pt11 , Pd 11 and
Au111 -although rates are much greater for the last two. The same cannot
be said of square planar complexes of nickel(ll), because nickel(II) occurs
in a wide range of coordination numbers and geometries, features which
set it apart. In general, however, the only safe viewpoint seems to be that
'where reactions of complexes of two different elements are superficially-
stoichiometrically- similar on paper this is no guarantee of similar kinetic
patterns'. This is also seen in the second word of caution. Most reactions
have been studied with water as solvent. As will be seen, molecules of the
solvent are seldom disinterested spectators of a reaction-they commonly
get involved, unless they are both nonpolar and without features, such as
carbon-carbon double bonds or donor oxygen atoms, which can function
as ligands. Changing solvents in a way which reduces solvent participation
commonly reveals features of the kinetics concealed when less innocent
solvents are used. So, much of that which follows is a simplification. We
shall often ignore a back-reaction, for example. Yet in the right situation
such a reaction may become of importance. Quite a few workers find a real
challenge in engineering molecule and solvent in such a way as to create the
'right situation', not only for this but also to force a D mechanism in
platinum(II) chemistry, for example. However, the resulting exotic species
are difficult to fit comfortably into a relatively brief account such as this.
14.4 Substitution reactions of square

planar complexes
As has just been said above, there seems no doubt that the substitution
reactions of square planar complexes, particularly of platinum(II) (the
most-studied set of complexes) are A, associative, in type, the substitution
being by a nucleophile. However, even for platinum(II), in rare and unusual
3 For instance, as will be seen later in the text, there is general agreement that substitution
reactions of pentaammine-Co 111 complexes fo11ow mechanisms which are mostly Id with a
leaning towards D; however, there is not the same agreement over the corresponding reactions
of, say [Cr(NH 3 ) 5 H 2 0] 3 + for which I., borderline I,/I, and I, have all been proposed. The
most recent evidence favours the latter.
Substitution reactions of square planar complexes 1 329
systems, electrophilic A-type substitution and even dissociation-controlled

pathways can occur. Dominance of A-type nucleophilic substitution is what
one would expect, for in a square planar complex there is no large steric
constraint opposing bonding of the incoming ligand whilst the outgoing
ligand is still attached. In practice, the rate law most generally observed for
substitution reactions of platinum(II) complexes for the reaction:
[Ptlal(] + Y -+ [Ptl3Y] + X
is
Here, for simplicity, the ligands not involved in the substitution have been
represented by L 3 ; however, they need not all be identical. Although in a
particular study only one term of the rate law may be important, changing
the reaction conditions somewhat can lead to the other becoming evident.
The use of polar solvents leads to dominance of the k 1 term, of apolar
solvents to the k 2 • Conversely, when Y is a strong nucleophile the k 2 term
is favoured; weak nucleophiles favour the k 1 • As the ligands L 3 are made
more and more bulky, so both k 1 and k2 are reduced in magnitude. Similarly,
when Y is made more bulky both k 1 and k2 are again reduced. Although,
from the rate expression alone, one might guess that the k 1 term represented
a D mechanism, the evidence just presented strongly suggests that the k 1
and k2 mechanisms are very similar. The detailed data are consistent with
very similar mechanisms indeed; that in k2 Y, and in k1 solvent, are involved
in the rate-determining A step.
In this reaction sequence, no back-reactions are included. With varying

degrees of ingenuity they can be made to be important (for example, by
carrying out the reaction in the presence of a large excess of X). The kinetics,
accordingly, become more complicated.
An incompletely understood aspect of these reactions is that they proceed
with a remarkably high retention of configuration. This is surprising. An A
mechanism implies a five-coordinate intermediate and so a, basically, square
pyramidal or trigonal bipyramidal geometry is likely. The energy difference
between the two geometries in most five-coordinate systems is found to be
small and the systems fluxional. 4 If this were the case here, retention of
configuration would be rare. However, as has been seen in Section 4.2.7,
the retention is so reliable that it is given a name and offers the possibility
of planned inorganic syntheses. As Section 4.2.7 illustrates, not only is
configuration retained but, given a choice, the ligand displaced is determined
not so much by its own nature as that of the ligand trans to it-the
4 Fluxional ~ stereochemically non-rigid; see Section 14.8.
so-called trans effect. Because the trans effect is kinetic-that is, it is a

generalization of the relative rates of the various alternative substitution
reactions-explanations of the effect which do not take account of the A
nature of the processes involved must be incomplete. As seen briefly in
Section 4.2.7, the current popular theology of the trans effect distinguishes
two contributions. The first is a static, ground state, effect, the trans influence.
It is seen, for example, as metal-ligand bond lengths being sensitive to the
trans ligand. The simplest explanation is to note that, if there is competition
for electron density along a common (L-Pt-L') axis, what one ligand gains,
the other loses. This model offers an explanation for the large trans effect of
the H- ligand, for which the second contribution, from n-bonding, is scarcely
likely to be relevant. Then bonding model suggests that a major contribution
to the trans effect comes from preferential stabilization of a particular
trigonal bipyramidal five-coordinate species by n bonding from the trans
ligand. The high trans effect of many ligands is thus explained and, properly,
the model is a kinetic one, focusing on the five-coordinate species. Even so,
the model is not without its problems. In particular, crystal structures of
genuine five-coordinate platinum(II) complexes tend to show them to be
square pyramidal rather than trigonal bipyramidal, although the model
requires the latter. This is because in the n bonding model the trans group,
the entering (Y) and leaving group (X), occupy the trigonal planar portions
of the trigonal bipyramid. So, for the three ligands in this plane, the
out-of-plane n orbitals of the trans ligand have a direct interaction with the
n orbitals of both X and Y (and only X and Y). There is no comparable
pattern possible in a square pyramidal complex.
A particularly unsatisfactory feature of the two models for the trans effect
is that there is no way of combining them. They are introduced to the extent
that they are needed to explain the experimental results. There are other,
minority view, explanations of the trans effect and it is unlikely that
the last word has been said. For instance, as we have mentioned, the effects
of strong magnetic fields have shown that the reaction intermediates in
electron-transfer reactions (Section 14.2) generally have magnetic properties
rather different from those of both paramagnetic reactant and product
species. Perhaps here, too, it could be that in going to the reaction inter-
mediate, the distortion mixes higher electronic states into the ground state,
thus introducing the possibility, for example, of significant relativistic
effects, effects which are known to be generally large for platinum. 5 One
final word, on [Pt(NH 3 ).CI 4 _.Jr•- 2 >+ species. For these, systems in which
the trans effect operates, it has been found that, when isomers exist, the
isomer formed most rapidly is the thermodynamically unstable one. That is,
for these systems at least, synthetic application of the trans effect depends
on a careful choice of time at which the products are separated. Leave the
reaction mixture too long and it appears that the trans effect is no more!
The relevance of such observations-which are some 30 years old-to the
explanations offered for the trans effect awaits full discussion, although it
has been recognized that successive displacements can lead to isomerization,
as shown in Fig. 14.8. A relevant recent discovery is that the cyanide anion
5 Note that a (relativistic) rr orbital automatically contains a (rotating) n component, suggesting

a simultaneous explanation for a and rr trans effects.
Substitution reactions of octahedral complexes 1 331
NH3 + NH 3
Fill. 14.8 A mechanism for the slow
conversion of cis-diamminodichloroplatinum(ll) I I
into the trans isomer. NH3-Pt-C1 CI-Pt-CI
I I
Cl NH 3
not only exerts a trans effect but can exert a cis effect also. One is left
with the distinct feeling that the trans effect, long a cornerstone in synthetic
inorganic chemistry, merits more detailed investigation into its real nature
and origin.
14.5 Substitution reactions of octahedral

complexes
As implied in Section 14.3, the substitution reactions of octahedral transition
metal complexes are in large measure dissociation-controlled (Id and D) and
so reaction rates are very dependent on the ligand displaced but not on the
entering ligand. However, the field covered by the title of this section is a
wide one, not confined to transition metals, and includes an enormous
diversity of behaviours. Another aspect of this complexity is that caution is
always necessary because simple kinetic data cannot reliably distinguish
between A, I., Id and D mechanisms. Our discussion will be restricted to
areas which have been the subject of much work and for which some
considerable insight has been gained. An important general theme will be
the importance of-almost invisible-pre-equilibria and of solvent participa-
tion. For the former, work in non-aqueous solvents has shown that a complex
which, formally, is positively charged may have closely associated with it so
many anions that, if these are included, the overall charge changes sign, and
with it our expectations! In contrast to the statement at the beginning of
this section, the observation that the rate of substitution of transition metal
ion complexes is often proportional both to the complex concentration and
to the concentration of the incoming ligand seems to point to an associative
mechanism:
slow fast
MI.;;+ L ---+ [intermediate] ---+ products
However, the actual rates show surprisingly little dependence on the chemical
nature of L and other quantities-entropies, enthalpies and, to a lesser
extent, volumes of activation-are also surprisingly constant. This sort of
inconsistency is typical in the field, pointing to a hidden complication. The
data can best be understood by a mechanism proposed by Eigen, Tamon
and Wilkins and nowadays referred to as the Eigen-Wilkins mechanism. This
mechanism is one in which the complex C, and the incoming ligand Y, diffuse
together to form a weakly bonded encounter complex, a rapidly established
equilibrium existing between this encounter complex and the free com-
ponents:
C + Y ~ CY; K - [CY] (14.1)

E- [CJM
Values of KE can seldom be measured, but, perhaps surprisingly, they can

be calculated, at least approximately. Two rather different models of the
situation give rise to the same final mathematical equation-and this
model-independence adds confidence in the result. The general conclusion
is not surprising: large ions bump into each other more often than do small
ones (and so give rise to larger KE values), ions of opposite charge bond
more strongly than do those of the same charge. In the second step of the
Eigen-Wilkins mechanism, the encounter complex occasionally rearranges
to give the final products in the rate-determining step
slow
CY ----;;- products
The observed rate of reaction is, then

rate= k[CY] (14.2)
From Equation 14.1 and rearranging,
(14.3)
Now the concentration of the total complex, [CJr, is the sum of that which
is free, [C], and that which is in the encounter complex, [CY]:
[C]r = [C] + [CY]
= [C] + K.[CJM
Rearranging,
[C]=~ (14.4)
1+K•M
Combining Equations 14.2 and 14.3,
rate= kKE[CJM (14.5)
and, from Equations 14.4 and 14.5, we have the final result:
rate = kKE[CJrM
1+K.M
It can immediately be seen that when the product KE[Y] is small compared
to 1 then the commonly observed proportionality of the rate to both [CJr
and [Y] is explained. It transpires that the-often unexpectedly small-
variations in rate constant mentioned above are the consequence of changes
in KE, not of the intrinsic rate k. So, for nickel(ll) complexes the observed
formation rates with respect to [Ni(H 2 0) 6 ]2+ vary from about 200 to
7 000 000 M- 1 s -~, a factor of over 30 000, but the corresponding k values
only vary by a factor of about 2. This means that the observed reaction rate is
controlled by the amount of encounter complex in solution. Relatively
constant is the dissociation step, the rate at which a ligand (here, H 2 0) leaves
the coordination sphere, to be replaced by the ligand in the encounter
complex-a ligand which can be thought of as lurking, waiting for the
opportunity to insert. This interpretation of the reaction sequence is entirely
in accord with the statement at the beginning of this section-and is a
warning not to jump too quickly to mechanistic conclusions from rate
expressions alone.
Substitution reactions of octahedral complexes I 333
Another general aspect of importance is the fact that it is almost invariably

found that in aqueous solution substitution of one ligand by another
proceeds through the intermediate formation of an aqua complex (exceptions
are found in platinum(II) chemistry, where, it will be recalled, a direct
substitution may occur). In fact, this generalization can itself be generalized-
solvent participation is more the norm than the exception. Even if there is
no directly observable so/vento complex, one may well be kinetically important.
The reason is not diflicult to see. First, the solvent is far more abundant
than any other species present. Secondly, the majority of solvents do not
form strong complexes so that a solvento complex, once formed, is a reactive
species. A name has been introduced to cover this situation: cryptosolvolysis,
solvent-mediated substitution in a situation in which no measurable amount
of the solvento species is present. The phenomenon of cryptosolvolysis
dovetails with the Eigen-Wilkins mechanism described above. In the
language of that discussion, the species CY is saturated; that is, because here
Y is a solvent molecule, all of the complex species C is.associated with-at
least -one solvent molecule. The first product species is, then, that in which
the leaving group has been replaced by a solvent molecule. The leaving group
may manage to regain the coordination site by displacing the solvent
molecule and thus reform the starting material. Alternatively, the (new)
incoming ligand may displace the solvent molecule to form the final product.
One can at once sec why there should be but little sensitivity to the incoming
ligand but much to the ligand displaced. Once the latter has been displaced
a reactive, low concentration, species is produced. It is obtaining this reactive
species which is the difficult step.
As has been seen, this sequence is compatible with the Eigen·-Wilkins
mechanism but, nonetheless, has a different emphasis which finds expression
in a different mathematical treatment-the steady-state approximation is
applied to the concentration of the solvento species. However, evidence for
the presence of the solvento species is usually indirect, frequently dependent
upon seeking a rationalization, for example, for the pattern obtained from
the study of a variety of closely related reactions in a series of closely related
solvents. Not surprisingly, the role of water as a solvent has been the object
of particular study; cations are commonly commercially available as their
hydrates and so the interchange of coordinated and solvent water is a topic
of great interest. Focusing on water at this point in the text has the advantage
of bringing in ions other than transition metal species and so widening the
scope of our discussion. Exchange of water between a coordination sphere
and bulk at room temperature can be extremely fast through to extremely
slow; the half-life of a coordinated molecule can be anything from 10- 9 s
through to 10 5 s. The geometry of the coordination sphere does not seem
to have any great effect. At the fast end are ions such as Cs +, Li +, Pb2+,
Cu 2 +, Ca 2 +, Cd2+, Ba 2 +, TJH and Gd3+. At the slow end (half-lives greater
than 1 s) are Al3+, Pt 2 +, Cr3+, Ru3+ and Rh3+ . One would expect to
find Co3+ in this region, too, but, of course, the ion is not stable in
water. Across the first transition series volume of activation (pressure
dependence of rates) measurements show a smooth change from an associative
mechanism at the beginning (T?, VH) to a dissociative mechanism towards
the end (Co2+, Ni 2 +). The change-over point occurs between d 5 (Mn 2 +,
FeH) and d 6 (Fe 2 +) configurations. These data do not correlate with the
rates of reaction and, again, one is warned of the danger of predicting the
behaviour of one species from a knowledge of that of another. Nonetheless,
quite simple ideas do seem to have some validity. Thus, as would be expected,
for a given formal charge, larger ions exchange more rapidly than smaller.
So, the rate constants vary: Cs+ > Rb+ > K + > Na + > Li+ ~ Ba2+ >
Sr2 + > Ca2+ » Mg2+ » Be2+ and so on; it is also usually true that for a
given ionic size an increase in formal charge is associated with a decrease in
reaction rate. The observed tendency of a change from dissociative to
associative activation on moving down a Group in the Periodic Table
becomes, therefore, rather more understandable.
It is not always a simple matter to follow solvent exchange-the overall
reaction is one in which there is no apparent change. Somehow, a label has
to be introduced by means of which exchange can be followed. The obvious
technique is that of isotopic exchange-for instance, preparing a complex
containing coordinated H/ 7 0 and following the exchange with solvent
H 2 16 0, perhaps by mass spectrometry. Clearly, such a method is only
applicable to slow exchanges for otherwise exchange will be complete before
measurements begin. As we have seen, another, much more widely used, set
oftechniques is based on NMR measurements, where, effectively, a spin label
is used. A nucleus which exchanges between bulk and coordination slowly
relative to the instrumental timescale will give two basic peaks (ignoring fine
structure), the relative intensity of which is dependent on the ratio of bulk to
coordinated solvent molecules (and so a ratio which can be used to obtain
the coordination number). A nucleus which exchanges rapidly relative to the
instrumental timescale gives a single peak at an average position; when
exchange and instrumental timescales are similar, broadened peaks result.
The NMR spectra can be studied as a function of temperature, pressure and
time. The latter-by watching the evolution of an NMR signal over a period
of time-enables the technique to be used for exchanges which are too slow
to be studied by the line-broadening technique. These same techniques find
applicability in the study of fluxional systems, a topic which forms the subject
of a subsequent section of this chapter.
First however, a brief discussion of reaction intermediates. As has been
emphasized, the majority of the reactions of octahedral transition metal
complexes are Id (the most common) or D in type. Let us confine our
discussion to the latter, for this is the more clear-cut. A D-type intermediate
must be five-coordinate, which means that, just as for platinum(II) reactions,
square pyramidal and trigonal bipyramidal intermediates have to be con-
sidered, together with a possible interchange between them. Here, however,
our concern is that of adding a ligand to the five-coordinate intermediate,
not of losing one from it, as in platinum(II) chemistry. Is it possible to say
anything about the geometry of the intermediate? What influence do the
other ligands have? Is there a trans effect? It is clear that ligands other
than that being replaced are important. Indeed, steric effects arising from
these ligands are increasingly being invoked as a determining factor. In
general, ligands trans to that being replaced seem to be no more important
than those cis, although there are exceptions-in rhodium(III) complexes,
for example, although the data are limited.
The series cis and trans [Co LX( enh] 2 +, where X is the leaving group and
L the ligand which is either cis or trans to it (X = Cl-, Br-; L = Cl-, OH-,
Base-catalysed hydrolysis of cobalt(lll) ammine complexes 1 335
NCS-, NO-), provide a good example of another phenomenon. This is that

the cis complexes show no tendency to isomerize to the trans on substitution
whereas the trans give some cis product. A possible explanation which has
been put forward is that the cis compounds dissociatively form a five-
coordinate intermediate which is close to a square pyramid. The vacant
position which accommodates the incoming ligand is unique and so offers
no opportunity for rearrangement. If, on the other hand, the trans species
dissociatively forms a near-trigonal bipyramidal intermediate then several
non-equivalent insertion sites would exist, offering the possibility of rearrange-
ment. This is a seductive argument, the more so when a possible n bonding
stabilization of the trigonal bipyramid by the trans ligand L is added (see
above). However, the comment at the end of Section 14.4 has relevance: it
is unlikely that there is no free energy difference between cis and trans
products. If the cis were inherently the more stable, perhaps for steric reasons,
then an alternative interpretation becomes available-the trans form tends
to give some of the more stable cis whereas the cis does not give the less
stable trans. If this alternative were valid, then it is by no means evident that
different geometry intermediates need be postulated.
14.6 Base-catalysed hydrolysis of cobalt(lll)

ammine complexes
It is common for studies of substitution reactions of cobalt(III) ammine
complexes in aqueous solutions to be carried out under slightly acid
conditions, for then protonation of the displaced ligand can occur, inhibiting
the back-reaction and simplifying the kinetics. The usual observation under
these conditions is of a rate which is proportional to the complex ion
concentration and to nothing else. In contrast, when alkaline solutions are
used, relatively rapid substitution occurs and reaction rates are usually
proportional to the hydroxide ion concentration as well as to the concen-
tration of the complex. The explanation for these observations was long
disputed but is now resolved for cobalt(III) ammines and related species.
What remains unclear is why the explanation is not more general-why
other metal ammine species do not follow the same pattern (a few do. but
only with complications). For the cobalt(III) ammines, the OH- rcmoYes a
proton from a NH 3 (or, more generally, a NR 2 H) ligand to give the conjugate
base. The reaction is of the type
This pre-equilibrium is well over to the left-hand side and is rapidly

established. However, the deprotonatcd species is up to ca. 10 13 times as
reactive as the species from which it is derived.
There is now a wide variety of evidence supporting this conjugate base
mechanism, of which just three pieces may be mentioned. First. there is the
unique character of the hydroxyl anion. In aqueous solution there are few
cases where another substituting anion appears in a rate expression and so
a unique mechanism involving the hydroxyl ion is indicated. Secondly.
supporting evidence comes from the fact that exchange of the protons in
[Co(NH 3 ) 5 Cl] 2 +. for example. with those of the solvent water is se\eral
orders of magnitude faster than the rate of base hydrolysis. Finally, the failure
of complexes without such protons, bipyridyl complexes, for example, to
undergo rapid base hydrolysis supports the conjugate base mechanism.
When the stereochemistry is studied, for example by working with cis and
also with trans disubstituted cobalt(III) ammines, then it is found that both
scramble; stereochemistry is not retained. This is taken to mean that the
five-coordinate intermediate is close to a trigonal bipyramid. A major
effect of n bonding involving the deprotonated ammine on the behaviour of
this intermediate has been postulated. However, although plausible, the
model is not proven and it does not easily explain all of the available data.
This, then, is another example in which a pre-equilibrium has a major
effect on a rate law-in this case, introducing a proportionality to the
hydroxyl ion concentration. It is not difficult to see how this arises. In the
example given above, let the forward rate be k 1 and the backward k_ 1 :
Suppose the rate-determining step involves the expulsion of a chloride ion

from this amide complex to give a five-coordinate intermediate, followed by
rapid addition of water to give the final product. That is:
followed by
fast
[Co(NH 3)4(NH 2 )f+ + H2 0 --+ [Co(NH 3) 50H] 2 +
Because the [Co(NH 3 ) 4 (NH 2 )] 2 + aquates as soon as it is formed, the

reaction rate is equal to
(14.6)
Now,
d[Co(NH 3)4 (NH 2 )CI]
dt
is equal to
k 1 [Co(NH 3)5CI][OH]- k. 1 [Co(NH 3) 4(NH 2 )CJ][H 2 0]- k 2 [Co(NH 3)4(NH 2 )CI]
(14.7)
Because the deprotonated species is present in such a small amount, its

concentration will remain approximately constant (approximately zero!)
and so
d[Co(NH 3) 4(NH 2 )CI] "' O
dt
It follows from Equation 14.7, then, that
[Co(NH3)4(NH2)CI] = k,[Co(NH3)5CJ][OH] (14.8)

k . 1 [H 2 0] + k 2
Mechanisms of ligand substitution reactions: postscript 1 337
For convenience, the [H 2 0] term-a constant since we would be working

in a very dilute aqueous solution-can be included in the k_ 1 •
Putting Equation 14.8 into 14.6 gives the final result:
where the experimentally observed rate constant is seen to be equal to
and the observed rate law explained.
14.7 Mechanisms of ligand substitut-ion

reactions: postscript
In the preceding sections a wide variety of reaction mechanisms and kinetic
patterns have been met. It has been emphasized that it is difficult to draw
firm conclusions which cover more than one coordination geometry of one
metal ion in one valence state. Can any general statements be made1 Are
there any general theories? The answer is yes to both questions.
It is found that inert and labile transition metal complexes are associated
with quite specific d orbital occupancies. So, inert complexes are formed by
ions with the configurations ti., ti., ti. and ~~ •. Labile complexes are
those in which the metal ion has electron configurations in which the e"
orbitals are occupied by one or more electrons, together with those with the
configurations ti. and ~~ •. The discussion of Section 13.5 is to be recalled-
site preference energies are related to crystal field stabilization energies. It
seems reasonable to expect that, provided that their reactions are mechanis-
tically similar and that other energies vary smoothly along the series, the
differences in kinetic behaviour associated with different d clcctron configur-
ations may be explicable in terms of crystal field stabilization energies. That
is, in going from the ground state of the reactants to the activated complex,
the contribution made by the change in crystal field stabilization energy
to the potential hill may be of importance. To test this theory one has to
know the detailed geometry of the activated complex, not only its shape, but
also the metal-ligand bond lengths (for these may not be the same as in the
ground-state complex), so that the crystal field stabilization energy of the
activated complex may be calculated. Unfortunately, our ignorance of the
activated complex is profound. Calculations have been carried out assuming
idealized geometries for activated complexes in which the metal ion is
seven- or five-coordinate and give reasonable agreement with experiment. It
is predicted that ti 9 , ~~. and t~ 9 e; ions will always react rapidly (that is,
they do not lose crystal field stabilization energy in forming the activated
complex). For t~"' t'j 9 ,tL and t~ 9 configurations a loss of crystal field
stabilization energy in forming the activated complex predicts, as observed,
relatively slow reaction. For other configurations, the predictions usually
depend on the geometry assumed for the activated complex.
14.8 Fluxional molecules

So far in this chapter organometallic and low metal-valence state systems
have not been mentioned, although such systems have been much studied.
Early work had a parallel with some sections of this chapter-there was
interest in the ligand substitution reactions of transition metal carbonyls, for
instance. It was soon found that the kinetics of reactions in the light could
be very different from those in the dark; in the next section the kinetics of
photochemical reactions of complexes will be covered, although the metal
carbonyls have not been the main focus of these studies. Within the
organometallic field there has been much work stimulated by the recognition
that many of these molecules are internally rather floppy-that they are
fluxional, stereochemically non-rigid. Of course, this is not a phenomenon
confined to this field-the turning-inside-out, the umbrella, motion of
ammonia and pseudorotation in the trigonal bipyramidal PF5 , which
interchanges axial and equatorial fiuorines, are two well-known examples
from main group chemistry. In the organometallic field there are many
examples but we will concern ourselves with just one, very important, class.
This class parallels the example of PF5 : those compounds in which atoms
which might reasonably be considered equivalent, are, in fact, not equivalently
bonded. Examples are u-cyclopentadiene complexes, of both main group and
transition elements, which contain a
M-CH
'\
OI
/ CH" CH
_......CH
CH
unit. If the C 5 H 5 ring moves round by one or two steps then the final
molecule is equivalent to the starting one; if the interchange occurs readily,
as it does, then it indicates that the barrier presented by an intermediate
bonding position, such as
/CH\
QcH
CH
M-1
CH /
""- CH
is not high. Not surprisingly, there has been much discussion on the nature
of this intermediate (it could, for example, be one in which the C 5 H 5 ring
is 11 5 , n-bonded with all CH groups equivalent). In fact, it seems that 1,2
shifts provide a general mechanism. Examples of this behaviour are pro-
vided by Hg(C 5 H 5 h ( =Hg('1 1 -C 5 H 5 h), Fe(C0h('1 5 -C 5 H 5 )('1 1-C 5 H 5 ) and
Cu(PEt 3 h('1 1 -C 5 H 5 ).
A second general example is provided by CO ligands, in complexes
containing this ligand bonded in a variety of ways. Crystal structure
measurements have revealed that there is an almost continuous range of
Photokinetics of inorganic complexes 1 339
H bonding positions for CO, from bonding uniquely to one metal atom
H I H (terminal CO), increasingly being bonded to a second until it is equivalently
"crc'-c/ bonded to two (bridged CO), and then an increasing interaction with a third
I \\ until it is equivalently bonded to three (face-bonded CO). This adaptability
H-C C-H has been recognized in some models of metal cluster carbonyls, such as
\\ I
/c-.....crc, Ir4 (C0) 12 , where a central metal cluster (here a tetrahedron) is surrounded
H I H by a polyhedron of CO groups (here a cube-octahedron), the arrangement
H within the polyhedron apparently as much being determined by packing
(a) constraints within the polyhedron as by metal-CO bonding considerations.
From such models it is but a short step to expect that when not all of the
CO groups are equivalently bonded, as in Co 4 (C0) 12 and Rh 4 (C0) 12 -
~~
molecules which contain three bridging CO groups-then the CO positions
will scramble rather rapidly. Indeed this is just what is found.
Finally, a ligand such as the (cyclic) cyclooctatetraene is too big for all of
6
(b) (c)
Fe
(C0)3
the carbons to be equivalently bonded to most metal atoms and in complexes
the ligand commonly adopts an asymmetric position (Fig. 14.9). Again, not
surprisingly now, the particular carbons bonded to the metal often change
rapidly. The phenomenon has been referred to, rather colloquially but
FIIC. 14.9 (a) cyclooctatetraene, drawn as a warmly, as 'ring whizzers'. There is one question which may be asked-does
planar molecule with four localized double
bonds as might be appropriate for it acting as a the ring rotate or does the metal hop?-which is less meaningful than it
symmetrically bonded ligand. (Actually, the appears. We are not concerned with the translations and rotations of the
molecule is non-planar; if it were planar the entire molecule. It follows that any internal motion must have zero linear
double bonds would be delocalized.) (b) And
(c) are two complexes containing momentum and zero angular momentum. Consequently, the motion involved
cyclooctatetraene as a ligand. in ring whizzers must involve both ring rotation and metal hopping, in
opposite directions-ring rotation on its own would mean that the motion
had angular momentum. The study of such phenomena parallels that
discussed in Section 14.5 for metal hydrates. Nuclear magnetic resonance
spectroscopy is used and the broadening and coalescence behaviour of a
signal as a function of temperature studied. Commonly, the experimental
data are compared with computer-generated model spectra; in many cases
the experimental and theoretical spectra are distinguished only by the noise
in the experimental! In such cases the confidence in a correct interpretation
is high, although this confidence is dependent on the correct assignment of
peaks in the slow exchange limit spectrum. The ability to make such studies
has depended critically on developments in NMR spectroscopy, particularly
Fourier transform techniques which have enabled low-abundance nuclei,
such as 13 C and 17 0, to be studied. Nowadays it is not uncommon for several
different nuclei in one molecule to be accessible for study. Such work has
made it clear that there is often more than one fluxional process occurring
simultaneously in a molecule. We are left with the pattern presented at the
beginning of this section-of molecules that are very floppy, perhaps to be
pictured as rounded lumps of jelly stuck together, as if by surface tension,
but quite free for the various bits to wobble, slide and rotate, rather than as
collections of spheres of various sizes, rigidly locked together.
14.9 Photokinetic& of inorganic complexes

Inorganic pigments have been used since antiquity, most of them having the
attribute that their colours do not change with time, any deterioration in
the paint resulting from changes in the medium supporting the pigment. In
contrast to this stability there are other inorganic systems which are very
sensitive to the action of light. Although the silver-halide photographic
process is outside the scope of the present book, other photographic
processes fall within it. For instance, iron(III) salts are readily reduced to
iron(II) and the oxalate anion is a useful reducing agent. So, although the
complex anion [Fe(oxlJ] 3 - is easy to prepare-concentrated aqueous
solutions of ammonium oxalate and almost any soluble iron(III) salt give
pale green crystals containing the anion-it is not surprising to learn that
it is not very stable, decomposing under the action of light to give iron( II)
oxalate and carbon dioxide. This behaviour is readily explained by the light
exciting an electron in a ligand-to-metal charge transfer transition (LCMT,
see Section 8.1 0.2). This electron transfer oxidizes the ligand and reduces
the metal. The phenomenon is exploited in two ways. First, by impregnating
paper with a mixture ofK 3 [Fe(oxlJ], or a similar species, and K 3 [Fe(CN) 6 ].
When the paper is placed under a mask and then exposed to light-
ultraviolet light is best-and the paper sprayed with water, the areas not
protected by the mask turn blue because of reaction between the photo-
produced Fe(II) and [Fe(CN) 6 ] 3 - to give the blue pigment Turnbull's blue,
KFe(II)[Fe(CN) 6 ]. This simple technique for producing blueprints is used
up to the present day for engineering drawings and the like because it can
easily be applied to sheets of paper much larger than can be accommodated
by ordinary photocopying machines. The salt K 3 [Fe(oxh] is also used in
chemical actinometry-when an aqueous solution of the salt is exposed to
light under standard conditions, the amount of Fe" produced is a measure
of the total amount of light that has passed through the solution. The
quantum yield of Fe"-the number of Fe" produced for each quantum
absorbed-is almost independent of the intensity of the incident light and
of the concentrations of Fe111 and Fe" species. However, the system is
only useful for light of shorter wavelength than yellow. Photolysis of
[Cr(NH 3 j,(NCS) 4 r, which causes release of NCS-, is effected by all
visible light except deep red and is also used in actinometry. For simplicity,
all ofthe above examples concern complexes containing just one metal atom.
A recent example of a more complicated photochemical system is the anion
[Os 18 Hg 3 C 2 (C0) 42 ] 2 -, a molecule in which, essentially, two Os 9 clusters
are joined by a triangle of mercury atoms. Photolysis leads to the expulsion
of a mercury atom, giving [Os 18 Hg 2 C 2 (C0) 42 ] 2 - where the Os 9 clusters
are joined by just two mercury atoms. In the dark and the presence of
mercury this latter complex adds a mercury atom to regenerate the original
compound.
In recent years, inorganic photochemistry has received much attention
and changed considerably from that typified by the historically important
examples just given. There have been three main directions of research. First,
the study of shorter and shorter timescales. This means using short time-pulse
lasers and exploring their immediate (or, if of interest, their not-so-immediate)
consequences. In this way, the sequence of steps following absorption of
radiation can, hopefully, be followed. In practice, there are such intricate
networks of finely separated energy levels, electronic, vibrational and
rotational, that the energy absorbed when an electron is excited usually
very quickly reappears as heat, the system rapidly changing levels when they
cross, until the excitation is dissipated.
Photokinetics of inorganic complexes I 341
The second line of research is in the opposite direction. If energy can be

pumped into a molecule by shining light on it and that energy, in large
measure at least, retained in the molecule, then there is the possibility of
using the energy to carry out chemistry. So, for example, there has been a
persistent search for systems, perhaps several combined, that will photo-
decompose water. If such a process could be made economically viable it
could be an excellent way of harnessing solar energy. It is scarcely likely that
a system could be found in which absorbed photoenergy does not degrade-
indeed, there are good theoretical arguments that such a system could not
exist. So the search is for a system which, as energy is lost and the energy
ladder descended, eventually finds itself in an excited state from which there
is no easy escape route. If, for instance, the excited state trap had a different
spin multiplicity from the ground state then photoemission, at any rate,
would be spin forbidden. Similarly, if the excited state were of the same parity
(g or u) as the ground state then the transition used would be orbitally
forbidden also. If it were a relatively simple system the npmber of vibrational
modes, and so the possibility of vibrational deactivation, would also be
reduced. Clearly, these requirements could well be met by simple, high-
symmetry transition metal complexes. Of course, there is still the non-trivial
problem of subsequently carrying out the required chemistry-but our first
task must be to find a molecule that remains excited long enough for it to
make contact with any molecule with which we would wish it to interact. It
has been estimated that for an excited state to have a chemistry it must live
for at least ca. 10- 9 s. Such systems are known. The most studied species is
[Ru(bpy)J] 2 + (bpy = 2,2'-bipyridine) which has an excited state lifetime in
solution at room temperature of about 10- 8 s and, at lower temperatures,
10- 6 s-or to-ss in some similar species containing ligands with substituents
on the pyridine rings.6 An unexpected result is that in the excited state of
the molecule the excitation is localized in a single bpy ligand molecule (all
the evidence is that the excited state has a lower symmetry than the D3 of
the ground state), that ligand being reduced (and the ruthenium oxidized to
ruthenium(III)). Such electron transfer is just what is involved in metal-to-
ligand charge transfer (MLCT) transitions; they were first met in Section
8.10.3, although the idea that ligands equivalent in the ground state may not
be so in the excited state was not mentioned there. There is no doubt that
excited [Ru(bpy)J]2+ is long-lived enough for it to react with other species in
solution. However, as yet no viable system for obtaining H 2 and 0 2 from
water containing [Ru(bpy) 3 ] 2 + has been discovered, although the energy
available in the photoexcited state is sufficient. Although there seems to be
no connection with the behaviour of the ruthenium compound, the complex
[Ir(bpy)J] 3 + undergoes an isomerization of one of its ligands on photolysis,
as shown in Fig. 14.1 0.
As implied above, it is generally true that the reactions undergone by
photoexcited species are different from those of the ground state species.
Most work has been done on octahedral complexes of d 3 and d 6 con-
figurations (half full and full t 29 shells) and species derived from them-
[Ru(bpy)J]2+ is a d 6 ion. Complexes of Cr 111 and Co 111 have been particularly
studied. Of course, the aquation of [Cr(NH 3 ) 6 ]3+ involves NH 3 loss by both
Fig. 14.10 The photoisomerization of
[lr(bpy)3 ] 3 +. Only one bpy ligand is shown in 6 This explains the interest in Creutz-Taube type Ru 11-Ru 111 species mentioned in Section
detail in the starting material. 14.2.
thermal (normal) and photochemical routes. However, for species of general

formula [Cr(NH 3 ) 5X]3+, where X is a halide or pseudohalide, thermal
aquation leads predominantly to loss of X but photochemical aquation
largely leads to loss of NH 3 . Similarly, whereas [Cr(NH 3 ) 6 ]3+ shows little
photosensitivity, not only are the species [Cr(NH 3 )sX]2+ much more
sensitive, but the favoured process depends on the wavelength of the
radiation. Excitation of spin-allowed transitions (particularly charge-transfer)
preferentially leads to aquation by replacement of NH 3 , whereas Cl- tends
to be replaced when the excitation is into a band that is spin forbidden
(usually, d -+d) from the ground state.
Although transient excited states living ca. 10- 12 s have been observed
in some cases, relatively little is known about the details of the processes
involved in the photochemistry outlined above. Although they are the only
ones we have so far mentioned, internal deactivation and chemical reaction
are not the only processes that occur. Excited states may re-emit radiation,
essentially, immediately. This is fluorescence and involves a transition from
an excited level from which emission to give the ground state is a spin-
allowed process. If there is deactivation to give an excited state from which
emission to the ground state is spin-forbidden, then phosphorescence occurs
with lifetimes of from milliseconds to minutes for solid-state species. In
solution, the corresponding lifetimes are much shorter. When there is only
a small energy difference between the two (spin-allowed and spin-forbidden)
transitions just discussed then there is the possibility that the thermal energy
of the system may be sufficient, for example, to promote molecules from a
lower, spin-forbidden transition, level to an upper spin-allowed. When this
happens fluorescence with a long lifetime is observed, a lifetime which is
characteristic more of a spin-forbidden than a spin-allowed process (but the
nature of which is usually evident from the absorption spectra). It is now
easy to understand why d. and tg• ions have proved of photochemical
interest. For the t~. configuration there are in addition to the ground state,
4 A 2 ., term a host of spin doublets eE., 2T 1., in particular, see Table 7.5
and Appendix 7; these have energies that have approximately the same d
dependence as the ground state; the former is involved in the light emission
process of the ruby laser). Similarly, cobalt(III) complexes, with a 1A,.
ground state, have low-lying spin triplets, in particular a 3 T1., which is the
one usually implicated (again, see Table 7.5 and Appendix A.7). A diagram
which is appropriate to [Cr(NH 3 ) 6 p+ is given in Fig. 14.11. It shows at the
top a (spin quartet) electronic excited state derived from the configuration
t~. e; (and so, remembering that e. electrons are weakly antibonding, with
slightly longer Cr-N bond lengths than the ground state). Associated with
this level is a host of vibrational energy levels. Intersystem crossing from the
spin quartet to a spin doublet derived from the t~. configuration occurs
(because of the spin change, spin-orbit coupling is implicated in this process).
Again, vibrational deactivation occurs until some process to give the ground
state is all that remains. The possibilities for this are discussed above.
A final example of how the relative stability oft~. systems in their lowest
excited state can be exploited is provided by the [Fe(CN)sN0] 2 - anion.
A neutron diffraction study has been carried out on a salt of this anion whilst
the crystal was under intense laser irradiation. The intensity of the light and
the lifetime of the lowest excited state (at liquid helium temperature) were
Further reading I 343
............................................................................................................................................................................................
Fig. 14.11 On the left is shown a simplified

Tanabe-Sugano diagram for the octahedral d3
case. The spin-allowed •r.., ....
4A
24 transition
is detailed on the right-hand side, where
vibrational levels are included. There is
intersystem crossing between 4 T24 and
2 T'-~fE, (only one of the latter is represented).
In contrast, the spin-forbidden step from the
latter to the ground state is more difficult.
such that approximately half of the anions in the crystal were in the excited
state throughout the diffraction experiment. After measurements had been
completed, the laser was turned off and the measurements repeated, this time
with all anions in their ground state. Comparison of the two sets of
measurements revealed that in the excited state the Fe-N bond (of Fe-NO)
was lengthened relative to the ground state, all other bond lengths being
unchanged within error. Clearly, the result requires an excited state local-
ization which is not consistent with the octahedral ~~. configuration Uust as
does the ground state structure), but the essential point-that the lowest
excited state involves an electron in a metal-ligand antibonding orbital-
remains valid.
Further reading
Chern. (1989) 34, 219; Electron Transfer Reactions by R. D.
An older book which is both so easy to read and so forward- Cannon, Butterworth, London, 1980.
looking that it deserves mention is Inorganic Reaction Mech- A Nobel Lecture gives both a fascinating and enlightening
anisms by M. L. Tobe, Nelson, London, 1972. A good contem- account: 'Electron Transfer Reactions in Chemistry: Theory
porary text is Kinetics and Mechanisms of Reactions of Transition and Experiment' R. A. Marcus, Angew. Chern. Int. (1993) 32,
Metal Complexes by R. G. Wilkins, VCH, New York, 1991. 1111.
Up-to-date accounts will be found in Comprehensive Co- For a touch of the unexpected: 'Dissociation Pathways in
ordination Chemistry G. Wilkinson, R. D. Gillard and J. A. Platinum(II) Chemistry' R. Romeo, Comments Inorg. Chern.
McCleverty (eds.), Pergamon Press, Oxford, 1987, Vol. 1: (1990) 11' 21.
Chapter 7.1 by M. L. Tobe 'Substitution Reactions'; Chapter An up-to-date and easy-to-read overview is provided by
7.2 by T. J. Meyer and H. Taube 'Electron Transfer Reactions'; Reaction Mechanisms of Inorganic and Organometallic Systems
Chapter 7.3 by C. Kutal and A. W. Adamson 'Photochemical R. B. Jordan, Oxford University Press, Oxford, 1991.
Processes'. A wide-ranging review is 'Ru(II) Polypyridine Complexes:
Other references worth mentioning are ·An appraisal of Photophysics, Photochemistry, Electrochemistry and Chemi-
square-planar substitution reactions' R. J. Cross, Adv. Inorg. luminescence' A. Juris, V. Balzani, F. Barigelletti, S. Campagna,
P. Belser and A. von Zelewsky, Coord. Chern. Rev. (1988) Complexes', L.S. Forster, Chern. Rev. (1990) 90, 331.
84, 85. 'Fluxionality of Polyene and Polyenyl Metal Complexes'
Dovetailing well with the general level of presentation in B.E. Mann, Chern. Soc. Rev. (1986) 15, 167.
the present book is 'The Photophysics of Chromium(III)
Questions 14.3 A useful rule of thumb is that for suitable systems

(usually d 3 or d6 ) light emission is the more favoured the
greater the number of vibrational steps separating the low-lying
14.1 Distinguish between the order and molecularity of a
excited state from the bottom of the ground state. This means
reaction. Give examples of how pseudofirst- and pseudosecond-
that simple anions such as [MI 6 ] 3 - would be favoured for
order kinetics may arise in inorganic substitution reactions.
emission studies. Yet they have not been discussed in this
chapter. Suggest plausible reasons for this omission.
14.2 [Fe(CN) 6 ] 2 - and [Fe(CN) 6 ] 3 - are inert complexes
yet they react with each other rapidly. [Co(NH 3 ) 6 ]l+ is inert 14.4 Detail the experimental distinctions between A, I., I0
and [Co(H 2 0) 6 ]2+ is labile yet these two react with each and D mechanisms. Suggest the tactic you might employ
other slowly. Detail the current explanation for this different in choosing ligands to force an ion such as Pt", for which A-type
behaviour. kinetics are typical, to follow a D-type mechanism.
Bonding in cluster compounds
15.1 Introduction
The study of cluster compounds is an active area of current research. The
scope is widening and systematic methods of synthesis beginning to appear
based, for example, on the isolobal principle first mentioned in Chapter 10
and which will be met again later in this chapter. It seems likely that it will
eventually be found that such compounds may -involve almost any element,
perhaps with the exception of the most electropositive. Some representative
examples are given in Fig. 15.1. Despite all this activity, our understanding
of the bonding in cluster compounds is relatively primitive. Most of them
are beyond the scope of detailed ab initio calculations and resource has to
be made to more approximate methods, the Xot and its refinements probably
being the most reliable (see Section 10.3.1). However, it seems that even the
simplest of methods, in particular the extended Hiickel (Section 10.5), can
give the highest occupied orbitals with a tolerable accuracy, though being
far less reliable for more deeply lying orbitals. Unfortunately, cluster
molecules have such a plethora of bonding molecular orbitals that any
photoelectron spectroscopic data (Section 12.7) can only be interpreted with
the greatest difficulty. So, it is difficult to find any reliable external check on
the available calculations.
Before commencing a detailed study it will be found useful to consider a
well known and simple molecule which may be regarded as a cluster-the
tetrahedral molecule P4 . It will prove helpful to compare this with another
tetrahedral molecule which has been the subject of study- the boron chloride
B4 CI4 , which consists of a B4 tetrahedron with a chlorine pointing radially
outwards from each boron; it is made from gaseous BC1 3 passed through
an electrical discharge. The argument developed will be symmetry-based and
so in Table 15.1 is given the character table for the Td point group. Three
different models of the bonding in P4 will be compared, the object of the
exercise being to show the relationship between the three models. Rather
similar relationships occur between the extant descriptions of the bonding
346 1 Bonding in cluster compounds
Fig. 15.1 Some examples of cluster

compounds. (a) Os 5 (C0) 16-at first sight this
appears a high-symmetry cluster. It is not. It
contains one Os(C0) 4 group and four Os(C01 3 .
At the moment, all that can be hoped for with
such structures is a retrospective rationalization.
(b) [~ 5 -C 5 H 5 Fe(C0)] 4 -a tetrahedral cluster
with a CO group above each tetrahedron face.
The ~ 5 -C 5 H 5 group attached to the closest iron
atom has been omitted, as has the most distant
CO. (c) Rh 6 (C0) 16-an octahedron of rhodiums
but four face-centering CO ligands reduce the
symmetry of the cluster to Td. To each Rh are
attached two terminal CO groups (not shown)
(d) Co 3 (C0) 9 CCH 3 -this cluster is typical of
those taken to model the attachment of a
reactive organic molecule to a bare metal surface.
in more exotic, more complicated, clusters although these relationships tend

to be more difficult to see than in the cases now to be considered.
The three models that follow all lead to the same qualitative result but
they do so by such different routes that one might reasonably conclude that
they are different. As will become evident, these apparent differences are not
fundamental. Although minor differences may persist, in the things of
Table 15.1 The Td character table
importance, the three models are equivalent. The P4 molecule has been
Td E 8C3 3C2 6S4 6ud chosen because just s and p valence shell atomic orbitals are involved; it is
the additional involvement of d orbitals that leads to the more complicated
Al 1 1 1 1 1 situation in clusters involving transition metals.
A2 1 1 1 -1 -1
15.2.1 'Simple ammonia' model for P4
E 2 -1 2 0 0
In NH 3 , one may think of three equivalent N-H bonding orbitals and of
Tl 3 0 -1 1 -1 an additional lone pair of electrons on the N atom (Appendix 2). The
T2 3 0 -1 -1 1 simplest picture of the bonding in P4 is to think of the N of NH 3 being
replaced by P and each H also being replaced by a P, the resulting overall
Bonding in P4 (and B4 CI 4 ) 1 347
bonding network leading to the P4 molecule. In this model, each P has a

lone pair of electrons, just like the N in NH 3 , and is also involved in three
P-P bonds, just as N in NH 3 is involved in three N-H bonds. In this model
each of the six edges of the P4 tetrahedron is associated with a P-P bond.
Such a bond need not lie along the edge, perhaps a pattern such as that in
Fig. 15.2 is more probable. The symmetries of the (delocalized) bonding
molecular orbitals in P4 can be obtained from the local orbitals in Fig. 15.2
by using them as basis functions for the generation of a reducible repre-
sentation in the Td point group. This development is detailed in Table
15.2 where the conclusion is reached that they have A 1 (0) + T 2 (l) + E(2)
symmetries. The numbers in brackets refer to the number of nodal planes
inherent in each molecular orbital. To obtain these numbers explicit
mathematical expressions for the orbitals are needed; they are readily
Fig. 15.2 A symbolic representation of four of obtained by the methods described in Appendix 6, and representative
the six P~P edge bonds in P4 • The exact orbital examples of the molecular orbitals are shown in Fig. 15.3. The more nodes,
makeup and detailed onentation in space of
these edge bonds are not important as they do the higher the energy, and so we conclude that the stability sequence is
not change the symmetry properties of A 1 > T 2 > E. These are stabilities relative to the stability of the P-P bonds
members of the set. that we started with; as interactions between these bonds will, as a first
approximation, lead to a splitting-and not an overall shift in energy-we
Table 15.2 The transformation conclude that the A 1 orbital is stabilized and theE pair destabilized relative
properties of the tetrahedron edge to the P-P bonding orbitals with which we started. The overall picture is
orbitals of P4 . Characters generated by that given in Fig. 15.4, where a small displacement of the T 2 set, given by a
the six orbitals under the operations of more detailed analysis, has been included. Feeding electrons into the bonding
the Td point group (see Fig. 15.2) which molecular orbitals gives the scarcely surprising result that all are filled. Of
is the sum of the characters of the
irreducible representations A1 + E + T2 course, had theE orbitals in Fig. 15.4 been much higher in energy then they
would perhaps have been too high in energy to be occupied. But that is
E 8C3 3C2 65 4 6ad another story.
6 0 2 0 2
Fig. 15.3 Representative examples of the edge

group orbitals of P4 . The signs indicate relative (a)
phases. So, (a) the A1 is
1/, 6(a + b + c + d + e +f); (b) one of theE
is 1/2(a - b + c -d); and (c) one of the 72 is z
1/v 2(e- f).
(b) (c)
P atom \
;=:;::=E
Fig. 15.4 A schematic energy level diagram for

the six edge molecular orbitals of P4 . P- P edge
bonding
(diatomic)
\
P4 edge
molecular
orbitals
15.2.2 'Twisted ammonia' model for P4

One sometimes encounters the statement that a rotational axis which is
threefold, C3 , or above (C4 , C5 etc.) shows cylindrical symmetry. Now, the
view down the C3 axis of a molecule such as ammonia shows something
more bumpy than cylindrical, so just what does cylindrical symmetry mean?
It means that, as long as they remain mutually perpendicular, the x and y
coordinate axes can be placed anywhere perpendicular to the C3 , C4 , C5
etc. axis (this axis is conventionally chosen as z). Of course, some choice
of x, y directions may be more convenient-they lead to simpler mathematics
and simpler pictures, but any other choice would be just as acceptable
(a) and would in no way change the final result. In a cylinder, x and y can
be chosen anywhere perpendicular to the principal rotational axis; so, too,
here. What applies to x and y coordinate axes also applies to Px and Py
atomic orbitals, and, for us, this is the important aspect of cylindrical
symmetry.
When, in Section 15.2.1, we talked of the N-H bonding orbitals of
ammonia, we did not enquire into the nitrogen orbitals involved in this
bonding. Had we done so, they would have been mixtures of nitrogen 2s,
2pz, 2px and 2py orbitals and would have looked something like those shown
in Fig. 15.5(a). But nitrogen 2s and 2pz are axially symmetric and, as we have
seen, 2px and 2py can be chosen to point anywhere perpendicular to z. This
means that as an alternative to Fig. 15.5(a) we could have worked with the
(b)
orbitals of Fig. 15.5(b ). They look different but, remembering that they show
contour diagrams, the actual electron density distributions implied by Figs.
Fig. 15.5 A set of sp 2 or sp 3 hybrids viewed
down a threefold axis of P4 . They may be 15.5(a) and (b) are identical.
orientated towards (a) the tetrahedron edges or Return now to the P4 molecule, and regard the contribution to the
(b) faces. Although these alternatives appear bonding from each phosphorus to arise from orbitals such as those of Fig.
rather different, they both correspond to
cylindrical symmetry of electron density 15.5(b ). Because they point towards the centres of faces of the P4 tetrahedron
distribution around the threefold axis. they combine to give three-centred bonding orbitals of the sort shown in
Bonding in P4 (and B4 CI,J 1 349
Fig. 15.6. There are four faces of the tetrahedron and so just four orbitals
like that in Fig. 15.6. These orbitals can be used to obtain delocalized
molecular orbitals; the steps in the sequence are detailed in Table 15.3. The
orbitals that result are of A 1(0) and T2 (1) symmetries; again, their explicit
forms may be derived using the methods of Appendix 6 and the numbers
in brackets refer to the number of inherent nodal planes; A 1 is more stable
than T 2 . The final molecular orbital energy level diagram that results is
shown in Fig. 15.7. It is immediately clear that this figure poses problems for
P4 • Placing a lone pair of electrons on each P leaves six bonding electron
pairs to be fed into the four orbitals of Fig. 15.7. It cannot be done. So what
of the statement made earlier that 'the actual electron density distributions
implied by Figs. 15.5(a) and (b) are identical'-if this is so, why do the
FJi:, 15.6 An approximate picture of two of the different choices lead to different results? Clearly, this is a point to which we
four tetrahedron face bonding orbitals of P4 •
Each of these orbitals is totally symmetric with must return. Although the present model seems to have failed for P4 , we end
respect to rotation about the threefold axis on this section on a more encouraging note: the model works for B4 Cl 4 , for
which it lies. which it is a simple matter to see that there are jus~ .eight valence shell
electrons available for the cage bonding. Three-centred face bonding orbitals
Table 15.3 The transformation of the of the type just encountered are commonly used in discussions of the bonding
tetrahedron face orbitals of P4 . in the boron hydrides and related compounds. Such species are cage (or
Characters generated by the four cluster) molecules and their apparent electron deficiency is bypassed by using
orbitals under the operations of the Td such three-centred orbitals. Allocating two electrons to each terminal B-Cl
point group (see Fig. 15.6) which is the bond, in B4 Cl 4 there are two electrons from each boron to allocate to cage
sum of the characters of the irreducible
representations A1 + T2 bonding molecular orbitals, a total of eight. Figure 15.7 is, now, clearly
appropriate. But, conversely, Fig. 15.4, which requires 10 electrons, is not.
E 8C3 3C2 6S4 6u" Or is it? At the end of Section 15.2.1, it was suggested that if the highest E
4 1 0 0 2 orbitals are sufficiently destabilized, then they will be unoccupied. It seems
that this is the case in B4 Cl 4 , leading the three-centred bonding model to be
preferred.
P atom
Fig. 15.7 A schematic energ; level diagram for

the four face molecular orbitals of P4 •
,P,
P-P
bonding
(triatomic)
'------Al
P4 face
molecular
orbitals
Zt
Fig. 15.8 The 3p, orbitals of the P atoms in

P4 • Each local z axis is directed radially
outward, away from the centre of gravi1y of the
tetrahedron.
15.2.3 Atomic orbital model for P4

Table 15.4 The transformation
properties of the phosphorus u orbitals If one were to wish to carry out a detailed molecular orbital calculation on
(p,, pointing towards the centre of the P4 one would consider, at least, all of the valence shell atomic orbitals on
tetrahedron) in P4 • Characters each phosphorus. The local C3 axis would be chosen as z, and so defining
generated by the four orbitals under the
p., whilst, as the previous section indicated, there would be considerable
operations of the Td point group (see
Fig. 15.8) which is the sum of the freedom about the choice of local x andy. To simplify the problem we shall
characters of the irreducible ignore the phosphorus 3s orbitals. This is not too bad an approximation.
representations A1 + T2 As in ammonia, each phosphorus has a lone pair of electrons and the major
contributor to this lone pair orbital is 3s; we are simply ignoring the small
E 8C3 3C2 6S4 6ud
3pz contribution. In this model, each phosphorus 3pz orbital in P4 points
4 1 0 0 2
towards the centre of the tetrahedron (Fig. 15.8); simple group theory,
detailed in Table 15.4, shows that they combine to give A 1(0) and T2 (1)
orbitals, the former being bonding and the latter, correspondingly, anti·
bonding. The A 1 orbital is schematically illustrated in Fig. 15.9. We now
have to consider the 3px and 3py orbitals on each phosphorus atom; here
the group theory is not so simple but is outlined in Table 15.5. It is found
that T 2 (most stable)+ £(intermediate) and T 1 (least stable) orbitals result.
Even with the detailed form of the orbitals (and these are given in the
reference given in Table 15.5), it is no trivial matter to obtain the relative
energies of these T 1 , E and T 2 sets, although that just given seems to be the
most general. The detailed pattern is rather dependent on the actual element
forming the tetrahedral cluster. These three levels are combined with the
A 1 + T 2 , arising from the orbitals of Fig. 15.8, in Fig. 15.10(a). This diagram
is adapted to P4 and B4 Cl 4 in Figs. 15.10(b) and (c), respectively, where the
results of photoelectron spectroscopic measurements on the former species
are included (see Section 12.7). In Figs. 15.10(b) and (c) the electron pairs
Fig. 15.9 A schematic picture of the totally
SYmmetric (A 1 ) combination of the four 3p,
excluded in the cluster bonding discussion have been added (phosphorus
orbitals of Rg. 15.8. lone pairs and B-Cl bonding pairs, respectively).
Table 15.5 The transformation properties of the Px and Py orbitals of each
phosphorus atom in P4 . Characters generated by the eight orbitals under the
operations of the Td point group which is the sum of the characters of the
irreducible representations T1 + T2 + E
E 8C3 3C2 6S4 6rr"

8 -1 0 0 0
Note: it is not a trivial task to generate the above reducible representation. First, the character of -1 under
8C3 . The -1 is a sum of ( -~} + (-~);the calculation is very similar to that detailed in Appendix 4. Secondly,
the character of 0 under 6S4 . To obtain this, it is helpful to remember, as discussed in Section 15.2.2, that
there is considerable freedom of choice of x andy axes at any phosphorus atom. The value of 0 ( = + 1 - 1)
is most readily obtained if x and y are chosen one to be symmetric and one antisymmetric with respect to
reflection in the local crd mirror plane. See Appendix 4 of Symmetry and Structure, S. F. A. Kettle, Wiley,
Chichester, 1995 for a detailed derivation.
T1
T2
n (tangential)
cr (radial) E
T2
(a) A1
Fig. 15.10 (a) A combined schematic
molecular orbital energy level diagram arising
from the radially directed a orbitals (a(radial)) of
a tetrahedron together with the n that are t2 (3) (-1.6)
tangential to the local z axes (n(tangential)).
(b) Fig. 15.10(a) adapted to P4 . The results of
calculations/photoelectron measurements on tl (1) (-1.3)
P4 are indicated in units of eV. Values in
brackets are calculated.
n (tangential)
cr (radial) e(1) 9.6
t2 (2) 10.3
a1 (2) 11.5
~'''"
19.3
cr (non-bonded)
:
(b) a1 (1) 29.0
e(1)
1t (tangential)
Fig. 15.10 (continued) (c) Figure 15.10(a)

adapted to s.cl •• where the e(1) level is
empty.
a (radial)
~~~
cr(B-CI) ·~
(c) 81 (1)
15.2.4 Unity of the three models of P4 bonding

Although they have features in common, Figs. 15.4, 15.7 and 15.10(a), all of
which purport to describe the same molecule, P4 , are rather different-why?
In Section 15.2.3, the atomic orbital model, all the valence shell atomic
orbitals of the phosphorus atoms were included. In Sections 15.2.1 and 15.2.2,
describing localized bonding orbitals, just the face and edge bonding
molecular orbitals, respectively, were included: their antibonding counterparts
were ignored. We shall not discuss the derivations associated with these
antibonding orbitals (they involve complications very similar to those
mentioned in Section 15.2.3 in connection with the phosphorus 3px, 3p,
orbitals) and merely quote the results. These are given in Fig. 15.11. In Fig.
15.11(a) are given approximate energy level diagrams arising from edge and
face antibonding orbitals. Note particularly that the stabilization resulting
from in-phase interactions between these orbitals can be such as to more
than compensate for their original antibondingness and so give overall
bonding. In Fig. l5.11(b) the levels are combined with their bonding
counterparts. The resulting face-basis and edge-basis energy level diagrams
are very similar, not only with each other but also with Fig. 15.10(b). At
long last, all three models agree-at least, superficially. In particular, we can
now use the three-centred-bond model to explain why P4 is stable. An E
molecular orbital, derived from what was originally an antibonding orbital,
becomes sufficiently stabilized for it to become bonding (Fig. 15.11(b)).
Comparison of the two models represented in Figs. 15.11(b) make under-
standable the fact that sometimes this orbital is occupied (P4 ) and sometimes
Wade's rules 1 353
····························································································································································································
(a)
Fig. 15.11 (a) Representative edge (upper) and unoccupied (B 4 Cl 4 ); in Section 15.5 further insights into this E orbital will
face (lower) antibonding orbitals of a tetrahedron. be gained.
The face antibonding orbitals are doubly
degenerate and the figure shows beth
One final point. All three models agree that the lowest lying orbital is of
components, one in the left-hand face and the A 1 symmetry but picture it rather differently. In the edge-bond model it is
other in the right. If both components are to composed of edge bonding orbitals (Fig. 15.3), in the face-bond model it is
refer to a single face one component (either)
should, mentally, be reflected in a vertical
an in-phase combination of all four of the face-bonding orbitals (we have
mirror plane containing the vertically drawn not attempted to picture it because such a picture would be rather convoluted
edge in the lower diagram. and complicated) whilst in the atomic-orbital model it is as shown in Fig.
15.9. Yet all three are attempting to describe the same molecular orbital!
Are the differences real? No, not if all three models are taken to the same
limit-the models differ in the sequence in which interactions are 'switched
on', not in the final result. Our conclusion is clear; different models of cluster
bonding may produce results which, superficially, are very different but these
differences may be illusory.
15.3 Wade's rules

A focal point of interest of clusters such as those shown in Fig. 15.1 is the
answer to the question: how do they hold together and what is the bonding?
This question seems almost impossibly difficult. One might reasonably
assume that the 18-electron rule holds for each transition metal atom but
how many electrons can be regarded as metal-metal bonding and how many
h h
T2
T2
E E Edge-anti bonding
derived
T2
Edge-bonding
derived
(b)
Fig. 15.11 (continued) (b) A qualitative
energyo level diagram which indicates the result
of combining edge bonding and antibonding
orbitals to give (tetrahedral) molecular orbitals.
(c) The corresponding diagram for 'face' h h
bonding and antibonding orbitals, which
combine to give (tetrahedral) molecular orbitals.
T2
In both upper and lower diagrams the final
\
resu~ is akin to Rg. 15.10(a).
E E
\ T2
Face-antibonding
derived
T2
A1
Face-bonding
derived
(c)
metal-ligand? There might even be orbitals significantly involved both in

metal-metal and in metal-ligand bonding. In the event, significant progress
has been made. The first step is to decide how many electrons are to be
associated with metal-metal bonding, at least as a first approximation.
The insight that provides an answer to this question comes from boron
chemistry. There are several hundred boron hydrides and their derivatives;
structurally they are based on polyhedra with boron atoms at their apices.
Wade's rules 1 355
Characteristically, each boron has a terminal hydrogen atom bonded to it,

that is, one in which the boron-hydrogen bond points away from the centre
of the polyhedron. Some examples are shown in Fig. 15.12. The existence of
such a large number of closely related molecules based on a simple common
unit, B-H, invites attempts to provide a description of their bonding,
applicable to all. Apart from commenting that, just as for B4 Cl 4 in Section
15.2, empty orbitals are important, we shall content ourselves with the
simpler question-how many electrons are associated with boron-boron
bonding? The number is easy to calculate. Each B-H unit is associated with
four electrons, three from each boron and one from each hydrogen; two of
these electrons we allocate to the terminal B-H bond. This leaves two
electrons per B-H unit for cluster bonding, to which have to be added any
formal charge on the molecule (many boron hydrides are anionic). Making
use of the experimental observation that anions of general formula B.H;-
are always closed complete polyhedra with triangular faces, we conclude that
such polyhedra have (n + 1) electron pairs involved in skeletal bonding. Such
species are termed c/oso (closed cage). If a molecule differs from a c/o so
species by a single B-H unit being absent, it is called a nido (nest-like) species.
They all contain bridging hydrogens which 'sew up' the edges of the hole
caused by the omitted B-H. Such molecules are usually neutral species, of
general formula B.H.+ 4 , but are conveniently thought of as the composite
(B.H:- + 4 protons). As the sewn up comment above implies, the four
protons are associated with orbitals also involved in cage bonding; we
conclude that nido species are characterized by (n + 2) skeletal electron pairs.
If two cage B-H units are missing from a close polyhedron, one has the
arachno (web-like) series, of general formula BnH.+ 6 and (n + 3) skeletal
electron pairs. A rare series is the hypha (net-like) with three B-H units
missing to give B. H.+ 8 and (n + 4) skeletal electron pairs. Clearly, this can
be worked backwards; given the formula of a boron hydride species the
structure can be deduced with some reliability. Thus, B5 H 9 is B.H.+ 4 with
n = 5 and so a nido species with one B-H unit missing from the basic
polyhedron. The basic polyhedron is therefore an octahedron; we correctly
deduce that B5 H 9 has C4 , symmetry. One important aspect of these rules
(called Wade's rules, after the person who first formulated them) is that they
may also be applied to metal clusters.
Important in this extension is the recognition that, in the boron hydride
discussion above, each B-H unit supplied not only two electrons to skeletal
bonding but also three atomic orbitals from the boron. We could think of
these as three sp 3 hybrids (the fourth being involved in B-H bonding).
However, it will prove more convenient to think of the three orbitals as one
sp hybrid (the other sp hybrid being involved in B-H bonding) and the
(Pxo p,) pair. The key point is that the set of three should transform as A 1 + E
under local C3 , or C4 , symmetry (other cases are actually similar-but
different symmetry labels apply). Provided that this condition is met, a wide
variety of units, with varying electron count, can substitute for B-H in boron
hydrides. Indeed, and as is usually the case in transition metal cluster
compounds, there need be no B-H units in the cage molecule. Nonetheless,
recognition of the boron hydride pattern reveals the general cluster pattern
also. So, a C-H unit is isoelectronic with a (B-H)- unit and has a similar
orbital pattern (and, perhaps more important, C-H+ and B-H are related
2- 2-
812 Hu Bto Hto
Fig. 15.12 Representa~ve boron hybride

molecules. Three are closo and two are nido .
2-
Ba Ha
Wade's rules 1 357
in the same way but with one electron fewer). Following the argument
above, it would be expected that B5 H~- is a closo molecule, and so a
trigonal bipyramid. In fact, this anion has not been prepared. Writing it
as B 3 H 3 (BH)~-. and substituting two BH- by CH, one concludes that
B3 H 3 (CH) 2 , C 2 B 3 H 5 , which has been prepared, is also expected to be a
trigonal bipyramid, as indeed it is.
Application to transition metal cluster compounds starts with the 18-
electron rule, in the sense that we take nine as the number of valence shell
metal orbitals of relevance. We reserve three of these, satisfying the (A 1 +E)
symmetry requirement given above, for metal-metal bonding, leaving six for
bonding of ligands, back-bonding, non-bonding-or anything else. Not
surprisingly, we do not attempt to enquire into the detailed characteristics
of the six orbitals' All we do is to allocate 12 electrons to them. Any electrons
in excess of this number are to be associated with the three orbitals we have
reserved for cluster bonding. Consider the Fe(COh unit as an example. The
neutral iron atom has a ... 3d 6 4s 2 configuration-that is, eight valence shell
electrons. In applying the 18-electron rule to the iron atom we add two (O")
electrons from each carbonyl ligand to give a total of 14 electrons. Of these,
12 are allocated to the six non-cluster orbitals, leaving two electrons to be
associated with the three, A 1 + E, cluster orbitals. We conclude that the
Fe(COh unit, like B-H, has two electrons and three orbitals associated with
cluster bonding. This relationship between B-H and Fe(COh is expressed
in the statement that the two units are isolobal, a term already met in Section
10.4.3. Ideally, isolobal units have the same number of orbitals of the same
effective symmetries, of similar size and energies and, formally, are associated
with the same number of electrons. So, because B.H,;- and C 2 B._ 2 H. have
closo structures, it follows that if we were to add another BH, another CH +,
or another Fe(COh (all three are isolobal) that the resulting molecule would
also have a closo structure. In particular, it is to be expected that species of
general formula C 2 BnHn+ 2 Fe(COh will be closo. Such molecules have been
prepared and, indeed, are closo, the C, B and Fe atoms lying at the apices
of a (slightly distorted) polyhedron with (n + 3) vertices.
As an example of the application of Wade's rules to a purely transition
metal cluster consider the tetrahedral molecule lr 4 (C0) 12 , where each Ir
atom is part of an Ir(COh unit (Fig. 15.13); the preparation of this compound
was briefly described in Chapter 4 and, in more detail, in Question 4.4.
0 °0
c c c
Because the electron configuration of an Ir atom is ... 5d 7 6s 2 , there are
9 + 6 = 15 valence electrons associated with each Ir(COh group (the six
oc,/'\/'
'-I/ coming from COO" orbitals). Placing 12 in the six non-cluster orbitals, three
electrons arc left to be associated with the three Ir orbitals allocated to cluster
bonding. This gives a total of 12 electrons for bonding within the tetrahedron,
exactly the same number as for P4 in Section 15.2. Clearly, the general
arguments used in Section 15.2 can be applied to Ir 4 (C0) 12 , although the
OC- l r - - - - l r - C O composition of the various orbitals involved will be less clear, for each may
oc
/-............
oc-lr-co / "co have a 4d orbital component. In summary, then, there are 60 (4 x 12 + 12)
l electrons associated with the cluster lr 4 (C0) 12 , although only a few of them
c arc formally involved in metal-metal bonding. It is found that the number
60 is a characteristic of tetrahedral transition metal clusters. The recognition
0
Fig. 15.13 The tetrahedral structure of of this fact, together with the generalizations implied by Wade's rules, leads
lr4 (C0) 12 . us to take an interest in the number of cluster valence electrons (often referred
Table 15.6 Cluster valence electron (CVE) counts for some common
polyhedra
Geometry CVE Example

Tetrahedron 60 lr4 (C0) 12
Butterfly (a tetrahedron with one edge open) 62 [Re.(C0) 16 f-
Square pyramid 74 Fe 5(C0)1 5C
Trigonal bipyramid 72 Os5(C0) 16
Octahedron 86 Ru 6(C0) 17 C
Trigonal prism 90 [Rh 6(C0) 15C] 2 -
to as the CVE; an alternative name is skeletal electron-pair count) in a

molecule. Indeed, it transpires that this is a useful exercise-different
geometries are associated with different CVE counts. Some examples are
given in Table 15.6. In all cases it is possible to develop arguments similar
to those given above for Ir4 (C0) 12 to justify the CVE count. So, in the case
of 'butterfly' clusters, with two electrons more than the 60 required for a
tetrahedron, the additional two are taken to fill an antibonding edge orbital,
such as one of those shown in Fig. 15.1l(a), thus breaking the edge bond
associated with this edge and giving the observed structure.
In this section, Wade's rules have been introduced and the bonding in
some tetrahedral cluster molecules considered in outline. In a sense, these
two topics are incompatible. Wade's rules do not work for tetrahedral
clusters! The tetrahedral molecules we have considered are closo, and so,
according to Wade's rules, should be derived from the unknown anion
B 4 H~- or, equivalently, B 4 Cl~-. But, of course, we have found that such
tetrahedral molecules actually have eight (B 4 Cl 4 ) or 12 (P4 , Ir4 (C0) 12 )
cluster bonding electrons, not the 10 indicated by formulae such as B4 Hi-
and B 4 Cl~-. Both the strength and limitations of Wade's rules are at once
evident. They are powerful, enabling the majority of simple clusters to be
given a common rationalization. However, they do not attempt to provide
detailed insights into the bonding in individual molecules and so exceptions
to the rules must be expected. Not surprisingly, one of the objects of detailed
calculations has been to discover just why Wade's rules work and what are
their limitations.
In Table 15.7 are listed the number of cluster electrons associated with
units which may be regarded as building blocks for cluster compounds. In
each case, three orbitals are allocated for cluster bonding; closo structures
have (n + I) skeletal bonding pairs associated with n building units.
Similarly, following Wade's rules, nido structures have (n + 2) skeletal
electron pairs and arachno, (n + 3) skeletal pairs, all associated with n cage
atoms. An advantage of the use of Wade's rules is evident when a cluster
contains bridging ligands-one simply includes in the cluster electron count
an appropriate number of electrons per bridging ligand-two for a bridging
CO, for example. The calculation is otherwise unchanged.
Wade's rules only apply to clusters which may be regarded as derived
from a cage, closo, molecule. However, many exotic clusters exist which are
much more complicated. When such molecules may be regarded as resulting
from the fusion of molecules which, separately, are derived from closo
Topological models I 359
Table 15.7 Transition metal cluster building blocks (Cp = ~ 5 - C5 H5 )
Molecular fragments associated with -1 skeletal electrons per unit

MnjTcjRe(CO),; CrjMo/WCp
Molecular fragments associated with 0 skeletal electrons per unit
FejRujOs(C0) 2 ; MnjTcjReCp; CpjMojW(C0) 3
Molecular fragments associated with 1 skeletal electron per unit
Co/Rhjlr(CO),; Fe/RujOsCp; MnjTcjRe(C0) 3
Ni/Pd/Pt(C0) 2 ; CojRhjlrCp; FejRujOs(CO),; Cr/Mo/W(C0) 4
Ni/Pd/PtCp; Co/Rh/lr(C0) 3 ; MnjTc/Re(C0) 4
Ni/Pd/Pt(COJ.; Co/Rh/lr(C0) 4
structures then, again, electron counting rules may be applied. Suppose the
fusion occurs because one transition metal atom is common to two cages
(an example is given in Fig. 15.14). Remembering that the 18-electron rule
(Section 10.1) is most probably applicable, the 18 electrons associated with
the common atom have been counted twice~once in each cage. The electron
count for the entire cluster is, then, the sum of those for the component,
separate, cages less 18. When fusion occurs between two transition metal
atoms which are bonded to each other and common to both cages then the
count is the sum of those for the component, separate, cages less 34
(34 =twice 18less 2 electrons for the bond between the two atoms common
to both cages). Similar counts are applicable to more complicated fusion
Fig. 15.14 A schematic representation of
single-atom fusion between two cages. Only the patterns but will not be discussed~the above two examples serve to establish
fusion atom is shown explicitly. the pattern.
15.4 Topological models

Surprising as it may seem, aspects of topology and chemistry are closely
related. When one draws a line between the carbon and a hydrogen atom
in methane, one is indicating a 1:1 correspondence between a topology~
a connectedness, a line~and a bonding interaction. This goes even deeper. It
is a simple application of group theory to use Table 5.1 to show that the
four C-H bonds in methane, or rather the four lines drawn from carbon to
hydrogens, span the A 1 + T 2 irreducible representations of the Td point
group. The four topological lines lead us to conclude that there is a singly
degenerate (A 1 ) C-H bonding molecular orbital in methane and a triply
degenerate set (T2 )~and that there are no others. Similarly, for P4 , in
the simple ammonia model, Section 15.2.1, we effectively used the six edges
of a tetrahedron to conclude that the interactions responsible for the bonding
were of A 1 + T 2 + E symmetries. Not surprisingly, this rather simple and
apparently straightforward way of obtaining insights into cluster bonding
has been the subject of many enquiries. Unfortunately, the model is not
always quite as simple as in the examples above. When discussing P4 we
360 I Bonding In cluster compounds
............................................ ······················································· ................................................................. ......................
Table15.8 The octahedral character table and octahedral clusters
o" E 6C4 3C2 6C2 8C3 6S4 3uh 6ud 86
A,.. 1 1 1 1 1 1 1 1 1 1
A2Jt 1 - 1 1 -1 1 1 - 1 1 - 1 1
Eg 2 0 2 0 -1 2 0 2 0 -1
r,.. 3 1 -1 -1 0 3 1 -1 - 1 0
T,g 3 - 1 -1 1 0 3 -1 -1 1 0
Alu 1 1 1 1 1 -1 -1 - 1 -1 -1
A2u 1 -1 1 - 1 1 -1 1 - 1 1 - 1
Eu 2 0 2 0 -1 - 2 0 -2 0 1
Tlu 3 1 -1 -1 0 -3 -1 1 1 0
r,u 3 - 1 -1 1 0 - 3 1 1 -1 0
Each threefold axis of an octahedron passes through the centre of a pair of opposite faces and each ud mirror
plane bisects four faces, facts that are apparent in the characters generated by the transformations of the
eight faces of an octahedron under the 48 operations of the Oh point group:
E 6C4 3C2 6C2 8C3 i 6S4 3uh Gad BS6
800020004 0
= All +A2u + T1u + T2t
Similarly, each pair of opposite octahedral edges are bisected by a c; axis; these edges are contained in a
ah mirror plane and are perpendicular to a u0 , leading to the reducible representation
E 6C4 3C2 6C2 8C3 i 6S4 3uh 6ud 8S6
12 0 0 2 0 0 0 4 2 0
= A1.g + Eg + TlLJ + T2g + T2u
were able to relate bonding orbitals either to tetrahedron edges (Section

15.2.1) or to tetrahedral faces (Section 15.2.2), but not both. For transition
metal clusters it is necessary to invoke both simultaneously. Further, as we
saw in Section 15.2.2, there is the possibility of delocalization stabilizing
what, locally, are antibonding orbitals- so that they may become occupied.
This general approach is commonly referred to as the topological equivalent
orbital (TEO) model. As a further example of its application we consider
the two cluster halide species [Mo 6 CI8 ] 4 + and [Ta 6 CI 12] 2+, shown in
Fig. 15.15, both occurring in solids made by reduction of a higher
chloride (so, [Mo 6 CI 8 ] 4 + occurs in MoCI 2 , a compound which is actually
[Mo 6 CI8 t +·4CI - ). In [Mo 6 CI8 ] 4 + the chloride ligands are centred one
above each octahedral face- and so each is involved in a bonding with a
face localized orbital. The metal- metal bonding involves localized orbitals
which are in a 1:1 correspondence with octahedron edges. It is a simple
[ Ta6 Cl12] 2+ matter to show (Table 15.8) that the 12 octahedral edges span the A 1 • + T 1 u +
E9 + T 29 + T 2 ., irreducible representations of Oh and so these are the
Fl•. 16.15 The octahedral clusters [Mo6 C18 ] " symmetries of the metal- metal bonding molecular orbitals (a conclusion
and [Ta6 CI1 2 ] 2 + (there are several examples
known of each type). In both cases metal generally supported by more detailed calculations). Use of Table 15.8 also
atoms are black and chlorines are white circles. shows that the octahedral faces span the A 19 + T 1 u + T 2 9 + A 2 u irreducible
Topological models 1 361
representations, so these are the symmetries of the chlorine-cluster u

delocalized molecular orbitals, a conclusion similarly supported by detailed
calculations. In [Ta 6 Cl 12 ] 2 +, where the chlorines are located above octa-
hedron edges, the pattern is reversed. The metal-metal delocalized bonding
molecular orbitals, associated with octahedron faces, are of A,,+ T 1u +
T 29 + A 2 u symmetries and the chlorine-cluster u delocalized molecular
orbitals of A 1 , + T 1 " + E 9 + T 2 , + T 2 u symmetries. Electron counts on the
two molecules indicate that they are both, efTectively, 40-clectron systems
with all of the orbitals listed above-2A 19 + 2T 1 u + E, + 2T2 , + T 2 u + A 2 u
-being filled with either metal electrons or chlorine.
Although a simple topological model works for the cases just considered,
it has to be recognized that these are high-symmetry cases and, consequently,
contain symmetry-enforced degeneracies. For low-symmetry molecules (which,
for comparison, it is often simplest to think of as an isomer of a high-
symmetry case whenever possible) the degeneracies will be split and a scatter
of energy levels will occur. In such cases, a component of a (high-symmetry)
weakly bonding orbital may become non-bonding or antibonding; similarly,
a low-lying (high-symmetry) antibonding orbital may become non-bonding
or bonding. Equally, the basis becomes unclear-just how far apart do two
metal atoms have to be before one no longer associates an edge bond with
them? So, ambiguities weaken the immediate applicability of the model.
Nonetheless, the approach remains useful and, in particular, enables face-
and edge-bonded ligands to be incorporated within a general discussion of
metal-metal bonding.
Another aspect of topology which has been studied and which is
not too different, at the beginning at least, from that which has just
been discussed, is graph theoretical. Essentially, one draws up a table, a
matrix, in which connectivities are listed. Each cluster vertex atom is listed
across the top and down the side of the matrix; in the matrix itself a 1 is
entered whenever atoms are connected and 0 when they are not; the
mathematical properties of the matrix are then considered. If, for example,
an additional atom, an additional vertex, is added to a cluster then the size
of the matrix increases and a modified connectivity pattern is represented.
Comparison of the 'before' and 'after' patterns potentially provides insights
into whether such an addition is likely. Although much work has been
done, the present impact of the method is not great. This is largely because
the precise link between the properties of a matrix and properties of
the corresponding molecule is not always clear. However, it is likely
that the situation will improve and that the method will assume a greater
importance.
One final general conclusion which results from simple topological
models-but which reappears in the more complicated models considered
in the next section-is to be noted. In a topological model, similar bonds
are regarded as equivalent. So, if molecular stability is to be maximized, the
more bonds the better. Consider an array of three atoms. If they form a
linear chain, M-M-M, there are just two bonds. If they are arranged at the
corners of a triangle, then there are three. So, we would expect to find that
molecular cages will be built of triangular faces, and, indeed, triangular faces
are found to be by far the most common-square faces. for example, are
relatively uncommon.
15.5 Free-electron models

One example of the use of something rather akin to a free-electron model
(a model in which the electrons are free to wander over some surface, a
surface which is free from nuclei and the complications of their associated
point charges), has already been met in Section 10.2, where metal-·fullerene
complexes were considered. This partial duplication of the discussion makes
available to the reader an alternative (and simpler) treatment of much the
same material as that which follows. It was felt that this would be helpful
since the theory to be developed can pose conceptual difficulties for the
student.
One of the problems encountered by the newcomer to bonding in cluster
molecules is a change in language. Although the labels 'a' and 'rr' are
encountered-the former was used in the last section-they are rather rare.
The reason is that they are often not helpful. Figure 15.16 shows an
octahedron of metal atoms, together with p orbitals on three of them.
Between one pair of p orbitals the interaction is largely a, although it also
has a rr component, between another pair it is pure rr and between the third
pair it is non-bonding! Although in an octahedron, application of group
theory would help simplify the problem (in the symmetry-adapted linear
combinations that result from its use, the non-bonding pattern is not found),
the basic ambiguity problem remains, and becomes worse in low-symmetry
molecules. Bonding interactions often cannot be spoken of as simple a or
rr. In the present section, in particular, such labels will be little used to
describe bonding interactions between atoms, although they will be much
used in a different way, to differentiate between the orbital types of any one
atom.
In closo, complete cage, structures all of the atoms of the cluster that are
bonded to each other lie approximately, and often accurately, equidistant
from the centre of the molecule. That is, they lie on the surface of a sphere.
Electrons which are responsible for cluster bonding will lie between the
bonded atoms and, so, approximately, will be largely located close to the
surface of the sphere. Now, if the sort of orbital which can lie on the surface
of a sphere has any particular characteristics, these characteristics are likely
to be reflected in the bonding of the cluster atoms. Indeed, as a first
Fig. 15.16 Examples of the variety of

interactions that can occur between adjacent
n(tangential) orbitals in a tetrahedron.
Free-electron models 1 363
Fig. 15.17 The projection of the atomic

orbitals at the centre of a sphere onto the
surface of the sphere: (a) a p orbital, (b) a d
orbital. In each case a corresponding orbital of
an octahedral cluster is shown: (a) a t 1 u,
(b) an e,.
approximation, we might ignore the cluster atoms altogether and look at

the wavefunctions of a free electron on a sphere. Later, of course, we would
have to add the cluster atoms, whereupon the symmetry would drop from
spherical to that of the cluster and there would be a splitting of some of the
high degeneracies which are associated with spherical symmetry. This is the
approach which will now be developed. Now, an isolated hydrogen atom
has spherical symmetry and so we would expect some sort of connection
between the possible wavefunctions of the electrons of a hydrogen atom-
orbitals-and the wavefunctions of an electron free to move over the surface
of a sphere. Our expectation is justified. There is a I: 1 relationship between
the two. In Fig. 15.17 is shown, schematically, how the projection of an
atomic orbital onto the surface of a sphere enables one to obtain 'free
electron on a sphere' functions from those of the corresponding atomic
orbital. The degeneracies of the hydrogen orbital functions, 1(s), 3(p), 5(d),
7(f) etc. are also the degeneracies of the free electron on a sphere functions.
To determine how these degeneracies are reduced by the presence of the
cluster atoms data such as that in Table 7.6 can be used (replacing upper
case symbols by lower case). However, it has been found possible to make
considerable progress even if this apparently essential step is omitted!
This progress is dependent on the recognition that there is an enormous
weakness in the free electron on a sphere model that has so far been
developed. To see this weakness, readers are asked to imagine themselves at

the centre of the sphere, looking out at the atoms lying on the sphere which
form the cluster molecule. The atomic orbitals of each atom are classified
from this viewpoint. A common classification is thus used for all of the cluster
atoms, which is clearly an advantage. Further, at each atom, the surface of
the sphere is perpendicular to our direction of vision. This means that our
classification is particularly appropriate to a 'surface of a sphere' model.
From this viewpoint, the atomic s, p and d orbitals are either u in type (s, p,
or d,,, where we take our axis of view as the z axis appropriate to the atom
at which we are looking), n (px, p,; dw d,,) orb (dxy, dx'-y')-and this is
the meaning that will be attached to the labels u, n and (i for the remainder
of this section. It is for the latter two types that our model is deficient. They
are characterized, from our viewpoint, by having nodal planes which contain
the atom. Now, in our model, the cage atoms are scattered, usually in a fairly
regular way, over the surface of the sphere. This relationship provides no
guarantee that the electron-on-a-sphere wavefunctions such as those shown
in Fig. 15.17 will have the right number of nodal planes in the right places
to conform with the inherent nodal plane requirements of n and (i type
orbitals. The wavefunctions of Fig. 15.17 are only applicable to u orbitals,
with no inherent nodal requirements perpendicular to the surface of the
sphere. To be able to proceed we need functions akin to those of Fig. 15.17
but which conform to the nodal requirements of n and (i orbitals on atoms
placed at any point on the surface of the sphere. Such functions have been
calculated by Stone, and carry the rather off-putting name of tensor surface
harmonics. We shall give pictures of some of them shortly, but first turn to
a point which will become of key importance.
The point is the rather obvious one that, on each atom, each n set has
two members-px and pY, for example, as too does each (i set. Members of
a n set are related to each other by a 90° rotation about the z axis
(centre-of-sphere to metal atom); members of a (i set by a 45° rotation about
the same axis. To see the significance of this it is simplest to consider a
particular example. It turns out that there is one that we have already
considered in sufficient detail. In Chapter 6 (Section 6.2.2) we discussed the
ligand n orbitals in an octahedral metal complex, ML 6 . If we remove the
metal we are left with the n orbitals, defined appropriately for the present
application, that of a M 6 octahedral cage. There are two n orbitals on each
M, a total of twelve. They transform as T 19 + T 1 • + T 2 • + T 2 u under the
operations of the Oh point group and representative examples of the first
three were given in Figs. 6.13(a), 6.13(b) and 6.16 (the T 2 • can be obtained
from the T 1• by changing the phases of two trans Pn orbitals). If the reader
refers to these figures, reproduced and modified as Fig. 15.18, he or she will
be able to see an interesting point related to the 90° rotation that interconverts
the two separate Pn orbitals. Consider the t 1• combination of Fig. 15.18. If
each Pn orbital in it is rotated 90° in the same sense (clockwise or
anticlockwise, about the centre-of-sphere to metal axis) then the t 19 combi-
nation results (possibly with all signs changed). The t 29 and t 2 • combinations
are similarly interconverted. The symmetry-adapted combinations of Pn
orbitals all occur in pairs, t 1 • +-+ t 1 • and t 2 u +-+ t 29 • Now, the interactions
between adjacent orbitals in the t 1• combination is in-phase and so bonding.
Because of this, the interactions between adjacent orbitals in the paired, t 19 ,
Free-electron models I 365
Fig. 15.18 Rotation in the same sense by 90'

(indicated by small arrows in the upper
diagrams) converts a (bonding) t1 u combination
into an (antibonding) t:~g. Conversely, an
(antibonding) t2 " is rotated into a (bonding) t:~g.
combination are anti bonding. Similarly, t 2 • is bonding but, as a consequence,

t 2 u is antibonding. Not only do the symmetry-adapted combinations appear
in pairs, but one member is bonding and the other antibonding. One should
not argue too much from a single example-and, after all, we are placing
no restrictions on the position of atoms on the surface of the sphere-but
it transpires that this conclusion is general, applying to rc, b and all higher
functions. There is a pairing, such that for every symmetry-adapted combi-
nation that is bonding there is another which is antibonding, unless
rather special circumstances-which will immediately become apparent-exist.
To illustrate the insight that this pairing pattern provides we consider
the tetrahedron. Here, as has been seen (Section 15.2.3) the metal rc orbitals
transform as T 1 + T 2 + E. Clearly, the T 1 +-+ T 2 sets can be (and are)
interrelated by 90° rotations, but what of the E? Here, it is the two
separate E functions that are related one to another by a set of 90° rotations,
as shown in Fig. 15.19. But, since the two E functions are symmetry-required
to be degenerate, it is ridiculous to think of one as bonding and the other
as its antibonding counterpart. The only way out is for them, as a pair, to
be non-bonding. In a real-life molecule the E functions are unlikely to be
exactly non-bonding but, rather, slightly bonding or slightly antibonding.
We can at once understand why sometimes theE set is empty (as in B4 Cl 4 )
and sometimes full (as in P4 and Ir4 (CO)n).
So far we have proceeded by means of examples. What of the general
case? As far as the 'from the centre of the sphere defined' rr orbitals are
concerned, the 'free electron on a sphere' functions are simply projections of
atomic orbitals, like those shown in Fig. 15.17. In energy order sequence,
which is just the order of nodal counts, they are
S.(O), P.(1), D.(2), F.(3), ...
(b)
Fig. 15.19 (a) An e orbital combination which Here, S, P and D etc. have their usual meanings but (hopefully), for clarity,
transforms like the d,, of a (hypothetical) atom the number of inherent nodal planes have been added in brackets (the
at the centre of the tetrahedron. It is viewed
perpendicular to the z axis. (b) The same
number is of those that are perpendicular to the surface of the sphere). Again,
conbination but viewed along z. The direction of as usual, the degeneracies are given by (2L + 1), where L is the total number
rotation (see the text) is indicated by small of nodal planes. For each orbital on a cluster atom that is u-like when viewed
arrows. This rotation, by 90', gives (c), a
combination which transforms as d?-r2 of an
from the centre of the sphere, the above sequence holds. So, for each cluster
atom at the centre of the tetrahedron. atom, which we shall take to be a first row transition metal, it holds
separately for 4s, 4pz and 3dz, (z being the axis from the centre of the
polyhedron to the atom). As we have seen, then orbitals (4Pxo 4p, and 3dw
3d,.zl are both more complicated and interesting. In particular, they have
the pairing properties discussed above. Because the n orbitals are inherently
nodal they can never give a node-free pattern. Nodes can be added but not
lost. Now, n orbitals are rather like vectors (a clear analogy can be seen
between a EB-8 pattern and the symbol <--) and so we shall use a vector
arrow representation in our diagrams. It is important to recognize that on the
surface of a sphere, as on any other surface, the vectors in an array do not
have to be either colinear or parallel; they can be rotated relative to each
other. In our case this freedom allows the members of a vector set to turn
round-for one member to point in a slightly different direction to another,
to rotate about an axis-without the intervention of a nodal plane.
Simultaneously, the relative amplitudes of adjacent vectors can vary, reflected
in the size of the arrows, so that pictures of the tensor surface harmonics
tend to appear frighteningly complicated (but are not, unless the reader
panics). Taking just one member of each pair of p, orbitals the following
are spanned (again including the number of nodal planes); note the way that,
because we are dealing with n functions, the label S does not appear
P,(1), 0,(2), F,(3), ...
where one node is inherent in each case (we are dealing with p, orbitals)
and the others are global. The degeneracy is again (2L + 1), where L includes
all nodes, be they inherent or not. In Fig. 15.20(a) we show the three members
of the P,(l) surface tensor harmonic. To make these more understandable,
against each left-hand diagram (which applies to all polyhedral clusters, no
Free-electron models I 367
Fig. 15.20 (a) The P.(l) set of tensor surface

harmonics together with (right) their application
to an octahedral cluster.
matter how many atoms they contain) is, on the right, given its application
to the octahedral case. We shall discuss this latter case in more detail shortly.
For the moment, before continuing further in this chapter, the reader should
carefully study the relationship between corresponding left- and right-hand
diagrams in Fig. 15.20(a); it may be helpful to look back at Fig. 15.17.
The corresponding, paired, rotated rr tensor surface harmonics are
conventionally distinguished from the original by the addition of a bar above
each symbol and so are
.0.(1l; o,(2l; F,(3l, ...
Again, degeneracies are (2L + 1). In Fig.l5.20(b) is shown the P,(l) set and,
again, at the right, the corresponding octahedral orbitals. For both parts of
Fig. 5.20 on the right-hand side it is p, orbitals of the octahedral cluster
which have been shown~d, could have been used as they cover similar sets.
but the pictures would have been more complicated. In Fig. 15.20 note the
way that for each vector (arrow) diagram in Fig. 15.20(a). in the corre-
sponding diagram in Fig. 15.20(b) the vectors (arrows) are rotated by 90'.
As has been noted above, concomitantly, the octahedral r 1, set becomes r 1".
A similar pattern is shown in Figs. 15.2l(a) and (b). which show the D.(2)
Fig. 15.20 (continued) (b) The i'.(1) set

of tensor surface harmonics. Notice that
each arrow is perpendicular to that in the Vy1
corresponding P.(1) function. To the right is
shown the octahedral cluster application. Again,
orbitals are 90" rotations of those in (a).
and the D.(2) functions. In both Figs. 15.20 and 15.21 approximate Cartesian
labels for the individual surface tensor harmonics are given, although there
is no accepted convention for these labels.
To complete the picture, b orbitals (d.,_,,, d.,) on the cluster atoms
have to be included. Again, there is an original-rotated set separation (but,
for these functions the rotation needed to interconvert the two sets is 45°).
Because (j orbitals are characterized by two nodal planes perpendicular to
the surface of the sphere, we cannot have combinations with fewer. The
combinations are therefore
o.(2). t=;,(3l, GMl •...

and
o.(2). F.,(3), ii.(4> •...

with degeneracies of (2L + 1), all nodes, local and global, being included in
the count. Of the above sets, in Fig. 15.22 only representative D6 (2) and D6(2)
combinations are shown, along with corresponding orbitals for an octahedral
cluster. In Fig. 15.22 the initial T used in the labelling of the functions may
be thought of as a distorted form of + (the lobe pattern of a (j orbital) but
eg
Fig. 15.21 (a) The 0,(1) set of tensor surface

harmonics together with, where appropriate
(light) their application to an octahedral cluster.
Vyz,
eg
Vxyu tl I y
X
)Z~
Fig. 15.21 (continued) (b) The corresponding

51t(1) set of surface tensor harmonics and
(right) their application to an octahedral cluster.
Note the 90" rotation patterns between
corresponding 0"(1) and 0"(1) functions.
X
z
Fig. 15.22 Representative (a) 0,(2) and

(b) 0,(2) tensor surface harmonics together with
(right) their application to an octahedral cluster.
The Cartesian labels given refer to the axes of
the first diagram. However, the orbitals at each
cluster atom refer to local axes. In each case, z
local z is radial (i.e. directed outwards from the
centre of gravity of the octahedron) and local x
andy perpendicular to z. These local axes are
taken to be parallel to the axes in the first
diagram, although axis labels are not necessarily
maintained. So, a local x may be parallel to
global z. This pattern means that the eg and t2 u
octahedral functions are based on local d,2 -?
orbitals, the eu and t2 g on local dXY (see also
Table 15.10).
actually is a shorthand for tensor, just as V was for vector in Figs. 15.20
and 15.21. Beware of confusion: all the functions under discussion in this
section are collectively called 'tensor surface harmonics' but the name
'tensor' also reappears to describe the specific functions with the T label.
Although, within this model, all functions of a given type, the seven F0 (3)
for example, are treated as degenerate, not all of them may exist for a given
cluster. The actual symmetries spanned by the a, :n: and {J orbitals on the
cluster atoms has to be worked out for each case. Correlations then have to
be made with the tensor surface harmonic functions given above, or rather
their components, in the symmetry of the actual complex. Let us look at a
specific example. In Figs. 15.20 and 15.21 an octahedral cluster was used as
an example and we will continue with this choice. In doing so we will tidy
up some loose ends left at the end of our topological equivalent orbital
description of [Mo 6 CI 8 ] 4 + and [Ta 6 Cl 12] 2 +. To this end two things have
to be done. First, the general discussion has to be applied to the octahedral
case. Second, the symmetry properties of the a, :n: and {J orbitals on the
transition metal atoms have to be determined. The first of these tasks is not
difficult. The data in Table 7.6 has to be applied to the spherical tensor
harmonic functions listed above, although our notation must first be
extended to distinguish between the different sets arising from the various
metal orbitals of :n: symmetry, for instance. This is easy. Instead of talking
about D.(2), for example, we will now call this D~, v:or D~, where we
have added p, d or f to indicate the sort of orbitals on the metal atoms
which are involved (although not much use of D~ is to be expected!) and
compensated by dropping the (2). The results are given in Table 15.9, which
is quite horrendous in appearance. This is because it applies to all octahedral
clusters, including octahedral clusters with dozens of metal atoms. We are
only concerned with the simplest octahedral cluster and so only a small
amount of the information in Table 15.9 is relevant. However, the complete
Table 15.9 has a basic pattern which both makes it worthy of inclusion and
of some study. The second thing which we have to do, the determination of
the symmetry pattern of the a, :n: and {J orbitals, is not too difficult either. We
have already discussed in Sections 6.2.1 and 6.2.2 the a and :n: bonding
associated with an octahedral array of atoms. The fact that it is ligand
orbitals which are discussed in these sections is irrelevant-transformation
properties do not depend on the chemical nature of an atom and so the
results needed are those given there. The discussion has, however, to be
extended to include {J functions. There are two of these on each atom, dx'-y'
and dxy• using the coordinate axes defined in the caption to Fig. 15.22. No
Table 15.9 The symmetry properties of the surface tensor harmonics in the
octahedral group Oh
Tensor surface harmonic Corresponding rotated-basis

tensor surface harmonic
S,(O) [S:, S~, S~]: A1g

P,(l) [P:. P~. P~]: T,
0,(2) [D:, D~, D~]: T2g + Eg
P,(l) [P~, ~. P~]: T, P,(l) [P~. P~. P~]: T1g
0,(2) [D~. D~. D~]: T2 g + Eg 0,(2) [0~, 0~, 0~]: T2u + Eu
F,.(3) [P., ~. P.J: A2u + T1u + T,u F,(3) [~. F~. f;:]: A2g + T1g + T2g
D,(2) [D~, D), ... ]: T2g + Eg 0,(2) [0~, 0), ... ]: T2u + Eu
F,(3) [Ff, f), ... ]: A2u + T1u + T2u F,(3) [Ff, P,, ... ]: A,.+ T1g + r,.
G,(4) [G~, G), ... ]: A1g + Eg + T1g + T2g G,(4) [G~, G), ...]: A1u + Eu + T1u + T2u
symmetry operation relates members of the dx'-y' set to members of the

Table 15.10 Transformation properties dxy• so they can be considered separately. The collected results are given in
of metal orbitals in an M6 octahedron
(see the caption to Fig. 15.22 for a Table 15.10.
definition of x and y axes) Some progress can be made by bringing Tables 15.9 and 15.10 together.
Only entries in Table 15.9 which appear in Table 15.10 are to be considered.
Symmetry relative Symmetry species Thus, because in Table 15.10 the u basis contains no T 29 entry we can ignore
to an axis from spanned
the T 29 entry against D.(2) in Table 15.9. Further, because of the nodal
the octahedron
centre pattern sequence of these entries in Table 15.9, it can be concluded that the
energy level stability pattern of the orbitals in Table 15.10 is A 19 > T 1• > E 9 •
A1g + Eg + T1 u
We are making progress! To make more progress, only the d orbitals on
T1g+ T1u+ T211 + T2u
A211 + Eg + T2 u the six metal atoms will be considered. This simplifies the problem and is
A2u + Eu + T2g actually much less restrictive than it seems. Inclusion of metals and p orbitals
would simply repeat (at higher energies) symmetry species that we shall meet
with a d orbital-only basis. The way that we will handle the problem with
this latter basis is the way that we would have handled it also for s and p
functions.
Of the two species [Mo 6 CI 8] 4 + and [Ta 6 CI 12] 2 + shown in Fig. 15.15
only the former will be considered. Surprisingly, it is the more complicated
and the latter, the simpler, is set as a problem at the end of this chapter.
Actually, the discussion is the same as far as d. and dn orbitals are
concerned-both correspond to cylindrical electrical density distributions
about the local C4 axis at each metal atom. Unlike the d., the da functions
on each atom do not transform as a pair and so it is necessary to decide
which of the pair it is appropriate to use in a particular discussion. For
[Mo 6 Cl 8 ] 4 + the dx, (pointing towards the faces of the Mo 6 octahedron) are
clearly involved in Mo-CI bonding (the Cl atoms lie above these octahedron
faces). So, as our concern is the direct metal-metal bonding we shall only
include the dx'-y' functions in our discussion. This discussion is conveniently
developed from Fig. 15.23. To the left in this figure are indicated the six d.,
12 d. and six d 6(x 2 - y 2 ), in order of increasing nodality. To their right
are drawn the levels arising (Table 15.8) modified by the requirement that
only entries appearing in Table 15.9 are acceptable (the unacceptable are
indicated in brackets). The pictures of Figs. 15.20 and 15.21 have been used
as a guide in placing the bonding orbitals below their antibonding counter-
parts, the greatest stability coming from interactions with au component. So,
for the d. orbitals the T 29 level is placed below the T 1 • because the
former involves illixed u----n bonding between adjacent cluster atoms, whereas
the latter is n-only bonding (see Fig. 15.16). Finally, to the right of Fig. 15.23,
levels of the same symmetry are allowed to interact with each other. An
example of this is shown in Fig. 15.24 where the T2 .(n)-T2 .(o) interacting
orbitals are shown. It is clear that they overlap and that a bonding and an
anti bonding combination will result. This conclusion is general-interaction
between levels leads to the lower being stabilized and the higher being
destabilized. The energy level pattern, and orbital occupancy, for the
metal-metal bonding in the [Mo 6 Cl 8] 4 + cation shown at the right of Fig.
15.23 results. This pattern contains one point which we could not have
predicted-a crossing over between T 2 • and T~u levels such that the former
is the HOMO and the latter is the LUMO. Although such a crossing is
entirely consistent with the argument that has been developed, the strong
argument in its favour, confirmed by detailed calculations, is the fact that it
is predicted by the topological model developed in the previous section. Not

for the first time we find that different models complement each other. The
advantage of the spherical tensor model is its general applicability-it can
be expected to give insights into the cage bonding in any near-spherical
cluster species. Its disadvantage is the fact that it includes no atoms! It seems
that this disadvantage is surprisingly unimportant and that fine-tuning of
the sort used at the end of our discussion of the bonding in [Mo 6 CI 8 ] 4 +
adequately compensates.
We conclude this section by using the results obtained to justify the
Wade's rule count of (n + 1) skeletal electron pairs needed to bond a closo
cluster of n atoms. The justification rests on two features. First, that the n
orbital interactions give rise to two sets (we initially called them Pn(l ),
Dn(2), ... and Pn(l), Dn(2), ... ); when the members of one set are bonding, the
corresponding members of the other set are antibonding. Now the total
number of n orbitals is 2n, where n is the number of cluster atoms,
Detailed calculations 1 375
so we may reasonably anticipate that we have just n potentially bonding

orbitals derived from the n basis (these orbitals spanning subsets within the
general P., D, etc. functions available for the geometry of the particular
cluster). Comparison with the transformations of the O" (which give rise to S.,
P., D., ... ) shows that the only symmetry type within the latter which is not
present from then orbitals iss •. Because this is spherically symmetric it is an
in-phase combination of all of the O" basis functions and so is certainly
bonding and therefore occupied. All of the other symmetries arising from
the n set are found in the O"; within each pair with the same symmetry label
one will be bonding and one antibonding; similar arguments hold for the ii.
We have then, (n + 1) bonding pairs-the number required by Wade's rules.
It is worth dwelling on the reason why the P;, Dff etc. are unoccupied. As
just indicated, the answer is something which we recognized earlier when
talking of [Mo 6 CI 8 ] 4 + -that when interaction occurs between two orbitals
of the same symmetry species, the result is that one becomes more bonding
and one more anti bonding. So, what we have so far called P. bonding orbitals
will contain some P, component (and also D,) but the actual number of
bonding molecular orbitals is unchanged-it is (n + 1), and this is just
Wade's rule applied to a closed cage molecule. It is to be recognized that the
argument that has just been used is one that enables bonding orbitals to be
counted. It is not really an argument about bonding.
A partial check on the correctness of the approach used in obtaining the
(n + 1) bonding electron pair count, Wade's rule, is provided by the
Fig. 15.24 The T2 " combinations of n- and tetrahedral and octahedral molecules that have been considered in some
b-type functions that are shown as interacting
in Rg. 15.23. Note that, for simplicity, the detail in this chapter. If consideration is confined to cage-bonding orbitals,
n-type function shown is Pn; in fact, a d,. is the in no model was the same symmetry species spanned twice. This is an
one involved in Fig. 15.23. expression of the argument used above. Of course, in lower symmetry
molecules a similar statement could probably not be made but, even here,
if the relevant orbitals are correlated back to their counterparts in Figs. 15.18
and 15.22, it will almost always be found that the exclusion we have invoked
holds at the low-symmetry level. Further illustrations of the way that levels
of like symmetry interact so as to lead to only one bonding combination
will be found in the way that figures such as Fig. 15.4 combine with Fig.
15.11(a) to lead to a final pattern such as that in Fig. 15.10(b); similarly Figs.
15.7 and 15.11(a) also combine to give the second pattern in Fig. 15.11(b).
15.6 Detailed calculations

As has been mentioned already, the task of carrying out detailed molecular
orbital calculations on cluster compounds is difficult in the extreme. They
contain far too many atoms and far too many electrons! The common way
that has been followed to attempt to circumvent the problem is that of
carrying out a calculation on the metal cage (with no ligands) and also one
on the ligand cage (with no metals) and then bringing the two together. It
is interesting that two distinct approaches along these lines have developed-
one coming from physics and the other from chemistry (that of this chapter).
The physicists have been interested in the answer to questions such as: how
small can a cluster of metal atoms be and yet show metallic properties? For
instance, how many iron atoms are needed before they show magnetic
properties akin to those of bulk iron? The answer to such questions seems to
Metal
en
8
en
8
Fig. 15.25 Density of state (DOS) energy
patterns for (a) vanadium metal and V6 (b)
silver metal and A&;. Adapted and reproduced
with pemnission from G. Seifert and H. Eschrig
Phys. Stat. So/. (b) (1985) 127, 573.
A&;
cluster
en
en 8
8
Energy- Energy-
(a) (b)
depend on the quantity used as a criterion-anything from four (electrical

conductivity) to a hundred (ionization energy). It has been commented that
'the solid does not emerge all at once, but like the Cheshire cat, it fades into
view slowly, with the smile appearing first'. The physicists have used models
rejoicing in names such as 'the droplet model' and 'the jellium model'-
names which are reasonably self-explanatory and which find experimental
justification in the metal-atom mobility within some clusters observed by
(metal atom) NMR. For the chemist, there is one problem with this
approach. Because in bulk metals there are so many energy levels one gives
up all attempts to list them and talks instead of density of states (DOS); so,
for small metal clusters the same language is often adopted by physicists. As
a result, the chemist may well find when reading the physics literature that
no use is made of the symmetry labels used throughout this chapter. Figure
15.25 shows a comparison of DOS patterns for bulk metal and octahedral
clusters, calculated using the X<X method. Although there are clear differences,
the gross similarities are perhaps more evident.
Assuming that most bare metal clusters can be taken to have some
metal-like properties, the next question is the extent to which these properties
persist in metal cluster compounds. We are immediately plunged into an
area of controversy. Whilst no ferromagnetic cluster compounds of iron are
known (although some osmium cluster complexes display paramagnetism)
this is temperature independent paramagnetism (TIP) in origin- see Section
9.3); this is scarcely relevant to the real issue. In bulk transition metals almost
all of the cohesive energy-the bonding energy-comes from interaction
between d orbitals. The s and p have little direct involvement. Is this also
true of cluster compounds? Extended Hiickel calculations tend to say no
while X<X (and the related, but better, density functional) method tends
Detailed calculations 1 377
oc,,c
0 0
c
I_;CO
M M
FJC. 1&.26 The D3h structure of M3 (C0) 12 •
M = Ru or Os (the iron compound has a
oc,..,I~O/''co
c c c
different, but related, structure which is 0 I 0
oc,.., t'co
discussed in Section 12.5.3). M
towards yes. Both, speaking from their own corner, offer explanations of
Wade's rules. Does this matter? Quite possibly not; it depends on the detail
one wishes to explain. So, for the compounds Os 3 (C0) 12 and Ru 3 (C0) 12
(shown in Fig. 15.26) the Xoc method gives the sequence of highest occupied
molecular orbitals (in D 30 symmetry) as
a;, < a2 < e" < a;, < e" < e' < e' < a;,
whereas extended Hiickel gives the same set in a different order
a;, < a2 < e' < e" <a;, < e' < e" < a;,
Of these, the highest e' and a~ are largely involved in metal-metal bonding
(note that these are also the irreducible representations generated by the
M-M bonds between the three M(C0)4 groups in M 3 (C0) 12 ). In the
photoelectron spectra (that of Os 3 ( CO) 12 is shown in Fig. 15.27) ionization
from the metal-metal bonding orbitals is associated with the lowest energy
peak, arrowed in Fig. 15.27. Ionization from the other levels listed above
accounts for the two peaks on the immediate right of that arrowed; the
orbitals involved are predominantly d, with significant CO bonding involve-
ment. The intense peaks with energies greater than 12 eV in Fig. 15.27 are
largely associated with ionizations from the CO groups. Looking at spectra
such as that in Fig. 15.27, which is for a small, high-symmetry cluster,
together with the orbital energy sequences listed above, one can understand
the attractions of a density-of-states approach! However, if one wishes to
interpret electronic spectra then a detailed knowledge of the occupied and
lowest unoccupied orbitals becomes essential.
Because of the inherent problems associated with calculations on large
clusters, most of the detailed calculations that have been performed concern
diatomic systems such as the [Re 2 CI 8Y- anion. This anion has a short
metal-metal bond and the two (ReCI 4 ) units are eclipsed (Fig. 15.28(a)).
Since the short metal-metal bond can only increase steric interactions
between the two sets of Cl4 ligands, a staggered configuration might
confidently be predicted! The only evident explanation for the observed
6 12 18 eclipsed arrangement is the presence of significant [J bonding between the
Ionization potential (ev) - two rhenium atoms (Fig. 15.28(b)). Does it occur? The answer given by the
FIJI. 1&.27 The photoelectron spectrum of best calculations (usually Xoc) is yes ... but! The 'but' occurs because of a
Os3 (C0),2 . Adapted and reproduced with problem we have met several times in this book-electron correlation.
pennission from J. C. Green, D. M. P. Minges
and E. A. Seddon, lnorg. Chern. (1981) 20, Indeed, one of the workers in the field has written an article entitled
2595. 'Problems in the theoretical description of metal-metal multiple bonds or
how I learned to hate the electron correlation problem' (see the Further
Reading at the end of the chapter). Essentially, electron repulsion forces
electron density from bonding orbitals, which may be a, nor b, into low-lying
antibonding orbitals, which may be a*, n* or b* The bonding orbital
occupancy is less than given by the simple model and the antibonding orbital
2.24A occupancy greater. Both vary with change in metal-metal internuclear
distance. Larger bond lengths usually mean a greater occupation of anti-
bonding orbitals (in the limit, where the bond breaks, bonding and anti-
~Re~,---CI bonding orbitals are equally occupied). This can be placed in some perspective
Cl,......... Cl by the outcome of some calculations that have been carried out on the bare
(a) Mo 2 molecule. It might reasonably be assumed that the 12 valence electrons
in the molecule might be distributed over the two a bonding molecular
orbitals (one originating in the s and one in the dz' orbital on each atom),
the two (degenerate) rr, d orbital in origin and the two b (degenerate). The
observed bond length in the gaseous molecule is 1.93 A; in principle a
sextuple bond could be involved! In fact, it was only with considerable
difficulty that the calculations could be induced to show any bonding. They
did so only after inclusion of considerable electron correlation. The absence
of ligands does not seem to be the vital factor; in calculations on the dimeric
diamagnetic molecule chromium(II) acetate the presence of the ligands did
not give results greatly different from those on Cri+; again, a correlation
energy contribution was required to give bonding. Note particularly that, as
has been indicated above, the correlation energy contribution is expected to
(b) increase with increase in metal-metal distance. Yet further perspective can
be added to this by the recognition that the metal-metal n bonding overlap
Fig. 15.28 (a) The eclipsed structure of seems invariably less than the carbon-carbon n overlap in benzene-and
[Re 2Cid] 2 -. (b) The Re-Re ii bond which would
be broken were the anion to adopt the the metal-metal n and c5 are presumably yet smaller. It has even been
sterically-preferred staggered configuration. In claimed, on theoretical grounds, that a Cr-Cr quadruple bond is to be
this case the Re-Re bond is short (2.24 A) and expected to be as weak as a Cr-Cr single bond if there are no bridging
so electron correlation relatively unimportant
(see text). ligands in the molecule (only one compound with a Cr-Cr bond but without
bridging ligands is known). The spectroscopic evidence is unambiguous.
In [Re 2 Cl 8 ] 2 - and related anions the electronic transition from the
b bonding to the b antibonding orbital lies at about 7000 A, an energy
corresponding to a fairly typical crystal field splitting (Chapter 8). So,
returning to larger clusters, where the bond lengths are much greater than
in species such as [Re 2 Cl 8 ] 2 -, must we expect electron correlation to be of
any importance? It seems that the answer is yes, although the delocalized
nature of most orbitals, a phenomenon which means that electron density
is spread out and so the effects of repulsion reduced, offers hope that the
importance is limited. Nonetheless, it is the difficulty of including electron
correlation which is a limiting factor in carrying out accurate calculations
on cluster compounds. However, one must proceed with extreme caution.
Not only is the problem a difficult one but it has been found that agreement
with experiment is not invariably a reliable guide to the accuracy of a
calculation. So, as has been mentioned in Chapter I 0, there are cases known
in which a calculation gives good apparent agreement with experiment-a
bond length is accurately predicted, for instance-but, when the calculation
is improved, perhaps by the inclusion of more orbitals on some atoms or
the inclusion of more excited states (which can interact with and thus modify,
the ground state) the apparent good agreement is lost. This is not all. There
Clusters and catalysis, a comment 1 379
is the problem of relativistic effects. These have briefly been mentioned

before (Section 11.12, their origin is outlined in the footnote to page 243);
they are particularly important for the heavier elements and particularly for
Pt, Au and Hg. So, it has been calculated that in the bare Pt 3 equilateral
triangle the non-relativistic bond energy is 7.1 kJ mol- 1 (1.7 kcal mol- 1 )
but that the relativistic is 36.4 kJ mol- 1 (8.7 kcal mol- 1 ); where the general
pattern is more to be noted than the particular energies. The origin of
the particular strength of the Hg-Hg bond in [Hg 2]2+ compounds is
almost certainly relativistic, particularly when electronegative atoms, such
as halogens, are attached. It seems that advances in our understanding of
the bonding in cluster compounds may involve the introduction of yet more
novel ideas.
15.7 Clusters and catalysis, a comment

Transition metals, either pure or in mixtures, are of wide industrial use
in catalytic systems. They are used as wire meshes or finely divided on a
support, the object being that of increasing surface area. Metal cluster
compounds, including many of those shown in Fig. 15.1, may often be
regarded, structurally, as a fragment of the bulk metal to which ligands are
attached. Further, the majority of the metal atoms are on the surface of the
cluster. Could not these atoms chemically resemble those on the surface of
the bulk metal? If so, could not cluster compounds be a rich source of
catalytic molecules? In that they also offer the possibility of fine-tuning by
appropriate choice of ligands-both the steric and the electronic properties
of the ligands can be varied-the prospect is attractive indeed. In practice,
whilst important cluster catalysts certainly exist, the high hopes have been
little fulfilled. Although the vibrational characteristics of ligands on clusters
have proved very similar to those of the same ligands absorbed on bulk
metal, catalytically, clusters have not modelled metal surfaces particularly
closely. Why? There are, no doubt, many reasons. The surfaces of (chemists')
clusters are usually saturated with ligands, metal surfaces often are not. Metal
surfaces have surfaces, edges, defects and corners which are not matched by
small clusters. However, we can also speculate on an electronic difference
affecting reaction between two coordinated molecules. We have seen that,
although the details remain a matter for debate, ligands significantly modify
the electronic properties of bare metal clusters. There are no known
paramagnetic cluster compounds of iron! In contrast, metallic iron, perhaps
dispersed on an alumina surface, remains paramagnetic despite the presence
of ligands. It seems possible that in the latter case, the reservoir of electrons
and electron energy levels associated with the bulk metal provides a damping
which enables individual ligands to undergo the gymnastics which are an
essential part of most catalytic reactions, relatively independently of one
another. For a small metal cluster, the gymnastics of one ligand have a
sufficient influence on the orbitals and energy levels of the entire cluster for
the gymnastics of a second ligand to be constrained. If this explanation is
correct, then complexes of a type which are just beginning to be made-
consisting of two or more concentric cages of metal atoms-are likely to be
more catalytically interesting, for, if it is meaningful to make the distinction,
the inner cages will surely have properties not far removed from those of the
bulk metal.
Further reading Transition-Metal Clusters and Related Compounds' by A. J.

Stone, Inorg. Chern. (1981) 20, 563.
A brief overview of cluster structure and, to some extent, The problem of electron correlation is discussed in 'Prob-
bonding, is contained in Metal clusters revisited, by J. Lewis, lems in the Theoretical Description of Metal-Metal Multiple
Chemistry in Britain (1988) 795. A more complete picture is Bonds or How I Learned to Hate the Electron Correlation
given in The Chemistry of Metal Cluster Complexes by D. F. Problem', M. B. Hall. Polyhedron (1987) 6, 679.
Shriver, H. D. Kaesz and R. D. Adams (Eds.), VCH, New A brief review of graph-theoretical ideas is to be found in
York, 1990. 'Mathematical methods in coordination chemistry: topological
A review of bonding theories, including those mentioned in and graph-theoretical ideas in the study of metal clusters and
the early part of this chapter, is contained in 'Theoretical polyhedral isomerizations' by R. B. King, Coord. Chern. Rev.
Models of Cluster Bonding' by D. M. P. Mingos and R. L. (1993) 122, 91.
Johnson, Struct. Bonding (1987) 68, 29. See also 'Bonding A book that covers both experiment and theory is Multiple
Models for Ligated and Bare Clusters' by D. M. P. Mingos, Bonds Between Metal Atoms, 2nd edn., by F. A. Cotton and
T. Slee and L. Zhenyang, Chern. Rev. (1990) 90, 383. R. A. Walton, Oxford University Press, Oxford, 1993.
Wade theory is detailed in Transition Metal Clusters B. F. G. A useful source is Volume I of Comprehensive Coordination
Johnson (Ed.), Wiley, 1980, Chapter 3 'Some Bonding Con- Chemistry G. Wilkinson, R. D. Gillard and J. A. McCleverty
siderations', by K. Wade. (Eds.), Pergamon Press, Oxford, 1987, Chapter 4 'Clusters and
Stone theory is described in 'New Approach to Bonding in Cages', by I. G. Dance).
Questions 15.4 To what extent do the molecules shown in Fig. 15.12

conform to the predictions of Wade's rules?
15.1 'Insights into cluster bonding come more readily from 15.5 Using [Mo 6 Cl 8 ]4+ as an example, apply Stone's surface
an approach in which the details of atomic structure are not tensor harmonic model to the species [Ta 6 Cl 12 ] ' +. To what
given prominence.' Comment on the validity of this statement. extent does the topological equivalent orbital model facilitate
15.2 In the compound Rh 6 (C0) 16 (Fig. 15.1(c)) the rhodium the application?
atoms have 0, (octahedral) symmetry but the CO groups 15.6 Review those areas in coordination chemistry in which
reduce the symmetry to T, (tetrahedral). Outline and compare it seems electron correlation is important for a detailed under-
the pictures of the cluster bonding in the O, and T, point standing (use the index at the end of the book).
groups.
15.7 In Section 15.4, [Mo 6 Cl 8 ]4+ and [Ta 6 Cl 12 ]2+ were
15.3 The molecule B4 Cl 4 exists but Al 4 Cl 4 does not. The presented as 40-electron systems although in Table 15.6 such
molecule N 4 does not exist but P4 does. Use these examples to octahedral species were associated with 86-electron counts.
highlight the limitations of our present qualitative under- Explore the rapprochement between these two numbers. (Hint:
standing of bonding in cluster species. consider the symmetry implications of the 86-electron count.)
Some aspects of bioinorganic
chemistry
16.1 Introduction
Coordination compounds are central to living processes. Although the
cations of Group I (Na +, K +) are usually thought of as very mobile, as
Section 5.7 shows, it is possible to have relatively stable complexes of
them when there is a matching between their size and that of a cavity in a
ligand; change the cavity size and the stability changes. Perhaps, in part,
this lies behind the function of these mobile cations in biology, for example
in the transmission of nerve impulses. Group 2 cations are also important,
presumably also through complexes that they form - Ca 2 + in physical
structures such as teeth, skeleton and shell, as well as functioning as a trigger
in neurotransmission, as a messenger in initiating hormone action and in
initiating blood clotting. The presence of Mg2+ in chlorophyll is well known .
Zn 2+ is present in many enzymes; sometimes it seems to have a structural
role, maintaining a protein structure in a particular conformation for
example, in others it is intimately involved in the chemistry. Complexes of
transition metal ions are of vital biological importance. They provide
mechanisms for electron storage, for electron transport and for catalysis-in
particular involving small molecules and ions such as 0 2, N 2, NO, NO:Jand
SO~-. Is it understandable that bioinorganic chemistry is currently one of
the fastest growing parts of inorganic chemistry 1 In the present chapter this
growth is not the focus of attention. Rather, attention is focused on the way
that the subject throws up fascinating problems which offer a great challenge
and require careful and, often, great ingenuity and subtle methodology to
solve. This, then, is a problem-solving chapter. We start with a general
overview of the subject which will provide a background for the specific
problems that will be considered.
1 A very readable paper which develops and enlarges the theme of this paragraph is 'The
Chemical Elements of Life" by R. J. P. Williams, J. Chern. Soc., Dalton Trans. (1991) 539.
382 1 Some aspects of bioinorganic chemistry
Bioinorganic chemistry is in large measure concerned with the study of

metal complexes in which one ligand (perhaps, the only ligand) is a large,
naturally occurring organic molecule~a protein, a sugar or a heme,
for example. There are many factors contributing to its growth in recent
years. First, X-ray crystallographic and related techniques (not least, the
art of obtaining single crystals of complicated bioinorganic molecules) have
developed to the point at which it has been possible to obtain accurate crystal
structure data on many, if not all, species. This means that the local structure,
composition and geometry around the metal ion is known for many
compounds. Secondly, spectroscopic methods of study have been developed
which are both sensitive and selective, probing the properties of, for instance,
the metal ion and its immediate environment, notwithstanding the low
concentration of this ion~ usually much less than 1% on a molar basis. The
present chapter will place some emphasis on a selection of these techniques.
Thirdly, it has been recognized that bioinorganic species play a key role in
the most important of biological reactions~respiration, energy transfer,
photosynthesis and nitrogen fixation to name but a few. Our understanding
of such processes is dependent on our first understanding the bioinorganic
chemistry involved. Fourthly, bioinorganic species are often remarkably
effective catalysts. Enzymes, and enzymatic reactions, are at the heart of
biochemistry. As an illustration of this area, consider the fact that at the
present time, the nitrogen fixation ability associated with bacteria present
on the roots of some plants~clovers, peas and beans, for example~cannot
be matched or mimicked in the laboratory. Perhaps one day it will be,
threatening Haber ammonia plants with closure and possiblf changing
the basis of agricultural production. Fifthly, it often turns out that in
bioinorganic species the metal ions have unusual coordination numbers
and/or geometries. One of the best known is the square planar coordination
of the Mg2+ ion in chlorophyll. This point is surely connected with the
fourth; it has stimulated attempts to make relatively simple inorganic
complexes having coordination patterns similar to those found in nature.
Such model compounds, when prepared, are likely to be much more readily
available in quantity and much cheaper than their naturally occurring
counterparts. However, they may well display some of the chemical properties
of the naturally occurring materials, catalytic properties, in particular. This
line of research has stimulated the synthesis of exotic ligands, mimicking
naturally occurring ligands in the local environment that they provide the
metal atom. Some of these ligands were met in Chapter 2. Sixthly, some
metal species have proved to be effective therapeutic agents. Whilst the
treatment of deficiency diseases (for instance treating anaemia, lack of iron,
with complexes of Fen or Fem) are important, perhaps best known is the
use of cisplatin, cis-[Pt(NH 3 hC1 2 ], in the treatment of some cancers.
Similarly, gold(lll) complexes, often with ligands bonded through sulfur,
are used in the treatment of some arthritic conditions. However, these drugs
can have bad side-effects~cisplatin, used alone, can have a destructive effect
on kidneys. Although these side-effects can be mitigated by administering
the metallic compound along with suitable additional, compensating,
2 'Possibly', because an alternative line of research is to incorporate into
wheat, for instance,
the gene involved with mediating nitrogen fixation, thus making the species self-sufficient and,
ultimately, removing the need for nitrogen-containing fertilizers in agriculture.
Introduction I 383
compounds and much water, these are only temporary expedients. Much
better would be the use of metal complexes without the side-effects. Although
the buck-shot approach (synthesize a lot of likely-looking compounds and
try them) can be used, the desired goal is more likely to be reached by a
detailed understanding of the mode of action of a drug, the in vivo
transformations that it undergoes and thus, ultimately, the nature of the
species that is actually leading to the desired effect. This calls for all the
apparatus of bioinorganic chemistry-the identification, isolation, handling
and study of biologically active species, the preparation of model compounds
and so on.
As already indicated, it is not the objective of the present chapter to
attempt to give either a comprehensive account of, or even an overview of,
bioinorganic chemistry. Rather, we shall explore the application of subject
matter in earlier chapters to the topic. From this viewpoint, bioinorganic
chemistry provides a series of novel applications of this subject matter.
Further, some advances in bioinorganic chemistry have. involved specialized
techniques and ideas which, to date, have been less applied to more
traditional areas of coordination chemistry; we shall look at some of these
also. In that such applications will surely become more commonplace,
bioinorganic systems provide an indication of the results that will be
obtained and of the information to be learned from them.
We start with four important points, some of which contain lessons in
their own right. First, although it seems that seldom, if ever, does a metal
ion occur in a biological molecule and not have a function, it should not be
thought that the sole function available is a catalytic one, although this aspect
is often emphasized. Quite the contrary, for instance, a number of cases are
known where the function of Zn 2 + is classified as structural. In some of
these, Zn 2 + forms rather strong bonds to sulfur ligands (usually from the
amino acid cysteine, HS-cH 2-cH(NH 2 }-COOH) which are part of a
protein amino acid sequence. By so doing, the protein chain, or several chains
together, are locked into a particular geometrical arrangement. It seems that
the resulting global, quaternary, structures are favourable to the action of
the protein (and this action could be a catalytic one) and are thus stabilized.
Other non-catalytic functions of metal ions include the transmission of
impulses along nerve fibres (each impulse is associated with a sudden change
from a 'high K + concentration inside the fibre, high Na + outside' pattern.
Immediately after the impulse, a pump mechanism restores the original
concentrations). Another function is that played by Ca 2 + as a messenger in
nerve action (from one nerve cell to the next) and in blood clotting. It seems
that Ca2+ serves such functions because the complexes it forms in these
situations are rather balanced-not too weak, not too strong; not too labile,
not too inert.
The second point is related to the first and echoes one that has already
been mentioned. Metal ions in biological systems are often in rather unusual
environments. This is not just because of the geometrical arrangements of
the coordinating atoms but because of the influence of what one might think
are the more distant parts of the coordinating molecule. Typically, these are
convoluted around the metal ion and its immediate ligating atoms and,
effectively, play the role of the solvent in simpler systems. Except, of course,
that the solvent is a rather rigid one, one that severely limits access of
molecules to (and egress from) the metal ion; seldom, if ever, is it as innocent
as it may appear! Further, the physical bulk of the main ligating molecule
both limits access to, and the mobility of, ligands attached to the metal ion.
By the hydrophobic or hydrophillic surfaces that it presents, it to some extent
selects the species that can approach the metal ion. These features combine
to lead to a rather different chemistry. A link with classical coordination
chemistry is forged by the observation that the coordination equilibria of
complexes of polyfunctionalligands are very sensitive to changes in solvent
composition.
Thirdly, we note that several of the ions mentioned in this paragraph are
not too easy to study. Zn2+ and Ca 2 +, in particular, lack convenient and
sensitive spectroscopic 'handles'. They have no low-lying electronic transitions
by which they can be conveniently characterized; they have no high-
abundance isotope which can be studied by NMR; they do not give
Mossbauer spectra, and so on. The problem is addressed by a technique
which would be almost unthinkable in conventional coordination chemistry.
Such ions are studied by replacing them with a similar ion which does have
suitable spectroscopic characteristics. The reason for this that, although the
biological macromolecule may distort a little when a look-alike ion is
substituted, the distortion is small and good insight into the environment of
the original metal ion may be obtained. So, ifbioinorganic species containing
Zn2+ are treated with an aqueous solution of 1,10-phenanthroline (phen,
Table 2.3) the zinc is extracted because it forms a very strong complex with
phen. If the zinc-free macromolecule is separated it is then readily coordinated
to another cation. Co2+ is a popular replacement ion, but Mn 2 +, Cu2+ and
Cd2+ have also been used. Replacements which have been used for Ca 2 +
are lanthanides, particularly Nd3+ and Eu3+.
The fourth and final point is that species containing two or more metal
ions, linked by bridging atoms, are fairly common in bioinorganic chemistry.
The two ions can be of different metals, so that one is immediately faced
with the question of whether chemistry takes place at both of them, perhaps
simultaneously, or whether one fine-tunes the chemistry of the other.
Commonly, the ions are of transition metals and it is then usual to find that
interaction occurs between what would otherwise be unpaired electrons at
each centre. We shall not develop this particular topic here because all the
important points have already been covered in Section 9.11.
16.2 Myoglobin and hemoglobin

In this and the following two sections a particular emphasis will be placed
on iron-containing species. This is a distortion of the subject of bioinorganic
chemistry. However, these systems are rich in the spectroscopic methods
which have been applied to them, making them particularly suitable
for a study of such techniques in bioinorganic chemistry.
So common is talk about 'the hemoglobin in the blood' that one is
tempted to believe that it is the only species involved in oxygen transport
from the lungs to the place where it is needed. It is not, myoglobin is also
involved. One can perhaps see a reason for a two-component process,
because oxygen transport is, in a sense, self-contradictory. One needs a
species with a high affinity for oxygen in order to extract it from the air but
Myoglobin and hemoglobin I 385
FIC. 18.1 The structure of myoglobin. In (a)

the protein sheath is black and the heme, iron
and histidine indicated. (b) The Fe-heme group
and (c) the attachment of the iron to the
protein through a histidine group.
(b)
one with a low affinity in order that the oxygen can readily be made available
where it is needed, within a muscle, for example. Hemoglobin has such
properties, but the final transport is carried out by myoglobin. Myoglobin
consists of a protein with a molecular weight of just under 18 000 and has
just one iron- porphyrin unit, a heme (Fig. 16.1). Hemoglobin, however, has
four heme units, each with its own protein chain, the whole intermeshed to
give a single molecule (Fig. 16.2). The unusual oxygen-affinity properties of
hemoglobin arise from a cooperative behaviour between the four heme
groups, a cooperation that must surely be mediated by the protein chains,
for the heme units are from 25 to 40 A apart. The heme cooperation in
hemoglobin means that when one or two 0 2 molecules are bonded, the
oxygen affinity ofthe other (0 2 free) heme sites is enhanced. The more oxygen
it has, the more it wants! Clearly, this is a useful property to have when
scavenging for oxygen in the lungs. Conversely, when transfer of oxygen to
myoglobin is under way, the less oxygen the hemoglobin has, the less it
wants. Just the necessary characteristics! Oxygen is absorbed by attachment
at iron atoms in myoglobin and hemoglobin (Fig. 16.3), interestingly, neither
uniquely u nor n bonded, but rather at an intermediate, angular orientation
386 1 Some aspects of biolnorganic chemistry
Fig. 18.2 The hemoglobin molecule; the

attachment of the iron (black dots) to the
protein through histidines is indicated.
Fl._ 18.3 The heme unit in myoglobin

(substituents on the porphyrin ring have been
omitted). In the absence of 0 2 the Fe is about
0.4 A below the plane of the ring, as shown.
When coordinated to 0 2 (shown dotted) the Fe
moves into a position almost coplanar with
the ring.
(when CO is absorbed in place of oxygen it bonds end-on through carbon

to the iron, just as in most metal carbonyls ).
In this chapter the use of model compounds (which are cheaper and
available in much greater quantity than most naturally occurring molecules
of interest) to explore the properties of naturally occurring species will often
be mentioned, but here we see a limitation. We will see models for myoglobin
but a model for hemoglobin which begins to mimic its cooperative behaviour
is another matter; its design and synthesis will require enormous chemical
insight and ingenuity. Perhaps it is as well that hemoglobin is a relatively
easy material to obtain and to work with-it was first crystallized in 1849
and its oxygen-transport properties recognized in 1864 (only 40 years later
did its C02 transport ability become known). Just to complicate things,
oxygen release from hemoglobin is not solely determined by cooperative
factors. It is also pH-dependent (protonation and deprotonation of amino
acid residues may well change the preferred geometry of the peptide chains).
Low pH leads to a more ready release of oxygen, so muscle tissue, where
C0 2 and lactic acid are generated, is an especially favoured site for oxygen
release.
The salts of iron(II) are susceptible to oxidation by the oxygen of
the air-iron(II) sulfate (ferrous sulfate), pale green crystals when freshly
Myoglobin and hemoglobin I 387
prepared, tend to become brown after having been stored for years;
the iron(II) salt used in volumetric analysis as a reducing agent, ferrous
ammonium sulfate is so used, in part, because it is much less prone to
oxidation than most iron(II) alternatives. So, too, with iron(II) porphyrins
(and the iron(II) found in myoglobin and hemoglobin is in a porphyrin ring,
Figs. 16.1-16.3). They are rapidly oxidized to their corresponding iron(III)
compounds by oxygen. Such oxidation can occur, although much less readily,
with myoglobin and hemoglobin, when met-myoglobin and met-hemoglobin
are formed (these names should not be written with a hyphen; one is
included here solely to facilitate a correct pronunciation). Metmyoglobin
and methemoglobin do not transport oxygen! About 3% of the total
hemoglobin in the blood is oxidized to methemoglobin each day, but
mechanisms exist within the red blood cells for regenerating hemoglobin
from methemoglobin. The reason that the loss is so low-and 3% is low-is
reasonably well understood, as follows. Kinetic studies show that the rate
of oxidation of Fe" porphyrins is second order in the iron(II) species and
first order in oxygen. The most probable explanation is that a 1: l complex
is first formed between an iron(II) porphyrin molecule and 0 2 which
subsequently reacts with another iron(II) porphyrin to give an 0 2 bridged
complex-an example of such a compound of cobalt is given in Table 2.2.
The decomposition of the bridged dimer leads to the oxidation products. In
myoglobin and hemoglobin the protein provides a protective shroud for the
0 2 molecule such that formation of a dimeric species is scarcely possible,
whilst remaining sufficiently open to enable ready 0 2 access and egress. The
low rate of formation of the met derivative is thus explained.
The object of models of the myoglobin molecule has been to provide a
similar environment to that found in the natural material; some examples
are shown in Fig. 16.4. One of these ligands, the picket fence molecule, was
given in Table 2.4 and it will be given again in detail in Fig. 16.5. What
characteristics do the models have to attempt to emulate? In myoglobin and
hemoglobin the iron(II) is in a square pyramidal environment, the four
square-planar coordination sites being associated with the porphyrin ring
and the axial with a histidine (see Fig. 16.1(c)) from the protein chain. Oxygen
enters the sixth site, completing an octahedral coordination around the
iron(II). Whereas the five-coordinate iron is high-spin-four unpaired
electrons-the 0 2 complex, oxyhemoglobin, is low-spin, no unpaired elec-
trons. The reduction in ionic radius of iron(II), high-spin, to low-spin
(Section 13.2) is sufficient to enable the iron(II), initially a bit too big to fit
in the porphyrin hole and so sitting slightly out of the ring and towards the
histidine ligand, to drop more or less into the plane of the ring in
oxyhemoglobin. The consequent tug on the histidine connecting the iron(II)
to the porphyrin presumably contributes to the cooperative oxygenation
effect seen in hemoglobin. How well is this behaviour of myoglobin mimicked
by the model compounds? To those shown in Fig. 16.4 we have to provide
a ligand, for obvious reasons a histidine derivative, to occupy the open
coordination site at the bottom of each species shown. With this addition,
comparison is possible. Of course, the models can offer no information on
the detailed involvement of the protein chain-this problem is being explored
by site-directed mutagenesis (creating mutants in which specific amino acids
are changed-and this is more than just an academic exercise; a number of
FJg. 16.4 Some examples of models of

myoglobin. Only schematic structures are
shown-there are usually other heteroatoms
(0, N) within the molecules but not shown here:
(a) a capped porphyrin, (b) a crowned porphyrin
(c) picket fence porphyrin.
R R
R
R
Myoglobin and hemoglobin 1 389
Fig. 16.5 The crystal structure of picket fence

porphyrin with a histidine coordinated to the
iron. The 0 2 molecule also coordinated is
disordered over four equivalent sites. The detail
of the 'fence posts' is shown as an insert.
diseases are known that arise from hemoglobin mutants). We concentrate,

then, on probes that investigate the iron(II) and its immediate environment,
particularly those discussed in Chapter 12.
Although the crystal structures of myoglobin and oxymyoglobin are
available-and their realization was a considerable achievement-they are
not as well resolved as those of model compounds, although the iron-atom
slip and tilted Oz binding, mentioned above for the natural species, are both
clear. In Fig. 16.5 is shown the crystal structure of Colman's picket fence
porphyrin model of hemoglobin, one that will be considered in some detail.
Notice that although the bottom of the Oz molecule is anchored to the Fe
along the latter's fourfold axis, the top of the Oz, as drawn, is tilted to one
side and disordered over four sites. Oz is a diradical, two unpaired electrons,
one in each of the degenerate n* orbitals of Oz. There is theoretical
evidence that it is reasonable to think of the two n* electrons pairing up in
just one of the two n* orbitals and as being involved, along with the 11
electrons on the Oz, in 11 donation to the Fe11 d,, orbital (so that the latter
receives electron density from two sources). For the d,, to be empty in order
to receive them, spin pairing occurs on the iron(II), t1 9 ei -+ t~ 9 eg, leading
to a diamagnetic complex with an obliquely coordinated Oz.
How strong is the Fe-0 bond? Is its strength in the model comparable
with that in oxyhemoglobin? How accurate is the bonding description that
we have just presented-in that it invokes n* electron density being removed
from Oz it might reasonably be expected that the 0-0 bond would be
strengthened by complexation. The available crystallographic work provides
a partial answer; in Table 16.1 are listed some relevant comparisons.
390 1 Some aspects of bloinorganic chemistry
Reasonable accord is found, although contrary to the model of the bonding

Table 16.1 Geometrical features of
oxyhemoglobin and model compound presented, the 0--0 bond length increases on complexation. This discrepancy
is easily explained by adding n back-donation from iron to the empty n*
Feature Picket Oxyhemoglobin orbital of the oxygen (and the empty orbital included is purely n* in
fence character) to the picture. In this way, the Fe--0 bonding is strengthened and
model the 0--0 weakened.
(Fig. 16.5)
Another way of answering questions about bond strengths is provided by
Fe-(0-<l) 1.75A 1.83A vibrational spectroscopy, provided that the vibrational spectra can be
distance sufficiently well understood. Not an easy task for complicated molecules! In
Fe-0-<l 136' ca. 130' the present case two ameliorating features exist. First, the Fe--02 unit is
angle
simple. Secondly, resonance Raman spectroscopy (Section 12.3) enables
0-<l' 1.25A
small parts of a molecule to be studied. With any luck, they are the parts
distance
one is interested in! However, before running a resonance Raman spectrum
' In 02 , 1.21A; 02 1.28A.
one would invariably first run an infrared. Unfortunately, at first it proved
impossible to find any infrared feature associated with the 0 2 unit in the
model compound; spectra of 16 0 2 and 18 0 2 species looked the same.
However, low-temperature work eventually led to the discovery of a sharp
peak at 1385 em- 1 , which broadened almost out of sight at room temperature
(perhaps this is associated with the disorder seen in the crystal structure).
For comparison, a peak at 1103 em - I has been reported for oxymyoglobin,
quite close to the value of 1145 em -I reported for the 0 2 - anion. Although
1385 em - I is well below the frequency reported for the v(O--O) stretching
vibration in gaseous oxygen (1556cm- 1 )-and this difference is in accord
with the modified picture of the Fe--02 bonding just presented-the difference
between it and the oxyhemoglobin value is more than one might wish.
However, by changing the trans ligand (to a dimethylimidazole) it proved
possible to observe the v(O--O) band in solution, where it proved to have a
value of 1159cm- 1, in much better accord with the hemoglobin value. It
has been found that vibrational coupling, and consequent frequency shifts,
can occur between coordinated 0 2 and the trans ligand. Even so, the shift
from 1385 to 1159 cm- 1 seems too large to be explained in this way alone.
The problem remains unresolved. As anticipated above, it has proved
possible to obtain data from resonance Raman measurements. In addition
to providing a (solution) value for v(O--O) of 1140cm- 1, the method has
also shown v(Fe--0 ) to be at 568 em - I in the picket fence model, in excellent
accord with the value of 567 em- 1 found for oxymyoglobin.
What other measurements can be made which offer a comparison between
model and biological compounds? Iron is an element which can be studied
by Mossbauer spectroscopy (Section 12.5.3). Mossbauer spectroscopy pro-
vides a way of determining whether, for instance, the electric field gradient
at the iron is similar in model and biological species. The comparison is
detailed in Table 16.2. Without going into a detailed discussion of the
meaning of the quantities in this table (although Sections 12.5.2 and 12.5.3
provide an indication), it is evident that a reasonable accord exists and this
conclusion is confirmed by a more detailed analysis-the iron atoms, indeed,
are in similar environments.
Whereas Mossbauer spectroscopy probes the immediate environment of
the iron atom, electronic spectroscopy is sensitive to both metal and
ligands-all ligands, not just the 0 2 . It is a characteristic of porphyrin rings
containing a metal atom that they show three bands in their visible and
Search for reaction intermediates 1391
Table 16.2 Comparison of Mossbauer spectroscopic data for oxyhemoglobin and

model compound
Compound Centre shift from Quadrupole splitting
metallic iron
(mmsec- 1 ) (mmsec- 1 )
4.2K 195K 4.2K 195K

Picket fence model 0.29 0.25 2.10 1.34
Oxyhemoglobin 0.24 0.20 2.24 1.89
Table 16.3 Electonic absorption specta of

oxyhemoglobin and a picket fence model compound
Compound Absorption maxima (nm)
Soret
Oxyhemoglobin (human}'
"582 fl
544 418
Picket fence model 548 429
a The data vary little with source of oxyhemoglobin.
Table 16.4 Oxygen uptake of myglobin and a picket fence model compound
Compound Fraction of saturation AH 0 AS 0

at equilibrium (20 oc) (kJ mol- 1 ) (JK- 1 mol- 1 )
Picket fence model 97% -65.3 -15.9

Myoglobin 93% -56.1 -13.4
near-ultraviolet spectra associated with n:* +- n: transitions in the rings. The

strongest band (by a factor of about ten) is in the near-ultraviolet and is
known as the Soret band. The other two, in the visible and of comparable
intensities, are called the a and fJ bands. The a band is at the longer
wavelength. In Table 16.3 data for the picket fence model are compared with
those for oxymyoglobin. As might have been anticipated, although there is
a general similarity, the differences in the porphyrin ring substituents, in
particular, are evident. The picket fence data are for !-methyl imidazole as
trans ligand, but varying this has little effect on the numbers in Table 16.3.
Finally, a simple, but important, question can be asked. How does the model
compare with the biological when it comes to oxygen uptake? Again, not at
all badly, as the data in Table 16.4 show.
16.3 Search for reaction intermediates

As has been seen, bioinorganic species tend to be complicated and one
must always be wary that in concentrating on the metal ion and its immediate
environment one becomes blind to other changes in the molecule. Even
definitions can become difficult If a sugar, peptide or other part of the

molecule changes orientation or conformation by a series of steps, is each step
to be regarded as a reaction intermediate? Such problems are resolved
experimentally, if at all possible. Try to find the intermediate and study it
'on the fly'. Such searches have led to a wide use of time-resolved methods.
Flash photolysis, in which the photolysis products are studied immediately
after they are generated, was one of the first More recently, laser pulse
techniques, in which one laser pulse prepares-excites-the system and
a second probes-studies-the excited molecules, have been extensively
applied. As an example of the tricks that can be used, it is possible to study
very short time intervals by deriving the probe pulse from the exciting pulse
but delaying it very briefly by allowing it to travel away from the apparatus
and then reflecting it back with a mirror. The delay, the time taken for the
round trip, is varied by moving the mirror.
Many of the techniques discussed in Chapter 12 can be adapted to give
such time-resolved information-and it seems that the complexity of many
bioinorganic systems is such that they can only be properly understood once
these sort of data are available. A typical example of the method, if not of
the result, is provided by circular dichroism (CD; Section 12.4). As is well
known, hemoglobin absorbs carbon monoxide. The complex formed, which
is often written HbCO, can be photodecomposed by irradiation by light with
a wavelength corresponding to an absorption band of hemoglobin. In Fig.
16.6 are shown the CD of HbCO, Hb and HbCO 10 J.lS after a laser pulse
at 532 nm. The interpretation is clear-the data provide no evidence for an
intermediate. If there is one, experiments with a shorter time scale will
be needed to find it (to date, none has been found). One reason that the
10 J.lS and Hb traces in Fig. 16.6 are not quite identical becomes evident from
Fig. 16.7, which shows that 20 J.lS after the laser pulse, HbCO is being
reformed. Of the other methods described in Chapter 12, the time resolution
of NMR methods is well known and has been discussed in Section 14.8. In
general, spectroscopies in which Fourier transform methods may be used
are particularly well adapted to time-resolved studies because they study
the entire spectral range simultaneously. One scan/pulse gives the entire
spectrum; with luck, effort, or both, the signal to noise ratio of the final
spectrum will be good enough (although, in general, in Fourier transform
FJC, 18.8 Thee time-resolved circular dichroism

(TRCD) of hemoglobin-CO (HbCO; dashed line),
hemoglobin (Hb; solid line) and HbCO 10 115
after laser flash photolysis (dotted line).
Wavelength (nm) ----

Peroxidases 1 393
Fig. 18.7 As Fig. 16.6 except that the dotted co

line is now HbCO 20 ~s after laser flash
photolysis.
+
400 450
Wavelength (nm) -
methods repetitive scans/pulses are accumulated to improve the final signal

to noise ratio).
16.4 Peroxidases
The peroxidases are heme enzymes that catalyse the oxidation of organic
substrates, the actual oxidant that is catalysed being hydrogen peroxide. In
a typical peroxidase-horseradish peroxidase is one of the best studied-
it seems that iron(III) is coordinated by a heme ligand, a fifth coordination
site being occupied by the nitrogen of a histidine. The action of hydrogen
peroxide on this enzyme gives an intermediate, usually called compound I,
which carries out part of the oxidation of the organic substrate, itself being
reduced to compound II. Compound II completes the oxidation of the
substrate, the native enzyme being regenerated. The question is: 'what are
compounds I and II?' Before seeking experimental evidence, it is sensible to
make some intelligent guesses. First, it seems likely that they contain both
heme and histidine. If the histidine were lost during the oxidation cycle then
it would have to be replaced before the cycle could repeat. Such a need to
scavenge for histidine would introduce a bottleneck which is scarcely
compatible with an effective catalyst. Secondly, they are likely to contain the
iron and/or heme in an oxidized form. For the iron this means Fe'v or
perhaps Fev, both unusual and therefore interesting valence states. Finally,
since hydrogen peroxide is the ultimate oxidant, it may be that some
fragment of the H 2 0 2 molecule is present in either or both of compounds
I and II.
The sixth coordination site around the iron is formally vacant in the native
enzyme and seems an obvious place for interaction between the enzyme and
H 2 0 2 to take place. Of particular interest in this field have been studies of
the absorption of white X-rays. In this context X-rays may best be thought
of in the sequence
visible light -+ ultraviolet -+ vacuum ultraviolet -+ X-rays
All excite electrons, the X-ray excitation being the most energetic. As is well
known, X-rays are emitted when an outer electron drops into a hole in a
low-lying incompletely filled orbital (this is the basis of everyday X-ray
394 1 Some aspects of biolnorganic chemistry
Fig. 11.8 The absorption of white X-rays at a

wavelength chosen because it is one
characteristic of the element one wishes to
study. The fine details contain useful information.
Edge, contains Absorption - EXAFS,

chemical contains structural
information information
Energy-
- - Wavelength
generators, the holes being produced by electron impact-thus, the domestic

colour television is a potential source of soft X-rays). So, the reverse: X-rays
of the right wavelength excite inner electrons into an outer orbital, the
wavelength region being a characteristic of the element under study.
Provided that access can be gained to one of the national or international
synchrotron radiation facilities at which high intensity white X-rays are
available, one can decide to study just the iron atom in compound I or
compound II provided that the molecule of interest is available in sufficient
quantity and sufficient purity. The details of the absorption observed indicate
not only the valence state of the element studied (the higher the formal
charge, the greater the energy needed to excite the inner electron) but also
something of the symmetry of the ligand field.
A typical absorption spectrum is shown in Fig. 16.8, where it is contained
within the (absorption) edge. For compounds I and II such data indicate
that the iron is Fe1v in the sort of tetragonal ligand field that we expect.
This is not all, however. What of X-rays that posses enough energy to ionize
the atomic species under study? An electron will be emitted. One might think
that, once the incident X-ray energy is sufficient to cause ionization, higher
energy X-rays would be no less effective. Figure 16.8 shows that it is not this
simple, there are wiggles, fine structure, in the X-ray absorption. This is called
extended X-ray absorption fine structure or, invariably EXAFS.
In EXAFS, an electron is photoionized by the incident X-rays. As is well
known, the electron sometimes behaves like a particle (usually, when large
distances are involved), sometimes like a wave (usually, when short distances
are involved). The EXAFS wiggles arise because of the wave-like nature of
FJa!. 18.9 An electron photcionized in an the emitted electron. The surrounding atoms reflect the electron wave back
EXAFS experiment may be thought of as giving to the atom from which it originates (Fig. 16.9). If there is constructive
rise tc a transient standing wave. In the upper
diagram the wave from the emitted electron interference between reflected and original waves an enhanced probability
(ejected from the larger black, metal, atom) of electron emission results. Destructive interference leads to reduced
constructively interferes along the metal-ligand
axis with the reflected wave from the ligand.
emission. Hence the wiggles, as the energy (and therefore wavelength) of the
The same happens along all other metal-ligand emitted electron varies. Evidently, if the distances between the central and
bonds of the same length. The phases of the reflecting atoms are changed, so too will the wiggle pattern. The wiggles
waves from each centre are indicated along the
arrow emanating from that centre. The phases contain structural information. We have a method by which we can probe
of the overlaps along the internuclear axis is the local geometry around an individual atom in a macromolecule! Whilst
indicated in circles. The lower diagram similarly
illustrates how, for a different wavelength,
this is true, it is also true that extraction of the structural information from
destructive interference can occur. the experimental data is often far from trivial.
Peroxidases 1 395
Fig. 18.10 The EXAFS of (a) horseradish

peroxidase compound I and (b) that of the
model compound shown in Fig. 16.11. Data
processing has removed the gradient evident in
Rg. 16.8 and by amplifying the right-hand side
of the spectra to give the same signal level as
the left-hand side has also increased the noise.
Energy-
- Wavelength
Fig. 18.11 The model compound referred to in

Rg. 16.10.
Life is made much simpler if a good model compound is available. In Fig.

16.10(a) is given the EXAFS of compound I and (b) that of the model
compound shown in Fig. 16.11. The similarity of the two EXAFS spectra is
such that one has considerable confidence in the accuracy of the model
compound's mimicry. It is seen that this model compound has an Fe=O
bond. Detailed analysis of the EXAFS data gives an Fe=O bond length of
1.61-1.64 A in compound I and 1.62-1.66 Ain the model; the Fe-N(porphyrin)
bond length is found to be 1.99-2.00 A in both compound I and the model.
The agreement is excellent. Compound II has been similarly modelled
and is believed to be simply compound I without a positive charge (i.e. with
one additional electron). Several weaknesses of EXAFS should be noted,
however. Although it gives bond lengths it only indirectly gives bond angles,
so it is limited in what it can say about coordination geometries. Second, it
cannot distinguish between elements which are close together in the periodic
table-N and 0; S and Cl. Finally, it is at its best when there is only one
type of individual atom present, one sort of Fe, for example. When there are
several chemically different Fe atoms, it will only yield an average structure.
Another technique that has been used to probe the structures of com-
pounds I and II is magnetic circular dichroism (MCD). Circular dichroism
(CD) has been met both earlier in this Chapter and also in Chapter 12. It
has the disadvantage that it is a phenomenon confined to molecules which
are optically active. MCD is a similar phenomenon but is applied to
molecules which are not optically active. The trick is this. All molecules
distort very slightly in a magnetic field and the distortion is always one that
changes the optical activity of a molecule. A previously inactive molecule
has no alternative, it becomes slightly optically active. The magnetic field
distorts the molecule so as to destroy any centre of symmetry and any mirror
planes. So, in the intense magnetic field of a superconducting magnet, optical
activity is the norm. MCD is particularly useful when the field-free molecule
has some symmetry, for then the magnetic field splits degenerate electronic
energy levels. This means that it is particularly useful for exploring transition
metal ions and their environments. In MCD there are two cases, A and C,
of particular importance:
• case A: the electronic ground state is non-degenerate but the excited is

degenerate in the absence of a field;
• case C: and the opposite pattern, in which the ground state is degenerate
but the excited state non-degenerate in the absence of a magnetic field.
Because of the theory used to describe the phenomenon, the spectral patterns
resulting (Fig. 16.12) are described as A term and C term, respectively. As
the reader may have guessed, B terms also occur but are much less important
(they relate to magnetic-field mixing of wavefunctions and so are akin to
TIP, mentioned in Section 9.3). Figure 16.12 shows the energy level splittings
which are associated with A and C terms. Because the ground-state splittings
tl:_
A term
-c:::C
Fig. 16.12 The origin of A and C terms in

MCD. A triplet, degenerate in the absence of a
magnebc field, is split apart in the presence of
No field
l 1
Magnetic field
A term spectrum
such a field. Two transitions are allowed; as

the A tenm case shows, they are ecual in Cterm
magnitude and opposite in sign (which is why
II ~
no spectrum is observed in the absence of
a field). In the C term case the thermal
populations of the split levels differ leading to
bands which, although of opposite sign, are of
unequal intensity.
No field
-c:::C Magnetic field
C tenm spectrum
Peroxidases 1 39 7
of C term spectra are small, the thermal populations of the split levels vary
significantly with temperature; the lower the temperature, the stronger the
spectral bands. In contrast, A terms are essentially temperature-independent.
MCD appears to be a somewhat esoteric technique but, in fact, it is of
great value. A complementarity between CD and MCD should be noted. CD
explores a static optical activity and this usually means exploring a property
of the ligand system around, but relatively remote from, a metal ion; MCD
offers the possibility of exploring the properties of the immediate environ-
ment of the metal ion for this is, in part, responsible for degeneracies. Further,
it is not surprising that for paramagnetic species, this paramagnetism is
reflected in their MCD spectra (the technique, after all, is concerned with
the splitting of degeneracies and this is what Chapter 9 was all about, too).
Indeed, MCD can be used to measure magnetic properties and, for this, has
advantages. It is much less sensitive to impurities than are the methods
described in Appendix 8. EPR (Section 12.6) also provides magnetic data
but, really, only for systems with an odd number of electrons. MCD provides
data for both even and odd. Just as a detailed understanding of the magnetic
properties of a complex is none too simple (Appendix 9 gives the simplest
case), so too with MCD. No attempt, therefore, will be made to give a
detailed analysis, but simply to give in Fig. 16.13 the low-temperature MCD
spectra of compound I and compound II. The differences, believed to result
from the one extra electron in compound II, are obvious. Just as for CD, it
is possible to carry out time-resolved MCD measurements and to use the
technique to follow the effects of photolysis, for example. It is in the area of
using spectroscopic techniques to follow such changes that many advances
in bioinorganic chemistry are presently being made.
Fig. 16.13 (a) The MCD of horseradish

peroxidase, compound I, and (b) horseradish
peroxidase, compound II.
500 750
Wavelength (nm)
398 I Some aspects of bloinorganic chemistry
0
N-11-N
,;;}e•v1+
Fig. 16.14 The current view of the catalytic //1\ Compound I
~
cycle in which peroxidase (top left-hand) is 1
oxidized by hydrogen peroxide to give compound H is
I. This oxidizes the substrate AH 2 to give AH"
(dots indicate free radicals), H+ and compound
11. This similarly oxidizes more substrate to
/\ 'AH2
regenerate the starting material (His = histidine).
0 •AH+W
N\11 /,N
V/ Fe'I\v
N - -1-N
7 Compound II
His
This section will be concluded with a survey of other evidence on

compounds I and II. First, in that it takes two one-electron reduction steps
(the formation of compound II is the first) to regenerate the Fe 111 compound,
peroxidase, it would appear reasonable that compound I might be regarded
as containing Fev. However, the evidence from Mossbauer and magnetic
susceptibility measurements, as well as EXAFS and MCD, suggest otherwise.
It appears that it is more accurately described as low-spin Fe1v, spin = 1,
coupled to a porphyrin n-radical cation. Both NMR and visible/ultraviolet
spectroscopic results are compatible with the presence of a n-radical cation.
This interpretation is not unreasonable. One must expect that the porphyrin
does more than just provide a suitable ligand field and a set of geometric
constraints. Indeed, there is evidence that the oxidation of organic substrates
by compounds I and II is carried out through the edge of the heme group
and not through the Fe=O unit. To complete the structural data, resonance
Raman work and oxygen isotope labelling indicate that there is an Fe=O
unit in both compounds I and II. Finally, in Fig. 16.14 is shown, schemat-
ically, the catalytic cycle as it is presently believed to occur. Although it
leaves many questions unanswered, the cycle is remarkably simple, notwith-
standing the relative complexity of the molecules involved. Simplicity in the
midst of apparent complexity is not uncommon in bioinorganic chemistry
and has acted as an additional attraction of the field.
16.5 Blue copper proteins

In the previous sections, the availability of model compounds often, although
not always, provided important contributions to the understanding of the
native material. What if it proves difficult to prepare suitable models? Such
a situation is encountered for the blue copper proteins and provides an
opportunity of demonstrating the use of some of the experimental methods
and their interpretations which have been the subject of previous chapters.
The blue copper proteins form a series of compounds which are involved in
Blue copper proteins 1 399
long-range electron transfer reactions, so that the interplay of Cu11 and Cu'
is of the essence. Although crystallographic work has confirmed the ultimate
conclusion it is interesting to trace the non-crystallographic path leading to
the structure of the site of the copper ion. The compounds are called blue
because of their intense colour, about 400 times more intense than that of
most copper(II) salts. The conclusion drawn was that the Cu11 in the protein
is in a tetrahedral site, because as seen in Section 8.8, such a site makes more
allowed d-d transitions because of mixing of d and p orbitals by the ligand
field. The conclusion was right but the reasoning wrong. The blue colour is
due to a charge-transfer transition, not d-d, although the suggestion of a
(very) distorted tetrahedral geometry was correct. Evidence for the charge-
transfer assignment comes from a comparison of the electronic spectrum
with the MCD spectrum (Fig. 16.15). The electronic spectrum can be
decomposed into a sum of up to eight individual transitions, as many charge
transfer as d-d (there is separate excitation from each of the components of
d sets which would be degenerate in T• symmetry), a spectrum which has
reasonably well been interpreted by XIX calculations. The lower energy bands
are the ones showing the greatest MCD activity and so are likely to be the
most metal-localized. The blue band also persists when the Cu" is replaced
by Ni" or Co", but it moves into the near-ultraviolet, consistent with a
low energy ligand-to-metal charge-transfer assignment.
In almost 30 blue copper proteins, all of which contain about 100 amino
acids, the amino acid sequence has been determined. About a quarter of the
amino acid sequence is common to all and so this sequence almost certainly
contains the ligands which bond to the copper. In this way, the amino acids
4000
2000
(a)
+0.025
Fig. 18.15 A comparison of (a) the visible and
(b) the MCD spectra of the blue copper protein
plastocyanin.
-0.025
..::; -0.075
15000 10000 5000

(b) cm- 1
0 0 0
-NHCHC-
II II II
-NHCHC- -NHCHC-
1 1 1
CH2 CH2
~N
Fig. 18.18 Chemical structures of amino acids
bonded to copper in blue copper proteins. I I
SH CH2
I
SMe
NH_jj
Histidine Cysteine Methionine
Fig. 18.17 1he coordination environment of

the copper in blue oopper proteins.
histidine (two), cysteine and methionine (Fig. 16.16) have been implicated
as involved in the coordination. This conclusion has been confirmed by X-ray
studies, studies that have also shown that the site at which the copper is
located changes little when the copper is removed. The copper is at a site
with a ligand-imposed geometry (Fig. 16.17). This at once provides one
reason why it has proved difficult to devise really good model compounds.
Copper(II) is rapidly reduced by thiols
2Cu 11 + 2RS- -+ 2Cu 1 + RSSR
and it is necessary to arrange that this fate does not befall the cysteine, which
contains an RSH unit. Some apparently promising model ligands form stable
complexes with Co" but react with Cu11 •
Another interesting characteristic of the blue copper proteins is the
visibility of the histidine ligands in their EPR spectra. Like many copper(II)
species, the blue copper proteins give strong EPR signals. When there is a
nitrogen ligand bonded to Cu" it is usual for its presence to be indicated by
nitrogen hyperfine structure (see Section 12.6) and this is seen in the EPR
Fig. 18.18 1he EPR spectrum of plastocyanin;
the wiggle pattern at the centre-left of the spectra of blue copper proteins (Fig. 16.18). Finally, EXAFS measurements
spectrum is the nitrogen hyperfine structure. on the proteins provided further evidence of the environment of the copper
Nitrogen fixation 1 401
····························································································································································································
ions at about the same time as the results of the first crystal structure
determination became available. Information on the long-range electron-
transfer properties of the blue copper proteins has come from several sources;
the use of [Ru(NH 3 )s(OH 2 )]3+ is particularly interesting. It has been found
that this ion binds to the surface of the protein surrounding the copper in
at least one blue copper protein, coordinating to an exposed histidine
(presumably losing the H 2 0 ligand) and thus being held almost 12 A
from the copper. Flash photolysis of [Ru(bpyiJJ2+, added to a solution
of the blue protein/rutheniumammine adduct gives [Ru(bpy)J]J+ and the
[Ru(NH 3 ) 5 ]2+ -protein adduct. The [Ru(NH 3 ) 5]3+ adduct is reformed by
electron transfer from the Cu11, which becomes Cu', the protein losing its
blue colour, providing a convenient method of measuring the electron-
transfer process. The observation of a rate constant (about 2 s- 1 ) which is
temperature-independent over the range studied indicates the absence of any
significant activation energy although there is an entropic factor, probably
associated with reorganization of the water molecules.-around the Ru site;
the site of the copper is optimized for electron transfer.
16.6 Nitrogen fixation

Just as Cu11-Cu1 oxidation-reduction is used in the blue copper proteins, so
the Fe 111 -Fe11 pair is used in a number of others, of which examples have
already been met. Some particularly important systems contain the Fe 111 -Fe 11
system together with sulfur ligands (Fig. 16.19). In the rubredoxins a single
iron atom is tetrahedrally surrounded by the sulfurs of four cysteine
ligands. In plant ferredoxins there are two iron atoms, again tetrahedrally
coordinated and associated with four cysteine sulfurs but bridged by two
so-called 'inorganic' sulfide, S2 -, ligands. These sulfides can readily be
displaced and so they are also called labile sulfides. Finally, in bacterial
(Cys)S .., /S(Cys) ferredoxins there are four iron atoms, again tetrahedrally coordinated and
''''Fe with four cysteine ligands, bound together into a tetramer by four triply
(Cys)S ~ 's(Cys) bridging inorganic S ligands; a relationship with some of the species discussed
(a) in Chapter 15 may be noted. The rubredoxins and ferredoxins have been the
subject of intense study using most of the methods described already and
could well be used to provide further examples of these applications.
However, the present section is more speculative.
The mystery of the mechanism by which biological nitrogen fixation is
achieved has recently, to some extent, been removed by the publication of
some structures based on crystallographic and related work (EXAFS,
(Cys)S, mutagenisis results, peptide sequencing and some spectroscopy). 3 Two
Fe--S different molecules are involved; one is a protein with two identical subunits
STFe,S(Cys) bridged by a [Fe4 S4 ] cluster of the ferredoxin type, bonding to each protein
I s--J--Fe
Fe·---s,.. 's(Cys)
subunit being through two cysteines. The other molecule contains two pairs
of clusters, the members of one pair themselves each contain a pair of
ferredoxin [Fe4 S4 ] clusters, bridged through two of the cysteines (Fig.
(Cys)S/
16.20(a)). The other pair of clusters in the second protein also each contain
(c) two bridged clusters, one a modified ferredoxin [Fe4 S4 ] cluster (not all of
Fig. 16.19 The typical environment of Fe in the sulfur-containing ligands attached to Fe are cysteine) and the other a
(a) rubredoxins, (b) plant ferredoxins and
(c) bacterial ferredoxins. 3 J. Kim and D. C. Rees, Science (1992)257, 1677.
i'
Cys
I
/s'-.....
I " ,,
s ---Fe Fez---s
Cys-S~ ...,.s-Cys
ser- o ---:~~-~~~~~~y Sl 5 ----s1___ .Je 1 Fie
,s,.....
c - s ,,,,,\\ """'" ", '- ·.!''''''
ys ' ',, ' ', ""S-Cys
s----Fe Fe~·s
Fig. 16.20 (a) The structure of the FeMe
cofactor Fe8 cluster in nitrogenase. (b) The I
structure of the so-called P-cluster (Fe 7 Mo) in (a) Cys
nitrogenase (the identity of Y is not known).
similar cluster but with one of the Fe atoms replaced with a Mo. The Mo,
perhaps deceptively, is located at one corner of this latter cluster and seems
well away from the site of likely action in the nitrogen fixation process (Fig.
16.20(b)). To what extent had these structures been anticipated by the
spectroscopic studies on the system? Rather well-consider the more
complicated Fe 7 Mo cluster (Fig. 16.20(b)). It had proved possible to isolate
this species, along with enough of its surrounding groups to preserve its
integrity, although it has proved impossible to obtain the crystals needed for
a structure determination. Has the extraction process destroyed the cluster?
EPR spectra of the native protein and the extracted cluster were both similar
and unusual; the cluster had survived. Analysis of the cluster species gave a
Fe:Mo ratio of not less than 6:1 and not more than 8:1, in accord with the
observed 7:1. Mossbauer spectra showed that most, if not all, of the Fe atoms
were involved in the interactions which gave rise to the unusual EPR signal.
Measurements related to the EPR showed that at least five different types
of Fe atom are present; in the proposed structure each of the seven is unique,
although this does not mean that they will necessarily be distinguishable.
Finally, EXAFS not only showed that the nearest neighbours of the Mo are
S and Fe but that it is also coordinated to two atoms which are either N
or 0 (as mentioned above, EXAFS cannot distinguish). For the (average)
Fe, EXAFS revealed, correctly, that each Fe is bonded to three S atoms.
Actually, six of the seven iron atoms present are three-coordinate, a very
unusual situation which must surely be connected with the coordination of
N 2 as the first step of the fixation process. Not surprisingly, this is another
case in which no model compounds have yet been prepared, although one
may be confident that they will. This does not mean that the biological
Protonation equilibria in bioinorganic systems 1 403
process will soon be replicated by some model system. In the biological

system, only the protein containing the bridging ferredoxin [Fe 4 S4 ] cluster
has been found capable of activating the protein containing the Fe 7 Mo
clusters. No substitute has been found and there is no detailed understanding
of this selectivity.
16.7 Protonation equilibria in

bioinorganic systems
The molecules involved in many bioinorganic systems can be very complicated,
they can interact not only with metal ions but also other species, such as
protons, in the system. It is all too easy for the system to be so complicated
that it appears that progress and understanding are scarcely possible.
Fortunately, the situation is not as bad as this. In this section just one
example is considered but it is sufficient to d~monstrate the fairly self-evident
conclusion that the more data there are available and, equally important,
the more accurate these data, the greater the progress that can be made.
In Section 5.3 there was described the classical approach to equilibria
in systems involving complexes. Although the existence of chelate effects
was recognized, the section was largely based on monodentate ligands
coordinating to a single metal ion. However, in a typical amino acid there
will be at least two potential donors-a nitrogen and an oxygen; in peptides
a plethora of possibilities will exist. In real-life biological systems some of
the potential donor sites are likely to be protonated and, indeed, the nature
of the complexes formed is pH-dependent. This pH dependence may
well be crucial to the biological role of the system and so the question
arises as to how the metal-ligand interactions vary with pH and how
these interactions change the pH-dependent properties of the ligand itself.
Although the former question is the more relevant and more interesting it
is inseparable from the latter and in this section only the later problem, the
pH-dependent properties of the ligand itself will be considered. This is
complicated enough and it illustrates the general approach.
No problem arises if the ligand protonation is step-wise. One site is
essentially fully protonated before protonation of another starts. More
interesting is the case in which several sites are being protonated simul-
taneously, for it is reasonable to expect the presence of metal ions to
significantly modify such equilibria. Consider the case of three competing
sites, A, B and C, arranged along a chain. Using the notation of Section 5.2
the (macro) stepwise protonation constants will be (omitting in-coming
protons for simplicity)
K1 K2 K3
[A, B, C] :;=:[A, B, C]H :;=: [A, B, C]H 2 :;=:[A, B, C]H 3
However, these K values, whilst in principle obtainable by following pH

changes as a function of added acid (or alkali), make no recognition of the
various protonation sites. Let us assume that A, B and C behave as
independent sites, that the protonation of the group A is independent of the
protonation of the rest of the molecule, for example. Denote the protonation
constant kA
The above equilibria may therefore be rewritten as
fr~:-f? ~
/
A~ ~ tA ft~:/\\ tAH
t -
~
k8 BH B _
k "\
k" -
tAH
BH
•'~ [: /,:~ r~ /
CVCHkA CH
CcH CcH
It follows that
K1 = kA + ks + kc
K2 = 2(kA + k8 + kc)
K3 = kA + ks + kc
so that K 1 , K 2 and K 3 are linearly related in this model. Clearly, determine

these K values as accurately as we may, we will never be able to obtain the
individual k values. Additional data on individual species-perhaps from
NMR or UV spectroscopy-are needed. However, the errors in such addi-
tional data are very important. By definition of our problem, kA ;::; kB ;::; kc
so the errors in the additional data must be small enough to enable the small
difference between kA and kB, for example, to be determined.
Not surprisingly, the relationship K 1 = l/2K 2 = K 3 is never accurately
followed. The groups A, B and C are not totally independent. The protonation
equilibrium at one site, A, say, is sensitive to the protonation at the others.
Indicate the already protonated sites by suffixes so that instead of a simple
kA we now have k~, k~, k~ and k~c (the dash indicates no protonated
sites). We have been making the approximation that the group constant kA
is equal to all of the so-called microconstants, k~, k~, k~ and k~c· Although
the move to microconstants has increased the number of unknowns by a
factor of four, the situation is not as bad as it may seem. Because of the
change, K 1 , K 2 and K 3 contribute three pieces of information, not just one.
Further data are obtained from the detailed variation of the pH as acid is
added in a titration. Similarly, the requirement that we reproduce not just
one NMR spectrum but all its variants with pH, all provide more data,
4-(0H)C6H4CH2CH(NH2) C02H = [4-(0H)C6H4CH2CHJ(NH2)[CO]OH

= R(NH2)0H
Fig. 16.21 Proto nation of tyrosine.
although sometimes at the cost of yet more unknown-extinction coefficients

in UV spectra for instance (although, for these, first approximations can be
obtained from data on simpler molecules). Although to date they have been
little exploited, infrared and Raman spectra, for instance, could provide yet
more data. However, the differences between k-":., k~ and k~, for example,
are likely to be smaller than those between kA and kB We need even more
data at even greater accuracy to obtain yet more protonation constants. Lest
the faint-hearted be tempted to give up the task, it should be pointed out
that the group constants are of little use to us-the assumption that they
exist is a negation of the hypothesis at the beginning of this section-that
what happens at one particular donor site will influence the behaviour of
the others.
In practice, of course, what happens is that the various data sets are
brought together in a computer program, weighted according to their relative
accuracies and some sort of least-squares fit made to these data. As more
than one set of acceptable solutions may well be found, it is necessary to
invoke some goodness-of-fit parameter to distinguish between the possibilities.
It is here that errors become important-a unique solution is only possible
if the noise (error) in the data fails to blur the distinction between it and
alternative solutions. An example of such an analysis is for the protonation
of tyrosine, 4-(0H)C 6 H 4 CH 2 CH(NH 2 )C0 2 H (Fig. 16.21)-although the
analysis above was for a more complicated example, most of the data
available refer to, effectively, bidentate ligands.
The important lesson to be learnt from this example is the fact that the
microconstants are, indeed, rather similar. It is not unreasonable to hope
that when there is no alternative to assuming that such microconstants are
identical that the error introduced is not too great.
Further reading contained in The Biological Chemistry of the Elements by J. J. R.

Frausto da Silva and R. J. P. Williams, Clarendon Press,
As has been made clear in the text, this chapter does not Oxford, 1991, although parts of the book are not really for the
attempt a balanced overview of bioinorganic chemistry. Almost newcomer to the field. More information on the material in the
any current inorganic chemistry text will contain a section present chapter will be found in the VCH series on Physical
which details some of the areas excluded. A good, broad-based Bioinorganic Chemistry, A. B. P. Lever and H. B. Gray, Eds.,
and forward-looking series of articles which is still worth particularly Volume 4, 1989. Other relevant data will be found
reading is to be found in the November (pages 916-1001) issue in Metal Complexes with Tetrapyrrole Ligands II J. W. Buchler
of J. Chern. Educ. (1985) 62. An overview of the subject is (Ed.), Springer-Verlag Berlin, 1991.
Other useful sources are o 'Active-site properties of the blue copper proteins' by A.G.
Sykes, Adv. Inorg. Chern. (1991) 36, 377.
o 'Magnetic circular dichroism of hemoproteins' by M. R.
Cheesman, C. Greenwood and A. J. Thomson, Adv. Inorg. o 'Long-range Electron-transfer in Blue Copper Proteins' by
Chern. (1991) 36, 201. H. B. Gray, Chern. Soc. Rev. (1986) 15, 17.
o 'Natural and Magnetic Circular Dichroism, Spectroscopy on o 'The Synthetic Approach to the Structure and Function of
the Nanosecond Timescale' by R. A. Goldbeck and D. S. Copper Proteins' by N. Kitajima, Adv. Inorg. Chern. (1992)
Klinger, Spectroscopy (1992) 7, 17. 39, 1.
o 'Calculating Equilibrium Concentrations for Stepwise Binding
o Biochemical Applications of Raman and Resonance Raman
of Ligands and Polyprotic Acid-Base Systems' by E. Weltin,
Spectroscopies by P. R. Carey, Academic Press, New York, J. Chern. Educ. (1993) 70, 568.
1982.
o 'X-ray Structure Analysis of FeMo Nitrogenase-Is the
0 EXAFS Spectroscopy, Techniques and Applications B. K. Teo Problem ofN 2 Fixation Solved?' D. Sellmann, Angew. Chern.,
and D. C. Joy (Eds.), Plenum, New York, 1981. Int. Ed. Engl. (1993) 32, 64.
o More recent is a review article 'EXAFS' by H. Bertagnolli o 'The Iron-Molybdenum Cofactor of Nitrogenase' B. K. Bur-
and T. S. Ertel. Angew. Chern., Int. Ed. Engl. (1994) 33, 45. gess, Chern. Rev. (1990) 90, 1377.
o 'Electronic Structures of Active Sites in Copper Proteins: 0 Biocordination Chemistry: Coordination Equilibria in Bio-
Contributions to Reactivity' by E. I. Solomon, M. J. Baldwin logically Active Systems K. Burger (Ed.), Ellis Horwood,
and M. D. Lowery, Chern. Rev. (1992) 92, 521. Chichester (1990).
Questions
16.1 Mushrooms of the Amanita family can contain sur-
prisingly high concentrations of vanadium (in fly agaric up to
325 mg kg- 1 ). It was originally believed that it was present as
the compound shown in Fig. 16.22(a) but, currently, the
compound shown in Fig. 16.22(b) is favoured. Given adequate
supplies of the pure material, what measurements would enable
a distinction between the two alternatives?
16.2 Two spectroscopic methods that have been used in
bioinorganic chemistry but not discussed in the text are
time-resolved infrared spectroscopy and time-resolved resonance
Raman spectroscopy. Speculate on the types of investigation
that might be made with these methods.
16.3 Starting with the A and C term spectra in Fig. 16.12
suggest how each would change as (a) the temperature of
measurement is reduced, (b) the temperature of measurement
is increased and (c) the magnetic field used in the experiment
is increased.
fbi. 16.22 Question 16.1.

Introduction to the theory
of the solid state
17.1 Introduction
There have been many points in this book at which it has been apparent
that the properties of aggregates of transition metal ions can differ from
those of isolated molecules. This, in some measure, was a theme in the
previous chapter and very much that of Chapter 15; it was met when
discussing magnetic properties (Section 9.11), visible spectra (Section 8.11)
and vibrational spectra (Sections 12.2.1 and 12.2.3). What, then, of the logical
limit, the solid state? There has been an increasing interest in inorganic
chemical aspects of the solid state in recent years, although it is a topic which
is sometimes called material science. The subject covers the whole of the
periodic table, not just transition metal species. Fortunately, the basic
understanding and theory are common to all. It is the purpose of this chapter
to give an introduction to this theory. Although it is a topic widely taught
and studied, it seems the theory of the solid state is not always well
understood by chemists. There are several reasons for this. First, crystalline
materials have a great deal of symmetry, that of the translation operations
that interrelate all the basic building blocks in the crystal (the term building
blocks is preferred to molecules because not all crystals are molecular solids).
The impact of group theory on chemistry is now so great that, given
symmetry, chemists will look for the corresponding character table. Where,
then, are the character tables appropriate to crystalline lattices? They are
rather difficult to find. Not that they do not exist-a fragment of one is
shown in Table 17.1-but the fact that, for an (idealized!) infinite crystal
there is an infinity of translation operations, coupled with the fact that all
character tables with which the chemist is familiar are square (as many rows
as columns), means that we might well expect the full character table to be
of dimensions infinity by infinity! Secondly, usually, chemists have learnt
about crystals from crystallographers. This means that they have ad~pted
408 1 Introduction to the theory of the solid state
Table 17.1 A fragment of the character table of a one-dimensional translation

group T (not to be confused with the point group T). That for a three-dimensional group
would be superficially similar. For the latter case the k and t would be vectors (a
direction has to be specified in the three-dimensional case) and so they have been
given as vectors in the table. The complex exponents indicate travelling waves (see
the text); because the power of the exponentials must be (complex) numbers and t
is a vector in real space, k must be in reciprocal space in order that the implied dot
(scalar) products in the exponents (k·t) be dimensionless. The t, t2, t 3 , . . . at the
head of each column are operators and correspond to translations t, 2t, 3t, ....
The square, cube etc. form is used for the operators so that they operate properly
on a function 1/J, say. For the k irreducible representation above
t21j! = t(tlj!) = t(eikti/1) = eik1,eik11/1 = e2ik11/1
which is correct, whereas the use of the translation itself gives
(a)
2tlj! = 2 · e 1k1w
which is wrong (the answer has to be the character given for t2 in the table)
T E
0 1 1 1 1
1 eikt e2ikt e3ikt
k
-k 1 e-ikt e -2ikt e -3ikt
1
1
the language and approach of crystallographers. So, every self-respecting

chemist knows the difference between a primitive cubic, a body centred cubic
and a face centred cubic lattice. It perhaps comes as a surprise to learn that
from the point of view of the theory of the solid state all lattices are primitive!
The use of centred lattices is very convenient for the crystallographer but
not for the theoretician or spectroscopist! As an illustrative example in
(b)
Fig. 17.1 there are shown three representative centred lattices and the
corresponding primitive unit cells. The present chapter has been written in
such a way that this particular problem will be minimized. It is mentioned
here because it is an important distinction, one that will certainly be
encountered by the reader who wishes to pursue the subject in more detaiL
There remains the problem of the character tables of space groups and it is
to these that we now turn. In doing so a rather unusual approach
to the theory of the solid state will be followed but it is adopted in the hope
that a different viewpoint will help circumvent the problems referred to
above.
17.2 Nodes, nodes and more nodes

(c)
When discussing the bonding between planar, cyclic C. R. systems and
Fig. 17.1 Three centred lattices and
associated primitive unit cells. The primitive is
transition metals, in outline in Section I 0.1 and in more detail in Appendix
one·quarter of the volume of the face-centred 13, it was recognized that the n electron orbitals of the c.R. molecule are
cubic (a). For the body-centred tetragonal (b) associated with different nodal patterns. Appendix 13 shows, but does not
and end-centred orthorhombic (c) it is one-half
the volume.
emphasize, the fact that different nodal patterns are associated with different
Nodes, nodes and more nodes 1 409
irreducible representations of the relevant point group (different labels of

symmetry species in Appendix 13). In fact, this is general. Symmetry
labels-and character tables-describe different nodal patterns (generally,
three-dimensional) in a very concise and convenient way. Given the problem
of the very large size of the character tables of space groups (the infinite size
problem disappears because real crystals are not infinite-but a full character
table would still be enormous), can progress be made by talking in terms of
nodal patterns, rather than character tables? Fortunately, the answer is yes.
In Section 10.1 (and Appendix 13) it was seen that as the number of
CR units in a planar cyclic hydrocarbon C.R. increases, so too does the
possible number of nodes in the p, molecular orbitals. The maximum
distinguishable number is that obtained when there is one node bisecting
each C-C bond. Suppose that one C-C bond is broken and the c.R.
Fig. 17.2 On the left are shown the nodal molecule straightened out so as to become linear (Fig. 17.2). When the bond
patterns of the six molecular orbitals of benzene
{they are shown in more detail in Rg. 17.6). is broken, the corresponding nodal plane must be deleted and the maximum
These patterns cany over into the diradical number becomes (n- 1). As we shall see, there is a contjnuity between cyclic
generated when a c-c bond is broken {centre) and linear C.R. systems in that just as the different nodal patterns in the
and thus into the linear structure {right). lhe
actual numbers of nodes in the initial and final cyclic system are associated with different symmetry species, so too for the
patterns should be compared. linear-provided that the ends of the molecule can be ignored. Since, in the
0-C- +
0 nodes 0 nodes
B--E-
+
1 node 1 node
$-[- +
-it-:{ - +
1 node 2 nodes
+
3 nodes 5 nodes
present case, in which our concern is with the solid state, the 'molecule' is an
almost infinite lattice the ends can safely be ignored-as was done in Fig.
17.2. This figure contains the nodal patterns for a linear polyene and these
will now be used to obtain the nodal patterns appropriate to crystals. The
theory we shall develop is applicable to more than orbitals and electronic
structure but, for simplicity, we will confine our discussion to this case and
relate the crystal orbitals to the molecular orbitals of polyenes. Although the
molecular orbitals of polyenes consist of p, atomic orbitals there is no reason
to expect that the crystal orbitals will also be :rc, but for the moment it is the
nodal plane patterns that are of interest, not the precise nature of the orbitals
involved. In the simple picture of a linear molecule presented above, all the
nodal planes were parallel. So, too, in a crystal. Crystal molecular orbital
nodes lie in parallel sets. Consider the two-dimensional crystal shown in Fig.
17.3(a). Despite its clearly finite size, it will be taken to be infinite; its
rectangular shape is meant to indicate that the primitive translation vectors
(the vectors that relate adjacent translationally related points) along the two
different axis directions are not of the same size.
If a crystal molecular orbital had just one nodal plane it could be
represented as shown in Fig. 17.3(b ). Extending this pattern, two nodal
planes could occur in the way shown in Fig. 17.3(c). Equally, however, a
b-y----'
X
(b)
(a)
+
FIJI. 17.3 Some of the possible nodal patterns
in a two-dimensional crystal. The crystal is
indicated in (a); the translation vector in they
+
direction has a magnitude which is one and a
half times that in the x direction. If a phase +
(+ or -) were added to (a) it would show the
no-node combination. One-node combinations
are shown in (b) and (d), a two-node in (c). (c)
(d)
However, nodal patterns are not limited to
those with nodal planes perpendicular to x or y;
a three-node inclined pattern is shown in (e).
\-
\
+
(e)
Nodes, nodes and more nodes 1 411
single nodal plane could be drawn as shown in Fig. 17.3(d)-or oriented in

quite a different direction, as in Fig. 17.3(e) where, to be different, three
·nodal planes are shown. Are all of these possible alternatives equally
acceptable? If so, how can this complexity be handled? It seems that we must
not only somehow systematically classify a large number of different nodal
plane patterns but also the large number of different possible directions in
which they may lie. Following from the generalizations made at the
beginning of this section, it seems likely that each different nodal pattern/
direction combination corresponds to a different symmetry species. This
speculation proves to be correct, so that we are moving in the right direction,
but this recognition is no help in resolving the problem of classifying the
multitude of different nodal patterns that can arise in a three-dimensional
crystal. It is necessary to use some cunning!
Our concern is with the electronic structure of crystals and it is our
hope that it will prove possible to describe the symmetry properties
of the crystal orbitals of a crystal, for example, by something akin to
the irreducible representations used to describe the molecular orbitals of
molecules. By analogy with the molecular case it would be expected that
the symmetry species of the crystal orbital which contains no nodes would
be important (in the molecular case all spectroscopic selection rules are
related to it, for instance). This will be the crystal counterpart of a point
group totally symmetric irreducible representation (which contains no
symmetry-required nodes). For the case of the crystal, represent this
zero-node symmetry pattern by a dot on a piece of paper. Now do the
same for corresponding parallel one-node, two-node, three-node and so
Number of nodal planes on patterns (a direction will have to be chosen to which the nodes are
•0 •1 •2 •3 •4 •5 all perpendicular but this apparent arbitrariness will pose no problem) .
Clearly, the dots should not be drawn randomly scattered around but, rather,
Fig. 17.4 A dot map which can be used to related to one another in a sensible way. In Fig. 17.4 is shown how it can
symbolically represent the entire set of nodal
plane patterns at the right-hand side of Fig. 17.2. be done for the linear polyene of Fig. 17.2. The task is now to do the same
for the crystal of Fig. 17.3. A difference is that each dot will now carry two
indices, indicating the number of nodes in each of the two directions in the
two-dimensional crystal of Fig. 17.3. This is easy when the nodes are
perpendicular to either the x or y axes.
The general pattern is evident from Fig. 17.5(a), which should be carefully
studied. It is easy to complete the pattern by extrapolation, as shown in Fig.
17.5(b), although extrapolation may not be enough. This is because although
it is easy to say, for example, 'two nodes cutting the y axis and one cutting
the x', it is much more difficult to picture this-how does one node disappear
somewhere between the y and x axes 1 This is an important problem which
will be explored shortly but it is helpful first to make some comments on
Fig. 17.5(a). Notice that the greater the distance between nodes, the shorter
the distance of the corresponding dot from the dot representing the origin
(which we take to be 0,0). Indeed, Fig. 17.5(b) could well have been drawn
with an exactly reciprocal relationship. Had it so been drawn, then Fig.
17.5(b) cannot be quite correct, for it has already been seen (Fig. 17.3(a)) that
in our two-dimensional crystal the translation vectors along y are longer
than those along x. Evidently, to be accurately reciprocal, in our pattern of
dots the separations along y must be appropriately shorter than those along
x. Figure 17.5(c) is the corrected form of Fig. 17.5(b). Here, each point
+ + + + +
~\~_
• • • •
( :::~ +
0,2
~ '----------1
.------------.
+
+
-
+
(a)
0,0 1,0 2,0 3,0

0,0 1,0 2,0 3,0 • • • •
• •
• •
0,1 1,1 2,1 3,1
0,1 1,1 2,1
• • •
3,1
• • • • •
0,2 1,2 2,2 3,2
• • • • 0,2
•
1, 2
•
2,2
•
3,2
•
0,3 1,3 2,3 3,3
• • • • 0,3 1,3 2,3 3,3
(b)
• • • •
{c)
Fig. 17.S (a) Some of the nodal pattems of represents a unique nodal pattern and this, of course, is just what irreducible
the two-dimensional crystal of Fig. 17.3 with all representations do also. It is not surprising to learn that there is a one to
nodes perpendicular to either x or y, together
with a dot-map representation of them. (b) The
one correspondence between the dots and the irreducible representations of
dot map of Fig. 17.5(a) extended to include the translational group of the crystal. We may not have full details
nodal pattems inclined to x andy. (c) Figure of these irreducible representations but, at least, we have a way of representing
17 .5(b) corrected for the fact that the
magnitude of the translation along y is 1.5
them; each irreducible representation corresponds to a dot in a figure such as
times that alongx in Fig. 17.3. Here, the Fig. J5.5(c). It is easy to see, in a general way, how to extend the pattern of
separations are in the inverse ratio, 0.67:1. Fig. 17.5(c) into three dimensions. We shall return to this task shortly, when
it will lead us to the important concept of the Brillouin zone.
Travelling waves and the Brillouin zone I 413
17.3 Travelling waves and the Brillouin zone

There is a major defect in our argument so far and removing it will also
remove the problem ofrepresenting the 'two nodes along one axis, one along
the other' nodal combination in dot patterns such as that in Fig. 17.5(c). This
defect becomes evident when diagrams such as those in Fig. A13.l(b) are
carefully studied. One of these is reproduced in Fig. 17.6; it shows the n
molecular orbitals (MOs) in benzene. In Fig. 17.6 there are two singly
degenerate MOs. In them, each carbon p, orbital makes an equal contri-
bution (albeit, with differences of phase). In the individual members of the
other, degenerate pairs, of MOs, equal carbon p, contributions do not occur.
However, if the carbon p, contributions to the degenerate pair, taken
together, are calculated, then the carbon p, orbitals are found to make equal
contributions to the pair. To actually check this requires explicit expressions
for the MOs and so it can only be qualitatively checked for Fig. 17.6. 1 An
alternative, quantitative, check is provided by the kno'Yledge that the one-
node functions vary round the benzene ring as cos 0 and sin 0, respectively,
so that the contribution of any carbon p, orbital, the sum of the squares of the
coefficients at that carbon atom, varies as
cos2 IJ + sin2 IJ = 1
Fill- 17.6 The P. molecular orbitals of

benzene. The size of the circles approximately
show the. magnitude of the coefficient of the
molecular orbital at each atom.
1 For a linear polyene such as that shown in Fig. 17.2, some symmetry species appear more
than once. It is then necessary to sum over all functions transforming as the same irreducible
representation before all carbon P. orbitals are found to make equal contributions.
which of course is a constant independent of 8 (the two-node functions vary

as cos 28 and sin 28 and so lead to the same conclusion)-and any concerns
about how 8 is defined vanish. This pattern is a very general one, applying
whether it is a one-, two- or three-dimensional case which is under
consideration. As an example of a three-dimensional case (and real crystals
are three-dimensional), explicit expressions for the ligand group orbitals in
an octahedral complex are given in Tables 6.1 and 6.2. If the reader wishes,
the pattern can be checked for these (again, sum the squares of coefficients;
there can be one, two or three orbitals contributing to a sum).
How does all this apply to the nodal patterns associated with the
two-dimensional crystal under study? In a crystal molecular orbital, the
relevant orbital on an atom that is situated close to a node will appear with
a small amplitude, whereas those on atoms well away from nodes will have
large amplitudes-amplitudes are what nodes are all about, be it here or
in a vibrating violin string. How then, if they have different amplitudes, can
all corresponding atomic functions make equal contributions, for their
contribution is just the square of their amplitude? Within our model, they
cannot, so something is wrong with the model. The way out is simple.
The waves, and nodes, are travelling, not stationary (and so the wave-
functions complex, involving i, .J( -I)). So, over a period of time, as the
waves pass through the crystal, all the atomic orbitals do make equal
contributions.
There are two immediate consequences of this refinement. First, for every
wave going in one direction there must be another, equivalent, going in the
opposite direction-it leads to the same conclusion in terms of orbital
involvement and so the two cannot be distinguished, one is as good as the
I other. Now, an important point. Although the waves are closely related, they
do not transform as a pair (if they did, they would have a pattern that in a
point group would carry an E irreducible representation label). Remember,
we are working with translations and only translations. There is no
combination of pure translations which either interchanges or mixes the two
wave motions. Functions which transform as doubly degenerate irreducible
representations of a point group are always mixed or interchanged by at
least one operation of a group-this is why they are degenerate? It follows
that the two travelling waves, although related, transform as different
irreducible representations of the translation group.
Secondly, the problem posed by a difference in the number of nodes along
two perpendicular axes disappears. This is demonstrated in Fig. 17.7 which
shows some time-sequenced pictures of a travelling 4 + 2 nodal pattern, most
readily seen for a finite crystal, although the general pattern can clearly be
extended in principle to an infinitely large crystal. The meaning of the
additional points included in Figs. 17.5(b) and (c) compared with Fig.
17.5(a) becomes clear; they correspond to travelling waves moving at an
angle to both x and y axes. An important lesson, therefore, is that in a
crystalline solid one should think in terms of travelling waves, not stationary
ones, because it is possible to attach symmetry labels to travelling waves but
not, in general, to stationary waves.
Fig. 17.7 Time-sequenced pictures of a
travelling wave in the crystal of Rg. 17.3. Such
travelling waves can be along axial directions as 2 The case of separable degeneracy is not covered by this statement but it is true for
all of
well as inclined to them. the point groups that normally concern the chemist.
Travelling waves and the Brillouin zone 1 415
•
•
•
• . .• +
-
• •
•
•Can there ever be stationary waves in a crystalline solid? The answer is
yes, under very special circumstances. So, if, along an axis, the nodal planes
•
exactly bisect interatomic planes, there is no need to invoke travelling waves
•
•
• . .
• • + • •
- •
•
to obtain equal atomic orbital contributions. Each atom is at a point of
•
greatest amplitude (which may be positive or negative) and so all orbitals
appear with the same coefficient although this is either positive or negative,
matching the sign of the amplitude. Pictures for three such waves in the
two-dimensional case are shown in Fig. 17.8. All three satisfy the requirements
imposed on functions which transform as an irreducible representation of a
space group. The example considered is that of a two-dimensional space
group but the conclusion is general. Standing wave irreducible representations
exist only when there is a matching between the nodal separation of the
wave and the magnitude of the corresponding translation vector.
An important point follows. Within the context of the present discussion,
it is meaningless to have wavelengths shorter than those shown in Fig. 17.8 in
• • • • • •
any of the three nodal plane directions shown there. A shorter wavelength
mean that somewhere in space we would have a pattern akin to that
• +
• • +• • •+ would shown in Fig. 17.9. However, in Fig. 17.9, for two of the four regions with
• • • • • • positive amplitude there are no atoms within the region and so no orbitals
• • • • • •
that could have this amplitude! So, those amplitudes are meaningless, and
with them the nodal pattern that gave rise to them. Another way of looking
Fig. 17.8 Some standing-wave patterns for a at this is to compare Fig. 17.9 with Fig. 17.10. Qualitatively, at least, the
two-dimensional crystal. The nodal planes lie at atomic orbital amplitudes are the same in Figs. 17.9 and 17.10, and a detailed
exactly half the distance between atomic layers. study shows that it is not difficult to ensure that they are also quantitatively
the same. If the pattern in Fig. 17.9 is not meaningless, it certainly is
• -1 I. • .I •
redundant. We see, then, that nodal patterns such as those shown in Fig .
17.8 are the limit that we need to consider, be it two-dimensional or
• .I • • • • three-dimensional space about which we are talking. Such nodal patterns
+ + + + can be incorporated in dot maps of the type given in Fig. 17.5(c)-enormously
• • • • • • extended-and represent the limit of the dot patterns. Further dots can be
• • • • •I • added outside this limit but, within the present context are either meaningless
or redundant. 3
Wave patterns in very similar directions will terminate at very slightly
displaced, adjacent, points. A family of such points will define a surface which
Fig. 17.9 A wave pattem in which the
separation between nodes is less than the turns out always to be planar, except where two or more such planes
separation, in the same direction, between intersect. Most important is the three-dimensional case, which we now
atoms. The result is that in two regions
(arrowed) there are phases but nothing to which
consider and which follows, without complications, by extension of the two
these phases can refer. dimensional. The three-dimensional limiting surfaces define some rather
beautiful shapes, the shape depending on the particular lattice under
consideration. Examples of some of them are given in Fig. 17.11. The shapes
• • • • • • shown in Fig. 17.11 are pictures of the Brillouin zones ofthe solids. A Brillouin
zone can be thought of as containing a multitude of closely packed points,
•+ • • +
• • • + each with a unique set of labels and each point corresponding to a unique
• • • • • •
• • • • • • 3 The present context is one in which there is a single set of points, presumably representing
atoms, interrelated by translation vectors. Had there been units made up of several points,
Fig. 17.10 A wave pattern which gives the grouped together and interrelated between themselves by, say, point group operations, then the
same amplitudes as those in Fig. 17.9 but which 'empty spaces' in figures. such as Fig. 17.9 would not be empty. In such a case there would be
has an internode separation which is greater a distinction between Figs. 17.9 and 17.10; the additional node patterns would be neither
than the interatomic separation in the same meaningless nor redundant. They are associated with the second, third, and so on, Brillouin
direction. zones.
(travelling) wave pattern. Each point corresponds to the translation group
@~
equivalent of an irreducible representation in a point group character table,
so the Brillouin zone, containing all points, is, in a sense, the equivalent of
a translation group character table. As we have seen, the further that one of
the closely packed points is from the origin (the origin is at the centre of the
Brillouin zone and is the no-node combination), the shorter the distance
OJ@
between nodes in the corresponding travelling wave pattern in real space.
Because of this reciprocal relationship one commonly talks of reciprocal
space; all Brillouin zones are in reciprocal space. Conventionally, the vector
drawn from the origin to a particular point is called the 'k vector' for that
point. As we have also seen, for every vector k there exists a vector - k. The
origin is often referred to as k = 0 in phrases such as 'At k = 0, .. .'. It is
important because k = 0 is the equivalent of the totally symmetric irreducible
p@
representation of a point group. As a result, just as for point groups, the
spectroscopic selection rules for crystals correspond to non-zero integrals
and this means integrals that they transform as k = 0, which is why phrases
such as at k = 0, ... will be encountered. The label k, with a suitable plethora
of subscripts to indicate the point in the Brillouin zone to which it refers,
is the equivalent of a point group irreducible representation label such
as A" B 2 ., ••• with which the reader will be familiar. Not surprisingly,
instead of reciprocal space the term 'k space' will be encountered. Just to
FJC, 17.11 Drawings of the shape of some add a bit of physics, and remembering that we are talking about travelling
Brillouin zones: (a) face centred cubic, waves, the name momentum space will also be found in books on solid
(b) trigonal, (c) tetragonal, (d) orthorhombic, state physics.
(e) monoclinic and (f) triclinic. In each case an
axis of highest symmetry is arranged It is to be emphasized that in the above discussion we have been talking
approximately vertically in the plane of the about translations, and only translations. A particular pattern of translations
paper. In principle, there is one general shape may well lead to the automatic generation of rotational axes. So, it takes
of Brillouin zone for each of the 14 Bravais
lattices but, because of the freedom allowed by nothing but translation operations to generate a cubic lattice. However, when
changes in relative axis size for the lower we look at the lattice so generated we find that it also has threefold, fourfold
symmetry groups, there is a total of 24 and twofold rotation axes. These are something additional. For instance, the
qualitatively different shapes.
way that a cubic lattice is brought into coincidence with itself by a C4
rotation, the way that the points map onto one another, cannot be replicated
by any combination of translation operations applied to the lattice. Inclusion
of operations other than pure translation operations is important for those
interested in the details of the theory of space groups, but is not essential
for our discussion. The lattices are always of the highest symmetry compatible
with any crystal type,4 be it cubic, tetragonal, orthorhombic, hexagonal,
trigonal, monoclinic or triclinic; addition of point group operations can never
increase the symmetry although the absence of such operations can lower
it. Clearly, it is convenient to restrict discussion to the high-symmetry cases.
These cases have given us the Brillouin zone and a myriad ofk vectors-and
that's enough! The point group operations serve to interrelate k vectors that
are otherwise distinct. So, in any cubic lattice, invariably of o. symmetry, a
k vector in a general position ink space has 47 symmetry-related equivalents
(there are 48 operations in the o. point group). Diagrams showing energy
levels across a Brillouin zone (which will be met shortly) would only show
one of the 48.
4 For instance, all lattices are centrosymmetric, contain centres of symmetry, although not
all crystals are centrosymmetric.
Band structure 1 417
17.4 Band structure

The content of the three previous sections is sufficient to enable us to
consider a specific example which will form the basis for application of the
basic theory. Let us consider a crystal composed of a diatomic molecule AB,
with just one molecule in a unit cell 5 (if there were several, they would be
interrelated by point group-type operations and these we have just excluded).
All of the A are interrelated by pure translations, as too are the B. We first
consider just the A; most of the development we need will follow~the
inclusion of B will not add greatly to it. Each and every atomic orbital of
A spans the entire Brillouin zone of the space group. That is, there is a
separate combination of each and every orbital of A corresponding to each
and every possible nodal pattern, every k vector. As the nodal patterns vary,
as k varies, so too the corresponding (crystal) molecular orbital energies
must be expected to vary. Each and every single energy level in the isolated
A atom gives rise to an energy band in a crystalline s.olid. Let us look at
this in more detail; we do this by considering just one orbital of A, and select
one that leads to the revelation of all of the important phenomena; we select
a p, orbital. To simplify the pictures that will arise, but for no other reason,
the discussion will be confined to the xy plane, these two axes, of course,
being mutually perpendicular. There is only one A atom in the (two-
dimensional) unit cell and so just one p, orbital in each unit cell.
Figure 17.12 shows pictures of the combinations of these p, orbitals
appropriate to three points in a two-dimensional k space. To avoid having
to work with travelling waves the points have been chosen to be at the centre
(k = 0) and two surface points of the Brillouin zone, but the intermediate
points follow by interpolation. At the centre of the zone, where, as shown
in the left-hand diagram, the pattern is the same in all unit cells, along the
y axis there is an in-phase n overlap; that is, a n bonding interaction.
However, along the x axis the interaction is CJ anti bonding. The edge of the
Brillouin zone along the x axis has a pattern in which there is an alternation
of phase between unit cells along the x axis. This is pictured at the top right
in Fig. 17.12. Examination shows that the interactions are now n bonding
along the y axis (the same as before, nothing has changed along they axis)
but now CJ bonding along x. Because a CJ antibonding interaction has become
bonding, the x axis zone edge pattern is lower in energy than that at the
zone centre. Conversely, at they axis zone edge (lower right in Fig. 17.12),
the CJ and n interactions are both antibonding and the pattern has an energy
which is higher than that at the zone centre.
These energy patterns are brought together in Fig. 17.13 which shows
plots along two directions in the Brillouin zone. These are the directions
corresponding to the x and y axes of Fig. 17.12 but, because the Brillouin
zone is in reciprocal space, they are called x* and y*, respectively, to
5 This is the first time that the name unit cell has been used, a name that seems innocuous
enough. [n fact, some care is needed in its use because there is no unique definition of unit cell
for any crystal structure. For any crystal structure there is an infinity of acceptable choices of
unit cell-the only requirement on it is that it contains one primitive structural unit and that
it generates the entire crystal by translation operations alone. A unit cell need not have six
faces and its faces need not be planar Oust as tiles with curved edges can cover a surface). It is
often convenient to choose as unit cell a volume defined by the primitive translation vectors
themselves but, equally, for other purposes this can be an inconvenient choice.
I I I I
~~~~~~
~~~~~ + I +
i
I I
I~
+ + I +
~~~ ~
I
Fig. 17.12 Standing-wave patterns for the + 1tb(Y)
~I~
k = x* edge; a,(x)
~
example described in the text. To the left is the
I
zone-centre pattern and two zone-edge patterns X I
to the right. The type of bond1ng (u or n) is
indicated together with whether the interaction I
is bonding (,) or anti bonding(,) and the axis k = y* edge; aa(X) + 1ta(Y)
along which these interactions are directed.
~ ~ ~
~ ~ ~
i ~ ~ ~
Fig. 17.13 An energy-level d1agram for the two E

directions in the Brillouin zone shown in Fig.
17.12. For simplicity, a linear interpolation has
been used between the points pictured in Fig.
17.12 and this gives rise to the simple block
form of the density of states on the right. The
reader is cautioned that there are good reasons
why this linear interpolation is not valid; it is
used here solely on grounds of simplicity.
-x• y*-
Density of
Brillouin k=O Brillouin states (DOS)
zone edge zone edge
distinguish them from x and y.6 Because the change from bonding to
antibonding is expected to have larger energetic consequences for u than 1t
interactions, the slope of the plot to the x* zone edge is greater than that to
they*. For simplicity, the interpolations from zone centre to edge have been
made linear, although in reality they would be curved. To the right of Fig.
17.13 is shown the resulting density of states plots (DOS; these were first
met in Fig. 15.25). It is clear that what was an individual level in the isolated
atom has become a band in the solid-a somewhat unrealistic rectangular-
sectioned band, but a band nonetheless. Further, the energy varies across
the band (in jargon, the 'dispersion of the band'); the energy varies with
position in k space. The solid-state physics literature is full of plots of
dispersion curves, usually along directions in reciprocal space that correspond
to high-symmetry directions in the crystal; we shall meet examples shortly.
It is evident that the greater the interaction between adjacent translation-
related orbitals, the broader will be the resulting band. Inner orbitals will
usually give narrow bands, outer orbitals will tend to give broad bands.
Increase in pressure on a crystal will tend to increase the interorbital
interactions and thus to broaden the bands, this not only affecting the
outermost orbitals but, with sufficient pressure, the inner ones too. So, it is
calculated that the ls orbitals of atomic hydrogen give rise to a band such
that hydrogen becomes a metal if it is subjected to ca. 10 6 atmospheres
(10 11 N m - 2 ) pressure. Returning to our example, if the Px orbital of A was
empty, so too will be all levels of the band that have been generated from
it. If the Px orbital contained a single electron the corresponding band would
be half-filled (if the bottom half of the band was filled then the electrons
would be paired up). If the A Px orbital were filled, so too will be the band.
Unless, of course, something happens to change the situation.
There are two important things that can happen, the first of which will
have to be discussed at some length. This arises from the recognition that
bands adjacent in energy can overlap with each other. So, if a band
corresponding to a doubly filled A orbital overlaps with that corresponding
to an empty A orbital, what results is a single band (really, composed of two
overlapping bands) which is half-filled. Although such overlapping is likely
to be of common occurrence, as we shall see, one has to be careful-
overlapping can be avoided. In the example considered above it was seen
that interactions between Px orbitals changes from antibonding to bonding
with increase ink along the x* axis. As shown in Fig. 17.14, the corresponding
change in s orbitals is the opposite, from bonding to antibonding. Now, it
is generally true that s atomic orbitals lie below the corresponding p, so the
different behaviour of s and Px along x* means that they are potentially set
on a collision course! In fact, they interact; this interaction can lead to
a mutual repulsion, a so-called avoided crossing, as shown in Fig. 17.14 (the
energy levels do not cross although the wavefunctions behave as if they did).
In such cases, increase in pressure increases the separation between two
bands. Although avoided crossings are not uncommon, more general is band
overlap. Such overlap is not limited to pairs of orbitals-there is no limit
on the number of energy bands that can overlap with each other. Just as in
6 In the general case, where x and y are inclined at an angle of other than 9W, care has to
be taken in defining the corresponding axes in k space.
FIJI. 17.14 An avoided crossing in a simple

one-dimensional case. At k = 0 the interactions
between adjacent s orbitals is bonding and
between the P. ~ is anti bonding. At the zone
boundary the s orbital interaction is anti bonding
and the P. is bonding. Instead of the energy
levels crossing, the s and P. orbitals mix so that
in the middle of the diagram (at the mixture
indicated by the crossing of the dotted lines)
there are two sp hybrids involved. By the zone
boundary they have become unmixed again.
Zone Zone
centre edge
a molecule there is a highest occupied molecular orbital (a HUMO) which

is of importance, so there is a crystal equivalent, of no less importance.
The crystal equivalent is called the Fermi Surface. This may be pictured
for the Px example discussed above and which formed the subject of Figs.
17.12 and 17.13; such a picture is given in Fig. 17.15 for the case of an
incompletely filled band. Notice particularly the way that the Fermi surface
depends on direction in k space. In some directions the Fermi surface shown
in Fig. 17.15 is quite close to k = 0; in others it is close to the zone surface.
When the Fermi surface lies within a band a conductor results (travelling
waves help to explain how electron migration occurs under the influence of
an electrical field). When the Fermi surface occurs at the top of a band, the
band is full and an insulator results. When there is a small gap between the
top of a full band and the bottom of an empty one, a semiconductor results;
semiconductors are usually classified in terms of their band gap. Two
important semiconductors, Si and GaAs, have band gaps of 1.1 and 1.4 eV
respectively; for comparison that of diamond is 6.0 eV.
The Fermi surface defined above was defined at a very low temperature,
formally 0 K. As the temperature is raised the sharp horizontal line in Fig.
17.15 which represents the Fermi surface becomes increasingly blurred as
electrons are thermally excited into levels which at 0 K were empty. In
semiconductors, with a small band gap by definition, electrons are similarly
thermally excited from the top of the highest filled band (the valence band)
into the bottom of the lowest empty band (the conduction band). Their
electrical conductivity increases with temperature. Add the possibility of
incorporating into such a semiconductor impurity atoms which have energy
levels which fall within the band gap, together with the possibility of bringing
materials with different band gaps into physical contact, and one has the
basis of modern semiconductor technology. Metals, of course, have con-
ductivities which decrease with temperature. This may most readily be
Zone edge
along y*
Fig. 17.15 A possible Fermi surface obtained

by partially filling the band structure indicated
in Fig. 17 .13. Because the example is
tiNa-dimensional the Fermi surface is here a
line. For simplicity the zone edge which runs
from the x* to they* axis has been shown as a
straight line. Realty, it should be more faceted,
Filled
as a cross-section of any of the Brillouin zones band
shown in Fig. 17.11 demonstrates.
Zone edge
along x*
understood somewhat colloquially. Electrical conductivity occurs because

the application of a voltage difference changes the energy level pattern at the
Fermi surface, travelling waves that serve to transfer electron density towards
the positive pole being preferentially populated. They have defined k values.
One may think of one such wave setting out from the negative pole. Because
of thermal vibrations of the atoms the lattice does not quite have the
regularity required for the accurate definition of its k value. Mismatches
occur and the wave is attenuated on its way to the positive pole. Clearly,
the higher the temperature the greater the thermal vibrations and the greater
the attenuation. The same phenomenon occurs for semiconductors, of course,
but the increase in thermal population of the conduction band with
temperature dominates. One final word: the Fermi surface of Fig. 17.15 has
been constructed in such a way as, hopefully, to make understanding easier.
It is a highly distorted, unreal Fermi surface. Reality arrives in the next
section!
So far we have restricted our discussion to the A atoms and considered
those things that can modify the pattern originally derived, one in which the
occupancy of a band is determined solely by the occupancy of the orbital
from which it is derived. Of course, everything that has been said about the
band structure derived from the orbitals of A hold for the orbitals of B also.
Next it has to be recognized that there can be interaction between the bands
derived from the orbitals of A and B-this is the equivalent of bonding in
an isolated AB molecule. In molecules, interaction only occurs between
orbitals of the same symmetry. So, too, in solids. Interaction between crystal
orbitals can only occur when the crystal molecular orbitals (which are
commonly referred to as Bloch functions , although the use of this term is not
confined to electronic wavefunctions- it applies to any function which
transforms as an irreducible representation of the translation group) have the
same symmetry. That is, they have the same k vectors. It follows that such
interactions are not required to be constant across the Brillouin zone. For
example, the energy difference between two Bloch functions at k = 0, the
zone centre, may be very different from the corresponding energy difference
at a point on the surface of the Brillouin zone. It is worthwhile at this point
to emphasize that a Brillouin zone is not, of itself, an energy surface, although
such an interpretation is sometimes encountered. Equally, it is not a hollow
shell, although sometimes drawn in this way. As we have seen, it is perhaps
best regarded as a compact- and perhaps solid - mass of points of the sort
shown in Fig. 17.5(c) in which each individual point represents a unique
irreducible representation.
17.5 Fermi surface

In the above section the concept of the Fermi surface, as the solid state
counterpart of a HOMO, was introduced. Just as HOMOs are important
in molecules, so Fermi surfaces are of vital importance in crystalline solids.
The object of the present section is to attempt to answer questions such as
'what does a real Fermi surface look like'? (the one introduced in the
previous section was highly unreal) and ' how are they determined- if they
can be'? It will avoid complications of the type encountered towards the end
of the previous section if we consider only solids with a single type of atom.
Because they show the way the bands are filled, it is more illuminating also to
confine the discussion to metals. It is clear that the more electrons that there
are and the more valence shell electrons, the more complicated will be the
behaviour of energy levels and their occupancy over the Brillouin zone. We
therefore start with a simple case, that of metallic sodium. Sodium has one
relevant valence shell orbital, a 3s, and just one valence shell electron. The
structure of the metal is body centred cubic (bee), for which a picture of the
Brillouin zone is shown in Fig. 17.16(a). We would expect the band to be
half-filled and the electrons paired; this is just what happens. The Fermi
surface is shown in Fig. 17.16(b ); it is, with but minor deviations, spherical
and occupies half of the volume of the Brillouin zone. As indicated in Fig.
17.14, s orbital interactions become increasingly antibonding with k and so
the Brillouin zone is filled from the centre outwards; for other orbitals with
K
other interaction patterns it may be the boundaries of the Brillouin zone
which are filled first. Note from this last sentence how easy it is to equate
Brillouin zone with energy, something which, as has been said, is to be
(b)
avoided. There is one other lesson to be learnt from Fig. 17.16(b). This is that
it is customary in solid-state theory to label various parts of the Brillouin
Fig. 17.16 (a) The Brillouin zone for a body
centred cubic (bee) lattice and (b) the Fermi zone. For all Brillouin zones the centre is labelled r; the other labels, like
surface of sodium, which has a bee structure. the labels for the irreducible representations of point groups, are perhaps
This surface is essentially spherical and
best thought of as having a unique meaning for each Brillouin zone. For the
nowhere touches the zone surface. Notice how,
to facilitate discussion of Brillouin zones and bcc 7 Brillouin zone, the point P is on the zone boundary, along a threefold
Fermi surfaces, labels are given to some points axis, and the point H is similarly on the boundary, along a fourfold axis.
in the Brillouin zone (those shown are an
incomplete list of those used for the bee 7 bee= body centred cubic; even though one has to work with the corresponding primitive
Brillouin zone). cells, the language of the crystallographer is used!
Fermi surface I 423
Ftg. 17.17 (a) The face centred cubic (fcc)

Brillouin zone together with some of the labels
used for special points. Although special points
may appear several times in the figure they are
only labelled once. (b) A calculated band
structure for metallic copper, which has a fcc
structure. The Fermi level is shown dotted. For
further comment on this figure see the text.
(c) A picture of the Fermi surface for copper. It
is to be remembered that such a Fermi surface
is concerned with a large number of electrons.
The electron density is not uniform throughout
the Brillouin zone~ for example, below the zone
surface there are the electrons occupying the d
orbital originating levels shown in Fig. 17.17(b).
(d) Two cross~sections of the zone surface
shown in Fig. 17.17(c). Figures (a) and (c),
taken together, should enable this figure to be X K
understood.
The next example is more complicated; it is copper, each atom having six
valence shell orbitals (five d and one s~the configuration of the isolated Cu
atom is 3d 1 0 4s 1 ) to be included together with, of course, 11 electrons. In
the language of Chapter 7, there is only one hole to consider. Copper has
the face centred cubic (fcc) lattice, the Brillouin zone of which was shown in
Fig. 17.11(a). This is repeated in Fig. 17.17(a) with the labels of special
positions indicated. In Fig. 17.17(b) are shown the results of calculations on
the energy bands in copper and their occupation at 0 K; the Fermi level is
indicated. Figure 17.17(b) merits close inspection. Most evident are the peaks
at W and K; they correspond to the antibonding combination of the 4s
orbitals at the zone surface. The band of levels lying together originate in
the 3d orbitals, although the way that they get mixed with the 4s is evident
on close inspection. Thus, the line representing the peak at W is not
continuous with that representing the peak at K; an avoided crossing has
occurred. Even this is not simple~the single low-lying level at r involves
the bonding combination of 4s orbitals. Of particular interest is the way that
at L, all nearby points in the Brillouin zone lie below the Fermi surface, the
only labelled point in Fig. 17.17(b) apart from r so to do. The consequence
of this is evident in Fig. 17.17(c), which shows the Fermi surface for copper.
A double cross~section of this surface is given in Fig. 17.17(d), a figure which

needs reference to Fig. 17.17(a) for its interpretation. In the L direction
copper is an insulator! That is, if in some (thought) experiment, we took a
perfect single crystal of copper and constrained electrons to attempt to move
through the crystal in straight lines, then in the L directions (there are four
such directions, corresponding to the four threefold axes of a cube) the
material would be an insulator. Support for this comes from a comparison
between the electrical conductivities of sodium and copper at very low
temperatures. Sodium is the better conductor.
The diagram shown in Fig. 17.17(b) was referred to as calculated. The
question immediately arises as to whether there exists any experimental
method by which such calculations can be checked. The answer is that there
is a host of methods which give information on the details of the Fermi
surfaces of metals, at least. They are all methods which the chemist
traditionally regards as in the province of physics, but it is only there
that the answer to the question is to be found. Almost all relate to a
remarkable and unexpected phenomenon. That single crystals of metals at
low temperature and in strong magnetic fields display oscillatory behaviour.
Oscillatory behaviour in almost any measurement that is made on them, the
oscillations occurring as the strength of the magnetic field is varied and the
value of the measurement plotted against the magnetic field strength (or,
better, the reciprocal of the magnetic field strength). The most important
and famous of these measurements is of the de Haas-van Alphen effect. This
is the result of measurement of the magnetization induced within a single
crystal of a metal as the magnetic field is varied (the phenomenon also varies
with the sample orientation). The plot obtained for a copper single crystal
when the magnetic field makes equal angles with each of the coordinate axes
is shown in Fig. 17.18(a).
The simplest explanation of the phenomenon requires that the reader
makes a brief reference back to Section 9.3 and the discussion there
about the orbital magnets induced by a magnetic field, which has to be
extrapolated to the case of solids.8 In a solid, as the magnetic field is varied,
the splittings into clockwise and counter-clockwise rotating components for
each level also varies and so the precise nature of the level at the Fermi
surface can change also. As the clockwise component is replaced by counter-
clockwise, so the local field alternates between reinforcing the applied field
and reducing it. So, oscillations are observed. The relevant crystal orientation-
and the path of the induced currents-for the copper case shown in Fig.
17.18(a) is given in Fig. 17.18(b). As the crystal is rotated, the pattern of
oscillations changes also, thus providing information on the Fermi surface.
17.6 Solid state and coordination compounds

Where do coordination compounds come in? At several points in this text
there have been discussed either molecular aggregates-Chapter 15 is
devoted to them-or crystalline species. In this section some general points
8 One point of extrapolation is clear. The possibility that there be circular electron orbits in
a solid has not been covered by the discussion in the text. However, since the Fermi surface,
by definition, is a surface of equal energy, it is evident that, as required by the discussion in
Section 9.3, there is no energetic barrier to such circular orbits.
Solid state and coordination compounds I 425
Fig. 17.18 (a) The de Haas-van Alphen

oscillations in metallic copper when the
--
magnetic field is orientated as indicated as
shown in (b). The two periods of oscillation may
be correlated with the concurrent existence of 1
the two different circulatory pathways shown. (a) H
A different orientation of the crystal with respect
to the magnetic field would lead to a different
oscillation pattern and thus provide information
on the topology of the Fermi surface (adapted
from Ashcroft and Mennin).
will be made and then the above discussion will be applied to some topics
covered in earlier chapters. Clearly, when we turn to complexes, we have
to start with the AB pattern, covered in the previous section, and develop
it. The relevant situation is one in which either or both of A and B are
coordination compounds. Were we to follow the pattern developed so far in
the chapter, we could consider each atomic orbital of every atom of A and
do the same for B. This, however, would not be sensible. A major theme of
many earlier chapters has been the exploitation of the molecular orbitals
appropriate to A and/or B when these are coordination compounds. The
question which immediately faces us is the extent to which these discussions
have to be modified by the introduction of interactions between translationally
related sets of A and also of B. It is not difficult to see the general
outcome. When an orbital of a coordination compound, A, is largely located
on a central metal atom, then the intervening ligands serve to insulate these
orbitals from each other. The orbitals do not interact and the discussion of
this chapter becomes largely irrelevant. The crystalline material behaves like
a set of superimposed atoms, each unaware of the others' existence. Such a
behaviour is much less likely to occur for molecular orbitals delocalized over
the whole molecule. Here, the contribution from orbitals on the periphery,
on the 'outside' of the complex, is important. Such orbitals enable an inter-
action between adjacent molecules, the interaction giving rise to dispersion
patterns of the type that have been discussed. Evidently, some-external-
ligand orbitals will be the most affected.
We can at once understand two things. First, why ligand field theory
works so well (in a typical coordination compound the d orbitals of the
transition metal ions are physically well separated). Secondly, why our
understanding of ligand orbitals is much less detailed and less general than
that of metal ions. The energy levels that ligand orbitals subtend are likely
to be environment-sensitive. The (complex) species B will only partially
insulate the ligands of one (complex) A molecule from those of another.
Change B and these residual A-A interactions will also change. Further,
interactions (strictly, over k-space) between ligand orbitals of A and external
orbitals of B may occur.
With this discussion in mind, it is helpful to reconsider the distinction
between charge-transfer bands introduced by Robins and Day, the class 1,
class 2 and class 3 dis<inction that was discussed in some detail in Section
8.11. An example of class 2 behaviour (clear interaction but chemical identity
retained) is provided by the coordination compound KFe"[Fem(CN) 6 ).
It can at once be understood why the Fe" and Fem ions retain their
individuality. They are insulated from each other by the eN- ligands.
However, a little consideration suggests that it is a delicate matter. They
have t~ 9 and t~ 9 configurations, respectively, and, following the discussion
of Section 6.6, these t 29 electrons are partially delocalized into the CN-
n antibonding orbitals. It follows that they have a non-zero probability of
being at the periphery of the complex. A mechanism for metal-metal
interaction does exist; evidently, its effect is small. Of course, although this
explains in outline why this compound is class 2 and not class 1 (no
interaction), there is ample evidence for small metal-metal interactions of
the type that have just been considered. Their effects are most important in
magnetochemistry and were discussed in Section 9.11. A typical class 3
compound is Ag 2 F (Table 8.4). The contrast with KFe[Fe(CN) 6 ] is clear.
In the crystal, the orbitals involving the Ag and Ag + ions must be delocalized
over the entire 'molecule' (Ag 2 F has an extended, not a molecular, structure).
Interactions between the silver ions and atoms will surely occur, leading to
dispersion, the observed broad band absorption and a black compound
results. Any characteristics of Ag and Ag+ are completely lost.
There are aspects of the discussion in the last paragraph which can be
extended well beyond the compounds covered by the Robins and Day
classification. Consider the following problem. NiO, when pure, is a pale
solid. TiO and VO both have the same structure as NiO but are black and
are almost metallic conductors of electricity, whereas NiO, when pure, is an
insulator. That there is some electronic explanation for this pattern is made
evident by the observation that when suitably doped with an impurity (Li 2 0
is the one usually cited) NiO also becomes black and highly electrically
Solid state and coordination compounds I 427
Fig. 17.19 11tanium(ll), with a ti, configurabon,

in a simple structure such as that of NaCI {that
adopted by TiO, shown here), has the possibility
of direct metal-metal interaction through the
overlap of t2g orbitals as shown. Such
interaction has the effect of giving a broad,
incompletely filled, band and so bhe compound
is a conductor of electricity and is black.
conducting. What is the explanation for this pattern? To avoid the problem
of the electronic levels introduced by an impurity we will consider just the
origin of the differences between TiO, VO and pure NiO. All three
compounds crystallize in the NaCl, rock salt, structure-which means that
each metal ion is octahedrally surrounded by oxide anions, so that the
ideas of simple crystal field theory should be applicable. The electronic
configurations of the three ions are d 2 , d 3 and d 8 , respectively. In an
octahedral crystal field these will become ~~., ti. and ~~.e;. Only the first
two have incompletely filled t 2• shells. As Fig. 17.19 shows, in the TiO
structure overlap can occur between such t 2 • orbitals, leading to an
incompletely filled band structure and an explanation for the electrical
conductivity. For NiO, a comparable band structure will be less available.
First, the partially occupied d orbitals are e., which do not overlap
each other in the same way-they point towards an oxygen, not a nickel; the
electrons are more likely to be localized on the metal atoms. Secondly, the
band structure in NiO will involve participation of the 0 orbitals. Finally,
it turns out that the Ni d orbitals are a bit more contracted than are the Ti
or V, reinforcing the idea that they will tend to be relatively insulated from
one another. We can confidently predict that NiO will have a lower electrical
conductivity than the other two oxides. As we will see, the difference in colour
also follows. One important point: because it was important to show the
interaction between several metal atoms, Fig. 17.19 does not show just a
single primitive unit cell. When multiples of the primitive unit cell are shown
in this way it is a good idea to check whether the interaction between
adjacent units is in-phase (as here). If so, then moving to the zone boundary
the interactions will become out-of-phase and so the interaction shown in
Fig. 17.19, at k = 0, at r, is at the bottom of the band (assuming that overlap
with other bands can be neglected).
Some generalizations have been made about those features which enhance
the probability of band formation of the type just discussed in transition
metal compounds:
1. the cation occurs early in the transition metal series;

2. the cation is in the second or third transition metal series;
3. the ligand is not too electronegative;

4. the actual charge on the cation is not great;
5. the ligand is small.
Of these features, 1, 2 and 5 are either evident from the discussion above or
are simple extensions of that discussion. Points 3 and 4 derive from the fact
that highly ionic materials (hard cations and anions) have large band gaps.
That in NaCI, for instance, is ca. 8.5 eV; that in KF is approximately 11 eV.
In contrast, a typical semiconductor such as GaAs, composed of soft species,
has a band gap of 1.4 eV. In the present context, these generalizations find
some reflection in the observation that unlike NiO, the compounds NiS,
NiSe and NiTe are quite good electrical conductors and black in colour.
Finally, a comment on the content of Chapter 15. In Fig. 15.25,
density of state diagrams for Ag and V were given and compared to M 6
clusters of the same elements. The general conclusion was that there are
distinct similarities between M 6 and the bulk metal. Perhaps this was not
altogether surprising. After all, the isolated-atom orbital energies are the
same and similar metal-metal interactions are involved. The extrapolation
enabled by our present discussion is not too surprising either. The more
metal-metal interactions in a cluster, the more it is likely to resemble the
solid metal. The data in Fig. 15.25 were for isolated octahedra. We may
safely predict that in a solid composed of these octahedra the corresponding
density of states would even more closely resemble those of the pure
metal.
17.7 Spectra of crystalline materials

In the previous section the spectra of crystalline materials were briefly
alluded to. In this section the key question to be addressed is what are the
spectroscopic selection rules for a solid; how do they differ from those for
isolated species?
When dealing with molecules, and therefore point groups, the general
selection rule is simple: for a transition to occur the relevant integral
must be totally symmetric-that is, transform as the irreducible representation
of the relevant point group. The totally symmetric irreducible representation
is the one which has a character of + 1 for all operations of the point group
(all other irreducible representations, effectively, have equal amounts of
+ 1 and -1 values so that integrals having any of these other symmetries
have integrals over all space of zero). The general selection rule in the solid
state is similar and for the same reason. The relevant integral must have the
symmetry properties ofk = 0. If it did not, in some parts of space the integral
would be positive and in others negative. The orthogonality properties of the
different irreducible representations ensures that in such an integral over
all space the positive and negative contributions combine to give an answer
of precisely zero. But this is not all. The vast majority of spectroscopies
of interest to the chemist have one feature in common. This is that they
involve radiation with wavelengths which are enormous compared with the
magnitude of a typical translational vector of a crystal. Such a vector
might be of the order of 5-10 A, and that is the wavelength of a rather
soft X-ray. The nearest that the chemist normally gets is in ultraviolet
Spectra of crystalline materials 1 429
spectroscopy, with wavelengths of perhaps 1000 A or more. So, whatever the

light wave does to the contents of a unit cell (however the unit cell is defined),
it will effectively be the same as that which it does in adjacent unit cells. It
takes a great many unit cells before a difference becomes apparent. And only
if our reference unit cell is sensitive to what happens thousands of Angstroms
away will it make any difference. In general, no such sensitivity exists9 so
that, effectively, in all spectroscopies of interest to the chemist, all unit cells
respond in the same way. The phenomena observed are translationally
invariant, and transform as k = 0. 10 However, it has just been seen that the
entire integral must transform as k = 0. Although for ease of explanation it
was convenient to talk of 'the light wave' earlier in this paragraph, as far as
an integral is concerned it is the corresponding operator which is of
importance. It, too, must transform as k = 0. We are left with the ground
and excited state wavefunctions. The picture just developed clearly indicates
that the most important transitions will be those in which they too transform
as k = 0 (since all unit cells respond in the same way).
One exception, which is more than a mathematical curiosity, is to be
noted. If the ground state wavefunction transforms ask, and the excited state
wavefunction transforms as - k,, then the integral transforms as k = 0. A
travelling wave moving in one direction (k,) is cancelled out by a travelling
wave moving in the opposite direction (- k, ). It can at once be understood
why the spectral bands of solids are so often broad (and the compounds are
black if they are also electrically conducting-there is a low-lying excited
band). For the ground state the k, can span the entire Brillouin zone, together
with all the attendant dispersion. Similarly, the excited state - k, can span the
entire Brillouin zone, together with its attendant dispersion. In Fig. 17.20 is
shown for a one-dimensional example how this can even give rise to an
apparent band splitting if ground and excited states have matching, but
different, high densities of states in the appropriate regions of k space.
Although complications of the sort that have just been described are well
known, they are least likely to occur when the ions or molecules making up
the crystal are well separated from each other and/ or the structural
interactions between the structural units rather weak. This is the case with
most coordination compounds. For 'concentrated' species such as anhydrous
metal oxides and halides complications are expected and, indeed, occur-as
was seen in the last section. What of the case where no complications are
found? Here, the effective building block of the crystal is the (primitive) unit
cell. Whatever happens in one unit cell happens in all others. This is the
ultimate justification for the use of unit cells in diagrams such as Figs. 9.8
and 12.4. In considering the spectra of solids just the contents of one
(primitive) unit cell need to be considered. When interactions between
9 A problem arises for some members of a series of related crystal structures for some
compounds- ZnS and CSi (carborundum) are well-known examples-in which translational
vectors of up to thousands of Angstroms can occur (so-called polytypism). However, the
long-range communication between unit ce11s which must be involved may be rather differently
mediated although the mechanism is not known (for a speculation on one possibility see S. F. A.
Kettle, J. Cryst. Spect. Res. (1990) 20, 59).
10 It is important to remember that the theory developed in this section concerns only pure
translation operations. It contains no rotation or reflection operations; inclusion of these into
the theory is required to explain the anisotropies associated with the absorption and emission
of radiation by real crystals.
t
E
(c)
0 ki~
(Ground (b)
state)
-k;-
(Excited individual structural units becomes important (as in TiO and VO) the band
(a) state)
structure of the solid means that quite a different approach has to be adopted,
Fig. 17.20 (a) The rather different dispersions one that takes account of the dispersion of the energy bands across the
of the ground and excited states involved in an Brillouin zone. Clearly, success in this latter endeavour is more likely to come
electronic transition, together with (b) the for simple species-and these are just the ones most likely to show the phenom-
corresponding density of states. The resu~ing
spectrum is shown as (c); the observed maxima ena which require it. Not surprisingly, the work published by those concerned
and minima can be correlated with features in with simple species tends to be rather different in content and interpretation
the dispersion curve. ln principle, similar to that published by those working with large coordination compounds.
patterns apply to all tonns of spectroscopy
carried out on solids, not just electronic Throughout this chapter the assumption of an infinite, perfect and pure
spectroscopy. crystal has been made. In general, these assumptions are sufficiently valid
for the general results obtained to apply to real-life crystals. Two important
exceptions are to be noted, however. The first has already been mentioned
but not developed (it would require a chapter of its own). This is the presence
of impurities. If these, usually by intention but sometimes by accident, have
energy levels which interleave the band gaps in the host material then the
properties of the host material are dramatically changed. The model
developed in this chapter is an adequate starting point for a discussion of
these changes but, of itself, is not adequate to deal with them. The assumption
of an infinite and perfect crystal is usually valid. When the size either of the
crystal or of the perfect blocks within a crystal become too small, however,
dramatic changes occur. For instance, cadmium phosphide, Cd 3 P2 , is a
semiconductor with a band gap of only 0.5 eV. Not surprisingly, it is black
(the dark colours of many main group element sulfides similarly reflect
small band gaps). If small particles of the compound are prepared as colloids
and then separated the colour of the product depends on the particle size.
So, at 30 A they are brown; as the particle size diminishes further, in rapid
succession they become red, then orange, then yellow and finally white at
about 15 A. The band gap has increased to 4 eV. These colour changes have
nothing to do with the lighter colours commonly displayed by coloured
materials when they are finely ground (this is a purely optical effect which
may be cancelled by changing the refractive index of the material surrounding
the finely ground crystals). Such particle size effects, and the breakdown in
the simple model presented in this chapter, are likely to become of increasing
importance as the search continues for molecular electronic devices.
Questions 1 431
Further reading and Zinc' by T. D. Brennan and J. K. Burdett, Inorg. Chern.

(1993) 32, 746. To fully understand this paper, it would be
The solid state is covered in most current inorganic texts but helpful to have available a picture of the first Brillouin zone
usually more at a factual level than a theoretical. Some parts of a fcc metal, such as that given in Fig. 17.11.
of the present chapter would be explored better in texts on A good general reference is Solid State Chemistry and
solid state physics. For rather obvious reasons, the two following its Applications by A. R. West, Wiley, Chichester, 1984.
references provide good stepping stones between the content of Fermi surfaces (and some beautiful diagrams of them)
the present chapter and selected wider literature references
will be found in The Fermi Surfaces of Metals by A. P.
(given in the reference sections of these papers):
Cracknell, Taylor and Francis, London, 1971.
• 'The Brillouin Zone, an interface between Spectroscopy and Methods of determining Fermi surfaces are particularly
Crystallography' by S. F. A. Kettle and L. J. Norrby, J. Chern. well described in Chapter 14 of Solid State Physics by N. W.
Educ. (1990) 67, 1022. Ashcroft and N. D. Mermin, Saunders College, Philadelphia,
1976.
• 'Really, Mr. Bravais, your lattices are all primitive' by S. F. A.
Band Theory of Solids. An Introduction from the Point of
Kettle and L. J. Norrby, J. Chern. Educ. (1993) 70, 959.
View of Symmetry by S.L. Altman, Oxford Science Publications,
Also relevant (even though the title name has not been used Oxford, 1991, develops the approach begun in this chapter.
in this chapter) is 'The Wigner-Seitz Unit Cell' by S. F. A. The interface between metal clusters (Chapter 15) and
Kettle and L. J. Norrby, J. Chern. Educ. (1994) 71, 1003. bulk metals is explored in 'Bonding in Molecular Clusters
An excellent and readable account which follows a devel- and Their Relationship to Bulk Metals' by D. M. P. Mingos,
opment quite similar to that of the present chapter is given in Chern. Soc. Rev. (1986) 15, 31 (this reference uses some of the
Solids and Surfaces, a Chemist's view of bonding in extended language of Chapter 15), and 'Colloidal Semiconductor Q-
structures by R. Hoffmann, VCH, New York, 1988. Particles: Chemistry in the Transition Region Between Solid
A short, easy-to-read article which has many examples State and Molecules' by H. Weller, Angew. Chern., Int. Ed. Engl.
of band structures is 'Electronic Structure of Elemental Calcium (1993) 32, 41.
Questions the compound was studied in solution. What conclusions

follow from these observations?
17.1 In Fig. 17.21 is shown a centred two-dimensional lattice
with two structural dots to each unit cell. Show that it is 17.3. In the text two-dimensional crystals have often been
possible to redefine the array in terms of a lattice of which the discussed in preference to real-life three-dimensional crystals,
(primitive) unit cell contains only a single dot. simply because it is so much easier to draw and understand
pictures of two-dimensional crystals. Fortunately, the extension
to three-dimensional crystals is almost trivial-no new principles
• • • • • arise. You are given the task of rewriting Section 17.2 in a
• • • • • manner such that it deals exclusively with the three-dimensional
case. Make a list of the key points at which you would need
• • • • • to make changes to Section 17.2 in such a rewriting.
Fig. 17.21 Question 17.1. 17.4. The Brillouin zone for the simple (primitive) cubic
lattice is a (simple!) cube. The Brillouin zone for the face-
17.2. When studied in the crystalline material the Raman centred cubic lattice is pictured in Fig. 17.11; that for the
band associated with the totally symmetric (breathing) v(Cr...CO) body-centred cubic is shown in Fig. 17.16. Show that all three
mode of Cr(C0) 6 was little different from the same band when have 0, symmetry.
Appendix
Consider the ligand ethylenediamine, en, which is typical of many. When it

chelates to a metal atom a five-membered ring is formed, as shown in Fig.
A 1.1. Is this ring planar? The answer is no, it is puckered. It is easy to see
why this should be so. In a regular pentagon each internal angle is 108°, not
far from the regular tetrahedron value of 109.SO. However, if the chelate ring
is part of an octahedral complex then the N- M- N angle will be close to 900.
This will open up the bond angles at the four atoms of the en ligand in
the ring to an average of 112.5°; add to this the fact that the M- N bonds
will be longer than the C--C and C- N and it is clear that some steric strain
is likel y to exist. This strain can be relieved at little or no cost if the ring
puckers. Having recognized the possibility of puckering, the next thing is to
recognize that two different puckered forms exist; much of this appendix is
concerned with them.
Any three points in space define a plane, unless the three points lie in a
straight line. Consider the plane defined by the metal atom and the two
nitrogen atoms of an en ligand in a complex. Our concern is with the two
carbon atoms in the ligand. Relative to the plane, these can be arranged in
either of the two ways shown in the top part of Figure A1.2. The two
configurations are denoted i, and c5, as indicated. An easy way of remembering
which configuration is which is to view the chelate ring with the metal atom
closest to the observer, as shown in Figure A 1.2. From this viewpoint, in
the c5 configuration the bonds involving the two carbon atoms roughly trace
out a c5 and for the i. configuration they trace out a i.. This is shown at the
bottom of Figure A 1.2.
This is not the end of the conformational possibilities that have to be
considered. In an octahedral complex there are two different ways of
arranging three chelate rings, denoted by the upper case letters A and tl..
They are shown in the upper part of Figure A 1.3. The simplest way
of remembering the difference between A and tJ. is to put arrows on the
tJ. in the natural, clockwise sense, as is done at the bottom of Figure A 1.3.
These arrows serve as a reminder that the tJ. configuration resembles a
normal, right-handed screw (the A configuration resembles a left-hand
screw).
Finally, the possible configurations of the three individual ligands (i,, c5)
Fig. Al.l Typical puckered five-membered ring are brought together with the two possible arrangements of a set of
fonmed by chelated en ligand. three of them in an octahedral complex (A, tl.). A set of eight possible
Appendix 1 1 433
c c
Q:,b r1;.0
\I M \ I
c c
=P- _ \'_
FIJI. AL2 The 2 and b ring configurations.
L-CJ---~---~' L--~-~--~---~\
A
Fig. AL3 The A and 6 configurations of three
chelate rings in an octahedral complex.
( J
configurations is obtained, all of which must be assumed to be present in a
solution of a complex such as [Co(enh]3+:
A(J.,V,) A(Mb)
A(J.J.b) t.(bbJ.)
A(lM) A( b).).)
A(bM) t.().I.J.)
The first member of each pair (those on the same line) is the mirror image
of the second (they are enantiomorphic pairs). Generally speaking, members
of families of such compounds rapidly interconvert. However, they have been
extensively studied, usually by NMR (often with somewhat more exotic
ligands than en) because they provide an opportunity of investigating the
434 I Appendix 1
effect of steric and other influences on the relative stabilities of octahedral

complexes. In crystals of salts of the [Co(enh]J+ cation, only one of the
eight forms listed above will usually be present. But which, and why? These
questions do not seem to have general answers, but they are fascinating
nonetheless.
For simplicity, this appendix has considered only five-membered rings.
Six-membered chelate rings are also quite common and follow a similar
pattern; Figure Al.2 would be adapted to a six-membered ring by inclusion
of an additional atom in the plane shown in each of the diagrams in this
figure. The subsequent discussion would have been essentially unchanged.
Finally, the ligand that has been considered is a rather symmetrical one,
en. A related but less symmetrical ligand has one C-H bond of en replaced
by a C-CH 3 . This ligand is 1,2-diaminopropane (usually denoted by pn); it
contains an asymmetric carbon atom (that bonded to the methyl group). In
principle, therefore, it is capable of separation into optically active isomers
and our discussion could be elaborated to include this. This is not some
academic exercise, a study of an increasingly remote problem. The naturally
occurring ligands of bioinorganic chemistry are invariably optically active
and an understanding of the properties of their complexes would not be
possible if the possibilities for as ligand such as pn were not well understood.
The subject matter of this appendix is comprehensively treated by C. J.
Hawkins in Absolute Configuration of Metal Complexes, Wiley, New York,
1971.
Appendix2
Valence shell electron pair
repulsion (VSEPR) model
This appendix contains an outline of the most successful of the simple

approaches to molecular structure, one that is described in most current
introductory and inorganic texts. In contrast to the treatment given in such
texts, the present is more concerned with an analysis of it, in order to assess
whether it correctly highlights the key factors involved in determining
molecular geometry. Whether, for main group complexes at least, it may be
said to provide an explanation (as opposed to a prediction) of their
geometries. The model is based on the postulate that repulsion between
valence electron pairs surrounding an atom lead to the observed atomic
arrangement around this atom, an arrangement in which these electron pairs
are as far apart as possible. That is, it is postulated that bonding forces are
not the vital factor in determining molecular geometry. This view finds
support in the observation that relatively little energy is required to excite
those vibrations which correspond to bond angle changes. These, of course,
change the geometry around the central atom . The actual energy is about
one half of that required to excite the corresponding bond stretching
vibrations.
The simplest way of looking at the VSEPR approach is to regard each
valence electron pair as being represented by a point, the points being
constrained to move over the surface of a sphere drawn around the central
atom . As the points (i.e.valence shell electron pairs) repel each other, the
most stable molecular arrangement will be that in which the points are as
far apart as possible. If each valence shell electron pair is called P, then the
following geometries of the Ps around the central atom M are at once
predicted:
MP2 linear
MP3 equilateral triangular
MP4 tetrahedral
MP6 octahedral
Comparison with the geometries which are discussed in Chapter 3 will show
that for main group elements these predictions are rather good. So,
[BF3 · NMeJ , in which there are four valence electron pairs around the
boron, is approximately tetrahedral, the distortion being consistent with the
436 1 Appendix 2
fact that one of the valence electron pairs (that involved in the B-N bond)
would not be expected to have exactly the same spatial distribution as the
other three, so that (B-N)--(B-F) bond repulsions will be slightly different
from (B-F)--(B-F) repulsions. Similarly, extending the discussion to a
compound which would not normally be regarded as a complex, the oxygen
atom in water is surrounded by four electron pairs (two from the 0-H bonds
and two from the lone pairs on the oxygen atom). Repulsion between these
electron pairs leads to a tetrahedral distribution and, in accord with this,
the H--()-H bond angle in water has roughly the tetrahedral value, the
deviation from regularity being in accord with the suggestion that the centre
of gravity of the electron density in the 0-H bonds is further away from the
oxygen atom than is the case for the lone pair electrons. Such arguments
lead to the prediction that the relative magnitude of the electron repulsion
is in the order:
lone pair-lone pair> lone pair-bonding pair> bonding pair-bonding pair
an order which almost invariably explains distortions from symmetric

structures.
So far, coordination numbers five, seven, eight, nine and above have not
been discussed. The case of nine-coordination is simple, for the predicted
geometry is that discussed in Section 3.2.7, the three-face-centred trigonal
pyramid, and is the one commonly observed. The others are not so clear-cut,
because the valence electron pair arrangements which involve minimum
electron repulsion are not intuitively obvious. This difficulty can be removed
by assuming some form for the repulsive energy between the points
constrained to move over the surface of a sphere (the points which represent
the valence electron pairs). For instance, it could be assumed that the energy
varies as 1/r", where r is the interpoint distance, and the arrangement of
lowest energy thus determined. Unfortunately, the results depend to some
extent on the value given ton, (a value of between 6 and 12 is usually chosen)
and in any case it is not certain that the real energy varies as 1/r"-an
exponential might be more appropriate, for instance. In any case, it is far
from certain that deformable and diffuse electron densities interact in a way
that is well modelled by undeformable charges located at precise points. Even
with the VSEPR model, it seems safest to conclude that, for coordination
numbers five, seven, eight, ten and above, factors other than valence shell
electron repulsions may play a part in determining molecular geometry.
These are cases for which, as discussed in Chapter 3, there seems little
energetic difference between several possible geometries.
Although the bones of the VSEPR model have been given above, it can
be elaborated further-for instance by arguing that double bonds are bigger
than single and that electronegative atoms suck electron density out of
bonding pairs, making them smaller, so that they can be squeezed more
tightly together. This last refinement is used to explain why the bond angle
in NF3 is smaller than that in NH 3, a problem to which we shall return.
On the whole, the VSEPR method predicts the geometries of main group
compounds and complexes rather well. This is not the same thing as saying
that it provides a correct explanation of molecular geometry. Indeed, in the
opinion of some, its status and use is just that of prediction, that of an aid
to getting the right answer. The VSEPR method only really works for main
Appendix 2 1 437
group elements (the d electrons of transition metals do not consistently have

the steric effects envisaged by the model; the f electrons of the lanthanides
and actinides have no generally recognized steric effects). Amongst those
main group molecules for which it fails are some, such as Li 2 0 (gaseous),
which is linear and not bent like water. The explanation which is offered is
rather naive: these molecules are ionic and so the model does not apply.
What such an explanation is really saying is that the geometry is determined
by nuclear-nuclear repulsion (lithium-lithium in Li 2 0; this repulsion is a
minimum for the linear molecule) and this interaction is ignored in the
VSEPR model. However, it is difficult to believe that this is the explanation
for the observation that the anion BrF6, with seven valence shell electron
pairs around the bromine, is octahedral in both solution and solid. In
contrast, the related species CIF6 is believed and IF6 known to be a
distorted octahedron, as is XeF6 , all species with seven electron pairs around
the central atom. For heavier elements, in particular, other effects come into
play-correlation effects (which are mentioned frequently in this book) and
relativistic effects (which are also mentioned, but less frequently). So, the fact
that a species which may be regarded as derived from the NH: ion,
N(AuPH 3 ):, is tetrahedral but its arsenic counterpart, As(AuPH 3 ):, is
square pyramidal, is probably to be associated with the operation of these
two effects in the latter compound (and not to Au-Au bonding, although
this explanation has been suggested).
Although the VSEPR model is commonly applied to complexes formed
by main group elements, data do not exist which would enable us to assess
its accuracy for these. Instead, we are forced to use data on simpler molecules
and we shall consider NH 3 and NF3 • Table A2.1 gives the results of some
accurate calculations on these two molecules. Of particular interest is the
origin of the energy barrier to inversion in them-at room temperature an
isolated ammonia molecule turns inside out, like an umbrella in a high wind,
many, many times a second. NF3 does not. The problem is relevant to our
discussion because at one point in the umbrella motion all of the atoms of
ammonia are co-planar; at either extremity they have the observed pyramidal
shape. Here, then, we have two geometries which can be, and have been,
compared in detail. One is the most, and the other the least, stable of all of
the entire set of geometries mapped out by the umbrella motion. The problem
was investigated theoretically by carrying out accurate calculations on the
planar molecules (corresponding to the top of the energy barrier) and also
on the pyramidal species. In order to identify the origin of the energy
Table A2.1 Change in energies between planar and most stable pyramidal geometries of NH 3 and NF3 . Positive quantities mean
that the pyramidal form is the more stable. All energies are in atomic units (1 au = 2626 kJ mol- 1 )
Molecule Change in Change in Change in Change In

bonding energy repulsive energy electron-electron repulsion nuclear-nuclear repulsion
NH 3 -0.475 0.489 0.158 0.290
NF3 7.626 -7.519 -3.789 -3.495
The 'Change in repulsive energy' entries are not the sum of the corresponding two following repulsion entries because of small changes in kinetic energy which have
not been explicitly included. There are very small changes in bond lengths between the two geometry sets which have been ignored in the discussion in the text.
Data from A. Schmiedekamp, S. Skaarup, P. Pulay and J. Boggs, J. Chern. Phys. (1977) 66, 5769.
438 I Appendix 2
barrier-and hence the reason why the molecules are pyramidal rather than
planar, for each of the three energy terms identified in this appendix-bonding,
electron-electron repulsion and nuclear-nuclear repulsion-the change in
going from a planar to a pyramidal molecule was calculated. These data, the
numbers in Table A2.1, can then be used to understand why NH 3 and NF3
are pyramidal rather than planar. In Table A2.1 a positive quantity means
that this contribution stabilizes the pyramidal molecule relative to the planar.
The first thing to be noted about Table A.2.1 is that the entries in the
'change in bonding energy' column are non-zero. Contrary to the VSEPR
model's assumption, bonding is a function of bond angle. Even more striking
is the different pattern of the signs of the entries in the 'change in bonding
energy' and 'change in repulsive energy' columns. The lesson is a salutary
one. The sum of the two entries for NH 3 and also for NF3 is positive-the
pyramidal molecule is the stable one-but the two molecules achieve this
result in quite opposite ways. Pyramidal NH 3 is stable because this geometry
minimizes the repulsive forces-bonding is a maximum in the planar
molecule. In contrast, NF3 is pyramidal because this geometry maximizes
the bonding; repulsive forces are a minimum in the planar molecule, in
contradiction to the VSEPR model's predictions. The most charitable thing
that can be said about the second set of entries in Table A2.1 is that they
show that electron-electron and nuclear-nuclear repulsion energy changes
are of comparable importance in determining molecular geometry (for NH 3
the nuclear-nuclear are the more important). It can be argued that the
electron- electron terms in Table A2.1 include all the electrons in the
molecules and not just those which are the focus of the VSEPR model.
However, such an objection misses a key point. If other electron-electron
repulsion interactions are important and yet ignored by the model then this,
too, is a weakness. One should not argue too much from one example, yet
that which we have given is a dramatic one-in both cases considered,
electron-electron repulsion makes a smaller contribution to determining the
molecular shape than do either bonding or nuclear-nuclear repulsion. One
is forced to remember a difficult lesson; a model that leads to a correct
prediction is not necessarily a correct model. A further indication of the
existence of concealed assumptions within the VSEPR model is provided by
some other molecules related to ammonia. The C-N-c bond angle in
triisopropylamine, N(CHMe 2 )J, is 119.2°-the N-c 3 framework is essen-
tially planar, presumably because of steric interactions between the isopropyl-
amine groups. In trimethylamine, N(CH 3 )J, the C-N-c bond angle is 111 o
but in the corresponding fluorinated compound N(CF 3 h it is 118°. This
change is surprising because fluorination has little effect on bond angles in
alkanes. In N(C 2 F5 h the C-N-c bond angle is 119.3°. Clearly, there are
factors of importance in determining these angles that are not included in
the simple VSEPR model. Nonetheless, the value of this model as a usually
correct predictor of the molecular geometries of compounds of main group
elements is unquestioned. Further, the sort of analysis which has been given
above for NH 3 and NF3 is not the only one possible. It has been much more
common to ask a different question: as to whether there is theoretical support
for the electron repulsion inequality sequence:
lone pair-lone pair> lone pair-bonding pair> bonding pair-bonding pair

Appendix 2 1 439
Most workers who have considered this question have concluded that there
is. Crucial is the way that a single entity, the electron distribution in a
molecule, is divided up to give distinct lone and bonding pairs. What has
usually been done is to adopt some criterion which seems to lead to a division
as close as possible to that made, more qualitatively, by the VSEPR model
itself.
We have reached a point at which opinions differ; the reader is entitled
to his or her own. Those who would like the benefit of a view which is
intermediate between the two extremes presented in this appendix might find
it helpful to read an article by A. Rodger and B. F. G. Johnson in Inorg. Chim.
Acta (1988) 146, 37. The most recent exposition of the VSEPR model is to
be found in The VSEPR Model of Molecular Geometry by R. 1. Gillespie and
I. Hargittai, Prentice-Hall (Allyn and Bacon), New York, 1991.
Appendix3
Introduction to group theory
Figure A3.1 shows a square planar complex and something of the symmetry
which it possesses. Most evident are the pictorial representations of mirror
planes- reflection in any of these infinitely thin, double-sided mirror planes
turns the square planar complex into itself. The solid black shapes in the
figure indicate rotational axes-- shaped for twofold and • for fourfold.
Although Fig. A3.1 shows symmetry elements- mirror planes and rotational
axes-in group theory it is the corresponding operations which are of interest;
the consequences of the act of reflecting or rotating. If the operations
indicated in Fig. A3.1 are counted one finds a total of 16. In Table A3.1
these 16 operations are listed across the top. Some of the operations which
are very closely related are grouped together- rotation by ± 90° about the
c4 (fourfold rotation) axis; rotation by 180° about equivalent c2 (twofold
rotation) axes; reflection in equivalent mirror planes (denoted a.). Symmetry
operations which are grouped together are said to be in the same class. Each
number in the table is known as a character and the whole table is called a
character table. This group of 16 symmetry operations is called the D4 , group,
and the character table is the D 4 , character table. Down the left-hand side
of the character table are listed some of the symmetry symbols used in the
text (in Table 7.2, for instance). All such symbols originate in character tables.
Each symmetry symbol is associated with a unique row of characters-no
Table A3.1 The D4h character table
E 2C4 c2 2C2 2C2 2$4 2u;,

D4"
"" 2ud
A'<~ 1 1 1 1 1 1 1 1 1 1
A2S 1 1 1 -1 -1 1 1 1 -1 - 1
B'<~ 1 -1 1 1 -1 1 -1 1 1 -1
B2g 1 -1 1 -1 1 1 -1 1 -1 1
Eg 2 0 -2 0 0 2 0 -2 0 0
Alu 1 1 1 1 1 -1 -1 -1 - 1 -1
A2u 1 1 1 -1 -1 -1 -1 -1 1 1
Blu 1 -1 1 1 -1 -1 1 -1 -1 1
B2u 1 -1 1 -1 1 -1 1 -1 1 -1
Eu 2 0 -2 0 0 -2 0 2 0 0
Appendix 3 1 441
This, the axis or highest symmetry, is rather

complicated because it is the axis or three
distinct symmetry operations. The first is the
operation of rotation by 90i , either clockwise or
antlclockwise, each denoted by C4 . The second
is the operation or rotation by 180i. denoted C, .
The third is a composite operation: rotate by 90i
and then reftect in the " " mirror plane (see
below). Denoted S4 , this operation takes the top A pair of equivalent
of one ligand Into the bottom of the next (cl. c•. mirror planes, " •
which sends top into top).
A pair of equivalent
mirror planes, a ' ll
Horizontal
mirror plane, "•
The other C2
A pair of equivalent
A pair of equivalent two-fold axes, C2. two-told axes, Ciz. The second is
The other is perpendicular to this one. on the other side of the diagram.
FIC. A3.1 Symmelly elements and operations two rows are identical, although inspection reveals that the characters block
associated with a square planar complex. There into four 5 x 5 squares, the characters within one square being very simply
are two other symmetry operations, not shown
on the diagram. The first, apparen~y trivial related to the corresponding characters within another. This fundamental
operation, is 'leave everything alone ' and is building unit of the D4 , character table is, in fact, the character table of the
denoted E. The other is the operation of D4 group (which has only the E, 2C 4 , C 2, 2c; and 2C~ operations of the
inversion in the centre of symmetry (which is at
the central atom) and is denoted i. D4 , group). In Table A3.1 any one character corresponds to a symmetry
symbol, (its row) and to one or more symmetry operations (its column). The
character has the property of telling us how something which carries its
particular symmetry symbol behaves under its particular symmetry operation.
So, something of A 2 • symmetry is multiplied by I (i.e. turned into itself) both
by a C 4 rotation (be it clockwise or anticlockwise) and also by the C 2
rotation. However, it is multiplied by -I (i.e. turned into itself with all signs
reversed) by the c; and q rotations (any of either pair of operations). The
442 I Appendix 3
&u
z
Rg. A3.2 (a) The p, orbital has A2 u symmetry

while (b) the Px and P, orbitals together have
Eu symmetry. Because Px and P, are
interconverted by some of the D4 h group (e.g.
C4 ), they must be considered as a pair.
p, orbital shown in Fig. A3.2(a) has A 2 u symmetry and the reader should
check that it behaves under these operations in the way that has just been
described. Those symmetry symbols which, for the leave-alone operation (E),
have characters of 2, describe the behaviour of two independent things
simultaneously. The number 0, for these species, means either that the two
things are interchanged in some way or that one is changed into itself ( + 1)
and the other into the negative of itself ( -1) so that, together, a character
of zero is obtained ( + 1 - 1 = 0). The number -2 means that under the
symmetry operation each of the two objects is turned into itself with all signs
reversed. So, Px and p, (Fig. A3.2(b)) together have E" symmetry. The reader
should check that the D4 • character table gives an accurate description of
their behaviour under the symmetry operations.
The direct product of two symmetry species, say A 29 x B 19 , is formed by
multiplying together in turn their characters under each symmetry operation,
thus
D4h E 2C4 c2 2C2 2C2 254 "h 2ud 2ad
A2g 1 1 1 -1 -1 1 1 1 -1 -1
a,. 1 -1 1 1 -1 1 -1 1 1 -1
A2g x a,. 1 -1 1 -1 1 1 -1 1 -1 1
Comparison with Table A3.1 then sht>ws that the set of characters produced
is, in fact, that labelled B 29 • We say that 'the direct product of A 29 and B 19
is B 2 ;. ForE-type symmetry species direct products are a bit more difficult.
Consider E 9 x Eu:
D4h E 2C4 c2 2C2 2C2 254 (Jh 2ud 2ud
Eg 2 0 -2 0 0 2 0 -2 0 0
Eu 2 0 -2 0 0 -2 0 2 0 0
Eg X Eu 4 0 4 0 0 -4 0 -4 0 0
Appendix 3 1 443
This direct product does not correspond to a symmetry species in Table

A3.1. It does, however, correspond to a sum of them. Consider the sum
Alu + A2u + Blu + B2u:
o•• £ 2C4 c2 2C2 2C'2 2S4

"• 2trd 2rr;,
Alu 1 1 1 1 1 -1 -1 -1 -1 -1
A:~g 1 1 1 -1 -1 -1 -1 -1 1 1
alu 1 -1 1 1 -1 -1 1 -1 -1 1
a2u 1 -1 1 -1 1 -1 1 -1 1 -1
Alu + A2u + alu + a2u 4 0 4 0 0 -4 0 -4 0 0
This is the same as that generated by E9 x Eu, so we can write:
Eg x Eu = A1u + A2u + 81u + B2u

'the direct product of E9 and E" is A 1u plus A 2 u plus B1u plus B 2 .' (note that
the product of suffixes g x u gives u). What have been called 'symmetry
species' are often also called irreducible representations (of the group), so,
'the A 2 u irreducible representation of the D4 h group'. The two expressions are
interchangeable and often mixed with each other to avoid excessive repetitions.
Similarly, one may well meet statements such as 'the P. orbital of the central
metal atom transforms as the A 2 u irreducible representation'. The direct
product E9 x E" is an example of a reducible representation, which, as has
just been demonstrated, may be expressed as a sum of irreducible components
(which themselves cannot be reduced further, which is why they are called
irreducible).
In Table A3.2 is given a complete table of direct products for the D 4
group. That for the D4 h group would be four times as large; it can be obtained
from that for D4 by adding suffixes according to the rules that
gxg=UXU=g; gxu=u
Notice that in Table A3.2 A 1 only appears on the diagonal and that it appears
on every diagonal element; had we been working in the group D 4 h then it
would have been the A 19 irreducible representation which would have
behaved in this way. This is important; it can be seen from Table A3.1 that
A 19 is the only symmetry species which never goes into its negative under a
symmetry operation. If we were interested in an integration over all space
and the mathematical function being integrated had A 19 symmetry then it
Table A3.2 Direct products for the 0 4 group
o. A, A2 a, a2 £
A, A, A2 a, a2 £
A2 A, A, a2 a, £
a, a1 a2 A1 A2 £
a2 a, a1 A, A1 £
£ £ £ £ £ (A 1 + A2 + a 1 + a2)
444 I Appendix 3
would not automatically equal zero. If it had any symmetry other than A,.
Table A3.3 The 0 character table
then the negative and positive contributions to the integral would exactly
0 E 6C4 3C2 6C2 8C3 cancel (this is easy to see for all except the Es in Table A3.1; for all the other
As and Bs there are as many -I entries as there are +I; the statement is
Al 1 1 1 1 1 also true for the Es). This conclusion is very relevant to quantum mechanics
A2 1 -1 1 -1 1 for there one is frequently interested in integrals over all space. So, if we are
E 2 0 2 0 -1
1 -1
interested in an integral such as
rl 3 -1 0
T2 3 -1 -1 1 0
f 1/1 1 1/12br
which is an overlap integral, we can discuss its symmetry properties (in
particular, whether it is zero or not) by considering the direct product of the
symmetry species of the two orbitals concerned. In this particular case it
leads to the important result that only orbitals of the same symmetry species
have non-zero overlap integrals. So, in those places in the text that direct
products have been used, it is because, really, we are interested in products
of orbitals; this is true even in places such as Section 8.7 where, ostensibly,
the subject is the intensities of spectral bands.
Finally, in Table A3.3, is given the character table for the octahedral group
0. It can be used to check that tables such as Table 6.3 are correct. The full
octahedral group, which has symmetry species with g and with u suffixes, is
denoted o. and is four times as large. The two groups are related in that
the character table for the group o. is built up from that of the group 0 in
the same way that the character table of the group D4 • is built up from that
for the group D4 .
The reader will find simple applications of group theory in Chapters 10
and 15. A non-mathematical introduction to the subject is given in Symmetry
and Structure by S. F. A. Kettle, Wiley, Chichester and New York, 1995. Also
relevant, useful and/or relatively non-mathematical are: Symmetry in Coordi-
nation Chemistry by J. P. Fackler Jr., Academic Press, New York, 1965;
Symmetry in Chemistry by H. H. Jaffe and M. Orchin, Wiley, New York, 1965;
Molecular Symmetry and Group Theory by A. Vincent, Wiley, London, 1977;
Symmetry in Chemical Bonding and Structure by W. E. Hatfield and W. E.
Parker, Merrill, Columbus, Ohio, 1974; and Group Theory for Chemists by
G. Davidson, Macmillan, London, 1991. More mathematical is Chemical
Applications of Group Theory by F. A. Cotton, Wiley, New York, 1990.
Appendix4
Equivalence of dz2 and dx2 _y2
in an octahedral ligand field
In Section 6.3 it was noted that the action of a C 3 rotation operation

converted dx'-y' into d,,_,, and d,, into dx, (or, for a C 3 rotation in
the opposite sense, into d,'-x' and d,,). In order to see how the new
orbitals are related to dx, _ ,, and d,,, we must first recognize that the d,,
is more correctly written as d./ (l /JH 2 ,,_x'-y'> (this gives it the central annulus
that d,, does not). Working with the labels alone (this is a perfectly valid
simplification) and after a lot of trial and error, knowing the answer in
advance or by use of some systematic method, we find that
y2
- z2 = - .1(x2
2 -y 2) - ...}3
2 x.J13 (2z2
- x2- y 2)
(hint: show that simplification of the right-hand side expression gives the
left-hand side) and
1 ...}3
- (x 2 - y 2 ) - -1 1
2 2 2 2 2
- (2x - y - Z ) = X - (2z -X - y 2)
...}3 2 2 ...}3
That is,
and
In other words the new orbitals are simply linear combinations of the old.
The readers should check that this is also true of d,' - x' and d,, (they
only have to change two of the signs in the right-hand side expressions
given above).
A final word, for the reader of Appendix 6, who has returned to this
appendix in search of a character of -I. As shown above, when dx, _,, is
rotated into d,, _,, its contribution to the latter is --!- Similarly, d,,
contributes --! to dx'· As the reader may well have discovered, the same
result is obtained if rotation into d,, -x' and d,, is considered. In both cases
the aggregate character is -I.
Appendix 5
Russell-Saunders coupling
scheme
Consider a free atom or molecule in which several electrons occupy but do

not fill a set of orbitals, which need not all be degenerate. There will usually
be many ways in which the electrons can be distributed, some of which will
be of lower energy than others. The differences in energy will be determined
by electron repulsion and the more stable arrangements will be those with
the least electron repulsion destabilization. So, electron- electron repulsion
will be relatively small if the electrons occupy orbitals which are spatially
well separated; it will be reduced yet more if the electrons have a high spin
multiplicity, i.e. have parallel spins, because two electrons with parallel spins
can never be in the same orbital.
If we are to discuss the energetics of free atoms or ions we have to separate
the possible electron arrangements into sets, the members of each set all
being degenerate. There are two schemes commonly adopted for this. The
first, which is the subject of the present appendix, is the Russell-Saunders
coupling scheme. The second, the j - j coupling scheme, is used when an
interaction which we shall neglect for the moment, spin- orbit coupling, is
large. It is of much less general applicability than the Russell-Saunders
scheme, although relevant to the content of Chapter II .
In the Russell- Saunders coupling scheme, the orbital motion of the
electrons are coupled together, as are their spins. Within the simplified
scheme discussed here-neglect of spin- orbit coupling- there is no co upling
between these two sets. For any particular a rrangement of the electrons
within the orbitals the coupling between their orbital motions gives rise to
a resultant; similarly, their spins couple to give a resultant, the resultants
simply being the vector sums of the individual orbital or spin motions. It
turns out that it is possible to measure not only the magnitude of the
resultants but also something about their orientatio n in space. F or each a tom
or ion there is a t any particular instant a n axis- let us call it the z
axis- along which the z components of the orbital a nd spin resulta nts have
well-defined values (unless something happens, absorption or emission of
radiation, for example, which changes them). Not any z component can exist,
however. If the largest component is M, then the others are (M, - I ),
(M, - 2), .. . , (M,- 2M, ). That is, the components run from M, to -M,
a total of (2M,+ 1) values in all. Each resultant will correspond to a
particular arrangement of electrons in the orbitals and, for resulta nts related
Appendix 5 1 447
in the way we have just described, the arrangements will also be related.
Note that if we wish to hold the z axis fixed in space we have to apply some
fixed axial perturbation, an electric field, for example.
So far, the nature of the vectors has not been specified. They represent
angular momenta (there is more on this topic in Chapter 9) and one talks
about the orbital and spin angular momenta, their magnitudes and their z
components. The quantization of orbital angular momentum, introduced
arbitrarily in the Bohr theory of atomic structure, appears in present-day
quantum mechanics in the way just described. The various z components
of orbital and spin angular momentum are, within our approximation,
energetically unimportant. The energy of a particular arrangement is deter-
mined solely by the absolute magnitude of the orbital and spin angular
momenta, and is independent of their z components. The reason that the z
components have been introduced is the following. For each of the (2L + 1)
z components of an orbital angular momentum vector and for each of the
(2S + 1) z components of a spin angular momentum vector there is an
individual orbital or spin wavefunction. Since a complete wavefunction
contains both orbital and spin parts and any of the orbital functions may
be combined with any of the spin functions, (2L + 1)(2S + 1) distinct
combinations exist. All of these wavefunctions are degenerate. It is the
removal of this degeneracy by a crystal field which is discussed in Chapter
7, by a magnetic field (amongst other things) in Chapter 9 and by spin-orbit
coupling in Chapter II. An individual energy level of a free atom or ion could
be characterized by specifying (2L + I) and (2S + I )-they are both integers
-or, alternatively and more simply by specifying L and S. L is always an
integer butS may not be, so a combination of the two is used. Land (2S + 1)
are given. Rather than quoting Las a number it is replaced by a letter: L = 0
is indicated by S; L = I by P; L = 2 by D; L = 3 by F; L = 4 by G;
L = 5 by H; L = 6 by I and so on. A parallel with orbitals should be evident;
for example, there are five d orbitals; similarly, the label D indicates a
(2 x 2 + 1) =five-fold degeneracy. The value of (2S + 1) is given as a
number, written as a pre-superscript to the L symbol. Thus, 1 S, 3 P, 1 D, 4 F,
etc. When one is talking one says singlet, doublet, triplet, quartet, quintet.
So, 2 D is 'doublet dee'. One talks of a 2 D term. When there are, say, two
electrons distributed within a set of three (degenerate) p orbitals, this is
spoken of as a p 2 (pee two) configuration, the fact that all three p orbitals
are involved being implicit. Such a configuration will usually give rise to
several terms. The p 2 configuration gives rise to 1 D + 3 P + 1 S terms.
There is a simple check that can be applied to a list of terms such as this,
which it is claimed, arise from a given configuration. Consider the p 2
configuration. The first of the two electrons can be fed into the set of three
orbitals in any one of six ways (allowing for spin) and the second in any
one of five (it can only go into the same orbital as the first electron with the
opposite spin). Recognizing that we have counted each arrangement twice
because j t is the same as t j, the number of distinct arrangements, and
therefore wavefunctions, is (6 x 5)/2 = 15. Now, the 1 D, 3 P and 1 S terms
correspond to 5, 9 and I wavefunctions respectively, again a total of 15. The
number of wavefunctions is the same whichever way it is calculated, as it
should be. In Table AS.! the process is worked out in some detail for the d 2
configuration, because this is of importance in Chapter 7. The z component
448 I Appendix 5
Table A5.1 The terms arising from the d2 configuration 1 G + 3 F + 1 0 + 3 p + 1s

Function 1G 3F 10 3p 15 Function 1G JF 1o 3p 15
Ti n ll ii n H Ti n H ii H H
22
2 1 io
2i i6
20 i -1
26 i -i
2 -1 i -2
2 -i i -2
2 -2
2 -2
06
21 0 -1
2i 0 -i
2o 0 -2
26 0 -2
2 -1 6 -1
2 -i 6 -i
2 -2 6 -2
2 -2 6 -:2
1i
10 -1 -i
16 -1-2
1-1 -1 -:2
1 -i -i -2
1 -2 -i -:2
1 -2 -2 -:2
It is to be emphasized that this table is a book-keeping exercise only; it cannot be assumed that the actual functions are as indicated in it (the actual 1S function,
for instance, is a linear combination of all functions listed with resultant spin and orbital angular momenta of 0).
of the orbital angular momentum of each electron in each distinguishable

arrangement is indicated by a number. Numbers without a bar indicate that
the electron has rx spin; numbers with a bar, such as 2, indicate a f3 electron
spin. In the d 2 case, we have (10 x 9)/2 = 45 distinguishable arrangements
and so the left-hand column contains 45 entries In the body of the table
each of these is allocated to a component of one of the terms arising from
the d 2 configuration, 1 G + 3 F + 1D + 3 P + 1S. A further discussion of this
general topic is contained in Section 11.3. The job of working out the actual
wavefunctions of the terms is more difficult than the job of determining just
what these terms are-and it is the latter that we have done. The job of
working out the relative energies of the terms is rather hard work.
A final word about spin-orbit coupling. Its effect is to cause small
splittings between functions which we have so far regarded as degenerate.
These splittings may be observed spectroscopically and values of spin-orbit
coupling constants thus obtained (Section 9.3).
Further reading An excellent treatment which goes well beyond that
given above but which is both easy and, often, fun to read is Orbitals, Terms
and States by M. Gerlach, Wiley, Chichester and New York, 1986.
Ligand G group orbitals of an
octahedral complex
In this appendix are presented three different methods of obtaining the linear
combinations of Table 6.2, repeated as Table A6.1, the ligand u orbital
symmetry-adapted linear combinations for an octahedral complex. There is
much to be learnt from a careful comparison of all three methods.
Method 1
The first method is scarcely a derivation. Knowing the symmetries spanned
by the ligand u combinations (A 19 + E9 + T 1u), one finds an object of each
of the symmetries in turn and asks what ligand combination matches the
nodal pattern of the object? The most obvious reference objects to choose are
the appropriate orbitals of the central metal atom. Labelling the ligand u
orbitals as in Fig. 6.7, repeated as Fig. A6.1, and referring back to Figs. 6.8
(s orbital), 6.9 (p orbitals) and 6.10(a) (d orbitals) it is evident that the
matching combinations are of the form
s: (ul + "2 + U3 + "• + "s + "s)

Px: (ul- "s)
Py: (u2- "•)
p,: (u 3 - us)
dx'-y': + 113 - "• + "s)

( -u2
d,, : (2ul- "2- "3- "•- "s + 2us)
remembering that the full label of d,, is d 2 z, _•' _,,. These apart from
normalization, and in one case a -I factor, are the expressions found in
Table A6.1.
Method 2
When we have two equivalent atomic orbitals, 1/1 1 and 1/1 2 say, the former
on atom I and the latter on a symmetry-equivalent atom, 2, symmetry
demands that the electron density at a given point in 1/1 1 is equal to the
electron density at the corresponding point in 1/1 2 . That is, 1/!i = 1/1~ at these
450 I Appendix 6
Table A8.1 Ligand group orbitals of six octahedrally

orientated a ligands

1
a:~g ~ 6 (u 1 +u 2 +u 3 +u 4 +u 5 +u 6 )
1
~ 2 (u1- Us)
1
~2 (<12- <14)
1
~ 2 (a3- <1s)
Fig. A6.1 Numbering system adopted for

ligand u orbitals in an octahedral complex.
points. It follows that t/1 1 = ± t/1 2 . In other words, t/1 1 and t/1 2 may be either
in-phase or out-of-phase with each other. If t/1 1 and t/1 2 combine to form a
molecular orbital, r/J say, then an analogous argument shows that r/J may
either have the form (t/1 1 + t/1 2 ) or (t/1 1 - t/1 2 ), all other combinations leading
to unequal electron densities on atoms 1 and 2. Normalizing, that is,
multiplying by a factor so that the sum of the squares of the coefficients
multiplying t/1 1 and t/1 2 equals unity, but neglecting overlap between t/1 1 and
t/1 2 , gives the orbitals
and
Appendix 6 1 451
B d
1 2 which, if atoms 1 and 2 are part of a larger, polyatomic, system are called
group orbitals.
In what follows it will, for simplicity, be assumed
1. That all orbitals labelled 1/1 are of a type with respect to a suitably placed
4 3 metal ion, although the discussion has to be little modified to cover 1t
and c5 type interactions.
FIJI. A8.2 Numbering system adopted for
ligand u orbitals of a square planar complex. 2. That all such orbitals have the same (positive) phase. That is, that
is a combination of two orbitals of identical phase; in
1/1 2 has the opposite phase to 1/Jt because of the negative sign, not because
it is inherently of opposite phase.
What if we have four identical a orbitals 1/Jt, 1/1 2 , 1/1 3 and 1/1 4 on atoms
arranged at the corners of a square? The a orbitals on the ligands of a square
planar complex are an example of this situation. The simplest way of
discussing this case is as follows. Label the orbitals cyclically as in Fig. A6.2
and consider the pairs 1/Jt and 1/1 3 , and 1/1 2 and 1/1 4 . We can treat each trans
pair as if the other were not present and use the discussion above to obtain
group orbitals for each pair. These are
1
t/>1 = ../2 (1/11 + .Pal
1
t/>2 = ../2 (1/11- .Pal
1
4>a = ../2 (1/12 + 1/14l
1
4>4 = ..}2 (1/12- 1/14)
All that we have to do now is to form suitable combinations of 4>t• </> 2 , </> 3
and </> 4 to obtain the ligand group orbitals that we are seeking. But how do
we know which combinations to take? The answer is that the transformations
of the final group orbitals must be described by the irreducible representations
of the D4 h point group. If the language of the last sentence seems obscure,
reread Appendix A3. This requirement boils down to the fact that any nodal
planes contained in <Pt. </> 2 , </> 3 and </> 4 , must be compatible with each other
for any interaction to occur. For example, neither 4>t nor </> 3 contain any
nodal planes; because they are equivalent to each other we simply, again,
take their sum and difference
1
ll1 = ../ 2 (t/>1 + 4>al = !(1/11 + 1/12 + .Pa + 1/14)
1
82 =- (t/>1- 4>al = !(1/11- 1/12 + .Pa- 1/14)
..}2
452 1 Appendix 6
However, ¢ 2 contains a nodal plane which passes through 1/1 2 and 1/1 4 and
so ¢ 2 cannot combine with the group orbitals involving these atoms, ¢ 3 and
¢ 4 . Similarly, ¢ 4 has a nodal plane which means that it cannot combine with
either ¢ 1 or ¢ 2 . We therefore conclude that the ligand group orbitals of a
square planar complex are
1
111 = 2 (1/tl + l/t2 + l/t3 + l/t4)
1
e2 = 2 (1/tl- l/t2 + l/t3- l/t4)
1
Eu(1): </>2 = .J 2 (1/tl- l/t3)
The reader is urged to check that these orbitals have the symmetries indicated
by using the D4 , character table given in Appendix 3.
Finally, the problem of the ligand group orbitals appropriate to an
octahedral complex ion. It is convenient to regard the octahedron as a
synthesis of a square planar complex, with the combinations given above,
and two additional ligands with orbitals 1/1 5 and 1/1 6 , which give rise to group
orbitals
1
<l>s = .J 2 (1/ts + 1/ts)
1
<l>s =-(1/ts- 1/ts)
.J2
Applying the nodal plane compatibility requirement it is found that only 11 1
and ¢ 5 , neither of which has a nodal plane, combine together. What are the
correct combinations? It is a general rule that, for the type of ligand group
orbital which we have considered, there is always a combination
1
- (ljt1 + l/t2 + · · · + 1/tn)
.Jn
which has the symmetry given by the first row of the appropriate character
table (the totally symmetric combination, a row of Is). In the present case
this means that one combination is
It is not difficult to show that this combination is simply
JG) x 81 + JG) x </> 5
There must be an orthogonal combination of 11 1 and ¢ 5 and this is

Appendix 6 1 453
(a combination of wavefunctions (c 1 p 1 + c2 p 2 ) is always orthogonal to

(c 2 p 1 - c 1 p 2 ) provided that p 1 and p 2 are, separately, orthogonal and
normalized). Written in detail, this orthogonal combination is
In summary, the ligand <I group orbitals of an octahedral complex ion are
1
.jB (1{11 + o/2 + 1{13 +f.+ fs + 1{16)
1
T1u(1): - (1{11- 1{13)
.j2
1
r1u(2): .j2 (1{12- "'·)
1
r1u(3): - (fs- 1{16)
.j2
~(1{11- "'2 + "'3- "'·)

1
- (2fs- o/1- o/2- 1{13- f.+ 21{16)
.j12
Again, the transformations of these combinations should be checked using

the 0 character table in Appendix 3 (the g and u suffixes will have to be
dropped from the expressions above). The only point of any difficulty in this
check is the transformation of the two E. orbitals under the C 3 rotation
operations, where the character is -I. In fact, this is the same problem as
that discussed in Appendix 4, to which reference should be made.
There is one final problem. Because of the square planar --> octahedral
sequence that was used above, the ligands are not labelled in the same way
as in Chapter 3 (see Fig. 3.4) and in Chapter 6 (see Table 6.1 and Fig. 6.7).
These latter were repeated as Table A6.1 and Fig. A6.1. The correlation
between the .j!s used above and the <IS of Table A6.1 is
(just the labels 1 and 5 have to be interchanged between the two sets).
Method 3
The final method is group-theoretical, a full derivation has been given
elsewhere. 1 That which follows is the shortest known to the author; we use
Fig. A6.1. From the full octahedral group choose a set of operations
consisting of complete classes, one of which is the identity, such that the
total number of operations in the full group (48) is an integer times the
number in the selected set. Further, choose a set such that the number of
operations in the set is an integer times the number of objects to be
considered (here, we have <I 1 --> <I 6 so this latter number is 6). This selection
1 SeeS. F. A. Kettle, J. Chern. Educ. (1966) 43,21 or S. F. A. Kettle, Symmetry and Structure,
Wiley, Chichester, 1995.

454 I Appendix 6
exercise is designed to make the final derivation involve the minimum of

work. In the present case, a suitable set of operations is E, 8C 3 , 3C2 , a total
of 12. This set also falls within the group 0 so, again for simplicity, we will
work in this smaller group. Now apply the formal projection operator
method by considering u 1 and applying the twelve operations to it in tum
(which turns it into a u which more often than not is one of the other us)
and then summing the results. This gives
E: (u 1 )
8C3 : (2u2 + 2u3 + 2u4 + 2u5 )

3C2 : (u1 + 2u6 )
Multiply the quantities in brackets by the appropriate characters from the
0 character table (Appendix 3) and add:
1. for A 1 the characters are E: 1, 8C3 : 1, 3C2: 1, so we get
u1 + 2u2 + 2u3 + 2u4 + 2u5 + u1 + 2u6
and the sum is
2(u1 + "2 +"a+ <14 + "s + "sl
2. for E the characters are E: 2, 8C3 : -1, 3C2 : 2, so we get
2u1 - 2u2 - 2ua- 2u4 - 2u5 + 2u1 + 4u6
and the sum is
2(2u1 - u2 - "a- u4 - u5 + 2u6 )
3. for T 1 the characters are E: 3, 8C3 : 0, 3C2 : -1, so we get
3u1 - u1 - 2u6
and the sum is
2(u1 - u6 )
Apart from normalizing factors, these are all entries in Table A6.1; the
additional T 1 functions can be obtained by starting with u 2 and u 3 in place
of u 1 . Either of these will also serve to generate the second E function but
give ligand group orbitals corresponding to either dx' or d,,. The function
corresponding to dx'-y' is obtained either by taking the difference between
dx' and d,, or by the equivalent, but more formal, procedure described in
Appendix 4. In Oh the A 1 function above becomes A,., the E becomes Eu
and the T 1 becomes T 1u.
Appendix 7
Tanabe-Sugano diagrams and some
illustrative spectra
80,----------------------,
80.----------------------,
50
50
t./8 AjB
10 000 20 000 30 000 10 000 20 000 30 000

cm-1 cm-1
RC. A7.1 d2 ; y ~ 4.42 and the spectrum of [V(urea) 6 ] 3 • . Rg. A7.2 d3 ; ;· ~ 4.50 and the spectrum of [ Cr(H2 0)6 ] 3 •.
456 1 Appendix 7
5T2g 1
T1g
1A2g
i3A2g
"'i:U- 40
:~~
11 3Eg
1A1g
80r-------------,---------- .
1Eg
1T2g
5Eg
3T1g
•Eg
50 •r2g
!VB 4 A1g, 4 Eg
2A1g
10
10 000 20 000 30 000

cm- 1
Fig. A7.3 d4 ; y ~ 4.61 and the spectrum of [Mn(H 2 0)s] 3 +.
w 0.02
10 000 20 000
cm-1
Rg. A7.4 d5 ; y ~ 4.48 and the spectrum of [Mn(H 2 0) 6 j>+ (the ground
state for all the transitions in the spectrum is 6 A.tg).
Appendix 7 1457
80
'Eg
lEg
3Eg
3A2g
1A2g
1T2g
"'i:iJ 40
30,11 1T1g
3T2g 80
3TlJI
12A2g
4 T2g
1AlJI
50
!:./8
"'i:iJ 2r2g
2r1g
1TlJI<-1AlJI
w 50
0 •rlJI
10 000 20 000 30 000
cm-1
Fig. A7.5 d6 ; y = 4.81 and the spectrum of [lrC1 6 ] 3 -. 2Eg
50
t:.jB
10 000 20 000
Fig. A7 .6 d 7 ; y = 4.63 and the spectrum of [Co(H 2 0) 6 ] 2 +.

458 I Appendix 7
80,-----------------------,
50
li/B
10 000 20000 30000

cm-1
FJa:. A7.7 d8 ; y = 4.71 and the spectrum of [Ni(H 2 0).,] 2 •.
Append·xs
Group theoretical aspects of
band intensities in octahedral
complexes
The most general selection rule for electric-dipole transitions, the type of
transition that is relevant to almost all of the bands observed in the electronic
spectra of transition metal complexes, is a group theoretical statement of the
excitation process described in qualitative terms in Section 8.1. It is the
following:
An electronic transition is electric-dipole allowed if the direct product of the
symmetries of the ground and excited electronic terms of a molecule contains
the symmetry species of one or more coordinate axes, both ground and excited
terms having the same spin multiplicity.
This selection rule sounds somewhat formidable but in practice it is not
difficult either to understand or apply. The coordinate axes come in because
a +--dipole has the same symmetry as a coordinate axis. Direct products
arise because we need to consider, simultaneously, the ground and excited
terms together with the mechanism of excitation (electric dipole). A triple
direct product is therefore needed. The real selection rule is that this triple
direct product must contain the totally symmetric irreducible representation-
that for which all of the characters are +I in the character table. For the
case of octahedral complexes this is A 19 • However, the totally symmetric
irreducible representation only arises when (any) irreducible representation
is multiplied by itself. This means that we can split the triple direct product
into two parts and compare them to see if the two parts contain the same
irreducible representation. The most convenient division is to form the direct
product of the ground and excited terms and to compare the result with the
symmetry of the coordinate axes. Let us consider a specific example, the
3T
29 +- T 19 (F) transition in the d
3 2 case. Here the ground term orbital
symmetry is T 19 and the excited term is T 29 • The direct product T 19 x T 29 is

A 29 + E9 + T 19 + T 29 (the relevant direct products are given in Table A8.1;
g suffixes have to be added). Now, the Cartesian coordinate axes x, y, and
z have the same symmetry properties as the corresponding p orbitals, which
in our case is T 1•• As T 1• is not contained in the T 19 x T 29 direct product
(this is A 29 + E9 + T 19 + T 29 ) the 3T 29 +- 3T 19(F) transition is orbitally
forbidden. We could have seen this result without all of this work. The triple
460 1 Appendix 8
Table A8.1 The Oh direct product table
oh A1 Az E Tl Tz
Al Al A2 E Tl T2
Az A2 Al E T2 Tl
E E E A1 +A2 +E Tl + T2 Tl + T2
Tl Tl T2 Tl + T2 A1 + E + T1 + T2 A2 +E+T1 +T2
T2 T2 Tl Tl + T2 A;,+E+T1+T2 A;,+E+T1+T2
This table is actually the product table for the 0 point group; for the point group Oh the usual suffiX rule applies:
gX g= U XU= g; g XU= U.
direct product must contain A 19 for the transition to be allowed. We focus

on the g in this symbol. Both ground and excited terms are g and the
coordinate axes u-but any triple direct product containing g x g x u can
never give g and so can never give A 19 and can never give an allowed
transition. As shown in Section 8.7, a vibration of T 1 • symmetry can remove
this forbiddenness by mixing some p orbital component into the d.
To investigate this in more detail we have to answer the question 'what
orbital can a vibration of T 1 • symmetry mix with ad orbital?'. At a deeper
level we are concerned with the question 'what excited terms can a vibration
to T 1 • symmetry mix with a ground term of T 19 symmetry?' again confining
our discussion to the d 2 case. The answer to the first question will help with
the second. A vibration will mix one orbital with a second if the direct
product of the symmetry species of the vibration and the first orbital contains
the symmetry species of the second orbital. Because the orbitals with which
we are primarily concerned are d they are g and so aT,. vibration can only
mix a u orbital with them. The only ones available are p orbitals, with T,.
symmetry. As Table A8.1 shows, the direct product (vibration x orbital) is
T1u x T1u = A:tg + Eg + T:tg + T2g (adding suffixes)
so we conclude that a T 1 • vibration will mix some p orbital into a d, be the

latter e9 or t 29 (since both of these feature in the direct product). That is, the
vibration can introduce into a configuration such as ~~. a tiny component
of a configuration such as t} 9 t: •. Into an excited state configuration such as
t~ 9 e; we can also introduce a small ~~.~:. component and also an tl. e;
component. Clearly, we could now extend our discussion to answer the same
question about excited states by working out the terms arising from these
configurations, much as is done for strong field complexes in Chapter 7, but
we shall use a simpler approach. Again consider the 3T29 +-- 3T 19 (F) transition
in the d 2 case as an example. We can include the effect of T 1 • vibrations by
simply including T 1• in the direct product. That is, we now take the triple
direct product
T:tg X T2g X T1u = (A2g + Eg + T:tg + T2g) X T1u
= (A2g X T1u) + (Eg X T1u) + (T:tg X T1u) + (T2g X T1u)
= A1u + A2u + 2Eu + 3T1u + 4T2u

Appendix 8 I 461
This final direct product contains T 1 • three times so, as T 1 • is the symmetry
species of the coordinate axes to which the selection rule directs our attention,
the intervention of the T 1 • vibration not only makes the 3T29 <-- 3T 19 (F)
transition allowed but also splits it into three separate transitions. These will
differ slightly in energy and so contribute to the breadth of the observed
peaks.
The picture is further complicated when it is recognized that there are
two distinct sets of T 1 • metal-ligand vibrations in an octahedral complex
and, further, there is a set of T 2 • vibrations which may also give intensity to
the d-d transitions by way of four subpeaks. We therefore conclude that we
must expect the 3T 29 <-- 3T 19(F) transition to be composed of no less than
ten (3 + 3 + 4) subpeaks, which may be separated by energies of the order
of magnitude of the energy of metal-ligand vibrations (ca. 200 em- 1 or
more). Each of these peaks is broadened by the vibrational modulation of
~ discussed in Section 8.7 and also, probably, by spin-orbit coupling and
by the Jahn-Teller effect. The large half widths observed in d-d spectra are
understandable!
Append·xg
Determination of magnetic
susceptibilities
Recent years have seen a considerable increase in the number of techniques

available for the determination of the magnetic susceptibility of a complex.
Traditional methods involve a measurement of the change in weight of a
sample when it is placed in a strong magnetic field, a field which can contain
a uniform region (the Gouy method, to be described later) or have a constant,
and large, field gradient (the Faraday method). A method based on the
regular reversal of such a gradient by the use of electromagnets driven by
an alternating current is the basis of the alternating force magnetometer.
Induction methods are based on the measurement of the change in the
magnetic field itself, which may be either static or alternating, when a sample
is inserted. The vibrating sample magnetometer also measures a field change,
this time when a sample is vibrated- for instance, by attaching it to the cone
of a loudspeaker driven by an ac signal. More recently, very sensitive- and
very expensive- measurements have become available by use of a super-
conductive quantum interference device (SQUID). This depends on effects
related to the magnetic fields generated within a superconducting coil and
their variation when a sample is placed within them. Finally, for solutions,
the effects on a reference NMR signal produced by the presence of dissolved
paramagnetic species may be used to measure paramagnetism. The reference
at the end of this appendix gives more details of most of these methods.
The method for determining magnetic susceptibilities most likely to be
met by the reader is the Gouy method. The description which follows is for
the conventional arrangement, although an inverted arrangement- in which
the sample is fixed and the change in weight of the magnet measured-is
becoming increasingly popular. In the Gouy method, a cylindrical sample
of the complex under study is suspended between the poles of a magnet. The
bottom of the cylinder is halfway between the two poles and on the interpole
axis (see Fig. A9.1). It is thus at the position of maximum field intensity.
Ideally, this maximum field is constant within a small volume, so that small
changes in the position of the bottom of the cylinder can be ignored. The
top of the cylinder is, essentially, out of the magnetic field. In this way the
cylinder experiences the full field gradient, the observable quantity being an
apparent change in weight of the sample when it is placed in the magnetic
field, either by switching on an electromagnet or bringing up a permanent
magnet. Paramagnetic materials are usually attracted into a magnetic field-
Appendix 9 1 463
Suspension -
Filling mark. scribed on
the sample tube
FIC. A9.1 Apparatus to detennine magnetic

Flat-bottomed
susceptibility by the Gouy method.
sample tube ............._
Draught-proof enclosure,
fitted with a themnometer
pocket (not shown)
they show an apparent increase in weight- the only exception being when
the inherent diamagnetism of the sample swamps its paramagnetism.
Experimentally, the sample is either a solution or a finely ground solid
within a silica or Perspex (Lucile) tube (ordinary glass contains paramagnetic
impurities). The sample length is usually from 5 to 10 em and its diameter
2 to 10 mm. It is suspended from the beam of a balance in a suitable
draught-proof enclosure. More complicated arrangements are used to study
magnetic behaviour over a temperature range. The sample tube is filled to
the same height for all measurements and, for solids, every effort must be
made to ensure that it is uniformly filled (it has been shown that loose
packing can lead to the crystals twisting around in the magnetic field so that
the assumption of random orientation is not valid; this can have a significant
effect on the results). A compound with several unpaired electrons per
molecule would be expected to show a greater paramagnetic effect than a
compound with only one, assuming that the same number of molecules of
each is studied. That is, the quantity
weight increase in the field ~w
weight taken + molecular weight w~M
would be expected to be larger for the former. The quantity on the right-hand
side of this equation is proportional to the molar magnetic susceptibility of
the compound, XM:
y~w x M
XM = - - -
w
The constant of proportionality varies from apparatus to apparatus- the
more powerful the magnet used, the greater .1w. Hence, y is called the tube
calibration constant. One small addition has to be made to the equation for
XM-the sample tube is diamagnetic and will lose weight in the field. The
actual magnitude of this correction is determined by a trial run with the tube
empty, when its weight drops by J. The value of J must be added to the
experimental values of .1w so as to make .1w more positive (or less negative).
464 1 Appendix 9
The corrected expression for XM is,

Table A9.1 Pascal's constants;
diamagnetic corrections, all
y(~W+ ii)M
values x 10- 6;gatom XM=----
w
u+ 0.6 F- 11 H 3
Na+ 5 cl- 26 c 6 It is convenient to measure Ll.w and i5 in milligrams and w in grams. This,
K+ 13 eN- 18 N 2 and multiplying y by 10 3 -so that it becomes of the order of unity-leads
NH,t 11.5 CNS- 35 0 5 to a modified relationship which is the one usually quoted
Fe 2 + 13 so~- 40 s 15
Co2 + 12 co~- 29.5 p 26 y(~w+ ii)M
6
Ni 2 + 12 ow 12.0 10 X M = - - - -
W
Note that this equation refers to the whole molecule. To find out the number
of unpaired electrons on the transition metal atom in a complex an allowance
must be made for all of the paired electrons present in the molecule. These
effects are approximately additive and are listed for some of the more
common species found in complexes in Table A9.1. A more complete list is
given in a reference at the end of this appendix. To make the correction
one sums the appropriate values. So, for [Ni(H 2 0) 6 ]2+Soi- · H 2 0, the
corrections are: Ni2+ 12; SOi- 40; 7H 2 0 7(2 x 3 + 5) = 77-a total of 129
( x 10- 6 /g atom). The correction calculated in this way, D, say, is added to
XM and the sum is x;.., the susceptibility corrected for diamagnetism. That is,
XM = ZM +D
The diamagnetic correction can be more elaborate than indicated above, but
in practice-and particularly when a complicated ligand is involved-it is
best to measure the diamagnetic correction, either directly on the ligand
(using Pascals constants for the remainder) or by measuring the susceptibility
of an analogous but diamagnetic complex. The reader is reminded that in
some cases (Con provides the most important example) it is better not to
make a diamagnetic correction unless a similar allowance is being made for
temperature independent paramagnetism (TIP), see the end of Section 9.3.
With a value for x;.., Jl,rr is obtained using a relationship derived in
Chapter 9
Jlert = .)(3RjN't,{3 2 J...IzM X T = 2.83-JXM X T
Jl,rr can then be treated using whichever approach is deemed appropriate.

For example, if the spin only model is used, we have that
Jlert = Jn(n + 2)
Bringing all this together, gives
[y(~W + ii) X M X 10- 6 + D]

n(n + 2) = 8.Q7T.::.:...__ _ _ _ _ _ _ __c
w
Since n = 0, 1, 2, 3, ... , the left-hand side of this equation equals 0, 3, 8,

15, .... All that remains is to determine y; this can be done from measurement
of the applied magnetic field and the length of the sample experiencing the
field gradient, but simplest is to calibrate the apparatus using a standard
material.
Appendix 9 1 465
References
A good review of reasonably recent developments is C. J. O'Connor, Prog.
Inorg. Chern. (1982) 29, 203. There have been several useful articles dealing
with the measurement of magnetic susceptibilities in J. Chern. Educ. (1972)
49, pages 69,117 and 505; (1983) 60, pages 600 and 681.
AppendixlO
Magnetic susceptibility of a
tetragonally distorted t~g ion
The object of this appendix is to expand the discussion in Section 9.9 on the
magnetic susceptibility of a t1 9 ion such as Ti 111 • It has been written in the
expectation that the reader will be more interested in the general arguments
than in the detail. As far as is possible at a simple level, the individual
mathematical steps will be justified, although the reader should allow neither
the mathematics nor the sometimes off-putting equations to inhibit a general
scan of the appendix. The treatment is algebraic, although today a computer
would be used to treat the problem and would do so more accurately than
does the presentation in this appendix. However, the technical language used
in discussions of magnetism is based on the algebraic approach and so an
understanding of this approach is essential.
The example chosen is almost the simplest possible-a single d electron
confined to the t 29 set of orbitals in a tetragonally distorted octahedral
complex. The effect of the octahedral ligand field is to give three degenerate
orbitals, each of which may be occupied by an electron with spin up or spin
down. There are six wavefunctions, all degenerate. A tetragonal distortion
is now applied and, as Table 6.5 indicates, this separates the d orbitals into
the degenerate pair (dzxo d,,) and dx,, taking the axis of distortion as the z
axis. If the tetragonal splitting is t then retention of the centre of gravity
requires that the energy of the degenerate pair be -t/ 3 and that of dx, be
2t/ 3, assuming that the distortion is such that the latter orbital is destabilized.
The octahedral ligand field stabilization is the same for all three, 2/ 511, and
is omitted from this point on. Including spin, we have four wavefunctions
of energy -t/3 and two of energy 2t/3. Each successive step at each point
in the development of the energy-level argument involves including the next
largest interaction and at this point this is the spin- orbit coupling. The form
of the result of including it follows from the rule given in Section 9.5.
Spin- orbit coupling causes the degenerate orbitals d,x and d,, to interact
and thus to lose their degeneracy. Further, each of these orbitals interacts
with dx, · Detailed calculations give the energies below-check that when (
(zeta, the spin-orbit coupling constant) = 0 they give the energies above.
Once spin-orbit coupling has been included it is no longer possible to refer
to orbitals or to use labels such as d,x·
Appendix 10 1 467
E3 = ~[~I; + ~~ + (t2 - tl; + ~e) 1 i 2 J (doubly degenerate)
(doubly degenerate)
Each of the above levels is doubly degenerate, and each pair is a Kramers
doublet, with a degeneracy that can only be removed by a magnetic field.
The effect of such a magnetic field on this set of energy levels is conveniently
divided into two parts. The first part is the effect on each individual
degenerate pair; the second part is the magnetic-field-induced interactions
between pairs, interactions which mix them. They are called the first- and
second-order Zeeman effects, respectively. The first-order Zeeman effect leads
to the following energy levels, where, for simplicity, it has been assumed that
the field is applied along the z axis. The quantity k is the orbital reduction
factor.
E3 (b) = ~[~I; + ~~ + (t2 - tl; + ~eJ112] + 1 - w2(1 2+ k) X fJH

1+w
E (a) = 1[1( + 1t + (t2- tl," + 9[2)1/2] - 1- w2(1 + k) x fJH

3 ::1::1. 3 4. 1 +w2
E2 (b) = -~t- ~~ + {1- k)fJH

E2 (a) = -~t - ~~ - (1 k) fJH
In these expressions, w is a numerical coefficient which is a complicated

function of both I; and t. H is the field strength-that is, the first-order
Zeeman splitting is proportional to the field strength-and f3 is the Bohr
magneton. The first-order Zeeman effect has removed all degeneracies.
Inclusion of the second-order Zeeman effect gives
E3 (b) = ~[~I; + ~~ + (t2 - tl," + ~~2)112] + 1 - w2(1 2+ k) x fJH

1+w
+ 1 x [w(k + 2)]2 X f32H2

(t2 _ tl," + ~1,"2)1;2 1 + w2
E3 (a) = ~[~I; + ~t + (t2 - tl," + ~~2)112] - 1 - w2(1 + k) x fJH

1+ w2
1 [w(k + 2)] 2 2 2
+ (t2 - tl; + ~1,"2)1/2 x 1 + w2 x f3 H
E 2 (b) = -~t- ~~ + (1- k) fJH

E2 (a) = -~t- ~~- (1- k) f3H
468 I Appendix 10
w2 (k+ 1)
£1 (b) = ~[~C + ~~- (t2 - tC + ~C 2 ) 112] + - X PH
1+ w2
1 [w(k + 2)]2 2H2

- (t2 - tC + ~e)1/2 X 1 + w2 Xp
w2 (k+1)
£1 (a) = ~[~C + ~~ - (t2 - tC + ~C 2 ) 112 ] - - x PH
1 +w2
1 [w(k + 2)]2 2H2
- (t2 - tC + ~C2)1/2 x 1 + w2 x p
These equations are quite horrific in appearance and we have considered a
particularly simple case! The important thing to remember is that the
second-order Zeeman effect is characterized by terms which depend on the
square of the field strength. The splitting of the six t 2 • functions we have just
discussed are shown schematically in Fig. AIO.l, which has previously
appeared as Fig. 9.2.
The final step is to consider the thermal population of these energy levels.
Clearly, given the fearsome equations we have obtained, it makes sense to
simplify the problem-we have seen what the real-life algebraic problem
looks like, and that is enough. First, we revert to an octahedral ligand field,
i===-'"/ ==:=<<=======
< E3(b)
E3 (a)
Fig. A10.1 Schematic energy level diagram

illustrating how the various interactions
discussed in this appendix determine the
ground state derived from a t~ configuration in
a magnetic field. Crystal
field
Tetragonal
splitting <c:...----....==-........,---
"1S:p:in:-o:rib:i:.t E1 (b)
Fi E1 (a)
coupling 1rst-order Second-order
Zeeman
Zeeman
effect
effect
Magnetic field
Appendix 10 1 469
by setting t = 0; this simplification has the bonus that w becomes equal to

2. Secondly, we revert to a pure crystal field model by setting the orbital
reduction factor equal to I. With these simplifications, the energy levels
become
E2 (b) = -~~
E2 (a) = -~~
4{J2H2
E1 (b) = -~~ - - -
31;
4{J2H2
E1 (a) = -~[ - - -
. 31;
These expressions were first derived by Kotani (by a less circuitous route!)
and the simple theory which neglects distortion and the orbital reduction
factor is called Kotani theory. Note that levels which are split apart in the
more detailed analysis are degenerate in the Kotani treatment.
We now have to derive an expression for the temperature variation of the
magnetic susceptibility in terms of the thermal population of these levels. To
do this, the energy levels are conveniently rewritten in the form
En = En(O) + En(1)H + En(2)H 2 + · · ·
Table AlO.l gives the above set of energy levels in this format. We will return
to this table when we have the mathematical expressions that require the
data in this form.
An important property of the magnetic moment of a molecule, Jln, is that
it is proportional to the decrease in energy of the molecule with increase in
the applied magnetic field. In atomic units the proportionality constant is
unity:
Combining the last two mathematical expressions gives

Jln = -En{1) - 2HEn(2) ...
Table A10.1 Set of energy levels
corresponding to Kotani theory an expression that will be used shortly.
In Chapter 9 we saw that
En En(O) En(1) En(2)
N(average magnetic moment per molecule)
ZM = H
f3(b) ( fJ 4{3 2 /3(
E3 (a) ( -{3 4{3 2/3(
f2(b) -(/2 0 0 The average magnetic moment per molecule will be given by an expression
E2 (a) -(/2 0 0 of the form
f1 (b) -(/2 0 -4{3 /3(
2
E 1 (a) -(/2 0 -4{3 2 /3( I;(magnetic moment of a molecule) x (number with this moment)
4 70 1 Appendix 10
where the sum is over all molecules. Because the population of the set of
energy levels will be governed by a Boltzmann distribution law, it follows that
We now expand the exponentials as a power series in H, a quantity on which

we know that the energy depends:
-En(O)) ( -En(1)H)
=exp ( ~ xexp ~ ···
where further terms, omitted, will-except at very low temperatures or high

fields-be approximately
exp( :~) = exp(O) = 1

and so may be neglected. Using the expansion
and putting this, together with the expression for Jln given earlier, into the
equation for XM gives, approximately,
We now expand the numerator but first note that when the magnetic field
is zero the average magnetic moment per molecule must be zero and so the
numerator must be zero also. It follows that
L -En(1) -E (0)) = 0
x exp ( _n_
n kT
We have, then,
N L [(En(1)) 2H- 2HE (2) + 2H2 En(1)] X exp( -En(O))

kT kT
(-E (0))
n kT n
XM =
HL:exp _n_
n kT
Cancelling the Hs and neglecting the term 2HE.(l)jkT, which will be very
Appendix 10 1 4 71
small except at low temperatures or high fields we have, finally,
NL [E"(i)) 2- 2E (2)] x exp( - En(O))

kT kT
(-E
n n
1/JM =
L:exp -"- (0))
n kT
Using the relationship derived in Section 9.2:
2 3RTxM 3kTXM
/lett= N2f32 = N/32
we have, for Jl;H:
3kTL: [(En(i)) 2- 2En(2)] exp(-E"(O))

x
n kT kT
f32L:exp (-E(O))
2
!lett=
_n_
n kT
These expressions for XM and Jl;rr are known as the van Vleck relationships.
It is a simple matter to apply these equations to our problem; we have
only to sum over the six different E, states using the values for E,(O), E,(l)
and £.(2) given in Table AIO.l. This gives
2
3kT[2(~-
kT
B~)
3(
exp(-.£) + exp(i_) + +8~() exp (i_)]
kT
2(0 -0)
2kT
2(0
3 . 2kT
llett=
P exp( -~) + exp(~) + expC~r) J
2[ 2 2 2
where the factors of 2 arise because E 1 (a) and £ 1 (b), for example, make the
same contribution to the summation. This expression simplifies to
(~-8)exp(- 3 ')+8
ll~tt = k~ ex{d): 2J
kT
[
2kT
or, by putting (/KT = x,
2 (3x-8)exp( -3~)+8
/lett= x[exp( -3~)+2]
another equation first derived by Kotani and discussed in Section 9.9. This
is the equation used to give the theoretical plot in Fig. 9.4(a); for Fig. 9.4(b),
the corresponding but complete expression, including k and t, was used.
Appe ixll
High temperature
superconductors
At the present time, several different, but frequently related, classes of high
temperature (high T,) superconductors are known . The first class that will
be considered is based on La 2 Cu0 4 . This compound has a structure which
may be thought of as consisting of planes of octahedra of oxygen atoms, the
octahedra sharing corners, with a copper atom at the centre of each
octahedron, as shown in Fig. All.l. The lanthanum ions are sandwiched
between the oxygen/copper octahedra layers. The system becomes super-
conducting when Sr 2 + ions are substituted for La 3 + To maintain overall
charge neutralityl one Cu 2+ has to become Cu3+ for every La3+ substituted
by Sr 2 +. Superconductivity occurs, it seems, because of the presence of Cu 3 +
ions in the copper- oxygen planes. It also seems that the superconductivity
is localized in these planes.
The second class of superconductors, the so-called 1,2,3 compounds, are
of approximate formulae YBa 2 Cu 3 0 7 , although the oxygen content is again
variable. Neither yttrium (Y3+) nor barium (BaH) have variable valency so
charge neutrality requires that, for the above formula, of the three copper
ions, two are Cu2+ and one Cu 3 + This class of compound is structurally
more complicated than the first although they seem to have in common that
the superconductivity is associated with the coexistence of Cu2+ and Cu3+
in an oxide lattice. In the 1, 2,3 compounds the lattice is three-dimensional.
Further, there are two distinct copper sites. One, in which the copper ions
are square planar, seems an obvious place for the dB ion Cu3+ (the d B ions
Fig. A11.1 Structure of La2 Cu04 showing the

octahedral copper environment.
1 The present discussion neglects the fact that the substituted compound may lose oxygen,
although the evidence is that this loss may be important in the phenomenon of high temperature
superconductivity. At the present time, the oxygen loss is regarded as a fine-tuning mechanism;
for every oxygen atom lost two Cu 3 + must revert to Cu 1 +, to compensate for the fact that
o' - ~o.
Appendix 11 I 473
Pt 11, Pd" and, to a lesser extent, Ni 11, characteristically form square planar
complexes). In the second site the copper ions are five-coordinate, being
surrounded by oxygens at the corner of a square pyramid. Such an
environment is not uncommon in copper(Il) compounds and so this seems
to be an obvious site for the Cu 2 + ions. The ratio of occurrence of the square
to square pyramidal sites is 1:2. This is also the ratio of Cu3+ to Cu 2 +,
supporting the model presented for the location of the two different copper
ions.
The relative complexity of the 1,2, 3 and other, yet higher temper-
ature, superconductor structures has meant that the majority of theoretical
work has been concerned with superconductors based on La 2 Cu0 4 , the
La 2 _xSrxCu0 4 superconductors, and only these will be considered in this
appendix (the highest temperature superconductor reported at the time of
writing this appendix contains Sr, Ca and Bi as well as Cu and 0-it is
reported to superconduct at -23 'C). However, to highlight the problem of
explaining the superconductivity of the high T, materials we shall give an
outline of model which for the past 30 years has been the accepted model
of superconductivity. This is the Bardeen, Cooper and Schrieffer (BCS) model.
The simplest explanation of the BCS model is an anthropomorphic one.
Suppose that the reader (an electron) enters at one corner of a large, but
extremely crowded, room in which an informal party is in progress. The
guests (atoms) are making small, to some extent random, movements (atomic
vibrations). The reader (electron) is presented with the task of crossing to
the opposite corner of the room (this simulates an electron's contribution to
an electrical current). In crossing the room the reader will find themselves
frequently deflected-this simulates electrical resistance. Now, suppose that
an ultra important person (UIP)-much more important than a YIP-
together with their partner enters the room, in place of the reader. As the
pair cross the room they will experience no resistance, the guests part to
allow them free passage. Their movement models the movement of super-
conducting electrons through a lattice-these electrons also move as pairs,
so-called Cooper pairs. The essential of the BCS model is a coupling between
the movement of the superconducting electrons and the vibrational motion
of the lattice through which they move. Normally, BCS theory is applied to
materials which are superconducting at very low temperatures-at these
temperatures the vibrational motion of the atoms will be very small anyhow.
At higher temperatures, such as those relevant to the present discussion,
the apparently2 random motion of the atoms will be of much greater
amplitude. Does the BCS model still apply? Apparently not, and therein lies
the problem.
It is easy to demonstrate the problem of explaining the superconductivity
of the high-temperature systems. In Fig. A11.2 is shown a copper-oxygen
plane in La 2 Cu0 4 ; there is an additional oxygen above and below each
copper which, for simplicity, are omitted. Each copper(II) ion in Fig.
Al1.2 has a single unpaired electron and these are antiferromagnetically
coupled, just as is described in Section 9.11, giving the arrangement shown
2 In fact, the atomic excursions may be expressed as a linear sum of thermally populated
lattice modes combined with appropriate phase differences. In addition, there will be the zero
point vibrational amplitude of each mode to be considered, but this contribution becomes less
important as the temperature rises.
474 1 Appendix 11
I
0 0 0 0
I2 I2 I2 I
- 0 - Cu • - 0 - Cu • - 0 - Cu •-o- Cu 2• - 0 -
I I I I
0 0 0 0
I I I I
Fig. A11.2 The copper-oxygen plane in - 0 - Cu 2•-o- Cu 2 •-o- Cu 2 •-o- Cu 2•-o-
La2 Cu04 •
l I I I
0 0 0 0
I2 I2 I2 I
- 0 - Cu + - 0 - Cu + - 0 - Cu + - 0 - Cu 2+ - 0 -
I I I I
0 0 0 0
I I I I
Fig. A11.3 Antiferromagnetic coupling between

unpairec electrons on copper(ll) rows in La 2 Cu0 4 .
Fig. A11.4 Effect of Cu 2 + becoming Cu 3 +,

leading to the possibility of an electron hop.
schematically in Fig. All.3. When Sr2+ substitutes for La3+ this substitution
occurs between the copper-oxygen layers-and this is rather remote from
where the superconducting action takes place, or, rather, is believed to take
place-remote from the copper-oxygen planes of Fig. A11.2. When Cu2+
becomes Cu3+ an electron is lost; consider just one such substitution, a
substitution which leaves the structure shown schematically in Fig. A11.4.
In all forms of electrical conductivity, superconductivity not excluded,
electrons have to move from site to site; in the present case the obvious
electron hop, such as that indicated by a curved arrow in Fig. A11.4 leads
to the arrangement shown in Fig. A11.5. The key point is that as far as spin
arrangements are concerned, Figs. A11.4 and A11.5 are not identical; parallel
spins occupy adjacent sites in Fig. A11.5 but not in Fig. A11.4. The discussion
of Section 9.11 makes it quite clear that the arrangement of Fig. A11.5 is
energetically unfavourable. If it is correct, then superconductivity requires
energy to make it occur. Of course, this is just the opposite of reality and
so something is wrong with the ideas that led up to Figs. A11.4 and Al1.5.
Appendix 11 1 475
Rg. A11.5 The result of the electron hop

shown in Fit. A11.4 is that parallel spins occupy
adjacent sites.
Rg. A11.6 Two-electron synchronous hop with (a)

spin change (a) before and (b) after.
(b)
The model is wrong; the problem is how to correct it. At the present time
the answer is not known, although it is possible to make some relevant
comments.
First, the problem posed by Figs. All.4 and Al1.5 could have been
avoided had the conduction electron hopped across the diagonal of a square
of copper ions in Fig. A11.4 rather than along an edge. This alternative can
be excluded because the theory is unambiguous-there has to be some
overlap between the orbital of the before and after sites. The distance
involved in a diagonal hop is too great, particularly when it is remembered
that the orbitals of a Cu3+ ion will be contracted compared with those of
a Cu2+. Secondly, we have adopted a one-electron model. At several points
in this book it has been recognized that a one-electron model may not
provide an adequate explanation; is this another example? Almost certainly,
the answer is yes, but there is no agreement on the model to be adopted.
Perhaps the simplest two-electron model is illustrated in Fig. All.6, which
parallels Figs. A11.4 and A11.5. Two electrons hop synchronously, along the
476 I Appendix 11
sides of a square, exchanging spins in the process (this spin change is

indicated by a loop in Fig. A11.6). A single-spin flip, on its own, would
require energy.
Finally, the discussion has implicitly considered Cu 2 + in an octahedral
environment of oxygen atoms. If the Cu3+ were in the same environment it
would have two unpaired electrons (it would be t~ 9 e;). The electron lost in
moving Cu2+ ..... Cu3+ would not be the one shown as lost in figure
A11.4-because this would leave no unpaired electrons on the Cu3+ -but,
rather, one of the other Cu 2 + e9 electrons which have not been included in
the figures. Alternatively, and much more likely, the Cu3+ would undergo
a distortion (not a Jahn-Teller distortion because the Cu3+ would have a
3 A 29 ground state-see the detailed discussion of the d 8 case in Section 8.2)
which splits the e9 orbitals. Presumably, such a distortion would be

tetragonal, involving the oxygens above and below each copper ion in our
figures, the oxygens that were omitted for simplicity. Such a distortion, which
would be associated with a lattice vibration, brings us comfortably closer to
the BCS model with which the appendix started but, on its own, it leaves
unresolved the problem posed by Figs. All.4 and A11.5.
Further reading
A rather different approach from that given above will be found in 'Three
theories of superconductivity' by F. A. Matsen, J. Chern. Educ. (1987) 64, 842.
The structural patterns associated with superconductivity are well illustrated
in 'The Crystal Chemistry of High-Temperature Oxide Superconductors and
Materials with Related Structures' by H. Miiller-Buschbaum, Angew. Chern.,
Int. Ed. Engl. (1989) 28, 1472. A mixture of structure and some theory is to
be found in a very readable, non-mathematical, article 'Some Structural-
Electronic Aspects of High Temperature Superconductors', by J. K. Burdett,
in Adv. Chern. Phys. (1993) 83, 207.
A very readable review of recent development in both experimental and
theoretical aspects of chemical superconductors is to be found in the
September 1994 (page 722 on) issue of Chemistry in Britain.
Appendix12
Combining spin and orbital
angular momenta
In Section 11.3 the problem of combining spin and orbital angular momenta
to give the correct resultant was encountered. Put another way, how do
terms such as 3 H, as occurs for an f 2 ion such as Pr 111 and 4 I, appropriate
to an f3 ion such as Nd 111 , split when the spin and orbital motions couple
together? The solution of the general problem will be indicated by considering
these two examples in detail. This is done in Tables A12.1 and A12.2. In
each table down the left hand side are given all of the components of the
orbital angular momentum; at the top are listed all of the components of
the spin angular momentum. The entries in the body of the tables are total
angular momenta, obtained by adding the appropriate spin and orbital
momenta (those at the top and side of the table). In Table A12.1 it is not
difficult to see that all of the components of the state with total angular
e
momentum of 6 H 6 ) are covered by the left-hand column and final row
entries.
If these entries are deleted from the table, the components of the state
with total angular momentum 5 eH5 ) are similarly obtained from the table
remaining. Repeating this process gives the components of the total angular
e
momentum 4 H4 ) state. That this pattern is not accidental is evident from
Table A12.2, where it is repeated. Note particularly that each set of total
angular momenta (the entries within the shaded panels in Tables A12.1 and
A12.2) contain an entry from the figures in the first row of the table. Further,
that these entries, together, are a complete listing of the magnitudes of the
permissible angular momenta. It follows that, as described in Section 11.4,
the allowed values of the total angular momentum are obtained by taking
the maximum value of the orbital angular momentum and adding each
component of the spin angular momentum to it.
One final word of caution, one that also appears in Appendix A14 and
which explains the appearance in the tables of the word ' schematic'. What
has been described in this appendix is really a method of counting; it is not
a method of obtaining wavefunctions. So, it is not true that the component
of the total angular momentum = 6 state which has an angular momentum
of 0 along the z axis is the function with orbital angular momentum = -I
combined with spin= I, although this is the association made in the table.
The correct wavefunction is a linear combination of all three functions with
478 1 Appendix 12
Table A12.1 The (schematic) origin of the 3 H4 , 3 H5 and

3 H6 1evels of the 3 H term arising from the f 2 configuration
aH Spin angular momentum
1 0 -1
5 6 5 4
4 5 4 3
3 4 3 2
2 3 2 1
Orbital 1 2 1 0 . _ 3H4
angular 0 1 0 -1
momentum - 1 0 -1 -2
-2 -1 -2 -3
-3 -1 -3 -4
- 4 -3 -4 - 5 +-- 3Hs
-5 -4 -5 -6 ,._____ JHe
Table A12.2 The (schematic) origin of the 4 / 912 , 4 /

1112 , 4 / 1312 and 4 / 1512 levels of
the 4/ term arising from the f 2 configuration
., Spin angular momentum
3/ 2 1/2 -1/ 2 - 3/ 2
6 15/2 13/2 11/2 9/2
5 13/2 11/2 9/2 7/2
4 11/2 9/2 1/2 5/2
3 9/2 7/2 5/2 3/2
2 7/2 5/2 3/2 1/2
1 5/2 3/2 1/2 - 1/2 +--- 4/9!2
Orbital 0 3/2 1/2 - 1/2 - 3/2
angular - 1 1/2 - 1/2 - 3/2 - 5/2
momentum - 2 - 1/2 - 3/2 - 5/2 - 712
-3 - 3/2 - 5/2 - 7/2 - 9/2
-4 - 5/1 - 7/2 - 9/2 - 11/2 +--4111/2
-5 - 7/2 - 9/2 - 11/2 - 13/2 +--4/up
-6 -9/2 -11/2 -13/2 - 15/2 .._4,15/2
Problem A12.1
Construct a table similar to Tables A12.1 and A12.2 for the 5 I (f4 ) case and
show that the levels 5 I 4 , 5 I 5 , 5 I 6 , 5 I 7 and 5 I 8 are given by the procedure
described in this appendix (see also Section 11.4).
Bonding between a transition
metal atom and a Cn Rn ring,
n = 4, 5 and 6
In this appendix are merged two closely related approaches to the bonding
between a cyclic C,R, (n = 4, 5 and 6) unsaturated hydrocarbon and a
transition metal atom. The first approach is that developed in Chapter 10.
The axis between the centre of the C, R, ring and the metal atom is taken
as z and the orbitals of the metal atom classified according to the number
of planar nodes which they have and which contain the z axis (cutting the z
axis does not count). The 1t orbitals of the C,R, ring are classified using the
same criterion. Interaction- be it bonding or antibonding-only occurs
between orbital pairs with the same nodal count. All interactions between
orbitals with different nodal counts are zero. This approach is summarized
in Figure A13.1 . In this figure the metal orbitals are drawn in a direction
perpendicular to the z axis (thus enabling their identity to be established, s,
p, and d,, look much the same when viewed down z). However, the
hydrocarbon n molecular orbitals are viewed down z, because this enables
their nodality to be clearly seen. As is the common pattern-from orbitals
through to acoustics- the more nodes, the higher the energy. Of course, for
this simple statement to be true the vibrations under discussion must be
immediately related to each other. So, the statement is applicable to the n
orbitals of hydrocarbons, because different combinations of carbon p,
orbitals are involved. It does not apply to the metal orbitals because these,
s, p and d, have no such simple common basis. For unsaturated hydrocarbons
it means that the orbitals of highest nodality- of highest energy- are empty
in the isolated molecule. These higher orbitals, then, are regarded as 1t
electron acceptors in a transition metal complex and the (occupied) orbitals
of lower nodality are 1t electron donors. Schematic molecular orbital
diagrams based on these arguments are given in Figure A13.2. We do not
indicate the final occupancy of the molecular orbitals because this will
depend on which particular metal atom is involved.
The second, equivalent, approach is group theoretical (see Appendix 4).
The local symmetry of the transition metal- C,R, unit in a complex is C,v·
The transformation under the operation of the C,v character table of the n
carbon p, orbitals in the ring gives the irreducible representations subtended
480 I Appendix 13
by these Pn orbitals. The metal orbitals are similarly classified; inter-

actions occur between metal and n molecular orbitals of the same symmetry
species.
In Table A13.1 this general pattern is detailed for each of C4 RrM,
C 5 R 5-M and C 6 R 6-M rings symmetrically bonded to a transition metal.
Alongside each of the three character tables, for the C4 ., C5 • and C6 • point
groups, are given the species which describe the symmetry of transition metal
orbitals-so, s, p, and d,, orbitals transform as At in all three groups. The
tables show that there is always a c.R. n molecular orbital of At symmetry,
so interaction between metal and ligand orbitals of At symmetry can always
occur. Similar arguments apply to the other orbitals; it can therefore be seen
that the metal d,y orbital is strictly non-bonding in C4 RrM complexes
(there is no ligand orbital of B2 symmetry). Conversely, the ligand orbital
of B 2 symmetry in C 6 R6 complexes is non-bonding (there is no valence shell
metal orbital of B 2 symmetry).
z z
¢~ ~ Pz
0 nodal planes
z z
~~$-·,$ Py dzx dyz

1 nodal plane
FJg. A13.1 Classification of (a) metal and

(b) hydrocarbon" orbitals according to the
number of nodal planes they have which
contain the z axis (the axis of highest
(a) 2 nodal planes
symmetry) (continued).
Appendix 13 I 481
N~~m
0-0 6=0 0--Kb
_cp±y2
6-t-0
0~~~. -
- +
+ +
- -
+
~~
FJg. A13.1 (b) (continued).
~1Wr
'--r----'
2
0-e=
~
. . : I .
(b)
482 1 Appendix 13
(n+ l)p
(n+ l)s
(0)
(a)
Fig. A:l3.2 Schematic molecular orbital energy The connection between the two approaches described above is also
level diagrams for transition metal complexes of shown in Table Al3.1. At the right-hand side of each character table is listed
(a) C4 R4 , (b) C5 R5 and (c) C6 R6 • The ligand
and final molecular orbitals are classified by the number of nodal planes containing the z axis associated with each type
their symmetry species under the relevant point of interaction. There is something to be said in favour of each model in
group. The final molecular orbitals are also comparison with the other. Thus, the 'number of nodal planes' criterion does
classified by the number of nodal planes they
have which contain the z axis. In all cases, for not discriminate between B 1 and B2 (in the C4 • and C6 • cases), yet only one
simplicity, only interactions between d orbitals is involved in the bonding. Conversely, although in each case the number
and the ligand orbitals are shown. However, as of nodal planes quoted is the lowest, for each case higher numbers exist also
Table A13.1 shows, p orbitals must be expected
to be involved in the (1)-node interactions and (we give these for the A 1 case). In such cases, the number of nodal planes
both p and s orbitals in the (0)-node interactions. criterion provides a distinction that the symmetry criterion does not. Only
orbitals with the same number of nodal planes interact.
Appendix 13 1 483
TableA13.1
(a) C4 R4 : P. orbitals of C4 R4 transform as A1 + E + 81 (taking the C4 atoms to lie in the 2crv
mirror planes)
C4v E 2C4 c. 2u. 2a~ Metal orbitals C4R4 T<? Nodal planes
A1 1 1 1 1 1 S, Pz• dz2 yes 0 or 4

A• 1 1 1 -1 -1 no 8
81 1 -1 1 1 -1 dxLy2 yes 2
82 1 -1 1 -1 1 d, no 2
E 2 0 -2 0 0 Px• Py; dz:<, dyz yes 1
(b) CsRs: P. orbitals of C5 R5 transform as A1 + E1 + E2

c•• E 2C5 2~ Suv Metal orbitals CsRs Tt? Nodal planes
A1 1 1 1 1 S, Pz• dz2 yes 0 or 5
A• 1 1 1 -1 no 10
E1 2 2 cos 72 2 cos 144 0 Px• Py; dzx, dyz yes 1
E2 2 2 cos 144 2 cos 72 0 dx2-y2• dxy yes 2
(c) C6 R6 : p, orbitals of C6 R6 transform as A1 + E1 + E2 + 8 1 (taking the C6 atoms to lie in the

3crv mirror planes)
c•• E 2C6 2C3 C2 3av 3a~ Metal orbitals CsRsT<? Nodal planes
A1 1 1 1 1 1 1 S, Pz• dz2 yes 0 or 6
A2 1 1 1 1 -1 -1 no 12
81 1 -1 1 -1 1 -1 yes 3
82 1 -1 1 -1 -1 1 no 3
E1 2 1 -1 -2 0 0 Px• Py; dzx, dyz yes 1
E• 2 -1 -1 2 0 0 dx2-y2• dxy yes 2
Hole-electron relationship in
spin-orbit coupling
The argument used in this appendix is that of providing a simple illustrative

example. The general procedure used follows that of Appendix 12, which,
if necessary, should be read before proceeding further.
Consider the 7 F term arising from the f 8 configuration. We denote
electrons by arrows and holes by circles. The septet spin state may be
considered as arising from the following arrangement, either of eight
electrons or of six holes:
!
0
0
0
0
0
0
Working with holes and following the procedures of Appendix 12 we

have
7F Spin (holes) angular momentum
3 2 1 0 -1 -2 - 3
5 +--- 3Fo
3 6 4 3 2 1 0
2 5 4 3 2 1 0 - 1 ._3Ft
Orbital 1 l!\ 3 2 1 0 -1 -2 +--- 3F,
angular 0 3 2 1 0 -1 -2 -3 +--- 3F3
-3 +-- 3F4
momentum -1 2 1 0 - 1 -2 -4
- 2 1 0 -1 -2 -3 -4 - 5 + - 3Fs
0 -1 -2 -3 -4 -5 -6 ._JF6
-3
Appendix 14 1 485
The entries in this table can be classified according to the levels that result.
So, the 7 F6 level has components with resultant angular momentum of 6, 5,
4, 3, 2, 1, 0, -1, -2, -3, -4, -5 and -6. Deleting entries with these values
from the table leaves 5 as the highest value. This must come from a 7 F5 level,
with components 5, 4, 3, 2, 1, 0, -1, -2, -3, -4 and -5. Proceeding in
this manner it is concluded that the 7 F6 , 7 F5 , 7 F4 , 7 F3 , 7 F2 , 7 F, and 7 F0 levels
of the 7 F term are obtained. Had the 7 F arising from the f 6 configuration
been under discussion, we would have been working with electrons and the
7 F0 level would have been the most stable-the orbital and spin magnets
would have been arranged in opposed fashion,

N-S
s-N
but because we are considering the f 8 configuration, and so working with
holes, it is the 7 F6 level which is the ground state because it is this which
has the most stable arrangement, that with orbital and (hole) spin magnets
parallel.
Index
a band 391 B centre of symmetry (and selection rules) 271

p band 391 B (Racah parameter) 157, 163 character 440
p hydrogens 66 balanced basis sets 221, 224 character table 440
y (ratio ofTanabe-Sugano parameters) 164 band contours 170 charge transfer to solvent bands
b bonds 58, 277, 377 band intensities 171 (CTTS) 180
b conformation (of chelate rings) 432 band intensities in octahedral complexes charge-transfer spectra 156, 178
Ll. (delta) 127 459 Chait-Duncanson model 17, 212
Ll. conformation (of chelate complexes) 433 band structure 417 chelate effect 9, 84; 86
Ll.S' (entropy of activation) 327 Bardeen, Cooper and Schreiffer (BCS) circular dichroism (CD) 280, 392
Ll.V' (volume of activation) 327 model 473 circular polarization 278
((zeta) 165,192 base-catalysed hydrolysis of cobalt( III) cis effect 331
~ (eta) 28 species 335 cisplatin 382
"(kappa) 26 Berry pseudorotation 35 class a 80
A (lambda) 192 bidentate I 0, 11 class b 80
;_conformation (of chelate rings) 432 bioinorganic chemistry 381 classical ligands
A conformation (of chelate complexes) 433 bite 10 closo (cages) 355
I' (mu) 27 bite angle 10 cluster valence electron (CVE) counts 358
"'" 189, 196, 471 Bloch functions 422 complete neglect of differential overlap
l'orr (f election systems) 266 blue copper proteins 398 (CNDO) 223
n bonding ligands 17 blueprints 340 conduction band 420
n orbitals in common geometries 112 body centred cubic 408 cone angle 22, 91
rJ orbitals in common geometries 112 Bohr magneton 189 configuration interaction 221
10 Dq 127 Boltzmann distribution 186, 195 configurations 122
IS-electron rule 18, 211, 353, 357 bond-stretch isomerism 2, 43 conformation isomerism 43
bonding in cluster compounds 345 conjugate base mechanism 335
bonding; C.R. ring-transition metal 479 connectivities 361
A Born-Oppenheimer approximation 173 controlled potential electrolysis 60
A (associative mechanism) 326 boron hydrides 349, 354 coordination 2
A (Racah parameter) !57 Brillouin zone 412, 415, 418, 422, 429 coordination compound 1
ab initio 220, 267, 292, 345 Brillouin zone (body centred cubic) 423 coordination isomerism 44
absolute configurations 281 Brillouin zone (face centred cubic) 416,424 coordination number 4, 31
absolute hardness 83 Brillouin zone (monoclinic) 416 coordination number one 32
actinides 238 Brillouin zone (orthorhombic) 416 coordination number two 32
actinometry 340 Brillouin zone (tetragonal) 416 coordination number three 32
addition reactions 51, 52 Brillouin zone (triclinic) 416 coordination number four 34
adiabatic demagnetization 266 Brillouin zone (trigonal) 416 coordination number five 35
alternating force magnetometer 462 Butterfly clusters 358 coordination number six 38
ambidentate 26 coordination number seven 38
angular momentum 245 c coordination number eight 39
angular overlap model 226 C (Racah parameter) 157, 164 coordination number nine 41
antiferromagnetic linking 206 c4, character table 483 coordination number 10 and above 41
antiferromagnetism 203 C5 , character table 483 coordination numbers 7-12 (in
antisymmetry 140 C6 , character table 483 lanthanides) 243
arachno (cages) 355 case A (in MCD) 396 coordination position isomerism 44
atom-superposition electron-delocalization case C (in MCD) 396 Cotton effect 280
molecular orbital (ASED-MO) catalysis (and clusters) 379 counterintuitive mixing 224
method 224 catenand 3 covalency in transition metal complexes
avoided crossing 419 cellular angular overlap model 226 295
488 1 Index
Creutz-Taube species 323, 341 electroneutrality 116 group theory 95, 122, 440
cross-reactions 322 electronic spectra 156 Guoy method 462
crown ether 3, 14 electronic spectra off electron systems 257
cryptosolvolysis 333 electronic spectra of V111 and Ni 111 157
cryptands 14, 91 enantiomers 46, 433 H
encounter complex 331 hapticity 28
crystal field splittings 123
crystal field stabilization energies 306, endohedral species 70, 215 hard acids 82
309, 337 energy level diagram for f 2 256 hard and soft acids and bases (HSAB) 81
hard bases 82
crystal field stabilization energies (of enthalpy 87
lanthanides) 312 entropy 87 heats of ligation 305
entropy of activation 327 hemoglobin 305, 384
crystal field theory 121
ethane 17 heterogeneous catalysis 21
crystal molecular orbitals 410
ethenetetracarbonyliron 232 hexacarbonylchromium 229, 293
cubic setoff orbitals 241, 242
exchange coupling and spectra 182 hexadentate 13
Curie-Weiss law 203, 205
excited states off electron systems 254 high spin 128, 130
cyclic voltammetry (CV) 297
exohedral 216 high temperature superconductors 208, 472
cylindrical symmetry 348
extended Hucke! method 223,232,345,376 highest occupied molecular orbital
extended X-ray absorption fine structure (HOMO) 3, 83, 212, 423
D
(EXAFS) 394, 400, 402 hole-electron relationship (in spin-orbit
d orbitals in an octahedral crystal field 124
coupling) 484
D (dissociative mechanism) 326
F
holes 133
D3, character table 120
f electron charge-transfer transitions 262 Hund's rules 138, 142
D4 direct product table 443
f electron systems 238 hydrate isomerism 45
D4 , character table 440
f orbital covalency 219 hyperfine structure (in EPR) 289
de Haas-van Alphen effect 424
hypersensitive transitions 261
density functional method 222, 267, 376 f orbital involvement in bonding 267
f orbital shapes 240 hypha (cages) 355
density of states (DOS) 376, 418
determination of magnetic susceptibilities f orbitals in an octahedral crystal field 126
462 f--> d transitions 257, 262
diamagnetism 186 f--> f transitions 257, 260 1 (interchange mechanism) 326
didentate 10 face bonds 349 I, (associative interchange mechanism) 327
differential thermal analysis (DTA) 62 face centred cubic 408 '• (dissociative interchange mechanism)
direct product 138, 442 Faraday method 462 327
dispersion 419, 429, 430 Fenske-Hall method 223 icosahedral 216
dissociation reactions 51 Fermi surface 420, 423 incongruent solubility 55
distorted octahedral crystal fields 153 ferredoxins 401 inelastic neutron scattering 270
distortional isomerism 43 ferrimagnetism 203 inert complexes 9, 52, 73, 318
donor-acceptor 1 ferrocene 212, 227, 292, 294 infrared spectroscopy 270
double groups 249 ferromagnetic linking 206 inner-sphere mechanism 323
droplet model 376 ferromagnetism 203 insertion reactions 66
dynamic exchange in NMR 319 first-order Zeeman effect 196, 467 intermediate field complexes 143
dynamic Jahn-Teller effect 170, 200 flash photolysis 392 intermediate neglect of differential overlap
flexidentate 26 (INDO) 223
E fluorescence 342 intraligand charge-transfer spectra 178,
edge bonds 34 7 fluxional isomerism 4 7 181
Eigen-Wilkins mechanism 331 fluxional molecules 21, 329, 338 ion replacements 239, 384
electric dipole allowedness 156, 270, 459 flyover complex 19 ionic potential 81
electric field gradient 287 formal oxidation states 115 ionic radius 81, 91, 239, 303
electric quadrupole moment 285 free-electron models 362 ionization isomerism 45
electrochemical methods 296 free energy 87, 311 irreducible representation 443
electron correlation 107, 118,220,228, 377 fullerenes 215, 362 Irving-Williams order 81, 311
electron deficiency 349 isoleptic 27
electron density difference maps 117 G isolobal principle 234, 345, 357
electron paramagnetic resonance (EPR) g tensor 289 isomer shifts 287
288, 397, 400, 402 general set off orbitals 241, 244
electron repulsion 107, 118, 128, 130, 138, geometrical isomerism 44, 209
143, 220, 223, 236, 245 gerade 96, 108, 131
electron spin resonance (ESR) 288 graph theory 361 Jahn-Teller effect 164, 166,293,311
electron-transfer reactions 317, 320 group (symmetry) 440 J ahn-Teller splitting 190
electron transfer 1 group constants (in protonation jellium model 376
electronegativity 82 equilibria) 404 Job's method 78
Index 1 489
K magnetism and lanthanides and actinides 0

k-space 426 265 0 character table 98, 444
Koopman's theorem 230, 292 Marcus-Hush theory 322 o, character table 360
Kotani theory 197, 469 matrix isolation 69 0, direct product table 139, 460
Kramer's doublet 193, 196, 290, 467 Maxwell model of light 259 optical activity 281
mechanisms of ligand substitution optical electronegativities 180
reactions 325 optical isomerism 46
L
metal clusters 22 optical rotary dispersion (ORD) 279
labile complexes 9, 52, 318
metal orbitals in common geometries 112 optically active molecules (spectroscopy
lanthanide colours 258
metal to ligand charge transfer transitions of) 277
lanthanide contraction 239
(MLCT) 179 orbital and spin angular momenta; their
lanthanide ion ground state terms 247
metal-fullerene complexes 215 combination 477
lanthanide luminescence 263
metal-ligand " interactions in octahedral orbital contribution to a magnetic
lanthanides 238
complexes I 03 moment 189
Laporte allowedness 163
metal-ligand rr interactions in octahedral orbital degeneracy 123
laser pulse techniques 340, 392
complexes 97 orbital energies 132
lattice energy 89, 307
metal-metal bonds 377 orbital magnets 189
lattice vibrations 199
metal to ligand charge transfer orbital reduction factor 195
Lewis acid 3
(MLCT) 341 Orgel diagrams 145
Lewis base 3
methemoglobin 387 outer-sphere mechanism 321
ligand 2
metmyoglobin 387 overall formation constants 74
ligand " group orbitals in a tetrahedral
microconstants (in protonation oxidation reactions 51
complex 110
equilibria) 404 oxidation state 4, 25
ligand " group orbitals in an octahedral
Mobius strip 251 oxidation-reduction reactions 58
complex 104
model compounds 382, 386, 388, 395 oxidative addition 60
ligand rr group orbitals in a tetrahedral
molar conductivities 296
complex 108
molecular mechanics 314
ligand rr group orbitals in an octahedral p
molecular orbital energy level scheme for
complex 100, 449
a tetrahedral complex Ill pairing energy 129
ligand field theory !53
molecular orbital energy level scheme for paramagnetism 186
ligand isomerism 46
an octahedral complex 102 particle size effects 430
ligand to metal charge transfer transitions
molecular orbital energy level scheme for Pascal's constants 464
(LMCT) 178
ferrocene 213, 229 Pauling's electroneutrality principle 116
ligand to f charge transfer bands 257
molecular orbital energy level scheme for peroxidases 393
linear free-energy relationships 325
hexacarbonylchromium 218, 231 pH change 269, 403
linkage isomerism 45
molecular orbital energy level schemes phosphorescence 342
low spin !29, 130
low-lying excited states (f electron) 266
c. R. ring-transition metal; n = 4-6 482 phosphors 264
molecular structure (through vibrational photochemical preparative methods 70
low-symmetry ligand fields 192
spectroscopy) 274 photoelectron spectroscopy (PES) 291,
lowest unoccupied molecular orbital
Miissbauer spectroscopy 286, 390, 402 345, 377
(LUMO) 3, 83, 212, 235
muffin tin 222 photokinetics 339
multidentate 30 picket fence porphyrin 3, 14, 389
M myoglobin 384 polarization 260, 261
Madelung constants 310 polarized light 278
magnetic circular dichroism (MCD) 396, N polarized neutrons 205
399 natural order 81 polarizing power 81
magnetic coupling 187 Nee! point 205 polydentate 30
magnetic coupling and spectra 182 nephelauxetic series 163 polyhedral symbol 29
magnetic dilution 187 nerve impulses 383 polymerization isomerism 45
magnetic dipole allowedness 173, 259 neutron diffraction 300 polytypism 429
magnetic fields (influence on reaction nido (cages) 355 porphyrin 14, 15
kinetics) 327, 330 nitrogen fixation 401 pre-equilibria (in reaction kinetics) 331,
magnetic moments of first row transition nuclear magnetic resonance (NMR) 282, 336
metal ions 194 392 precursor complex 326
magnetic non-dilution 203 nuclear magnetic resonance and reaction primary valency 4
magnetic properties of complexes 185 kinetics 334 primitive lattice 408, 429
magnetic space groups 205 nodal plane patterns 214,217,409,412,479 protonation equilibria (in bioinorganic
magnetic susceptibility 187 nomenclature 24 systems) 403
magnetic susceptibility of a d, ion 466 non-crossing rule 141 pseudorotation 35
magnetic unit cells 205 nuclear quadrupole resonance (NQR) 285 puckering (of chelate rings) 432
490 !Index
Q spin allowed transitions 254 Tanabe-Sugano diagrams (d'-7 ) 147

quadrupole (metal-metal) bonds 378 spin and orbital angular momenta; their r. character table 237, 346
quadrupole allowedness 261 combination 477 temperature independent paramagnetism
quantum size effects 430 spin contribution to a magnetic moment (TIP) 191, 265, 284, 376, 396
quaternary structures 383 191 template reactions 21, 67
quenching 189, 190 spin degeneracy 123, 131 tensor surface harmonics 363
spin density difference maps 118 term energies 132
R spin equilibria 209 terms 122
Racah parameters 157 spin forbidden transitions 162, 163, 173, tetradentate 12
Raman spectroscopy 270 182, 254 tetrahedral complexes 107, 148, 175
reaction coordinate 325 spin isomerism 44, 48, 209 thermal dissociative reactions 61
reaction intermediate 326, 391 spin-coupled valence bond theory 236 thermodynamics and ligand fields 303
reaction kinetics of coordination spin-only model 201, 265 thermogravimetric analysis (TGA) 62
compounds 316 spin-orbit coupling 123, 164, 173, 191, three-centred orbitals 349
reactions of coordinated ligands 65 221, 235, 238, 247, 255, 293, 484 time-resolved circular dichroism
reciprocal space 416 spin-orbit coupling constants 192 (TRCD) 392
reciprocal lattice 411, 416 spin-orbit coupling in pictures 249 topological equivalent orbital (TEO) 360
redox spectra 180 spin-orbit levels arising from crystal field topological models 359
reducible representation 443 terms 165 trans effect 68, 330
reduction reactions 51 spin-orbit states in f electron trans influence 92, 330
reductive elimination reactions 61 systems 248, 255 transition state 326
relativistic effects 221, 235, 243, 267, 379 spinels 206, 310 translational invariance 429
relaxation phenomena 319 square planar complexes 149 translation group character table 408
resonance Raman spectroscopy 275, 390 stability constants 74, 311 travelling waves (Brillouin zone) 413
ring whizzers 48, 339 stability correlations 80 tridentate 10, 12
Robins-Day classification 182, 325, 426 states 122 triflate 57, 297
rotating charge densities 245 statistical effect 84
rotational barrier 48 stepwise formation constants 74
rotational invariance 224 steric effects 90 u
rubedoxins 401 steric interactions 22 ungerade 96, I 08, 131
Rule of average environment 130, 177 stoichiometric stability constants 80 unit cell 417
Russell-Saunders coupling 126, 446 strong field 130 unit cell (and vibrations) 273
Russell-Saunders terms (f electron) 255 strong field complexes 136 unpaired electron densities 283
Russell-Saunders terms from d" strong field substates 137
configurations 144 structural isomerism 47
Russell-Saunders terms; crystal field substitution reactions 51, 54 v
splittings 144 substitution reactions of octahedral valence band 420
complexes 331 valence bond 97
s substitution reactions of square planar valence shell electron pair repulsion
Schlenk tube techniques 63 complexes 328 (VSEPR) model 43, 435
Schri:idinger equation 220 successor complex 326 van Vleck relationship 471
second-order Zeeman effect 196, 467 superconducting quantum interference vibrating sample magnetometer 462
secondary valency 4 device (SQUID) 462 vibrational spectroscopy 270, 314, 390
self-consistent charge and configuration supercritical fluids 70 vibronic coupling 173, 200
(SCCC) 225 superexchange 208 vibronic transitions 259
self-consistent field 221 symmetry element 95, 440 volume of activation 327
semiconductor 420 symmetry operation 95, 440
semiempirical methods 222 symmetry species 443
sensitizers 264 synergic 17 w
sepulchrate 3, 14, 67 symbiosis 82 Wade's rules 353, 374
shift reagents 284 weak field 130
site preference energies 308, 311 T weak field complexes 130
site-directed mutagenesis 387 Tanabe-Sugano diagram (d 2 } 159 Werner, A. 4
soft acids 82 Tanabe-Sugano diagram (d 3 ,
soft bases 82 expanded) 343
solid state theory 407 Tanabe-Sugano diagram (d 3 } 172 X
solvento complexes 333 Tanabe-Sugano diagram (d 8 ) with X-ray crystallography 47, 205, 209, 287,
Sorel band 391 spin-orbit coupling 174 299, 315
spectra of crystalline materials 428 Tanabe-Sugano diagrams X-rays 393, 428
spectrochemical series 129 (collective) 162, 455 Xa 220, 267, 345, 376, 399

Physical Inorganic Chemistry - A Coordination Chemistry Approach PDF

Uploaded by

Copyright:

Available Formats

Physical Inorganic Chemistry - A Coordination Chemistry Approach PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Physical Inorganic Chemistry - A Coordination Chemistry Approach PDF

Uploaded by

Copyright:

Available Formats

What are coordination compounds?

What are coordination compounds?

What are some common coordination numbers and geometries?

What are some common coordination numbers and geometries?

Springer-Verlag Berlin Heidelberg GmbH

Library of Congress Cataloging-in-Publication Data

1. Physical inorganic chemistry 2. Coordination compounds.

Copyright© 1996 S. F. A. Kettle

Set by KEYWORD Publishing Services, London

Foreword xiii 3.3 What determines coordination number

3.4 Isomerism in coordination compounds 43

1 3.4.1 Conformation isomerism 43

2.1 Classical ligands, classical complexes 7 3.4.6 Hydrate isomerism 45

Stability of coordination 7.9 Square planar complexes 150

compounds 73 7.10 Other stereochemistries 152

5.2 Stability constants 74 8

5.5 Statistical and chelate effects 84 8.1 Introduction 156

5.6 Solid complexes 89 8.2 The electronic spectra of ym and Ni 11

6 8.5 Jahn-Teller effect 166

Molecular orbital theory of transition 8.6 Band contours 170

metal complexes 95 8.7 Band intensities 171

6.4 Complexes of other geometries 110 9

9.1 Introduction 185

7.1 Introduction 121 9.4 Spin contribution to a magnetic

7.1 Non-octahedral complexes 148 9.9 An example 195

10 12.1 Introduction 269

Beyond ligand field theory 211 12.2 Vibrational spectroscopy 270

Reaction kinetics of coordination 16.4 Peroxidases 393

14.1 Introduction 317 16.6 Nitrogen fixation 401

15.7 Clusters and catalysis, a comment 379

I am no doubt representative of a large number of current inorganic chemists in having

picking up an advanced text on magnetochemistry, for example, will probably not

Textbooks on physical inorganic chemistry can, during their preparation,

although it is to be emphasized that, once formed, there is no difference

Further reading nial, published in 1967 by the American Chemical Society as

Question cobaltous oxide. Peculiar compounds are produced in this solu-

2.1 Classical ligands, classical complexes

Donor atoms from group 17(7) F- flu oro

Donor atoms from group 16(6) o2- oxo

Donor atoms from group 15(5) cw cyano

that was previously regarded as difficult, or the preparation and character-

2.2 Novel ligands, novel complexes

this recommendation has yet to gain general acceptance.

Name Common abbreviation Structure

Table 2.3 (continued)

Name Common abbreviation Structure

2 ,2' ,6' ,2"-Terpyridyl terpy

Tetradentate• ligands /CH2-CH2

Table 2.3 (continued)

p; porphyrins are usually

Table 2.4 Some more exotic polydentate ligands

Table 2.4 (continued)

Table 2.5 Selection of complexes formed by some polydentate ligands

The square planar complex [Ni(dmg)2 ]: note

The K+ complex of dibenzo-18-crown-6

Table 2.5 (continued) +

Rb+ 2,2,2-cryptate: oxygen and

The crystal structure of the M = Ag+ species Cu 2• phthalocyanin: a complex which