Chemical Applications of Group Theory 3rd Edition

Copyughled Materral
Copyright e 1990 by F. Albert Conon

Copyright e 1%3. 1971 by John WiiC)' & Sons. lnc.
AU rights rcscrved. Pubhshed simuhaneously in Canada.
Rcproduction or translauon of any part of this wo rk
bcyond that pc:nmncd b)' S(clion 107 or 108 of the
1976 United Statcs ("opyright Acl thc pcrmission
of the copynghr owncr is unlawful. for
pcrmission or rurther information should be 10
the Dcparlment. John Wiley & lnc.
l. ibrury of Congnss Caruloging in Pub/icario' Dura:
Conon. F. Albert (frank Albert ). 1930-
Chcmical applicarions of thcory 1 L Albcn Couon.-J rd ed.
p. cm.
" A Puhlication.
p.
ISBN 0-471-51094-7
1. Molecular thcory. 2. Group rhcor)'. l. Tllc.
1990
541' .22'015122-<k20
Printed in thc Uniled of Amcrica
20 19 18 17 16 IS l4 13
89-16434
CIP
CHE MI CAL
APPLICATIONS
OF GROUP THEORY
F. Albert Cotton
Department of Chemistry, Texas A&M University
College Station, Texas
THIRD EDITION
11
WILEY
A WILEY-INTERSCIENCE PUBLICATION
JOHN WILEY & SONS
New York 1 Chichester 1 Brisbane 1 Toronto 1 Singapore
CONTENTS
Part l. Principies
l. INTRODUCTION
2. DEFINmONS ANO THEOREMS
OF GROUP THEORY
2.1 The Defining Properties of a Group, 6
2.2 Sorne Examples of Groups, 8
2.3 Subgroups, 12
2.4 Classes, 13
Exercises, 16
3. l\'IOLECULAR SYMMETRY AND THE
SYMMETRY GROUPS
3.1 General Remarks, 17
3.2 Symmetry Elements and Operations, 18
3.3 Symmetry Planes and Reflections, 18
3.4 The Inversion Center, 22
3.5 Proper Axes and Proper Rotations, 22
3.6 Improper Axes and lmproper Rotations, 27
3. 7 Products of Symmetry Operations, 29
3.8 Equivalent Symmetry Elements and Equivalent
Atoms, 32 .
3.9 General Relations Among Symmetry Elements and
Operations, 33
3
17
xi
xii CONTENTS
3.10 Symmetry Elements and Optical Isomerism, 34
3.11 The Symmetry Point Groups, 39
3.12 Symmetries \Vith Multiple High-Order Axes, 44
3.13 Classes of Symmetry Operations, 50
3.14 A Systematic Procedure for Symmtry Classification of
Molecules, 54 .
3.15 Illustrative Examples, 56
Exercises, 61
4. REPRESENTATIONS OF GROUPS
4.1 Prefactory Comments on Matrices and Yectors, 68
4.2 Representations of Groups, 78
4.3 The "Great Orthogonality Theorem" and lts
Conseq uences., 81
4.4 Character Tables, 90
4.5 Representations for Cyclic Groups,. 95
Exercises, 99
68
5. GROUP THEORY AND QUANTUM MECHANICS 100
5.1 Wave Functions as Bases for Irreducible
Representations, 100
5.2 The Direct Product, 105
5.3 ldentifying Nonzero Matrix Elements, 109
Exercises, 113
6. SYMMETRY-ADAPTED LINEAR COMBINATIONS 114
6.1 Introductory Remarks, 114
6.2 Derivation of Prpjection Operators, 115
6.3 Using Projection Operators to Construct SALCs, 119
Exercises, 129
Part ll. Applications
7. MOLECULAR ORBITAL THEORY AND ITS
APPLICATIONS IN ORGANIC CHEMISTRY
7.1 General Principies. 133
7.2 Symmetry Factoring of Secular Equations, 140
7.3 Carbocyclic Systems, 142
7.4 More General Cases of LCAO-MO n Bonding, 159
7.5 A Worked Example: Naphthalene, 172
133
7.6 Electronic Excitations of Naphthalene: Selection Rules and
Configuration lnteraction, 176
CONTENTS xi
7. 7 Three-Center Bonding, 180
7.8 Symmetry-Based Selection Rules for Cyclization
Reactions, 188
Exercises.. 201
8. MOLECULAR ORBITAL THEORY FOR INORGANIC
AND ORGANOMETALLIC COMPOUNDS 204
8.1 Introduction, 204
8.2 Transformation Properties of Atomic Orbitals, 205
8.3 Molecular Orbitals for a Bonding in ABn MoJecu)es:
The Tetrahedral AB_. Case, 209
8.4 Molecular Orbitals for a Bonding in Other ABn
Molecules, 214
8.5 Hybrid Orbitals, 222
8.6 Molecular Orbitals for n Bonding in AB, Molecules, 227
8.7 Cage and Cluster Compounds, 230
8.8 Molecular Orbitals for Metal Sand\vich Compounds, 240
Exercises, 251
9. LIGAND FIELD THEORY 253
9.2 Electronic Structures of Free Atoms and Ions, 254
9.3 Splitting of Levels and Terms in a Chemical
Environment, 260
9.4 Construction of Energy Level Diagrams, 265
9.5 Estimation of Orbital Energies, 281
9.6 Selection Rules and Polarizations, 289
9.7 Double Groups, 297
Exercises, 302
10. MOLECULAR VIBRATIONS
10.2 The Symmetry of Normal Vibrations, 305
10.3 Determining the Symmetry Types of the Normal
Modes, 309
304
10.4 Contributions of Particular Interna( Coordinates to Normal
Modes, 315
10.5 How to Calcula te Force Constants: The F and G Matrix
Method, 317
10.6 Selection Rules for Fundamental Vibrational
Transitions, 324
10.7 Illustrative Examples, 328
10.8 Sorne Important Special Effects, 338
Exercises, 346
xiv CONTENTS
11. CRYSTALLOGRAPHIC SYMMETRY
. 11.1 Introduction, 348
11.2 The Concept of a Lattice-In Two Dimensions, 350
11.3 Two-Dimensional Space Symmetries, 358
11.4 Three-Dimensional Lattices and Their Symmetries, 368
11.5 Crystal Symmetry: The 32 Crystallographic Point
Groups, 375
11.6 Interrelating Lattice Symmetry, Crystal Symmetry, and
Diffraction Symmetry, 380
11.7 Additional Symmetry Elements and Operations: Glide
Planes and Screw Axes, 384
11.8 The 230 Three-Dimensional Space Groups, 388
11.9 Space Groups and X-Ray Crystallography, 400
Exercises, 410
Part lll. Appendices
l. MATRIX ALGEBRA
IIA. CHARAcrER TABLES FOR CHEMICALLY
IMPORTANT SYMMETRY GROUPS
liB. CORRELATION TABLE FOR GROUP O,
111. SOME REMARKS ABOUT THE RESONANCE
INTEGRAL p
IV. THE SHAPES OF j ORBITALS
V. CHARAcrER TABLES FOR SOME DOUBLE
GROUPS
VI. ELEMENTS OF THE g MATRIX
VII. STEREOGRAPHIC PROJECTIONS FOR THE 32
CRYSTALLOGRAPHIC POINT GROUPS
VIII. THE 230 SPACE GROUPS
IX. READING LIST
lndex
348
417
426
437
438
441
443
444
448
450
452
457
CHEMICAL
APPLICATIONS
OF GROUP THEORY
PART 1
PRINCIPLES
1
INTRODUCTION
The experimental chemist in bis daily work and thought is concerned with
observing and, to as great an extent as possible, understanding and inter-
preting bis observations on the nature of chemical compounds. Toda y, chem-
istry is a vast subject. In arder to do thorough and productive experimental
work, one must know so much descriptive chemistry and so much about
experimental techniques that there is not time to be also a master of chemical
theory. Theoretical work of profound and creative nature, which requires a
vast training in mathematics and physics, is now the particular province of
specialists. And yet, if one is todo more than merely perfonn experiments ..
one must ha ve sorne theoretical framework for thought. In arder to formulate
experiments imaginatively and interpret them correctly, an understandng of
the ideas provided by theory as to the behavior of molecules and other arrays
of atoms is essential.
The problem in educating student chemists-and in educating ourselves-
is to decide what kind of theory and how much of it is desirable. In other
words, to what extent can the experimentalist afford to spend time on the-
oretical studies and at what point should he say, uBeyond this I ha ve not the
time or the inclination to go?" The answer to this question must of course
. vary with the special field of experimental work and with the individual. In
sorne areas fairly advanced theory is indispensable. In others relatively little
is useful. For the most part, however, it seems fair to say that mo]ecular
quantum mechanics, that is, the theory of chemical bonding and molecular
dynamics, is of general importance.
As we shall see in Chapter 5, the number and kinds of energy levels that
an atom or molecule may have are rigorously and precisely determined by
the symmetry of the molecule or of the environment of the ato m. Thus, from
3
4 PRINCIPLES
symmetry considerations alone, we can always teJI what the qualitative fea-
tures of a problem must be. We shall know, without any quantitative calcu-
lations whatever, how many energy states there are and what interactions
and transitions between them n1ay occur. In other words, symmetry consid-
erations alone can give us a complete and rigorous answer to the question
uWhat is possible and v.rhat is completely impossible?" Symmetry consider-
ations alone cannt;>t, however .. tell us ho\v likely it is that the possible things
will actually take place. Symmetry can tell us that, in principie, two states of
the system nzust differ in their energy, but only by computation or ineasure-
ment can \Ve determine how great the difference \vill be. Again .. symmetry
can tell us that only certain absorption bands in the electronic or vibrational
spectrum of a molecule may occur. But to learn where they "'ill occur and
how great their intensity will be, calculations must be made.
Sorne illustrations of these statements may be helpful. Let us choose one
illustration from each of the five major fields of application that are covered
in Part II. In Chapter 7 the symmetry properties of molecular orbitals are
discussed, with emphasis on the n: molecular orbitals of unsaturated hydro-
carbons, although other systems are also treated. It is shown hov.' problems
involving large numbers of orbitals and thus, potentially, high-order secular
equations can be formulated so that symmetry considerations simplify these
equations to the maximum extent possible. It is also shown how symmetry
considerations permt the development of rules of great simplicity and gen-
erality (the so-called Woodward-Hoffmann rules) goveming certain concerted
reactions. In Chapter 8, the molecular orbital approach to molecules of the
ABn type is outlined. In Chapter 9 the symmetry considerations underlying
the main parts of the crystal and ligand field treatments of inner" orbitals in
complexes are developed.
In Chapter 10, it is shown that by using symmetry considerations alone \Ve
may predict the number of vibrational fundamentals, their activities in the
infrared and Raman spectra, and the way in which the various bonds and
interbond angles contribute to them for any molecule possessing sorne sym-
metry. The actual magnitudes of the frequencies depend on the interatomic
forces in the molecule, and these cannot be predicted from symmetry prop-
erties. However, the technique of using symmetry restrictions to set up the
equations required in calculations in their most amenable form (the F-G
matrix method} is presented in detail.
In Chapter 11 of this edition, the symmetry properties of extended arrays,
that is, space group rather than point"group symmetry, is treated. In recent
years, the use of X-ray crystallography by chemists has increased enonnously.
No chemist is fully equipped todo research (or read the literature critically)
in any field dealing with crystalline compounds, without a general idea of the
symmetry condtions that govem the formation of crystalline solids. At least
the rudiments of this subject are covered in Chapter 11.
The main purpose of this book is to describe the methods by which we
can e ~ t r a c t the information that symmetry alone will provide. An understand-
INTRODUCfiON 5
ing of this approach requires only a superficial knowledge of quantum me-
chanics. In several of the applications of symmetry methods, ho\vever, it \VOuld
be artificial and stultifying to exdude religiously all quantitative considera-
tions. Thus, in the chapter on molecular orbitals, it is natural to go a few
steps beyond the procedure for determining the symmetries of the possible
molecular orbitals and explain ho\v the requisite linear combinations of atomic
orbitals m ay be written down and how their energies may be estimated. lt
has also appeared desirable to introduce sorne quantitative ideas into the
treatment of ligand field theory.
lt has been assumed. necessarily, that the reader has sorne prior familiarity
with the basic notions of quantum theory. He is expected to know in a general
way what the wave equation is, the significance of the Hamiltonian operator.
the physical meaning of a wave function, and so forth, but no detailed knowl-
edge of mathematical intric!lcies is presumed. Even the contents of a rather
qualitative book such as Coulson's Valence should be sufficient., although, of
course, further background knowledge will not be amiss.
The following comments on the organization of the book may prove useful
to the prospective reader. lt is divided into two parts. Part 1, which includes
Chapters 1-6, covers the principies that are basic to aH of the applications.
The applications are described in Part 11, embracing Chapters 7-11. The
material in Part I has been written to be read sequentially; that is, each chapter
deliberately builds on the material developed in all preceding chapters. In
Part 11. however. the aim has been to keep the chapters as independent. of
each other as possible without excessive repetition, although each one .. of
course, depends on all the material in Part l. This plan is advantageous to a
reader whose immediate goal is to study only one particular area of appli-
cation, sin ce he can proceed directly to it, whichever it m ay be; it al so allows
the teacher to select which applications to cover in a course too short to
include all of them, or, if time permits, to take them all but in an order
different from that chosen here.
Certain specialized points are expanded somewhat in Appendixes in order
not to divert the main discussion too far or for too long. Al so, sorne use fui
tables are given as Appendixes. Finally, Appendix IX provides a reference
list for each of the five chapters in Part II. indicating where further discussion
and research examples of the various applications may be found.
2
DEFINITIONS AND THEOREMS OF
GROUP THEORY
2.1 THE DEFINING PROPERTIES OF A GROUP
A group is a collection of elements that are interrelated according to certain
rules. We need not specify what the elements are or attribute any physical
significance to them in order to discuss the group which they constitute. In
this book, of course, we shall be concerned with the groups formed by the
sets of symmetry operations that may be carried out on molecules or crystals,
but the basic definitions ~ theorems of group theory are far more general.
In order for any set of elements to forma mathematical group. the following
conditions or rules must be satisfied.
l. The product of any two elentents in tite group and the square of each
ele1nent must be an elenzent in the group. In order for this condition to have
meaning, we must, of course., have agreed on what we mean by the terms
"multiply'' and "product." They need not mean what they do in ordinary
algebra and arithmetic. Perhaps we might say "combine" instead of hmulti-
ply" and "combination" instead of "product" in order to avoid unnecessary
and perhaps incorrect connotations. Let us not yet commit ourselves to any
particular law of combination but merely say that, if A and B are two elements
of a group, we indicate that w ~ are combining them by simply writing AB or
BA. Now immediately the question arises if it makes any difference whether
we write A B or BA. In ordinary algebra it does not, and we say that multi-
plication is commutative, that is xy = yx, or 3 x 6 = 6 x 3. In group the-
ory, the commutative law does not in general hold. Thus AB may give C
while BA m ay give D, where C and D are two more elements in the gro u p.
6
DEFINITIONS ANO THEOREfviS OF GROUP THEORY 7
There are sorne groups, however, in which combination is commutative, and
such groups are called Abelian groups. Because of the fact that multiplication
is not in general commutative, it is sometimes convenient to have a means
of stating whether an element B is to be multiplied by A in the sen se A B or
BA. In the first case we can say that B is left-ntultiplied by A, and in the
second case that B is right-ntultiplied by A.
2. One element in the group must conznzute with al/ others and leave them
unchanged. It is customary to designate this element with the letter E, and
it is usually called the identity elenzent. Symbolically we define it by writing
EX= XE =X.
3. The associative law of tnultiplication ntust hold. This is expressed in the
following equality:
A(BC) = (AB)C
In plain words, we may combine B with C in the order BC and then combine
this product., S, with A in the order AS, or we may combine A with B in the
order AB, obtaining a product, say R, which we then combine with C in the
order RC and get the same final product either way. In general, of course,
the associative property must hold for the continued product of any number
of elements, namely,
(AB)(CD)(EF)(GH) = A(BC)(DE)(FG)H = (AB)C(DE)(FG)H
4. Every element ntust ltave a reciproca/, which is also an element of tite
group. The element R is the reciprocal of the element S if = SR = E,
where E is the identity. Obviously, if R is the reciproca[ of S, then S is the
reciproca! of R. Also, E is its own reciproca!.
At this point we shall provea small theorem reciprocals which
will be of use later. The rule is
The reciproca! of a produGt of two or more eletnents is equal to tite product
of tite reciprocals, in reverse order.
This means that
(ABC XY)-
1
= y-rx- c-rB-
1
A -t
PROOF. For sirnplicity we shall prove this for a temary product, but it will be
obvious that it is true generally. If A, B, and C are group elements, their
product, say D, must also be a group element, namely,
ABC = D
8 PRINCIPLES
If no\v we right-multiply each side of this equation by c-
1
B -rA -, we obtain
ABCc-B-
1
A -t = DC-
1
8-'A-
1
ABEB-
1
A _, = DC-
1
B-
1
A _,
Sin ce D times c-
1
B _,A_, = E, c-
1
B -
1
A - is the reciproca) of D, and since
D = ABC, we have
which proves the above rule.
2.2 SOME EXAMPLES OF GROUPS
Most of our attention in this book, until we reach Chapter 11, will be con-
centrated on a type of symmetry group called a point group. The significance
of these terms, "symmetry group" and "point group. '' need not detain us
here (see Chapter 3). Mosto( them contain a finite number of elements, but
two (to which linear molecules belong) are infinite. The number of elements
in a finite group is called its order, and the conventional symbol for the order
is h. To illustrate the above defining rules, we may consider an infinite group
and then sorne finite groups.
As an infinite group we may take all of the integers, both positive and
negative, and zero. If we take as our law of combination the ordinary algebraic
process of addition, then rule 1 is satisfied. Clearly, any integer may be
obtained by adding two others. Note that we have an Abelian group since
the order of addition is immaterial. The identity of the group is O, since O +
11 = 11 + O = 11. Also, the associative Jaw of combination holds, since .. for
example, [(+3) + (-7)] + {+1043) = (+3) + [(-7) + (+1043)]. The
reciproca} of any element, n., is ( -n) .. since ( +n) + ( -n) = O.
Group Multiplication Tables
If we have a complete and nonredundant list of the h elements of a finite
group and we know what all of the possible products ( there are lt
2
) are, then
the group is completely and uniquely defined-at Jeast in an abstract sense.
The foregoing information can be presented most conveniently in the form
of the group multiplication table. This consists of h rows and h columns. Each
column is labeled with a group element, and sois each row. The entry in the
table under a given column and along a given row is the product of the
elements which head that column and that row. Because multiplication is in
DEFINITIONS AND THEOREMS OF GROUP THEORY 9
general not commutative, \Ve must ha ve an agreed u pon and consistent rule
for the order of multiplication. Arbitrarily, we shall take the factors in the
order: (column element) x (row element). Thus at the intersection of the
column labeled by X and the row labeled by Y we find the element, which
is the product XY.
We now prove an important theorem about group multiplication tables,
cal1ed the rearrangentent theorenz.
Each row and each column in the group 1nultiplicarion table lists each of
the group elentents once and only once. Front this, it follol'l'S that no two rows
ntay be identical nor ntay any t1110 colun1ns be identical. Thus each row and
each colunuz is a rearranged list of the group elen1ents.
PROOF. Let the group consist of the h elements E, A
1
, A
3
, , A
11
The
elements in a given row. say the nth row, are
Since no two group elements, A; and A for instance, are the same, no two
products, A;An and AAn, can be the same. The h entries in the nth row are
all different. Since there are only lt group elements, each of them must be
present once and onJy once. The argument can obviously be adapted to the
columns.
Groups of Orders 1, 2, and 3
Let us now systematically examine the possible abstract groups of low order,
using their multiplication tables to define them. There is .. of course, formally
a group of order 1, which consists of the identity element alone. There is only
one possible group of order 2. lt has the following muJtiplication table and
will be designated G
2

G2 E A
E E A
A A E
For a group of order 3, the multiplication table will have to be, in part,
as follows:
E A B
E E A B
A A
B B
10 PRlNCIPLES
There is then only one way to complete ~ h table. Either AA = B or AA =
E. If AA = E, then BB = E and we would augment the table to give
E A B
E E A B
A A E
B B E
But then we can get no further, since we would ha ve to accept BA = A and
AB = A in order to complete the last column and the last row, respectively,
thus repeating A in both the second column and the second row. The alter-
native, AA = B, leads unambiguously to the following table:
E E A B
A A B E
B B E A
Cyclic Groups
G
3
is the simplest, non trivial member of an important set of groups, the cyclic
groups. We note that AA = B, while A B ( = AAA) = E. Thus we can consider
the entire group to be generated by taking the element A and its powers,
A
2
( = B) and A
3
( =E). In general, the cyclic group of order h is defined as
an element X and all of its lt powers up to Xh = E. We shall presently examine
severa) other cyclic groups. An important property of cyclic groups is that
they are Abelian, that is, all multiplications are comml:ltative. This must be
SO, since the various group elements are all of the form Xn, Xm, and SO 00,
and, clearly, X" X"' = X'" X'' for all nz and n.
Groups of Order 4
To continue, we ask how many groups of order 4 there are and what their
multiplication table(s) will be. Obviously, there will be a cyclic group of order
4. Let us employ the relations.
X= A
X"!.= B
X
3
=e
x.s =E
From this we find that the multiplication table, in the usual fonnat, is as
follows:
DEFINITIONS ANO THEOREMS OF GROUP THEORY 11
G ~ l l
E A B
e
E E A B e
A A B e E
B B e E A
e e E A B
That there is a second type of G
4
group, Gfl .. is fairly obvious .. We note
that for Gi
1
> only one element, namely B, is its own inverse. Suppose, instead,
we assume that each of two elements, A and B, is its own in verse. We shall
then have no choice but to a1so make e its own inverse, since each of the
four E's in the table must lie in a different row and column, Thus, we would
obtain
E A B e
E E A B e
A A E
B B E
e e E
A momenfs consideration will show that there is only one way to complete
this table:
G ~ : u
E A B e
E E A B e
A A E e B
B B e E A
e e B A E
lt is also clear that there are no other possibilities. * Thus, there are two
groups of arder 4, namely G ~
1
and G1
2
), which may be considered to be
defined by their multiplication tables.
Groups of Orders 5 and 6
It is left as an exercise (Exercise 2.2) to show that there is only one group of
order 5. Similarly, a systematic examination of the possibilities for groups of
lf we make up a table in which only one element ( other thnn E) is its own inverse and let thnt
element be A or C instead of B as in the G u table given. we are 1101 inventing a different G"'.
We are only permuting the arbitrary symbols for the group elements.
12 PRfNCIPLES
order 6 is also left asan exercise (Exercise 2. 9). To provide illustrative material
for several tapies that we shall take up next9 the multiplication table for one
of the groups of arder 6 is given.
Gf
1
E A B e D F
E E A B e D F
A A E D F B e
B B F E D e A
e e D F E A B
D D e A B F E
F F B e A E D
2.3 SUBGROUPS
Inspection of the multiplication table for the group
will show .that within

this group of arder 6 there are smaller groups. The identity E in itself is a
group of arder l. This will, of course, be true in any group and "is trivial. Of
a non trivial nature are the groups of order 2 .. namely. E, A; E .. B; E, e; and
the group of arder 3 .. namely, E, D, F. The last should be recognized also
as the cyclic group G
3
, since D
1
= F, D
3
= DF = FD = E. But to retum
to the main point, smaJier groups that may be found within a larger group
are called subgroups. There are, of course, groups that have no subgroups
other than the trivial one of E itself.
Let us now consider whether there are any restrictions on the nature of
subgroups, restrictions that are logicaJ consequences of the general definition
of a group and not of any additional or special characteristics of a particular
group. We may note that the orders of the group C1
21
and its subgroups are
6 and 1, 2, 3; in short, the orders of the subgroups are all factors of the arder
of the main group. shall now prove the following theorem:
Tlze order of any subgroup g of a group of order h 1nust be a divisor of h.
In other ltl g = k where k is an integer.
PROOF. Suppose that the set of g elements, A., A
2
, A
3
, , AR, forms a
subgroup. Now let us take another element B in the group which is not a
member of this subgroup and forro all of the g products: BA
1
, BA
2
, ,
BAg. No one of these products can be in the subgroup. If, for example,
DEFINITIONS AND THEOREMS OF GROUP THEORY 13
then, if we take the reciproca( of A
2
, perhaps A
5
, and right-multiply the above
equality., we obtain
BA
2
A
5
= A-4A5
BE= A-4A
5
B = A-4As
But this contradicts our assumption that B is not a member of the subgroup
A
1
A
1
, , Ag, since A-4As can only be one of the A;. Hence, if all the
products BA; are in the large group in addition to the A; themselves, there
are at least 2g members of the gro u p. If h > 2g" we can choose still another
element of the group, namely e, which is neither one of the A; nor one of
the BA;, and on multiplying the A; by e we will obtain g more elements, all
members of the main group, but none members of the A; or of the BA; sets.
Thus we now know that h must be at least equal to 3g. Eventually. however.,
we must reach the point where there are no more elements by which we can
multiply the A; that are not among the sets A;, BA;, CA;. and so forth, airead y
obtained. Suppose after having found k such elements, we reach the point
where there are no more. Then h = kg, where k is an integer, and hlg =
k., which is what we set out to prove.
Although we ha ve shown that the order of any subgroup, g, must be a
divisor of h, we have not proved the converse, namely, that there are subgroups
of all orders that are divisors of h, and, indeed, this is not iri general true.
Moreover, as our illustrative group proves, there can be more than one
subgrou p of a given order.
2.4 CLASSES
We have seen that in a given group it may be possible to select various smaller
sets of elements, each such set including E, however, \\{hich are in themselves
groups. There is another way in which the elements of a group may be
.separated into smaller sets, and such sets are called classes. Before defining
a class \Ve must consideran operation known as sinlilarity transfornzation.
If A and X are two elements of a group, then x-
1
AX will be equal to
sorne element of the group, say B. We have
B = x-Ax
We express this relation in words by saying that B is the silnilarity transfonn
of A by X. We also say that A and B are con.jugate. The following properties
of conjugate elements are important.
14 PRlNCIPLES
(i) Every elentent is co1tjugate with itself. This means that if we choose
any particular element A it must be possible to find at Jeast one element X
such that
A= x-Ax
If we left-multiply by A - \Ve obtain
A -A = E= A -x-'AX = (XA)-
1
(AX)
which can hold only if A and X commute. Thus the element X may always
be E, and it may be any other element that commutes with the chosen element,
A.
(ii) If A is conjugate with B, then Bis conjugate vitlz A. This mearis that
if
A= x-'BX
then there must be sorne element Y in the group such that
B = y-AY
That this must be so is easily proved by carrying out appropriate multipli-
cations, namely.,
XAX-
1
= xx-Bxx- = B
Thus, if Y = x-r (and thus also y- = X), we have
B = y-AY
and this must be possible, since any element, say X, must have an inverse,
say Y.
(iii) Jf A is conjugate with B and C, then B and C are conjugate with each
other. The proof of this should be easy to work out from the foregoing
discussion and is left as an exercise.
We may now define a class of elements.
A co1nplete set of ele1nents that are conjugare to one anptlzer is called a class
of the group.
In order to determine the classes within any particular group we can begin
with one element and work out all of its transforms, using all the elements
in the group .. including itself, then take a second element, which is not one
of those found to be conjugate to the first, and determine all its transforms,
and so on until all elements in the group have been placed in one class or
another.
Let us illustrate this. pro ce dure with the group Gf). All of the results given
below may be verified by using the multiplication table. Let us start with E.
DEFINITIONS AND THEOREMS OF GROUP lHEORY 15
E-
1
EE = EEE = E
A -EA = A -AE = E
B-
1
EB = B-
1
BE = E
Thus E must constitute by itself a class, of order 1, sin ce it is not conjugate
with any other element. This will, of course, be true in any group. To continue,
E-
1
AE =A
A-
1
AA =A
B-
1
AB = C
c-'AC = B
D-
1
AD = B
F-
1
AF =e
Thus the elements A, B,_and e are all conjugate and are therefore members
of the same class. lt is left for the reader to show that all of the transforms
of B and C are either A, B, or C. Thus A, B, and C are in fact the only
members of the class.
Continuing we have
E-
1
DE = D
A-
1
DA = F
B-
1
DB = F
c-Dc = F
D-nn = D
F-
1
DF = D
It will also be found that every transform of F is either D or F. Hence, D
and F constitute a class of r ~ r 2.
It will be noted that the classes have orders 1, 2, and 3, which are all
factors of the. group order, 6. It can be proved; by a method similar to that
used in connection with the orders of subgroups, that the following theorem
is true:
Tlze orders of al/ classes 1nust be integral factors of tlze order of the group.
16 PRINCIPLES
We shall see later (Section 3.13) that in a symmetry group the classes have
useful geometrical significance.
EXERCISES
2.1. Prove that in any Abelian group, each element is in a class by itself.
2.2. Show that there can be only one group of order h. when h is a prime
number.
2.3. Write down the mu1tiplication table for the cyclic group of order 5.
Show by trial and error that no other one is possible.
2.4. Why can we not ha ve a group in which A
2
= 8"1 # E?
2.5. If we start \Vith the multiplication table for group G
3
and add another
element, C, \Vhich commutes with both A and B, what multiplication
table do we end up with?
2.6. Show that for any cyclic group, X, X
3
, X''( =E), there must
be one subgroup corresponding to each integral divisor of the order h.
Give an example.
2. 7. lnvent as many different noncyclic groups of order 8 as yo u can and
give the multiplication table for each.
2.8. For each of the groups of order 8, show how it breaks. down into
subgroups and classes.
2.9. Derive the multiplication table for aiJ other groups of order 6 besides
the one in the text. This will require you to show that a group
of order 6 in which every element is its own inverse is impossible.
2.10. For the groups
and the cyclic group of order 6, show what

classes and subgroups each one has.
3
MOLECULAR SYMMETRY AND
THE SYMMETRY GROUPS
3.1 GENERAL REMARKS
It is perhaps appropriate to begin this chapter by sketching what we intend
to do hereo lt is certainly intuitively obvious what we mean wlten we say that
sorne molecules are more symmetrical than others, or that sorne molecules
ha ve high symmetry whereas others ha ve low symmetry or no symmetry. But
in order to make the idea of molecular symmetry as use fu) as possible, we
must develop sorne rigid mathematical criteria of symmetry o To do this we
shall first consider the kinds of sy1nmetry ele1nents that a molecule may have
and the sym1netry operations by the symmetry elementso We shall
then sho\v that a complete but nonredundant set of symmetry operations (not
elements) constitutes a mathematical group. Finally, we shall use the general
properties of groups, developed in Chapter 2, to aid in correctly and system-
atically determining the symmetry operations of any molecule we may care
to consider. We shall also describe here the system of notation normaJly used
by chemists for the various symmetry groupso An altemative system used
primarily in crystallography is explained in Chapter 11.
It may aJso be worthwhile to offer the following advice to the student of
this chapter o The use of three-dimensional models is helpful in
leaming to recognize and visuaJize symmetry elements. Indeed, it is most
unlikely that any but a person of the most exceptional gifts in this direction
can fail to profit significantly from the examination of models. At the same
time, it may also be said that anyone with the intelligence to master other
aspects of modero chemical knowledge by the use of models, surely
succeed in acquiring a good working knowledge of molecular symmetry.
17
18 PRINCIPLES
3.2 SYMMETRY ELEMENTS AND OPERATIONS
The two things, symmetry elements and symmetry operations, are inextricably
reJated and therefore are easily confused by the beginner. They are, however,
di:fferent kzds of things, and it is important to grasp and retain, from the
outset, a clear understanding of the difference between them.
Definition of a Symmetry Operation
A symmetry operation is a movement of a body such that, after the movement
has been carried out, every point of the body is coincident with an equivalent
point ( or perhaps the same point) of the body in its original orientation. In
other words, if we note the position and orientation of a body befare and
after a movement is carried out, that movement is a symmetry operation if
these two positions and orientations are indistinguishable. This would mean
that, if we were to loo k at the body, tum away long enough for someone to
carry out a symmetry operation, and then look again, we would be completely
unable to tell whether or not the operation had actually been performed,
because in either case the position and orientation would be indistinguishable
from the original. One final way in which we can define a symmetry operation
is to say that its effect is to take the body into an equivale1zt configuration-
that is, one which is indistinguishable from the original, though not necessarily
identical with it.
Definition of a Symmetry Element
A symmetry element is a geometrical entity such as a line, aplane, ora point,
with respect to which one or more symmetry operations may be carried out.
Symmetry elements and symmetry operations are so closely interrelated
because the operation can be defined only with respect to the element, and
at the same time the existence of a symmetry element can be demonstrated
only by showing that the appropriate symmetry operations exist. Thus, since
the existence of the element is contingent on the existence of the operation(s)
and vice versa, we shall dscuss related types of elements and operations
together.
In treating molecular symmetry, only four types of symmetry elements and
operations need be considered. These, in the arder in which they. will be
discussed, are listed in Table 3.1.
3.3 SYMMETRY PLANES AND REFLECfiONS
A symmetry plane must pass through a body, that is., the plane cannot be
completely outside of the body. The conditions which must be fulfiUed in
order that a given plane be a symmetry plane can be stated as follows. Let
MOLECULAR SYMMETRY ANO THE SYMMETRY GROUPS 19
TABLE 3.1 The Four Kinds of Symmetry Elements and Operations Required in
Specifying Molecular Symmetry
Symmetry Element
1. Plane
2. Center. of symmetry or
center of inversion
3. Proper axis
4. Improper axis
Symmetry Operation(s)
Reftection in the plane
Inversion of all atoms. through the center
One or more rotations about the axis
One or more repetitions of the .sequence: rotation
folJowed by reftection in a plane .l to the rotation axis
us apply a Cartesian coordinate system to the molecule in such a way that
the pfane incJudes two of the axes (say X and y) and is therefore perpendicular
to the third (i.e., z). The position of every atom in the molecule may also be
specified in this same coordinate system. Suppose now, for each and every
atom. we leave the x and y coordinates fixed and change the sign of the z
coordinate: thus the ith t o m ~ originally at (x;, y;, z;), is moved to the point
(x;, y;, - z;). Another way of expressing the above operation is to say, "Let
us drop a perpendicular from each ~ t o m to the plane, extend that line an
equal distance on the opposite si de of the plane, and m ove the ato m to this
other end of the line." If, when such an operation is carried out on every
atom in a molecule, an equivalent configuration is obtained, the plane used
is a symmetry plane.
Clearly. atoms lying in the plan e constitute special cases, since the oper-
ation of reflecting through the plane does not move them at all. Consequently.
any planar molecule is bound to ha ve at Jeast one plan e of symmetry, namely,
its molecular plane. Another significant and immediate consequence of the
definition is a restriction on the numbers of various kinds of atoms in a
molecule having a plan e of symmetry. AJI atoms of a given species that do
not lie in the plane must occur in even numbers, since each one must have
a twin on the other side of the plane. Of course, any number of atoms of a
given species may be in the plane. Furthermore, if there is only one atom of
a given species in a molecule, it must be in each and every symmetry plane
that the molecule may have. This means that it must be on the line of
intersection between two or more planes or at the point of intersection of
three or more planes {if there is such a point), since this atom must lie in all
of the symmetry planes simultaneously.
The standard symbol for a plane of symmetry is a. The same symbol is
also used for the operation of reflecting through the plane.
l_t should be explicitly noted that the existence of one symmetry plane gives
rise to, requires. or, as usually stated, genera tes one symmetry operation. We
m ay al so note he re, for future use, that the effect of applying the same
reftection operation twice is tb bring all atoms back to their original positions.
Thus, while the operation a produces a configuration equivalentto the original,
20 PRINCIPLES
the application of the same a twice produces a configuration identical with
the original. Now we can conveniently denote the successive application of
the operation a 11 times by writing an. We can then also a
2
= E, where
we use the symbol E to represent any combination of operations which takes
the molecule to a configuration identical with the original one. We call E, or
any combination of operations equal to E, the identity operation. It should
be obvious that a" = E when n is even and an = a when n is odd.
Let us now consider sorne illustrative examples of symmetry planes in
molecules. At one extreme are molecules that have no symmetry planes at
all. One such general class consists of those which are not planar and which
ha ve odd numbers of all atoms. An example is FCISO, se en below.

t....
1 .......
c1-----
At the other extreme are molecules possessing an infinite number of sym-
metry planes, that is, linear molecules. For these any plane containing the
molecular axis is a symmetry plane, and there is obviously an infinite number
of these planes. Most small molecules fall between these extremes; that is,
they have one or a few symmetry planes. If, instead of FCISO, we take F
1
SO
or ChSO, we ha ve a molecule with one symmetry plane, which passes through
S and O and is perpendicular to the Cl, Cl .. O or the F, F, O plane. The H
2
0
molecule has two symmetry planes. One is, of course, coextensive with the
molecular plane. The otber includes the oxygen atom (it must, since there is
only one such ato m) and is perpendicular to the molecular plane. The effect
of reflection through this second plane is to leave the oxygen atom fixed but
to exchange the hydrogen atoms, while reflection through the first plane lea ves
all atoros unshifted. A tetrahedral molecule of the type AB
2
C
2
(e.g., CH
2
CI
2
)
also has tWo mutually perpendicular planes of symmetry. One contains AB
2
,
and reflection through it leaves these three atoms unshifted whle interchang-
ing the C atoms; the other contains AC
2
, and reflection through it interchanges
only the B atoms.
The molecules NH
3
and CHCI
3
are representative of a type containing
three pJanes of symmetry. For NH
3
, any plane of symmetry would ha ve to
incJude the nitrogen atom and either one or all three of the hydrogen atoms.
Snce NH
3
is not planar, there can be no symmetry plane including N and all
three H atoms; hence we loo k for planes including N and one H and bisecting
the line between the remaining two H atoms. There are clearly three such
planes. For CHCI
3
, the situation is quite analogous except that the hydrogen
atom must also Jie in the symmetry planes.
The NH
3
is only one example of the general class of pyramidal
AB
3
molecules. Let us see what happens as we begin flattening su eh a molecule
by pushing the A atom down toward the plane of the three B atoms. It should
MOLECULAR SYMMETRY ANO THE SYMMETRY GROUPS 21.
be easily seen that this does not disturb the three symmetry planes, even in
the limit of coplanarity. Nor does it introduce any new planes of symmetry
except in the limit of coplanarity. Once AB
3
becomes planar there is then a
fourth symmetry plane, which is the molecular plane. Molecules and ions of
the planar AB
3
type possessing four symmetry planes, three perpendicular to
the molecular plane, are fairly numerous and important. There are, for ex-
ample, t ~ boron halides, CQj-, N03, and S0
3

A plnar species of the type [PtCI.a]
2
- or [AuCI
4
]- possesses.five symmetry
planes. One is the molecular pjane. There are also two, perpendicular to the
molecular plane and perpend\cular to each other, which pass through three
atoms. Finally, there are two more, also perpendicular to the molecular plan e
and perpendicular to each other, which bisect CI-Pt-CI or Cl-. Au-CI
angles.
A regular tetrahedral molecule possesses six planes of symmetry. Using
the numbering system illustrated in Figure 3.1., we may specify these symmetry
planes by stating the atoms they contain:
A regular octahedron possesses, in aH, nine symmetry planes. Reference
will be made to the numbered figure on page 22 in specifying these. There
are first three of the same type, namely, those including the -following sets of
atoms: AB
1
B:!B
3
B.h AB
2
B
4
B
5
B
6
, and AB
1
B
3
B
5
B
6
There are then six more of
a second type, one of which indudes AB
5
B
6
and bisects the B
1
-B
2
and
Figure 3.1 A tetrahedral AB_. molecule.
22 PRINCIPLES
B
3
-a_. lines, a second which includes AB
1
B
3
and bisects the B ~ B
5
and B.c-B
6
lines, and so forth.
Bs
3.4 THE INVERSION CENTER
If a molecule can be brought into an equivalent configuration by changing
the coordinates (x, y, z) of every atom, where the origin of coordinates lies
at a point within the molecule, into (-x, -y, - z), then the point at which
the origin lies is said to be a center of symmetry or center of inversion. The
symbol for the inversion center and for the operation of inversion is an italic
i. Like a plane, the center is an element that generates only one operation.
It may be noted that, when a center of inversion exists, restrictions are
placed on the numbers of aJI atoms, or all but one atom, in the molecule.
Since the center is a point, only one atom may be at the center. If there is
an atom at the center, that atom is unique, since it is the only one in the
molecu)e that is not shifted when the inversion is performed. All other atoms
must occur in pairs, since each must have a twn with which it is exchanged
when the inversion is performed. From this it follows that we need not bother
to look for a center of symmetry in molecules that contain an odd number
of more than one species of atom.
The effect of carrying out the inversion operation n times m ay be expressed
as i". It should be easily seen that n = E when 11 is even, and in = i when
n is odd.
Sorne examples of moJecules having inversion centers are octahedral AB
6
,
planar AB
4
, planar and trans AB
2
C
2
, linear ABA, ethylene, and benzene.
Two examples of otherwise fairly symmetrical molecules that do not have
centers of inversion are CsHs (plane pentagon) and tetrahedral AB
4
(even
though A is at the "center" and B's come in even numbers).
3.5 PROPER AXES AND PROPER ROTATIONS
Befo re discussing proper axes and rotations in a general way, let us take a
specific case. A line drawn perpendicular to the plane of an equilateral triangle
and intersecting it at its geometric center is a proper axis of rotation for that
MOLECULAR SYMMETRY AND THE SYMMETRY GROUPS 23
triangle. U pon rotating the triangle by 120 (2n/3) about this axis, the triangle
is brought into an equivalent configuration. It may be noted that a rotation
by 240 (2 x 2n/3) also produces an equivalent configuration.
The general symbol for a proper axis of rotation is en, where the .subscript
n denotes the order of the axis. By order is meant the largest value of 11 such
that rotation through 2nln gives an equivalent configuration. In the above
example, the axis is a e
3
axis. Another way of defining the meaning- of the
order n of an axis is to say that it is the number of times that the smallest
rotation capable of giv.ing an equivalent configuration must be repeated in
order to give a configuration not merely equivalent to the original but also
identical to it. The meaning of "identical" can be amplified if we attach
num bers to ea eh apex of the triangle in our example. Then the effects of
rotating by 2n/3, 2 x 2rc/3, and 3 x 2n/3 are seen to be
27r/3
I n
2 x 2r/3
I
m
2
16.3
3 X 27r/3
I
IV=I
Configurations II and 111 are equivalent to I because without the labels
(which are not real, but represent only our mental constructions) they are
indistinguishable from 1, although with the labels they are distinguishable.
However, IV is indistinguishable fro"m 1 not only without la beis but also with
them. Hence, it is not merely equivalent; it is identical.
The e
3
axis is also caUed a threefold axis. Moreover.. we use the sym-
bol c3 to represent the operation of rotation by 2n/3 around the e3 axis.
For the rotation py 2 x 2n/3 we use the symbol Cj, and for the rotation by
3 X 21CI3 the symbol e ~ . Symbolically we can write e ~ = e:h and hence only
C
3
, e ~ and e ~ are separa te and distinct operations. However, e ~ produces
an identical configuration, and hence we may write C ~ = E.
After consideration of this example, it is easy to accept sorne more general
statements about proper axes and prOJler rotations. In general, an n-fold axis
is denoted by en and a rotation by 2n/ n is al so represented by the symbol
24 PRINCIPLES
en. Rotation by 2n/ n carried out successively m times is represented by the
symbol C':. Also., in any case, e: = E, e:+' = en., e:+
2
= e;., and so on.
In discussing planes of symmetry and inversion centers, attention was di-
rected to the fact that only one operation, refiection, is generated by a sym-
metry plane, and only one operation, inversion, by an inversion center. A
proper axis of order n, however, generates n operations, namely en, e;,
... , e:+l, e:(= E).
One last general consequence of the existence of a en axis cohcerns the
requirement that there be certain numbers of each species of atoro in a
molecule containing the axis. Naturally, any atom that lies on a proper axis
of symmetry is unshifted by any rotation about that axis. Thus there may be
any number., even or odd, of each species of atom lying on an axis (unless
other symmetry elements impose restrictions). However, if one atom of a
certain species lies off a en axis, there must automatically be 11 - 1 more, or
a total of n such atoms, since on applying en successively n times, the first
atom is moved to a total of n different points. Had there not been identical
atoms at all the other n - 1 points to begin with, the new configurations
would not be equivalent configurations; this would mean that the axis would
not be a Cn symmetry axis, contrary to the original assumption.
The symbol C:;' represents a rotation by m x 2nln. Let us consider the
operation e, which is one of those generated by a e
4
axis. This is a rotation
by 2 x 2n/4 = 2n/2, and can therefore be written justas well as e
2
Similarly,
among the operations generated by a e6 axis, we find .. and which
m ay be written, respectively., as e
3
.. C
2
, and Cj. lt is frequent though not
invariable practice to write an operation e: in what, by considering the
fraction (nzln) in (m/n}27r., can be called Jowest terms, and the reader should
be familiar with this practice so that he immediately recognizes, for example,
that the sequence C
6
, e
3
, e
2
, E is identical in meaning \Vith eb, Ci,
e
3 es eh
. 6' 6., 6' 6
Let us now consider sorne further illustrative examples chosen from com-
monly encountered types of molecules. Again we may begin by considering
extremes. Many molecules possess no axes of proper rotation; FCISO, for
example, does not. (Actually, FCISO is a gratuitous example, since as we
saw earlier, it possesses no symmetry elements whatsoever.) Neitner CI
2
SO
nor F
2
SO possess an axis of proper rotation. At the other extreme are linear
molecules which possess oo-fold axes of proper rotation, colinear with the
molecular axes. Since all a toms in a linear molecule lie on this axis .. rotation
by any angle whatever, and hence by all ( oo number) angles, lea ves a config-
uration indistinguishable from the original. Again .. as with planes of symmetry,
most small molecules possess one axis or a few axes, generally of low orders.
Among examples of molecules with a single axis of order 2 are HJO and
CH
2
Cl
2
No molecules possess just two twofold axes; this will be later
to be mathematically impossible. There are many examples of molecules
possessing three twofold axes, for example., ethylene (CJH.). One C
2
is col-
inear with the C-C axis. A second is perpendicular to the plane of the
Figure 3.2 A tetrahedral molecule inscribed in a cube.
molecule and bisects the c-e line. The third is perpendicular to the first
two and intersects both at the midpoint of the C-C line. A regular tetrahedral
molecule also possesses three twofold axes, as shown in Figure 3.2.
Threefold axes are quite common. Both pyramidal and planar AB
3
mol-
ecules possess threefold proper axes passing through the atom A and per-
pendicular to the plane of the three B atoms. A tetrahedral molecule, AB.h
possesses four threefold axes, each passing through the atom A and one of
the B atoms. An octahedral molecule, AB
6
, also possesses four threefold
axes, each passing through the centers of two opposite triangular faces and
the A atom.
The planar AB
3
molecule possesses three twofold axes perpendicular to
the threefold axis as shown in the diagram. The existence of the C
3
axis and
............ _________ _
--
--
--
--
--
--
--
--
26 PRINCIPLES
one C
2
axis perpendicular to the C
3
axis means that the other two C
2
axes"
at angles of 2n/3 and 4n/3 to the first, tnust exist. For, on carrying out the
rotation C
3
, we generate the second C
2
axis from the first, and on carrying
out the rotaton Cj, we generate the third C
2
axis from the first.
The effect of the operations en, ... , e:- in replicating other symmetry
elements may p_rofitably be discussed more ful1y at this point. The other
symmetry elements of interest are planes and axes. It will also be sufficient
to limit the discussion to axes perpendicular to the axis of the replicating
rotations and to planes that contain the axis of the replicating rotations.
A plane perpendicular to the axis of the replicating rotations is obviously
not replicated, since all rotations carry it into itself. Although a completely
general discussion might be given, it seems more instructive to consider
separately each of the replicating axes that can be encountered in practice
(Cn, 1 < n 8).
An axis perpendicular to a C
2
axis or a plane containing a C
2
goes into
itself on carrying out the operation C
2
; hence no further axes or planes of the
same type are required to exist in this case. We have just seen that from one
axis perpendicular to a C
3
axis two similar ones are generated. The same is
true for a plane of symmetry containing a C
3
axis. We may also deal with the
es and C
7
cases (and, indeed, any C, where 11 is odd) for they all behave in
the same way. One axis perpendicular toa C
5
or C
7
axis or one plane containing
a C
5
or C
7
will be made to generate four or six more separate and distinct
axes or planes by the operations that the C
5
or C
7
axis makes possible.
For cases where n in Cn is even, the results are less straightforward. Suppose
that we have one axis C
2
(1) perpendicular to a C
4
axis. On carrying out the
rotation C
4
, C
2
(1) is rotated by 2n/4 and a second C
2
axis, C
2
(2), is thus
produced. On carrying out the rotation Cl( = C
2
) about the C
4
axis, however,
C
2
(1) merely goes into itself, and C
2
(2) also goes into itself. Tbe operation
C3 takes C
2
(1) into C
2
(2} and e
2
(2) into C
2
(1). Hence, because C1 is reaUy
c2 and is only c4 followed by el, the c .. axis requires only that the
axis e
2
(1) be accompanied by one other such axis and not three others.
A completeiy argument holds regarding planes. In the C
6
case,
using the same line of argument, it will easily be seen that, if one axis per-
pendicular to a C
6
or one plane containing a C
6
exsts, it must be accompanied
by two more of the same type. Similarly, a C
8
axis will replicate a C
2
axis
perpendicular to it so as to produce a set of four such C
2
axes.
Continuing with examples of proper axes in typical molecules, we may cite
the planar PtCii- ion, which has a C
4
axis perpendicular to the plan e of the
ion and four c2 axes in the plane of the ion. The cyclopentadienyl anion,
CsH5, possesses a C
5
axis perpendicular to the molecular plane and five C
2
axes in the molecular plane. Benzene possesses a C
6
axis and two sets of three
C
2
axes. Probably the only known example of a molecule with a C
7
axis is
the planar [C
7
H
7
]+, the tropylium ion. An example of a molecule with a C
8
axis is (CaH
8
)
2
U {uranocene).
3.6 IMPROPER AXES ANO IMPROPER ROTATIONS
An improper rotation may be thought of as taking place in two steps: first a
proper rotation and then a reflection through a plane perpendicular to the
rotation axis. The axis about which this occurs is called an axis of improper
rotation or, more briefly, an ilnproper axis, and is denoted by the symbol S,,
where again 11 indicates the order. The operation of improper rotation by
2n/n is also denoted by the symbol Sn. Obviously, if an axis C, anda perpendic-
ular plane exist independently., then Sn exists. More important, however, is
that an sn may exist when neither the e, nor the perpendicular u exist sep-
arately.
Perhaps this can best be emphasized by taking an example. Let us consider
ethane in its staggered configuration. The C-C line defines a C
3
axis, but
certainly not_ a C
6
axis. Yet there is an S
6
, as the diagram shows. Observe
that 11 and III are equivalent to each other but that neither is equivalent to
1; that is, neither u nor C
6
is by itself a symmetry operation. But the com-
bination of both, in either order, which we call S
6
, is a symmetry operation
since it produces IV, which is equivalent to l.
1 ~
6""i/4
32
~ 6
3%2
m
5 ~ 4
3+1
6 4
5
~
3""!7'1
~
2
1
52l.4
IV
n
lt will be shown later that the operations of rotation and reflection in a
plan e perpendicular to the rotation axis always give the same result regardless
of the order in which they are performed. Thus the definition of improper
rotation need not specify the order. _
As another important example of the occurrence of improper axes and
rotations, let us consider a regular tetrahedral molecule. We have already
noted in Section 3.5 that the tetrahedron possesses three C
2
axes. Now each
of these C
2
axes is simultaneously an S
4
axis, as can be seen in the diagram
on page 28.
The element Sn in general genera tes a set of operations Sn, s;, S ~ ....
However, sorne important features of these operations should be noted. There
are differences in the sets generated for even and odd n, so these two cases
28 PRINCIPLES
will be considered separately. Let us assume that the Sn axis is colinear with
the z axis of a coordina te system and that the plane to \Vhich the reflection
part of the operation S, is referred is the xy plane.
An improper axis, Sn, of even order generates a set of operations S,, s;,
... , s:. Let us first that (for n even) s: = E. s: means that we
carry out the operations e,,, a, en, a, ... until, in all, Cn and a have each
been carried out n times. Sin ce n is an even number, tl repetitions of a is an
identity operation, so that s: = e:; but e: is also just E. TheFefore, s: =
E, and S:+
1
= Sn; s:+
2
= s;, and so on. Now by the same argument will
be equal to C:;' whenever nz is even. Thus, in any set of operations generated
by an even-order Sn, certain of the S:' may be written in other ways. Consider,
for example, the set S
6
, s:, The operation S
6
can be written in
no other way. = = e
3
= S
2
= i. = ej. The operation can
be written in no other way. s: = E. Hence, the complete set of operations
generated by the element S
6
can, and normally would, be written: S
6
, e
3
., i,
Ci, St E. Having written the set in this way, however, we can readi1y make
another useful observation. The set contains e
3
, ej, and E, whiGh are just
the operations generated by a e
3
axis. Hence the existence of the S
6
axis
automatically requires that the e
3
axis exist. It should not be difficult to see
that, in general, the existence of an Sn axis of even order always requires the
existence of a Cn
12
axis.
Let us now turn to mproper axes of odd order. Their most important
property is that an odd-order Sn requires that en and a a perpendicular to it
must exist independently. This is easily proved. The eJement Sn generates
operations Sn, s;, S!, .... Let us examine the operation s: when 11 is
odd. It must have the same effect as will application of e: followed by un =
u. But since e: = E, we see that s: = (J. In other words, the element Sn
generates a symmetry operation u. But if the symmetry operation u exists,
the plane to which it is referred must be a symmetry element in its own right.
Now, the operation Sn requires us to reflect in the plan e u, thus carrying a
configuration 1 into another configuration, II, and then to rotate by 2n/n,
thus carrying 11 into IIL Because S, is a symmetry operation, 1 and 111 must
be equivalent configurations. However, when n is odd, u is itself a symmetry
operation, so that 11 is also equivalent to l. Then II is also equivalent to III,
and we see that rotation by 2nln has carried 11 into an equivalent configu-
ration, Ill. Thus the operation Cn is also a symmetry operation in its own
right.
ro gain further familiarity with odd-order improper axes, let us consider
how many distinct operations are generated by sorne such axis, say S
5
The
sequence begins S
5
., S3, S ~ S ~ .... Using relations and conventions previously
developed, we can write certain of these operations in altemative ways, as
follows:
Ss - Cs then u ( or u then e
5
)
S ~ - e ~
S ~ - e ~
then u
Si
)
- C4
- S
S ~
) =a
/
S ~ - Cs
S?.
) = e ~ then u
S ~
- C3
- S
S ~ = e ~ then u
S!O
=E
su
5
= Cs then u
We see that for S
5
through S!
0
(in general, Sn through S;:a), the operations
are all different ones., but commencing with S!n+
1
repetition of the sequence
begins. Of the 10 operations, however, 4 plus E can be expressed as a single
operation only by using symbols Sg, whereas the other 5 can be written either
as Cg or as u. Thus there are operations which, although they may be ac-
complished by using Cg and a successively, cannot be represented as unit
operations in ariy other way than Sg. We also see that in general the element
sn with ll odd generates 2n operations.
3.7 PRODUCTS OF SYMMETRY OPERATIONS
In Sections 3.3-3.6 we have often discussed the question of how we can
represent the net effect of applying one symmetry operation after another to
a molecule, but only in a Jimited way. In this section we shall discuss this
question with regard toa broader range of possibilities. First, we shall establish
a conventional shorthand for stating that "operation X is carried out first and
then operation l'-, giving the same net effect as would the carrying out of the
single operation Z., This we express symbolically as
YX= Z
Note that the order in which the operations are applied is the order in which
they are written from right to left, that is, Y X means X first and then Y. In
general, the order makes a difference although there are cases where it does
not. When the result of the sequence XY is the same as the result of the
30 PRINCIPLES
sequence YX, the two operations, X and Y, are said to co1n1nute. It is also
normal to speak of an operation that produces the same result as does the
successive application of two or more others as the product of the others.
One way in \Vhich we may approach the problem of finding a single op-
eration which is the product of two others is to consider a general point with
coordinates [x., y
1
, z
1
]. On applying a certain operation, this point will be
shifted toa new position with coordinates [x
2
, y
2
, z
2
]; if stiiJ another operation
is applied, it will again be shifted so that its coordinates are now (x
3
, y
3
, z
3
].
The net effect of applying the two operations successively is to shift the point
from [x., y., z
1
] to [x
3
, y
3
, z
3
]. We now look for a way of accomplishing this
in one step. The operation which does so will be the product of the first two.
Let us iUustrate this proce_dure by proving the statement made earlier that,
if there are two twofold axes at right angles to one another, there must
necessarily be a third at right angles to both. Suppose that the two given axes
coincide with the x and y axes; we can designate them C
2
(x) and C
2
(y). On
applying first C
2
(x) and then C
2
(y) to a general point, the following trans-
formations of its coordinates take place:
That is , the value of x
3
is - x., the value of y
3
is -y, and the value of z
3
is
z
1
If now we apply C
2
(z) to the general point, it is shifted to [ -x, -y., z
1
].
Thus we may write
Thus, whenever C
2
(x) and C
2
(y) exist, C
2
{z) must also exist, because it is
their product.
As a second example of how the existence of two symmetry elements may
a:utomatically require that a third one exist; we shaU considera case having
a C
4
axis and one plane containing this axis. We have already seen that the
operation C
4
will generate a second plane from the first one at right angles
to it. lt is also true, however' though less obvious, that when the c4 axis and
one of these planes exist there must then be a second plane also containing
C
4
at an angle of 45 to the first one. We can p r v ~ this by the method just
used. The effect of reflecting a general point [x, Yb z
1
] through the xz plane
is given by
whereas the effect of a clockwise c4 rotation about the z axis upon the point
is given by
From these relations we can determine the effect of applying successively
u(xz) and then C
4
(z), namely,
Now let us consider the effect of reflecting the point through aplane CT, which
also contains the z axis and bisects the angles between the +y and - x axes
and the + x and -y axes. This transformation is
We see that
which means that the existence of C
4
(z) and u(xz) automatically requires that
Ud exist. The C4 rotation then generateS from UJ another plane, Ud, which
passes through the first and third quadrants. The final result is that, if there
is one plane containing a C
4
axis, there is automatically a set of four planes.
It may be shown in a very similar way that if C
4
(z) and C
2
(y) axes exista
C
2
axis lying in the first and third quadrants of the xy plan e at 45 to C
2
( y)
must also exist. This is left as an exercise.
Examination of the shifts in a general point m ay also be employed to show
a commutative relation, for example, that C
2
(z) and u(xy) commute. Thus,
we may write, in a notation that uses x instead of - x, y instead of -y, and
z instead of - z:
and
C
2
{z)(x, y, z]-+ [x, y, z]
u(xy)[x, y, z]-+ [x, y, z]
C
2
u[x, y, z]-+ C
2
[x, y, z]-+ [x, y, z]
uC2[x, y, z]-+ u[x, y, z]-+ [:T, y, z]
We see also that the product in each case is equivalent to i.
In these examples, where only C
2
and C
4
rotations and certain kinds of
planes are concemed, the transformation of the coordinates [x, y, z] to
[x, y, z] by a twofold rotation about the x axis, for example, is fairly obvious
by inspection. lt is also obvious that a fourfold rotation about the x axis will
transform the coordinates into [x, z, y]. lt is also easy to see by inspection
the effects of the inversion operation, an improper rotation by 2rc/2 or 2n/4
aqd reflection in a plane t ~ t is the xy, xz, or yz plane or a plan e rotated by
32 PRINCIPLES
45 from these. However, the transformations effected by more general sym-
metry operations, such as rotation by 2n/n or m2rcl n and reflections in planes
other than those just mentioned, are not easily handled by the simple methods
and notation used above. Further discussion along this line will therefore
employ sorne geometrical methods and also the more powerful methods of
matrix algebra, which will be introduced in Chapter 4.
3.8 EQUIVALENT SYMMETRY ELEMENTS AND
EQUIVALENT ATOMS
If a symmetry element A is carried into the element B by an operation
generated by a third element X, then of course B can be carried back into
A by the application of x-
1
The two elements A and B are said to be
equivalent. If A can be carried into still a third element e, then there will
also be a way of carrying B into e, and the three .elements, A, B, and e,
form an equivalent set. In general, any set of symmetry elements chosen so
that any member can be transformed into each and every other member of
the set by application of sorne symmetry operation is said to be a set of
equivalent symmetry elements.
Por example, in a plan e triangular molecule such as BF
3
, ea eh of the twofold
symmetry axes lying in the plane can be carried into coincidence with each
of the others by rotations of 2n/3 or 2 x 2n/3, which are symmetry operations.
Thus all three twofold axes are said to be equivalent to one another. In a
square planar AB
4
molecule, there are four twofold axes in the molecular
plane. Two of them, C
2
and C ~ lie a long BAB axes, and the other two., ej
and e;', bisect BAB angles. Such a molecule also contains four symmetry
planes, each of which is perpendicular to the molecular plane and ntersects
it along one of the twofold axes. Now t is easy to see that e
2
may be carried
into C ~ and vice versa, and that e ~ may be carried into e'i' and vice versa ..
by rotations about the fourfold axis and by reflections in the symmetry planes
mentioned, but there is no way to carry e
2
or ei into eithere; ore;' or vice
versa. Thus e
1
and C2 form one set of equivalent axes, and e ~ and C2' form
another. Similar} y, two of the symmetry planes are equivalent to each other,
but not to either of the other two., which are, however, equivalent to each
other.
As other illustrations of equivalence and nonequivalence of symmetry ele-
ments, we may note that all three of the symmetry planes in BF
3
that are
perpendicular to the molecular planeare equivalent, as are the three in NH
3
,
whereas the two planes in H
2
0 are not equivalent. The six twofold axes lying
in the plane of the benzene molecule can be divided into two sets of equivalent
a.Xes, one set containing those that transect opposite carbon atoms and the
other set containing those that bisect opposite edges of the hexagon.
Equivalent atonzs in a nzolecu/e are those that may all be interchanged '"'ith
one another by syntmetry operations. Naturally .. equivalent atoms musf be of
the same chemical species. Examples of equivalent atoms include all of the
hydrogen atoms in methane, ethane, benzene, or cyclopropane, all of the
fluorine atoms in SF
6
,' and all of the carbon and oxygen atoms in Cr(CO)b
Examples of chemically identical atoms which are not equivalent in molecular
environment are the apical and equatorial fluorine atoms in PF
5
; no symmetry
operation possible for this molecule ever interchanges these fluorine atoms.
The a and P hydrogen and carbon atoms of naphthalene are not equivalent.
All six carbon atoms of cyclohexane are equivalent in the chair configuration,
but four are different from the other two in the boat configuration.
3.9 GENERAL RELATIONS AMONG SYMMETRY ELEMENTS
AND OPERATIONS
We present here sorne very general and useful rules about how different kinds
of symmetry elements and operations are related. These deal with the way
in which the existence of ~ o m two symmetry elements necessitates the ex-
istence of others. and with commutation relationships. Sorne of the statements
are presented without proof; the reader should profit by making the effort to
verify them.
Products
l. The product of two proper rotations must be a proper rotation. Thus,
although rotations can be created by combining reflections (see rule 2), the
reverse is not possible. The special case C
2
(x)C
2
(y) = C
2
(z) has already been
examined (page 30).
2. The product of two reflections., in planes A and B, intersecting at an
angle cP.
48
., is a rotation by 2cfJAB about the axis defined by the line of inter-
section. The simplest proof of this is a geometric one, as indicated in Figure
3.3. It is clear that this rule has sorne far-reaching consequences. If the two
planes are separated by the angle ePAs, a Cn axis, where n = 2n/2cfJAs, is
required to exist. Here n must be an integer, and the Cn axis will then assure
that a total of n such planes exists. Thus., the two planes imply that the entire
set of operations constituting the C,v group (see below) is present.
3. When there is a rotation ruds, Cn, and aplane containing it, there must
be n such planes separated by angles of 2n/2n. This follows from rule 2.
4. The product of two C
2
rotations about axes that intersect at an angle {)
is a rotation by 20 about an axis perpendicular to the plane of the C2 axes.
This can be proved geometrically by a diagram similar to Figure 3.3. It also
implies that a en axis and one perpendicular c2 axis require the existence of
a set of n C
2
axes and thus generate what we shall soon recognize as the Dn
group of operations.
34 PRINCIPLES
Figure 3.3 A geometric proof that the two reHection planes A and 8 require the
existence of a en axis along their line of intersection with ll = cP:t.ll = a +
p, X = a + a + p + P = 2( a + /1). :. X =
A proper rotation axis of even order and a perpendicular reflection
plane generate an inversion center, that is = = C
2
a = aC
2
= i.
S
. 1 en . en e . e
lffil ar y 2nl = l ln = 'J.l = l 2 = u.
Commutation
The following pairs of operations always commute:
l. Two rotations about the same axis.
2. Reflections through planes perpendicular to ea eh other.
3. The inversion and any reflection or rotation.
4. Two C
2
rotations about perpendicular axes.
5. Rotation and reflection in a plane perpendicular to the rotation axis.
3.10 SYMMETRY ELEMENfS AND OPTICAL ISOMERISM
Although we have follo\ved conventional practice-and for general purposes
will continue to do so-in setting out four kinds of symmetry elements and
operations, u, i, Cn, and Sn, we should note that the list can in principie be
reduced to only two Cn and Sn. A re:Hection operation can be. regarded as an
SJ operation. that is, the (trivial) rotation by 2nl1 to
0
ethcr witb rcfiC(;tiOll.
The 0\lta\\O\\ sl \he \o\\ow\ng e\lect on a genera\ point, X, y, z. We
MOLECULAR AND THE SYMMETRY GROUPS 35
suppose that the axis coincides with the z axis of a Cartesian coordinate
system; the reftection component then takes place through the xy plane:
S
2
(x, y, z) = uC
2
(x, y, z) u(x, y, z) --7 (x, y, z)
But it is also true, by definition, that
i(x, y, z)--+ (x, y, z)
Thus S
2
and i are merely two symbols for the same thing.
All symmetry operations that we wish to consider can be regarded as either
proper or improper rotations.
Molecules that are not superimposable on their mirror images are termed
dissyn1metric. This term is used rather than asymmetric, since the latter means,
litera)) y, have no symmetry; that is, it is applicable only to a molecule be-
Jonging to point group e,. Al/ asynnzetric 11lO!ecules are dissynznzetric, but the
converse is not true. Dissygtmetric molecules can and often do possess sorne
symmetry. lt is possible to give a very simple, compact rule expressing the
relation between molecular symmetry and dissymmetric character:
A Molecule That Has No bnproper Rotation Axis Must Be DissyJnJnetric.
Since improper rotation axes include S
1
= a, and S
2
= i, the more
(but incomplete!) statement about optical isomerism existing in molecules that
lack aplane or center of symmetry is subsumed in this more general one. In
this connection, the tetramethylcyclooctatetraene molecule (page 37) should
be examined more closely. This molecule possesses neither a center of sym-
metry nor any plan e of symmetry. It does ha ve !" S
4
axis, and inspection will
show that it is superimposable on its mirror image.
To show the validity of this rule, we first prove that if an Sn axis does exist
the molecule cannot be dissymmetric; that is, it must have a superimposable
mirror image.
l. A molecule has one and only one mirror image. It makes no difference
where or in what orientation we place the mirror plane; we may place and
orient it wherever convenient. We may therefore aiJow it to pass through the
molecule4
2. If the molecule has an Sn axis, we may place the plane so it coincides
with the plane through which the reflectional part of an sn operation takes
place. If the Sn axis is of odd arder, the pure reflection operation (S
1
or s:)
will actually exist as a symmetry operation. The molecule is then obviously
superimposable on its mirror image.
3. Iftne mpro11er axis is of even arder and ah does not exist independent\)',
refiection wiU g\ve a figure that is not superlmposed on the origina\ but needs
36 PRINCIPLES
(a)
2
(e)
B' A'
/ /
C' ~ D ~
1 .
, ~ - - ~ e
A"' B
3
Figure 3.4 The labeled cube 1 has as its only symmetry element an inversion center.
In step (a) the inversion operation is carried out on 1. In step (b) we rota te by x about
the z axis and thus orient it in space so it can be directly superimposed on 3. The
labeled cube 3 is the mirror image of 1 obtained by reflection, step (e), in a plane
perpendicular to z.
only to be rotated by 2n/n in arder to come into coincidence. This rotation
of the molecule as a whole does not change its structure, and thus the molecule
and its mirror image are superimposable.
If a molecule is superimposable on its mirror image, that mirror image will
be produced by applying one of the symmetry operations in the point group
for the molecule. This is obvious for all point groups having an Sn (n odd
axis), sin ce these ha ve the u operation as su eh.
For the speciafcase of S
2
= i, the mirror image is produced by the inversio
operation, but must be rotated by 180 to bring it into an exact reflective
relationship to the original. This can be seen in Figure 3.4 and is conveniently
expressed by using the matrices for the coordinate transformations. (Readers
unfamiliar with matrix algebra may consult Appendix 1.) Thus, we represent
the operation S
2
= i by the first matrix shown below and a rotation by 1r
around the z axis by the second one. The product of these is the third matrix
and this clearly describes the process of reflection in a mirror plane perpen-
dicular to z. Since the i operation gives a molecule indistinguishable from the
original, and the rotation by 1r cannot change that indistinguishability, it is
clear that the symmetry operation i generates a mirror image superimposable
on the original.
[
- - ~ ] [-~ - ~ ] - [ ~ ~ o]
o o -1 o o 1 o o ~
We now tum to the only remaining cases, those that have only S, axes of
even arder ~ 4 . In practice only S
4
is likely to be encountered, although rare
examples of S
6
and S
8
are known. We shall analyze that specific case, but it
will be evident that the analysis is easily adaptable to aH Sn (n even) axes.
The matrix describing the action of an S
4
rotation about z on any point in
the molecule, (x, y, z) is shown at the top of page 38. If this is multiplied by
the matrix for a proper rotation by n/2, we obtain the matrix for reflection
in a mirror plan e perpendicular to z. Thus, the operation s ~ converts the
molecule into its superimposable mirror image. For the specific case of 1.,3,5,7-
tetramethylcyclooctatetraene, which belongs to point group S.h this is shown
pictoria11y in Figure 3.5.
3
!
Rrllrrt
in. hutitntlllll
nurmr
4
2
Figure 3.5 Drawings showing that L3,5,7-tetramethylcyclooctatetraene, which be-
longs to point group s ~ . is converted by the s ~ operation into its mirror image. The
direct result of the s ~ operation needs to be turned 90 to become coincident with the
mirror image at lower left.
38 PRINCIPLES
-
o o 1 o o -1 o o -1
Finally, we turn to the principie that if a has no Sn axis, it cannot
be superimposable on its mirror image. The symmetry group. of a molecule
must include every operation capable of converting the molecule into one
indistinguishab)e from tbe original. lf its mirror image is indistinguishable, it
must result from at least one of the symmetry operations. We have shown
that all improper rotation axes generate operations that produce a mirror
image that is superimposable. However, proper rotations are incapable of
producing any mirror image, beca use all they do, siogly, or however combined
with one another, is reorient the molecule in space. If a molecule has a
superimposable mirror image, it must belong to a symmetry group that con-
tains at least one operation that will make that manifest, and only sorne kind
of Sn operation is able todo so. If, a molecule belongs toa symmetry
group having no Sn operation, that is, to a group consisting only of proper
rotations, its mirror image must be nonsuperimposable.
Are Tbere Exceptions?
There are no genuine exceptions to the above prescriptions conceming dis-
symmetric molecules, but when we deal with molecules in which internal
rearrangement, say by rotation about bonds, may occur, these prescriptions
may not tell the whole story from the point of view of the chemist. Such cases
are rare, but let us mention one that has been in the chemical literature for
sorne years [cf. K. Mislow and R. Bolstad, 1. Am. Chem. Soc., 77, 67t2
(1955)]. It has been stated-correctJy-that the compound shown as 1 in
Figure 3.6 is c.coptically inactive and configurationally pure', even though its
"individual molecules are asymmetric." This moJecule is, however, dissym-
metric, as the drawings in Figure 3.6 clearly show: Molecule 2 is the mirror
image of 1, and in 3 it is simply flipped top to bottom and rotated 90 clockwise.
This causes the lower half to coincide with the lower half of 1, but the upper
half does not then coincide. In short 1 and its mirror image do not coincide,
and these molecules 'are members of a dissymmetric pair.
The reason 1 (and, equally 3) are not optically active is because interna!
rotation about one C-C bond can take place so as to interconvert 1 and 3.
While that is chemically of importance, it has nothing to do with the math-
ematical question of the symmetry properties of these molecules.
The fact that optical isomers of amines of the type NRR'R" cannot be
isolated is the same sort of nongenuine "'exception." Such a mo)ecule is truly
dissymmetric. The only reason optical isomers are not isolable is that its
intemal structure is not rigid. Here it is not an internal rotation but an
1 2
e
o' 'oR
3
d
Figure 3.6 Structure 1 is a molecule that has been shown experimentaUy to be
optically inactive, but which is dissymetric. In step (a) it is reflected in a vertical mirror
perpendicular to the plan e of the paper. In step (b) the mirror image 2 is flipped
vertically and rotated to make the bottom half of 3 coincide with the bottom half
of l.
umbrella-type inversion that allows the two dissymmetric isomers to inter-
convert.
3.11 THE SYMMETRY POINT GROUPS
Suppose that we have, by inspection, compiled a list of all of the symme-
try elements possessed by a given molecule. We can then list all of the sym-
metry operations generated by each of these elements. Our first objective
in this section is to demonstrate that such a complete list of symmetry
operations satisfies the four criteria for a mathematical group. When this
has been established, we shaJI then be free to use the theorems concem-
ing the behavior of groups to assist in dealing with problems of molecular
symmetry. .
Let us first specify what we mean by a complete set of symmetry operations
for a particular molecule. A complete set is one in which every possible
product of two operations in the set is also an operation in the set. Let us
consideras an example the set of operations which may be performed on a
planar AB
3
molecule. These are E, C
3
, Cj, C
2
, C, C2, a,., u:, ah, S
3
, and
Sj. It should be clear that no other operations are possible. lf we
number the B atoros as indicated, we can systematically work through all
binary products; for example:
40 PRINCIPLES
1
i
'
.
B
u,
"8!! B2
1/,/
,
1 1
Ca
cr.,
A' )'
A
):
A
B/'"B->
/'"
Br "Bt B 1 Ba
/ 3 -
.
.
/
1
1
1
CTu
'
(1'11
Hence we see that a,.C
3
= a:.. Proceeding in this way, we can check aJl the
combinations and we will find that the set given is indeed complete. This is
suggested as a useful exercise.
Now, we can see that, because our set of operations is complete in the
sense defined above, it satisfies the first requirement for mathematical groups ..
if we take as our law of combination of two symmetry operations the successive
application of these operations.
The second requirement-that there must exist a group element E such
that for every other element in the group, say X, EX = XE = X-is also
seen to be satisfied. The "operation" of performing no operation at aH, or
that which results from a sequence of operations which sends the molecule
into a configuration identical with the original (e.g., cr, e:), is our identity,
E, and we have been calling it that all along.
The associative law is obviously valid for products of symmetry operations.
The final requirement, that every element of the group have an inverse,
is also satisfied. For a group composed of symmetry operations, we m ay define
the inverse of a given operation as that second operation, which will exactly
undo what the given operation does. In more sophisticated terms, the recip-
roca} S of an operation R must be such that RS = SR = E. Let us consider
each type of symmetry operation. For" a, reflection in a plan e, the inverse is
clear1y (J itself: (J X a = cr = E. For proper rotation, c:a' the inverse is
c:-m, for C': X c:-m = e: = E. For improper rotation, the reciprocal
depends on whether nt and n are even or odd, but a reciprocal in each
of the four possible cases. When n is even, the reciproca} of S': is s:-m whether
1n is even or odd. When n is odd and nz is even, = C':, the reciprocal of
which is c:-m. For with both 11 and In odd we may write = C':a. The
reciprocal would be the product whichis equal to and which
in turn may be written as a single operation,
We have shown that complete sets of symmetry operations do constitute
groups. Now we shall systematically consider what kinds of groups will be
obtained from various possible collections of symmetry operations.
In the trivial case where there are no symmetry operations other than E,
we ha ve a group of arder 1 called C
1
*
Let us next consider molecules whose sole symmetry element is a plane.
This element genera tes only two operations, namely, a and u2 = E. Hence
the group is of arder 2. The symbol normally given to this group is Cs. lt is
also possible to have a molecule whose sole symmetry element is an inversion
center. The only operations generated by the inversion center are i and i
1
=
E. Again we have a group of order 2; this one is conventionally desig-
nated C;.
Let us consider now the cases where the only symmetry element is a proper
axis, Cn. This generates a set of operations Cn, e ~ e!, ... , e: = E. Hence
a molecule with en as its only symmetry element would belong to a group of
arder 11, which is designated en. lt may be noted that a en group is a cyclic
group (see Section 2.2) and hence also Abelian.
When an improper axis is present, we must consider whether it is even or
odd. When the axis, S,., is of even arder, the group of operations it generates
is called S,. and consists of the n elements E, S,., C,.
11
, S ~ ... , s:-
1
The
group s2 is a special case because, as shown earlier, the symmetry element
s2 is equivalent to i. Thus the group that might be caiJed the s2 group is
actually called C;. The group of operations generated by an S,. axis when n
is odd has been shown to consist of 2n e-lements, including ah and the oper-
ations generated by e,.. By convention such groups are denoted C
111
,. This
symbol emphasizes that there is a en axis and a horizontal plane; this com-
bination of symmetry elements of course implies the existence of Sn just as
Sn (n odd) implies the existence of cll and U,. The Cnh groups will be discussed
in more detail shortly.
Next we tum to groups that arise when two or more symmetry elements
are present. In so doing, we shall divide the discussion into two parts. First,
we deal with cases where there is no more than one axis of order higher than
2. Then, in the next section, we shali take up the groups that arise when
there are several high-order (n > 2) axes. In each part, we shall proceed in
a systematic way which should suggest strongly, if not actually pro ve rigor-
ously, that all possibilities ha ve be en included.
We have aJready seen that, if a molecule possesses a proper axis .. C,., and
also a twofold axis perpendicular to it, there must then necessarily be 11 such
twofold axes. Tbe 11 operations, E, C,. .. e;, ... ,. e:-', plus the n twofold
rotations constitute a complete set of symmetry operations .. as m ay be verified
This symbol and the other ones for the symmetry groups, for example, C,.. D,., C,.,.. c ..... D,,..
D,.J, ... which will be introduced are called the Schonfties symbols after their inventor. The
symmetry groups are also frequently called point groups, since all symmetry elements in a
molecule will intersect ata common point, which is not shifted by any of the symmetry operations.
There are also symmetry groups, called space groups, which contain operations in\'olving trans-
latory .motions. The latter are considered in Chapter 11.
42 PRINOPLES
by actually carrying through aH of the binary products. Thus such a group
consists of a total of 2n elements. The symbol for a group of this kind is Dn
We ha ve now reached a point of departure in the process of adding further
symmetry elements toa Cn axis. We shall consider (1) the addition of different
kinds of symrnetry planes to the en axis only, and (2) the addition of symmetry
planes to a set of elements consisting of the en axis and the n el axes per-
pendicular to it. In the caurse of this development it will be useful to have
sorne symbols for several kinds of symmetry planes. In defining su eh symbols
we shall consider the of the en axis, which we call principal axis
or reference axis, to be vertical. Hence, a symmetry plane perpendicular to
this axis will be called ,a horizontal plan e and denoted ah. Planes that include
the en axis are generally called vertical planes, but there are actually two
different types. In sorne molecules aH vertical planes are equivalent and are
symbolized a, .. In others there may be two different sets of vertical planes (as
in PtCil-; cf. page 32), in which case those of one set will be called avand
those of the other set ad, the d standing for dihedral. It will be best to discuss
these differences more fully as we meet them ..
If to the Cn axis we add a horizontal plane, we expand the original group
of 11 operations, en, C, ... , E, to include all of the products uhen,
u he!, . . . , a hE = ah, making 2n operations in all. Now the operation
ahC': = e: ah, sin ce ah affects only the z coordinate of a point while C': affects
only its x and y coordina tes, so that the order in which a C:' are performed
is inconsequential. all of the new operations of the type uC':
can be expressed as single operations, namely, as improper rotations. This
new set of 2n operations can easily be shown to be a complete set and hence
to constitute a group. Such a group has the general symbol Cn11.
Let us look next at the consequences of adding a vertical ptane to the en
axis. First we recall (Section 3.5) that the operations generated by en when
n is odd will require that an entire set of 11 such vertical planes exist. All of
these planes are properly called vertical planes and symbolized u .. When n
is even, however, we have seen (Section 3.5) that only n/2 planes of the same
type will exist as a direct consequence of the Cn axis. However, we ha ve also
shown (Section 3. 7) that another set of n/2 vertical planes must exist as the
various products C:'av. These vertical planes in this second set are usually
called dihedraJ planes, since they bisect the dihedraJ angles between members
of the set of uv's, and they are denoted u d. Obviously, it is completely arbitrary
which set is considered vertical and which dihedral. In either case, n even or
n odd, the set of operations generated by the Cn and by all of the a's constitutes
a complete set, and such a group is called enp
We might naturally ask now about what happens when we add both the
horizontal plane and the set of n vertical planes to the en. This gives a group
called Dnh' which we shall now develop by a different procedure.
Let us consider the consequences of adding a ah to the group Dn. The
group generated is denoted Dnh We must first look at all of the products of
a
1
, with the operations generated by the e
2
axes and by the Cn axis. Suppose
we choose a coordinate system such that the Cn axis coincides with the z axis,
and one of the C
2
axes, C
2
(x), coincides with the x axis. We can indicate the
effect of rotation about the C
2
(x) axis follo\ved by ah on a general point
[x, y, z] thus:
[
C2(.t) _ _ n, [ _
x, y, z] [x, y, z]----+ x., y, z]
The effect of reflection in the xz plane on the same point will be
Thus we can write
where the second equality simply states that the rotation and the ah commute,
which we have previously shown to be generally true. Of course, it now follows
that, if one of the C
2
axes lies in a vertical symmetry plane, so must all of
the others. There must then be a set of n operations a, .. We may now left-
multiply the above equation by ah, obtaining
and we see that all of the products of a
1
, with the a.,'s are C
2
's. Thus we might
also take the simultaneous existence of Cn, a
1
, and the as the criterion
for the existence of the group Dnh It is only by reason of convention and not
because of any mathematical requirement that we take instead the simulta-
neous existence of Cn, 11C
2
"s, and a
1
, as the criterion.
We have now shown that the operations in a group Dnh include E,
(n - 1) proper rotations about en, 11 reflections in vertical planes, ah., and 11
rotations about C
2
axes. These 3n + 1 operations still do not constitute the
complete set, however. lt will be found that among the products C';a,. =
uhC': are n - 1 additional operations which are all improper rotations. For
. the general case where n is even, we obtain the new operations: Sn., Sn
12
, .,
i ( =

... , s:-. In the group D6h, for example, we have S
6
,
S3, i, and When n is odd we obtain in the general case the following
n - 1 improper rotations: Sn, ... , s;n-
3
, except s: (=ah);
S': where n1 is even being, of course, either E or one of the proper rotations
which we have aJready recognized. Thus, we now have a total of 4n operations
in the group Dnh Systematic examination will show that the set is complete.
Our next and final task is to consider the consequences of adding to Cn
and the n C
2
's a set of dihedral planes, u's. These are vertical planes that
bisect the angles between adjacent pairs of. C
2
axes. The groups generated by
this combination of symmetry elements are denoted Dnd The products of a
U with the various C;:r operations are all other a operations. However, among
44 PRINCIPLES
the various products of the type uC
1
there is a set. of n new operations
generated by an S
2
" axis colinear with en. These 4n operations now constitute
the complete group Dnd
Linear Molecules
These constitute a somewhat special case, although their possible point groups
are closely related to the scheme just developed. Any linear molecule has an
axis of symmetry coinciding vvith all the nuclei. The order of this axis is oo;
that is, rotations by any and all angles about this axis constitute symmetry
operations. Also., any plane containing the molecule is aplane of symmetry.
There is an infinite number of such planes, all intersecting along the molecular
axis. Proceeding from this, we ha ve just two possibilities: ( 1) the molecule is
oi the type OCO, NCCN, and so on, such that it consists of t\\'0 equivalent
halves., or (2) it is of the type NNO .. HCN .. and so on., and does not consist
of two equivalent halves.
In the first case, the equivalence of the two halves means that any line
which is a perpendicular bisector of the molecular axis is a symmetry axis;
there is an infinite number of such C
1
axes. The equivalence of the two
halves of the molecule also means that there is a plane of symmetry perpen-
dicular to the molecular axis. Since there is an infinity of rotations about a
unique, vertical axis, there is an infinity of axes perpendicular to
e-z., and there is a horizontal plane of symmetry. The group is, very reasonably,
designated Dxh
For linear molecules which do not consist of equivalent halves, the only
symmetry operations are the rotations about Cx and reflections in the vertical
planes. The group is called Cz,
3.U SYMMETRIES WITH MULTIPLE IDGH-ORDER AXES
We have so far considered how groups of symmetry operations may be sys-
tematically built up by beginning with one proper axis of rotation (the ref-
erence axis) and the operations that it generates and adding to this group (a
pure rotation group, en) the operations generated by additional symmetry
elements, these being restricted to planes or twofold axes. We have not yet
inquired about the possibility of adding to the operations generated by one
higher-order (n > 2) axis additional high-order axes. That is the subject we
niust now investigate.
It tums out that there are not actually very many possibilities ( only seven,
to anticipate); sorne of these, however, are among the most important point
groups we shall encounter in nature, and thus they deserve careful consid-
eration. An interesting and systematic way to approach the subject is to
recognize that groups involving several equivalent, intersecting, higher-order
axes will be represented by polyhedra having faces perpendicular to such
axes. For example, the tetrahedron, with four equilateral triangular faces,
must have four equivalent., intersecting C
3
axes. By first making a demon-
strably complete list of all such polyhedra., and then systematically considering
their symmetry groups and all subgroups thereof in which the multiple axes
are retained, we may expect to obtain a complete list of the symmetry groups
with multiple high-order axes.
The Five Platonic Solids
To implement the above plan \Ve consider the regular polyh_edra, sometimes
called the Platonic solids, of \Vhich there are five. By a regular polyhedron
we mean a polyhedron
(1) whose faces are all sorne regular polygon (i.e., equilateral triang1e.,
square, regular pentagon, hexagon., etc.) and equivalent to one an-
other;
(2) whose vertices are all equivalent; and
(3) whose edges are all equivalent.
By "equivalent" we mean, as usual, interchangeable by symmetry operations.
The five regular polyhedra are depicted and their essential characteristics
listed in Table 3.2.
Our first task is to show that the five Platonic solids do, in fact, represent
all the possibilities. This is quite easy to do.
In order to construct a polyhedron, three or more of the desired faces must
meet at a point so asto produce a closed, pyramidal (not planar) arrangement.
Using equilateral triangles, we have the following possibilities:
l. Three triangles with a common vertex.
2. Four triangles with a common vertex.
3. Five triangles with a common vertex.
If six equilateral triangles share a common vertex, the sum of the angles
around the vertex is 6 x 60 = 360. The arra y is planar and cannot form
part of a regular polyhedron. lt is clear that the three possibilities listed give
rise to the tetrahedron, the octahedron, and the icosahedron., as shown in
Table 3.2.
For the next higher regular polygon, the square, there is only one possi-
bility, namely, three squares with a common vertex, and this gives rise to the
cube. Four squares having a common vertex would alllie in one p1ane.
With regular pentagons {inte.rnaf arigle, 108), t h ~ r is only one possibility,
namely, three pentagons meeting ata common_ vertex (3 x 108 = 324),
since four or more could not be fitted together ( 4 x lOBo = 432). This single
conjunction of pentagons can be replicated to produce the dodecahedron.
46 PRINCIPLES
TABLE 3.2 The Five Regular Polyhedra or Platonic Solids
Tetrahedron
Faces: 4 equilateral triangles
Vertices: 4
Edges: 6
Cube
Faces: 6 squares
Vertices: 8
Edges: 12
Octahedron
Vertices: 6
Edges: 12
Oodecahedron
Faces: 12 regular pentagons
Vertices: 20
Edges: 30
lcosahedron
Vertices: 12
Edges: 30
With bexagons, there is no way to construct a regular polyhedron, snce
even three hexagons sharing a vertex lie in the same plane. With all higher
polyhedra not even three can be fitted together at a common vertex.
lt is clear, therefore, that the five Platonic solids are the only regular
polyhedra possible. Let us now examine them to see what symmetry opera-
tions may be performed on each one.
Inspection of the tetrahedron (see Fig. 3. 7) reveals the following symmetry
elements and operations.
(i) Three S
4
axes coinciding with the x, y, and z axes. Each of these
generates the operations S
4
, = C
1
, and
(ii) Three C
2
axes coinciding with the x, y, and z axes., each of which
generates an operation C
2
These operations have already been
however, by the S
4
's.
(iii) Four C
3
axes, each of which passes through one apex and the center
of the opposite face. Each of these generates Cj and Cj operations, that is,
eight operations in all.
(iv). Six planes of symmetry, each of which generates a symmetry opera-
tion.
The entire set of operations thus consists of the following 24, which are
listed by classes (as will be explained in Section 3.13):
This group is called TJ.
The octahedron has the following symmetry elements.and operations.
(i) Three S
4
axes, each passing through a pair of opposite apices. Each
generates the operations S
4
, C
2
, Sl.
z
y
Figure 3.7 A tetrahedron within a cube., showing a set of Cartesian axes.
48 PRINCIPLES
(ii) Three C
2
axes colinear with the S
4
's. The C
2
operations generated
by these axes, however, are airead y accounted for under (i).
(ii) Three C
4
axes colinear with the and C
2
's. Each generates a set
of operations C.a, C2, and Ct but only C4 and are ne\V.
(iv) Six C2 axes, which bisect opposite edges. Each generates an oper-
ation
(v) Four S
6
axes, each passing through the centers of a pair of opposite
triangular faces. Each generates a set of operations S
6
,
i, Cj,
(vi) Four C
3
axes co1inear with the S
6
's. Each generates two operations,
C
3
and Cj, which are also generated by colinear S
6

(vii) An inversion center which generates an operation i, also generated
by each of the S
6
axes.
( viii) Three planes of symmetry ea eh of which passes through four of the
six apices and generates an operation u
1
,.
(ix) Six planes of symmetry each of which passes through two
bisects two opposite edges and generates an operation UJ.
The entire set of operations thus consists of the following 48, grouped by
classes (as will be explained in Section 3.13):
This is the group caiJed 0
1
,.
Inspection will show that the cube has precise( y the same set of symmetry
operations as the octahedron; it, too, belongs to the point group oh. It is
worthwhile noting that the cube and the octahedron are very closely related.
Each is obtainable from the other by shaving off corners, as shown in Figure
3.8 for the cube-to-octahedron conversion. In a cube the faces are penetrated
by C.a axes and the vertices by C
3
axes; in the octahedron the vertices lie on
e_. axes and the faces are penetrat.ed by the C
3
axes. The nonregular poly-
hedron obtained as an intermediate when the triangular faces just meet is
called a cuboctahedron. It, too, has oh symmetry' as does every polyhedron
which is transitional, in the sense .of Figure 3.8, between the cube and the
octahedron. The cuboctahedron occurs in nature as B;
2
cages in certain borides
and has an interesting relatonship to the icosahedron (cf. Exercise B3.1).
Finally, we turn to the pentagonal dodecahedron and the icosahedron.
These two polyhedra ha ve the same symmetry. They are related to each other
as the cube and octahedron are related. The symmetry elements and oper-
ations are as follows.
(i) Each polyhedron has a set of six Sm axes. In the dodecahedron these
pass through opposite pairs of pentagonal faces; in the icosahedron they pass
through opposite vertices. Each S
10
axis generates these operations: S
11
.,
S
, e S
1
- e' S
5
- sb - C
3
S
7
sx - S"' E
JIJ = 5, ju, 111 - 5, ifl - 111 - 5 llh IU - 5, ICh

/

MOLECULAR SYf\.1METRY ANO THE SYMMETRY GROUPS 49
1
J------
Cube
Cuboctahedron
-----:;:o
- :
1 1 1
' 1 1
1 J
1 ,
1 1
L ____ J.-"
Octahedron
(inscribed in the
original cube)
Figure 3.8 Conversion of the cube to the octahedron via the cuboctahedron.
(ii) Each polyhedron has 10 S
6
axes. In the dodecahedron they pass.through
opposite pairs of vertices; in the icosahedron they pass through pairs of
opposite faces. Each of these generates these operations: S
6
, =
C
3
, S
2
= i, S! = Cj .. .. E. Of these. i andE have already been noted.
(iii) There are 6 C
5
axes, colinear with the S
111
axes. They generate C
5
,
operations, which have already been counted under Sw.
(iv) There are 10 C
3
axes, colinear with the S() axes. These generate C
3
and Ci operations, which have already been counted under S
6

(v) There are 15 C
2
axes, which in each case bisect opposite edges. These
generate 15 c2 operations.
(vi) There are 15 mirror planes, each containing 2 axes and 2 C
5
axes.
They generate 15 reflection operations.
Altogether there are 120 operations, which form the following classes:
E, 12Cj, 20C
3
, 15C
2
, i, 12Sw., 12S:,., 20S,.,. 15a
The group that they constitute is called l,.
By direct inspection of the five regular polyhedra we have discovered three
50 PRINCIPLES
point. groups: TJ, O,, lh. There are, however, severa) more, which we \Vill
now obtain straightforwardly from these. As noted earlier (page 33). products
of rotations can only be rotations. Thus there are pure rotation groups. If
from any group containing reflections we remove t ~ reflections and all their
products with the proper rotations, there will remain a subgroup consisting
entirely of proper rotations.
Thus the group TJ has apure rotational subgroup, T, of order 12. It consists
of the following classes:
The group 0
1
has apure rotational subgroup O of order 24. It consists of
the following classes:
The group Jh has a pure rotational subgroup, 1, consisting of the foUowing
60 operations:
Finally, there is one more group, called 7J,. This can be derived by addjng
to T a set of planes, a
1
,, which contain pairs of C
2
axes (as opposed to planes
aJ, which contain one C
2
axis and bisect another pair, thus giving Td). When
all the distinct products of these planes with the operations of T are enum-
erated and collected into classes. we have
All together, we now have the following seven groups containing multiple
high-order axes:
From the systematic way in which they were obtained it should be clear that
this is an exbaustive list.
3.13 CLASSES OF SYMMETRY OPERATIONS
In Section 2.4 the concept of classes of elements within a group was intro-
duced. This GOncept is utilized in dealing with symmetry groups. As we shall
see in Chapter 4, it is convenient and customary in writing what is called the
character table of a group to consider all the e1ements of a given class together,
MOLECULAR SYMMETRY ANO THE SYf\1METRY GROUPS 51
z

Figure 3.9 The relationship of a,. and UJ planes in
since they all behave identically in the properties covered by the character
table. It is the purpose of this section to explain the manner in which the
symmetry operations are arranged into classes and to discuss the geometrical
significance of the classes.
Of the general definition of a class and the method of arranging
the elements of a group into classes given in Section 2.4 is perfectly applicable
to a symmetry group. Let us consider, for example, the group c .. , .. This group
of operations arises when the following symmetry elements are present: c4
and av. There are eight operations in the complete set generated by these
symmetry elements, namely, E, C
4
, = C
2
, Cl, 2a,., 2ad. The a/s are planes
perpendicular to one another, intersecting along the C
4
axis, and so are the
u/s. The a/s make 45 angles with the a,.'s (see Fig. 3.9). By methods pre-
viously explained and illustrated for determining the products of symmetry
operations, a multiplication table for this group can be worked out.
Then, using this to carry out all of the possible similarity transformations,
we find that there are the following classes:
E
n(l) aC2)
V l' ' V
uOl a(2)
d ' d
52 PRINCIPLES
It may be noted parenthetically that this result provides a good example of
the fact although the orders of all classes must be integral divisors of
the group order, not al/ integra) divisors of the group order need be repre-
sented among the orders of the classes. Observe that, while 4 is an integral
divisor of 8., there is no class of order 4 in this group.
With symmetry groups the classes have a geometrical significan ce. that m ay
be stated as follO\\'S: Two operations belong to the same class when one may
be replaced by the other in a new coordinate system which is accessible by a
SJ'nunetry operation. The italicized part of this prescription is quite importan t.
Let us consider the group C.,. and its subgroup C
4
to see what this means.
The operation shifts every point in the mo)ecule by 3 x 2n/4 in., let us
say, the dockwise direction. This., however is the same thing as shifting every
point by 21f/4 in the counterclock\\'ise direction. Let us then for the moment
think of the operation c .. as rotation by 2n/4 clockwise and = as
rotation by 2n/4 counterclockwise. Now suppose that the coordinate system
in which we have been \Vorking is (a) such that clockwise rotation by 2n/4
y
(a) (b)
converts a point [x, y] into [y, -x], while counterclock\vise rotation by
27f/4 converts [x, y) into [-y., x]. Symbolically
c .. (z)[x, y]--+ [y, -x]
y]--+ [-y, x]
In coordinate system (b), however, the effects of C
4
(clockwise) and (coun-
terclockwise) are
C..(z)[x, y]--+ [-y, x]
y]--+ [y, - x]
In short, the roles of C
4
and are interchanged in coordinate systems (b)
from what they are in coordinate system {a). No\v (and this is the important
point) there is a symn1etry operation in the group .. which \Vill convert
coordinate system (a) into coordinate system (b), namely,
Thus, in the
SYMf\.1ETRY ANO THE SYMMETRY GROUPS 53
group C
4
,., C
4
and = are in the same class. However, in the group C
4
(which contains only the operations E, e
4
, e"!, they are not in the same
class because none of these four operations has the effect of transforming
coordinate system (a) into system (b). Of course, since any Cn group is cyc\ic
and hence Abelian, we can see that all operations must be in different classes,
since each operation is conjugate only with itself in an Abelian group.
Returning to the group e
4
,. again, we note that C
2
is in a class by itself.
This is so on geometrical grounds, because clearly there can be no way of
shifting the coordinate system so that the effects of a rotation by 180 can be
produced by a rotation by 90, or by a reflection of any kind. Also, the a,.'s
and a/s form separate classes. Only a rotation by 2n/8 couJd change the
orientation of the coordinate system to a new one in which ad \Vould accom-
plish what a,. did in the old one, and rotation by 2TC/8 is not a symmetry
operation that occurs in the group.
lt might be suspected intuitively that there would be a clase relationship
between the classes of operations and the various sets of equivalent operations
in a group. In fact, the classes correspond directly to the sets of equivalen!
operations. reason is easy to see. The geometrical criterion for putting
two operations, A and B,. in the same class is that there be sorne third op-
eration, e, which can be applied to the coordinate system so that operation
B in the transformed coordinate system is analogous to operation A in the
original coordinate system. At the same time, we say that operations A and
B are equivalent if one is converted into the other (in the same coordinate
system) by applying operation e to operations A and B. Now to say that
operation C interchanges operations A and B \\'hen applied to them, leaving
the coordinate system fixed, is perfectly equivalent to saying that operation
e interchanges the functioris of A and B when applied to the coordinate
system, 1eaving the operations fixed in space. Hence, the silnp/est way of
arranging the operations of a synunetry group into classes is to arrange the1n
into sets of equivalen! operations. These sets will be the c/asses.
A practical consequence of collecting all operations in the same class when
writing down the complete set, for example, at the head of a character table,
is that the notation used is a little different from what we have beep using
thus far. This new and final form of notation will now be explained and
illustrated for the four kinds of symmetry operations.
(i) Jnversion. Only one inversion operation is possible in a molecule. If
one exists it is denoted i. It will always be in a class by itself.
(ii) Reftections. Reflection in a horizontal plane is denoted a
1
,. This op-
eration \vill always be in a class by itself. When there is a set of n vertical
planes all in the same class, we write simply 11a,1, and for a set of aJ,s, naJ.
When there are sorne vertical planes in one class and sorne in another, sorne
may be cal1ed a,.'s and the set will be indicated .by na,., whi1e the second set
may be denoted na:. or naJ (the use of a:. or aJ for the second set is somewhat
arbi trary).
54 PRINCIPLES
(iii) Proper rotations. In the cyclic groups each of the operations, en,
e;, .. C! ....... e: -t .. constitutes a dass by itseJf and we continue to use this
notation. However, in aH other groups of higher symmetry, the number of
dasses spanned by these operations will be reduced in the following way. A
C'; will fall into the same dass with We ha ve seen an example of this
in the group C.,, ... where C
4
and are in the same class. In these cases, we
use the notation illustrated below for the various operations generated by a
c7 and a e6 axis.
Old Notation
(Grouped by classes)
{
c7 ..
c;"s
e
] C-4
, 7
cm's C
2
= C C
4
= e
1
6 6 ), 6 3,
Ne\\' Notation
2e1
?C2
- 7
.,C3
7
{
cf.,
C2 C2
In short, when two operations such as e
7
and are in the same dass, one
will be the same as the other only in the reverse direction, so that C
7
and
e' may both be called simply c7 ... and so on.
(iv) bnproper rotations. Justas with proper rotations, when two improper
rotations fall in the same class it wiJI be because one is really the same as the
other except that the rotation is in the opposite sense. Thus S
6
and may
both be considered Sb's and are so written.
3.14 A SYSTEMATIC PROCEDURE FOR SYMMETRY
CLASSIFICATION OF MOLECULES
In Section 3.11 we have shown that a complete and nonredundant set of
symmetry operations for any molecule constitutes a mathematical group, and
the various groups or kinds of groups (i.e., Cn., Dn, Sn, Cm,, C,,,, Dnd' T, . .. )
that we may expect to encounter among real molecules have been described.
In this section we shall present a systematic procedure for deciding to what
point group any molecule belongs. This will be done a practical, "how-to-
do-it" manner, but the cJose relationship of this procedure to the arguments
used in deriving the various groups should be evident. The following sequence
of steps will lead systematically to a correct classification.
l. We determine whether the molecule to one of the .. special'"
groups, that is, C=o,., or one of those with multiple high-order axes. Only
linear molecules can belong to C=v or D-z:h, so these cannot possibly involve
any uncertainty. The specially high symmetry of the others is usually obvious.
AH of the cubic groups, T, Th, TJ, O, and 0
1
,, require four C
3
axes, while 1
and /
1
, require ten C
3
's and six C
5
's. These multiple C
3
and C
5
s are the key
things to look for. In practice only molecules built on a central
cuboctahedron, cube .. pentagonal dodecahedron .. or icosahedron
will qualify, and these figures are usually very conspicuous.
2. If the molecule belongs to none of the special groups, \Ve search for
proper or improper axes of rotation. If no axes of either type can be found,
\Ve loo k for a plane or center of symmetry. lf a plan e only is found, the group
is Cs. lf a center only is found (this is very rare), the group is C;. If no symmetry
element at all is present, the group is the trivial one containing only the
identity operation and designated e,.
3. If an even-order improper axis (in practice only Sb, and Ss are likely)
is found bu t no planes of symmetry or any proper axis except a colinear one
(or more), whose presence is automatically required by the in1proper axis,
the group is S4, sb .. SR, ... - An s4 axis requires a c2 an sh axis requires
a C:. axis; an SR axis requires c4 and c2 axes. The important point here is that
the Sn (n even) groups consist exclusively of the operations generated by the
S, axis. lf any additional operation is possible, we are dealing with a D,,
1
or
D,
1
, type of group. Molecules belonging to these S, groups are relatively rare,
and the conclusion that a molecuJe belongs to one of these groups should be
checked thoroughly befare it is accepted.
4. Once it is certain that the molecule belongs to none of the groups so
far considered, we look for the highest-order proper lt is possible that
there will be no one axis of uniquely high arder but instead three C"! axes.
In such a case, we look to see whether one of them is geometrically unique
in sorne sense, for example, in being colinear with a unique molecular axis.
This occurs with the molecule aliene, which is one of the examples to be
worked through later. If all of the axes appear quite similar to one another,
then any one may be selected at random as the axis to which the vertical or
horizontal character of planes will be referred. Suppose that C, is our ref-
erence or principal axis. The crucial question now is whether there exists a
set of 11 C
2
axes perpendicular to the Cn axis. lf so, we proceed to step 5. If
not, the molecule belongs to one of the groups C,, Cm., and Cnh If there are
no symmetry elements except the e, axis, the group is e,. If there are n
vertical planes, the group is c,L.. If there is a horizontal plane, the group
is Cnh
5. lf in addition to the principal C, axis there aren C
2
axes Jying in a plane
perpendicular to the e, axis, the molecule belongs to one of the groups D,,
/ D,h, and lf there are no symmetry elements besides C, and the n C
2
axes, the group is Dn. If there is also a horizontal plane of symmetry, the
group is D,,. A D,h group will also, necessarily, contain 11 vertical planes;
these planes contain the C
1
axes. If there is no a
1
, but there is a set of 11 vertical
planes which pass between the C
2
axes, the group is D nJ
The five-step procedure just explained is summarized in the flow chart of
Figure 3.10.
56 PRINCIPLES
Start
(step 1)
Special Groups
{a) linear molecules: c-:tJIJ'
DtG,
{b) Multiple high-order axes:
T, T,., Td, O, 0
11
, /, /,
(step 2)
1--.----------- ~ No proper or inlproper rotation
axes: C
1
, C
1
, C;
(step 3)
n ~ v S, (ll eren) axis: S4-, S
6
, S
8
, ,
e,. axis (not simple consequence of s2,.)
..
n C
2
'sJ..to C,.
a, n uv's no u's u,. n u/s no u's

~
+

c.,, c.u c. D,"
D,, D,.
Figure 3.10 A five-stage procedure for the symmetry classification of molecules.
3.15 ILLUSTRATIVE EXAMPLES
The scheme just outlined for allocating molecules to their point groups will
now be illustrated. We shall deal throughout with molecules that do not be long
to any of the special groups, and we shall also omit molecules belonging to
C
1
, Cn and C;. Thus, each illustration wiiJ begin at step 3, the search for an
even-order S, axis.
3. H
2
0 possesses no improper axis.
4. The highest-order proper axis is a C
1
axis passing through the oxygen
atom and bisecting a line between the hydrogen atoms. There are no other
C'! axes. Therefore H
2
0 must belong to C
2
, C
2
,., or C
21
,. Since it has two
vertical planes, one of which is the molecular plane, it belongs to the
group e2L'"
Example 2. NH
3
3. There is no improper axis.
4. The only proper axis is a C
3
axiS.; there are no e
2
axes at all. Hence,
the point group must be e), e),., or e)IJ_ There are three vertical planes, one
passing through each hydrogen atom ... ['he group is thus e
3
,
Example 3. Aliene (Fig. 3.11)
3. There is an S
4
axis coinciding the main, molecular (C e C) axis.
However, there are also other symmetry elements besides the C
2
axis that
1
is a necessary consequence of the S
4
obv1ous, perhaps are the planes
of symmetry passing through the H
2
C==e C and C C CH
2
sets of atoms.
Thus, although an S
4
axis is present, t te additional symmetry rules out the
point group s4. .
4. As .noted, there is a e
2
axis lyingl along the e e C axis. There is no
higher-order proper axis. There are tvto more C
2
axes perpendicular to this
one, as shown in Figure 3.11. Thus, the group must be a D type, and we
proceed to step 5.
5. Taking the e
2
axis lying along the e C C axis of the molecule as the
reference axis, we look for a ah. There i; none, so_the group D
2
h is eliminated.
There are, however, two vertical planes (which lie between C axes), so the
group is D
2
d
H
3.11 The aliene molecule.
58 PRINCIPLES
~
H
Figure 3.12 The hydrogen peroxide molecule in its nonplanar form.
Example 4. H
2
0
2
(Fig. 3.12)
A. The nonplanar equilibrium configuration
3. There is no improper axis.
4. As indicated in Figure 3.12, there is a C
2
axis and no other proper axis.
There are no planes of symmetry. Tbe group is therefore C
2
Note that the
C2 symmetry is in DO way related to the Value of the angle e except when ()
equals 0 or 90, in which case the symmetry is higher. We shall next examine
these two nonequilibrium configurations of the molecule.
B. The cis-planar configuration (O = 00)
3. Again there is no even-order Sn axis.
4. The C
2
axis, of course, remains. There are still no other proper axes.
The molecule now lies in a plan e, which is a plan e of symmetry, and there
is another plane of symmetry intersecting the molecular plane along the C
2
axis. The group is C
2
v
C. The trans-planar configuration (e = 90)
3. Again, there is no even-order Sn axis. (except S
2
= i).
4. The C
2
axis is still present, and there are no other proper axes. There
is now a ah, which is the molecular plane. The group is C
2
h
Example 5. 1,3,5,1-Tetramethylcyclooctatetraene (Fig. 3.13)
3. There is an S
4
axis. There are no additional independent symmetry
elements; the set of methyl groups destroys all the vertical planes and hori-
zontal C
2
axes that exist in C
8
H
8
itself. The group is therefore S
4

lt may be noted that this molecule contains no center of symmetry or any
plane of symmetry and yet it is not dissymmetric. It thus provides an excellent
illustration of the rule developed in Section 3.10.
Figure 3.13 The 1,3,5, 7-tetramethylcyclooctatetraene molecule.
Example 6. Cyclooctatetraene (Fig. 3.14)
3. There is an S
4
axis. However, there are also numerous other symmetry
elements that are independent of the S
4
axis. We thus proceed to step 4.
4. Coincident with the S
4
axis there is {by necessity) a C
2
axis. No proper
axis of higher arder can be found, but there are two more, equivalent C
axes in a plane perpendicular to the S
4
-C
2
axis. Thus we are dealing with a
D
2
type of group.
5. There is no ah, thus ruling out D
21
,. There are, however, vertical planes
of symmetry bisecting opposite double bonds. These pass between the C
axes, and the point group is D
2
d
Figure 3.14 The cyclooctatetraene molecule.
60 PRINCIPLES
Example 7. Benzene
3. There s an S
6
axis, perpendicular to the ring plane, but there are also
other symmetry elements independent of the s6 axis.
4. There is a c6 axis perpendicular to the ring plane and six c2 axes lying
in the ring plane. Hence the group is a D
6
type.
5. Sin ce there is a a
1
, the group is D
6
h. Note that there are vertical planes
of symmetry, but they contain the C
2
axes.
Example 8. PF
5
(Trigonal Bipyramidal)
3. There is no even-order Sn axis.
4. There is a unique C
3
axis, and there are three C"J. axes perpendicular
to it.
5. There is a a
1
,; the group is D
31
,.
Examp/e 9. Ferroce11e (Fig. 3.15)
A. The staggered configuration
3. There is an even-order improper axis, S
10
, as indicated in Figure 3.15,
but there are also other unrelated symmetry elements, so the group is
not S
10

Figure 3.15 The ferrocene molecule in its staggered configuration.
Figure 3.16 The ferrocene molecule in its eclipsed configuration.
4. The unique, high-order, proper axis is a C
5
axis, as shown. Perpendicular
to this there are five c2 axes.
5. Because of the staggered of the rings there is no a
11
There
are, however, five vertical planes of symmetry which pass between the c2
axes. The group is thus Dsd
B. The eclipsed configuration ( Fig. 3 .16)
3. There is no even-order S,, axis.
4. There is a C
5
axis as seen in Figure 3.16. There are five C
2
axes per-
pendicular to the C
5
axis.
5. There is a ah, so the group is D
511

C. Any internzediate configuration
3. There is no Sn axis at all.
4. There is a C
5
axis, as in Parts A and B. There still remain five C
2
axes
perpendicular to C
5

5. There are now no planes of symmetry. Hence, the point group is D
5

Thus, as one ring in ferrocene is rotated relative to the other the symmetry
changes from D
5
d (staggered) through a continuum of D
5
conformations to
Dsh (eclipsed).
EXERCISES
Set A. Properties of Symmetry Groups
A3.1 What are the highest-order pure rotational subgroups of C
811
, D2d, C
5
,.?
62 PRINCIPLES
A3.2 What group is obtained by adding to or deleting from each of the
following groups the indicated symmetry operation?
(a) C
3
plus i (d) S
6
minus i (g) D3d minus S
6
(b) c3L plus i (e) Td plus i (h) s4 plus i
(e) C
5
,. plus ah (f) C
3
plus S
6
(i) C
3
h minus ~
A3.3 What is the conventional designation for the group of operations gen-
erated by an Sn axis when n is odd?
A3.4 Write out all the operations generated by S
5
and S
8
axes, and express
each one in conventional notation.
Set B. Symmetry of Polyhedra and Other Nonmolecular Objects
B3.1 Show how a cuboctahedron may be transformed into an icosahedron
by converting each square face into a pair of triangular faces with a
common edge.
83.2 There are three relatively common types of dodecahedron. In addi-
tion to the pentagonal dodecahedron (point group Jh) tbere are the
trigonal dodecahedron (a) and the rhomboidal dodecahedron (b ). To
what point group does each of these belong?
(a)
(b)
83.3 To what symmetry is a tetrahedron with aH black edges reduced if
two edges that do not intersect are made red?
83.4 If you begin with an octahedron having eight black faces and paint
four of them, no two of which have a common edge, w h i t e ~ to which
symmetry group does the octahedron then belong?
83.5 What is the symmetry of a cube when a line is drawn across each of
its faces in the manner shown below?
83.6 Suppose that the line on each face of the cube shown in Exercise
B3.5 is rotated by O, where O < O < 45, in the clockwise direction
as seen from the outside. What is the point group now? If the angle
of rotation is 45, what is the point group?
83.7 If alternate vertices of the marked cube in Exercise B3.5 are painted
black, what is the symmetry?
83.8 Suppose that we begin with an octahedron, on each face of which
is drawn a smaller equilateral triangle oriented so that each vertex
of the smaH triangle points directly toward a vertex of the larger one;
the small triangles do not disturb the 0
11
symmetry. If each small
triangle is now twisted clockwise by an angle O, where O < (} < 60,
what symmetry does the figure possess?
83.9 Yet another dodecahedron, called a pyritohedron is shown below. To
what point group does it belong? Why is it not one of the Platonic
solids?
Pyritohedron.
83.10 What is the point group of the polyhedron called a hexagonal sca-
lenohedron shown below?
Hexagonal scalenohedron.
83.11 A tennis ball (including the seam).
64 PRINCIPLES
Set C. Sorne Relatively Easy Molecules
(1) SF
5
CI
(2) Chlorobenzene
(3) 1 ,2-Dichlorobenzene
(4) 1,3-Dichlorobenzene
(5) 1 ,4-Dichlorobenzene
(6) 1 ,3,5-Trichlorobenzene
(7) 1 2 ~ 3 Trichlorobenzene
(8) 1 ,2,4-Trichlorobenzene
(9) trans-[CrCI
2
{H:!O)-t)+ (ignore H atoms)
(10) 00
(11) OPCI
3
(12) trans-Pt(NH
3
)
2
Cl
2
(ignore H atoms)
(13) cis-Pt(NH
3
)
2
Ch (ignore l-1 atoms)
(14) BICIF
(15) What is the point group fo.r each of the following substituted cyclo-
butanes? Assume that C
4
H
8
itself has D
41
, symmetry and that replacing
an H by X or Y changes no other structure parameters.
X X
(a)
d
(d)
X
X
X
(b)
o
(e)
L=1
X X
X
(e)
u
(/)
X
MOLECULAR SYMMETRY AND THE SYMMETRY GROUPS
X
(g)
y
X X
(h)
u
X
(i)
X
X
X
Set D. Sorne Less Obvious Molecules
(1)
(2)
(3)
(5)
fac-Cr(CO)J(PR
3
)
3
(ignore R groups)

<>
Cl
,CI
,
Cl-Re -CI
CI_,..]Il , ,Cl
Cl-Re-Cl
l ~ .
(4)
(6)
X X
(k)
X X
X
(/)
X
X
y
(m)
y
X
65
66 PRINCIPLES
(7) (8)
(9) Ethylene
(10) Ethane (staggered)
(11) Be(CH
3
COCHCOCH
3
)
2
(ignore H atoms; Be is "tetrahedral")
(U) 1,3-Dichloroallene
Set E. Sorne Tricky Molecules
e
(1)
(2)
(all bonds of each type, a, b, e, are equal)
C>
(3) Cr
C>
o
Cr
o
(5) 0
(4)
(6) (7)
(8) (9)
(10) The crown-shaped S
8
molecule (11)
(U) (13)
rt ......
L-M-L
L...,...-1
"-_L
4
REPRESENTATIONS OF GROUPS
4.1 PREFACfORY COMMENTS ON MATRICES AND VECTORS
Because the representations of groups are in general made up of matrices,
and because certain properties of representations can be advantageously for-
mulated by using certain properties of vectors, this chapter will begin with
an account of several special aspects of matrix and vector algebra essential
to an understanding of the foHowing discussion of representation theory. For
those totally unfamiliar with matrices, background material antecedent to that
covered here is presented in Appendix I.
A Special Case of Matrix Multiplication
A special case of matrix multiplication occurs when we deal with matrices
having all nonzero elements in square blocks along the diagonal, such as the
following two:
1 o o o o o 4 1 o o o o
1 2 o o o o 2 3 o o o o
o o 3 o o o o o 1 o o o
o o o 1 3 2 o o o o 1 2
o o o 1 2 2 o o o 3 o 2
o o o 4 o 1 o o o 2 1 1
68
REPRESENTATIONS OF GROUPS 69
The product of these two matrices taken in the above arder is
4 1 o o o o
8 7 o o o o
o o 3 o o o
o o o 13 3 10
o o o 10 3 8
o o o 2 5 9
The most conspicuous feature of this product matrix is that it is blocked
out in exactly the same way as are its factors. It is not difficult to see that this
sort of result must always be obtained. Moreover, it should also easily be
seen that the elements of a given block in the product matrix are determined
only by the elements in the corresponding blocks in the factors. Thus, when
two matrices which are blocked out along the diagonal in the same way are
to be multiplied, the corresponding blocks in each may be considered inde-
pendently of the remaining blocks in each. Specifically, in the abo ve case,
u = [:
[3) X (1] = [3)
n u = !
Characters of Conjugate Matrices
An important property of a square matrix is its characcer. This is simply the
sum of its diagonal elements, and it is usually given the symbol X (Greek chi).
Thus
We shall now prove two important theorems concerning the behavior of
characters.
lf fi = si.J3 and q. = the characters of ({ and are equal.
70 PRINCIPLES
PROOF
Conjugare nzatrices have identical characters.
Conjugate matrices are related by a similarity transformation in the same way
as are conjugate elements of a group. if matrices .41 and _..,>are con-
jugate, there is sorne other matrix .:1 such that
Since the associative law holds for matrix multiplication, the theorem is pro ved
in the following way.
PROOF
X of . .!/( = x of..:? -l .-IJ> .1 = x of ( ..1-
1
.-r>) ..1
= x
= x of (..1..1-
1
).-f)
= x of.-1'
Matrix Notation for Geometric Transformations
One important application of matrix algebra is in expressing the transfor-
mations of a point-or the collection of points that define a body-in space.
We have employed previously five types of operations in describing the sym-
metry of a molecule or other object: E, a, i, Cn, Sn. Each of these types of
operation can be described by a matrix.
The ldentity. Wben a point with coordinates x, y, z is subjected to the
identity operation, its new coordina tes are the same as the initial ones, namely,
x, y, z. This may be expressed in a matrix equation as
Thus, the identity operation is described by a unit matrix.
Rejlections. If a plane of refiection is chosen to coincide with a principal
Cartesian plane (i.e., an xy, xz, or yz plane), reflection of a general point
has the effect of changing the sgn of the coordinate measured perpendicular
to the plane while leaving unchanged the two coordinates whose axes define
the plane. Thus, for reftections in the three principal planes, we may write
the following matrix equations:
a(xy): [ ~ ] [ ~ ] = [ ~ ]
a(xz): [ ~ - ~ ] [ ~ ] [ ~ ]
a(yz): [ ~ J [ n -~ ]
Jnversion. To simply change the signs of all the coordinates without per-
muting any, we clearly need a negative unit matrix, namely,
Proper Rotation. Defining the rotation axis as the z axis, we note first
that the z coordinate will be unchanged by any rotation about the z axis.
Thus, the matrix we seek must be, in part,
The problem of finding the four missing elements can then be solved as a
two-dimensional problem in the xy plane.
Suppose that we have a point in the xy plane with coordinates x
1
and y"
as shown in the diagram. This point defines a vector, r
1
, between itself and
the origin. Now suppose that this vector is rotated through an angle O so that
a new vector, r
2
, is produced with a terminus at the point x
2
and y
2
We now
inquire about how the final coordinates, x
2
and y
2
, are related to the original
coordina tes, x
1
and y
1
, and the angle O. The relationship is not difficult to
work out. When the x component of r
1
, x
1
, is rora red IJy O, ir IJecornes a vector
x' which has an x component of x
1
cos (J and a y component of x
1
sin O.
72 PRINCIPLES
y
Similarly, the y component of r
1
, y
1
, upon rotation by O be comes a new vector
y', which has an x component of - y
1
sin O and a y component of y
1
cos O.
Now, Xz and y
2
, the components of r
2
, must be equal to the sums of the x and
y components of x' and y', so we write
x
2
= x
1
cosO - y
1
sin O
Y2 = X sin O + y
1
cos O
(4.1-1)
The transformation expressed by 4.1-1 can be written in matrix notation in
the foiJowing way:
[
cos ()
sin O
-sin fJJ [x
1
] = [x2]
cosO Yt Y2
This result is for a counterclockwise rotation. Beca use cos c/J = cos(- c/J) while
sin c/J = -sin( - c/J), the matrix for a clockwise rotation through the angle ljJ
must be
[
cos c/J sin c/JJ
-sin ljJ cos ljJ
Thus, finally, the tota) matrix equation for a clockwise rotation through ljJ
about the z axis is
lmproper Rotation. Since an improper rotation through the angle ljJ about
the z axis produces the same transformation of the x and y coordinates as
does a proper rotation through the same angle, but in addition changes the
sign of the z coordina te., we may infer directly from the equation just derived
that the matrix for clockwise rotation is
[
cos ljJ sin ljJ
-sin ljJ cos ljJ
o o
J]
It will be clear that one could also have obtained this matrix by explicitly
multiplying the matrices for rotation and reflection in the xy plane.
In general, the matrices that describe symmetry operations can be multi-
plied together so that the product of any two is the matrix for sorne ( usually
other) operation. For instance., we previously showed (page 33) somewhat
tediously that the line of intersection of two perpendicular planes of symmetry
must be a twofold axis of symmetry. We may employ the matrices to show
the same thing very neatly. Thus, for a:c::.' and C
2
(z) we have
[
- [ - = [ - [ - = [ - -
o o 1 o o 1 o o 1 o o 1 o o 1
a:rz a_v:: ayz a:cz Cz(Z)
Symbolically, if a set of geometrical operations, A, B, C, D, ... , applied
successively gives the same net effect as a single operation X, that is,
DCBA =X
then the products of the matrices representing these operations will multiply
together in the same order to give a matrix corresponding to X. namely,
(r /JJ._r./ = ,:r
The in verse, .r ;-, of a matrix, .' 1, is defined by the equation
= ,r- 1 ,r/ = (!e
where cf is the unit matrix.
All of the matrices we have just worked out, as well. as all others which
describe the transformations of a set of orthogonal coordina tes by proper and
improper rotations, are called orthogonal matrices. They ha ve the convenient
property that their inverses are obtained merely by transposing rows and
columns. Thus, for example, the inverse of the matrix
IS
74 PRINCIPLES
as confirmed by
[
~ ] [ ~ ~ -] - [ ~ ~ -] [ ~ ] - [ ~ ~ ~ ]
-1 o o o 1 o o 1 o -1 o o o o 1
Since rotations by ljJ clockwise and counterclockwise are in verse operations,
their matrices must be in verse to each other. Thus the relation between the
matrices for these two operations could now be deduced simply by saying
that one must be the transpose of the other.
As a more general illustration of how matrices can be used to express
symmetry operations, consider the eight C
3
operations of a tetrahedron as
shown in the following Figure 4.1. Let us first consider the effect on a general
point, with coordinates x, y, and z, of a clockwise rotation by 2n/3 about the
axis C ~
0
This sends y into x, z into y, and x into z; that is, [x, y, z] becomes
[y, z, x]. Writing the two sets of coordinates as column matrices, we see that
the rotation operation can be described by the matrix equation
Similarly, counterclockwise rotation ( or C ~ in a clockwise direction) is de-
scribed by
[! ~ ~ ] [ ] -[f]
z
Figure 4.1 A tetrahedron inscribed in a cube, with the four threefold axes shown.
The matrices for C
3
and Cj rotations about the other C
3
axes are as follows:
c(2)
3
[
o

[J
-1
]
c3 =
o
C2-
o
3 -
1 o

[J
1
-!]

o

c3 =
o
C2-
o
3 -
o -1

[
o

-1

CJ =
o Cj = o
-1 o
Since the rotations C
3
and Cj in any pair about a given C
3
axis are inverse
to each other, the matrices representing them should also be in verse. More-
over, sin ce we are dealing with orthogonal matrices, it should be true that
each matrix in ech pair is the transpose of the other. It will be seen that this
IS SO.
With this set of matrices it can be shown that the product of any two
threefold rotations about different axes is either a twofold rotation about one
of the Cartesian axes or a threefold rotation about another C
3
axis. The
following three matrix equations iJiustrate how the C
2
operations, C
2
(x),
C
2
(y), and C
2
(z), respectively, arise as products ofcertain pairs of C
3
oper-
ations:

1

o

o

o o
-
-1
o -1 o

1
n[ -!
o

[
o

o o - 1
o -1 o

1
n[ -!
o

[
o

o o -
-1
o -1 o
It can easily be shown that any product of a C
2
and a C
3
operation is
another C
3
operation and any product of one C
2
operation with another gives
the third C
2
operation. Thus, the existence of 4 C
3
axes as shown in Figure
76 PRINCIPLES
4.1 (page 74) Jeads toa set of 8 c3 3 andan identity
operation. These 12 operations constitute the pure rotation group T.
What we ha ve just done is to substitute the algebraic process of multiplying
matrices for the geometric process of successively applying symmetry oper-
ations. The matrices multiply together in the same pattern as do the symmetry
operations; it is clear that they must, since they were constructed to do just
that. It will be seen in the next section that this sort of relationship between
a set of matrices and a group of symmetry operations has great importance
and utility.
Vectors and Their Scalar Products
Since many of the basic arguments in Section 4.3 will lean heavily on the
concept of orthogonal vectors in generalized, multidimensional space, a brief
summary of the essentials will be included here.
A vector in p-dimensional space may be defined by the lengths of its
projections on each of a set of p orthogonal axes in that space. For instance,
a vector A, in real space, with the coordinates xh y., z
1
for its outer
has a projection Ax of length x
1
on the x axis, a projection A.v of length y
1
on
the y axis, and a projection Az of length z
1
on the z axis.
One type of product of two vectors is called the scalar product because it
is merely a number, a scalar. This m ay be defined as the product of the lengths
of the two vectors times the cosine of the angle between them. The scalar
product is indicated by pJacing a dot between the symbols. We denote a vector
as A, its length by A, and its projections on coordina te axes by, for example,
A . .., A.,., ....
If two vectors C and D are orthogonal, their scalar or "dot'' product will
be zero, for
C D = CD cos 90 = O
If they are parallel or colinear, their scalar product is equal to the product
of their lengths since cos oo = l.
There is an equivalent but more generally useful way of writing the scalar
product of two vectors. Suppose that we have two vectors A and 8, both
lying in the xy plane. Let A make an angle c/J to the x axis and 8 a greater
angle lf/. The angle between A and B, O, is then ( lf/ - 4>). Thus
A B = A B cos O = A B cos (VI - e/>) (4.1-2)
Now the components of A ~ that is, its projections on the x and y axes., are
and similarly for B
Ax = Acose/>
A ~ = A sin 4>
Bx = B cos VI
B-" = B sin tJ
Using a trigonometric identity, we can write 4.1-2 as
A 8 = AB(cos 4> cos VI + sin 4> sin lf!)
which may be rearranged to
A 8 = A cos e/> B e os VI + A sin e/> B sin VI
Substituting the relations 4.1-3 and 4.1-4, we obtain
(4.1-3)
(4.1-4)
Thus the scalar product of vectors A and 8 in two-dimensional space is equal
to the su m of the products of their components with no cross terms ( e.g.,
A.rBy). This result is actually only a special case of the general rule in p-
dimensional space:
p
A B,= 2: A;B;
i=l
We can now restate the rule for orthogonality of two vectors in p-dimen-
sional space as requiring that
p
: A;B; = O
i= 1
and the square of the length of a vector m a y be written as
p
A
2
= L Af
i= l
78 PRINCIPLES
4.2 REPRESENTATIONS OF
A representation of a group of the type we shall be interested in may be
defined as a set of matrices, each corresponding to a single operation in the
group, that can be combined among themselves in a manner parallel to the
way in which the group elements-in this case, the symmetry operations-
combine. Thus, if two symmetry operations in a symmetry group, say C
2
and
a, combine to give a product then the matrices corresponding to C
2
and
a must multiply together to give the matrix corresponding to But we have
already seen that, if the matrices corresponding to all of the operations have
been correctly written down, they will naturally have this property.
As an example, let us work out a representation of the group e
2
..,, which
group consists of the operations E, e
2
, a,., a:.. Let us say that the e
2
axis
coincides with the z axis of a Cartesian coordinate system, and let a" be the
xz plan e anda:. be the yz plane. The matrices representing the transformations
effected on a general point can easily be seen to be as follows:

o

[-1
o

1
Cz:
-1
o o
a.:
o

[ -1
o

-1
a;:
1
o o
Now the group multiplication table is:
E
c2 a,.
,
a,.
E E e! a..,
,
a,.
e2 e'1
E
,
a"
a,.
a" a"
a:. E e1
'
a,.
,
a,. c2 E
It can easily be shown that the matrices multiply together in the same fashion.
For example,
and
-
o o 1 o o 1 o o 1
Again, each element in the group C
2
" is its own inverse, so the same must be
true of the matrices. This is easily shown to be so; for example,
-! -! - !
We ha ve now, by one procedure, namely, by considering the transfor-
mations of a general point, generated a set of matrices which form a repre-
sentation for the group C
2
, It will be recalled that we have also done the
same thing (pages 74-76) for the group T.
A question that natura1ly arises at this point is: How many representations
can be found for any particular group, say C
2
v, to continue with that as an
example? The answer is: A very large number, limited only by our ingenuity
in devising ways to generate them. There are first sorne very simple ones,
obtained by assigning 1 or -1 to each operation, namely,
E
Cz
a,, a'
V
1 1 1 1
1 -1 1 -1
1 -1 -1 1
1 1 -1 -1
Then there are many representations of high arder. For example, if we were
to assign three small unit vectors directed along the x, y, and z axes to each
of the atoms in and write down matrices representing the changes and
interchanges of these upon applying the operations, a set of four 9 x 9
matrices constituting a representation of the group would be obtained. Using
CH
2
Ch in the same way, we could obtain a representation consisting of
15 x 15 matrices. However, for any group, only a limited number of rep-
resentations are of fundamental significance, and we shall now discuss the
origin and properties of these.
Suppose that we ha ve a set of matrices, .r{, .-'A, t , ... , which form a
representation of a group. If we make the same similarity transformation on
ea eh matrix, we obtain a new set of matrices, namely,
,., = l.d
,r:{' = ..:1 -1
.-'11, = ! -1 /IJ.!!
80 PRINCIPLES
It is easy to prove that the new set of matrices is also a representation of the
group. Suppose that
.r:l.!/3 -
then
.. r.i, .43, = ( .d-
1
.d -l .-11.1) = ..1 -l ,r./ ( ,d.d -l) .4ld
= --1)-1( r:J.J!J) .J) = :? -1 r.if<v) = r.:/.'
....,;. 1..":1.7) ........ """ J
Clearly, all products in the set of matrices .r:( ', .4-3 ', . , will run parallel
to those in the representation .r:l, .43, ... ; hence the primed set also con-
stitutes a representation.
Let us now suppose that, when the matrix ,t:/ is transformed to si' using ..1
or sorne other matrix, we find ,e{' to be a block-factored matrix, namely,
-
,
1
r'
2
( 1,
3
' il
r /
' ,
...
1
for example. If now each of the matrices .r 1 ', .-tJ f. ', and so forth is blocked
out in the same way, as shown on page 69, corresponding blocks of
each matrix can be multiplied together separately. Thus we can write such
equation as:
r' 4:>' t'
. .. . }"" - - ""
Therefore the various sets of matrices
f ' r ' .//) ' / ' r 1 '
(t 1 , 1 ' ., J 1 ,
1
' 1 - 1 ,
(
1
( 1
1
..1/) f F l
(r ::! , ::! , , 7).!, 1' :! ,
( . ,
_1 :!_,
are in themselves representations of the group. We then call the set of matri-
ces, r, . reducible representation, because it is.possible,
using sorne matrix, ,;!/ in this case, to transform each matrix in the set into a
new one so that all of the new ones can be taken apart in the same way to
give two or more representations of smaller dimension. (The dimension of a
representation is the order of the square matrices that constitute it.) It if is
not possible to find a similarity transformation which will reduce all of the
matrices of a given representation in the above manner, the representation
is said to be irreducible. It is the irreducible representations of a group that
are of fundamental importance, and their main properties will now be de-
scribed.
4.3 THE "GREAT ORTHOGONALITY THEOREM"
ANO ITS CONSEQUENCES
All of the properties of group representations and their characters, which are
important in dealing with problems in valence theory and molecular dynamics,
can be derived from one basic theorem conceming the elements of the matri-
ces which constitute the irreducible representations. of a group. In arder to
state this theorem, which we shall do without proof, t sorne notation must be
introduced. The arder of a group \Vill, as before, be denoted by h. The
dimension of the ith representation, which is the order of each of the matrices
which constitute it, will be denoted by l;. The various operations in the group
will be given the generic symbol R. The element in the mth row and the nth
column of the matrix corresponding to an operation R in the ith irreducible
representation will be denoted f;(R)mn Finally, it is necessary to take the
complex conjugate (denoted by*) of one factor on the left-hand side whenever
imaginary or complex numbers are involved.
The great orthogonality theorem may then be stated as follows:
(4.3-1)
This means that in the set of matrices constituting any one irreducible rep-
resentation any set of corresponding matrix elements, one from each matrix,
behaves as the components of a vector in h-dimensional space such that all
these vectors are mutually orthogonal and each is normalized so that the
square of its length equals hll;. This-interpretation of 4.3-1 will perhaps be
more obvious if we take 4.3-1 apart into three simpler equations, each of
which is contained within it. We shall omit the explicit designation of complex
conjuga tes for simplicity, but it should be remembered that they must be used
t The proof, which is not trivial, may be found elsewhere. for example. H. Eyring. J. Walter.
and G. E. Kimball, Quanlum Chemistry, John Wiley & Sons, New York. 194-t, p. 371.
82 PRINCIPLES
when complex numbers are involved. The three simpler equations are as
follows:
2: f;(R)mnf(R)mn = O if i # j (4.3-2)
R
L f;(R)mnf;(R)m'n' = O if 1n # 111' and/or n # n' (4.3-3)
R
2: f(R)mnf;(R)mn = h/1; (4.3-4)
R
Thus, if the vectors differ by being chosen from matrices of different repre-
sentations, they are orthogonal ( 4.3-2). If they are chosen from the same
representation but from different sets of elements in the matrices of this
representation, they are orthogonal ( 4.3-3). Finally, 4.3-4 expresses the fact
that the square of the length of any such vector equals lzl 1;.
Five Important Rules
We shall now discuss five important rules about irreducible representations
and their characters.
l. The sum of the squares of the dimensions of the irreducible representa-
tions of a group is equal to the order of tlze group, that is,
'Llf = Lt + L ~ + Lj + = h (4.3-5)
PROOF. A complete proof is quite lengthy and will not be given. It is, however,
easy to show that 2fT :<:: h. In a matrix of arder L there are f2 elements. Thus
each irreducible representation, f;, will provide It h-dimensional vectors. The
basic theorem requires this set of 11 + ~ + /j + vectors to be mutually
orthogonal. Since there can be no more than h orthogonal h-dimensional
vectors, the sum Lf + li + Lj + may not exceed h. Since X;( E), the character
of the representation of E in the ith irreducible representation, is equal to
the arder of the representation, we can also write rule 1 as
(4.3-5a)
2. The sum of the squares of the characters in any irreducible representation
equals h, that is,
(4.3-6)
PROOF. From 4.3-1 we may write
Summing the left si de over m and m', we obtain
L L L f;(R)mmf;(R)m'm' = L [L f;(R)mm L f;{R)m'm']
m' m R R m m'
= L X;(R)x;(R)
R
while summing the right side over m and 1n', we obtain
h h
-
1
. 2: L n u n = -
1
. f; = lz
1 m' m 1
thus proving the equality 4.3-6.
3. The vectors whose co1nponents are the characters of two different irre-
ducible representations are orthogonal, that is,
L X;(R)X(R) = o when i=/=j (4.3-7)
R.
PROOF. Setting m = n in 4.3-2, we obtain
L f;(R)mm f(R),nm = O if i =/=j
R
L X;(R)X(R) = L [2: f;(R)mm L f(R)mm]
R R m m
= ~ [ f f;(R)mm f(R)mm] = O
4. In a given representation (reducible or irreducible) the characters of all
matrices belonging to operations in the same class are identical.
PROOF. Since all elements in the same class are conjuga te to one another,
all matrices corresponding to elements in the same class in any representation
84 PRINCIPLES
must be conjugate. But we have sho\vn on page 70 that conjugate matrices
have identical characters.
5. The nunzber of irreducible representations of a group is equal to the
number of classes in the group.
PROOF. As for rule l. a complete proof will not be given; we can, however.
easily prove that the number of classes sets an upper limit on the number of
irreducible representations. We can combine 4.3-6 and 4.3-7 into one e q u ~
tion, namely,
2: X;(R)X(R) = lu5; (4.3-8)
R
If now we denote the number of elements in the nzth class by gm, the number
in the nth class by gn, and so on, and if there are k classes altogether, 4.3-8
can be rewritten:
k
2: X;(Rp)X(Rp)gp = lu5; (4.3-9)
p=l
where RP refers to any one of the operations in the pth class. Equation
4.3-9 implies that the k quantities x
1
(Rp), in each representation f
1
behave
like the components of a k-dimensional vector and that these k vectors are
mutually orthogonal. Since only k k-dimensional vectors can be mutually
orthogonal, there can be no more than k irreducible representations in a
group which has k classes.
Illustrations of the Five Rules
Let us now consider the irreducible representations of several typical groups
to se e how these rules apply. The group C
2
v consists of four elements. and
each is in a separa te class. Hence (rule 5) there are four irreducible repre-
sentations for this group. But it is also required (rule 1) that the sum of the
squares of the dimensions of these representations equal h. Thus we are
looking for a set of four positive integers, 1
1
, 1
2
, /
3
, and /
4
, which satisfy the
relation
Lr + ~ ~ + 15 + 1 = 4
Clearly the only solution is
Thus the group C
2
" has four one-dimensional irreducible representations.
We can actually work out the characters of these four irreducible repre-
sentations-which are in this case the representations themselves because the
dimensions are 1-on the basis of the vector properties of the representations
and the rules derived above. One suitable vector in 4-space which has a
component of 1 corresponding toE will obviously be
1 1 1
for
: [x,(R)]2 = P + P + 1
2
+ F = 4
R
thus satisfying rule 2. Now all other representations will have to be such that
2: [x,(R)]2 = 4
R
which can be true only if each X;(R) = l. Moreover, in order for each of
the other representations to be orthogonal to ft (rule 3 and 4.3-7), there will
ha ve to be two + 1 's and two - 1 's. Thus
(1)( -1) + (1)( -1) + (1)(1) + (1)(1) = o
Therefore we will have
,
a,,
1 1 1
-1 -1 1
-1 1 -1
1 -1 -1
All of these representations are also orthogonal to one another. For example,
taking r2 and r4, we have
(1)(1) + (-1)(1) + (-1)(-1) + (1)(-1) =0
and so on. These are then the four irreducible representations of the group
C2v
As another example of the working of the rules, let us consider the group
C3v This consists of the following elements, listed by classes:
86 PRINCIPLES
We therefore know at once that there are three irreducible representations.
If we denote their dimensions by lb 1
2
, and
we have (rule 1)
1 + + /j = 11 = 6
The only val u es of the l; that will satisfy this requirement are 1, L and 2.
Now once again, and always in any group, there will be a one-dimensional
representation whose characters are all equal to l. Thus we have
r, 1
3a,,
1 1
Note that (from 4.3-9)
2 + 2( 1 )
2
+ 3 ( 1 )
2
= 6
We now look for a second vector in 6-space all of whose components are
equal to 1 which is orthogonal to r
1
The components of such a vector must
consist of three + 1 's and three -1 's. Since x(E) must always be positive and
since all elements in the same class must have representations with the same
character, the only possibility he re is
1
1
1
-1
Now our third representation will be of dimension 2. Hence x:.(E) = 2.
In arder to find out the values of z
3
(C
3
) and x
3
(a,.) we make use of the
orthogonality relationships (rule 3, 4.3-7):
2: x,(R)x3(R) = [1][2] + + 3[I][x3(a,.)] =o
R
2: X1(R)xJ(R) =.[1][2) + 2[I][x:t(CJ)] + 3[ -I][xJ(a,,)] = O
R
Solving these, we obtain
2x3( C3) + 3x3(a,.) - -2
- [2x3( C3) - 3x3( a,.) - -2]
6x3(ar) - o
XJ(a,.) - o
and
2x3( C3) + 3(0) = -2
xiCJ) = -1
Thus the complete set of characters of the irreducible representations is
E 2C3 3a,.
r, 1 1 1
r1
1 1 -1
r3 2 -1 o
We m ay note that there is still a check on the correctness of r
3
: the square
of the length of the vector it defines should be equal to h (rule 2), and we
see that this is so:
An lmportant Practical Relationship
We shall conclude this section by deriving a relationship between any reducible
representation of a group and the irreducible representations of that group.
In terms of practical application of group theory to molecular problems, this
relationship is of pivotal importance. We know already that for any reducible
representation it is possible to find sorne similarity transformation which will
reduce each matrix to one consisting of blocks along the diagonal, each of
which belongs to an irreducible representation of the group. We also know
that the character of a matrix is not changed by any similarity transformation.
Thus we may express x(R), the character of the matrix corresponding to
operation R in a reducible representation, as follows
x(R) = 2: aX(R)
j
( 4.3-10)
where a represents the number of times the block constituting the jth irre-
ducible representation will appear along the diagonal \Vhen the reducible
representation is completely reduced by the necessary similarity transfor-
mation. Now we do not need to bother about the difficult question of how
to find out what matrix is required to reduce completely the reducible rep-
resentation in arder to find the values of the a
1
We can obtain the required
relationship by working only with the characters of all representations in the
following way. We multiply each side of 4.3-10 by X;(R) and then sum each
88 PRINCIPLES
si de over all operations. namely,
2: x(R)x;(R) = 2: 2: a,x(R)x;(R)
R R 1
= 2: 2: aX(R)x;(R)
i R
Now for each of the terms in the sum over j, we have from 4.3-8
2: aX(R)x;(R) = a 2: X(R)x;(R) = alu5;
R R
since the sets of characters x,(R) and l.;(R) define orthogonal vectors, the
squares of whose lengths equal h. Thus, in summing over all j. only the sum
over R in which i = j can survive, and in that case we have
2: x( R)x;( R) = ha
R
which we rearrange to read
1
a; = -
1
2: x(R)x;(R)
1 R
(4.3-11)
Thus we ha ve an explicit expression for the number of times the ith irreducible
representation occurs in a reducible representation where we know only the
characters of each representation.
1/lustrative Examples
Let us take an example. For the group C
3
l. we give below the characters of
the irreducible representations. r,, r2, and r3, and the characters of two
reducible representations, r,, and rb
C3v
E 2C3 3a,.
r.
1 1 1
r2
1 l -1
r3
2 -1 o
fa
. 5
2 -1
rb 7 1 -3
Using 4.3-11, we find for fa
and for rb
a
1
= 1[1(1)(5) + 2(1)(2) + 3(1)( -1)] = 1
a
1
= 1[1(1)(5) + 2(1)(2) + 3( -1)( -1)] = 2
a
3
= 1[1(2)(5) + 2(-1)(2) + 3(0}(-l)] = l
a
1
= ![1(1)(7) + 2(1)(1) + 3(1}(-3}] =O
a
2
= ![1(1)(7) + 2(1)(1) + 3( -1)( -3)] = 3
a
3
= '[1[(2)(7) + 2( -1)(1) + 3(0)( -3)] = 2
The numbers in italics are the numbers of elements in each class. The results
obtained above will be found to satisfy 4.3-10. as of course they must. For
rll we have
E 2C
3
3a,.
r. 1 l 1
r2 1 1 -1
r2
1 1 -1
r:\ 2 -1 o
fa
5 2 -1
and for rb
E 2C
3
3a,,
r2
1 1 -1
r2
1 1 -1
r2 1 1 -1
r3
2 -1 o
r3 2 -1 o
r/J 7 1 -3
lndeed, in simple cases, a reducible representation may often be reduced
very quickly by using 4.3-10, that is. by looking for the rows of characters
which add up to the correct total in each column. For more complicated cases.,
it is usually best to use 4.3-11; however, 4.3-10 then provides a valuable check
on the results.
90 PRINCIPLES
4.4 CHARACTER TABLES
Throughout all of our applications of group theory to molecular symmetry
we will utilize devices called character tables. A set of these for all symmetry
groups likely to be encountered among real molecules is given in Appendix
IIA. * In this section we shall explain the meaning and indicate the source of
the information given in these tables. For this purpose we shall examine in
detail a representative character table, one for the group C
3
,., reproduced
below. The fourt m a in are as of the table ha ve be en assigned Roman numerals
for reference in the following dicussion.
C:"'
E 2C:l
3a,,
A, 1 1 1 z
x2 + y2, z2
A2
1 1 -1
R:
E 2 -1 o (x, y )(Rx, R_,.) (x
2
- y
2
, xy)(xz, yz)
Il l 111 IV
In the top row are these entries: In the upper left comer is the Schonflies
symbol for the group. Then, along the top row of the main body of the table,
are listed the elements of the group, gathered into classes; the notation is the
kind explained in Section 3.13. ,
Area l. In area 1 of the table are the characters of the irreducible repre-
sentations of the group. These ha ve been fully discussed in preceding sections
of this chapter and require no additional comment here.
Area.ll. We have previously designated the ith representation, or its set
of characters, by the symbol r in a fairly arbitrary way. Although this practice
is stil1 to be found in sorne places and is common in older literature, most
books and papers-in fact, virtually all those by English-speaking authors-
now use the kind of symbols found in the C
3
" table above and all tables in
Appendix 11. This nomenclature was proposed by R. S. Mulliken, and the
symbols are normally called Mulliken symbols. Their meanings are as follows:
l. All one-dimensional representations are designated either A or B; two-
dimensional representations are designated E; three-dimensional species are
designated T ( or sometimes F).
2. One-dimensional representations that are symmetric with respect to
rotation by 2n/ n about the principal C" axis [ symmetric meaning: x( C,) = 1]
Appendix IIA will be found as a separate booklet in a pocket in the back of this book.
t Sometimes areas III and IV are combined.
REPRESENTATTONS OF GROUPS 91
are designated A, while those antisymmetric in this respect [x( C,) = - 1]
are designated B.
3. Subscripts 1 and 2 are usually attached toA 's and B's to designa te those
which are, respectively, symmetric and antisymmetric with respect to a C
2
perpendicular to the principal axis or, if such a C
2
axis is lacking, to a vertical
plan e of symmetry.
4. Primes and double primes are attached to allletters, when appropriate,
to indicate those which are, respectively, symmetric and antisymmetric with
respect to a
1
,.
5. In groups with a center of inversion, the subscript g (from the German
gerade, meaning even) is attached to symbols for representations which are
symmetric with respect to inversion and the subscript u (from the German
ungerade, meaning uneven) is used for those which are antisymmetric to
inversion.
6. The use of numerical subscripts for E's and Ts also follows certain
rules, but these cannot be easily stated precisely without sorne mathematical
development. 1t will be satisfactory here to regard them as arbitrary labels.
Area 111. In area 111 we will always find six symbols: x, y, z, Rx, R_v, R=.
The first three represent the coordina tes x, y, and z, while the R's stand for
rotations about the axes specified in the subscripts. We shall now show in an
illustrative but by no means thorough way why these symbols are assigned
to certain representations in the group C
3
,,, and this should suffice to indicate
the basis for the assignments in other groups.
Any set of algebraic functions or vectors may serve as the basis for a
representation of a group. In order to use them for a basis, we consider them
to be the components of a vector and then determine the matrices which
show how that vector is transformed by each symmetry operation. The re-
sulting matrices, naturally, constitute a representation of the group. We ha ve
previously used the coordina tes x, y, and z as a basis for representations of
groups C
2
,. (page 78) and T (page 74). In the present case it will be easily
seen that the matrices for one operation in each of the three classes are as
follows:
E
[
cos 2n/3
sin 27!/3
o
Now the first thing we can observe about these matrices is that they never
mix z with x or y; that is, z' is always a function of z only. Hence z by itself
forms an independent representation of the group. On the other hand C
3
mixes up x and y to give x' and y', so x and y jointly form a representation.
92 PRINCIPLES
This is equivalent to observing that the three matrices are all block-factored
in the same way, namely. into the following submatrices:
E

[
cos 2n/3
sin 2n/3
-sin 27r/3]
cos 2n/3
1
We see that f: is the A
1
irreducible representation. This means that the
coordinate z forms a basis for the A
1
representation, or, as we also say, "z
transforms as ( or according to) A
1
., If we examine the characters of r.r.y' we
find them to be those of the E representation (2 cos 2n/3 = - 1 ), so that the
coordina tes x and y together transform as or according to the E representation.
It is important to grasp that x and y are inseparable in this respect, since the
representation for which they form a basis is irreducible.
A thorough treatment of how the transformation properties of the rotations
are determined would be an unnecessary digression from this discussion. In
simple cases we can obtain the answer in a semipictorial way by letting a
curved arrow about the axis stand for a rotation. Thus such an arrow around
the z axis is transformed into itself by E, it is transformed into itse)f by C
3
,
and its direction is reversed by a, .. Thus it is the basis for a representation
with the characters 1, 1, -l. and so we see that R:: transforms as A
2

Area IV. In this part of the table are listed aJI of the squares and binary
products of coordinates according to their transformation properties. These
are quite easy to work out using the same procedure as for x. y, and
z, except that the amount of algebra generally increases, though not always.
Por example, the pair of functions xz and yz must have the same transfor-
mation properties as the pair x, y, sin ce z goes into itself under all symmetry
operations in the group. Accordingly1 (xz. yz) are found opposite the E
representation.
The Character Table for D
4
: An lllustration of all the Concepts in Sections
4.1-4.4.
l. The pure rotation group D
4
consists of all operations generated by the
symmetry elements C
4
and four C
2
axes perpendicular to the C
4
We may
place the C
4
axis in the Z direction, tWO of the C2 axes along X and y, and the
other axes, C's bisectng the angles between the C
2
's, as shown below. The
reader should be able to confirm that this group comprises the following eight
operations:
E C4 C3 C2(x) Cly) C(xy) Ci(xy)
This list may be rewritten with the operations collected into five classes:
E 2C.a C
2
(z) = Cl 2C2 2C2
2. To produce the character table for this group we as k: (a) How many
irreducible representations? (b) Of what dimensions? Rule 5 tells us that since
z
1
c4

C2(y)
----....,.--Cix)-+ X
y

.
there are five classes there will be five irreducible representations. The di-
mensions of these, /
1
, 1
2
, , 1
5
, must o bey rule 1, namely,
Ir + li + 15 + 1 + = h = s
The only set of positive integers that satisfies this requirement is 1, 1, 1,
1, 2.
3. It is relatively easy to deduce the four one-dimensional representations.
As in every group, there must be the so-called totally sym1netric representation,
in which every symmetry operation is represented by the one-dimensional
matrix l. At this point, we have in hand the following part of the character
table:
E 2C4 C
2
(z) 2C2 2Ci
1 1 1 1 1
1
1
1
2
4. The characters of the remaining three one-dimensional representations
may be obtained by the orthogonality requirements, rule 3. To make a rep-
resentation orthogonal to the first one we shall have to assign + 1 to the
operation C
2
(z) and to those in one of the other classes, with - 1 being
assigned to the remaining two classes. Thus, we have
E 2C4 C2{z) ZC1 2C2
1 1 1 1 1
1 1 1 -1 -1
1 -1 1 1 -1
1 -1 1 -1 1
2
94 PRINCIPLES
I t can be shown that all other possible pairs of these are orthogonal, for
example, the next to last and the last:
(1)(1) + 2( -1)( -1) + (1)(1) + 2(1)( -1) + 2( -1)(1) -
1 + 2 + 1 2 2 =0
5. To find the two-dimensional representation, we might take a point (x,
y, z) and examine the matrices describing its transformation by one operation
in each class. A little prior reflection will show that we can omit z because it
is carried only into 1 times itself by all operations and thus forms a basis
for one of the one-dimensional representations. (The reader should be able
to see, without pencil and paper, that this is the second one.) Working, then,
only with a point (x, y) in the xy plane, the matrices and their characters are
as follows:
E
[ ~ ]
X= 2 C
2
(x)
[
-n
X= O
C4(z)
[ ~ ]
X= O C2(xy)
[ ~ ]
X =O
Cz(z) [ -
-n
X= -2
Since these matrices are not all blocked out, this representation cannot be
separated into two one-dimensional representations. In addition, we can easily
see that it is normalized (i.e., obeys rule 2., 4.3-6):
2 ~ + 2(0)
2
+ (- 2)
2
+ 2(0)
2
+ 2(0)2 = 8
Moreover, its set of characters is orthogonal to those of all four one-dimen-
sional representations, for example,
(1)(2) + 2(-1)(0) + (1)(-2) + 2(-1)(0) + 2{1)(0) =o
Hence, we have found the two-dimensional irreducible representation and
we may add its characters to the table.
6. We can also assign Mulliken symbols to each representation. The first
two, being symmetric with respect to C
4
are A
1
and A
2
while the next two
are 8
1
and 8
2
The unique two-dimensional representation is called E. We
ha ve now developed the table to the extent shown below.
D4
E 2C-t C2(z) ZC2 2 C ~
A, 1 1 1 1 1
Az
1 1 1 -1 -1
B.
1 -1 1 1 -1
82 1 -1 1 -1 1
E 2 o -2 o o
7. The remaining task is to assign the Cartesian coordina tes, the rotations
and certain algebraic functions of Cartesian coordinates to their representa-
tions. We ha ve already se en that z belongs to the A
2
representation ( or, as
is commonly said, "has A
2
symmetry') and x and y jointly belong to the E
representation. By using curved arrows about the three Cartesian axes, it is
not difficult to determine the symmetries of the rotations.
The functions of Cartesian coordinates normally included in character ta-
bles are those associated with the d orbital wave functions, namely z
2
x
2
+
y
2
, x
2
- y
2
, xy, xz, and yz. Sin ce we already know how each of the individual
coordinates behaves, it is simple to work out the results for these combina-
tions. To illustrate let us take x
2
- y
2
:
e4 This causes X ~ y and y ~ -X
C
2
(z) This causes x ~ -x and y ~ -y
Cz(x) This causes x ~ x and y ~ -y
C
2
(xy) This causes x ~ y and y ~ .x
:. x2 _ y2 ~ y2 _ x2 = _ (x2 _ y2)
:. x2 _ y2 ~ x2 _ y2
1 7 , 7
:. x- - y - ~ x- - y-
:. x2 _ y2 ~ y2 _ x2 = _ (x2 _ y2)
The function x
2
- y
2
therefore has the following transformation properties
under the symmetry elements of the group in the order E e: 1, -1, 1,
1, - l. It is therefore assigned to the B
1
representation. It is left as an exercise,
which can be done in one 's head, to assign the other five functions listed
above. The results can be checked using the D
4
character table in Appen-
dix 11.
4.5 REPRESENTATIONS FOR CYCLIC GROUPS
As noted earlier, a cyclic group is Abelian, and each of its h elements is in
a separate class. Therefore, it must have h one-dimensional irreducible rep-
resentations. To obtain these there is a perfectly general scheme which is
perhaps best explained by an example. lt will be evident that the example
may be generalized. Let us consider the group e
5
, consisting of the five
commuting operations e
5
, e ~ e ~ e ~ e ~ = E; we seek a set of five one-
dimensional representations, f
1
' f
2
' f
3
' f
4
' f
5
, which are orthonormal in the
sen se
m=S
2: [fP( e:r) )[fq( e:r)] * = hJpq (4.5-1)
m=l
We shall use the exponentials
exp(2nip/5) = cos 2np/5 + i sin 2np/5 (4.5-2)
96 PRINCIPLES
as the fP(C
5
). Abbreviating these exponentials as eP [i.e., e = exp(2n:i/5)],
we write the first column of the following table:
Cs
e;!
5
C3
5
~
'
es-
5 -
E
r
..,
e3 e4 es
e e-
f2
..,
e4 ;6 es ew
e-
f3 eJ eb e9 eJ2 els
f4 e4 es el2
e'n
e2o
rs eS ew
ets e2o
e
25
The remaining columns follow from the group multiplications. It will now be
shown that these representations satisfy the orthonormalization condition of
4.5-1.
Consider any two representations, say fP and rq' where q - p = r. The
left-hand side of 4.5-1 takes the form
This may be rewritten as
Since for exponentials in general
expression 4.5-3a reduces to the relatively simple sum
(4.5-3b)
It is then clear that the representations are normalized, because if fP =
rq, r = O and 4.5-3b is simply five times e
0
= 1, namely, 5.
If fP and rq are different, r is sorne number from 1 to 4. Because e
5
= 1
(see 4.5-2), we have such equalities as
(4.5-4)
Therefore any sum of the type 4.5-3b reduces to
n=5
e + e
2
+ e
3
+ e
4
+ e
5
= 2: exp(2n:in/5)
(4.5-5)
n=l
Because of the trigonometric identities
n=l
L cos 2nnll = O
n=l
n=l
L sin 2nnll = O
11=1
it follows that the sum in 4.5-5 equals O, thus proving the orthogonality of
the representations.
If we now replace all tn's, such as e
5
, e
10
, e
15
, , which are equal to unity
by the number 1, and reduce all other exponents in excess of 5 to their lowest
values as indicated in 4.5-4, we can rewrite the table in the following form:
e-
"
e2
5
e ~
J
C4
5
e ~
fl
e
e2 e3 e4
1
f2
.,
e4 e3
1 e-
e
f3 e3
e
e4
..,
e- 1
f4 e4 e3
.,
1 e- e
rs
1 1 1 1 1
Inspection of ( 4.5-2) wi11 show that t\VO ei's whose exponents add up to 5
are complex conjugates of each other. For example,
(e
4
)* = (cos 2n! + i sin 2n,)* = cos 2n! - i sin n ~
= cos n ~ + i sin 2nA = e
We now rewrite the table once more, replacing e
3
by (e
2
)* and e
4
by e*
and aJso rearranging the rows and colun1ns, and obtain
New Old
Designation E Cs
e!
5
eJ
5
e ~
Designation
A 1 1 1 1 1 r
{
1
.,
e2* e*
fl
E,
e e-
1
e* e2*
.,
f4
e- e
{
1
.,
e*
el* f2
E2
e- e
1
el*
e*
..,
rJ
f. e-
The new arrangement of columns is such as to place the e ~ = E column
first. The new ordering of rows places the totally symmetric representation
at the top; it is denoted A. The remaining representations are then associated
98 PRINCIPLES
in pairs, such that the elements of one row in each pair are the complex
conjugates of the other. Each member of these pairs must be regarded as a
separate representation in order to satisfy the requirements as to the number
and dimensions of the irreducible representations. However, for certain ap-
plications to physical problems we shall prefer to add them, thus obtaining a
set of characters for a representation of dimension 2. Beca use we are always
adding a complex number to its complex conjugate .. the sums are invariably
pure real numbers. Because of this property of adding to give characters like
those for a t wo-dimensional representation, the pairs of representations are
jointly denoted by the Mulliken symbol E.
Let us take the group C
3
as an example. Its correct character table is, in
part,
A
E
1
g
1
e
e*
e = exp(2ni/3)
and when we combine the two parts of the "E representation" we obtain
A 1 1 1
E 2 2 cos 2n/3 2 cos 2n/3
To show that the table in this form is serviceable for physical problems, Jet
us work out the transformation properties of the x, y .. and z coordina tes in
the same way as we did above for the group C
3
, Here we obtain the matrices
E
. [cos 2n/3
sin 2n/3
o
-sin 2n/3 o ~
cos 2n/3
o
[
cos 4n/3
sin 4n/3
o
-sin 4n/3 o ~
cos 4n/3
o
Again we see that these matrices give a reducible representation which can
be reduced on inspection to two representations having the following char-
acters:
E c3
Cj
r=
1 1 1
r.t._,. 2 2 cos 2n3 2 cos 4n3
Thus x and y transform according to E, and z transforms according to A
1
,
which are the results stated in the complete form of the mathematically correct
table. (Note: cos 4n/3 = cos 2n/3.)
EXERCISES
4.1 Using matrix methods, verify all the rules concerning the products and
the commutation of symmetry operations given on pages 33-34.
4.2 Derive the complete matrices for all representations of the group C
3
, .
Hint: Write out and reduce matrices for expressing the transformations
of the general point (x, y, z), and use matrix multiplication.
4.3 Using matrix multiplication, carry out the required similarity transfor-
mations to determine the arrangement of the twelve operations in the
group Tinto classes.
4.4 For the group O show that the sets of functions (x, y, z), and (xy, xz,
yz), correspond to the irreducible representations T
1
and T
2
, respec-
tively.
4.5 Prove that all irreducible representations of Abelian groups must be one
dimensional.
4.6 Work out the character table for the group C
4
Present it in the con-
ventional format, reducing all exponentials to their simplest form.
4. 7 Consider the group D
3
Let the C
3
axis coincide with the z axis and one
of the C
2
axes coincide with the x axis. Write out the complete matrices
for all irreducible representations of this group. Derive from these the
character table.
4.8 Write the matrices describing the effect on a point (x, y, z) of reflections
in vertical planes which lie halfway between the xz and yz planes. By
matrix methods determine what operations result when each of these
reflections is followed by reflection in the xy plane.
5
GROUP THEORY AND
QUANTUM MECHANICS
5.1 WA VE FUNCfiONS AS BASES FOR
IRREDUCIBLE REPRESENTATIONS
lt will be appropriate to give first a brief discussion of the wave equation. lt
is not necessary that the reader have any very extensive knowledge of wave
mechanics in order to follow the development in this chapter, but the infor-
mation given he re is essential. The wave equation for any ph ysical system is
.ll''l' = E'l! (5.1-1)
He re ,/t'is the Hamiltonian operator, which indica tes that certain operations
are to be carried out on a function written to its right. The wave equation
states that, if the function is an eigenfunction, the result of performing the
operations indicated by l f w i l 1 yield the function itself multiplied by a constant
that is called an eigenvalue. Eigenfunctions are conventionally denoted '1',
and the eigenvalue, which is the energy of the system, is denoted E.
The Hamiltonian operator is obtained by writing down the expression for
the classical energy of the system, which is simply the sum of the potential
and kinetic energies for systems of interest to us, and then replacing the
momentum terms by differential operators according to the postula tes of wave
mechanics. We need not concern ourselves with this construction of the Ham-
iltonian operator in any detail. The only property of it which we shall have
to use explicitly concerns its symmetry with respect to the interchange of like
particles in the system to which it applies. The particles in the system will be
electrons and atomic nuclei. lf any two or more particles are interchanged
by carrying out a symmetry operation on the system, the Hamiltonian must
100
GROUP THEORY AND QUANTUM MECHANICS 101
be unchanged. A symmetry operation carries the system into an equivalent
configuration, which is, by definition, physically indistinguishable from the
original configuration. Clearly then, the energy of the system must be the
same before and after carrying out the symmetry operation. Thus we say that
any symmetry operator, R, commutes with the Hamiltonian operator. and
we can write
(5.1-2)
The Hamiltonian operator also commutes with any constant factor e. Thus
)'f\}1 = c.li''l' = cE'I' (5.1-3)
lt has been more or less implied up to this point that for any eigenvalue
E; there is one appropriate eigenfunction '1';. This is often true, but there are
also many cases in which severa) eigenfunctions give the same eigenvalue,
for example,
Yf''l! . ., = E-'1!-., ,_ 1 ,_
(5.1-4)
In such cases we say that the eigenvalue is degenerate, and in the particular
example given above we would say that the energy. E; is k-fold degenerate.
Now in the case of a degenerate eigenvalue, not only does the initial set of
eigenfunctions provide correct solutions to the wave equation, but any linear
combination of these is also a solution giving the same eigenvalue. This is
easily shown as follows:
= E;ail'l';
1
+ E;a;
2
'1!;
2
+ + E;a;k'l!;k
= E; 2: a;'Y;i
(5.1-5)
One more important property of eigenfunctions must be mentioned. Ei-
genfunctions are so constructed as to be orthononnal, which means that
J
'{!:'='{!. dr = b
1 J IJ
(5.1-6)
where the integration is to be carried out over all of the coordinates, collec-
102 PRINCIPLES
tively represented by r, which occur in 'l'; and '1'. When an eigenfunction
belonging to the eigenvalue E; is expressed as a linear combination of a set
of eigenfunctions, we have. from 5.1-6,
Now all products \vhere j #- j' will vanish, for example,
I
a. 1 ~ U ~ dr = a a' J 1 ~ 1 dr = O
IJ 1/ IJ J} 1} IJ IJ IJ
and, assuming that each '1';
1
is normalized, we are left with
J 2: a;'l'ta;'l'; dr = 2: a;/ = 1
i j
(5.1-7)
We can now show that the eigenfunctions for a molecule are bases for
irreducible representations of the symmetry group to which the molecule
belongs. Let us take first the simple case of nondegenerate eigenvalues. If
we take the wave equation for the molecule and carry out a symmetry op-
eration, R, upon each side, then, from 5.1-1 and 5.1-2 we have
(5.1-8)
Thus R'l'; is itself an eigenfunction. Since '1'; is normalized, we must require,
in order that R '1'; al so be normalized,
R'l'; = 1 '1';
Hence, by applying each of the operations in the group to an eigenfunction
'1'; belonging to a nondegenerate eigenvalue, we generate a representation
of the group with each matrix, f;(R), equal to l. Since the representations
are one dimensional, they are obviously irreducible.
If we take the wave equation for the case where E; is k-fold degenerate.,
then we must write, in analogy to 5.1-8.
(5.1-9)
But he re R'l'u m ay in general be any linear combination of the '1';
1
, that is,
k
R 'l'u = 2: r
1
,'1'; (5.1-10)
= l
For sorne other operation, S, we ha ve similarly,
k
S'l'; = L Sm'l';m
(5.1-11)
m=l
Because R and S are members of a symmetry group, there must be an element
T = SR; its effect on 'l'u can be expressed as
k
TfJfu = L lmllflim
(5.1-12)
m=l
However, by combining the preceding expressions for the separa te effects of
S and R, we obtain
k k k
SR f/1,, = S rf/1; = 2: Smtlfl;m
j=l j=l m=l
Comparing 5.1-12 with 5.1-13, we see that
k
lml = 2: Smrjl
i=l
(5.1-13)
But this is just the expression that gives the elements of a matrix . ~ 7 which
is the product .1 /1? of two other matrices. Thus the matrices that describe the
transformations of a set of k eigenfunctions corresponding to a k-fold de-
generate eigenvalue are a k-dimensional representation for the group. More-
over, this representation is irreducible. If it were reducible we could divide
the k eigenfunctions 'l';, '1';
2
, , 'l';k, or k linear combinations thereof, up
into subsets such that the symmetry operations would send one member of
the subset into a linear combination of only members of its own subset. Then
the eigenvalue for members of one subset could be different from the eigen-
value for members of another subset. But this contradicts our original as-
sumption that all of the 'llu must have the same eigenvalue.
To illustrate this explicitly, Jet us consider the 2px and 2p_,, orbitals of the
nitrogen atom in ammonia, which belongs to the group C
3
v. These orbitals
are represented or described by the following eigenfunctions:
Px = '11 r sin O cos </>
p-" = 'I'r sin O sin 4J
where '1', is angle independent and thus a constant insofar as symmetry op-
erations are concerned; we shall therefore omit it henceforth. O and denotes an angle
measured in the counterclockwise direction from the x axis in the xy plane.
Let us now work out the matrices that represent the transformations of these
functions by each of the symmetry operations in the group C
3
t We consider
what happens to a line whose direction is fixed initially by the angles 0
1
and
</J
1
First of all we note that none of the operations in the group will affect O.
so that 0
1
, the value of O after application of a symmetry operation .. will always
equal 0
1
Hence
sin 0
2
= sin O,
If we rota te by 2n/3 about the z axis, however, we ha ve
= c/J
1
+ 2n/3
and hence
cos 41
2
= cos ( c/J
1
+ 2n/3) = cos l/>
1
cos 2n/3 - sin 41
1
sin 2n/3
= -! cos 41
1
- (v'3/2) sin c/>
1
sin 41
2
= sin ( </>
1
+ 2n/3) = sin c/J
1
cos 2n/3 + cos </J
1
sin 2n/3
= -l sin </J
1
+ (v'3/2) cos </J
1
If we reflect in the xz plan e, we ha ve
and hence
E:
cos c/>! = cos cjJ,
sin c/>
2
= -sin </J
1
We can now use use this information to out the required matrices.
Ep.'C = E(sin O, cos </J) = sin 0
2
cos c/J
2
= sin O, cos cf>t = P.'(
Ep_r = E (sin 0
1
sin </J
1
) = sin 0
2
sin </>
2
= sin 0
1
sin </>
1
= P.v
C
3
px = C
3
(sin 0
1
cos </>
1
) = sin 0
2
cos </>
1
- (sin 0
1
)(- !)( cos </J
1
+ v'3 sin </JJ)
= - sin 0
1
cos c/J
1
- (v'3/2) sin O, sin <f>,
- (Y3/2)p_,.
C
3
p-" = C
3
(sin 0
1
sin c/>
1
) = sin 0
1
sin </>2
= (sin OJ)(- !)(sin </J
1
- v'3 cos </Jd
= (V3/2) sin 0
1
cos c/>
1
- ! sin 0
1
sin c/>
1
= (Y3/2)px - !p_,.
a,.p, = a,.(sin 0
1
cos c/>
1
) = sin 0
2
cos c/>
2
= sin O, cos cP1 = Px
a,.p_,. = a,.(sin 0
1
sin </J
1
) = sin 0
2
sin 41
1
= -sin 0
1
sin cp, - - p_,.
Expressing these results in matrix we write
[
1 O] [P.r] = E [Px]
O 1 p)' P.v
x(E) = 2
- [P.r] = c
3
[P.t]
2 p_.., p'}'
[
1 O] [Px] = a,.[P.r]
O -1 p.\. P.r
The characters are seen to be those of the E representation of C
3
, . Thus we
see that the p_, and p-" orbitals, as a pair, provide a basis for the E represen-
tation. It will be noted that the coordinates x and y are shown as transforming
according to the E representation in the character table for the group C
3
, .
Thus the functions sin O cos </> and sin O sin </> transform in the same way as
x and y. For this reason the p orbital that has an eigenfunction sin O cos 4> is
called Px and the one that has an eigenfunction sin O sin e/> is called p, ..
5.2 THE DIRECT PRODUCT
Suppose that R is an operation in the symmetry group of a molecule and X
1
,
X2, ... , Xm and Y
1
, Y
1
, , Yn are two sets of functions (perhaps eigen-
functions of the wave equation for the molecule), which are bases for re-
presentations of the group. As shown earlier, we may write
It is also true that
,
RX- = "XX
1 L., ,, 1
j ... l
"
RYk = 2: y,k Y,
1= 1
, "
RX;Yk = 2: 2: Y, = 2: Lz,,_;kX;Y,
=11=1 j 1
106 PRINCIPLES
Thus the set of functions X; Y k, called the direct product of X; and Y k, al so
forms a basis for a representation of"the group. The z
1
,;k are the elements of
a matrix .1? of order (mn) x (mn ).
We now have a very important theorem about the characters of the .Y.
matrices for the various operations in the group:
The characters of the representation of a direct product are equal to the
products of the characters of the representations based on the individual sets
of functions.
PROOF: This theorem is easily proved as follows:
m "
~ (R} = 2: Zt,t = 2: L XHYII = X" (R)x ,, (R)
ji j=J 1=1
Thus .. if we want to know the characters x(R) of a representation that is
the direct product of two" other representations with characters x
1
(R) and
x
2
( R), these are given by
(5.2-1)
for each operation R in the group.
For example, the direct products of sorne irreducible representations of
the group c4,. are as fo11ows:
c.".
E e:! 2C4
?
_(]\,
2ad
A, 1 1 1 1 1
A1 1 1 1 -1 -1
B, 1 1 -1 1 -1
8.2 1 1 -1 -1 1
E 2 -2 o o o
A
1
A
1
1 1 1 -1 -1
B
1
E 2 -2 o o o
A
1
EB
2
2 -2 o o o
E2
4 4 o o o
It should be clear from the associative property of matrix multiplication
that what has been said regarding direct products of two representations can
be extended to direct products of any number of representations.
In general, though not invariably, the direct product of two or more ir-
reducible representations will be a reducible representation. For example,
GROUP THEORY ANO QUANTUM MECHANICS 107
the direct product representations of the group given above reduce in the
following way:
A,A
2
= A
2
B
1
E =E
A
1
EB'!. =E
How Direct Products Are Used
We shall now explain the importance of direct products in the solution of
problems in molecular quantum mechanics. Whenever we have an integral
of the product of two functions, for example,
the value of this integral will be equal to zero unless the integrand is invariant
under all operations of the symmetry group to \Vhich the molecule belongs
or unless sorne term in it, if it can be expressed as a sum of terms, remains
invariant. This is a generalization of the familar case in which an integrand
is a function of only one variable. In that case, if y = .f (x), the integral
equaJs zero if y is an odd function, that is, if f(x) = - f(- x). In this simple
case we say that y is not invariant to the operation of reflecting all points in
the second and third quadrants into the first and fourth quadrants and vice
versa.
Now when we say that the integrand .f Af
8
is invariant to all symmetry
operations, this means that it forros a basis for the totally symmetric repre-
sentation. But from what has been said above, we know how to determine
the irreducible representatOnS OCCUfring n the representatOn f AB for Which
f_.,J R forroS a baSS if We knOW the irreducible representatiOOS for Which .fA
and f
8
separately form bases. In general:
fAB = a SUffi Of irreducible representatOOS
Only if one of the representations occurring in the sum is the
totally symmetric one will the integral have a value other than zero. There
is a theorem concerning whether the totally symmetric representation will be
present in this sum:
The representar ion of a direct product, r Afl, will containthe total/y synunetric
representatOfl 011/y if fhe irreducible fA = the irreducible f R
108 PRINCIPLES
PROOF: Equation 4.3-11 tells us that the number of times, a;, the ith irre-
ducible representation, f,,. OCCUfS n a reducible representation. say fAB' S
given by
(5.2-2)
lf a
1
and x
1
refer to the totally symmetric representation, for which all x
1
(R)
equal 1, we ha ve
But from 5.2-l
hence
1
a, = -
1
2: XAB(R)
l R
1
a, = h 2: XA(R)xB(R)
R
(5.2-3)
According to the properties of characters of irreducible representations as
components of vectors ( 4.3-8), we obtain
Q.E.D.
We also see that if f
1
occurs at all it will occur only once. It is very easy to
check this theorem by using the character tables in Appendix IIA, as the
following illustrative example shows. Many other examples may be selected
by the reader to develop familiarity with the manipulation of direct products.
11/ustrative Example
Group D
4
The character table f9r this group should be very familiar, sin ce it was
derived in toto earlier. lt should be obvious that the direct product of any
two different irreducible representations must yield a representation contain-
ing two -1 's, which will not, therefore be an A
1
representation. It is also
apparent that no direct product of A
1
, A
2
, B,, or ~ with E can be other than
E. The only remaining direct product is E x E. This has the characters
4 o 4 o o
lt should contain the A
1
representation, and it does:
aA, = i (4 + O + 4 + O + O] = 1
From the foregoing discussion of the integrals of products of t\vo functions
it is easy to derive sorne important rules regarding integrands that are products
of three, four or more functions. The case of a tripl_e product is of particular
importance. In order for the integral
to be nonzero, the direct product of the representations of fA, fn, and fe
must be or contain the totally symmetric representation. That this can happen
only if the representation of the direct product of any two of the functions is
or contains the same representation as is given by the third function follows
directly from this theorem. This corollary is applicable chiefly in dealing with
matrix elements of the type
J lf/;Pif/
1
dr
(5.2-4)
where tJ; and 11
1
are wave functions, and P is sorne quantum mechanical
operator.
5.3 IDENTIFYING NONZERO MATRIX ELEMENTS
Integrals of the general type 5.2-4 occur frequently in quantum mechanical
problems. They are often termed 1natrix ele1nents, sin ce they occur as su eh
in the secular equations which commonly provide the best way of formulating
the problem (see Chapters 7, 8, and 10 for examples of secular equations).
In order to give the results just presented in Section 5.2 sorne concrete mean-
ing, we shall discuss here the two commonest examples of the type of matrix
element represented in 5.2-4.
Energy Elements
If we take the wave equation
(5.3-1)
left-multiply it by 11,, integrate both si des, and rearrange slightly, we obtain
f tJ;H 11
1
dr = E
f 11, 11
1
dr
(5.3-2)
110 PRJNCIPLES
We thus ha ve an explicit expression for an energy, which can be thought of
as the energy of interaction between two states described by the wave func-
tions lfl; and f/1,. lf the integral that occurs in the numerator of the left-hand
side of this equation is in fact required to have a value identically equal to
zero, it will be helpful to know this at the earliest possible stage of a calculation
so that no computational effort will be wasted on it. This information may
be obtained very simply from a kno\vledge of the irreducible representations
to which the wave functions 11, and tJ belong.
We note first that the Hamiltonian operator must have the full symmetry
of the molecule; it is simply an operator expression for the energy of the
molecule, and clearly the energy of the molecule cannot change in either sign
or magnitude as a result of a symmetry operation. The Hamiltonian operator
then belongs to the totally symmetric representation, and the symmetry of
the en tire integrand tJ;H tJ depends entirely on the representations contained
in the direct product of the representations of lfl; and lf/j The totally symmetric
representation can occur in this direct product representation only if tJ; and
11
1
be long to the same irreducible representation. Thus we ha ve the completely
general and enormously useful theorem that:
44n energy integral f 1./I;H 11
1
dr tnay be nonzero only if lfl, and 11
1
be long to
the sa1ne irreducible representation of the nzolecular point group.
Spectral Transition Probabilities
Perhaps the second commonest case in which the simple question of whether
or not a n1atrix element is required by syn1metry considerations to vanish
occurs in connection with selection rules for various types of transition from
one stationary state of a system to another with the gain or loss of a quantum
of energy. If the energy difference between the states is represented by
E; - E, then radiation of frequency t' will be either absorbed or emitted by
the transition, if it is aJiowed, with ,. being required to satisfy the equation
ht = E - E
1 J
In general the intensity, 1, of a transition from a state descri bed by lf/; to
another described by 11, is given by an equation of the type
1 "' J 11 Jllf!
1
d r
(5.3-3)
The symbol JI is a transition 1nonwnt operator, of which there are various
kinds, namely, those corresponding to changes in electric or magnetic dipoles ..
higher electric or magnetic multipoles, or polarizability tensors.
The con1monest type of transition, and the only one to be considered right
GROUP THEORY AND QUANTUI'vl MECHANICS 111
now, is the electric dipole-allowed transition. In this type the charge distri-
butions in the two states differ in a manner corresponding toan electric dipole.
Such a transition can therefore couple with electromagnetic radiation by in-
teraction with the oscillating electric vector and thereby transfer energy to or
from the electromagnetic field. Both theory and experiment show that nor-
mally this is by far the most powerful intensity-giving process, and the state-
ment that a transition is uallowed," without further qualification, means ex-
plicitly uelectric dipole aliO\\'ed. ''
The electric dipole operator has the form
t = 2: e,x; + e;y, + 2: e;z; (5.3-4)
; i i
where e; represents the charge on the ith particle, and X;, y;, and Z; are its
Cartesian coordinates. When this expression is introduced into 5.3-3, we
obtain a result which is usefully expressed as three separate equations beca use
of the orthogonality of the Cartesian coordinates:
lx <X J lf/XIf/ dr
ex J lf/; y tJ dr
1, ex J tJ; z lf/ dr
(5.3-5)
In these equations scalar quantities such as the e; have been omitted, and
summation over all particles is assumed.
These equations mean that the transition from the ith to the jth state ( or
the re verse) m ay acquire its intensity in any of three ways, namely, by in-
teracting with an electric vector oscillating in the x, the y, or the z direction.
lf it is only the integral f tJ;Xf/1; dr that is nonzero, we say that the transition
is polarized in the x direction or that it is x polarized. In cases of sufficiently
high molecular symmetry, where x and J' jointly form the basis for an irre-
ducible representation, the integrals involving x and y will not be independent
of each other. If both are nonzero, the transition is said to be xy polarized.
In problems with sorne kind of cubic or higher symmetry (point groups T,
T,, TJ, O, 0
1
,, /, /
11
), where the three Cartesian coordinates jointly forro a
basis for sorne three-dimensional representation, no polarization effect exists.
Radiation with an electric vector in any direction will excite the transition, if
it is allowed at all.
Thus the problem of deciding whether a certain transition is electric dipole
and what the polarization is. reduces to that of deciding -..vhich. if
any, of the three integrals in 5.3-5 are nonzero. We can always ascertain to
what representations the Cartesian coordina tes belong by inspection of the
112 PRINCIPLES
character table of the molecular point group. Then, according to the consid-
erations of Section 5.2, we have the following rule:
An electric dipole transition will be allowed with x, y, or z polarization if
the direct product of the representations of the two states con cerned is or contains
the irreducible representation to which y, or z, respective/y, belongs.
11/ustrative Example
Metal-Based Electronic Excitations of
Species
The
ions, for example,
have the square parallepiped

structure shown below .. and therefore belong to point group D
41
,. Their highest
occupied molecular orbital (HOMO) is a J orbital and the lowest unoccupied
molecular orbital (LUiviO) is a J* orbital. These two orbitals are composed
almost entirely of the atomic d.n orbitals, combined as shown belo\v.
z
l
Cl---------CI
1 ', .,.,"' 1
1 'Mo"' ,'
:,-1111' : y
CI--------- Cl /

1 ' .,, 1
1 'M ., ,
,' o :
"1
CI--------- Cl

J*
We can ask whether the --+ * transition of an electron is allowed and,
if so .. with what polarization (a coordinate system s shown). We n1ust first
determine the irreducible representations for \Vhich each of the orbitals., c5
and J*, forms a basis. This can easily be done using the D-t
1
, character table.
Each of these orbitals forms the basis for a one-dimensional representation
(and thus an irreducible one); when subjected to each of the symmetry op-
erations ( or one from ea eh class) the following results are obtained. 1 means
the orbital goes into itself (no signs changed) and - 1 means it goes into the
negative of itself (all signs changed):
E 2C4 c"1 2C2 2C2 i 2S4 , 2a,. 2ad
1 -1 1 1 -1 1 -1 1 1 -1
* 1 -1 1 -1 1 -1 1 -1 1 -1
Reference to the D., character table shows that these are of
and 8
1
,(l5*)
symmetries.
GROUP THEORY ANO MECHANICS 113
The intensity of a c5 6* transition will be governed by the magnitude of
the integral(s).
J 'l',,(x. y, z )'1',
1
- dr
Our task here is to determine whether any of the three Cartesian com-
ponents is nonzero. Since, in D
41
, the z vector transforms according to the
A:!u representation and x and y jointly according to the E, representation, we
need to know whether either of the direct products, B l.t: x A
1
, x 8
1
, or
x E, x 8!,, contains the A representation. It is a simple matter to show

that the first one is equal to A lg while the second is equal to Eg. Thus. the
(5 (5* transition is electric-dipole allowed \Vith z polarization and forbidden
for radiation \Vith its electric vector in the xy plane.
EXERCISES
5.1 Write out the characters of the representations of the following direct
products .. and determine the
representations which comprise

them for group Dflh: A lg x
A
1
,, x A
1
,; 8
1
, x E
1
.t:;
x E
2
,;
X X X E
111

5.2 In determining vibrational selection rules (Section 10.6) we shall need
to determine whether integrals of the types f 1f1 dr are nonzero,
where the function f is x .. y, z .. x!, y
2
.. z.:!, xy .. yz, zx., or any combinations
or sets thereof. Also, is totally symmetric and 1f1 :. m ay be long to any
irreducible representation. ldentify the irreducible representations to
which 1f1 :, m ay be long in arder to give nonzero integrals for tnolecules
of symmetry C.k and D
3
c1.
5.3 In connection with certain forms of spectroscopy ( e.g., circular dichroism
and magnetic circular dichroism) it is necessary to know what electronic
transitions are magnetic dipole allowed. The operators for this have the
symmetry properties of R_p R,., and R=. For a molecule of T,_
1
symmetry ..
determine what pairs of states could be connected by a magnetic dipole
allowed transition.
6
SYMMETRY-ADAPTED
LINEAR COMBINATIONS
6.1 INTRODUCfORY REMARKS
In nearly all of the ways in which chemists employ symmetry restrictions to
aid them in understanding chemical bonding and molecular dynamics-for
example. constructing hybrid orbitals, constructing molecular orbitals (MOs),
finding proper orbital sets under the action of a ligand and analyzing
the vibrations of molecuJes, to name those subjects which will be covered
explicitly in Iater chapters of this book-there is a common problem. This
problem is to take one or more sets of orthonormal functions, which are
generally either atomic orbitals (A Os) or internal coordinates of a molecule,
and to make orthonormallinear combinations of them in such a way that the
combinations form bases for irreducible representations of the symmetry group
of the molecule.
It will be obvious from the content of Chapter 5 why such combinations
are desired. First, only such functions can, in themselves, constitute acceptable
solutions to the wave equation or be directly combined to form acceptable
solutions, as shown in Section 5.1. Second, only when the symmetry properties
of wave functions are defined explicitly. in the sense of their being bases for
irreducible can we employ the theorems of Section 5.2 in
order to determine without numerical computations \vhich integrals or matrix
elements in the problem are identically zero.
The kind of functions we need may be called synunetry-adapted linear
conzbinations (SALCs). It is the purpose of this chapter to explain and illus-
trate the methods for constructing them in a general way. The details of
adaptation to particular classes of problems will then be easy to explain as
the needs arise.
114
SYMMETRY-ADAPTED LINEAR COMBINATIONS 115
The fundamental, universally applicable tool for constructing SALCs is the
projection operator. A complete projection operator is capable of generating
a complete set of SALCs without, soto speak, human intervention. You just
put the m achine in gear, push the start button anq collect the answer. How-
ever, complete projecton operators require a knowledge of the entire matrix
for each operation. In practice, we prefer to employ the usual character tables,
which provide only the characters of the matrices. There is an Hincomplete''
projection operator that functions by using only the characters, but since
there is no such thing as a free lunch, (not even with the help of group theory)
one cannot obtain complete SALCs with these operators. To
obtain the complete result in cases of two- and higher-dimensional represen-
tations, ''human intervention'' is required. In the next section. we shall show
how both the complete and the incomplete projection operators are derived.
We shall then show how they work, in progressively more elaborate cases,
with particular emphasis on showing how the incomplete operators in con-
junction with standard character tables can always provide a complete set of
SALCs.
6.2 DERIVATION OF PROJECTION OPERATORS
Let us assume that we ha ve an orthonormal set of l; functions ... ,
c/J}, which form a basis for the ith irreducible representation ( of dimension
l;), of a grpup of arder h. For any operator, R, in the group we may then, by
definition, write
(6.2-1)
This equation is then multiplied by [f(R H, ]*, and each si de summed over
all operations in the group., giving
2: = 2: 2: (6.2-2)
R R S
We note that the are functions independent of R; hence the right si de of
6.2-2 may be rewritten as
2:
J R
Thus, we have a series of L; terms, each of which is a multiplied by a
each coefficient is itself expressed as a sum of products over the
operations R in the gro u p. These coefficients, however, are governed by the
116 PRINCIPLES
great orthogonality theorem (page states that
Thus all except have coefficients of zero, and only when i = j and
t = t' can even the term in lf>s survive. Thus 6.2-2 simplifies to
(6.2-3)
We now introduce the symbol
p;, = [f(R)i, ]* R
.fl hL; SI
(6.2-4)
and rewrite 6.2-3 as
p, ,,ki = A-,i,(). ()'
J r 'l'r o/.f 11 rr
(6.2-5)
The operator is called a projection operator. It may be applied to an
arbitrary function e/>!, and only if that function itself or sorne term in it happens
to be rbJ _ will the result be other than z_em. If is a component of the
arbitrary function, lf>{ will be "projected'" out of it and the rest will be
abolished. Thus, we have
In the very important special case where we use P{, we have
(6.2-6)
which means that P{, projects out of an arbitrary function cJ>:. Thus, by
using the 1 projection operators based on the 1 diagonal matrix elements, we
m ay generate from sorne arbitrary function, the functions that form a
basis for the jth irreducible representation.
An lllustrative Application of the Complete Projection Operator
To illustrate how 6.2-6 works, let us consider the general function xz +
yz + z
1
, in the group C:"' (which is isomorphous to We shall use the
pro jection operators to obtain from this arbitrary function a pair of functions
which form a basis for the E representation. The matrices for this represen-
tation are given in Table 6.1. Table 6.2 shows how the arbitrary function
xz + yz + is transformed by each of the six symmetry operators in the
gro u p.
Table 6.1 Matrices for the E Representation of the Group C.
1
..
Operation Matrix Operation Matrix
E
1
2 2
V}
1
2
v'3
2
2 2
Y3
2
1
2
a,.(xz)
O':.
#Ttr
u,.

tV3
2 2
VJ
-
2 2
2
V3
2
V3
2
-
2
We shall first use the projection operator Pf,.
Pf1 (xz + yz + z
2
) = 1f {(1)(xz + yz + z
2
)
+ (- !)[- !(1 + \13)xz + i(\13 - 1)yz + z
1
]
+ ( -l)[!(\13 - l)xz - 4(1 + \13)yz + z
2
]
+ (l)(xz - yz + z:?)
+ (- !)[- 4(1 + V3)xz + 4(1 - \13)yz + z
1
]
+ (- !)[!(\13 - l)xz + + \13)yz + x
2
]}
We now collect terms. The coefficients of the xz, yz, and z
2
terms are as
follows:
xz: i[l + !(1 + \13) - !(\13 - 1) + 1 + !(1 + \13) - !(\13 - 1)]
= if[l + t + t + 1 + t + t + V3(t - ! + ! - t)]
= 1,(3 + O) = 1
yz: ii[l - t(\13 - 1) + l(l + \13) - 1 - !(1 - \13) - !(1 + \13)]
= i[ 1 + t + ! - 1 - l - t + \13( - t + t + - t)]
= i(O) = O
z
2
: i( 1 - ! - ! + 1 - ! - !) = = O
Table 6.2 Transformations of So me Simple Functions of x, y, and z
Operator
E
e)
C1
.:l
Functions
X y z xz + yz + z:!
x y z xz + yz + z:!
H- X+ v1_v) H- y- V3x) z H-(1_+ V3)xz + (VJ -_l)yz] + z:!
H- x- Y3y) M- y+ vJx) z 4((Y3- l)xz- (1 + Y3)yz] + z
1
a, (xz) x - v z xz - vz + z:!
a:. H- x- YJy) !(y - -V3x) z H -(l + YJ)xz + (l - YJ)yz] + z
1
H- x + V)y) HY + \l3x) z !((\IJ - I)xz + (1 + \.IJ)yzJ + z:!
118 PRIN CIPLES
Next we use the projection operator,
(xz + yz + z
2
) = i{(l)(xz + yz + z
2
)
+ (- !)[- !(1 + \.3)xz + !(V3 - l)yz + z!]
+ (- - l)xz - !(1 + \.3)yz + z
2
]
+ (- 1 )(xz - yz + z:!)
+ (!)[- !(1 + V3)xz + !(1 - \.3)yz + z
2
]
+ (!)[!(\.3 - l)xz + !(1 + \.3)yz +
Again, we collect terms and evaluate the coefficients, obtaining
xz: i[l + !(1 + \.3) - !(\.3 - 1) - 1 - !(1 + \.3) + !(\.3 - 1)]
= f;[ 1 + l + ! - 1 - - t + V3 (! - t - t + ! ) ]
= = O
yz: - !(\.3 - 1) + + \.3) + 1 + !(1 - \.3) + t(l + V3)]
= i[ 1 + t + ! + 1 + l + ! + \.3(- t + ! - t + ! ) ]
= i(3) = 1
z
2
: i(l - - k - 1 + 4 + 4) = i(O) = O
Thus we have projected out of the function xz + yz + z
2
the two functions
xz and yz, which form a basis for the E representation. The component z
2
has been abolished; it cannot, in \vhole or in part, contribute to a basis set
for the E representation.
The "lncomplete" Projection Operators
It will be clear from the foregoing discussion that in arder to use the type of
projection operator we ha ve developed so far we need to know the individual
diagonal elements of the matrices. This is inconvenient, since normally the
only information readily accessible is the set of characters-the sum of all the
diagonal matrix elements-for each matrix of the representation in question.
For one-dimensional representations, this is a distinction without a difference,
but for two- and three-dimensional cases it is advantageous to have a
projection operator that employs only the characters. It is not difficult to
derive the desired operator, beginning with the explicit expression for P!,,
nan1ely,
(6.2-7)
If we su m each si de over all values of t' we obtain
p, = " p,
L ,,
r'
= L L
h t' R
= f { [f(R)i,]* }R
p = !J. 2: x(R)iR
h R
(6.2-8)
In this development we have employed the interchangeability of the order of
the summations and the definition of the character of the matrix.
Let us now see what happens when we apply f>E to xz + yz + z
2

PE(xz + J'Z + z:!) = i{(2)(xz + vz + z
2
)
+ (- 1 )[ -- 4(1 + v'3)xz + !(\13 - 1)yz + z
2
]
+ (- 1)[4(\13 - l)xz - 4(1 + V3)yz +
+ O + O + O}
= 1;{[2 + !(1 +_ \G) - !(\/3 - 1!Jxz
+ (2 - !(V3 - 1) + !(1 + V3)]yz
+ (2 - 1 - l)z
2
}
= i(3xz + 3yz + Oz
2
)
= xz + yz
We see that this operator has abolished the irrelevant part of the function
and projected out a linear combination of the two separate functions, xz and
vz, which we were able to obtain by employing the projection
.'f
1
and Pr,_. This should not be surprising. We obviously cannot get two
separa te results with only one operator. Moreover, si nce the operator is
derived by adding the individual operators, a sum of the results given by the
individual operators is what we must expect. the projection operators
of the type p; cannot be as powerful and explicit as those of the type
However they usually suffice for solving practical as we shall dem-
onstrate in the next section.
6.3 USING PROJECTION OPERATORS TO CONSTRUCT SALCs
The most important and frequent use for projection operators is to determine
the proper way to combine atomic wave functions on individual atoms in a
molecule into M Os that correspond to the molecular symmetry. As pointed
out in Chapter it is essential that vald MOs form bases for irreducible
represenrarions of the molecular point group. we encounrer rne proOicm or
writing SALCs when we deal with molecules having sets of symmetry-equiv-
120 PRINCIPLES
alent atoms (e .g., the C atoms of benzene, the F atoms of SF
6
, or the Cl
atoms of
ea eh contributing atomic wave functions to the formation

of M Os. What we need to know are ( 1) the irreducible representations to
which any given set of equivalent AOs will contribute. and (2) the explicit
form of the linear combinations of atomic orbitals (LCAOs) that satisfy the
symmetry properties of each representation. In we need SALCs coP-
structed from atomic orbitals. In this section we shall present a number of
examples of how this is done.
SALCs Belonging Only to One .. Dimensional Representations:
Sigma Bonding in C
2
H
4
Clearly, for a one-dimensional representation the character and the full matrix
are the same thing. Hence, the Hincomplete'' projection operator is Hcom-
pJete,. in these cases. and will provide the appropriate SALC unambiguousJy
and automatically. Let us illustrate by asking what SALCs can be formed by
the ls orbitals of the four hydrogen atoms in ethylene.
The three preliminary steps befare the SALCs can actually be constructed
are, in this and all other the following:
l. We identify the point group: D
111

2. We use the four H ls orbitaJs as the basis for a representation; we eh ose
the coordinates and label the a functions as shown below. It is necessary to
se e how the set of four basis functions, that is, a,
is affected by each
symmetry operation. If we list the four basis functions as a column vector
befare and after the application of sorne symmetry operation, say this
is simply a way of stating that the C
2
(z) operation takes a
1
from its initial
position to the third position. a'!. to the fourth position. a
3
to the first position,
and a
4
to the second position.

a-...

Befo re After
SYMMETRY-ADAPTED LINEAR COMBlr'JATIONS Ul
We can now write a matrix that expresses this transformation:
o o 1 o u,
UJ
o o o 1
u!
X= O
1 o o o
-
UJ U
o 1 o o u4 a:.
Clearly the only way that any such matrix can ha ve 1 rather than O at any
diagonal position is when it carries one of the basis functions into itself. The
C:!(z) operation moves every basis function from its original position to some-
where else. and thus has all of its 1 's in off-diagonal positions. lts character
could thus have been recognized to be O without the bother of writing out
the entire matrix. The criterion is: Any basis function that ntoves contributes
nothing to the character. Thus the only operations in D
2
1r that leave any basis
function in place are the two that lea ve them all in E and u(xy), and
these have a character of 4. Hence we obtain the following reducible rep-
resentation:
E C
1
(z) C
1
(y) C:!{x) 1 u(xy) a(xz) a(yz)
4 o o o o 4 o o
3. We reduce this to its irreducible components: Ag +
+ B
111
+ 8
3
,.
With these preliminaries settled, we now apply the appropriate projection
Pi, (6.2-8) for each of these representations to one of the members
of the basis set, that is. to any one of the four u functions. In this and sub-
sequent examples, we shall drop the numerical factor, 1/ h sin ce we are in-
terested only in the functional form of the SALC; its normalization is a trivial
matter that can be attended to at the end. For each of the representations
we have the following results:
E C2(Y) C:!(x)

- (l)a
1
+ (1)u
3
+ (1)u4 + (l)u2
i u(xy) u(xz) a(yz)
+ (l)u
3
+ (l)u
1
+ (l)u
2
+ {l)u_.
= 2(a, + a
2
+ a
3
+ a
4
) = a
1
+ a
2
+ a
3
+ a_.
P
8
r: (aJ) = (l)u
1
+ (l)a
3
+ ( -l)u
4
+ ( -l)u:!
+ (l)a
3
+ (l)u
1
+ ( -1)u
2
+ ( -l)u ..
= 2(u
1
- u
2
+ u
3
- = u, - u:! + a3 - U.
P
8
2u (a
1
) = (I)u
1
+ ( -1)a
3
+ (l)u
4
+ ( -1)u2
+ (-l)u
3
+ (l)u, + (-1)a
1
+ (l)a ..
= a, - a
1
- u
3
+ a_.
U2 PRINCIPLES
As a check on these results, each one can be subjected to the symmetry
operations to be sure it responds ( i.e . ., goes into + 1 or - 1 times itself) as
required by the representation to \vhich it belongs.
A Case of Involving a Two-Dimensional Representation:
Sigma Bonding in PtCH-
We again take the first three steps and obtain:
Point Group: D-tlt
Irreducible representations: A ~ + B
1
..: + E,
For the two one-dimensional representations. the procedure is routine, but
note that we must explicitly apply every one of the 16 operations in the class:
P A ~ : (a
1
) = a
1
+ a'l + u
3
+ a_.
P
8
e (a.) = a
1
- u'1 + a
3
- a_.
Now let us go after the E, SALC. Note that only the operations E, C l ~ i,
and u
11
have nonzero characters, so these are the only ones that need to be
considered. We obtain
E Cl i a,
pE" (u
1
) = (2)a
1
+ (- 2}u
3
+ (- 2)u
3
+ (2)a
1
For a two-dimensional representation, we require two orthogonal func-
tions, which jointly form a basis for the representation. We have one, but we
require its partner. To obtain it, we recall that any member of a set of functions
forming the basis for a representation must be affected by the symmetry
operations of the group in one of two ways:
l. It \Vill go into 1 times itself.
2. lt will go into another member of the set or a combination of members
of the set.
lf we look at the effect of the operations E, C:!, 2 C ~ i, a
1
, and 2a,., \Ve
see that they give + l(a
1
- a
3
). This is as it should be,. but uninformative.
However, the remaining operations ha ve the second type of effect, and thus
enable us to find the partner function:
.t (u
1
- aJ) - (a:! - a.t) or -(a2- a.t)
C ~ (a
1
- O"J) - (a'! - a.t) or -(a:! - 0'-t)
s.t (u, - aJ) - (u2 - a.t) or -(a2 - U.t)
{d (a, - aJ) - (a
1
- a")
or -(a
2
- a-4)
Since only two orthogonal functions are needed to provide a basis for the
E, representations, we ha ve clearly, in this simple case, reached the end of
our quest. The two functions, in normalized form, that we require are
and
1
v'2 (a-y_ - u
4
}
An Example of SALCs for a Three-Dimensional Representation
In a hypothetical MH(l molecule with 0, symmetry. it can be sho\vn that the
set of six M-Ha bonds provide the basis for the irreducible representations
A lg + E ~ + T
111
(As an exercise, sho\v that this is so.) What are the expres-
sions for the three SALCs corresponding to the T
1
, representation?
To solve this problem, we first recognize that we need not employ all of
the 48 operations of Oh; instead, we can deal with the T
1
representation of
the pure rotational subgroup O, which has only one half as many operations.
Let us label axes and basis functions as shown in the sketch below:
z
l
124 PRINCIPLES
If we apply the projection operator pr, to u
1
and u ~ we obtain:
pr, (u
1
) = 3u
1
+ (2u
1
+ 2a
2
+ 2u
2
) - (u
1
+ 2u2)
- (u
3
+ a
4
+ u
5
+ u
6
+ 2u
2
)
= 4a
1
- u
3
- U4 - u
5
- a
P T, (U
2
) = 4u
2
- U
3
- U. - U; - U h
By subtracting one of these from the other we obtain 4u
1
- u ~ = u, -
u
1
Clearly, by proceeding in the same way with u
3
and u .. and then with u
5
and a
6
, we can obtain the following set of normalized SALCs:
These are clearly mutually orthogonal, and it is easy to show that they do.
indeed, transform according to the T
1
u representation of 0
11
This last step is
left as an exercise.
A Cyclic Tr System: 1T Orbitals for the Cyclopropenyl Group
The cyclopropenyl group .. C
3
H:h is the simplest carbocycle with a delocalized
n system and can serve as a prototype for this class of molecules. Let us see
how the pn orbitals of the individual carbon atoms can be combined in to
MOs-or at least the immediate precursors of actual rr M Os. Cyclic rr systems
will be discussed in general in Chapter 7, and this illustration is intende<;J only
to demonstrate the use of projection operators in making SALCs of A Os on
different atoms.
The usual preliminary steps befare the SALCs can actually be set up give
the following results.
Point group: DJ
1
,
The representation (see sketch below)
3 o -1 -3 o 1
Irreducible components: A ~ + E"
SYMMETRY-ADAPTED LINEAR COMBINATIONS U5

____ ,

8

Let us first apply the projection operator p.-t; to 41
1
We thus obtain
pA; l/J, = 411 + QJ"!. + 41J + 4J, + QJ.l + 411 + cp +
qJ'! + 41J + ePI + l/JJ + 411
= 4(41, + 41"!. + 413) = 4J, + 411 +
The reader may demonstrate by applying the 12 group operations to it that
this function does indeed form a basis for the representation.
Clearly. if we multiply the SALC by 1/YJ. it will be normalized. Thus
the final result for is ( 1 1\13 )( l/J
1
+ + l/J
3
}.
For the E" representation .. using again 41
1
, we ha ve
ftt:" q1
1
= (2)E4J, + ( + ( +

+ (O)Ci4J, + (O);c/J, + ( -2)u,4J, + (l)S_,cfJ,
+ + + (O)u:.41, +
= 241, - (/>2 - 41J + 241, - 4J:! - l/13 = 241, - 4J1 - (/JJ
It is easy to show that the norn1alized function is ( 1 !v'6)(24J, - 41
2
-
As befare. this is but one of two functions. which together form a basis for
the E" representation. We find its partner following the same method as
befare.
lf we carry out a symmetry operation on one of the two functions, it will
either go into + 1 times itself, into its partner .. or into a linear combination
of itself and its partner. Let us chose an operation which does not convert it
into + 1 times itself, namely CJ.:
U6 PRINCIPLES
It may easily be shown that the second function is not 1 times the first. but
in this case it is also not orthogonal to as the partner must be. The se.cond
function must therefore be a linear combination of the first and its partner
and we can find the expression for the partner by subtracting an appropriate
multiple of the first one out of the second one, leaving the partner as the
remainder.
This is most easily done by for the moment, normalization, with
the idea of attending to that at the end. Thus we proceed as follows:
This may be normalized to (l/v'2) (r/JJ - rp
3
). It is orthogonal to the first
function and is thus an acceptable partner:
The reader may demonstrate that the pair of functions
do in fact forma basis for the matrices ofthe E" representation
that each of these is orthogonal to the SALC of A; symmetry.
A General Simplification
The procedure just used, although routine and reliable, is lengthy, particularly
for the two-dimensional representation. The results could have been obtained
with less labor by recognizing that the rotational symmetry .. the behavior of
the SALCs upon rotation about a principal axis, of order 3 or more, alone
fixes their basic form. Their behavior under the other symmetry operations
is a direct consequence of the inherent symmetry of an individual orbital
toward these operations, ah or a c2 passing through it, being added to the
symmetry properties under the pure rotations about the principal axis. Let
us consider specifically the C
3
H{ case, where this can be seen by inspecting
the D
3
h character table. For all A -type A .. A';, and
A 2, the characters are the same for the C
3
and operations; similar) y. the
E' and E" representations are identical within the subgroup C
3
The thing
which decides that we are dealing specifically with and E" SALCs is the
inherent nature of the pn basis functions.
On the strength of the above considerations, a procedure that restricts
attention to the pure rotational symmetry about the principal axis may be
used to construct the SALCs. For C
3
H
3
\Ve use the group C
3
This group,
like all uniaxial pure rotation groups, is Abelian. Its three operations fall into
three and it must ha ve three irreducible representations of dimension
l. In general, a group C, has n one-dimensional representations ( cf. Section
4.5), so that what we show here for c3 will be generalizable to all en groups.
In the subgroup C
3
, the set of pn orbitals of C
3
H
3
spans the A and E
representations. The latter. however, appears in the character table as two
associated one-dimensional representations; a projection operator may be
written for each of these one-dimensional components individually. Thus. we
shall be able to obtain each of the SALCs belonging to the E representation
directly and routinely, using projection _operators. This is the advantage of
using only the principal axis rotational symmetry. Let us now work through
the algebra and see how much this trick expedites our task of constructing
the SALCs.
Application of the projection operators pA, pE<'>., and pE<l) to rp
1
( neglecting
constant numerical factors) gives:
PAljJl = (l)Erp, + (l)C\ljJ, + (l)jrp,
= (l)rp, + (l)l/12 + (l)l/JJ
= ljJ, + QJl + l/13
PE(l)rp
1
= (l)Erp
1
+ (e)C34J, + (e*)54J,
= l/1
1
+ erp'! + e*ljJ3
jJE<'!lljJ, = (l)Erp, + (t:*)C3ljJ, + (e)CjljJ,
= l/1
1
+ e*r/J2 + eljJ3
The A SALC has exactly the same form as we previously obtained for the
SALC. using full D
3
h symmetry. The two E SALCs are actually satisfac-
tory in the sense of being proper basis functions and being orthogonal to each
other. However, we prefer to ha ve real rather than complex coefficients. This
change can be accomplished very simply because of the fact that the two sets
of coefficients are arranged as pairs of complex conjugates ( cf. Section 4.5).
Thus, if we add them term by term, the imaginary components of each pair
are eliminated, leaving a SALC with real coefficients. Also, if one set is
subtracted, term by term, from the other, a set of pure imaginary coefficients
will be obtained and the common factor i may be to leave another
128 PRINCIPLES
set of real coefficients. These addition and subtraction procedures are simply
a case of forming new linear combinations of an initial set .. and this is an
entirely proper and rigorous thing to do. Thus, we add the two E SALCs
obtained above:
(/1
1
+

+

+ (/1
1
+

+

+ (e + + (r. + e*)/1
3
e + e* = (cos 2n/3 + i sin 2n/3) + (cos 2n/3 - i sin 2n/3)
= 2 cos 2n/3 = 2(-!) = -1
The first new SALC is therefore
Next, we subtract the original E SALCs and divide out i:
+ +
- (/J, + +
(e - e* )/1
2
- (e - e*

(r. - e*) (cos 2n/3 + i sin 2rc/3) - (cos 2rc/3 - i sin 2rc/3)
i i
= (2i sin 2n/3 )/ i
= 2 sin 2rr/3 = z('7) = v3
The second new SALC, which should be orthogonal to the first and therefore
its proper partner in forming a basis for the E thus has the
form
Clearly, when the SALCs we ha ve just obtained are properly normalized,
they are identical to those previously obtained by using full D
3
1r symmetry.
This second procedure, which was much simpler, may be summarized as_
follows:
l. An initial set of SALCs may be written down by inspection of the
character table. Each one is of the form +

+
with coef-
ficients that are the characters for E, C
3
, Cj. .
2. The pairs of SALCs for the E representation are added and subtracted
( dividing the result by i) to get two new orthogonal SALCs which ha ve
all real coefficients.
3. The SALCs are normalized.
SYMMETRY-ADAPTED LINEAR COMBINATIONS U9
EXERCISES
V
l
7(DH 9 1 2
lO 3
5 4
6.1 Consider the naphthalene molecule (point group D
21
,). Each carbon
ato m has a P: orbital to con tribute to the n M Os of the molecule. Take
these 10 P: orbitals as a basis set and answer the following questions:
(a) How do they divide into subsets of symmetry-equivalent orbitals?
(b) What representations are spanned by each subset? (e) What are the
10 normalized SALCs?
6.2 Use the projection operator method to determine the SALCs required
for the formation of M-X u bonds by the eight ligands in an

ion of symmetry D
4
1r.
6.3 Consider a regular pentagonal dodecahedral molecule, C
1
,,H
2
.,. What
irreducible representations are spanned by the set of C-H bonding
orbitals? By the set of (30) c-e bonding orbitals?
6.4 What are the SALCs made up of a orbitals on F atoms that can be
employed in forming P-F bonds in PF
5
?
6.5 Return to the octahedral MH
6
where the a bonding can be
described by MOs of
and T
1
, symmetry. The SALCs for the
T,, set were derived n the illustration (page 123). Determine t_hose for
the A and components.
PART 11
APPLICATIONS
7

M
O
L
E
C
U
L
A
R

O
R
B
I
T
A
L

T
H
E
O
R
Y

A
N
D

I
T
S

A
P
P
L
I
C
A
T
I
O
N
S

I
N

O
R
G
.
A
N
I
C

C
H
E
M
I
S
T
R
Y

7
.
1

G
E
N
E
R
A
L

P
R
I
N
C
I
P
L
E
S

I
n

t
h
e

v
a
l
e
n
c
e

b
o
n
d

t
h
e
o
r
y

a
s

d
e
v
e
l
o
p
e
d

b
y

S
l
a
t
e
r
,

P
a
u
l
i
n
g
,

a
n
d

o
t
h
e
r
s
,

a
l
l

b
o
n
d
s

a
r
e

c
o
n
s
i
d
e
r
e
d

t
o

b
e

t
w
o
-
c
e
n
t
e
r

b
o
n
d
s
,

t
h
a
t

i
s
,

t
o

s
u
b
s
i
s
t

b
e
t
w
e
e
n

t
w
o

a
t
o
m
s
.

I
t

i
s

o
n
l
y

a
s

a
n

a
f
t
e
r
t
h
o
u
g
h
t
,

m
o
r
e

o
r

l
e
s
s
,

t
h
a
t

a
c
c
o
u
n
t

i
s

t
a
k
e
n

o
f

a
n
y

i
n
t
e
r
a
c
t
i
o
n

b
e
t
w
e
e
n

s
u
c
h

b
o
n
d
s
.

I
n

a
d
d
i
t
i
o
n
,

t
h
e
r
e

i
s

n
o

a

p
r
i
o
r
i

w
a
y

t
o

d
e
c
i
d
e

w
h
i
c
h

p
a
i
r
s

o
f

a
t
o
m
s

s
h
o
u
l
d

b
e

c
o
n
s
i
d
e
r
e
d

a
s

b
o
n
d
e
d

a
n
d

w
h
i
c
h

n
o
n
-
b
o
n
d
e
d
;

w
e

m
a
k
e

c
h
o
i
c
e
s

t
h
a
t

a
r
e

s
u
p
p
o
r
t
e
d

b
y

c
h
e
m
i
c
a
l

k
n
o
w
l
e
d
g
e
.

T
h
e

m
o
l
e
c
u
l
a
r

o
r
b
i
t
a
l

t
h
e
o
r
y

b
e
g
i
n
s
,

a
t

l
e
a
s
t

i
n

p
r
i
n
c
i
p
i
e
,

w
i
t
h

t
h
e

i
d
e
a

t
h
a
t

a
l
l

o
r
b
i
t
a
l
s

i
n

a

m
o
l
e
c
u
l
e

e
x
t
e
n
d

o
v
e
r

t
h
e

e
n
t
i
r
e

m
o
l
e
c
u
l
e
,

w
h
i
c
h

m
e
a
n
s

t
h
e
r
e
f
o
r
e

t
h
a
t

e
l
e
c
t
r
o
n
s

o
c
c
u
p
y
i
n
g

t
h
e
s
e

o
r
b
i
t
a
l
s

m
a
y

b
e

d
e
l
o
c
a
l
i
z
e
d

o
v
e
r

t
h
e

e
n
t
i
r
e

m
o
l
e
c
u
l
e
.

T
h
e

t
h
e
o
r
y

d
o
e
s
,

n
a
t
u
r
a
l
l
y
,

a
d
m
i
t

o
f

t
h
e

p
o
s
s
i
b
i
l
i
t
y

t
h
a
t

o
n
e

o
r

m
o
r
e

o
f

t
h
e
s
e

t
n
o
l
e
c
u
l
a
r

o
r
b
i
t
a
l
s

(
M
O
s
)

m
a
y

h
a
v
e

s
i
g
n
i
f
i
c
a
n
t
l
y

l
a
r
g
e

v
a
l
u
e
s

o
f

t
h
e

w
a
v
e

f
u
n
c
t
i
o
n

o
n
l
y

i
n

c
e
r
t
a
i
n

p
a
r
t
s

o
f

t
h
e

m
o
l
e
c
u
l
e
.

T
h
a
t

i
s
,

i
o
c
a
l
i
z
e
d

b
o
n
d
i
n
g

i
s

a

s
p
e
c
i
a
l

c
a
s
e

w
h
i
c
h

i
s

a
d
e
q
u
a
t
e
l
y

c
o
v
e
r
e
d

b
y

M
O

t
h
e
o
r
y
,

b
u
t

l
o
c
a
l
i
z
a
t
i
o
n

i
s

n
o
t

u
b
u
i
l
t

i
n
"

a
s

a

p
o
s
t
u
l
a
t
e

a
t

t
h
e

o
u
t
s
e
t

a
s

i
n

t
h
e

v
a
l
e
n
c
e

b
o
n
d

t
r
e
a
t
m
e
n
t
.

B
e
c
a
u
s
e

M
O

t
h
e
o
r
y

t
r
e
a
t
s

o
r
b
i
t
a
l
s

w
h
i
c
h

a
r
e

i
n

g
e
n
e
r
a
l

s
p
r
e
a
d

o
v
e
r

t
h
e

e
n
t
i
r
e
t
y

o
f

a

m
o
l
e
c
u
l
e
,

c
o
n
s
i
d
e
r
a
t
i
o
n
s

o
f

m
o
l
e
c
u
l
a
r

s
y
m
m
e
t
r
y

p
r
o
p
e
r
t
i
e
s

a
r
e

e
x
t
r
e
m
e
l
y

u
s
e

f
u
i

i
n

t
h
i
s

t
h
e
o
r
y
.

T
h
e
y

m
a
k
e

i
t

p
o
s
s
i
b
l
e

t
o

d
e
t
e
r
m
i
n
e

t
h
e

s
y
m
-
m
e
t
r
y

p
r
o
p
e
r
t
i
e
s

o
f

t
h
e

M
O

w
a
v
e

f
u
n
c
t
i
o
n
s
.

W
i
t
h

t
h
e
s
e

k
n
o
w
n
,

i
t

i
s

o
f
t
e
n

p
o
s
s
i
b
l
e

t
o

d
r
a
w

m
a
n
y

u
s
e
f
u
l

c
o
n
c
l
u
s
i
o
n
s

a
b
o
u
t

b
o
n
d
i
n
g

w
i
t
h
o
u
t

d
o
i
n
g

a
n
y

a
c
t
u
a
l

q
u
a
n
t
u
m

c
o
m
p
u
t
a
t
i
o
n
s

a
t

a
l
l
,

o
r

b
y

d
o
i
n
g

o
n
l
y

v
e
r
y

s
i
m
p
l
e

o
n
e
s
.

I
f

e
l
a
b
o
r
a
t
e

c
a
l
c
u
l
a
t
i
o
n
s

a
r
e

t
o

b
e

c
a
r
r
i
e
d

o
u
t
,

t
h
e

u
s
e

o
f

M
O

s
y
m
m
e
t
r
y

p
r
o
p
-
e
r
t
i
e
s

c
a
n

i
m
m
e
n
s
e
l
y

a
l
l
e
v
i
a
t
e

t
h
e

l
a
b
o
r

i
n
v
o
l
v
e
d

b
y

s
h
o
w
i
n
g

t
h
a
t

m
a
n
y

i
n
-

m
u
s
t

b
e

i
d
e
n
t
i
c
a
l
l
y

e
q
u
a
l

t
o

z
e
r
o
.

1
3
3

134 APPLICATJONS
Although the fundamental principies are the same regardless of the type
of molecules treated, there are differences in the practica) procedures em-
pJoyed in handling and organon1etallic molecules. In this
chapter the basic concept will be introduced and the most con1mon applica-
tions to organic molectiles described. In the next chapter \\'e shall deal with
the major types of inorganic systems ( metal-metal bonds, clusters)
and organometallic compounds, where the organic and inorganic methodol-
ogies will be blended.
The LCAO Approximation
By far the commonest approximation employed to reduce the notion of an
MO to an explicit, practical form is the linear combination of atomic orbitals
(LCAO) approximation. Each MO is written as a linear combination of atomic
orbitals on the various atoms. Denoting the ith atomic orbital 4>; and the kth
molecular orbital l.fls., we write
l.fl k = C;k 4>;
;
(7 .1-1)
The 4J/s are a basis set, and it is convenient to choose or adjust them so that
they are normalized. This property, which we shall henceforth take for granted
is defined by the equation
J r/>1c/J; dr = 1
(7.1-2)
By using LCAO-MOs, a particular form of the wave equation, called the
secular equation is developed in the following way. The wave equation is
written in the form
.!ti - Ef/1 = (. i'f'- E)l.fl = O (7.1-3)
An LCAO expression for 1.f1 is no\v introduced, giving
C;(._l('- E)QJ; = o (7.1-4)
For clarity, without loss of generality, it is easier to continue deve)-
opment explicitJy for the case of a two-term LCAO-MO; thus 7.1-4 takes
the form
(7.1-5)
MOs IN ORGANIC CHEMISTRY 135
Equation 7.1-5 is no\v multiplied by J>h and the left si de integrated o ver all
spatial coordinates of the \Vave functions:
c
1
J l/>1(X- E)l/>
1
dr + c2 J lj>,(X- E)l/>
2
dr = O (7.1-6)
To simplify notation, the following definitions will now be introduced:
H;; = J l/>;. ~ f i p dr
H; = J f/J;. Yt i/J dr
s .. = J >..>. dr
IJ 'f'1 'f'
(7.1-7)
(7.1-8)
(7.1-9)
The integral H;; gives the energy of the atomic orbital {j>. The H;
1
integrals
give the energies of interaction between pairs of AOs. The S; are called
overlap integrals. Beca use the energy, E, is simply a number
J l/>;Ef/J dr = E J f/J;f/J dr = ES; (7.1-10)
Equation 7.1-6 m ay now be written
(7.1-11)
Equation 7.1-5 could also ha ve been multiplied by 4>
2
and integrated, leading
to
(7.1-12)
These two equations forro a system of homogeneous linear equations in e
1
ande!. They obviously have the trivial solutions e
1
= e
2
= O. It is proved in
the theory of homogeneous linear equations that other, non trivial solutions
can exist only if the matrix of the coefficients of the e
1
's forms a determinant
equal to zero (Cramer's theorem). Thus, we have the so-called secular equa-
tion:
Hn - ES1'1 =
0
H22- E
(7.1-13)
The numerical values of the H;;'s and H;/s and S;'s can be guessed, estimated,
or computed at sorne level of approximation, and the secular equation then
136 APPLICATIONS
solved for the values of E. From the algebra of determinants (cf. Appendix
1) it follows that an n x n determinant will give rise toan nth order polynomial
equation in the energy. Equation for gives the quadratic
equation
where the relations S; = S; and H;; = H;; have been invoked. This equation
can be solved to give two roots, E, and E
2
, which can be shown (by the
variational theorem) to be upper limits to the energies of the ground and first
excited states.
If the value E, is inserted into 7.1-11 and these equations may then
be solved for the coefficients c
11
and c
12
, which give the MO tJ
1
having the
energy E,. Similarly .. E
1
m ay be substituted to obtain equations for the coef-
ficients c:!l and c
12
giving the MO
which has the energy E

2

Thus, in summary, by using values of the various integrals ( often called
matrix elements) H;;, H;. S;, the energies of MOs may be calculated in the
LCAO-MO approximation without knowing the explicit form of the LCAO-
MOs. After the energy values are it is possible to determine the
coefficients, e;, and thus obtain the explicit expressions for the LCAO-MOs.
The Hckel Approximation
The LCAO-MO approach just outlined is in itself an approximation. Even
so, if no further approximations are the evaluation of the integrals
7.1-7, 7.1-8, and 7.1-9 can be time consuming. Sorne simplifications and fur-
ther approximations are often made, the most drastic of which are, as a
called the Hckel approximation. For our purposes this relatively crude ap-
proximation will suffice, and it has the advantage of permitting us to carry
treatments in which symmetry arguments can be employed through all stages,
even to numerical results, without becoming bogged down in algebraic and
computational problems.
Two points should be stressed. First, the symmetry arguments themselves
are rigorous and would serve equally well in calculations of whatever degree
of computational rigor one might care to make. Second, the Hckel approx-
imation happily, a successful approximation; for all its apparent crudity it
gives useful results.
The Hckel approximation assumes that all S; = O and that all HJJ = O
unless the ith and jth orbitals are on adjacent atoms .. The setting of all S, =
O means that the normalizing factors for LCAOs are obtained very simply.
If
MOs lN ORGANIC CHEMISTRY 137
and we require that
we obtain
J 1/f;l/f; dr = 1
- J ( a,cP;r dr
- J cfo;cfo; dr + f.t a;
1
a;, J cfo;cfok dr
( j#ok)
The second sum is equal to zero because overlap is assumed to be zero. The
first sum is just equal to '22 since the 4>/s are assumed to be normalized.
Thus
1

-=
N;
LJ a,
'
or
N- =
1
'
af;
For the special but not uncommon case in \vhich all a;' S are 1, N is just
1 where n is the number of atomic orbitals in the linear combination.
When the Hckel approxirnation is applied to the rr orbitals of hydrocar-
bons, the follo\ving abbreviations are conventional:
a = the energy of an electron in a carbon prr orbital befare interaction
with others
P = H;l' the energy of interaction between orbitals on adjacent atoms
The appearance of the secular equations can be further sin1plified if a is taken
as the zero of energy (i.e., set equal to zero) and {J taken as the unit of energy.
It can be sho\vn that {J is inherently a negative quantity. Thus an MO whose
energy is positive in units of fJ has an absolute energy that is negative. An
electron in such an MO is therefore n1ore stable than an electron in an isolated
pn orbital.
138 APPLICATIONS
Energy Level Diagrams
In many instances it is helpful to plot the calculated energies of MOs as in
Figure 7 .l. The energy sea le is and ea eh orbital is represented as a
short horizontal line placed at the position corresponding to the energy of
the orbital. For doubly degenerate orbitals a pair of closely spaced lines is
used. The occupation of the orbitals by electrons is indicated by placing small
circles on the lines.
Hund's Rule and the Exclusion Principie
The order of filling of MOs for the ground state of a molecule follo\VS the
same rules as does the filling of orbitals in the ground state of an atom. Thus
an electron will go into the lowest unfilled level subject to the following
restrictions: Only two electrons may occupy a single and their spins
must be of opposite sign ( exclusion principie). When electrons are to be placed
in a pair of degenera te orbitals, they will (as shown for th in Figure 7.1)
occupy each of the two degenerate orbitals singly giving a total spin of 1
rule}.
Bonding Character of Orbitals
It is convenient, as already noted, to choose the zero of our energy scale to
be that of the system in the hypothetical condition in which no interaction
between the separate atomic orbitals is occurring. In the actual state of the
molecule, then, sorne MOs will be of lower energy, sorne of higher energy,
and, in certain cases, sorne of the same energy as this state which we take as
zero. The MOs that are more stable than the separate, noninteracting atomic
orbitals have absolute energies less than zero and are called bonding orbitals.
Those with absolute energies greater than zero are called antibonding orbitals,
-2
1/Js
"'4
t:Q.
-l

o
U)
....
"2

::S
o

bQ
'""
V
e

UJ
2
---o--o-- "'1
Figure 7.1 An example of an energy level diagram.
MOs lN ORGANIC 139
and any orbitals having energies of precisely zero are called nonbonding
orbitals. In Figure 7.1, we have a strongly bonding MO, t1
1
; a less strongly
bonding, doubly degenerate MO, tJ
2
; a nonbonding, doubly degenerate MO,
'1'
3
; a moderately antibonding, doubly degenerate MO, and a strongly
antibonding, nondegenerate MO, 'lis
To see how the Hckel approximation simplifies the treatment of a mod-
erately complex problem, consider the rr orbitals of naphthalene. The 10 prr
orbitals, numbered as in Figure 7.2, can be combined into 10 linearly inde-
pendent rr MOs; thus, a 10 x 10 secular determinant can be written, as
follows:
Hr!. - ES.-1 H13 - ESn . . . H1.1u - ES1.1u
Hu- E
Hw.w - E
=o
(7.1-15)
We now introduce the Hckel approximation, employing the a, {J notation,
the convention that a is the zero of energy, and using {J itself as the unit of
z
y
2
--------..- X
6
Figure 7.2 Cartesian axes and atom numbering for the naphthalene molecule.
140 APPLICATIONS
energy. Explicitly we ha ve
Hu = H22 = H33 = = H99 == Hw.tu = a = O
s .. =
lJ ,,
H;
1
= H; = O, except for H
12
.. H
23
, H34 .. Hst. .. H61, H1th
Hm, H9.llh H
1
9, H.uu .. and H
5
10
,
which are all equal to p,
which is equal to 1 unit of energy
The secular equation is then:
-E 1 o o o o o o 1 o
1 -E 1 o o o o o o o
o 1 -E 1 o o o o o o
o o 1 -E o o o o o 1
o o o o -E 1 o o o 1
=o
o o o o 1 -E 1 o o o
o o o o o 1 -E 1 o o
o o o o o o 1 -E 1 o
1 o o o o o o 1 -E 1
o o o 1 1 o o o 1 -E
(7.1-16)
This is simpler than 7.1-15, in that many of the terms in the 10th-order
polynomial equation which will result on expanding the determinant will now
be equal to zero. Nevertheless, the basic, awkward fact is that a 10th-order
equation still has to be solved. This is not a task- to be confronted with
pleasurable anticipation; without the use of a digital computer it would be a
protracted, tedious job. Fortunately, in this case and all others in which the
molecule possesses symmetry, the secular equation can be factored-that is,
reduced to a collection of smaller equations-by using the symmetry prop-
erties in the right way. The method of symmetry factoring will now be ex-
plained and illustrated.
7.2 SYMMETRY FACTORING OF SECULAR EQUATIONS
Even with the simplifications that result from a drastic approximation such
as the Hckel approximation, the secular equation for the MOs of an n-
atomic molecule will, in general, involve at least an unfactored nth-order
determinant, as just illustrated in the case of naphthalene. It is clearly desir-
able to factor such determinants, and symmetry considerations provide a
systematic and rigorous means of doing this.
A secular equation su eh as 7.1-15 is derived from an arra y of the individual
MOs IN ORGANIC CHEMlSTRY 141
atomic orbitals of the basis set. Thus, in general, all of the H; and S; are
nonzero. Asan approximation, sorne may be set equal to zero, as in 7.1-16,
but still there is no logical, a priori reason why any entire classes or sets of
integrals involving the individual AOs as such should systematically vanish.
Suppose that, instead of \vriting the secular determinant from an n x n
array of atomic orbitals, we use an n x n array of n orthonormal, linear
combinations of the basis set orbitals. Suppose, furthermore-and this is the
key-we require these linear combinations to be SALCs'L lhat is, each one
is required to be a function which forms a basis for an irreducible represen-
tation of the point group of the molecule. Then, as shown in Chapter 5, all
integrals of the types
J lfllfl dr
and
are identically equal to zero unless f/1; and f/1 belong to the same irreducible
representation.
The foregoing considerations Jead to a three-step procedure for setting up
a symmetry-factored secular equation:
l. Use the set of atomic orbitals as the basis for a representation of the
group, and reduce this representation to its irreducible components.
2. Combine the basis orbitals into linear combinations to
each of the irreducible representations. These SALCs can always be con-
structed systematically by using the projection operator technique developed
in Chapter 6.
3. List the SALCs so that all those belonging to a given representation
occur together in the list. Use this Iist to label the rows and columns of the
secular determinant. Only the elements of the secular determinant that lie at
the intersection of a row and a column belonging to the same irreducible
representation can be nonzero, and these nonzero elements willlie in blocks
along the principal diagonal. The secular determinant will therefore be
factored.
As a very persuasive illustration of the effectiveness of symmetry factori-
zation in reducing a computational task that would be entirely impractical
without a digital computer to one that is a straightforward pencil-and-paper
operation, we shaiJ again consider the naphthalene molecu)e. It has been
shown in Section 7.1 that the secular equation for the n M Os is the 10 x 10
determinantal equ'ltion, 7.1-15, if the set of 10 pn orbitals is used directly for
constructing LCAO-MOs.
The naphthalene molecule belongs to the point group D
2
,. The set of 10
pn orbitals may be used as the basis for a fn, of this group.
142 APPLICATIONS
This reducible representation can be decomposed into irreducible represen-
tations as follows:
When the 10 pn orbitals are combined into SALCs, f/1
1
to f//Jth which are then
listed in order of their symmetry t y p s ~ namely,
and these 10 SALCs are used to construct the secular determinant, it takes
the form shown in Figure 7 .3, where all the empty, off-diagonal places must
contain zeros. for rigorous, symmetry reasons.
It follows from the properties of determinants that, if the en tire determinant
is to have the value zero. each block factor separare/y must equal zero. Thus
the 10 x 10 determinantal equation has been reduced to two 2 x 2 and two
3 x 3 secular equations. For example, the energies of the two MOs of A,
symmetry are given by the simple secular equation
We shall return Iater (page 172) to the symmetry-factored secular equation
for the n MOs of naphthalene and solve for the energies, LCAO-MO coef-
ficients, and other useful results.
7.3 CARBOCYCLIC SYSTEMS
We begin by considering the most famous and important of such systems,
benzene. This molecule belongs to the point 'group DM When the set of six
pn orbtals, one on each carbon atom, is taken as the basis for a representa-
tion of the group D
61
, we obtain the result
r1f 6 o o o -2 o o o o -6 2 o
We have defined C axes and a,,'s as those passing through opposite carbon
atoms, and C2 axes and (J/S as those bisecting opposite edges of the hexagon.
The above characters can be obtained at once by remembering that the matrix
describing the etfect of a given operaran on the set of six orbitals obtains a
contribution to its diagonal from each basis orbital according to the following
schedule: O if the orbital is shifted to a ditferent position; + 1 if the orbital
"
"
"
"
'
e

A
u

B
l
u

"
'
6

B

l
o

B
3
,

A

u

B
,
.
,

B
J
,

"
'
1

"
'
"

t
/
1
3

"
'
4

"
'
5

t
/
1
6

i
f
,

t
/
l
a

t
/
1
9

t
/
J
,
o

H
1
1
-
E

H
u

H
2
1

H
2
2
-
E

H
3
3
-
E

H
J
4

H
3
5

H
4
3

H
4
4
-
E

H
4
s

H
5
J

H
s
4

H
s
s
-
E

H
6
6
-
E

H
6
,

H
7
6

H
1
1
-
E

'
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
1
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
1

H
a
a
-
E

H
8
9

H
a
,
1
o

'1
H
9
a

H
9
9
-
E

H
q
,
1
o

_
_

_
_
_

_

_

_

H
t
o
.
s

H
t
o
,
t
o
-
E

J
F
i
g
u
r
e

7
.
3

S
y
1
n
m
e
t
r
y
-
f
a
c
t
o
r
e
d

f
o
r
m

o
f

t
h
e

s
e
c
u
l
a
r

e
q
u
a
t
i
o
n

f
o
r

t
h
e

n
:

o
r
b
i
t
a
l
s

o
f

n
a
p
h
t
h
a
l
e
n
e
.

144 APPLICATJONS
goes into itself; -1 if the orbital goes into the negative of itseiL which means,
simply that it is turned u psi de down.
This representation is reduced ( cf. Section 4.3) as follows:
Thus we need to form LCAOs of the indicated symmetry types, and this can
be done by using projection operators for these representations of D
611
How-
ever, it is advantageous to approach this task from a less direct point of view
in order to arrive ultimately at an easier and more general approach to the
type of problem it represents.
As pointed out in Section 6.3, for the (CHh case, all the essential symmetry
properties of the LCAOs we seek are determined by the operations of the
uniaxial rotational subgroup, C
6
When the set of six p7C orbitals is used as
the basis for a representation of the group e
6
, the following results are ob-
tained:
eh
E
c6 eJ C2
C2
3
~
A 1 1 1 1 1 1
B 1 -1 1 -1 1 -1
E,
n
e
-:;* -1 -e
:*}
e* -e -1 -e*
E1
G
-e* -e 1
-e*
-e}
-:; -e* 1 -F. -e*
f,, 6 o o o o o
rr/J =A + B + E
1 + E2
Note first that x( E) = 6 while all other characters are zero. The reason
is that the operation E transforms each l/J; into itself while every rotation
operation necessarily shifts every cp; to a different place. Clearly this kind of
result will be obtained for any n-membered ring in a pure rotation group C,,.
Second, note that the only way to add up characters of irreducible represen-
tations so as to obtain x = 6 for E and x = O for every operation other than
E is to sum each column of the character table. From the basic properties of
the irreducible representations of the uniaxial pure rotation groups (see Sec-
tion 4.5), this is a general property for all en groups. Thus, the results just
obtained for the benzene molecule merely illustrate the following general
rule:
In a cyclic ( CH),. molecule with rotational symnzetry C", tlzere wi/1 always
be n re molecular orbitals., one belonging to ea eh irreducible representation of
the group en.
The (CH)
3
system discussed in Section 6.3 provides another illustration of
this rule.
The A, B, E
1
, and E
2
representations spanned in the pure rotation group
C
6
become the A
2
u, B
11
p E
1
-,:, and E
1
u representations in the group Dh
1
, when
the full symmetry of the benzene molecule and the inherent symmetry of the
individual p1C orbitals are considered. The great advantage of working with
the pure rotation group, which has already been illustrated (Section 6.3) in
the very simple case of ( CH)
3
, is apparent when we come to determine the
LCAO expressions for the MOs. Since each of the required linear combi-
nations belongs uniquely toa single one-dimensional representation, the pro-
jection operator technique is extremely simple to apply-so simple, indeed.
that the result can be written down as fast as one can write, simply by in-
spection of the character table.
Consider the following statement of the effect of applying the projection
operator for any representation of C(l to l/1
1
, the prc orbital on carbon ato m 1:
= x(E)El/J + x( Ch)Chl/Jl + x( + x(
+ x( + x(
= x(E)l/1, + x(Ch)l/1:! + +
+ x( c)l/Js + x(
The second expression is simply a list of the six l/J;'s, in numerical order, each
multiplied by the character for one of the six operations, in the conventional
order E, C(, ... , This must be true for each and every representation.
Hence, the sets of characters of the group are the coefficients of the LCAO-
MOs. The argument is obviously a general one and applies to all cyclic (CH),,
systems belonging to the point groups D,
11
, each of which has a uniaxial pure
rotation subgroup. en.
For the sake of concreteness, let us continue to use benzene asan example
and write out the tJ's:
A: lfl = l/J, + lP1 + cPJ + cP4 + l/J; + l/J
B: f//2 = l/11 - l/11 + lPJ - c/1-t + l/Js - lPfl
E: {lf/3 = l/J, + r.l/11 - r.*c/>3 - l/J.t - 1-:l/15 + r.*l/Jh
1
lf/4 = 4> 1 + r.* l/12 - el/13 - lP.t - t:* l/J; + cc/J6
E . { lf/:. = l/J, - e* c/11 - el/JJ + lP.t - r.* l/Js - t:l/16
2
f//n = l/J
1
- r.l/>
1
- Q1
3
+ Q1_. - eQ1; - e* Qlh
From a practica) point of view there are two disadvantages in these LCAO-
MOs. First, they contain imaginary coefficients. Second, they are not nor-
malized to unity.
As explained in Section 6.3, we can easily convert the pairs of SALCs
146 APPLJCATJONS
belonging to each pair of E-type representations into new linear combinations
with real numbers as coefficients by (1) adding them, and (2) subtracting them
and dividing out i. Thus, by adding tJ
3
and l.j/4 we obtain
which reduces to
The second linear combination, (t1
3
- is explicitly

= [(e - e*)l/>
2
- (t:* - e)l/1
3
- (e - e*)l/>:. + (e* - t:)Q>ll]li
= - V3Q>2 - V3ci>J + V3cp; + V3Q>6
In a similar way I.Jis and tJ
6
may be combined to give
I.JI( E2a) = I.Jis + 'l'n = 2ljJ, - c/>2 - QJ3 + 2cp4 - lP5 - QJ
t1(E:1b) = (I.JI:. - f//
6
)/i = -V3cp! + V3Q>3 - V3cb:. + V3Q>h
We now normalize these MO wave functions as described in Section 7 .1,
neglecting overlap, and obtain the following final expressions:
1
= v'6 ( c/11 + l/11 + ifJ 3 + e/>" + l/Js + c/16)
1
tJ(B) = v'6 ( c/J, - c/12 + QJ3 - + QJs - c/Jh)
1
f/1( E
1
a) = Yl
2
(2l/>t + c/J! - l/>3 - - l/15 + l/J6)
tJ( E
1
b) = l/>:1 + QJ3 - l/1; -
1
tJ( E1a) = v'i2 (2f/J, - lP:! - cPJ + 2cjJ4 - cP; - cP6)
IJI( E:!b) = - QJ3 + l/Js -
In addition to being normalized the MOs should be mutually orthogonal.
That this is true of those given above is easy to verify. although it foJJows
from the procedure used to construct them that they must be.
It is instructive to examine sorne diagran1s showng how the sons of tl1e
w's vary around the ring. For benzene, using the orbitals above,
we can make the following drawings:
tf(A)
1/J(B)
Note that the E
1
orbitals have one nodal plane and the E
2
orbitals two nodal
planes.
The energies of these MOs may now be calculated, using the Hckel
approximation as discussed in Section 7 .l. For 'I'(A) we obtain
E A = k( 6a + .12{J) = a + 2P
Proceeding in a similar way, we find the energies of the other LCAO-MOs
for benzene to be
En= a - 2{1
EE
2
a = E E ~ / J = a - {J
Taking a as the zero of energy and fJ as the unit of energy, we m ay express
these results in the form of an energy level diagram, namely,
-2 B
t
-1
El
E
o
1
--o-o-
~ o o
2 A ----o-o--
148 APPLICATIONS
Delocalization (Resonance) Energy
If each of the six 7C electrons in benzene occupied a single atomic re orbital
and there were no each would have an energy of a. The total
energy would then be 6a, which is zero if we assume, as above, that a is the
zero of our energy scale. However, when the atomic orbitals interact to
produce the MOs, the six electrons will now occupy these MOs according to
Hund's rule and the Pauli exclusion principie. The first two will enter the A
orbital, and the remaining four occupy the E
1
orbitals. The total energy of
the system is then
ET = 2(2P) + 4(JJ) = sp
Remembering that p is negative. we see that re bonding has stabilized the
molecule by sp. Energies expressed in units of pare not very informative,
however, unless we can estima te the val u e of p. We shall next turn to the
calculation of the delocalization energy of benzene in units of fJ. Since the
delocalization energy m ay be estimated experimentally, we shall then be able
to evaluate p. It is not practicable to do this by computation.
Strictly speaking, the concept of delocalization or resonance energy belongs
in valence bond theory. It is defined there as the difference between the
energy of the most stable canonical structure, which is one of the Kekul
structures, and the actual energy. The actual energy is assumed to be cal-
culable, according to valence bond theory, by taking into account resonance
between all possible canonical structures. Of these, usually only five are
considered sufficiently low in energy to be significant. There are the two
equivalent Kekul structures and three Dewar structures:
00
Kekul structures Dewar structures
However, it m ay be assumed that the difference between the energy of one
Kekul structure and the actual energy as calculated by MO theory will also
be the resonance energy except that, in keeping with the basic concept of
MO theory, we shall call it the 4delocalization energy.
We have already shown that the energy of the system of six 1C electrons in
benzene is equal to SfJ. We must now ca)cu)ate, in units of /1. the energy of
a Kekul structure, that is, the energy of the hypothetical molecule cyclo-
hexatriene. In cyclohexatriene there are three localized n bonds. When two
atomic re orbitals, say l/1
1
and c/J
2
, interact to form a two-center bond. two
MOs, f//
1
and f//
2
, are formed. In order that these be real, normalized, and
orthogonal, they must be
1
f{/1 = Y2 (c/J + c/12)
1
f{/"!. = V2 ( cp, - c/Jz)
Their energies are readily seen to be
E
1
= J f/1
1
.lrip, dr = i (J r/1
1
.lrip, dr + J r/1
1
.lr(/1
1
dr
= i(2a + 2P) = P
E2 = -p
+ J r/1 2 .ltip1 dr + J r/J 2 .i'tjJ2 dr)
Sin ce tJ
1
is the stable MO. the two n electrons will occupy it and their com-
bined energy will be 2p. Thus each of the pairs of localized n electrons in a
Kekul structure contributes 2P to the energy of the n1olecule, making the
total n electron energy of the localized, cyclohexatriene structure 6p.
But the actual energy is BP; hence the resonance or delocalization energy is
2p. .
Experimentally .. the delocalization energy of benzene is estimated in the
following way. The actual enthalpy of formation of benzene can be deter-
mined by thermochemical measurements. The energy of the hypothetical
molecule cyclohexatriene can be estimated by using the bond energies for
C-C .. C C, and C-H found in other molecules such as ethane and ethyl-
ene. The difference between these energies is the value
of the delocalization energy. We then evaluate IPI, sin ce
2IPI = "experimental" delocalization energy
The value of IPI obtained for benzene is 18-20 kcal/mol, depending on
choices of bond energies. * Practically the same value is obtained when other
aromatic molecules, such as naphthalene and anthracene, are treated in the
same way, a fact which lends support to the belief that the LCAO method
is at least empirically valid.
Other carbocyclic systems of the type (CH),, belonging to the point groups
D""" which are of interest are those with n equal to 3, 4 .. 5, 7, and 8. In sorne
cases the actual systems (which may be cations, neutral molecules, or anions)
are not necessarily of Dn" symmetry, but there is usually sorne purpose in
* See Appendix 111 for sorne mportant qualifications concerning the evaluation of {J.
150 APPLICATIONS
System
(symmetry)
C4H4
(D4rr)
6
LCAO-MO Expressions
1
.P<Ea) = v
6
(2c/>, - c/>2 - t/>J)
1 -
fo(Eb) = v'i (cp2- c/>3)
a/J(A) = i(cfot + c/>2 + c/>3 + e/> ... )
t
#.Ea>= v2. (e/>. - c/>3)
1
ap(Ea) = V
2
(t/>l - c/>4)
.p(B) = !(cp, - c/>2 + c/>J- cf,,..)
TABLE 7.1
Energies
a.-P
a.+ 2/3
l
- a.
1
l
a.- 2/3
MOs IN ORGANIC CHEtviiSTRY 151
Summary of Hckel MO Calculations for (CH),. Systems
Energy Leve) Diagram
(populated for
neutral species)
-2
t ~ E
2 -o-o--- A
-2
B
t -1
O E
E
o
---o-
2
--oo-A
Orbital Shapes
Ea
A
Ea
A
Eh
B
Eb
Delocalization
Energies,
Units of IPI
1
CJI4 o
152 APPLICATIONS
System
( sym metry) . LCAO-MO Expressions
CsHs
(Dn)c
w = 21T/5
r/J(E,a} =A(</>,+... cos w + .., cos 2UJ)
+ c/>4 cos 2w + cfo5 cos w)
r/J(E,b} = ~ (</> sin w +..,sin 2w
- q,,. sin 2w - 4>s sin w)
r/J(E. a}= ~ (tf>, +.p. cos 2w + t/> cos w
+ 4>4 cos w + 4>s cos 2w)
r/J(E. b} = ~ (t/>2 sin 2w- t/> sin w
+ q,,. sin w'- 4>s sin 2w)
TABLE 7.1
Energies
+2P
+
(2 cos w)P
+
(2 cos 2w)p
(continued)
Energy Level Diagram
(populated for
neutral species)
-2t
- 1
_____ E2
o
2 --oo-A
Orbital Shapes
A
Delocalization
Energies.
Units of IPI
CsHs+ 4 COS w
CsHs 6 COS w
CsHs- 8 cos w
154 APPLICATIONS
System
(symmetry) LCAO-MO Expressions
C,H, 1
(Du) ap(A} = v'7 (cpl + rPz + rPJ + rP4 + cps + rP6 + cp,}
W = 2TT/7 e
,P(E,a) = J ~ (</>1 + r/>z cos w + r/>3 cos 2w + .,4 cos Jw
+ rPs cos 3w + cp6 cos 2w + cp, cos w)
r#(E,b) = J ~ (r/>z sin w +.,,sin 2..; +.,.sin 3w
- cps sin 3w - cp6 sin 2w - rf>, sin w}
r#(Ez a) = J ~ (rf>, + r/>z cos 2w + .,, cos 3w + r/>4 cos w
+ cps COS w + cp6 cos 3w + q,, cos 2w)
,P(E, b) = ~ (r/>z sin 2w- .,.sin 3w- .,4 sin w
+ cp5 sin w + cP6 sin 3w- cp, sin 2w)
,P(E. a) = J ~ (rf>, + .,, cos 3w + r/>3 cos w + .,. cos 2w
+ cps e os 2w + r/>6 cos w + cp1 cos 3w)
r#(E, b) = J ~ (rf>, sin 3w - r/>3 sin w +.,.sin 2w
- cps sin 2w + r/>6 sin w- q,, sin 3w)
TABLE7.1
Energies
(X+ 2/1
+ 2P cos w
+ 2P cos a ~ ~
+ 2/1 cos 3w
(ontin ued)
(populated for
neutral species)
_____ EJ
-o-o--
----o-o-- E
-o-o--A
Orbital Shapes
E
2
a E
2
b
E
1
a E
1
b
A
C,H,-
Delocalization
Energies.
Units of IPI
8 cos (JJ
- 2cos 2w- 2
8 cos (JJ
- 4cos 2w- 2
156 APPLICA TIONS
System
(symmetry)
Ca He
(Ds,.)d
LCAO-MO Expressions
1
r/I(A) = VB (r/>J + cP1 + cPJ + rP4 + cPs + ep6 + ep, + r/>a)
1
r/I(B) = VS (ept- ep1 + epJ- ep4 + eps- ep6 + ep7- epa)
tfo(E,a) = ) ( V2q,, + <{> 2 - <{>4 - V2q,, - q,. +</>a)
1 - -
t/J(Elb) = vs (ep1 + V2epl + ep4- ep6- V2ep7- epa)
r/I(El a)= i(ep - ep3 + eps- ep7)
ap(E2 b) = i(ep2- ep4 +ep6- epa)
}
tfo(E,a) = ~ S { V2 q,, - q,, + q,,.- V2 q,, + q,. - <{>a)
tfo(E, b) = ~ <.Pz- V2 .,, + .,,.- .,. + V2 ..,- <{>
8
)
TABLE 7.1
Energies
cx+2P
ex -2/1
cx+V2fJ
IX
cx-v2p
"For a discussion of the C
3
H
3
systems. see R. Breslow, Angew. Chem . In t. Ed. Engl., 7. 565
(1968).
"The system C
4
H., is only mctastable with minimum energy in a rectangular (D:!h) shape; cf.
P. Reeves. T. Devon, and R. Pettit, J. Am. Chem. Soc . 91, 5890 (1969).
'"In obtaining the results for the C,H!' and C,H, systems the relation
n-I ~ k2n n-I , k2n n
2: cos-- = )2 sm-- =-
~ .. ., n ~ .. u n 2
has been used, along with other, more familiar trigonometric identities. Although the energies
of the doubly degenerate orbitals can be obtained by using the expressions given above for them
t
(continued)
(populated for
neutral species)
-2
------Bz
______ EJ
-1
E O
2
Orbital Shapes
Delocalization
Energies,
Units of IP\
CaHa 4( v2-1)
and the method described for benzene, sorne rather messy trigonometric algebra is encountered.
A simpler method, which, however, involves sorne quanturn theory not treated in this book. is
available. See H. Eyring, J. Walter, and G. E. Kirnball, Quamum Chemistry, John Wiley &
Sons, New York, 1944, pp. 254-255.
"Cyclooctatetraene, CtH,h is of course well known to be nonplanar and nonaromatic. lt is properly
described as a conjugated but nonarornatic tetraolefin. An MO treatment of the hyp01hetica/
planar (CH)
11
is of sorne interest. however, in respect to the questions of the instability of this
configuration for the free molecule, as well as of the possibility of the stabilization of this
configuration by formation of the anions CKH, and ~ H ~ . There is evidence that the dianion is
planar with a closed shell configuration [cf. M. J. S. Dewar, A. Harget, and E. Haselbach. J.
Am. Chem. Soc., 91, 7521 (1969), and references therein].
158 APPLICATIONS
treating them as though they are, if only as a starting point. The treatment
of benzene just developed is applicable, nzucatis nzutandis, to all of them. The
results are summarized in Table 7 .l. The student can obtain practice in ap-
plying the treatment by verifying these results.
The 4n + 2 Rule
From the results we have obtained for the systems C
4
H
4
.. C
6
H
6
, and CHHK we
can infer a rule .. first discovered by Hckel and now rather well known,
concerning the aromaticity of planar .. carbocyclic systems of the type (CH),..
According to valence bond theory, any su eh system in which the number
of carbon atoms is even would be expected to have resonance stabilization
because of the existence of canonical forms of the type illustrated below for
the first three members of the homologous series:
D
o
o
D
o
o

Plus severa! less stable forms
Plus less stable forms such
as the Dewar structures
Plus many less stable forms
Since a calculation of the resonance energy of benzene by the valence bond
method shows that the greater part of it is a result of the resonance between
the two Kekul structures shown, we might suppose that its homologs would
also have significant resonance stabilization energies. Such conclusions are at
variance with experimental fact, ho\vever, since cyclobutadiene appears to
be too unstable to have any permanent existen ce, and cyclooctatetraene exists
as a nonplanar tetraolefin, having no resonance stabilization of the sort con-
sidered.
Simple LCAO-MO theory provides a direct and natural' explanation for
the facts. It may be seen that for C
4
H
4
, C
6
H
6
, and CxHR the energy Ievel
diagrams ha ve the same general arrangement of levels, namely, a symmetrical
distribution of a strongly bonding, nondegenerate A level and a strongly
antibonding, nondegenerate B level .. with a set of E levels between them. lt
can be shown that such a pattern will always develop in an even-membered
C,Hn system. Now, in o r ~ r to attain a closed configuration (i.e., one with
aJJ electrons paired) of the general type ('!'tt)
2
('f/E
1
)
4
('PE)", it will be nec-
essary to fiJI the Jowest nondegenerate A Jevel and then to fill completely the
first x pairs of degenerate levels above it. For this. 4x + 2 electrons will be
required. It therefore follows that only for systems in which n. which is both
the number of re electrons and the ring size of C,H
11
, is a number expressible
as 4x + 2 (x = 1, 2, 3, ... ) can we obtain a closed configuration. The
numbers, n, meeting this requirement are 6, 10, 14, .... For the other even
integers, namely, 4, 8, 12, ... , we shall always ha vean electron configuration
of the sort ( '1' A)
2
( '1' E.)
4
( lfl E)
2
.
The systems with 4n electrons ( e.g., C
4
H
4
and C
8
H
8
), which as planar
systems with D,,, symmetry would be diradicals. are more stable \Vith a set
of alternating single and double bonds. Thus C
4
H
4
is apparently a reactive
rectangular molecule with two short (double) and two long (single) bonds.
while CsHR is definitely known to contain four single and four double bonds.
Because of angle strain and H H repulsions, the planar form of CKHN is
something like 17 kcal/molless stable than the boat form. * E ven C,llH,h has
been shown to be a polyolefin rather than an aromatic system. while C ,KH IH
and certain related macrocyclic CnHn systems ( called show physical
evidence of aromatic character. t
7.4 MORE GENERAL CASES OF LCAO-MO n BONDING
Tetramethylenecyclobutane
This molecule provides an interesting and entirely genuine illustration of the
predictive power of simple Hckel MO theory; it also serves as a good first
example of how to apply symmetry methods in more general cases. The
molecule was predicted** to have -30 kcal/mol of delocalization energy in
1952; in 1962 it was synthesiied and shown to be stable.tt
The molecule m ay oe assumed to be planar, and we shall adopt the num-
bering scheme shown below.
1 4

2 3
7 6

H2C CH2
* M. J. S. Dewar. A. Harget. andE. Haselbach. J. Am. Chem. Soc .. 91. 7521 (1969).
t For an extensive review of annulenes. see F. Sondheimer, Proc. R. Soc. London, 297A, 173
(1967) .
.,. J. D. Roberts. A. Streitwieser, Jr .. and C. M. Regan. J. At!l. Chem. Soc., 74. 4579 (1952).
it G. W. Griffin md L. J. Peterson. J. A m. Chem. Soc., 84. 339R (1962).
160 APPLICATIONS
The molecule belongs to the point group D
4
,. By using the set of eight
carbon pn orbitals as a basis we can obtain a reducible representation which
will contain the irreducible representations to which the 1r MOs must be long:
r1l s o o o -4 o o -8 o 4
There is a very important feature of this situation, which we can turn to
advantage. It will be observed that the set of four methylene carbon atoms.
numbers 1, 4 .. 5, and 8 .. possess D
41
, symmetry by themselves and that the set
of four carbon atoms in the ring, numbers 2, 3 .. 6 .. and 7, also by themselves
constitute a set having D
4
, symmetry. Furthermore, the atoms in one set are
not equivalent symmetrically to any of those in the other. Non e of the outer
atoms is ever interchanged with any of the inner atoms by any symmetry
operation. Thus each of these sets can be used separately as the basis for a
representation of the group, and if this is done we obtain from each set a
representation, r ~ which reduces as follows:
This means that, if we write an expression for an A
2
u MOas a combination
of all eight orbitals., namely,
we can separa te it into two components .. one made up only of orbitals of the
inner set and one made up only of orbitals of the outer set:
f/1 A!M = N(a2l/J2 + a3ljJ3 + a6l/J6 + a1l/J1) + N(atl/JJ + a4ljJ4 + asl/Js + ascfJK)
Inner set Outer set .
Since the symmetry operations cannot interchange orbitals of the two sets,
each of the subsets in the expression for the A
2
u MO must itself have A2u
symmetry. Thus, in order to construct an orbital of A
2
, symmetry for the
en tire molecule, we can first construct partial orbitals from the atomic orbitals
in each of the subsets and then combine them into a complete MO. A similar
line of reasoning applies to M Os of any other symmetry.
Our immediate problem .. then, is to combine the four outer orbitals into
linear combinations having A
2
u, B
1
,, .. and ER symmetry and also to combine
the four inner orbitals into linear combinations having these same symmetries.
As in the case of the carbocyclic rings, this process can be simplified by using
only the corresponding rotation group C
4
instead of D
4
, .. since the former can
discriminate between the orbitals. That is, if we construct an orbital of A
symmetry in the group C
4
it wiiJ automatically turn out to have A
2
, symmetry
in D
4
h beca use of the inherent symmetry of the pn orbitals namely.
their antisymmetric character with respect to reflection in the molecular plane.
With these considerations in the process of constructing the correct
linear combinations of the subsets proceeds exactly as in the case of the
carbocyclic systen1s. The correct coefficients of the atomic orbitals are simply
the characters of the representations. For the E orbitals we will obtain sorne
imaginary coefficients, but these may be eliminated by taking the appropriate
linear combinations. We can thus write, almost by direct inspection of the
character table of the C
4
group:
'I'A' = + QJ;, + QJll + QJ7)
'I'A'' = l(cPt + QJ4 + cP:; + cPN)
'l'n' = !(l/J2- lP3 + QJ- QJ7)
tJ B" = 4( QJ 1 - cP4 + QJ; - lPK)
1
tJ Ea' = v2 ( lP2 - l/J(,)
1
tJ Eh' = v2 ( QJ 3 - lP 7)
1
lfl Ea" = v2 ( QJ 1 - cP:.)
1
tJ Eh" = v2 ( QJ4 - QJK)
where we use the superscripts i and o to indicate that the combination is made
up of inner or outer orbitals.
We may now solve the secular using these symmetry-correct
MOs, and obtain the MO energies. Thus, for the A orbitals we have the
equation
=o
HA"A"- E
The elements of this determinant are easily evaluated by using the Hckel
approximation:
HA'A' = J f/1 A' A' dr = J ( f/12 + ifJJ + f/Jh + fjJ,) X
X (l/J2 + QJ3 + QJ6 + l/>1) dr
162 APPLICATIONS
= (J c/>2 dr + J c/>2 .7fi/JJ dr
+ J t/J2 .7r ijJb dr + .. + J c/>1 .YfijJ1 dr)
= !(a + p + O + + a)
= !( 4a + BP) = a + 2p
HA"A" = J lfl A". 7f"lp A" dr = J ( t/J1 + + t/Js + tfJR) .ii'
X ( l/J 1 + l/J 4 + l/J 5 + l/J K) d t
= !(4a) = a
HA'A" = HA"A' = J lf!A"I('If!A' dr
= J {e/> + t/J4 + t/Js + t/JR) X(t/J2 + t/JJ + t/J6 + tjJ 7) dr
= i(4P) = P
As before, it is convenient to choose a as the zero of energy and to take
p as the unit of energy. The secular equation for the A orbitals then takes
the form
2- E 1
1 -E =O
which is expanded into a quadratic equation and solved:
E
2
- 2E - 1 = O EA = (1 + V2), (1 - V2)
Following an analogous procedure for the orbitals of B symmetry, we obtain
Hn'B'- E
Hn"B'
The solutions are
Hn'B"
Hn"B"- E
En= (Vz- 1) and
-2- E
1
1 = o
-E
(-V2-1)
For the E orbitals we will obtain two two-dimensional determinants, one
involving the Ea orbitals and the other invo)ving the Eb orbita)s. lt is necessary
to solve only one of them. Choosing the Ea determinant, we have
HE'E'- E
H;oE'
which has the roots
HE'E"
Hr;-r;- - E
-E
1
1 =o
-E
Remembering that p is intrinsically negative, we may use these results to
construct the following energy level diagram, in which the eight electrons
have been added to the lower four orbitals:
----( -1 - ..[2) 13
1/;(2)j -fJ
El----
----(1- .2)fj
(1) M
1/; 8 ---o--o-- ( -1 + -v 2) {3
It can be seen that the order of the levels is such that all of the bonding levels
(those with energies <O) are just filled, and all electrons must have their spins
paired.
The delocalization energy can be calculated easily. The most stable ar-
rangement of the four electron pairs in localized double bonds would un-
doubtedly be the one labeled (a) below, all other arrangements, such as (b)
or (e), containing fewer than four short, strong double bonds. The energy of
this arrangement, taking a = O, can easily be seen, by the argument used in
Section 7.3 for benzene in a Kekul structure, to be 8p. The total energy of
the eight electrons occupying the M Os as shown in the energy level diagram
164 APPLICATIONS
e e
:0:
e e
?)::(: 1 1
1 1
1 1
1 -- 1
1 1
e e
e.
p:
e e
(a) (b) (e)
above is
(2(1 + v'2) + 4(1) + 2(v'2 - l)]p = 9.656P
Hence the resonance or delocalization energy is 1.66/J, taking l/11 =
20 comes to about 33 kcal/mol.
We turn, finally, to the task of finding the actual expressions for the oc-
cupied M Os in arder that we may compute such properties of the groundstate
electron distribution as the n bond orders. For example, which has the
energy (1 + v'2){1, is neither tf1 A'' which has the energy 2/1, nor tf1 A'', which
has the energy O. It is a linear combination of both, and the problem is to
find the appropriate mixing coefficients. As explained earlier, we do this by
returning to the simultaneous equations from which the secular equations
a rose. For the orbitals of A symmetry, we ha ve
and
Either of these equations m ay be used to express the ratio of e; to C
0
, namely,
c;/C
0
= -1/(2 - E)
or
For the correct values of E, these two equations must give the same ratio. If
we insert the energy of (1 + v'2), we obtain
C;/C
0
= - 1/(2 - 1 - Vz) = -1/(1 - Vz) = 1/0.414 = 2.414
CIC = 1 + Vz = 2.414
1 o
which are indeed equal as they must be. When this relationship is combined
with the normalization condition
e?. + c
2
= 1
' o
the actual values of e; and C
0
can be obtained:
C
0
= 0.382 and C; = 0.924
The final expression for the MO is then
= (0.924)(!)(c/Jz + cPJ + cP6 + 4h) + + + cPs + c!J,j
= 0.19l(c/J, + cP4 + cPs + cPR) + 0.462(cP2 + cPJ + cP + cP1)
Proceeding in the same way, we obtain the following expressions for the
other occupied MOs:
= 0.462(c/J, - 4>-t + cPs - c/Js) + 0.19l(c/J:! - cPJ + cP6 - cP7)
= 0.500(c/J, + cP2 - cPs - cP6)
= 0.500( cPJ + cfJ,. - cP1 - cPs)
As examples of the uses of such LCAO-MOs, let us calculate the bond
orders in tetramethylenecyclobutane. 1_he _Qrder of the bond between two
atoms is defined as the sum of the products of the coefttc1ents of the atomic
orbitals. of the Lwo atoms in each of the occupie_d each product
weighted with the number of electrons o.c.c.upying_the MO- Tbus we have for
one of the equlvaient ring bonds, say the one between C
2
and C
3
:
2 X (0.462)(0.462) 0.428
2 X (0.191)( -0.191) = -0.074
2 X (0.500)(0) - 0.000
2 X (0.500)(0) 0.000
0.354
For one of the exo bonds, say the one between C
1
and C
2
:
2 X (0.191)(0.462) = 0.176
2 X (0.462)(0.191) = 0.176
2 X (0.500)(0.500) = 0.500
2 X (0)(0) = 0.000
0.852
From these numbers we can see that the n electrons are much more heavily
localized in the exo bonds than in the ring bonds.
166 APPLICATIONS
Bicyclooctatriene
The presumed structure of this compound, which was first reported in 1960, *
is shown in Figure 7 .4a. It belongs to the point group D
31
,. An LCAO-MO
treatment of it has be en described, t but not in as m u eh detail as we shall give
he re.
Figure 7 .4b shows a numbered set of six pn atomic orbitals which will be
used to construct the TC MOs. Using these AOs as a basis for a representation
of the group D
3
h, we obtain the following results:
r1f 6 o o o o -2
r 11 = A + A'{ + E' + E"
It is to be noted that in this molecule all of the pn orbitals are members of
one equivalent set; there is sorne symmetry operation that will exchange any
two of them. Thus we must consider all six in making up M Os of the appro-
priate symmetries; in fact, it will be- entirely impossible to make up orbitals
of the correct symmetry using any fewer than the entire six. We now show
how this can easily be done.
1 e .
i a aXIS
1 H
HC 1
arll'c-aH
HC
4
HC
\ 5
rvc---cH
BC 1
H
(a) (b)
Figure 7.4 (a) The molecular structure and the numbering of the carbon atoms for
bicyclooctatriene. (b) A sketch showing the orientation of the pn orbitals (<J>,'s) used
in the MO treatment.
:H. E. ~ m m e r m a n and_R. M. Paufter. J. Am. Chem. Soc . 82. 1514 (1960).
C. F. Wtlcox. Jr . S. Wmstein. and W. G. McMillan. J. A m. Chem. Soc . 82. 5450 (1960).
We first note that all types of A orbitals (in D
3
h) have the same symmetry
properties with respect to the rotations constituting the subgroup C
3
; also,
both E' and E" orbitals have the same properties with respect to these ro-
tations. Thus we can use the group C
3
to set up sorne linear combinations
that will be correct to this extent. Since these rotations about the C
3
axis do
not interchange any of the orbitals l/>r,

with those of the set c/J
5
,

we can, temporarily, treat the two sets separately. We thus first write down
linear combinations corresponding to the A and E representations of C
3
As
shown in Section 7.3 for su eh cydic systems, the characters are the correct
coefficients, and we can thus write, by inspection of the character table for
the group C3:
and
and
Again using the procedure explained in Section 7 .3, we take linear combi-
nations of the above expressions for the E orbitals so as to afford real coef-
ficients, obtaining
We have not bothered to normalize these, since they are not yet actually
wave functions.
We now turn back to the character table for D
3
h and note that an Ai orbital
must go into itself on reflection through the horizontal symmetry plane. The
effect of this symmetry operation on the individual atomic orbitals is as follows:
O',{c/J) cP4
c/Js
u,,(
cP4) --7
uh( l/Js) cP2
0',( l/J6) --7 cP3
Thus we must combine the two sums which have A symmetry with respect
to the threefold rotations into one which goes into itself on reflection through
uh, making use of the above transformation properties of the individual A Os.
It is obvious that the correct result must be
We can see from the character table that an A f orbital would a/ so ha ve the
symmetry properties as those which we have so lar consciously built
tnto this LCAO. However, A; and A orbitals differ in their behavior upon
168 APPLICATIONS
rotation about a twofold axis or upon reflection in a, .. The inherent symmetry
of the p orbitals is responsible for the fact that the type of orbital we require
is A and notA i without our having explicitly looked after this. We can easily
confirm that an orbital is needed. If we reflect through the a,. which passes
through carbon atoms 1 and the atomic orbitals transform as follows:
Therefore
- cPt
a,.( QJz) -<!JJ
a,.( cP 3) -
cP4) - cP-t
a,.( l/Js) - QJh
a,.( cPf,) - cPs
a,.(f/l..t') = a,.(c/J + cP1 + QJ3 + cP4 + 4J:. + cPtl)
= (- cP1 - cPJ - cP2 - cP.J - 4J6 - cPs)
- - (c/JJ + cP2 + cP3 + cP4 + cPs + cPtJ)
Thus, as stated, this orbital is an orbital. The correct normalization constant
in the Hckel approximation is 1iv6.
The A'{ orbital, which must change sign on reftection through a
1
, and also
on reflection through a,., obviously has the form
Similarly, the characters of the E' and E" representations under a,, are 2
and -2, respectively, rrieaning that ea eh member of an E' pair will go into
itself on reflection through ah while each member of an E" set will go into
the negative of itself on reflection in ah. These requirements are by
combining the expressions which have E symmetry with respect to the
threefold rotations as follows:
tl E' a = 2cfJJ - cP2 - cP3 + 2c/J4 - cPs - cP6
f/JE'h = c/>2 - cf>J + cPs - lP6
t/IE .. a = 2c/J - cP2 - cP3 - 2c/J4 + cPs + cPt1
t/1 E"h = cP2 - cP J - cPs + cP6
Now collecting all of the above results and normalizing ea eh one we write
the following final list of the LCAO-MOs for bicyclooctatriene:
1
tfiA; = v6 (cPt + cP2 + lPJ + c/J_. + cPs + c/Jh)
1
'lfA = v6 (ePI + q,J. + cPJ - QJ4 - cP5 - cPtJ
1
tf/E'a = V2 (2c/JI - cP2 - cP:. + 2c/J4 - cP:. - 4J,)
tf/E'b = !(c/J2 - cPJ + cP:, - cP)
1
tl E"a = vT2 (2c/J 1 - cP:! - cP 3 - 2c/J. + cPs + cPt,)
tfl E"h = !( cP1 - cPJ - cP; +
We now consider the energies of these M Os. If we calcula te these energies
using the Hckel approximation, we set all resonance integrals other than
H1.t =
H
15
= H
5
:!, and H)(, = Hh:. equal to zero. We then obtain the
following results:
Orbital Energy
a- fJ
a- fJ
Orbital
A
E'
Energy
a+ fJ
a+ fJ
Thus, in this approximation. the A'
1
' orbital is accidentaJly degenerate with
the E" orbitals and the A orbital is accidentally degenerate with the E'
orbitaJs.
To employ the Hckel approxmation in this case, however. is to make
the entire process of using an MO treatment pointless. for we then obtain
exactly the same answer as we would obtain by assuming the molecule to
contain three isolated double bonds. Each double bond can be regarded as
resulting from the formation of two hvo-center MOs, one of energy a + fJ
(bonding) and one of energy a - fJ (antibonding). In the Hckel approxi-
mation, therefore, we find that bicyclooctatriene has no resonance stabili-
zation energy.
The advantage of the MO treatment is that we can rather easily extend it
to take account of interaction between the double bonds. To do this. we
recognize that the energy of interaction between two orbitals such as c/1
1
and
c/J:!, namely. the integral f 4J
1
Ir c/J
2
dr, will not be exactJy zero but will ha ve
sorne finite value, say P'. Also, there will be son1e finite value for the integrals
170 APPLICATIONS
such as f cp
1
.'11' cp
5
dr, which we may call {1'. If we recalculate the energies
including these quantities, we will get somewhat different results. Thus for
'11 Ai we obtain
EA = i J ( cf>t + cf>z + cfJ1 - cfJ4 - c/Js - c/J")
X .'lf(c/J + cP2 + cP3 - - cPs - cP6) dr
= (J c/Jt .'tfi/>1 dr + J c/J 1 dr + J c/J 1.lti/J3 dr - J cfJ
1
.iti/J
4
dr
- J c/J 1.'tti/J5 dr - J c/J 1.1ti/J, dr + J cfJ2 .lti/J1 dr + )
= A(a + P' + P' - fl- P" - P'' + P' + )
= a - p + 2P' - 2P"
Similarly, for '11 E"f, we obtain
E E"" = J (c/Jz - c/J1 - c/J
5
+ c/J
0
) .lt(c/J
2
- c/J 3 - cfJ
5
+ c/J
0
) dr
= (J dr - J c/J
2
.'1ti/J
3
dr - J c/J
2
.fti/J
5
dr
+ J th lti/J. dr - J c/J.1 . 'ltijJ
2
dr + ... )
= t( a - P' - P + fJ" - fJ' + )
= a - p - {1' + P''
Thus, when allowance is made for these additional interactions, we find
that the accidental degenerad es are removed. In the same way, we find that
the energies of the A and E' orbitals are
EA; = a + P + 2P' + 2P"
E E' = a + fJ - P' - fJ"
Let us now see what effect this allowance for interaction between the double
bonds has on the calculated resonance energy. .
The quantity p js intrinsically negative, since it is a measure of the int.er-
action between adjacent pH orbitals so oriented (Fig. 7.5n) asto give a bond1ng
interacti?n. !he integral P' measures the interaction bet\veen p orbitals ori-
ented as n FJgure 7.5b. It can be seen that ths interaction will be antibonding,
(a) (b)
Figure 7.5 (a) The relative orientation of pn orbitals on adjacent, bonded carbon
atoms of bicydooctatriene. (b) The relative orientation of pn orbitals on two non-
adjacent carbon atoms of bicyclooctatriene.
so that P' is positive and also that it should be smaller in absoJute magnitude
than p, since the orbitals concerned are q1uch farther apart and overlap less.
The ratio - P' 1 P can be roughly estimated by using overlap integrals: it is
-0.1. Also by means of overlap integrals, it may be shown that {!' is still
smaller and for the present we shall neglect it entirely. With these consid-
erations in mind we can draw the following energy level diagram, in which,
as usual, we take a as the zero of energy:
-13
o
Ai -----fJ + 2fJ'-:=-l.2{3
E"l---- -fJ- /3-::-0.9f1
E' j-o-o-- (3- t3'= 1.1/3
l--o-o--
The energy of the six electrons occupying these orbitals as sho\\'n is given
by
4(/1 - P') + 2(P + 2P') = 6P
172 APPLICATIONS
The same answer is obtained even when the complete expressions using P"
are employed. Thus the conclusion is that, although interaction between the
double bonds causes certain displacements of the energy levels, it does not
result in any increased stabilization of the molecule. The delocalization energy
rema1ns zero.
7.5 A WORKED EXAMPLE: NAPHTHALENE
In order to summarize and illustrate the methods of MO theory as they have
been developed so far, an example that is elaborate enough to be general
and yet is within the scope of noncomputerized numerical solution may be
useful. The naphthalene molecule is suitable for this purpose, and an outline
of an MO calculation in the Hckel approximation is presented here. In
addition, the naphthalene molecule affords an excellent vehicle for introducing
sorne basic ideas concerning the electronic spectra of unsaturated organic
molecules, including the concept of configuration interaction. These matters
will be considered in the next section. The reader who has mastered the
material in Sections 7.1-7.4 should ha ve no difficulty in verifying all of the
results given in this section, which in tum provide a basis for Section 7 .6.
The naphthalene molecule, as mentioned previously, belongs to the point
group D2h. A set of coordinate axes and a numbering scheme for the atoms
ha ve airead y been shown in Figure 7 .2. The pn orbitals c/>
1
, c/J
2
, , cf>ac,
form three subsets; the members of each are symmetry equivalent to each
other but not to those in other sets. These sets and the irreducible represen-
tations for which they form bases are as follows:
Set 1: cP , cP4, cf>s, cPR
Set 2: cP2, cP3, cP6, cP1
Set 3: c/19, cPw B
11
, 83
8
Thus, there are two A, MOs, two B
28
MOs, three B
1
u MOs, and three BJ
8
MOs. By constructing SALCs corresponding to these representations, the
well-nigh hopeless problem of solving a 10 x 10 determinantal equation is
reduced to the tractable task of solving two quadratic and two cubic equations.
This has airead y be en illustrated in Section 7 .2.
Projection operators may be used to obtain the SALCs. The procedure is
straightforward, since only one-dimensional representations are in volved. For
example,

= + (

+ + ( -1)Cc/>2 + (1)lcfJ2
+ ( + (l)&r'c/>2 + ( -1)aYzcP2
= c/J"1 - cP6 - cP7 + cfJJ - c/Jtt + c/J2 + cPJ - cP7
= cP2 + cPJ - cP6 - c/J7
= i(cfJ2 + cfJJ - c/J6 - c/J7)
The complete set of SALCs (which constitute the solution to Exercise 6.1)
is as follows:
A,: f/1 = !(e/>. - cP4 + e/>; - lPR)
f//2 = l(l/12 - lPJ + cP - l/11)
B 1r1: f//3 = !( l/1 1 + lP4 + lPs + cPK)
f//4 = i(cf>z + l/JJ + cP6 + l/11)
1
f/Js = Vz (l/JIJ + cf>w)
B2g: f//6 = !(l/J, + cP4 - cf>s - cPK)
f/17 = !(l/12 + cPJ - 4>6 - cP7)
B3g: f//s = !( cP1 - cP4 - lPs + c/Jx)
f//9 = !(c/>2 - cfJJ - lPfl + cP7)
1
film = Vz ( cP9 - cfJ w)
The following secular equations may then be set up:
A,: a-E p
=o
p
a-P-E
B.,: a-E p Vzp
p
a+P-E o
Vzp o
a+P-E
B ~ : :
a-E p
=0
p
a-p-E
BJg: a-E p Vzp
/!
a-p-E o
Vip o a-p-E
=o
=o
These determinants can be expanded into polynomial equations ( with ener-
174 APPLICATIONS
gies measured in units of p and referred to a as the zero of energy):
-1 + V5
A,: E
2
+ E - 1 = O; E =
2
= - 1.618, + 0.618
B 1,: (E - 1 )( E
2
- E - 3) = O; E= 1
B2g: E
2
- E - 1 = O;
-3
-2
t
-1
<tl.
-
o
.!:1
e:
o ::S
>;
1:10
'-
cu
e:
l&J
1
2
3
1 VTI
E= 2 = 2.303, -1.303
1 V5
E= 2 - 1.618, -0.618
B(3}
3g
A<2>
u
B(3)
lu
BC2)
3g
BC2>
2g
A(l)
"
BC2>
lu
B(l)
3g
B(l)
2g
BU>
lu
E= -1
-lVTI
E = ---- -2.303, 1.303
2
-2.303
-1.618
-1.303
-1.000
-0.618
1
0.618
1
1.000
1
1.303
1
1.618
1
2.303
Figure 7.6 Energy level diagram for 1l orbitals of naphthaJene.
These results lead to the energy Ievel diagram shown in Figure 7 .6. The
delocalization energy in units of P is given by
2(2.303 + 1.618 + 1.303 + 1.000 + 0.618) - 10 = 3.684
The SALCs f/1
1
through lflw may then be combined into MOs. For the A,
MOs, the simultaneous equations from which the secular equation originates
are as follows:
c
1
(a - E) + c
2
P = O
c
1
P + c
2
(a P - E) = O
Taking a as the zero of energy and p as the unit of energy, we obtain an
expression for the ratio c
1
/ c
2
:
For the Au orbital of energy 0.618, we have c
1
/c
2
= 1.618. Normaliza-
tion requires that ci + = l. Solving for c
1
and c
2
, we obtain c
1
= 0.850
and c
2
= 0.526; the bonding A, MO may thus be written
= 0.85f/1 + 0.526f//2
= 0.425(4>1 - 4>4 + <Ps - <f>s) + 0.263(4>2 - QJ3 + QJ6 - <f>1)
Proceeding in the same way for all the MOs of naphthalene, we obtain the
results shown in Table 7 .2. The student m ay test his understanding of the
procedure by verifying sorne of these results.
There are four nonequivalent types of C-C bonds in naphthalene, these
represented by C
1
-C
2
, C
2
-C
3
, C
4
-Cu, and C
9
-C
10
By using the
MO expressions in Table 7 .2, the n bond order of each type may be computed.
As before, in treating tetramethylenecyclobutane, the bond order, p'""' of the
bond between atoms m and n is defined as the sum of contributions from
each occupied MO, each contribution being given by twice (for two electrons)
the product of the coefficients of l/Jm and <f>, in that MO. For p
12
in naphthalene
we have
2 X (0.301 X 0.231) = 0.139
111(1). 2 X (0.263 X 0.425) = 0.225
't' B2g.
2 X (0.400 X 0.174) = 0.139
flliL: 2 X (0.000 X 0.408) = 0.000
lfl;t: 2 X (0.42.) X O.ZJ) = O.ZZJ
N --
0.728
176 APPLICATIONS
TABLE 7.2 Molecular Orbitals of Naphthalene
MO
"'
tJ n,,
r/11
.
821(
Cll
tJ BJ:
tJ("J.)
n,,
tJOI
-"u
t/(21

t/(2.)

tJIJI
B,
lf/("1)
.-1,
lf/IJI
n.\t
Energy,
Units of P
(a = O)
2.303
1.618
1.303
1.000
0.618
-0.618
-1.000
-1.303
-1.618
-2.303
LCAO Expression
0.301(qJ
1
+ + QJ:, + c/Jx) + 0.231(cf>:! + cPJ + t/J., + qJ;)
+ 0.461 ( c/J 11 + c/Jw)
0.263(t/Jt + t/J .. - t/J; - t/Jx) + 0.425(c/>:! + cP:l - c/Jn - c/J,)
0.400( c/Jt - cP4 - QJ; + cfJH) + 0.174( - c/J:. - t/Jn + t/>
7
)
+ 0.347(t/J.) - c/JJII)
0.408( + cPJ + c/J,. + QJ;) - 0.408( c/J., + e/> m)
0.425( c/Jt - t/>4 + e/>; - c/JH) + 0.263{ c/J! - cp3 + cPt. - 1J)
Q.425(cp, + cP4 - e/>; - c/>H) - 0.263(cp2 + cP:. - c/Jt. - cP;)
0.408( cP2 - cPJ - cf>,., + t/J,} - 0.408{ QJq - c/J,u)
0.400(cp, + c/J4 + cP; + cPx) - 0.174(c/J.! + c/J, + cPt. + cP;)
- 0.347(cf>IJ + c/J, .. )
0.263( c/Jt - cP4 + cP:; - cPH) - 0.425( c/J2 - t/J:. + c/Jt. - lp;)
0.30l(c/J- c/>4- cP:; + cPH)- 0.23l(c/J2.- c/J.,- c/Jt. + t/>;)
- 0.461(cpq - cP
111
)
The orders of the other bonds m ay be figured similarly, to give the results
expressed by the following diagram:
7.6 ELECTRONIC EXCITATIONS OF NAPHTHALENE:
SELECTION RULES AND CONFIGURATION INTERACTION
From the energy level diagram, Figure 7 .6, it can be se en that the lowest
energy transition for naphthalene might be expected to involve excitation of
an electron from the A!,
0
orbital to the orbital. The energy should be
equal to the energy difference between the two orbitals, which is 0.618 -
(- 0.618) = 1.236p. * The next two transitions, involving to and
to excitations, \Vould be expected to have identical energies, namely,
1.618p, but this is not actually the case. In this section we shall look more
closely at the three lowest electronic excitations of naphthalene and shall
discuss selection rules, polarizations, and the effect of what is called config-
uration interaction on two of these transitions.
* At this point the comments on the value of fJ made in Appendix 111 should be noted.
In order to understand the electronic spectra of molecules we must first
recognize that, although we ha ve so far found it convenient to think exclusively
in terms of the configurations of the electrons (i.e., the way in which the one-
electron orbitals populated), this does not directly provide a satisfactory
basis for describing electronic transitions. One-electron orbitals and the elec-
tron configurations are, in themselves, fictional; it is the states arising out of
electron configurations which are real. Wave functions, to be genuine, must
describe states, not individual orbitals (unless only one electron is present)
or configurations. Therefore, as we apply symmetry arguments to the analysis
of electronic transitions in naphthalene or any other molecule, it is the sym-
metries of the states that must ahvays be considered, not directly the sym-
metries of the orbitals to which we assign individual electrons. We can, how-
ever, determine the symmetries of the states from the syn1metries of the
occupied orbitals. In this elementary discussion we wish to emphasize spatial
symmetry properties. We shall therefore omit explicit consideration of elec-
tron spin by dealing only with excited states which, like the ground state,
have a spin quantum number of zero. The designation of state will then refer
only to the spatial or orbital distribution of the electron density.
Finally, it is convenient to introduce a form of notation that allows us to
specify electron configurations as simply as possible. An electron occupying
a B
1
, orbital will be represented by b
1
,, the lower-case letter denoting that
this is the symmetry of the orbital of one electron, not of a true state wave
function. When two electrons occupy the same orbital we shall write bi,. In
this way, the electron configuration of the naphthalene molecule in its ground
state is written
b
, b, b, b'"' .,
, jg ,;;a;;
Note that the electrons are listed from left to right in order of the
energies of the orbitals they occupy and that different b
1
, orbitals are distin-
guished oy using a prime instead of the more cumbersome superscript num-
bers, which might now be confused with the superscripts indicating the pres-
ence of two electrons.
lt is important to realize that any configuration of electrons in which there
are only completely filled orbitals, such as that just considered. gives use to
only one state and that state is totally symmetric. * Thus, the ground state of
the naphthalene molecule has A symmetry.
The Lower, Singly Excited Configurations
In an excited configuration one or more electrons occupy orbitals other than
the lowest ones available. In a singly excited configuration only one electron
* This can easily he demonstrated. using direct product representations for orbitals alone. when
only nondegenerate orbitals are involvcd. The same is true when degenera te orbitals are in volved.
but more sophisticated methods of proof are required.
178 APPLICATIONS
is promoted from the orbital it occupies in the ground configuration to an
orbital of higher energy. Most observed electronic transitions are from the
ground state to one arising from a singly excited configuration. In
the transitions most likely to be in the conveniently observable spectral regions
(visible, near UV) are to states arising from the excited configurations with
the lowest energies. These energies may be estimated in an approximate
manner by considering the energies of the orbitals whose populations are
changed in going from ground to excited configurations.
For naphthalene, the three lowest-energy" singly excited configurations and
their energies" relative to the ground-state energy as zero" are as follows:
B2u:
E = 0.618 - ( -0.618)
-
1.236
8 b
2
b., b
2
b'., b
Ju , lg Jg a, 3:
E = 0.618 - ( -1.000) - 1.618
83,: bfubigbjgb E= 1.000 - ( -0.618) - 1.618
Each of these configurations gives rise to a state with the symmetry that
is specified at the left. The state symmetries are determined in the following
way. First'l we neglect all of the electrons that are in fully occupied orbitals,
since that part of the entire configuration is in each case totally symmetric.
Second, we invoke the fact that for the remaining electrons a product wave
function can be written. These product wave functions are aub
2
g, a,b
31
P and

in the three cases above. The symmetry of each is found by forming

the direct product representation, as explained in Section 5.2.
Selection Rules and Polarizations
Transitions from the ground state to each of the excited states just discussed
may or may not be allowed, in the sense discussed in Section 5.3. We can
employ the criterion presented there to find out which transitions are allowed
and to ascertain their polarizations.
The criterion for a transition being electric dipole allowed is that the direct
product representation for the ground and excited states be or contain an
irreducible representation to which one or more of the Cartesian coordinates
belongs. Since the ground state belongs to the totally symmetric represen-
tation, the direct product representations will in each case be the same as the
representations to which the excited states belong; these are 8"11, 8
3
,, and
8
3
,. As the D"ll, character table shows, the y coordinate belongs to the B
21
,
representation and the x coordinate belongs to the 8
3
., representation.
Thus, the A tg --+ 8
2
, and both of the A --+ 8
3
u transitions are electric
dipole allowed. In all cases (se e Fig. 7 .2) the transitions are "in-plane" po-
larized, the first one occurring by absorption of radiation with its electric
vector vibrating along the y or short axis of the molecule" the other two having
x or polarization.
Experimentally, three transitions are observed in the near ultraviolet, and
polarization measurements indicate that one is polarized while
the other two are Hlong-axis" polarized. The results are given in Table 7 .3.
Configuration Interaction
Although the results in Table 7.3 seem to be in generally good agreement
with theory, there is one notable discrepancy. From orbital energies we would
ha ve estimated that the two A tg B
3
, transitions ha ve the same energy and
that this energy is higher than that for the A B
2
u transition. Experiment
shows, however, that the energy of the A 8
2
, transition lies between the
energies of the two A Jg B
3
, transitions. The cause of this is configuration
interaction.
Configuration interaction is simply a manifestation of the rule (Section 5.3)
that an integral of the form
may ha ve a nonzero value only if lf/
1
and lf/
2
belong to the same representation.
Previously, we ha ve emphasized the converse, namely, that the integral1nust
equal zero when lf/
1
and rp
2
belong to different representations. Now we are
concerned with the fact that only in special cases-by accident, as it were-
will such an integral be zero if the two wave functions do have the same
symmetry. In the present case this means that the two excited configurations
of 8
3
, symmetry will ha ve a net interaction with each other. The result is that
neither of the actual 8
3
, states will be derived purely from one of the excited
B
3
, configurations, as we have thus far tacitly assumed. Instead, there will be
a mixing and splitting, described by a second-order secular equation,
with
TABLE 7.3 Electronic Transitions in Naphthalene
Energy (cm-') Polarization Assignment
31,800 Long axis A,K B_lJI
34,700 Short axis
45,200 Long axis

180 APPLICATIONS
Obviously, this leads to two different energy values, E
0
H th instead of the
same one, E", for both states. The en tire situation is formally analogous to
the occurrence of a 2 x 2 determinantal equation for the energies of two
individual orbitals of the same symmetry. In practice, however, it is a little
different, since there is no simple way to estmate the magnitude of Hr
2
, the
interaction energy. This is caused by interelectronic repulsion and is difficult
to compute accurately.
The way in which the two excited configurations mix and split apart in
energy is called configuration interaction. It is depicted in the energy Ievel
diagram of Figure 7. 7 for the case at hand.
Actually, the example of configuration interaction that \Ve ha ve just ex-
amined is special in the sense that the two interacting configurations have the
same energy before interaction. In its most general form, configuration in-
teraction involves any two configurations with the same symmetry. The result
of the interaction is always to produce an energy difference between the two
states which is greater than that between the two configurations.
When this kind of interaction occurs between vibrational states instead of
electronic states it is called Fermi resonance; we shall discuss this later (Sect.
10.8). In fact, the whole qualitative concept of resonance stabilization as used
in the valence bond theory is just the same principie in still another guise.
7.7 THREE-CENTER BONDING
It is now recognized that there are many molecules in which bonding must
be treated with three-atom units as the smallest ones considered. In other
words, the three atoms must be regarded as one indivisible entity instead of
as a pair of two-center systems with an ato m in common. We will thus assume
50
7 45
o
-
X
- / 1 /
E 40 ""
u ;/
>. Bau = = = = ~ bO
Cii .............
e '
/
/
/
/
/
-----Bau
/
lJ.J .....
35 B 2 u - - - - - - - - - - ~ - - - - - - - - B ~
.......
.......
.....
'-----Bau
Configuration s
S tates
Figure 7.7 Energy level diagram showing how configuration interaction in the B.'"
excited configurations of naphthalene leads to two widely separated 8,., states. The
energies are measured from the ground-state energy as zero.
that electrons may be delocalized over the framework of three atoms instead
of Iocalized between only two of them. If our three-atom framework consti-
tutes a complete molecule or ion in itself, then our treatment will be an MO
one in the strict sense of the word. lf the three-atom entity is instead only a
portion of a larger molecule, then our analysis will not be an MO one in the
strict sense of considering the possibility of delocalization of electrons over
the entire molecule. Nevertheless, its principies and results as they apply to
the selected group of three atoms will not, of course, differ in any essential
way from those obtained when the three atoms are the entire molecule.
As examples of three-center bonding, we will take the following:
(i) Open three-center bonding as found, for example, in the n system of
the allyl ion, [H
2
CCHCH
2
]-, and in the bridge bonding in diborane.
(ii) Closed three-center bonding as found, for in one of the types
of B-B-B bonding occurring in certain boron hydrides.
Open Three-Center Bonding
The nuclear framework, pn .A Os, and a set of reference axes for the a1lyl ion
are shown below:
-
t
Lx
The ion belongs to the C
2
,, point group, and the set of three pn orbitals forms
a basis for the following representation:
c2 ..
1
E C2 Gyz G:ry
r1[ 3 -1 1 -3
r1t = A2 + ZB1
We must now write SALCs that belong to these representations. By ap-
plying projection operators to 4>
1
and then normalizing, we readily obtain
1
lfJA
1
= \12 - cPJ)
1
lf/8 = \12 (</>. + c/J3)
182 APPLICATIONS
Since 41
2
is unique, it must, by itself, form a basis for the B
1
representation.
It is easy to confirm this by examining its behavior under each symmetry
operation. Thus, the other SALC of B
1
symmetry is
, - A,.
'lf 8
1
- t(J2
The secular equation to be sol ved for the B
1
orbitals is then
Has - E Hns' =
0
HnB' Ha'n' - E
The elements of the determinant are evaluated as follows:
H88 = J cfJ2 ./f(/>
2
dr = a
H88' = J (c/J2) ,/f(c/J + cfJJ) dr
= (J cfJ 2./f(/>1 dr + J cfJ 2.11(/l3 dr)
1
= Vz (/1 + p) = Vip
H 8' 8' = i J ( c/J1 + c/JJ)./f ( c/J 1 + c/J3) dr
= i (J c/J 1 ,/t(/l
1
dr + J c/J
1
./t(/l
3
dr
+ J cfJ3 .1tiJ1 dr + J c/J3./f(/l3 dr)
= !(a + O + O + a) = a
Taking a = O, as usual, and inserting into the determinant, we obtain
which has the roots + Vip.
-E Vzp =O
Vzp -E
The single A
2
orbital, which contains only the nonadjacent pair of orbitals
tPt and tPJ., can be seen on inspection to have the energy a ( = 0).
It is now possible to draw the following energy level diagram for the allyl
anion:
---- -J2f3
--oo--
J2P
The energy of the allyl ion in the localized form H
2
C-CH=CH
2
would be
2a + (2a + 2P) or, with a = O, 2p. The energy of the electrons distributed
as shown above is 2v'2p. Hence the delocalization energy is calculated to be
(2v'2 - 2)P = 0.828P = 16 kcal/mol.
To find the actual forro of the occupied B, orbital, we first write the si-
multaneous equations from which the B, secular equation is obtained:
c,(H
88
- E) + c
2
HaB' = O
c,Hnn' + c2(HB'B' - E) = O
The appropriate value of E(v'2 in units of P) and the values of the H's may
then be inserted into either one of these equations, leading to an equation
fClr the ratio e,/ c
2
For example, with the first of the two equations we obtain
c,{O - v'2) + c
2
V = O
-v'2c
1
+ Vc
2
= O
Normalization then requires that
We thus obtain
' 1 ' ~ = }z [ (<f>, + <b,) + tb,]
= i( fjl, + v2rp1 + if>J)
184 APPLICATIONS
The antibonding 8
1
orbital, f / 1 ~ , can similarly be shown to be
Generalization of the Results
We have so far treated only the specific case of rr bonding in a three-center
system of identical atoms using identical rr orbitals. lt is easy to make a
semiquantitative generalization of these results. A few examples should suffice
to show how this is done.
Suppose that we have a system of three atoms each with a rr orbital but
with the center atom different from the end atoms .. as in N0
2
or NOi. The
symmetry is still C
2
,., and so we still expect MOs belonging to the represen-
tations A
2
and 28
1
of C
2
, . The expression for the A
2
orbital will still be, for
reasons of symmetry alone"
and its energy will be a
0
, where the subscript shows that this is the energy
of an oxygen p orbital. We can again set up the same two linear combinations
of orbitals having B
1
symmetry. The elements of the secular determinant \Vill
have the following values:
aN- E Vip
v'ip ao- E
=o
Because we now have two different a's, we cannot obtain the extremely sin1ple
result that we previously did, but qualitatively the results will be very similar ..
as indicated in Figure 7 .8. The two 8
1
orbitals are no longer symmetrically
placed with respect to the A
2
orbital. The degree to which the diagram be-
<:!1 -----E= -
2
d f + -lfF+S)p
J B \
1
~ + 1.7 for f = 0.5
ao=O =======:: -- ___ .._,A
2
-----E= ao=O
........
'
.......
'
.......
.......
'!,!- (1)
n
1
----E= !(f- -Jf2+s)p
z - 1.2 j for f = 0.5
Figure 7.8 Energy Jevel diagram for n bonding in N0
1

comes unsymmetrical depends on how much aN and a
0
differ. In Figure 7.8
the difference has been expressed in units of P as fp. Experimental data show
that the value of f is probably in the range O < f < l. If it is assumed that
f = 0.5, the relative placement of levels turns out as shown in Figure 7 .8.
Let us now look at sorne cases of a bonding instead of rr bonding. A three-
center approach to a bonding finds application in the boron hydrides, in
certain cases of very strong hydrogen bonding .. and in many compounds of
the heavier post-transition elements .. su eh as XeF:! anc.l SF_.. Basically two
cases arise, as shown in Figure 7. 9. 1 n case 1 the center ato m uses an orbital
which is symmetric to reftection in a plane perpendicular to the B B line;
in case II .. the center ato m uses an antisymmetric orbital. lt should especially
be noted even if the system happens to be linear, the results of interest
here are unchanged. The only orbital symmetry property of importance to
us is the behavior upon reflection in a plane passing through the center atom
and perpendicular to a line connecting the terminal atoms. Whether that line
happens to go through the center atom is irrelevant. In effect .. then .. we employ
the group C.., which is a subgroup of both C
2
,. and Dx,, one of which will be
the full symmetry group of the system. Although C\ symmetry allows only
the distinction between symmetric (A) and antisymmetric ( B) orbitals, that
is as much symmetry information as we need. For cases 1 and 11 the following
results are easily obtained, assuming that all three orbitals have the same a ..
which we set equal to zero:
Case 1
l. SALCs
1
A: V (a
1
+ a
3
), a2
1
B: V (a, - a
3
)
2. Secular Equations and Energies
A: -_E Y2fJa2 = O
Y2fJn. {J
13
- E
E = V/JI'2(pl3 = O)
B: E = O(PIJ = O)
3. Expressions for Orbitals
= !(a
1
+ V2a
2
+ a
3
)
Case JI
1
A: V (a1 + a3)
1
B: --;::: (a
1
V2
A: E = OUJIJ = O)
B: -E
= O
v'zpa2 /la.1 - E
E = Y2/la2(/JJJ = O)
1
filA = V (al + a3)
= !(a
1
+ Va
1
- a3)
,/,c2) - l(a
'f'B - 2 1
186 APPLICATIONS
Case I
Case U
or
'
Figure 7.9 The two general cases of three-center bonding. Case 1: center atoro
employs a symmetric orbital. Case 11: center atom employs an antisymmetric orbital.
4. Energy Leve/ Diagrams
o ----rpB o
~ 2
lt will be perfectly obvious on inspecting the results that the two seemingly
different cases are essentially similar in their ultimate results; only the Jabeling
differs. Moreover, bearing in mind that the effect of changing a for the center
atom to a different vaJue from that for the end atoros has only a small qual-
itative effect on the end results (unless the difference in a's is made very
great), we see that in all open three-center bonding situations we shall be
deaJing with essentially the same set of energy levels; one bonding level, one
approximately nonbonding level, one antibonding leve l. lt m ay be observed
that the question of what happens if the a values differ a great deal is not
really very important because a large difference would lead to poor bonding
and instability.
There are two general types of three-center bonding, based on the number
of electron pairs involved. If only one pair of electrons s available, as in the
B-H-B bonds for B2H6, the electrons will occupy the bonding three-center
MO. The expression for the bonding MO, l(a
1
+ Va
2
+ a
3
), shows that
the distribution of electronic charge is roughly even for this case. We thus
have an essentially nonpolar system with two bonding electrons serving to
unite two pairs of atoms.
In the second important type of three-center bonding a total of four elec-
trons must occupy the M Os. One pair will enter the bonding orbital, tending
to give a fairly even distribution of charge anda strongly bonding contribution.
The second pair will enter the nonbonding MO, where the electrons have
little effect on the bond strength but ha ve a marked effect on the bond polarity.
The nonbonding orbital, described by the expression
concentrates its electrons entirely on the end atoms and thus makes them
more negative than the center atom. The presence of electrons in the non-
bonding as well as the bonding MO leads to a situation where one might
speak of "ionic character" in the bonds. For general and fruitful applications
of these ideas, the reader may consu1t various articles in the research and
review literature. *
Closed Three-Center Bonding
In the cases treated so far, we ha ve neglected any direct interaction between
the end atoms. Open three-center bonding can be defined as the situation in
which such neglect is justified. When the effect of P
13
on the results becomes
too great to neglect, we have closed three-center bonding. In the secular
equations written earlier for cases I and ll of open three-center bonding, P
13
was included but then set equal to zero in solving for the energies. Suppose
that we now go to the other extreme for case 1 and assume that P
13
= P
12

The secular equation for the A orbitals takes the form
-E v'2p =O
v1p p-E
which has the roots - p and 2p. Meanwhile the energy of the B orbital is also
changed, becoming - p instead of O (for a = 0). Since there must be a
continuous change in the energies of the M Os as the magnitude of P
13
changes,
we can draw the following correlation diagram:
* An especially good place to start is the followino: R. E. Rundle Rec. Chem t o ~ o 23 J95
(1962). l:J ' ' ' b. ' '
188 APPLICATIONS
o
~ 1 2
1 3 1 3 ~
fJ 12
It will be seen that, when we set P
13
= Pn and all atoms are assigned the
same a value, we are imposing at least C
3
symmetry on the system. Thus,
the results in this limit could have been obtaineq directly by treating a system
of three equivalent a orbitals on three atoros related by a threefold axis. In
that case we should regard the two orbitals with the same energy (-{J) as
components of a doubly degenerate pair. An entirely analogous situation,
differing only in unimportant details, arises if we rework the treatment of the
allyl radical, including P
13
= {J
12
The limiting case here is equivalent to the
n system of the cyclopropenyl radical.
7.8 SYMMETRY-BASED "SELECTION RULES" FOR
CYCLIZATION REACTIONS
It has been shown in Section 5.1 that acceptable molecular wave functions
must form bases for irreducible representations of the point symmetry group
to which the moJecule in question belongs. In earlier sections of this chapter
this most basic of symmetry restrictions has be en employed to find satisfactory
wave functions for a variety of molecules in their ground or excited states.
The science of chemistry deals not only with what molecules are but also with
what they do-that is, with chemical reactions. In principie a reacting system
must conform at all stages to the requirements of the quantum theory .. in-
cluding symmetry restrictions, but generally the explicit analysis of a reacting
system is forbiddingly complex. There are certain important cases, however,
in which symmetry considerations en ter relatively straightforwardly, and by
proper analysis very powerful and general rules may be discovered. Although
rules of this kind have been discussed by several workers, the most compre-
hensive studies have been done by R. B. Woodward and R. Hoffmann, and
such rules are often called ''Woodward-Hoffmann rules.,. An extended ac-
count of the subject has been published by these workers. *
Symmetry-based rules" of the Woodward-Hoffmann type are
relevant when two conditions are fulfilled. ( 1) The reaction must ha ve as its
rate-determining step a concerted process. By this is meant a process in which
the reacting entities come together and are transformed into products in one.
continuous, progressive encounter, without any intermedia tes or any inter-
vention by nonreactng species, such as catalysts. (2) During the entire course
of the concerted process one or more symmetry elements of the en tire reacting
system must persist. Wave functions for the system must then continuously
conform to the requirements imposed by these persisting symmetry elements.
As a practica) matter .. the second condition need not be fulfilled rigorously.
If the reacting skeleton of atoms by itself has certain symmetry elements,
rules based on these symmetry elements may often be expected to hold even
if substituents on sorne of the skeletal atoms formally destroy the skeletal
symmetry. This will be true \vhen the substituents do not differ much from
each other in an electronic sense. For instance, the diene 7.8-1 has C:!a sym-
metry. If one methyl group is replaced by an ethyl group, to give 7.8-11, the
symmetry is reduced to no more than Cs. However, the electronic similarity
(7.8-1)
(7.8-11)
of CH
3
and C
2
H
5
means that the electronic structure of. the skeleton of
7.8-ti will differ very little from that of 7.8-1. Symmetry-based rules pertaining
rigorously to 7 .8-I will hold in practice also for 7 .8-Il. Clearly, if one methyl
group in 7.8-1 were replaced by a group very different electronically from
CH
3
, say COOH or F, then the transferability of rules based on C
2
,. symmetry
to the substituted compound might be questionable. The chemist must use
his judgment and be cautious in dealing with symmetry-destroying substitu-
ents, but experience suggests that a good deal of deviation from the idealized
symmetry is tolerable.
Among the many types of reactions which may be treated by Woodward-
Hoffmann rules, cyclizations in which open-chain olefins are converted, either
thermally or photochemically, to cyclic species are especially important and
can serve well to illustrate the principies in this type of analysis. We
R. B. Woodward and R. Hoffmann, Angew. Chem. lnt. Ed. Engl., 8, 781 ( 1969). ldem .. The
Conservation of Orbital Symmetry. Verlag Chemie, GmbH. Weinheim. FRG, 197(J.
190 APPLICATIONS
shall therefore discuss bimolecular and unimolecular cyclizations, beginning
with the dimerization of ethylene and proceeding to the important and cele-
brated Diels-Aider reaction.
Dimerization of Ethylene
This is the practical example of an olefin cyclization. The reaction
is represented schematically as follows:
11 + (7.8-1)
If the reactant molecules approach each other with their molecular planes
parallel and then pass into the produce cyclobutane, with a planar c4 skeleton,
there are a number of persisting symmetry elements. Until the product mol-
ecule is actually reached, the symmetry is D
21
,; for the final product with a
planar ring the symmetry is D
4
h. Since D
21
, is a subgroup of D
41
, the symmetry
elements giving rise to D
21
, are the persisting ones.
The group D
21
, can be generated by only three symmetry elements, namely,
the three planes of symmetry. Each pair of reflections in the planes genera tes
a c2 rotation about the axis formed by the intersection of the thus
the symmetry properties of a wave function with respect to the planes will
determine its symmetry with respect to the rotations, and the
latter need not be explicitly considered.
The reaction we are considering transforms two rr bonds into two a
bonds-or, more precisely, transfers two electron pairs from n bonding or-
bitals to a bonding orbitals. All other bonds in the molecules remain more
or less unaltered and can be ignored. We focus atterition on the electrons and
orbitals that undergo major change. The same strategy will apply in all similar
analyses. Moreover, since all the atomic orbitals to be considered lie in the
plane containing the four carbon atoms, all atomic and molecular orbitals, a
or n, involved will be symmetric to reflection in this plane. Hence, symmetry
with respect to this plane is invariant and may be ignored. Only the symmetry
properties relative to the other two planes, perpendicular to the plane of the
carbon need 'be considered. This means that the symmetry group C
2
,,,
a subgroup of both D2h and D
41
,, is sufficient, and the arguments to follow
will be framed using only the two mutually perpendicular planes which define
the group e:!,
Each C-C n bond is formed by overlap of pn orbitals on the carbon atoms;
positive overlap leads to the bonding orbital, stabilized by the energy IPI and
containing two electrons. There is also the empty antibonding orbital in which
there is negative overlap; this is destabilized by the energy IPI- When the two
ethylene molecules are so far apart that interaction between them is
the energy level diagram shown at the left in Figure 7.10 is appropriate. The
two bonding and the two antibonding n orbitals are degenerate. As the eth-
-----------1/;A2
_____ lf;A2-
_--------- .., 1/ls2
---1/IB--
Figure 7.10 An orbital correlation diagram for ethylene dir:nerization. Left: two
widely separated ethylene molecules. Center: two ethylene molecules close enough
for significant interactions to occur. Right: cyclobutane; electron configurations cor-
respond to the ground state for ea eh stage.
ylene molecules however, and interaction begins, the will
change in the way shown in the center of Figure 7 .10.
The reason for the changes in energy and the basis for the symmetry
designations of the orbitals can be easily understood by referring to Figure
7 .11. It is clear from comparison of these sketches with the
character table
that the functions shown have the symmetry properties required by the rep-
resentations to which they have been assigned. lt will also be clear that, as
the ethylene molecules approach closely enough for sorne intermolecular
overlap to come into play, the positive overlap in volved in f/1 A and the negative
overlap in IJI
82
will cause their energies to diverge as shown in Figure 7.1 O.
We now turn to the a orbitals, which must arise in order to form the cyclic
product. In the sketch es in Figure 7.12 these are drawn schematically. The
right si de of Figure 7.10 shows the relative energies of the a orbitals in a
semiquantitative way. Their energies relative to one another follow from the
relative degrees of overlap, which are obvious. Their energies relative to the
energies of the rr orbitals have been adjusted to take qualitative account of
the fact that the a overlaps are greater than the n overlaps and, hence, a
bonds are stronger than n bonds.
Inspection of Figure 7.10 as it is now drawn reveals a highly significant
inconsistency. In its present form the figure shows each stage of the system
with the electron configuration of lowest energy that is, in its ground state.
Since the two symmetry planes axz and av= exist continuously, as the relative
importance of the different interactions changes, leading from the left to the
right side of Figure 7.10 (or vice versa), there should be a continuous cor-
192 APPLICATIONS
Uyz
ffi ffi
ffi ffi
Most
strongly
bonding
1/IBl
Uxz
Antibonding
Cly.:
ffi ffi
ffi ffi
Less
strongly
bonding
~ 7
Cfx::
Most
antibonding
Figure 7.11 Diagrams of the p1t orbitals of two approaching ethylene molecules.
relation of the orbitals of each symmetry type. Lines representing these cor-
relations have been drawn. We s e e ~ however, that the 8
1
orbital changes its
nature from bonding at the left to antibonding at the right. Therefore, the
two electrons occupying an orbital of B'! symmetry would end up in an anti-
bonding orbital, according to the orbital correlations shown, and the cyclo-
butane molecule would be formed in a high-energy excited state instead of
in the ground state. 1t could then, of course, lose energy, radiatively or
otherwise, and reach the ground state, but the point is that sufficient en-
ergy to reach the excited state in the first place would have to have been ac-
quired in the course of the concerted transformation. The amount of energy
required is so great as to be unavailable from thermal excitation, and thus
the reaction should be "thermally forbidden." This is in agreement with
the experimental fact that ethylene dimerization to cyclobutane and other
essentially similar, intermolecular reactions of monoolefins do not proceed
at a useful rate under purely thermal activation, even though they are ther-
modynamically favorable.
Although the preceding argument is correct as far as it goes, and leads to
a satisfying insight into the source of the inhibition of ethylene dimerization,
it is possible and desirable to carry it a step further and thus obtain an even
better understanding of this kind of problem. As stressed airead y in Section
Most
strongly
bonding
Y Antibondmg
Less
strongly
bonding
00
1/;A
2
----+------Ux=.
y {:) Most
o antibonding
Figure 7.12 Diagrams of the new a orbitals that form in a planar C.aHK molecule.
7.6, orbital descriptions leading to the specification of electron configurations
are a Iimited and not entirely adequate basis fQT understanding the electronic
structures of molecules. It is frequently necessary to Iook explicitly at the
statcs which arise from the configurations. When degeneracies are not in-
volved, the symmetries of the states can easily be found by forming the direct
product representations of the various occupied orbitals. Let us now do this
for the lowest energy states of both reactants and products in the ethylene
dimerization reaction. In the process we shall employ the shorthand notation
previously (Section 7 .6) introduced, whereby an electron in, say, an A
2
orbital
is represented by a
2
and two electrons occupying this orbital are represented
by
The lowest energy configuration of the n electrons in two weakly interacting
ethylene molecules is, as shown in Figure 7 .10, ai The representation of
the direct product A
1
x A
1
x B
2
x B
2
is easily seen to be A
1
The next most
stable configuration must be aib
2
b
1
, which uniquely defines a state of A
1
symmetry. The only other configuration of interest to us he re is the one which
would correlate directly with the ground configuration of the product, namely
the ayby configuration, which gives a state of symmetry A
1
In a similar fashion,
we find the lowest and next to lowest configurations for the product, as
well as the of the product which correlates directly with the
194 APPLICATIONS
ground configuration of the reactants, and the states they give are as follows:
"b' A a --+ 1
1t is now possible to draw a correlation diagram between the states,
as shown in Figure 7.13. The crucial feature to note he re is that the A
1
to
A
1
correlations which would seem to follow from direct orbitaLcorrelations
cannot and do not actually occur, beca use of what is called the noncross-
ing rule. Two states of the same symmetry cannot cross, in the manner
indicated by the dotted Iines, because of electron repulsion. Instead, as they
approach they turn away from each other so that the lowest A
1
states on each
side are correlated with each other as shown by the full lines. The repulsive
interaction is similar in essence to that involved in configuration interaction
in naphthalene, as discussed in Section 7 .6. lndeed, the noncrossing rule is
no more than a special but straightforward instance of configuration inter-
action.
> 2 A1
afb1----\
Figure 7.13 A correlation diagram for electronic states in the ethylene dimerization
reaction. Left: for the n electrons in the two ethylene molecules. Right: for the new
a electrons in the cydobutane molecule.
From Figure 7.13 it is clear that the thermally activated dimerization, that
is, a ground-state to ground-state process, is inhibited by a substantial energy
barrier. We reach the same conclusion as befo re, namely, that the thermal
reaction is "forbidden." But Figure 7.13 tells us something more. We see
that, if one of the reactant molecules is photoexcited, so that the system
comprised of the two reactant molecules is formed in its first excited state,
of A
2
symmetry, it may cross directly, without any further electronic excita-
tion, to the first excited state, also of A
2
symmetry, of the product. Thus, the
reaction is photochemically "allowed." Although this conclusion could ha ve
been drawn from Figure 7 .lO, it is shown in a more obvious and certain way
in Figure 7 .13. It is an experimental fact that intermolecular olefin dimeri-
zations of the type in question proceed at a useful rate under irradiation.
The Diels-Aider Reaction
In its simplest form this may be shown as follows:
4 4
:( + 11!
:Q:
(7.8-2)
Experimental data indicate that this reaction proceeds thermally under mild
conditions ( often at temperatures below 0C)' apparently in a concerted man-
ner, and that it is not, in general, accelerated by irradiation with visible or
ultraviolet light.
In order for a concerted reaction to occur, the diene and the ole fin must
approach each other in the way shown in Figure 7 .14. In this configuration
"
/
1
"
\
1
' 1 \
1 l
1
' 1
\
1 \
1
\
1
\
1
\
1 \
1 l
'
Figure 7.14 The approach configuration required for a concerted Diels-Aider re-
action.
196 APPLICATIONS
the only persisting symmetry element is a plane of symmetry which is per-
pendicular to the single bond in the diene and to the double bonds in the
reactant and product monoolefins. All orbitals that change in the course of
the reaction must now be classified with respect to this symmetry element.
The analysis here will obviously follow along lines similar to those used for
the ethylene dimerization reaction and can be presented in a less discursive
fashion.
As the student will be asked to show in Exercise 7 .2, the n M Os for
butadiene ha ve the forms and relative energies shown belo\v. Symmetry des-
ignations are relative to the perpendicular plane, A signifying symmetric and
B anti-symmetric.
-2
+
J/.'s(.4)
f23\
+ -
-1
1 4
1j (3)
A
+/\+
o
+ -
1j (2J
8
+/\-
1
+ +
f23\
l/J A(l) + +
1 4
2
For the n bonds in both the reactant and the product monoolefins, the
bonding orbitals are of A symmetry and the antibonding orbitals are of B
symmetry. Finally, the symmetries of the new a orbitals, between atom pairs
1, 6 and 4, 5, and their relative energies are as follows:
~ B
+16
~
1
4 -
~ A
+7
)o
\+
~
+l
\_
B
)o:
00
~ A
')r
+;
\+
AH of the foregoing discussion of orbital symmetries and relative energies
is summarized in Figure 7 .15. From this orbital correlation diagram it is seen
that all filled bonding orbitals in the reactants correlate with filled bonding
-2
t/;8(4)
-
1/;B *(u)
t/;A*(u)
t -1
+JB
JtB *(7r)
t/; }')
ctl.
o
.!3
e:
o ::l
>;
DO
Qj
e:
UJ
v,cn
00
~
B
1
+t/;A. ~
t/;A (7r)
~
0<1
1/IB (u)
t/lll) oc
oc
1/IA (u)
2
Figure 7.15 An orbital correlation diagram for the Diels-Alder reaction. The "'"
and f/ln orbitals at the Jeft are for ethylene, while the others at the left are for
butadiene. The orbitals on the right are for the product.
orbitals in the ground state of the product, and thus the reaction would be
expected to be thermally allowed, in agreement with experiment.
1t is possible to show very generally that for two olefins, having m
1
and m!
n electrons, coming together to form a cyclic olefin with (111
1
+ 1 1 1 ~ - 4)/2 n
bonds, as shown below, the reaction will be thermally allowed when n1
1
+
m2 = 4n + 2 (e.g., 4 + 2 = 6 in the case of the Diels-Alder reaction). On
the contrary, when n1
1
+ 1n
2
= 4n ( e.g., 2 + 2 = 4 for the ethylene dimer-
ization) the reaction is thermally forbidden but photochemically allowed. For
a discussion of this generalization the article of Woodward and Hoffmann
should be consulted.
e-----
~
198 APPLICATIONS
Intramolecular Cyclization: The Butadiene-Cyclobutene Interconversion
One further example of selection rules for reactions is provided by the intra-
mo1ecular conversion of an open-chain, conjugated polyene to a cyclic olefin
with one less pair of 7C electrons. The simplest example is the butadiene-
cyclobutene interconversion:
( []
while in general we have the following reaction:
C=C-(: )-C==C
-2
When a reaction like 7.8-3 is carried out thermally with a 1 ,4-disubstituted
butadiene, the stereochemistry is very specific. Thus we have:
H
H
X
H:QH

H
H
where X = CH
1
or COOR, but not
X
HyA
HA:
X
H
"0x
HAx
H
or
X
"J"
----x-( "H
H
HJx
XrH
H
X
HJH
H-{'H
X
A steric explanation is not adequate, since the second isomer is sterically
as good as the one obtained or even better. Also, the -COOR and -CH
3
groups have quite ditferent inductive properties, so that the course of the
reaction seems to depend on sorne inherent characteristics of the olefin sys-
tems.
The product obtained in reaction 7.8-5 arises by a concerted bond-breaking/
rotatory rearrangement, as shown in Figure 7 .16a. In this process the two
rotations are in the same direction; the process is therefore called conrotatory.
The processes shown in Figure 7 .16b ande would lead to the other geometric
isomers, which do not form to any significant extent in the thermal reaction.
The latter processes involve rotations in opposite directions and hence are
called disrotatory. We could sum up the experimental results neatly by saying
that the ring opening is thermally allowed by a conrotatory process and ther-
mally forbidden by a disrotatory process. We now seek an explanation for
these simple and striking observations.
In the disrotatory processes, the system has one persisting symmetry ele-
ment; a plane perpendicular to the skeleton of carbon atoms. In the conro-
tatory process the one persisting element of symmetry is a C
2
axis, which is
a bisector of the original double bond. The symmetry groups concerned are
Cs and C
2
, respectively, and the wave functions must in each case be either
symmetric (A' or A) or antisymmetric (A!' or B) to the relevant symmetry
operation. Figure 7.17 shows the orbital symmetry correlations for the two
cases. The symmetries and relative energies of the orbitals should be evident
without further explanation. The energies and symmetries of the n orbitals
for butadiene have been given on page 196.
Figure 7.17 shows that in conrotatory ring opening bonding orbitals cor-
H H
H H
}----(
~ "-x
X H
(a) Conrotatory
cA)
w
X ~ "X
H H
(b) Disrotatory
w
~ 1'H
X X
(e) Disrotatory
Figure 7.16 Conrotatory and disrotatory ring openings.
200 APPLICATIONS
u* A"

B--- ---A"

o o
B , __
A ---cox-o,__ A'
--oo-- A''
oa A"
Conrotatory Disrotatory
Figure 7.17 Orbital correlation dagram for conrotatory and disrotatory ring openings
of cyclobutenes.
relate only with bonding orbitals; thus the reaction should be thermally al-
lowed. In disrotatory ring opening, on the contrary, the product cannot be
produced in the ground state because of correlations of bonding with anti-
bonding orbitals and such a process should be thermally forbidden, again in
agreement with experiment.
Finally, let us loo k at the corresponding state correlation diagram, as shown
in Figure 7 .18. In constructing this, we use the direct product rules:
A X A= A}
B X B =A for C
2
AxB=B
A' x A' =A'}
A" x A"= A' for Cs
A' X A"= A"
This diagram confirms our earlier conclusion, namely, that conrotatory
ring opening is thermally allowed and disrotatory ring opening is thermally
forbidden. lt shows further, however, that a singly excited cyclobutene mol-
ecule can undergo an allowed disrotatory ring opening but is unlikely to
undergo a conrotatory one.
It is possible to extend the results just obtained to the general case of a
cyclic conjugated olefin with m n electrons (say cyclohexadiene, m = 4)
opening to a straight-chain polyolefin \Vith m+ 2 n electrons (hexatriene),
whence the following rules are found. For m = 4n + 2 (n = O, 1 2, ... ) the
ring opening ( or is thermally conrotatory and photochemically dis-
rotatory (as we have just seen for n = 0), whereas for 1n == 4n (n = 1,
2, ... ) the reaction is thermally disrotatory and photochemically conrotatory.
For further see the article of Woodward and Hoffmann.
MOs IN ORGAN1C CHEMISTRY 201
2 2
a a A'
Conrotatory Oisrotatory
Figure 7.18 S tate correlation diagram for conrotatory and disrotatory ring openings
of cyclobutenes.
EXERCISES
7.1 Derive the results given in Table 7.1 for the systems C
4
H
4
and CxHx.
7.2 Show that for butadiene [which may be treated either in its trans-planar
( C
21
,) or cis-planar ( C
2
,.) form ], the four M Os ha ve energies of
0.62{J, - 0.62{J, and - 1.62/J, the delocalization energy is 0.48 IPL and
the n bond orders are 0.88 and 0.46.
7.3 The following molecules will afford additional practice in using the Hckel
approximation for hydrocarbons. In each case the n bond orders, the
delocalization energy (in units of 1/JI>' and an indication ( n of whether
the molecule has a triplet ground state are given.
H

0.6tlll:)i.37
0
46
C==CH
1
H
1.21<n
'>'
'-'"
H C=C 0.47
2
0.83 1-1
H
C==CH
2
1.15{T)
2.25
0.71
o.ooTI=r o.so
1.66
0.45 0.72
0.7211
0.28
2.47
202 APPLICATIONS
CP
21
0.73 0.90

2.38
4.51

2.68(T)
3.07
0.39 CH2

1.21

0.54
0
39
CH
2
1.95

{.:)6&"
0.96
{.:)').q

0.79
1.46
7.4 Tetramethyleneethane can be detected spectroscopically [P. Dowd, et al.,
J. A1n. Chem. Soc., 108, 7416 (1986)]. Assuming it to be planar, what
does a simple Hckel calculation predict about its delocalization energy
and spin multiplicity?
7.5 lnvestigate the symmetry restrictions on the reaction of the allyl anion
with the ethylene molecule to form the cyclopentadienyl anion. If the
\s not therma\\y a\\owed, which reactant should be singly excited
for an allowed photochemical reaction?
7.6 The Cope rearrangement is as follows:
o
o
Jut the appropriate symmetry analysis to show that ~ is thermally
l.
8
MOLECULAR ORBITAL THEORY
FOR INORGANIC AND
ORGANOMETALLIC COMPOUNDS
8.1 INTRODUCfiON
In Chapter 7 molecular orbital theory was introduced and discussed with
emphasis on its application to organic molecules in which n systems extending
over planar skeletons constitute the major problem. We now turn to inorganic
compounds of t\VO major types and to one class of organometallic com-
pounds.
We shall begin with the very broad class of inorganic molecules (\vhich
also includes a few organic molecules su eh as methane) consisting of a central
atom A surrounded by a set of n other atoms, B,, BhCn-b' BbC,D,._b-" all
of which are bonded to the central ato m but not to each other. lncluded in
this class are all of the mononuclear coordination complexes, polynuclear
complexes in which direct metal-to-metal interactions are negligible, all of
the oxo anions such as NO) and SOi-, and all of the molecular halides,
oxides, sulfides, and so on, such as BF
3
, PF
5
, SF
4
, XeF
4
, S0
3
, OsO .. and
PC1
3
S.
Another class of molecules that will be discussed contains cluster com-
pounds su eh as the polyhedral borane anions, B H ~ - and sorne metal con-
taining species such as the metal carbonyl clusters.
Finally, we shall loo k briefly at the type of organometallic compound in
\vhich unsaturated carbocyclic systems are combined with metal atoms, as in
ferrocene, ( C
5
H
5
hFe, and related compounds.
204
INORGANIC ANO ORGANOMETALLIC COMPOUNDS 205
8.2 TRANSFORMATION PROPERTIES OF ATOMIC ORBITALS
The symmetry group to which an C, .. . )
11
molecule belongs is deter-
mined by the arrangement of the pendent aton1s. The A being unique
must le on all planes and axes of symmetry. The orbitals that ato m A uses
in forming the A-(B. C, ... ) bonds must therefore be discussed and clas-
sified in terms of the set of symmetry operations generated by these axes and
planes-that is, in terms of the overall symmetry of the molecule. Thus. our
first arder of business is to examine the wave functions for A Os and consider
their transformation (symmetry) properties under the various operations
which constitute the point group of the A(B. C, ... ), molecule.
The wave functions for the hydrogen ato m are known exactly. They are
functions of the three spatial coordinates of the electron and take their most
simple form when we choose these coordinates to be the polar coordinates
shown in Figure 8.1 in relation to a set of Cartesian axes. The point at x, y.
and z in Cartesian coordinates is fixed by r, the radial distance. O P, from
the origin of the coordinate system ( always considered positive) O, the angJe
between the z axis and the line O P; and r/J, the angle between the x axis and
the projection of O P on the xy plane.
The wave functions for the electron in the hydrogen atom are all products
of two functions. First there is the radial function R(n, r), which depends on
the principal quantum number n and the coordinate r. Then there is the
z
Figure 8.1 Diagram showing the relation of polar coordinates, r, O. lf>, to Cartesian
coordinates for the point P.
206 APPLICATIONS
angular part A(O, r/J ), which is independent of both n and r but is a function
of O and r/J. Both R(n, r) and A(O, r/J) are assumed to be separately normalized*
to unity, that is,
J.
x [ R ( n, r) ]2 r
2
d r = 1
(1
f.
'lz f.ll
- [A(O, r/J)J2 sin O dO dcp = 1
n o
Their produce the complete orbital wave function, is then al so normalized
to unity.
Because no symmetry operation can alter the value of R(n, r), we need
not consider the radial wave functions any further. Symmetry operations do
alter the angular wave functions, however, and so we shall now examine them
in more detail. lt should be noted that, since A(O, r/J) does not depend on n,
the angular wave functions for all s, all p, all d, and so on, orbitals of a given
type are the same regardless of the principal quantum number of the shell to
which they beJong. Table 8.1 lists the angular wave functions for s, p, d, and
f orbitals.
In an example worked out at the end of Section 5.1 it was noted in passing
that the p orbital with an angular dependence on sin O cos r/J was called a P:r
orbital beca use the function sin O cos r/J has the same transfonnation properties
as does the Cartesian coordinate x. At this point we shall discuss the trans-
formation properties and hence the notation for the various orbitals more
TABLE 8.1 Angular Wave Functions, A(O, </J), of s, p, d, and f Orbitals
(Normalized to Unity)
Orbital A O ~ <jJ)
Letter Full Simplified Normalizing
Type Polynomial" Polynomial Factor Angular Function
S
1/v;
--
2
V3"hc
z coso
2
p V3/n
sin O cos <P X --
2
V3"hc
sin O sin <P y
2
* It shouJd be recalled that the ditferential element of volume in polar coordinates is , ~ sin fJ dr
dO d</J.
INORGANIC AND ORGANOMETALLIC COMPOUNDS 207
TABLE 8.1 ( Continued)
Orbital A(O, </J)
Letter Full Simplified Normalizing
Type Polynomial" Polynomial Factor Angular Function
v5hc
(3 cos
2
O - 1) --
2zl - xz - yz zl
4
v'i5Ti
sin O cos O cc;>s q, xz
2
d
v'i5Ti
sin O cos O sin q, yz
2
v'i5Ti
sin
2
O cos 2q,
' .,
x--y-
4
v'l5Tii
sin
2
O sin 2</J xy
4
~
sin
2
O cos O sin 2</J xyz
4
x(z2 - y2)
Vl05/n
sin O cos </J( cos
2
O - s i n ~ O sin
2
<P)
4
y(z:z - x2)
~
sin O sin </J( cos
2
O - sin
2
O cos:! <P)
4
fb
z(xl - y2)
~
sin.! O cos iJ cos 2</J
4
x(5x
2
- 3r
2
)
xJ
v'7ir
sin O cos </1(5 sin
2
O cos:! <P - 3)
4
y(5y2 - 3rz)
y3
v'7ir
sin O sin </J(5 sin
2
O sin
2
<P - 3)
4
z(5z
1
- 3r
1
)
z:l
v'7ir
5 cos' O - 3 cos O --
4
., ~ = xl + y2 + z2
, See Appendix IV for further discussion of f orbitals.
fully. Todo so we should recognize that the x, y, and z coordinates of a point
(se e Fig. 8.1) are related to its polar coordinates in the following way:
x = r sin O cos l/J
y = r sin O sin 4>
z = r cosO
(8.2-1)
208 APPLICATIONS
These relations mean that, since x is equal to sin O cos 4> times a constant,
which is of course unaltered by any transformation of the kind occurring in
a point group, sin O cos 4> must transform in the same way as does x. On this
basis the assignment of the subscripts x, y. and z to the p orbitals is clear.
The notation of the d and f orbitals may be deduced- by using relations
8.2-1, as shown in ~ h following examples:
( 1) sin
2
O sin 2r/J = 2 sin
2
O sin 4> cos 4>
= 2(sin O cos ljJ )(sin O sin ljJ)
= 2(xlr)(ylr) = (2/r
2
)xy
= constant x y
(2) 3 cos
2
O - 1 = 3 cos
2
O - cos
2
0 - sin
2
0
= 2 cos
2
O - sin
2
0
Now
hence
(x!rf = sin
2
O cos
2
O
(y 1 r )2 = sin
2
O sin
1
r/J
. ")o
= stn-
Therefore we can write
3 cos
2
O - 1 = 2(z
2
/r
2
) - (l/r
2
)(x
1
+ y
2
)
= constant (2z
2
- x
2
- y
2
)
Thus the d orbital whose angular wave function is a constant times 3cos
2
O
- 1 should be written d
2
=1-x"!-y:! Since in most groups z
1
and x
2
+ y
2
transform
in the same way, 2z
2
- x
2
- y
2
will transform in the same way as z
2
and the
shorter notation d=1 is used.
( 3) si n
1
O e os O sin 24> = sin
2
O cos 0(2 sin 4> cos r/J)
= 2( sin O cos ljJ )(sin O sin 4>) cos O
=2xyz
Hence the f orbital with the above angular functions is called the f.n= orbital.
As a result of the fact that the polynomial subscript to an orbital symbol
tells us that the orbital transforms in the same way as the subscript, we can
immediately determine the transformation properties of any orbital on an
atom lying at the center of the coordinate system by looking up its subscript
in the appropriate column on the right of a character table, if it is a p or d
orbital. An s orbital always transforms according to the totally symmetric
representation, since it has no angular dependence. For f orbitals. the usual
character tables do not give the desired information directly because f orbitals
are not used sufficiently often. However, the assignment of each f orbital
to the appropriate irreducible representation can be worked out for any given
group when needed. Consider, for the f.tr:: orbital in the group D
21
,.
The function xyz is transformed into + 1 times itself by the operations E,
C
2
(z), C
1
(x), and C
2
(y) but into -1 times itself by the operations a(xy),
u(xz), and u(yz). Thus it is a basis for the A, representation. In the following
sections we shall not explicitly discuss f but it is to be emphasized
that their inclusion would not require any new principies.
As an example of using the character tables directly for p and d orbitals,
we will consider the phosphorus atom in PCI
3
By looking at the character
table for the group C
3
,. we immediately learn that the phosphorus orbitals
belong to the following representations:
E: (d.rJ., dy:), (Px Py)
It should be. recalled that, when we say a certain orbital ( or group of orbitals)
"belongs" to a certain irreducible representation, we mean that it is a basis
for that irreducible representation.
The preceding discussion of the symmetry properties of A Os has referred
explicitly to the one-electron orbitals of the hydrogen atorn. However. the
principies can be carried over to the treatment of many-electron atoms. The
wave functions for these atoms may be written as products of one-electron
wave functions. For each electron in a many-electron atom we write a wave
function consisting of an angular function which is the same as the angular
function of an analogous electron in the hydrogen and a radial function
which differs from the radial function the electron wou)d have in a hydrogen
atom because of mutual shielding and repulsion effects among the electrons.
The important point is that, since the angular properties of an electron in a
many-electron atom can be taken as being the same as those of a correspond-
ing hydrogen electron, the symmetry properties of the one-electron wave
functions used to build up the total wave functions for atoms have the same
transformation properties as the simple and exact one-electron wave functions
obtained by solving the wave equation for the hydrogen atom.
8.3 MOLECULAR ORBITALS FOR u BONDING IN
AB" MOLECULES: THE TETRAHEDRAL AB4 CASE
The set of n A-B a bonds in AB, molecules are often thought of as inde-
pendent entities, and no doubt this point of view is pragmatically useful. It
210 APPLICATIONS
is also partly true, in the sense that electron pairs really are concentrated in
the n regions between the A atom and- the n B atoms. However, total reliance
on the concept of localized bonds is neither correct, nor As already
noted, the idea of MOs allows us to begin with a very general and nonre-
strictive framework that is in accord with all symmetry requirements. It is
then possible to learn, by computations, what the wave functions actually
look like. The possibility that they may be highly localized between pairs of
metal atoms is not in any way excluded, but the analysis is a1so not biased
toward such a result.
Let us first see how we determine the symmetries of the a MOs, and,
having done thac the A Os on A and on the B atoms that will be suitable for
forming the proper MOs. It is perhaps easiest to deal with these questions
by going directly toan example, anda tetrahedral AB.s molecule will do very
well.
We first need to find the reducible representation for which the entire set
of a orbitals forms a basis. The AOs on the central A atom (together with
corresponding SALCs on the B atoms) that can be u sed to form M Os will
ha ve to be long to (i.e., form bases for) the representations that con tribute
to this reducible representation. Thus, we begin by applying all of the op-
erations in the molecular point group to the set of a orbitals. For this purpose
we may represent each a orbital by a vector pointing from A to a B atom,
and number these vectors r., r
2
, r .. , as shown in Figure 8.2. Applying the
identity operation, we obtain
r
1
--+ r
1
+ Or
2
+ Or
3
+ Or
4
r
2
Or
1
+ r:! + Or
3
+ Or.s
r
3
Or
1
+ Or
2
+ r
3
+ Or
4
The matrix of the coefficients on the right is a unit matrix of dimension 4 and
hence x(E) = 4.
If we rotate the set of vectors by 2n/3 about the C
3
axis which is coincident
with r
1
, we get
r
1
r
1
+ Or.:! + Or-' + Or
4
r
2
Or
1
+ Or
2
+ r
3
+ Or
4
The character of the matrix of the coefficients he re, x( is equal to l.
INORGANIC AND ORGANOMETALLIC COfvfPOUNDS 211
z
y
Figure 8.2 A set of vectors, r,. rh r
3
. and r." representing the four a bonds from A
to the B atoms in a tetrahedral AB,. mo)ecule.
Proceeding in the same way with a C
2
, an S
4
, and a ath we obtain the
following set of characters for the representation generated:
rtemt
1: 2
3C::! 6S4
o o 1
Ref::rence to the Td character table shows that this representation can be
reduced in the following way:
This means that the four MOs that will be equivalent to the set of four a
orbitals must be eh osen so as to include one orbital of A
1
symmetry and a
set of three orbitals belonging to the T'! representation. The character table
also tells us that AOs of atom A falling into these categories are as follows:
A
1
Orbitals T
2
Orbitals
S (p,f, p ~ . P::)
(d:cy, d,..;:, d.,.z)
We therefore see that to form a complete set of tetrahedrally directed a bonds,
atom A will have to provide an s orbital, as weU as a set of p orbitals or a
set of d orbitals, or both. Let us assume that the identity of the central atom
2U APPLICATIONS
A is such that a set of p orbitals is appropriate. Thus. \Ve might suppose we
are dealing with C H . ~ SiF_., AICI, or ZnCH-.
We now knO\V \Vhich AOs of the central atom \Vill be used to form the
M Os of A
1
and T:! symmetry. We next need to construct SALCs from ap-
propriate a orbitals on the four B atoms that belong to these same represen-
tations. As the u orbital on a B ato m we can choose any AO that is cylindrically
symmetrical about an axis directed to the A atom. This could be an s orbital
or a p orbitaL as shown below.
A 0 or
A(f)B8
We also kno\v how to generate the SALCs by employing the projection
operator technique. This wiJI be very simple in the case of the A
1
SALC but
considerably more time consuming for the T"! SALC. In this case. and in n1any
others, there is an easier way to obtain them.
Consider first the A
1
SALC. It n1ust ha ve the same symmetry as the s
orbital on atom A. This requires that it beeverywhere positive and unchanged
by all symmetry operations, and thus it must be u
1
+ a
1
+ a
3
+ a .... The T
2
SALCs must match the symmetries of the p orbitals on ato m A. By u match.,
\Ve mean, ha ve positive amplitude where th p orbital is positive and negative
amplitude where it is negative. To match the p orbitals, on a 1:1 basis, the
combinations clearly must be
P::: a a
-
a1
-
O'J +
0'.
Px:
U -
0'2 + O'J
- a.
P.,.: u. + a'!
- aJ - a.
A similar set of three SALCs could be obtained to match the d orbitals .. had
that been required. These three SALCs are all orthogonal to each other as
well as to the A
1
SALC, just as the three p orbitals and the s orbital are all
mutually orthogonal.
We are now in a position to form the M Os by allo\ving. overlap of the AO
on A with the SALC of corresponding symmetry on the four B atoms. In
each case. \Ve may bring the central orbital and the SALC together to give
positive overlap or negative overlap. thus forming a bonding oran antibonding
MO. Both of these \Vill have the same symmetry but their energies will be
quite different. The bonding combination, VJh, \vill go do\vn in energy and the
antibonding one, f/1,. up by (toa first approximation) the same amount, each
change being measured from the mean energy of the interacting AOs. This
is illustrated for the A
1
M Os in Figure 8.3.
In a similar diagram for the T
1
component of the bonding the p orbitals
will le at a higher energy. but \Ve shall put the T
2
SALCs at about the same
energy as the ~ 4
1
SALC since w_e treat the direct interaction of the a;'s \Vith
1
S
INORGANIC AND ORGANOMETALLIC COfvtPOUNDS 213
1
1
1
rp,
s - la
1
+ a1 + a3 + \
\
1
1 \
1 \
1 \
\
A atom
\ \
\
\ \
\ \
\ \
\
, ' a
1
+ a., + a
3
+
\
\ 1
\ 1
\ 1
\ 1
\ 11
\ 1
\ 1

S + (a
1
+ 0
2
+ 11
3
+ G.a)
M Os B atoms
Figure 8.3 By combining the central s orbital with the Jigand orbital SALC to give
either positive overlap or negative overlap, we get f/Jh and f/1,, respectively. The expres-
sions written for these are only meant to this sign the actual
expressions for "'" and Y'h contain different coefficients.
I[I.(Tz)
,
\
1
\
1
Y'a(Aa)
\
1
/
\
1
1 \ \
/
1 \ \
p
{ }(
/
\ \
1
,
,,
\ ,
\1
,,
A
,,
S
1 \
,,
\
,,
\ \
,,
\ \
,,
e,o \ \
,,
t) \ \ \
e::
\ \
'
r + u, + u, + u,
t.tl
\ \
) ( } O' - 0'2, + a3 - Cl.a
\ \
/
\ \ /1 etc.
\ \ //1
\ \
'f/b{Tz)
/ 1
\ \ /
1
\
1
\ 1
\
VJ,(A,)
1
'
1
A atom M Os
B atoms
Figure 8.4 An MO energy leve) diagram for a tetrahedral n1olecuJe showing
both the A
1
and T
1
type interactions.
214 APPLICATIONS
Figure. 8.5 Oualitative representations of the three T2 type a M Os for an AB .. mol-
ecule.
each other as negligible. lf we add the T
2
contribution to Figure 8.3 we get
a complete MO energy Jevel diagram, as shown in Figure 8.4. In Figure 8.5
are sketches showing in a qualitative way the MOs.
8.4 MOLECULAR ORBITALS FOR a BONDING IN OTHER
AB,. MOLECULES
Octahedral AB
6
With the case of tetrahedral AB
4
molecules now fully explained, the principies
required to deal with a bonding in any AB,. molecule should be clear. We
take first the important case of an octahedral AB
6
molecule. lt can easily be
shown that the set of six a bonding orbitals gives rise to the follo\ving rep-
resentation:
Bs
..
as
>
y
tC7 6 o o 2 2 o o o 4 2
This representation was determined "easily" because there is a simple way
to determine each character by inspection, without writing out any complete
matrix. If, on carrying out a symmetry operation on a set of (J orbitals a
certain orbital remains unshifted, there appears in the matrix a diagonal
element equal to l. lf, however, that vector and sorne other one are inter-
changed by the operation, two corresponding diagonal elements are equal to
O. Hence, to determine the character of.the matrix corresponding toa given
operation we can use the following simple rule:
The character is equal to tite ntunber of vectors tlzat are unshifted b_v tite
operation.
For the octahedral set of (J orbitals, the operation E leaves all of them un-
shifted, and thus x = 6. All the rotation operations that do not coincide with
any l!onds shift aH of them and thus give x = O. The rotations about the
x, y, and z axes, the bond axes, leave two bonds unshifted but move all
the others (as does a (Jd reftection) and for them x = 2. The (Jh type reftec-
tion plane contains four bonds, which are thus not shifted, and hence for (J
1
,,
X= 4.
The reducible representation can now be reduced giving
To form a complete set of a bonds, we therefore require AOs on the
central atom A belonging to these representations as well as SALCs on the
B atoms belonging to them. When we examine the right-hand side of the 0
1
,
character table, we see that there ares, p, or dAOs matching each of these
representations, as follows:
Ag: S
Eg: (d.?., dx!-,-2)
T,: (Pn p,., P:)
It is, therefore, going to be possible to make the full set of M Os, beca use all
of the necessary orbitals on the central atom are available. There is, of course,
no question that the necessary SALCs can be formed from B atom a orbitals
because that set of basis orbitals itself gives the same representation as the
one worked out above.
To write out the actual expressions for the SALCs, we can, of course, ha ve
recourse to the projection operators. For the T
1
, SALCs this has already been
216 APPLlCATIONS
done in Section 6.3. Rather than do it here for the E,: SALCs (although this
can be recommended as a good exercise for the student), we shall turn instead
to the matching procedure airead y en1ployed in the tetrahedral AB_. case. In
this proce_dure, we look for those combinations of basis orbitals that have
patterns matching the AOs on the centra) atom.
For the A t.r: SALC,!_ which must match the totally symmetric atomic s orbital.,
it is obvious (as it was in the tetrahedral case) that all basis orbitals must
en ter with positive signs and equal weight. Thus., the normalized A .r: SALC
has to be
For the E.'= SALCs we require combinations that match the and dx"!_,.:!.
orbitals. The former has positive lobes along the z axis that have twice the
amplitude of the toroidal negative region in the xy plan e. To match this we
need the following normalized SALC:
To match the d.r- ,.2 orbital it should be obvious that we require
. .
For the T,, SALCs, we require each one to match one of the p orbitals of
the central atom .. and thus \Ve can write by inspection the following ones ..
which are identical to those more laboriously derived by the projection op-
erator technique in Section 6.3:
Vl(a, - a3)
v1(a2 - a4)
V!(as - a,,)
The central AOs and the matching ligand atom SALCs must now be com-
bined, in-phase (added) to give bonding M Os and out-of-phase (with opposite
signs) to give antibonding MOs. The final results may be illustrated in a
schematic way for a case. typical of the transition metals, where the d orbitals
of the central atom are of lower energy than the s and p orbitals, as shown
in Figure 8.6.
Trigonal Bipyramidal AB
5
A trigonal bipyramidal molecule belongs to point group

lt differs sig-
nificantly from the tetrahedral AB.a and octahedral ABh in that all of the
ligands. the B atoms. are not equivalent. If we use the numbering scheme
1 \
1 \
1 * \
1 ___ l..;.;;)":...---. \
...,-
-- 1 \ \
' ' e,* \'
\ 1 \ \
( 1 \ \\
1\ 1 ,,,
S 1 \ 1 \\\
-----({ " \ \\
1\ \ ,,
\ 1 \ \ ,,
\ 1 \ ,,,
\
1\ \ \i
1 ' ' '\ {a SALCs of A1,
1 ..... \ ' r,. ----)r---- E T
..... \ 1' ...
..... , e, Symmet.ry
,..... /
\ a /
' ., /
A orbitals M Os 8
6
orbitals
Figure 8.6 A schematic MO diagram for an octahedral AB,, molecule in which only
u bonding occurs. Lower case letters are used for M Os as for A Os, and the * denotes
an antibonding orbital.
shown in Figure 8. 7, \ve can show that for the en tire set of a bonds we obtain
Paotal (the total representation), but since the two axial bonds and the three
equatorial bonds constitute entirely independent sets, (i.e., are not inter-
changed by any symmetry operation) they may be used separately to give the
representations r:.x and \vhich must, of add up to

Then, by
the usual procedure, we resolve each of these three representations into their
irreducible components:
D3h E 2C
3
JC2 a,, 2S
3
3a,,
ftolal
11
5 2 1 3 o 3 ( = 2A + A; + E')
rax
tT
2 2 o o o 2 ( = A +
req
11
3 o 1 3 o 1 E')
We next look at the D
3
h character table to see what central atom orbitals
are available t meet these requirements and we obtain the following results:
A A1
E
S
P:
(p:r, p..,)
( d.Q'' y!)
Thus, in the usual valence shell, cons.isting of s, and d orbitals, we can find
all the AOs needed to form a complete set of but there are two am-
218 APPLICA TIONS
z
Figure 8.7 Coordinate axes and numbering system for P F ~
biguities about how to do so. First, and most obviously, we ha ve our choice
of two sets of E orbitals. In reality, both pairs would probably make sorne
contribution, but in many cases one or the other might be markedly dominant
owing to its having properties such as shape and energy that best match the
corresponding properties of the E type SALC that is provided by the B atoms.
For PF
5
it is the p orbitals that should be best, and for simplicity we shall
assign them the entire role.
The other ambiguity is a little more subtle. We require an A; orbital for
both an equatorial MO andan axial MO, and we have two A; AOs available.
But, which one is to be used in ea eh place? Actually, each orbital may
contribute in each place, and the real problem is to determine the distribution.
Symmetry cannot provide an answer to this question, and it is only by carrying
out an actual calculation that the distribution can be determined. As a very
rough guess, we might expect the d::! orbital to play a greater role in the axial
A ~ MO since it has its major amplitude in the z directions., whereas an s
orbital has the same amplitude in all directions and hence no preference.
The ~ SALC on the B atoms may be written down by inspection for the
axial bonding despite this ambiguity, sin ce the form must be o-
4
+ o-
5
regardless
of whether the central atom uses a d
1
'-, an s or a mixture of the tw orbitals.
The Ai SALC has to match the Pz orbital and is thus of the form o-
4
- a
5

For the inplane M Os, the one of A symmetry must be symmetric to all
symmetry.operations and thus of the form cr
1
+ u
2
+ u
3
The E-type SALCs
can be deduced either by writing combinatons to match the P.r and p-'. orbitals
(with due regard to trigonometric factors) or by borrowing the results (Section
6.3) already derived by the projection operator technique for the case of c3
symmetry. The en tire set of SALCs, in normalized form, together with the
orbitals of the central atom with which they may be combined to form the
INORGANlC AND ORGANOMETALLIC COMPOUNDS 219
MOs, are thus:
A (axial): v1(cr
4
+ u
5
) d::2
A! (axial): Vl(a4 - o-s) Pz
(equatorial): \/'i(o-
1
+ o-
2
+ o-3) s
E (equatorial): \,1'(2ul - 0"1 - o-3)} + {P.r
\1'1(u2 - o-3) P>
Other ABn Molecules in Brief
The three systems just treated in detail, tetrahedral AB
4
, octahedral AB
6
..
and trigonal bipyramidal AB
5
, should have afforded a thor.ough exposition of
how the MOs for u bonding in ABn molecules general1y are obtained. Listed
below,. in summary form, are results for sorne other important cases. The
student may obtain valuable practice by deriving these results. It should be
noted that the important case of a trigonal planar AB
3
(D
3
h) molecu]e is
implicit in the discussion of trigonal bipyramidal AB
5
, but nevertheless, the
results are given explicitly below.
AB
3
, Trigonal (BF
3
, Coj-, S0
3
)
Dy, 1 E 2C3 3C2 uh 2S1 3u,.
rn 3 o 1 3 o 1
lz /Y

1' .....
1 ' ........
1 A-B ---+x
.JI' .... 1
B
..,-
3 ....
Irreducible
Representation
A Os
on A SALCs on B Atoms

E'
s,
(p:c, p.,)
(d.r.v' dr-.v:!)
W(a. + 0"2 + o-3)
vl(2ut - 0"1 - o-3) and
AB .. , Square [ AuCI, XeF
4
, Ni(CN)l-]
E 2C4 C2 2C2 2C:! i 2S.. u,, 2o-,,
tz /y
B ------- B
, 3, , , 2
, ... ,, ,
, .. , ,
i
B ----- B
4 l
o o o 2 o 4 4 o 2 o
. "
X
220 APPLICA TIONS
Irreducible
Representation
A Os
on A

d.r-."'!
(P.n P.v}
SALCs on B Atoms
! (a, + a
1
+ a
3
+ a
4
)
i (a
1
- a"!. + a
3
- a
4
)
vi (a
1
-- a
3
) and v1 (a
2
- a
4
)
AB
5
, Square Pyramidal (Ni(CNH-, lnCIJ-)
r
c4,.
flutal
n
rax
Q
fbilSC
a
Irreducible
Representation
r:x
E
2C4
5
1
4
. l
1
o
A Os
onA
P::
S
d.r-i
(px., p_..)
(d., . , d_..:)
c2
1
l
o
2a,.
?
-a,
B. /y
3 1
B B
1 4, .; 1 3
, ... .;' 1
1 .. , 1
1 1
fii"A'
2 o
B ---B
S 2

X
SALCs on B Atoms
a
!(a2 + a3 + a. + as)
- a3 + a. - a5)
\l!(a
2
- a
4
) and \I(a
3
- a
5
)
AB
6
, Trigonal Prism [MoS
2
, WS
2
, Re(PhC(S)C(S)Ph)J]
D3h E 2C3 3C
2
a,, 2S3 3a,.
z
o o
Irreducible A Os
Representation on A
A; S, d:.-:
Ai
Pz
E'
(Pn
o o 2
SALCs on B Atoms
'V1(a
1
+ a
1
+ a3 + a .. + as + a6)
Vl(a
1
+ a
2
+ a
3
- a .. - a
5
- a
6
)

+ 2a4 - a! - a3 - as - ab)
and l(a
2
- a
3
+ - a
6
)
E"
AB
8
, cubic
E
8
2 ... o ...
Irreducible AOs
Representation on A
J41g S
T 1g ( d.rz, d)':, d.r_v)
A !u
"Vt;(2a
1
- 2a
4
- a
2
- a
3
+ a
5
+ ah)
and l(a! - a
3
- a
5
+ ab)
r
4
SALCs on B Atoms
Vi (a t + + a K)
'\ll(a
1
- a
2
- a
3
+ O".c - as + att + a1
- a
8
), etc.
W(a
1
- a
2
+ a
3
- a .. - a
5
+ a& - a1
+ as)
V{a
1
- f!
2
- a3 + a .. + as - a6 - a1
+ aR), etc.
AB
8
, Dodecahedral Ta04(dmpe)t] r
D211 E 2S
4
C1 2C 2ad
o o o 4
--;,
Irreducible A Os
Representation on A SALCs on B Atoms
At S !(a
1
+ u
1
+ a3 + u
4
)
Al
d-1 !(as + a
6
+ a1 + as)
B2
d,
:r-y-
!(a
1
- a
2
+ a
3
- a
4
)
B2
P:
i(as + a6 - a1 - aH)
E
(Pn Pv)
V(a
1
- a
3
) and v'!(a2 - a.a)
E
(d:rz, d,-z)
v1(a
5
- a6) and v1(a7 - aH)
11
dmpe = 1,2-bis(dimethylphosphino)ethane
222 APPLICATIONS
This example has several features \Vorthy of comment. The eight B atoms
comprise two nonequivaJent subsets, B t - B4 and B
5
- Bs. However, each
subset gives the same representation ( 4 O O O 2) and hence gives rise to SALCs
of the same form which will central A Os of the same symmetry types
with which to interact. In the above results, we have placed each SALC
opposite to the AO it seems tikely to overlap with best when the shapes of
the AOs are considered. For example, the E SALCs formed by the atoms of
the flattened tetrahedron (B
1
- B
4
) will overlap better with the Px and p,
orbitals than the E SALCs formed by the atoms of the elongated tetrahedro
(B
5
- B
8
). Of course, this is only a crude approximation anda real calculation
would doubtless show that all orbitaJs of the same symmetry undergo sorne
degree of mixing.
8.5 HYBRID ORBITALS
The MO approach to (J bonding in ABn nioJecules is widely regarded as the
most generally useful one for two reasons. First, it is rigorous with regard to
the symmetry properties of both the basis orbitals and MOs. Second., within
this symrnetry-based framework, the numericaJ accuracy of the results can
then be taken to any level desired if su:fficiently elaborate computations are
done.
On the other hand., the results of an MO analysis might seen1 at odds with
the obvious fact that in molecules such as CH
4
or SFb all of the (J bonds are
equivalen t. Actually, there is no inconsistency. If the electron density in the
mo)ecule is computed from the wave functions for all of the filled MOs (e.g.,
the A
1
and T
1
M Os in CH.) the equivalence of all the bonds will be evident.
At the same time, the fact that these equivalent bonds are a result of the
presence of electrons in nonequivalent M Os is also experimentally verifiable
by the technique of photoelectron spectroscopy.
If all the vaJence shell electrons of methane (i.e., all but the carbon ls
electrons) were equivalent, the photoionization spectrum of methane would
show only one peak. In fact, as seen in Figure 8.8, it shows two, which have
energies and intensities appropriate to the A
1
and T
1
M Os.
Long before it was .possib1e to perform MO calculations on even the sim-
plest molecules, the equivalence of the bonds led to the development of a
different conception of the bonding in ABn molecules, in which nonequivalent
AOs on the central atom are combined into hybrid orbitals. These hybrid
orbitals provide a set of equillalent Job es directed at the set ( or subset) of
symmetry equivalent B atoms. lt is therefore obvious that all A-B bonds
to all equivalent B atoms will be equivalen!.
In spite of the advances in computational chemistry, which ha ve made the
MO approach highly feasible and widely used, the hybridization approach is
still of vaJue and interest. Moreover, it too has a firm in the
symmetry properties of the molecule. 1 t is therefore worthwhile and appro-
JNORGANJC ANO ORGANOMETALLIC COMPOUNDS 223
12 15 18 21 24
lE(eV)
Figure 8.8 The photoelectron spectrum of CH,.. Adapted by permission from A. W.
Pons and W. C. Price, Proc. R. Soc. London, A326, 165 (1972). The ionization
( -14 e V) is more in tense than the A
1
ionization (-23 e V) partly beca use of the 3: l
ratio of populations. It is much broader because of a J ahn-Teller effect.
priate to explain and illustrate it. To do let us take the case of a trigonal
planar AB
3
molecule of D
3
h symmetry.
To determine how to form a set of trigonally directed hybrid orbitals., we
Legin in exactly the same way as we did in the MO treatment. We use the
three a bonds as a basis for a representation, reduce this representation and
obtain the results on page 219. However., we now employ these results dif-
ferently. We con elude that the s orbital may be combined with two of the p
orbitals to forrn three equivalent lobes projecting from the central atom A
toward the B atoms. We find the algebraic expressions for those combinations
by the following procedure.
The hybrid which we m ay designate as <P
1
, <P
2
, and <11
3
., (Fig. 8.9)
will be expressed in terms of the atomic s, P.n and p-'. orbitals by the following
set of equations:
Our problem now is to evaluate the coefficients C; and the procedure for
224 APPLICATIONS
::
(a)
''
%
(b} (e)
Figure 8.9 (a) Orientation of AB
3
molecule in Cartesian coordinate system. (b) Set
of equivalent hybrid orbitals, cl>
1
, 4 l
~ <1>
3
(e) Set o ~ equivalent a orbitals on penden!
doing so is as follows. The set of coefficients we seek forms a matrix., and the
above set of equations can be written in matrix form:
This matrix te lis us how to take. a set of atomic wave functions, each belonging
to a particular irreducible represen tation and listed in a specified order, and
combine them into a set of three equivalent functions. It should evidently be
INORGANIC AND ORGAN0f\1ETALLIC COMPOUNDS 225
possible to carry out the inverse transformation and to express it in matrix
form, namely,
where the matrix is the inverse of the (f. matrix. Thus, one way to determine
the elements of the fi matrix would be to ha ve the q. matrix and take its
inverse. This is a very simple process because the inverse of an orthogonal
matrix is the transpose. Thus, we can ohtain the (, matrix easily if we know
how to obtain the !1 matrix. A little reftection \Vill show that \Ve already
know hov.' to obtain the matrix. The matrix describes the transformation
of a set of three equivalent basis into a set of linear combinations
having the symmetry of the AOs, which, in turn, have symmetry correspond-
ing to certain irreducible representations of the molecular symmetry group.
As we know., projection operators genera te su eh symmetry-adapted linear
combinations, the coefficients of which are the elements of the desired matrix.
Thus, if we use the projection operator technique to transform a set of equiv-
alent a orbitals-either the hybrid orbitals on the central atom or the (J orhitals
on the pendent atoms-into SALCs we obtain a set of coefficients which
constitute the matrix. We can now state and illustrate the three steps in
a systematic procedure for forming hybrid orbitals.
l. Fonn SALCs froln the set of equivalent orbitals on the penden/ atonzs. As
noted above and emphasized in Figure 8. 9, we could use either the hybrid
orbitals on atom A or the (J orbitals on the B atoms, since their symmetry
properties are the same. We choose the orbitals on the pendent atoms because
the application of projection operators to these is exactly as previously ex-
plained in Chapter 6. The results obtained are
lS
lfi
1
(AD = + th + o-3)
lf/2(E;) = (2u, - u2 - uJ)
lf'l(E.) = (u2 - O:J)
2. The nzatrix of the Cf:?efficients is written and its in verse taken. The matrix
[
1/\1:3 1/\13 I/\/3]
2/'6 -1/'6 -1/'6
o I/V2 -1/Yz
226 APPLICATIONS
and the in verse of this matrix (its transpose) is
[
I/v3 2/\16 O J
1tv3 -tt\16 Itvz
1/v3 - 1/\16 -ltv'2
3. matri-r so obtained is applied to a colunuz vector of the A Os (in the
correct order of the representations to whclz they belong) to generate the hy-
brids. We therefore write
-
I/v'3 -1rv6 -11Vi P.v
= (1/v3) s - (1/\16) Px + (1/v2) P.r = l>2
[
(1/vJ) s + (21\I'ID Px J = epi
(I/v3) s - (It\16) P.r - (I/v2) Py = ep3
These hybrid orbitals are commonly designated sp
2
hybrids, it being under-
stood that the explicit form of mixing is that shown above. In a very similar
way sp
3
hybrids give rise to four equivalent orbitals directed to the vertices
of a tetrahedron. The explicit forms here can be written virtually by inspection.
With the coordinate system in Figure 8.2 and the hybrids numbered as are
the vectors r;, we can write
epi = (s + Px + p,. + pz)
<l>2 = ! (s - P:r + Py - P:).
epJ = i (s + Px - p_, - P::)
ep 4 = l (s - P x - P J' + Pz)
Hybrid Orbitals in Other lmportant Cases
There are several other symmetries of ABn molecules for which hybrid orbitals
on atom A are often wanted. The results for these are summarized below.
AB
4
, planar. We require A
18
+ B
1
g + E, atomic orbitals. The two possible
combinations are + d:r2_,,'! + P:r + Pv and s + dx'!-,.2 + Px + Pv The latter
is usually eh osen and briefly as dsp
2
hybr{dization. However, not
just any choice of d or p orbitals is implied, but rather only the specific one
just mentioned.
AB
5
, trigonal bipyramidal. We require AOs corresponding to 2A +
Ai + E' (of group D
3
h}. There are two possible orbitals (s, one
orbital (p.) and two choices for the E' orbitals, (Pn p_,.) or (dx-''' We
ca.n have either dsp
3
or tPsp hybrids.
AB
6
, Octahedral. This is perhaps the best known case, where the A
1
.t: +
Eg + T
1
u set of orbitals is made up of s, d.r"! __ ,,!) and (P.P p_,, Pz) and d
1
sp
3
hybrids are often cited in the chemical literature.
8.6 MOLECULAR ORBITALS FOR 7C BONDING IN
ABa MOLECULES
For a great many ABn molecules 1r bonding as well as (J bonding is important.
This is particularly true in certain classes of compounds such as metal car-
bonyls and oxometallates. For tetrahedral and octahedral .. for
example; MnO, Ni(C0}
4
, and Cr(C0)
6
, as well as for ions such as
MoOCI MO treatments that do not include n: bonding between the metal
atom and the ligands would be utterly valueless. We therefore tum now to
this topic. Although 7C MOs can be handled by employing the same funda-
mental principies as were used for a bonding, it is instructive to look at n:
bonding in detail, especially for the important cases of tetrahedral AB
4
and
octahedral AB
6

Tetrahedral AB
4
To determine the possibilities for n: bonding we must
first determine the representations spanned by a complete set of n orbitals
on the set of B atoms. Figure 8.10 shows vectors representing P.r and p,.
orbitals, that is the p orbitals perpendicular to the A-B (J bonds, on each
B atom. The orientation of each vector pair about its own z axis has been
chosen so that the y vectors are all parallel to the xy plane of the coordinate
z
\ 1
\ 1 /
\ 1 /
1/0::

X//{/
/ J-------
/ ,.

//
"""'

83
Figure 8.10 A set of vectors representing the x-type p orbitals on the four B atoms
of a tetrahedral AB,. molecule.
228 APPLICATIONS
system for the complex as a whole. With this vector set as a basis, we obtain
the following results:
-1 o o o
fl( = E + T. + T2
(The character - 1 unqer C
3
comes from the matrix describing the mixing of
X and y under a el about z. See Table 6.1.}
This means that SALCs of E.. T
1
, and T
2
symmetries can be constructed
from the eight n: orbitals of the B atoms. But, n MOs can be formed only if
there are appropriate A Os on atom A with which they can interact. (Strictly
speaking., a SALC on the B atoms alone is also a MO, but it is a nonbonding
one.) Inspection of the TJ character table shows that the following AOs are
available.
T
1
: None
This result has two important consequences. First, it is impossible to form
a con1plete set of 7C bonds (i.e., two to each B atom), because there are no
orbitals on atom A of T
1
symmetry (assuming, as always, that \Ve consider
only s-, p-, and d-type orbitals). Second, the n: bonding will not be entirely
independent of the a bonding, since T
2
MOs are required for both. There
are two sets of T
2
type A Os and thus it is possible to have both a and x M Os
of T
2
symmetry. In the general case, there will be both p and d orbital
contributions to both, although in specific cases it is possible that little mixing
will occur.
An energy level diagram _qualitatively applicable to an MCH- complex of
a metal from the tirst transition series is given in Figure 8.11.
Octahedral AB,. As before, we employ a set of vectors (Fig. 8.12) rep-
resenting n: orbitals on the B atoms to determine the representation corre-
sponding to a complete set (12) of A-B n: bonds. We obtain the following
results:
0
1
, E SC; 6C
2
6C
4
3C
2
t 6S_. 8Sn 3a,, 6a,
r, 12 o o o -4 o o o o o
INORGANIC ANO ORGANOMETALLIC 229
Aatom
orbitals
4p:7'2---
-4s: A
3d: E.T2

T(r)
At(IT). T2(a) (
E(r), T2(r)
Molecular
orbitals
At(u)
T2(u, r)
E(r*)
V
B atom
orbitals
{ pcr SALC"s:
At. T2
p1r SALC's:
E, T, T2
us SALC's:
--- At,T2
Figure 8.11 An approximate MO diagram for a tetr.ahedral AB,. molecule or complex,
where A is a + 2 ion from the first transition series and the are O or Cl atoms.
We now examine the right side of the 0
1
, character table to see what AOs
on A are available to match these requirements. Within an s, p, d manifold,
there are none for _T
1
g or T
21
,. For
and T
1
u we have, respectively (d.r_,., d.r=,,
d.,.:) and ( Pn p,., P:) If we loo k back at our previous treatment of a bonding
in octahedral AB
6
we find that the T
1
, set was also needed there. In view of
the need for strong a bonds and the fact that the set of p orbitals is very well
shaped to form a bonds, this is normally their primary role. We are then left
with only the T
2
r type of A-B n: bonding.
SALCs to match the
orbitals of ato m A, namely, the dX)., d.r= and

orbitals, can easily be written by inspection, as illustrated in Figure 8.13, the
result being:
When these two sets of T
28
orbitals are combined with positive overlap (as
230 APPLICATIONS
l,
X ~
Figure 8.U Coordinate system for an octahedral ABh molecule or complex ion.
in Fig. 8.13) to give bonding MOs and with negative overlap to give
antibonding MOs, we have completed our development of n: bonding
in the octahedral AB
6
molecule. There remain SALCs of T
11
,, T
21
,, and T
~ :
symmetry that are nonbonding in character. When all of these M Os are added
to the diagram we previously had (Fig. 8.6) for the a bonding, we get
the complete MO diagram for an octahedral AB
6
molecule as shown in
Figure 8.14. Shown here is the case .where tilled ligand n: orbitals lie lower
than the metal T2N orbitals. Such n:-donor ligands destabize the metal T2R
orbitals and transfer sorne n: density to the metal atom. There are also ligands
(CO, NO, phosphines) with empty n: orbitals that le above the metal T
2
~ :
orbitals. These stabilize the metal T
2
,. orbitals and receive sorne n: density
from them.
8.7 CAGE AND CLUSTER COMPOUNDS
Cluster compounds are those in which a group of atoms forms a polygonal or
polyhedral array to which ligands are attached around the outside. In most
cases, there is nothing in the center, although important cases are known
where a small central atom such as H, Be, B, C, N, or Si is present. For the
empty clusters, the problem of setting up M Os that conform to the symmetry
of the cluster is formally analogous to that of determining the SALCs for a
set of ligands, that is, the B atoms in an AB" molecule.
In this section we shall deal with a few of the principal classes, choosing
for each one a prototypal example with which to illustrate how symmetry
considerations are brought to bear.
INORGANIC ANO ORGANOMETALLlC COMPOUNDS 231
z
t
Bs
86
(a)
(e) (el)
Figure 8.13 Diagrams showing how the T,., SALCsof B atom x orbitals are derived
by matching to the T
21
orbitals (dzr' dn, d ... :) on the A atom.
Polyhedral Boranes: B
6
H ~ as Prototype
Much of the current literature on metal atom cluster species employs bonding
concepts that are derived from MO treatment of the polyhedral borarie anions,
B n H ~ . We thus begin by discussing these species, of which the most important
examples are shown in Figure 8.15. We shall deal with the BJ-ij- ion in detail
to illustrate the general approach to these systems.
Each boron atom has four valence shell orbitals, s, P.n p,., and P:. We
choose a local coordinate system for each boron atom so that its z axis is
directed toward the center of the octahedron and the x and y axes are arranged
as for the ligand atoms in Figure 8.12 where we were dealing with an octa-
232 APPLICATIONS
Aatom
orbitals
4p:Tl,---
4s: A1g
3d: En, T2# ---
Molecular
orbitals
---E
11
(CT*)
7}R(7T). TJ,(1r) [:---
T2t,(7T) l---
B atom
orbitals
su SALC's:
---
Alg E
6
, T
1
,,
A
1
.,(a), E
11
(rr), /
_/
TJu(a)
Figure 8.14 An approximate MO diagram for an octahedral ABh molecule or i o n ~
where A is a + 2 or + 3 ion from the first transition series and the B's are F. O, or
Cl atoins.
hedral AB
6
complex. CJearly, the representations for which the six P: orbitals,
the six s orbitals, ~ n d the six pairs of p .. , and p-'. orbitals form bases are exactly
the same as we obtained when working out SALCs for a and n bonding in
an octahedral AB
6
molecule, and from these we obtain the same irreducible
representations, namely, the following:
The 6 s orbitals:
A,R + ER + Ti u
}
radial
The 6 Pz orbitals: A1
8
+ E ~ + Tau
The 12 Px' P.r pairs: T
1
K + T
2
: + T
1
, + T2,
tangential
1 1
4
2
5 2
6
5
6 7
1
1
1
4
1
2
10
6
12
Figure 8.15 The shapes of the el oso BnH;- ions for val u es of n from 5 to 12. The
conventional numbering scheme and point group for each one are shown.
Here we designate orbitals that point directly in or out of the cluster as radial
and the ones mainly on the surface as tangential.
We next note that to form the set of B-H bonds either the set of s orbitals._
the set of P: orbitals or sorne mixture of the two must be used, thus leaving
only one set of radial orbitals that point in towards the center of the
octahedron. The
SALC formed from these orbitals (Fig. 8.16) will be a

strongly bonding MO because each of the AOs overlaps positively with its
four nearest neighbors as well as with the one opposite to it. The other SALCs ..
of E1: and T
1
, symmetry .. as shown in Figure 8.16, are either distinctly anti-
bonding because they entail negative overlaps of adjacent orbitals (E:J:) or
slightly antibonding because of weak negative overlaps across the octahedron
(T,,).
Finally, we -turn the M Os formed by the 12 tangential (P.: and p_,.) orbitals.
For the
type we already have the shapes from our discussion of the n

bonding in an octahedral AB
6
molecule. The others can be obtained by use
of projection operators._ and one representa ti ve of each set is shown in Figure
8.16. For these orbitals the overlaps implied by these drawings show that the
T2x and the T
11
, sets are bonding and the
and T
1
, antibonding.
The overall result of the foregoing analysis is that the B
6
cluster itself has
the following bonding MOs: A
1
n, T"!.r:" and T
1
,. Since these can hold._ 2 .. 6, and
6 electrons, respectively, a total of 14 electrons are required to occupy the
234 APPLICATIONS
T-q
Bonding Antibonding
and two
similar
ones
Figure 8.16 Schematic representations of the M Os of BJf-. The upper set are those
formed by inwardly directed radial orbitals (s, Pn or sp% hybrids). The lower ones
show one each of the tangential sets.
bonding orbitals completely. Each boron ato m has 3 valence shell electrons,
one of which is used to form a B-H bond. Therefore the 6 boron atoms can
supply 12 electrons to the cluster bonding M Os. Complete filling of all bonding
M Os requires the addition of2 more electrons thus accounting for the stability
of the B
6
H
6
unit as a dianion.
From tms simple beginning, a Iarge body of less rigorous but useful bonding
theory has been eJaborated. It can be shown that for all boron clusters of the
BnHn type which are closed polyhedra (Fig. 8.15), with n from 5 to 12, the
optimum number of bonding electron pairs .within the cluster is n + 1, jus1
as we have found for B
6
H ~ . Therefore, all of these closo boron clusters arf
predicted to occur as dianions, and all of them do.
There are a number of other important boron cage compounds as well a ~
structurally similar metal atom cluster molecules that are not closed poly
hedra. Generally, these may be regarded as derived from closed polyhedrL
by removal of one or two vertices. The diagram below illustrates how removal
of one or two vertices generales the so-called nido and arachno relatives of
a closo octahedral structure.
In terms of bonding, the removal of one vertex results simply in the elim-
ination of one of the AOs that cdntribute to the SALCs. The principal effect
INORGANIC ANO ORGAN0l\1ETALLIC COMPOUNDS 235
Vertex
removal
Nido
octahedron
D
1

2
Arachno
octahedra
1 = Square
2 = Butterfly tetrahedron
on the bonding orbitals is that they become slightly less bonding in character.
Most importan ti y, however, the total number of bonding orbitals remains the
same, n + 1 (or n + 2 if 11 is the numberof vertices in the nido polyhedron).
The antibonding orbitals become less antibonding, and in fact, sorne of these
disappear. All species of the type BnHn+h which can be viewed as ions
for the purpose of electron counting, exhibit nido structures. The reasoning
behind vertex removal can be extended to the bonding orbitals of arachno
structure, and electron counting procedures usually result in a correct predic-
tion of the structure. We will discuss this in more detail presently when we
develop the concept of the total electron count. .
Polyhedral Metal Carbonyl Clusters:
as Prototype
Bonding in the Os
6
(CO)i; ion is obviously more complicated than that in
but sorne years ago, in an ingenious intuitive Ieap, K. Wade pointed
out a useful analogy between them. Working from this analogy he then de-
veloped a set of guidelines, called Wade's rules, that help to rationalize tbe
myriad of structures and compositions adopted by the high-nuclearity metal
carbonyl clusters. We shall not present Wade 's rules here ( there being good
discussions available elsewhere*), but we shall describe the basis for the
important analogy between an Mc,(C0)
1
R species and the ion.
The metal atom in an M(C0)
3
group having C
3
,. symmetry uses three A Os
to forrn a bonds to the CO groups and must also be engaged in sorne degree
of n bonding as well. By employing the u and n orbitals for the M-C bonds
K. Wade in Transition Metal ClusLers, B. F. G. Johnson. Ed., John Wiley & Sons. New
pp. 193-264.
236 APPLICATIONS
as the following representations .. and their component irreducible rep-
resentations are obtained:
o
o
1
o
fa= A1 +E
f" = A1 + + 2E
From the character table we see that there is no s, p, or d orbital of the
A:! type .. but for A
1
and E \Ve ha ve the results
A 1: S' p:.. d
E: { p."), (dx:' d_v:) .. (dr- . .. d.r_r)
On the basis of M-C overlaps, the orbitals best suited to forro the M-C u
bonds are undoubtedly the s and the (dxz .. dy:) orbitals .. while the best 1r bonds
can probably be formed by using the d=2 and (d.r:! __ ,.:!, d.-c_r) sets. None of these
allocations is rigorous, because to so1ne degree all pairs of E-type orbitals can
play a part in both a and n: bonding, nor is the P: orbital entirely excluded
from u bonding. However, if this prescription is accepted., we see that the
remaining orbitals on the metal atom are the three p orbitals. As the drawings
in Figure 8.17 then show, the M( CO h group has the same ( or at least a very
similar) set of orbitals available for cluster bonding as did the B-H unit. We
mny note in passing that species exhibiting this kind of similarity of orbital
structure have been termed by Hoffmann isolobal.
From this point on, the entire analogy of
to Mt,(CO)i; is straight-
forward. The formation of MOs occurs in the same way and we see that the
ideal electron count for cluster bonding will again be 14. This number is most
easily satisfied by employing M(COh units that (like B-H) provide two
electrons ea eh and then adding two more to give the dianion. The metal we
seek is one that can use six electron pairs to fill the three n:-bonding orbitals
ca
oc 1
Figure 8.17 Schematics showing the similar. orbital structures (isolobality) of a BH
and an M(COh group.
and have two left: we need a metal atom from the Fe, Ru .. Os group.
The anion Osn(CO)ii is a stable, well-characterized example.
It is possible to employ other M(CO).t fragments and obtain similar results
for other M, species. Thus,
and Rh
6
(C0)
1
,. also utilize 14 elec-
trons for cluster bonding and possess octahedral arrangements for the 6 metal
atoms. The
fragment can be considered analogous to M(C0)

3

As with the polyhedra, it can be shown that the optimum number
of bonding electron pairs should be n + 1 for other deltahedral clusters (i.e.
those having entirely triangular faces) of metal atoms M, (n 5). For
[Os
5
{C0)
12
]2- should have 12 electrons available for cluster bonding and its
metal atoros are arranged in the form of a trigonal bipyramid.
Addition of Cappilzg Groups. * Many metal clusters can be enlarged by
the addition of a metal-containing group to a triangular face of a c/oso poly-
hedron. The frequently referred to as "cappingH is illustrated in
Figure 8.18. Multiple capping is also possible.
The addition of a capping group affects the bonding orbitals of the cluster
in the following way. The capping group con tributes three one radial
and two tangential, to the cluster bonding. The principal effect upon the
bonding SALCs is that they will contain an additional component. To the
first approximation, they will beco me more strongly bonding in character.
Sorne of the antibonding orbitals may be stabilized and sorne new orbitals
will appear, but the most strongly bonding orbitals will be the same n + 1
orbitals that existed in the closo polyhedFon. The capped polyhedron will
have n pairs of bonding where n is the number of vertices of the
add a
1
cap

add a
second
cap

2
Figure 8.18 Schematic illustration of the capping" process.
"' D. M. P. Mingos and M. l. Forsyth. J. Chem. Soc. Dailon Tran.f . 610 ( 1977).
238 APPLICATIONS
closo polyhedron plus one for the capping group. The compound Os
6
(C0)
18
can be viewed as a combination of Os
5
(CO)i2 plus Os(CQ)j+. The latter group
wouJd contribute no electrons to the cluster bonding and should assume a
capping site on the [Os
5
(C0)
12
]l- group. Thus, we obtain the structure shown
below. This is a monocapped trigonal bipyramid although it is often described
in the Jiterature as a bicapped tetrahedron. Tbe latter description, while not
incorrect, fails to emphasize the relationship of the structure -to the bonding
principies.
Oss(COhs
Monocapped
trigonal bipyramid
The Total Electron Count. Mingos* has formulated the electron counting
procedure for cages and clusters so as to focus on the total number of valence
electrons in the molecule or ion. For elements. from the main groups, there
are only four valence orbitals, one s and three p's. Three of these from each
atom are combined to form the n + 1 bonding orbitals and these will contain
2n + 2 electrons when they are filled. The one remaining orbital on each ato m
will contain 2 electrons ( either a Ion e pair or a bonding pair) and 2n electrons
will be required to fill these n orbitals. Thus, the total electron count for a
closo cluster containing 11 vertices is 4n + 2. Tbe total electron count for the
capped cluster is 4n, for the derived nido cluster 4n + 4, and for the derived
aracltno cluster 4n + 6.
A transition element has 5 additional valence orbitals, the 5d orbitals, and
therefore 10 additional electrons are required per atom to fill the valence
shell of each metal atom. A closo cluster consisting only of transition metal
atoms should have a total of 14n + 2 valence electrons. A capped luster should
have 14n, a nido cluster 14n + 4, andan arachno cluster 14n + 6. The combined
formula 4n + 2 + lOnt would represent the total electron count for a closo
cluster, A,-mMm., of n atoms that contains m transition metal atoms and n -1n
main group atoms .. Table 8.2 summarizes the main rules, and the fol1owing
examples show how the total electron counting scheme is applied.
D. M. P. Mingos. Acc. Chem. Res . 17, 311 (1984).
TABLE 8.2 Formulas Representing the Total Electron Count for Polyhedra
Consisting o' Main Group Elements, A, and Transition Metals, M
Example 1
n = 5
Example 2
n=1
Example3
n = 5
Capped Closo Nido Aracluzo
4n
14n
4n + 2
14n + 2
4n + 4
14tr + 4
411 + 6
14n + 6
4n + lOnz 4n + 2 + lOnz 14n + 4 + 10m 4n + 6 + lOn1
B
5
H
9
5B x 3e-IB = 15e-
9H x Ie-/H = 9e-
Total electron count 24e-
4n + 4 = 24
14n = 98
14n + 4 = 74
Predict a nido octahedron, as observed.
7Rh x 9e-!Rh = 63e-
16CO x 2e-/CO = 32e-
charge = 3e-
Total electron count 9Be-
Predict a capped octahedron, as observed.
5Fe X se-/Fe= 40e-.
15CO x 2e-/CO = 30e-
1C x 4e-/C = 4e-
Predict a nido octahedron .. as observed.
N o t e ~ The carbido carbon atom is semi-interstitial and is regarded as a ligand
and not a vertex site, as shown below:
240 APPLICATIONS
ll = 5
Example 4
3Fe X se-/Fe = 24e-
9CO x 2e- /CO = Ise-
2S x 6e-IS = 12e-
(3Fe + 2S) 4n + 4 + 30 = 54
Predict a nido octahedron, as observed.
8.8 MOLECULAR ORBITALS FOR METAL
SANDWICH COMPOUNDS
The term "metal sandwich compounds"' applies strictly to compounds of the
type (C,H,)
2
M, such as (C;H
5
hFe and (C
6
H
6
)
2
Cr, in which a metal atom is
"sandwiched" symmetrically between two paralle1 carbocyclic ring systems,
but is commonly used in a broader sense to include, in addition, all com-
pounds in which at least one carbocyclic ring, C,H,, such as C
4
H
4
, C
5
H
5
,
ChHt., C
7
H
7
, is bound to a metal atom in such a \Va'f that the metal atom lies
along the n-fold symmetry axis of the ring and is thus equivalently bonded
to all of the carbon atoms in the ring. Thus, in addition to. the highly sym-
metrical molecules above, the term refers also to such mono-
ring compounds as
C
6
H(1Cr(CO):h [C1H
7
Mo(CO)J]+, and
C
5
H
5
Fe(COhC
2
H
5
, and such mixed-ring systems as (C
5
H
5
)(C1H
7
)V and
(C
5
H;)(ChHtJ)Mn. lt should be noted that there are also compounds, such as
(C
5
H
5
):!MoH
2
, in which the rings are not exactly parallel, although it is believed
that the metal-ring bonding is still symmetrical about the symmetry axis of
each ring. The bonding in such cases will have essentially the same features
as in the more symmetrical molecules but cannot of course be treated with
the same degree of rigor with regard to symmetry.
A metal sandwich compound does not strictly fit our previous concept of
an AB,-type molecule, since the ligand atoms interact strongly with each
other as welJ as with the central atom. lt is desirable to extend the discussion
to these molecules, however .. since they pro vide clear and important examples
of how to treat the situation in which the ligands in a complex are themselves
polyatomic entities with an intemal set of M Os perturbed by interaction with
the AOs of the central atoro.
Using ferrocene., (CsH
5
)
2
Fe, as an example., we can demonstrate all of the
basic ideas in the MO treatment for the whole class of molecules. Accordingly,
we will first treat ferrocene in detail then briefly outline the application of
the method to a few other selected cases.
Ferrocene
The basic strategy is to construct linear combinations of aH of the prc orbitals
of the two C
5
H
5
rings belonging to the irreducible representations of the
molecular point group D
5
J, * to classify the orbitals in the valence shell of the
metal atom according to their symmetry in the point group, and then to
combine metal and ring orbitals into MOs of the entire molecule. The im-
portant difference from what has been done earlier in treating ABn molecules
arises from the strong interaction of the various pendent atoms ampng them-
selves. Because of this, it is best to first work out the hlocal" MOs and their
energies and then formulate the A B interactions in terms of the established
local MOs for groups of ligand atoms-the C
5
H
5
rings. in this case-rather
than between A and each B as an individual atom.
In arder to construct the proper linear combinations of pn orbitals, we can
make use of the results we have already obtained for a single C
5
H
5
ring. For
such a ring we have constructed LCAO-MOs transforming correctly under
the rotations belonging to the group C
5
(see page 152). These are of A, E ...
and E
1
symmetry. The set of 10 prc orbitals provided by two su eh rings oriented
The energy dilference between the staggered ( and eclipsed (D.;,.) rotomers of ferrocene
is apparently very small {SI kcal/mol}. with the latter being perhaps the more stable (cf. R. K.
Bohn and A. Haaland. J. Organomet. Chem . S, 470 ( 196)). The two symmetries are equally
suitable and convenient for a discussion of bonding. Since most of Lhe research literature has
used D5tJ, we make the same choice here.
242 APPLICATIONS
as they are in the ferrocene molecule spans the following representations of
Dsd:
r. 10 o o o o o o 2
Thus we see that for the system of two rings we require two A orbitals, one
symmetric and the other antisymmetric to inversion in the center, two E
1
orbitals, one symrnetric and the other antisymmetric, and finally, two E
2
orbitals, one symmetric and one antisymmetric to inversion in the center. It
is rather easy to write down expressions for these by making appropriate
combinations of the orbitals that we airead y ha ve for the individual rings .. In
doing this, we will refer to the rings and orbitals as they are shown and labeled
in Figure 8.19. lt is very important here to note that we ha ve eh osen the
directions of the pn orbitals such that all of their positive lo bes point in toward
the metal ato m. Thus the + z axis for ring 1 is in the opposite direction to
the + z axis for ring 2.
Referring to the character table for group D
5
J, we se e that an A ~ orbital
must be symmetric to inversion in the center of symmetry. This requirement.
RingB
Figure 8.19 Sketch showing the p1C orbitaJs on the two rings used to construct M Os
for a bis( cycJopentadienyl)metal molecule.
INORGANIC ANO ORGANOMETALLIC 243
will be satisfied by the following combination of the A orbitals of the two
rings:
where we use subscripts 1 and 2 to refer to the rings. lt can also easily be
seen that the tp(A
1
.r:) so obtained satisfies all other symmetry requirements.
To obtain an orbital of the system of two rings which is antisymmetric to
inversion we take the linear combination
Again, a simple check will show that this orbital, rp(A
21
,), satisfies all the
symmetry requirements of the A
211
representation.
For the E
1
and E
2
orbitals we proceedin exactly the same way, choosing
normalized combinations of the E
1
and E
2
orbitals of the individual rings so
as to obtain functions which are symmetric and antisymmetric to inversion,
namely,
1
f!I(E
1
ga) = V2 ['l'a(E
1
a) + f//
2
(E
1
a)]
1
rp(E
11
h) = Y2 ('1'
1
(E
1
b) + f//:!(E
1
b)]
1
fii(E
1
,a) = Y2 [fii
1
(E
1
a) - rp
2
(E
1
a)]
1
fii(Etuh) = Y2 [lfi(E.b) - lf/2(E.b)]
1
fii(E2,a) = Y2 ['I'(Ela) + f//2(E2a)]
1
rp(E
2
,b) = Y2 [tflt(E:!b) + lf/2(E2b)]
1
rp(E
21
,a) = Y2 [rp(E2a) - lf/2(E2a)]
1
r(Emb) = \1'2 [ '1'1(E2b) - '1'2(E2b)]
244 APPLICATIONS
For the metal atom, iron, the valence shell orbitals are the five 3d orbitals,
the 4s orbital, and the three 4p orbitals. The transfonnation properties of
these orbitals may be ascertained immediately by inspection of the character
table for D
5
J, the results being
A
18
: 4s, d ~
E
1
g: (3d.r: 3d_,.::)
E
28
: (3d.r,., 3d.r-_,.l)
A2u: 4p::
Thus we have a total, counting the degeneracies, of nineteen orbitals, but
beca use of their symmetry properties, as shown in Section 7.1, we do not
have to salve a 19 x 19 secular determinant. Instead we have only the
following small detenninants:
One 3 X 3 for the A
1
g MOs
Two 2 x 2 for the E
1
g MOs (same roots for both)
Two 2 x 2 f ~ r the E
2
R MOs (same roots for both)
One 2 x 2 for the A2u MOs
Two 2 x 2 for the E
1
, MOs (same roots for both)
The E
21
, M Os on the rings are in themselves E2u MOs for the whole molecule,
since there are no E2u metal orbitals with which they might iQteract.
The problem is now reduced to one of evaluating the matrix elements. We
shall not discuss the details of this process; the reading list, Appendix IX.
cites several such calculations. Figure 8.20 shows the energy level diagram
for ferrocene produced by one of them.
The energies of the ring orbitals relative to one another have already been
estimated in the Hckel approximation in units of p (page 152); as explained
in Appendix 111, we choose here the "spectroscopic" value of p, -60 kcal/
mol, for use in constructing the energy level diagram. Moreover, sin ce the
rings are -4 apart in the molecule, the reasonable assumption is made that
there is no significant direct interaction between them; the g and u orbitals
of the same rotational symmetry are thus taken to ha ve the same energy.
1 t will be se en that there are 9 more or Jess bonding orbitals, which are
just filled by the 18 electrons originating in the 1r systems of the rings and in
the valence shell orbitals of the metal atom. The exact ordering of the levels
varies in so me respects from orie calculation to another, depending on the
approximations used in evaluating the matrix elements. For other (C
5
H;)2M
5ev
Ring
orbitals
a
16
,a2u
INORGANIC AND ORGANOMETALLIC COfviPOUNDS 245
Molecular
orbitals
ee
lron
orbitals
Figure 8.20 An energy level diagram for ferrocene. The MO energies are those
calculated by Shustorovich and Dyatkina, using a self-consistent fie1d procedure. The
positions of the ring and iron orbitals on this diagram are only approximate.
246 APPUCATIONS
compounds, energy level diagrams having the same qualitative features would
be anticipated, but the relative order of the least stable bonding M Os is subject
to change because of variation in both the relative energes of metal and
ligand orbitals and the reJative magnitudes of the different interaction ener-
gies. Consequently, sorne caution must be exercised in attempting to predict
from a diagram constructed specifically for one ( C
5
H
5
)
2
M compound the elec-
tronic structure of another containing a different metal.
Dibenzenechromium
If this molecule is assumed to consist of two benzene rings placed on either
side of the chromium atom so that tJleir planes are parallel and their CtJ axes
colinear, the molecular symmetry m ay be D6J or D
61
, depending on whether
the rings are staggered or eclipsed. X-ray study of the crystalline compound
shows that the chromium atom is at a center of inversion so that in the
crystalline st_ate at least the molecular point group is Dbh. There is evidence
from infrared and Raman spectra that the molecuJe has DtJr symmetry in
solution.
Assuming D
611
symmetry we can easily develop an MO bonding scheme
similar to that given above for ferrocene. Again we begin with the LCAO-
MO n: orbitals of a single ring and combine them to obtain symmetry orbitals
appropriate to the entire molecule. Using the 12 carbon .Pn: orbitals as the
basis for a representation of the group D6h., and breaking this down into the
irreducible representations, we find that the symmetry orbitals must belong
to the following irreducible representations:
By combining the A, B, E
1
, and E
2
orbitals for the individual rings into
combinations that are symmetric and antisymmetric to inversion .. we readily
obtain expressions for symmetry orbitals of the required types. The relative
energies of these symmetry orbitals in units of P (again using the spectroscopic
val u e) are those airead y calculated (page 147) for benzene.
For the metal atom we find from the table for the group DM that
its valence shell orbitals belong to irreducible representations as follows:
A 1.'=: S'
Elt: (d.r:' dr:}
E2g: ( d.fJ"'
Figure 8.21 shows an energy level diagram for dibenzenechromium. It will
be noted how similar the MO scheme is to that in ferrocene. This is not too
surprising in view of the fact that the only qualitative difference between the
two systems is in the presence .of high energy orbitals of B symmetry in C
6
H, ..
which, however, do not participa te in the ring-metal interactions sin ce the
metal has no valence shell orbitals of B symmetry.
Benzenechromium Tricarbonyl
In this molecule we have two halves in which the symmetries are different
( C
6
., and C
3
..), but they share certain symmetry elements, namely, all those
(a)
of C
3
l. for either the staggered or eclipsed orientations, (a) or (b), but only
C
3
for the intermediate case, (e). Actually, only the behavior of wave functions
under C
3
rotations is required to sort them out sufficiently to recognize the
bonding possibilities. Rigorously, the en tire molecule shouJd be treated in
(b)
this way. However, it is helpful to handle (C
6
flt,)Cr portian ofthe molecule
: according to its "local rotational symmetry," namely, C
6

We can thus start with the four 7C M Os of benzene, as determined in Section
7.3 .. namely, A2u,
E2u., and
For the metal atom the AOs can also be

(e)
classified as to their symmetries under C, rotations. On the basis of these
results we can set up an energy level diagram for the hexagonal pyramidal
(CJit;)Cr moiety as shown on the left side of Figure 8.22.
248 APPLICATIONS
5 ev
e2c t'2fl
Ring
orbitals
eiu
a
C2u
,
O!!u
Molecular
orbitals
Chromium
orbitals
fp
Figure 8.21 An energy level digram for dibenzenechromium. The positions of the
ring and chromium orbitals on this diagram are only approximate.
b2g
f!lu
eg
02u
e2
//
.,..,.
a
........
.......
et
..........
Cdxz,dyz)
\
\
a
\
dz2 --oo--, \ {

/
. ez /
Cdxy.dx2-y2) -oooo-/
'
'
'
e
'-oooo-
'
'
'
a
'--oo--
4p
4s
3d
Cr
a
/
/
/
/
1
1
1// } 1
e
Cd;u.,dyz}
a
dz2
e
Cc4:y.dx2-y2)
l{
a 1
---oo-'
e 1
-<>-000-
COa
SALCs
Figure 8.22 A schematic energy level diagram for CttH,Cr(COh.
}a.
The interaction of the Cr atom with the trigonal set of CO groups s
governed by the symmetry properties of orbitals under C
3
rotations. The set
of three (J donar orbitals on the carbon atoms will give SALCs of a and e
symmetry. It must be noted that in C
3
symmetry there is no e
1
versus e
1
distinction; both sets of d orbitals, (d.r:n d,.=) and d.r,.) belong to the E
representation. However, the former have directionl properties that make
them far better able to overlap with the CO (J orbitals than the latter or the
(P.r p-'.) pair. Thus, the (J bonding in the Cr(C0)
3
unit can be qualitatively
described by the ene.rgy level diagram on the right side of Figure 8.22.
It can be seen that the set of 18 valen ce shell electrons of the ( C
6
H
6
)Cr( CO h
molecule are distributed so that 6 are primarily engaged in Cr-CO (J bonds,
6 in ( C
6
lit,)-Cr bonding, and 6 others are still essentially u metal d electrons.,
This last result is a little unrealistic since the e
2
electrons would be seen to
play a role in Cr CO n backbonding if we included this aspect of the
bonding in our diagram. The (C
6
Ht,)Cr(C0)
3
has been treated in the above
fashion in the literature* and has also been the subject of an ab initio cal-
culation. t
M: Elian. M. M. L. Chen. D. M. P. and R. lnorg. Clrem .. 15, 1 148 ( 1976).
t M. F. Gucst. l. H. Hillicr. B. R. Higginson. and D. R. Lloyd. Mol. Phys . 29, J 13 ( 11J75).
250 APPLICATIONS
Cyclopentadienylmanganese Tricarbonyl
This molecule provides a significant example of the type in which the true or
overaJI molecular symmetry is very low but the bonding in parts of the mol-
ecule may be treated, at least qualitatively, in tenns of relatively high local
symmetries. In this case the ( C
5
H
5
)Mn part of the molecule m ay be considered
to have C
5
,. symmetry and the Mn(C0)
3
part taken as having C
3
,. symmetry,
although the molecule in its entirety can ha ve no more than Cs symmetry,
and that only for two particular orientations of the C
5
H
5
ring relative to the
Mn(C0)
3
grouping. In treating the (C
5
H
5
)Mn and Mn(C0)
3
bonding sepa-
rately .. each in terms of its own ideal local symmetry, we make the assumption ..
inter ala, that degeneracies permitted in C
5
,. symmetry will not be greatly
by the presence of C
3
,. symmetry in the other part of the molecule and
vice versa. Because of the particular shapes of the d orbitals such an as-
sumption probably. has sorne validity in this case, but it cannot always be
taken as true.
Figure 8.23 shows a schematic energy level diagram for (C
5
H
5
)Mn(C0)
3

In the center the valence shell orbitals of manganese are shown. They are
labeled on the right with their symmetries in C
3
and on the left with their
symmetries in C
5
At the extreme left are the n: M Os of the CsH
5
ring, labeled
with their symmetries in C
5
and with the energy of the e
1
orbitals placed at

---a

======te
a-oc
e 1 ::::g::g::::
CsHs
Mn
(CO)a
Figure 8.23 A schematic MO energy level diagram for

about the same level as the d orbitals of manganese. On the extreme right
are shown the energy levels of the (J orbitals for the three CO groups, vvith
symmetry designations appropriate to el. These designations \Vere obtained
by taking the set of three a orbitals on the carbon atoms as a basis for a
representation of the group c3 and decomposing the representation into its
component irreducible representations. The energy of these symmetry orbitals
aonstructed from the carbon (J orbitals has been assumed to be about the
same as the energy of the a orbital of C
5
H
5
No account has been taken of
the interaction of the metal with the n: orbitals of the CO groups. This could
be done by finding the representations spanned by the six n: orbitals and then
permitting them to interact with metal orbitals of the same symmetry types.
Although it is certain from the vibrational frequencies of the CO groups in
this molecule that the lvf-C n: interactions are substantial, we have chosen
to omit them here .. since they can be only crudely estimated and their inclusion
in the energy level diagram would make it extremely unwieldy.
The only real justification for drawing an entirely schematic diagram such
as this one is that it helps with the "bookkeeping. "' The diagram makes it
somewhat easier to see how the orbitals of different components of the mol-
ecule may interact in order to produce a satisfactory set of bonding M Os than
is possible by merely inspecting a list ofthese components. For example, the
shapes of the metal e
1
orbitals, d:c: and d_r::' are such that their overlap with
the ring e
1
orbitals should be fairly substantial and the diagram accordingly
shows a sizable interaction of these orbitals. On the other hand .. the metal.e
2
orbitals are not strongly directed toward the ririg e
1
orbitals and the energy
difference is initially greater, so that a much weaker interaction \Vould be
expected, as indicated.
EXERCISES
8.1 Confirm all the results given on pages 219-221 for AB
3
(D
3
h), AB .. (D .. h),
ABtt(DJh), ABK(Oh), and type mo)ecules. -
8.2 For an MX. molecule having a flattened tetrahedral ( or ruffled planar)
structure of Du symmetry, determine the SALCs which must be formed
by the set of X atoms and the AOs of the central metal atom that are
required to form a set of M-X (J bonds.
8.3 Assume that M is a transition metal and that the MX
4
molecule in
Problem 8.2 is only slightly ruffled. Draw an approximate energy level
diagram, similar to those in Figures 8.4 and 8.6.
8.4 What s, p, and d orbitals of a central atom can be used to forma hybrid
orbitals for an AB
8
molecule having a square antiprism structure?
8.5 For each of the two important geometries, a trigonal
prism centered on all rectangular faces (D
311
) and a square antiprism
252 APPLICA TIONS
centered on one square face, show how hybrid orbitals suitable for u
bonding can be formed on a central ato m by employing an s, p
3
, d
5
set
of valence shell orbitals.
8.6 Givea qualitative analysis of the bonding to be expected in
where C
4
H
6
is trimethylenemethane, (CH:!)
3
C, with intrinsic threefold
symmetry.
9
LIGAND FIELD THEORY
9.1 INTRQDUCfORY REMARKS
For conciseness, the title of this chapter is simply "Ligand Field Theory. u
However, many of the principies which wil1 be developed are as much a part
of crystal tield theory and the molecular orbital theory of transition metal
complexes as they are of ligand field theory. lndeed the three theories are
very dosely related, and hence it seems advisable to begin this chapter with
a brief, historically oriented discussion of the nature of these theories.
The beginning of all three theories can be traced to the year 1929, when
Hans Bethe published his classic paper entitled of Terms in Crys-
tals. "* There are really two completely separa te parts to Bethe's paper. The
first is concerned purely with the qualitative consequences of the symmetry of
the surroundings of a cation in a crystal lattice. In this part, Bethe showed
that, in general, the states arising from a particular electronic configuration
of an ion which are degenerate when ion is free of perturbing influences
must break up into two or more nonequivalent states when the ion is intro-
duced into a lattice. He showed how it is possible, using the methods of group
theory, to determine just what states will result when an ion of ny given
electronic contiguration is introduced into a crystalline environment of detinite
symmetry.
The second part of Bethe's paper describes a method by which the mag-
nitudes of the splittings of the free-ion states may be calculated, assuming
H. Bethe, Ann. Physik, 3, 133-206 (1929). An English translation of this paper is available
from Consultants Bureau Enterprises. 227 W. 17 Street, New York, N.Y. 10011.
253
254 APPLICATIONS
that the surroundings effect these splittings by purely electrostatic forces. This
assumption that all interactions between the ion and its surroundings may be
treated as electrostatic interactions between point charges is the defining
feature of the crystal field theory. It has the consequence that all electrons
which are in metal-ion orbitals in the free ion are treated as though they
remain in orbitals which are 100% metal-ion orbitals.
lt was first pointed out by Van Vleck that the symmetry part of Bethe s
approach will remain entirely val id if we change the computational part from
a purely electrostatic approach to one that admits the existence of sorne
chemical bonding between the metal ion and its neighbors. If we do so .. the
orbitals with which we must deal will be no longer pure metal orbitals, but
only partly metal orbitals. This means that in principie we cannot write down
the same fairly simple expressions for because the orbitals involved
are no longer simple. In practice, however, if the covalence of the metal
ligand bonds is relatively energies will be given by equations jdentical
in form to the equations derived in the crystal field theory. This modified
crystal field theory, which admits that there is sorne covalent as well as elec-
trostatic interaction between the ion and its neighbors, is called ligand field
theory.
Van Vleck also pointed out that even in the case of very highly covalent
bonding [as in, e.g., Ni(CO)_. or which is best treated by using
MO theory, the symmetry properties and requirements remain exactly the
same as for the crystal field model and the ligand field model.
Thus, in qrder to gain an understanding of any one of these theories .. the
same symmetry considerations are required at the outset.
9.2 ELECI'RONIC STRUCTURES OF FREE ATOMS ANO IONS
We intend in this chapter to consider the manner in which the symmetry of
the chemical surroundings of an ion determines the effect of this environment
on the energy levels of the ion. In the crystal field and ligand field theories
we often wish to regard the effect of tlle environment as a small perturbation
on the states of the free ion. For the benefit of readers not acquainted with
certain general features of the electronic structures of free atoms and ions., a
.
brief rsum of the subject is given in this section.
Wave Functions and Quantum Numbers for a Single
The wave function '1' for a single electron, in a hydrogen atom for example,
may be written as a product of four factors. These are the radial function
R(r), which is dependent only on the radial distance r from the nucleus; two
angular functions 8(0) and l>(c/J), which depend only on the angles O and c/J.
LIGAND FIELD THEORY 255
(cf. Fig. 8.1); and a spin lflp which is independent of the spatial
coordinates r, O, and rp. Thus we write
'11 = R(r) 8(0) cll(l/>) f//.r (9.2-1)
This overall wave function and each of its factors separately have a parametric
dependence on certain quantities called quantum numbers, of which there
are four: 11, 1, n1, s.
The principal quantum number 11 takes all integral values from 1 to infinity.
1t determines the nature of the radial part R(r) of the wave function only.
The quantum number 1 occurs in the E>( O) factor of the wave function. lt
may be thought of as representing the angular momentum of the electron, in
units of h/2n, because o'f its orbital motion, and we shall call it the orbital
momentum quantum number. * lt m ay take all val u es O, 1, 2, ... , n- 1,
where n is the principal quantum number. Thus, in the first principal shell,
there exist only wave functions with 1 = O; in the second shell there are wave
functions with l = O and 1; and so on. For historical reasons, letter symbols
are given to orbitals according to the value of l, as shown in the following
scheme:
1 = o 1 '2 3 4 5 6
Letter symbol: s p d f g h i
If we continue to take the classical view of an electrn as a discrete charged
particle having angular momentum as a result of its orbital motion, we must
also conclude that because of its charge its orbital motion wiJI generate a
magnetic dipole. The vector representing this magnetic dipole \Vill be colinear
with the vector representing its angular momentum (both are perpendicular
to the plane of the orbit), and the value of this orbital magnetic dipole JL
1
is
directly proportional to the anguJar momentum.
The quantum number m occurs in both the E>(O) and 4>(cp) parts of the
wave function. lt indica tes the tilt of the plane of orbital motion with respect
to sorne reference direction. It can take all integral values from 1 to -1, or
2/ + 1 values in all. Its relation to the tilt of the orbital plane may be stated.
more explicitly as follows. If the plane is perpendicular to the reference
direction, the length of the projection of the vector representing 1 on the
reference line is numerically equal to /. The next Jargest angle of tilt permitted
by quantum mechanics is such that the length of the projection equals 1-1,
the next such that the length of the projection is /-2, and so on, until the
value of -1 is reached. The quantum number 111 is simply the length of the
To be precise. the total orbital angular momentum is not /(h/2n) but V/(/ + 1 but
this point need not concem us here.
256 APPLICATIONS
projection of 1 on the reference line. The situation is illustrated for the case
of 1 = 2 in the sketch below.
m=l=2 m=l
m= -1
For given n and 1 = O there is only one possible orbital .. namely .. one \Vith
n1 = O. Thus there is only one s orbital of each principal quantum shell. For
given n and 1 = 1 there are three possible n1 values. Hence each principal
shell has three different p orbitals. Similarly, d orbitals come in sets of five ..
f orbitals in sets of seven, and so on. In the absence of a11y external forces,
the energy of an orbital s independent of its nt value. Hence aH three np
orbitals, aJI five nd orbitals .. and so on, are of the same energy.
The electron spin quantum number s is the number on which f/ls depends,
and it m ay take only the values +! and - !. lt may be nterpreted, classically,
as a measure of the spin angular momentum of the electron, that is, as a
measure of angular momentum resulting from the rotation of the electron
about its own axis. We may think of the electron as intrinsicially having spin
angular momentum equa) to k{h/2n)* and consider that this may be oriented
with respect to a reference direction so as to produce components of +! or
-! along the reference direction. Again .. however, in the framework of this
classical picture, we must also expect that, if a charged body is rotating, a
magnetic dipole is generated. Thus every electron has associated with it a
spin magnetic dipole Jln the direction of which depends on s.
Although these four quantum numbers are always sufficient to specify .
completely the state of an electron, there is another quantum number, j,
which is useful in accounting for the energy of the state. In units of h/2n .. j
gives the total angular momentum of the electron, which is a vector sum of
the orbital angular momentum and the spin angular momentum. Quantum
mechanics requires., however, that the vector su m can be made only in certain
ways. The value of j may be either 1 + i or 1 - i. So long as we regard the
Again. strictly V!(! + l}(h/2x).
spin wave lfls .. as entirely of the orbital \Vave function,
R(r)9(0)<l>(c/J), these two j states have the same energy.
Actually, spin and orbital magnetic moments do interact, so that the state
with j = 1 - i is of lower-energy than the state with j = 1 + This would
be expected classically, sin ce in the j = 1 - ! state the rbital and spin
moments are opposed. In the hydrogen atom, in -all hydrogen-like ions, and
in all atoms and ions having one electron outside of a closed core .. the splittings
due to this phenomenon of spin-orbit coupling are very small compared to
the energy differences between orbitals differing in their 1 values. Thus the
effect of spin-orbit coupling in such cases is justifiably regarded as a small
perturbation on an energy leve) pattern which is basically determined only
by the val ues of 11 and /.
Quantum Numbers for Many-Eiectron Atoms
Although various general cases come under this heading .. we need consider
only one: that in which most of the electrons in the atom or ion are in closed
shells and the others are in the sa1ne partly filled sheJI. The closed shells are
spherically symmetric, and the only effect they have on the other electrons
is to diminish the strength of the nuclear attraction for these electrons. This
means that the wave function for the one electron in a partly filled shell
containing only a single electron will have the same angular functions, 8(0)
and P(Q> ), as ft would if this electron were the only one in the atom, but its
radial function will be different according to the ueffective" nuclear charge
which it feeJs.
To a first approximation each of several electrons in such a partly filled
shell may be assigned its own prvate set of one-electron quantum numbers ..
n, 1, m, and s. However, there are always fairly strong interactions among
these electrons .. which make this approximation unrealistic. In general the
nature of these interactions is not easy to describe .. but the behavior of real
atoms often approximates cJosely to a limiting situation called the L"-S or
Russeli-Saunders coupling scheme.
In L-S or Russeii-Saunders coupling, a quantum number L .. which gives
the total orbital angular momentum of all the electrons, and a quantum
number S, which gives the total spin angular momentum of all the electrons,
are used. Note the use of capital letters for quantum numbers characteristic -
of tht entire configuration, in contrast to the use of lower-case letters for
quantum numbers of individual electrons. This practice is carried further in
assigning letter symbols to states of different L. We have the following
which is completely analogous to the scheme for single electrons:
L = O 1 2 3 4 5 6
Letter symbol: S P D F G H 1
258 APPLICATIONS
In the Russel1-Saunders coupling scheme it is assumed that the angular mo-
menta of the individual electrons are cotipled to 9ne another and the spins
of the individual electrons also coupled to one another to give, respec-
tively, the L and S values for the configuration. Consider .. for example .. an
atom which has, outside of closed shells, a 3d electron and a 4d electron.
Each of these has an l value of 2 and an S value of i. Now the two 1 vectors
may combine to give integral vector sums as shown in Figure 9.1, where 1
1
refers to the 3d electron and /
2
to the 4d electron, or vice versa. Thus L values
of O, 1, 2, 3 .. 4 are possible.and hence states symbolized S, P .. D, F, and G
may arise from two nonequivalent d electrons. By "nonequivalent" we mean
differing in the principal shell to which they belong. Similarly, the s vectors
may combine to give S va)ues of 1 or O. Since the electrons differ in their
principal quantum number, any L value may be combined \Vith any S value
without violating the exclusion principie. Thus there are altogether ten states
or., as they are called, ternzs, which may arise when two d electrons are present
but are in different principal shells.
With a single electron the total angular momentum j is the vector sum of
1 and s. Here there are only two possibilities, namely, j = l + s = 1 -r ! and
j = 1 - s = 1 - !. More generally, the total angular momentum, denoted
J for a multielectron configuration .. may take all of the values L + S, L +
S - 1, L + S - 2, ..... IL - SI, or altogether 2S + 1 different values.
The number 2S + 1 is called the multiplicity of a term, and t is placed as a
left superscript to the term symbol. Justas with one electron, states with the
same value of L and S but different values of J will differ somewhat in energy.
If these differences in J value are important and must be specified, this is
done by. putting the J value as a right subscript to the term sym bol. Thus all
of the different terms, including those differing in their J values .. for an ndmd
L=4
l
l
l2 L=O
Figure 9.1 Integral vector sums, L .. of two 1, and 1
2
each of length 2.
configuration are
'-
'Su
1
P
1
1
D'!.
3s. 3p" 3D,
3p1 3D'!.
:tp1 lDJ
for the
3
S state the triplet character is not actually realized be cause .. when
L = O .. J can only equal S. There is no finite value of L_ with \Vhich S
may combine vectorially.
Now that we ha ve seen how two d electrons in different principal shells
couple to give terms in the Russeii-Saunders or L-S coupling scheme .. let us
turn to the more directly interesting problem of what terms may arise when
the two d electrons belong to the sanze_ principal shell. Straightforward but
lengthy procedures for making certain that we do not violate the exclusion
principie are necessary .. sin ce no\v we cannot count on a difference in n values
to prevent this. * lt is found that for a d
2
configuration only the following
states may arise:
This L-S coupling scheme m ay be considered a useful approximation when
the components of a given multiplet term, that is, states with the same S and
L values but different 1 vaJues .. differ in energy by amQunts which are small
compared to the differences between one multiplet term as a whole and
another. Among the transition elements, to which we wish to apply ligand
field theory .. the L-S coupling scheme works well enough for many purposes
in the ions of elements in the first and second transition series. In the third
transition series .. however, it is not a very good approximation, although it
can serve as a starting point for more elaborate treatments. In general .. the
adequacy of the L-S coupling approximation diminishes steadily as the atomic
number increases. For the actinide elements it is of no use at all.
For convenience, the states that may arise by Russeii-Saunders coupling
from all dn configurations are listed in Table 9.1. For quantitative applications
of ligand field theory we must know not only the nature of the states but also
their relative energies in a particular ion. For a great number of the ions of
practical interest these energies are known from experimental measurements.
The standard tabulation of such data is C. E. Moore's ~ t o m i c Energy
Levels.'"t
These considerations are treated in many books on atomic structure. See. for example. H. E.
Whiteslntroduction toAtomic_Spectra. McGraw-Hill. New York, 1934 .. Sections 12.1 and 13.11.
t Circular 467, National Bureau of Standards. for sale by the Superintendent of Documents,
U.S. Govemment Printing Office. Washington 25, DC. Volume l. Hydrogen-Vanadium; Volume
11, Chromium-Niobium; Volume liT, Molybdenum-Lanthanum and Hafnium-Actinium.
260 APPLICATIONS
TABLE 9.1 States for d" Systems in Russeii-Saunders Coupling
d'
cf!
cP
~
~
dll
d1
dK
d'
dlll
2(D)
'(S, D .. G) J(P. F)
:!(D)
!(P .. D. F, G, H) "'(P .. F)
'(S, D. G) J(P .. F) '(S .. D, F, G., /) :t(P. D. F, G, H)
!(D)
2
(P, D, F, G .. H) "(P. F) !(S, D ... F .. G, 1)
Same as d"'
Same as d
3
Same as d'!.
Same as d'
'(S)
9.3 SPLITTING OF LEVELS AND TERMS lN A
CHEMICAL ENVIRONMENT
='(D)
"(D, G)
We may use the set of five d wave functions as a basis for a representation
of the point group of a particular environment and thus determine the manner
in which the set of d orbitaJs is split by this environment. Let us choose an
octahedral environment for our first illustration. In order to determine the
representation for which the set of d wave functions forms a basis .. we must
first find the elements of the matrices which express the effect upon the set
of wave functions of each of the symmetry operations in the group; the
characters of these matrices will then be the charac:ers of the representation
we are seeking.
Although the full symmetry of the octahedron is Oh .. we can gain all re-
quired information about the d orbitals by using only the pure rotational
subgroup o because oh may be obtained from o by adding the inversion, i.
However .. we already know that d orbitals are evento inversion, so that it is
only the pure rotational operations of O which bring us new information.
We assume that the wave functions of a set of d orbitals are each of the
general form specified by 9 .2-1. We shall further assume that the spin function
'l'.r is entirely independent of the orbital functions and shall pay no further
attention to it for the present. Since the radial function R(r) involves no
directional variables, it is invariant to all operations in a point group and need
concem us no further. The function 8(0) depends only upon the angle O.
Therefore, if all rotations are carried out about the axis from which O is
measured (the z axis in Fig. 8.1), 8(0) will also be invariant. Thus, by always
choosing the axes of rotation in this way ( or, in other words, always quantizing
the orbitals about the axis of rotation), only the function <l>(Q>) will be altered
by rotations. The explicit form of the 4l( 4>) function, aside from a normalizing
constant, is
(9.3-1)
'(S)
and the five d orbitals are those in which nz takes the val u es 1, 1 - 1, ... ,
O, ... , 1 -/, -1, namely9 29 1, O, -1, -2.
If we take the function and rota te by an angle a, the function becomes
eim(c/l+cr>. Thus we can easily see that the set of<l>(c/J) wave functions .. 1, pecomes
11 on rotation by a:
e1;,, eliCt/l+)
eic/1 rot:stion
ei(t/l+a)
e o
e"
e-;.. e-i(,P+cr)
e -u,p e-2i(.,6+a)
1 11
The matrix necessary to produce this transformation is
e2ia
o o o o
o
eia
o o o
o o
en
o o
o o o
e-ia
o
o o o o
e-lia
This five-dimensional matrix is only a special case for a set of d functions,
and clearly in the (21 + 1}-fold set of functions (l = O for an s leveL 1 for a
p level, 3 for an f level, and so on) we shall have
e la
o o
e(l-l)ia
e(l-l)ia
o
o o
e-lia
The sum of the diagonal elements x(a} can be shown to be*
_ (a} = sin(/ + !)a
X sin a/2
(a :;6 O) (9.3-2)
We now have the necessary formula to determine the characters of the
representation we seek. Let us proceed to them out.
"'The proof of this will not be given here, but is suggested as a useful excrcise. The quantities
being summed form a geometric progression.
262 APPLICATIONS
For a twofold rotation, a = n and hence
. (C.,) = Sn/2 = ! =
1
X - . s1 n n /2 1
Similarly for the threefold and fourfold rotations:
. (C) =sin 5n/3 = -sin n/3 __
1
l
3
sin n/3 sin n/3
sin 5n/4
x(C.,) - - -1
sin n/4
The general formula above is inapplicable if a = O; however 9 it is obvious
that in this case each diagonal element is equal to 1 and the character is equal,
in the general case, to 21 + 1; in the present instance, z(E) = 5. Referring
to the character table for the group O and using the methods developed in
Chapter 4, we easily see that the representation we have derived is reducible
to E + T
1
In the group 0
1
, we will have, since the d wave functions are
inherently g in their inversion property:
Thus we have shown that the set of tive d wave functions, degenerate in the
free ato m or ion ( or, more precisely, under conditions of spherical symmetry)
does not remain degenerate when the atom or ion is placed in an environment
with Oh symmetry. The wave functions are split into a triply degenera te set,
T
28
, and a doubly degenerate set, E
8

lt is easy to apply the same treatment to electrons in other types of orbitals
than d orbitals. The results obtained are collected in Table 9 .2. lt will be seen
that an s orbital is totally symmetric in the 0
11
environment. The set of p
orbitals remains unsplit, transforming as t
1
u; this same conclusion could have
TABLE 9.2 Splitting of One-Eiectron Levels in an Octabedral Environment
Type Irreducible
of Represen tations
Level 1
x(E) x(C!) x.(C.a) Spanned
S o 1 1 1 1 A,,
p 1 3 -1 o 1 T,.,
d 2 S 1 -1 -1

f
3 7 -1 1 -1 A2u + T,., + T2u
g 4 9 1 o 1 A
11
+ E
11
+ T,s: + T
2
s:
h 5 11 -1 -1 1 E, + 2T
1
, + T2u
i 6 13 1 1 -1
A,J + A
2
n + E!: + T,l! + 2T21
been obtained directly from the. oh character table' where it is seen that (x '1
y, z) form a basis for the t tr representation of Oh. All orbitals with higher
values of the quantum number 1, however, are split into two or more sets;
this must be so, since the group Oh cannot allow any state to be more than
threefo)d degenerate.
In a similar manner, we could determine the splitting of various sets of
orbitals in environments of other symmetries which we may encounter in
complexes, such as Td, D ..
1
, .. D
2
d, and C
2
, ... and indeed for any sort of symmetry
we may encounter. An a)ternative and simpler way of obtaining this infor-
mation is to use the results we have obtained for the octahedral case in
conjunction with the correlation table given in Appendix liB. In Table 9.3
are the results for a few point groups of particular interest.
The results we have obtained so far for single electrons in various types
of orbitals apply also to the behavior of terms arising from groups of electrons.
For example, just as a single d electron in a free atom has a wave function
that belongs to a fivefold degenerate set corresponding to the five values
which m may take in the tfl{cp) factor of the wave function, so a D state arising
from any group of electrons has a completely analogous fivefold degeneracy
beca use of the five values that the quantum number M may take. Moreover ..
the splitting of a D term will be just the same as the splitting of the set of
one-electron d orbitals. This is so because the el>( cfJ) factor of the wave function
for a D term is eiMt/J in exact analogy to the el>( t/J) factor, eimf/1, in the wave
function for a single d electron. Exactly the same relationship exists between
f orbitals and F states, p orbitals and P states, and so on. Thus all of the
results given in Table 9.2 for the splitting of various sets of one-electron
orbitals apply to the splitting of analogous Russeli-Saunders terms.
In Table 9.3 we have used small letters to represent the states for a single
electron in the environments of various symmetries, corresponding with the
use of the smallletters, s, p, d, f, ... , to represent their states in the free
atom. Similarly, we shall use capital letters to represent the states into which
the environment splits terms of the free ion. Thus, for example., an F state
of a free ion will be split into the states A
2
, T
1
, and T
2
when the ion is placed
in the center of a tetrahedral environment.
In Table 9.3 the use of subscripts g and u is governed by the following
rules. If the point group of the environment has no center of symmetry, then
no subscripts are used, since they cannot have any meaning. When the en-
vironment does ha ve a center of symmetry, the subscripts are determined by
the type of orbital, all A Os for which the quantum number 1 is even (s, d, g ..
. . . } being centrosymmetric and hence of g character, and all A Os for which
lis odd (p, h, ... ) being antisymmetric to inversion and thus of u character.
In using Table 9.3 for term splittings the following rules apply. Again, if the
environment does not bave a center of symmetry. tbe g and u subscripts are
inapplicable. For point groups having a center to which the inversion operation
may be referred, the g or u will be determined by the nature of the
one-electron wave functions of the individual electrons making up the con-
264 APPLICATIONS
TABLE 9.3 Splitting of One-Eiectron Levels in Various Symmetries
Type of
Level
p
d
f
g
"
i
a,,
lau
ex + 12,
alu + ftu + 12u
a + et: + 1
11
+ t
21
e, + 21
1
, + t2u
a1r + + e.( + 11t: +
a.
11
e + t
1
Symmetry of
a2 + ' + t2
a
1
+ e + 1
1
+ t2
e + t
1
+
a1 + a"! + e + ' + 212
figuration from which the term is derived. We shall be interested only in terms
derived from dn configurations .. and all of these will give g states in point
groups possessing a center of symmetry.
One point needs to be mentioned regarding the splitting of terms of
the free ion in chemical environments, and this concerns the spin multiplicity.
The chemical environment does not interact directly with the electron spins,
and thus all of the states into which a particular term is split have the same
spin multiplicity as the parent term.
In arder to illustrate the splitting of terms of a d" configuration, the states
for a ion in several point groups are shown below. The free-ion terms ha ve
been given on page 259.
States in Point Groups
Free-
Ion Terms
oh TJ

IS
A ln
1
A1
lAlR
lA IR lT'!t
lAl ,T2
.,,A
, lg
IG lE
R
lE
,A2n
?lE
- R
lTlR
,Tt
3p
3Tln
JTl
JA
2R
lE
R
E
g
lE

lE
K
ID
IT2g
lT2 lB,g
lB2R
3A1R
JA2
3A
2
3
E
2g
R
3F

'JT,
lB
lg
3T 3T2 3B"!.t
2R
Environment
D.Jh
a.,
a"Ja + e,
a
11
+ b
11
+ b1, + ex
alM + b,, + b2u + 2e ..
2a
11
+ a1, + b,, + b
1
x + 2e,
a,. + 2o:!ll + b1u + b!u + 3e"
2a
11
+ 0 1; + + 2b1_e +
LIGAND AELD THEORY 265
a,
a
1
+e
a,+ 2e
a
1
+ 2a
1
+ 2e
2a
1
+ a
1
+ 3e
a, + 2a2 + 4e
3a
1
+ 2a
1
+ 4e
a,
b
2
+e
a
1
+ b, + bz + e
a. + a1 + b"1 + 2e
2a
1
+ o
2
+ b, + b
1
+ 2e
a
1
+ a
1
+ b, + 2b:! + 3e
2a, + a
1
+ 2b, + 2b
1
+ 3e
9.4 CONSTRUCfiON OF ENERGY LEVEL DIAGRAMS
We have seen in Section 9.3. that all free-ion terms having L > 1 are split by
chemcal environments of symmetry 0
1
,. TJ, or lower symmetries into t\\'0 or
more states that we label according to the representation of the point group
describng their transformation properties. We now turn to the question of
what the relative energies of these states are and how these energies depend
on the strength of the chemical interaction of the ion with its surroundings.
Obviously, these energies can be straightforwardly calculated by setting up
and solving the requisite secular equations. It is also possible, however .. to
obtain a great deal of information about the energies, especially the relative
energies .. abnost entirely by use of arguments based on the symmetry prop-
erties of the states, and this is the subject of this section.
Of course .. from symmetry arguments alone, quantitative information on
energies cannot be obtained. The procedures we are about to describe require
one piece of quantitative information obtainable only by a calculation. In
Section 9.5 \Ve shall show how this item of information is obtained, )?ut for
the present we \Vill accept it without proof and proceed to the
of energy level diagrams.
It will be demonstrated in Section 9.5 that the relative energies of the
doubly and triply degenera te sets of d orbitals into \\'hich the set of five d
orbitals is split in a tetrahedral or octahedral envitonment are as shown in
Figure 9 .2. Thus., when there is a single d electron in an ion in an octahedral
environment .. it will occupy one of the t
2
n orbitals and the energy required to
promote it to an et: orbital is d
0
; for the same ion in a tetrahedral environment.,
the electron will occupy an e orbital and the energy required to pron1ote it
to a t
2
orbital will be
The energy level diagram we wish to construct will show how the energies
of the vari_ous states into which the free-ion terms are split depend on the
strength of the interaction of the ion with its environment. The separation of
two sets of orbitals into which the group of five d orbitals is split can be
taken as our measure of this interaction. Thus our diagram will have the
266 APPLICATIONS
t
6'D-
...
GJ
r::
lLI
Free ion
Figure 9.2 Diagram showing the relative energies of e and t
2
orbitals resulting from
the splitting of the set of d orbitals by octahedral and tetrahedral environments.
magnitude of do or 6., as abscissa and energy as ordinate. At the extreme left,
where do or 6., is zero, we shall ha ve the free-ion term energies. At the right
side of the diagram we shall have the energies of states which will exist when
the interaction produces sucti a great separation of the e and t
2
orbitals that
the en_!!rgies resu1ting from interelectronic interactions become negligible by
comparison.
We will now explain the method of constructing an energy level diagram
by treating the particular case of a d
2
ion in an octahedral environment. For
this system the free-ion in order of increasing energy, are
In the limit of an extremely large splitting of the d orbitals, the following
three configurations, in order of increasing energy, wi11 be possible:
The usage of the symbols here is the same as that for showing free-atoni
configurations. Thus these are (1) the configuration in which both e)ectrons
are in t
28
orbitals,
(2) the configuration in which one electron is in a

orbital while the other is in an orbital., t
2
e
8
; and (3) the highest energj
configuration, e, in which both electrons are in e
8
orbitals. The energy increase
from one of these to the next highest is 6.
0

Now Jet us consider what will happen as we begin to relax the strong
interaction of the environment with the ion, so that the electrons start to feel
one another's presence. wilJ begin to couple in certain ways, giving ris'
toa set of states of the en tire configuration. The symmetry properties of these
states can be determined by taking the direct products of the representations
ofthe single electrons. Thus for the configuration we take the direct product
x t
2
R, and then decompose it into A'" + E
8
+ + T
2
.e Similarly the
direct product t
2
g x el: gives T
1
n + T
2
.t:, and the direct product gives A
1
..: +
A:zg + E.r: These are the symmetries of the orbital states produced by the
interaction of the electrons. However, we ha ve yet to determine the spin
multiplicities of these states. Clearly, with two electrons involved, they must
be either singlets or triplets, and we must be careful to observe any restrictions
placed on the multiplicities by the exclusion principie.
Later in this section we shall describe a rigorous and completely general
method of determining the multiplicites of the strong-field states . but for the
present we shall work out the d
2
case by means of a less elegant but instructive
method. Consider first the configuration We may regard the
levels as
a set of six boxes as shown below:
s=!
Spin degeneracy = 2
S= -t
Orbital degeneracy = 3
The number of ways in which two electrons may occupy these six boxes is
given by (6 x 5)/2, where the 2 in the denominator takes account of the
indistinguishability of the electrons. Thus the total degeneracy of the con-
figuration is 15. Now as the field is decreased, giving rise to the separate
orbital states .. A
18
, E
8
,
and T
2
g, the total degeneracy must remain 15.
Thus, if we write
we may say that the total degeneracy equals 15 by writing
1 a + 2 b + 3 e + 3 d = 15
a, b, e, and d must each be either 1 or 3. It is not difficult to see that
this restriction the equation has only three solutions:
a b e d
1 1 1 1 3
II 1 1 3 1
111 3 3 1 1
268 APPLICATIONS
Similarly for the ei configuration we have two electrons to place in four
equivalent and this m ay be done in ( 4 x 3)/2 = 6 distinguishable
ways. Thus we may write
and the equation
la+1b+2c=6
which admits of only two solutions:
a b e
1 1 3 1
11 3 1 1
Now for the
configuration we may place one electron in any of six

boxes while we independently place the second electron in any of four boxes,
giving a total of 24 possible arrangements. We also note that there is no
possibility of two electrons being in the same box, so that for all arrangements
their spins may be either paired or unpaired. Thus both the
and
states
derived from the
configuration may be both triplet and singlet. We thus

get the unique answer that the configuration
gives
1

1
T
21
P and
3
T
2
g. The total degeneracy of these four states is 24, in agreement with our
count of the number of arrangements in the boxes.
Now we can determine which of the possible assignments of the multi-
plicities in the and configurations are correct by proceeding to correlate
the states on the two si des of the diagram. To do this we shall use two
principies .. neither of which will be proved, but both of whic_h are important
and rather easily remem be red. As we go from the weak to the strong inter-
action with the environment, we do not in any way change the symmetry
properties of the system. Thus there must be the same number of each kind
of state throughout, and we may accept, almost as an axiom, this principie:
There n1ust exist a one-to-one correspondence between the states at the two
extremes of the abscissa.
The second principie, which has been used earlier (page 194) in constructing
the correlation diagrams for the Woodward-Hoffmann rules, and which has
its ultimate origin in the phenomenon of configuration interaction (page 179)
is called the noncrossing rule:
As the strength of the interaction changes, states of the sa1ne spin degeneracy
and symmetry cannot cross.
In Figure 9.3 we have shown on the extreme left the states of the free ion.
Immediately to the right we have shown the states into which these free-ion
states split under the influence of the octahedraJ environment. Here we know
the spin multiplicities of all states. Now at the extreme right are the states in
the (hypothetical) case of an infinitely strong interaction with the environ-
ment, and immediately to the left of them are the distinct states which we
have just shown to exist in the case of a very strong, but not infinitely strong ..
interaction. In order that each state on the left go over into a state of the
same kind on the right without violation of the noncrossing rule. the con-
necting lines can be drawn only in the manner shown. The way in which this
was done m ay be briefly recapitulated.
We note that there are two 'A
1
..: states on the left and no
3
A ~ states. Thus
both A
1
..: states on the right must be singlets. This immediately settles the
multiplicities of the states coming from the e ~ configuration and rules out
ts 1At
lA
------------
}
'E
e2
-----
3A2
---
--
----
--
{2----
lG
1T1
-"""
lT2 ........_
----""" .........
1 .......... .........
-, """ ..............................
'\. ...............................
"
...........
"
1T1
}
3p
'\.
1T2
t2e
~
\ 3Tt
3
T2
tE
lD
{--:;,-"
-,"-
,,
,,
'
,,
'
{
" ' '\
3p
' ' '\.
' ' '\ 3T1
',' \ 'At
-
,, '
}
,, tE
t2
2
"
1
T2
lTJ
Free Weak
Strong coStrong
ion
interaction
interaction interaction
Figure 9.3 A correlation diagram for a d
1
ion in an octahedral environment. All
states and orbitals are of g type. and this subscript has therefore been omitted.
270 APPLICATIONS
possibility III for the configuration. Now \Ve note that there are two
states at the left. The higher one n1ust connect to the

3
T
1
g state com-
ing from the
configuration. There is only one state below this,

namely.. that from so this state must be a triplet, and this settles the
assignment of spin multiplicites of the states coming from the configuration.
The remaining connections are now dra\vn in accord with the noncrossing
rule.
The complete diagram is often called a correlation diagra1n. It shows how
the energy levels of the ion behave as a function of the strength of the chemical
interaction with an octahedrally symmetric environment.
The Method of Descending Symmetry
_ In the foregoing pages we have constructed the complete correlation diagram
for a d"!. ion in 0
1
, symmetry. In the course of doing so we ran into the problem
of determining spin multiplicities of the orbital states as they arise from
interelectronic interactions in the configurations e; and but we were able
to obtain a solution by the somewhat oblique procedure of requiring the states
to be just those necessary to correlate with the set of weak-field states. Such
a procedure would be rather impractical in more complicated cases. but there
is a straightforward and general method, due to Bethe and known as the
method of descending symmetry, which can always give us the required in-
formation. We shall explain this method by showing its appJication in the d
2
case, and it should then be obvious how the same procedure can be used in
any case.
Let us begin with the configuration. We have shown that this must go
over into the states A
and as the electron interactions take effect.

Suppose now that we take the environment of the ion, which has 0
11
symmetry,
and lower this symmetry to D_.
1
,., say by taking a trans pair of ligands in an
octahedral MX
6
complex and moving them out to a greater distance than the
other four. The degeneracy of the one-electron orbitals is now lifted, and,
as may be seen from the correlation table in Appendix IIB, we get two
nondegenerate levels of symmetries
and b
1
g:

0
1
, D.u,
Now the number of ways in which \Ve can place the two electrons in the two
Ievels, a
1
g and b
18
, are the following:
.,
a.r:

big
Direct
Product
Possible Spin
Multiplicities
IAt.r:
lBIIP 3Btg

Clearly the exclusion principie requires that the A
18
states resulting from the
configurations aj
8
and must be singlets, that is, the spins of the electrons
must be different; for the configuration
however, since the two electrons

have different orbital states, there is no restriction on their spins, and the
resulting states
1
B and
3
B t.r: are both permitted. lt should be noted that the
total number of arrangements of electrons in the four eg-type boxes was six
and the total is still six in D
41
, symmetry, as it must be.
Now just as the one-electron level eg in 0
1
, symmetry goes over into the
levels a
18
and b
1
.r: when the symmetry is lowered to D
4
h, so the states deriving
from the configuration in Oh symmetry. namely, A
18
, A
2
.P and must go
over into states appropriate to D
411
symmetry. lnspection of the correlation
table shows that the relationship is
O,, D4/r
At.r:--+ A,n
A
2
.r:--+ B
18
E --+ {A
g Bl.r:
Lowering the symmetry cannot change the spin degeneracies; hence, if the
A state in Oh is a singlet, then the corresponding A
18
state in D
must al so
be a singlet, and so on. Moreover, whatever is the multiplicity of state
in 0
1
,, both the A and B
1
g states which arise from it on lowering the symmetry
to D
41
, must ha ve that same spin multiplicity. Since the only A
1
K states available
in D
4
h are
1
A
1
.r:, it immediately follows that the correlation between the Oh
states and the D
41
, states must be with the spin multiplicities as shown below:
0,
'Aig
38
l.r:
'E --+ {IA,g
R 'B
\. lg
272 APPLICATIONS
This fixes the multiplicities of the states in 0
11
and, of course, gives the same
assignments of spin multiplicities that we previously deduced.
Let us now proceed to the states arising from the configuration, namely,
+ EK +
We found previously, it will be recalled, that only

three assignments of multiplicities were possible, consistent with the total
degeneracy (15) of the configuration, which must be conserved. These are
enumerated again in the chart below for convenient reference.
X l1g =
{f

T,g
T'!t:
Possible spin 1 1 3
multiplicity 1 3 1
assignments 3 1 1
Corresponding ,41! Ag
Ax
representations
Bn
Bp,
AR
in C
21
,
B.t:
lt is now necessary to look for a subgroup of 0, such that each of the
representations A E.'P T
1
g, and T
2
R of 0
1
, goes over into a different one-
dimensional representation or sun1 of one-dimensional representations of the
subgroup. Unless these are all different, it will not be possible to obtain a
complete and unambiguous result. lnspection of the correlation table for 0
1
,
in Appendix IIB shows that the subgroups C
2
1r and C
2
,. will be satisfactory.
We shall use C
21
, here; the reader may obtain practice in appJying the method
by using C
2
,. to verify the results. In the chart above we have listed under
each of the 0
1
, representations the representations that correspond to it
as obtained from the correlation table.
Since (21: in o,, goes over to al: + ag + in the direct product X
goes o ver into the sum of the six direct products of + a, + namely,
ar. x = An
ar. x bR = Bg
bJ: X br. = Ar.
The first of these represents the occupation of one orbital by both electrons
and so must be a singlet,
1
AR. The second corresponds to the placement of
each electron in a different aR orbital and can therefore give rise to both singlet
and triplet states, +
3
The third and the fifth al so correspond to placing
the electrons in different orbitals, and these too give rise to both triplet and
LlGAND FIELD THEORY 273
singlet states. namely, 2
1
+ 2:.B,r The fourth and the sixth correspond to
placing the hvo electrons in the same orbital and thus give rise only to singlet
states,
In summary. then. the direct product t

1
g x which goes o ver
into (ag + a
8
+ bg) X (ag + a.r: + b:) in C!/, gives rise to the following states
in c'1/,:
It may be noted that the sum of all the degeneracies, 4(1 x 1) + (3 x 1) +
2(1 x 1) + 2(3 x 1). equals 15, as it must if no errors have been made.
We can now obtain the result we desire. We can immediately make a unique
assignment of multiplicities to the states listed in the lower part of our chart
by noting that there is only one state and two
states. These must

therefore be assigned to the Ag and two Bg states arising from
thus
establishing that the T
1
n state is a spin triplet and. hence. that the A E1: ..
and T
1
x states are all spin singlets.
Energy Level Diagrams in Tetrahedral Environments
Energy level diagrams for ions in tetrahedral environments can be constructed
by the same procedures as those described in the preceding pages for the
octahedral case. We shall briefty outline he re the procedure for d
1

To obtain the Ieft (weak interaction) side we look up each of the free-ion
terms in Table 9.3 and find that these terms split as follows:

3
A::! +
3
T
1
+
3
T
1

1
E + 'T'!
Turning our attention now to the right (strong interaction) side ofthe diagram ..
we observe in Figure 9.2 that the one-electron e orbitals are more stable (by
A,) than the one-electron t"!. orbitals. Hence the three configurations under
the inftuence of a strong interaction with the tetrahedral environment \Vill
in increasing arder of energy; e:!, et'!. Taking the direct product represen-
tations of these and obtaining their constituent irreducible representations.
we see that interelectronic coupling will cause the following states to arise:
274 APPLlCATIONS
We can now assign the correct spin multiplicity, 1 or 3, to each of these states
by the same methods described in detail for the octahedral case and then
correlate the states on the two sirles of the diagram to obtain the complete
correlation diagram shown in Figure 9.4. -
The Hole Formalism
The methods illustrated above for working out the correlation diagrams of
d
2
ions in octahedral and tetrahedral environments can be applied to all dn
configurations for 2 < n ;5; 9. However, the labor involved increases extraor-
dinarily fast as the number of electrons increases. Fortunately, severa) kinds
of relationships make it possible to obtain certain diagrams from others which
are more readily constructed.
One of these kinds of relationship is the hole formalism. According to this
Free
IOn
Weak
interaction
Strong co Strong
interaction interaction
Figure 9.4 A correlation diagram for a d'! ion in a tetrahedral environment.
principie, which is thoroughly rigorous, a dw-, configuration will at
all points along the abscissa of the energy level diagram, in the same way as
the corresponding d'' configuration, except that all energies of interaction \Vith
the environment will have the opposite sign. A physical way of looking at
the problem is to say that n hales in the d shell may be treated as n positrons.
In their interactions with one another, n positrons will behave the same as 11
electrons. However, where the environment tends to repel an electron, it \vill
with the same force attract a positron and vice versa.
The consequences of this relationship are most easily seen by considering
the right side of the correlation diagrams. For n electrons in an octahedral
environment the configuration containing as many electrons as possible in the
t"! orbitals is the most stable, \Vhereas for n positrons such a configuration is
the least stable one. It should be easy indeed to see that quite generally the
states at the extreme right of the diagram for a d" configuration {in either
octahedral or tetrahedral environments) will be in verted for the corresponding
d
1
u-n configuration. Since the energy order of the states which arise from the
free-ion terms of a d" configuration is dictated by the energy arder in which
they go into the various configurations on the it follows that the
splitting patterns of the free-ion terms of a d" system will be just inverted for
the corresponding dw-rr system.
Thus, for example, the correlation diagram for a dx system in an octahedral
field is obtainable from Figure 9.3 simply by inverting the vertical arrangement
of the configurations et
2
, e
2
and redrawing the connecting lines.
More General Relations
The prescription given in the preceding paragraph for obtaining the corre-
lation diagram of a d
8
configuration in an octahedral environment from that
for a d
2
configuration in an octahedral environment is exactly the same as
one for obtaining the diagram of the d
2
ion in a tetrahedral environment from
that for the d
2
ion in an octahedral environment. Changing the environment
from octahedral to tetrahedral inverts the energies of e and t
2
orbitals, and
so also does the change of n electrons to n positrons while keeping the
symmetry of the environment the same.
We may thus state the following very general in which we use the
symbol d"(oct) to mean the energy leve) arder for a d'' system in an octahedral
field, with the other symbols having analogous meanings:
d"(oct) = dw-"(tetr) are inverse to dn(tetr) = d
10
-n(oct)
Thus for the 18 possible cases, that d
1
- d
9
each in tetrahedral and
octahedral environments, correlation diagrams can be obtained by explicitly
working out only those for the following cases, which are the simplest:
276 APPLICATIONS
o
d
2
B = 860 cm-
1
for V(lll)
10 20 30 40
ll/B
f
830 cm-
1
for Cr(II)
d
4
B = 1140 cm-
1
for Mn(III)
~ 4 4 cm-
1
for Fe(lV)
50
10 20 30
ll/B
40 50
10
{
766 cm-
1
for V(ll)
B-
- 1030 cm-
1
for Cr(lll)
20 30
ll/B
40 50
{
960 cm-
1
for Mn(JI)
d
5
B = --1100 cm-
1
for Fe(lll)
10 20 30
ll/B
40 50
Figure 9.5 Energy level diagrams [after Tanabe and Sugano, J. Phys. Soc. Jpn., 9,
753 (1954)] for the d
1
-ds configurations, in octahedral symmetry. Note: For ti" this is
the correct diagram-not the incorrect one often encountered [A. C. Hormann and
C. F. Shaw. J. Chen1. Educ., 64., 918 (1987)].
-1
B = 970 cm for Co(Il)
B _ {1058 cm:-
1
for Fe(ll)
- 1065 cm-
1
for Co(lll)
8 0 ~ ~ ~ ~
!A(tz'J
10 20 30 40 50 10 20 30 40 50
ll/B A/B
d8
B = 1080 cm-
1
for Ni(II)
80
70
60
50
CQ
40
.........
~
'Al
30
tE
o 10 20 30 40 50
A/B
Figure 9.5 ( Continued)
The d
5
case is special in that all of the four related diagrams he re are identical.
The d" and d
10
-" are of course the same configurations when 11 = 5 and
d
5
(oct) = ds(tetr).
In Figure 9.5 are shown energy evel diagrams for all ofthe d" configurations
in octahedral environments. These diagrams are plotted in a manner requiring
sorne comment. Instead of using absolute units for the ordinate and abscissa
278 APPLICATIONS
scales, which would restrict the application of each diagram to just the one
case in which the separations of the free-ion terms matched those in the
the energy unit is the interelectronic repulsion parameter B. On
each diagram are given the values of B for the common free ions of the
corresponding d" and dw-n configurations. In the diagrams are so
drawn that the energy of the ground state is taken as the zero of energy for
all values of L\
0
Thus. in cases where the ground state changes there are sharp
changes in the slopes of all lines. lt is to be emphasized that these
are artifacts of the diagrams and do not represent real discontinuities in the
energies of the states.
Energy Level Diagrams for Lower Symmetries
We have so far considered only the most highly symmetrical situations of
common occurrence. We have seen that for octahedral and tetrahedraJ sym-
metry the d orbitals split into only two sets" and thus only one parameter,
or is required to describe the energy pattern. In cases in \vhich the sym-
metry is high and the nun1ber of free parameters is small, symmetry consid-
erations are, as we ha ve seen, highly informative. When the symmetry is
lower" there are more splittings and more parameters, and hence less m ay
be learned fron1 symmetry considerations alone regarding the relative arder
of the levels. Symmetry considerations then tend to become less an end in
themselves and more a necessary preliminary to setting up the equations for
a calculation in the simplest and most expedient form. With regard to the
interpretation of spectral intensity measurements and especially the polari-
zation of absorption pure symmetry considerations remain immensely
useful, as will be seen in Section 9.6.
One of the most commonly occurring of the lower symmetries in coordi-
nation compounds is D_.,. This is the point group of atoms surrounding and
directly interacting with the metal atom in square planar complexes. of oc-
tahedral complexes which are distorted by elongation or compression along
one of the fourfold and of trans-disubstituted octahedral complexes such
as trans-[ Co(NH
3
)
4
Ch] +. We shall discuss this point group as one example to
illustrate the general nature of the problems which arise when the symmetry
is lower than cubic.
As shown in Table the d orbitals form a basis for a representation of
the group D.u, that contains the irreducible representations A
B
2
g, and
E:. By referring to the character table for we can obtain the more specific
information that the d orbitals correspond with these representations in the
following way:
B
. d ,
1: .r- ,.-
Since there are four different symmetry species of orbitals, we now require
three parameters to specify the energy differences between them. The relative
values of these parameters must be known in arder to find the relative energies
of these four types of orbital, so that at least two actual nuntbers, ratios of
two of the parameters to the third, perhaps, must be known before any sort
of energy level diagram can be constructed. Where the deviation from perfect
octahedral symmetry is small, it can be assumed that the a
1
n and b
1
g Jevels
arising from the splitting of the e P. Jevels in Oh wiJI be separated by an energy
comparable in magnitude to the separation of the b
2
n and en leveJs arising
from the t'lg levels in 0
11
, and that both of these energies will be small relative
to the energy difference, L\
0
, initially existing between the eg and
levels in
the octahedral environment. Nevertheless, there remains the question of the
relative energies of
versus b In and of b'1P. versus

Relation of Energy Level Diagrams to Spectral and Magnetic
Properties of Complexes
Among the most important applications of energy level diagrams of the sort
we have been discussing is their use in the interpretation of spectral and
magnetic properties of complexes and other con1pounds of the transition
elements. For detailed discussion of these applications the reader is referred
to several books dealing more broadly with ligand field theory, * but it is
appropriate to give here a summary account.
The visible and near UV spectra of transition metal ions in chemical en-
vironments are a consequence of transitions from their ground states to the
various excited states., as these are shown on the energy level diagrams in
Figure 9.5. As will be discussed more fully in Section 9 .6, these transitions
are nominally forbidden by selection rules in first approximation but appear
weakly because of breakdown of these selection rules in higher approxima-
tions. Transitions to excited states with the same spin multiplicity as the ground
state are sorne 10
2
times stronger than those to states differing in spin qua!Jtum
number, as might be expected. Thus the spin-forbidden transitions cause
absorption bands which nearly always are too weak to be observed in ordinary
measurements.
By inspection of the energy level diagrams it is possible to see directly what
sort of spectrum the ion should have in the given environment. For example.,
it can be seen from Figure 9.3 that a d
2
ion in an octahedral complex., say
[V(H
2
0)nJ3+, should have three spin-allowed transitions, from the
3
T
1
ground
state to the upper states
3
T
2
,
3
T, and
3
A
1
Experimentally, two absorption
bands have been found at --17,000 and -24,000 cm-, and these may be
assigned to the
3
T
1
---?
3
T
2
and
3
T
1

3
T
1
transitions if is taken as -21,500
cm -- The lT
1
-- transition would be expected to be at still higher energy
at this value of L1
0
and has not been definitely observed.
As another example, Jet us consider the d
7
ion Co(ll) in a tetrahedral
* See reading list, Appendix IX.
280 APPLICATIONS
for example, in [CoCI_.F-. The energy level diagram is the same
as that for a d
3
ion in an octahedral complex, and from Figure 9.5 we observe
that three spin-allo\ved bands are to be expected. Again only two have been
observed at -5500 and --14,700 cm-
1
From the energy level diagram it
follows that these must be assigned as the
4
A
2

4
T
1
(F) and
4
A
1

transitions, and it would then be predicted that the
4
A
1

4
T
2
transition should
Iie in the range 3000-3500 cn1-
1
In a few other tetrahedral complexes such
a transition has been observed but is extremely weak. The reason for the
weakness is discussed in Section 9 .6. J t is to be noted that A, he re is only
3000-3500 cm-
1
, as compared to the value of cm- for
found earlier. This great difference results from the combination of two effects.
One is that, for a particular metal ion and Iigands, A
0
for the octahedral
complex M X
6
is about twice as great as 6., for the tetrahedral complex lvlX
4

Second. for octahedral complexes, an increase of 1 unit in the oxidation
number of the metal causes Ac
1
to rise by a factor of 1.5 to 2.0.
Insofar as magnetic properties are concerned, the energy Jevel diagrams
provide a ready explanation of the way in which the symmetry and strength
of interaction of the environment determine the spin multiplicities of the
metal ions in their compounds. Basically, two different situations arise. In
one of these, exemplified by the octahedral d
1
, d'l, d
3
, d'\ and the
ground state is derived from the lowest term of the free ion for all values of
the parameters L\
0
, however large. Hence the number of unpaired electrons
must be the same as that in the free ion, however strongly the ion m ay interact
with its environment. In the other cases, namely, d
4
d
5
, d
6
, and d
1
, the
ground state is derived from the lowest free-ion term only out to a certain
critical value of A
11
, beyond which a state of lower spin multiplicity originating
in a higher free-ion term drops below it and hence becomes the ground state.
For these systems, therefore, we can predict that for complexes and other
compounds in which the perturbing effect of the environment, as measured
by An, is weak there will be the maximum number of unpaired electrons,
whereas for compounds in which the perturbing effect of the environment is
very strong, greater than the critical value of Ac1- there will be fewer ( two or
four fewer) unpaired electrons. Predictions of this sort have been found to
be in remarkable accord with experimental observations. Although similar
predictions could be made for the various d" systems in tetrahedral environ-
ments .. they would be of little practica} value., since in real tetrahedral systen1s
the val u e of A, never seems to exceed the critical val u e, and hence all tetra-
hedral complexes kno\vn ha ve the highest possible spin multiplicity.
Another way of deducing whether a given d'
1
ion in an octahedral envi-
ronment will ha ve only one possible spin multiplicity or severa), and what
these multiplicities will be, s so simple that it does not require reference to
any pu blished energy leve) diagrams or indeed e ven the use of pencil and
paper. We consider the set of one-electron d orbitals split as shown in Figure
9.2 into the lower-Jying t
1
g subset and the upper-lying el! subset. When an
electron is to be placed in ths set of d orbitals, two energy terms must be
considered. lf the electron enters the
subset, it will be more stable by an

amount du than if it enters the ex subset. However, if in arder to en ter the
subset it will have to enter an orbital that is already occupied by one

electron. there will be a repulsive energy, usually called a pairing energy, P.
If this pairing energy is greater than d
0
, the electron will go into an orbital
despite the fact that this costs an energy d
0
Thus the critica! value L.\u can
be equated (approximately) to the pairing energy P. The latter can be esti-
mated from spectroscopic data on the free ion.
With these considerations in mind \ve can make the following statements
about the d electron distributions for the various d" configurations in octa-
hedral environments. For the d
1
, d
2
, and d
3
cases the electrons can en ter the
t
2
.a: orbitals without any need of double occupancy of any orbital. Hence these
ions will ha ve one, two, and three unpaired electrons, respectively. regardless
of the magnitude of L\
0
For the dK and dtJ ions all possible configurations
require double occupancy of three and four orbitals, respectvely, and the
lowest energy configurations will always be those, and with the

orbitals fully occupied regardless of the magnitude of dn. Thus the ground
states of d', d
2
d
3
, and d
9
ions in octahedral environments must ha ve the
maximum number of unpaired electrons irrespective of the magnitude of A
0

For d\ d
5
, d
6
, and d
7
ions, however. the electron distributions will depend
on the magnitude of du compared to P, as indicated in the following table.
The numbers in parentheses indicate the numbers of unpaired electrons.
Configuration p a..> p

(2)
(r
tlt:ei(5)
('

(1)
dfl
4)
1('
:!,t:
(O)
d1
tt
11
ei(3) (1)
1t will be seen that all these results are in accord with the condusions
can be drawn from the energy level diagrams of Figure 9.5.
9.5 ESTIMATION OF ORBITAL ENERGIES
The numerical evaluation of the ehergies of orbitals and states is fundamen-
tally a matter of making quantum mechanical computations. As indicated in
Chapter 1, quantum mechanics per se is not the subject of this book. and
indeed we have tried in general to avoid any detailed treatment of methods
for solving the wave equation, emphasis being placed on the properties that
the wave functions must ha ve purely for reasons of symmetry and irrespective
of their explicit analytical form. However, this discussion of the symmetry
aspects of ligand field theory would be artificial and unsatisfying without sorne
282 APPLICATIONS
brief outline of the various models which may be used to make computations
and also to visualize the nature of the interaction bet\\'een the metal ion and
its chemical environment.
Our discussion here of computational procedures will be very superficial
and aimed at bringing out the physical features of the models. For a full
treatment of this subject with references to the original and for
further discussion of the interpretation of the chemical behavior of transition
metal compounds in terms of ligand field theory., the reader is referred to the
publications cited in Appendix IX.
As noted in Section 9.1, there are three dosel y related theories of the
electronic structures of transition metal complexes., all making quite explicit
use of the symmetry aspects of the problem but employing different physical
models of the interaction of the ion with its surroundings as a basis for
computations. These three theories, it will be recalled, are the crystal field,
ligand field, and MO theories. There is also the valence bond theory which
makes less explicit use of symmetry but is nevertheless in accord with the
essential symmetry requirements of the problem. We shall now briefly outline
the crystal field and ligand field treatments and comment on their relationship
to the MO theory.
The Crystal Field Theory
This model of a complex or of a crystalline salt of a metal ion in a compound
such as a halide or oxide is of an electrostatic, point charge, or point dipole
type. The ligands or neighbors of the metal ion are treated as
orbital-less point charges., which set up an electrostatic fiel d. The effect of this
field on e)ectrons in the d orbitals of the metal ion is then investigated.
Group theory alone has shown us that a single d electron in an ion that is
at the center of an octahedron may be in either of two states. In one of these
it will have either of two wave functions or one which is a linear combination
of both, which together provide a basis for the representation of the group
0,. In the other state it may have one of three wave functions or so me linear
combination of these, this set of three being such as to provide a basis for
the T
2
g representation of the group 0,. If we refer to the character table for
the group 0
1
, we note that the wave functions d.r1 __ ,,2 and d::'! form a basis for
the Eg representation. Hence we may consider these to be the orbitals which
may be occupied by an ex electron. Similarly a
electron may be assumed

to occupy one of the orbitals d_ .. Y' d_vz, d.lz, since these forro a basis for the T
2
g
of the group. We shall now show how the energies of electrons
in these orbitals are estimated according to the electrostatic model of the
crystal field theory.
We assume that each of the six ligands is either an anion such as 0
2
-, F-,
CJ-, ... , or a dipolar molecule such as Qti- ... , having its
negative end clase to the cation. In either case we may then look upon the
environment of the cation as being that shown in Figure 9.6 in relation to a
z
y
X
Figure 9.6 The electrostatic environment of a cation, C, surrounded by an octahedral
array of anions or dipoles.
specific set of Cartesian axes. Befare considering how, on the basis of such
an electrostatic model, the actual magnitud e of the energy difference between
the e
8
and
states of the d electron might be computed, Jet us first consider

the simpler. qualitative question of which state, en or t
28
, is the more stable.
Group theory tells us that both eg orbitals have the same energy and that
all three t
18
orbitals ha ve the same energy. Hence we need only compare
either of the orbitals with any one of the t
2
g orbitals to obtain the answer.
Let us choose the and orbitals for our comparison. The shapes of
these orbitals are indicated in Figure 9. 7 by drawings intended to show surfaces
cnclosing a major fraction, say 90%, ofthe electron density. Since the electron
has a negative charge and the ligands either are negative or appear so to the
electron, it is apparent that because of electrostatic repulsive forces the elec-
z
z
y y
X
X
dxz _y'! (eg) dxy (t
2
g)
Figure 9.7 Sketches showing the d,.l __ and d". orbitals, representative of the eg and
t2
8
orbitals, respectively, in relation to the coordinate axes.
284 APPLICATIONS
tron is more stable in the d."C\' than in the d.'C:!_,,:z orbital. The interaction of the
electronic charge with the charges of the ligands lying along the z axis will
be the same in either case, but it is evident that the electronic charge is much
more concentrated in the region of the other negative ligands in the d.r"!.-
orbital than it is in the d.ty orbital. Thus we know, qualitatively, that, insofar
as the eJectrostatic model is a faithful representation of the true situation.
the en orbitals are of higher energy than the t
28
orbitals.
It is easy to see the correctness of the group theoretical result that the d:r"'
dxz, and d.v:: orbitals must all ha ve the same energy. Each of these is identical
in form to the other two, differing only in the plane in which its maxima Iie.
That the and dzz orbitals ha ve the same energy is certainiy not so obvious,
merely on inspection. However, it is easy to grasp this equivalence by using
the following line of reasoning. We know from wave mechanics that there
can be only five linearly independent solutions to the wave equation with the
same value of the quantum number n (;;:::::3) and with 1 = 2. Howeverl we
may write an infinite number of solutions initially and select any five linearly
independent combinations we desire (see page 101). Accordingly. let us con-
sider the following six functions for the angular parts of nd orbitals:
= xy
, ., .,
t/4 = x- - y-
fii = yz
lf/3 = xz
, , .,
lf/h = z- - x-
Of these six the following set of five is usually selected:
f//1 = lf/ = xy lf/J = fi/J = xz
lf/2 = = yz
lf/4 = = x2 - y2
1 ( , ') ? ., ., ., .,
f/ls = v'2 f/ls + fllta = _z- - x- - y- = z-
Thus we see that the dz1 orbital can be regarded as a normalized linear com-
bination of dz'--y"!. and dz'!-x:! orbitals. Now the dx'--y1, d='--_,.1. and d='--x'! orbitals
are obviously all of identical energy in an octahedral field, and, as shown on
page 101, any linear combination of two degenerate wave functions has the
same energy as each of its constitutents. Thus we can see by geometrical
reasoning ihat the d:-z and d:r2-_..,.2 orbitals are degenerate in an octahedral field,
despite the fact 'that they may not clearly appear to be equivalen t.
We m ay, on the basis of the foregoing argument, draw the simple energy
level diagram, Figure 9.8, showing the relative energies of the eg and t
2
c
orbitals. We know that the levels are of higher energy than the levels,
and we have denoted the magnitude of this difference by Llu or lODq, which
are the symbols commonly used in the literature.
We may now consider whether it is possible to calculate the value of An
eg{=========
ll
0
or lODq
t2g======
Figure 9.8 Splitting of d orbitals in an octahedral crystal field.
by using an eJectrostatic model. To do so we consider anion ligands as point
charges and dipolar ligands as point dipoles. For the latter we should also
have to estmate the effective value of the dipole moment, which would be
equal to the permanent moment plus that induced by the positive charge of
the metal ion. Finally, we should require knowledge of the metal-ligand
distance and a proper radial wave function for the d electron. 1 t \vould then
be possible to perform a calculation of the
separation. Such caJculations

ha ve been carried out in a number of cases. For di polar Jigands, it is necessary
to assume very unrealistic values for the effective dipole moments in order
to obtain correct values of d
0
For ionic ligands the model gives results which
are of the right order of magnitud e but not much better than that. However,
it is now generally recognized that this purely electrostatic model is too simple
to be taken literally, since in all cases the value of de, is determined by
interactions other than purely electrostatic ones. Thus Llu is best regarded as
a phenomenological para meter to be determined from experiment rather than
as one to be calculated from first principies using the crystal field model.
One further aspect of the splitting of the one-electron d orbitals in an
octahedral field must be noted. Suppose that we consider the foiJowing Ge-
danken experiment. We surround an atom or ion by a concentric spherical
shell of uniformJy distributed negative charge, the total charge being 6q units.
A set of 10 d electrons in this ion will now ha ve an energy Es, which is higher,
because of repulsive forces between the electrons and the outer shell of
negative charge, than its energy Eu, in the free ion. However, since the charge
distribution is spherical, the d electrons all ha ve the same energy. Now sup-
pose that the total charge on this spherical shell is redistributed, but moving
only on the surface of the sphere, so as to place six point charges, each q
units in magnitude, at the six apices of an octahedron. This redistribution
cannot change the energy of the d w configuration as a whole, and yet we
know that now six electrons are in
orbitals and four are in e,: orbitals and

that these orbitals differ in energy by d
0
These relationships are depicted in
Figure 9.9. In order for the total energy of the d
10
configuration to be the
same in the octahedral field as it is in the spherical field, the following equa-
tions must hold:
6(Es - B) + 4(Es + A) = lOEs

286 APPLICATIONS
t
E o ..Q.2.. ..2.2.. ..Q.Q.. .ruL ..Q..Q...
Free ion
E ...Q.Q... ..Q.Q_ .Q.Q.. ...Q.Q.. ...9..Q..
s Ion in spherical field
..2.2.. _Q_Q_
TT
------+t- .1o
..Q.Q.....Q.Q....QQ.._i_j_
Ion in octahedral
field
Figure 9.9 An energy diagram showing how the energies of the electrons in a d'"
configuration are affected by spherical and octahedral electrostatic fields.
Solving, we obtain
Hence
38 = 2A
A
_ 3A
- 5Llo
1t should be noted that the splitting d
0
is gene rally of the order of 1-3 e V,
whereas the elevation of the set of d levels as a whole is of the order of 20-
40 e V. Thus it should always be borne in mind that the crystal and ligand field
theories focus attention on only one relatively small aspect of the overall
energy of formation of a complex.
For a d electron in the electrostatic field of four anions or dipoles arranged
tetrahedra11y around it, the splitting pattern can be derived by an analogous
line of reasoning. Group theory tells us that the fivefold degenerate state of
the d electron in the free ion is split into two states, one twofold degenera te,
E, and one threefold degenerate, T
2
Reference to the character table for the
Td group shows that the former state will be one of the two orbitals d:'1 and
ora linear combination thereof, and that the T
2
state will be one of the
dx_v, dxn and dyz or sorne linear combination of these.
In order to find out the relative energies of the e and t
1
orbitals, let us
place our tetrahedral complex in a set of coordinate axes as shown in Figure
9.10. Once more, we can make a comparison between either of the e orbitals
and any one of the t
2
orbitals, and perhaps the best selection in order to
visualize the relative electrostatic energies is again the pair d:c) and _ },:!. The
difference here is much less striking than in the octahedral case, but it can
be seen that qualitatively an electron in the d:cy orbital will have a higher
X
Figure 9.10 The electrostatic environment of a cation surrounded by a tetrahedral
array of anions or dipoles.
poten ti al energy than one in the dx'!- If we call the energy separation
between the e and t
2
orbitals d,, then by an argument exactly analogous to
the one used in the octahedral case we can show that the energies of the
and e orbitals relative to their energies in a spherical shell of the same total
charge will be as shown earlier in Figure 9.2.
lt is worth noting that, when detailed expressions are written for du and
ll, in the purely electrostatic crystal field approximation, it turns out that for
equal charges and metal-ligand distances these two quantities are in the ratio
E ven though the absolute values calculated by this modelare, as noted befo re,
grossly inaccurate, the ratio is in approximate agreement with experiment.
Tbe Ligand Field Theory
This is a modification of the crystal field theory in which we drop the as-
sumption that the partially filled electron shell is one consisting of pure d
orbitals. Instead it is admitted that there is overlap between the d orbitals of
the metal and the orbitals of the ligand atoms.
Various lines of evidence indicate that even in complexes in which the
binding might be expected to be most ionic, for example, in hexafluoro com-
plexes such as and in aquo ions such as [Fe(H
2
0)
6
]2+, overlap of metal
and ligand orbitals occurs toa small but significant extent. Thus there is direct
evidence from ESR and NMR studies that spin density of the "d" electrons
288 APPLICATIONS
has a nonzero value at nuclei in the ligand molecules, and that in fact actual
delocalization of the electron into ligand orbitals occurs. Numerical estimates
from these data of the time spent by a "d .. electron in ligand AOs in species
such as (in Mnf
2
) and are -2-5% per ligand atom. From this
it can be concluded that the "d" orbitals have only about 80 10% d character
and 10-30% ligand orbital character. In reasonable accord with these ob-
servations is the fact that in complexes the interelectronic repulsion energies,
which are responsible for the energy differences between the different terms
of a dn configuration, are only -70% of their magnitudes in the free ions.
There are two practical consequences of the recognition of overlap between
metal and ligand orbitals. First. we abandon all hpe of making a priori
calculations of orbital splittings by a pure point-charge electrostatic model
using pure d wave functions. The electrostatic expressions for the H d" orbital
energies remain valid in their general form, but the crystal field parameters,
the effective charge or dipole of the ligands, the metal-ligand distance, and
the radial part of the d orbital wave functions now lose their literal physical
significance and must be considered as fictitious adjustable parameters.
Second, in constructing the energy level diagrams for dn configurations,
we must le ave the separations between the various free-ion terms as functions
of adjustable interelectronic repulsion parameters, rather than simply setting
them down at the free-ion values. Practical energy level diagrams _such as
those in Figure 9.5 are interpreted using free-ion term separations equal to
--75% of the separations spectroscopically observed for the free gaseous ions.
Comparison with Molecular Orbital Theory
Inspection of Figurs 8.14 and 8.11, the MO energy level diagrams for oc-
tahedral and tetrahedral complexes, respectively, shows how the crystal field
or ligand field treatment of the electronic structure of a complex fits into the
complete picture which an MO treatment provides. The d orbitals of the
metal ion provide the majar contribution to the eg and t
2
g (octahedral) ore
and t
1
( tetrahedral) orbitals lying in the center of the diagrams. M oreo ver,
all MOs lower in energy than these are filled \Vith electrons that originate in
ligand orbitals, and indeed the MOs in which these "ligand" electrons now
reside are made up mainly of the originalligand orbitals. The e and t
2
orbitals
contain the electrons that were originally in the pure d orbitals of the un-
complexed metal ion.
The crystal and ligand field theories were developed to deal with only those
properties of the complexes that are derived directly from the set of electrons
originally occupying the d orbitals of the metal ion. Since these orbitals are
the principal parents of the e and of the complex it is not unreasonable
to treat the latter as though they were nothing more than split ( crystal field
theory) or split and somewhat diluted (ligand field theory) metal d orbitals.
It is clear, however, that such a view can be only an approximation-indeed,
a fairly ruthless one. Yet. with judicious empirical choice of one or more
disposable parameters (the splitting energy, at least, and perhaps also the
interelectronic repulsion energy and the spin-orbit coupling constant) the
approximation has great practical utility for certain purposes. Specifically, it
is rather good for fitting electronic spectra when only Hd-d" transitions are
involved and for interpreting magnetic behavior when the unpaired eJectrons
occupy only ud" orbitals. The crystal and ligand field theories are .. however,
formalisms and cannot ever give complete and literal descriptions of the en tire
electronic structures of complexes.
9.6 SELECfiON RULES AND POLARIZATIONS
In Section 5.3 the general symmetry restrictions on transitions occurring by
dipolar interaction \\'th electromagnetic radiation were discussed. Here they
will be invoked with particular reference to electronic transitions on metal
ions in ligand fields.
Centrosymmetric Complexes, Vibronic Coupling
In a complex that possesses a center of symmetry, all states arising from a
d" configuration have the g character inherent in the d orbitals. Since the
dipole moment vectors belong to odd representations, all of the integrals such
as f fll;x r
8
dr are identically zero beca use the direct product of two g func-
tions can never span any u representations. On this basis alone, we would
predict that transitions between the various states arising from d" configu-
rations in octahedral environments would ha ve zero absorption intensity. In
fact, these transitions do take place but the absorption bands are only -I0-
3
times the intensity expected for symmetry allowed electronic transitions. Thus
the prediction we have made is substantially correct, but at the same time
there is obviously sorne intensity giving mechanism that has been overlooked.
lt is generally accepted, following Van Vleck, that this mechanism is one
called vibronic coupling-that is, a coupling of vibrational and electronic
wave functions. In a qualitative sense we may say that sorne of the vibrations
of the complex distort the octahedron in such a way that the center of sym-
metry is destroyed as the vibration takes place. The states of the d" config-
uration then no longer retain rigorously their g character, and the transitions
become "slightly allowed." Figure 9.11 shows the approximate nature of
several of the modes of vibration of an octahedron which do destroy the
symmetry center.
This phenomenon of vibronic coupling can be treated very effectively by
using group theoretical methods. As will be shown in Chapter 10, the vibra-
tional wave function of a molecule can be written as the product of wave
functions for individual modes of vibration called n9rnzal modes, of which
there will be 3n - 6 for a non linear, n-atomic molecule. That is, we can
290 APPLICATIONS
Figure 9.11 Two of the normal vibrations of an octahedral AB,, molecule in which
the displacements of the atoms destroy the center of symmetry. Another type of T
1
u
vibration. not shown here, has the same property.
express the complete vibrational wave function, f/1,., as a product of 3n - 6
functions ea eh pertaining to one of the normal modes, namely,
),-6
f/1 ,. = TI f/1
i= 1
1t is further shown in Chapter 1 O that, when each of the normal m o des is in
its ground state, each of the f/1; is totally symmetric and hence f/1,. is totally
symmetric. If one of the normal modes is excited by one quantum number,
the corresponding f/1; m ay then be long to one of the irreducible representations
other than the totally symmetric one, say f;, and thus the entire vibrational
wave function f/1,. will belong to the representation f;. Simple methods for
finding the representations to which the first excited states of the normal
modes belong are explained in Chapter 10. In this section we will quote
without proof results obtained by these methods.
Toa first approximation, and usually a rather good one, the complete \Vave
function '1' for a molecule can be written as a product of an electronic wave
function lfle, a vibrational wave function f/1,., and a rotational wave function
f/1 ,:
1t is then assumed that none of these factors of the complete wave function
are interdependent, so that instead of having to sol ve one large wave equation
, 'lt .'fl = E'fl
it is possible to salve three simpler ones:
.ltiJ, = E,f/1,
and write the total energy as a simple sum of the electronic, vibrational, and
rotational energies, namely,
E = Et' + E,. + Er
This is, of course, only an approximation. Although it works well for many
purposes, one of its limitations is that it cannot explain the low but nonzero
intensity of the transitions between states of d" configurations in centrosym-
metric environments as we ha ve shown abo ve.
The way out of the difficulty is to drop the assumption that fllt and f/1,. are
entirely independent, though retaining the approximation that f/lr can be treated
as independent of these other two. Thus it is not the values of integrals like
which we must consider, but rather the values of the integrals
It is easy to show by symmetry arguments that the latter do not in general
vanish. First, we note that, if we assume that in the lower state, f//t'f/1,., the
molecule is in its vibrational ground state, flll is totally symmetric and we can
ignore it. Our problem then is to decide whether there are any vibrational
wave functions belonging to representations such that, although the direct
product representation of f/l;x f/1" does not contain the totally symmetric rep-
rese:-:tation, the direct product representation of f/l;f/I;.Xf/1" does. Whenever
this is so the transition will be vibronically atlowed. According to the results
of Section 5.2, the integral will be nonzero if there is any normal mode of
vibration whose first excited state, f/1;., belongs to one of the representations
spanned by f/I;X f/1 e
In arder to show ho\v this is done Jet us take a simple example. For the ion
[Co(NH
3
)
6
]3+, the ground state lfle transforms as 'A
1
g. There are two excited
states with the same spin (S = 0), which belong to the representations T
1
~ :
and T l ~ : In the group olr the coordinates X, y' z jointly form a basis for the
T,u representation. Thus, for the
1
A tx ~
1
T
1
~ : transition the direct product
representation of fll;(x, y, z )f/1 e is given by
This can be reduced to
292 APPLICA TIONS
Thus, if there are any normal vibrations whose first excited states belong to
any of these representations, there will be nonvanishing intensity integrals.
By the methods of Chapter 10 it is easily found that the symmetries of the
normal modes of an octahedral AB
6
molecule are
Thus, while the pure electronic transition
1
A
18
-+
1
T
18
is not allowed, all the
transitions in which there is simultaneous excitation of a vibration of T," or
T
2
, symmetry are allowed.
Similarly, for an
1
A
18
-+
1
T
2
g transition, we find
= A2u + E, + T
1
, + T
21
,
Thus the
1
A t.r:-+
1
T
28
transition can also occur so long as there is simultaneous
excitation of a T
1
, or T 2u vibration.
Vibronic Polarization
For an octahedral complex we see that the direction of vibration of the electric
vector of the light makes no difference, for the directions x, y, and z are
equivalent in the sense that they are interchangeable by the symmetry op-
erations of the molecule. However, in less symmetrical complexes in which
x, y, and z do not all be long to the same representation, we encounter the
phenomenon of polarization.
Let us suppose that we place a polarizing prism between the light source
and the sample. lf the sample is a single. crystal in which all of the molecules
have the same orientation relative to the crystallographic axes, we can so
orient the crystal that the direction of the electric vector of the light will
correspond to the x, y, or z direction in a coordina te system for the molecule.
It is then possible that sorne transition may occur for only one or two of these
orientations but not for all three.
As an example, let us consider a trans-substituted octahedral complex such
as trans-dichlorobisethylenediaminecobalt(III), which is pictured in Figure
9.12 with a set of coordinate axes. Although the symmetry is no longer cubic,
there is still a center of inversion so that d-d transitions can have nonzero
intensity only if there is vibronic coupling. Since it is only vibrations in the
pait of the molecule induding and immediately surrounding the coba)t ion
\Vhich would be expected to have any appreciable interaction with the e)ec-
tronic wave functions of the metal ion, we shall restrict attention to the normal
vibrations of a group, which has local symmetry D
41
,. The
tz
Cl

(2

Cl
Figure 9.12 A sketch of the trans-dichlorobisethylenediaminecobalt(lll) ion showing
a set of reference axes.
methods of Chapter 10 will tell us that for such a group the first excited states
of the normal modes have the following symmetries:
For the trans-dichloro complex the ground state will be a
1
A state, as in
a strictly octahedral complex, but the excited singlet states
1
T
1
1! and
1
will
be split as follows (cf. the correlation table, Appendix IIB):
T,K: A:!: + EK
T2K : B'!.t: + Eg
Thus the possible transitions from the ground state to excited states will be
of the following types so far as the symmetry of the electronic states is con-
cerned:
l. A,K -7 A"lg
2. A,K-+ B2g
3. A,g-+ EK
For these transitions we obtain the following results for the representations
of the purely electronic dipole integrals:
A,
8
-+ A2g A t..:-+ B'2
8
EK
J lfi;Zifl, dr A,u
B,, E,
J lfl;.(x. Y)lfl, dr
E" E"
A,, + A:!, + B," + B:!,
294 APPLICATIONS
Comparing these results with the list of the symmetries of the first
states of the normal vibrations, we can immediately write the following pre-
dictions for the polarizations of the transitions:
Transition
A
18
A:!g
Alg
A,g
Polarization with
Vibronic Coupling
z (x, y)
Forbidden
Allowed
Allowed
Allowed
Allowed
Allowed
These results have actually been used to analyze experimental data. Figure
9.13 shows the experimental observations of Y amada et al.* on trans-
[Co(enhCI:!]CIHCI2H:!O. It can be seen that there are three regions of
absorption. From -27,000 to -40,000 cm-
1
the absorption exhibits no sig-
nificant polarization. At -22,000 cm-
1
there is a strongly polarized band that
is absent for light parallel to z but present for light perpendicular to z, while
at -16,000 cm -l there is a band that shows so me difference in intensity for
the two directions of polarization but is clearly present in both. The results
of the above analysis permit a simple interpretation of these observations.
"'
DO
.9
(\
1 \
1 '
1 \
1 \
1 '
1
1
1
1
1
1
Frequency, cm -l
Figure 9.13 Dichroism of trans-[Co( enhCI:a .. after the results of Y amada et al. on
The fullline shows the spectrum with light polarized
parallel ( or nearly paralle1) to the C1-Co-Cl axis, and the dashed line shows the
spectrum with 1ight polarized perpendicular to the CJ-Co-C1 axis.
* S. Yamada et al., Bull. Chem. Soc. Jpn., 25, 127 (1952); 28, 222 (1955).
The bands at ---16,000 and -22,000 cm-
1
are assigned. respectively, to tran-
sitions to the Eg and A
2
R components of the T
1
g state existing in 0
11
symmetry,
since an A !g ~ En transition is vibronically allowed for both directions, but
an A l,c: ~ A : ~ : transition is vibronically forbidden for z polarization. The broad
absorption abo ve -27,000 cm -l m ay be assigned to the unresolved transitions
from the A
1
..: ground state to the En and B
1
g states coming from the T
2
~ : state
for 0, symn1etry. An interpretation a long these lines was first given by Ball-
hausen and Moffitt, * who also showed that independent calculations and
experimental evidence would lead to the expectation that the order of the
excited levels shou1d be as postulated in order to explain the polarization
data.
Noncentrosymmetric Complexes
When the complex Jacks a center of symmetry even in its equilibrium con-
figuration. the ud-d" transitions become allowed as simple changes in the
electronic wave functions. The breadth of the absorption bands shows that
the electronic transitions are still accompanied by vibrational changes, but
these vibrational changes are not in themselves essential to the occurrence
of the transition. It is believed that the way in which the noncentrosymmetric
Iigand field alters the d wave function so that the states of the "d"" config-
uration no longer have rigorous g character is by making possible the mixing
of d and p orbitals. We have already shown on page 211 that in a tetrahedral
field the p orbitals and the d.t:_,., dx::' and d_,.= orbitals belong to the same ( T
1
)
representation .. and that to sorne finite degree each of the two sets of T:!
orbitals which will be found among. the orbitals arising from the valen ce shell
atomic orbitals of a transition metal ato m must ha ve both p and d character.
Then, if two different electronic states of the ud"" configuration of a metal
ion r:ontain different amounts of p character, a transition from one to the
other will be to a certain extent a d ~ p or p ~ d transition, which is highly
allowed even in the free atom since the d orbitals are even to inversion and
the p orbitals odd. The exact extent to which this mixing occurs and the
resulting intensity of the transition must, of course .. be computed using explicit
wave functions, but symmetry consiqerations alone can tell us whether it is
possible for a particular transition to acquire any intensity at all in this way.
As a first illustration let us consider the optical transitions in a tetrahedral
complex of Co(II). The ground state belongs to the A
1
representation of the
tetrahedral point group Td, and there are two excited states of T
1
symmetry
and one of T
2
symmetry. The character table for Td te1Js us that the coordina tes
X, y, and z forma basis for the r2 representation. For the : ~ T, transitions
we then see that the intensity integral will span the representations in the
direct product of A
1
x T
1
x T
2
, and this reduces as folloY.rs:
A
2
X T, X T
2
= A
1
+ E + T
1
+ T
2
"'C. J. Ballhausen and W. Moffitt. J. /norg. Nucl. Chem., 3, 178 (1956).
296 APPLICATIONS
Since the A
1
representation is it follows that these transitions are
allowed by the symmetry of the purely electronic wave functions. For the
A:! transition we must consider the direct product A1 X T'! X
which reduces as follo\vs:
We see that the A
1
---+ T
2
transition is not allowed by the symmetry of the
pure electronic wave functions and that whatever intensity it may have must
be attributed to vibronic interaction. In agreement with this prediction, it has
been found that the A
1
T
2
transition is observed but with an intensity 10
to 100 times smaller than the intensities of the A
1
T, transitions in the
systems [e. g., Co(ll) in ZnO] that ha ve been studed.
Polarization of Electronically Allowed Transitions
J ust as with vibronically allowed transitions, in symmetry groups in which all
Cartesian axes are not equivalent (noncubic it is found that. in
general, transitions will be allowed only for certain orientations of the electric
vector of the incident light. One cJass of compounds in which this phenomenon
has been studied both theoretically and experimentally consists of trischelate
compounds such as tris(acetylacetonato)M(III) and tris(oxalato)M(III) com-
plexes. In these complexes the six ligand atoms form an approximately oc-
tahedral array but the true molecular symmetry is only D
3
Since there is no
center of symmetry in these molecules, the pure electronic selection rules
might be expected to be dominant.
For the tris(oxalato)Cr(III) ion, [Cr(C
2
0 .. hP-., the electronic states are
those into which the states of a complex of O, symmetry are reduced when
the symmetry is reduced to D
3
From the correlation table (Appendix IIB)
we see that the correlation of the 0
11
and D
3
states is as follows:
A
28
A
2
(ground state)
T,g A
1
+ E
A
1
+ E
Thus in D
3
symmetry we want to know the polarizations of the following
types of transitions: A
2
---+ A A
2
A
2
, and A
1
E. Noting in the character
table for that z belongs to the A
2
representation and (x, y) to the E
representation, we obtain the following results for the irreducible represen-
tations spanned by the dipole integrals for each of these transitions:
J lfi;Zifl, dr
J lfJ;(x,y)lfJ,dr
Thus the selection rules are
Transition
A,

E
E
A"t
E
Polarization of
lncident Radiation
z (x, y)
Allowed Forbidden
Forbidden Forbidden
Forbidden Allowed
E
lt can be seen that these are very powerful selection rules indeed. On the
other hand, we might have assumed that the symmetry of the environment
of the metal ion could have been adequately by considering
only the six coordinated oxygen atoros. In this case. the symmetry would be
DJJ, in which there is a center of inversion and the transitions would be
governed by vibronic selection rules. When these are worked out, it is found
that all of the transitions are vibronically permitted. Thus. experimental study
of the polarizations should provide clear-cut evipence as to the correct effec-
tive symmetry and selection rules. Such a study has been reported* and shows
conclusively that the selection rules followed are those given above for pure
electronic transitions in D
3
symmetry.
9.7 DOUBLE GROUPS
In Section 9.3 we that for an orb.ital or state wave function having
angular momentum quantum number 1 (or L) the character of the represen-
tation for which this forms a basis, under a symmetry operation that consists
of rotation by an angle a, is given by
( )
sin(/ +
l. a =
sin a/2
(9.3-2)
"'T. S. Piper and R. L. Carlin, J. Chem. Phys .. 35. 1809 ( 1961 ). These authors also give selection
rules and experimental data for the oxalato complcxes of the trivalent ions of Ti. V. Mn, Fe.
and Co.
298 APPLICATIONS
We then applied this formula to various types of single-electron wave functions,
for example s, p., d, f, g, and to wave functions for various Russeii-Saunders
terms characterized by integral values of the quantum number L.
There are., however, many cases of interest in which we may want to
determine the splitting of a state that is well characterized by its total angular
momentum, J. This will in fact be the only thing of importance in the very
heavy for example, the rare earth ions, where states of particular
L cannot be used since the various free-ion states of different J are already
separated by much greater energies than the crystal field splitting energies.
Now J = L + S, and for ions with an odd number of electrons S and
hence J must be half-integral numbers. For states in which J is an integer,
the characters can be obtained by using the above formula with 1 replaced by
J. However, when J is half-integral a difficulty arises. We know that a rotation
by 2n is an identity operation and therefore it should be true that
x(a) = x(a + 2n)
It can easily be seen that this is true when L or J is an integer. However,
when J is half-integral we have
(
? ) _ sin (J + !)(a + 2n) _ sin [(J + i)a + 2n)
l a + _n - . - [
- s1n (a + 2n)/2 sin a/2 + n]
sin (J + !)a
=
-sin a/2
= - x(a)
Since the characters of a representation must be uniquely defined, we see
that those we would obtain by the above procedure when J is half-integral
cannot belong to true representations.
A simple device for avoiding this difficulty was proposed by Bethe. We
introduce the fiction (mathematically possible but not physically significant)
that rotation by 2n be treated as a symmetry operation but not as an identity
operation. We must then expand any ordinary rotation group by taking the
product of this new operation, which we shall call R, with all of the existing
rotations. The new group will therefore contain twice as many operations and
more classes and representations (though not twice as many) than the simple
rotation group with which we start. This new group is called a double group.
In working out the various products C:'R and Re:, we first note that two
rotations about the same axis commute so that C':R = If n = 2 we
have RC
2
, which is a special case since
z(n) = z(3n) = o
For rotation by any other angle, namely, ln2nln, it is not difficult to show
that the following equality holds generally:
x.(n12nln + 2n) = z[(n - n1)2nln]
In order to evaluate the characters of E and R, that is x(O) and z(2n ), \Ve
must evaluate the limit of an indeterminate form, for as O or 2n,
sin (J + O

sin a/2 O
This may easily be done by using l'Hpital's rule, and the results are
x(O) = 21 + 1
(? ) - {2' + 1
X -'Ir - - (21 + 1)
when J is an integer
when J is a half-integer
After we have worked out the characters for all of the new operations,
C';iR, of the double group, we will then collecr them into classes, using the
same rule as for simple groups, name1y, that all operations having the same
characters are in the same class. Th us in general we shall find the following
classes in double rotation groups:
l. E
2. R
3. C:1 and C:1R
4. C, and
5. C': and
We can then determine the number and dimensions of the irreducible rep-
resentations by using the familiar rules that there are as many irreducible
representations as there are classes and that the sum of the squares of the
dimensions of the irreducible representations must equal the group order.
In order to illustrate this procedure Jet us consider the group D"' and the
corresponding double group D4. The 8 operations of are E, e_., Ct e
1
,
2C;. According to the general results given above, the 16 operations of
may be arranged into the following classes:
E R
Since there are 7 classes, there must be seven irreducible representations,
300 APPLICATIONS
and their l;, must satisfy the equaton
lj + + /j + ll + + + = 16
It s easy to convince oneself that the only combination of positive integers
satisfying this equation is 1, 1, L 1, 2, 2 .. 2. Thus there are four one-dimen-
sional and three two-dimensional irreducible representations of the double
group
Double groups are of greatest importance for transition metal complexes.
In Appendix V we give the character tables for the double groups and O'
corresponding to the simple rotation groups D
4
and O. Severa) features of
these tables should be noted. First, there are two systems for labeling the
representations. One is an adaptation of the Mulliken system for simple
groups, in which we use primed Mulliken symbols. The other is Bethe's
original system, in which we use a serially indexed set of f/s. Second, it will
be noted that among the representations of the double groups are all the
representations of the simple group. Whenever we form a representation using
a wave function having an integral value of angular momentum, /, L, S, or
J, it will either be one of these representations or it will be reducible to a
sum of only these representations. In other words, when the angular mo-
mentum quantum number used is integral, we ha ve no need of the double
group. However, when the angular momentum quantum number, s, S, or J,
is half-integral, we will obtain one of the new representations not occurring
in the simple group, or a representation that can be reduced to a sum con-
taining only these new irreducible representations. It will be noted that all
of these new representations ha ve an even order, 2, 4, and so on. Thus all
wave functions of a system must be at least twofold degenerate. This is a
manifestation of Kramer's theorem that in the absence of an externa) magnetic
field the spin degeneracy of a system having an odd num ber of electrons
must always persist even when the low symmetry of the environment lifts all
other degeneracies.
The direct products of representations of double groups can be taken in
the usual vvay and reduced to sums of irreducible representations.
In order to illustrate the utility of double groups Jet us consider several
examples. Suppose that we have an ion with one d electron in a planar
complex. Real examples of this case are represented by complexes of Cu(II)
and Ag(II) (where we virtually have one positron, but this behaves as one
electron except in the signs of the energies). In each case there will be two
states with J values l l = 2 t namely, J = i and J = i. Using 9.3-2,
we find that these form bases for the following representations:

r3,2
rs,2
E
4
6
R
-4 o o o o o
-6 -Vz v'2 o o o
These can be reduced in the standard way, giving
r3:2 = r6 + r1
rs,2 = r6 + 2r7
The procedure we have used would be particularly appropriate in the case
of Ag(II), where the two J states are already well separated in the free ion
because of a very large spin-orbit coupling. An energy level diagram of the
following sort could then be drawn:
J=l.
.---..;..2 __
1

--rs
2D l= 2
=t \ /-r7

"
2 --r
6
This illustrates how the two J states would be further split by an environment
of D
4
h symmetry.
If the spin-orbit coupling is relatively small. we might wish to consider first
the splitting of the
2
D state by the environment and then the further splitting
of the resulting states by spin-orbit coupling. To do this we first use the D
411
character table to find that the
2
D state splits into A
1
g,
and E:,. These

representations in D
4
, can easily be se en to correspond1 respectively 1 to A
1
,
B,,
and E in D
4
, ano to f,, f
3
, f
4
, and f
5
in This analysis has dealt
with the orbital part of the wave function (se e Section 9 .2, especially 9.2-1 ).
Fvr the spin part, VIs in 9.2-1, we find the representation for which the spin
angular momentum, s = i, forms a basis in the double group Using the
formulas given above1 we easily obtain
E R 2C
4
R ZC2 4C;
rl/2
2
-2. V2
-v2 o o o
Thus
rl,! = r
Now, to find the representation of a wave function that is a product of two
other functions we must obtain the representation of the direct product of
the two functions. The characters of this representation are the products of
302 APPLICATIONS
the characters of the representations of the two functions. Thus, using the
character table for D ~ we obtain the results
r, x
rb = rh
r
3
x
rh = r1
r
4
x
r6 = r7
f; X
r6 = r6 + r7
Of course the final results are the same as those previously obtained. The
manner in which we obtain them is unimportant so far as pure symmetry
considerations are concerned. Normally, however, we would choose the first
method in a case where we expected the splitting between the free-ion states
with J = and J = ~ to be greater than the further splittings caused by the
environments. and !he second method when we expected the splitting of the
Russell-Saunders term, !D, by the environment to be much larger than the
splittings resulting from spin-orbit coupling. In the latter case our energy level
diagram for one electron might look somewhat as follows:
Bia<ra>
r1
A
11
(f
1
)
rs
2 l:::;; 2
B
21
cr
4
>
TJ--
r1 s=.!.
2
r1
E
8
(r
5
)
--<
rs
In this example the relative order of the orbitals is somewhat arbitrary.
Only by solving an appropriate wave equation could the actual order be
determined in a particular case, but the result would have to correspond with
this diagram in regard to the types and the number of each type of wave
functions obtained.
EXERCISES
9.1 Use the subgroup C
2
,. in the method of descending symmetry to verify
the correlation of Figure 9.3.
9.2 Select an appropriate subgroup of Td and verify Figure 9.4 by the method
of descending symmetry.
9.3 The selection and polarization rules for magnetic dipole-allowed tran-
sitions are given by a set of integrals of the type 5.3-3 in which the
transition moment operators have the same symmetry properties as the
rotations, R.''' R=. The magnetic dipole mechanism is weaker by
several orders of magnitude than the electronic dipole one, but in cen-
trosymmetric situations, where d-d transitions are not electric dipole
allowed, magnetic dipole and vibronic mechanisms both need to be
considered to explain the weak bands. Work out the magnetic dipole
polarizations for trans-[Coen
2
Ch] +. Are they consistent with the obser-
vations or not?
9.4 How will the d orbitals split in a trigonal bipyramidal environment?
Using a crystal field (pure point charge) model, determine what the
relative energies of the orbitals will be.
9.5 If an octahedron is distorted by stretching it along one threefold axis,
it becomes a trigonal antiprism with DM symmetry. The d orbital splitting
pattern becomes (d.r::' dv:J llE
1
above d:'!, and d:'! llE
2
above (d.n,,
We now choose the z axis to coincide with the C
3
and Sn axes f
the trigonal antiprism. Assume that llE, = 31lE
2
and work out the
complete weak-field/strong-field correlation diagram.
10
MOLECULAR VIBRATIONS
10.1 INTRODUCTORY REMARKS
A molecule possesses three types of internal energy. These are, in the usual
decreasing order of their magnitudes, electronic, vibrational and rotational
energies. In preceding chapters we have dealt with the use of symmetry
properties for understanding the electronic states of various kinds of mole-
cules. The rotational energy states ha ve no symmetry properties of importance
in ordinary chemical processes and will not concern us directly in this book.
We are left, then, with the subject of molecular vibrations, to which symmetry
arguments may be very applied.
Every molecule, at all temperatures, including even the absolute zero, is
continually executing vibrational motions, that is, motions in which its dis-
tances and internal angles change periodically without producing any net
translation of the center of mass of the molecule or imparting any net ang-
ular momentum (rotatory motion) to the molecule. Of course the molecule
m ay, and if free certainly will, be traveling through space and rotating. but
we may divorce our attention from these motions by supposing that we are
seated at the center of gravity of the molecule and that we travel and
rotate with it. Then it will not appear to us to be undergoing translation
or rotation and our full attention m ay be focused on its internal or
genuine vibrations.
Although a cursory glance at a vibrating molecule might suggest that its
vibratory motion is random, close inspection and proper analysis reveal a
basic regularity and simplicity. 1 t is the. underlying basis for this simplicity
that we shall formulate in this chapter. We shall also develop working methods
304
MOLECULAR VIBRATIONS 305
by which all of the analysis of molecular n1otions \Vhich symmetry alone allows
may be rapidly and reliably performed.
10.2 THE SYMMETRY OF NORMAL VIBRATIONS
The complex, random, and seemingly aperiodic internal motions of a vibrating
molecule are the result of the superposition of a number of relatively simple
vibratory motions known as the nonnal vibrations or nornzaln1odes of vibra-
tion of the molecule. Each of these has its own fixed frequency. Naturally
then, when many of them are superposed. the resulting motion must al so be
but it may have a period so long as to be difficult to discern.
The first question to be considered regarding the normal modes is that of
their number in any given molecule. fortunately, is a very easy question1
and doubtless many readers know the answer airead y. An ato m has three
degrees of motional freedom. It may move from an initial position in the x
direction independent of any displacement that it may or may not undergo
in the y or z direction, in the y direction independent of \vhether or not it
moves in the x or z direction1 and so on. In the molecule consisting of n atoms
there will thus be 3n degrees of freedom.
Let us now suppose that aH n atoms m ove simultaneously by the san1e
amount in the x direction. This will displace the center of mass of the entire
molecule in the x direction without causing any alteration of the interna}
dimensions of the molecule. The-same may of course be said of similar motions
in the y and z directions. of the 3n degrees of freedom of the molecule,
three are not genuine vibrations but only translations. Similarly, concerted
motions of all atoms in circular paths about the x, and z axes do not
constitute vibrations either but are instead, morecular rotations. Thus, of the
3n degrees of motional freedom, only 3n - 6 remain to be combined into
genuine vibratory n1otions.
We make note he re of the special case of a linear molecuJe. In that instance
rotation of the molecule may occur about each of two axes perpendicular to
the molecular axis, but "rotationH of nuclei about the molecular axis itself
cannot occur since al1 nuclei lie on the axis. Thus an n-atomic linear molecule
has 3n - 5 normal modes.
Let us now look at the normal modes of vibration of a molecule which is
as simple as possible and yet exemplfies all general features ordinarily en-
countered. The planar ion COi- will serve for this purpose. As a nonlinear
four-atomic it must ha ve 3( 4) - 6 = 6 norn1al modes. In Figure 10.1
we have depicted these vibrations. In each drawing the Jength of an arrow
relative to the length of another arrow in the same drawing shows how much
the atom to which it is attached is displaced at any instant relative to the
simultaneous displacement of the atom to which the other arrow is attached.
306 APPLICATIONS
Figure 10.1 The six normal modes of vibration of the carbonate ion.
The lengths of the arrows relative to interatomic distances in the dra\vings,
however, are exaggerated.
As m ay be seen in the particular case of COj-, which we are using as an
illustration, normal modes ha ve two important properties:
l. Each of the vectors representing an instantaneous atomic displacement
may be regarded as the resultant of a set of three basis vectors.
2. Each of the normal m o des forms a basis for or "belongs to" an irre-
ducible representation of the molecule.
Let us first consider the ways in which we might regard the displacement
vectors in the normal modes as resultants of sorne set of basis vectors. There
are many ways in which the sel of bttsis vectors might be chosen, but only
two are of interest. In the tirst., we t t ~ c h a separate Cartesian coordinate
fviOLECULAR VIBRATIONS 307
system to each atom, with the atom at origin and all x axes, all y axes,
and all z axes parallel and pointing in the same direction (cf. Fig. 10.3). In
each small coordinate system we place unit vectors along the x, y, and z axes.
Now, the vector representing the displacement of a given atom, the ith atom,
may be expressed as the vector sum of Cartesian displacement vectors of this
atom, X;, y;, and Z;. We may call this process resolution of a general displace-
ment into Cartesian displacements. lt will be noted that the three translational
motions and the three ( or two) rotational motions m ay also be resolved into
vector sums of Cartesian displacements. Thus all 3n degrees of motional
freedom of the molecule may be represented by suitable combinations of the
3n Cartesian displacements.
The second important way of resolving the displacement vectors of the
normal modes is to use basis vectors related to the internal coordinates of
the molecule, that is, the interatomic distances and bond angles. There is no
unique way of doing this, Normally, however, we choose first the changes in
interatomic distances between bonded atoms and then as many changes in
bond angle (taking care that those chosen are all independent) as are necessary
to provide a set of 3n - 6 internal vectors. For example., in the
carbonate ion we require six interna( displacement vectors to represent the
six normal modes. We choose first changes in the three C-0 distances. Next
we may choose changes in two of the three OCO angles. Our sixth choice
might be a change in the remaining OCO angle or a change in the angle
between a C-0 bond axis and the molecular plane.
Let us consider now the second important property of the normal modes,
namely, their symmetry. It is easy to see by comparison of the diagrams in
Figure 10.1 and the character table for the group D
3
h, to which the carbonate
ion belongs, that each normal mode ( or pair of normal modes) transforms
exactly as required by the characters of the representation to which it belongs.
These representations are noted in parentheses on Figure 10.1. Clearly the
set of vectors representing v
1
is carried into itself by all operations; hence it
belongs to the A i representation. lt is equally obvious that the set of vectors
representing v
2
is carried into itself by the operations E, C:"' and a,. but into
the negative of itself by C
2
, S::" and a1r. Thus this mode belongs to the A2
representation, as stated in Figure 10.1.
The l'Ja and v
3
b modes together form the basis for the E' representation of
the group D
311
Clearly the identity operation takes each component into itself,
as required by the character of 2. We can express this symbolically as follows:
E( v
3
a) = v3a + 0l'
3
b
E( l'Jb) = 0\'3a + l'Jh
The matrix of the coefficients on the right side of this set of eQuations is
308 APPLICATIONS
It is, of course1 a two-dimensional unit matrix with the character 2. The effect
of a threefo)d rotation on either l'Ja or vJb is to take the mode into a linear
combination of both v
3
a and l'Jb Figure 10.2 illustrates this for the case of a
clockwise rotation of v
3
a by 2n/3, to give a mode which we have labeled
v ~ . The lower part of Figure 10.2 shows in detail how each of the displacement
vectors in v), is the vector su m of - !vJa and iv
3
,. Thus \Ve can write
lt can simiiarly be shown that application of a clockwise rotation by 2n/3 to
vJb would produce a mode which could be expressed as the following linear
combination of l'J., and 1'
3
b;
1 1
3 3 2
-f(uaa) + (l13b)
'
=
ll3f1
l.
1
+
1
=
1
2.
~
+
1
=
)
3. V +
<
=
~
4.
+
=
Figure 10.2 Vector diagrams showing how a threefold rotation transforms l'31r
into 1 ~ and how the latter is a linear combination of v
311
and v
3
b; specifically, 1 ~ ==
- !v:w + ~ l'Jb
Now the matrix of the coefficients of these two transformations is
and its character is - 1 as required by the character table.
It is easy to see that the operation e! transforms VJo into the negative of
itself and 1'
3
b into itself. Thus a matrix is obtained which has only the diagonal
elements - 1 and 1 and the character O as required by the character table. It
is equally easy to see that a
11
carries each component of 1
3
into itself, so that
the matrix of the transformation has onJy the diagonal elements 1 and 1 and
hence a character of 2. We could carry out similar reasoning for the remaining
operation applied to l'Ja and v
3
b and also with respect to the application of all
of the operations in the group to ''4a and v
4
b, and it would be found that they
satisfy the requirements of the characters of the E' representation in every
respect.
10.3 DETERMINING THE SYMMETRY TYPES OF THE
NORMAL MODES
The two characteristic features of normal modes of vibration that ha ve been
stated and discussed above lead directly to a simple and straightforward method
of determining how many of the normal modes of vibration of any molecule
will belong to each of the irreducible representations of the point group of
the molecule. This information may be obtained entirely from knowledge of
the molecular symmetry and does not require any or by itself
provide any information, concerning the frequencies or detailed forms of the
normal modes.
We have illustrated in detail for the case of CQj- how the normal modes
of genuine vibration have symmetry corresponding to one or another of the
irreducible representations of the molecu1e. This is true for every molecule,
though we shall not offer proof he re.* I t is al so true that the
vibrations, the translational and rotational motions, transform according to
irreducible representations of the molecular point gro u p. Moreover the en tire
set of 3n normal modes may be expressed as functions of a set of 3n Cartesian
displacements, as described in the preceding section. lt is also evident that
we may use the 3n Cartesian displacement vectors as the basis for a reducible
representation of the molecular symmetry group. This representation will
contain ( or, is sorne times said, span) the set of irreducible representations
to which the normal modes, genuine and belong.
* Rigorous proof, which involves more quantitative discussion of the mechanics of the normal
vibrations and the use of explicit expressions for the kinetic and potential energies, may he found
in more specialized texts. for example. E. B. Wilson, J. C. Decius. and P. C. Cross. Molecular
Vibrations, McGraw-HiJI. New York. 1955.
310 APPLICATIONS
We shall now illustrate this, using asan example. A number offurther
examples will be found in Section 10.7. The first step must be to determine
the symmetry group to which the molecu)e belongs, as described in Chapter
3, especially Section 3.14. We find that COj- belongs to the D
3
, group. Figure
10.3 shows the COj- ion with the sets of Cartesian displacement vectors
attached to. each ato m. There are of course 3n = 12 in all, and the repre-
sentation will therefore be of dimension 12.
We turn now to the character table for the group D
31
,. The first operation,
naturally, is the identity operation. When this is applied to the set of vectors,
each remains in place, that is, remains identical with itself. We may express
this as shown in Figure 10.4. We assume that the symmetry operation is applied
only to the set of vectors, moving them but Jeaving the nuclei themselves
fixed. Thus we may specify or label each vector befare the operation by stating
its direction and the number of the atom to which it is attached, namely, X,
or Z
4
For the same vector after the symmetry operation we use the same
symbol primed, whether the vector has moved in any way or not. The left
vertical column in Figure 10.4 lists the vectors after application of the sym-
metry operation, and the top horizontal row lists the original set. The purpose
of Figure 10.4 and a similar device for each symmetry operation is to express
the composition of the primed vectors in terms of those in the original un-
primed set. In this case the results are trivial: Each primed vector is identical
with its counterpart. The square array of numbers so obtained is
a matrix describing the effect of the symmetry operation upon the set of
vectors, and its character is the character corresponding to the particular
operation in the reducible representation that we are seeking. Thus, for the
identity operation, we have here a character of 12.
We now apply a threefold rotation to the set of Cartesian displacement
vectors with the results pictured in Figure 10.5. Again we wish to construct
the matrix expressing these results. This procedure is a trifle tedious but
requires no more than the simplest trigonometry. For example, as Figure 10.6
shows, X can be expressed as - iX
2
- (\13/2) Y
2
, and this result has been
Figure 10.3 The set of 3n = 12 Cartesian dispJacement vectors used in determining
the reducible representation spanning the irreducible representation of the normal
modes of COj-.
x.
yl
z. x2
yl
z2 xl
y3
z3 x4 y" z4
x; I o o o o o o o o o o o
Y:
o 1 o o o o o o o o o o
z; o o 1 o o o o o o o o o
X ~
o o o 1 o o o o o o o o
y ~ o o o o 1 o o o o o o o
z ~
o o o o o 1 o o o o o o
x; o o o o o o 1 o o o o o
y ~ o o o o o o o 1 o o o o
z ~
o o o o o o o o 1 o o o
X ~
o o o o o o o o o 1 o o
y;
o o o o o o o o o o 1 o
z ~
o o o o o o o o o o o 1
Figure 10.4 The matrix expressing the effect of the identity operation on the set of
Cartesian displacement coordina tes (Fig. 10.3) for CO:(!.
entered in the .first row of the matrix, Figure 10.7. The reader should have
no difficulty in verifying the other en tries in Figure 10.7. The character of
this matrix has the value zero.
Befo re proceeding further, we take note of a labor-saving procedure. The
operation C
3
shifts all of the vectors originally associated with atom 1 to atom
2. Thus, when we write down the components of X ~ Yi, and Z ~ we find
Zz
Ca
Figure 10.5 Diagrams showing the etfect of a threefold rotation on the set of Cartesan
displacement vectors.
3U APPLICATIONS
Figure 10.6 Diagram showing the resolution of the displacement vector x: into its
and y:! components.
that they come entirely from the set X"!., Y
2
, Z
2
This being so., all of the diag-
onal elements in the first three rows are zero. Since all of the vectors ini-
tially on atom 2 are rotated to atom 3, there are no nonzero diagonal elements
in the next three rows. For the same reason., there are non e in the next three
rows either. Only the vectors X"', Y
4
., and Z
4
, which the operation C
3
merely
shuffies among themselves but does not shift to a different atom.,
nonzero diagonal elements. Thus we could ha ve determined the character of
the C
3
matrix by ignoring all the vectors that are shifted from one atom to
x. Y. z. x2 y2
z2 x3 y3
ZJ x4 y4 z ...
x; o o o
-!
,.
o o o o o o o
y;
o o o

-! o o o o o o o
z; o o o o o 1 o o o o o o
o o o o o o -!
_,l
o o o o
o o o o o o
-! o o o o
z; o o o o o o o o 1 o o o
x;
-! - v'i o o o o o o o o o o
"\/:t.
..
-4
o o o o o o o o o o

o o J o o o o o o o o o
x; o o o o o o o o o -!
_,;
o
y;
-
o o o o o o o o o
v';,.
-! o

o o o o o o o o o o o 1
Figure 10.7 The matrix expressing the effect of the operation C
3
on the set of Carte-
sian displacement for
another when the mo1ecule is rotated and taking account of only those that
remain associated with the same atom. Henceforth, we shall approach the
task in this way.
When the molecule is subjected toa twofold rotation, we see that vectors
on two of the oxygen atoms, let us say numbers 2 and are shifted. Thus
we know that these vectors will contribute nothing to the character of the
matrix. For each of the other two 0
1
and C
4
, however the Z vectors
are transformed into the negatives of themselves, two Y vectors go over into
their own negatives, while the X vectors remain- unchanged. These results
are expressed in the abbreviated matrix of Figure 10.8, in which only the
elements relating to the vectors on atoms 1 and 4 are given. The value of the
character is seen to be - 2.
The character of the matrix corresponding to the operation a, may be
found without writing out any part of the complete matrix itself. We note
that the operation a, does not shift any vectors from one atom to another.
Hence no set of vectors m ay be summarily ignored. We note further that each
set of vectors will be affected by a
1
, exactly the same way. Thus whatever
contribution to the character is made by one of the four sets may simply be
multiplied by 4 in order to get the total value of the character. In any one
set, a, transforms the X and Y vectors into themselves and the Z vector into
its own negative. Thus the submatrix for this set of vectors wiiJ be diagonal
with the elements 1, 1, and. - 1 and hence a character of l. The character of
the entire matrix corresponding to the operation a, is thus 4.
The operation S
3
shifts all the vectors on oxygen atoms, so we know that
these nine vectors can be ignored. The effect of S
3
on the vectors of the carbon
atom can be found quickly by recalling the effect of C
3
on the same set, as
shown in the lower right portion of Figure 1 O. 7. Since S
3
is simply C
3
followed
by a,, its effect on X
4
and Y
4
is the same as that of C
3
However, whereas C="
left Z
4
unaffected, S
3
transforms it into its negative. Thus S
3
acting on the
displacement vectors of the carbon atom causes the following elements to
appear along the diagonal of the matrix: -!, -!, -l. The character of the
en tire matrix is therefore - 2.
Finally, we ha ve to consider the operation al .. Let us choose the plane
x. yl
z. x4
y4
z4
x: 1 o o
y;
o -1 o
z: o o -1
1 o o

o -1 o
o o -1
Figure 10.8 Abbreviated matrix for the operation C
2
on the Cartesian displacement
vectors of e
314 APPLICATIONS
passing through 0
1
and C
4
The ve<;tors on atoms 2 and 3, which are shuffted
by reftection through this plane, may be dismissed from consideration. It is
obvious by inspection that X
1
, Z
1
, X
4
, and Z.h go into themselves, whi1e Y
1
and Y
4
go into the negative of themselves by reflection of these sets of vectors
through the a,. considered. Thus the diagonal elements generated are 1, 1, 1,
1, -1, and -1, and the character of the en tire matrix corresponding to a,,
has the value 2.
In Figure 10.9 we reproduce the character table of the group D
311
and
append to it the results just obtained for the characters of the operations in
the reducible representation for which the 12 Cartesian displacement coor-
dinates form a basis. This may be reduced by the methods of Section 4.3 with
the following result:
r, = A ~ + A ~ + E ~ + 2A2 + E"
We know that of the 12 normal modes of the molecule only 6 are genuine
vibrations while 3 are translations and 3 are rotations. We can easily strike
the nongenuine ones from the above list by reference to the information on
the right side of the character table. The translatory motions must belong to
the same representations as do the coordinates x, y, and z. Thus we delete
from our list one of the E' and one of the Ai species. We see that rotation
about the z axis is a motion having ~ symmetry and that rotations about the
x and y axes are a degenera te pair having E'' symmetry. We therefore strike
the A; and the E" from our list. This then leaves the following list of the
representations to which the 6 genuine normal modes of the molecule be long:
f = A1' + 2E' + A ~
g ~
These results are, of course, in agreement with information given in Figure
10.1.
We conclude this section by summarizing the procedure developed and the
D3,. E 2C3 JC2
a,
2S3 3av
A ~ 1 1 1 1 1 1
xl + yz, zl
A ~ 1 1 -1 1 1 -1 R:
E' 2 -1 o 2 -1 o (x, y) (xl- yl, xy)
Af 1 1 1 -1 -1 -1
A; 1 1 -1 -1 -1 1 z
E" 2 -1 o -2 1 o (Rx, R,) (xz, yz)
r. 12 o -2 4 -2 2
Figure 10.9 The character table for the group D
3
h with the characters for the rep-
resentation generated from the 12 Cartesian displacement coordinates appended.
meaning of the results that it provides. In order to find out how many genuine
normal modes of vibration of a molecule there are belonging to each irre-
ducible representation of the molecular symmetry_group, we merely have to
form the reducible representation,. \ Y ~ h k h ts generated on applying fhe op-
erations of the group (actu.a.Hy j-Hst ane. o_peration from each cfass 1 to 'fhe set
of Cartesian <!!5J'lacement vectors. This gives directly the symmetry types of
both the genuine and the nongenuine modes, and from this list those of the
translations and rotations m ay be removed immediately by use of information
explicit in the character table. The process of finding the characters of the
reducible representation is greatly simplified by recognizing that none of the
vectors which are shifted to different atoms by a symmetrv operation makes
any contribution to the value of the character of the matrix corresponding to
tbis operalion. The final result is a Jist of irreducible representatons ot the
-molecular symmetry group such that the su m of their dimensions equals the
number of internal or genuine modes of vibration of the molecule. There is
a genuine normal mode having symmetry corresponding to each of the ir-
reducible representations in the list.
10.4 CONTRIBUTIONS OF PARTICULAR INTERNAL
COORDINATES TO NORMAL MODES
We noted in Section 10.3 that the normal modes could be expressed not only
as functions, that is, vector sums, of a set of Cartesian displacement vectors
but also as functions of a set of interna! displacement vectors. We have seen
how the former relationship is utilized in determining the total number of
normal modes of each symmetry type. We use the second relationship mainly
to gain information on how the stretching of bonds and the bending of bond
angles contribute to the normal modes according to their symmetry types.
For instan ce, inspection of Figure 10.1 reveals that the A i vibration in vol ves
purely the stretching of C-0 bonds while the A1 vibration involves only
bending or deformation of the molecule out of the equilibrium plane. The
E' vibrations, however, are clearly not of this Hpurity"; both involve a mixture
of C-0 stretching and in-plane deformation of the OCO angles. We may
ask: Could these facts ha ve been deduced from symmetry considerations alone
without knowledge of the actual forms of the normal modes? The answet to
this question is in the affirmative, and it is our purpose in this section to
explain how such information is obtained.
Suppose that we choose as a set of three internal displacement coordina tes
stretchings of the three C-0 bonds, and use these as the basis for a three-
dimensional representation of the symmetry group. The irreducible repre-
sentations spanned by this representation will include only those to which
belong normal modes involving C-0 stretching. From Figure 10.1, we an-
ticpate that the answer will be A; + E'. Let us carry through the"procedure
outlined and see whether we do obtain this representation. In so doing we
316 APPLICATIONS
shall make use of the short cut developed above in handling Cartesian dis-
placement vectors. Any vector that is shifted to a different position by a given
symmetry operation will contribute nothing to the value of the character of
the corresponding matrix.
The various symmetry operations will affect our set of C-0 stretchings
in the same way as they will affect the set of e-O bonds themselves. With
this in mind we can determine the desired characters very quickly as follows.
For the operation E the character equals 3, since each e-O bond is carried
into itself. The same is true for the operation a
1
,. For the operations C
3
and
s3 the characters are zero because all e-o bonds are shifted by these op-
erations. The operations C
1
and a,. ha ve characters of 1, sin ce each carries
one e-o bond into itself but interchanges the other two. The set of char-
acters, listed in the same order as are the symmetry operations at the top of
the D
3
, character table, is thus as follows: 3 O 1 3 O l. The represen-
tation reduces toA i + E' .. as it should according to our previous discussion.
Thus we ha ve shown that normal modes of symmetry types A ( and E' must
in vol ve sorne degree of C-0 stretching. Sin ce there is only one A mode,
we can state further that this mode must involve entirely e-o stretching.
As a second set of internal displacement coordinates we may choose the
increments or decrements to the three OCO angles. Before using this set to
form a representation which will tell us the normal coordinates involving in-
plane OCO bends, we must be careful to note that all of the coordinates in
the set are not independent. If all three of the angles were to increase by the
same amount at the same time. the motion would ha ve A i symmetry. It is
obviously impossible, however, for all three angles simultaneously to expand
within the plane. Thus the A ( representation which we shall find when we
have reduced the representation is to be discarded as spurious.
This problem of spurious or, as they are conventionally called, redundan!
coordinates always arises when there are sets of angles that form a closed
group, as in the case just considered, in planar cyclic molecules and also in
three dimensions (e.g., tetrahedral and octahedral molecules). Several of the
examples discussed in Section 10.7 will illustrate the point further. Redundant
coordinates can usually be recognized without much difficulty, though trou-
blesome cases sometimes arise.
A set of three increases in the three OCO angles will be affected by the
symmetry operations in the same way as the three angles themselves, therefore
we may make use of the angles in determining the characters. The operation
E transforms each angle into itself, giving a character of 3 for this operation.
So also does the operation ah. Since the operations C
3
and S
3
shift all of the
angles, these operations have characters of O. The operations C
2
and al. each
leave one angle unshifted but interchange the other two, thus having char-
acters of l. The complete set of characters, again in the order in which the
symmetry operations are listed, is as follows: 3 O 1 3 O l. This rep-
resentation reduces toA + E'. Djsreg_arc!_ing the ~ for the reason explained
abo ve, we ha ve the result that in-plane bending of the OCO angles contributes
tviOLECULAR VIBRATIONS 317
to vibrations of E' symmetry. This result is seen to be in agreement with the
actual nature of the normal modes as shown in Figure 10.1.
Sin ce there remains only one other kind of internal coordina te, namely,
changes in the angles between the C-0 bonds and the plane of the ion, and
only one normal mode is unaccounted for, we may naturally conclude that
this remaining mode of ~ symmetry. must consist entirely of out-of-plane
bending of the ion, which is, of course, correct. However, we may bring more
positive reasoning to bear on the problem. lt is evident that normal modes
involving C-0 stretching, OCO in-plane angle bending, or both must by
nature be symmetric with respect to reflection through the plane of the mol-
ecule. All vectors used to indicate the displacements in such vibrations will
lie in the plane and cannot be affected by reflection through the plane in
which they lie. Since a mode of A! symmetry must be antisymmetric to ah.
according to the character table, we might have seen that the ~ mode would
not involve C-0 stretching or in-p1ane OCO bending without having any
knowledge as to which vibrations did involve such internal displacements.
Moreover, it is obvious that only a vibration in which all displacements are
perpendicular to the molecular plane can belong to the ~ representation.
10.5 HOW TO CALCULATE FORCE CONSTANTS: THE F ANO
G MATRIX METHOD
Qualitative ways of analyzing a problem in molecular vibrations, that is,
methods for determining the number of normal modes of each symmetry type
which wi11 arise in the molecule as a whole and in each set of equivalent
internal coordinates, have been developed. There is also the quantitative
problem of how the frequencies of these vibrations, which can be obtained
by experiment. are related to the masses of the atoms, the bond angles and
bond lengths, and most particularly the force constants of the individual bonds
and interbond angles. In this section we shall show how to set up the equations
which express these relationships, making maximum use of symmetry to sim-
plify the task at every stage.
For this purpose we shall adopt, without proof of its validity. Wilson 's
method of F and G matrices.* All of the required relations are combined in
the master equation
IFG - E).l =o
(10.5-1)
in which F, G, andE are matrices and the entire left-hand side of the equation
is a determinan t. F is a matrix of force constants and thus brings the potential
For a full exposition of this method see. E. B. Wilson, Jr., J. C. Decius. and P. C. Cross,
Molecular Vibrations, McGraw-Hill, New Y o r k ~ 1955.
318 APPLICATIONS
energies of the vibrations into the equation. G is a matrix that involves the
masses and certain spatial relationships of the atoms and thus brings the kinetic
energies into the equation. E is a unit matix, and J., which brings the frequency
v into the equation, is defined by
(10.5-2)
If the atomic masses in the G matrix elements are expressed in atomic mass
units rather than in grams and the frequencies are expressed. in reciprocal
centimeters (cm-
1
), 10.5-2 may be \vritten
The plausibility of 10.5-1 may be appreciated by comparing it Yth the
equation obtained by treating a diatomic molecule AB as a harmonic oscil-
lator, namely,
(10.5-3)
in which )_ is defined as above, f is the force constant, and 11 is the reduced
mass [i.e., MAM
8
/(MA + M
8
)]. Equation 10.5-3 is evidently the limiting case
of the F-G matrix equation when the F, G, and E matrices are one dimen-
sional.
Once the elements of the F and G n1atrices are known for the molecule
in question, the determinantal equation (10.5-1) may be written out explicitly.
This is the secular equation for the vibrational problem, analogous to the
secular equation for energy, which has airead y been examined in connection
with LCAO-MO theory in Chapter 7. Using known frequencies, we may
apply the secular equation to calculate force constants, or, utilizing force
constants, to calculate the frequencies of the normal m o des. However, for a
nonlinear molecule of N atoms this equation will involve a 3N x 3N deter-
minant, equivalent to a polynomial of order 3N of which only six roots will
equal zero. It is then obviously advantageous to be able to factor this equation
so that several smaller determinantal equations, rather than one large one,
are to be solved. The great virtue of the F and G matrix method is that it
affords a convenient and systematic means of employing the symmetry prop-
erties of the molecules to achieve maximum factorization of the secular equa-
tion.
Symmetry Coordinates
In order to use the F-G matrix method to maximum advantage, it is first
necessary to set up linear combinations of internal coordina tes which provide
the proper number of functions, transforming according to each of the irre-
ducible representations spanned by the 3N - 6 normal modes of genuine
vibration. Todo this, the methods of Sections 10.3 and 10.4 are used to obtain
the fulllist of irreducible representations involved and the internal coordina tes
which con tribute to each. Then, using the sets of interna) coordina tes as basis
functions, we set up the SALCs employing the projection operator technique
developed in Chapter 6. The SALCs should be normalized.
The details of the F-G matrix procedure are best explained by working
through a simple example, such as the water molecule. This belongs to the
point group C
2
, The nine Cartesian displacement vectors, three on each atom,
give rise to the representation
E C
2
a,,(xz) a,.(yz)
r 9 -1 3 1
where we have placed the molecule in the xz plane. This reduces to
From the character table we can tell at once that translations and rotations
account for A
1
+ A
2
+ 28
1
+ 28
2
Thus, the three genuine internal vibrations
span the irreducible representations 2A
1
+ 8
1

Displacements of the interna] coordinates of the water molecule, the two
0-H distances (d
1
and d
2
), and the HOH angle (.6.0) are bases for the
following irreducible representations:
tld, tld
2
: A t + 81
tlO: A1
Writing symmetry coordinates as SALCs of the intemal coordinates is very
easy in this simple case, where tlO itself is an A
1
symmetry coordinate. Ap-
plying projection operators to a member of the set tld, tld
2
, we obtain
PA tld1 = (1)E lldt + 1 ) ~ tld1 + (1)a,. tld1 + {I)a,. tld1
= tld, + tld2 + tldl + .6.d2
= lld, + tld2
P
8
1 tld
1
= (1) tld, + ( -1)(2 tld
1
+ (l)a-,. tld
1
+ ( -1)a-,. tld
1
= tldl - .6.d1 + tldl - tld1
= tldl - tld2
The normalizing factor for each of these, assuming tld
1
and tld
2
to constitute
an orthogonal set (lid; tld = ;), is 1/\12. Thus, the complete set of
320 APPLICA TI ONS
symmetry coordinates for vibrations is
(10.5-4)
The F Matrix
We as sume that the nuclei vibra te harmonically. Thus the poten ti al energy
V of the molecule may be written
2 V = : /;kS;Sk
i.k
(10.5-5)
where S; and s" are changes in internal coordinates, /;k - f ki and the sum
extends over all values of i and k. A term such as /;;SJ represents the potential
energy of stretching a given bond or bending a given angle, while the cross
terms represent the energies of interaction between such motions. The f;t's
are called force constants.
For H
2
0 there are three intemal coordinates and hence nine force con-
stants. A systematic way to list these is to make a square arra y, where the
rows and columns are labeled by the internal displacements. Thus we have
the f matrix, a matrix of the f;k,s; note that this is symmetrical about its
diagonal because f;k = fki For H
2
0 the f matrix is, explicitly,
fl.d 1 f d f dd fcm
fl.d2 fdd fd fdo
!lO f JtJ f dJI ffl
where the force constant for stretching an 0-H bond is f J, the one for
bending the HOH angle is f
0
, and the constants for interaction of one bond
stretch with the other and with angle bending are fdd and fd
0
, respectively.
This is equivalent to writing the appropriate form of 10.5-5 as
2V = fd(lldt)
2
+ fd(l1d2)
2
+ fo(b.0)
2
+ 2fdd(lld, lld2)
+ 2fdtJ(fldJ !lO) + 2fclo(!ld2 f10)
(10.5-6)
It is also possible to express the potential energy in terms of symmetry
coordinates. We write 10.5-7 instead of 10.5-5, namely,
2 v = : Fj,sjs,
j,l
(10.5-7)
Here the F
1
are again force constants but pertan to vibrations described by
the symmetry coordinates S, S, and so on. From the standpoint of physical
insight, it is the fil.: that have meaning for us, whereas mathematically the F
1
and the associated symmetry coordinates provide the easiest route to calcu-
lations beca use of symmetry factorization of the secular equation. Clearly, if
we could express the in terms of the .fil, 's we would ha ve an optimum
situation. The following considerations will show how to do this.
Each of the two equations for the potential energy, 10.5-5 and 10.5-7, can
be expressed in matrix notation:
2V = s'fs
2V = S'FS
(10.5-6a)
(10.5-7a)
by writing the s;'s as a column matrix (vector) s, and the S/s as a column
111atrix S, and taking s' and S' as the corresponding row matrices. To illustrate,
for H
2
0 we have
(8d1 8d2 80) [fd !Jc1 !Jo] [8d
1
]
fdd fd fdo ild2
f JO f Jo f o 8o
= (8d, 8d2 80) [fd 8dl + .fdd lldz + !do 80]
6.d1 + fd 8d2 + fdo !lO
!Jo 8d1 + fdo 8d1 + .fo 80
= 8d) + !JJ(6.d
2
8d) + fdo(80 8d,) + !Jd(8d
1
8d
2
) +
+ . . . .f o( 80 80)
= fJ(!ld)l +

+ fo(80)
2
+ 2fdd(8d
1
8d2) + 2f,J0(8d
1
80)
+ !lO)
which is equiva]ent, as shown, to 10.5-6.
It is also possible to write the relationship between the internal coordinates
and the symmetry coordinates, 10.5-4, in matrix form:
S= Us (10.5-8)
where the matrix U is
322 APPLICATIONS
The inverse of the matrix U is simply its transpose, U', beca use U describes
a linear. orthogonal transformation. Thus, 10.5-8 may be rewritten as
s = = U'S
(10.5-Sa)
and we have
s' = (U'S)' = S'U
(10.5-9)
since for matrices, as for group operations in general (page 7), the inverse
( or transpose) of a product is the product of the in verses ( or transposes) in
re verse order.
We may now equate the right-hand sides of and 10.5-7a and employ
relations 10.5-8 and 10.5-9:
s'fs = S'FS
(S'U) f(U'S) = S'FS
S'(U fU')S = S'FS
U fU'= F
-.._____
We thus obtain a simple matrix equation for transforming the f matrix into
the F matrix. Let us apply this to the water molecule.
F = UfU'

o
1] [!J
!JJ !Jt] [o 11V2
1/4
= 1/Vl
O fdcl fd fdtl o 1/Vl
-16V2
1/Vz -1/Vz
o .fdtl f,m f, 1 O

o
1] [!Jt/ l!Vz(fd + !JJ) l!Vl(fd - !JJ)]
=
1/Vz
o .fJ/1 1/Vl( f,fd + .fJ)
l!Vz(f(/ - !J)
1/Vz -1/Vz O fo 2/Vz(fdtJ)
[ !ti
Vz,,ll
fd /J
= '1JJo
fd + fdd
o
We see that the F matrix is symmetry-factored into a 2 x 2 block for the
two A
1
vibrations and a one-dimensional block for the single B
1
vibration and
that each element of it is expressed as a linear combination of the various
interna( force constants, fJ, fr, fdch feto
The G Matrix
The G matrix may be set up by a procedure formally analogous to that used
for the F matrix. Thus, in general we have
G = UgU'
(10.5-10)
For the water molecule the explicit equation is
where the g matrix elements are given general symboJs except that 8
21
is
replaced by g
12
, g
31
by 8
13
, and g
32
by 8n because of the general requirement
that the n1atrix be symmetrical about its diagonal. We may also make the
substitutions
beca use interna} coordinates 1 and 1 are equivalent (Ad
1
, ~ d
2
. With these
further substitutions, the result of matrix multiplication is
[
833 v'2813 o ]
G = v'28r3 811 + 812 O
o o 811 - 812
lt now remains to consider how the various 8;i terms are expressed with
respect to the atomic masses and the dimensions of the molecule. The pro-
cedure, although not difficult or profound, s definitely tedious; it is explained
[n Wilson, Decius, and Cross and other specialized textbooks. Fortunately,
it is possible to tabulate the most commonly used ones in the form of general
expressions into which the specific parameters of any molecule may be in-
serted. Such a tabulation and directions for its use are given in Appendix VI.
lt is found that
8u = JlH + Jlo
812 = Jlo cosO
813 = - (.tolr) sin O
833 = 2(JLH + Jlo - Jlo cos O) 1 r
2
where JlH = the reciproca! of the mass of the H atom, and so on.
The G matrix, which is symmetry-factored in conformity with the F matrix,
then becomes
[
2{)lH + Jlo - f1o cos O)lr
2
- (v'iL
0
1r) sin O O ]
- (v'ip
0
/ r) sin O P.H + .t
0
( 1 -+- cos O) O
O O JIH + JLo(1 - cos O)
., When the molecular structure parameters are known, or estimated, this
matrix may be reduced toa set of numbers. Thus, for H
2
0, using O = .10431'
324 APPLICATIONS
and r = O. 9580 , we obtain
[
2.332
G = -o.gs93
-0.0893 o ]
1.0390 o
o 1.0702
The problem of expressing. for the H
1
0 molecule. the relationship between
the frequencies of the fundamental modes and a set of force constants has
now been solved in such a \\'ay that the equations are as simple as symmetry
will permit them to be. In this case, the secular equation is factored into one
of second order for the two A
1
vibrations and one of first order for the sol e
B
1
vibration. The explicit forms of these separate equations are
A
. [ [o
l \/2f
dfl
v2J,w J [ 2.332
.fd + .fttd -0.0893
- o.o893] _ [;_ o] =
0
1.0390 o )_
This is the end of the symmetry analysis; from he re on the problem is a purely
computational one.
The computational problem is not a proper subject for discussion here,
but a few remarks about it may be pertinent. In practice one is nearly always
interested in computing force constants from observed frequencies. However,
analytical expressions for force constants as explicit functions of the frequen-
cies and geometric parameters cannot generally be obtained. The on1y efficient
way to calculate force constants from frequencies is to use an iterative ap-
proach, implemented by a digital computer. To do this a starting set of as-
sumed fore constants is refined by successive approximations until the set
which yields calculated frequencies in best agreement with the observed ones
is obtained.
An additional problem in computing force constants is that usually-and
H
2
0 as treated here is a case in point-a force field that includes the principal
interaction constants, as well as the stretching and bending constants, wiH
have more constants to evaJuate than there are fundamental frequencies. The
most generally applicable procedure for coping with this problem is to measure
frequencies in isotopically substituted molecules, thereby providing new sets
of equations involving different frequencies and sorne different G matrix ele-
ments, but the same force constants. For light molecules, which can be ob-
served in the gas phase, centrifugal distortion effects in rotational fine structure
afford another source of data on force constants.
10.6 SELECTION RULES FOR FUNDAMENTAL
VIBRATIONAL TRANSITIONS
The wave functions of normal modes can be written in the simplest possible
way by using the normal coordinates as the variables. As is shown in many
books on elementary quantum mechanics, these wave functions, assuming
that the oscillations are harmonic (potential energy a function only of the
square of the displacement coordina te), ha ve the form
(10.6-1)
where N; is a normalizing constant, a; = 2nv;l h (in which V; is the frequency
of the ith norma) mode), Hn is a Hermite polynomial of order n, and ~ ; is
the ith normal coordinate expressed as a SALC of interna) dispJacement
coordinates. The number n is the vibrational quantum number; it is O in the
ground state, 1 in the first excited state, and so forth.
The first few Hermite polynomials, Hn(x), are
H
0
(.r) = 1 H
1
(x) = 2x
(10.6-2)
H'!(x) = 4x
2
- 2 H3(x) = 8x
3
- 12x
As shown in Section 5.1, the wave functions must form bases for irreducible
representations of the symmetry group of the molecule, and the same holds,
of course, for all kinds of wave functions, vibrational, rotational, electronic,
and so on. Let us now see what representations are generated by the vibra-
tional wave functions of the normal modes. Inserting H
0
(Va,.c;;) into 10.6-1,
we obtain
In the event that ~ ; represents a nondegenerate vibration .. all symmetry op-
erations change it into 1 times itself. Hence ~ T is unchanged by all symmetry
operations. Thus f//;(0) is invariant under all symmetry operations and forros
a basis for the totally symmetric representation of the group. If <!; is a normal
coordinate describing one of a set of degenerate vibrations, any symmetry
operation will change it into + 1 times itself or into a linear combination of
all members of the set (cf. Section 10.2). Let us suppose that ~ a and ~ h are
the normal coordinates of a pair of degenerate vibrations and that sorne
symmetry operation, R, acting on ~ a has the effect:
lf a <!b are normalized, ra and rh will be such that ~ ~ will also be normalized.
We then have, because normal coordinates are orthogonal and normalized ..
.. ., 1
<;; = '
Thus, in the degenerate case also, f//;(0) is invariant to al1 symmetry opera-
tions.
326 APPLICATIONS
We may therefore state the following important rule:
Al! wave functions for nonnal vibrations in their ground states, tJ;(O), are
bases for the total/y synzmetric representation of the point group of the 1nolecule.
For the excited vibrational states, r;(n), the wave functions are products
of the same exponential as in the ground state, which we ha ve se en is always
totally symmetric, and the nth Hermite polynomial. Therefore 'IJ;(n) has the
symmetry of the n th Hermite polynomial.
For the first excited state, we then find that the wave function, lf/;( 1), has
the same symmetry as <!;: for the second excited state, f//;(2) has the symmetry
of which means that it is totally symmetric, and so on.
Consider a molecule with k normal modes of vibration. At any time each
of these modes will be in a certain quantum the wave function for the
ith mode in the nth state is r;(n;). For the total molecular vibrational wave
function, r,., we m ay use the product of the r;(n;), sin ce the variables in each,
the are linearly independent. Thus we write
(10.6-3)
When each of the n; is equal to O, the molecule is in its vibrational ground
state. If it absorbs radiation so that the ith normal mode is excited to the
state with n; = 1, while the remaining k - 1 normal modes remain in their
lowest (n = O) states, the molecule is said to ha ve undergone a Junda1nenta/
transition in the ith normal vibration. The k different transitions of this kind
are ca1led the fundamentals of the molecule.
Sin ce the fundamental transitions gene rally give rise to IR absorption bands
and Raman lines which are more intense by at least an order of magnitude
than any other kinds* of transition. they are of the greatest interest and we
shall deal only with the fundamentals here. Selection rules for other types of
transition can also be obtained by arguments of the type we shall use, but
the reader is cautioned that where degenerate modes are concerned subtle
complications often arise. t
For a fundamental transition of the jth normal mode we may write
rr r;(O) w;(1) n rJ;(o)
1
* If one mode alone is doubly excited, this transition is calJed the first overrone of the fundamental;
if one mode alone is triply excted, this is called the second Ol'ertone. and so on. Transitions in
which two or more modes are simultaneously excited (by. one or more units each) are called
combinotion tones. If the molecule initially has one or more fundamentals in an excited (n >O)
state, ony transitions that it makes are called hot transitions.
t For details regarding the selection rules for other types of transitions, Molecular Vibrations_.
by Wilson. Decius. and Cross. should be consulted. especially pages 151-155, where comph-
cations arising because of dcgeneracy are treared.
VIBRATIONS 327
Abbreviating the total molecular vibrational wave functions in the ground
and excited states by and f/1{, respectively, we can also write
For a fundamental transition to occur by absorption of infrared dipole
radiation (see Section 5.3) it is necessary that one or more of the integrals
J (. dr
f
() . d
lfl,.y '11 {. r
be nonzero. The x, y, and z in these integrals refer to the orientation of the
oscillating electric vector of the radiation relative to a Cartesian coordinate
system fixed in the molecule.
Now, by using the results of Section 5.3, it is extremely easy to determine
whether or not such integrals vanish. Sin ce belongs to the totally symmetric
representation, the coordinate x, y, or z and lf/{. must belong to the same
representation in order that the representation of their direct product will
contain the totally symmetric representation, whereby the representation given
by the entire integral may contain the totally symmetric representatioq. Since
'11;(1) must ha ve the same symmetry as the jth normal coordina te, and
since all of the other wave functions, f/1;(0), are totally symmetric, '11 {. belongs
to the same representation as the normal mode which is undergoing its fun-
damental transition.
We therefore have the following very simple rule for the activity of fun-
damentals in infrared absorption:
A fundamental will be infrared active (i.e., will give rise to an absorption
band) if the normal nzode which is excited belongs to the same representation
as any one or severa/ of the Cartesian coordinates.
Of course the character tables show to which representations the Cartesian
coordinates belong, so that this rule can be used with the utmost ease.
For Raman scattering, it is necessary that at least one integral of the type
J rp?Prp {. dr
be nonzero. In these integrals P is one of the quadratic functions of the
Cartesian coordinates, namely, x
2
., y
2
., z
2
, xy, zx, all of which (simply or
in combinations such as x
2
- y:!) are Jisted opposite the representarions that
they generare in the character tables. These P's are components of the po-
328 APPLICATIONS
larizability tensor, and the requirement that the abo ve integrals be nonzero
means physically that there must be a change in polarizability of the molecule
when the transition occurs. By exactly the same reasoning as used above in
regard to infrared absorption, we obtain the following simple rule for the
Raman activity of fundamentals:
A[undan1ental transition will be Raman active (i.e., will give rise toa Ran1an
shift) if the nonnal mode in volved belongs to the same representation as one
or more o[ the components o[ the polarizability tensor o[ the molecule.
We may illustrate these rules, using the carbonate ion. We see in the D
3
h
character table that (x, y) form a basis for the E' representation and z for
the A ~ representation. For the polarizability tensor components we see that
one or more of these be long to the ~ E', and E" representations. Thus. for
any molecule of D
31
, symmetry, we ha ve the following selection rules:
Raman-active only: A , E"
Infrared-active only: A1
Both Raman- and infrared-active: E'
In the particular case of the carbonate ion:
v
1
(A i): Raman only
v
2
(A2): Infrared only
l'
3
(E'), v
4
(E'): Infrared and Raman
10.7 ILLUSTRATIVE EXAMPLES
The Pyramidal AB
3
Molecule
This type of molecule will be used to illustrate all of the procedures and
techniques discussed earlier in this chapter, including setting up the vibrational
secular equation by the F-G matrix method.
Symmetry Types of the Normal Modes. For this nonlinear four-atomic
molecule there are 3(4) - 6 = 6 genuine interna[ vibrations. Using a set of
three Cartesian displacement coordinates on each atom, we obtain the fol-
lowing representation of the group C
3
,.:
r 12 o 2
which can be reduced, giving
r = 3A + A2 + 4E
The C
3
~ character table shows that the translations account for A
1
+ E and
the rotations for A
2
+ E, leaving, for the genuine, internal vibrations,
2A
1
+ 2E
Selection Rules. The character table shows that A
1
and E vibrations are
both IR and Raman active. Thus, all four fundamental modes for such a
molecule should be observable in both the IR and the Raman spectra.
lntemal Coordinate Contributio11s. The nature of the vibrations in terms
of changes in internal coordinates may be found by using the three A-B
bond lengths and the three ABA angles as internal coordinates. In this case
no spurious or redundant results will be obtained because all of these six
intemal coordina tes may change independently. The representation f.4
8
, given
by the bond lengths, and f
15
, given by the bond angles, are readily found to
be the following:
fAB 3 0 1
rd 3 o 1
The two representations are identical; each reduces to A
1
+ E. Therefore it
folJows that each of the A
1
normal modes will involve both bond stretching
and deformations of the bond angles, and so also will the E modes.
F and G Matrices. The internal coordinates are shown in the sketch below.
The SALCs, which constitute symmetry coordinates, have already been de-
rived for an AB
3
molecule in Section 6.3 and are reproduced here.
330 APPLICATIONS
1
S3a = v6 (2Ar
1
- Ar
2
- Ar
3
)
1
s3b = v2 -
E
1
S4a = \16 (2 1101 - 1102 - 1103)
1
s4b = v2 - 11o3)
1t should be noted that the displacement coordinates have been labeled and
then combined into the E-type symmetry coordina tes in a particular way. The
angles have been indexed so that O; s opposite to r;. This assures that O; and
the change therein, 110,., are related to the molecular symmetry elements in
the same way as are r; and Ar,.. Then, when the SALCs are written, 110; and
flr; occupy positions in the expressions. U nless this is done the
symmetry factorization will not work out.
We can now write the U matrix. In doing so, the corresponding types of
E symmetry coordinates, S
3
a, S4a and S
3
b, S
4
b, are placed in successive rows.
The U matrix is as follows:
'
r2 r3 o. 02 03
SI
l/v'3 1/v'3 l/v'3 o o o
Sz
o o o l/v'3 l/v'3 1/v'3
s3a
21\16 -1/v6 -l/v6 o o o
s4a o o o 21\16 -1/\16 -l/v6
s3b
o 1/\12 -1/\12 o o o
s4b
o o o o 1/\12 -1/\12
The f matrix that will be adopted in this illustration is the following.
11r
1
11r
2
11r
3
fl01
1103
11r
1
.f, frr' frr'
o
!rO frtl
11r
2 J, .fr ", f,o o
frtl
11r
3 fn-' f"t
f, f rfl frtl o
&o. o
!rO .f riJ
fo
o o
1102 frll
o
!rO
o ffl o
1103 fro
frfl o o o ft,
It may be noted that a complete f matrix (that is, one containing no zeros)
would ha ve six different force constants: f, f,, f,, f rt, f rll', and foo'. The last
two, pertaining to the interaction between llr; and llfJ (interaction between
stretching a bond and deforming an angle to which that bond does not be long)
and the interaction between two angle deformations, respectively, ha ve be en
omitted. Since there are only four fundamental frequencies, it would not be
possible to solve for al1 six force constants by using the data for only a single
isotopic species. In this case, there are two force constants which seem very
likely to be smaller and less important than the others, and these have been
omitted from the outset, thus bringing the number of force constants carried
into line with the number of experimental data likely to be readily available.
Using the above U and f matrices, we find that
fr + 2frr' 2.f rll
o o o o
2frO fo
o o o o
..
F= unr-
o o f, - f,' - frtl o o
o o - frl) fo o o
o o o o fr - frr' - .f rll
o o o o - .f r(J
ftJ
The first block, at the upper left, is for the two A
1
modes, and the two
remaining blocks, which are identical, are for the a and b components of the
E modes. Thus, the factorization is into two second-order equations, as ex-
pected. lf the interna! coordinates and the symmetry coordinates (SALCs)
had not been selected and arranged so that the 11r; and 110; sets matched
correctly, as specified earlier, the factorization of the 4 x 4 E block into two
identical 2 x 2 blocks would not have been accomplished.
Finally, the G matrix must be set u p. Using the notation defined in Figure
AVI.l of the appendix and omitting elements related by the symmetry about
the diagonal to those given we write the g matrix as follows.
r
r2 r3
o,
02 03
rs
.,
g;,
.,
g;{l

')

.,

.,
,2 g;,. g;(/ g;(J
2
.,
g:.0(2) rJ g, g;,,
o.

gJO'( 1)
02
3
glltl
03
3

;From this we obtain the equation on page 332.
C
.H

C
.H

N

G

=

U
g
U
'
-
g
;
,

+

2
g
;
r
'

+

2
g
;
,
,

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

g
2

-
g
l
,

r
r

r
r

g
:
o
(
2
)

-

o

o

o

o

o

o

g
2

-
g
l
,

r
r

r
r

g
;
{
/
(
2
)

-
2
g
;
o

o

o

-
2
g
;
t
l

-
2
g
7
1
{
1
1
(
1
)

The F and G matrices may now be combined to give two two-dimensional
equations, one for the A
1
modes and one for the E modes. To illustrate what
these modes actually Iook like in a real case, they are depicted for ND
3
in
Figure 10.10. These drawings are based on calculations (cf. Wilson, Decius,
and Cross for details) from experimentally observed frequencies. The mol-
ecule ND
3
is used rather than NH
3
because the inverse mass dependence of
the amplitudes would make the vectors on the nitrogen atom of NH
3
im-
practically small compared to those on the hydrogen atoms.
lt will be seen that one each of the A
1
and E modes involves abnost purely
bond stretching, while each of the others involves almost. purely angle bending.
Such a situation is not uncommon and makes it easy to interpret qualitatively
the shifts in observed frequencies when molecules or parts of molecules (e.g.,
CF
3
or CH
3
groups) change their chemical environment. lt must be empha-
sized, however, that the occurrence of su eh convenient separations is a con-
sequence of the particular values of the masses and force constants, and no
such separation into "stretching modes" and ubending modes" is in general
required. Quite often the normal modes involve complex mixtures of several
ntemal coordinates; in that case," naive discussions in terms of pure Hstretches'"
and "bends', can be quite misleading.
E
E.
1'4b
Mainly N-O stretching
Mainly DND bending
Figure 10.10 The normal vibrations of ND
3
(after Herzberg).
334 APPLICATIONS
Trans-N
2
F2
This molecule has the following planar, nonlinear structure:
F

1
N=N
1 .
F
It belongs to the point group C2h. Since it is a nonlinear four-atomic molecule
it has 3(4) - 6 = 6 degrees of internal freedom.
The set of 12 Cartesian displacement vectors for the entire molecule gen-
erates the following reducible representation:
r 12 o o 4
This can be reduced in the following manner:
r = 4A.r: + 2B
8
+ 2Au + 4B,
Inspection of the character table shows that the translations and rotations span
the representations Ag + B ~ : + A, + 2B,. On deleting these from the total
number constituting f, we are left with the following list of the representations
spanned by the genuine normal vibrations:
The selection rules for the fundamentals of these modes are obtained
immediately from the right columns of the character table and are
IR active: Au, B,
Raman active: Ag
The nature of these six vibrations may be further specified in terms of the
contribution made to each of them by the various interna) coordinates. We
first note that A.r: and B, vibrations must involve only motions within the
molecular plane, since the characters of the representations A
8
and 8, with
respect to ah are positive. The A,, vibration will, however, involve out-of-
plane deformation, since the character of A, with respect to a, is negative.
Thus we may describe the normal mode of A
11
symmetry as uthe out-of-plane
de{ormation.'' In order to treat the remaining five in-plane vibrations we need
a set of five internal coordinates so chosen that changes in them may occur
entirely within the molecular plane. A suitable set, related to the bonding in
the molecule, consists of the two N-F distances, the two NNF angles, and
the N=N distance.
It is found that the two N-F distances form the basis for the representation
rNf, the two NNF angles for r NNf, and the N=N distan ce for r NN, ea eh of
which is given below:
c2h
E c2
i
fNF
2 o o
fNNF
2 o o
fNN
1 1 1
It is then easily shown that
r NF = Ag + B,
rNNF = AR + B,4
ah
2
2
1
It therefore follows that the three Raman-active vibrations (AR) will be com-
pounded of symmetric N-F stretching, symmetric NNF bending, and N=N
stretching, the relative amounts of each involved in each normal mode de-
pending, of course, on the actual values of the force constants and atomic
masses. Similarly, the two B, vibrations in the IR will involve asymmetric
N-F stretching and NNF angle bending. Again the proportion of each of
these in each of the true normal modes will depend on force constants and
atomic masses.
Tetrahedral Molecules, Such as Methane
Such moJecuJes hav7 Td symmetry. As nonlinear, five-atomic species they
have 3(5) - 6 = 9 degrees of internal freedom.
The set of 12 Cartesian displacement vectors forms a basis for the following
representation:
r 15 o -1 -1 3
This reduces as follows:
r = A t + E + T, + 3 T1
The character tabJe shows that the rotations transform as T
1
and the trans-
336 APPLICATIONS
lations as T
2
; hence the symmetry types of the genuine vibrations are
The character table also shows that the activities of these fundan1entals are
as follows:
Raman only: A
1
and E
IR and Raman: T
2
To find the contributions of the interna( coordinates, C-H bond lengths
and HCH angles, to these vibrational modes we first use the set of four C-H
bond (engthS 3S the basis for a representation, Obtaining r CI-J shown below.
fcH 4 1 O O 2
fHCH 6 0 2 0 2
This is easily found to reduce as follows:
fcH =Al + T-_
Similarly, we can use the six in ter bond angles to genera te a representation,
Obtaining f HCH' Which reduces thus:
rHcH = A, + E + T'2
It will be seen that the total dimensionality of these two representations,
ten, is one in excess of the correct number, and specifically, that there is an
extra A
1
representation. It is easy to determine that the spurious or redundant
representation S the One in r HCH' for, although t S. possibJe for all fOUf
of the C-H distances to change independently. it is not possible for all
six angles to change independently. If any five are arbitrarily altered, the
alteration of the sixth one is then automatically fixed. For an A
1
vibration
all six angles would have to change in the same way at the same time
(i.e., all increase or alJ decrease ), and this is clearly impossible. Hence we
obtain the results that the A
1
vibration of CH_. consists purely of C-H
stretching. and the E vibration purely of HCH angle deformations, while
both bond stretching and angle bending contribute to each of the normal
vibrations of T"!. symmetry.
Octahedral Molecules, Su eh as SF
6
Such n1olecules belong to the point group 0
1
,. They have 3(7) - 6 = 15
degrees of interna( freedom. The 21 Cartesian displacen1ent vectors generate
the representation r given in the table below:
o, E
8C3
6C"'
= 1 65_. ssh
..,
6rT,
r 21 o -1 3 -3 -3 -1 o 5 3
fsF
6 o o 2 2 o o o 4
')
fFSF
12 o 2 o o o o o 4 2
The representation f reduces as follows:
r = + + + 3T,, + T'Y.t: + T'Y.II
The character table shows that the_ rotations and translations be respec-
tive( y. to the T,J: and T
111
representations. After deleting these, we obtain the
following list of genuine normal modes, grouped according to the activities
of their fundamentals:
IR-active: 2 T,"
Raman-active: A l..:
lnactive:
We encounter here for the first time the occurrence of a normal vibration
which is completely nactive as a fundamental. This phenomenon is not com-
monplace but is encountered occasionally in relatively symmetrical molecules.
In the table above we also give the representations generated by the set
of six S-F bonds, rSF and the set of 12 FSF angles, rFSF These represen-
tations can be reduced as follows:
fsF = 4
1
,a: + + T
11
,
r FSF = + EJ: + T11: + T,, +
Dbviously there s redundancy since the S-F coordinates are
WholJy independent, the redundancy ffiUSt be entirely in r f-"SF (see the pre-
ceding discussion of CH.-). By comparing the total of rsF and r FSF with the
correct Jist of genuine internal modes \Ve see that the A Jg and occurring
in r FsF are the spurious ones. Thus we conclude that each of the two T,,
modes wiH in'volve a combination of bond stretching and angle deformation,
the A
1
..: and modes will involve only bond stretching, and the
and T'!,
modes will involve only angle deformation.
338 APPLICATIONS
10.8 SOME IMPORTANT SPECIAL EFFECTS
The Exclusion Rule
Considera molecule that has a center of symmetry. When one of the Cartesian
coordina tes, x, y. or z, is in verted through this center, it goes into the negative
of itself. Hence all representations generated by x. y, or z or any set of these
must belong to a u representation. On the other hand, a binary product of
two Cartesian coordina tes, say xy or z
2
, does not change sign on inversion,
since each coordinate separately does change sign and -1 x - 1 = l. lt
therefore follows that all such binary products, which represent components
of the polarizability tensor, belong to g representations.
From these rules, which can be verified by inspection of the character
tables, we conclude that in centrosymmetric molecules only fundamentaJs of
modes belonging to g representations can be Raman active and only funda-
mentals of modes belonging to u representations can be infrared-active. 1t is
also obvious that the same must be true for other transitions besides funda-
mentals, since the reasoning is completely general.
Another way <;>f stating this result. the so-called exclusion rule, is as follows:
In a centrosym1netric n1olecule no Ranu1n-active vibration is also infrared-
active and 110 infrared-active vibration is also Ran1an actil'e.
The reader may refer to the preceding section to see this rule exemplified
for trans-N
1
F
1
and SFh.
Fermi Resonance
The vibrational secular equation as it is normally set up ( e.g., by the F-G
matrix method) deals only with the fundamental modes of vibration. The n
fundamental modes with a given symmetry are governed by the appropriate
11 x n dimensional block factors of the F and G matrices. When 11 2 the
frequencies of the various modes are mutually interdependent to a degree
determined by the off-diagonal matrix elements. The usual secular equation
does not take account of overtones or combinations and hence neglects any
influence these may have on the fundamentals. In most cases this approxi-
mation is very satisfactory, but there are instances in which a combination or
overtone interacts strongly with a fundamental. This may happen \vhen the
two excitations give states of the same symmetry. The phenomenon of in-
teraction is called Fenni resonance because it was first recognized and ex-
plained by Enrico Fermi in the vibrational spectrum of carbon dioxide.
Fermi resonance is only a special example of the general phenomenon
caJJed configuration interaction, which has been discussed earlier in connec-
tion with electronc spectra (page 179). In the latter case each o{ two ditferent
excited electronic conflgurations gave rise to states which then mixed and
interacted with each other to give the two actual states, each of which in volved
contributions from both electron configurations. the energies of the two
actual states were further separated than the energies calculated for each of
the pure configurations. We will see now that exactly the same type of in-
teraction is involved in Fermi resonance.
Let us suppose that the frequencies, 1';, 1'i, and of three fundamental
vibrational transtions are related as follows:
V + 1' = \'k
and the symmetry of the double excited state
'1'; = f/1 ;( 1) f/1 ( 1) TI "'(O)
/-;i.j
is the same as that of the singly excited state
'l'k = f//k{l) I1 f//r(O)
/y!. k
This will be true when the direct product representation for the ith and jth
normal modes is or contains the irreducible representation for which the kth
normal mode forms a basis. These two excited states of the same symmetry
will then interact in a way \Vhich can be represented by the usual sort of
secular equation, namely,
( 1' + 1') - V W;.k
W;i.k l'k - "
=o
where the roots, 1'., will be the actual frequencies. The magnitude of the
interactions is given by
W;.k = f '1'; W'l'k dr
The interaction operator W has a nonzero value because the vibrations are
anharmonic. It is also totally symmetric, and for this reason the two states,
'JI; and 'l'b must belong to the same representation in order for the integral
to have a finite value.
1t is clear that one of the roots of the secular equation wilJ be greater than
either ( 1'; + v) or vk and that the other will be less than either l'; + v;). or
''k This is an obvious consequence of the nature of the quadratiC equatJon
in determinantal form. The c/oser (v t IJ) and lk are lo begin
WJth, the more the actual roots will diverge from these values. Thus, one
340 APPLICATIONS
effect of Fenni resonance is to displace the two expected frequencies a\vay
from each other.
Fermi resonance also has an effect on the intensities of the two transitions.
Overtone or combination transitions generally have less than one tenth the
intensity of fundamental transitions. However, the interaction between ':V;
1
and 'V b which affects the actual energies, al so leads to new wave functions
which are linear combinations of these two. as explained for secular equations
gene rally in Section 7 .l. Thus, the actual excited state whose energy is closest
to ( v; + v) will be described by a wave function
where N is a normalizing factor and x :s; 1, while the actual excited state
whose energy is closest to \'k will be described by a wave function
Therefore, e ven though a transition from the ground state, f//u to f/1;, has an
inherenily low intensity. proportional to the small integral
J 'l'u(x, y, z )'1' ; dr
the transition to the actual excited state, '1!;. has an intensity proportional to
J '1'
11
(x, y, z )'1';
1
dr + x J '1
1
11
(x. y, z )'1'
1
dr
where the second integral must be Jarge, beca use the 'l'u '1' k transition is
intense. The weak overtone transition "borrows'' intensity from the strong
fundamental transition because it is close to it and has an excited state with
the same symmetry. The closer are the energies ( v; + 1') and ''k. the more
the total intensity tends to be shared equally between them. The situation is
illustrated schematically in Figure 10.11.
In the actual case of the C0
1
molecule. which provides an excellent ex-
ample, the three fundamental transitions ha ve frequencies of 667, 1300, and
2350 cm -t. The first overtone of the 667-cm _, vibration, which is doubly
degenera te, has a frequency of 1334 cm-', which is quite close to that of the
1300-cm _, fundamental. Now it can be sho\vn that the excited state for the
1300-cm - fundamental and one component of the representation generated
by the excited state corresponding to the first overtone of the 667-cm -l vi-
bration do belong to the same representation of the group D-x
11
, and hence a
Fermi resonance occurs. Thus, in the Raman spectrum of C0
1
two strong
bands at 1285 and 1388 cm-
1
are observed, instead of just one at -1300 cm-.
+
(a)
(h)
Figure 10.11 (a) The spectrum expected for the cqmbination ( l', + 1'
1
) and the fun-
damental ''A, in the absence of Fermi resonance. (b) The actual spectrum where the
two bands have diverged and shared intensity as a result of Fermi resonance.
This example also serves to emphasize that Fermi resonance occurs in
Raman as wen as IR spectra, although, for the sake of definiteness, the
immediately preceding intensity integrals were written as appropriate for
electric dipolar activity in the IR. Fermi resonance has been unambiguously
recognized in various other th<?roughly analyzed molecules. such as cct_..
C
6
H
6
, and CHCI
3
Undoubtedly, it occurs frequently in very complex mole-
cules. where there is considerable chance of a near coincidence of one of the
many combinations of lower frequency modes with one of the higher fre-
quency fundamentals of appropriate symmetry. Nevertheless. it should be
noted that papers reporting hasty and superficial vibrational analyses of com-
plex molecules display a deplorable tendency to en1ploy uFermi
as an escape from difficulties or inconsistencies without adequate evidence.
Solid-State Effects
The vi.bratiqns of an individual molecule in the gas phase are subject only to
the symmetry restrictions on its own intrinsic point symmetry, and this
ch:Ipter has so far been concerned exclusively with syn1metry conditions of
that kind. When a molecule resides in a crystal it is, in principie, subject only
to the symmetry restrictions arising out of its crystalline environment. To be
entirely rigorous, the molecule cannot even be treated as a discrete entity:
instead the en tire arra y of molecules must be analyzed. However, su eh a
completely rigorous approach is essentially impossible for practical reasons
and unnecessary for most purposes, and therefore approximations are justi-
fiably made. Two Ievels of approximation have frequently been used: (1) the
342 APPLICATIONS
site symmetry approximation, and (2) the correlation field (sometimes called
ufactor group'') approximation. The first is conceptually very simple and verv
often is entirely adequate. Sometimes, however, it fails. and then the o r ~
abstruse correlation field treatment must be employed. Each of these ap-
proximations will now be explained and illustrated.
Site Symmetry .4pproximation. This approximation assumes that the de-
tailed dynamical nature of the interactions between one molecule and the
molecules surrounding it may be ignored. The surroundings are treated as a
static environment whose only relevant property is its symmetry. as seen by .
the molecule occupying a site within it.
In general a crystal lattice has various elements of symmetry, as will be
explained in detail in Chapter 11. Sorne of these are of the same kind as those
occurring in molecules, that is. they generate symmetry operations which
cause no translation of the center of mass of the object t> \Vhich they are
applied. Thus a given point in a crystal may be a center of inversion, it may
lie on a proper or improper axis. or it may lie in a reflection plane. Of course
severa) such symmetry elements may intersect at one point in the crystal
lattice. The collection of all the operations made possible by the symmetry
elements intersecting at such a point in the crystal constitute a point group
which can be used to specify the site sy1nnzetry of a molecule whose center
of mass is situated at this point. In a rigid, ordered crystal, the site symmetry
can never be such as to contain symmetry elements not belonging to the free
molecule, * and in general the si te symmetry is lower than the molecular
symmetry. In other words, of a number of symmetry elements present in the
free molecule, only sorne will correspond with symmetry elements of the
lattice. The symmetry elements of the free molecule. \Vhich do not correspond
with symmetry elements inherent in its crystal environment, are not part of
the si te symmetry. In principie, the molecule loses these symmetry elements
when it enters the crystal, and its vibrational behavior must be treated only
in terms of the si te symmetry. In general, then, the point group for the si te
symmetry will be a subgroup of the molecular point group, though occasionally
the two groups are the same [ e.g., for the (SiF
6
f- ion in the cubic form of
K
2
SiF].
In many cases molecules in a crystal occupy what is called a general position
in the unit cell. This is a position through which no symmetry element passes.
* There are sorne apparent exceptions. In the ammonium halides. which have the rock salt
lattice. the ammonium ions. which do not themselves ha ve centers of symmetry. lie at lattice
points which would. were the ammonium ions not present, be centers of inversion. Depending
on the particular halide and the temperature, either the ammonium ions are freely rotating. so
that the time-averaged symmetry is centric, or el se ammonium ions at different si tes ha ve different
orientations so that. although any particular site is not centrosymmetric, the crystal as a whole
appears to be centrosymmetric. In this brief discussion we shafl consider only cases in which the
crystals are completely ordered and contain no rotating molecules.
. The site symmetry for a molecule on a general position is C
1
i . e . ~ nil), and
in principie al/ selection rules and degeneracies for the molecule under its
intrinsic symmetry are lost.
It should be noted that the degree to which the vibrational spectrum of a
molecule in a site of low symmetry will deviate observably from the behavior
of the free molecule depends on just how strongly the mo)ecule interacts with
it surroundings in the crystal. Symmetry considerations alone can of course
tell us nothing about this, and the degree of deviation vares from one case
to another. A few examples now to be considered will give sorne indication
of the magnitude of such effects in these typical cases.
Qualitatively, the effects of low si te symmetry are of two general types:
(1) changes in selection rules and (2) splitting of degeneracies. A nonde-
generate vibration may be inactive in the high symmetry of the free molecule
but active in the symmetry of one or more subgroups of the same molecule.
For example, the A; mode of the carbonate ion ( totally symmetric C-0
stretching, Fig. 10.1) is not infrared-active under the full D
3
, symmetry of
COj-. The compound CaC0
3
occurs in two crystallographically ditferent forms"
cal cite and aragonite. In the former, the si te symmetry of the COj- ion is
D
3
, in the latter, Cs. Reference to the character tables for these two groups
show that in D
3
the total1y symmetric mode is still not infrared-active, whereas
in the group Cs totally symmetric vibrations are infrared-active. In agreement
with these expectations, the symmetric C-0 stretching mode of COj- (known
from the Raman spectrum of solutions of carbonates) is not observed in calcite
but appears weakly in aragonite.
The effect of low site symmetry in splitting degeneracies is also nicely
demonstrated in the forms of CaC0
3
The E' representation of the group D
311
correlates with the E representation in D
3
Hence, in caJcite, both ,,
3
and 1'-t
(Figure 10.1) are observed as single peaks. In the group C.'t there are no
representations of order greater than 1; this means that the degenerate vi-
brations of COj- must be split by the Cs site symmetry of aragonite. Actually
1
3
s still observed as a single peak, indicating that the magnitude of the
splitting is too small to permit resolution or that one component has very low
intensity, but v . ~ is distinctly split into two peaks separated by 14 cm -l.
As another example of the effect of low site symmetry in splitting degen-
eracy we may consider the thiocyanate ion in KNCS. The SeN- ion is linear
and in the absence of perturbing influences has as one of its normal modes
a doubly degenerate bending vibration. The degeneracy exists in the isolated
ion because bending of the molecuJe in any given plane containing the mo-
lecular axis is entirely equivalent to bending in a plane perpendicular to the
first one. However, if the site symmetry of the ion in a crystal is such that
these two planes are not equivalent, then the frequencies of bending in the
two planes need not be identical. In Figure 10.12, which shows the structure
of KNCS, it can be seen that planes parallel to the plane of the drawing are
not equivalent to planes perpendicular to the drawing. lt was found that,
while the bending vibration gives rise to a single peak at 470 cm-
1
in the
344 APPLICATIONS
Figure 10.12 Sketch of the unit cell of KNCS.
infrared spectrum of NCS- in solution, crystalline KNCS has two absorption
peaks at 470 and 484 cm_,.
The Corre/ation Field Approximation. In sorne cases it is not possible to
explain experimental observations in terms of the site symmetry approxi-
whereby the surroundings of a given molecule are treated as static.
A clear example is provided by the crystalline form of the trans isomer of
[(C
5
H
5
)Fe(COhh, which has the centrosymmetric structure and the IR spec-
trum shown in Figure 10.13. The trans molecule (other isomers exist) has
inherent C'1
11
symmetry when rotational orientation of the C
5
H
5
rings about

80
lj': 60
c

.E 40
Vl
e:
ro
....
t-
20
o
1950 1850
Frequency. cm -I
1750
Figure 10.13 The structure and infrared spectrum of trans-[C
5
H
5
Fe(COhh (The two
weaker bands are CO satellites.)
their C
5
axes is ignored. With this centric form of symmetry it turns out
straightforwardly that the only infrared-active C-0 stretching modes are
those arising from antisymn1etrically coupled vibrations of the terminal pair
and the bridged pair.
In the crystalline form of
each molecule lies on a

crystallographic center of symmetry; the site syn1metry is C,. In this site
symmetry it should stll be true that vibrations of the molecule can be rig-
orously classified as symmetric and antisymmetric with respect to inversion
and that only the antisymmetric ones can be infrared-active. Therefore, ac-
cording to the si te symmetry approximation, only the two antisymmetric C-0
stretching vibrations should be observed in the crystal spectrum. As Figure
10.13 shows, however, four C-0 stretching bands are actually observed.
This can be explained by treating the entire set of molecules in the unit
cell (two, in this case) as the vibrating unit, subject to symn1etry restrictions
arising from the symmetry the entire unit cell. Thus, we have a total of
eight oscillators, four bridging and four terminal, arranged in four pairs. The
members of each pair are related by a center of syn1metry that les between
the molecules. Each pair of centrically related oscillators gives rise to a sym-
metrically and an antisymmetrically coupled mode, and the antisymmetric
ones are infrared-active. This is illustrated schematically for the bridging CO
groups in Figure 10.14. Within each molecule there are symmetric and anti-
O=CC=O
O=C C=O
l',,s :>-::::
Figure 10.14 Upper: two pairs of CO groups centrically related both within each pair
and between the pairs. e denotes center of inversion. Lower: symmetric and anti-
symmetric modes of vibration within each pair.
346 APPLICATIONS
symn1etric m o des, vs, vas, v;, v ~ These can then couple inte1molecularly as
folJows:
1' + v'}
s + s, Symmetric
l'as l'as
1' - ,,'}
' J Antisymmetric
l' - t'
as as
This way of expressing the overall modes for the pair of molecular units is
only approximate, and it assun1es that intramolecular coupling exceeds in-
termolecular coupling. The frequency difference between the two antisym-
metric modes arising in the pair of molecules jointly will depend on both the
intra- and intermolecular interaction force constants. Obviously the algebraic
details are a bit complicated, but the idea of intermolecular coupling subject
to the syn1metry restrictions based on the symmetry of the entire unit cell is
a simple and powerful one. It is this symmetry-restricted intermolecular cor-
relation of the molecular vibrational modes which causes the correlation field
splittings.
The site symmetry method is most likely to fail in just such a case as we
have been discussing, where the crystal contains neutral molecules that lie
close to each other and which also contain highly dipole-active oscillators
(CO groups are among the most strongly dipole active of all oscillators). In
the several CaC0
3
crystals and other ionic con1pounds, the polyatomic ions
are separated from each other by intervening counter ions; since the corre-
latan field effects depend on dipole-dipole interactions, which vary as ,-\
where r is the separation of the dipoles, it is reasonable that effects which
amount to -10 cm -l for adjacent molecules are negligible when the separation
is 2 to 3 times as great: 10 cm-
1
x 2.5 -J = 10/15 cm -l = O. 7 cm-
1
! Correlation
field effects have been generally observed in crystalline metal carbonyl com-
pounds. *
The preceding discussion of correlation field effects is mathen1atically very
superficial. The reader interested in further details should consult the paper
by Vetter and Hornig cited in Appendix IX.
EXERCISES
10.1 Set up the symmetry coordina tes for the five in-plane vibrations of the
trans-N
1
F
2
molecule and construct the F matrix. Include force constants
for N-F and N-N stretching (fNF and !NN) N-F/N-F interaction
(fNF.NF), and N-FIN-N in.teraction (fNr.NN), bending (Jo) and bend-
bend interaction (fm,).
10.2 Benzene is a good example of an important molecule \Vith reJatively
high symmetry (D
61
J. (a) Determine the symmetry types ofthe genuine
* See. for example. papers by H. J. Buttery et al.. J. Chem. Soc. (A), 1970, 471.
vibrations. (b) Determine the activities of each type of vibration. (e)
Classify aH the in-plane vibrations as to \vhether they involve C-H
stretching, C-C stretching, or angle bending. ( d) What vibrations
involve out-of-plane C-H wagging? (e) Derivethe normalized expres-
sions for the symmetry coordinates of the C-H stretching modes.
10.3 Octahedral metal hexacarbonyls can be substituted by other ligands in
a number of ways, to give MLn(C0)
6
_, n1olecules. These have char-
acteristic IR spectra in the CO stretching region. The frequencies of
the observed CO stretching modes can be related to one another by a
simple force field model (Cotton-Kraihanzel) in whch only the C-0
stretching constants (a different one for each type of CO) and one
interaction constant, k,, between cis-CO groups are used. It is assumed
that the interaction between a pair of trans-CO groups is given by 2k;.
It is also assumed that mechanical coupling is negligible so that the G
matrix is diagonal with the inverse reduced n1ass of CO, Jl, for each
of its elements. With these assumptions, derive the secular equations
for CO vibrations and determine their activities (IR, Raman, in active)
for each of he following species:
M(C0)
6
, ML(C0)
5
,
1ner-ML
3
(CO)J,
and fac-ML
3
(C0)
3
10.4 The ion can serve as a ligand. When it does so. the effective
symmetry is lowered and this changes the number of S-0 stretching
n1odes and their activities. For each of the following, give the symmetry,
number of S-0 stretches, and their activities:
M-O-SO
3
M-O
M-0
'
/S0
2
11
CRYSTALLOGRAPHIC SYMMETRY
11.1 INTRODUCTION
All of the other chapters in this book deal with the symmetries of finite
(di serete) objects. We now turn to the symmetry properties of infinite arrays.
The end use for the concepts to be developed here is in understanding the
rules governing the structures of crystalline solids. While an individual crystal
is obviously not infinite, the atoms, ions, or n1olecules within it arrange them-
selves as though they were part of an infinite arra y. Only at, or very el ose
to, the surface is this not the case; this "surface effect" does not, in practice,
diminish the utility of the theory to be developed.
Our ultirnate concem will, of course, be with three-dimensional (3D) ar-
rays, but it is very useful to consider first the properties of one-dimensional
(ID) and two-dimensional (20) arrays. Practically all concepts of importance
pertaining to 3D arrays can be illustrated by using lD and 2D arrays and the
exposition can be done more simply.
The type of array we are concerned with, be it in one, two, or three
dimensions, is obtained by the repetition of sorne object or unit in a regular
way throughout space of that dimensionality. Let us take, to begin with, as
our object or unit something that has no symmetry of its own: This is called
the .nzotif ( or, in crystallography, the asymnzetric unit). We repeat it by a
process of translation, namely, by shifting it a defined distan ce in a certain
direction, and then doing this again and again by the same distance, in the
same direction. The distance is called the unit translation. If there is only one
such direction, the result will be a ID arra y. When there are two such di-
rections, ea eh with its own unit translation, we ha ve a 2D arra y. ~ t us now
examine lD and 2D arrays and their properties systematically and in detail.
348
CRYSTALLOGRAPHIC SYMMETRY 349
One-Dimensional Symmetries. These will not detain us long, but they do
merit examination beca use they introduce in the simplest possible way several
new concepts. There are se ven types of lD symmetry. and they are shown
in Figure 11.1, where a triangle is used as the motif. The simplest symmetry,
class 1, is that having only the translation operation. The unir translation is
the distance from any given point on one triangle to the identical point on
the nearest triangle.
If we introduce a longitudinal reflection operation (i.e., the line of trans-
iation is also a mirror line) we get class 2.
Class 3 is obtained by introducing a twofold axis of rotation, symbolized
by 1 below the motif on the line of translation. The important thing to note
here is that in addition to the c2 operation explicitly introduced (and all those
just like it obtained by unit translation) a second set of C
2
operations. with
axes halfway between those in the first set is created. In space symmetry (even
in ID space) the introduction of one set of (equivalent) symmetry elements
con1monly creates another set, whi<;h are not equivalent to those in the first
set. It should also be noted that had we chosen to introduce explicitly the
,1
,.... ,....
---- ---- --t-t----t-3
....... ....... ..... .....
Figure 11.1 The seven classes of one-dimensional symmetry. [Adapted from 1. Har-
gittai and G. Lengyel, J. Chem. Educ., 1984,61, 1033.]
350 APPLICATIONS
second set of C
2
axes, the first set would automatically ha ve arisen. The
relationship is a reciprocal one.
CJass 4 is obtained by introducing a transverse mirror line. Not only is this
line reproduced by the translation but a second set of transverse
mirror lines is created. This is similar to what occurred in class 3. The second
set is not equivalent to the first, but is brought into existence by it.
Class 5 arises by a natural extension of what has be en done to make classes
2, 3, and 4. If both types of mirror line are introduced simultaneously, we
obtain this class. Justas in point groups, the intersection between two mirrors
generates a c2 axis and this class therefore also contains two sets of c2 axes.
We could also have generated class 5 by introducing the C
1
axes and only one
type of mirror line; the other type would then have arisen as the product of
these. The relationships are exactly as in the case of point group C
2
,
With class 6 \Ve encounter a new type of translational symmetry, the glide
reflection. This is a con1bination of reflection and translation by one half the
unit translation. The glide line is represented by a broken ( dashed) line,
whereas the mirror is represented by a solid line.
Finally, in class 7 we ha ve four types of symn1etry operation: ( 1) simple
(unit) translation; (2) transverse reflection; (3) hvofold rotation; and (4) glide
reflections. As in class 5, not all of these symmetry are indepen-
dent. If we begin with class 1 and introduce explicitly only the glide reflection
and one transverse reflection, all the other operations will arise as products
of these. Again, this is analogous to the way point groups behave.
11.2 THE CONCEPT OF A LATIICE-IN TWO DIMENSIONS
Befare we proceed toa discussion of the 20 symmetry classes, we must deal
with sorne of the simpler consequences of having two unit translations. Con-
sider the portian of an infinite 20 array shown in Figure 11.2a. The whole
thing is built up by taking one of the triangular objects-the asynunetric unit-
and repeating it along ea eh of two directions by the unit_ translation pertinent
to that direction. What are these directions and the pertinent distances of
translation? The question has no unique answer. To understand this, Jet us
get rid of the repeated object itself and just mark the set of positions where
it is found, using a dot for each one. This is shown in Figure 11.2b. This array
of dots is called a lattice. *
Shown in Figure ll.2c are several pairs of vectors, each pair providing a
way to genera te the en tire lattice from one point. Clearly, there is an infinite
nun1ber of other ways to choose such pairs of vectors. Each pair defines two
directions and the pertinent unit translations. For practical reasons the pre-
The word lattice is one that suffers terrible abuse at the hands of chemists (and other scientists)
writing about crystals. A Jattice is not-a physical thing; it is simply an abstraction. a collection
of points whereon real objects may be placed. A term like .. lattice water" is as absurd as it is
common.
--
.....
--
(a)

(b)
b
a/./ :, /l
1 / 1
--- --- / . .,.,,, / /
// / / /
............. /

(e)
Figure 11.2. (a) A regular two-dimensional array of objects. (b) The lattice cor-
responding to this array. (e) Any pair of noncolinear translation vectors can be used
to generate the lattice from one point.
ferred choice is usually the two shortest vectors; in the example shown these
are labeled a and b and the angle between. them is called y.
The _yectors a and b can be considered as two edges of a parallogram called
a unit cell. The lattice can also be thought of as an infinite array of unit
all neatly fitted together so as to leave no gaps.
To summarize, a lattice is an infinite arra y of identical points (i.e. each
one has exactly the same environment of other points) and the points are
obtained one from another by translations only (no rotations or reflections
being required).
The /Vumber of Distinct 2D Lattices. We shall now show that there are
only five different types of lattice in a plane. The one we have already con-
352 APPLICATIONS
sidered is the most general one in the sense that no special relationships are
demanded of the defining vectors. The ratio of a to b may have any value
and the angle j' can be chosen arbitrarily. This called an oblique lattice,
is shown in Figure 11.3a. By drawing lines between the points along the
directions of minimum repeat the array of unit cells is defined.
(a) (d}

(b) (e)
{e)
{/)
Figure 11.3. The five distinct plane (20) lattices (a) oblique, (b) primitive rectan-
gular, (e) square, (d) and (e) are both centered rectangular but show alternative
choices of unit cell, (f) hexagonal.
If we introduce one restriction. namely }' = 90 .. we get a rectangular
lattice. This is the second of the five types, Figure 11.3b. For a reason to be
made clear shortly we describe this more precisely as a prinzitive rectangular
lattice.
If we not onJy restrict J' to be 90 but require that a = we get a square
lattice, Figure ll.3c.
Suppose we require that a = b, but leave the value of j' arbitrary. We then
get the type of lattice shown in Figure 11.3d. It is clear that all the points on
this lattice are identical (or equivalent), so that it is indubitably a mathe-
matically proper lattice. However, it is a lattice that can be looked at in a
different way and there are advantages to this different point of view. This
is shown in Figure 11.3e. It is se en that we now ha ve a rectangular unit cell.
and are thus able to benefit from the trigonometric advantages of dealing
with right angles .. whose sine and cosine are simply 1 and O. However. \Ve
also have a lattice point in the center of the cell. We therefore call this the
centered rectangular lattice, to distinguish it from the primitive rectangular
lattice previously considered, Figure 11.3b.
These terms., primitive and centered, are widely used in naming 3D lattices
as well. By a primitive lattice has ponts only at the corners of the
cell, whereas a centered cell has a lattice point at the center of the cell ( or.
in 30, at the centers of sorne faces of the cell. as we shall see later). The
important idea here is that the point at the center is not different from the
corner points, even though at first glance it might seem to be. lf it were, the
lattice would not in fact be a properly defined lattice. Note that in the centered
rectangular lattice we could have used the centering points to define the unit
cell edges, thus putting the other points into the centers of the new cells. But
the arrangement would have been identical.
Another idea to be clear about is that a centered cell or centered lattice
need not necessarily be used. An equivalent primitive lattice can always be
chosen, as illustrated by the relationship between parts (d) and (e) in Figure
11.3.
We have now delineated four lattices by a systematic process of using
various restraints and combinations of restraints on the defining vectors and
the angles between them. It might appear that all possibilities should have
been covered, but in there is one more .. namely, the special case of
Figure 11.3d where }' = 60 or 120 ( which are just two different orientations
of the same thing). This fifth case s shown in Figure 11.3f and it can be seen
that the reason it differs from all the others is that it has sixfold rotational
symmetry about each lattice point. It is called a hexagonallattice.
In addition to the primitive rhombic unit cell shown, it is easily seen that
a centered hexagonal cell can also be eh osen. This might at times be considered
advantageous because of the higher symmetry it displays, as compared with
the primitive rhombic cell. Note also that we could treat the hexagonallattice
as though it were a. centered rectangular lattice. However .. there would nor-
mally be no reason to do this as well as one very good reason why not to do
354 APPLICATIONS
this, namely., that we should thereby obscure rather than illuminate the high-
est (i.e., hexagonal} symmetry of this lattice.
In considering the hexagonallattice, our attention is strongly drawn to the
question of the symmetry of lattices. lt is a question that must eventually be
addressed in more detail for this as well as the other four plane lattices and
we shall do so shortly. However, we shall first de al with a geometrical aspect
of plan e lattices that hinges on just one of their possible symmetry properties,
namely, rotational symmetry. When we ha ve done this it will be clear why
the five lattices just described are the only ones possible. We shall understand
why it is that we need not look for sorne special value of }' that would allow
for fivefold or sevenfold, eightfold, and so on rotational symmetry.
Rotational S_vmmetry of 2D Lattices. Each of the five lattices has rota-
tional symmetry about axes perpendicular to the plane of the lattice. For the
oblique lattice and both the primitive and centered rectangular lattices these
are twofold axes, but there are several types in each case. The standard symbol
for a twofold rotation axis perpendicular to the plane of projection is t . In
the case of the square lattice there are fourfold as well as twofold axes. The
symbol for a fourfold axis seen end-on is . For the hexagonal lattice there
are two-, and sixfold the latter two are represented by ., and
, respectively. In Figure 11.4 are shown all of the rotation axes possessed
by each lattice.
In Figure 11.4 we also give the symbols used to specify the symmetries of
these lattices. This type of notation will be fully explained in Section 11.4,
but we can point out here that a rotation axis of order n is represented simply
by the number n and mirror lines (or planes) by 111. In addition, p and e
specify primitive and centered lattices, respectively.
We shall now pro ve that the orders of these axes, namely, two- to four-
and sixfold are the only ones possible for a lattice. Consider two adjacent
points, P and Q., of a lattice (Figure 11.5} and assume that there is an n-fold
rotation axis through each one. Rotation by 2n/ 11 about Q \vill genera te P'
from P and, similarly, rotation by - 2nln about P will generate Q' from Q.
Now in order that the 2n/ 11 rotation operations be val id symmetry opera-
tions, Q' and P' must coincide with other lattice points. For this to happen,
it must be true that the distance between them, /, is equal to, or an integral
multiple o, the lattice spacing a between P and Q. It is easily se en that the
distance 1 is given by
1 = a - (2a cos 2n/n) = a(l - 2 cos 2n/n)
This means that only those val u es of 1 - 2 cos 2nl n that are integers are
allowed. For what angles, 2n/n, is the value of the cosine equal toO, !, or
an integral multiple thereof. Clearly, only those listed in Table 11.1 ha ve the
appropriate values. We therefore conclude that the only permissible axes of
rotational symmetry for a lattice or for any array of real objects distributed

r

L

_
j

r

L

7

p
2

'

.
.

L

1

_
j

p
4
(
m
m
2
)

1

r

1

1

"

_
j

n
.

L

_
j

p
r
n
m
(
2
)

c
m
m
(
2
}

r

7

L

_
_

p
6
(
m
r
n
3
2
)

F
i
g
u
r
e

1
1
.
4
.

T
h
e

s
y
m
m
e
t
r
y

e
l
e
m
e
n
t
s

o
f

t
h
e

f
i
v
e

2
D

l
a
t
t
i
c
e
s
.

F
o
r

e
a
c
h

p
a
i
r

t
h
e

l
a
t
t
i
c
e

i
s

r
c
p
r
e
s
e
n
t
e
d

o
n

t
h
e

J
e
f
t

b
y

t
h
e

p
o
i
n
t
s

d
e
f
i
n
i
n
g

o
n
e

u
n
i
t

c
e
l
l

a
n
d

o
n

t
h
e

r
i
g
h
t

a
r
e

t
h
e

s
y
m
m
e
t
r
y

e
J
e
m
e
n
t
s
.

D
i
f
f
e
r
e
n
t

o
r
i
e
n
t
a
t
i
o
n
s

o
f

t
h
e

s
y
m
b
o
l
s

a
r
e

u
s
e
d

t
o

d
i
f
f
e
r
e
n
t
i
a
t
e

n
o
n
e
q
u
i
v
a
l
e
n
t

a
x
e
s

o
f

t
h
e

s
a
m
e

o
r
t
l
e
r
.

S
y
m
m
e
t
r
y

s
y
m
b
o
l
s

i
n

p
a
r
e
n
t
h
e
s
e
s

a
r
e

r
e
d
u
n
d
a
n
t
,

t
h
a
t

i
s
.

i
n
d
i
c
a
t
e

t
h
o
s
c

t
h
a
t

a
r
i
s
c

a
u
t
o
m
a
t
i
c
a
l
l
y

f
r
o
r
n

t
h
o
s
e

p
r
e
c
e
d
i
n
g

t
h
e
m
.

356 APPLICATIONS
Q' P'

1 \
1 \
1 \
1 \
1 \
1 \
1 \
/ \
1 \
1 \
1 \
/'\
1 - 21rln 2rln \

p Q
Figure 11.5. A geometrical construction used to show how rotation axes in a lattice
are limited to those with orders 1, 2, 3, 4, and 6.
on a lattice are C
1
, C
2
, C
3
, C
4
, and C
6
A C
5
axis and all higher ones produce
results incompatible with the defining properties of a lattice.
Another way to appreciate this point is to ask what kinds of polygons can
be used to create a continuous 20 array of identical polygons that has no
gaps. In everyday terms, it is the question of how many ways there are to
tile a floor with regular polygonal tiles that are all identical. Clearly rhom-
buses, rectangles, squares, and triangles may be used; these have, respec-
tively, C
2
, C
2
, C.h and C
3
axes. Hexagons, with C
6
axes, m ay also be used. 1t
is not at all difficult to convince oneself by tri al and error that it is impossible
to tile a floor exclusively with identical tiles the shape of regular pentagons,
heptagons, or any higher regular polygon.
Additional Comments on Centering. In working out systematically the
existence of just five distinct planar lattices, we arrived at the centered rec-
tangular lattice by regarding it as a centered alternative to a primitive, non-
TABLE 11.1
Angle
60 = 2n/6
90 = 2n/4
120 = 2n/3
180 = 2n/2
0(=360t = 2n/1
Cosine
1/2
o
-1/2
-1
1
Order of Rotation Axis
6
4
3
2
1
rectangular formulation of the same lattice. We could also have obtained it
by simply taking the primitive rectangular lattice and adding the centering
points-being careful, of course, to assure ourselves that the centering points
were indeed equivalent to the ones already present. This line of thinking
naturally raises the question of whether we may ha ve overlooked other planar
lattices that might be obtained by centering the other three primitive ones.
In fact, we ha ve not, but let us se e exactly why.
Suppose we were to center the oblique lattice of Figure 11.3a. This does
not in any way improve its symmetry. All we ha ve is another, denser oblique
lattice, which would be properly defined as having a smaller set of defining
translation vectors and a unit cell with half the are a., as shown in Figure 11.6a.
If we were to center the square lattice we would again fail to bestow any
Original
"Centered"
a

j)

V
1
1

1
-----

' 1

' 1
....

(a)

ro_l
(;.

(b)
. a..

;;,120" /
-f- .

60 1
----

(e)
Figure 11.6. Drawings showing the consequences of attempting to produce centered
lattices of oblique (a), square (b), and hexagonal (e) types.
358 APPLICATIONS
improvement-or indeed make any change whatever in its symmetry. We
should simply have another, denser square lattice for which a smaller unit
cell, with half the volume and edges equal to 11v'2 times those we had
previously (see Figure 11.6b). If we were to center the rhombic cell of the
hexagonal lattice, we should simply destroy the hexagonal symmetry and
create a primitive rectangular lattice (see Figure 11.6c). We would degrade
the symmetry without creating anything new or useful.
11/ustrative Example The array shown below is certainly regular and could
obviously be extended to infinity in aH directions. However, it is nota Jattice.
Why not?
a a# b
Answer
The points are not all equivalent, that is, in identical environments. There .
are in fact three distinct kinds. (Selecta representative example of each kind
and label them a, p, and y.) The array shqwn is three interpenetrating lattices,
of a points, p points, and }' points. All three lattices are identical but displaced
from ea eh other by the shortest horizontal and/ or shortest vertical distance
shown.
11.3 TWO-DIMENSIONAL SPACE SYMMETRIES
A Building-Up Scheme. There are 17 of these symmetries. We shall first
show how they are built up, and then examine their symmetry elements in
greater detail. The process of building them up is similar to, but more elab-
orate than, the process of building up the 7 lD symmetry classes. The starting
point is analogous, however, namely, an arra y generated only by translations.
With nothing but two unit translations that are unequal and at a random
angle (i.e., not necessarily 60, 90, or 120), we get the symmetry designated
p 1 shown in Figure 11.7. The letter p is used to indica te that the pattem is
primitive and 1 to indicate that no rotation other than 2n/n with n = 1 is a
symmetry operation.
We may create several new 2D space symmetries by adding rotational
symmetry. As we have seen, only axes of orders 2, 3, 4, and 6 are possible.

...
f

.... ......

p2
. ,. ,.. ,.. p: P"
f ,.. ,.. ,..
r ,.. ,..
r ,.. ,.. ,.. "'
,
pl
E E E E t
E E E E E

,. ,..
-...:--..::--... ......

-.:--.:--..., ......
, ,.
... ... .... .....
pg
,.
pgg
Figure 11.7 The 17 two-dimensional space groups; continuation on page 360.
[Adapted from l. Hargittai and G. Lengyel, J. Chem. Educ., 1985, 62, 35.]
360 APPLICATIONS
E E ____ E ____
(
E ~
( (
cm
~
~
(
!"'
p4m
Figure 11.7. ( Continued)
~ ~
~ ~
~ ~ ~
" .
/ ...
U pon adding each one of these we ere ate. respectively, the symmetries p2,
p3, p4, and p6. In each case it can be seen in Figure 11.7 that the combined
effect of the explicitly introduced rotation axis and the translational operations
generates further symmetry axes. Thus. in p2, in addition to the set of twofold
axes explicitly introduced, two other sets, lying at the midpoints of the trans-
lations connecting them, arise automatically because of the repetitive nature
of the lattice. In the case of p3 a second, independent set of threefold axes
arises. The reader can see by inspection of Figure 11.7 the additional axes
that arise in p4 and p6.
Another way to create new 20 space symmetries is to introduce reflection
lines to the preceding ones. This can be done only for the rectangular, square,
trigonal, and hexagonal lattices. If one set of planes is introduced, parallel
to one of the translation directions. we get the symmetry p1n. lt is to be noted
that a second set of reflection lines, interleaving the introduced set arises
automatically.
If two perpendicular sets of reflection lines are introduced, we obtain the
symmetry pnun. As always, twofold axes arise wherever the reflection lines
intersect, and therefore the symbol p1nm unambiguously specifies this group.
even though it does not explicitly mention the twofold axes.
Continuing with the rectangular lattices, we can add a set of glide lines.
thereby obtaining the symmetry pg. We can, further, introduce mutually
perpendicular sets of glide lines, whereby the symmetry pgg is obtained. Note
that a set of twofold axes arises bellveen the glide fines.
Finally. there are three combinations of m and g symmetry elements to
consider. lf we ha ve mutually perpendicular sets of 1n and g lines, we get the
symmetry pmg, where twofold axes arise on the g lines. lfwe ha ve alternating
sets of parallel m and g lines we get the symmetry cm. Here we see that the
combined effect of both kinds of lines has been to give a centered rectangular
lattice. We could equally well have arrived at the same result by starting with
the symmetry pm (or pg) and introducing the centering condition. Since the
symbol for centering is e, a perfectly rationa) designation for this symmetry
is cm, and that, by custom, is the one used.
In a similar way. the symmetry cmnz is obtained by adding g lines to pmm.
Twofold axes, of course. again arise.
The remaining five 20 space symmetries are all obtained by adding re-
flections to the groups p3, p4, and p6. On adding reflection lines to p3, there
are two ways to do it. If they are added so as to pass through all threefold
axes, we obtain a symmetry called p31n 1, whereas if they are added so as to
pass through only alternate threefold axes. we obtain a different one, P31m.
The numeral! does not denote any symmetry but instead serves as a notational
device to allow the formation of different symbols for the two trigonal sym-
metries.
There are also two ways to add reflection lines to symmetry p4. In one
case they pass through all fourfold axes, thus giving the symmetry designated
p4nz. In this symmetry (as we shall see in more detail presently) there are
362 APPLICATIONS
also glide lines lying between the reftection lines. If we add reflection lines
so that they pass between the fourfold axes of p4 (but through the twofold
axes) we automatically genera te a network of glide lines that Ji e between the
reftection lines, sorne of which do pass through the fourfold axes. This sym-
metry is called p4g.
Finally, if reflection lines passing through all the sixfold axes of p6 are
added, we get the symmetry p61n, which, of course, has other automatically
generated symmetry elements.
Diagrams for 2D Symmetries. The drawings in Figure 11.7 help to show
how the 20 symmetries can be developed from the simplest (pl, which has
only general lattice translations) to rather elaborate ones such as p4g and
p6m. However, for clear and rigorous representations of these symmetries,
we now introduce the kind of diagrams used by crystallographers. Diagrams
.similar to these will later be used for 3D symmetries, and thus the present
discussion serves as a gentle and useful introduction to our ultimate goal of
understanding the symmetries of real, 3D crystals.
The diagrams depict the outline of a unit cell (light solid lines) and show
within it all the symmetry elements that occur. The symbols u sed ha ve all
been introduced airead y, namely" those for the various axes perpendicular
to the plane, the reflection lines (heavy solid lines), and glide lines (heavy
broken lines). The set of diagrams is shown in Figure 11.8. We shall now
discuss several of them.
For pl we have only the outline of the cell since there are no symmetry
elements. For p2 the twofold axes are shown. For pm we see the parallel
reflection lines at the top and bottom edges and through the middle of 'the
cell, while in pg we see a similar display of the glide lines.
The diagrams for the remaining symmetries show clearly all of the sym-
metry elements in these more elaborate cases. For instance, in p41n we see
the twofold axes that were required by the presence of the fourfold axes of
p4 as well as the glide lines that arise automatically when the reflection lines
are introduced. In p4g we see there are actually two networks of glide lines.
These symmetry diagrams dramatize the fact that from a minimal set of
defining symmetry elements, others arise. This is, of course, in the nature of
all groups, as we have already seen for point groups in Chapter 3. In point
groups, however, additional symmetry elements always appear at the inter-
section of those initially specified. For example, two mirror planes generate
a twofold axis along their line of intersection. The new feature that we en-
counter with space symmetry is that new symmetry elements can be generated
elsewhere. Let us take a conspicuous and important example of this and see
how it works before we end this section.
In symmetry pgg, we see that the network of perpendicular glide lines gives
rise to a set of twofold axes that lie in the interstices of the network, not at
its intersections. To see how this happens, and that it is generaL we m ay refer
to Figure 11.9, which shows how an asymmetric object at an initial general
point, x, y, when subjected to all the operations made possible by the glide
lines, generates a set of fout such objects \Vithin the cell. Also shown are the
closest objects in adjacent unit ce lis, each of which is equivalent by translation
to one of those in the cell outlined.
Let us first make certain that all of these 10 objects can be interrelated to
pl p2
--- .
1 1

1 1

pm pg
cm pmm

~
1
1 1

1-----------
1 1 1

-------
pmg 1 1 1
J 1 1
pgg
1
~ : ~ i
1 1
~ ~ i i

1
t---'--1
1 1 1

1 1 1
1 1 1
1 1 1

cmm _ 1 j l ,
r- ~ 1 ~
~ ~ t- -f---i
1
Figure 11.8. Diagrams showing all symmetry elements for the 17 two-dimensional
symmetry classe.s; continuation on page 364. {Adapted from the International Tables
t?r X-ray Crystallography. 1965.)
364 APPLICATIONS
p4
. ---
p4m
. ---
l 1
!
-------
p4g
p3
p3ml
p3lm
p6 p6m
1\ 1
A
, ...

1
...
\/

each other and to the initia1 one at x, y by using only the glide operations
and, of course, the lattice translations. In the following list, we use an arrow
to show that one object goes to another and place a symbol over the arrow
to specify the glide plan e g
1
, g
2
, g
3
, g
4
, or lattice translation, a, b, used to
effect the shift.
b
4"
C3 C4
1 1
1 1
4"'
@x,y 1 1 ~ 1
1 1 1
gl-- ---,- ---- -,- ---- g
a
1 '2 1
1 3
1
. ':1 1
g2 ______ 1 ------+-----g2
1 1
1 1
4'
1 1
1 1
1,
83
1"
Figure 11.9. A diagram showing how an entire set of objects is generated from an
initial one (No. 1) ata general position (x, y) by the combined action of glide lines
and the Jattice translations.
1
.I:J ?
---+-
1 ~ 3
3 ~ 4
2 ~ 4
4 ~ 4 '
4 ' ~ 4
4" ~ 4"'
etc.
This diagram clearly shows that objects 1, 1 ', ... , 2 .. 2', ... , 3 .. 3' ..... ,
4, 4', 4'', 4"' are all obtainable with only glides and lattice translations and
also that 1, 2, 3, and 4 are the only ones within a unit cell that are so related.
We can now easily see that 1 and 4" are related also by twofo)d rotation
about an axis at the origin .. while 1 and 4' are related by an axis at a = t b
= 0., and 1 and 4"' by an axis at a = O, b = i, and so forth. Thus, we see
that a net of glide Iines coupled with the basic translational symmetry gen-
erates an entire set of twofold axes, one tn the center of each rectangle
bounded by the glide lines.
366 APPLICATIONS
Examples of 2D Symmetry Classes. One thing that distinguishes 2D sym-
metries from 3D symmetries is that examples of the former are all around us
in everyday life. To this, we shall take four examples from the
realm of brickwork. These are shown in Figure 11.10. It will be obvious that
these are patterns seen very often, but it may not, be obvious just
how much symmetry these pattems have.
The reader will derive maximum benefit from these examples by covering
up columns B and C and looking only at the patterns themselves in column
A. He should then try to assign each one to its 20 symmetry class. The other
columns and the text below will give the correct results.
A B
e
g2'
B'
'
A'
g}--
A
'
g
'
B
C' O' e
&
'
'
' g2
Figure 11.10. Four examples of masonry patterns and their symmetries. Column A
shows the pattems; B shows the corrector incorrect choices of lattice vectors and unit
cells; e shows the syminetry elements.
D L

1 1 e
1
(3)] D 1 1 [

For each case we show in column A a sample of the pattern itself, in
column B the vectors that define the pertinent lattice, and in column C the
symmetry elements that are present.
Examp!e l. The net of vertical and horizontal reflection lines is rather ob-
vious, as are the twofold axes. The fourfold axes may be slightly less obvious.
Once they are found, however, it beco mes clear that we need to turn the
pattern 45 in order to put it into the standard orientation for one of the
square symmetries, p4, p4m, or p4g. Since we have seen the net of reflection
lines we know it must be either p4m or p4g, and when we note that the
reflection lines pass between, not through, the fourfold axes we conclude that
it is p4g. The presence of the two different nets of glide lines, only <_?ne net
passing through the fourfold axes, is not obvious. The reader should convince
himself that they are there. The second dagram in column C shows one
of each type of glide. g
1
takes brick AB to A'B' while g"1 takes brick
368 APPLICATIONS
CD to C'D'. This example shows that the symmetry of the pattern can be
higher than that of the object used to create it.
Example 2. This example shows that the symmetry of the pattern can be less
than that of the object used to create it. The blocks may be perfectly square,
but the pattern does not have fourfold symmetry. Again .. a net of reflection
lines is perhaps mqst obvious and, of o u r s e ~ there are twofold axes at their
intersections. If this were all, we would have pmtn symmetry. However, it
can be seen that the array is centered, and closer inspection will show that
there are horizontal and vertical glide lines and an additional set of twofold
axes, making this an example of cnun symmetry.
Example 3. This is another example of p4g symmetry, as indicated by the
diagrams in columns B and C. The pattern looks very different from that in
Example 1, but its symmetry is exactly the same. The diagram in colun1n C
shows only the axes. Reflection lines and glide lines are also present.
Example 4. This rather simple-looking arrangement is se en in brick sidewalks
the world over. At first sight it may not seem very symmetrical; it may seem
that the asymmetric unit may be two bricks and the lattice might be taken to
be an oblique one. In fact, the correct lattice is rectangular (a and b) and
there is a net of glide fines. The correct group is pgg. Also shown fr this
case are two more "'obvious" but incorrect choices of lattice vectors (a' and
b') that would give oblique unit cells and symmetry p2. With this choice the
glide lines are omitted. To emphasize their presence they are shown in a
separate diagram at the bottom. The g
1
-type line carries brick AB to A'B'
while the g
2
-type line carries brick CD to C' D '.
11.4 THREE-DIMENSIONAL LATIICES ANO
THEIR SYMMETRIES
For one who has fully understood the preceding development of lattices.,
symmetry, and symmetry groups in 2D, the same fundamental concepts in
3D should be e as y to understand. The principies are the same; only the
dimensionality in which they are to be implemented will change.
Again, our first concern must be to see how many ways there are in which
the translation vectors can be related to one another (relative lengths, angles
between them) to give distinct, space-filling patterns of equivalen! points. We
have seen (Section 11.2) that in 20 there were only 5 distinct lattices. We
shall now see that in 3D there are 14. These are often designated eponymously
as the Bravais lattices and are shown in Figure 11.11., in the form of one unit
cell of each.
A lattice will be formed in 3D by applying three translational motions to
an initial point. This is shown in Figure 11.12 where we label the vectors and
CRYSTALLOGRAPHJC SYMMETRY 369
Triclinic
Monoclinic
Orthorhombic
Tetragonal
U. JJ lsometric
DlJ
Hexagonal
Figure 11.11. The 14 Bravais lattices arranged into the 6 crystal systems.
the angles between them in the way that is conventional in crystallography.
We shall no\v discuss why the 14 Bravais lattices are the only ones.
Centering in 3D. A few prefactory remarks are necessary on the property
of centering .. possessed by several of the Bravais lattices. In the 20 lattices
we encountered only one case of centering {the centered rectangular lattice) ..
but it well illustrated the principie involved., that is .. that a centered lattice is
used when it has the advantage of explicitly preserving the highest symmetry
of the array. We must remember also that the centering points must be
370 APPLICATIONS
e
e
a
A
e
e B e
a
(b) .
Figure 11.12. (a) Conventional labeling of translation vectors and angles used to
generate a 30 lattice from an initial point P. (b) Conventional Jabeling of unit cell
edges and faces.
equivalent to the others, even if this is not immediately obvious in sorne
drawings. Were they not equivalent the array would not be a true lattice.
In 20, there is only one way to introduce a centering namely, in
the middle of the rectangle that constitutes the unit cell. In 3D there are
several ways to introduce centering points. If one occurs in the center of the
cell (body centering) the sym bol 1 is u sed. If only one pair of opposite faces
are centered the symbols A, B, or C are used, depending on which pair of
faces (Figure 11.12b) is eh osen. If all six faces are centered, the symbol F is
used. There is also one case in Figure 11.11 in which two points lie within
the unit cell; this is not really "centering" and the reason for it will be discussed
fully at the appropriate place.
Let us now develop systematically the 14 lattices shown in Figure 11.11.
Clearly, if we impose no special requirements on the set of defining vectors
(Figure 11.12a), b anda# p #y, we have the 3D analog
of our 2D oblique lattice. It is called a triclinic lattice. As with the 2D oblique
lattice, there is no unique. way to choose the vector set, but normally one
would choose the three shortest vectors. Even if one angle happens to be 90
or two of the three vectors happen to be equal, the lattice is still triclinic
because these special relations do not enhance its symmetry. The triclinic
lattice has inversion centers as its only symmetry elements. Moreover, a
triclinic lattice is necessarily primitive, since if any additional points were
introduced at the center of the cell or at any of the face centers, we would
have to redefine our vector set in order to include them in a true lattice.
We next turn to monoclinic lattices, of which there are two types. A
monoclinic lattice is one in which we require one vector to be perpendicular
to the plane of the other two. The lattice then has t\vofo1d rotational symme
try about this unique vector and planes of symmetry perpendicular to it. A
monoclinic lattice (so-called beca use there is only one nonorthogonal pair of
translation directions) may be centered as well as primitive. In Figure 11.11
we show this in the form of body J. However. it is possible to retain
the same e axis ( which preserves the twofold rotational symme1ry and thus
maintains the monoclinic nature of the lattice) but to eh o ose the a and b axes
differently. These other choices, shown along with the original one in Figure
11.13, provide unit cells that are A centered or B centered. These are purely
arbitrary ( and equally va lid) eh o ices and the lattice itself is the same in all
cases. There are only two monoclinic lattices, P and 1 (the latter also definable
as A or B).
Note that there would be nothing to gain in definng a C-centered mono-
clnic cell since the resulting lattice would still be a primitive monoclinic lattice
with shorter a and b translation vectors, a smaller volume, and the same
symmetry.
We can continue to apply restrictions to the defining vector set so as to
obtain an orthorhombic lattice in which all three vectors are of different
but are required to be orthogonal. The lattice now has considerable
symmetry, namely, three mutually perpendicular sets of twofold axes, and
three sets of mutually perpendicular reflection planes.
In addition to the primitive orthorhombic lattice, there are three centered
types, /, C, and F, as shown in Figure 11.11. Because of the unequallengths
of the a, b, and e translation vectors, none of these three could be recast in
primitive form without loss of one, or of the right angles and hence
degradation to monodinic or triclinic symmetry. Note that A or B centering
are possible, but constitute nothing new since the assignment of labels to the
axes (and hence to the faces) is arbitrary. The important point is that either
any one pair or all three pairs of opposite faces can be centered.
If in addition to orthogonality of the translation vectors we also require
two vectors to be of equal length, say a = b, we have a tetragonallattice.
This now has the same mirror planes and twofold axes as an orthorhom bic
lattice but has fourfold axes parallel to the e direction. In this case there is
only one form of centering possible, namely, 1 centering .
./
./
a
Figure 11.13. Alternate ways of choosing a unit cell for the centered monoclinic
lattice. a, b, e, define the body-centered ([) cell; a. b', e define the A-centered cell;
a', b, e define the B-centered cell.
372 APPLICATIONS
Worked Example: Why is a e-centered tetragonal lattice not possible? Be-
cause the e faces are square, e centering would be to redefining
the a and b vectors (as a' and b') to produce a primitive tetragonal lattice
with a unit cell of one half of the volume, as shown below .

b

A C-centered tetragonal lattice would be nothing more than a primitive te-
tragonal lattice with two incorrectly chosen vectors.
Worked Example: Why is an A-centered tetragonall_attice not possible? Be-
cause centering only on the A faces (or only on the B faces) would destroy
the fourfold symmetry and hence the lattice would not be tetragonal. The
question of why centering on both the A and B faces is also disallowed is left
as an exercise for the reader.
If we proceed from the restrictions defining the tetragonal-type lattice to
the severest limitation, namely, a = b = e as well as a = p = y = 90 .. we
ha ve a cubic or isometric lattice. The symmetry is now very high. 1 n addition
to a set of mutually orthogonal fourfold axes, there are all the other symmetry
elements of a cube (point group 0
11
). Besides the primitive lattice it is possible
to ha ve 1 centering and F centering .. as shown in Figure 11.11.
Lattices with 3 or 6 Axes. The occurrence of six- and/ or threefold axes in
actual crystals occasions certain difficulties in classification. So long as we are
concerned only with lattices, however, there need to be no ambiguity, if we
proceed carefully and rigorously. For that reason we shall take a different
(and more fundamental) approach in demonstrating that there are only two
distinct lattice types consistent with the presence of three- or sixfold axes that
are oriented in one direction only. The latter qualification is necessary to
distinguish the present case from that of the isometric lattices where there
are sets of intersecting threefold axes. AH of the preceding 3D lattices could
also ha ve been treated in the following way, but for them the more heuristic
procedure gives correct results unambiguously.
A 30 lattice can be built up by stacking 20 lattices. If a 20 lattice is defined
by two transJation vectors, t
1
and t
2
, we need to introduce a third translation
vector, t
3
, that defines the stacking pattern. For example, if we stack a set of
(identical) oblique lattices (defined by t
1
and t:!) employing a vector t
3
that is
not orthogonal to the 2D lattice planes, we generate the triclinic lattice, while
if we require t
3
to be orthogonal to the 20 lattice planes and connect each
plane with a point in the nearest neighboring plane we get the primitive
monoclinic lattice.
To build up a 3D lattice that has six- and/ or threefold symmetry we
must take a set of hexagonal 20 lattices that ha ve symmetry p6 (Fig. 11.4)
and stack them in such a way as to retain the six- and/or threefold axes of
symmetry. Obviously, this means that the stacking vector t
3
must be perpen-
dicular to the planes of the 2D lattices, and sorne or all of the axes in each
of the nets must coincide. There are only two distinct ways to do this, as
shown in Figures 11.14 and 11.15.
lf, as in Figure we make the sixfold axes of all 20 nets coincide, \Ve
obtain a prin1itive lattice that retains a.ll the symmetry present in the 20 lattice
p6. We call this the primitive, hexagonallattice. However, we can also choose
the stacking pattern shown in Figure 11.15a, where we place the origin of the
cell in the nth layer over the point :\, A in the (n - l)th layer. The result of
this stacking scheme is se en in elevation in Figure 11.15b. I t has several
important properties.
l. Hexagonal symmetry is lost: Since the sixfold axes of the 2D net have
been made coincident with threefold axes, only the threefold symmetry sur-
vives. The reason that this is nevertheless still called a hexagonal lattice will
be given in Section 11.5.
2. If we wish to have a primitive cell, we must choose the one defined by
the vector t
3
, and two others of equal Jength related to it by threefold rotation.
We then obtain a rhon1bohedral cell bounded by six rhombuses.
3. We can, however, retain the original t
1
and t
2
vectors and use as the
third one, whereby we obtain a prismatic cell with two internallattice points-
Figure 11.14. The formation of a primitive hexagonal lattice by stacking 2D Iattces
of p6 symmetry.
374 APPLICATIONS
(a), plan
Figure 11.15. The formation of a primitive rhombohedral or triply primitive hex-
agonal lattice by stacking of 20 lattices of p6 symmetry.
a so-called triply primitive cell that has three times the volume of the primitive
rhombohedral cell. For practica) r e s o n s ~ crystallographers nearly always pre-
fer to employ the prismatic cell.
Now that we ha ve enumerated all of the 3D lattices. the 14 Bravais lattices,
we can look in more detail at their symmetries. First of all, it must be rec-
ognized that every lattice point is a center of symmetry. The translation vectors
t
1
t:!, and t
3
are entirely equivalent to - t
1
, - t:!, and - (h respectively. There-
fore, in determining the point symmetry at each lattice point (which is what
"symmetry of the lattice" means) we must include the inversion operation
and all its products with the other operations.
For the triclinic lattice there is no symmetry other than the inherent in-
version symmetry already noted.
For a monoclinic lattice (primitive or centered) the unique axis, that is,
the one perpendicular to both of the others is a twofold symmetry axis. When
the inversion property is added to this, we have the point symmetry group
Cu,.
For the orthorhombic lattices, each translation vector les on a C
2
axis.
These three axes plus the center of inversion result in the point symmetry
group D2h
For all the tetragonallattices one of the translation vectors is a C
4
axis and
the other two are C
2
axes. These elements plus the center of inversion give
us the point group D.t,.
For all the isometric lattices each translation vector lies along a e ~ axis,
there are C
2
axes bisecting each angle between them, and all the body di-
agonals of the cube are S
6
axes. Combining all the operations resulting from
these symmetry elements with the inversion operation, we obtain the group
0/r.
The two hexagonal lattices, however, do not ~ a v e the same symmetry.
Clearly .. when we stack the 20 nets of p6 symmetry without any offset, we
conserve not only the sixfold axes that pass through the lattice points but also
the two sets of three vertical reflection planes. These alone give rise to the
group c6, When the inversion operation is added we have the group DM
On the other hand .. the off-set stacking pattern that gives the other Hhexag-
onal'' lattice destroys the C
6
axes, leaving only C
3
axes and three vertical
mirrors at each lattice point. This gives rise to the group C
3
,.. When the
inversion operation and all its products are added, we get the group D3.
1

Table 11.2 summarizes these symmetry properties of the 14 crystallattices.
1t is seen that they are grouped into the six crystal systen1s, although one
system has two subdivisions. Also given in Table 11.2 are the characteristics
of the lattice vectors for each crystal system and another set of group symbols,
the meaning of which will next be explained.
11.5 CRYSTAL SYMMETRY: THE 32 CRYSTALLOGRAPHIC
POINT GROUPS
A crystal consists of an ordered 3D repetition of a fundamental unit ( the
asymmetric unit), which m ay be one molecule or several molecules ( all the
same or of several kinds). This ordered array consists of a large number of
unit ce lis { effectively, an infinite number) that fit together to fill space and
TABLE 11.2 Properties of 3D Lattices
Crystal
Lattice Symmetry
Axial CeJJ
System Schonflies CrystallbgraphiC'' Relations Types
Triclinic e, 1 a#b#c
p
a#p#y
Monoclinic c'J.h
21111 a#b#c P, 1 (orA or 8)
J' # a = p = 90
Orthorhombic D1/, 11ll1l11l a#b#c P, l. A. F
a=p=y = 90
Tetragonal
D4h
4/ 11ll1ll11 a=b#c P, 1
a = /1 = j' = 90
Isometric
oh
nz3nz a=b=c P ~ 1 .. F
(cubic)
a= P = )' = 90
Trigonal-
D.,J
3n1 a=b#c P or rhombohedral
hexagonal t t ~ t
6/nzn1n1 a= fJ = 90
)' = 120
'The use of these symbols is explained in Section 11.5.
376 APPLICATIONS
that are defined by three translation vectors. The termini of these vectors,
re pea ted o ver and o ver, form an arra y of points called the crystallattice. We
have devoted the first four sections of this chapter to the properties, especially
the symmetry of the lattices available to 2D crystals (monolayers)
and the more usual 3D crystals.
However, the symmetry properties of the crystals themse)ves are more
complex than those of the lattices, and we now turn to these. There are, of
course, clase connections between lattice symmetries and crystal symmetries
and we shall presently bring our knowledge of lattice symmetries into use in
exploring crystal symmetries.
Point Symmetry Elements and Operations for Crystals
Crystallographers use different conventions and symbols to specify symmetry
than those we have so far employed. There are good reasons for preferring
each system in certain fields (i.e., Sch6ntlies in molecular spectroscopy and
the new ones when dealing with ordered arrays of molecules, namely, crys-
tals). In developing the crystallographic conventions and we shall
go first to those required to handle point group symmetry. The most crucial
difference is that crystallographers, unlike spectroscopists, define improper
rotation as a combination of rotation and inversion rather than rotation and
retlection. More formally put, they use rotoinversion axes and operations
rather than rotorejlection axes and operations. This, of course. changes noth-
ing insofar as the realities of symmetry are but it does change the
descriptors markedly, and it is necessary to explain and illustrate the crys-
tallographic system of notation and see how it relates to the one we have so
far used in this book (the Sch6nflies system used by molecular scientists).
Associated with this change from rotoreflections (S,'s) to rotoinversions,
ttiere are other notational changes and it is most efficient to deal with them
all at once. To describe point symmetry in a crystal we use the following
symmetry elements and operations.
First, there are five pure rotations, by 2nln where n = l. 2, 3, 4, and 6.
Previously these axes were symbolized C
1
, C
2
, C
3
, C_., and C
6
They are
represented by 1, 2, 3, 4, and 6 in crystallographic notation.
Second, there are five rotoinversion operations, symbolized I, 2, 3, and
6. We need to examine these more closely to see how they relate to the
Schonflies symbols S,, 1n, and i (where it is well to recall that m = S, and
i = S2).
The 1 Operation: This entails no net rotation but inverts all the coordi-
nates. In short, it converts a point x. y, z to the point x, y, z. Thus T is simply
.!_he crystallographers equivalent of i. The crystallographic notation employs
l, not i.
The 2 Operation: This is a product of a rotation by 2rr/2 ( which accom-
plishes the transformation x, y, z x, y, z if we rotate about the z axis) and
inversion (which carries x, y. z to .X, y, z). Overall, it transfonns x, y, z to
x, y., z and the effect of 2 about a given axis is the same as reftection
-in aplane perpendicular to that axis. The 2 axis can be, and usually is, written
as a mirror, tn. by crystallographers.
The 3 Operation: The easiest way to see how 3 is related to our previous
notation is to examine Figure 11.16a and compare it to Figure 11.16b. These
diagrams show all the points generated by clockwise rotation from an initial
point 1 lying some\vhere on the upper half (solid points) of a sphere. The
open points lie on the lower half of the sphere as viewed from above. The
numbers 1 to 6 show the order in which the points are generated. which is
different in the two cases. The set of however, is exactly the same and
thus 3 is the crystallographer's equivalent for s6.
2
1
4
(a)
1
3
(e)
1,4
(e)
4 4
3,6
1
4
{b)
1
3
(d)
1.4
(/)
2 .
2,5
Figure 11.16. Diagrams comparing rotoinversion and rotoreflections operations.
378 APPLICATIONS
The 4 Operation: In parts (e) and ( d) of Figure 11.16 the sets of points
generated by the 4 and S
4
operations are shown. Clearly the two operations
give the same result and are equivalent.
The 6 OperatiQn: In parts (e) and (f) of Figure 11.16 we see that the set
of points generated from an initial point by the set of 6 operations is matched
(not by S(, which we already sho\ved is equivalent to 3) but by S
3

Let us now illustrate how sorne familiar point groups are denoted in the
crystallographic notation. To begin at the very beginning, C
1
and C; are called
1 and I, and Cs is called In. The groups C2, e)., e,., and c6 are called 2, 3, 4.,
and 6.
As we know., in groups with a symmetry axis and one or more symmetry
planes we need a way to indicate whether there is a plane perpendicular to
the axis or a set of planes that intersect along the axis. Previously we used
subscripts to the symbol a, namely, a
1
,, a,., ad, or in the group symbols, namely,
C
3
, or C
3
, . In crystallographic notation we use a slash befare the m if it is
perpendicular, as in 2/ m, which denotes the C
21
, point gro u p. This symbol is
read in words as u2 upon m" or u2 on m." When there are planes that intersect
along the axis we write one m for each independent set directly following the
axis designation. Thus, CM and C
6
,. are, respectively, 611n and 61nm. When
there are both vertical planes and a horizontal plane, as in D(J,, we write
6/ rnnun, where the slash applies only to the first n1.
The 32 Crystallographic Point Groups
Beca use only rotations (proper or improper) by 2rr./ n \Yith n = 1., 2, 3, 4,
and 6 can occur in 3D lattices, only those point groups that comprise these
and no other rotations are found in crystalline solids. They can be identified
by turning to the list of character tables and selecting all of those that meet
the above qualification, dius leading to the list given in Table 11.3. It is worth
pointing out that while one might ha ve been tempted to put D -td and D6d into
the list based on the group symbol, closer inspection shows that they are
excluded because they contain operations implying that S
8
(=S) and S
12
(=12) axes exist.
Let us now look more closely at these 32 point groups., for each of which
Table 11.3 gives both the familiar Schontlies sym bol and the crystallographic
one. For many of the simpler cases we have already discussed the basis for
the crystallographic symbol, but it is worthwhile to do this for a few more.
Let us begin with C
1
,., D
1
, and D'lh. The group C
2
,. is described as mm
because two intersecting mirror planes generate a proper axis along the line
of intersection. The symbol 2nun could have been used but contains a re-
dundancy. D
21
, is represented as nunm, indicating there are three different
mirror planes. A 2 axis is necessarily generated along each of the three fines
of intersection, as well as an inversion center where all planes and axes mee t.
Thus, the en tire group of eight operations comprising the group D
2
h is implied
TABLE 11.3 The 32 Crystallographic Point Groups
Schontlies CrystaHographic Crystal
Number Symbol Symbol System
1 el
1
}
Triclinic
2 e, 1
3 el
111
}
4 e"!.
2 Monoclinic
5 e ~
21n1
6 e:!,. 111111
}
7
D ~
222 Orthorhombic
8 D211
11li1U7l
9 c ..
4
10 s ~
4
11 c4,, 41n1
12 C4,. 4nznz Tetragonal
13 D"J..I
42n1
14 D4
422
15 D4h
41111111111
16
c.l
3
17
S"
3
18 C.:,. 3nz
19
D"
32
20 D ~ 31n
21
C"',
6 Trigonal-
22 e,. 6 Hexagonal
23
CM
6/nz
24
DJh
6n12
25 e,_,. 6nznz
26 Dn
622
27 D111r
6/111111111
28 T 23
29 Th
1113
30 T,, 43nz Cubic
31 o 432
32 oh
nl3nz
by the symbol nun1n. The group D
2
is a pure rotation group consisting of
only three C
2
(or 2) operations and the identity (1) operation. lt is uniquely
specified by the symbol 222. Actually this symbol is redundant since two
perpendicular twofold axes e reate the third ( cf. page 30) but it is the one
conventionally used.
For the groups D"' and D
6
the symbols 422 and 622 state the defining
properties, namely a unique principal axis ( 4 or 6) and two sets of twofold
380 APPLICATIONS
axes perpendicular to it. Again these conventional symbols strictly speak-
ing, redundant.
The crystallographic symbols 421n and 6m2 each convey the same three
pieces of information but in different sequences. This difference is arbitrary
and without significance as far as point symmetry per seis concerned, although
in a crystal the sequence in which the 2 and the m are listed may be significant.
In each of these group symbols the necessary and sufficient defining symmetry
elements are named, namely a principal improper axis (4 or 6), a (set of)
mirror plane(s) m that intersect along the principal axis, and a set of per-
pendicular twofold axes. What is most striking to one familiar with the cor-
responding Schonflies symbols, namely, D2d and D
31
p is how different they
appear. This is beca use the crystallographic point of view always emphasizes
the highest axis of rotation, be it proper or improper. In the group D"J.J there
are the operations arising from an S
4
axis (S
4
, s = C
2
, and Sj = E); since
4 = 'S
4
.. the crystallographer begins with 4 and adds the mirrors and twofold
axes necessary to specify the group. Similarly, in the group D
3
, there are S
3
,
Sj (= Cj), Sj(a
11
), S1 (=C
3
), and operations and this implies the existence
of an S
3
axis. For the crystallographer this is a 6 axis and this highest-order
symmetry axis is featured in the symbol, along with the a,.'s (as m) and the
C
2
's (as 2).
The remaining crystallographic group symbols that might cause any puz-
zlement are the last five, those .for the tetrahedral and octahedral groups.
Here, again, a little thought {while looking at the pertinent character tables)
will show that the crystallographic symbols are models of precision and econ-
omy. They state just enough symmetry elements to imply all the rest and,
hence, all the operations that arise therefrom. The reader should have no
difficulty, though it m ay take a little time, to convince himself of this. The
use of m3m to imply the existence of the entire set of 48 operations in the
group having the Schonflies symbol 0
11
is impressive, since the existence of
fourfold axes is not explicitly mentioned.
11.6 INTERRELATING LA'ITICE SYMMETRY, CRYSTAL
SYMMETRY, AND DIFFRACTION SYMMETRY
In Section 11.4 the fourteen 30 lattices (Bravais lattices) were derived and
it was shown that they could be grouped into the six crystal systems. For each
crystal system the point symmetry of the lattice was determined ( there being
one point symmetry for each, except the hexagonal system that can have
either one of two ). These seven point symmetries are the highest possible
symmetries for crystals of each lattice type; they are not only ones.
Befo re we can address the question of how many other (i.e., lower) point
symmetries may be assigned to each crystal class, it will be helpful to specify
what symmetry is essential (i.e., the lowest symmetry) to defining each crystal
class. Let ns take each class in turn and see what is essential.
Triclinic. The essential requirement is that there be no symmetry (other
than, possibly, T). lf a crystal possessed even one twofold axis or one mirror
plan let alone any more symmetry, it would not be triclinic.
Monoclinic. The essential symmetry for this class is one proper or im-
proper twofold axis, that .is, either 2 or nz ( =2). This guarantees that one axis
is perpendicular to the plane defined by the other two. lt is important to
appreciate that the mere occurrence of an axis that appears to be perpendic-
ular (to within \vhatever accuracy however great) does not make the crystal
monoclinic. The perpendicularity must be the rigorous result of the symmetry
2or m.
Orthorhombic. This system has three mutually perpendicular unit trans-
lations, whose perpendicularity is rigorously required by symmetry. This can
only be assured if there are three mutually perpendicular twofold axes, either
all proper (222) or the combination 222 (= mm2). The lattice itself has n1mnz
symmetry because when the centering condition is added to either of the
above, it expands them from 222(D
2
) or 2mm(C
2
,.) to mmnz(D
21
,).
Tetragonal. The essential requirement he re is fourfold symmetry either
4 or 4. The only symmetry elements that can be added are those perpendicular
or parallel to 4 or 4, so as not to generate additional 4 or 4
the symmetry would in crease to sorne type of cubic symmetry.
Trigonal-Hexagonal. The essential symmetry elements are 3 or 3 (for
trigonal) and 6 or 6 (for hexagonal).
lsometric (cubic). The essential symmetry here is a set of 3 axes that are
carried one into another by rotations about 2 axes (and vice versa), the 2 and
3 axes not being perpendicular to each other.
In Table 11.4 we list the crystal classes along with the minimum symmetry
necessary for each, and the maximum (i.e., lattice) symmetry possible for
each.
In Section 11.5 the 32 point groups to which crystals can belong were
presented. Naturally, the 7 point groups found for lattices are among them.
However, that lea ves 25 point groups that we need to associate with the lattice
type with which they are consisten t. Of these, 11, Jisted in column 2 of Table
11.4, are those that have the bare mnimum required for that crystal class.
The remaining 14 ha ve intermediate degrees of symmetry.
For the triclinic system, only the two point groups l(C
1
) and l(C;) are
possible since all others have too much symmetry.
t
.

:
3
C

T
A
B
L
E

1
1
.
4

S
y
s
t
e
m

T
r
i
c
l
i
n
i
c

M
o
n
o
c
l
i
n
i
c

O
r
t
h
o
r
h
o
m
b
i
c

T
e
t
r
a
g
o
n
a
l

h
e
x
a
g
o
n
a
l

I
s
o
m
e
t
r
i
c

S
y
m
m
e
t
r
y

G
r
o
u
p
s

f
o
r

t
h
e

S
i
x

C
r
y
s
t
a
l

S
y
s
t
e
m
s

E
s
s
e
n
t
i
a
l

L
a
t
t
i
c
e

S
y
m
m
e
t
r
y

S
y
n
1
1
n
e
t
r
y

N
o
n
e

1

2

o
r

n
z
(
=

2
)

2
/
n
t

2
2
2

o
r

2
n
u
n

1
1
l
n
l
l
1
l

4

o
r

4

4
/
n
m
u
n

3

o
r

3

J
1
n

6

o
r

6

6
/
n
t
n
u
n

2
3

n
t
3
n
t

D
i
f
f
r
a
c
t
i
o
n

P
o
i
n
t

G
r
o
u
p
s

(
L
a
u
e
)

S
y
m
m
e
t
r
y

H
e
l
)
.

T
(

e
,
)

T

1
n
(
C
,
)
,

2
/
m
(
C
2
,
)

2
/
n
t

2
2
2
(
D
:
!
)
,

2
n
u
n
(

C
:
!
,
.
)
,

1
1
1
1
n
n
z
(

D
2
1
,
)

n
u
n
n
z

4
(

C
4
)
,

4
(
S
,
.
)
,

4
/
n
t
(
C
4
,
,
)

4
/
n
t

4
2
2
(

D
,
,

4
m
m
(
C
,
,
.
)
,

}

4
2
n
z
(

4
1

4
/
n
z
n
u
n

3
(

C
J
)
,

J
(
S
t
t
)

3

3
2
(
D
3
)
,

3
n
z
(

C
,
,
,
.
)
,

J
n
1
(
D
3
,1
)

3
n
z

6
(
C
,
,
)
.

6
(
C
J
1
,
)
.

6
/
n
z
(
C
(
,
,
,
)

6
1
1
1
1

6
2
2
(
0
,
,
)
.

6
m
m
(
c
.
.
.
)

}

6
n
z
2
(

D
3
1
,
)
,

6
/
1
1
l
l
7
l
1
1
1
(

D
f
l
l
r
)

6
/
1
1
1
1
1
1
1
1
1

2
3
(
T
)
.

1
n
J
(
T
h
)

n
z
3

4
3
2
(

O
)
.

4
3
n
t
(

T
c
t
)
.

n
z
3
n
z
(

O
,
,
)

1
1
1
3
1
1
1

For the monoclinic system it is essential to have one twofold axis, either
2(C
2
) or 2(1n), and it is permitted, of course, to have both. When both are
present the point group is that of the lattice, 2/m( C21t) There are no inter-
mediate symmetries. By proceeding in this way .. we can arrive at the results
shown in column 4 of Table 11.4, where each of the 32 crystallographic point
groups (i.e., crystal classes) has been assigned to its appropriate crystal system.
To illustrate further these relationships, let us loo k at the tetragonal system.
The defining characteristic is the occurrence of a single 4 or 4 axis. To this
may be added further symmetry elements so long as they do not create any
more 4 or 4 axes. Thus, to 4 we may add a horizontal plane. two sets of
vertical planes. or both, thus converting 4( to 4/nt( C
4
,) .. 4mtn ( C
4
,.), or
4/ntmm (D
4
,). We may also add only two sets of perpendicular twofold axes
and obtain 422 (D
4
). Beginning with 4 (S.) we can add elements leading to
42nt (D
2
").
It remains now to discuss the entries in the last column of Table 11.4. the
diffraction (Laue) symmetries. It was observed many years ago by Friedel
that the diffraction pattern given by a crystal will display a center of symmetry
whether the crystal actually has one or not. In other words, the diffraction
pattern cannot distinguish among all the 32 point symmetries, but only among
those that differ by something other than having or not having centers of
inversion. The diffraction pattern will show which crystal system we are deal-
ing with, but within each crystal system it cannot distinguish among those
subsets that all become the same when the inversion operation is added to
those that do not already have it.
The term diffraction symmetry, which is often called La u e symmetry be-
cause it can be most conspicuous in a type of X-ray photograph called a Laue
photograph, is applied to those point groups that are recognizably different
in the diffraction pattern.
Consider, for example, the groups 2( C
2
) and 1n( C
5
). If the inversion op-
eration (and all ts products) is added to either of these it becomes the group
2/ m( C
21
,). Among the five noncentric point groups of the tetragonal system,
however, there are two subsets that respond differently to the addition of the
inversion operation. Addition of the inversion operation and all its products
to 4( C
4
) or 4( crea tes 4/ m ( C
4
,.), whereas addition of the inversion operation
to 422 4mnt( C
4
v), and 42m (D2c1) gives rise in each caseto

These, and all of the other relationships of point groups to Laue groups shown
in Table 11.4 can be verified by consulting the group character tables.
To summarize, crystals may have any of 32 point symmetries, but beca use
of the limitations imposed by the translational repeat patterns, the lattices
themselves can belong to only 7 point groups. By means of X-ray diffraction
it is possible to determine the point group only insofar as the 32 can be sorted
into 11 subsets, each containing 2, 3, or 4 point groups. Those noncentric
point groups that all become the same upon addition of inversion are indis-
tinguishable from one another or from their centrosymmetric parent group
by X-ray diffraction alone.
384 APPLICATIONS
11.7 ADDITIONAL SYMMETRY ELEMENTS AND
OPERATIONS: GLIDE PLANES ANO SCREW AXES
Thus far we have addressed the symmetry of crystalline arrays only in terms
of the proper rotations and the rotation-inversion operations ( the latter in-
cluding simple inversion, as T, and reflection. as 2) that occur in point sym-
metries, along with the lattice translation operations. However. for a complete
discussion of symmetry in crystalline solids, we require two more types of
operation in-which translation is combined with either reflection or rotation.
These are, respectively. glide-reflections ( or, as commonly called, glides) and
scre\v-rotations.
Glide-Reftection. This operation is referred to a symmetry element called
a glide plane. We have already employed a glide line (its 20 equivalent) in
developing the 2D space groups.
In 3D a glide operation combines the results of reftecting a point through
a plan e and translating it in the direction of sorne lattice vector ( not necessarily
a, b, or e) by one half of the length of that vector. The restriction to one half
of the Iength arises because a repetition of the operation has to bring the
initial point into coincidence with an equivalent point. Since the net effect of
doing the reflection part twice is to do nothing, the effect of doing the trans-
lation part twice must be equivalent to carrying out a lattice translation.
Otherwise, the assumption that there is a lattice would be violated.
lt turns out that three types of glide plane can be differentiated. In the
first type, the translation is in the direction of a principal lattice vector, that
is .. it is given by one of the vectors a/2, b/2, or c/2. For each of these, the
plane must be parallel to the plane defined by the translation direction and
one of the other two principal directions. Thus, if we have a glide plane
parallel to the plane of a and e, the glide component may be either a/2 or
c/2. Planes of this type are called axial glide planes and are symbolized a, b,
or e, according to the direction of the glide.
There is also a possibility of a diagonal glide plane (or operation). Suppose,
again, that our plane is parallel to that defined by lattice vectors a and c. but
the glide part of the operation is given by a/2 + c/2. This means a motion
parallel to a face diagonal of the unit cell. The symbol for a diagonal glide
is n. In rare cases a glide operation may entail a translation halfway along a
body diagonal, that is., by a/2 + b/2 + c/2.
Finally, there is a rare type of glide called a diantond glide (plan e or
operation).- This can occur only for 1 or F lattices where a translation of, for
example. a/4 + b/4 can move a corner lattice point toa face centering lattice
point (or vice versa). The symbol for a diamond glide is d.
Table 11.5 summarizes the symbols and translations for each type of glide.
Screw Axes. An operation in which a point is rotated by 2rr 1 n about an
axis and also translated by sorne distance r parallel to that axis is called a
TABLE 11.5 The Types of Glide Planes and Opera-
tions
Glide Type
Axial
Diagonal
Diamond
Translation
a/2
b/2
c/2
(a + b)/2
(a + c)/2
(b + c)/2
(a + b + c)/2
(a :!: b)/4
(a c)/4
(b c)/4
(a b c)/4
Symbol
ll
b
e
ll
el
screw-rotation. lf the operation is to be a symmetry operation (i.e . move an
initial point toan equivalent point) it is necessary that r be of such a magnitude
that after n operations (which comple-te the rotational cycle) the accumulated
translations are equal to sorne lattice translation t or an integral multiple 111
of that translation, nzt. If we are to satisfy this requirement. namely. that
nr = 1nt, the only possible values of r are (nlln )t. For n., we may ha ve any
of the values allowed for rotational symmetry in a lattice, namely, 1., 2, 3, 4,
6, and in each case we limit m to the integers ~ n. Thus, the possibilities for
screw axes are restricted to 2., 3., 3
2
., 4t. ~ 4
3
, 6, 6
2
, 6
3
, 6
4
, and 6
5
The
symbols just used are of the form nm.
To grasp the effect of the operations arising from each of these screw axes
il is useful to examine Figures 11.17 and 11.18. In Figure 11.17 we introduce
the conventional symbols (of which, more later) for the screw axes as seen
end-on. We use a dot to represent each of the points that are generated, one
from another by the screw operations. One of the points is assumed to be at
an (arbitrary) initial height of zero and the height of each of the others is
given as a fraction of the total translational distance covered by n applications
of the nm operation.
For a 2
1
screw axis, the initial point genera tes another point at ~ ~ but
repetition of the 2
1
operation takes this to the point that would be obtained
by simply employing the lattice translation t. We are able to see all the
consequences of the successive 2
1
operations within one unit cell. The same
will be true for the 3
1
, 4
1
, and 6
1
operations, but for the others we shall find
the initial point being carried into the cell above or even beyond. However.,
we must keep in mind that the simple lattice translations, t or - t, will always
generate equivalent points in al/ cells, including the one we start \Vth. For
the screw axis designated 3
1
it takes three operations to move the initial point
by the lattice repeat vector t. The two points generated along the way, at
386 APPLICATIONS
1

t
2

2
1
2
fr.
*-

t 3
3 32

:f.

1
-
-
2 2

--

4 2 4
41 42 43
1 2
1

2. 6
3
2
l
3 s
3 2
6 6 3 62
Figure 11.17. Diagrams showing the effects of screw rotations in projection down
the screw axis. Fractions represent the fractional distances of each new point above
the level of the initial point (which carries no fraction).
t/3 and 2t/3 are shown. For the 3
2
axis. the first application generates a point
at 2t/3 and the second gives a point 4t/3 = 1 + t/3. This is a point at t/3 in
the next unit cell, but the translation - t gives us the equivalent point in the
unit cell under consideration. lt is clear from the diagrams for 3
1
and 3
2
that
they can be regarded as having the same translational component but coupled
with rotations in opposite directions.
Clearly the same relationship exists for the 4
1
/4
3
and 6
1
/6
5
pairs.
With we have still another situation, is perhaps best shown by
the diagram in Figure 11.18. Here we number the points from 1 for the initial
point through 2, 3, 4 to show the arder in which they are generated.
In a similar way, the effects of 6
1
, 6:h and 6
4
operations are evident in
Figures 11.17 and 11.18.
1'
1
2 2
3
3
4
4
6
6
1
1,
1
3
13 -,-+--1
42
5
1'
1
4'
1 -4,--+--1 1 --+--1
62 5' 6.3 64
1
6s
Figure 11.18. A second type of diagran1 showing the way screw operations replica te
an initial point 1, until finally it reaches its original position but translated up by one
vector distance (point 1 ').
Standard Symbols for Symmetry Elements
We have now introduced all of the symmetry elements required to build up
the 30 space groups, and in the next section, we shall introduce these space
.groups. Befare doing so, it is convenient to define the complete set of symbols
that are required to represent all of the necessary symmetry elements, as seen
from all possible directions, when representing them in a diagram of a unit
celL Table 11.6 displays these symbols.
This table is essentially self-explanatory. Each plane that is perpendicular
to the page is shown by a line drawn where it intersects the page. The type
of Hne shows the type of plane. When the plane is in or parallel to the page
it is shown by a pair of line.s, without arrowheads for a simple mirror and
with one or more arrowheads (oran additional arrow) to indica te the direction
of a glide operation. To show the height of a plane above the base of the
cell, its fractional distance will be given.
388 APPLICATIONS
TABLE 11.6 Standard Symbols for Space Group Symmetry Elements
l. to Projection ln or 11 to
Element Symbol Plan e Projection Plane
Sin1ple mirror 111
1
or\
(glide in projection
1
or--
Axial glide a,b,c,
plane)
..........
(glide l. to projection
or
pJane)
Diagonal glide ll
~
Diamond glide d --E---
~
---+-
Center of inversion 1
o
Rotation axis 2, 3, 4, 6

A.+
)
Rotation-inversion axis 3,4,6
A 4 > ~
Screw axis
21

3,, ~
4,, 42, 4;\

6, 62, 6.h 6.s. 6 ~
.......
As for the remaining symbols, many ha ve already been used, namely. those
for the rotation axes, 2, 3, 4, and 6 and the various screw axes seen end-on.
Symbols not previously used are those for the I axis (inversion center) and
the other three rotation-inversion axes 3, 4, and 6. Recall that 2 is equivalent
to m. Fin aH y, there are the symbols for rotation and screw _axes that lie parallel
to the page, which are distinguished by use of full and half-arrowheads,
respectively.
Fractional Coordinates. In specifying the location of a point in a crystal
lattice it is customary to employ coordinates, x, y, z, that give the fraction
of each principal vector distance (a, b, e), which define the unit cell. Thus,
a point at the origin has the fractional coordinates 0,0,0 while the center of
the cell has the coordinates ttl. The face centers are O,l,i, !.,O,! and !,!,0
for the a, b, and e faces, respectively. lt is to be emphasized that these
fractional coordinates are not Cartesian except for isometric cells and are not
even orthogonal for triclinic, monoclinic, or hexagonal lattices.
11.8 THE 230 TOREE-DIMENSIONAL SPACE GROUPS
In Section 11.3 it was shown ho\V the 17 20 space groups could be built up
by combining, in a nonredundant fashion, all the possible symmetry opera-
tions with each of the fundamental sets of lattice translations, that is, the 5
. 20 lattices. In an analogous way we can go from the 14 Bravais lattices through
the 32 crystaJ classes to develop a total of 230 3D space groups. The derivation
of these 230 space groups is a process that can be carried out in an entirely
rigorous, systematic fashion. In fact, the derivation was carried out as a purely
mathematical exercise around 1890, long befare the discovery of X-ray dif-
fraction {1912) provided a way to determine the internal symmetry (as con-
trasted to that portian revealed by face development) of real crystals.
We cannot present here the complete derivation, as it is very lengthy. *
However, we shall discuss sorne representative cases in detail and these should
serve to convey the essential ideas. For each illustrative case, we shall give
the space group symbol and the conventional diagrams and tables used by
X-ray crystallographers. On the basis of these specific examples the general
rules for notation and diagrams will be relatively e as y to appreciate.
Triclinic Space Groups. The triclinic crystal system allows no axis of ro-
tation of arder higher than one, namely, 1 or T. Since neither o f. these can
gve rise to any additional symmetry, there are just two triclinic space groups.
Both are primitive and are designated Pl and Pl.
PJ. This group is so trivial, having no symmetry apart from the three
independent lattice translations, that we shall not present the diagrams or
coordinate tables.
Pl. The action of the lattice translations (i.e., the symmetry of the lattice
itself) upon any one inversion center (l) that we introduce is to generate
others (cf. the 20 group p2). It is conventional to place one inversion center
at the origin of the unit cell. The translational symmetry of the lattice then
generates another one at the center of the cell (!,t!), three more at face
centers (e.g., and three at the midpoints of the edges (e.g., for
a total of eight inversion none of which are equivalen t.
We now give our first illustration of how the essential features of a given
space group are presented compactly in the master tabulation employed by
X-ray crystallographers throughout the world. t The three most fundamental
features (although there is much additional information given for each space
group ), and the only ones we shall discuss he re, are
l. A symntetry diagrant showing a projection of the unit cell and the
location of all its symmetry elements.
2. A commensurate cell diagram showing equivalent general points that
arise when any one initial point is replicated by the symmetry.
* An excellent recapitulation of the total derivation is given by M. J. Buerger, Elemenlary
Crystallograplzy, John Wiley & Sons, New York, 1956.
t lnternationa/ Tables for Crystallography, Volume A. S pace Group Symmetry, International
Union of Crystallography, D. Riedel, Dordrecht, Holland, and Boston. 1983.
390 APPLICATIONS
3. A list of positions of aiJ possible types within the cell and the symmetry
that exists at ea eh one.
Symmetry Diagram for PI:
In this diagram the origin is at the upper left, the +a axis running down,
and the + b axis running horizontally to the right. The e axis must be imagined
to be coming upwards at the appropriate angle. Thus the a, b, ande axes (as
well asan associated set of fractional coordinates, x, y, z) forma right-handed
system.
Centers of symmetry, represented by the conventional symbol (Table 11.5)
are also shown. This diagram can show only those in the ab plane and those
above this plane (at !,!,!, 0,0,!, i,OJ, and O,!J) are not explicitly shown.
These omissions are covered by the list of positions to be discussed below.
General Point Diagran1jor PI:
-8 8-
-0
-0
O+ O+
At the upper left within the cell is an open circle with a + sign beside it.
This represents an initial point lying in the octant defined by 0,0,0 and the
vectors a/2, b/2, c/2. The + sign shows that e > O. lnversion of this point
through l,!,O gives the other point within the cell. This point is accompanied
by a - sign to show that it has a coordina te of - z. It also has a comma in
it to show that the inversion operation turned the first point upside down.
The remaining points outside the perimeter of the cel1 are translation equiv-
alents of those inside. They are shown to emphasize that while there are seven
other inversion centers besides the one at !,!,O, no further points are generated
that cannot be obtained by lattice translations from the first two.
List of Positionsfor Pl:
Number of positions,
Wyckotf notation,
and point symmetry
l 1 x,y,z; x,y;z.
1 h 1
lll.
2,2,2
1 g 1 O,k,!.
-
1
f
1 !,O,i.
1 e 1 !,!,O.
1 d 1 !,0,0.
1 e 1 0,!,0.
1 b 1 0,0,!.
1 a 1 0,0,0.
This list is reproduced exactJy as it appears in the International Tables. It
tells us all the different kinds of locations that exist within one unit cell. In
each instan ce we are given the "multiplicity" of the type of point, namely,
how many of them there are that are equivalent and obtainable from each
other by application of symmetry operations. There is also an italic letter,
called the Wyckoff letter. This is simply an arbitrary code letter that sorne
crystallographers sorne times find useful; these letters need not concern us
further. Next there is the symbol for the point symmetry that prevails at the
site. Finally, there is a list of the fractional coordinates for each point in the
se t.
The point given at the top of this list (and all such lists) is called the general
position for that space group. It does not lie on any symmetry element and
392 APPLICA TIONS
therefore its multiplicity is the highest possible for that space group. All the
other points are called special positions. They lie on one ( or, in more elabora te
groups. at the intersection of several) symmetry element(s). Consequently
they have the symmetry characteristic of that element and there are fewer of
them because., unlike the general position. a special position does not get
replicated by the operation(s) on which it lies. In PI there are eight distinct
special positions, one for each of the eight distinct inversion centers.
Monoclinic S pace Groups. There are 13 monoclinic space groups and space
limitations will not permit us to discuss all of them, Jet alone present a sys-
tematic derivation of them. We shall, however, use this set of groups to
illustrate in a general way how the process of systematic derivation works.
We have seen in Section 11.6 that the monoclinic system embraces three
point symmetries: 2 .. 111, and 2hn. and that a monoclinic lattice may be either
primitive or centered. The centering may be treated as body centering or
centering in the middle of one of the rectangular faces. The standard choice
among X-ray crystallographers is the latter, and we shall discuss only that
alternative. If the unique direction in the lattice (i .e., the one that is per-
pendicular to the other two) is called e then we ha ve chosen A or B centering
as opposed to l. It is, of course. always possible to redefine any face-centered
monoclinic cell as an J-centered one by redefining the principal translation
vectors and there are (rare) when this offers sorne advantage. A
and B centering are interchangeable by simply redefining the a and b lattice
vectors. We shall work only with A centering.
From the foregoing prefactory remarks. we can describe the task of finding
all of the space groups in the monoclinic system in the following explicit
terms. For each lattice type (Por A) we must add one or more of the
symmetry elements, which, for this system are the various forms of twofold
symmetry: 2, 2 (=n1) and their related translational elements 2
1
anda glide.
plane a (or b) or n.
Just as we have previously done with point groups and 20 space groups,
we must test all the possibilities, beginning first with those where we add only
one symmetry element, and then those with combinations. For the latter there
can (and will) be redundancies. Two combinations may seem different, but
because a pair of symmetry elements in combination generates a third one.
they may be no more than two ways of defining the same final result.
Let us note before proceeding that we need not consider both a and n
glides, since they are interconverted by simply redefining the a and b axes.
Table 11.7 lists all of the distinct possibilities, that is, the 13 monoclinic
space groups. This table should suggest two questions, one involving the logic
and another pertaining to the notation. The first question is: What is wrong
with the obvious combinations of A with 2
1
, with 2
1
and 1n, and with 2
1
and
a? That is, why are there not three more monoclinic space groups A2.,
A2
1
hn. and A2
1
/a? The reader can, no doubt. anticpate the general answer:
Because these are simply alternative symbols for groups already listed. It is,
TABLE 11.7 The 13 Space Groups in the Monoclinic System
Lattice
Type
p
A
Added Space Group Symbol
Symmetry
Elements
2
21
11l
a
2 and 111
2
1
and 111
2 anda
2
1
anda
'J
k
111
a
2 and 111
2 anda
Standard
n
P21
Pnz
Pa
Pllnz
Pl/111
P2/a
P21/a
A2
A1n
A a
AZ/n1
A2/a
Alternate
Pe
C2
Cn1
Ce
C21nz
C2/c
in fact, not difficult to show that A2
1
is the same as A2., from which it fo11ows
that A2
1
1nz and A2
1
/a are equivalent to A211n and A2/a, respectively.
Figure 11.19 shows the diagrams for each of the groups P2, P2
1
, A2, and
A2
1
For P2 we have a set of 2 axes parallel to the e direction. The symmetry
of the lattice generates all the types of 2 axis from any one that is originally
introduced ~ e e Section 11.3 ). Similarly for P2
1
we ha ve only the several types
of 2
1
axes arising from any one type in conjunction with the fundamental
translation symmetry of the lattice. Now Jet us turn to the diagrams for A2
and A2
1
We can think of the former as being derived from P2 by adding the
A centering and the latter from P2
1
by adding the A centering. However. it
turns out that each one has interdigitated sets of 2 and 2
1
axes. In fact., they
are identica/ and the diagrams appear different only because we have chosen
the origins differently.
Why are A2 and A2
1
the same group by two different names? We can give
both a general answer and a specific answer. The general answer is beca use
of the closure property of all groups (space groups just as surely as point
groups). The operations generated by the principal translations, the twofold
rotations and the A centering are not a gro u p. The products of the A centering
and the rotations genera te additional operations, namely, screw rotations
about a set of 2
1
axes lying between the 2 axes. Conversely., for A2
1
, the
translations, the 2
1
operations, and the A centering are not a closed group.
The products of the 2
1
operations and the A centering are a set of twofold
proper rotations about a set of axes that lie between the 2
1
axes.
To be more specific, let us look at a diagram showing the behavior of a
general point in the group A2" as shown in Figure 11.20. The initial point
(number 1) becomes point 2 upon rotation about the 2 axis through the origin.
394 APPLICATIONS

P2
~ O
1
2+0
O+
O+
f
1
2+0
O+
P2
+O
1
--+Q +O
O+
O+
O+
A2
A2
Figure 11.19. Diagrams showing symmetry elements and general point positions for
space groups P1, P2
1
, A2, and A2
1
(which is not different from A2 except for placement
of the origin).
If we now invoke the A centering (i.e., point 2 is translated by b/2 + c/2),
point 2 becomes point 3. But clearly, point 1 will go directly to 3 if a 2t
operation is carried out about a 2
1
axis cutting the b edge of the cell at b/4.
In a similar w a y ~ every twofold rotation can be coupled with the A-centering
translation to generate one of the 2
1
axes.
Turning now to the question of the group symbols in Tab1e 11.7, we may
note lirst that all of those in the column labeled "standard" arise in a simple,
logical way. First, the symbol tells us whether the Iattice is primitive or
b
Figure 11.20. A diagram showing how the operations 2, 2
1
, andA centering move
an initial point 1 to other positions, 2 and 3 (see text).
centered. Second, it tells us what symmetry element, or elements, are nec-
essary and sufficient to specify the group unequivocally. This is a scheme that
is followed for all the remaining space groups.
The list of alternate symbols arises because at an earlier stage in the history
of X-ray crystallography, it was the accepted convention to call the unique
axis b. The reader can easily see that if this is done, the alternate symbols
become correct. Since in the other crystal systems with a unique direction
(i.e., tetragonal and hexagonal) the unique direction is called e, the formally
correct practice is now to do the same for the monoclinic system. However,
the literature is still replete with the old choice of axes and it is necessary to
be cognizant of both systems and of their relationship.
To conclude our discussion of monoclinic space groups, let us examine the
group P2
1
1 c-probably the one most commonly found out of all the 230. *
We use the older, more common choice of axes just to illustrate that such a
choice is entirely practica} even if it is not now officially sanctioned. We show
the standard diagrams (as they appear in the 1983 Edition of the lnternational
Tables) and table of positions for P2
1
1 e in Figure 11.21.
The projections shown are on the ab plane. The screw axes are seen running
parallel to the b axis, and it is indicated that they lie at z = ! . The c-glide
planes, which are perpendicular to the diagram, are represented by dotted
lines at b = ! and 1. As a conseq uence of the presence of these defining
symmetry elements there is also a set of inversion centers, as also shown.
Those shown in the ab plane are, of course replicated in the plane at z = l
by the screw and glide operations.
The way in. which a general point is replicated by the operations of this
group is seen in the projection to the left. The multiplicity of the general
point in P2
1
1 e is 4 and the coordinates of the four members of the set are
More thcm three-quaners of all crystals whose space groups have been established IJelong to
only -6, of which P2,1c is most common. lt appears that at / e a ~ t 16 and possibly more than 30
of the 230 space groups have never been observed. (Sorne claims are probably erroneous.)
396 APPLICATIONS
-0
O+
-0
t+0
1
02-
Numhcr of positiuns.
Wvckolf notation.
and poinl symmctry
4 e
2 d
2 e
2 b
")
a
1
4 ......=::_ ) o
-0
O+

v t + ,. l_ -
.. \. 2 _,, ,:z 4
1 J+
X,2-y.2 Z
1 tto.
1 O,l.O.
1 tO,O;

1 0.0.0; o.t!.
Figure 11.21. The diagrams and list of positions for space group P2
1
/c, as given in
the lnternationaJ Tables for X-ray Crystallography ( 1965). The diagrams are projec-
tions on the ab plane.
listed below. There are also four special positions, ea eh having 1 symmetry
and a multiplicity of 2. While at first glance more such positions n1ight have
been expected, a little thought will show that the pairings shown are in accord
with the fact that members of any one pair would be interchanged simply by
choosing the origin difieren ti y. These origin changes would not change any
of the spatial relationships of the symmetry elements to one another.
ILLUSTRATIVE EXERCISE
Draw the symmetry elements for a n
1
/c unit cell in its other two projections.
The results as shown draw attention to several points. First, for the
be projection, the choice of origin ( upper righL not left) is dictated by the
b o
be a e
need to maintain a right-hand axis since the Hn1issing" axis is always
assumed to ha ve its positive direction upward from the plane of the paper.
The half-arrows for the screw axes now lie in the plane of projection (also at
x = !) and therefore carry no fractional coordina te. Finally, the glide planes
are now indicated by a dashed rather than a dotted line since the glide direction
is parallel to one of the axes shown rather than parallel to the one 1zot shown
as in the ab projection.
In the ac projection, the origin and axes are again chosen to conform with
a right-handed axis set. The screw axes are now represented by the symbol
that shows them to be perpendicular to the projection plane. The glide planes
are now shown by the symbol at the upper left, which gives both the elevation
(y = t and, by and glide direction (e).
Orthorhombic Space Groups. There are 59 of these space groups divided
among three crystal classes: 222(D
2
), m1n2( and 1nnun(D
21
,). Within each
class there is at least one group associated with each of the four types of
orthorhombic Bravais lattice, P, C (orA), F, l. We shall make no attempt
to derive these systematically, but a few examples and sorne use fui obser-
vations are warranted. The complete list of the 230 space groups given in
Appendix VIII should be consulted at this time.
For the orthorhombic system, the assignment of letters to the axes is ar-
bitrary, except that they should form a right-handed system. We thus have
three possible choices. On the other hand, the three axes are physically quite
distinct, being, in general of different lengths. Moreover, the symmetry ele-
ments are oriented in only one particular way relative to the actual axes. This
means that for each assignment of letters to the axes, a different symbol may
arise. This is illustrated in Figure 11.22 for the case of the group conventionally
designated Pba2. Physically, the b-glide plan e is ..L to the shortest axis a ..
while the a-glide is .1 to the medium axis b. This is an immutable physical
fact. Thus, if we wish to relabel the axes as shown in the other two drawings,
we must also change the group symbol (as shown) so asto be always describing
the same physical reality.
This question of axis labeling can arise in practice if the researcher first
determines the axial lengths for the unit cell and only la ter, after all data
ha ve be en collected and indexed according to that choice, finds out (by meth-
398 APPLICATIONS
ods to be discussed in Section 11.9) where the symmetry elements lie. There
is one chance in three that the choice made corresponds to the conventional
one. If not, either the axes may be relabeled (and all data reindexed) or the
unconventional choice (called by crystallographers a nonstandard setting) mav
be retained, but all of the diagrams and positional coordina tes in the standard
table will ha ve to be changed appropriately.
To derive the 9 space groups in crystal class 222 is no more complicated
than was the derivation of the 13 monoclinic space groups. We take, in turn,
each lattice type and associate all possible combinations of 2 or 2
1
axes with
it, and then weed out the duplications. For the primitive ones we get 4, with
no duplications:
With the C, F, and 1 lattices we would find that of those we might write in
a similar way, only 5 would be distinct. The process is not a difficult one, but
it takes time. Let us examine just one illustration.
Shown below ~ r the symmetry diagrams for C222 and C222
1
, the only two
C-centered orthorhombic space groups. C2
1
2
1
2 does not exist because C222
already contains the sets of 2
1
axes parallel to the a and b axis. C2
1
2
1
2
1
is, in
fact, C222
1
, possibly with a shift of origin. Final1y, C2
1
22 and C22
1
2 are simply
C222, with relabeled axes, while C22
1
2
1
and C2
1
22
1
are relabeled versions of
C2
1
2
1
2 {which, as already noted, is C222}.
t t t
1
1 1
..
L
1
--
..
-
---
-- J

1
J
1
---
---

1
J
1
- --
--- 4
4

1 ___
J J
__l
4 4
--
!
-
1_ L
-l
4
;,
4
~
~ t
+
~
1
'
C222
C222
Similarly, we could begin a search for the 28 groups in crystal class n1nzm
systematically by considering all distinct combinations of mirror or glide op-
erations. In this way we would get the 16 listed in the table, almost by
inspection. In so doing, it must be remembered that since the a, b, and e
axes are arbitrarily labeled, none of the unlisted combinations, Pnnb, Pbnn,
Pcnn, Pnan, and Pncn, are actually different from Pnna, the one listed. These
others merely represent physica\\y meaningless re\abeling of axes.
Question: Why are the "possibilities" Pann, Pnbn, and Pnnc excluded?
a-glide ,.,
.l.b /
\f
a
1/
1
1
1
e
----,
1
b-glide
V .l.a
1
1
Pba2
1
1
1
e
b a
b
Pc2a P2cb
Figure 11".22. Diagrams showing how three different choices for labeling the axes
lead to different symbols for the same space group. At left, where the positions of
the glide planes are shown is the standard setting, that is, the conventional, tabulated,
choice.
Let us examine one specific space group, C1nnz2, in more detail and use
it to illustrate sorne further important points. The necessary information is
shown in Figure 11.23. The symbol te lis us that the cell is e centered, possesses
mirror planes perpendicular toa and b, and has twofold axes in the e direction.
Sorne twofold axes are along the intersections of the mirror planes but there
is an additional set at x = y = ! , and so on, and al so two sets of glide planes
c. Perhaps the easiest way to see how these additional symmetry elements
arise is to look at the diagram showing the set of general points.
All eight members of the set of general points may be obtained by using
only the two m 's and the e centering. However, it is then seen that the glide
operations and the additional set of twofold rotation operations will also
generate the set of eight general points. All of this behavior is, in fact, exactly
as seen in the 2D group cmtn! Moreover, we could ha ve defined this group
as Cgg2 and proceeded to develop the same total symmetry.
There are two important points to note about the list of special positions
and their symmetries. The point t t z is identical to O, O, z., beca use they
are related by the centering translation. The point at !, O .. z may not appear
equivalent to the listed one, O, 4 .. z since the a and b directions in the lattice
are not equivalen t. However, it is because the two twofold axes at the centers
of the a and b edges are interchanged by the twofold rotations about the axes
ata = b = !.. and so on.
Space Groups in Other Crystal Systems. There are 68 tetragonal, 52 tri-
gonal-hexagonal, and 36 isometric space groups. Though they outnumber the
triclinic, monoclinic, and orthorhombic ones by more than 2: 1, they are less
commonly encountered. Fundamentally, they involve no new principies al-
though the very high and complex symmetries they often present can provide
sorne very severe tests of one's understanding of the principies. In any event,
space does not permit us to discuss them any further.
400 APPLICATIONS
+O 0+ +O 0+
+ 0 o+
t---__ +_,0 0 +
+Q O+
+O 0+
+0 O+
Numhcr of pnsilinns.
Wvckoff notoation.
anJ point !>ymmctry
+O 0+
+ 8 O+
1 1
1 1
--------
--------
1 1
1 1
1 1
1
'
-------- --------
'
r
1 1
8
J
x.y,z; x.y . .:; x.y,.:.
4 e 111 O,y.z; O.y, :::.
4 d 111 x.O .:; x,O.z.
4 e 2
! 1 - 1 ...
..... ... '
... .....
2 b 11ll1l o.tz.
2 a 111111 U,O.z.
Figure 11.23. The standard diagrams and list of positions for the space group Cnun2.
11.9 SPACE GROUPS AND X-RAY CRYSTALLOGRAPHY
S pace groups (like point groups) constitute a very pretty branch of mathe-
matics, but that (presumably) is not why chemists study them. Space groups
are important to a chemist because they are essential to solving and inter-
preting crystal structures. Therefore, we conclude this chapter with three
topics that relate space group theory directly to the use of X-ray crystallog-
raphy to obtain chemically useful information.
Our topics will be
l. Systematic absences-these are essential to solving crystal structures
and hence defining molecular and solid state structures.
2. The relationship of molecular symmetry to crystal symmetry.
3. Chirality in molecules and crystals.
Distinguishing Space Groups by Systematic Absences. From the symmetry
and metric properties of an X-ray diffraction pattern we can determine which
of the 6 crystal systems and, further, which of the 11 Laue symmetries we
are dealing with. Since we need to know the specific space group in order to
solve and refine a crystal structure, we would still be in a highly unsatisfactory
situation were it not for the fact that the X-ray data can tell us still more.
Crystal symmetries that entail centering translations and/or those syn1-
metry operations that have translational components (screw rotations and
glides) cause certain sets of X-ray reflections to be absent from the diffraction
pattern. Such absences are called systen1atic absences. A general explanation
of why this happens would take more space and require use of more diffraction
theory than is possible here. Thus, after giving only one heuristic demon-
stration of how a systematic absence_ can arise, we shall go directly to a
discussion of how such absences enable us to take a giant step toward spec-
ifying the space group.
Every diffracted X-ray beam can be regarded as a reflection of the incident
beam from a set of planes defined by the lattice points. Each such set of
planes is characterized by a set of numbers that specify its orientation relative
to the Jattice and, indirectly, give the distance between planes, d. As shown
by W. L. Bragg .. a diffracted beam will appear when a set of planes is so
oriented relative to the incident beam that the condition shown below Figure
11.24a, the Bragg equation, is fulfilled. For a given X-ray wave length, )., a
set of planes with a spacing d will give a diffracted ray at an angle 20 to the
incident beam. The reason for this is that the path difference between rays
reflected by adjacent planes is equal to (that is all the Bragg equation says),
and thus they are in phase and reinforce one another.
The situation shown in Figure 11.24a involves a primitive rectangular 20
lattice. Suppose we have a centered rectangular 20 lattice, as shown in Figure
11.24b. We can see that for the 1.1 set of planes, no new ones are
the centering points lie on the planes already present. Therefore, the Bragg
condition for refJection is unaffected for these planes. For the 2,1 planes ..
however, the centering points introduce a whole new set of planes (shown
by broken lines), each lying exactly halfway between the ones already there
(shown by fulllines). Sin ce the spacing, d
21
, in the case of the primitive lattice
creates a phase difference of exactly 2n between adjacent rays (which is the
same as no phase difference) a halving of the spacing means that adjacent
rays will be n: out of phase and thus cancel each other out.
In summary, the introduction of the centering points will cause the 2,1
reflection to disappear, but will not cause the 1,1 reflection to disappear. The
reasoning used can readily be extended. Al/ sets of planes that pass through
the center of the cell (which requires them to have h + k even, that is, h +
k = 2n) continue to produce reflections when centering is introduced. AJI
402 APPLICATIONS

2, 1 Diffracted beam
lncident
beam
X = 2d;,k sin 8nk
(a)
--
-- . --
--
--
{b)

1, 1 Diffracted
be a m
Figure 11.24. (a) A primitive 2D rectangular lattice with two sets of planes shown,
along with constructions leading to the Bragg equation. given beJow. (b) A centered
2D rectangular lattice showing that additiona) planes are formed in the 2.1 case but
not in the 1.1 case.
sets of planes that ha ve h + k odd (h + k :1: 2n) will miss the centering
points and the number of planes in the set will be doubled. This will extinguish
all reflections from these planes.
What we have here is the simplest example of a systematic absence. All
reflections from the centered lattice whose indices obey the relationship
h + k# 2n
must ha ve zero intensity.
TABLE 11.8 Conditions for Systematic Absences
Condition
l. Lattice Centering
A-centered lattice (A)
B-centered lattice (B)
C-centered lattice ( C)
Face-centered lattice (F)
Body-centered lattice (/)
2. Glides Planes
perpendicular to a
translation b/2 (b glide)
e/2 (e glide)
b/2 + c/2 (n glide)
b/4 + e/4 (d glide)
perpendicular to b
translation a/2 (a glide)
e/2 (e glide)
a/2 + c/2 (n glide)
a/4 + e/4 (d glide)
perpendicular to e
translation a/2 (a glide)
b/2 (b glide)
a/2 + b/2 (n glide)
a/4 + b/4 (d glide)
3. Serew Axes
Twofold screw (2
1
)}
Fourfold screw ( 4:!) along
Sixfold screw ( 63)
Threefold screw (3" 3,) }alon
Sixfold screw (62o 6"') g
Fourfold screw ( 4
1
, ~ a long
Sixfold screw ( 6
1
6
5
) a long
H
e
a
b
e
e
Absent Reftections
ltkl k + 1 = 2n + 1
h + 1 = 2n + 1
h + k = 2n + 1
h + k = 2n + 1}that is, h, k, 1 not
h + 1 = 2n + 1 all even or all
k + l = 2n + 1 odd
h + k + 1 = 2n + 1
Okl k = 2n + 1
1 = 2n + 1
k + l = 2n + 1
k + 1 = 4n + 1, 2, or 3
hOI h = 2n + 1
1 = 2n + 1
h + 1 = 2n + 1
lz + 1 = 4n + l. 2. or 3
hkO h = 2n + 1
k = 2n + 1
h + k = 2n +
h + k = 4n + 1, 2, or 3
hOO lz = 2n + 1 = odd
OkO k = 2n + 1
00/ 1 = 2n + 1
001
1 = 3n + 1, 3n + 2,
that is, not evenly divisible by 3
hOO h = 4n + L 2, or 3
OkO k = 4n + 1, 2, or 3
001 1 = 4n + 1 , 2, or 3
00/ 1 = 611 + 1, 2, 3, 4. or 5
404 APPLICATIONS
In 30 lattices, face centering willlead to exactly analogous results: Planes
that do not pass through the centering points cannot give rise to diffraction.
In 3D, for a C-centered lattice we ha ve the following systematic absence
condition for a reflection with in dices Jzkl: h + k =F 2n. Similarly., for A and
B centering, the conditions must be k + 1 -:rf 2n and h + 1 -:rf 2n .. respectively.
A lattice centered on all faces will be subject to all three of these conditions
simultaneously. By a straightforward extension of the reasoning, we can sho\v
that for an /-centered lattice, the absence condition is h + k + 1 -:rf 2n.
Not only do centering operations give rise to systematic absences, but so
also do those other symmetry operations that entail translations, namely ..
glides and screw rotations. Geometrical derivations of the conditions for
absences .. similar to that used above for the centering operations, are cum-
brous in these cases. However, the conditions are easily discovered if the
proper algebraic expressions for diffraction intensities as a function of the
reflection ndices. k, h, l, are inspected. It is well beyond the scope of this .
chapter to derive these expressions and we ~ h r f o r simply present the re-
sults, along with those for the centering absences, in Table 11.8.
1t will be se en that glide planes give somewhat similar, but less restrictive
conditions. They do not affect completely general reflections but only those
from planes parallel to the glide plan e. Similarly, screw axes reveal their
presence in X-ray diffraction by still more limited conditions. Only reflections
from planes perpendicular to the screw axis are affected.
Let us now see how these conditions for systematic absences are used.
Suppose we have established from the X-ray diffraction data that a crystal is
monoclinic. See Table 11.7 for the monoclinic space groups. We can next see
if the unit cell is primitive or centered. If we choose the unique axis to be e,
we look for absences indicative of A centering (hkl, k + L # 2n) or B centeriPg
(hkl, h + 1 -:rf 2n). If we find the latter, we can reassign axes so as to obtain
the standard settng. If we find neither, we ha ve a primitive cell.
Let us suppose we have a primitive cell. There are then two other types
of absences to loo k for: ( 1) 00/, 1 -:rf 2n, which will indicate a 2
1
axis, and/ or
(2) hkO, h -:rf 2n, which will indicate an a-glide. We might find hkO, k # 2n.
indicating a b-glide, but we can reassign axes. The possible results and con-
clusions to be drawn are as follows:
Systematic Absence Found
Non e
OOL, 1 -:rf 2n
hkO, h # 2n
Both
Possible Space Groups
P2, Pn1, P2/nz
P2, .. n,hn
Pa, P2/a
P2,1a
Symmetry considerations and systematic absences can take us no further.
It is impossible to tell anything about which nontranslational symmetry ele-
ments are present. The crystallographer will have to resolve the choices be-
tween, say, Pa and nla or P2, Pm, and P2/nt, by actually proceeding with
CRYSTALLOGRAPHIC SYI\IIMETRY 405
the solution and refinement of the structure. Often, it is simply a matter of
tri al and error, with only one group allowing a satisfactory solution and/ or
refinement. These matters are beyond the scope of this discussion.
One other possibility might ha ve arisen in the case of a primitive monoclinic
crystal. A systematic absence hkO, lz + k -:F 2n might have been seen. This
(Table 11.8) would indicate a diagonal glide, a/2 + b/2. Thus. space groups
designated Pn and P2/n, (as well as P2
1
/n if the screw axis is also present)
would be possibilities. These could be retained as su eh or converted to Pa,
P21 a, or P2
1
1 a, respectively, by redefining the a and b vectors as
a' = a + b b' = b
The question of \Vhether to do this or not is often decided, in practice, by
which choice gives a J' angle closer to 90, and the literature is filled with
comparable numbers of structures in both P2
1
/a (or P2
1
1c) and P2
1
/n.
In all other crystal systems we encounter the same general situation., namely ..
that a few space groups (69, in fact) can be uniquely identified from a knowl-
edge of diffraction symmetry and systematic absences, while the rest form
mostly pairs, or small groups that are indistinguishable in this way. Table 11.9
lists for the triclinic .. monoclinic, and orthorhombic crystal systems the uniquely
determined space groups and the sets with identical systematic absences.
One final word on this subject is pertinent. In the monoclinic groups. the
systernatic absences of different types are independent and observable sep-
arately. This is not always the case. Note, in Table 11.9, that Pnnza and Pna2
1
are indistinguishable, even though the latter contains 2
1
axes as well as the
n- and a-glides. The reason that they remain indistinguishable is that the 00/,
l #: 2n absences that would signal the presence of the screw axis are already
demanded by the n-glide. for which the condition is Okl, k + 1 # 2n. This
latter condition includes the 00/ reflections.
Molecular Synzmetry aiJd Crystal In sorne cases, and with ca re,
a knowledge of the space group together with the number of molecules in
the unit cell can indicate forthwith something about the symmetry of the
molecule. This can happen when the rnolecule is required to reside on one
of the special positions and hence on a symmetry element ( or several). It is
then said to ha ve Hcrystallographically required" ( or "imposed,) symmetry.
Actually, such statements are unjustified, although the literature abounds
with them. We shall see why presently.
lt is never possible for a discrete molecule to lie on a screw axis or glide
plane (although a polymer can). so we are concerned only with havng point
group symmetries (i.e., one of the 32 appropriate for crystals). A couple of
illustrations will convey the essential concept.
In that commonest of space groups, P2
1
/a, the general position has a
multiplicity of 4, but there are also four, twofold special positions, each on
a center of inversion. lf a molecular compound crystallizes in this space group
406 APPLICATIONS
TABLE 11.9 Classification of Space Groups Based on Systematic Absences for the
Triclinic, Monoclinic, and Orthorhombic Systems
System
Triclinic
Monoclinic
Orthorho m bic
Uniquely
Determined
P222,
nt2,2
P2,2,2,
C222,
Pnnn
Pban
Pnna
Peca
Pe en
Pbcn
Pbca
Ceca
Fdd2
Fddd
lbca
Sets with ldentical Absences
PI
P2, Pnz, P21 nz
P2h P2,1rn
Pa, P2/a
A2, A1n, A2/nz
Aa, A2/a
P222. Pnun2, Pnznznz
C222, Cnun2, Cnunm
F222, Fnznz2. Fnznun
/222. /2
1
2
1
2
1
, lnun2, Jnunm
Pmc2
1
Pnta2, Prnnza
Pcc2, Pccnz
Pca2
1
Pbcnz
Pnc2, Pnzna
P1nn2., Pnunn
Pbarn
Pna2
1
, Pnnta
Pnn2, Pnnnz
Cnzc2
1
Anta2. Cnzcnz
Ccc2, Cccnt
Abnz2, Cnznza
Aba2, Cnzca
Jba2, lbant
bna2, Jnzma
(which is uniquely specified by diffraction symmetry and systematic absences)
and the unit cell volume and density show that there are only two molecules
per unit celL it follows that they must reside on one of the pairs of inversion
centers. For example. an "octahedral" molecule MX
4
Y
2
forming su eh crystals
would have to be the trans rather than the cis isomer.
As an example showing that even higher molecular symmetry can be in-
ferred from space group symmetry together with the number of formula units
present, we may look at
which crystallizes in the

tetragonal space group i4, \Vith only two formula units in the unit ceH. This
requires the cations to reside on a special position of 4( = symmetry.
Moreover, it also requires that the Mo-Mo axes of all molecules be paralJel.
1 t is important to be aware that there are al so traps for the unwary in this
sort of reasoning. It need hardly be said that when the symmetry demanded
by the crystal is low., e ven nil, the molecule may ha ve inherently higher
symmetry. This is, the usual case. Of course, as a symmetrical molecule
sits in the crystal, amidst surroundings of lower symmetry, it is subject to
packing distortions that to sorne extent, ho'Never smalL degrade its inherent
symmetry. Thus, had the MX
4
molecule mentioned above be en found to
occupy a general position in its crystal \Ve would learn nothing about the
molecular symmetry. Both isomers have symmetry or C
2
,.) but the crystal
need not reflect it, or any part of it.
There are occasionally more subtle traps. Sorne arise out of faulty reasoning
and are thus, in principie, entirely avoidable. lf, for example1 a crystal is
found to contain a number of molecules equal to the multiplicity of the general
position for its space group, that does not necessarily mean that the molecules
must be on gene_!"al positions. Consider the simplest possible a crystal
in space group Pl with two molecules in the unit cell. While in most instances
there are two equivalent molecules on the twofold general position, cases are
also known in which ea eh molecule resides on a different center of inversion.
In every such case this is a logical possibility and only an actual structure
solution can rule it in or out.
By far the most insidious source of error1 and also a common one1 in
making inferences about molecular symmetry from space group symmetry is
the occurrence of disorder. This usually takes the form of what we m ay call
a systematic disorder 1 as opposed to a random one: A molecule, or a com-
ponent of the formula unit, will lie on a crystallographic symmetry element
that would seem to require more symmetry of it than it is of having.
lt accomodates to this by systematically taking ea eh of two ( or more) ori-
entations an equal number of times in different unit cells.
A classic case is Fe
3
(C0)
12
, which was shown to form monoclinic crystals
in space group P2
1
/c with two molecules in the unit cell. It was concluded
that the molecule had to have a center of symmetry and hence that the set
of iron atoms was linear and not triangular. This seemed conclusive, and yet
it was very difficult to reconcile with much of the other data on the molecule.
In fact1 the crystallographic inference is false because the structure is disor-
dered. At each inversion center there is a molecu)e containing a triangular
Fe
3
unit buried in a roughly spherical shell of CO groups. The triangle is
oriented up or down, in the sense below1 at each site1 such that both orien-
tations occur equally often. On average then the iron atoms appear to define
a "cross of David" consisting of six one-half iron atoms:
Fe
/o" +
Fe -Fe
Fe

Fe '-\/Fe
Fe
In many cases an entity that cannot inherently have a true plane of sym-
metry will reside on crystallographic plane of symmetry and usatisfy'' the
requirement of this site by occupying one half of them with one orientation
and the other one half of them with the opposite orientation. This is likely
408 APPLICATIONS
to occur for molecules whose t\vo halves are similar in nature and even
though not identical.
Nearly planar molecules occurring as solvent of crystal-
is a notorious example) often "lie on a symmetry meaning,
in that so1ne of the atoms are actually in the plane while others are
disordered on either si de 1 as shown below:
One last example is shown in Figure 11.25. In [N(C.H
9
)
4
h[Re:!Ix], both the
cations and anions Ji e on threefold crystallographic axes, whereas, clearly.
neither can actually accommodate such an axis. For the cation the threefold
symmetry is satisfied by the lower part of the molecule,
while the
remaining

chain is partly disordered over three positions. The lx portian
of the anion is very nearly a cube and it sits with one of its body diagonals
coincident with the crystallographic threefold axis. lnside, the Re-Re units
are disordered over three orientations, each equally occupied, so that there
appears to be an octahedron of one-third rhenium atoms.
We conclude by returning to the earlier animadversion about the use of
the phrases "crystallographically required" or "crystalJographically
symmetry. While they are a convenient1 and comrnon, code for sorne of the
situations we have just been discussing, it is well to remember that they
actually state the cause-effect relationship entirely backwards. Space groups
neither require anything or nor impose anything upon the chemical substances
that form crystals. It is the substances themselves, that the sizes,
polarities. and so on of their component molecules or ions, that lead to a
certain packing pattern in the solid state. All such patterns then display one
of the 230 space group symmetries. The molecular structure causes the crystal
and not vice versa.
Chirality in Crystals. When chiral molecules form crystals the space group
symmetry must conform with the chirality of the mplecules. In the case of
racemic mixtures there are two possibilities. By far the commonest is that the
racemic mixture persists in each crysta1
1
where there are then pairs of opposite
enantiomorphs related by inversion centers or mirror planes. In rare cases,
spontaneous resolution occurs and each crystal contains only R or only S
molecules. In that event or 1 obviously, when a resolved optically active com-.
pound crystallizes, the space group must be one that has no rotoinversion
axis. According to our earlier discussion (page 34) the chiral molecule cannot
itself reside on such an axis. Neither can it reside elsewhere in the unit cell
unless its enantiomorph is also present.
This requirement exdudes PI in the triclinic system as well as all those
monoclinic space groups associated with crystal classes 111 and 2/ nz. Thus,
106
Figure 11.25. The disordered cation and anion in (N(C.,H
11
).,b[Re
1
IK] Each one lies
on a threefold axis of the unit cell. The hatched atoms are ordered, while the open
circles show the sets of atoms (C above, Re below) that are disordered over three
positions about the axis.
410 APPLICATIONS
only P2, P2
1
, and C2 are possible monoclinic space groups for chiral com-
pounds. It is quite easy to see by examination of Appendix VIII that the only
possibJe orthorhombic space groups are the nine belonging to crystal class
222, that the only tetragonal ones are those belonging to crystal classes 4 and
422, and so forth.
lt is important to note that not all space groups that can accommodate
chiral molecules are necessarily chiral. For example, it is clearly possible to
place 2 nonchiral molecules in a monoclinic unit cell in, say n, and have a
nonchiral crystal. On the other hand there are 11 enantiomorphous pairs of
space groups that must give chiral crystals because they are inherently chiral,
regardless of what is in them. These are the following, which are all based
on screw axes, and the pairs simply have axes of the same type spiraling in
opposite directions:
P3
1
(P3
2
)
P4
1
(P4
3
)
P6
1
(P6
5
)
P6
2
22( 6 ~ 2 2
P4
1
32(P4
3
32)
P3
1
12(P3
2
12)
P4
1
22(P4
3
22)
P6
2
(P6
4
)
P3
1
2l(P3
2
21)
P4
1
2
1
2(P4
3
2t2)
P6
1
22( P6
5
22)
While a collection f molecules that are all of the same chirality (e .g. a
o- or L-amino acid ora naturally occurring protein) must forma chiral crystal,
inherently nonchiral molecules are not barred from doing so, if they crystallize
in one of the 11 pairs of enantiomorphous space groups. In that event, which
is rather rare, there will, of course, be an equal probability of forming either
enantiomorph and a batch of crystals will normally contain both. A couple
of real examples are (NH
4
}3Tc
2
Cl
8
3H
2
0 (P3
1
21 and P3
2
21) and Sn(Ta
2
CI
9
h
(P6
1
22 and P6
5
22).
EXERCISES
11.1. Start with blank paper and a closed book and develop, systematically,
the seventeen 20 space groups, or selected subsets ( e.g., those with
only g and m lines, those with fourfold axes).
11.2. The 20 space group p6 arises by explicitly introducing one set of sixfold
axes. Show with a diagram the other symmetry elements that arise
automatically.
11.3. In the following sketches and patterns (a-i) you will find several of
the seventeen 20 space symmetries represented. Examine each sketch,
identify its symmetry group., and draw a diagram that shows all the
symmetry elements.
.ATA TATA
TATATAT
.AT.AT.ATA(a)
T.AT.ATAT
AT.AT.ATA
412 APPLICATIONS
~ ~ ~ ~
~ ~ ~
c_-=J ~ ~ ~ (h)
~ ~ ~
c _ = J ~ ~ ~

(i)

414 APPLICATIONS
11.4. Why must a glide-reflection operation entail a translation of i the
repeat distance and no other fraction of it?
11.5. Where are the inversion centers in a triclinic lattice? How many distinct
ones are there?
11.6. Write a matrix that converts an (incorrectly} C-centered monoclinic
lattice to the proper primitive one.
11.7. There is no space group that could be called Pna2. Show why. (Hint:
draw the symmetry diagram implied and examine its effect on a general
point.)
11.8. Why are space groups Cmm2 and A1nm2 truly different and not in-
terconvertible simply by relabeling axes?
11.9. To which crystal system must each of the following be long: Pn3n,
P3
2
21, P42
1
m, Ceca, P2/n, Pcn1b, 14/mmm, and Ama1n?
11.10. What systematic absences do you expect for each of the following
space groups? ldentify any pairs that are on this basis:
Pbc2
1
, Peca, Cmca, Pbc1n, C2cb, 14/tncm, and F222.
PART 111
APPENDICES
APPENDIX 1
MATRIX ALGEBRA
For those with inadequate knowledge of the fundamentals of matrix algebra ..
a brief introduction is given here.
Definition of a Matrix
In the most general sense a matrix is a rectangular array of numbers, or
symbols for numbers, which may be combined with other such arrays ac-
cording to certain rules. When a matrix is written out in full, it has an ap-
pearance of which the fo11owing is typical:
4 -7 6 o
2 9 -1 -8
2 o 5 4
-8 7 o -3
6 3 -4 7
Note the use of square brackets to endose the array; this is a conventional
way of indicating that the array is to be regarded as a matrix (instead of,
perhaps, as a determinant).
In order to discuss matrices in a general way, certain general symbols are
commonly used. Thus we may write a symbol for an entire matrix as a script
letter, for example .. .r 1, which stands for
a u a12 aJJ aln
a:H a22 an a2t,
a31 a32 a33 a3n
aml a,r2 a,u a,n
417
418 APPENDICES
We may also represent this matrix by [a;
1
]. The vertical sets are called coltunns,
and the horizontal ones rows. The symbol a; represents that element of the
matrix .r( that stands in the ith row and the jth coltunn. The 111 and n tell us
the order of the matrix; m gives the number of rows and n the number of
columns. A matrix in which m = n is called a square matrix and will be of
special importance to us. The elements in the set a; with i = j, that is, a
11
,
a
22
, a
33
, and so on, in a square matrix are called the diagonal e/en1ents be cause
they lie entirely on the line running diagonally from upper left to lower right
corners. A square matrix in which all of the diagonal elements are equal to
1 and all of the other elements are equal to O is called a unit nzatrix and
conventionally represented by the symbol cf.
A type of matrix that is of considerable importance is the one-column
matrix. To have the convenience of writing such a matrix all on one line, it
is sometimes written out horizontally but enclosed in braces, { }, so as to
distinguish it from a one-row matrix, which is normally written on one line
in square brackets. The chief significance of the column matrix, at least for
our purposes, is that it affords a way of representing a vector. Indeed it is
sometimes actually called a vector.
Let us consider a vector in ordinary three-dimensional space. We can
specify the length and direction of this vector in the following way. We arrange
to have one end of the vector lie at the origin of a Cartesian coordinate
system. The other end is then at a point which may be specified by its three
Cartesian coordinates, x, y, z. In fact. these three coordinates completely
specify the vector itself provided it is understood that one end of the vector
is at the origin of the coordinate system. We can then write these three
coordinates as a column matrix, in this case one with three rows, {x y z},
and say that the matrix represents the vector in question.
Obviously this notation can easily be generalized for vectors in abstract
spaces of any dimension. In p-dimensional space a vector can be specified by
a column vector of order (p x 1). The geometrical significance of the elements
of this vector matrix is the same as in real space: They give the orthogonal
( Cartesian in a general sen se) coordinates of one end of the vector if the
other end is at the origin of the coordinate system.
It should be noted that each of the coordinates of the outer terminus of
the vector is numerically equal to the length of a projection of this vector on
the axis concerned. Thus the set of numbers which define the vector in the
sense discussed above may also be thought of as defining it in the sense of
specifying its projections on a set of p orthogonal axes in the p-dimensional
space in which it exists.
Combination of Matrices
There are certain rules for adding, subtracting, and dividing matrices; these
are the rules of matrix algebra. It should be noted first that two matrices are
equal only if they are identical. If.( 1 = . ~ then a; = b; for all i and j.
To add or subtract two matrices, say .r./ and .41, to give a sum or difference fr,
MATRIX ALGEBRA 419
the three matrices must be of the same dimensions. The elements of? are
given by
A matrix may be multiplied by a scalar number or by another matrix. For
multiplication of a matrix [ C;] by a sea lar, a ~ we ha ve
Multiplication of a matrix by a matrix is somewhat more complicated. In
the first place, it can be done only if the two matrices are conformable. This
means that, if we wish to take the product .ri.4J = f,. the number of columns
in .rt must be equal to the number of rows in .-tJ. lf this requirement is satisfied,
so that.r( is of order (n x h) while ~ is of order (h x 1n ), then t will be of
order (n x 1n). Each element of the product matrix is given by the following
expression:
(AI-1)
This sum may be written out explicitly as follows:
where a;h is the last element in the ith row of ,rf, and bh
1
is the last element
in the lth coJumn of ..41. Perhaps this will be still clearer if we explicitly write
out the process of multiplying a 3 x 2 matrix into a 2 x 4 matrix.
420 APPENDICES
A mnemonically helpful way of summarizing the process is to say that the
ijth element of the product is obtained by taking the ith row of the first matrix
into the jth column of the second, \Vith emphasis on the "row-into-column"
aspect. From this discussion of the process of multiplication the conforma-
bility requirement is readily obvious. lf a row of matrix .r.t is to be multiplied
into a column of .43 then clearly the number of elements in that row, which
is the number of columns in the matrix must be equal to the number of
elements in a column of .43, which is the number of rows in the matrix .JIJ.
1t should be noted specifically that matrix multiplication is not in general
commutative. If the matrices .r.l and .41 are conformable in the sen se.' l-YJ
they need not necessarily be conformable in the sense .!IJ.r/. They will always
be confonnabJe in both ways when both are square and of the same order.
But even when the conformability requirement is commutation is
not in general possible. For example, consider the following two products:
Matrix multiplication does, however, always o bey the associative law. This
can easily be proved by extension of AI-1, and working through this proof
is a recommended exercise.
The quotient .' 11.41 m ay be equally well regarded as the product .r /.41-
1
,
that is, as .' 1 multiplied into the inverse of /IJ. We thus reduce the question
of how to carry out a division to the question of how to find an inverse. Tn
order to find the inverse of a matrix certain properties of the corresponding
determinant must be used. The subject is treated in detail la ter.
The product of a matrix and its inverse is commutative and equals a unit
matrix
Use of Matrices to Express Geometric Transformations
In Chapter 4 this use of matrices will be explored in great detaiL but prior
to that (Section 3.10) a few applications to coordinate transformations are
given. We give here the rudiments necessary. We may designate any point in
Cartesian space by a column matrix that gives its coordina tes, x, y, namely,
MATRIX ALGEBRA 421
Suppose we wish to reftect this point x, y, z through the origin. lts coor-
dinates will now be - x, -y, - z. We m ay express this by the following matrix
equation
If we want to rota te the point around the z axis by n/2, or reftect it through
the xy plane we can write
and
If we want to do all of these things. one after the other, we m ay obtain the
matrix to do it by multiplying the individual matrices to get one matrix that
does it all. Since matrix multiplication is associative it does not matter how
we choose the pairs. Thus, we may write
[
1 o o] [o - o] [ -1 _ ~ o]
~ ~ o ~ ~ o ~
- [ ~ - ~ ] [ ~ ~ ] - ~ ]
o o -1 o o -1 o o 1
We can now apply this to the original point and get, as before,
[-! ~ ~ ] [ ~ ] = [ ; ]
Evaluation and Expansion of Determinants
To find the inverse of a matrix we shall have to employ the corresponding
determinan t. Determinants are, by definition, square. They consist of a square
422 APPENDICES
of numbers. The number of ro\vs or columns is called the order n of the
determinant, and the determinant contains a total of n
2
elen1ents.
A determinant, unlike a matrix, is a scalar quantity. Its val u e is given by
the sum of n! different products each containing n elements so chosen that
each row and each column is represented but once. The sum of these products
is called the expansion of the determinant. That there are n! such products
is easily shown. To form one of them we select an element from the first row,
which may be done in n ways. An element from the second row cannot be
chosen from the column to which the element from the first row belongs;
thus for this choice we have only n - 1 possibilities. We can then choose an
element from the third row in only n - 2 ways and so forth. Thus there are
n(n - 1 )(n - 2)(n - 3) 2 1 = n! different ways of composing a product.
In addition. each product is given a sign, + or -. The choice of sign is
determined as follows. We write down all of the factors in such a way that
the row ( column) in dices run serially and count the number of transpositions
(exchanges of neighbors) that are required to put the column (row) indices
in serial order. If this number is even. the sign is +; if it is odd, the sign
is -. Consider, for example, the following product, which would be obtained
from a third-order determinant:
Written with the row in dices in serial order, it is
To put the column indices in serial order we must perform two successive
interchanges of adjacent factors, namely,
(""'\
a12a23a31 at2a31a2.l

a
31
a
12
a
13
Hence this product is given a positive sign.
The value of a determinant lA 1 of order 2 is given by
and that of a third-order determinant by
The results for these two simple and frequently occurring cases can easily be
remembered. The positive terms are products obtained by selecting elements
along diagonal lines running from upper left to lower right, and the negative
MATRIX ALGEBRA 423
terms are products of elements on lines running from upper right to lower
left. For determinants of the fourth and higher orders the number of products
(24 for n = 4) exceeds the number which may be enumerated in this way (8
for n = 4). and more labor is required to write down the complete expansion
of the determinant.
Determinants of order ;:?:4 may conveniently be evaluated by the 1nethod
of cofactors. Inspection of the list of six products whose algebraic sum is the
value of a determinant of order 3 shows that we may rewrite it in the follov -
ing way:
Each of the terms in parentheses is the expanded form of the determinant
made up of the elements of the original determinant which remain after we
strike from it the elements belonging to the row and column of the element
in front of the parenthesis. lt is given a + sign if the sum of the ndices of
the element before it is even and a negative sign if the sum of these indices
is odd. The terms in the parentheses are called the cofactors of the elements
in front of the parentheses. Thus we see that the third-order determinant can
be evaluated by finding the sum of the products of each element in the first
row with its cofactor.
A little reftection will show that we could just as well have arranged the
six terms in the expansion so as to have a sum of the products of any ro.w or
any column with their cofactors. For instance, choosing the second column,
we can write
It should also not be difficult to see that a similar process can be carried out
on a determinant of any order. Thus. using A; to represent the cofactor of
a;. we can write
!Al-
(for any j) (fnr any i)
One further important property of any determinant is that if any two rows
or colu1nns are identical the value of the detenninant is zero. This is easily
proved. Suppose that the pth and rth rows are identical. Then for any term
in the expansion, say
there must be another which is identical except that it will contain a" and ap
Now suppose that the column in dices are arranged serially in the term shown
424 APPENDICES
above, that p > r, and that x transpositions are required to put the row indices
in serial order. Then, if exactly the same x transpositions are carried out in
the term
it will still be necessary to make an additional2(p - r - 1) + 1 transpositions
to put a" and ap in their proper places, making x + 2(p - r - 1) + 1
transpositions in all. Thus, if x is even, x + 2(p - r - 1) + 1 must be odd
and vice versa. lt therefore follows that all the terms in the expansion will
cancel out in a pairwise fashion. Obviously a similar argument could be made
if we assume two columns to be identical.
The Adjoint Matrix
Before defining the adjoint matrix we must define the transpose of a matrix.
This is a matrix of which the columns are the rows, and vice versa, of the
original matrix. Symbolically, the transpose of the matrix (a;] is [a;]. Now,
the adjoint matrix of a matrix [a;] is defined as follows:
Adjoint of [a;] = [Aii]
That is, we treat the array of elements constituting [a;] as a determinant,
write the cofactor Aii of each eJement in place of the element giving the matrix
[Aii], and then make the transpose of [Aii]. The matrix adjoint to si will be
sym bolized si .
THE INVERSE OF A MATRIX
The inverse .. r:{-
1
of a matrix ,_r:. is, by definition, such that
-
Let us now loo k at the product ._r.i:ni for a square matrix of order 3:
We see that each diagonal element is the expansion of the determinant !Al
in terms of a row and its cofactors. On the other hand, each off-diagonal
MATRIX ALGEBRA 425
element is the sum of products of the elements of a certain row, say the ith,
with the cofactors of the elements of so me other row, say the jth. Such a su m
is, in fact, the expansion in the elements of the ith row with their cofactors
of a determinant in which the ith and jth rows are identical. Since we have
already seen that the value of such a determinant must be zero, all off-diagonal
of the product ._r.l._r:} are zero. lt is also easy to see that .. r:{,r:f =
.__r:i..r.{.
Thus we have the result
!Al
o o o
' /._r'
- r)r =
o
IAI
o o
- .......
o
1 o o o
=!Al
o 1 o o
o
- IAI cf
Now, referring to the definition of ._r-t, we see that
ri
=
o
o
IAI
o
o
1
That is, each element of .'1-
1
is the element of .:.i divided by !Al. Since division
by zero is not defined, only matrices for which the corresponding determinants
are nonzero can ha ve in verses. A matrix .. r:l such that 1 A 1 = O is sad to be
singular (no inverse), whereas matrices of which the corresponding deter-
minants are nonzero are said to be nonsingular. Only nonsingular matrices
can occur in representations of a group. It is also clear that since only square
matrices can have corresponding determinants, only square matrices can ha ve
1nverses.
APPENDIX IIA
CHARACTER TABLES FOR
CHEMICALLY IMPORTANT
SYMMETRY GROUPS*
1 . The N onaxial Groups
C, E
A 1
C:s E ah e, E i
A' 1 1 x, y, R:
x2, y2,
A, l 1 R)C, Ry, R:
2
Z, X}'
A" 1 -1 z, Rx, R
7
yz,xz
A. 1 -1 x,y,z
2. The Cn Groups
el E c2
A 1 1 z, R:
xz, y2, zz, xy
B 1 -1 X, y, Rx, RY yz,xz
c3 E el C3
2
e = exp (2TTi/3)
A 1 1 1 z, R:
xz+yz,z2
E
/1
E
:
(x, y)(R)C, Ry) (x
2
- y
2
, xy}(yz,.xz)
\1
e*
x2, y2, zz
xy,xz,yz
* Appendix IIA is presented in two places: ( 1) he re, in its proper location in the sequence of
appendices, and (2) as a separa te section in a pocket inside the back cover.
426
CHARACTER TABLES 427
The Cn Groups ( continued)
c4 E c4 c2 C4
3
A 1 z, R:
x2 + yz, z2
B 1 -1 l -1 X
2
- y
2
, xy
E
{:
i -1
-;} (x, y)(Rx, R,) (yz, xz)
-i -1
E Cs Cs
2
Cs
1
Cs
4
e= exp (2TTi/5)
A 1 1 l 1 l z, R: x2 + yz, z2

E
ez e2*
e*
(x, y)(Rx, Ry) (yz, xz)
e*
EJ.* E2
:2. el
e* E
(xz- y2, xy)
el*
E e*
Ez
c6 E c6 el c2 C3
2
C6
5
E = exp (2rr/6)
A 1 1 l l l z, R:
x2 + yz, z2
B 1 -1 1 -1 -1
E1
{:
E -e* -1 -e
:
(x,y)
e* -E -1 -e* (Rx, R,)
(xz, yz)
Ez
{:
-E* -E 1 -e* -E
(x2- yz, xy)
-e -e* 1 -E -e*
e, E e, C,2 C,l C,4 c,s c,6
e = exp (2TTi/7)
A 1 1 1 l z, R: x2 + y2, z2
{:
E
e2 El el* s2*
: }
(x, y)
E1
e*
e2* e3* e3 ez
(R;r, RF)
(xz, yz)
E2
{:
E2 El*
e* e
el
e*}
(x2- y2, xy)
e2* E3
E e*
el* E2
E3
{:
E3
e*
E2 e2*
E
e"*}
E3*
e ez* Ez
e*
El
el( E
e"
c4 ~ c2 ~ ~ ~ e = exp(2ni/8)
A 1 1 1 1 1 1 1 1 z, R:
x2 + y1 z2
B 1 -1 1 -1 1 -1 1 -1
E,
g
e i -e* -1 -e -i
:*}
(x, y)
(xz, yz)
e* -i -e -1 -e* (Rx, R_,.)
E "J.
g
i -1 -i 1 i -1 -i
}
(x2 - y2, xy)
-i -1 i 1 -i -1 i
E)
g
-e l e* -1 e -i
-E*}
-e* -i e -1 e* i -e
428 APPENDICES
3. The Dn Groups
D2 E C2(z) C2(y) C2(x)
A 1 l
x2, y2, z2
BJ J 1 -1 -1 z, R: xy
B2 1 -1 1 -1 y, R, xz
Bl -1 -l x, R.'f yz
D3 E 2Cl JC2
A1 1 x2+y2,z2
A2 -1 z, R:
E 2 -1 o (x, y)(Rx, R,) (x
2
- y
2
, xy)(xz, yz)
D4 E 2C4 C2(=C.a.
2
) C ~ C ~
A1 1
xl + y2, z2
A2 1 -1 -1 z, R:
BJ J -1 1 -1
x2-y2
B2 1 -1 1 -1 1 xy
E 2 o -2 o o (x, y)(Rx, Ry) (xz, yz)
Ds E 2Cs 2Cs
2
SC2
A a 1 1 1
x2 + yl, z2
A2 1 -1 z, R:
E1 2 2 cos 72 2 cos 144" o (x, y)(R:c, Ry) (xz, yz)
E! 2 2 cos 144 2 cos 72 o (x.z- y2, xy)
D E 2C
6
2Cl e! 3 C ~ JC;
A1 1
xl+yz,z2
A2 -1 -1 Z, R:
B, 1 -1 -1 -1
B2 1 -1 -1 -1 l
E. 2 -1 -2 o o (X, y)(Rx, R,) (xz, yz)
Ez 2 -1 -1 2 o o (x2- yz, xy)
4. The Cm, Groups
C!r E c.! a,,(xz) a:.(yz)
A. 1 z
x2, yz, zz
A! -1 -1 R: xy
BJ -1 -1 X, RJ. xz
B!. -1 -1 y,Rx )'Z
C3r E 2CJ 3 a ~
A a 1 z xl + y2, z2
Az 1 -1 R:
E 2 -1 o (x, y)(Rx, Ry) (x
2
- y
2
, xy)(xz, yz)
c4,. E 2C4 Cz
2u,. 2ad
A1 z
x2 + yz, z2
A2 1 -1 -1
R:
BJ -1 1 -1
x2 _ yz
B2 -1 -l 1 X)'
E 2 o -2 o o (x, y){Rx, RF) (xz, yz)
Csr E 2Cs 2Cs
2
Scrr
At J 1 z
x2+y2,z2
A1 l -1 R:
E. 2 2 cos 72" 2 COS 144 o (x, y)(Rx, R>.) (xz, j.z)
E:z 2 2 cos 1 4 4 ~ 2 cos 72? o
(xz- y2, xy)
eh E 2C6 2CJ c.! 3at, 3ad
A, 1 1 z
x2 + y2, z2
Az -1 -1 R:
Bt -1 -1 -1
B2 1 -1 1 -1 -1 1
Ea 2 1 -1 -2 o o (x, y)(Rx, R,.) (xz, yz)
E2 2 -1 -1 2 o o
(x2- y2, xy)
5.
430 APPENDICES
The Cnh Groups
Cn E Cl
;
A, 1 1 1
B, 1 -1 1
A., 1 1 -1
Bu 1 -1 -1
c3, E c3
Cll
A' 1 1 1
E'
{!
t t*
t* t
A" 1 1 1
E"
n
t
e
t* t
c4, E c4 c2
A, 1 1 1
B, 1 -1 1
E,
n
i -1
-i -1
A., 1 1 1
B., 1 -1 1
E.,
{!
; -1
-i -1
s ~ r
E Cs Cs
1
A' 1 1 1
Et'
u
t
tl
e*
t2
E,_'
1 J
el e
\l
e2+
t
A" 1 1 1
Et"
\l
e
2
t*
el
E1"
el e
el
e
e 6ft
E c6 CJ
Ag 1 1 1
B, 1 -1 1
E1
1
1 t -e
1 e -t
El,
1 -e -t
1 -t -e
A u 1 1 1
Bu 1 -1 1
El u
1 t -e
1 e* -e
E2u
1 -e -t
1 -e -e
a,
1 R:
xl, y2, =l, xy
-1 Rx, R
1
x:, .v=
-1
;:
1 X, J'
a,. s3
sJs t= exp
1 1
)
R:
xl+y
l -l
1 t
:}
(x, y) (xl- J'
1 t*
-1 -1 -1
-1 -t
-e*)
-1 -e* -E
C4
3
i s43
ah
1 1 1 1
-) 1 -1 1
-i 1 i -1
; )
-i -1
1 -1 -1 -1
-1 -1 1 -1
-i -1 -i 1
i -1 i 1
Cs
3
Cs
4
a,
Ss
)
1 1 J
t2 t* 1 t
el
E 1 t*
e
el
1
El
e
tl
1
tl
1 J -1 -1
el e -1 -e
tl
t -1 -e
t
el
-1
-el
t*
El
-1
-el
c2
Cll c6s ;
1 1 1
(
-1 1 -1 1
-J -e
e 1
-1 -e e 1
1 -e -e 1
1 -e -e* 1
1 1 1 -1
-1 1 -1 -1
-1 -e t* -1
-1 -e* t -1
1 -e -t -1
1 -e -e -1
=
(R.'C, Rr) (x:, y:)
s4
J
-1
-i\
if
-1
1
i\
-il
Ss 7
1
El
t2
t*
E
-1
-el
-el
-e
-t
sls
1
-1
E
e
-t
-t
-1
1
-e
-e*
t*
E
R:
(Rx, Rl.)
=
(x, y)
Ss
3
1
tl*
tl
E
e
-1
-el
-el
-e
-e
s6s a,.
1 1
1 -1
-e -1
Ss
1
E
E
E
E
. ,
~ }
-1
-e
-e
-t
-e
;.
-t
x2 + yl, :l
x2- yl, xy
(xz, Y=)
R:
(x, y)
sl
1
-1
R:
-e -1 -e
:}
(Rx, R
1
)
-e 1
-e 1
-e )
-e
-e*
-e
-1 -1 -1 -1
=
-1 1 -J 1
e 1
e 1
-e
(x, J')
-e
t -1 e
e -1 e
t = exp (21ri/5)
(x=, .v:)
e= exp (2TTi/6)
xl+y2,:l
(xz, yz)
(xl - J'l, xy)
. The Dnh Groups
E C
2
(z) Cz()') C
2
(x) a(xy) u(x=) a(y.:)
1 1 1 1 1 1 1 1
x2, )'2, z2
A,
Ba,
Blg
B3
1
A u
Bau
B2u
B3u
1 1 -1 -1 1 .
1 -1 -1 R:: x_v
Aa
1
A2,
B ~ a
B 2 ~
E,
A1u
A2u
B ~ u
B2u
E u
A a'
A2'
Ea'
E2'
A a"'
A2"'
Ea"'
E2"'
1
1
1
2
2
1
1
2
2
1 -1 1 -1 1
1 -1 -1 1 1
1
)
J 1 -1
1 1 -1 -1 -1
1 -1 1 -1 -l
1 -1 -l
)
-J
E 2C3 3C
2
a, 2S3 3a
11
1 1 1 l
1 -1 1 -1
2 -1 o 2 -1 o
1 J 1 -1 -J -1
)
1 -J -1 -1 1
2 -1 o -2 1 o
E
\
2C4 el
2C
2
' 2c2
;
1 1 1 1 1 1
1 J 1 -1 -1 1
1 -1 1 1 -1 1
1 -1 1 -1 1 1
21 o -2 o .o 2
1 \ . 1
J 1 1 -1
1 i 1
1 -1 -1 -1
q -1 1 1 -1 -1
1\ -1
J -1 1 -1
2 o -2 o o -2
\
!2C
S
2C
2
5
st
l
l 1 1
1 1 -1
2 cos 72 2 cos 144 o
2 cos 144 2 cos 72 o
1 1 1
1 1 -1
2 cos 72 2 cos 144 o
2 cos 144 2 cos 72 o
E 2C, 2C
3
c.l 3C
2
' 3C
2
"'
1 t J
1 J 1 1
1
1
1
)
1
~
-1
1\ - 1
1 -) 1 -1
1\ -1
1 -1 -1 1
2i 1 -1 -2 o o
2 l -1
-1 2 o o
1 1 1 1 1 1
)
1 1 1 -1 -1
1 -1 1 -1 1 -1
J -1 1 -1 -1 1
2 1 -1 -2 o o
2 -1 -1 2 o o
-1 1 -1 R, xz
-1 -1 1 Rx
\''"
--
-1 -1 -1
-1 1 1
=
1 -1 1 y
1 1 -1
"'
x2 + y2, :2
R:
(x, y)
(xl _ _..2 . Y:y)
=
{Rx, R
1
) (x:, yz)
2S
4 a,. l 2a"
!
2ad
1
1 l 1
1 x2 + y2, =
1 1 -1 -1
R:
-1
1 1 1
-]
x2 _ y2
-J 1 -1 1 xy
o
-2 o
o (R:n R,.) (xz, y:)
-1 -1 -1 -1
-1 -1 1
: 1 =
1 -1 -1
1 -1 1 -1
o
2\ o
O (x, y)

l
a,. 2Ss 2Ss
3
S a
1 1 1 1 xl + y.l, =2
1 1 1 -1 R:
2 2 cos 72 2 cos 144 o (x, y)
2 2 cos 144 2 cos 72 o (x2- y2, XJ')
-1 -1 -1 -1
-1 -1 -1
1 ,.
-2 -2 cos 72 -2 cos 144
g (Rx R,)
(xz, y;)
-2 -2 cos 144 -2 cos 72
;
2Sl 2S
6
a,
3ad 3 a ~
1
1 1 1
:
1 1 x2+y2,:2
1 1 1 -1 -1
R:
1 -1 1 - 1 -1
1 -1 1 - -1 1
2 1 -1 - o o (Rx, R,) (x=. y:)
2 -) -1 o o {x2- y2' xy)
-1 -1 -1
-
-1 -1
-1 -1 -1 - 1 1
=
-1 1 -1 -1 1
-1
]
-1 1 1 -1
-2 -1 1 J o o (x. y)
-2 1 1
~
o o
432 APPENDICES
6. The D,
11
Groups (Continued).
A a,
A2
9
Ba,
B2,
El,
Ezg
E31J
A1u
A:!u
B,.
B2u
E u
El u
EJu
1 1 1 1 1
1 1 J J -1
-1 -1 1 1 1
-1 -J 1 1 -1
2 '' -\
1
o -2 o
2 o o -2 2 o
2 - ,, i ,;i o -2 o
1 1 1 1 1 1
1 1 1 1 1 -1
1 -1 -1 1 1 1
1 -1 -1 1 1 -1
2 ,,.. 2 - v' 2 o - 2 o
2 o o -2 2 o
2 -\
1
2 \li o -2 o
1 1 1
-1 1 1
-1 -1 -1
1 -1 -1
..
o 2 v'2 - v 2 o
o 2 o o -2
o 2 - v' 2 ,; 2 o
1 -1 -1 -1 -1
-1 -1 -1 -1 -1
-1 -1 1 1 -1
) -1 1 J -1
o -2 -v'2 v2 o
o -2 o o 2
o - 2 v' 2 - , , ~ 2 o
1 1
-1 -1
1 -1
- 1 1
-2 o o
2 o o
-2 o o
--1-1 -1
- 1 1 1
-1 -1 1
-1 1 -1
2 o o
-2 o o
2 o o
7. The Dnd Groups
2
1 1 1 1
1 1 -1 -1
-) 1 1 -1
-1 1 -1 1
o -2 o o
=
(x, ;);
(fl ll\
1 , ,., 1
1 1
J -1
1 1
1 -1
x2 _ y2
X)'
(x.:, y:)
2 -1 o 2 -1 o (xl - yl' xy),
(x.:, y.:)
1 1 -1 -1 -1
1 -1 -1 -1 1
=
2 -1 o -2 1 o (x, .1')
1
1
-1
-1
2 \/2
2 o
2 -v'2
2C
5
1
1
2 2 cos 72
2 2 cos 144
1 1
1 1
1
1
1
1
o
-2
o
1
1
1
1
-1
-1
-\
1
2 -2
o 2
\/2 -2
1
2 cos 144
2 cos 72"
-1
o
o
1 1
1
1 1
-1 -1
1 -1
-1 1
o o
o o
o o
1
1
=
x ~ y)
2
2
-1
2 cos 72
2 cos 144
-)
-1 -1
(xl- J'2 xy)
(x:,y:)
2S
10
1
1
2 cos 144
2 cos 7ZO
-1
-1
1
-1
o
o
=
=
2 2 cos 72
2 cos 144
2 cos 144
2 cos 72';)
-1
o
o
-2 -2 cos 72 -2 cos 144
-1
1
o
o
(x, _1')
-2 -2 cos 144 -2 cos 72
(x:, y:)
(xl- y2, xy)
(x:, y:)
(xl - )'2' xy)
CHARACfER TABLES 433
7. The Dn,-J Groups (Continued).
J 1
1 J
1 -1
1 - 1
2 \/3 1
2 l -1
2 o -2
2 -1 -1
2 -V
1
J
8. The Sn Groups
1
1
-J
-1
o -1
-2 -1
o 2
2 -1
o -1
s4 E s4. el S4.
3
A
B -1
1
l
-1
-l
-v'j
1
1 1 J
1 -1 -1
1 1 -1
1 -1 1
-2
2
o -2
-1 2
o
o
o
o
o
o
o
o
\-'3 -2 o o
=
(x, J')
(xl -- _..2, xy)
(x:. y:)
E
1 -1
( J i
ti -i
(x, y); ( Rx., Rr) (x=, yz)
-1 -i\
-1 iJ
s6 E CJ
eJ.l i S6
5
s6 E = exp (27Ti/3)
Ag R:
xl + yl, zl
{.
e*
e*l
(x2- J'2, xy);
E,
J J
e e
(R.", R-r)
\1
E* e E*
e J
(xz, J'Z)
A, 1 -1 -1 -1 z
E,
(:
e e* -1 -E -E*\
(X, y)
e* E -1 -e*
-E J
Ss E Ss c4 Sa
3
e2 Sa
5
C4
3
Ss
7
E = exp (27Ti/8)
A l 1 1 R:
xl + .vl, zl
B 1 -1 -l 1 -1 -1 z

E i -e* -1 -E -i
e*l
(x y)
E.
' . ,
e* -i -E -1 -e* i
e J
(Rx., R,.)
E2
i -1 -i i -J -i
\
(xl - yl' xy)
l 1
-i -1 i 1 -i -1 i
1
EJ
1 1
-e* -i E -1 e* i
-E}
(xz, yz)
ll
-e i E* -1 e -i -E*
. The Cubic Groups
T E 4C
3
4C
3
2
JC2 e= exp (27Ti/3)
A t 1 xl+ yl+ :l
E{
1 e e* 1
(
2
:2 _ xl _ yl,
l e e 1
xl- yl)
T 3 o o -1 (Rx., R.", R:); (x, y,:) (xy, x:, y:)
434 APPENDICES
9. The Cubic Groups (Continued).
T,. E ..te,
..te'
3C:! i ..JSh 3a
1
, 1: =
1 1 1 1 1 1 1 1
+ + :z:!

{:
1:
;*
1 1 r. r.
:}
( 2.:;: - .r;: - .r;: - y!)
t: t: 1 1
;*
1:

3
() {)
-1 1 o
()
-1 (R .. R,, R:) (xz. y:. xy)
A, 1 1 1 1 -1 -1 -1 -1
E.,
{:
t: t: 1 -1 -; -r.*
-1}
t;*
r. 1 -1
-e -;
-1
T,, 3 o o -1 -1 o
()
1 (x. y. z)
T,, E HC1 a,,
A, 1 1 1 1 1 + y
2
+ z
2
A.:!
l l l -1 -l
E 2 -l 2
()
o (2.::! - .\-1 - . \.-1 -
T, 3
(}
-l l -1 (R" R,. R:)
T.:! 3 o -1 -1 J (x, y. :) (xy. x:, y:)
o E xcl JC2< =en hC
1
A, l 1 1 1 1
x1 +y.:! + =:!
A:! 1 1 l -1 -1
E 2 -1 2 o o (2:! - x
2
- y
1
, x
2
- y
2
)
T, 3 o -1 1 -1 (R,. R,. (x, y. z)
1 3 o -1 -1 J. 1 1 (.ry. XZ, y:)
o,, E HC\ 6C
2
3C
2
(=Ci> i HSt. 3ah 6nJ

1 1 1 1 1 1 1 1 1 1 x.! + y.:!+=:!

1 1 -1 -1 1 1 -1 1 1 -1
E e
1 -1
()
o 2 2
() -1 2
() (2.::! - x:! - y.!. - y
2
)
3
(}
-1 1 -1 3 l o -1 -1 (R,. R, . R=)
T.!.:
3
()
1 -1 -1 3 -l o -1 1 (xz, y.:. xy)
A tu
1 1 1 1 1 -1 -1 -1 -1 -1
Az, 1 1 -1 -1 1 -1 1 -1 -1 1
E, 2 -1 o o 2 -2 o 1
_..,
()
T,,, 3
()
-1 1 -1 -3 -1 o 1 1 (.r. y .:)
T.!ll 3 o 1 -1 -1 -3 1 o 1 -1
::Jroups CJu and Da:Jh for Linear Molecules
E 2 C ~

OOau
+
1 1 1 z xl + yl, zl
1 1 -1 R:
2 2 cos C1> o (x, y); CR.x. R
11
) (xz, yz)
2 2 cos 2Cl> o (xl - yl' X)')
2 2 cos 3<1> o
E 2Cao
0
ex>a
11
2Sao
0
ooC
2
1 1 1 1 1 1 xl+ y2, zl
1 1 -1 1 1 -1
R::
2 2 cos <1> o 2 -2 cos cJ> o (Rx, R
11
) (xz, yz)
2 2 cos 21 o 2 2 cos 21> o (xl- yl, xy)
...
1 1 1 -1 -1 -1 z
1 1 -1 -1 -1 1
2 2 cos 1> o -2 2 cos cJ> o (x, y)
2 2 cos 2Cl> o -2 -2 cos 21> o
1
1
.

T
h
e

I
c
o
s
a
h
e
d
r
a
l

G
r
o
u
p
s
*

!
,

E

1
2
C
,

1
2
C
s
2

2
0
C
3

1
5
C
z

i

1
2
S
l
o

2
0
S
6

1
5
0
'

A
,

1

1

1

1

1

1

1

1

1

1

x
2

+

y
l

+

z
l

T
t
,

3

!
O
+

v
i
s
>

H
1
-
V
5
)

o

-
1

3

1
0

-
v
'
s
>

1
0

+

v
'
5
>

o

-
1

R
z
)

T
1
,

3

!
O

-
v
'
s
)

!
O

+

v
'
s
>

o

-
1

3

}
(
t

+

v
5
)

!
O

-
v
'
5
)

.

o

-
1

G
,

4

-
1

-
J

l

o

4

-
1

-
1

1

o

H
,

5

o

o

-
1

1

5

o

o

-
1

t

(
2
z
l
-
x
l
-
y
l
,

x
z

_

y
l
,

x
y
,
y
z
,
z
x
)

A
u

l

1

1

1

1

-
1

-
1

-
1

-
1

-
1

T
1
u

3

i
O

+

v
'
s
>

!
O

-
v
s
>

o

-
1

-
3

-
H
1
-
,
1
5
>

+

v
'
s
>

o

1

(
x
,

y
,

z
)

T
z
u

3

l
O

-
v
'
s
>

!
O

+

v
5
)

o

-
J

-
3

-
!
O
+

v
'
s
>

-
!
O
-
v
'
5
>

o

1

G
u

4

-
1

-
1

1

o

-
4

l

1

-
1

o

H
u

S

o

o

-
1

l

-
5

o

o

1

-
1

'
'
f
o
r

t
h
c

p
u
r
e

r
o
t
a
t
i
o
n

g
r
o
u
p

/
.

t
h
e

o
u
t
l
i
n
e
d

s
e
c
t
i
o
n

i
n

t
h
e

u
p
p
e
r

l
c
f
t

i
s

t
h
c

c
h
u
n
u
c
t
e
r

t
h
e

g

s
u
h
s
c
r
i
p
t
s

s
h
o
u
l
d
,

o
f

c
o
u
r
s
c
.

b
e

d
r
o
p
p
e
d

a
n
d

(
x
,

y
.

z
)

a
s
s
i
g
n
e
d

t
o

t
h
e

T
1

r
e
p
r
c
s
e
n
t
u
t
i
o
n
.

:
>

"
'
t
t

"
"
C
C

m

z

o

m

e
n

oh
At,
A2,
E,
Tt,
T2,
A,,
A2u
E,
r.,
T2,
APPENDIX IIB
CORRELATION TABLE FOR
GROUP Oh
This table shows how the representations of group 0
11
are changed or decom-
posed into those of its subgroups when the symmetry is altered or lowered.
This table covers only representations of use in dealing with the more common
symmetries of complexes. A rather complete collection of correlation tables
will be found as Table X-14 in Molecular Vibrations by E. G. Wilson, Jr.,
J. C. Decius, and P. C. Cross, McGraw-Hill, New York, 1955.
o D411 D211 C.a.u c2, DJd Dl c2,
A
Al At, At At Al
A1"
At A,
A2 A2 Bt, Bt Bt A2 A2, A2 B,
E E A1
9
+ B
19
A
1
+ B
1
A 1 + B
1
A1+A2 E, E A,+ B
9
r. Tt A2,+E, A2+E A2+E A2+B
1
+B2 A2+E A
9
+2B,
Tl T2 B2,+E, B2+E B2+E A 1+B1 +B2 A 1,+E, At+E 2A
9
+ B,
Al Al
B A1 A2 A., Al A,
A2
A
Blu A1 B2 A A2u A2
B,
E E A
1
,+ B1, At+ Bt A2+ B1 A1 + A2
E, E A,+ Bu
Tt T2 A2,+ E, B2+E At+ E A 1 + B1 + B2 A2u+ E, A2+E
A,+2B,
T2 r. B2u+Eu A2+E Bt+E A2 +B
1
+B2
A
1
+E 2A,+ Bu
437
APPENDIX 111
SOME REMARKS ABOUT THE
RESONANCE INTEGRAL p
On pages 148-149 we described a seemingly straightforward and obvious
method for evaluating the integral p. This involved equating the so-called
experimental or empirical resonance energy of benzene, defined as the energy
difference between "real" benzene and Kekul benzene, to a multiple of p.
This procedure is widely used and is of at least empirical validity, as shown
by the fact that it gives essentially the same value of P in various molecules.
Sorne results illustrative of this point are given in the following table.
Table of Experimental Resonance Energies and the Derived Value of p
Compound
Benzene
Diphenyl
Naphthalene
Anthracene
Phenanthrene
Observed Theoretica) Resonance
Resonance Energy,
Energy, Hckel
(kcal/mol)
36
70-80
75-80
105-116
110-125
Approximation
2{1
4.38/1
3.68/1
5.32{1
5.45P
Vertical Resonance Energy
-P
18
-17
-21
-20
-21
\n the process }ust described we are actual\y considering not only the energy
of delocalization but also the energy required to stretch and compress the
C-C bonds in Kekul benzene from the lengths 1.54 and 1.34 to the
438
RESONANCE INTEGRAL 439
common length 1.39 found in real benzene. Thus the experimental reso-
nance energy Rexp is related to the true delocalization energy, or vertical*
resonance energy and to the combined energies of bond stretching, bond
compression, and other changes (as, for instance, in repulsive forces between
nonbonded atoms) collectively denoted by E e, by the following equation:
Rcxp = R,. + Ec
Now it can be argued that the MO calculation actually applies to and that
in order to evaluate p we should equate the multiple of p given by theory
to R,. and not Rcxp In order todo this, Ec must .be calculated. For benzenet
Echas been estimated to be -37 kcal/mol. From this we estima te that R,, =
36 + 37 = 73 kcal/mol, and hence P must be about 37 kcal/mol, or about
twice the value used in the calculations of experimental resonance energies.
Thus, for estimating the actual separations of energy levels in the real mol-
ecule, this larger value of P should be more nearly correct.
The Spectroscopic P
One might reasonably observe that the Hckel assumption of negligible over-
lap between neighboring prr. orbitals seems physically unlikely, sin ce just su eh
overlap would be required to allow formation of multicenter molecular or-
bitals. This point has been considered in more detail by Mulliken, Rieke, and
Brown,t and we summarize here their discussion.
Without the assumption that all S; = 6; our secular equations will take
form (a) rather than form (b), which has been used in the Hckel approxi-
mation:
(a)
HJJ -E Hr1 - ES,2
=0
H'!., ES21 H11- E
(b)
Hn E
Ht'!.
=0
H21
Hn - E
Thus the true value of Pis not Hr
1
= H
21
but Hr!. - ES
1
"1 = H
11
- ES"lt It
can be shown that the value of E (referred to the true zero of.energy, which
is the energy of an electron when completely separated from the molecule)
is -50 e V (1 e V = 23.06 kcallmol). Moreover, S
12
must have a value between
0.2 and 0.3 so that ES
12
must be between 10 and 15 eV. Mulliken et al. note
The term vertical is used with respect to a diagram in which we plot energy on the ordinate
against internuclear distances on the abscissa. Thus, if the internuclear distances are held the
same in the hypothetical. nonresonating and the real. resonating molecule, the change in energy
is a purely vertical change on the plot.
t See R. S. Mulliken and R. O. rarr, J. Chem. 19, 1271 (1951 ).
:t R. S. MuJJiken, C. A. Rieke, and W. G. Brown, J. Am. Chem. Soc., 63, 41 (1941).
440 APPENDICES
that the integral H
12
should also be of about this magnitude, which is consistent
with the fact that p turns out to be only 1-2 eV, that is, a small difference
bet\veen two larger numbers.
lt is therefore rather surprising that p remains substantially constant from
one molecule to another sin ce relatively small variations in Hr
1
or Er
1
could
cause relatively large variations in p. Empirically, however, this approximate
constancy is observed.
On the other hand, we should perhaps not be too surprised if values of fl
derived from different kinds of measurements do not agree exactly, and indeed
this is the case. We ha ve se en abo ve that P obtained from the vertical resonance
energy of benzene is about 37 kcallmoJ, whereas Platt* first showed that the
best overall fit to the spectra of benzene and other unsaturated hydrocarbons
was obtained in the framework of the Hckel approximation by taking fJ to
be 55-60 kcal/mol. This high value, which has subsequently been widely
adopted to estmate actual differences in energies between MOs, is generally
known as the spectroscopic value of p.
J. R. Platt, J. Chem. Phys., 15.419 (1947).
APPENDIX IV
THE SHAPES OF f ORBITALS
When the wave equation for a hydrogen-like atom is solved in the most direct
way for orbitals with the angular momentum quantum number 1 = 3 .. the
following results are obtained for the purely angular parts (i.e., omitting all
numerical factors):
tp
0
: (5 cos
3
0 - 3 cos O)
l/1:
1
: sin 0(5 cos
2
O - 1 )e:;,.
l/1 -:
2
: (sin
2
O cos O)e:uq,
'1' :
3
: (sin
3
O)e-rJ;,p
m
1
=O
m
1
= 1
1n
1
= 2
1n
1
= 3
The seven functions are grouped into sets having projections of the orbital
angular momentum on the z axis of O, 1, +2, and 3. Each of the functions
in the pairs with m
1
equal to 1, 2, and 3 is complex as written above,
but by taking linear combinations of each pair, for example,
and
the imaginary _parts are eliminated. In this way the seven real, normalized
(to unity) orbitals listed in Table AIV.l are obtained. As given in the table,
the f orbitals are in a convenient form for problems involving only a single
high-order symmetry axis. For instance, in treating bis(cyclooctatetraene)-
metal compounds, where the point group is D
8
h, we note that the orbitals are
already grouped into sets belonging to the representations A
2
," E
11
" E2u, and
441
442 APPENDICES
TABLE AIV.l
True Simplified Normalizing
Polynomial Polynomial Factor Angular Function
z(5z
2
- 3r)
z3
\17/n
5 cos
3
O - 3 cos O
4
x(5z
2
- r) xz
2
\f42/n
sin 0(5 cos
2
O - 1) cos cp
8
y(5z
2
- r)
\1'42"/n
sin 0(5 cos
2
O - 1) sin q,
yzl
8
z(xy)
YI05/n
sin
2
O cos O sin 2cp
4
z(x
2
- y ~
Y105/n
sin
2
O cos O cos 2qJ
4
x(x
2
- 3y
2
) V70/n
si n
3
O e os 3cp
8
y(3x
2
- y
2
)
v=;o;
sin' O sin 3lf>
8
E
311
However, for problems involving cubic symmetry, the functions given in
Table AIV.l are awkward to use since they do not directly form triply de-
generate sets, despite the fact that the entire set of f functions spans the
representations A
21
, T
1
u, and T
211
in the group Oh.
A set of functions directly useful in problems with cubic symmetry can be
obtained by taking the following linear combinations of those in Table AIV. 1.
A2t,:
It is these functions that are given in Table 8.1.
APPENDIX V
CHARACTER TABLES FOR SOME
DOUBLE GROUPS
c4
c43
c2 2c; "C''
- 2
D ~
E R c43R C
4
R C
2
R 2C;R 2CiR
rl ~
1 1 1 1 1 1 l
r2 A; 1 l 1 1 1 -1 -1
r3 B't 1 1 -1 -1 1 1 -1
r4
B; 1 1 -1 -1 1 -1 1
rs E; 2 2 o o -2 o o
r6 E
2 -2
J2 -J2
o o o
r, E; 2 -2 --\/2 Ji
o o o
4C
3
4C32
JC2 3C
4
3C43 6c;
O' E R 4C
3
2
R 4C
3
R 3C
2
R 3C
4
3
R 3C
4
R 6C;R
rl
A's l 1 l l 1 1 1 1
r2
~ 1 1 1 1 J -1 -1 -)
r3 E ~
2 2
-) -1 2 o o o
r4 T ~
3 3 o o -1 1 l -1
rs
r; 3 3 o o -1 -1 -1 1
r6
E; 2 -2 1 -1 o
J2 -J2
o
r, E; 2 -2 1 -1 o
-J2 J2
o
rs
G' 4 -4 -1 1 o o o o
443
APPENDIX VI
ELEMENTS OF THE g MATRIX
Each element of a g matrix is identified by a double subscript that specifies
the kinds of internal coordinates involved, using r for interatomic distances
and l/J for angles. Thus there will be three main classes of g matrix elements:
g,,. gq,,p. g,r/1. Within each of these classes subclasses arise based on the number
of atoms which are shared by the interna) coordinates; a superscript specifies
this number of shared atoms. For example g'4 is a matrix element involving
a bond, r, and an angle, c/J, with two atoms in common; the common atoms
are those forming the bond, r, and one side of the angle. Even this notation
is not entirely unequivocal, and sorne further distinctions must be made. The
cases to be treated he re are represented and assigned symbols in Figure A VI.l.
In these sketches the atoms common to both coordinates are represented by
double circles.
In sorne of the formulas for the g-matrix elements it will be necessary to
use two types of dihedral angles, as well as cp and r. In the expressions for
and g:r/1{2) there occur dihedral angles denoted 1/f afl' while in the expres-
sions for and g:,,(l) a dihedral angle denoted r will
be found.
The '11-type dihedral angles are defined by the fol1owing diagram:
p

t .. p
a '-- / Y
c/Jdy
444
g MATRIX 445
. - e os cp Oti - cos cP oJ{J cos cP /M7
COS 1/f uJI - ../,.
Sin cPocS(I Sin o/fM
The r-type dihedral angle is defined as that between the planes formed by
atoms 1, 2, 3 and 2, 3, 4 \Vhen atoms 1, 2, 3, 4 are bonded in sequence. When
only two such planes are involved, as in and r \vithout sub-
scripts denotes the unique dihedral angle. In the other two cases, subscripts
denote the first atom of the set defining the first plane and the last atom of
the set defining the second plane. For example, in r
25
denotes the
angle between the planes defined by the overlapping ato m triplets 2, 1, 4 and
1, 4, 5. It is necessary to ha ve a sign convention for r angles. These angles
are restricted to the range - n < r n and r
1
_. is positive if, on viewing the
atoms along the bond 2, 3 with 2 nearer the observer, the angle from the
projection of 2, 1 to the projection of 3, 4 is measured clockwise.
g;,.

g;_
g,..(l)
g!.( 1)
g..(2)
Figure A VI.l Diagrams showing how various common g-matrix elements are defined.
446 APPENDICES
In the following expressions Jl; represents the reciprocal mass. (in atomic
units) of the ith atom, and P;i represents the reciprocal of the distance between
the ith and jth atoms.

1)

Jll + Jl"!.
Jl.l cos rP
PhJt1 + + (PT2 + - 2pi'!.Pn cos e/> )112
(PI2 cos '1'31-t)Jll + [(Pl2 - Pn cos c/J1n - P24 cos c/J124)P12
x cos lf/
314
+ (sin c/>
123
sin

sin
2

+ cos r/>324 cos
- P12 cos r[ (PI2 - Pl4 cos cP1)Jl1 + (PI2 - P2J cos cP2)Jl.:!]
-
r
25
sin r
3
"' + cos r
25
cos r
34
cos c/>
1
)tt
1
Pl'!llt [ .
S
n (stn r/>21-l cos c/>
41
:. cos
- stn c/>'!.
15
cos r
35
)p
14
1
+ (sin c/>
2
1s cos qJ-t1s cos r
35
- sin qJ'!.t4 cos r
34
)p
15
]
Jl1
A. [ ( cos c/J41; - cos

cos rf>:.us - cos cP:!I-t cos cP:!IS
stn

s1n 'P.,Is
+ COS c/J213 COS cP214 COS cPJis)PI2PI3
+ ( cos

- cos

cos l/>
5
n - cos c/>
214
cos qJ
213
+ cos cP21s cos r/>214 cos r/>:.tJ)P12P1s
+ (cos c/>
215
- cos c/>
312
cos cP:us - cos l/>
412
cos c/> ..
15
+ cos cP.JIJ cos lP-t12 cos cP31s)PI-tPI3
+ ( cos c/>!1.1 - cos c/>:.12 cos cP513 - cos cP-tl2 cos c/J-413
+ cos r/>
415
cos c/>
41
2 cos

- P2JIL"!. sin e/>
P13J11 sin r/J
1
cos r
- (p
13
sin r/>
213
cos + p, .. sin r/>
214
cos

APPENDIX VII
448 APPENDICES
STEREOGRAPHIC PROJECTIONS
FOR THE 32 CRYSTALLOGRAPHIC
POINT GROUPSa
Triclinic Monoclinic (1st setting) Tetragonal
f--
X
o CDO EBE9
1 2 4
x -
00
ffi_E9
(even)
m(=2) 4
X
(])
WO EBE9
(even)
plus
centef
end
x
I
2/m
4/m
(odd)
Monoclinic (2nd setting) Orthorhombic
1--
con
CDfi\
EB
X2 11(--6
--
\ 1"- 1
222'-.LJ
'- .,

422
X m
CDCD CDEB E9
m mm2 4mm
X2
E9f16
(even)
or
- -
\.IY
X m
- ..
(even)
42m
X2
CDffi
CDfi\
E9 ;_,
011
#. 1
X m
1 "/ \
plus
t-t-

CenW
'-.l/
1 .,
and
........
X m
2/m
m mm
4/mmm
(odd)
POINT GROUPS 449
Trigonal Hexagonal Cubic
3 6 23
_
x
(even)
6
X m
3m 6mm
APPENDIX VIII
THE 230 SPACE GROUPS
Crystal Crystal
Space Groups
System Class
Triclinic
1 PI
1 Pl
Monoclinic 2 P2 n, A2(C2)
m Pm Pa(Pc) Am(Cm) Aa(Cc)
2/m P21m P2,1m A2/m(C2/m) nta(P2/c) P2,!a(n,tc) A2/a(C2/c)
222 P222 P222, P2,2,2 P2,2,2, C222, C222
F222 1222 /2,2,2,
mm2 Pmm2 Pmc2, Pcc2 Pma2 Pca2, Pnc2
Pmn2, Pba2 Pna2, Pnn2 Cmm2 Cmc2,
Ccc2 Amm2 Abm2 Ama2 Aba2 Fmm2
Orthorhombic Fdd2 lmm2 lba2 lma2
mm m Pmmm Pnnn Pe cm Pban Pmma Pnna
Pmna Peca Pbam Pe en Pbcm Pnnm
Pmmn Pbcn Pbca Pnma Cm cm Cm ca
Cm mm Ce cm Cm m a Ceca Fmmm Fddd
/mm m Iba m Jbca lmma
4 P4

/4 14,
4
p
/4
4/m P41m P41n P4 ..
1
1n 14/m /4
1
/a
Tetragonal
422 P422 P42,2 P4
1
22 P4,2
1
2 P4
2
22

P4
3
22 P412
1
2 /422 /4,22
4mm P4mm P4bm P4:!cm P4:!nm P4cc P4nc
P4
2
bc /4mm 14cm /4
1
md 14,cd
42m P42m P42c P42,m P42,c P4m2 P4c2
P4b2 P4n2 i4m2 i4c2 i42m i42d
4/mmm P41mmm P41mcc P4/nbm P41nnc P4/mbm P41mnc
P4/mnm P4/ncc P4
2
1mmc P4'11nbc P4/nnm
P4"Jimbc P4
2
1mnm P4
2
/ncm 14/mmm 14/mcm
14,/amd /4
1
/acd
450
SPACE GROUPS 451
3 P3 P3, P3! R3
3
pj
RJ
32 P312 P321 P3,12 P3,2l P3
2
12 P3!21
R32
3m P3m1 P31m P3cl P3Jc R3m R3c
3m P31m PJlc PJml PJct RJm R)c
Trigonal
6 P6 P6, P6, P6! P 6 ~ P6,
hexagonal
6 P6
6/m P6/m P6,1m
622 P622 P6
1
22 P 6 ~ 2 2 P 6 ~ 2 2
P6 "')?
-'--
P6J.2
6mm P6mm P6cc P6,cm P,mc
6m2 P6m2 P6c2 P62m P62c
6/mmm P61mmm P61mcc P 6 ~ m c m P6,1mmc
23 P23 F23 /23 P2,3 11,3
m3 Pm3 Pll3 Fm3 Fd3 1m3 Pu3
la3
432 P432 P4;!31 F-B2 F..t,32 /..t32 P4,32
Cubic
P.f,32 14,32
..J3m P-:i3m F43m 143m
p],
fJc [:i3d
m 3m Pm3m Pn31t Pm3n P11Jm Fm3m Fm3c
Fd3m Fd3c /m 3m la 3d
APPENDIX IX
READING LIST
Presented here is an annotated list of references to more complete or more
advanced treatments of the tapies covered in this book. References to basic
theory relevant to aH the chapters in Part 1 are given first, and then references
pertinent to each of the areas of application covered in Part 11 are presented
separately for each chapter.
Basic Theory
Numerous books on quantum mechanics present the fundamentals as ap-
propriate for chemicaJ problems. Severa) which would provide a good basis
to precede the study of this book are listed below. It is not necessary, however,
to be conversant with aH of the material in the more advanced of these books
in order to study the present one.
Quanttun Chemistry, W. Kauzman, Academic Press, New York, 1957.
Valen ce Tlreory, 2nd ed., by J. N. Murrell, S. F. A. Kettle, and J. M. Tedder, John
Wiley & Sons, New York, 1970.
Quantlun Meclzanics in Chenzistry, 3rd ed., M. W. Hanna, Benjamin/Cummings,
Menlo Park, CA, 1981.
Quantutn Chenzistry, D. A. McQuarrie, University Science Books, Mili Valley, CA
1983.
Many books cover the application of group theory to quantum mechanics
at a more advanced and sophisticated level than this one, and the aspiring
452
REAOING LIST 453
theoretician will find it necessary to consult sorne of these. A few of the best
are listed below:
Group Theory, E. P. Wigner, Academic Press, New York, 1960.
Group Tlteory and lts Applications to Plzysical ProblentS, by H. Hammermesh. Ad-
dison-Wesley, Reading, MA, 1962.
Synunetry: An lntroduction to Group Theory and lts Applications. R. McWeeny.
Pergamon Press. London. 1963.
Group Tlzeory and Quantunt Mechanics, M. Tinkham, McGraw-Hill, New York .. 1964.
Group Theory and lts Physical Applications, L. M. Falicov, The University of Chicago
Press, Chicago, IL, 1966.
Chapter 7
Information on the details of various approximations besides the simple Hckel
approximation for organic and heteroorganic molecules and on applications
of these calculations will be found in:
Molecular Orbital Theory for Organic Che1nists, A. Streitwieser. Jr., John Wiley &
Sous, New York, 1961.
The Theory of the Electronic Spectra of Organic Molecules, J. N. Murrell, Methuen.
London, 1963.
The Molecular Orbital Theory of Conjugated Systenzs, L. Salem, W. A. Benjamin.
New York, 1966.
The Jvlolecular Orbital Theory of Organic Chenzistry, M. J. S. Dewar. McGraw-Hill,
New York, 1969.
Approxilnate Molecular Orbital Theory, J. A. Pople and D. L. Beveredge. McGraw-
Hill, New York, 1970.
A tabulation of the results of Hckel calculations the n systems of many
organic molecules is available:
Dictionary of n-Electron CalculatiottS, C. A. Coulson andA. Streitwieser, Jr., W. H.
Freeman, San CA, 1965.
The symmetry-based seJection rules for cyclizations and other rearrange-
ments and reactions of otganic molecules developed by Woodward and Hoff-
mann are covered in detail in their book:
Conservation of Orbital Synunetry, R. B. Woodward and R. Hoffmann, Academic
Press, New York, 1970.
Chapter 8
A few articles and research papers dealing with the quantitative aspects of
MO calculations for complexes and organometallic compounds are:
454 APPENOICES
C. J. Ballhausen and H. B. Gray, Inorg. Chent., 1, 111 (1962) [VO(H
2
0)s]
2
+.
H. B. Gray and C. J. Ballhausen, J. Am. Clzem. Soc., 85, 260 (1963) (square
MX7- complexes).
L. L. Lohr and W. N. Lipscomb, lnorg. Chem . 2, 911 (1963). (An approach to Jahn-
Teller effects va semiempirical MO theory.)
R. F. Fenske, lnorg. Cltenz., 4, 33 (1965) (the T i ~ ion).
R. S. Berry, M. Tamres, C. J. Ballhausen, and H. Johnson, Acta Chenz. $candq 22,
231 (1968) (AB
5
molecules).
Several semiempirical MO calculations on the electronic structures of
(C
5
H
5
)
2
M and (C6H
6
)
2
Cr have been published:
E. M. Shustorovitch and M. E. Dyatkina, Dokl. Akad. Nauk. SSSR, U8, 1234 (1959);
J. Strukt. Che1n. ( USSR), 1, 98 (1960).
J. P. Dahl and C. J. Ballhausen, Kong. Danske Vidensk. Selsk., Mat. Fys. Medd.,
33, No. 5 (1961).
R. D. Fischer, Theoret. Chinz. Acta, 1, 418 (1963).
J. H. Schachtschneider, R. Prins, and P. Ros. Inorg. Chinz. Acta, l, 462 (1967).
M. F. Rettig and R. S. Drago, J. Ant. Chem. Soc., 91, 3432 (1969).
An MO treatment of C
6
H
6
Cr(C0)
3
is reported by:
D.G. Carroll and S. P. McGiynn, lnorg. Chetn., 7, 1285 (1968).
The type of symmetry-based MO analysis employing extended Hckel cal-
culations which R. Hoffmann and his students employ is well illustrated in
the following two important papers:
D. L. Thorn and R. Hoffmann, Inorg. Chem., 17, 126 (1978).
S. Shaik, R. Hoffmann, C. R. Fisel, and R. H. Summerville, J. A1n. Chent. Soc.,
102, 4555 (1980).
Chapter 9
The best general texts for chemists are:
lntroduction to Ligand Field Theory, C. J. Ballhausen, McGraw-Hill, New York,
1962.
lntroduction to Ligand Fields, B. N. Figgis, John Wiley & Sons, London, 1966.
Basic Principies of Ligand Field Theory, H. L. Schlafer and G. Glieman, Wiley-
lnterscience, New York, 1969.
A very sophisticated treatise is
The Theory of Transition Metal lons, J. S. Griffith, Cambridge University Press,
Cambridge, 1961.
REAOING LIST 455
Chapter 10
The definitive book on theory for discrete molecules is:
Molecular Vibrations, E. B. Wilson, Jr., J. C. Decius, and P. C. Cross, McGraw-Hill,
New York, 1955.
The inftuence of a crystalline environment on the vibrational spectrum of
a molecule is very lucidly covered by:
W. Vedder and D. F. Hornig, in Advances in Spectroscopy, Vol. II, H. W. Thompson,
Ed., Interscience, New York, 1961, p. 189.
Chapter 11
The following books by the late Martin Buerger can be highly recommended
for clear, detailed presentation of fundamentals. Sadly, sorne are out of print,
but most libraries have them:
X-Ray Crystallography, John Wiley & Sons, New York, 1942.
Elementary Crystallography, John Wiley & Sons, New York, 1956.
Vector Space, John Wiley & Sons, New York, 1959.
Crystal Structure Analysis, John Wiley & Sons, New York, 1960.
lntroduction to Crystallography, Krieger, New York, 1977.
Sorne other useful books on crystallography are:
X-Ray Analysis and the Structure of Organic Molecules, J. D. Dunitz, Cornell Uni-
versity Press, lthaca and London, 1979.
Crystal Structure Analysis, A Printer, 2nd ed., J. P. Glusker and K. N. Trueblood,
Oxford University Press, New York, 1985.
Structure DetemJination by X-Ray Crystallography, 2nd ed., M. F. C. Ladd and
R. A. Palmer, Plenum, New York, 1985.
Essentials of Crysta/lography, D. McKie and C. McKie, Blackwell Scientific, Oxford,
1986.
INDEX
AB, (planar, D'lh) molecules, a MO's. 219
sigma hybrid orbitals. 226
AB., (pyramidal. C,'") molecules, vibrational
analysis of. 328
AB .. (planar, D .. ") molecules, a MO's for.
219
sigma hybrid orbitals for. 226
AB., {tetrahedral, T
11
) molccules. a MO's for.
209
n MO's for. 227
AB, ( octahedral, 0
11
) molecules,
n MO's, 228
a MO's, 214
Abelian group, 7
Aliene. symmetry of. 57
Allyl group. LCAO-MO treatment of, 181
Ammonia molecule:
symmetry of, 57
vibrations of. 333
Angular momentum, coupling of, 258
orbital. :!56
spin. 256
Asymmetric unit. 348. 350
Atomic orbitals. 205-209, 254
Atomic wave functions. 205-209
Benzene. LCAO-MO's of, 142
resonance energy. 148
B H ~ - . 231
Bicyclooctatriene. LCAO-MO treatment of.
166
Bond order. definition in MO theory. 165
Boranes. 231
Bragg equation, 402
Bravais lauices. 369
Butadienc, LCAO-MO's for. 196
cyclization to cyclobutene, 198
Cage compounds. 230
Capping, of clustcrs, 237
e, axis, dcfinition, 22
Centering:
of 20 lattices. 356
of 3D lattices. 369
Center of invcrsion. 22
Character tables, compilation of. Appendix
II, 426-436
definition. 90
for' double groups, 297. 443
Chirality:
of crystals, 408
of molecules, 34
of space groups, 41 O
Classes:
definition, 13
orders of, 15
of symmetry operations, 50
Cluster compounds. 230
Complexes, magnetic properties, 279
spectra. 279
Configuraton interaction. 176
in naphthalene spcctrum, 176
457
458 INDEX
Conjugare group elements. 13
Conrotatory rng opening, 199
Cope rearrangement. 203
Correlation diagrams, for ligand field
splitting. 265
Correlation field, 344
Correlation table, for 0
11
and its subgroups.
437
Crystal field theory. 282
Crystallographic point groups, 378
table of. 379
projection diagrams. 447
Crystal systcms, 375, 380
Cuhoctahedron. 49
Cyclic groups:
definition. 10
representations for. 95
Cyclobutadiene, LCAO-MO treatment of.
ISO
Cycloheptatricnyl group, LCAO-MO
treatment of, 154
Cyclooctatetraene, LCAO-MO treatment of,
156
symmetry of, 59
1 ,3,5,7-tetramethyl, 58
Cyclopentadienyl group. LCAO-MO
treatment of, 152
metal complexes of, 241. 250
Cyclopropenyl gro u p. pi orbitals for. 124
Delocalization energy. definition. 148
of benzene, 148
Descending symmetry. method of. 270
Determinants. 421
Diels-Aider reaction, 195
Direct product:
applications, 107
definition, 105
representations of, 106
Disorder, in crystals, 407
of nearly-planar molecules . ..J.08
Disrotatory ring opening, 199
Dodecahed ron:
pentagonal. 46
rhombohedral, 62
symmetry of pentagonal, 48
trigonal, 62
d orbitals, wave functions. 207
splitting:
by octahedral field, 266, 282
by tetrahedral field, 266, 276, 283
Double groups, '297
Eigenfunctions. lOO
orthonormality of. 101
Eigenvalues, 1 OU
degenerate. 101
Electron count, of a cluster. 238
Energy level diagrams. 138
ligand field for d" configurations. '276
for pi molecular orbitals, ISO
Tanabe-Sugano type, 276
Equivalent atoms. 32
Equivalent symmetry elements, 32
Ethylene. dimerization of, 190
Exclusion rule, in vibrational spectra. 338
Fermi resonance. 338
Ferrocene:
MO treatment of. 241
symmctry of. 60
F-G matrix method. 317
applied:
to H;!O. 319
to NH,. 328
F matrix. 320
f orbitals. 441
Force constants. 320
4n + 2 rule. 158
Fundamental transition (vibrational),
definition, 326
selection rules, 3:!4
Glide plane:
definition, 350
axial, 3R4
diagonal, 384
diamond, 384
table of. 385
G matrix, 322
g matrix elements, 4-14
Great orthogonality theorem. 81
Group, Abelian, 7
classes of, 13
cyclic, lO
definition, 6
multiplication table for, 8
space, see Space groups
subgroups of, 12
Hamiltonian operator, 100
Hole formalism. 274
Hckel approximation. 136
for carbocyclic systems. 147
Hund's rule, 138
Hybrid orbitals, 222
AB
3
(planar). 223
AB. (planar), 226
octahedral, 227
tetrahedral. 226
trigonal bipyramidal. 226
Hydrogen peroxide molecule. symmetry of.
5H
lcosahedron. 46
symmetry of. ..JS
Interna! coordinates, 307
comributions to normal modes . .315
lnversion center:
definition, 22
matrix representation of. 71
Lattice. 20:
deflnition. 350
hexagonal. 353
oblique, 352
rectangular. centered, 353
rectangular. primitive. 353
rotational symmetry of. 354
symmetry of. 355. 3HO
types of. 351
Lattice, 30:
centering of, 370
chiral. 410
cubic. 372
isometric, 372
monoclinic. 370
orthorhombic, 371
properties of. 375
rhombohedral. 373
table of, 375
tetragonal, 371
triclinic. 370
types. 369
Laue symmetry, 383
LCAO approximation, 134
for allyl group. IR 1
for carbocyclic systems, 142
for tetramethylene cyclobutane, 159
Ligand field theory, 253, 287
Linear molecules, symmetry of. 56
L-S coupling scheme. 257
d" states in. 259
Matrices. adjoint, 424
block factored, 68
character of, 69
conjugate. 70
definition. 417
F. see F matrix
G. see G matrix
for geometric transformations. 420
inverse. 424
multiplication of, 418
non-singular. 425
orthogonal. 73
square. 62
Metal carbonyl clusters. ::!35
Methane:
INDEX 459
vibrational analysis, 335
photoelectron spectrum of. 223
Molecular orbital theory:
for AB, (planar) molecules, 219
for B ~ (tetrahedral)- molecules, 209
for AB,. (octahedral) molecules, 214
for AB" type molecules. 209
for carbocyclic systems, 142
for cyclobut:tdiene sandwich compounds.
2-lO
for dihenzenechromium. 246
for fcrrocene. 241
general principies. 133
for metal sandwich compounds. 240
Molecular vibrations, 304
selection rules. 324
wave functions for, 325
Motif, crystallographic. 348
Mulliken symbols. 90
Multiplicity, see Spin. multiplicity
Naphthalene:
electronic spectrum of. 176
LCAO-MO treatment of, 172
MO's, tahulation. 176
secular cquation for n bonding in, 139
NO,, vibrational modes. 333
N:F., vibrational analysis. 334
Normal modes of vibration:
of COj-. 306
gcnuine. 305
of NDh 333
number in a molecule. 305
symmetry of. 304, 309
Optical isomerism. 34
Orbital angular momentum. 255
Orthogonalty theorem. 81
Orthonormality. conditions for. 101
Orthorhombic space groups. 397
choice of axes. 399
Pi bonds. MO's for. 227
Planes of symmetry. 18
Platonic solids, 45
Polar coordinates. 205
Polarization. of d-d transitions. 289. 296
vibronic, 292
Polyhedra. regular. 45
460 INOEX
Polyhedral boranes. 231
MO"s for. 233
Products:
of group elements. 6
of symmetry operations. 29
Projection operators, 115
Ouantum mechanics, relationship to group
theory. lOO
Ouantum numbers, 254
for many-electron atoms, 257
for single electrons. 254
Reading list (Appendix IX). 451
Reciprocals. of group elements, 7
product of, 7
Redundant symmetry coordinates, 3l
Reflection operation, definition, 18
Reflection plane. 18
Regular polyhedra, 45
Representations. of groups, 78
irreducible. 81
reducible, 80
rules concerning. 81
Resonance energy, 148
Resonance integral. 438
Rotations:
improper, 27
matrix representation of, 71
proper. 22
rotation-inversion, 376
rotation-reflection. 27
Russell-Saunders coupling. 257
d" states arising from, 259
SALC's, 114
Schnflies notation. 41
sn definition. 27
Secular equations, derivation of. 135
symmetry factoring, 140
vibrational, 318
Selection rules:
for cydization reactions. 188
for spectra of complexes. 289
vibrational, 327
Woodward-Hoffmann, 189
SF
6
vibrational analysis of, 337
Similarity transformation. 11
Salid state. effect on vibrational spectra.
341
Space groups, 20, 358
diagrams, 362
examples. 366
list of, 359
symmetries. 355
Space groups, 30, 388
isometric. 399
monoclinic, 392
orthorhombic, 397
systematic absences for. 406
tetragonal, 399
triclinic, 389
trigonal/hexagonal, 399
S pace symmetry. 348
one-dimensional, 349
three-dimensional. 368
two-dimensional, 358
Spectroscopic fJ, 439
Spin. angular momentum, 256
multiplicity. 258
quantum number. 258
Stereographic projections, 448
Subgroups:
definition, 12
orders of, 12
Symmetry, of vibrational modes, 305,
309
Symmetry-adapted linear combinations
(SALC"s), 114
Symmetry classification. of molecules. 54
Symmetry coordinates. vibrational. 31R
Symmetry elements:
definition. IS
kinds. 19
Symmetry operations:
definition, IS
equivalent, 32
kinds. 19
products of. 29
rclations among. 33
Symmetry point groups. 39
Systcmatic absences. 400
conditions for. 403
Tanabe-Sugano diagrams. 276
Terms. in L-S coupling. "258
splitting by environment, 20
Tetramethylene cyclobutane. LCAO-MO
treatment. 159
Three-center bonding. 180
Transition probabilities, 110
electric dipole. 1 11
Translation, unit. 348
Unit cell, defining vectors of. 370
face labels. 370
20. 351
Unit matrix. 418
Vectors, 76. 418
orthogonal. 76
scalar products of. 76
Vibrational spectra. 304
selection rules for. 324
solid state effects on. 341
Vibronic coupling, 289
Wade s rules, 235
Water molecule:
symmetry of, 56
vibrational analysis of. 319
Wave equation, 100
INDEX 461
Wave functions, atomic, 206
vibrational. 325
Woodward-Hoffmann rules. 189
X-ray crystallography, 400
APPENDIX IIA
CHARACTER TABLES FOR
CHEMICALLY IMPORTANT
SYMMETRY GROUPS*
l. The Nonaxial Groups
Ca E
A 1
Cs E a,. e, E i
A' 1 1 x, y, R:
x1, y2,
A, 1 1 Rx, R,, R:
l
z, xy
A"
)
-1 z, Rx, R, yz,xz A, 1 -l x,y,z
2. The Cn Groups
el E el
A 1 1 Z, Rz
xl, yl, zl, xy
B J -1 X, y, Rx, R., yz,xz
CJ E CJ CJ
2
e = exp (21f'i/3)
A 1 1 1 z, R:
xl+ y2, zl
E
JI
E
:l
(x, y)(Rx, R,) (x
1
- y
1
, xy)(yz, xz)
li
e*
x2, yl, zl
xy, xz, yz
Appendix IIA is presented in two places: (1) he re, in its proper location in the sequence of
appendices, and (2) as a separate section in a pocket inside the back cover.
2 APPENDICES
The Cn Groups ( continued)
c4 E c4 C1
C4l
-
A 1 1 1 1 z, R: x2 + y2, z2
B 1 -1 1 -1 xl- yl, xy
E
:
i -1
-;} (x, y )(Rx, R,)
-i -1
(yz, xz)
Cs E Cs Cs
1
Csl
Cs
4
E = exp (27Ti/5)
A 1 1 1 1 1 z, R: xz + yl, zl
E1

E
el el*
E* l
e*
el* l
{x, y)(Rx, R,) (yz, xz)
:l.
E1
l
t* E
el*
E e*
2
(xl- yz, xy)
c6 E c6 el C1 C3
2
C6
5
e = exp (27T/6)
A 1 1 1 1 1 1 z, R:
xl+yl,zl
B l -1 1 -1 1 -1
g
E -e* -1 -E
E*l
(x, y)
(xz, yz) E1
e* -e -1 -e* E (Rx, R,)
E1
{:
-e* -e 1 -e* -E
(xl- yl, xy)
-e -e* 1 -e -E*
c7
E c7 C7
1
C7
3
C7
4
C7
5
C7
6
e.= exp {21Ti/7)
A 1 l 1
] ]
1 1 z, R:
xz + y2, zl
{:
E
E.l el el* el*
: }
(x, y)
(xz, yz)
EJ
t*
el* el* el l
(Rx, R,)
{:
E.l el*
e* E
el
E'*} (xl- yl, xy)
Ez
el* el
E e*
el* El
{:
el
e*
l el*
E
E'*}
El
El*
E
el* l
e*
l
es E Ca c .. e ~ c2 e ~ e ~
q e = exp(2ni/8)
A 1 1 1 1 1 1 1 1 Z, Rz
x2 + y2, z2
B 1 -1 1 -1 1 -1 1 -1
El
ft
E i -e* -1 -E -i
:}
(x, y)
(xz, yz)
lt
e* -i -e -1 -e* i (R:u R
1
)
El
n
i -1 -i 1 i -1
~ }
(x2 - y2, xy)
-i -1 i 1 -i -1
E3
g
-e i e* -1 e -i
-e*}
-e* -i e -1 e* i -E
CHARACTER TABLES 3
3. The Dn Groups
D2 E Cz(z)
L10'J
(. 2(X)
-
A l l l l
xz, yz, z2
B1
1 1 -1 -1 z, R: xy
B1 1 -l 1 -1 y, R, xz
B3
1 -1 -l 1 x, Rx y.z
D1 E 2C3 JC2
As
1
)
1 xz + yz, zz
A2
l 1 -1 z, R,
E 2 -1 o (x, y)(Rx, R,) (x
2
- y
2
, xy)(xz, yz)
D4 E 2C4 Cz(=C4
2
) C ~ 2c;
-
A1
1 1 1 1 1
xz+yz,zz
Az
1 1 1 -1 -l z, R:
Bs
1 -1 1 1 -1
xz -yz
Bz 1 -1 1 -l 1 xy
E 2 o -2 o o (x, y)(Rx, Ry) (xz, yz)
Ds E 2Cs 2Cs
2
5Cz
At
1 1 l 1 xz + yz, zl
Az 1 1 1 -1 z, R:
E, 2 2 cos 72 2 cos 144
~ 1 (x, y)(R., R,)
(xz, yz)
E2 2 2 cos 144 2 cos 72 (xz- y2, xy)
D6
E 2C
6
2C3 el 3 C ~ JC;
A1 1 1 1 1 1 l
x2 + yz, zz
Az
1 1 1 l -l -1 z, R:
B1
1 -1 1 -1 1 -1
Bz
l -1 1 -1 -1
]
E1
2 1 -1 -2 o o (x, y)(Rx, R,) (xz, yz)
Ez 2 -1 -1 2 o o (x2- yz, xy)
4 APPENDJCES
4. The C"., Groups
C2r: E Cz a . (xz)
A a 1 1 1 z
xz, yz, zz
A1 1 1 -1 -1 R: xy
B1 1 -) 1 -1 X, R., xz
Bz -1 -1 y, R;r: yz
z
-1 R:
-1 O (x, y)(R:r, RJ.) (x
2
- y
2
, xy)(xz, yz)
C4v E 2C4 Cz 2au
A1 1 1 1 1 z
xl + yz, zl
Az 1 1 -1 -1 R:
B1 1 -1 -1
xz-yz
Bz l -1 1 -1 1 xy
E 2 o -2 o o (x, y)(R;r:, R.,) (xz, yz)
E 2Cs 2Cs
2
Su u
1 1 1 1 z
x2 + y2, zz
1 l. 1 -1 R:
2 2 cos 72 2 cos 144 o (x, y)(R:r, R,) (xz, yz)
2 2 cos 144 2 cos 72 o
(x2- y2, xy)
C6u E 2C6 2C3 c2 3u., 3a,
A, 1 z
xz + yz, z2
A:t
1 -1 -1 R:
BJ
-1 -1 1 -1
Bz 1 -1 1 -) -1 1
E1 2 1 -1 -2 o o (x, y)(R;r:, R.,) (xz, yz)
E2 2 -1 -1 2 o o
(x2- yz, xy)
5. The Cnh Groups
Cu E el
;
A, 1 1 1
B, 1 -1 1
A a I l -1
B, 1 -1 -1
Cu E CJ C3
1
A' l 1 1
E'
g
e: e:*
t* e:
A" 1 1
]
E"
n
E t:*
t: t:
c .. , E c .. Cz
A, l 1 1
B, 1 -1 1
E,
n
i -1
-i -1
A u
1 1 1
Bu l -1 1
E,
n
i -1
-i -1
Cs,. E Cs
Cs2
A' 1 l 1
El'
n
E
El
t t2
E:'
n
t:l t
tl
1:
A" 1 1 1
El"

t
tl
t:*
tl
El"
tl t:
t:2
t:
c6,. E c6 CJ
A, 1 1 l
B, 1 -1 1
Et,

t: -t
t -e:
Ez
9
-t* -e:
-E -t*
A. 1 1 1
B, 1 -1 1
Elll
1 t -t:*
1 t -t
El,
1 -t -1:
1 -t -t
CHARACfER TABLES 5
a,.
1 R:
xl, _vz, z2, xy
-1 Rx,R, xz.y=
-1 z
1 x,y
a,.
s3
sJs
l 1 1
1 t
:}
J t:*
-1 -1 -1
-1 -t
-t}
-1 -e -t:
c .. 3 ; s.3 a,.
1 1 1 1
-1 l -1 1
-i 1 i -1
i 1 -i -1
1 -1 -1 -1
-1 -1 1 -1
-i -1 -i 1
;
-1 i 1
Cs3
Cs
4
a,.
Ss
1 1 1 l
tl
e:* 1 E
tl
t: 1 t:*
E
tl
1
t:l
t:*
t:l
1
t:l
1 1 -1 -1
t:l t -1 -t
tl
t -1 -t:
1:
t2
-1
-t:l
E*
t:2
-1
-tl
el
C32
C6
5
i
1 l 1 1
-1 1 -1 1
-1 -E ~ 1
-1 -E* t 1
1 -t -t 1
1 -t -t 1
1 1 J -1
-J 1 -1 -1
-1 -t: e: -1
-1 -t E: -1
1 -t -t -1
l -t -E* -1
Rz
(x, y)
z
e:=exp
xz+y2
(xl-y
{2ni/3)
.. l
....
(Rx. R,) (xz, yz)
s ..
1
-J
-n
-1
1
-!}
Ss
7
1
t:l
tl
t
t:
-1
-tl
-t2
-t:
-t
sls
J
-1
t
E*
-e:
-t
-1
1
-t
-e:
t
E
R:
(R", R,.)
z
(x, y)
Ss
3
Ss
9
1 1
t:l
t
t:2
t:
t t
~ t:
-1 -1
-t2
-E
-tl -t:
-t -e
-t: -t:
;.
S6
5
a,.
1 1
1 -1
-e: -1
-t: -1
-E 1
-t 1
-1 -1
-1 1
E* 1
t 1
t -1
t:* -1
xl + y2, zl
x2- yz, xy
(xz, yz)
Rz
(x, y)
z
(Rx, R,)
s3
1 R:
-1
=l
(Rx, R,)
-E
-tl
-J z
1
-t:*}
-t:
(x, y)
:.}
t = exp (2'1Ti/5)
(xz, yz)
t = exp (2'1Ti/6)
xz + y2, zl
(xz, yz)
(xl- yz, xy)
6.
6 APPENDICES
The Dnh Groups
Du,
A,
B1fl
B2fl
B3fl
A u
B,,
Bz.,
Bl,
Dl,.
A a'
Az'
E'
A''
Az"
E"
D4,.
A,
A z r ~
B,
B2,
E,
A tu
Azu
B..,
B:z,
E.,
Ds,.
A'
Az'
E'
E:z'
A"
A:z"
Ea"
El"
A a,
A2fl
B
B2fl
E,
E2,
Alll
A2u
Bau
Bl,
E tu
E2u
E
J
J
2
2
1
1
2
2
E C2(z) C2(y) Cz(X) i
1 1 1 1 1
1 1 -1 -1 1
1 -1 1 -1 1
1 -1 -1 1 1
l 1 1 1 -1
1 J -1 -1 -1
1 -1 1 -1 -1
1 -1 -1 1 -1
E 2C
3
3C
2
o,. 2S3 3a
11
1 J 1 1 J
J 1 -1 1 -1
2 -1 o 2 -1 o
1 1 1 -1 -1 -1
1 1 -1 -1 -J 1
2 -1 o -2 J o
E 2C4 Cz 2C
2
' 2C
2
"
J 1 1 1 1
1 1 -1 -1 J
-J J J -1 J
-) 1 -1 1 1
2 o -2 o o 2
1 J 1 1 1 -1
1 J
)
-1 -1 -1
l -1 1 1 -1 -1
J -1 J -1 J -1
2 o -2 o o -2
2Cs
2Csl se!
J J
1 -1
2 cos 1r 2 cos 144 o
2 cos 144 2 cos 72 o
1 1 1
1 1 -1
2 COS 72 2 cos 144 o
2 cos 144 2 cos 1r o
E 2Cb 2C
3 el 3C
2
' JC2"
1
)
J 1 1 1
J 1 1 1 -1 - 1
1 -1 1 -1 1 - J
1 -1 1 -1
-i
1
2 1 -1 -2

o
2 -1 -1 2 o o
1 1 1 1
j
1
1 1 1 1 -1 - 1
1 -J 1 -1 1 - 1
) -1 1 -1 -1 1
2 1 -1 -2 o o
2 -1 -1 2 o o
u(xy) a(xz) o(yz)
1 1 1 x2, yl, .:2
1 -1 -1
R=
xy
-1 1 -1 R., x=
-1 -1 1
Rx
Y=
-1 -1 -1
-J 1 1 -
-
1 -1 1 y
1 1 -1 X
xz + yl, =z
R:
(X, y) (x:z- yz, xy)
=
<Rx,R
1
) (x.:, _vz)
2S"
a,. 2a,. 2a
11
1 1 1 xz+yz,=:z
1 -) -J
R:
-J J -1
x1-y2
-1 -1 1 X)'
o -2 o o (R:t, R}.) (.\'.:',y.:)
-1 -1 -1 -1
-1 -1 J J
=
J -1 -1 1
1 -J 1 -J
o 2 o o (x, y)
u,. 2Ss 2Ss
3
5a
11
1 J J x:z + y:z, =2
J 1 -1 R:
2 2 cos 1r 2 cos 144 o (x, y)
2 2 cos 144 2 cos 72<J o (x2- yz, xy)
-1 -1 -1 -1
-1 -1 -- 1 1
=
-2 -2 cos 72 -2 cos 144 o
CR:t, RrJ <x=, Y=J
-2 -2 cos 144 -2 cos 72" o
i 2S3 2S
6
a,. 3a
11
3av
1 1 1 1 J 1
1 1 1 1 -1 -1 R:
1 -1 1 -1 1 -J
1 -1 1 -1 -1 1
2 1 -J -2 o o (R:t. R,.)
2 -1 -1
2
o o
-1 -1 -1 -1 -1 -1
-1 -1 -1 -1 1 1 -
-
-1 1 -1 J 1 1
-1 1 -) 1 l -1
-2 -1 1 2 o o (x, y)
-2 J 1 -2
{j
o
CHARACTER TABLES 7
6. The Dnh Groups (Continued).
D s ~ c
E 2Cs
2CsJ 2C
4
Cz 4Cz' 4c2 2Sa 2Ss
3
2S"
a,. 4a
11
4a
11
At,
1 1 1 J J 1 l 1
x2 + )'2, z2
A:z,
1 1 -1 -1 1 1 -1 -1 R,
Bt,
-1 -1 1 -1 -1 -1 1 -1
Bz,
-1 -1 -1 J -1 -1 -1 1
E,, 2 vi
_,lz
o -2 o o 2 v'2 -v'i o -2 o o R ~ , Ry) (xz, yz)
E:z,
2 o o -2 2 o o 2 o o -2 2 o o
(xl- y2, xy)
El,
2 -v'i v'i o -2 o o 2 -V
,12
o -2 o o
A su 1 1 1 1 1 1 1 -1 -1 -1 -1 -1 -1 -1
A2u
1 1 1 1 1 -1 -1 -1 -1 -1 -1 -1 1 1 z
Bu 1 -1 -1 1
)
1 -1 -1 1 1 -1 -1 -1 1
B:zu 1 -1 -1 l 1 -1 1 -1 1 1 -1 -1 1 -1
El u 2
,; i
-Vi o -2 o o -2 -v'2 vi o 2 o o (x,y)
E2u 2 o o -2 2 o o -2 o o 2 -2 o o
El u 2 -Vi vi o -2 o o -2 vi -'\/2
o 2 o o
7. The Dnd Groups
Dz, E 2S4 C2
2C
2
' 2a
4
A, 1 1 1 1 x2+y2,=2
Az
1 1 -1 -1 R,
Bt
-1 1 1 -1
x2 _ J'2
B:z -1 1 -1 1
-
xy
E 2 o -2 o o (x, J'); (xz, yz)
R ~ , Ry)
D3, E 2Cl 3C
2
2S
6
3ad
A., 1 1 1 1 1 1 x:z + y2, z2
A:ze
1 1 -1 1 l -1
R=
E, 2 -1 o 2 -1 o (Rx. Ry) (xl - y:z xy),
(xz, Y=>
Au 1 1 1 -1 -1 -1
Azu
1 1 -1 -1 -1 1 z
E u 2 -1 o -2 1 o (x, y)
D"d
E 2S
8
2C
4
2SaJ
Cz
4C
2
' 4crd
At
1 1 1 1 1 1 x2 + y2, zl
A2 1 1 1 1 -1 -1 R:
B, -1 1 -1 1 1 -1
Bz
-1 1 -1 1 -1 1 z
Ea 2 v2 o
_,Ji
-2 o o (x, y)
E:z 2 o -2 o 2 o o (xl - J'l xy)
E3 2 -'\12 o vi -2 o o R ~ , Ry) (xz, yz)
Ds, E 2Cs 2Cs
2
5C
2 2Sto
3
2S
10
Sad
A,, 1 J 1 J 1 1 x2 + y2, zl
A2,
1 1 -1 1 1 -J R,
E., 2 2 cos 72 2 cos 144 o 2 2 cos 72 2 cos 144 o R ~ , R}.) (x=. y.::)
E2, 2 2 cos 144 2 cos 1r o 2 2 cos 144 2 cos 1r o (x:z - y2, xy)
A1u
1 1 1 1 -1 -1 -1 -1
A2u
1 1 1 -1 -1 -1 -1 1 !
E., 2 2 cos 72 2 cos 144 o -2 -2 cos 1r -2 cos 144 o (X,)')
E2.
2 2 cos 144 2 cos 72 o -2 -2 cos 144 -2 cos 72 o
8 APPENDICES
7. The Dnd Groups (Continued).
D64
E 2S
12
2C6 2S4 2C
3
2S
12
S
c2 6Cz' 6a4
A a
1 1 l 1 1
x2 + yl, :2
Al 1 1 l -1 -1 R:
Ba
-1 -1 -1 1 -1
Bz
-1
.:._1
-1 -1 1 z
Ea 2 v3 1 o -1 -v3 -2 o o (x, y)
El 2 1 -1 -2 -1 1 2 o o (x2- y2, xy)
EJ 2 o -2 o 2 o -2 o o
E4 2 -1 -1 2 -1 -1 2 o o
Es 2 -v'3 1 o -1
v'j
-2 o o , R ~ , R y ) (xz, yz)
8. The Sn Groups
s4 E s4 c2 s4
3
A 1 l R:
x2 + yz, zz
B 1 -1 1 -1
....
xl- yl, xy
..
E
(:
i -1
-i\
(x, y); (Rx, R).) (xz, yz)
-i -1
iJ
s6 E CJ CJ
2
i S6
5
s6 e = exp (21Ti/3 J
A, 1
)
R:
xl + yz, zl
/1
E e
E
:}
R ~ , R,J
(xz- yz, xy);
EtJ
\ 1
t:"' E e* (xz, yz)
A., J -1 -l -1 z
J 1
E e -1 -E
-e*}
(x, y) E.,
\1
E* E -1 -e* -E
S a E Ss c4 Sa
3
c2 Sa
5
C4
3
Sa
7
e= exp (2rri/8)
A 1 1 R:
xl + yz, zz
B -1 -t 1 -:-1 -1 z
e i -E* -1 -e -i
:}
(x, y);
Ea
e -i -e -1 -e* R : ~ : , R,)
i -1 -i i -1 -i
\
(x
1
- .v
2
, xy)
E1
1 -i -1 i -i -1 i
1
{:
-e* -i E -1 e i
-E, (xz, yz)
El
-E i
e
-1 E -i -E*
9. The Cubic Groups
T E 4Cl
4C
3
:.i
3Cz E= exp (2n"i/3)
A 1 1 xl+yl+zl
E{
l E t 1
(
2
:2 _ xz _ yz,
1 E* e 1
xz _ yl)
T 3 o o -1 CRx. R,, R:); (x, y,:) (X}', XZ, yz)
CHARACTER TABLES 9
9. The Cubic Groups (Continued).
. T,
E 4C
3
4Cj 3Cz 3a,. e = exp(21li/3)
A, 1 1 1 1 1 + .r + z!
E,
{ l E
E*
1 1 t
,.

(2z
1
- r - y
1
, - Jrl)
1 e e 1 1
e
E
T,
J.O o -1 1 o o -1
(R .. R'", R:) (xz, yz. xy)
A., J 1 1 1 -1 -1 -1 -1
E.,
{;
e
e
1 -1 -e
-[
-1}
e* [ 1 -1 -e -e
-1
T., 3 o o -1 -1 o o J (x, y. z)
E 8C
3
3C
2
6S .. 6ud
A, 1 1 1
)
1 x2 + 1 + z2
A2 1 1 l -1 -1
E 2 -1 2 o o (2z
2
- x
2
- y!. r - y.!)
T, 3 o -1 1 -1
(R R'". R:)
Tz 3 o -1 -1 1 (x, y, z) (xy, xz. yz)
o E se) 3C
2
(=Ci)
6C"'
6Cz
A, 1 l 1 1 l .r + y2 + zl
Az 1 1 J -1 -1
E 2 -1 2 o o (2z2 - xz - x"! - y:!J
T, 3 o -1 1 -1 (R
6
, R:); (x. y. z)
r2 3 o -1 -1 1 (xy, xz, yz)
olt E 8C
3 6C2 6C.a 3C2(=d) 6S .. SS,. 3a
11 6ud
A,, 1 r + yl + zl
A2
1
1 -1 -1 -1 J -1
E, 2 -1 o o 2 2 o -1 2 o (2z
2
- r - y
1
x
2
- yl)
T,, 3 o -1 -1 3 o -1 -1 (R,. R'". R:)
T2r
3 o -1 -1 3 -1 o -1 1 (xz. yz. xy)
A,. 1 -1 -1 -1 -1 -1
A:z,
)
-1 -1 -1 -1 -1 1
E u 2 -1 o o 2 -2 o 1 -2 o

T,u 3 o -1 -1 -3 -1 o (x, y, z)
Tz, 3 o 1 -1 -1 -3 1 o -1
10 APPENDICES
10. The Groups ~ u and Da;,h for Linear Molecules
CIDI1
E 2CID coa.,
A= I:+ 1 l 1 z xz + yz, zl
Az = I:- 1 1 -1
Rz
E
1
:=ll 2 2 cos el> o (x, y); (Rz, R,.} (xz, yz)
Ez=A
2 2 cos 2<1> o (xl- yz, xy)
E1 =el>
2 2 cos JCz
l:+ 1 1 1 l 1 1 xz+yz,zz
l:'-
1 1 -1 1 1 -1
Rz
:
2 2 cos el> o 2 -2 cos o (Rz, R,) (xz, yz)
A, 2 2 cos 2cl> o 2 2 cos 2 o -2 2 cos o (x, y)
A. 2 2 cos 2<1> o -2 -2 cos 2cl> o
1
1
.

T
h
e

I
c
o
s
a
h
e
d
r
a
l

G
r
o
u
p
s
*

1
,

E

1
2
C
5

1
2
c
,
:
z

2
0
C
3

1
S
C
2

1

1
2
S
a
o

1
2
S
a
o
3

2
0
S
6

I
S
u

A
.

1

1

1

1

1

1

1

1

1

1

x
:
z

+

y
:
z

+

z
2

T
,
.

3

l
(
l

+

v
'
s
>

i
<
l

-
v
'
s
>

o

-
1

3

i
O
-
v
'
s
>

l
O
+

v
'
s
>

o

-
1

(
R
;
s

,
R
,
,

R
z
)

T
:
z
.

3

l
<
l

-
v
'
s
>

l
<
t

+

v
'
s
>

o

-
1

3

!
(
1

+

v
'
s
>

!
U
-
v
'
s
>

o

-
1

a
.

4

-
1

-
1

1

o

4

-
1

-
1

l

o

H
.

S

o

o

-
1

1

S

o

o

-
1

1

(
2
z
:
z

_

x
:
z

_

y
:
z
,

x
2

-
y
:
z
,

x
y
,
y
z
,
z
x
)

A
.

1

1

1

1

1

-
1

-
1

-
1

-
1

-
1

T
1
1
1

3

i
O

+

v
'
s
>

!
(
1

-
v
'
s
>

o

-
1

-
3

-
1
(
1
-
v
'
s
>

-
1
(
1

+

v
'
s
>

o

1

(
x
,

y
,

z
)

T
:
z
.

3

l
O

-
v
'
s
>

i
{
l

+

v
'
:
S
)

o

-
1

-
3

-
!
U

+

v
'
s
>

-
i
O

-
v
'
s
>

o

1

G
.

4

-
1

-
1

1

o

-
4

1

1

-
1

o

H
.

S

o

o

-
1

1

-
S

o

o

1

-
1

.
,
F
o
r

t
h
e

p
u
r
e

r
o
t
a
t
i
o
n

g
r
o
u
p

/
,

t
h
e

o
u
t
l
i
n
e
d

s
e
c
t
i
o
n

i
n

t
h
e

u
p
p
e
r

l
e
f
t

i
s

t
h
e

c
h
a
r
a
c
t
e
r

t
a
b
l
e
;

t
h
e

g

s
u
b
s
c
r
i
p
t
s

s
h
o
u
l
d
,

o
f

c
o
u
r
s
e
,

b
e

d
r
o
p
p
e
d

a
n
d

(
x
,

y
,

z

a
s
s
i
g
n
e
d

t
o

t
h
e

r
.

r
e
p
r
e
s
e
n
t
a
t
i
o
n
.

(
"
)

:
e

>

:
=
t1

>

:=
t1

t
x
!

r

m

V
l

.
.
.
.

"
"
"
"

Chemical Applications of Group Theory 3rd Edition

Uploaded by

Copyright:

Available Formats

Chemical Applications of Group Theory 3rd Edition

Uploaded by

Document Information

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

Chemical Applications of Group Theory 3rd Edition

Uploaded by

Copyright:

Available Formats

Copyughled Materral

Copyright e 1990 by F. Albert Conon

will show .that within

and the cyclic group of order 6, show what

ions, for example,

have the square parallepiped

x E, x 8!,, contains the A representation. It is a simple matter to show

representations which comprise

ea eh contributing atomic wave functions to the formation

which has the energy E

in the three cases above. The symmetry of each is found by forming

orbitals of ato m A, namely, the dX)., d.r= and

SALC formed from these orbitals (Fig. 8.16) will be a

type we already have the shapes from our discussion of the n

fragment can be considered analogous to M(C0)

For the metal atom the AOs can also be

(2) the configuration in which one electron is in a

configuration we may place one electron in any of six

configuration may be both triplet and singlet. We thus

states at the left. The higher one n1ust connect to the

configuration. There is only one state below this,

and as the electron interactions take effect.

however, since the two electrons

We found previously, it will be recalled, that only

In summary. then. the direct product t

states. These must

versus b In and of b'1P. versus

subset, it will be more stable by an

subset it will have to enter an orbital that is already occupied by one

electron may be assumed

states of the d electron might be computed, Jet us first consider

separation. Such caJculations

orbitals and four are in e,: orbitals and

and E:,. These

each molecule lies on a

which crystallizes in the

You might also like