ON THE MOLECULAR THEORIES OF

LIQUID

CRYSTALS

w
Claudio Zannoni
Dortore in Chimica Industriale

A dissertation submitted i n partial fulfilment of the requirements

fox the degree of Doctor of Philosophy a t the University of

Southamp ton.

Department of Chemistry

June, 1975.

AK O LD E E T C N WE GMNS

It i s a p l e a s u r e t o thank D r . G. R. Luckhurst f o r s o many u s e f u l

d i s c u s s i o n s and f o r s t i m u l a t i n g my i n t e r e s t i n l i q u i d c r y s t a l s . I should l i k e t o thank t h e o t h e r members of t h e Chemical Physics group and p a r t i c u l a r l y D r . S. G. Carr, M r . S. K. Khoo, M r . R. N. Yeates f o r reading p a r t s of t h e manuscript. Also M r . P. D. F r a n c i s f o r h i s

e s s e n t i a l maintenance of t h e Varian 620i computer. The experimental r e s u l t s analysed i n Chapters 5 and 7 were obtained by D r . M. Setaka and D r . R. Poupko. I am g r a t e f u l t o P r o f e s s o r G. F. P e d u l l i (Bologna U n i v e r s i t y ) f o r h i s encouragement during t h e course of t h i s work. Last b u t of course by no means l e a s t one of my g r e a t e s t d e b t s is t o my w i f e N i c o l e t t a f o r h e r continuous support and p a t i e n c e and f o r typing t h i s t h e s i s .

TABLE OF CONTENTS

Page
CHAPTER ONE

ORIENTATIONAL ORDER IN LZqUID CRYSTALS

1.1

Liquid Crystals
The Orientational Order Parameters

1.2
1.3

Determination of the Order Parameters

CHAPTER TWO
A

STATISTICAL THEORY OF THE NEMATIC P ~ S E

Introduction
A Molecular Theory of the Nematic Mesophase

2.1
2.2

2.3
2.4

The Maies

Saupe Model

The Nematic

Isotropic Phase Transition

CHAPTER THREE
A M O L E C W FIELD THEORY FOR UNIAXIAL NEMATIC

LIQUID CRYSTALS FORMED BY NOH-CYLINDRICALLY SYMMETRIC MOLECULES

3.1
3.2

Introduction
The Orientational Pseudo
Parameterisation Predictions and Reality

- Potential

3.3

3.4

CHAPTER FOUR
SPIN RELAXATION IN LIQUID CRYSTALS
4.1
Introduction
Spin Relaxation Theory

4.2 4.3
4.4(a)

The Static Spin Hamil tonian The Dynamic Spin Hamiltonian The Correlation Function

4.4 (b)
4.4 ( c )
4.4(d)

The Strong Collision Model

The Diffusion Model

Computational Problems

4 . 4 (e)

CHAPTER FIVE
ANGUIAR DEPENDENT LINEWIDTHS

5.1
5.2

Introduction Experimental

5.3

Angular Dependent Linewidths

The Strong Collision Model The Diffusion Model

5.4 5.5

CHAPTER SIX

A MORE COMPLETE THEORY OF ANGULAR DEPENDENT LINEWIDTHS

6 1 6.2
6.3

Introduction

Theory
Results

CHAPTER SEVEN
SPIN REUXKTZON FOR BIRADIW SPIN PROBES IN ANISOTROPIC ENVIRONMENTS

Intruduction
Theory Statics

Dynamics

The Experiment
S p e c t r a l Analysis

The Molecular Syrmnetry

The Angular Linewidth Coefficients

Additional Spectral Features

Quadrupole Relaxation

APPENDIX 1

IRREDUCIBLE TENSORS AND WIGNER ROTATION MATRICES

APPENDIX 2
SOME PROPERTIES OF THE KUMMER CONFLUENT HYPERGEOMETRIC

FrnCTIONS

ABSTRACT FACULTY OF SCIENCE CHEMISTRY Doctor of Philosophy

ON THE MOLECULAR THEORIES OF LIQUID CRYSTALS

by Claudio Zannoni

This thesis is concerned with the statistical theory of liquid crystals and the molecular reorientation in these anisotropic fluids. The first chapter reviews the definition and the experimental determination of the order parameters. We then develop, in the second chapter, a statistical theory of the nematic mesophase, starting from the total partition function for rod-like particles and arriving at an effective orientational partition function. The usual molecular field theory of nematics is obtained as the first term in a rather general expansion.

An extension of the molecular field theory to uniaxial nematics formed by

non-cylindrically symmetric molecules is discussed in Chapter 3 and comparison is made with available experimental data. The study of molecular reorientation in such anisotropic systems as liquid crystals or membranes by using electron resonance is the subject of the remainder of the thesis. In Chapter 4 some background material on

spin relaxation and the theory of reorientation in the presence of an ordering potential is reviewed. Methods for the computation of the spectral densities are also developed. The molecular reorientation of a monoradical spin probe in a highly ordered smectic A phase is studied in

Chapter 5 by analyzing the angular dependence of the electron resonance linewidths in terms of both a strong collision and a diffusion model. The determination of the order parameter
4

is discussed. The spin

relaxation theory employed is generalized in Chapter 6 to include the effect on the angular dependent linewidths of the pseudosecular terms in the static spin hamiltonian. The predictions of this theory are illustrated with model calculations. Finally, in Chapter 7 a theory is developed, and experimental data analyzed, for the angular dependence of the spectral linewidths of biradical spin probes.

- 1 -

CHAPTER ONE

ORIENTATIONAL ORDER IN LIOUID CRYSTALS

11 .

Liquid Crystals

A liquid crystal is an equilibrium phase intermediate between

a liquid and a crystal. At a microscopic level crystals formed of non-spherical molecules are characterized by a long-range order both in the position and the orientation of their constituents and normal liquids by the absence of both. often called a mesophase, In contrast a liquid-crystalline phase,

can still possess an orientational order,

12. while the positional one is greatly reduced or absent altogether ( , ) More precisely, the amount of positional order helps to classify liquid crystals into various types: nematics and cholesterics having no positional order, and smectics having the molecular centres of mass

3. arranged in layers ()

An idealized diagram of the molecular organi-

zation of nematics, cholesterics, and smectics is pictured in figure 1.1. The common feature in all these three types, as well as in the various subtypes, is the existence of a preferred or average molecular orientation, called the director. Liquid-crystalline phases can be produced in two basic ways: either by dissolving the mesogen in an appropriate solvent or by a thermal transition. Substances giving mesophases in just the first

way are called lyotropics ( 4 ) , an example being soaps; the others are termed thermotropics (5,6). We shall be only concerned with the

latter type in this thesis, even if many of our arguments, mutatis mutandis, could be applied to lyotropics as well.

Figure 1.1

Molecular organisation in (a) the nematic phase, (b) smectic phase, and ( c ) the cholesteric phase.

A t y p i c a l example of a t h e m t r o p i c nematic l i q u i d c r y s t a l i s

4,4'-dimethoxyazoxybenzene.

This i s c r y s t . a l l i n e a t room temperature

and g i v e s on m e l t i n g a t u r b i d phasewhichon f u r t h e r h e a t i n g y i e l d s an ordinary, c l e a r , l i q u i d . The mesophase owes i t s c h a r a c t e r i s t i c t u r b i d

appearance t o t h e f a c t t h a t t h e d i r e c t o r changescontinuously o r i e n t a t i o n from one p o i n t of t h e sample t o t h e o t h e r , t o p r e s e r v e t h e s p h e r i c a l symmetry of t h e i s o t r o p i c phase, and t h e r e s u l t i n g o r i e n t a t i o n a l inhomogeneity causes l i g h t s c a t t e r i n g (1,2). This f u l l rotaFor example

t i o n a l symmetry of t h e nematic mesophase i s e a s i l y broken.

t h e d i r e c t o r can be r e a d i l y o r i e n t e d by applying a magnetic f i e l d (7), because of t h e enhancement of t h e anisotropy i n t h e s u s c e p t i b i l i t y caused by t h e o r i e n t a t i o n a l c o r r e l a t i o n . The r e s u l t i s t h a t t h e motion

of every s i n g l e molecule with r e s p e c t t o t h e d i r e c t o r i s r e l a t i v e l y unaffected by t h e f i e l d , b u t t h e l i q u i d c r y s t a l a s a whole i s aligned (2,8). The o r i e n t e d mesophase i s an unusual s t a t e of m a t t e r i n t h a t

i t has; a s an example, t h e o p t i c a l and d i e l e c t r i c p r o p e r t i e s of a

u n i a x i a l c r y s t a l , t h e d i r e c t o r bzing t h e a x i s of symmetry, while s t i l l r e t a i n i n g t h e m o b i l i t y of a f l u i d . pic fluid. I n a sense i t i s then an a n i s o t r o -

Much of both t h e t h e o r e t i c a l i n t e r e s t and of t h e p r a c t i c a l stem from t h i s f a c t t o g e t h e r

a p p l i c a t i o n s of l i q u i d c r y s t a l s (9,10,11)

w i t h t h e i r e a s e of alignment by magnetic and e l e c t r i c f i e l d s ( l 2 ) a s w e l l a s p r o p e r l y t r e a t e d s u r f a c e s (13). From a chemical p o i n t of view t h e molecules of mesogens, i . e . substances g i v i n g r i s e t o l i q u i d c r y s t a l l i n e phases, can be q u i t e simple
/

and i t i s estimated t h a t , on average, one o u t of two hundred of o r g a n i c molecules has t h i s property. To g i v e some examples we have l i s t e d i n

Table 1.1 t h e s t r u c t u r e and t h e t r a n s i t i o n temperatures of some

TABLE 1.1

The structure and the transition temperature of

some nematogens.

NAME AND FORMULA

1)

4,4 '-dime thoxyazoxybenzene (PAA)

118

136

14

3)

anisaldazine

4)

2,4-nonadienic acid

. M 4 = \ /
OH

5) p-quinquephenyl

Figure 1.2

The molecular organisation i n (a) the smectic A phase,

(b) untilted smectic B phase and (c) the smectic C phase.

TABLE 1.2

The transition temperatures of some smectogens

(a)

Temperatures ('C) Name and formula S~

S~

for the transitiorr to

Ref.

s~

4,-4 '-di (n-hep ty 1)azoxybenzene

P
n-c7",5 -@N\ @ N

34

54

71

23

*-SH,,

Diethy 1 4,4'-azoxybenzendicarboxylate

114
"5

122.6

4,4'-di-n-heptyloxyazoxybenzene

"=7"l5

ixkN{O
rcC7n15

74

94

124

Ethyl 4-ethoxybenzylidene-4'-aminocinnamate
"5

o '*
~*\<5"5

81.7

118.5

156.5

159

Terephthal-bis (butylaniline)

113

144

172

200

236

(a) SA, SB, SC, indicate the smectic A, B, C types; N and I indicate the nematic and isotropic phase.

The first number from the left of every entry is the transition temperatute from the solid.

t y p i c a l nematics.

Common f e a t u r e s a r e a high length t o breadth r a t i o ,

r i g i d i t y , and anisotropy i n t h e p o l a r i z a b i l i t y t e n s o r , u s u a l l y ensured by conjugated r e l e c t r o n systems. They o f t e n c o n t a i n permanent

d i p o l e s b u t t h e s e a r e n o t e s s e n t i a l f o r t h e formation of t h e mesophase, a s proved by t h e e x i s t e n c e of v i r t u a l l y a p o l a r mesogens (e,g, the

l a s t e n t r y i n Table 1.1) and from t h e f a c t t h a t l i q u i d c r y s t a l s do n o t e x h i b i t t h e f e r r o e l e c t r i c i t y expected i f dipole-dipole were important. Nematogens, i n p a r t i c u l a r , a r e n o t o p t i c a l l y a c t i v e . I f the interactions

m l e c u l e i s c h i r a l , then t h e h y p o t h e t i c a l mesophase produced w i l l be cholesteric, i.e. a n a t u r a l l y t w i s t e d v e r s i o n of nematics having t h e shape, with a p i t c h of t y p i c a l l y about

d i r e c t o r wound i n a h e l i c a l 2000

(1,16,17). Molecules g i v i n g smectics, i . e . l a y e r e d l i q u i d c r y s t a l s , have

again t h e same g e n e r a l c h a r a c t e r i s t i c s b u t moreover they u s u a l l y have f l e x i b l e a l k y l chains a t t h e ends and s t r o n g d i p o l e moments more o r l e s s perpendicular t o the c e n t r a l p a r t . The d i f f e r e n t o r d e r i n g of

t h e s e d i p o l e s i n t h e smectic l a y e r has been invoked (18,19) a s a p o s s i b l e e x p l a n a t i o n of t h e e x i s t e n c e of many d i f f e r e n t smectic types (20,21,22). I n f i g u r e 1.2 we i l l u s t r a t e t h r e e of t h e s e types, t h e Notice t h a t t h e p r e f e r r e d o r i e n t a t i o n i s perpendi-

smectic A, B and C.

c u l a r t o t h e l a y e r i n t h e A type and t i l t e d with r e s p e c t t o i t i n t h e

C type.

The smectic B, i n c o n t r a s t w i t h t h e o t h e r two types, possesses The d i r e c t o r can be

some degree of s p a t i a l o r d e r i n s i d e each l a y e r .

p a r a l l e l o r t i l t e d t o t h e l a y e r normal i n analogy w i t h smectic A and

C types.
arelisted

The s t r u c t u r e and t r a n s i t i o n temperatures of some smectogens in Table 1.2.

To d e s c r i b e a l i q u i d c r y s t a l two concepts a r e p a r t i c u l a r l y relevant: t h a t of a d i r e c t o r and of a set of o r d e r parameters. The d i r e c t o r i s a u n i t v e c t o r defined a t every p o i n t of t h e sample and giving t h e l o c a l p r e f e r r e d orientation.
It i s assumed

t o vary slowly and continuously from one p o i n t t o another except i n c e r t a i n regions c a l l e d d i s c l i n a t i o n s , where t h e p r e f e r r e d o r i e n t a t i o n changes a b r u p t l y (1,24). The concept of t h e d i r e c t o r i s a macroscopic

one, even though i t can be given a molecular s t a t i s t i c a l d e f i n i t i o n , and a continuum theory of l i q u i d c r y s t a l s has been developed (1,24-

29) which d e s c r i b e s t h e behaviour of nematics under e x t e r n a l perturbat i o n s i n terms of t h e deformation energy of t h e d i r e c t o r f i e l d . Ty-

p i c a l examplesare t h e determination of t h e e q u i l i b r i u m conformation of a nematic under c e r t a i n boundary conditions o r i n t h e presence of an e l e c t r i c o r magnetic f i e l d , t h e study of i t s flow and s u r f a c e prop e r t i e s etc... While s u c c e s s f u l i n e x p l a i n i n g such e f f e c t s , the c o n t i -

nuum t h e o r i e s give no d i r e c t information on t h e behaviour of l i q u i d c r y s t a l s a t a molecular l e v e l , which i s our major concern i n t h i s t h e s i s . From a molecular p o i n t of view we s h a l l f i n d i t p a r t i c u l a r l y u s e f u l t o introduce a s e t of o r i e n t a t i o n a l o r d e r parameters d e s c r i b i n g t o what e x t e n t t h e molecules a l i g n with r e s p e c t t o t h e d i r e c t o r . The importance of t h e concept of an o r d e r parameter t r a n s i t i o n s i s w e l l known (30). i n studying phase

I n general it can be defined a s a quanti-

t y ( o r a s e t of q u a n t i t i e s ) which i s non zero below a c e r t a i n c r i t i c a l and zero above i t . temperature T k'

A s an example, f o r t h e common

liquid-gas t r a n s i t i o n t h e o r d e r parameter can be taken a s t h e d i f f e rence between t h e d e n s i t y of t h e gas and the l i q u i d pG -pLand f o r a ferromagnetic system i t can be taken a s t h e magnetization, which

again vanishes above a c e r t a i n temperature, t h e Curie p o i n t .

The way

the order parameter changes i n t h e course of a phase t r a n s i t i o n i s a l s o important s i n c e i t t e l l s us something about t h e c h a r a c t e r of t h e t r a n s i t i o n ; a d i s c o n t i n u i t y i n d i c a t e s a f i r s t o r d e r phase t r a n s i t i o n while a continuous change i n d i c a t e s a second o r d e r one. The f i r s t type i s

a l s o c h a r a c t e r i z e d of course by a d i s c o n t i n u i t y i n t h e f i r s t d e r i v a t i ve of an a p p r o p r i a t e thermodynamical p o t e n t i a l ( t h e Gibbs f r e e energy, say) and t h e o t h e r by a d i s c o n t i n u i t y o r o r d e r d e r i v a t i v e of t h i s p o t e n t i a l . I n t h e r e s t of t h i s f i r s t chapter we s h a l l t r e a t i n some det a i l t h e set of o r d e r parameters r e l e v a n t f o r t h e d e s c r i p t i o n of liquid c r y s t a l s . I n Section 1.2 t h i s set i s defined i n a general way an i n f i n i t y i n t h e second

and then s p e c i a l i z e d t o t h e important case of u n i a x i a l mesophases composed of c y l i n d r i c a l l y symmetric molecules. The way t h e o r d e r

parameters can be determined experimentally i s then discussed i n Section 1.3.

1.2

The Orientational Order Parameters It is usually assumed that the orientational order in uniaxial

liquid crystals is the average of just the second Legendre polynomial: 2 P2 = (3cos - 1)/2, where g is the angle between the axis of symmetry of the molecule, assumed to be rodlike, and that of the mesophase. The upper bar denotes an ensemble average (31,32). this choice is that The argument for 2 obviously reduces to the proper limits: P2 = 0

in the case of no alignment, i.e. an isotropic system, and P = 1 in 2 the case of a perfect alignment. However, Humphries, James and Luckhurst (33) have shown that in the molecular field limit one should more properly have not just one, but an infinite set of order parameters, given by the ensemble averages of all the Legendre polynomials:
-

p2, p4, P6,". In this section we shall define a set of orientational order parameters in a general way i.e. as the minimal set of quantities necessary to describe completely the singlet orientational distribution (B) function f a y . This distribution is defined as the probability

density that the orientation of a molecule is given by the three Euler angles (aBy) ( 4 . 3) The ensemble average of any single particle can be expressed in terms of f(aBy) as

orientational function A(aBy)

A = \A(aBy)

f (aly) da sing dB dy.

(1.1

As any other well behaved function of three Euler angles the singlet distribution can be expanded in a basis of Wigner rotation matrices

The definition and the main properties of the Wigner rotation matrices are described in Appendix 1. Multiplying both sides of

equation (1.2) by DL* and integrating over the angles, we find, qn

because of the orthogonality of Wigner matrices.

The averages

DL The expansion can be are our set of order parameters ( 5 3 ) 3.6. m,n simplified using the symmetry property of the mesophase or of the constituent particles. For uniaxial phases, as the nematic or smectic

A are found to be, the singlet distribution must be invariant underro.tation about the z axis if this is chosen along the director i.2. the
-

axis of symmetry.

This implies that m must be zero in D

L* m,n

.
To simplify

Another general property, for a uniaxial mesophase with D ah symmetry, is that only even L can appear in equation (1.2).

further one has to investigate the symmetry of the molecules forming the mesophase. We shall discuss this in some detail in Chapter 3.

By now we consider the simplest case of all, which arises of course when the molecules are cylindrically symmetric. Then a rotation about

their symmetry axis should not modify the distribution function, which implies n=O in equation ( . ) 13. In other words in this limit the distri-

bution function will depend only on the angle 6 between the director and the molecular symmetry axis:

fB ()

0 =

cLPL(cosfi)s

( L even )

(1.4

*ere

CL

(ZL+1) L 12. and the PL(x) are the Legendre polynomials

If we let

x = cosB then

and

For the most

conaPon

case of cylindrically symaetric molecules t h e

. order parameters are

just the averages

F2, P4

etc. and it is

clear

that a knowledge of all the

completely defines the s i n g l e t dirtri-

bution; while knowing f(d all the order paramters can be calculated.
An alternative way of characterising the singlet distribution

is through its moments un:

'n

- 7- -

xn f(3 dx.
1

These are sometims easier t o calculate as we shall see

in Chapter 2,

where we give a compact expression for the mments of the distribution

f (XIexp(Sx ), which is frequently used in atudiea of l i q u i d crystals. a

The representation of the distribution functian in term of order parameters is clearly equivalent to that in t e r m of moments, since

where [n] means the largest integer contained in n (Since L is even .

in our case [ ~ / 2 ]

L/Z).

The coefficieatg

t, are r

def iued as (37)

The first few example are

Averaging e . 1 7 q(.)

or eqs.(l.8)

gives the desired relation between

moments and order parameters. We have defined before the order parameters as averages of Wigner rotation matrices. It is worth mentioning that often in the

literature an equivalent formulation in terms of cartesian, as opposed to spherical, coordinates and tensors has been used to introduce parameters that measure partial alignment.. As an example, considering the case of second rank order parameters, an ordering matrix S is obtained ( 3 2 ) . The matrix S is symmetric and traceless so that in its proper frame it can be described by just two parameters: SZZ and (Sxx

- SW).

The

explicit relation between the elements of the ordering matrix and the set of D2 is m,n

1.3

Determination of t h e Order Parameters Let us suppose w e want t o measure a molecular 2nd rank tenso-

r i a l property F during a time s u f f i c i e n t l y long t o allow t h e observation of s t a t i s t i c a l l y averaged q u a n t i t i e s . r t ) I f F ( ~ a~ e ~ h e i r r e d u c i b l e

, s p h e r i c a l components of F, a s defined i n Appendix 1 and i f F i s s p e t r i c , then i t s only non zero components w i l l be those of rank L=O and L=2. I f t h e molecule i s r i g i d and i f we n e g l e c t l o c a l f i e l d problems,

t h e L=O component, being a pure s c a l a r , w i l l be unchanged going from an i s o t r o p i c t o an o r i e n t e d f l u i d , and we s h a l l n o t consider i t . W e

then n o t e t h a t F is measured i n a l a b o r a t o r y (unprimed) frame and t h a t t h e L=2 components of F can be w r i t t e n i n terms of t h e molecule f i x e d components F (2,m)'

as:

Taking an ensemble average we have, f o r a u n i a x i a l mesophase,

where we have i m p l i c i t l y taken t h e z a x i s p a r a l l e l t o the director.

of our l a b o r a t o r y system

I f t h e o r d e r parameters a r e d i f f e r e n t from z e r o , a s i n t h e nematic mesophase, i t then t u r n s o u t t h a t t h e molecular anisotropy of

F i s n o t completely averaged o u t by t h e molecular motion a s i n an

isotropic liquid. To stress this point the tensor F as given by 3) equation (1.11) is often called a partially averaged tensor ( 8 . To illustrate the argument let us imagine the following gedankenexperiment ( 9 : we have a rodlike molecule with only two 3) protons interacting via a certain dipolar tensor, this being our tensor F. The proton magnetic resonance spectrum of this (imaginary)

molecule in an isotropic, low viscosity, liquid would contain just one line, but on passing to the nematic state two lines, separated by an amount proportional to the partially averaged dipolar coupling tensor, will appear. The splitting increases as we lower the temperature,

corresponding to the rather obvious fact that in doing so the orientational order of the system increases.

Note also that the partially averaged tensor F has to reflect the symmetry of the mesophase.

and smectics A, F has to be cylindrically symmetric, as shown by equation (1.11), '' (0 F2)
=

If this is uniaxial, as for nematics

irrespective of the symmetry of F . '

The component

~'273

(ti-J) that we measure is a linear combination of the I

molecular components of F and of three order parameters:

Because of the structure of equation (1.12) it is clear that some simplification can be achieved when we work either in the principal 2 1 ' 0 coordinate system of the ordering tensor, i.e. the one in which Do +, 9or in the principal system of F', where F (2,+1) r=O and F (292)' equals (2 9-21 ' and is real. It is usual to make one of these two assumptions,

g u e s s i n g a p r i n c i p a l system from t h e molecular s t r u c t u r e i f t h e molecul e does n o t have s u f f i c i e n t symmetry elements t o determine i t s l o c a t i o n a priori. I n t h i s c a s e we f i n d :

where

2 O0,0 - d2 , o = 0

P2'

and t h e f a c t t h a t

DL (aBy) = d i V n ( B ) exp(-ima m,n

iny)

( c . f . Appendix 1 ) has been used. symmetry, F ( 2 ' 2 ) ' = 0 and we o b t a i n

I n p a r t i c u l a r , i f F' h a s c y l i n d r i c a l

which g i v e s immediately

W f i n d t h e same r e s u l t i f t h e molecule i s c y l i n d r i c a l l y symmetric, e

O2

i n the sense t h a t

0,2

= 0.

I n t h e s e two c a s e s t h e o r d e r parameter

i s t h e n o b t a i n e d a s a r a t i o of t h e a n i s o t r o p y i n t h e p a r t i a l l y averaged
q u a n t i t y F t o t h e molecular one.
-

It i s worth n o t i n g h e r e t h a t t h e o r d e r

a s g i v e n by e q u a t i o n (1.16) i s a measure of t h e alignment of t h e

A s we mentio-

z a x i s of t h e molecular system we have chosen t o work i n .

ned b e f o r e one h a s u s u a l l y two p r a c t i c a l

c h o i c e s : t h e p r i n c i p a l system

of t h e molecular t e n s o r F o r t h e p r i n c i p a l system of t h e

ordering t e n s o r , t h i s l a s t one having t h e molecular long a x i s p a r a l l e l to z. Since we a r e u s u a l l y i n t e r e s t e d i n t h e o r d e r of t h e long a x i s ,

i t i s u s e f u l t o go back t o equation (1.10) and suppose t h a t the

molecular primed system i s always t h e long-axis system. W then e f f e c t e

one more r o t a t i o n , of(a'B1y'),from t h i s primed system t o t h e p r i n c i p a l system of t h e t e n s o r F (double primed system) and then take t h e ensemble average.

I f F" has c y l i n d r i c a l symmetry F ( 2 y p ) " = F ( ~ * O ) 6 t e n s o r i s c y l i n d r i c a l l y symmetric a s w e l l then

PO '

If the ordering

and we f i n d

Even w i t h t h e s e assumptions of c y l i n d r i c a l symmetry we s e e t h a t we need t o know t h e angle B ' between t h e long a x i s and t h e p r i n c i p a l a x i s

An extreme example i s t h e case t h a t t h e two a x i s a r e perpendi-

c u l a r i.e.

6'

= n12.

Then

AFl AFW=-P2/2 and t h e measured apparent

o r d e r parameter i . e . goes from -112 t o 0.

t h e q u a n t i t y on t h e l e f t hand s i d e i s n e g a t i v e and This c a s e i s o f t e n encountered i n e l e c t r o n reso-

nance s t u d i e s f o r d i s c l i k e molecules l i k e vanadyl a c e t y l a c e t o n a t e (40,41).

Another l i m i t i n g case i s t h a t t h e two a x i s a r e a t t h e "magic" angle ( i . e .

2 c o s ~ ' = l / J 3 )so that D (fl1)=O and no measurable e f f e c t 0,o

AF i s

emerging.

A s an example i f F i s a

c13

chemical s h i f t t e n s o r t h e

corresponding l i n e s w i l l be u n s h i f t e d changing t h e temperature and even going t o t h e i s o t r o p i c phase (42). Of course i f t h e p r i n c i p a l

system of F is along t h e long a x i s w e recover t h e f a m i l i a r e x p r e s s i o n

A more complicated c a s e a r i s e s when t h e o r d e r i n g t e n s o r i s n o t

c y l i n d r i c a l l y symmetric (43).
n

W then have two independent o r d e r parae

n

meters t o determine: DL and Re (DL ) and a t l e a s t two independent 0,o 0,2 measures a r e obviously needed. Additional complications can a r i s e i f t h e molecules considered a r e not r i g i d ( 4 4 , 4 5 ) . More assumptions a r e then r e q u i r e d t o keep

t h e problem manageable, such a s assuming t h a t t h e i n t e r n a l motions a r e decoupled from t h e r e o r i e n t a t i o n of t h e molecule a s a whole and t h a t they a r e s o f a s t on t h e experimental time s c a l e t h a t only t h e i r average e f f e c t need be considered. W s h a l l n o t t r e a t t h i s problem i n more e

d e t a i l s i n c e i n t h i s t h e s i s we d e a l mainly with r i g i d molecules and we s h a l l i n s t e a d r e t u r n , f o r t h e sake of c l a r i t y , t o t h e s i m p l e s t case of a r i g i d c y l i n d r i c a l l y symmetric molecule with a c y l i n d r i c a l l y symmet r i c t e n s o r F. I f w e consider again equation (1.16) we have t h a t t h e q u a n t i t y

AF' i s a molecular c o n s t a n t , r e l a t i v e l y independent of temperature and

o t h e r s t a t e v a r i a b l e s , s o t h a t i t should only s c a l e

P: 2

i f we p l o t
-

AF
vs.T.

a g a i n s t temperature we w i l l have t h e same shape of t h e p l o t of P

However, i f t h e a b s o l u t e magnitude of t h e o r d e r parameter has t o be determined, t h e problem i s how t o f i n d t h e value of t h e molecular

quantity

F in the nematic phase because, even if the free molecule '

one is known, local field effects can a l t e r i t in an unpredictable manner. Up to now we have i n fact been talking of a single molecule quantity F,

while it i s clear that what we measure i s a macroscopic quantity

corresponding to it and containing contributions from a l l molecules in the sample.
Our discussion is thus fitting to the ease in which the

i n t e m l e c u l a r couplings have a negligible effect on the quantity F as, for example, can happen in practice for the magnetic susceptibility X , (1).

In thia case the macroscopic tensor is simply

where p i s the number density.

In other

cases i t i s clearly possible to measure a macroscopic

quantity ( d i e l e c t r i c constant, refractive index tensor e t c . . . )

but the

relation of this to the corresponding malecular property is much more complicated and usually unknown.

W a i s often done i s to invoke some ht

See as an

model for the local field ox some empirical expressions.

example the use of Vuks (46) and Neugeubauer (47) equatFons by

Chandrasekhar and his group when F i s the e l e c t r i c polarizability (48).
The experimental method mainly discussed in t h i s thesis will be electron
spin resonance spectroscopy [ 38 )
I (

We s h a l l d e a l with very d i l u t e
In our ease

~o-~H) solutions of s t a b l e free radicals ( s p i n probes).

then we can s a f e l y rule out intermolecular or local field problems but

we have of courde an additional complication in that we are measuring

properties (ordering parameters and correlation times) of a solute

/

, '

molecule and not of the liquid crystal itself.

We shall return to thia

problem in Chapter 4. It is maybe worth while to point out here that most of the techniques measure only the order parameter
P

2'

Apart from NMR

experiments involving solute molecules with the aim of determining their structure (43,44), the assumption of cylindrical symmetry is In this case the apparent

usually made and equation (1.18) is used.

(2,2)' ! , 2 /F(2.0)'1 092 We shall study the effect of deviations from cylindrical symmetry in more order parameter measured is in error of an amount

%IF

detail in Chapter 3. Even if the molecule is cylindrically symmetric it would be

-

of some importance to determine the higher rank order parameters P4,P6etc. However, as directly measurable quantities of rank higher than two are rather difficult to find, one has to turn to indirect methods. As an

example a technique able to measure the mean square value of a 2nd rank
-

molecular quantity:will yield values of P prove that

4'

It is in fact easy to

There are at present only two techniques used to measure P One is Raman scattering, used by Priestley and Pershan (50,51).

4'

The

molecular quantity F' is in this case the differential polarizabllity tensor for a certain localized Raman active mode. like N-(p'-butoxybenzylidene)-p-cyanoaniline For example a probe

is dissolved in a nematic

and the strong line coming from the C E N stretching mode is studied.

The other technique consists in analyzing the electron resonance linewidth of a radical dissolved in a liquid crystal and has been developed by our group (52-55). It is worth pointing out that both

techniques have been used to study solutes and not the liquid crystal directly and also that both of them rely on certain assumptions. In particular the first one is complicated by local field and multiple scattering problems () 1. We shall discuss in detail the second technique in this thesis; for the moment we note that 7 measurements on exactly comparable 4 systems, which would allow a cross check of the two different techniques, do not exist as yet. This comparison would be particularly interesting

in view of some surprising results obtained by using the Raman scattering method. For instance negative values of P have been obtained
4

(50,51), in contrast with the predictions of the current theories of the nematic phase (c.f. Chapter 2 . )

CHAPTER

TWC)

A STATISTICAL THEORY OF THE NEMATIC PHASE

2.1

Introduction

The task of developing a molecular theory for phases of anisotropic particles seems at first sight a formidabl~: one. de Gennes ( ) 1:

To quote

"The statistical mechanics of l i q u i d s is difficult;

While t h i s is

the statistical mechanics of nematics i s s t i l l worse".

certainly true in general, fortunately we can take liquid c r y s t a l s

.

also as an illustration of the so-called inventor's paradox ( 5 6 ) i . e .

that somtimes a more complicated problem is e a s i e r t o solve than a

simple one.

To be more s p e c i f i c , we have to define what

lar theory of l i q u i d crystals.

we

mean by a mlecu-

Let us f i r s t s t a t e that we are not

really aiming a t a theory capable of predicting, for example, all the

various phase transitions and i n particular t h e solid-nematic or the liquid-gas transition for an assembly of anisotropic p a r t i c l e s .

As is well known these problems are p r a c t i c a l l y untractable even for

systems composed of much simpler molecules.
What we are seeking

instead

is a theory able of describing the most characteristic

That is, first of a l l , that they

properties of liquid c r y s t a l s .

undergo an orientational transition from a disordered, isotropic phase, i . e . an ordinary liquid, to a phase where an orientatianal order parameter such as

P2

i s different from zero.

Apart from t h i s clualitative

prediction of the existence of such a phase transition, a theory of the nematic mesophase i s expected to predict, or a t least reproduce

the behaviour and the values of the relevant thermodynamical quantities at this transition. In particular the following experimental

observations should be explained by the theory: the isotropic-nematic

(I-N) transitions are of first order type with latent heats typically
of 1 K J mo1-I (32), small volume changes i .e. . entropy changes i e.
-

0.3% (57), small

s ( ~ ) / R 3 () The value of the order parameter a. 6.

3) 2 at the I-N transition is usually found to be about 0.4, ( 2 .

Finally values of the higher rank order parameters such as P 4 , though ) difficult to obtain experimentally (c.f. Chapter I , are very valuable in verifying theoretical predictions. Having said what we expect of a theory of liquid crystals, we shall proceed to make a brief survey of the existing theories. These

have a relatively long history dating back e.g. to Born in 1916 (58) and Zwetkoff in 1942 (31), although the first really successful attempt at developing a theory was made by Maier and Saupe (M-S) in 1960 (57). To have a liquid crystal it is quite clear that a necessary condition is a strongly anisotropic intermolecular potential. Maier and Saupe

based their approach on the assumption that the anisotropic potential between two molecules, considered rigid and cylindrically symetric, comes solely from London dispersion forces. This attractive intera-

ction is then treated in the molecular field approximation, generalizing the well known Weiss model of ferromagnetism ( 9 . 5) The spirit

of the molecular field treatment is to replace the detailed interaction between molecules (spins in the case of magnetism) by an effective potential or pseudopotential representing the average interaction of all other molecules with the one in question.

The M-S anisotropic pseudopotential obtained in this way has the form

where x is the cosine of the angle between the symmetry axis of the molecule and that of the mesophase, A is a molecular constant, propor-

a tional to the square of ( -a ), the anisotropy in the polarizability II I

tensor of the molecules and V is the volume of the system. The remining &anti ty , P2, is just the normal order parameter defined in Chapter 1. It is determined from the consistency condition as

2 -

-I
$

p2 (x) exp{-$u (x) ldx

expl-8U (x) ldx,

where

-1lkT.

It is easy to see that

Y2-0 -

is always a solution of equation (2.2),

corresponding to a disordered system. As we shall discuss later (for a mathematically equivalent model) there is another solution with

P2

#O, corresponding to the nematic mesophase, which exists below a

certain critical temperature T K'

The value of

P2 at the transition

is predicted to be 0.43 (57) in reasonably good agreement with experiment. Another prediction of the M-S theory is that

P2

should be a

universal function of the reduced temperature T* = T/T~,where TK is the absolute nematic isatropic transition temperature. we compare a plot of In figure 2.1

P2 vs.T*

according to M-S theory with some typi-

cal experimental results. We see that the agreement between theory and experiment is generally good but also that there is not really a universal curve. Another prediction is that the entropy at the

-1- 0

-0.8

-0.4

--O'*
-0

4,41-~imethoxyazoxybenzene o 4,4'-Diethoxyazoxy benzene A Anisaldazine r 2,4 - Nonadienic acid x 2;4-Undecadienic acid

1

90

92

1 0 0 TITK 94 96 1

98

100

Figure 2.1

The orientational order

?i2

plotted as a function of the The

reduced temperature for a number of nematogens (2).

continuous line is the prediction of the Maier-Saupe theory.

transition s(~)/R should be 0.42.Again this isin qualitative agreement with experiment. Although very successful and widely used, the M-S theory has been subject to some criticism. For instance, M-S theory does not

appear to consider steric or indeed any type of intermolecular forces except dispersion.

6) Kaplan and Drauglis ( 0 have noticed that an

order of magnitude calculation based just on dispersion forces gives a far too low value for the interaction parameter A as compared to the value determined from the transition temperature T K' Their calcu-

lation is itself open to criticism since the actual values of the molecular polarizability anisotropy cannot be extracted from the refractive index data with any certainty because of local field complications None the less there is some doubt the polarizability alone is responsible for the existence of liquid crystals. Other discrepancies, between the Maier-Saupe theory and experiment this time, come from the fact that the volume dependence of the transition temperatuis nearer to four (61) re, as measured by the value of (alnT/alnV)P2 than to the value of two predicted from equation(2.1). Apart from dispersion forces the other obvious:candidate as an orienting factor and as responsible for liquid crystal formation is a steric (repulsive) one in combination with ghe anisotropy in the molecular shape i e with the fact that nematogen molecules are rather .. elongated and similar to rigid rods. Many authors, starting from

Onsager in 1942 (62) have shown that systems of elongated objects interacting with just a hard potential i.e.

U12
=

if there is overlap
0 if there is no overlap

can give a transition to an oriented phase when the density of elongated objects becomes sufficiently high. however, have not been extremely successful. The steric theories (62,63), Firstly it is clear that

purely steric models give a phase transition independent of temperature and so they are rather inappropriate for the treatment of therrnotropic liquid crystals. Apart from this difficulty, they predict values

of the order parameter at the transition far too high ( 4 . 8 ) and also the entropy change is predicted to be much larger than observed.

A Monte Carlo calculation (64) has been performed on a system of

hard spherocylinders (65) but failed to exhibit an orientational phase transition even after over 5 0 x 1 0 rated. While then it seems clear that steric forces as well as dispersion are relevant to describe interactions between nematogen molecules it is likely that a blend of the two and possibly of other factors such as deviations from cylindrical symmetry, flexibility, effects of electrostatic multipoles will have to be taken into account before we have a proper understanding of the relation between the nematic behaviour of a molecule and its structure. In view of all these com-

configurations had been gene-

plications it may now become difficult to see why the Maier-Saupe theory is so successful. The reason is probably a very simple one,

namely that the M-S potential has the correct functional form.

A simple

heuristic argument to illustrate this point is that the potential for a rod in a (molecular) field of uniaxial symmetry will be a function of the angle 6 between rod and field symmetry axis only. This means

that we can expand U (x) , where x=cos B

, as

a sum of Legendre polynomi a1 :

and if we approximate t h i s series with the first sympetry af lowed term

then U(x)%Pp(r).

We do not know anything about the coefficient c

but on the other hand, since ~ ( x )has t o vanish in the isotropic phase,
it is obvious that the simplest correct a~sumptionis that c will be

proportional to the order parameter i,e.

which is essentially the Maier

Saupe result but where now c ' is

essentially an a d j u s t a b l e parameter.

A general mean field approch has been developed by Humphries,

Jams and Luckhurst (H-J-L) (33).

Their treatment consists of writing

d m the intermolecular potential between two molecules

as

a Pople

expansion i . e . expanding the p o t e n t i a l i n a complete product basis

set of spherical harmonics of the orientations of the two molecules

(66,671.

Three successive averages are then taken to obtain the The f i r s t i s over a l l t h e

pseudopotential for a s i n g l e m l e c u l e .

possible orientations of the intermolecular vector in space, the

second over a l l the orientations of the second molecule in the molecu-

lar f i c l d and the third average over the intermolecular distances.

Their result is

U(X) =

1
L

FL p L W ,

(t even)

where

L are the orientational order parameters determined self-

consisteatly by a s e t of equations similar t o equation (2.23 ;

L is an average interaction coefficient, which is a function of moleHumphries,

cular properties as well as of the volume of the system.

James and Luckhurst studied in particular the effect of retaining terms

up to L=4 on the properties of the mesophase,
They found that in many

cases the quartic term is necessary to reproduce quantitatively the

temperature dependence of transition ( 8 . 6)

5 and i n particular

its value at the

assumed to be uL

- qv-'

In the H-J-L theory the volume dependence of uL is

A fitting of the temperature dependence

of P gives a value of y=4 (691, in excellent agreement vi t h the 2 independent measurement of ( 8lnTldlnV ) - by Me Coll and Shih (61).

P2 Bath in M-S and H-J-L theory the calculation starts from t h e

paisvise intermolecular potential which is then averaged in a certain
way t o obtain

the molecular field.

While this seems a physically

plausible procedure, it is, in a sense, a somewhat unconventional

statistical calculation and a formal justification should be provided.

It is clear in fact that in general one has just a total potential

energy for the system and that the various averages have t o come from

a proper partition of this total potential.

It would be more

satisfactory t o start Prom the partition function for a system of anisotropic particles and proceed by successive, well-defined appro-

ximations to obtain an effective orientational distribution and

partition function. txeatmen t

In Section 2.2 we provide this more rigorous

2.2

A Molecular Theory of the Nmatic Mesophase

The configurational partition function for a system of N

interacting rigid particles with positions defined by a v e c t o r R and

orientation described by a set of three EuPer angles ( a , % , y ) ~ R can be

written as (70)

The notation {dR) and Cdn3 is anabbreviation for the volume elements
3 3 in configurational space i . e . {dR) = d R1... d
As usual, B d /

and {dR3=di71.

..a%m

{kT) where k is the Boltzmann constant and T the

The potential energy II(R

temperature.

I...s,PI.*.QN) o f the system

i s a function of both the positions and the orientations of all the

molecules,

In general these will be coupled in acomplicated way and

the problem will be practically insoluble but, on the other side, we

expect that liquid crystals can b e described e f f e c t i v e l y in terms of

an orientational potential only.

In this section

we start from

equation (2.6) and derive an effective orientational potential and

partition

function.

This is then compared with the man f i e l d

results of ~ttmphries-~ames-Luckhurst(33) and Maier-Saupe (571,

which were obtained in a mre ad hoc way.

One

of the aims w i l l be

to define the various approximations involved to obtain these results

and to indicate possible generalizations and improvements.

The spirit

is similar to t h a t adopted i n a paper by Schultz (713, w i t h the

difference however that he essentially poses the problem and then turns
to

a specific model where the molecules can only assume discrete

I n s t e a d we shall aim t o be as general as

positions and orientations,

possible.

As a starting point let us assume that the total potential can

be separated into a part U S depending on the position of the molecules

only together withananisotropic part UA depending on the distance between molecules as well as their orientation:

u = uS

+UA.

This assumption allows us to rewrite the partition function as

where

is the partition function relating to the position of the

molecules (their centres of mass say):

28 and the upper bar in equation ( . ) indicates a positional average.

We see then that even after separation of the potential in two parts
we are not yet able to factorize the totalpartitionfunction into positional and anisotropic contributionsas we should like. this Schultz (71) proves that To achieve

where

is an orientational partition function, and then simply takes (2.10) as an equality. This has the obvious disadvantage of not indicating

how to improve the approximation. Instead we shall then develop exp(-BU A

) in cumulants as follows:

exp (- BUA)

exp { cexp (- 8UA)- 1>C 1,

(2.12

where <.....>

denotes a cumulant average ( 2 . 7)

The first few terms

in this expansion are:

= exp(-=

a)

. exp(correction

terms)

We see then from equations (2.8) and (2.13) that the inequality (2.10) is indeed verified, being the quantity (correction terms) a non negative function.

We would now like to retain, for the sake of simplicity, only

the first non-zero cumulant, i e U ~ 'which amounts essentially to a ..

high temperature approximation. sight, since

A

This does not seem reasonable a t f i r s t

i s an energy r e f e r r e d t o t h e whole system and s o i s

On t h e o t h e r hanc? i t

p r o p o r t i o n a l t o t h e number of p a r t i c l e s N.

is

c l e a r t h a t t h e v a l i d i t y of t h e expansion i.e. t h e s t r u c t u r e of t h e equations t h a t we g e t should be independent of t h e number of molecules i n t h e system. The s o l u t i o n t o t h i s apparent paradox o f t e n appearing

i n t h i s s o r t of c a l c u l a t i o n i s simply t h a t t h e second term i n t h e se-

e .U

i s s t i l l p r o p o r t i o n a l t o N and n o t t o N~ because of
It then t u r n s

t h e f i n i t e range of t h e i n t e r m o l e c u l a r p o t e n t i a l (73).

o u t t h e c o n d i t i o n of a p p l i c a b i l i t y of t h e expansion i s t h a t t h e s i n g l e p a r t i c l e energy, i s small compared t o kT.

U A / ~ ,i . e .

For t h e moment (but o t h e r s

we s h a l l then r e t a i n only t h e f i r s t term i n equation (2.13) terms can be t r e a t e d i n t h e same way) and t r y t o g i v e
A

i n more d e t a i l .

Let us f i r s t i n t r o d u c e a pairwise a d d i t i v i t y assumption:

which allows

F to A

be w r i t t e n as:

The p a i r d i s t r i b u t i o n f ( i , j ) g i v e s the p r o b a b i l i t y of f i n d i n g t h e c e n t r e s of mass of molecules i and j r e s p e c t i v e l y a t R. and R

1

j'

Since we have assumed I1

IIti

to he isotropic, f ( i , j ) w i l l depentl only an

R j l = R.. and w e could write f ( i , j ) = ~(R..)/v~,where V is the 1J 1J

volume of the system and g(R) the pair distribution function .of the centres of mass.

For large separations f (i, j ) goes as f /V

Notice

that, the isotropy in f (i , j 3 comes d i r e c t l y from equation ( 2 . 7 ) without

further assumptions o r , viceversa, that within that approximation it

is not possible t o allow for deviations f r o m spherical

sya3metry

of

g(R) even though such anisotropy e x i s t s in liquid crystals.

Since the explicit forms of U ( 3 , j )

and f ( i , j ) in equation (2.16)

are not known, to proceed further we s h a l l try ta extract as much

information as possible from t h e i r general properties.

We can write

the anisotropic intermolecular potential ~ ( , ) in a (complete) product i j

basis set of Wignet rotation matrices (39) in the orientations of the

two molecules i . e . as a series of terms of increasing

tensorial rank

(66,671,

Choosing a coordinate system, which we call the intermole-

cular or IM system, with the z axis along the R.. :Ri 1J this has the form

- R. 3

vector,

where

bli,R

defines the orientation of molecule i in this system and L2 = 0 term is

the prime on the summation indicates that the L1 = 0 ,

excluded, since it constitutes part of the scalar potential.

To avoid the inconvenience of every p a i r of nmlecules defining

a d i f f e r e n t coordinate system we now rewrite equation (2,171 in a
laboratory system using the relation (see Appendix 1)

where QR .-L and Qi-L represent a rotation from a laboratory system

1 J

to the IM system and to a coordinate system fixed on molecule i respectively. This gives:

Substituting in equation (2.16) we obtain

(.) Since f R .
1J

is isotropic, the integral over the two Wigner

rotation matrices gives imnediately

and we can rewrite the positionally averaged anisotropic potential as

i<j

L,nn 1 2

' I

Here we have defined

and written

(Qi-L) as ( ) , since from now on all the orientations i

are measured from the laboratory frame and no confusion should arise. From equation (2.22) we see that F is a purely orientational A potential: each molecule can interact, through its orientation, with any other molecule, with an average "strength of interaction" coefficient c coming from the smear over all the molecules of the Ln1n2 local interactions. Thus, even very short range interactions give rise to an orientational potential not dependent on distance i.e. of infinite range, while of course the coefficients nitesimal in magnitude. are infitnl"2 For instance, if we suppose the molecule can

interact with only z neighbours and with an average potential E, the coefficients c will be of order ZE/~N. Ln1n2 Substitution of the expression for u in equation (2.11) shows
A

that the problem o the evaluation of the orientational partition f function has been reduced to calculating

where

J ' D

UA =

C'

cLn n 1 2

1 (-lq

,nl

() ! i D

Q 9n2

() j.

(2.22

The similarity between this potential and that used to treat orientational and magnetic transitions on lattices is quite striking (74-76). Eowever, a difference with the lattice treatments that is worth pointing out is that they usually restrict the orientational (or spin) interactions to nearest neighbours only, i.e. they have a local interaction, while here, since we have averaged over the distances, only infinite range orientational interactions appear/. We could now go on with the potential in equation (2.22) and keep the treatment quite general but, since this complicates the bookkeeping without changing the physics of the problem, we shall restrict ourselves to the case of cylindrically symmetric molecules (i.e. nl=n2=O) with second rank interactions only (i.e. L=2). potential will be Then the anisotropic

Here c2

22' c~~~ and the angle Qi-j in equation ( . 6 ) In equation ( . 6 ) 22'

. J

is the difference

of the two rotations Qi and Q..

we have rewritten

the potential by using the spherical harmonics addition theorem in a form that puts in evidence that the potential is invariant under rotation of the laboratory system and of the long axis of the molecules. We now write the Wigner rotation matrices in equatioi (2.26) as

where o (i) is a new orientational function and D a separation q parameter to be determined variationally later on. We shall show that this provides a way of decoupling the potential into a "molecular field" part, in agreement with the Humphries-James-Luckhurst one (33),and into a correction term. We shall also show that the separation

parameter D is, in the molecular field limit, just an orientational average of a Wigner rotation matrix i.e. an order parameter. To begin

with we substitute equation (2.27) in equation (2.26) and find

where

provided the number of particles is so large that N N I % (-)N

, and

We shall take Uo, which is linear in the orientational functions, as our zeroth order or unperturbed potential and U 1' which is quadratic

in the same quantities, as a correction. Accordingly the orientational partition function can be written as:

where the upper bar indicates the average over the unperturbed s t a t e

and where we have defined a zeroth order N-particle partition function

@N as

Let us now investigate more closely the parameter D.

We f i r s t notice

that t o obtain an effective potential for a mesophase having cylindrical sy-try

the potential
we have somehow t o break the

full rotational symmetry of

way of doing t h i s is t o include an external W field term of the proper symmetry as 5 P2(cosBi) where Bi is the

BA. A natural

i=l

angle between the field direction, taken as the laboratory z axis, and

the i-th molecule symmetry axis (75-77).

Of course, since we are

interested in the field-free system, we then let the coupling parameter

<

go to zero.

If we assume t h a t t h i s limiting process does not inter-

fere with our successive mathematical steps, we can take t h e limit a t

once i,e. in practice avoid

writing down the field term, remembering

however that the potential has now t o conform t o the broken symmetry. In our case this means to be invariant under rotation about the z axis

and under reflection in a plane perpendicular t o i t . After symmetriza-

tion the potential Uo becomes

This in turn gives

where

I,+}a is

the N-th power of the single particle pseudo-partition

function

The value of D is n w determined by minimizing the appropriate unperturbed themdynamic potential which, since we suppose the volume is essentially constant

, is

the Helmoholtz free energy A .

From t h e definition 0.f A we have

Ag

= -kT ln'PN

--1
2
A

N ~ DNkT ln+. ~ 2

necessary condition for a minimum is that

Thus, to minimize the free energy, the separation parameter satisfy the consistency equation

Tl

must

or, in other words, the best separation parameter is juyt the order parameter

c.
L

To confirm that

D=F,
L

is a true minimum the second de-

rivative of the free ' energy A has to be calculated and this )=2 DF will give the region of stability of the phase. We shall see an example in Section 2.4. Let us now consider the correction term exp (-BU, 1. In lattice

t:t2

problems a method for successive systematic approximations to this type of integral has been developed (74,75). The physical idea is to

consider progressively larger clustersof molecules in the molecular field produced by all the others. The unperturbed potential Uo corresponds from this point of view to a cluster of just one molecule i.e. to the mean field approximation. The other terms allow for the short range effects, since the interactions within a cluster are treated in detail. Since our orientational potential is of infinite range

we do not expect the correction term to be important, by analogy with, for example, ferromagnetic problems, where it has been shown by Kac

( 8 that the mean field solution becomes exact when the range of the 7)
interaction goes to infinity. To check this point consider the first cumulant of U1 i.e. approximate exp (-BU;) by exp (-$El) If we rewrite U1 as

we see imediately that U %O since at this level of approximation 1 as we have .shown before.

DZ~,

One should then retain at least terms up to f3 Z

and since we have rejected terms of this type during the course of the derivation, to be consistent we shall content ourselves in retai.

ning only the unperturbed term U as an approximation to the potential i J 0 A' In summary, we have shown that starting from the total partition function for a system of N particles with anisotropic interactions it is possible to obtain an approximate orientational partition function

QN.

Corresponding to this partition function we have an effective

0

orientational potential U

which, in the case of cylindrically symmetric

molecules with second rank interactions only is

where c2 (or Nc ) gives the average strength of the interaction, P

2
1

is

an order parameter and f3. is the angle between the symmetry axis of the molecule and that of the mesophase. It is maybe worth pointing out here that the decomposition scheme we have adopted to reduce the total partition function is not the only possible one. As an example we could have introduced the the integration over the intermolecular

separation parameters before separations.

In general it is difficult to compare different decompo-

sition schemes because they usually do not involve just a different ordering of the same operations but different sortsof approximations or physical assumptions as well. In any case, at the first level of

approximation which is the one explicitly treated in this Chapter,the molecular field result is obtained anyway.

We are now in a position to compare our results with those

obtained by Humphries, James and Luckhurst (H-J-L) and by Maier and
Saupe.
Ta do this, we notice first that H-J-L have not only retained

t e r n with L=2, but, at least formally, all the terms with even L in

the potential.

W can do the same, by having separation parameters e

The symmetry argument

which are L dependent i . e . DL, instead of D.

L* where L must be even if the meaophase has uniaxial symmetry. If we

consider now the single particle partition function

and the minimization of the free energy would then give D

-{9N}1'N

and
.

d e f i n e u c N we see that we have e s s e n t i a l l y recovered 8-J-L results 2 2

i.e. the e f f e c t i v e potential of a molecule in the field of all the

others 'is

1'
L

P P UL-LI (XI.

(L even)

(2.42

Restriction to the first term, L=2, as we did before givea a potential

formally analogous to the M-S one.

Having established the equivalence

of the treatments at this level of approximation, we may underline

some other points.

In t h i s treatment: we have obtained H-J-L as a f i r s t

order approximation in a rather general expansion.

In particular,

our partition function is the proper one in the sense that we can obtain

all the themdynamic quantities from i t by using the standard formulae

(70,?3).

This sheds light on a sort of paradox of the usual mean field

theory, where one has sometimes t o use what seem ad hoc relations.

In fact, since the mean field treatment concentrates on the single

molecule potential U(x),

what is naturally evaluated is not the partition

function but what we have already called the pseudopartition function

i.e.

4 and not

Thus, as an example, the quantity -kT In4 will not be just the free -2 energy, but a term u2P2/2 or in general added as is usually done in practice. .

1 uL1*/2 will L
L

have to be

2.3

The Maier

Saupe hdel

The simplest effective potential that gives a reasonably good description of the liquid crystalline phase has the angular dependence of a second Legendre polynomial. We have shown in the previous section how this potential can be obtained in a rather general way. The

potential is formally equivalent to the one originally obtained by Maier and Saupe (57) and it could also be obtained on purely heuristic grounds (c.f. Section 2.1) or by using the Humphries, James and 3) Luckhurst method ( 3 . In this section we treat in some detail the

problem of calculating the partition function, the moments and the order parameters for this Maier-Saupe like model. Since this has

already been studied, the justification for the following Section lies in the fact that many of the results, particularly on the order parameters, will be used again in this thesis and that to obtain them we give a number of new and hopefully useful expressions. The single particle partition function for the model is
@ ,

exp(~u~F$/2)4,

(2.43

where we have defined a (reduced) pseudopartition function ZO as

and 5=

-38

T u2P2

--3fiC

is the interaction parameter, which is positive

if the pairwise potential favours parallel alignment. Even for this simple angular dependence the pseudopartition function is not an

elementary function of tGe interaction 6 .

We can however express Z

in terms of known functions.

Rewriting ZO as

we see that thfd is j u s t t h e integral representation of the Kummer

confluent hypergeometric function M(a,b,z) (79,80)

if

we take

a=1/2, b=3/2, z=E.

In other words

As well as its compactness, t h i s expression has the advantage of

allowing us to use the large anmunt of resulta existing for these

special functions, as we shall see in t h e following.

Some of the

mst useful properties of the Kummer functions are listed in Appendix 2.

Since, as we said, 2 is not an elementary function of 6 it is

0 ,

difficult t o have a feeling for how i t behaves.

To t h i s purpose a

simple approximate expression for ZO would be useul

, and

one is obtained

in the following way.

Notice first that a trivial power expansion

2

of ZO is obtained from equation ( 2 . 4 5 ) by expanding em (Ex 1 and

integrating term by term i. e.

is given in Egure 2.2.

A compact expression for all the even moments of t h e orientatio-

nal pseudodistribution f (x)==exp(cx

perties of the moments are
Ktmrmer

can be obtained using the pro-

functions.

By definition (see Chapter 1) the

It is easy t o aee

that

or using the differentfation formula for M(a,b,z),

Substitution in equation ( 2 . 5 4 ) gives the desired expression for the

moments.
An expression for a l l the Legendre polynomial averages can

be obtained at once from equations (1.7)

and 1 2 . 5 4 ) .

Asymptotic expressions for high values o f the interaction

parameter can be obtained from the corresponding formulae for t h e Kummer

functions reported i n Appendix 2.

At high order

where (C),

= C(C+l)...

(C+n-l)

and ( C ) O = 1.

Figure 2.2

The orientational order parameter P (x), calculated for a 2 2 pseudopotential - -= Ex , is plotted as a function of kT the interaction parameter

5.

In particular one obtains

-

?=

2 1~(25

+o(~-~),

Or, equivalently,
-

P2
-

1 1

3/(2E)

2 3/(45 ) 2
+

0(~-~),

P4

515 + 25/(45

0(5-~).

We shall return to the very high order limit in Chapter 5, where we show that the same asymptotic results for the order parameters can be obtained also for the more general potential U(x) Even retaining just the term in 5
=

-1

1 CL
L

PL (x).

the asymptotic expression works 6.

rather well, especially for P 2 ' when[> of P 2

In Table 2.1 some values

and P obtained in this way are compared with values obtained 4

by numerical integration. Low order expansions for the parameters from equation (2.48).

5L

can be obtained easily

We just give the result for

P2 :

When the limiting expressions for high and low order are not applicable one has in general to resort to numerical calculation to determine the order parameters. Since it is sometimes necessary

(e.g. in linewidth studies, c.f. Chapter 5 and 6) to generate a large number of these order parameters P even of relatively high rank L (up to L-30 say) the problem of their efficient

TABLE 2.1

The order parameters

P2

and

54

calculated by numerical

integration and from the asymptotic expansion eqaation(2.59) truncated at the first term

computation poses.

One possibility is to generate the moments xL in

a recursive way, which is very easy to do,and then use equation (1.7)
-

to obtain the P from these. However this is not so convenient from L the numerical point of view since it implies the evaluation of the (small) coefficients a in equation (1.7) which are ratios of large L,r factorials. Moreover it is more satisfactory to have a recursion relation for order parameters without intermediaries. This prompted us to derive such a recurrence relation as we shall now describe. As it is well known (37)

L P (x) L

(2L-1)x PL-l(~)

(L-1) PL-2(~).

Multiplication of both sides by the distribution function and integration gives

The problem is then to evaluate

5L '

where L'

L-1 is odd.

Integrating by parts, using the fact that P (0)=0 for odd L', and L' remembering the relation (37)

(L1 odd) we finally obtain

(2.62

(L even)

(2.63

This allows recurrent computation of every

PL

in terms of Z

0'

As an

example we can write down the explicit expression for the first three order parameters i.e.

According to the Maier-Saupe model the knowledge of one order parameter permits the prediction of all the others. (e.g. by using equation (2.63). In figures 2.3, 2.4 the Maier-Saupe orientational order parameters and P
4

P2

are plotted one against the other for positive and negative It may be noticed that

values of the interaction parameter 5.

5 is

always predicted to be positive even when

F2 is

negative (e.g., for a

disclike solute molecule whose symmetry axis tends to be perpendicular to the director).

Figure 2.3

The relation between the order parameters to the Maier-Saupe model for positive

P2

and

% according

P2 '

Figure 2.4

The relation between the order parameters

t o the Maier-Saupe m o d e l , for negative

Y2

and

F4

according

P2.

2.4

The Nematic

Isotropic .Phase Transition

In the previous section we have studied the order parameter P 2' defined by equation (2.52), as a function of the interaction parameter

5.

We have seen that P ( 5 ) is a continuous and differentiable function 2

that vanishes only when 5=0. It is apparent that we cannot just look at its graph (figure 2.2) to locate where the nematic-isotropic transition occurs. What we should do instead is to remember (c.f. Chapter 1)

that a first order transition occurs when the slope of the free energy of the system has a discontinuity. Since we calculate only the orientational contribution to the thermodynamical quantities we can equivalently say that the nematic-isotropic phase transition occurs
I

when the free energy of the mesophase becomes higher than the free energyof the isotropic reference state, which is just taken as zero. In figure 2.5 we plot the orientational free energy of the Maier-Saupe like model (continuous line). It is now clear from this figure that

a phase transition does indeed occur when curve crosses the abscissa.
-

S=Sk

= 2.903, since there the

Remembering that 5= -u

P /kT and noticing 2 2

P at this point is 0.43, we obtain that 2

where T is the transition temperature. K

The transition is of first

order, since the slope of the free energy, when it crosses the abscissa, is not zero, and at that point the order parameter jumps discontinuously to zero. The peak in the free energy at E = . correspond physically ,22

to the maximum temperature, T', at which the mesophase can be thermodynamically stable, even if it is unfavoured as compared to the isotropic phase since T>T K' For T>T1 the mesophase is absolutely unstable ( 1 : 8)

T' represents the superheating limit of the mesophase.

Figure 2.5

The orientational molar free energy for the Maier-Saupe

model.

The continuous line is the free energy of the hypo-

thetical rwsophase and t h e dashed line that of the disordered

reference s t a t e .

A MOLECULAR FIEtD THEORY FOR UNIAXIAL NEMATIC LIQUID CRYSTALS FORMED BY NON-CYLINDRICALLY SYMMETRIC MOLECWLES

3.1

Introduction
An essential characteristic of compounds forming l i q u i d c r y s t a l s

is the rod-like shape of their constituent molecules, w i t h an attendant

high length to breadth ratio.

This comrmn feature has prompted both

experimentalist and theoreticians to suppose that the molecules possess

cylindrical symmetry; an assumption which has many attractions. example, the ordering matrix, often used to describe the partial

For

alignment in a mesophase, contains just one independent element and

this can be determined by a variety of techniques 6 4 3 , 4 9 1 .

The assumpti01

of mlecular cylindrical symmetry is appealing to a statistical mecha-

nician because the pairwise anisotropic intermolecular potential required in any calculation is particularly simple for such particles ( 6 6 ) . However, the molecules forming l i q u i d c r y s t a l s are, in fact lath-like

and certainly do n o t possess the high symmetry which is frequently

assumed for them.

Two p r i n c i p a l components of the ordering matrix are therefore

required to describe the orientational order of a uniaxial mesophase

composed of lath-like malecules although m s t investigators are content

to assme that one is sufficient.

However Alben et al. (82) have estiand f i n d the

mated the ordering matrix for 4,4'-dimethoxyaeoxybenzene

deviation of this matrix from cylindrical synrmetry t o be significant.

Their analysis is open to question (83) and it is regrettable that

there have been no furthex attempts t o determine the entire ordering

matrix for real nematogens.

Fortunately the importance of deviations

from cylindrical synaaetry may be inferred from unambiguous determinations of the ordering matrix for rodlike molecules, such as 4,4'-dichlorobiphenyl (841, dissolved 5n.a nematic mesophase.
In addition these

matrices are found t o be comparable t o that estimated by Alben et al.

for a pure mesophase, In unusual contrast t o the experimentalists, theoreticians have

been more concerned with exploring t h e consequences of deviations from
molecular cylindrical symmetry.

Thus Frieser (85) has shown that a

system composed of particles with a lower symmetry than

Dm h is capable
The

of existing either as a uniaxial or a biaxial liquid crystal.

p o s s i b l e existence of a b i a x i a l phase has been studied in some detail

f o r a s y s t e m of hard rectangular plates using a l a t t i c e model (861,

the Landau approach (87) and the molecular field approximation (88).

These calculations confirm the original prediction of the eventual transition to a biaxial phase although this transition has y e t to be discovered.

Indeed, all known nematics axe uniaxial so presumably

the mesophase freezes before the predicted t r a n s i t i o n to the b i a x i a l

phase can occur.

None-the-less although deviations from molecular

cylindrical symmetry have yet t o be evidenced by the observation of

a biaxial phase

such deviations should influence the order parameters

For example, the devia-

for the liquid crystal in the uniaxial phase.

tion of the ordering matrix from cylindrical synmetry is clearly determined by the molecular symmetry and the element of the i n t e m l e c u l a r
potential.

Straley (88) has reported calculations of the ordering

These calcula-

matrix for an ensemble of hard rectangular particles.

tions are of some interest but they may not be particularly realistic,

especially as recent experiments have indicated that dispersion forces

may make a dominant contribution to the anisotropic i n t e m l e c u l a r

potential ( 8 9 ) .

In this Chapter therefore we shall develop, within

the molecular f i e l d approximation, a theory for non-cylindrically synrmetric p a r t i c l e s interacting via a completely general intermlecular
potential.

We! then obtain a series expansion for the orientational

The expansion coefficients

pseudo-potential in the uniaxial phase.

are subsequently determined by assuming th-at dispersion forces make

t h e only contribution t o the anisatropic intermolecular p o t e n t i a l .

We are then a b l e t o investigate the influence of deviations from

molecular cylindrical symmetry on the nematic-isotropic transition

and the various order parameters.

Finally we compare these p r e d i c t i o n s

with the data available for the nematogen 4,4'-dimethoxyazoxyberizene.

3.2

The otientational Pseudo-Potential

As we have seen in the previous Chapter the simple procedure

employed by Humphries, James and Luckhurst (33) to obtain the orienta-

tionaf pseudo-potential yields the same results,at the first l e v e l

of approximation, as a more rigorous approch based on the factorization of the total partition function.
For the sake of simplicity we

shaf 1 then use in t h i s Chapter the H-J-L procedure, which amounts t o

taking three averages of the intermolecular potential.

The f i r s t of

these is over a l l orientations of the intermolecular vector, the

second over a l l orientations of one particle and finally over the intemofecular separation. We begin with the intermolecular potential which, for particles
of general shape, we expand i n a product basis of Wigner rotation

matrices ( 3 4 ) as

where R

12

is the separation between molecules 1 and 2 (67).

The

orientation of malecule i i n a coordinate system containing the inter-

molecular v e c t o r as the z axis is denoted by D

i-R'

The energy J12

is clearly invariant under rotation of the coordinate system about

t h i s z axis and so we restrict the summation t o those terms with m
equal t o -m

(67):

where u

L1L2mlnZ- U ~ 1 ~ 2 m w l n 2 .

It is convenient to define the molecular orientation in terms of a

conrmon coordinate system and so we transform t h e potential U

12

to a

laboratory frame.

The choice of this coordinate system is determined

by the symmetry of the liquid crystal phase and for a uniaxial mesophase we s h a l l take

the laboratory z-axis to be parallel t o the symmetry

to

axis of the mesaphase i . e .

the director.

The transformation of W

12

is carried o u t by performing the rotation from the intesmolecular

vector to the molecule in two steps using the relationship

Here the subscript R-L denotes the rotation from the laboratory to
t h e intermolecular frame while 1-L indicates that from the laboratory
t o the

molecule

coordinate system.

The intermolecular p o t e n t i a l

becomes

and we may now take the average over the orientation of the inter-

mlecular vector.

If the distribution function for the intermolecular

vector is orientation independent (see the discussion in Section 2.2) then w e may take advantage of the orthogonality of the rotation matrices
to evaluate the ensemble average

Ll*

Dql,m(R~-~) Dq2,-m R-L

' 2 *

(n

as

The partially averaged potential may then be written as

The next stage i s to average over the orientations adopted by particle 2 and so we require

$q *. . Within n,
7

the mlecular field

approximation t h i s average is taken t o be independent of the orienta-

tion of mlecule l and is to be identified with the normal ensemble

average.

Since we are only concerned with a b

average D

' 9 r n2

uniaxial mesophase the -h will vanish unless q is zero and L is even. The poten-

t i a l is now reduced to

where the prime denotes n sumation restricted t o even values of L

greater than zero.
The scalar co~itxiburiont o the potential is ignored

because we are only concerned with the orientational properties of

the mesophase.

The finat average is over a l l intermolecuZar separations

and gives the orientational pseuda-potential, for molecule 1, as

The expansion coefficients are defined by

where US (R1..

..%)

is the scalar potential of the N molecules in t h e

The pseudo-potential is conveniently written as

ensemble ( c . f . Chapter 2).

where

the Euler angles ( 8 , ~ ) define the orientation of the director in the

mlecular coordinate nya tern.

We have three checks on t h i s form of the pseudo-potential. The

f i r s t occurs for an ensemble of cylindrically symetric molecules for

then the summation in equation (3.2) for t h e pairwise intermolecular

potential is restricted to terns for which both n
we assumed in the previous chapter.

and n2 are zero as

With these restrictions the

pseudo-potential,

given in equation 13.10), reduces to

in complete agreement with the result derived by Humphries, James

and Luckhurst (33).

The second particular case is an infinitely

dilute solution of rrolecules ofarbitrarysbape in a liquid of: cylindrically symmetric particles.

The orientaticnal pseudo-potential for

the non-cylindrically symmetric molecule in such a system is obtained

formally from equation (3.10) by s e t t i n g p equal t o zero and yielding

in accord with t h e result obtained by B q h r i e s and Luckhurst 690).

Finally if the sunnnation in equation (3.10) is restricted to terms

with L equal t o 2, then the pseudo-potential is identical in form t o

that proposed by Stsaley on the basis of synrmetry arguments (88).

We conclude t h i s s e c t i o n by considering t h e symrrmetry properties

of the coefficients

, For i d e n t i c a l particles t h e permutation "LSP symmetry of the pairwise intermolecular potential ensures that (67)

If-

L1I 2m 1n 2 'R1~' .

= (-l)Ll+L~

UL2~~7rm2n1

mI2)
9

and so

-Ph'

The pair potential must also be real and this requires that

L L w n

(R

1 2 1 2

12

) = (-1)n~+"2

(RI2)*;
%lL27"-"1"2

(3.15

consequent 1y

3.3

Paxameterisation

One of our major objectives is to investigate the influence

of deviations from molecular cylindrical symmetry on the various

orientational order parameters for a uniaxial nematic mesophase.

We therefore seek to minimise t h e number of variables in the calculation while maintaining the essential physics.
As a f i r s t approxima-

tion we shall consider only those terms with L equal t o 2 in the expansion of the pseudo-potential.

WE expect the neglect of terms

higher than quadratic to be a realistic approximation since similar assumptions for cylindrically spmaetxic particles have l e d t o results

in reasonable agreement with experiment 132,331.

We can reduce the

number of expansion coefficients still further by appealing to some

specific mdel far the interactions or by imposing symmetry restxictions
on the molecules.

For example i f we follow Straley 188) and take the

particles t o be hard rectangular parallelopipeds then the coefficients

U

29P

are independent of the sign of either p or q and zero if either

of these subscripts is odd.

In fact

and
=

-L (w-B)

12,

where L is thelength, B the breadth and W the width of the p a r t i c l e . This particular parameteris ation is only approximate 188) and in

addition anisotropic repulsive fasces are probably not dominant in

determining the behaviour of real l i q u i d crystals.
derived expressions for the u

We have therefore

for mLecules interacting via dispersion 2qp forces and, aa we shov in Section 3.3(b) these coefficients again vanish

i f either p or q is odd and are also independent of the sign of p or q.

It is p o s s i b l e to place similar restrictions on the expansion

coefficients

%P

'

without appealing t o specific forms of t h e inter-

molecular interactions, by using more formal arguments based an the

molecular symmetry and its influence on the p a i r potential.
For example

if each mlecule has

a centre of

s y m e t r y then (67)

eonsequen t ly

but for this t o be compatible with equation (3.16) the coefficients

must be real.

Further, if the molecules also possess a plane of symmetry

orthogonal to t h e i r z-axes then

and so

However, equations ( 3 . 2 1 ) and (3.23) can only be consistent if both

p and q are restricted to even values ; in addition the coefficients

are independent of the sign of p and q. Given these general restrictions on the coefficients
may write the pseudo-potential as

"UPWe

u(Br) =

L {Do,p * Do,-p ltDO,q(B~) 1 kp ~ (even) l q l *lpla

Do,-q (BY))/

If

we now limit the summation t o those

terms with L equal t o 2 we have

where d2 IS) is a reduced Wigner rotation matrix ( 3 4 ) . The second 'Q,n and d2 cos2Y occurring in t h i s expression rank order parameters d Z

0,o

Q,2

are directly related ta the princi.pa1 elements of the ordering matrix

(c.f. equations 1 9 . .)

In particular

and

W see therefore that the order parameter d2 e

0 2

cos2y reflects the

deviation of the ordering matrix f r o m cylindrical symmetry

Other

order parameters have been introduced to measure the deviation of the

ordering matrix from cylindrical symmetry although they only differ

by simple numerical factors; thus ALben et a k . (82) use the symbol R
t o denote (6 d2 0,2

cos2y while Strakey (88) has chosen U which is equicos2y

valent to J 8 / 3 d 2

0,2

For calculations i t is convenient t o write

the pseudo-potential as

where

and

The pseudo-partition function for this single particle potential is then

where I (x) is an nth order modified Bessel function (80). n

The orienta-

tional molar potential energy U is just one half the total average energy (33),

the molar entropy S is

and so the orientational molar Helmholtz function is

This free energy will be a minimum provided the consistency equations for the order parameters

(3.35 and

fll

dg,2(8)

are satisfied.

Il{b d6,2(8)~ exp{a d2 ( $ ) I sinBdB/Z(a,b), 0,o (3.36

The two consistency equations may be solved in a variety of

ways.
For example we might f i r s t fix the three variables U ~ ~ / ~ T ,

220

ikT and uZz2/kT and then evaluate the integrals in equations (3.35)

and (3.36) for t r i a l values of the order parameters d2

a,o

and dg,2cos2y.

The trial values would then be varied, using a non-linear least squares

fitting routine (331, until the consistency conditions are s a t i s f i e d .

An alternative, and often more convenient: approakh, is to tabulate the

order parameters and thermodynamic functions for a range of values of

the variables a and b,

The ratios u

~ and urn ~ / u ~ ~ ~ 2 2 ' ~ 2 0 of the 0

expansion c o e f f i c i e n t s may then be extracted from the variables w i t h the a i d o f equations ( 3 . 2 9 ) and ( 3 . 3 0 ) .

This exercise is particularly

straightforward when the particles interact v i a dispersion forces because

t h e ratio u 2 2 2 / ~ 2 0 0 is j u s t (uZZO/~200)Z as we shall now show.

Section 3 . 3 (b)

Parameterisation for Dispersion Forces

Here we derive expressions for the expansion coefficients

u

(RI2) when t h e molecules interact via dispersion forces (91). L1L2mln2 Our starting p o i n t is the dispersion interaction energy for a p a i r of

molecules in their electronic ground states ( 9 2 ) ,

where t h e tensor convention of summation over repeated Greek s u b s c r i p t s

is adopted.
The proportionality constant d12 is given by

dI2 =

-E

1r 2f(cl

+ c ~ ) .

(3.38

where

E~

is an energy parameter often identified with an ionization

potential.

The tensarsA and

are defined in terms of the polariza-

b i l i t y tensorsul and a2, for the two molecules, by

and

The scalar product in equation 13.37) may also be written in irreducible

spherical tensor form ( 3 4 ) as

where L may only take values 0 l and 2 because A and B are second ,

rank tensor s The irreducible components A (L'mf

are constructed from

the tensors T and a according to t h e prescription I91b)

where W (abcd ; ef) and C (abc ;de) are Racah and Clebsch-Gordan
coefficients respectively (34).

The values of L1 and q may be

restricted by working in a coordinate system containing the i n t e r n -

lecular vector as z-axis; in this cage the

only non-vanishing

component is

The same restrictions hold for B (L*m) and so

where

Transforming the polarizability tensorsto their respective molecular coordinate system gives the dispersion energy as

U12

(-1)

m-n -n2

(R~ (1") (L2,n2)*] ' L ~ L 12~ a1L . 1 * a2

DL1 (Q1-R) m*nl

This result has the same form as the general expansion of the pairwise intermolecular potential given in equation (3.2) and allows us : o identify the coefficients in that expansion as

of course L

and L may only take values 0 or 2 because the

2

polarizabilisy a is a symmetric second rank tensor.

surviving coefficients % P
are found t o be

Consequently the

after evaluating t h e vector coupling coefficients.

Equations (3.29) and (3.30) now become

and

where

and A is the ratio u

W may therefore fix the parameter A , which is a measure of the e

deviation from cylindrical symmetry, a t the start of the calculation

and from the results, tabulated simply as a function of a , obtain the behaviour of the ensemble as the temperature is varied.

In the n e x t

section we shall describe the results of such calculations and compare

these theoretical predictions with experiment. For this final section we shall denote the order d2 by the more usual P p and t h e secondary 0,o parameter d2 eos2~ by 0,2

~6,~.

3.4 Predictions and Reality

The first stage in the calculation is the identification of the transition from the isotropic phase to the uniaxial nematic phase. This may be accomplished for a given molecular asymmetry by determining the value of a at which the orientational Helmholtz function vanishes, provided of course there is no volume change at the transition. value, aK is then related to the transition temperature TK by The

The results of such calculations are listed in Table 3.1 for a range of

X values including X=O which corresponds to the Maier-Saupe limit for

cylindrically symmetric particles. We can see immediately that increasing the deviation from cylindrical symmetry decreases a and so increases K the nematic-isotropic transition temperature for constant u200* At first

sight this trend is somewhat surprising although it is in accord with calculations based on a quite different model (88). To appreciate the

Table 3.1

The order parameters and entropy at the nematic-isotropic transition.

o r i g i n of t h i s e f f e c t i t i s important t o r e a l i s e t h a t varyingA while holding u~~~ f i x e d corresponds t o keeping t h e component a '

Zz

of t h e

t r a c e l e s s p o l a r i z a b i l i t y t e n s ~onstant, increasing a c ' by a given xx amount and decreasing a ' by t h e same amount. YY Under t h e s e conditions

t h e o r i e n t a t i o n a l energy of a p a i r of p a r t i c l e s , w i t h p a r a l l e l p r i n c i p a l co-ordinate systems,becomes more n e g a t i v e a s t h e d e v i a t i o n from c y l i n d r i c a l symmetry i n c r e a s e s .

It i s t h i s s t a b i l i z a t i o n of t h e p a r a l l e l

c o n f i g u r a t i o n which i s r e s p o n s i b l e f o r t h e increased range of t h e nematic mesophase. The values of s e v e r a l o r d e r parameters, a t t h e t r a n s i t i o n temperature, a r e a l s o given i n t h e t a b l e . The parameter

F2( ~ i s )

seen

t o decrease w i t h i n c r e a s i n g A , a behaviour which might appear t o be incompatible w i t h t h e g r e a t e r t e n d e n c y f o r p a r a l l e l molecular alignment. However, i n t h e l i m i t t h a t t h e components a '
ZZ

and a i X of t h e t r a c e l e s s

p o l a r i z a b i l i t y t e n s o r become i d e n t i c a l , a s f o r a d i s c - l i k e p a r t i c l e ,

i t i s t h e y-axis which w i l l tend t o be p a r a l l e l t o t h e d i r e c t o r .

Consequently t h e z-axis w i l l tend t o be orthogonal t o t h e d i r e c t o r and

w i l l be small o r even n e g a t i v e ; the o r d e r parameter

P 2

must there-

f o r e decrease w i t h i n c r e a s i n g d e v i a t i o n from c y l i n d r i c a l symmetry s o a s t o approach t h i s l i m i t . meter S i m i l a r arguments hold f o r t h e h i g h e r para-

F4( ~ a s )

A i s increased.

I n c o n t r a s t t h e o r d e r parameter D 2 (K)
0 ,2

which r e f l e c t s t h e deviazion from c y l i n d r i c a l symmetry, shows t h e expected i n c r e a s e w i t h i n c r e a s i n g A . t h e order-disorder with increasing A , F i n a l l y , t h e entropy change a t

t r a n s i t i o n , a l s o l i s t e d i n t h e t a b l e , decreases i n accord w i t h t h e decrease i n -(K) P 2

'

The temperature dependence of t h e o r d e r parameters

P2

and

for the long molecular axis is shown in figure 3.1 for several values
of A .
The results are p l o t t e d as a function of the reduced variable

which is identical to the reduced temperature TIT provided K the coefficients u are themselves independent of temperature. The
2qp l i n e s labelled h=O correspond to the Maiex-Saupe l i m i t and the effect
sylranetry

ofintroducing deviations from cylindrical

curves as well as changing their slopes.

is to lower these

The order parameter

5 is

found to be positive for a l l reassnabfe values of h and so deviations

from cylindrical symmetry cannot be invoked as an explanation for the:

negative values of

4 determined for 4-n-butyloxgbenzyl idene-4 '-cyan0

thoxybenzylidene-4 * -n-buty laniline (51 )

aniline dissolved in 4 -

The secondary order parameter D

2
Q,2

. is p l o t t e d as a function of 3 in

figure 3.2 for various X values.

These r e s u l t s exhibit an unusual

behaviour for the order parameter D is observed to increase with 0 92

increasing
t o zero

P2

pass through a m a x i m and then decrease, going finally

when the z-axis is completely ordered.

The temperature

dependence of the secondary order parameter D~ is shown in figure 3 . 3 0,2 for some values of A.

We now turn to a d e t a i l e d comparison of these predictions with

experiment.

The most d i r e c t evidence for the influence of deviations

from molecular cylindrical symmetry on the behaviour of the mesophase is

because this the observation of the secondary order parameter D Q,2

is identically zero f o r c y l i n d r i c a l l y symmetric particles,

Of course

the other arder parameters will also depart from their Maier-Saupe
values but similar deviations may also be observed for cylindrically
spmnetric particles if quadratic and quartic terms are retained in the

pseudo-potential given by equation (3.11) ( 3 3 , 9 3 ) .

Unfortunately the

Figure 3.1

The dependence of t h e o r d e r p a r a m e t e r s

P2

and

on t h e

reduced v a r i a b l e a / a f o r X e q u a l t o ( a ) 0 , (b) 0 . 1 , K ( c ) 0 . 2 , and (d) 0 . 3 .

Figure 3.2

The variation of the secondary order parameter D the order parameter

(b) 0.2 and (c) 0.3.

P2

with 0,2 calculated for X equal to (a) 0.1,

Figure 3 . 3

The temperature dependence of the secondary order parameter

' D

0,2

=(Sxx

YY

)/d6.

The curves are calculated f o r

equal to Ca10.1, (b10.2

(c)0.3

2 for a pure nematogen was when Alben et al. only attempt to determine P
0-2

(82) showed how the nuclear magnetic resonance experiments performed by

Rowell e t al. ( 9 4 ) on p a r t i a l l y deuterated samples of 4,4'-dimethoxyazoqbenzene might be interpreted t o yield D as well as Although 0-2 these experiments were performed throughout the nematic range the ana-

lysis was restricted to a single temperature (82).

We have therefore

applied an analogous procedure t o all of the data obtained by Rowel1

et al. ( 9 4 ) and found the values of

P2

and ' D p l o t t e d i n figure 3.2. 0,2

The results are reasonably well accounted for by the curve obtained
w i t h A equal t o 0 . 2 , especially as the values of
Q9

are particularly

sensitive t o the fine details of the geometry assumed for the nematagen.
Given a h of 0.2 we p r e d i c t the entropy of transition S ' K ) / ~ t o be 0 . 2 7

in surprisingly good agreement w i t h the observed value o f 0.28, although

be this does appear to subject to some experimental uncertainty (81).

This prediction is a considerable improvement on the Maier-Saupe value

for s ( ~ ) / R 0 . 4 2 . of
The order parameter

Y2

a t t h e transition p o i n t

may also be calculated for this particular A and we f i n d 0 . 3 4 in good the agreement with observed value of about 0 . 3 6 . This prediction should
be contrasted with the Maier-Saupe value of 0.43.
ter
The higher parame-

4 is p r e d i c t e d

t o be 0 . 0 8 a t the order-disorder transition but

as y e t there are no determinations of t h i s quantity although it is

available from light scattering experiments (503.

As a final t e a t of the theory we compare the predicted temperature

dependence of the order parameter

F2 with

the observed dependence.

This comparison is not straightfornard because the measurements are

always nude at constant pressure and as a consequence t h e volume

will also change as the temperature is varied.

The pair distribution

-6 function employed in evaluating the average R will then be temperature 12

dependent and so the coefficients u occurring in the theory will 2qp change with temperature. It is possible to allow for this variation by assuming a simple volume dependence of the form

where the u0 is temperature independent ( 3 . However, even though 3) 2qp the temperature dependence of the volume is known, such an assumption introduces an additional variable y into the calculation. This problem can be avoided if the order parameter is measured atconstant volume for then the coefficients u should, to a good approximation, Ll cp be independent of temperature. Such measurements are available for
4 , 4 ' -dimethoxyazoxybenzene (61) and the order parameter

F2

is plotted

as a function of the reduced temperature in figure 3.4.

The solid

line also shown in the figure was obtained from the theory withA set equal to 0.2. The agreement with experiment is good although the slope

of the theoretical curve is slightly greater than that observed. The theory based on deviations from cylindrical symmetry and treated within the molecular field approximation would appear to provide an acceptable account of the orientational properties of the nematogen 4,4'-dimethoxyazoxybenzene. However it is worth to recall that

the same properties have been interpreted, with comparable success, by the pseudo-potential, for cylindrically symmetric particles, given by equation(3.ll)with

L restricted to 2 and 4 ( 3 . 3)

The order parame-

is ter F2( ~ ) predicted to be 0.3, the entropy change S(K)/~ at the transition is calculated to be 0.32 and the temperature dependence of

Figure 3 . 4

The temperature dependence of

observed for 4,4 "dime thoxy-

azoxybenzene a t constant volume; the curve is calculated w i t h

1 equal to 0 . 2 .

the order parameter

experiment ( 6 9 )

0,2

at constant volume i n in complete accord with

Of course the theory f ai Is t o predict the order

parameter II2

since for cylindrically symnetrie particles t h i s must

be zero.

On present evidence it

is hard to distinguish between the

CK) two theories although it may be significant that the parameter P2

is usually less than the Maier-Saupe limit (69) which is a natural

result

of deviations from molecular cylindrical s y m ~ t r y . Clearly

future experimental studies of the orientatlonal order in nematics

should concentrate an the secondary order parameter Do,Z.

In the

event that this is shown, unambiguously, t o be non-zero, then the

appropriate terms must be included in the pseudo-potential,

However,

such observations may notexclude the retention of the quartic terms

in the expansion.

Indeed when

F2' ~ ) exceeds

the MaierSaupe value of

(68), then quartic terms

0 . 4 3 , as i t does for 4,4'-diethoxyazoxybenzeae

will have

to

be retained because such an effect cannot be accounted

for simply in terms of the pseudo-potential for non-cylindrically

synnnetric molecules when t h i s is restricted to quadratic terms. We

shall not attempt cafculetions w i t h the extended potential here because

the available experimental evidence does not warrant such an extension.

72

CHAPTER FOUR

SPIN RELAXATION IN LIQUID CRYSTALS

4.1

Introduction In this Chapter and in the following ones some static and dyna-

mic aspects of liquid crystals are investigated using electron resonance spectroscopy. This technique can in fact often be used to investigate a diamagnetic system simply by adding a radical, or spin probe, whose interaction with its environment is reflected in the electron resonance spectrum. The nature of the system should, of course, play a dominant role in the choice of spin probe. For example

if the system to be studied is a liquid crystal, or a membrane, then two properties of the spin probe are of prime importance. The molecules must deviate significantly from spherical symmetry so that they may be highly oriented by their anisotropic environment. Secondly there must be considerable anisotropy in the magnetic interactions in order for the extent of the partial alignment to be readily discernable in the electron resonance spectrum ( 8 . 3) These anisotropic interactions may
s

also inluence the widths of the lines as well as their positions because the anisotropy coupled to the molecular reorientation invariably constitutes a powerful spin relaxation process ( 8 . 3) Consequently the

spectrum of the spin probe may often exhibit a characteristic linewidth variation and, because of the anisotropic nature of the environment, this variation will depend on the sample orientation with respect to the magnetic field. An understanding of such linewidth effects is

important not only as an aid to a correct spectral analysis but also because the linewidths contain valuable information relating to both the dynamic and static behaviour of the system (38). obtainable includes and The static information

the orientational order parameters, as

P2

P4.

Their values and their tqmperature dependence provide a test

for the statistical theories of the mesophase (c.f. the previous Chapters). Several theories have been advanced to describe molecular reorientation in the presence of an ordering potential and they range from those in which the motion proceeds via small jumps (95-97) to the strong collision model where the jump angle can take any value (38,40,52). In the next Chapters both these models will be used to interpret some experimental results and some pieces of theory necessary to connect the microscopic process to the observable spectrum will be developed. Our aim will be twofold. Firstly to obtain a qualitative description of the reorientation and try to distinguish between the models. Secondly to obtain values for the physically significant parameters of the models, such as the correlation times for reorientation of the long axis and about the long axis in' the strong collision model, or the parallel and perpendicular components of the rotational diffusion tensor in the diffusion model. We shall treat in some detail the

problems encountered in this type of analysis later on. In this Chapter we shall instead discusp some background material necessary for the successive developments.

4.2

Spin R e l a x a t i o n Theory This s e c t i o n g i v e s a b r i e f summary of t h e t h e o r y of s p i n r e l a -

x a t i o n o r i g i n a l l y due t o Bloch ( 9 9 ) , R e d f i e l d (100) e t c . . (101). s t a r t w i t h , i t i s u s e f u l t o r e c a l l t h a t t h e s t a t e of any system,

To

c l a s s i c a l o r quantum, can be d e s c r i b e d i n terms of an o p e r a t o r c a l l e d t h e d e n s i t y o p e r a t o r o r t h e d e n s i t y m a t r i x of t h a t system (102). In

f a c t any o b s e r v a b l e A ( t ) can be c a l c u l a t e d from t h e d e n s i t y o p e r a t o r p ( t ) a t a given time t a s (103)

where T r i . . . I i n d i c a t e s an o r d i n a r y t r a c e o r an i n t e g r a l over t h e phase space a s appropriate. Liouville equation The time e v o l u t i o n of p i s governed by t h e

6 (t)

p ( t ) = -i [ H , p ( t ) ]

= -iH

p (t).

Here H i s t h e h a m i l t o n i a n of t h e system; t h e s q u a r e b r a c k e t s denote a commutator i n t h e quantum case (fi = 1 i s used h e r e ) and t h e Poisson b r a c k e t s i n t h e c l a s s i c a l c a s e .

4-1
X

i times
a c t s on

The o p e r a t o r H

o p e r a t o r s producing t h e i r commutator w i t h t h e h a m i l t o n i a n .

Operators

a c t i n g on o p e r a t o r s t o g i v e o t h e r o p e r a t o r s a r e c a l l e d s u p e r o p e r a t o r s and H
X

i s o f t e n c a l l e d t h e Liouville superoperator.

The a l g e b r a of

s u p e r o p e r a t o r s i s t r e a t e d i n d e t a i l elsewhere e.g. by Muus (104); h e r e we j u s t mention two p r o p e r t i e s which w i l l be used l a t e r , f i r s t l y t h a t t h e e i g e n v a l u e s of H

X

a r e energy d i f f e r e n c e s ( a s d i r e c t l y observed i n

spectroscopy) and secondly t h a t

.
(4.3

e G (H) A exp(-H) = ~ X ~ ( A. ~ ) H

I n a magnetic resonance experiment (105) t h e observable can be r e l a t e d t o a component M of t h e complex magnetization a t a c e r t a i n a frequency ( o r time). This i n t u r n can b e r e l a t e d t o t h e e x p e c t a t i o n

value of a p a r t i c u l a r s p i n o p e r a t o r S where a=x,y,z a' examp 1 , e

A s a simple

where g i s t h e g f a c t o r and

t h e e l e c t r o n Bohr magneton.

The l a s t

equation i l l u s t r a t e s t h e obvious f a c t t h a t t o c a l c u l a t e s p i n p r o p e r t i e s we do n o t need t h e complete d e n s i t y m a t r i x b u t j u s t a s p i n d e n s i t y m a t r i x a ; t h i s i s obtained from t h e complete d e n s i t y m a t r i x by summing over a l l degrees of freedom except t h e s p i n . t r a c e over t h e s p i n v a r i a b l e s .
Tr
S

i n d i c a t e s t h e remaining

I n t h e following we s h a l l t h e r e f o r e

only concern o u r s e l v e s w i t h t h e problem of f i n d i n g an approximate expression f o r a ( t ) .

A n a t u r a l s t a r t i n g p o i n t t o do t h i s i s t o d i v i d e

t h e t o t a l system i n two p a r t s : a s p i n system and a bath.

Neglecting

i n t e r m l e c u l a r s p i n couplings, t h e s p i n system w i l l be a s i n g l e r a d i c a l . The b a t h ( o r l a t t i c e ) i s weakly coupled t o t h e s p i n system; i t a f f e c t s t h e s p i n s and a s we s h a l l s e e l a t e r , causes t h e i r r e l a x a t i o n , b u t i t i s n e g l i g i b l y a f f e c t e d by t h e s p i n s . Corresponding t o t h i s s e p a r a t i o n t h e

total hamiltonian is

where H

is the hamiltonian for the spin system in the presence of

static magnetic field, Hb is the hamiltonian of the bath, dependent on position and orientation of the particles, H is the (weak) coupling 1 term.

In accord with these definitions

In the following, it should be remembered that we treat the bath as a classical system and the spins as a quantum one, even though a uniform notation is adopted. matrix becomes The Liouville equation for the total density

It is convenient to transform equation (4.5) to a new representation by writing e.g. p=exp{i(~'

0

+ q)t)pl.

Equation (4.5) then becomes

77

This has t h e formal s o l u t i o n

W now assume, a s u s u a h t h a t a t time z e r o t h e s p i n system and t h e b a t h e a r e n o t coupled (106,107) s o t h a t t h e t o t a l d e n s i t y i s

where

Pb

i s t h e e q u i l i b r i u m d e n s i t y of t h e b a t h i . e .

i t s Boltzmann

distribution.

S u b s t i t u t i n g i n e q u a t i o n (4.6) and remembering t h a t t h e

t r a c e of a product of o p e r a t o r s i s independent on t h e i r r e p r e s e n t a t i o n we f i n d

where we have i n t r o d u c e d t h e d e f i n i t i o n s

:exp(iHot)a

and

H = exp(i$ 1 -

t ) ~ This r e p r e s e n t a t i o n i s o f t e n c a l l e d t h e i n t e r a c t i o n ~ . The upper b a r i n d i c a t e s an average over t h e b a t h Trb {( . . . ) p b } a n d i n p r a c t i c e f o r o u r c a s e t h i s w i l l W e

representation. variables i.e.

mean an ensemble average o v e r the o r i e n t a t i o n s o f t h e r a d i c a l .

a r e now l e f t w i t h t h e problemof e v a l u a t i n g t h e r i g h t hand s i d e of e q u a t i o n (4.8). To do t h i s we s h a l l f o l l o w Freed (108) and u s e Kubo's According t o cumulant t h e o r y

cumulant method (72).

The cumulant function K t ()

is given by (108)

where <,,,>

is a cumulant average and T is a time ordering operator,

which ensures that the time dependent operators on its right are arranged in chronological order, Substitution of equation (4.9) in equation (4.8) and differentiation gives the equation of motion

The terns in equation (4.10) vanish if any of the ~ + ( t ) ~are 1 uncorrelated, so that the only nomranishing contributions come from times less than a characteristic correlation time
T.

It is then reavalid

sonable to look for an approximate asymptotic solution k t ) (-

for t > > ~ . If this condition is satisfied the upper limits in the time integration can be taken as infinity. Since we can define -+ Ho=Ho+Hl (t) and H1(t)=Hl ( ) H ( ) t-lt we shall take Hl(t)=Hl (t)=O in the following, Using this fact and cumulant algebra it is not difficult to show that, to second order,

We shall retain only this term in the expansion, an approximation which is expected to be valid only if I 1 t H()

l2 r 2 < < l where r

is a

characteristic correlation time. Transforming back from the interaction ~epresentation find we

k (-1
where

exp (iHo t) X

The (super) matrix R defined in this way is just the celebrated 'Redfield matrix. It is easy to show that in a basis of eigemrectors

a of Hop where H~ > = E

la >,

the matrix elements of R are explicitly

where 6ab ia a Kronecker delta and

The quantity Labblal(oab) is then defined as the Fourier-Laplace

- (E~-E~)/?I the dynamic hamiltonian of abcorrelation function, In practice the imaginary part of the transform
transform at frequency w can be neglected (109) and the Redfield matrix is just

To derive equation (4.18) we have used the fact that H operator and introduced the usual notation

is an Hermitian

The following relations are easily proved

If we now consider again the equation for the spin density matrix the following results are obtained immediately. In the interaction repreSentation

and in the ordinary representation

' nmatrix form we have i

If the relaxation term R is zero the solution is just an oscillation

about a eWect.
eq

, if R

is non zero the oscillation becomes damped, as we might

The real part of R is the damping term which relaxes the ini-

tially non-equilibrium situation. It is obvious from its physical

significance and also verifiable explicitly that R does not act on the equilibrium density matrix i . e . Ra =O. eq We could then write also

as

is sometimes found in the literature (110).

Equation ( . 3 describes the evolution of the spin density matrix 42) under the action of the bath. In an electron resonance experiment

We are usually interested in measuring the complex magnetization in

a plane perpendicular to the static magnetic field B0 '

A weak radio-

frequency field perpendicular to Bo is applied and the rate of change in the magnetization of the sample is monitored. To keep into account the interaction with the radiofrequency field B1 we assume that, being weak, it produces only a linear response in the density matrix. We then have

%
at

= - j . ~ a + ( - )

aa
+

at

relax.

(E )ref,

au

(7illX 0

+R

rillX r,f,( ) lo, t

is where R is the relaxation matrix previously determined and H r.f. the perturbing Hamiltonian:

82

By Fourier transformation of equation 4.26 we find

~(0) =(

io-R + iHo}

x -1

(-iw1sx o 2 ) -io1s~o(0) + (w

where

and o0 ()

is the equilibrium spin density matrix.

The non-resonant

term w1sXo(2w) on the right can usually be neglected and, if we assume + B1<< BO and gBBO << kT, the other term becomes (111)

where C is a real constaht. For an unsaturated absorption experiment the line shape is (111-113)

where S+ is the (2S+1) dimensional vector formed with the elements <a

IS+

1 b>.

Substitution of the expression already found for o (o) gives

where we have defined

-iH0 -iw+R

In electr~ measured. In

qce the first derivative of S ( w ) is usually

-t formalism this is just

Let us now examine the spectrum the R matrix is diagonal. With this rest^ and we can write

limit'ing situation that

' is diagonal as well

It follows that

where ga=(s+a)(s-a+l)

and ~ ( w ,a ,ra) is a lorentzian line shape i.e. u

In our case
L ( ,ma,ra) = - ~ e l/ [ (wa+1ua) -io + ~ e r 1 ] w i i ~

(4.39

Thus, when R is diagonal, the spectrum is a sum of lorentzians;

' Re Ra-l ,a ,a-1 ,a

is a linewidth and

Imra is a (dynamic)

frequency shift. A common more complex situation arises when a group of transitions is degenerate i.e. they have the same transition frequency
+t

but these are well separated from the others or, in other

words, when the M matrix is block diagonal. The block corresponding to the degenerate set has to be inverted explicitly. Sometimes the

inversion of the block can be done analytically (c.f. Chapter 7) but in general it is necessary to invert numerically the matrix for every frequency at which the spectrum has to be calculated. However, the situation, even in the general case of degenerate and non-degenerate transitions is not as complicated as it may seem. This is firstly because we are only interested in calculating certain elements of M

-1

-1
i.e. the Ma-l,a,b-l,b

An inspection of the M matrix shows that it

is possible to permute its rows and columns and rearrange all of these elements in a block matrix of dimension 2Sx2S, which is quite a dramatic
2x reduction in size from the original space of dimensions (2~+1) (2~+1)~.

For a triplet, as an example, the reduction is from a 9 x 9 matrix to a 2x2. The contraction of the total Liouville space to the 2S dimensional

"transition" space can be achieved because of certain restrictions on the subscripts of the relaxation matrix coming, physically, from energy conservation(100
)

S~mmarizing~the Liouville space formalism offers a compact and general way of calculating the spectrum. In the fast motion limit the relaxation operator has been shown to coincide with the Redfield matrix. A contraction of the total Liouville space is possible reducing its

dimension to 2s i.e. to the number of allowed transitions. It is worth pointing out that the formalism does not depend on having

degenerate or nondegenerate transitions as does the usual Redfield treatment. An expression is given for the total spectrum i.e.

s (o)a~e(s+M-' S-)

and it just reduces to the proper limiting cases

for degenerate and non-degenerate transitions. The method is particularly useful in problems like the one of the angular dependence of the electron resonance spectrum treated in this thesis since by changing the angle the transitions can pass from non-degenerate to degenerate and viceversa.

4.3 The Static S ~ i n Hamiltonian

The calculation of the spectrum outlined in the previous Section requires the knowledge of the transition frequencies and of the relaxation matrix of the system under consideration. The calculation of the transition frequencies is treated here. Consider the total time dependent spin hamiltonian H(t),that is the unperturbed spin hamiltonian H0 together with the spin-bath coupling hamil tonian H1' Since in the derivation of the 'ledf ield

matrix we have supposed that the dynamic hamiltonian has zero average, in the following we shall define H 5 Ht () (static hamiltonian) and H (t) 5 H t - ( ) ()Ht (dynamic hamiltonian). We shall use the notation 1

instead of H

0'

The transition frequencies are then obtained from

and the elements of the relaxation matrix from the correlation The total spin hamiltonian Ht, () since it contains

function of H ( ) lt.

the coupling between the spin system and the thermal bath will depend both on spin and spatial variables. We shall find it convenient to write the hamiltonian H t () as a contraction of irreducible tensors,

which at a certain instant of time is just,

In this expression

denotes the nature of the interaction, L its

rank and F its strength, while T is the corresponding spin tensor. Since we shall assume that the molecular reorientation is the only

relaxation mechanism for the spins, equation (4.40) gives also the hamiltonian at a given orientation. The choice of using irreducible tensors appears therefore as a very natural one, given their simple

transformation properties under rotation. The time dependence can be conveniently separated out by writing the tensorsT in the molecular coordinate system in terms of the same tensorsin a coordinate system fixed in the laboratory. It is convenient to choose the z axis of this laboratory system,which we call the d frame,along the director. The spin hamiltonian is then

Ht ()

I-P ()

p-(Ls~) l~

DL (t) 49-P

TF~~)-

Provided the molecular reorientation is fast on our experimental time scale, the static spin hamiltonian can be obtained from equation (4.41) by taking an ensemble average over the orientations.

The symmetry of the mesophase can now be employed to restrict the number of non-vanishing order parameters. then If the mesophase is uniaxial,

and

If the director is not parallel to the stationary magnetic field we require one more transformation to express the spin operators in terms of their components in a frame with the z axis along the magnetic field

(the B frame).

The result is

and

For all the practical cases only the angley between the z axes of the B frame and the d frame needs to be specified, so only the rotation Dr
( 1 )needs to be considered. 00 sq The static spin hamiltonian in equation (4.45) determines of
L

course the line positions in the spectrum.

It can be rewritten

formally as the static hamiltonian for a cylindrically symmetric species

where

is called a partially averaged tensor ( 8 . 3)

This useful result shows

that,irrespective of the symmetry of the probe,the line positions are formally that of a pseudomolecule with the symmetry of the mesophase (c.f. Chapter 1. ) The spectrum of a probe dissolved in a liquid crystal

can therefore be quite different from that in the isotropic phase. In fact, taking the isotropic limit of equation (4.46) i.e. letting

the arder parameter go to zero, ue see t h a t

In other words, in an isotropic solvent the anisatropy in the magnetic

interactions is completely averaged to zero, in the fast motion
limit that we have assumed so far.

As a trivial consequence the

By contrastpin an

spectrum is angular independent (isatropic).

ordered mesophase non-scalar interactions'can survive even in the fast

motion limit and the spectrum of an oriented probe shows a dramatic

change from the isotropic one.

The number of lines, their position,

their shape can change; furthermore the spectrum can exibit variations when the angle between field and director is changed.
as the

As far

line positions are concerned, it is not necessary t o go into

details of these changes: having reduced the static hamiltonian t o

that of a cylindrically symmetric pseudomolecule, these d e t a i l s can

be found in many textbooks (114,115).

We shall just report the

results for t h e important case of a spin probe containing one unpaired

electron which interacts with a single magnetic nucleus of spin I.

The observed resonant fields are, t o f i r s t order in la/ool.

where % is the operating frequency of the spect romefer , and

neglecting any nuclear quadrupole interaction.
The electron

resonance spectrum contains (21+1) hyperfine lines. dependent coupling constant and g factor are

The angular

and

The partially averaged components of the hyperfine tensor A are defined as

where a is the scalar coupling constant.

4.4 The Dynamic Spin Hamiltonian The general form for the dynamic hamiltonian H ( ) lt, in the

laboratory frame,is obtained by subtracting equation (4.44) and equation (4.45), as

We have defined

The spectral densities to be calculated are

Jabatb, (u)

(t) lb><al H ~ J (t+tt) lb'xos~t'dt'.

(4.58

Substituting equation (4.56) we find

where

J' ' G) = "rr

Q)

lJ

(L1 (t) F,,, sr')

(t+tl)* cosut'dt 1.

The problem of evaluating the Redfield matrix elements separates therefore in a spin part consisting in the calculation of certain matrix elements and in a properly dynamic part i.e. the calculation

of correlation function transforms J. " : rr

The spin part can be dealt

with quite simply by using angular momentum techniques. 1n practice the calculation of the spin part is the same as for isotropic systems provided the eigenstates of b are still the same. Therefore quite often the results for the spin part of the problems can be found in the standard works on spin relaxation in isotropic systems (109,116). We just mention here some well known results for a probe with a single electronic spin interacting with a nucleus of spin I. This includes the case of the popular nitroxide spin probes whose spectra we shall study in the next Chapters. As mentioned before,the calculation of

the matrix elements of the Redfield matrix ought normally to be performed in a basis of eigenvectors of the static spin hamiltonian 51. It is convenient however to invoke the so called high-field approximation which corresponds to retaining only that part of the static hamiltanian which commutes with S and to work in the simple product

basis

IS,^S> l1,?>

(see however Chapter 6 . )

When the correlation

times for the reorientation of the probe is greater than the inverse of the resonant frequency the terms which do not commute with S

(i.e. the non-secular terms) can be neglected also in the dynamic spin hamiltonian. In this case the linewidth of the transition labelled by mI = m is just the inverse of a Redfield matrix element and can be written simply as (116)

:T

(m) = A + B m + Cm

where

To obtain these expressions only contributions from the hyperfine A tensor and from the g tensor have been considered. A simplified notation for the spectral densities ,e.g.,
0

instead of p 0 adopted. () is
0

4.4 ( ) The Correlation Function b In the previous sections we have shown that in the so called fast motion limit, i.e. when the bath relax much more quickly than the spin system, the linewidths and in general the shape of the spectrum depends on the spectral densities Jpr:

From equations (4.60) and

(4.57) it is easy to see that the only dynamical quantity to be considered is the correlation function of the Wigner rotation matrices, that we write in genera1,for a stationary random process,as

In practise we shall only need here correlation functions with L=L1=2 so the superscripts L,L' will be left implied. restrict the number of subscripts in
I#I

It is possible to

by employing the symmetry Q P 'P' ~ of the mesophase and that of the probe (c.f. Chapter 5 . As an example, )

for a cylindrically symmetric probe and mesophase, the only case treated in this thesis,

The calculation of the correlation function requires some knowledge of the microscopic dynamic reorientation process. a stochastic, Markov process (117), so that We assume it is

where P Q ) is the equi'ibrium orientational distribution function (

0

and ~(Q~lfit) the probability that, if the molecular orientation is is 0

0

at time zero, it will be Q at time t; it is called the conditio-

nal probability. The equilibrium probability is given by

() where U Q

is the pseudo potential for the probe molecule.

According to a molecular field model of binary mixtures of rod-like particles (118) the pseudopotential for a solvent and solute molecule is

L ieven)

L (even)

where the labels (1) and ( ) indicate the solvent and the solute, 2
x is the mole fraction of solute, the coefficients 4ij)are interaction

parameters between molecules of species i and j.

In electron resonance

the solute (probe) concentration is vanishingly small. Taking the limit for x+O of equations (4.69), (4.70) one finds (119,120)

(even)

L (even)

L (even)
Thus, when the probe concentration is very low, the two simultaneous equations ( . 9 . 4 7 ) 46)(.0 become decoupled. The solute pseudopotential
L

u(~) depends

on the solvent-solute interaction u (I2) and on the solvent

L

order parameters ~(l); the solvent pseudopotential is the same as that of the pure mesophase. The quantities directly measurable in electron They can be

resonance are the solute order parameters F(*) ( - . ) 124. L

used, if the solvent order parameters are known, to extract some information on the solute-solvent interaction (118). By choosing the probe

to be as similar as possible to a nematogen, in the sense that )214 uil1) some insight can be gained on the mesophase itself (83).

From a more formal point of view, notice that the solute pseudopotential in equation (4.72) has the same angular dependence as the solvent one. In particular by truncating the series at the first term (L=2) the Maier-Saupe like potential studied in Section 2.3 is recovered. In this and the next Chapters the superscripts labelling solute and solvent will be omitted unless confusion can arise. Let us now turn to the calculation of the conditional probability ~("11"t). It is known that the conditional probability for a Markov

process obeys an integral master equation and that under certain conditions (117) this can be reduced to a differential equation. We shall

assume this to be the case and consider in brief two rather different limiting situations.

4.4 (c) The Strong Collision Model

The strong collision model (121,122) represents a process where the molecule of interest undergoes sudden changes in its orientation at time intervals
T.

The time taken for the transition from one

orientation to the other is supposed to be negligible; the orientational probabilities before and after the sudden change (collision) are assumed to be given by the equilibrium Boltzmann distribution. Under these is conditions the rate of change of the conditional probability ~(I"~lS2t) simp1y

The general integral of this first order linear differential equation, ( subject to the initial condition P D 0I Ro) = 6 (QO -Q) is
i I

This conditional probability obviously reduces to the equilibrium probability for very long times, as it has to be (117). general, the quantity Thus, in

will reduce to

ID^ 1 4,P

at very long times and the correlation function

(t) defined by equation(4.65) will then decay to zero. It is %,P easy to calculate the correlation function 4 at any instant of time 9,P t employing equation (4.74). The calculation shows that for a cylindri-

cally symmetric probe and mesophase equation (4.66) is obeyed, and that

$4,~

(t)

=I

D;

D ~ * ,P q,p

-2 P2 6 Oq 6 OP 1 exp(-t/~).

(4.75

Although this function has been employed to study the electron resonance spectrum of vanadyl chelates dissolved in a nematic mesophase ( 4 0 ) , the usi of a single correlation time ignores the anisotropic nature of the solute. This deficiencj has been removed by assuming (52) that the corre-

lation time depends on p but not 'q and so

In linewidth theory (c.f. the previous section) the second subscript turns out to indicate the component of a certain (2nd rank) %,P tensorial interaction so that this generalized strong collision model allows each component to decay with its own correlation time r Another form of equation (4.76) can be obtained by coupling the two Wigner rotation matrices as (34)

This gives

(even)

For linewidth studies it is often useful to have the zero frequency Fourier transforms of 41 () t, 4,P They are just i.e. the spectral densities j 9P
(0).

L (even)
After evaluation of the Clebsch-Gordan coefficients we find the explicit

expressions listed in Table 4.1.

TABLE 4.1

for the strong collision model The spectral densities j QP

(I

Spectral density j

QP

(0)

Notice that the following symmetry relations are obeyed:

4.4 (d) The Diffusion Model

The equation of motion for the conditional probability for a particle undergoing a rotational Brownian motion subjected to a poten() tial U Q is, in coordinate free form (123),

This is subjected to the initial boundary condition

'

In equation (4.81)

is called the stochastic operator, J is an

angular momentum operator in a particle fixed frame and in dimensionless form and finally D is the rotational diffusion tensor of the particle. The single particle potential U is a function of the molecular orientation. When U is a constant,equation (4.81) reduces to the well known equation of rotational diffusion in a isotropic medium; the eigenvalues and eigenfunctions of the stochastic operator are in that case (124)

and

if the diffusion tensor is cylindrically symmetric, as we shall assume from now on. The presence of the term containing the potential in

equation (4.81) causes in general I' to be non hermitian. However, it can be shown that, if the detailed balance principle holds, be made hermitian by the transformation

can

where P is the equilibrium orientational distribution function and

is the hermitian stochastic operator.

The existence of this diagonalized and that its

transformation guarantees that eigenvalues are real.

r can be

The eigenvalues of

r r
6

constitute a non-negative

sequence; the reciprocal of the eigenvalues are the relaxation times of the process. The first eigenvalue of is always zero, corresponding

to the equilibrium probability which, being unaffected by the stochastic operator, has an infinite relaxation time. We shall employ the symme-

trized version of the diffusion equation (98) in Chapter 5; by now we shall instead examine briefly the unsymmetrized problem, originally solved by Nordio and his group (96). where it has the form They express

in Euler angles,

where v2 is the Laplacian in terms of the Euler angles ( o ~ Y )descri-

bing the orientation of the molecule with respect to the director. The operator

is then given a representation in a Wigner rotation

2 R

matrices basis (i.e., as we have seen, the eigenfunctions of J D J =V ). For a cylindrically symmetric probe reorienting in a uniaxial mesophase it is clear that the angular momentum projection quantum numbers q

and p are still good labels for the eigenvalues of the total representation of by q and p. and one has

r.

In other words

factorizes in diagonal blocks labelled

In Nordio's notation the (q,p) block is called -Dl RqP

where the vector

cqP is defined

through the expansion

The explicit expression for the RqP matrix is (96)

The parameters XL are the expansion coefficients for the orientational potential of a molecule,

U(B)

= kT

1
L

X L PL (cos~).

(even) The axis of symmetry of the diffusion and of the ordering tensors are assumed to coincide. It has been shownby Nordio et a1 (96) that

the solution to equation (4.87) subject to the initial condition

equation ( 4 . 8 2 ) is

(4.91

where X is the transformation that diagonalizes R:

A lower case notation is introduced for the diagonal eigenvalues

matrix. The zero frequency spectral densities calculated with this

conditional probability can be written as ( 9 6 )

where

J ~ P

are time constants defined by

The spectral densities obtained from this model obey certain symmetry rules because of the properties of the Clebsch-Gordan coefficients, in fact (96)

4.4 (e)

Computational Problems

can be performed qP by inverting the matrix IZqP in equation (4.89) t o determine the time constants *r and then substituting in equation (4-93). The dimension J ~ P of R corresponds t o the numbers of terms retained.in the series

The calculation of the spectral densities j

expansion is 30.

(4.88) and is chosen to ensure convergence.

A typical value

The inversion of R bas t o be repeated for every (q,p), for

2 one is interested

any D l , /DL and, of course for any order parameter

in.

Computation t i m e requirements can then be quite heavy

for a

complete set of calculations.

on a GDC CYBER 74,16 computer.

Hemming& (128,129) quotes about 4 hours

Here we shall investigate how t o

reduce dramatically this time while getting a better insight into the theory.

As a starting point we decompose R in three parts.

where the matrices a , d , B are defined as

We use a lower case notation for diagonal matrices.

Ue assume, for

the sake of simplicity, that the pseudopotential is of the Maier-Saupe

type U(x) = kTA P2(x)

The follo~ingarguments can be generalized t o

more complex potentials. Suppose now we want to change the diffusion tensor. (i.e. if we calculate j no problem.
q0

If p=O

) there is no dependence on D l , /Dl and then

If p+O,notice that dJJ' is a multiple of the identity matrix and that it is the only part of R that contains DII/Dl. We can
write the matrix R as

(a + XB) + d

Now,if Q is the matrix that diagonalizes C,

then

since d commutes with Q. Therefore the inverse of R is

so that

in units of D-1 l

It is then not necessary to repeat the calculation

of R-I for every DI I /Dl : we just have to diagonalize the matrix C and invert its eigenvectors matrix Q once. densities j qp ' Let us now return to the spectral

Substituting equation (4.104) in (4.93) and exchanging

the order of the summations on K and J we find

where the array

zqP is

1
J=o

(J1( 2+)

1
J' (even)

C(22JV,q-q) C(2JJV,P-P)FJl

Equation (4.105) shows very clearly how the ratio D I l / D l affects the spectral densities. It allows a very efficient computation of the

spectral densities and then of the linewidth coefficients in the case when the order parameter is kept constant while D I , / D l is changed, whichis what is often required in practice. parameter P 2 For example, the order

can be determined from the line positions assuming a

Maier-Saupe potential the interaction parameter A can be determined e.g. from figure 2.1 (notice there 5=- 3 A is used) and the array ZQP 2 is calculated once and for all.
D l , /Dl

The diffusion tensor components ratio

is then determined by changing it until the linewidth coefficients

agree with experiment. A slightly more complicated problem arises when we want to change the order parameter P 2' This is because neither

matrix B nor matrix a are a multiple of the identity so that the

commutativity assumed in equation (4.102) is not guaranteed. We can overcome the problem operating some transformations on the matrix Ri.e.

Now, if S is the transformation that diagonalizes G, i.e. GS=Sg, then after some manipulations

-1 and in units of Dl ,
f

= -

J ~ P

2J+1 - -4

-4

d22 d~~

1 (1 +

' 2 k

-1 '~ k

Therefore, to calculate the time constants r

at various values of J ~ P (and then of P2) we have just to set up the matrix G, diagonalize

it and invert its eigenvector matrix once; then perform the summation in equation (4.109) for every X of interest. Notice that the formalism described here is even more convenient when the symmetrized version (98) of relaxation theory is used for then the inverse of the eigenvector matrix is just its hermitian conjugate.

CHAPTER FIVE

ANGULAR DEPENDENT LINEWIDTHS

5.1

Introduction The magnetic field present in the eiectron resonance spectrometer

is usually employed to align the director in the mesophase.

It is,
'

however, possible to vary the angle between the field and the director by proper application of an additional constraint to counterbalance the orienting effect of the magnetic field. A number of techniques have been developed for this purpose. They include spinning the sample
As the

(130-132) and the application of an electric field (133-135).

angle between director and magnetic field changes, both the positions and the widths of the spectral lines can vary (52-55). From an

analysis of the angular dependent linewidths in terms of the strong

I

collision and diffusion model some static and dynamic informations can
I

be obtained.

The development of such an analysis should be of value

I

also for the interpretation of spectra of spin probes dissolved in

!

biological membranes (128, 129, 136-141) in view of the similarity between these anisotropic system and a smectic mesophase. In this Chapter we report a study of the angular dependence of the electron resonance spectrum of a spin probe dissolved in a smectic A mesophase, The spectra are analysed to obtain the angular variation of the linewidth coefficients and hence the angular linewidth coefficients. These angular coefficients may be interpreted by

employing a theory of spin relaxation based on the strong collision

model for molecular reorientation.

Such an analysis yields values

for the correlation times for reorientation about the long and short

axes of the s p i n probe a s well as the magnitude of the solute order

parameter P

4'

The possibility of interpreting the angular linewidth

coefficients in terms of a theory based on the diffusion model for

molecular reorientation is also explored.

Relatively straighforward

expressions for these coefficients can only be obtained from the diffusion model when the orientational order is high, a s i t is in a
smectic phase.
'She analysis o f the angular linewidth coefficients then

provides values for the components of the diffusion tensor parallel

and perpendicular to the long axis of the spin probe.

The large

anisotropy in the diffusion tensor may be indicative of the unhindered

molecular motjon about the long axis.

Both models of molecular reorientation in a highly ordered

system appear to provide a satisfactory account of the observed angular

linewidth coefficients.

5.2

Experiments l
The mesogen employed in t h i s investigation was 3-N-(4'-

ethoxybenzi1id~nemino)-6-n-butylpyridin

This is reported to melt

at about 2 9 ' ~ to form a smectic phase which passes t o a nematic meso-

phase a t 4 9 ' ~ ; the isotropic f l u i d is subsequently formed at approximately 5 4 ' ~ (143). The spin probe, (3-spire- [2 'N-oxyl-3' ,3'-dimethyloxazolidine

] ) 5a-choles tane was synthesised f ram

5a -cholestan-3-one

using the standaxd route (144). The molecular structure of this spin
probe is s h m in figure 5.1 together with the p r i n c i p a l coordinate

system for the magnetic interaction tensors.

The transition tempera-

tures far the mesogen were depressed several degrees by t h e addition

of the s p i n probe but this will not affect our analysis of the linewidth coefficients. A monodomain sample of the smectic mesophase was
prepared in the following way.

The sample was f i r s t heated, within

the cavity of a Varian E-3 spectrometer, until the nematic phase was
formed and at this point the director, or preferred molecular orientation, is aligned parallel to t h e magnetic field of the spectrometer.

The field was then set to its maximum value of 6.5 kG and the temperature lowered slowly until the smectic

A phase

was

formed.

The surface

forces, which can inhibit the production of a monodomain sample, were

reduced by coating the tube with p l a s t i c (142). The electron resonance

spectrum of the sample a t 2 0 ' ~ was then measured as a function of i t s

orientation relative t o that of t h e magnetic f i e l d ,

The spectra obtained

for a11 orientations contained the three hyperfine lines expected f o r a

monodomain sample doped with a nitroxide spin probe.

In a d d i t i o n the
I

angular dependence of the nitrogen hyperfine spacing shows t h a t the

director remains parallel t o the magnetic field on passing from the nematic t o the smectic phase and so shows t h i s t o be smectic A 4142).

Figure 5.1

The structure of the spin probe and the principal coordinate system for the magnetic tensors.

The line shape is essentially gaussian because of inhomogeneous

broadening caused by unresoLved proton hyperfine structure. linewidth was equated with the peak-to-peak separation.
The

The widths

of the three hyperfine lines were determined from that of the central

line and the heights of the outer two lines relative t o the central
peak.

These linewidths were then used t o determine t h e linewidth

coefficients in the expression

where m is the nuclear quantum number.

The observed angular

dependence of these linewidth coefficients is shown in figure 5 . 2

and it is t h i s dependence which we shall now examine in greater d e t a i l .

5.3

Angular Dependent Linewidths

The dominant relaxation process for a spin probe, similar t o

t h a t employed i n this study, is its molecular reorientation, with

respect to the director, coupled t o the anisotropy in the magnetic

interactions.

The observation of motional narrowed spectra from this

system is important for it shows that the linewidths are directly

related to the dynamic spin hamiltonian.

In fact they are determined

by the Fourier transform of the correlation function of the dynamic

perturbation.

For this problem the dynamic spin hamiltonian is conve-

niently written in irreducible tensor notation as (52):

Figure 5 . 2

The angular dependence of the linewidth coefficients

A(.),

B(*) and C(m);

t h e linewidths are in Mrad.s-l.

The time dependence in H1 (t) is contained entirely in the Wigner 2 (t) relating the molecular coordinate system to rotation matrix D 49-P a space fixed system containing the director. This space fixed coordinate system is connected to a second, which contains the magnetic The orientational order of the field, by the rotation matrix D~ r99 spin probe in the mesophase is given by the ensemble average of the third rotation matrix, 2

Do -P
9

For douplet state species the only ani-

sotropic magnetic interactions of any importance are the second rank Zeemen and hyperfine terms.
U

The strength of these interactions is

denoted by F (29p) where the subscript IJ indicates a particular term. Finally the relevant spin operators are represented by T ( ' ) 2r.
IJ

The

dynamic aspects of the problem are contained therefore in the correlation function

This may be simplified to

by invoking the cylindrical symmetry of both the mesophase and the spin probe as we shall see later on.
I" function $ (J) r

The complete cross correlation

for the interactions IJ and IJ' then is

(even) P99

where C(22L;r-r)

is a Clebsch-Gordan coefficient (34).

The restriction

of the summation to even values of L results from the D symmetry wh of the smectic A phase and the properties of the Clebsch-Gordan coefficients require that the maximum value is four. The angular dependence of the correlation function is contained in the Legendre polynomial P (cosy) where y is the angle between the director and the
L

magnetic field. transform of $ t ()

Since the linewidths are proportional to the Fourier their angular dependence and hence that of the line-

width coefficients will be of the form (52)

A = AO + A

+ A P (cosy). 2 P2 (cosy) 4 4

It is important to realise that this relatively simple angular dependence only obtains if the nuclear spin is quantised in the direction of the magnetic field (137). to be the case. For the moment assume this

In general however, due to the effect of the pseudo-

secular hyperfine terms in the hamiltonian, the direction of quantization of the nuclear spin will change as we change the angle between the director and the field. The effect that this has on the linewidths

is examined in detail in the next Chapter. The observed angular linewidth coefficients, listed in Table 5.1, were obtained by a least-squares fit of the angular dependence of the linewidth coefficients to the theoretical expression in equation ( . ) 56. The theoretical angular dependence, calculated with these values, is plotted as the solid lines in Figure 5.2. The agreement between

theory and experiment is seen to be extremely good for both the A and B

TABLE 5 . 1

The angular linewid th coef f iciears (Mrad s-I )

Component

*L

n~

C~

coefficients b u t not quite as impressive for C.

The discrepancy

might be attributed to imperfections in the monodomain sample as

well as t o the nuclear spin quantization effect and we shall return

ta this point l a t e r .

In she following two Sections

we

shall see how

t h e s e coefficients may be anafysed using f i r s t the strong collision

model and then the diffusion model.

5.4

The Strong Collision Model As we saw in Section 4.4 (c) the 2nd rank Wigner rotation matrices

correlation function may be written, for the generalized strong collision model, as

The angular linewidth coefficients predicted by this relatively crude correlation function are (52)

In these equations o

is the microwave frequency of the spectrometer,

g is the g tensor, A is the hyperfine tensor and g is the partially cf(.1) averaged g factor ( . . 4 5 ) . The correlation time T 0 is associated

with fluctuations in the orientation of the long axis of the spin probe and r2 with rotations about this axis. The expressions for the

angular A coefficients differ slightly from those of Reference (52) which are in error. McFarland and McConnell (137)were the first to attempt a calculation of the angular dependence of the linewidth resulting from rotational diffusion subject to an anisotropic ordering potential. However.

it is difficult to compare their results with ours because they adopt a more general form for the correlation function. In fact they assume an exponential decay for
(t) as in equation ( 5 . 7 ) but where the ~ Y P correlation time is taken to depend on q as well as on p. We shall
$ I

not therefore attempt a comparison here. Hemminga and Berendsen (l40) have also implicitly employed the strong collision model in their calculation of the angular dependence of the linewidths for a spin probe. Their results are not directly

comparable with equations (5.8) to (5.16) since they assume complete

ordering and neglect the pseudo-secular hyperfioe terms in the dynamic

spin hamiltonian.
and
4

We can achieve some comparisons by setting both P 2

In this limit our results for

equal t o one in our equations.

the angular B coefficients are in complete accord with those o b t a i n 4

by Hennninga and Berendsen (140); there are however differences in the

A and C coefficients. These result from their neglect of the pseudosecular terms and so our results can be employed to gauge the magnitude of the errors r e s u l t i n g from this neglect.

Thus t h e i r expressions

for the angular A coefficients do not contain t h e contributions involving

the hyperfine tensor and their C coefficients are, in our notation,

We see then that ignoring t h e pseudo-secular terms creates considerable

error in C but is more successful for C and C 0 2 4'

We

m u s t now consider the analysis

of the observed coefficients

Of t h e

using the equations obtained from the strong collision model.

various parameters in equations ( 5 . 8 ) - (5.16) only three, T ~ , T * and P are unknown. Thus t h e g and hyperfine tensors have been determi4

ned for (3-spiro-

[2'~-ox~l-3 3'-dimethyloxazolidine] } 5a-androstane\

17f3101, whose structure is v i r t u a l l y identical to that of the s p i n

probe (136) and it is reasonable to employ these tensors.
The compo-

nents of the g tensor, in the coordinate system shown in figure 5.1, are

and those for the nitrogen hyperfine tensor are

and

= 540.35

Mrad.s

-1

When calculating the irreducible components F '2yp) from t h e s e

r e s u l t s i t is important to selise t h a t they are required in the

coordinate system which contains the supposed molecular symmetry axis.

Since t h e y axis makes an angle of about 2' 0

with t h i s symmetry axis

the required components of the irreducible tensors are (120) :

The remaining parameter is the solute order P which is found t o be

2
0.89 from the angular dependence of the hyperine spacing.

These

parameters were then used t o e x t r a c t the two correlation times to and

T* as

well as the solute order

p4

from the three angular C coefficients

A second s e t of values was

w i t h the results shown i n Table 5 . 2 .

determined from the B linewidth coefficients and these results are

also included in Table 5.2.

TABLE 5 . 2

The strong coZlisioa parameters

Parameter

From

From C

There is reasonable agreement between these two s e t s although i t is

not possible, a t this stage, to attribute t h e differences either t o
a failure of the strong collision model or t o experimental error.

The angular A linewidth coefficients have not been treated in this

manner because they do contain a contribution resulting from unresolved

proton hyperfine structure.

We have however calculated the A coeffithe two sets of parameters given

cients using equations(5.8-l0)and

i n Table 5.2; these calculated coefficients are compared w i t h the

observed values in Table 5.3.

TABLE 5.3

The theoretical and experimental A coefficients ( ~ r a ds-I)

~ o eiicient s f

Experimental

Prom B

From C

The calculated values of the A

and A linewidth coefficients are in 4 In

reasonable agreement with their experimental counterparts.

contrast there is a large discrepancy between the observed and mean predicted values of A ~ r a ds'
0 '

However this difference, which is about 10

, is

comparable to the inhomogeneous line broadening caused

by the unresolved proton hyperfine structure, and so does not indicate the presence of other relaxation processes. One of the important consequences of such linewidth studies is the ability to determine the solute order P difficult to determine in any other way.
4

which is extremely

A knowledge of this additional

parameter is important because it provides another test of the statistical theories of liquid crystal mixtures and, by implication of liquid crystals themselves. The cylindrical symmetry of both the spin probe and the smectic A phase dictate that the orientational energy of the probe is given by

UB ()

kT

AL PL(cosB), L
(even)

where p i s the angle between the molecular symmetry axis and the director. According to theory (c.f. Section 4 4 b ) .() when the solute

concentration is vanishingly small the expansion coefficients are

where

FL ( ' )

is the solvent order and u (I2) is a measure of the soluteL

solvent interaction.

The solute order P

is obtained by taking the

appropriate Boltzmann average:

Consequently if both P

and ?

4 are known it should be possible to

L

determine two of the expansion coefficients X

and so test the Unfortunately

expected (118-120) convergence of equation (5.20).

this determination is not possible when the solute order is high as we shall now discover. If the orientational order is high, as it often is in a smectic phase, then the Boltzmann factor will be a rapidly decreasing function of 6. Accordingly, in this high order limit, only small values of B

make an important contribution to the integrand and we therefore obtain a simple expression for P retaining terms up to B 2
L

by expanding the Legendre polynomials

The potential energy becomes

where

(even) and

The ensemble averages P

are independent of 5

and so we shall not be

concerned with its magnitude.

Since the integrand is insensitive to

high values of $ the upper limit of the integrals may be extended to infinity and this then gives

and

In this high order limit the magnitudes of both

P2

and

P4 are

determined by the single parameter 6 and so there is apparently little to be gained from knowing P that equations (5.26),(5.27)

as well as

P 2 ' Notice as an example

are in complete accord with the asymptotic

formulae (2.59) obtained for a simple Maier-Saupe potential such as

with X;-2513. of P 2 and P 4

In Table 2.1 we compared, not unfavourably, values obtained from the asymptotic formulae with values obtained

from numerical integration. For example when 5 is 14 the approximate

P 2 is less than 1% too high while P4 is about 4% too small. The

observed solute order of 0.89 corresponds to 5 equal to 13.0 and so the limiting relations for P
2

and P should be relatively good. 4

Indeed the value of P predicted by equation (5.27) is 0.62 which 4 agrees well with the experimental values of 0.63 and 0.73. We may

therefore conclude that U($) is indeed a rapidly decreasing function

of 6 as we might have anticipated. The two correlation times, given in Table 5.2 show that the neglect of non-secular terms in deriving equations (5.8)-(5.16)
0 0

is

a valid approximation for this system for w2 r2> 1 (52), although the extension to include these terms is quite straight forward (14). However it is difficult to comment further on these times since the strong collision model does not indicate how they should depend on the molecular structure or environment. In contrast the diffusion model

does predict the dependence of the correlation functions on both the diffusion tensor and the orientational order and we shall now analyse our results in terms of these predictions.

5.5

The Diffusion Model The solution of the rotational diffusion, in presence of an

ordering potential, is usually expressed (96) in a basis of eigenfunctions of the diffusion operator for an isotropic potential. As mentioned in Section 4.4 (d), these eigenfunctions are just Wigner rotation matrices. This method of solution is clearly perfectly

general but in practise the number of tepms to be retained in the relevant series expansions (4.91), (4.93) increases as the system becomes

more ordered and the potential more anisotropic. However in this case,

9) Polnaszek et al. ( 8 have obtained a simple limiting solution.

start by

They

a symmetrized version of the equation of motion (4.81) for

the conditional probability, i.e.

where @(Q

IQt) = P' -

P(P~IR~)

f.

P' -

I. P 1 and

P E P(Q) is the
A

equilibrium angular distribution. The zeroth eigenfunction of I? : GO(Q) has a zero eigenvalue and is just

A formal solution to equation (5.29) is

. when the orientation at zero time is Q

0'

Since the eigenfunctions

of
6

form a complete set they may be used as a representation of the

function and so

The formalism used up to here can be applied to a quite general Markov process. As an example, it can include the strong collision
T.

model involving a single correlation time lues of the stochastic operator are

In this case the eigenva-

As another aside we can prove, before passing to describe the asymptotic solution, the simplification of the correlation function in equation (5.3) to the 41 in equation ( . ) 54. From the '+'qpq 'P 4P formal expression for the conditional probability we find

where the matrix notation, introduced by Polnaszek et al.(98) is defined by

Since the rotation matrices form a complete set of functions which span the space of the Euler angles we can expand

Ix;R>

as

where

Of course the eigenfunctions of

must transform according to an

irreducible representation of the system which in this case is of the liquid crystal and the spin probe. When both of

these possess cylindrical symmetry the eigenfunctions are of the form

Ix;n> =

im'a in'y e G(3 e xf>

(5.39

Consequently the expression for the eigenfunction in equation (5.37) reduces to

and the zeroth eigenfunction is simply

The correlation function is then

OL* a 0 0 mn

mn

OL "' a
00

f7.42

The matrix elements are just integrals of three rotation matrices and are given (34)

L <Do,o(Qo)

I:~Q) D,(,

L' ID~,~(Q,)>

8n2 2L+1

q+m,O 6p+n,O C(L12L;mq)C (~'2~;np) (5.43

and

Combining the 6 functions from these two matrix elements gives the desidered result. When the spin probe is not cylindrically symmetric the eigenfunctions of

take the form

I x;n>

im'a

G~(B,Y)

Consequently the correlation function can only be reduced to

for such a spin probe. Let us return now to the asymptotic solution of equation (5.29). In view of equation (5.39) we may restrict our attention to the operator

1 a - sing - (sing k~ aB

*
d~

,(5.47

which may be written as

in the limit of high order.

This result is obtained using the complete

expansion of the ordering potential; none the less it is identical in

form to the stochastic operator derived by Polnaszek et al. (98) who restricted the potential to just the first term. of this limiting form of The eigenfunctions

r6

are found (98) to be

where v is Ip-q(; the normalization factor is

and the function L' ( 6 M (80).

B2)

is a generalized Laguerre polynomial

The eigenvalues of the stochastic operator in the high-order (98)

limit are

The normalized distribution function obtained from equation (5.49) f?r the zeroth eigenfunction is

which is, of course, identical to that resulting from the limiting form of the ordering potential given in equation (5.23). The matrix

elements of the rotation matrices required to determine the correlation function

I $

q-P

(t) have been calculated by Polnaszek et a1.<98);

they find

where

This expression for the matrix element differs from that in (5.23) where, presumably as the result of a printer's error, a factor 1/25 occurs instead of 112 5 1

(t), required q9-P in the calculation of the angular linewidth coefficients, from equarions (5.35)(5.49)
>

We have evaluated the correlation functions 4

and (5.51).

The zero-frequency Fourier transforms j ( 0 (4,-P are not taken

of these functions are given in Table 5.4; the terms with p = ' l shown because the principal axes of the magnetic tensors are

to be parallel to those of the diffusion tensor for this spin probe. We can now calculate the angular linewidth coefficients in terms of the three non-zero spectral densities jlO, j12 and j 2 2 ; the results are

TABLE 5 . 4

The spectral densities j

9t P

(0) for the d i f f u s i o n model

in the limit of high order.

Although the angular linewidth coefficients are a ftinction of three

spectral densities there are only two unknowns namely the principal

components of t h e d i f f u s i o n t e n s o r .

Consequently t h e problem i s over

determined when a l l t h r e e components of an a n g u l a r l i n e w i d t h c o e f f i c i e n t a r e known. W have t h e r e f o r e sought t o o b t a i n t h e b e s t v a l u e s f o r D e

II

and D l i n t h e f o l l o w i n g way.

The t h r e e s p e c t r a l d e n s i t i e s cannot be

e v a l u l a t e d from e i t h e r t h e B o r t h e C c o e f f i c i e n t s because t h e determinant of t h e a p p r o p r i a t e c o e f f i c i e n t s v a n i s h e s . which f o r t h e C l i n e w i d t h c o e f f i c i e n t s a r e W t h e r e f o r e d e f i n e new v a r i a b l e s e

and

(2,212

j12

; The

s i m i l a r v a r i a b l e s a r e employed i n t h e a n a l y s i s of t h e B c o e f f i c i e n t s .

b e s t v a l u e s f o r a and b a r e now o b t a i n e d from t h e l i n e w i d t h c o e f f i c i e n t s using a l i n e a r l e a s t squares f i t .

As a check we u s e t h e s e l e a s t - s q u a r e s

f i t t e d parameters t o r e c a l c u l a t e t h e c o e f f i c i e n t s ; c a l c u l a t i o n s a r e given i n Table 5.5.

t h e r e s u l t s of such

The agreement between t h e o r y and

TABLE 5.5

The observed and c a l c u l a t e d v a l u e s f o r t h e a n g u l a r B and C linewidth coefficients

Component L

Observed

Calculated

Observed

Calculated

experiment i s extremely good f o r t h e B c o e f f i c i e n t s b u t c l e a r l y l e s s impressive f o r the C linewidth c o e f f i c i e n t s ; discrepancy s h o r t l y . we s h a l l r e t u r n t o t h i s

Given t h e v a l u e f o r t h e v a r i a b l e a t h e s p e c t r a l

d e n s i t y j 2 2 can be c a l c u l a t e d and from t h i s we can e v a l u a t e t h e p a r a l l e l component of t h e d i f f u s i o n t e n s o r . Armed w i t h t h i s component i t i s

1

p o s s i b l e t o e x t r a c t t h e o t h e r component D values obtained f o r D

II

from a knowledge of b .

The

and D l

from b o t h t h e B and C l i n e w i d t h c o e f f i c i e n t s

a r e g i v e n i n Table 5.6.

The agreement between t h e v a l u e s e x t r a c t e d from we s h a l l r e t u r n t o

t h e two s e t s of l i n e w i d t h c o e f f i c i e n t s i s encouraging; the large anisotropy i n the d i f f u s i o n tensor s h o r t l y .

TABLE 5.6

The d i f f u s i o n t e n s o r

Diffusion tensor
D I I (Mrad s

From B 137 2.6

From C 122 2.6

-1

D L (Mrad S-')

A s a check on t h e v a l i d i t y of t h e d i f f u s i o n model, i n t h e l i m i t of

high o r d e r , t h e components of t h e d i f f u s i o n t e n s o r , given i n Table 5.6, have been used t o c a l c u l a t e two s e t s of t h e a n g u l a r A l i n e w i d t h c o e f f i c i e n t s . The r e s u l t s of t h e s e c a l c u l a t i o n s a r e l i s t e d i n Table 5.7 t o g e t h e r w i t h t h e e x p e r i m e n t a l v a l u e s of

q.

The agreement w i t h experiment i s seen t o

TABLE 5.7

The experimental and t h e o r e t i c a l A l i n e w i d t h c o e f f i c i e n t s (Mrad s -1

)

Coefficient

Experimental

From B

From C

be good especially as the inhomogeneous broadening caused by unresolved proton hyperfine structure might be expected to make a contribution of about 15 ~rad.' s to Ao. On balance this limiting form of the diffusion

model appears to provide the most satisfactory quantitative interpretation of all three experimental linewidth coefficients. However, it may be significant that the asymptotic spectral densities calculated with this Polnaszek et a1 (98) method appear not to satisfy all the symmetry requirements (96,129). This important problem is now under investigation.

It would be unwise to attempt to differentiate between the two models on the basis of our experimental results. However one clear advantage of the diffusion model is that the dynamic parameters occurring in the theory may then be the subject of further rationalisation. We shall now attempt such an analysis.

A combination of experimental (145) and theoretical (143) investigations

has shown that the principal components of the diffusion tensor are often given by (147):

In this expression a a and a

=

is a geometric factor which takes the values (124)

4ri (ri 4: ( r

r2)/3(2rZ x (2ri

-) ~ : r

(5.6.7

r3 :{ ) /

r2)s X

2rZl

,
Z

(5.68
=

for a symmetric top molecule with semi-axes r > r

r and where Y

The parameter T-I has the dimensions of viscosity and is frequently

identified with the coefficient of shear viscosity. When
K

is the unit

tensor then equation (5.66) is the expression obtained by Perrin (124) for

rX is approximately 3.5

8.

These dimensions yield a value for ol lol

of about 3 and so the ratio rl /rI, of the components of the anisotropic interaction tensor is approximately 17. This value is large and will naturally depend on the choice of the molecular dimensions, however these and hence the ratio
K K

are unlikely to be in serious error.

K

Theoretically the maximum value of

is unity and this occurs when the

intermolecular potential is dominated by the anisotropic component (146). At the other extreme
K

is zero which corresponds to a purely isotropic

K

intermolecular potential. Accordingly the value of 17 implies that is no larger than 0.06;

in other words rotation about the long molecular

axis is essentially unhindered. This result is in accord with the cylindrical symmetry of the ordering matrix observed for this spin probe dissolved in several nematic mesophases (120). Also, a recent independent work by Hemminga (128) on the same spin probe in a highly ordered

(P2 =

0.85) lyotropic system seems to confirm

K

/ that D I I D ~is between 40 and 50, and then that

is very small.

It is possible to check the ratio D I I /Dl by measuring the relative linewidth coefficients when the spin probe is dissolved in the isotropic melt of the mesogen. We have made these measurements and find that C/B

is approximately -2.1 and independent of temperature. This ratio is related to D l , /Dl by

provided the molecular motion is sufficiently slow to make the non-secular contributions to the linewidth negligible (126). By using the same

components of the anisotropic g and nitrogen hyperfine tensor as those encountered previously the ratio D l , /DL is calculated to be 44. This

resu1t;is in excellent agreement with the value determined for the spin probe dissolved in the smectic phase especially as D,,/Dl is rather sensitive to the magnitude of C/B. The close similarity in the ratios for the two phases is unexpected in view of the absence of any long range orientational order in the isotropic phase which might be expected to reduce
K K

in this phase.

Our results for D I I /Dl presumably imply that

is determined by the local molecular environment which is essentially Accordingly

unchanged on passing from the smectic to the isotropic phase.

if this interpretation is correct then Dl, /DL should be reduced when the solvent is composed of molecules more symmetric than those of the mesogen. We have therefore measured the ratio of C/B for the spin probe dissolved in o-terphenyl and find it to be -1.7. value for D This corresponds to the smaller

,, /Dl

of 19 and such a reduction is in accord with our

expectations. It is appropriate, in concluding, to comment on those areas in which the theory has been less successful. For example the predicted angular dependence of the B linewidth coefficient is in complete accord wich experiment whereas the agreement for the C coefficient was not so impressive. Similarly the diffusion model was able to provide an almost quantitative account of the angular B coefficients but only a qualitative description of the C linewidth coefficients. These discrepancies can be ascribed to two assumptions in the analysis; the first is implicit and the second explicit. We have assumed implicitly that the director is uniformly aligned throughout the sample of the smectic mesophase. However minor

deviations from complete alignment could result when the smectic phase is formed from the aligned nematic mesophase. Indeed the slight asymmetry

in the spectral line shape hints at such deviations in uniformity (52,148). To first order the imperfect alignment produces inhomogeneous broadening

of t h e s p e c t r a l l i n e s .

For a n i t r o x i d e s p i n probe t h e c o n t r i b u t i o n of

such broadening t o t h e B c o e f f i c i e n t depends on t h e a n i s o t r o p y i n t h e p a r t i a l l y averaged g t e n s o r whereas t h e c o n t r i b u t i o n t o C i s determined

by t h e a n i s o t r o p y i n t h e p a r t i a l l y averaged h y p e r f i n e t e n s o r Experimentally whereas

(52).

i s found t o be v i r t u a l l y i s o t r o p i c f o r t h i s system
Consequently t h e inhomogeneous l i n e

is highly anisotropic.

broadening w i l l a l t e r t h e a n g u l a r C l i n e w i d t h c o e f f i c i e n t s b u t n o t t h e B coefficients. I n a d d i t i o n t h e a n g u l a r dependence of C r e s u l t i n g from

inhomogeneous l i n e broadening i s p r e d i c t e d t o be d i f f e r e n t from t h a t caused by s p i n r e l a x a t i o n ( 5 2 ) . The behaviour of t h e B l i n e w i d t h c o e f f i c i e n t i s

t h e r e f o r e expected t o be i n b e t t e r accord w i t h t h e o r y t h a n t h a t of t h e C coefficient. The e x p l i c i t assumption concerns t h e q u a n t i z a t i o n of t h e n u c l e a r s p i n i n t h e d i r e c t i o n of t h e magnetic f i e l d . This approximation i s e x a c t

when t h e magnetic f i e l d i s e i t h e r p a r a l l e l o r p e r p e n d i c u l a r t o t h e d i r e c t o r b u t i s l e a s t r e l i a b l e when t h e f i e l d i s midway between t h e s e two extremes

as we s h a l l d i s c u s s i n t h e n e x t Chapter.

I n a d d i t i o n t h e assumption i s

l e s s a c c u r a t e when t h e r e i s a l a r g e a n i s o t r o p y i n t h e p a r t i a l l y averaged hyperfine tensor. Again t h e consequences produced by t h e inaccuracy of

t h e assumption w i l l a f f e c t t h e C r a t h e r t h a n t h e B c o e f f i c i e n t s and w i l l have a d i f f e r e n t a n g u l a r dependence t o t h a t given e.g. i n e q u a t i o n (5.6).

CHAPTER SIX

MORE COMPLETE THEORY OF ANGULAR DEPENDENT LINEWIDTHS

6 1

Introduction
The theory of angular dependent linewidths employed in the last

Chapter was based on an explicit assumption, namely that when the

angle between director and magnetic field is varied, the nuclear spin

is s t i l l quantized along t h e direction of the magnetic field.

This

is equivalent t o neglecting the effect on the linewidths of the

pseudosecular hyperfine terms in the static hamiltonian.
The existence

of the problem has been pointed out in the literature (137) but the

effect has simply been disregarded as we d i d in the previous Chapter.

Part of the reason for doing so is probably the absence of a simple

theory for the effect.

In t h i s Chapter

we

s h a l l develop such a

theory, showing that the linewidths can still be expressed as

However, the angular dependent coefficients A,B,C

written

can no longer be

in the simple form

where X = A,&,C.

6.2

Theory We shall start by reconsidering briefly the static hamiltonian

for a spin probe reorienting rapidly in a liquid crystal. Fast reorientation is intended here in the sense that Redfield (100) type theories (c.f. Chapter 4) are applicable. In a molecule fixed co-

ordinate system the spin hamiltonian is, in irreducible tensor notation,

where, as in the previous Chapters, the F magnetic and spin tensors.

lJ

and T are the relevant

lJ

A transformation of T to the director

lJ

frame confines the time dependence to a Wigner rotation matrix.

A further transformation to a coordinate system with the z axis

parallel to the static magnetic field gives the spin hamiltonian at time t as:

An ensemble average gives, for a uniaxial mesophase, the static hamiltonian as

where the partially averaged tensors $(L,O)

are, as usual,

The eigenvalues of the spin hamiltonian (6.6) determine the line positions and its eigenfunctions constitute the basis to be used in the calculation of the Redfield matrix elements. For a nitroxide

spin probe the so called high-field approximation, which amounts to neglecting that part of the static spin hamiltonian which does not commute with Sz ' can usually be invoked (109, 116). We shall then

neglect terms such as Sf I+ , S+z' BS+,i.e. the non-secular terms, I when considering the irreducible spin operators. In Table 6.1 we

give the explicit expressions for these spin operators and their corresponding spatial tensors. An inspection of the spin hamiltonian matrix in the product basis

I SmS> (ImI>

shows that it is justified

to neglect the non-secular terms as long as the partially averaged hyperfine terms are much smaller that the transition frequency. We shall call the static hamiltonian, after the elimination of the non-secular terms, the truncated hamiltonian. The time-independent

L transformation D ($2) connecting director and field involves in r,,o principle two Euler angles. However, in view of the cylindrical symmetry

about the director and field, it is sufficient to consider only angle between the z axis of the two frames, i.e. DL (OyO) where r,O is the angle between the director and the static magnetic field. the From equation (6.5) it is easy to see that if y=O or y=1~/2 truncated hamiltonian is already diagonal, since
y

TABLE 6.1

The irreducible components of the Zeeman and hyperfine tensors g and A together with that of their corresponding spin operators. field B
=

It has been assumed that the magnetic

B(001).

A(2.0)

~312 A'
ZZ

(290) T~

=fi/?i[~

zIz -(S+I

-+S - 1+)/4]

A2" (')

,;A (1

iA1

xz

yz1

T~ 9 1 = ( I S + 1z S+)/2 2 -

A (92 2')

= (A'

xx

A' + 2iA )/2 YY xy

12

(a)

The prime indicates the traceless tensor components.

Therefore, when the director is either parallel or perpendicular to

1I the magnetic field 8, the set I~=1/2,m~> , mI> is a good basis

set and can be employed for the calculation of the Redfield matrix. However, when O<y<n/2, the situation is far more complicated. In fact an inspection of the truncated hamiltonian shows that it is not diagonal i.e. it does not commute with S I any more.
Z Z

Its matrix

consists of two 3 x 3 blocks corresponding to m =+1/2. S -

The off-diagonal

elements in the two blocks are the pseudosecular hyperfine terms. As far as the static hamiltonian is concerned the problem is formally identical to that occurring in single crystal studies of the hyperfine structure of a radical spectrum (114, 150). two possibilities.
-

There one has essentially

One is to express the eigenfunctions of the truncated

H(y) in terms of the eigenfunctions of

H0 ()

as

using the fact that the eigenfunctions of an angular momentum (spin)

I transform according to the irreducible representation D ( ' )

(151).

The other, equivalent, possibility is to notice that terms such as

S I
z

can be eliminated by changing to a new set of axes for the nuclear Since the partially averaged hyperfine tensor has

spin I (114).

cylindrical symmetry, Q, the direction of quantization of I lies in the plane defined by the director d and the field B (114). In figure 6.1

the various relevant directions are shown. The angle x between Q and

B is so defined that the coefficients in front of terms like SzIf.

in the static hamiltonian vanish after the transformation. The explicit

Figure 6.1

The axes systems involved in the treatment of the nuclear spin quantization effect. In the figure,

d represents the director, Q the axis of quantization of the nuclear spin, B the direction of the static magnetic field.

expression for

is given by (114,137) 2 - - 2 sin y + g A cos y)

I I II

cosx

Isl Al
.-.

- -

,
,

sin^
where

= (g

1 1

g A hiny cosy /@
II
II

- -

This results can be further simplified by assuming that the partially averaged g tensor id essentially isotropic, an approximation expected to be very good for all the practical cases we have to 'deal with (52). This assumption has also the important consequence that as we change the angle y the electron spin is constantly quantized along the field. The nuclear spin quantization axis, instead, changes as the angle y changes and it deviates most from the field direction when

cos y

(A,,Al

- -

-2 - Al

/ (A,, - A l

-2

-2

At this point

The magnitude of the effect clearly depends on the anisotropy in the partially averaged hyperfine tensor and is therefore related to the order parameter. For example, for a spin probe like the one

used in the previous Chapter (see figure 5.1), with the symmetry

axis 0f.A perpendicular to the long axis of the molecule, we have

and therefore (A,, A l ) -

-3A'

I 1

P2 1 4 ,

where
II

P2

gives the orientational

order of the long molecular axis and A' is the component of the traceless hyperfine tensor parallel to its symmetry axis. The problem so far

has been mainlyan adaptation of known results, but of course in liquid crystals as compared to single crystals the interesting thing is that we have also anisotropic dynamic effects. To investigate the effect that the proper quantization of the nuclear spin has on the electron resonance spectrum we return to the time dependent spin hamiltonian in equation (6.4) ard rewrite it more explicitly as

~ ( t ) HO+ = ()

I(--) g ( 2 y p ) ~ 2
q9-P

( c ) D ~ (Y) T~ 2 (2,r) 9 q

where )H ' (

is the scalar part of the hamiltonian.

Now we decouple

T in its spin operator components (34) A

and reject the non-secular terms.

Then

(2*r)= ~(112;or)s (1,o) I(l,r) T~

We then transform I to its quantization frame which we indicate with an upper cap,

Substitution in equation (6.14) gives the truncated spin hamiltonian as

When we take an ensemble average the rotation of

causes only the

m=O term to survive in the 2nd line of equation (6.18)

The dynamic spin hamiltonian is

where we have written

as a tensor formally analogous to the truncated hyperfine tensor (6.16). With this definition we may now write the dynamic harniltonian as

Since the

I S m S> I I me> basis I

is a good basis and since the hamilto-

nian (6.21) is formally identical to that of the ordinary nitroxide problem (52) it follows that the linewidths are still given by the usual expression (6.1), with the coefficients A,B,C given by equations (4.62)-(4.64)
!J

in terms of the spectral densities J' " r

The tensors

F ( ~ * - ~ ) equation (6.21) are given by in

D~ (y) ~(112,Or)~l (x). r9q m,r

This leads to the following spectral densities

It is easy to check that when there is no nuclear spin quantization effect, i.e. X=O these spectral densities reduce to the ordinary ones (c.f. Chapter 4 . ) In particular when the director is parallel or The spectral

perpendicular to the field the spectrum is unchanged.

densities (6.24-6.27) show also that another important feature of the angular dependent linewidth coefficients is unchanged i.e. they still possess the symmetry

where X=A,B,C. x(y)=-~(n-y)

This can be easily verified from the fact that

and from the symmetry properties of the Wigner rotation In general it is

matrices and the Clebsch-Gordan coefficients (34).

no longer possible to write the linewidth coefficients in the form (6.2) as we did in the previous Chapter but, in view of the symmetry in equation (6.28), it is possible to expand the coefficients X(y)

in a series of even Legendre polynomials as

KY ()

1 XLPL(cosy)

L even

Therefore the angular dependence in equation (6.2) can be regarded as a special case of equation (6.29) where the series terminates on the L
=

4 term.

6.3

Results To investigate the effect that the non-quantization of the nuclear

spin along the magnetic field has on the linewidths we have performed a set of model calculations for a nitroxide spin probe with various order parameters. We have employed the same magnetic parameters as those used before (cf. page 118) for the cholestane spin probe whose structure is shown in figure 5.1. As far as the dynamic part of the problem is

concerned, we have used the strong collision spectral densities given in Table 4.1, with the following correlation times:

In figures (6.2) By, () Cy ()

(6.4) we compare the angular coefficients A y , ()

obtained by allowing for the non-quantization effect with those

obtained by neglecting this effect were

(X = 0 . The order parameters employed )

P2

0.9, 0.8, 0.6,

respectively, while the parameter

P4 was

obtained assuming a Maier-Saupe pseudopotential (cf. Table 2.1). It is apparent that the effect is most important for the A and C coefficients as we had anticipated in Chapter 5. Therefore, in general, the order parameter P

extracted from the B coefficients is likely to be

more accurate than that obtained from the C, if the non-quantization effect

Figure 6.2

The angular dependent linewidth coefficients A(Y),

E(Y)

,C

calculated considering the non-quantization effect (continuous

line) and neglecting it (dashed l i n e ) .

rg = 5 .
x

P2 = 0.90, P4

= 0.71,

1 0

-9

sJrad,

r2 =

1.

1 0

-9

sirad.

Figure 6.3

The angular dependent linewidth coefficients A(y),

B(y),

C(y)

calculated considering the non-quantization effect (continuous line) and neglecting it (dashed line).

- = P2

0.79, P4

0.48,

ro = 5.

lo'

slrad, ~2 = 1.

lo'

slrad.

Figure 6 . 4

The angular dependent linewidth coefficients A(y),

B ( Y ) , C(y)

calculated considering the non-quantizat ion effect (continuous

line) and neglecting it (dashed line).

TO = 5 . x

- = 0.60, P2

0.25,

lo'

slrad,

TZ

1.

lo-'

s/rad.

is neglected. The effect is clearly more important at high orders, since then the anisotropy in

is the largest.

The problem of the influence that the

non-quantization effect has on the correlation times, and on

P4

extracted

by analysing experimental angular dependent linewidth coefficients is now under investigation.

CHAPTER SEVEN

SPIN RELAXATION FOR BIRADICAL SPIN PROBES IN ANISOTROPIC ENVIRONMENTS 7.1 Introduction The large zero-field splitting of certain biradicals makes them important candidates for spin probes of anisotropic systems such as liquid crystals and membranes. The electron resonance spectrum of a

biradical dissolved in a liquid crystal may be influenced by the zerofield splitting in two quite distinct ways. Firstly the line positions

are affected because the spin probe is partially oriented by the liquid crystal solvent. In addition the zero-field splitting coupled with the

molecular reorientation constitutes a powerful spin relaxation process and so may determine the width of the spectral lines. Here we develop

a theory of such line broadening, within the limit of fast motion, and show how it results in an angular dependence of the linewidths. The application of the theory is illustrated by studying the line broadening for a nitroxide biradical dissolved in a nematic mesophase. An analysis of the angular dependence of the linewidths suggests that the diffusion model provides a better account of molecular reorientation in a liquid crystal than the strong-collision approach. Finally we draw attention to the potential value of the theory for understanding the linewidth variations in the deuteron magnetic resonance spectra of liquid crystals.

7.2 (a)

Theory

The majority of stable biradicals are formed by linking two nitroxide groups together; consequently the dominant magnetic interactions are the zero-field splitting together with the Zeeman and nitrogen hyperfine couplings. However, in developing the theory, we shall find it convenient to ignore all hyperfine interactions and such anassumption will be permissible provided the zero-field splitting exceeds the hyperfine interaction. In situations where the hyperfine terms are

important it would be a straight forward matter to extend the present theory although the book-keeping could become somewhat involved (152). We begin by obtaining the static spin hamiltonian for a biradical in an anisotropic environment and then proceed to the spin relaxation problem.

7.2 (b)

Statics

The spin hamiltonian for a given orientation of a biradica! may be written, in irreducible spherical tensor notation ( 3 4 ) , as

where

HO

contains the orientation independent Zeeman and exchange In the second term, D(~'P) is the pth component of the is the appropriate combination

interactions.

zero-field splitting tensor and T (2'-p)

of total electron spin operators. We list the components of the spatial and spin operators, which we use, in Table 7.1 because various definitions are encountered in the literature.

TAELE 7.1

The irreducible spherical tensor components

This spin hamiltonian is written in a molecular co-ordinate system and so the spin operators fluctuate in time as the molecule reorients; it is this motion which combines with the zero-field splitting to yield the spin relaxation process. The time dependence may be removed

from the spin operators by transforming to a space fixed co-ordinate system which we choose to contain the synrmetry axes of the ordering potential responsible for the partial alignment of the spin probe. This gives

where D~ (t) is a second rank Wigner rotation matrix which contains q 9-P the complete time dependence of the spin hamiltonian. The molecular motion is assumed to be sufficiently fast that the static spin hamiltonian may be obtained by taking a time or ensemble average of equation ( . ) 72. In other words the relevant rotational correlation

times are smaller than the inverse of the zero-field splitting. We

2 therefore require the average D and for an environment with D 49-P -h

symetry, such as a nematic ar smectic A liquid crystal, this quantity

vanishes unless q is zero.

The static s p i n hamiltonian is then

where the partially averaged zexo-field splitting is.defined by

P
and D 0 9-P

is a measure of the alignment of the s p i n probe by its This spin hamiltonian corresponds t o a triplet s t a t e

environment.

with a cylindrically syunnetric zero-field splitting tensor and the

magnetic behaviour of such a system is completely understood (115).

When the magnetic f i e l d is parallel t o the symmetry axis of D the

three triplet spin functions 11>,

and ]-I> are eigenfunctions of k.

For a l l other orientations of the sample these spin s t a t e s are mixed by the zero-field splitting; however,provided D is small in comparison

with t h e Zeeman splitting,then the t r i p l e t functions are good

approximations t o t h e eigenfunctions.

Within this high-field

approximation the two allowed transitions axe

and

The electron resonance spectrum therefore contains two lines with resonant fields

where D,,is the cartesian component of the partially averaged zerofield splitting tensor parallel to its symmetry axis and so to the director; the angle between the magnepic field and the director is y. The spacing between the two spectral lines changes dramatically with this angle and we shall now see that the linewidths should also

exhibit a pronounced angular dependence.

7.2 (c)

Dynamics

It is necessary to calculate the linewidths using the Liouville space or Binsch (112) formalism discussed in Chapter 4 because for certain orientations of the sample the separation between the two spectral components is comparable to the linewidth. Under such con-

ditions Redfield's relaxation theory alone is unable to predict the line shape in contrast to the situation when the transitions are either degenerate or well resolved (100). Since the motion is assumed

to be fast the line shape in a slow-passage, low-power electron resonance experiment is given by ( c . f . Section 4.2)

The vector S- is composed of the matrix elements <a1 S-1 a'> where la>denotes an eigenfunction of H. The matrix M is related to

Redfield's relaxation matrix R (100) and the transition frequencies by

Ma,'bb*

- i(uata -u)6 ab ' ' ' ab

+ Raa'bb'

'

(7.8

For the triplet-state problem the vector S- is proportional to 1,l) The resonant frequencies may be replaced by resonant fields

because, according to equation (7.6) there is a 1:l correspondence between magnetic field and frequency. The relaxation matrix is

calculated, in the usual way, from the dynamic spin hamiltonian H (t) 1 which is obtained by subtracting the static spin hamiltonian from its instaneous value Ht. () Thus

P 9q (7.9 where T (2yq) depends on the orientation of the director with respect to the magnetic field. This angular dependence is removed

by transforming the spin operator to a laboratory frame containing the field

which we write formally as

where F (27r) contains both the time and angular dependence of the dynamic perturbation. The relaxationmatrix is then found to be

where the basis functions correspond to the two allowed transitions: the matrix elements are

and

These spectral densities are related to theFourier transform of the correlation function for the spatial operator F (2,r) by

p(27r) (0) F(27r)*(t) r

-irwot dt.

The matrix M is then

within the same basis.

The angular dependence of the spectrum is

contained entirely within M for, as we have seen, the relaxation

matrix elements depend on the sample orientation and in addition

6 =

'0
2

11

P2 (cos

y)

(7.17

The general expression for the absorption spectrum is found to be

This adopts a particularly simple form for three limiting situations. One occurs when cosy equals 1/43 for then the two transitions are degenerate and setting 6 equal to zero in equation (7.18) shows that this single line has a lorentzian shape width a width

T ; ~= - A + B. ( )

(7.19

At the other extreme the two lines are well resolved and so the line separation 6 is large compared with the linewidth whose magnitude is of order A. In addition inspection of equations (7.13)

and (7.14) for A and B reveals that B is always less than A, consequently 6>B. The off-diagonal elements in M may therefore be

ignored and the spectrum calculated from equation (7.16) is found to be

a sum of two lorentzians separated by 6 and with equal widths:

For intermediate situations the entire line shape should be computed from equation (7.18); fortunately this is not always necessary.

When the molecular motion is slow in the sense that the rotational correlation time is large compared with the inverse of the microwave frequency
wg

then the non-secular spectral densities J

and J

are

small in comparison with the secular contribution J 0 '

Under these

conditions B may be ignored in comparison with A and the line shape is lorentzian for all sample orientations with a width

In the remainder of the linewidth calculation for the electron resonance spectrum of the biradical we shall take this to be the case. The calculation demands a knowledge of the spatial operator which is obtained by a direct comparison of equations (7.10) and (7.11). density J
0'

This operator may then be used to evaluate the spectral for a particular orientation as

The simplification of this multiple summation is entirely analogous to the procedure adopted in calculating the angular dependence of the

) linewidth for doublet state species (c.f. Chapter 5.

There it was

shown that terms in the summation vanish unless q equals q', because the mesophase is cylindrically symmetric, and so

where the product of rotation matrices has been replaced with a sum of Legendre polynomials by resorting to the Clebsch-Gordan series ( 4 . 3) The summation is restricted to even values of L becaude of the D mh symmetry of the environment and so the angular dependence of the linewidth is found to be

-1
T2

= X o + X 2 P2 (cosy) + X 4 P4 (cosy).

(7.24

To proceed further and obtain explicit expressions for the angular linewidth coefficients we shall require some model for the molecular reorientation process. Two quite different approaches

(strong collision and diffusion) are available, although both find it particularly convenient to restrict attention to those spin probes

for which the ordering tensor D2 is cylindrically symmetric, that 0,P is the only non-vanishing component has p=O. Of course the magnitude of the ordering tensor is determined by the solute-solvent interactions and we can only be certain that D~ is cylindrically symmetric when 0 ,P the spin probe has the same symmetry. In practice there are few spin probes with cylindrical symmetry although this does not guarantee that the ordering tensor for the remainder are not cylindrically symmetric, but now the symmetry must be established experimentally. Providing the

ordering tensor is cylindrically symmetric we may replace D 0,P

, in

and, more imporequation (7.23) for the spectral density JO by P 6 2 OP tantly, we can eliminate those terms with p not equal to p'; the expression for

reduces to:

is where the spectral density j q-P

The components of the zero-field splitting tensor must now be expressed in the co-ordinate system which contains the symmetry axis for the ordering tensor as one of the axes. There are several variants of the

first model which is based implicitly on the notion of a strong collision which results in instantaneous molecular reorientation through any angle (c.f. the previous Chapters). between reorientational collisions is
T

If, on average, the time then the correlation function,

which is the integrand in equation (7.26), decays exponentially with a single time constant
T;

accordingly

As usual we introduce correlation times dependent on the modulus of p, instead of T as a way to reflect the aniso=lPl tropy of the molecular reorientation (52). One of the major advantages
T~

of this approach is its mathematical simplicity, for example the angular

linewidth coefficients are readily calcufated to be

(7.28

Evaluation of the Clebsch-Gordan coefficients ( 3 4 ) then gives the

explicit expressions for

x, as

(7.30
and

(7.31

The linewidth coefficients thus depend on the three correlation

times

T ~ , T ~ , T ~To .

reduce the number of parameters we make the further

assumption that

in terms of T 1 0 and T and it is suggested by the comparison e.g. with the well know 2
expressions for the correlation times f o r rotational diffusion in an
isotropic system (126).
M course the validity of the assumption can

This relation provides a sort of interpolation for r

only be j u s t i f i e d by comparison with experiment and a t present t h e few

results available seem t o support these i n t u i t i v e expression for the

spectral densities in anisotropic systems.

As we mentioned in the Introduction, electron spin relaxation

for a triplet state is identical to nuclear spin relaxation, for a

nucleus of spin 1 , by t h e quadrupole interaction (153). Consequently

a partial check of equations (7.29-7.31)

is possible because Egozy

e t al. (1%) have calculated the deuterium relaxation times,

TI and

TZ, for perdeuterated benzene dissolved in

a nematic mesophase.

However, they restrict their attention to the situation when the

director is parallel t o the magnetic field, the f i e l d gradient a t the

deuterium nucleus is cylindrical symmetric and the molecular reorientation

can be described by a s i n g l e correXation time, as in the simple If we impose all of t h e s e

then we obtain t h e

version of t h e strong collisional model.

restrictions on equations ( 7 . 2 4 ) and (7.29-7.31)

secular coneriburion to the linewidth, T2 , which has the same form

as that found by Egozy et a1 b u t which is twice as large.

Since our

-1 results yield the correct limiting value for the secular part of T2

when both the system and the motion is i s o t r o p i c 0 5 3 ) we suppose their

equation is error.

At the other extreme the molecular reorientation

is assumed t o proceed via sum11 angle jumps, that is according t o the

diffusion model, discussed in Section 4 . 4

b )

According t o this

model the spectral densities at zero frequency are given by (96)

L'l (even)

(7.33

The time constants

depend on the components of the probe diffuL'qp sion tensor and on the orientational order; they have been are defined
T

9) in terms of the Nordio et a1 ( 6 ItqP matrix in Section 4.4 (dl.

The diffusion tensor D is assumed to be cylindrically symmetric and its symmetry axis to be coincident with that of the ordering tensor.

An explicit expression for the linewidth coefficients can be obtained,

remembering the symmetry rules

and

In the following section we shall compare the predictions of the two dynamical models with the angular linewldth coefficients measured for a nitroxide biradical dissolved in a nematic mesophase.

7.3 (a)

The Experiment

The b i r a d i c a l spin probe employed t o test the theoretical predictions was 1,4-bis(3-spiro-[2"N-oxyl-3'

,3'dimethyloxazolidine])

eyclohexane which has the structure shown in figure 7.1.

The spin

probe was synthesised from cyclohexan-l,4-dione using t h e route

described by Michon and Rassat (15 5 ) . The nematogen, 4,4 '-di-n-buty-

loxyazoxybenzene, was purchased from Eastman Kodak L t d , and purified

by recrystallisation from ethanol (TS+ 1 0 2 ' ~ ; TNWI 1 3 2 ' ~ ) .

In the

experiment the orientation of the director is controlled by an electric

field, consequently the sample, nade by doping the nematogen with the
spin probe, was contained in a cell made from two glass plates wpproxinately ImxlOcm
and held 125 p apart with a Teflon spacer.

The

inner surfaces of the plates were coated with tin o x i d e and t h i s allows
the application of an electric f i e l d perpendicular to the surface.

In the absence of an electric field t h e director is aligned parallel

t a t h e spectrometer's magnetic

field, however the magnetic forces

responsible f o r this alignment are overcome by a 14 kHz electric f i e l d

from an a p p l i e d voltage of 90 volts.

Under these conditions the

director is aligned perpendicular to t h e electric f i e l d and so parallel

to the plates of the cell because t h e anisotropic permittivity is

negative (39).
the director
to

The only effect of the magnetic field now is to constrain

l i e in the plane containing both the magnetic and

electric fields. is then 90'

field.

The angle y between the director and the magnetic field

minus the angle made by the e l e c t r i c field w i t h the magnetic

The cell was mounted in a goniometer and the angular dependence

of the spectrum for the spin probe in the nematic mesophase was recorded; the measurements were made at two temperature, 127OC and 1 0 4 ~ ~ .

Figure 7.1

The molecular structure of the biradical spin probe 1,4-bis

(3-spire-[2'

N-oxyl-3' ,3'-dimethoxazolidine] )

cyclohexane together with a co-ordinate system in which the ordering tensor is supposed to be cylindrically symmetric.

Typical spectra are shown in figure 7.2 for the sample at 104'~ with the director parallel and perpendicular to the magnetic field, these two spectra clearly demonstrate the angular dependence of the linewidth and we shall now describe the determination of these widths.

7.3 (b)

Spectral Analysis

The spectra in figure 7.2 are dominated by the dipolar splitting whose angular dependence is predicted by equation (7.6); for those

orientations where the dipolar splitting is well resolved there is complete agreement between the observed and theoretical line separa,..

The values of D determined from the angular dependence for II the two temperatures are given in Table 7.2. The spectrum, for the tions.

TABLE 7.2

The partially averaged magnetic parameters

, .

Temperature

11

A ,

A 11

Dl, I G,,-A,

director orthogonal to the magnetic field, also exhibits hyperfine structure from the two equivalent nitrogen nuclei, the spacing between the lines is just one half the nitrogen coupling constant because the electron-electron interaction greatly exceeds the hyperfine interaction(156). For many orientations the hyperfine splitting is

not resolved but does result in inhomogeneous broadening of the spectral

Y
gure 7.2

I 0 gauss 0

The electronresonance spectrum of the biradical in the

nematic meeophase of 4,4'-di-n-butyloxyatoxybenzene

at

1 0 4 ~ ~ w i tthe director (a) parallel and (b) perpendicular h

to the magnetic f i e l d ,

lines, consequently it is essential to allow for this broadening when obtaining the true linewidth. We have therefore simulated spectra which incorporate both the dipolar and hyperfine splittings, the linewidth was then varied until agreement with the observed spectrum was obtained. The angular dependence of the coupling constant, required

in the simulations, is given by (39)

since the anisotropy in the partially averaged g tensor is negligibly small. The values of the two components of the partially averaged

nitrogen hyperfine tensor employed in the spectral simulations are given in Table 7.2. Since a ( y ) could not always be obtained from

the spectrum of the biradical we have also measured the nitrogen coupling constant of the nitroxide monoradical, which is responsible for the central three lines in figure 7 . k The radical is almost

certainly the hydroxylamine derivative of the biradical formed by hydrogen abstraction from the solvent. In view of the great similarity

of the structures of mono and biradical we expect their partially averaged nitrogen hyperfine tensors to be the same and this is found to be so when both splittings can be measured simultaneously. The linewidths obtained by matching the theoretical and experimental spectra are plotted against the director orientation in figure 7.3 for the sample at 104'~; a similar angular dependence was observed at 127 C.
0

Figure 7.3

The angular dependence of the linewidth for the biradical dissolved in the nematic mesophase at 104 OC. line is the best fit to equation (7.25). The solid

7.3 (c)

The Molecular Symmetry

The expressions obtained for the angular linewidth coefficients are only valid when the ordering tensor D~ for the spin probe is 0,P cylindrically symmetric. The spin probe employed in the experiments does not possess cylindrical symmetry and so we must first show that D~ is the only non-zero component of the ordering tensor. There 0,o are five unknown order parameters and since there are only two

..
independent pieces of experimental information (A
II

and Dl,)all we

can hope to achieve is to show that they are consistent with an assumed cylindrical symmetry. The spin probe may exist in one of several conformations but solid-state electron resonance studies indicate that the equatorial-equatorial chair form of the trans isomer, shown in figure 7.1, is preferred (157). Examination of molecular models then suggests that the ordering tensor for the spin probe in the nematic mesophase should be cylindrically symmetric about the z axis shown in figure 7.1, together with a projection formula for the conformer. Provided the assumption of cylindrical symmetry is correct then the components of the partially averaged tensors are given by

and

A,,- Al = (312) 1 A (2,O) P2 '

(7.39

where the zeroth component of the appropriate anisotropic tensor is evaluated in the xyz co-ordinate system, shown in figure 7.1. The

nitrogen hyperfine tensor, for such oxazolidines, is cylindrically

symmetric about the x axis (157) and so

where A' is the component of the anisotropic tensor along this symmetry
I1

axis.

The zero-field splitting tensor is, to a good approximation,

also cylindrically symmetric (155,157) but about the z ' axis which makes an angle 8 with the z axis and so

Because the symmetry axes of the two magnetic tensors are not parallel . the ratio Dl, /(A,,-Al) may only be independent of temperature if the

ordering tensor is cylindrically symmetric. The values calculated for this ratio are essentially constant as we can see from the results given in Table 7.2.

for Dl, / (A,,-A! since, from equations (7.38-7.41) )

- -

The angle 8 can be extracted from these values

Calculation of 8 provides us with an additional check on the assumption of cylindrical symmetry because this angle may be estimated from molecular models. Thus the z' axis has been found to intersect

the N-0 bonds at about one quarter of the bond length from nitrogen (157) and so 8 should be about 20, given our assignment of the z axis. Since 8 is relatively small rather precise values of the parameters occurring in equation (7.42) are required to obtain a reliable estimate

of this angle.

The zero-field splitting tensor is known, with some

is found to be 230 MHz accuracy, from single crystal studies and D z' z' (157). Unfortunately the anisotropic hyperfine tensor is not known The

with such certainty and three reasonable values are available.

single crystal study of the biradical yields a value for A of 52.5MHz , ' , although the accuracy is said not to be high (157).A more reliable

value of 51.1 MHz has been determined for the structurally related 17. monoradical 2-N-oxyl,1,1,3,3,-tetramethyloxazolidine ( 5 )
I 1

Finally

A' is 48.7 MHz for another monoradical, (3-spiro- [2'N-oxyl-3' ,3'dimethyloxazolidine] )-5cr-androstane-17~-ol (1581, which, in view of the cyclic groups, should be most closely related to the biradical. The magnitudes of 8 and f' 2 obtained from the single value of D z'z'

together with these three values of A' are listed in Table 7.3 for

8 5 both temperatures. The angle e is seen to range from ' to 1' even
though the component of the anisotropic hyperfine tensor only changes by an amount, 3.8 MHz, not greatly in excess of the experimental error in A'
II

.
The derived parameters 8 and P
2

TABLE 7.3

Temperature A /MHz , ' ,

8

(OC)

14 0'

17 2'

P2

P2 0.29 0.29

52.5 51.1 48.7

' 9 1' 2 1' 5

0.46 0.47 0.49

'

11 15O

Oa31

In view of this extreme sensitivity of the angle to A;, we believe

the results for 9 are also in accord with the assumption of a c y l i n -

drically syunnetric ordering tensor.

In a d d i t i o n we shall take 0 to

be 15' in subsequent calculations because t h i s value is closest t o

our original estimate of 20
0

based on an examination of molecular

models.

7.3 (d)

The Angular Linewidth Coefficients

As we have j u s t seen our assumption of a cylindrically symmetric

ordering tensor is not unreasonable and so we may now t u n t o a detaile

analysis cf the linewidths.

The f i r s t test of the theory is to see if

the angular dependence of the linewidths is in accord w i t h that predict4

by equation ( 7 . 2 4 ) .

The theoretical angular dependence of T-1 is 2

shown as the s o l i d line in figure 7.3 and is seen to be in good agreemet

w i t h experiment; comparable agreement is also found for the linewidths

measured at 1 7 ~ 2'.

Of course such comparisons do not provide a p s r t i -

cularly severe test for the theory because the p r e d i c t e d angular

dependence of the linewidth seems d i r e c t l y f ran t h e D

-h

symmetry of the

anisotropic environment and t h e second rank nature of the zero-field splitting tensor.
In fact the good agreement between theory and

experiment may be taken as confirmation of our correct assignment of the dominant spin relaxation process.
A more searching test of the

theory is provided by an attempt t o p r e d i c t the observed angular linew i d t h coefficients

%;

these are fisted in Table 7 . 4 and were obtained

by fitting equation ( 7 . 2 4 ) for T ly) t o the experimental linewidths. 2

There are, of course, two theoretical expressions for the coefficients

-1

% and one

of the major objectives of any experimental linewidth

TABLE 7.4

The angular linewidth coefficients

Temperature

investigation is t o distinguish between the two extreme models for

molecular reorientation.
However in this study we shall be primarily

concerned with using the experimental r e s u l t s , which are not of high accuracy, simply t o illustrate the theory.

We now proceed to calculate the angular linewidth coefficients

using both models.

To make the comparison with experiment as e q u i t a b l e

as possible the ordering potential, given

in general by a sum of even

Legendre polynomials (c.5.

first term, i.e.

Chapter 21, will be restricted to j u s t the

in both calculations.

The magnitude of the s i n g l e parameter A 2 may

2;

then calculated from the observed value of the order parameter

the results of this calculation are in Table 7.5.

Because the potential

is restricted to a single tern both dynamic models lead t o s e t s of equations f o r

which contain j u s t two arbitary parameters.

In addi-

tion the r a t i o s %/XZ

and Xo/Xq of the angular linewidth coefficients

TABLE 7.5

The order parameters

Temperature
(OC)

F2

are seen to depend on the ratio of the appropriate parameters; in T ~ the case of the strong collision model this is T ~ / while for the diffusion model it is DII/DL. We begin with the strong collision model; the evaluation of the linewidth coefficients requires the order parameter calculated from the values for X 2 4 and this was

with the results given in Table 7.5.

This procedure contrasts with previous linewidth studies employing doublet states where the angular linewidth coefficients were employed to determine P (52-54). 4 The components of the zero-field splitti-ng

tensor in the principal co-ordinate system, xyz, for the ordering tensor were found to be

D(~'O)
1 ' 9 2 ( D)

252 MHz,
85 MHz,

and

The two ratios of the linewidth coefficients, X /X and X /X were 0 2 0 4 ' calculated from equations (7.29-7.31) for a wide range of values of the relative correlation time with experiment.
T

/r in an attempt to find agreement 0 2

However both X /X and XO/X4 proved to be virtually 0 2

independent of results.

T /T

0 2

and it was not possible to fit the experimental where the experimental data is most

For example at 104'~,

reliable, X /X was predicted to change only from 2.3 to 2.2 when 0 2 T ~ / was varied from 1 to 8. This prediction is in reasonable T ~ agreement with experiment, but, in marked contrast, X /X is calcu0 4

lated to be -3.7 which is over three times the observed value.

It is,

of course, possible to force agreement with the experimental value of X /X by changing the magnitude of the order parameter 0 4

%,

since

this affects just the X angular linewidth coefficient. Agreement 4 can only be achieved when P accord with small values of

is virtually zero; a result which is in

P4 determined by Raman studies (51).

However we suspect that the value is unreasonably small in view of the success of the single parameter potential in linewidth investigations employing doublet state spin probes (52-54). The analysis of the angular linewidth coefficients obtained for the biradical at 127'~ yielded similar discrepancies between theory and experiment. The angular linewidth coefficients were also evaluated using the diffusion model; again the ordering potential was restricted to a single term A 2 taking the values given in Table 7.5. 2 For these

relatively low values of A

it was found to be adequate to truncate

the series expansion of the spectral density in equation (7.33) after L"= 20. The relative linewidth coefficients were obtained for

various values of DII /D and although the results were not particularly I sensitive to D I D it did prove possible to obtain the ratios XO/X2 iI A and X /X in fair, but not complete, agreement with experiment. 0 4
II
-L

The best fit was found for a range of values for D /D and these are

listed in Table 7.6, together with the predicted linewidth ratios.

TABLE 7.6

The diffusion model predictions

Temperature
(OC)

As we expect the agreement between theory and experiment is best for the more accurate angular linewidth coefficients measured at 104'~. The agreement for the less accurate results at 127 is not so impressive although still an improvement on the predictions of the strong collisional model. The ratio DII /D- for the components of the diffusion
0

tensor is independent of temperature as we might anticipate (53). We can estimate this ratio in the hydrodynamic limit if the molecular dimensions are known (see Section5.5 ) ; using molecular models we gauge the major semi axis to be 12 about 4.5

2 while

the minor semi axis is

8 .

The hydrodynamic value of D,,/D, then found to be - is It is

2.7, which is about twice as large as the experimental value.

surprising to find a value below the hydrodynamic limit bacause similar studies of doublet-state probes have yielded value in excess of this

limit (53,128,141).

Indeed, for these systems it proved necessary

K

to introduce an anisotropic interaction tensor

(53) whose magnitude

is related to the anisotropy in the intermolecular potential (146) to account for the deviations from the hydrodynamic limit. Of course

analogous explanations cannot be invoked, for rod-like molecules,

5) when the ratio D,,/DL is less than the hydrodynamic limit ( 3 .

The diffusion model of molecular reorientation in an anisotropic medium provides an adequate description of the angular linewidth coefficients, even though the ordering potential is restricted to a single term. This limited success contrasts with the apparent failure

of the strong collision model which is only able to predict the ratio

X0/X2 when the ordering potential contains a single term.

In the

anisotropic system which we have studied the molecular reorientation would appear to occur via rotational diffusion although more results will be required before we can differentiate with complete certainty between the two models.

7.4

Additional Spectral Features When the director is orthogonal to the magnetic field the five

nitrogen hyperfine lines associated with either component of the dipolar doublet are well resolved. Close inspection of the appropriate spectrum, shown in figure 7.2, reveals a minor variation in the widths of the hyperfine lines. Such asymmetric line broadening cannot be

accounted for by the theory which we have just developed because the anisotropic hyperfine interaction was specifically excluded from the dynamic perturbation. However quite straightforward arguments (159)

show that this line broadening can be explained by inclusion of the hyperfine term in Hl(t), the asymmetric linewidth effect then originates

from the cross term between the anisotropic hyperfine interaction and the zero-field splitting tensor. We do not feel impelled to make

detailed analysis because this broadening is quite secondary to the other linewidth variation which we have observed. The spectral line shape is expected to be symmetric about the base line but examination of the spectrum for the director parallel to the magnetic field shows that this is not the case. Similar

asymmetry in the line shape has been observed for doublet state spin probes dissolved in liquid crystals and has been attributed to nonuniform alignment of the director, probably caused by thermal fluctuations ( 4 ) 19.

A slight departure from uniform alignment of the director

would also be expected to produce asymmetric line shapes for biradical spin probes and we have used spectral simulations to confirm this expectation. Of course we also expect the line shape to become aspetric when the rate of molecular rotation is comparable to the inverse of the

18. zero-field splitting ( 0 )

Unfortunately it is difficult to devise

experiments to distinguish between these two possible explanations of asymmetric line shapes. However the rate of motion may be obtained from the inverse of the diffusion tensor; from the values listed in

. Table 7 6 we can see that the ratio of the zero-field splitting to

either component of the diffusion tensor is just less than unity. Accordingly the assumption of rapid motion should be valid for this system and the asymmetric line shape presumably originates for nonuniform alignment of the director.

75 .

Quadrupole Relaxation

The problem which we have

j u s t solved

is foxmally identical to

the calculation of the spin relaxation times, for a nucleus of spin 1,

caused by a quadrupole interaction coupled to the molecular reorientation.
Consequently our results could be of value in understanding

the spin relaxation behaviour of nuclei, such as deuterium. when they

are incorporated in molecules forming part of a l i q u i d crystalline
system.

However some: caution must be exercised in the application of

For example although the neglect of the non-secular

these results.

terms in the dynamic perturbation is a good approximation for electron

resonance spectroscopy such an asaumption may not be valid for nuclear

magnetic resonance where the operating frequency is muck smaller.
If

the non-secular terms do have to be retained then t h e entire line-

shape will have to be calculated from equation (7.18) and a simple

analysis is only possible when the lines are well resolved or the transitions are completely degenerate.
In addition our assignment

of the dominant process responsible for changes in the molecular

orientation may be q u i t e irrelevant for nuclear magnetic resonance.

This situation obtains because the quadrupole interaction is considerably smaller than the zero-field splitting and so the motion with respect t o the director, which we have considered, is too fast t o make
a significant contribution to the nuclear spin relaxat ion rimes

(160 )

However, the molecular orientation is also changed by thermal fluctuations in the director and although this motion is slow on the electron
resonance time scale i t is sufficiently fast to influence the relaxation

times in nuclear magnetic resonance (161). Consequently the spectral densities employed in this Chapter would be inappropriate for calculating the deuteron relaxation times; it is however straightforward to obtain the relevant spectral densities from continuum theory (161). The calculation of the spin lattice relaxation time T also presents 1 certain difficulties especially when the two transitions are nearly degenerate. Indeed the general angular dependence of T is complicated 1

even when the spectral lines are well resolved and we shall not discuss this aspect of the problem here.

APPENDIX 1

IRREDUCIBLE TENSORS AND WIGNER ROTATION MATRICES

A tensor of rank n is a quantity that transforms under rotation as the nth direct power of a vector. The 3 dimensional representation "

of the rotation group realized in this way can be decomposed in a set of irreducible representations D(~) each of dimension (2L+1). If we

describe the rotations in terms of Euler angles (aBy) then the matrix elements of the irreducible representation in a basis where J (.. ie
2

, Jz

the angular momentum and its projection) are diagonal are (34)

The matrix elements D~ (uBy) are called Wigner rotation matrices m,n or Wigner functions or generalized spherical harmonics. Combinations of ordinary tensor components transforming according to the representation m) are called irreducible tensor components of rank L: T (L ,

Equation (A1.2) illustrates the main reason of the usefulness of irreducible tensors in problems involving rotations i.e. that their transformation properties are very simple. The set of (2L+1) compo-

nents, T(~), is called an irreducible tensor of rank L. The Euler angles E(af3y) in equation (A1.2) determine the rotations which carry the original (unprimed) coordinate system into the rotated (primed)

one.

We follow the convention of Rose (34) for the Euler angles, From equation (Al.l) it is

which is illustrated in figure Al.l.

apparent that we can express DL (af.3~)as (34) m,n

where the real quantities

are called reduced Wigner matrices. The functions D~ (06~) constitute a complete orthogonal set m,n spanning the space of the angles a,B,y. Some of their properties, which we use most often in this thesis, are: 1. Orthogonality

(A1 .5
2.

Integral of three Wigner rotation matrices

8n2 c(LL'L";~~')c(LL'L";~~') 2~"+l m+ml,m" 6n+n' ,n"

(A1 .6

where C(abc;de) is a Clebsch-Gordan coefficient (34).

Figure Al.l

The Euler angles a, 6 and y and the three Euler rotations which carry the initial (x, y, z ) coordinate system into the final (x"'

, y"' ,

z"' ) coordinate system (34).

3.

Closure

where (a&)

is the resultant of (c f.3 y ) then (a f.3 y ). 1 1 1 2 2 2

4 .

Symmetry

5.

Products

(A1 .9

6.

Special cases

D~

0,o

(OBO) = d

L (6) = PL(cosB), 0,o

(Al. 11

where Y is a spherical harmonic and P a Legendre polynomial. L ,m L

APPENDIX 2

SOME PROPERTIES OF THE KIJMMER CONFLUENT HYPERGEOMETRIC FUNCTIONS

We shall give here only the properties used in the text. A thorough discussion of the confluent hypergeometric function is given in References (79,80). The Kummer function M(a,b,z) can be defined as

where z,a,b, can take real and complex values and

A useful integral representation is

with Re b >Re a >O. Differentiating equation (A2.1) n times we find

A general asymptotic expression for large arguments is

where

The Kummer functions can be expanded in a series of modified Bessel functions

with

References P.G. de Gennes, The P h y s i c s of L i q u i d C r y s t a l s , 1974).

(Oxford U.P.,

G.R. L u c k h u r s t , Phys. B u l l e t i n ,
G. F r i e d e l , Ann. Physique,

23,

279, (1972).

19,

273, (19221.

P.A.
G.W.

Winsor, Chem. Rev.,

68,

1 , (1968).

Gray, M o l e c u l a r S t r u c t u r e and t h e P r o p e r t i e s o f L i q u i d New York, 1962).

N e t t , S t r u c t u r e and P h y s i c a l

C r y s t a l s , (Academic P r e s s ,

6.H. Brown, J . W .
V. Zwetkoff,

Doane,

V.D.

p r o p e r t i e s of L i q u i d C r y s t a l s ,

(C.R.C.

Butterworth, 1971).

A c t a Physicochim. U.S.S.R.,

-, 6

865, (1937).

W. Maier, A . Saupe, Z. Phys. Chem.,

-, 6

327, (1956).

J.L. L.A. J.A.

W.

Fergason, S c i . American,

77,

211, (1964).

Goodman, J. Vac. S c i . and Tech., van R a a l t e , Proc. I.E.E.E.,

10,

804, (1973).

56,

2146, (1968).

elfr rich, J . Chem. Phys., - 4092, (1969). 51,

105, (1943).

P. C h a t e l a i n , B u l l . Soc. F r a n c . Mineralog. C r y s t a l l o g r . , 66, W.Kas t , Land01 t ~ o r n s t e i n ,2, 266,

D. C o a t e s , G.W.

(1960).

H . K e l k e r , B. S c h e u r l e , Angew. Chem., 81, 903,

(1969).

Gray, Mol. C r y s t . L i q . C r y s t . ,

24, G.R.

163, (1973). Luckhurst 319, (1973). Meyer, W.L. McMillan, Phys. Rev., A9, 899, (1974).

,H . J .

Smith, Mol. C r y s t . L i q . C r y s t . ,

20, R.J.

W.L. McMillan, Phys. Rev., - 1921, (1973). A8,

H. Sackmann, 21, P.G.

G.R.

D. Demus, Mol. C r y s t . L i q . C r y s t . ,

239, (1973). de Gennes, Mol. C r y s t . L i q . C r y s t . , 21, 49, (1973).

L u c k h u r s t , M. P t a k 1952, (1973).

A. Sanson, J. Chem. Soc. Faraday 11,

69, -

J. Van d e r Veen, W. H. D e J e n , M.W.M.

Wanninkhof, (1973). (1973).

C.A.M.

Tienhoven, J . Phys. Chem.,

A. Saupe, Mol. C r y s t . L i q .

- 2153, 77, C r y s t . , - 211, 21,

C.W. Oseen, Trans. Paradny Soc., - 833, (1933). 29, F.C. Frank, Discuss. Faraday Soc., - 19, (1958). 25, F.M. Leslie, Quart. J. Mech. Appl. Math., - 357, (1966); 19, Archs. Ration. Mech. Anal., - 265, (1968). 28, 19;357, J.L. Ericksen, Archs. Ration. Mech. Anal., O.J. Parodi, J. Physique, - 581, (1970). 31, see, for example, H.E. Stantley, Introduction to Phase Transitions and Critical Phenomena, (Oxford U.P., 1971). (1960).

V. Zwetkoff, Acta Physicochim U.S.S.R., - 132, (1942). 16,

A. Saupe, Angew. Chem. Int. Ed., - 97, (1968). 7,

R.L. Humphries, P.G. James, G.R. Luckhurst, J. Chem. Soc. Faraday 11, - 1031, (1,972). 68, M.E. Rose, Elementary Theory of Angular Momentum, (Wiley 1957). 43, L.C. Snyder, J. Chem. Phys., - 4041, (1965). 43, A.D. Buckingham, Discuss. Faraday Soc., - 205, (1967). G. Sansone, Orthogonal Functions, (Interscience 1959). G.R. Luckhurst, in Electron Spin Relaxation in Liquids, Eds. L.T. Muus and P.W. Atkins (Plenum Press 1972); Progress in NMR Spectroscopy, Eds.J.W. Emsley, J. Feeney and L.H. Sutcliffe (Pergamon Press, to be published). G.R. Luckhurst, in Liquid Crystals and Plastic Crystals, vo1.2, EdS0G.W. Gray and P.A. Winsor, (Ellis. Horwood L.t.d. 1974). S.H. Glarum, J.H. Marshall, J. Chem. Phys., 46, 55, (1967). 16, D.H. Chen, G.R. Luckhurst, Mol. Phys., - 91, (1969); Trans. Faraday Soc., - 656, (1969). 65, A. Pines, J.J. Chang, J. Amer. Chem. Soc.,

96,

5590, (1974).

A. Saupe, 2. N a t u r f . ,

e, 161,
60,

(1964).
3599, (1974).

S. Maxcelja, J. Chem. Phys.,

J * Emsley, J - ~ y n d o n , G.R.Luckhurst,

Mol. Phys.,
Chem.

to be published;

J. Charvolin, P. ManneviXle, B. Deloche,

23 -, 3 4 5 , 11973).

Phys, Letters,

M.F. Vuks, Optics and Spectroscopy, - 361, (1966). 20,

H.E.J.

Neugebauer, Canad. J. Phys.,

18,

292, ( 1 9 5 0 ) .

S. Chandxasekhar, N . V . Madhusudana, J. Physique, - ~ 4 ,24, (1969). 30

P. Diehl, C,L, Khetrapal, NHR Basic Principles and Progress,

+

1, 1, (1969).
Priestley, P . S . Pershan, Mol. Cryst. L i q , C r y s t . ,

E.B. 23 -,

369, ( 1 9 7 3 ) .
Clark, P . S . Pershan, E.B.

S . Jen, N.A.

Priestley,

Phys. Rev. Letts., 26, 1552, (1973).

G.R.

Euckhurst, A. Sanson, Mol. Phys., 24, 1297, (1972).

G,R. Luckhurst, M. Setaka, C. Zannoni, Mol. Phys.,

28, -

49, (1974).

G . R . Lucki~urst, R. Poupko, Chem. Phys. L e t t e r s , 2 9 , 191, ( 1 9 7 4 ) .

G.R. Luckhurst,

R. Poupko, C. Zannoni, Mol. Phys.,(in

1945).

t h e press)

G. Polya, How to solve it, (Princeton U.P.

W. Maier, A. Saupe, 2. Naturf., E a , 564, (1959);

lfa, -

287, (1960) ; - 882, ( 1 9 5 9 ) . 14,

H. Bern, Sitz. Phys.

- Math.,

25,

6 1 4 , (1916);

27,
2,

1 0 4 3 , (1916).

P, Weiss, J. Phys. Radium, Paris, - 667, (1907). 6,

3.1. Kaplan, E. Drauglis, Chem. Phys. Letters,

6 4 5 , (1971).

J.R. McColf, C.S. Shih, Phys. Rev. Letters, 29, 8 5 , ( 1 9 7 2 ) .

L. Onsager, Ann. N. Y. Acad. Sci., - 627, (1949). 51, A. Isihara, J. Chem. Phys., - 1142, (1951); 19, G. Lasher, J. Chem. Phys., - 4141, (1970). 53, N. Metropolis, A.W. Rosenbluth, M.N. Rosenbluth, A.H. Teller, 21, E. Teller, J. Chem. Phys., - 1087, (1953). J. Vieillard-Baron, Preprint. J.A. Pople, Proc. Roy. Soc., - 498, (1954). A221, W.A. Steele, J. Chem. Phys., - 3197, (1963). 39, R.L. Humphries, G.R. Luckhurst, Mol. Cryst., - 269, (1974). 26, R.L. Humphries, G.R. Luckhurst, Chem. Phys. Letters, 17, - 514, (1972).

see, for example, R. K. Pathria, Statistical Mechanics, (Pergamon Press 1972). T.D. Schultz, in Liquid Crystals, 3, Eds. G.H. Brown and M.M. Labes, (Gordon and Breach 1972). R. Kubo, J. Phys. Soc. Japan, - 1100, (1962). 17, L. Landau, E.M. Lifshitz, Statistical Physics, 2nd Edn., (Pergamon Press 1969). 130, B. Strieb, H.B. Callen, G. Horwitz, Phys. Rev., - 1798,(1963); R.L.Humphries, G.R. Luckhurst, unpublished notes. J.C. Raich, R.D. Etters, L. Flax, Chem. Phys. Letters, 6, - 491, (1970). R.G. Priest, Phys. Rev. Letters, - 423, (1971). 26, P.A. Lebowhl, G. Lasher, Phys. Rev., A6, 426, (1972). M. Kac, in Statistical Physics, Phase Transitions and Superfluidity, Vol. 1, Eds. M.Chretien, E.P. Gross, S.Deser, (Gordon and Breach 1968).

79.
80.

187

mn f l r ~ e nit , (C:rentonexc
19 54 1.

F .G, Tricomi , Funzioni 3pergeometr i che

M. Abramovitz, I . A .

Segun.(Eds.),

Handbook of !lathematical

Functions, (Dover 1964).

8. Deloche,
-9

B. Cahne, D. Jerome, Mol. Cryst. L i q . Cryst.,

15

197, ( 1 9 7 1 ) .
S t a t e Comun.,

R. Alben, J.R. MeColl, C.S. Shih, Solid

11, -

1081, (1972).

G.R. Luckhurst, Wol. Cryst. L i q . Cryst., 21, 125, (1973).

W, Niederberger, P. Diehl, L Lunazzi, Mol. Phys. 2 6 , 571, (1973). .

M.J. Frieser, Phys. Rev. ~ e t t e k s ,2 4 , 1041, ( 1 9 7 0 ) ; in
Liquid Crystals 3, Eds. G . H . Brown and N.M. Labes, Part 1,
(Gordon-Breach 1972).

C.S. Shih, R Alben, J . Chem. Phys., 57, 3057, (1972). .

R. Alben, Phys. Rev. Letters,

30, 778,

(1975).

J.P. Straley, Phys. Rev., A10, 1881, (1974).

E. Sackman, P, Krebs, H.U. Rega, J. Voss, B. Hohwald,

Mo1. Cryst. Liq, Crpst.,
R.L.Bumphries,
69, -

24,

283, (1973).

G . R . Luckhurst, J. Chem. Soc. Faraday Trans. I f ,

1491, ( 1 9 7 3 ) .

G. Agostini, P.L. Nordio, G. Rigatti, U. Segre, ~ e m o r i eAccademia

Nazion. Lincei, in the press; G. Giacmetti, P.L. Nordio, G.Rigatti, U, Segre, J. Chem. Soc. Faraday, Trans II,E, 1815, (1973).
A.D,

Buckingham, Adv. ~ h k m . Phys., - 107, (1967). 12,

S. Chandrasekhar, N.Y. Madhusudana, Mol. Cryst. L i q . Cryst.,

10, -

151, (1970); Acta Cryst., 27A, 303, (1971).

J.C. Rowell, W.D. Phillips, L . R . Melby, M. Panar,

J. Chem. Phys., -, 3442, (1965). 43

H, Benoit, Ann. Phya.,

6,

561, (1951); J. Chem. Phys., 49, 517,(1952).

P.L.Nordio,

P. Busolin, J . Chem. Phys.,

55,

5485, (1971);

P.L. Mordio, G. Rigatti, U. Segre, J. Chem. Phys., 56, -

2117, (1972); Mol. Phys.,E,

129, (1973).

R Ullman, J. Chem. Phys., - 1869, ( 1 9 7 2 ) . . 56,

C.F. Polnaszek, G.V. Bruno, J.H.

Freed, J. Chem. Phys.,

58, - 3185, (1973).

F. Bloch, Phys. Rev.,
A.G.

2, 460,

11946).

Redfield, IFN J. Res. Develop., -, 1 9 , ( 1 9 5 7 ) ; 1

Adv. Mag. Res.,

1 1 (1965). ,,
U . P . 1961).

A. Abragam, The Principles of Nuclear Magnetisn,(Oxfwrd

P.A.~.~irac,The Principles of Quantum Mechanics,(Oxford UsP. 1930).

U . Fano, Rev. Mod. Pbys.,

2, 7 4 ,

(1957);

Lectures on the Many-body Problem, Vo1.2, Ed.E.R.Caianel10,

(Academic Press, 1 9 6 4 ) .

L.T. Muus, in Electron Spin Relaxation

in Liquids,

Eds. L.T. Muus and P.W. Atkins (Plenum Press 1972).

see, for example, A. Carrington, A.D. McLachlan, Introduction

to Magnetic Resonance, (Harper and ROW, 1967).

U. Fano, Phys. Rev.,

131,2 5 9 ,

(1963).

J, Deutcb, in Electron Spin Relaxation in Liquids, Eds. L.T. Muus and P.W. Atkins (Plenum Press 1972).
J.H. Freed, in Electron Spin Relaxation in L i q u i d s ,

Eds. L.T. Muus and P.W. Atkins (Plenum Press 1972).

J.H. Freed, G.K. Fsaenkel, J. Chem. Phys., - 326, (1963). 39,

R. tynden-Bell, Mol. Phys., 2 2 , 8 3 7 , (19713.

A, Baram, Z. Luz, 5. Alexander, J . Chem. Phys., 5 8 , 4 5 5 8 , ( 1 9 7 3 ) .

G. Binsch, Mol. Phys.,

2, 469,

(1968).

113. 114.

22, G.R. Luckhurst, G.F. Pedulli, Mol. Phys., - 931, (1971). A. Abragam, B. Bleaney, Electron Paramagnetic Resonance of Transition Ions, (Oxford U.P. 1970).

115. 116. 117.

W. Low, Paramagnetic Resonance in Solids (Academic Press 1960).

A. Hudson, G.R. Luckhurst, Chem. Rev., - 191, (1969). 69, see, for example, J.B. Pedersen, in Electron Spin Relaxation in Liquids, Eds. L.T. Muus and P.W. Atkins, (Plenum Press 1972).

118.

R.L. Humphries, P.G. James, G.R. Luckhurst, Symp. Faraday Soc., 5, - 107, (1971).

114. 120.

P.G. James, G.R. Luckhurst, Mol. Phys.,

20,

761, (1971).

G.R. Luckhurst, M. Setaka, Mol. Cryst. Liq. Cryst., 19, - 279, (1973).

121. 122. 123.

W. Kauzmann, Rev. Mod. Phys., - 12, (1942). 14,

J. Roberts, R. Lynden-Bell, Mol. Phys., - 689, (1970). 21,

L.D. Favro, Phys. Rev.,

119,53,

(1960); in Fluctuation Phenomena

in Solids, Ed.R.E. Burgess (Academic Press 1965). 124. 125. 126. 127. 128. 129. 130. 131. 132. 5, 7, F. Perrin, J. Phys. Radium, - 497, (1934); - 1,(1936).

D. Edwardes, Quart. J. Math., - 70, (1892). 26,

J.H. Freed, J. Chem. Phys., - 2077, (1964). 41, 48, W.T. Huntress Jr., J. Chem. Phys., - 3524, (1968). 6, M.A. Hemminga, Chem. Phys., - 87, (1974). M.A. Hemminga, Thesis, University of Groningen, (1974).

V. Zwetkoff, Acta Physicochim. U.S.S.R., - 557, (1939). 10,

H. Lippmann, Ann. Physik, - 157, (1958). 1,
F.M. Leslie, G.R.Luckhurst, H.J. Smith, Chem. Phys. Letters, 13, - 368, (1972).

133.

E.F. Carr, C.R.K. Murty, Mol. Cryst. Liq. Cryst., - 369,0972) 18,

M. Schara, M. Sentjurc, Solid State Comm., - 593, (1970). 8, G.R. Luckhurst, Chem. Phys. Letters, - 289, (1971). 9, W.L. Hubbel, M.M. McConnell, Proc. Nat. Acad. Sci. U. S., 64, - 20, 93, (1969); J. Amer. Chem. Soc., - 314, (1971).

B.G. McFarland, H.M. McConnell, Proc. Nat. Acad. Sci. U. S., 68, - 1274, (1971). H.M. McConnell, B.G. McFarland, Quart. Rev. Biophys., 3,91,(1970). P. Jost, A.S. Waggoner, O.H. Griffith, in Structure and Function of Biological Membranes, Ed. L.J. Rothfield, (Academic Press 1971). M.A. Hemminga, H.J.C. Berendsen, J. Mag. Reson.,

8, 133,

(1972).

H. Schindler, J.Seelig, J. Chem. Phys.

,z, (1973) ;65,29469 (1974) 1841,

G.R. Luckhurst, M. Setaka, Mol. Cryst. Liq. Cryst.,s,179,(1972). R.A. Champa, Mol. Cryst. Liq. Cryst., - 175, (1972). 16, J.F.W. Keana, S.B. Keana, D. Beetham, J. Amer. Chem. Soc., 89, - 3055, (1967). R.E.D. McClung, D. Kivelson, J. Chem. Phys., - 3380,(1968). 49, 146.

D. Kivelson, M.G. Kivelson, I. Oppenheim, J. Chem. Phys.,

52, - 1810, (1970).

147. 148. 149.

J.Hwang,D.Kivelson,W.Plachy,J.Chem.

Phys.,52,1810,(1970).

R.M. Golding, Trans. Faraday Soc., 59, 1513, (1963). S.A. Brooks, G.R. Luckhurst, G.F. Pedulli, Chem.Phys.Letters, 11, - 159, (1971).

150. 151. 152.

R.E.D. McClung, Can.J.Phys., 46,2271, (1968).

R. Lefebvre, J. Maruani, J. Chem. Phys.,C, 148, (1965).

G.R. Luckhurst, G.F. Pedulli, Mol. Phys., - 1043, (1971). 20,

A. Carrington, G.R. Luckhurst, Mol. Phys., - 125, (1964). 8, Y. Egozy, A. Loewenstein, B.L. Silver, Mol. Phys.,E, 177,(1970)
J. Michon, A. Rassat, J. Am. Chem. Soc., - 335, (1974). 96,
C.P. Slichter, Phys. Rev., - 479, (1955). 99,
0 Rohde, S.P. Van, W.R. Kester, O.H. Griffith, J. Am. Chem. .

SOC

., 96, 5311, (1974).

93,314,(1971). W.L. Hubbel, H.M. McConnell, J. Am. Chem. Soc., -

H. Lemaire, J. Chim. Physique, - 559, (1966). 64,

J.W. Doane, D.L. Johnson, Chem. Phys. Letters,6,291,(1970). P. Pincus, Solid State Comm., - 839, (1969). 6,