S.

Singh / Physics Reports 324 (2000) 107}269 107

PHASE TRANSITIONS IN LIQUID CRYSTALS

Shri SINGH
Department of Physics, Banaras Hindu University, Varanasi - 221 005, India

AMSTERDAM } LAUSANNE } NEW YORK } OXFORD } SHANNON } TOKYO

 Physics Reports 324 (2000) 107}269

Phase transitions in liquid crystals

Shri Singh
Department of Physics, Banaras Hindu University, Varanasi-221 005, India
Received March 1999; editor: D.L. Mills

Contents

1. Introduction 110 6. Re-entrant phase transitions (RPT) in liquid

1.1. Liquid crystals: an overview 110 crystals 204
1.2. Classi"cation of liquid crystals: symmetry 6.1. Examples of single re-entrance 205
and structures 114 6.2. Examples of multiple re-entrance 208
2. Order parameters 120 6.3. Theories for the RPT 209
2.1. De"nition of microscopic order parameters 122 7. Uniaxial nematic}biaxial nematic (N N )
6 "
2.2. De"nition of a macroscopic order phase transition 215
parameters 125 8. Phase transitions involving smectic phases 221
2.3. Relationship between macroscopic and 8.1. Smectic A}smectic C transition 221
microscopic order parameters 126 8.2. Nematic}smectic A}smectic C (NS S )
A C
3. Phase transitions in liquid crystals 126 multicritical point 223
3.1. Thermodynamics of phase transitions 129 8.3. Transitions involving hexatic smectic
3.2. Order of phase transition 130 phases 225
3.3. Experimental determination of some 9. Computer simulations of phase transitions in
phase transition properties: a critical liquid crystals 226
comment 131 9.1. Lebwohl}Lasher model 227
4. Nematic}isotropic phase transition 133 9.2. Hard-core models 228
4.1. Landau}de Gennes theory of the uniaxial 9.3. Gay}Berne model 232
nematic}isotropic (NI) phase transition 135 10. Phase transitions in liquid crystal polymers
4.2. The hard particle or Onsager-type theories 151 (LCPs) 234
4.3. The Maier}Saupe (MS) type theories 158 10.1. Nematic order in polymer solutions 235
4.4. The van der Waals (vdW) type theories 163 10.2. Nematic order in polymer melts 241
4.5. Application of density functional theory 10.3. Nematic polymers with varying degree of
(DFT) to the NI phase transition 178 #exibility 243
5. Nematic}smectic A (NS ) phase transition 189 11. Phase transitions in chiral liquid crystals 244
A
5.1. Phenomenological description of the NS 11.1. Non-ferroelectric liquid crystals 244
A
transition 190 11.2. Transitions in ferroelectric liquid crystals 251
5.2. Mean-"eld description of NS transition 193 12. Overview and perspectives 256
A
5.3. Application of density functional theory to Acknowledgements 257
NS transition 200 References 257
A

0370-1573/00/$ - see front matter ( 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 0 - 1 5 7 3 ( 9 9 ) 0 0 0 4 9 - 6
 S. Singh / Physics Reports 324 (2000) 107}269 109

Abstract

Mesogenic materials exhibit a multitude of transitions involving new phases. Studies of these phases are of
importance in a wide range of scienti"c "elds and as such have stimulated considerable theoretical and
experimental e!orts over the decades. This review article presents a comprehensive overview until this date of
the developments in this subject. An attempt is made to identify the essential key concepts and points of
di$culty associated with the study of phase transitions. The article begins with a brief introduction about the
symmetry, structure and types of liquid crystalline phases. This is followed by a discussion of the distribution
functions and order parameters which are considered as the basic knowledge essential for the study of
ordered phases. A brief discussion of the thermodynamic properties at and in the vicinity of phase transitions,
which are required to understand the molecular structure phase stability relationship, is given. The most
widely used experimental techniques for measuring these transition properties are critically examined. The
remaining parts of the article are concerned with the current status of the theoretical developments and
experimental studies in this "eld. The application of the various theories to the description of isotropic
liquid-uniaxial nematic, uniaxial nematic-smectic A, uniaxial nematic-biaxial nematic, smectic A}smectic
C phase transitions are reviewed comprehensively. The basic ideas of Landau}de Gennes theory and its
applications to study these transitions are discussed. Since the formation of liquid crystals depends on the
anisotropy in the intermolecular interactions, questions concerning its role in the mesophase transitions are
addressed. The hard particle, Maier-Saupe and van der Waals types of theories are reviewed. The application
of density functional theory in studying mesophase transitions is described. A critical assessment of the
experimental investigations concerning reentrant phase transitions in liquid crystals is made and the factors
which impede its proper understanding are identi"ed. A survey is given of existing computer simulation
studies of the isotropic to nematic transition, the nematic to smectic A transition, the smectic A to hexatic
S transition, the smectic A to reentrant nematic transition, and transitions to the discotic phase. The current
B
status of the study of phase transitions involving hexatic smectic, cholesteric, polymeric and ferroelectric
liquid crystals is outlined. Finally, a range of unexplored problems and some of the areas which are in
greatest need of future attention are identi"ed. ( 2000 Elsevier Science B.V. All rights reserved.

PACS: 64.70.Md
110 S. Singh / Physics Reports 324 (2000) 107}269

1. Introduction

1.1. Liquid crystals: an overview

The states of matter whose symmetric and mechanical properties are intermediate between those
of a crystalline solid and an isotropic liquid are called `liquid crystalsa (LC) [1}30]. The basic
di!erence between crystals and liquids is that the molecules in a crystal are ordered whereas in
a liquid they are not. The existing order in a crystal is usually both positional and orientational, i.e.,
the molecules are constrained both to occupy speci"c sites in a lattice and to point their molecular
axes in speci"c directions. Contrary to this, the molecules in liquids di!use randomly throughout
the sample container with the molecular axes tumbling wildly. Interestingly enough, many
molecular materials, in which the building blocks are anisotropic entities, exhibit more complex
phase sequences. In particular, as they are heated from the solid phase at the melting point the
following possibilities exist:

(i) Both types of order (positional and orientational) disappear at the same time and the resulting
phase will be an `isotropic liquid (IL)a (Fig. 1b) possessing T(3)]O(3) symmetry.
(ii) Only orientational order disappears leaving the positional order intact and the corresponding
phase is called a `plastic crystal (PC)a (Fig. 1c).

Fig. 1. The arrangement of molecules in various phases.

 S. Singh / Physics Reports 324 (2000) 107}269 111

(iii) The positional order either fully of partially disappears while some degree of orientational
order is maintained. The phase thus derived is called `liquid crystal (LC)a phase. A more
proper name would be mesomorphic phase (meaning intermediate phase) or mesophase
[1}30]. In this phase (Fig. 1d) each of the molecules has a tendency to align itself along
a speci"c direction de"ned by the unit vector n( which is known as the `directora.

The liquid crystal phase possesses some of the characteristics of the order, evidenced by X-ray
di!raction, found in crystalline solids (Fig. 1a) and some of the disorder, evidenced by ease of #ow,
existing in liquids. The molecules in liquid crystal phases di!use much like the molecules in a liquid,
but they maintain some degree of orientational order and sometimes some positional order also.
The amount of order in a liquid is quite small as compared to a crystal. There remains only a slight
tendency for the molecules to point more in one direction than others or to spend more time in
certain positions than in others. The value of latent heat (around 250 J/g) indicates that most of the
order of a crystal is lost when it transforms to a mesophase. In case of a liquid crystal to isotropic
liquid transition the latent heat is much smaller, typically about 5 J/g. Yet, the small amount of
order in a liquid crystal reveals itself by the mechanical and electromagnetic properties typical for
crystals.
Liquid crystalline materials in general may have various types of molecular structure. What they
all have in common is that they are anisotropic. Either their shape is such that one axis is very
di!erent from the other two or, in some cases, di!erent parts of the molecules have very di!erent
solubility properties. From the molecular structures liquid crystal can be divided into several types.
The liquid crystals derived from the rod-like molecules are called `calamiticsa. It is essential that
the mesogenic molecule be fairly rigid for at least some portion of its length (Fig. 2a), since it must
maintain an elongated shape in order to produce interactions that favour alignment. The liquid
crystals formed from disk-shape molecules (Fig. 2b) are known as `discoticsa [3,4,31}34]. Again the
rigidity in the central part of the molecule is essential. Intermediate between rod-like and disk-like
molecules are the lath-like species (Fig. 2c).
Transitions to the mesophases may be brought about in two di!erent ways; one by purely
thermal processes and the other by the in#uence of solvents. Liquid crystals obtained by the "rst
method are called `thermotropicsa whereas those obtained by the second one are `lyotropicsa.
Amphotropic materials are able to form thermotropic as well as lyotropic mesophases. The recipe
for a lyotropic liquid crystal molecule is combining a hydrophobic group at one end with
a hydrophilic group at the other end. Such amphiphilic molecules form ordered structures in both

Fig. 2. Geometrical structure of liquid crystalline molecules.

112 S. Singh / Physics Reports 324 (2000) 107}269

polar and non-polar solvents. Good examples are soaps and various phospholipids. Both of these
classes of compounds have a polar `heada group attached to a hydrocarbon `taila group. When
dissolved in a polar solvent (e.g. water), the hydrophobic tails assemble together and form the
hydrophilic `headsa to the solvent. The resulting structure for soap molecules is called a `micellea
(Fig. 3a) and for phospholipids is called a `vesiclea (Fig. 3b). Both soap and phospholipids
molecules also form a bilayer structure, with the hydrocarbon chains separated from the water by
the `heada groups. These lamellar phases are of extreme importance, for example, in the case of
phospholipids such a lipid bilayer is the structure unit for biological membranes. When these
amphiphilic molecules are mixed with a non-polar solvent (e.g. hexane) similar structures result but
now the polar `headsa assemble together with the non-polar `taila groups in contact with the
solvent. These are called the reversed phases (Fig. 3c).
Liquid crystals are also derived from certain macromolecules (e.g. long-chain polymers), usually
in solution but sometimes even in the pure state. They are known as `liquid crystal polymers
(LCPs)a [20,21,23}27]. The polymers are long-chain molecules formed by the repetition of certain
basic units or segments known as monomers. Polymers having identical, repeating monomer units
are called `homopolymersa whereas those formed from more than one polymer type are called

Fig. 3. Structures formed by amphiphilic molecules.

 S. Singh / Physics Reports 324 (2000) 107}269 113

`copolymersa. In a copolymer the monomers may be arranged in a random sequence, in an

alternating sequence, or may be grouped in blocks. The low molar mass mesogens can be used as
monomers in the synthesis of liquid crystal polymers. On the basis of molecular anisotropy two
types of structure are possible (Fig. 4) [20,21,35]. In the "rst type the macromolecule as a whole has
a more or less rigid form, whereas in the second class only the monomer unit possesses a mesogenic
structure connected to the rest of molecule via a #exible spacer. In the "rst kind, the mesomorphism
is due to rigid macromolecule as a whole or from several monomer units made up of rigid chain
segments. The dominant factor in the second class is the structure of the individual monomer units
which make up the #exible macromolecule.
On the basis of the location of the mesogenic group LCPs are divided into two kinds: main-chain
liquid crystal polymers (MCLCPs) and side-chain liquid crystal polymers (SCLCPs). Fig. 4 shows
all types of LCP which have been chemically synthesized or whose behaviour has been studied.
A rigid MCLCP can be derived if known low molar mass rod or disc-shaped mesogenic molecules
can be joined rigidly to one another. A calamitic macromolecule can also be derived from the disc
shaped monomer units. This type of structure also results from #exible macromolecules, which
form a rigid helical superstructure, via hydrogen-bonded compounds. When the linking units are
long and #exible a semi#exible polymer is derived. The degree of #exibility and the structural
composition both determine the mesomorphic properties of MCLCPs. The side-chain LCPs were
derived in 1978 by inserting a #exible spacer unit between the rigid mesogenic unit and the polymer
backbone. In SCLCPs the #exible polymer backbone has a strong tendency to adopt a random,

Fig. 4. Schematic representation of the di!erent types of liquid crystal polymers.

114 S. Singh / Physics Reports 324 (2000) 107}269

coiled conformation. When mesogenic units are attached to the #exible polymer backbone, they
will have a strong tendency to adopt an anisotropic arrangement. Clearly, these two features are
completely antagonistic, and where the mesogenic groups are directly attached to the backbone,
the dynamics of the backbone usually dominate the tendency for the mesogenic groups to orient
anisotropically; accordingly, mesomorphic behaviour is not usually exhibited. However, if a #ex-
ible spacer moiety is employed to separate the mesogenic units from the backbone, then the two
di!erent tendencies of the mesogen (anisotropic orientation) and backbone (random arrangement)
can be accommodated within one polymeric system. According to this theory, the backbone should
not in#uence the nature and thermal stabilities of the mesophases, but this is not true because the
spacer unit does not totally decouple the mesogenic unit from the backbone. However, enhanced
decoupling is generated as the spacer moiety is lengthened.
The nature of the mesophase depends sensitively on the backbone, the mesogenic unit and the
spacers. Polymers invariably consist of a mixture of chain lengths, the size and distribution of
which depend on the structural unit present and the synthesis procedure. Accordingly, there is no
such thing as pure polymer and the average size of each chain is often referred to as the `degree of
polymerization (DP)a. A polydispersity of one (monodisperse) denotes that all of the polymer
chains are identical in size.

1.2. Classixcation of liquid crystals: symmetry and structures

The symmetry of liquid crystalline phases can be categorized in terms of their orientational and
translational degrees of freedom. The nematic, smectic and columnar phase types (characteristic
features are given below) possess, respectively, 3, 2 and 1 translational degrees of freedom, and
within each type there can exist di!erent phases depending on the orientational or point group
symmetry. The point group symmetries of common liquid crystal phases are listed in Table 1. Only
those phases are included which have well-established phase structures; so-called crystal smectic
phases have been omitted. As discussed below, it can be seen from Table 1 that most nematics,
orthogonal smectics and columnar phases are uniaxial having two equal principal refractive
indices, while the tilted smectic and columnar phases are biaxial, with all three di!erent principal
refractive indices.

1.2.1. Non-chiral calamitic mesophases

Following the nomenclature as proposed originally by Friedel [36], the liquid crystals of
non-chiral calamitic molecules are generally divided into two types * nematics and smectics
[3}22,37].

1.2.1.1. Nematic phases. The nematic phase of calamitics is the simplest liquid crystal phase. In
this phase the molecules maintain a preferred orientational direction as they di!use throughout the
sample. There exists no positional order. Fig. 5a}d shows the nematic phases occurring in di!erent
types of substance. An isotropic liquid possesses full translational and orientational symmetry
T(3)]O(3). In case of isotropic liquid-nematic (IN) transition the translational symmetry T(3)
remains as in isotropic liquid, but the rotational symmetry O(3) is broken. In the simplest structure
the group O(3) is replaced by one of the uniaxial symmetry groups D or D and the resulting
= =)
phase is the uniaxial nematic phase N (Fig. 5a) with symmetry T(3)]D . The molecules tend to
6 =)
 S. Singh / Physics Reports 324 (2000) 107}269 115

Table 1
Symmetries of common liquid crystal phase types

Liquid crystal phase Point group and Optical

translational symmetry
degrees of freedom uniaxial (u(#),
u(!)), biaxial (b),
helicoidal (h)

I. Achiral phases:
Calamitic, micellar, nematic N or N D ]T (3) u (#)
6 =)
Nematic discotic (N ), columnar nematic N D ]T (3) u (!)
D C =)
Biaxial nematic (N ) D ]T (3) b
" 2)
Calamitic orthogonal smectic or lamellar phase (S ) D ]T (2) u (#)
A =)
Tilted smectic phase (S ) C ]T (2) b
C 2)
Orthogonal and lamellar hexatic phase (S ) D ]T (1) locally u (#)
B 6)
D ]T (2) globally
6)
Tilted and lamellar hexatic phases (S and S ) C ]T (1 or 2) b
F I 2)
Discotic columnar: hexagonal order of columns, ordered D ]T (1) u (!)
6)
or disordered within columns (D or D )
)0 )$
Rectangular array of columns (D or D ) D ]T (1) b
30 3$ 2)
Molecules tilted within columns (D or D ) C ]T (1) b
50 5$ 2)
II. Chiral phases:
Tilted nematic or cholesteric (N or NH) D ]T (3) b, h, locally
5 =
biaxial but
globally u (!)
Tilted smectic C phase (SH) C ]T(2) b, h
C 2
Tilted and lamellar hexatic phases C ] T (1 or 2) b, h
2
(SH and SH)
F I

Fig. 5. The arrangement of molecules in nematic phases of nonchiral calamitic mesogens: (a) uniaxial nematic phase, (b)
biaxial nematic phase, (c) nematic phase of a comblike polymer, and (d) lyotropic nematic of a sti! polymer.

align along the director n( . A biaxial nematic phase N may result due to the further breaking of the
"
rotational symmetry of the system around the director n( . The existence of this phase was originally
predicted on a theoretical basis [38,39]. The "rst N phase was observed in the ternary amphiphilic
"
116 S. Singh / Physics Reports 324 (2000) 107}269

system composed of potassium laurate, 1-decanol and D O [40] and later on it could also be
2
derived in some relatively simple thermotropic compounds [41,42]. In the N case di!erent
"
symmetry groups, which are subgroup of O(3), are in principle admissible, orthorhombic, triclinic,
hexogonal or cubic. Fig. 5b illustrates the biaxial nematic phase existing in compounds with
lath-like molecules (T(3)]D symmetry). In this phase the rotation around the long axis is
2)
strongly hindered. Fig. 5c shows the structure of the nematic phase of comb-like liquid crystal
polymer in which the mesophase moieties attached to the polymeric chain as the side groups have
parallel orientation. The main chain may possess a liquid-like structure. Fig. 5d illustrates the
lyotropic nematic phase of a rigid polymer in solution in which the rigid main chains are nearly
parallel and separated by the solvent.

1.2.1.2. Smectic phases. When the crystalline order is lost in two dimensions, one obtains stacks of
two-dimensional liquids: such systems are called smectics. The smectic liquid crystals have layered
structures, with a well-de"ned interlayer spacing which can be measured by X-ray di!raction. The
smectic molecules exhibit some correlations in their positions in addition to the orientational
ordering. In most smectics the molecules are mobile in two directions and can rotate about one
axis. The interlayer attractions are weak compared to the lateral forces between the molecules, and
the layers are able to slide over one another relatively easily. This gives rise to #uid property to the
system with higher viscosity than nematics.
A smectic can be de"ned by its periodicity in one spatial direction and by its point group
symmetry. A priori no point group is forbidden. As a result an in"nite number of smectic phases
can be expected. The observed smectic phases di!er from each other in the way of layer formation
and the existing order inside the layers. The simplest is the smectic A (S ) phase with symmetry
A
T(2)]D . In this phase the average molecular axis is normal to the smectic layers (Fig. 6a).
=)

Fig. 6. The arrangement of molecules in smectic phases of nonchiral calamitic mesogens.

 S. Singh / Physics Reports 324 (2000) 107}269 117

Within each layer the centre of gravity of molecules are ordered at random in a liquid-like fashion
and they have considerable freedom of translation and of rotation around their long axis. Thus, the
structure may be de"ned as an orientationally ordered #uid on which is superimposed a one-
dimensional density wave. The #exibility of layers leads to distortions which give rise to beautiful
optical patterns known as focal conic textures. When temperature is decreased, the S phase may
A
transform into a phase possessing even lower symmetry. The breaking of D symmetry may lead
=)
to the appearance of tilting of molecules relative to the smectic layers. The phase thus derived is
called smectic C (S ) (Fig. 6b) which possesses the symmetry T(2)]C .
C 2)
There exist several types of smectic phase with layer structures in which the molecules inside the
layer possess e!ective rotational symmetry around their long axes and are arranged in a hexogonal
(S ) or pseudohexagonal (S , S , S , S ) manner. In a S phase (Fig. 6c) the molecules are ortho-
B F G I J B
gonal with respect to the layer plane whereas in other phases they are tilted. The existence of
S phase and other phase types of higher order has also been observed in polymeric liquid crystals.
B
Several smectic phases in which the rotation around the long-molecular axes is strongly hindered
have also been observed. The highly ordered molecules produce an orthorhombic lattice if the long
axes of the molecules are orthogonal with respect to the layer planes (S , Fig. 6d) and a monoclinic
E
lattice in case of a tilted arrangement of the long axes (S and S ). Due to the three-dimensional
H K
order in these phases they are also considered as solids. Table 1 lists the various phases together
with a few of their most important properties. The smectic D (S ) phase is not included in this table
D
because it is a cubic phase and thus does not form layers. The S (hexatic), S and S phases are 2D
B I F
in character, i.e., inside the layers the molecules are oriented on a 2D lattice having no long-range
correlations. The higher order smectic phases S (or S crystal), S , S , S , S , S possess 3D
L B J G E K H
long-range order and therefore are also known as `crystal smecticsa. In S , S and S the rotation
L J G
of molecules around the molecular long axis is strongly hindered. In the S , S and S phases the
E K H
rotational hindrance around the long axes is so strong that only 1803 jumps between two favoured
positions are possible.

1.2.2. Chiral calamitic phases

If the liquid crystal molecules are chiral (lacking inversion symmetry), then chiral phases are
observed instead of certain non-chiral phases. The most important mesophases occurring in chiral
materials are listed in Table 2.

1.2.2.1. Non-ferroelectric phases. In calamitics, the uniaxial nematic phase is replaced by the
chiral nematic phase in which the director rotates about an axis perpendicular to the director
leading to a helical structure (Fig. 7a). Hence, the name twisted nematic (N or NH) or cholesteric
5
is given to this phase. Due to the helical structure this phase possesses special optical proper-
ties which makes the material very useful in practical applications. Below their clearing
point anomalous phases appear in many of them which are collectively known as `blue phasesa
[21,43,44]. Blue phases (BPs) occur in a narrow temperature range between the cholesteric and
isotropic phases. In many chiral compounds with su$ciently high twist upto three distinct blue
phases appear. The two low-temperature phases, blue phase I (BPI) and blue phase II (BP II), have
cubic symmetry, while the highest temperature phase, blue phase III (BP III), appears to be
amorphous.
118 S. Singh / Physics Reports 324 (2000) 107}269

Table 2
Mesophases in chiral materials

Phase type Ordering Molecular Molecular packing

orientation

Nonferroelectric types
N (or NH) Helical nematic Uniaxial
5
*
structure
Blue phases Cubic structure * *
Ferroelectric smectics
SH Helical, short- Tilted Random
C
range
SH No layer Tilted to side Pseudo-hexagonal
I
correlation
Short-range in-
plane corelation
SH Helical structure Tilted to apex Pseudo-hexagonal
F
SH Long-range layer Tilted to apex Pseudo-hexagonal
J
correlation
Long-range in-
plane correlation
SH No helical structure Tilted to side Pseudo-hexagonal
G
SH Long-range layer Tilted to side Herring-bone
K
correlation
Long-range in-
plane correlation
SH No helical structure Tilted to apex Herring-bone
H

1.2.2.2. Ferroelectric smectic phases. All of the smectic phases with tilted structure exhibit fer-
roelectric properties. Due to their low symmetry they are able to exhibit spontaneous polarization
and piezoelectric properties and are known as ferroelectric liquid crystals. It is well established that
any tilted smectic phase derived from chiral molecules should possess a permanent electric
polarization P which is oriented perpendicular to the director n( and parallel to the smectic layer
plane. The presence of permanent dipoles fundamentally alters the nature of the interactions
between the molecules themselves and with any cell wall or applied electric "eld. The ferroelectric
smectic-CH (SH) phase (Fig. 7b) has become very important for its fast electronic switching of the
C
twist of the layers with respect to tilt direction. The optical activity in this phase arises as a result of
molecular asymmetry. A macroscopic helical arrangement of molecules occurs as a result of
precession of the molecular tilt about an axis perpendicular to the layer planes. The tilt direction is
rotated through an azimuthal angle upon moving from one layer to the next. As the rotation is in
a constant direction, a helix is formed which is either left- or right-handed. The helical twist sense is
determined by the nature and position of the chiral centre with respect to the central core of other
mesogenic material. One 3603 rotation of the helix for the SH phase usually extends over hundreds
C
of layers and the azimuthal angle is generally found to be of the order of 0.1}0.013.
 S. Singh / Physics Reports 324 (2000) 107}269 119

Fig. 7. The arrangement of molecules in chiral calamitic mesogens: (a) twisted nematic or cholesteric (b) chiral smectic
C(S*), Arrow shows the direction of spontaneous polarization, (c) antiferroelectric, and (d) ferrielectic.
C
120 S. Singh / Physics Reports 324 (2000) 107}269

1.2.2.3. Antiferroelectric and ferrielectric phases. Two new associated phenomena [45}47] were
observed which show fundamental di!erences to the situations existing in SH phase. This led to two
C
di!erent arrangements of the molecules in their layer planes from that seen in SH phase with
C
associated variations in polarization directions. These new phases are known as antiferroelectric
(Fig. 7c) and ferrielectric phases (Fig. 7d). From the suggested structure (Fig. 7c) in a antiferroelec-
tric smectic CH phase the molecular layers are arranged in such a way that the polarization
directions in subsequent layers point in opposite directions which result in an average of the
spontaneous polarization equal to zero. This structure is evidenced by the fact that when a strong
electric "eld is applied to this phase, the layer ordering is perturbed and the phase returns to
a normal ferroelectric phase. In the switching of antiferroelectric phases three states are produced:
one antiferroelectric and two ferroelectric. This tristable switching occurs at a de"ned electric "eld
and thus the presence of a sharp switching threshold may be useful in display applications which
require multiplexing with grey scales. The structure of ferroelectric SH is repeated every 3603
C
rotation of the helix, whereas the helical structure of the antiferroelectric phase repeats every 1803
rotation. The phase, therefore, appears to have a relatively short pitch and the pitch appears to
change quite signi"cantly as the temperature is changed.
In a ferrielectric smectic phase (Fig. 7d), the layers are stacked in such a way that there is a net
overall spontaneous polarization. The number of layers of opposite polarization is not equal. There
could be, for example, twice as many layers where the polarization direction is opposite to that of
the other layer type. It also has been suggested that the stacking of the layers has two interpenetrat-
ing sublattices. There will be alternating layer structures, i.e., two layer tilted to the right and one to
the left, with this arrangement repeating itself through the bulk of the phase. Thus, the ferrielectric
phase will have a measurable polarization.

1.2.3. Discotic liquid crystals

In case of mesogens composed of disc-shape molecules, when the crystalline order is lost in one
direction (i.e. the system is melted in one dimension), one obtains aperiodical stack of discs in
columns; such systems are called discotics or columnar phases. The di!erent columns constitute
a regular 2D array and hence the structure has translational periodicity in two dimensions, but not
in three. A number of variants of this columnar structure have been identi"ed and analysed
systematically in terms of 2D (planar) crystallographic space groups. The fact that the system is
a 1D liquid indicates only the absence of correlations in the arrangement of the centres of mass of
the molecules along a straight line. On the basis of structures clearly identi"ed so far, the discotics
fall into two distinct categories: the nematic and the columnar (Fig. 8). A smectic-like phase is also
reported but the precise arrangement of the molecules in each layer is not yet fully understood. In
the columnar structures the molecules are stacked upon each other, building columns which may
be arranged in hexagonal, tetragonal and tilted variants. Along the axes of the columns, long-range
order or disorder of the molecules can exist.

2. Order parameters

As discussed in the previous section, the most fundamental characteristic of a liquid crystal is the
presence of long-range orientational order while the positional order is either limited (smectic
 S. Singh / Physics Reports 324 (2000) 107}269 121

Fig. 8. The arrangement of molecules in discotic liquid crystals.

phases) or absent altogether (nematic phases). One phase di!ers from another with respect to its
symmetry. The transition between di!erent phases corresponds to the breaking of some symmetry
and can be described in terms of the so-called order parameter (OP). It represents the extent to
which the con"guration of the molecules in the less symmetric (more ordered) phase di!ers from
that in the more symmetric (less ordered) one. In general, an order parameter describing a phase
transition, must satisfy the following requirements:

(i) Q"0, in the more symmetric (less ordered) phase, and

(ii) Q d 0, in the less symmetric (more ordered) phase.

These requirements do not de"ne the order parameter in a unique way. In spite of this
arbitrariness, in many cases the choice follows in a quite natural way. In the case of liquid}vapour
transition the order parameter is the di!erence in density between liquid and vapour phases and is
a scalar. In the case of ferromagnetic transitions without anisotropic forces, the order parameter is
the magnetization which is a vector with three components. In more complicated cases the choice
of order parameters requires some careful considerations.
122 S. Singh / Physics Reports 324 (2000) 107}269

2.1. Dexnition of microscopic order parameters

Order parameters constructed in relation to a speci"c molecular model which can give a micro-
scopic description of the system, are known as microscopic order parameters. By de"nition these
order parameters may contain more information than just the symmetry of the phase. Various
approaches have been adopted to de"ne them. Here we introduce the order parameters of di!erent
mesophases as expansion coe$cients of the singlet distribution in a suitable basis set [37,48,49]. In
case the distribution depends on the positions and orientations, we de"ne orientational, positional
and mixed (orientational-positional) order parameters. The single-particle distribution function
o(1) (,o(X)), which measures the probability of "nding a molecule at a particular position and
orientation, is the best candidate to be used for de"ning the order parameters. The singlet
distribution can be de"ned as

AB
N
o(r, X)" + + (2l#1)e~*G.rDl (X)Se~*G.r@DlH (X@)T . (2.1)
< G l m,n m,n
,m,n
Here r and X are "eld variables ("xed points in space) and therefore, unlike the dynamical variables
r@ and X@ are not a!ected by the ensemble average. The functions Dl (X) are the Wigner rotation
m,n
matrices, S T represents the ensemble average and G the set of reciprocal lattice vectors of the
crystalline phase.
Eq. (2.1) can be written as

o(r, X)"o + + Ql (G)e~*G > rDl (X) , (2.2)

0G l mn m,n
,m,n
where
Ql (G)"(2l#1)Se~G>rDlH (X)T
mn m,n
"QM lH
m,n

A BP
2l#1
" o(r, X)e~*G > rDlH (X) dr dX (2.3)
N m,n

are recognized as the order parameters.

From Eq. (2.3) the following order parameters can be de"ned:
Q (0)"1 , (2.4a)
000

A BP
1
Q (G)"k6 " dr dX o(r, X)e~*G > r , (2.4b)
000 G N

A BP
2l#1
Ql (G),(2l#1)q6 l" dr dX o(r, X)e~*G > rPl(cos h) , (2.4c)
00 G N

A BP
2l#1
Ql (0),(2l#1)DM lH" dr dX o(r, X)DlH(X) . (2.4d)
mn mn N mn
 S. Singh / Physics Reports 324 (2000) 107}269 123

Here k6 are the positional order parameters for a monatomic lattice which is characterized by
G
the set of reciprocal lattice vectors MGN. Ql (0) (or DM lH) are the orientational order parameters and
mn mn
Ql (G) (or q6 l) the mixed (orientational-positional) order parameters. It may be noted that there
00 G
can be upto (2l#1)2 order parameters of rank l. Exploiting the symmetry properties of
the mesophase and of its constituent molecules and applying the e!ects of all the operations
of the relevant space groups of the symmetry the number of order parameters can be drastically
reduced [50].
In the uniaxial mesophases (e.g. N and S phases) the singlet distribution must be invariant
6 A
under rotation about the director. If the director is chosen to be along the Z-axis, it follows that
m must be zero in Ql (or DM lH). In addition, if the mesophase has a symmetry plane perpendicular
mn mn
to the director (D symmetry) only terms with even l can appear in Ql . The most important
=) mn
order parameters are the orientational ones de"ned as

0,n P
DM lH " dX f (X)Dl (X) ,
0,n
(2.5)

where the orientational singlet distribution function, de"ned as

A BP
<
f (X)" exp[!b;(x , x ,2, x )] dx dx 2dx (2.6)
Z 1 2 N 2 3 N
N
is normalized to unity,

P f (X) dX"1 . (2.7)

;(x , x ,2, x ) is the potential energy of N particles and b"1/k ¹ with k the Boltzmann
1 2 N B B
constant. Z is the con"gurational partition function of the system,
N

N P
Z " dx dx 2dx exp [!b;(x , x ,2, x )] .
1 2 N 1 2 N
(2.8)

Here the x ("r , X ) specify both the location r of the centre of the ith molecule and its relative
i i i i
orientation X described by the Euler angles (h , / , t ). In Eqs. (2.6) and (2.8) we have used only one
i i i i
integration sign to indicate the possible multiple integration over all the variables whose volume
elements appear.
If one assumes that the mesogenic molecules also possess cylindrical symmetry, the rotation
about the molecular symmetry axis may not modify the distribution, i.e., n"0 and f (X) has to
depend only on the angle h between the director and the molecular symmetry axis. Accordingly, we
get

0,0 P
DM lH (,PM l)" dX f (h)Pl(cos h) , (2.9)

where the PM l, the ensemble average of the even Legendre polynomials, are known as the Legendre
polynomial orientational order parameters. Thus, from the knowledge of f (X) all the orientational
order parameters PM , PM , etc., can be calculated.
2 4
124 S. Singh / Physics Reports 324 (2000) 107}269

For uniaxial mesophase composed of molecules of non-cylindrical symmetry some of the

important orientational order parameters are given by
DM 2H "PM "SP (cos h)T , (2.10a)
0,0 2 2

T U
J3
DM 2H " sin2 h cos 2/ , (2.10b)
0,2 2
DM 4H "PM "SP (cos h)T . (2.10c)
0,0 4 4
It should be mentioned that PM l measures the alignment of the molecular e( axis along the
z
space-"xed (SF) Z-axis (or the director). The order parameter DM 2H is an indicator of the di!erence
0,2
in the alignment of the molecular axes e( and e( along the director. When the mesogenic molecules
x y
possess axial symmetry, the molecular axes e( and e( are indistinguishable and the order para-
x y
meters DM 2H , DM 4H , etc., vanish.
0,2 0,2
For the smectic phase with positional order in one dimension only, the singlet distribution
simpli"es to

o(r, X)"o + + (2l#1)DM lH (G)exp[iG z]Dl (X) , (2.11)

0 l mn z m,n
G ,m,n
where the Z-axis is parallel to the layer normal. Again Eq. (2.11) can be simpli"ed by using the
symmetry of the phase and the constituent molecules. For the S phase which has symmetry
A
T(2)]D and is composed of cylindrically symmetric molecules,
=)

o(r, X)"o + @ + (2l#1)DM lH (q)cos[2pqz/m ]Pl(cos h) , (2.12)

0l 00 0
q
where DGD"2pq/m , with m being the average interlayer spacing. The prime on the summation sign
0 0
indicates the condition that l is even. Here DM lH (q)"k6 and DM lH (0)"PM l represent, respectively, the
0,0 q 0,0
positional and orientational order parameters. DM lH (q) (,qN l) is the mixed (positional-orienta-
0,0 q
tional) order parameter.
The order parameters for the biaxial mesophases (e.g., biaxial nematic, smectic-C, etc.) can be
de"ned by Eq. (2.3). In de"ning order parameters for these phases it usually is assumed that the
ordered phases have the same symmetry as the constituent molecules. Thus, to characterize
a biaxial nematic (N ) phase the following orientational order parameters can be identi"ed [51,52].
"
PM "DM 2H "S1(3 cos2 h!1)T,SP (cos h)T , (2.13a)
2 0,0 2 2

T U
J3
g6 "DM 2H " sin2 h cos 2t , (2.13b)
2 0,2 2

T U
J3
k6 "DM 2H " sin2 h cos 2/ . (2.13c)
2 2,0 2
and
q6 "DM 2H "S1(1#cos2 h)cos 2/ cos 2t!cos h sin 2/ cos 2tT . (2.13d)
2 2,2 2
 S. Singh / Physics Reports 324 (2000) 107}269 125

The order parameter PM measures the alignment of the molecular e( -axis along the director,
2 z
whereas g6 measures the departure in the alignment of the molecular e( and e( axes along the
2 x y
director. The other two order parameters k6 and q6 are a measure of the biaxial ordering existing in
2 2
the system.
For the smectic-C phase with positional order in one dimension only, the singlet distribution is
de"ned by Eq. (2.11) and the corresponding order parameters are identi"ed as the positional k6 , the
q
orientational DM 2H , DM 2H , DM 2H , and DM 2H , and the mixed order parameter q6 l.
0,0 0,2 2,0 2,2 q
2.2. Dexnition of a macroscopic order parameters

In most cases, the microscopic order parameters, as de"ned above, provide an adequate
description of real mesogenic systems. However, in some cases, this microscopic description is no
longer adequate and some other means must be found for specifying the degree of order.
A signi"cant di!erence between the high-temperature isotropic liquid and the liquid crystalline
phase is observed in the measurements of all macroscopic tensor properties. Thus all macroscopic
properties, e.g., the diamagnetic susceptibility, the refractive index, the dielectric permittivity, can
be used to identify the macroscopic order parameter. As an example, we consider the diamagnetic
susceptibility. The relationship between magnetic moment M (due to the molecular diamagnetism)
and the "eld H has the form
M "s H , (2.14)
a ab b
where a, b"x, y, z, s is an element of the susceptibility tensor s. When the "eld H is static, the
ab
tensor s is symmetric (s "s ). In the isotropic phase it has the simple form
ab ab ba
s "sd . (2.15)
ab ab
For the uniaxial nematic phase, where the Z-axis is parallel to the nematic axis, s can be written
in the diagnonal form

K K
s 0 0
M
s" 0 s 0 . (2.16)
M
0 0 s
@@
s and s are the susceptibilities parallel and perpendicular to the symmetry axis, respectively.
@@ M
One "nds from a comparison between Eqs. (2.15) and (2.16) that the anisotropic part of the
diamagnetic susceptibility ful"lls the requirement imposed on an order parameter,
*s "s !1sd . (2.17)
ab ab 3 ab
Thus, an order parameter tensor Q can be de"ned as
Q "*s /*s , (2.18)
ab ab .!9
where *s is the maximum anisotropy that would be observed for a perfectly ordered mesophase.
.!9
The de"nition (2.18) of the order parameter Q covers a wider class of liquid crystals than simple
uniaxial nematics. In general, Q, is an arbitrary symmetric traceless second-rank tensor. Hence, it
126 S. Singh / Physics Reports 324 (2000) 107}269

has "ve independent elements. In a reference frame where Q is diagonal,

K K
!1(x#y) 0 0
2
Q" 0 !1(x!y) 0 . (2.19)
2
0 0 x

Here the condition of zero trace is automatically ful"lled. In addition, it allows the possibility that
all the three eigenvalues are di!erent, xO0, yO0. This corresponds to a biaxial nematic phase.
For the N phase, xO0, y"0, and x"y"0 corresponds to the isotropic liquid phase.
6
In an arbitrary reference frame, Q is given in terms of the parameters x and y, by
ab
Q "3x(n n !1d )!1y[m m !(n( ]m( ) (n( ]m( ) ] , (2.20)
ab 2 a b 3 ab 2 a b a b
where n( , m( and n( ]m( are the orthogonal eigenvectors of Q corresponding to the eigenvalues
x, !1(x#y) and !1(x!y), respectively.
2 2

2.3. Relationship between macroscopic and microscopic order parameters

When the molecules can be approximately taken as rigid, one may "nd a simple connection
between the macroscopic tensor associated with Q and the microscopic quantities de"ning the
microscopic order parameter QM lH (,QM). In fact, the level of knowledge is not the same for all the
m,n
macroscopic tensor properties. Like the macroscopic order parameter Q the microscopic order
parameter QM is the symmetric traceless second-rank tensor that can be brought on a principal
axis. We should mention that only in some simple cases there is a straightforward connection
between the macroscopic and microscopic approaches.
A relationship between Q and QM can be written for the case of a rigid rod model by noting that
the anisotropic part of the diamagnetic susceptibility is proportional to QM :
ab
*s "Ns QM , (2.21)
ab ! ab
where N is the particle density number and s is the anisotropy of the molecular magnetic
!
susceptibility. Since by de"ntion *s "Ns , within the framework of the rigid rod model for the
.!9 !
uniaxial nematic, one obtains

Q "QM . (2.22)
ab ab
For realistic models the relationship between Q and the microscopic order parameter QM may be
quite complicated. The relationship (2.22) obviously cannot be generalized.

3. Phase transitions in liquid crystals

Among the most spectacular and remarkable macroscopic events in nature are the transforma-
tions between the various states of matter. The theoretical and experimental study of phase
 S. Singh / Physics Reports 324 (2000) 107}269 127

transitions has a long and illustrious history spanning over a period of more than a century to the
present day. It is still a very active "eld of study with many unsolved problems. So the "eld has
expanded enormously; from embracing transformations between the classical states of matter, i.e.,
the solid, liquid and gaseous phases, to transitions to a variety of mesophases, and phases
characterized by such diverse properties as superconductivity, super#uidity, magnetic ordering,
surface structures, ferroelectricity, cosmological quark con"nement, chaos, topological ordering,
helix cooling of proteins, #uidity of biological membranes, etc.
Phase transitions are characterized by abrupt changes, discontinuities, and strong #uctuations.
It has been known for a long time that such singular behaviour is a consequence of a cooperative
phenomenon and thus intimately related to the interactions between the microscopic constituents
of matter. It is therefore obvious that a theoretical description of phase transitions at a microscopic
level will be very di$cult. The main di$culty with the theoretical methods is in the attempt to
include very many coupled degrees of freedom.
As discussed in Section 1, the liquid crystalline materials exhibit the richest variety of polymor-
phism. The transition between di!erent phases corresponds to the breaking of some symmetry (see
Table 1). As physically expected, the most common examples of phase transitions involve a trans-
formation from an ordered (lower symmetry) phase to a relatively disordered (higher symmetry)
phase (or vice versa) as the transition temperature is crossed. Usually, the system does not re-enter
the original state as only one of the various "eld variables is changed in a continuous manner. One
can predict the order of stability of the di!erent phases on a scale of increasing temperature simply
by utilizing the fact that a rise in temperature leads to a progressive destruction of molecular order.
Thus, the less symmetric the mesophase, the closer in temperature it lies to the crystalline phase.
This means that upon cooling the isotropic liquid, "rst the nematic, then smectic phases `without
ordera, smectic phases `with order-hexagonal structurea, smectic phases `with order-herring bone
structurea, and "nally solid/crystalline phases appear in a "xed sequence. If in a mesogen all of the
above phases can exist, they are expected to appear according to the sequence rule, which, in
a general form, may be written as

IL*blue*N*S *S *S
A C B()%9) I B(#3:) F
*S *S *S
D (3.1)
Crystal/Solid*S *S *S *S *S
H K E G J

No single compound is known yet in which the complete sequence (3.1) has been observed. In
most of the materials only a small part of the di!erent phases exist which are in complete agreement
with the predicted sequence. Some examples of real observed sequences are given in Table 3. In the
observed sequences within a homologous series there are pronounced variations for the respective
phase transition properties, e.g., for the clearing point, with increasing alkyl (or alkoxy) chain
length the clearing point decreases. An even}odd e!ect [53] is observed in the sense that clearing
points for compounds with even chain length have a higher value. In such cases, the transition
properties (e.g., transition entropy, order parameter at the transition, etc.) also show a similar
variation.
One of the basic reasons for the excitement and continued interest in liquid crystal phase
transitions is that these transitions provide numerous examples for much of the recent theoretical
128 S. Singh / Physics Reports 324 (2000) 107}269

Table 3

work on critical phenomena. Since the mesophase transitions are either weakly discontinuous or
continuous, they should display the behaviour associated with critical points, including strong
#uctuations and diverging susceptibility. One of the most signi"cant "ndings of the theoretical
works is that in the vicinity of such a transition the microscopic details of the system become
unimportant in describing the details of the transition. Instead, the range of the interactions, the
physical dimension of the system, and symmetry of the order parameter determine the behaviour of
the system very close to the transition.
Although the liquid crystals are soft systems on a macroscopic scale, they provide qualitative
solutions to the complicated, often unobservable, equations on a large scale that can be ob-
served using a polarizing microscope. Many mesophase transitions involve broken continuous
symmetries in real space and their interactions on a molecular scale are short range. As a result,
 S. Singh / Physics Reports 324 (2000) 107}269 129

#uctuations have long been known to be an important feature of LC phase transitions. Relatively
little is known about #uctuation phenomena (critical phenomena) at a "rst-order phase transition,
such as NI transitions, as compared to #uctuation-controlled second-order phase transitions. The
critical phenomena [54] in LC have unique features due to a variety of symmetries of the di!erent
phases and coupling of the di!erent order parameters. These features complicate theoretical
considerations and restrict the application of methods which are successful in other cases. Another
salient feature of mesogenic materials is that they have generic long-range correlation, even far
from critical points or hydrodynamic instabilities [55] that they could make it di$cult to access
critical regimes before being "nessed by a "rst-order phase transition. At these high temperatures,
externally supplied noise can supress the onset of macroscopic instabilities such as spatial
turbulence far from any phase transition.

3.1. Thermodynamics of phase transitions

Invaluable qualitative and quantitative information about the liquid crystalline phases can be
derived from the thermodynamic data at and in the vicinity of their phase transitions. The
molecular structure phase stability relations can usually be understood from detailed thermodyn-
amic data. In addition, reliable data are required for the de"nitive testing of the theoretical
treatments. The most frequently used data for testing theories of phase transitions are (Refs. [15](a)
and [6,56]);
(i) Transition temperatures: These are usually determined by optical microscopy or by calorimetry
and are important quantities characterizing the materials. The di!erence in transition temper-
ature between the melting and clearing points gives the range of stability of the liquid
crystalline phases.
(ii) Transition densities o (of more ordered phase o and less ordered phase o ) and fractional
03 $03
density (or volume) changes *o/o (*o"o !o ): These are determined by dilatometry and
03 03 $03
play a most important role in characterizing the phase transition.
(iii) Transition enthalpy and transition entropy: These may be measured by classical adiabatic
calorimetry or by dynamic di!erential scanning calorimetry (DSC). The transition enthalpies
and entropies between the solid and the liquid crystalline state, between the di!erent liquid
crystalline states and between the liquid crystalline state and the isotropic state are related to
the degree of internal order present in the system. When a mesogenic material is heated from
its crystalline state to above the clearing point a variety of transitions with accompanying
enthalpy and entropy changes might occur.
(iv) Order parameters at the transition: These can be measured by a number of methods.
(v) The ratio d¹/dp: Is determined by direct measurement of the e!ect of external pressure on the
transition temperature. Since the application of pressure has a strong in#uence on the range of
existence of mesophases, this ratio yields important information on the phase transition.
(vi) *C , *a, *K and C (¹), a(¹) and K(¹) on both sides of the transition: Here C (¹), a(¹) and
p p p
K(¹) are, respectively, the constant pressure heat capacity, expansion coe$cient, and isother-
mal compressibility and *x represents respective transition quantity.
(vii) Another useful quantity is

C"!o(LPM /Lo) /¹(LPM /L¹) "(L ln ¹/L ln o) M 2 . (3.2)

2 T 2 o P
130 S. Singh / Physics Reports 324 (2000) 107}269

This parameter was initially de"ned by Alben [57] as a particular sensitive probe of the relative
importance of attractive and repulsive interactions. McColl and Shih [58] were "rst to determine
C successfully. A plot of ln ¹ against ln o at constant PM is virtually linear.
2

3.2. Order of phase transition

In general, an nth-order phase transition is de"ned to be one in which the nth derivatives of the
Gibbs function (or chemical potential k) with respect to temperature and pressure change discon-
tinuously at the point of the transition. For an a}b equilibrium phase transition,

G (¹, p)"G (¹, p) , (3.3)

! "
i.e. the zeroth derivative is continuous. Thus, a "rst-order phase transition can be mathematically
formulated as follows:

G !G "0 , (3.4a)
! "

C D A B
LG LG ¸
! " # ! "R !R "*R" , (3.4b)
L¹ L¹ " ! ¹
p p

C D C D
LG LG
" ! ! "< !< "*< , (3.4c)
Lp Lp " !
T T
where *R and *< are, respectively, the changes in entropy and volume at the transition. Thus, for
the "rst-order transition, where Lk/L¹ and Lk/Lp are discontinuous, changes in enthalpy *H (or
entropy *R) and in volume *< are observed. The projection of the line of intersection of the
k surfaces onto the P}¹ plane provides a consistency relation between d¹/dp and ¹*</*H
through the Clausius}Clapeyron equation.
For a second-order phase transition, both *H and *< are zero, but discontinuities in the
second-order derivatives of Gibbs function (or chemical potential) with respect to temperature and
pressure lead to changes in the *C ,*a and *K, respectively. Mathematically, the second order
p
phase transition may be characterized as

G (¹, p)"G (¹, p),

! "

C D C D
LG LG
! " # ! "R !R "*R"0 , (3.5a)
L¹ L¹ " !
p p

C D C D
LG LG
" ! ! "< !< "*<"0 , (3.5b)
Lp Lp " !
T T

C D C D
L2G L2G 1
" ! ! " [C !C ]"*C /¹ , (3.5c)
L¹2 L¹2 ¹ p! p" p
p p
 S. Singh / Physics Reports 324 (2000) 107}269 131

C D C D
L2G L2G
" ! ! "<[K !K ]"<*K , (3.5d)
Lp2 Lp2 ! "
T T

C D C D
L2G L2G
" ! ! "<(a !a )"<*a . (3.5e)
LpL¹ LpL¹ " !

In the case of liquid crystals, changes in order parameters must also be considered. The
transition is said to be "rst-order or second-order depending on whether the order parameters
change discontinuously or continuously, respectively, at the transition point. Most of the me-
sophase transitions are characterized as weakly "rst-order in nature because they are associated
with a small value of *H and *o/o (or *</<).

3.3. Experimental determination of some transition properties: a critical comment

A knowledge of both the temperature and the heat of transition is necessary if the principles of
physical analysis are to be applied to study the mesophase transitions. From this information the
transition entropy may be calculated which may play a pivotal role for evaluating the type and
degree of order present in the system. When a material melts, a change of state occurs from a solid
to a liquid and this melting process requires energy (endothermic) from the surroundings. Similarly,
the crystallization of a liquid is an exothermic process and the energy is released to the surround-
ings. The melting transition from a solid to a liquid is an exothermic process and the energy is
released to the surroundings. The melting transition from a solid to a liquid is a relatively drastic
phase transition in terms of the structural change and relatively high energy of transition is
involved. The relatively small enthalpy changes show that the liquid crystal phase transitions are
associated with more subtle structural changes. Although the enthalpy changes at a transition
cannot identify the types of phases associated with the transitions, the magnitude of the enthalpy
change is proportional to the change in structural ordering of the phases involved. Typically,
a melting transition from a crystalline solid to a liquid crystal phase or the isotropic liquid
phase involves an enthalpy change of around 30}50 kJ/mol. This indicates that a considerable
structural change is occuring. The liquid crystal to liquid crystal and liquid crystal to isotropic
liquid transitions are associated with very much smaller enthalpy changes (K4}6 kJ/mol.)
Other smectic and crystal smectic mesophases are also characterized by the values of similar order.
The nematic-isotropic liquid transition usually gives a smaller enthalpy change (1}2 kJ/mol).
The enthalpy changes of transition between di!erent liquid crystal phases are also small.
For example, the S to S transition is often di$cult to detect because the enthalpy change
C A
is typically less than 300 J/mol. The enthalpy for S N transition is also fairly small (1 kJ/mol) and
A 6
the S N transition has enthalpy less that 1 kJ/mol. These changes can be readily detected by
C 6
optical methods.
The transition from one phase to another is not necessarily sharp but streches over a certain
range. The nematic-isotropic liquid transitions tend to be smaller than about 0.6}0.8K whereas
crystalline smectic or smectic-isotropic liquid transitions are generally much wider. Usually, the
transition of the melting point is less sharp than the clearing point leading to an absolute error of
$1 and $0.1 K, respectively.
132 S. Singh / Physics Reports 324 (2000) 107}269

A number of techniques (e.g. optical microscopy, mottler oven, microscope hot stage, di!erential
thermal analysis, calorimetry, Raman spectroscopy, etc.) are available for the measurement of
transition quantities. Most of these quantities may be determined for both pure mesogens and
mixtures. The main instruments which may be used for the measurement of transition temperatures
are mottler oven, microscope hot stage and di!erential scanning calorimetry (DSC). The mottler
oven can detect small changes in transmission as a function of temperature. For the liquid crystals,
one may expect a change of optical properties for every phase transition. Under these experimental
conditions, the transmission of a liquid crystal increases with decreasing order parameter. This
results, for solid and smectic states, in a low transparency, while the nematic and isotropic states
have a higher one. In order to perform a measurement, a melting tube is "lled with the respective
substance and placed into the oven. Any change in transmittance is monitored by a built-in
photodiode in the oven, giving a photocurrent which is proportional to the transparency of the
sample. It is the "rst heating process of a sample for which all phase transitions are quoted. With
increasing temperature, a general increase in transmission is observed due to a general decrease of
the order parameter. The determination of transition temperatures and the characterization of
liquid crystal phases can be done concurrently using a hot-stage under a polarising microscope. All
transition temperatures are measured upon "rst heating of the sample. For the melting and clearing
points the "rst change observed in the texture is of relevance. Often transition from the crystalline
to a smectic texture cannot be distinguished, because no change in optical appearance results. So in
these cases both techniques, oven and hot stage, fail and the transition temperature is deduced from
a DSC analysis. The existence of a temperature gradient in the hot stage may account for a possible
systematic error of $1}2 K. When a high precision is required, the mottler oven may be employed.
The two methods (oven and hot-stage) are the most common techniques which are applied in the
determination of phase transitions of liquid crystals.
The most widely used technique in mesophase research is the calorimetric study of the phase
transitions. From the molar heat of transition, q, the entropy change of transition is calculated from
the familiar relationship *R"q/¹. A large number of measurements using classical adiabatic
calorimetry are available. However, the bulk of thermodynamic data presently available has been
obtained by dynamic calorimetry. By the very nature of instrumentation, dynamic methods are
much less accurate than adiabatic calorimetry. The expected accuracy is usually not much better
than $1% and in some cases only $10%.
The application of the methods of di!erential thermal analysis (DTA) and di!erential scanning
calorimetry (DSC) has provided extremely rich data on the temperatures, heats of transition and
heat capacity of various phases. In DTA the sample and reference material are heated at some
linear rate. The absolute temperature and the di!erential temperature between sample and
reference are recorded. The area beneath the di!erential curve is related to calories via calibration
with a material of known heat of fusion. Since the area is due to temperature di!erence, factors such
as sample and instrument heat capacity are important. When adequately calibrated, the data may
be determined from the curves within an accuracy of $1%. The DSC involves a comparison of the
sample with an inert reference during a dynamic heating or cooling programme. It employs two
furnaces, one to heat the sample under investigation and the other to heat an inert reference
material (usually gold). The two furnaces are separately heated but are connected by two control
loops to ensure that the temperatures of both remain identical through a heating or cooling cycle.
The heating or cooling rate for each is constantly identical. A balance between the sample and the
 S. Singh / Physics Reports 324 (2000) 107}269 133

reference is maintained by adding heat via the "lament. When the sample melts, for example, from
a crystalline solid to a S phase, energy must be supplied to the sample to prevent an imbalance in
A
temperature between the sample and the reference. This energy is measured and recorded by the
instrument as a peak on a baseline. The instrument is precalibrated with a sample of known
enthalpy of transition, and this enables the enthalpy of transition to be recorded for the material
being examined. The sample is weighed into a small aluminium pan and then the pan is placed into
a holder in a large aluminium block to ensure good temperature control. The sample can be cooled
with the help of liquid nitrogen and a working temperature range of between !1803C and 6003C
may be achieved.
Although the DSC reveals the presence of phase transitions in a material by detecting the
enthalpy change associated with each phase transition, a precise phase identi"cation cannot
be made. However, the level of enthalpy change involved at the phase transition does provide
some indication of the types of phase involved. Accordingly, DSC is used in conjunction with
optical polarising microscopy to determine the mesophase types exhibited by a material. If
a transition between mesophases has been missed by optical microscopy, then DSC may reveal the
presence of a transition at a particular temperature or vice versa. After DSC, optical microscopy
should be used to examine the material very carefull to furnish information on phase structure, and
to ensure that the transitions have not been missed by DSC. This procedure hopefully may lead to
the likely identity of the mesophases. Accordingly, optical polarising microscopy and DSC are
important complementary tools in the identi"cation of the types of mesophase exhibited by
a material.
The DSC often has been successful in revealing the presence of chiral liquid crystal phases. In
case of blue phases and TGBA phase, the range of stability of mesophases are often too small to
provide a distinct enthalpy peak. The transition between the SH phase and SH phase and their
C&%33* C!/5*
transition with the SH phase involve extremely small enthalpy values. This makes detection by DSC
C
very di$cult. However, the latest DSC equipment may enable the detection of even such remark-
ably small enthalpy transitions.

4. Nematic+isotropic phase transition

The nematic liquid crystal is #uid and at the same time anisotropic because while preserving
their parallelism the molecules slide over one another freely. The experimental observations
[4,21,59] using various techniques [6,13,20] show that the order parameters decrease monotoni-
cally as the temperature is raised in the mesophase range and drop abruptly to zero at the
transition temperature. In case of uniaxial nematic the order parameter PM drops abruptly to zero
2
from a value in the range of 0.25}0.5 depending on the mesogenic material at the nematic-isotropic
(NI) transition temperature ¹ . Thus, the NI transition is "rst-order in nature, though it is
NI
relatively weak thermodynamically because only an orientational order is lost at ¹ and the heat
NI
of transition is only 1 kJ/mol. This in turn leads to large pretransitional [4,13,53,60] abrupt
increases in certain other thermodynamic properties, such as the speci"c heat, thermal expansion
and isothermal compressibility of the medium near ¹ . The changes of entropy and volume
NI
associated with this transition are typically only a few percent of the corresponding values for the
solid-nematic transition.
134 S. Singh / Physics Reports 324 (2000) 107}269

The molecular theory of liquid crystals aims to understanding the physical behaviour of these
materials in the mesophase range and in the vicinity of a phase transition. Moreover, the most
interesting feature of a theory must be its usefulness in predicting the behaviour of the system.
Accordingly, a prerequisite for a complete and satisfactory theory is a knowledge of the inter-
molecular interactions. However, such a knowledge is almost entirely lacking. The liquid crystal-
line molecules possess a strong anisotropy in both intermolecular repulsions and attractions. As
a result, in the construction of a molecular theory one faces the complications of dealing with both
spatial and angular variables of the molecules [61]. Even if the essentials of the intermolecular
interaction are known one might question the successful application of a theory because of the
enormous calculational problems. Owing to these di$culties model potentials have been introduc-
ed, i.e., the most relevant characteristics of the molecules and their mutual interactions are
represented in terms of simple models. However, this does not necessarily mean that a molecular
approach is out of question.
The theory for the nematic phase at and in the vicinity of their phase transitions has been
developing in several directions. In one of the most applied approaches one uses the phenom-
enological theory of Landau and de Gennes [62}68] in which the Helmholtz free energy is
expressed in powers of the order parameters and its gradients. In the process, "ve or more
adjustable parameters, associated with the symmetry of the system and physical processes, are
required to be determined by the experiments. While this theory is physically appealing and
mathematically convenient, it has many drawbacks, including the lack of quantitative predicting
power about the phase diagram. In another approach, mainly developed by Faber [96], the
nematic phase is treated as a continuum, in which a set of modes involving periodic distortion of an
initially uniform director "eld is thermally excited. All orientational order is assumed to be due to
mode excitations. For a system of N molecules, 2N modes are counted corresponding to all
rotational degrees of freedom. This theory works well near the solid nematic transitions but fails
close to nematic}isotropic transition. In the molecular "eld theories [97}145] one begins with
a model in the form of interparticle potentials and proceeds to calculate the solvent-mediated
anisotropic potential acting on each individual molecule. Such calculations require full knowledge
of pair correlation functions. For a potential which mimics all the important features of the
molecular structure this approach includes lengthy and complicated mathematical derivation and
numerical computation. As a consequence, too many simplifying approximations are made in the
choice of the models and in evaluation of correlation furactions and transition properties.
The initial version of mean-"eld theory is due to Onsager [97] which ascribes the origin of
nematic ordering to the anisotropic shape of molecules, i.e., to the repulsive interactions. The
Maier}Saupe (MS) theory [117] and its modi"cations and extensions [118}130] attribute the
formation of the ordered phase to the anisotropic attractive interactions. In reality, of course, both
of these mechanism is operative. Thus, in the van der Waals type theories [131}139] both
anisotropic hard core repulsions and angle dependent attractions are explicitly included. Another
type of molecular theory has been developed by Singh [49] and others [140}145]. These works are
based on the density functional approach which allows writing formally exact expressions for
thermodynamic functions and one-particle distribution functions in terms of direct correlation
functions. The molecular interactions do not appear explicitly in the theory. These types of theory
are physically more reasonable than those of Maier}Saupe or repulsion dominant hard particle
theories. However, the major di$culty with theories based on the density functional approach is
 S. Singh / Physics Reports 324 (2000) 107}269 135

associated with the evaluation of direct correlation functions which requires a precise knowledge of
intermolecular interactions. Only approximate methods are known for this purpose.
In the following subsections instead of summarizing all the research papers published so far we
discuss the basic ideas involved in the aforesaid theories and its application to the uniaxial
nematic}isotropic (NI) phase transitions.

4.1. Landau}de Gennes theory of the uniaxial nematic}isotropic (NI) phase transition

This section is devoted to the phenomenological model for the NI phase transition, the so-called
Landau}de Gennes (LDG) theory [62}68]. It is based on Landau's general description of phase
transitions and was "rst developed by de Gennes [62]. The strengths of the LDG theory are its
simplicity and its ability to capture the most important elements of the phase transition. In
addition, it has been applied to many other transitions [10] and embellished in many ways. First all
the essential ingradients of the Landau theory [65,66] will be discussed. The application of the
theory to the NI transition will be summarized. Next, attention will be paid to pretransitional
phenomena, for example, magnetically induced birefringence (or Cotton}Mouton e!ect) and light
scattering.

4.1.1. The basic ideas of Landau theory

The Landau theory [65,66] is concerned with a phenomenological description of a phase
transition. These transitions involve a change of symmetry. Generally the more symmetric (less
ordered) phase corresponds to higher temperature and the less symmetric (more highly ordered)
one to lower temperatures. The di!erence in symmetry between the two phases can be presented by
the order parameters which are constructed in such a way that they are zero in the more symmetric
phase. The mathematical description of the theory is based on the idea that the thermodynamic
quantities of less symmetric phase can be obtained by expanding the thermodynamic potential in
powers of order parameters and its spatial variations in the neighbourhood of the order}disorder
transition point and that su$ciently close to the transition, only the leading terms of the series are
important so that the said expansion becomes a single low-order polynomial. The motivation for
this simple and most elegant speculations is derived from the continuity of the change of state at
a phase transition of the second order, i.e., the order parameters show values near the transition
point. Hence, Landau's original procedure is, in principle, restricted to second-order phase
transitions. The thermodynamic behaviour of the order parameters in the less symmetric phase is
then determined from the condition that their values must minimize the postulated expansion of
the thermodynamic potential.
In order to clarify Landau's original arguments [65,66] and to discuss the main ingradients of
this model, we begin by considering a macroscopic system whose equilibrium state is characterized
by a spatially invariant, dimensionless, scalar order parameter Q. Though this situation is not
directly related to the NI phase transition, it permits to understand the basic ingradients of the
Landau approach. The general form of the thermodynamic potential G(p, ¹, Q) is postulated, near
the transition point, to be

G(p, ¹, Q)"G(p, ¹, 0)#h Q#1AQ2#1BQ3#1CQ4#2 , (4.1)

1 2 3 4
136 S. Singh / Physics Reports 324 (2000) 107}269

where G(p, ¹, 0) is the thermodynamic potential for a given temperature and pressure of the state
with Q"0. The numerical coe$cients are introduced for convenience. The coe$cients
h , A, B, C,2 are functions of p and ¹. The equilibrium state can be obtained by minimizing
1
G with respect to Q for "xed p and ¹. In other words, the thermodynamic behaviour of the order
parameter follows from the stability conditions,
dG d2G
"0, '0 . (4.2)
dQ dQ2
At the transition temperature ¹ these stability conditions are
#
dG d2G
"h "0, "A"0 (4.3)
dQ 1 dQ2
because of the coexistence of both the phases. This implies that (i) h "0, (ii) A"a(¹!¹ ) with
1 C
a"(dA/d¹) C'0. The "rst condition is derived from the requirement that the high-temperature
T
phase with Q"0 must give rise to an extreme value of G(p, ¹, Q). The second condition follows
from the behaviour of G at Q"0 for ¹ above and below the transition temperature ¹ . The
C
function G(p, ¹, 0) must have a minimum for ¹'¹ , i.e., A'0, and a relative maximum for
C
¹(¹ , i.e., A(0. The Landau theory postulates that the phase transition can be described by the
C
following expression for the di!erence in the thermodynamic potential of the two phases:
*G"G(p, ¹, Q)!G(p, ¹, 0)"1a(¹!¹ )Q2!1BQ3#1CQ4#2 . (4.4)
2 C 3 4
Here the negative sign with B has been chosen for the reason of convenience. It is further assumed
that the coe$cients B, C,2, must be weakly temperature dependent so that they can be treated as
temperature independent coe$cients.
The thermodynamic behaviour of the system follows directly from the stability condition,
dG
"0"a(¹!¹ )Q!BQ2#CQ3#2 . (4.5)
dQ C
Eq. (4.5) has the following solution near the transition point:
Q"0, the high temperature phase, (4.6a)
B$[B2!4aC(¹!¹ )]1@2
Q" C . (4.6b)
2C
Now the discontinuity of Q at the transition requires B"0, i.e., for the low temperature Q reads

C D
a(¹ !¹) 1@2
Q"$ C . (4.7)
C
Thus, for the reasons of stability the coe$cient C must be positive. For a phase which remains
invariant by replacing Q by !Q such as, for example, a binary alloy, only coe$cients belonging to
even powers in Q survive. Thus in case of a second-order phase transition, Eq. (4.1) takes the form
G(p, ¹, Q)"G(p, ¹)#1A(¹)Q2#1CQ4#2 . (4.8)
2 4
 S. Singh / Physics Reports 324 (2000) 107}269 137

It is important to mention that the Landau theory can be extended to the "rst-order phase
transition by two possible mechanism. While the "rst one corresponds to the condition B"0 and
C(0, the other considers the presence of a third-order term BQ3 in the expansion of G(p, ¹, Q). In
the former case a stabilizing sixth-order term with coe$cient E'0 in Eq. (4.1) is required.
So far only spatially uniform systems have been considered above. Eqs. (4.1) and (4.8) can be
generalized to include spatial variations of the order parameters by replacing Q by Q(r) and
including the contribution of the interaction term c[*Q(r), ¹] in a power series in *Q(r) and
retaining only the leading term
c[+Q(r)]+1h (¹)[+Q(r)]2 . (4.9)
2 2
In order for the spatially uniform state to be the state of lowest free energy, h (¹)'0 and near the
2
critical temperature, h (¹) must be approximated as a temperature-independent constant.
2
The above development is based on the hypotheses that the expansion (4.1) may be used and that
G is an analytic function of p, ¹ and Q. However, there exists, a priori, no reason to believe that
these requirements are true because in the neighbourhood of a critical point correlations between
#uctuations of the order parameter are of great importance. The problem with this theory is that
the coe$cients appearing in the expansion are phenomenological and their dependence on the
molecular properties are not determined. It is to be expected that these coe$cients have singular-
ities as a function of p and ¹. These singularities present great di$culties. As a result, it is assumed
that the presence of singularities does not a!ect the terms of the expansion. In addition, this theory
does not contain any information about the molecular interactions.
Despite the above-mentioned di$culties, Landau theory has been successfully applied to a great
variety of physical phenomena. It reveals the role of symmetry in physics. Mathematically, it is
simpler than the mean-"eld (MF) theory. The inclusion of spatial variation of order parameters
gives it a new dimension not found in MF theory.
The "rst indication about the extension of Landau theory to liquid crystals can be witnessed in
the original work of Landau [65] itself where a short paragraph is devoted to the form of the
probability density that de"nes a nematic state. Subsequently, the approach has been used
extensively [3,63}68] to provide a phenomenological justi"cation, explaining almost all observed
facts in the area of liquid crystals, for example, multilayer ordering in smectic phases, various
incommensurate modulations, ferroelectricity, elastostatics, interfacial phenomena. Indenbom and
others [69}72] applied the group theoretical and thermodynamic concepts forming the Landau
theory for studying the mesophase system.

4.1.2. The uniaxial nematic * isotropic (NI) phase transition: Landau}de Gennes theory
The nematic state is described by the symmetric tensor order parameter Q with zero trace, i.e.,
Q "0. In order to describe the "rst-order NI phase transition, it is su$cient, according to
aa
Landau}de Gennes [64], to expand the thermodynamic potential upto fourth or sixth order in the
tensor order parameter Q near the transition. Since the thermodynamic potential is a scalar, the
expansion can only contain terms that are invariant combinations of the elements Q of the order
ab
parameter. In general, the expansion reads
g"g #1A Tr(Q2)#1B Tr(Q3)#1C Tr(Q4)#1D[Tr(Q3)]
0 2 3 4 5
][Tr(Q2)]#1E[Tr(Q2)]3#E@[Tr Q3]2#2 , (4.10)
6
138 S. Singh / Physics Reports 324 (2000) 107}269

where g and g represent the Gibbs free-energy density of the nematic and the isotropic phases,
0
respectively. The term linear in Q does not appear in the expansion due to the di!erent symmetry
ab
of the two phases. In principle, the gradient terms have to be added.
From the general consideration, as discussed in Section 4.1.1, we note the following possibilities
about the expansion (4.10):

(i) The absence of a linear term in Q allows for the existence of an isotropic phase. In case an
ab
external "eld is present a linear term has to be included which makes the isotropic phase
impossible.
(ii) Since the NI transition is "rst-order, odd terms of order three and higher are allowed.
(iii) There are two independent sixth-order terms. The presence of the E@ term indicates the possible
occurrence of a biaxial nematic phase. Since for the moment we are interested only in the
uniaxial phase, the E@ term will be omitted.
(iv) The NI phase transition takes place in the neighbourhood of A"0. Therefore, it is assumed
that the temperature dependence of the free energy is contained in the coe$cient A alone and
that other coe$cients can be regarded as temperature independent. To describe the phase
transition we write
A"a(¹!¹H ) , (4.11)
NI
where a is positive constant and ¹H is a temperature close to the transition temperature ¹ .
NI NI
In order to make a comparison with molecular statistical calculations which are often performed
at constant density, we consider the Helmholtz free energy density f instead of g. For the uniaxial
nematic phase, we write the expansion
f"f #1A Tr(Q2)#1B Tr(Q3)#1C Tr(Q4)#1D[Tr(Q2)Tr(Q3)]#1E[Tr(Q2)]3 . (4.12)
0 2 3 4 5 6
For calculating the minimum of the free energy, we use the order parameter in diagonal form (Eq.
(2.19)).
The invariants of Q are
Tr(Q2)"1(3x2#y2) ,
2
Tr(Q3)"3x(x2!y2) .
4
Choosing the unixial ordering along the Z-axis (the director) and y"0, the free-energy density is
written as
9 9 9
f"3Ax2#1Bx3# Cx4# Dx5# Ex6 . (4.13)
4 4 16 40 16
Here the free energy is normalized such that f "0.
0
We consider two cases. In the simplest model D"E"0. For the minimum to be at "nite
x"x , C'0, and the sign of B is opposite to the sign of x . For a calamitic nematic B(0 and for
0 0
a discotic nematic B'0. It is convenient to express the minimum value of x in terms of PM (,3x ).
0 2 2 0
With these provisos, Eq. (4.13) reads
f"1a(¹!¹H )PM 2! 2 BPM 3#1CPM 4 . (4.14)
3 NI 2 27 2 9 2
 S. Singh / Physics Reports 324 (2000) 107}269 139

The equilibrium value of PM is obtained by minimizing the free energy (4.14) with respect to PM . This
2 2
means that PM is determined by
2
a(¹!¹H )PM !1BPM 2#2CPM 3"0 . (4.15)
NI 2 3 2 3 2
The solutions of Eq. (4.15) are
PM "0 the isotropic phase , (4.16a)
2

G C D H
B 24aC(¹!¹H ) 1@2
PM " 1$ 1! NI . (4.16b)
2B 4C B2
The correct solution which can describe the temperature dependence of the order parameter in the
nematic phase is the PM solution. The transition temperature ¹ can be calculated using the
2` NI
condition f"f , i.e.,
0
a(¹ !¹H )PM 2 !2BPM 3 #1CPM 4 "0 (4.17)
NI NI NI 9 2NI 3 2NI
and the second relation between PM and ¹ given from Eq. (4.15) as
2NI NI
a(¹ !¹H )PM !1BPM 2 #2CPM 3 "0 . (4.18)
NI NI 2NI 3 2NI 3 2NI
Eqs. (4.17) and (4.18) yield
1BPM 3 "1CPM 4 . (4.19)
9 2NI 3 2NI
Thus, the following two solutions are possible:
PM "0, ¹ "¹H (4.20a)
2NI NI NI
B B2
PM " , ¹ "¹H # . (4.20b)
2NI 3C NI NI 27aC

Obviously, the result PM "0 at ¹"¹H corresponds to the PM solution whereas the
2NI NI 2~
PM solution gives PM "(B/3C) at the higher temperature ¹ "¹H #B2/27aC. It turns out that
2` 2NI NI NI
PM solution represents the thermodynamically stable solution. Eq. (4.16b) determines a third
2NI`
temperature ¹` given by
NI
B2
¹` "¹H # (4.21)
NI NI 24aC

In case ¹'¹` , the solutions PM and PM no longer hold because of their complex behaviour.
NI 2` 2~
In conclusion, the LDG theory distinguishes four di!erent temperature regions:

(i) ¹'¹` : the minimum corresponds to an isotropic phase, PM "0.

NI 2
(ii) ¹ (¹(¹` : the minimum of free energy is still given by PM "0, i.e., the isotropic phase is
NI NI 2
the thermodynamically stable state. There exist a relative minimum at PM "PM and a relative
2 2`
maximum at PM "PM . As a result an energy barrier of height f (PM )!f (PM ) exists between
2 2~ 2~ 2`
the two minima PM "0 and PM "PM . It follows that a metastable nematic phase can be
2 2 2`
140 S. Singh / Physics Reports 324 (2000) 107}269

Table 4
Values of parameters in the Landau expansion for MBBA

Parameter Value

a 42]103 J/m3/K
B 64]104 J/m3
C 35]104 J/m3

obtained in this temperature region due to overheating. At ¹` the height of barrier becomes
NI
zero. For PM (¹` )"PM (¹` )"(B/4C) the associated point in the free-energy curve repres-
2` NI 2~ NI
ents a point of in#ection.
(iii) ¹H (¹(¹ , the minimum corresponds to a nematic phase. There exists a local minimum
NI NI
corresponding to a possible supercooled isotropic state.
The transition entropy at ¹ is given by
NI

K
L( f!f ) 1 aB2
*R"! 0 " aPM 2 " . (4.22)
L¹ 3 2NI 27C2
T/TNI

The latent heat per unit volume, *H, is given by

aB2¹
*H" NI . (4.23)
27C2
Thus, the coe$cients a, B and C can be determined from the experimental values of PM , ¹ and
2NI NI
*H. Table 4 lists the values of these parameters for MBBA.

(iv) ¹(¹H : the minimum corresponds to a nematic phase. The PM solution corresponds to the
NI 2`
lowest free-energy density, whereas the PM solution to a relative minimum and PM "0 gives
2~ 2
rise to a relative maximum.
The height h@ of the energy barrier at ¹"¹ between the isotropic PM "0 state and the nematic
NI 2
state PM "PM is given by
2 2NI
B4
h@" . (4.24)
11 664C3

4.1.3. The inyuence of external xelds on the NI phase transition

In the absence of external magnetic or electric "elds, nematic and isotropic liquids do not have
the same symmetry. The application of "elds may change the character of the NI phase transition.
The e!ect of an applied "eld is that it induces orientational order in the isotropic phase that grows
with increasing "eld intensity. For the case of positive dielectric anisotropy, the "rst-order phase
boundary in the temperature}applied "eld plane terminates in a "eld-induced critical point. With
increasing external "elds, the jump at the transition vanishes at the "eld-induced critical point. The
increase in the orientational order in the isotropic liquid results in an enhancement of the NI
transition temperature, d¹ . As a result, it is now possible to pass through the "eld-induced critical
NI
 S. Singh / Physics Reports 324 (2000) 107}269 141

point and observe the state beyond where nematic and paranematic states are indistinguishable.
The application of a magnetic "eld to nematics with a negative magnetic anisotropy induces, in
general, biaxial ordering and so a biaxial solution of Q is required. This case will be taken up in
Section 7.
In this section, the role of external "elds on the physical properties of the NI phase transition will
be examined within the framework of LDG theory. Suppose that a static magnetic "eld H is
applied to the system. Many applications of liquid crystals are strongly dependent on their
response to such external perturbations. The application of "eld leads to an extra orientation
dependent term in the free-energy density of Eq. (4.12):
f "!1H H s . (4.25)
. 2 a b ab
Expressing s in the order parameter elements (Eqs. (2.17) and (2.18)) this can be written as
ab
f "!1s6H2!1*s H H Q , (4.26)
. 2 2 .!9 a b ab
where s6"1s . The "rst term may be omitted because it is independent of the molecular ordering.
3 aa
The sign of *s must be positive. A negative sign of *s refers to the nematic ordering in which
.!9 .!9
the director is perpendicular to the "eld. Consequently, the "eld direction becomes a second axis
and the phase is biaxial. Table 5 shows the "eld e!ects on all possible phases. When an electric "eld
is applied to the system the contributing term to the free-energy density has the same form as Eq.
(4.26) with s6 and *s replaced by the average permittivity and the maximum permittivity
.!9
anisotropy, respectively.
When *s '0 the phase remains uniaxial. With the "eld direction H and the director n( along
.!9
the Z-axis and the order parameter de"ned by Eq. (2.20) with y"0, the LDC free-energy density
containing terms upto fourth order in order parameter, reads
1 9
f"!1*s H2#3Ax2# Bx3# Cx4 . (4.27)
2 .!9 4 4 16
Minimization of free energy gives
a(¹!¹H )"h/PM !1BPM !2CPM 2 , (4.28)
NI 2 3 2 3 2
where h"(1/2)*s H2. Since the PM (¹) curve (Fig. 9) has a negative slope, the order of uniaxial
.!9 2
phase with positive *s N` increases. It is clear that if hO0, the PM "0 is never a solution of
.!9 6 2

Table 5
The breaking of symmetry of a phase due to a magnetic "eld

Phase in zero "eld Phase in nonzero "eld

h"0 hO0

*s '0 *s (0
.!9 .!9
IL N` N~
6 6
N` N` N
6 6 "
N~ N N~
6 " 6
142 S. Singh / Physics Reports 324 (2000) 107}269

Fig. 9. The variation of order parameter with temperature for di!erent values of the "eld variables. The dashed line
represents the NI coexistence curve, cp is critical point and the numbers on the curves are the values of the "eld variables.

Eq. (4.28) and instead of an isotropic phase a small induced N` ordering is obtained. In order to
6
di!erentiate between the usual N` phase and the induced N` phase, the latter one is often called
6 6
the paranematic phase. The value of PM in the paranematic phase is small and can be obtained from
2
Eq. (4.28) by ignoring the coe$cients B and C:

h
PM (h)" . (4.29)
2 a(¹!¹H )
NI
Fig. 9 shows the variation of order parameter with ¹!¹H for di!erent values of the "eld variable
NI
h. It can be seen that the jump of the order parameter at the NI phase transition is directly related
to the value of the "eld. For smaller values of the "eld there exists a "rst-order phase transition
between the paranematic and the nematic phase. The order parameter jump decreases with
increasing "eld until the critical value h of the "eld is reached where there is no jump any more. At
#
this point the transition becomes second order. For "elds larger than h there is no phase transition
#
and the nematic and paranematic phases are indistinguishable.
The location of critical point is given by

f @(x)"f A(x)"f A@(x)"0 (4.30a)

or equivalently

B3 1
h "! " C(PM 0 )3 , (4.30b)
Cp 324C2 12 2NI
 S. Singh / Physics Reports 324 (2000) 107}269 143

1 1
¹ !¹0 " " (¹0 !¹H ) , (4.30c)
C NI 54aC 2 NI NI
1B 1
PM "! " PM 0 . (4.30d)
2# 6C 2 2NI
Here the superscript 0 refers to the zero "eld and ¹ the critical point. At the phase transition there
C
are two minima x ,x of equal energy with the conditions,
1 2
f (x )"f (x ) ,
1 2
f @(x )"f @(x )"0 .
1 2
The solution gives
x "x [1$J3!2q] , (4.31a)
1,2 C
where

A B
27aC h
q" (¹!¹H )"1# . (4.31b)
B2 NI 2h
C
The shape of the coexistence curve of the nematic and paranematic phase is parabolic. The shift in
the transition temperature is proportional to h:
2h
¹ (h)!¹0 " . (4.31c)
NI NI aPM 0
2NI
As is obvious from the above discussion three general predictions can be made from the theory:
(i) the existence of a paranematic phase with "eld-induced orientation order,
(ii) the increase of ¹ with the increasing "eld and
NI
(iii) the existence of a magnetic (electric) critical point.
However, it is important to mention that the related experimental e!ects are very small. Taking
maximum "elds and typical anisotropy of the order of D*s D+10~7 (CGS) one gets, for
.!9
¹ !¹H "1 K, the induced nematic order at ¹ of the order of 10~5}10~4 which is very small
NI NI NI
as compared to the typical values 0.3}0.5 at the other side of the NI phase transition. Helfrich [73]
was the "rst to observe an increase in ¹ by applying an electric "eld. Rosenblatt [74] reported
NI
a magnetic experiment. The observed e!ect is so weak that the possibility of observing the magnetic
critical point in thermotropic nematics is quite remote. Contrary to it, the "rst evidence for the
existence of an electrically induced critical point was shown by Nicastro and Keyes [75].
The possibility of the existence of critical region at the "rst-order transition line of the NI
transition has been investigated [76] in the context of epsilon (e) expansion. It has been observed
that the LDG expansion with tensorial order parameter, which has a BQ3 interaction in addition to
CQ4 (and h"0), has a critical value B"B (C, A) below which there is no transition. At the
C
critical value the system undergoes a second-order transition with no symmetry breaking. Above
the critical value of B the transition is of "rst-order. This is in contradiction with the prediction of
Landau's theory. The results hold also for d'4, since it depends only on the fact that A "0.
C
Thus, if at ¹ the behaviour is critical, assuming that in the scaling law [76] the non-analyticity
NI
144 S. Singh / Physics Reports 324 (2000) 107}269

appears at the critical point (#uid-like critical point) and also on the spinodal curve, the critical
indices of the absolute stability limit of the nematic phase (metastable) are b "b and
1
a "c "1!b . Thus, it is obvious that there is a possibility of critical behaviour with d"3 for
1 1 1
the NI phase transition. The experimental veri"cation of this argument has recently been given by
Rzoska et al. [77].
The anomalous parts of the speci"c heat capacity above the transition point have been
calculated [78] on the basis of the continuum theory of de Gennes. The excess speci"c heat capacity
at constant pressure (per volume) due to #uctuations is given by
*C (¹)"2.54]10~4¹2[¹!318.7]~1@2 J/mol/K . (4.32)
p
It is found that the empirical formula (4.32) explains the observations of Anisimov [79] for
MBBA. The amplitude of the order parameter #uctuation increases abnormally near ¹ and it
NI
brings about the anomalous increments in heat capacity.
Recently, the renormalization group (RG) technique has been used to calculate [80,81] the
¹ !¹H of the NI transition. The model free energy of LDG type can be written [80,81] as
NI NI

P C D
1
F" ddx (AQ2 #+ Q + Q )!BQ Q Q #C(Q Q )2!H Q . (4.33)
4 ij k ij k ij ij jk ki ij ij ij ij
Here ddx indicates a functional integral in d dimensions over the tensor "eld Q"Q(x). The model
(4.33) was studied extensively by the e ("4!d) expansion technique [82]. This method relies on
the fact that the MF approximation is exact for d'4. It is a perturbation expansion about the
solution for d"4. The "xed point of the RG corresponds to a second-order phase transition with
B"0. The cubic coupling was found to be a relevant term, so B was treated as perturbation. The
scaling form of the equation of state in the second order of e expansion is obtained [80]:
H b
# "f @(x@) , (4.34)
Qd Qu
where x@"t/Q1@b. The result for
f @(x)"1#x#e f (x)#e2f (x) (4.35)
1 2
is
d"3#e, u"1# 7 e, b"1! 3 e . (4.36)
13 2 26
Here t is the reduced temperature, t"(¹!¹H)/¹H . It is the temperature at which the second-
N NI
order phase transition would take place if B"0.
The strong point of this method over the other methods is that it needs only one kind of
experimental input data, namely the jump in the order parameter at ¹ . The results obtained are
NI
still far away from experimental "ndings. The calculation has been extended near the coexistence
curve [81] which is de"ned as the region of small external "eld and below the critical temperature.
Mukherjee and Saha [83] have also calculated the critical exponents numerically for the d"3
LDG model near the isolated critical point on the NI transition line from the recursion relations
[84] in the g"0 approximation from RG theory. It has been found that the critical exponents
c"1.277 and l"0.638 are in fair agreement with the best e expansion result, c"1.277 and
l"0.64.
 S. Singh / Physics Reports 324 (2000) 107}269 145

4.1.4. Density ewects on the NI transition

In order to investigate the density e!ects at the NI transition, the free-energy density can be
written [78,85}87] as

f [o, Q, ¹]"1AHQ2!1BHQ3#1CHQ4#1Eo2#j oQ2#j oQ3 . (4.37)

2 3 4 2 1 2
The term Eo2 is the free-energy density of the isotropic phase. The coe$cients j and j are the
coupling constants. 1 2
Minimization of free-energy results in the following renormalization [78] of the expansion
coe$cients:

A"AH#2(j /E)M , (4.38a)

1

B"BH!3(j /E)M , (4.38b)

2

j2
C"CH!2 1
E
.
A B (4.38c)

Here M is the quantity thermodynamically conjugate to density o. It is seen from Eqs. (4.38a) and
(4.38b) that the interaction between Q and o leads to the decrease of B and the increase of ¹H. It is
obvious that not only the director #uctuation, but also the density e!ects play an important role in
the NI transition. The density #uctuation alters the character of the NI transition and makes it very
weakly "rst-order. The modi"ed form of the temperature can be expressed [85] as
¹H (o)"¹H #a(o!o )2 . (4.39)
NI MF 0
Here a is a positive constant and o the equilibrium density without order parameter-density
0
coupling. ¹H is the MF absolute stability limit of the isotropic phase in the absence of any order
MF
parameter-density coupling. In this case, the free energy, given by Eq. (4.12), upto the fourth order
in order parameter reads [85]

f [o, Q, ¹]"f [o, ¹]#1a(¹!¹H )Q2!1BQ3#1CQ4#1j(j!j )2Q2 . (4.40)

0 2 MF 3 4 2
0
The transition temperature ¹ and the value of QH (at ¹"¹H ) are obtained as
NI NI
2B2 j(oH!oH)2
¹ "¹H # ! I , (4.41)
NI NI 9aC a

C A BD
B 4jC 1@2
QH" 1# 1! (o!o )2 , (4.42)
2C B2 0

where oH and oH are the nematic and isotropic density at ¹ , respectively.

I NI
146 S. Singh / Physics Reports 324 (2000) 107}269

The variation of ¹ with pressure and volume can be expressed as

NI

A B
d¹ !2j
NI " [(1!1/jEQ2 )2@3P~1@3!oH(1!1/jEQ2 )1@2P~2@3] , (4.43)
dp 3a NI N 1 NI N

d(ln ¹ )
NI "2joH(oH!oH)/a¹ (4.44)
d(ln <) I NI

with

E"(oH/oH)2/[1#(oH/oH)] . (4.45)
I I
Here P is the pressure at the nematic phase.
N
Using the Landau expansion parameter determined from experimental data, d¹ /dp and
NI
d(ln ¹ )/d(ln <) were calculated [85]. The calculated values of d¹ /dp ("41.50 K/kbar) and
NI NI
d(ln ¹ )/d(ln <) ("0.3964) agree well with the experimental values of 20}40 K/kbar and 0.39,
NI
respectively and also with the low value of ¹ !¹H "1 K. However, the value of QH!Q ,
NI NI NI
instead of being very close (K2% experimentally), remains about 50% as obtained from the LDG
expansion without incorporating any e!ect of density variation. This QH!Q discrepancy implies
NI
that the change in the value of the calculated order parameter over a small temperature interval
1 K (as ¹ !¹H K1 K) would be much higher than that observed in the experiments: The
NI NI
observed value [85] of (dQ/d¹) ("0.3998) is in gross disagreement with the observed value
T/TNI
0.008. Thus it is obvious that a complete resolution of the ¹ !¹H puzzle remains outside the
NI NI
realm of a simple MF analysis where #uctuations are ignored. These results support the molecular
MF results of Tao et al. [88] and also brings out the inadequacy in explaining the small value of
(QH!Q )/Q and (dQ/d¹) in the LDG framework.
NI NI T/TNI
It is seen from Eqs. (4.38a), (4.38b) and (4.38c) that as the temperature decreases and o increases,
the quantity LM/do decreases and the coe$cient C tends to zero, i.e., a tricritical point (TCP)
appears. On further variation of the temperature the density coe$cient changes sign. In general,
two coe$cients having the same symmetry in the Landau free energy vanish simultaneously, and
the corresponding point is called tricritical (TCP). Since the coe$cients C and A have same
symmetry, C"A"0 gives a TCP. In that case a positive stabilizing sixth-order term with positive
coe$cient has to be included in the free-energy expansion [64,86,87]. In such a situation, one way
of studying the weakly "rst-order NI transition occurring near a TCP would be to have a B with
a small non-zero value. The value of ¹ !¹H thus obtained, without density variation, was
NI NI
7.68 K and (QH!Q )/Q K26% which shows substantial improvement over the model
NI NI
(4.14). Taking into account the density variation, a substantial improvement in the value of
¹ !¹H (K0.9999 K) has been obtained [78] but the value of (QH!Q )/Q did not improve.
NI NI NI NI
In another approach [86], by assuming B to be so small that it can be discarded and taking small
negative value of C and sixth-order stabilizing term, the "rst-order nature of the NI transition and
the neighbourhood of the TCP both can be achieved. The value of the ¹ !¹H calculated from
NI NI
the scaling equation of state [86] is 2.55 K and that of (QH!Q )/Q K15.47%, which shows an
NI NI
improvement of 70% over the previous results. The critical exponents obtained are b"0.25 and
D"1.25, whereas experiments show these values to be b"0.247$0.01 and D"1.26$1.0.
 S. Singh / Physics Reports 324 (2000) 107}269 147

4.1.5. Inyuence of non-mesogenic impurities on the NI transition

In general, all real nematic liquid crystals contain impurities. It is well known [89] that the solute
impurities which are not su$ciently rod-like and rigid in molecular structure depress the NI
transition temperature. Hence, the addition of non-mesogenic impurities to the pure nematic leads
to a broadening of the NI transition temperature and the appearance of a two-phase region. In the
context of LDG theory this problem has been studied by Mukherjee [87]. It has been found that
this two-phase region indicates the "rst-order character of the NI transition and the depression of
the ¹ is related to the width of the two-phase region and the transition entropy *R of the pure
NI
nematic solvent

*x
¹ !¹x " , (4.46)
NI NI *R

where *x is the width of the two-phase region and ¹x is the NI transition temperature in the
NI
absence of coupling between the concentration of the non-mesogenic substance and Q. Obviously,
for *RO0, a two-phase region must exist at a "xed temperature. Furthermore, the lower the values
of *R, the larger the depression of the ¹ . The existence of two-phase region indicates the
NI
tricritical behaviour of the NI transition.

4.1.6. Landau theory of the NI transition: inclusion of yuctuations

Though the singular-like behaviour of various quanties at the NI transition is yet a puzzle, the
solution could be obtained within the framework of LDG theory. Since the nature of NI transition
is "rst order, the part of critical region, closest to the critical point is not accessible to the
experiments. It was suggested by Keyes and Shane [90] that the critical exponents for quantities
diverging towards ¹H before being cut o! by a "rst-order transition at ¹ , should be characteristic
NI NI
of a TCP. The di!erence between critical and tricritical behaviour of the NI transition is di$cult to
verify. The value of the exponent b ("0.247$0.01) obtained for 8 CB strongly supports the
suggestion of tricritical character of the NI phase transition. On the other hand, the experimental
value of speci"c heat near the NI phase transition of MBBA indicates that the behaviour of this
transition is near tricritical and does not appear to agree with the LDG model (4.12).
In the nematic phase the `directora #uctuations are critical. Nelson and Pelcovits [91] pointed
out that strongly developed director #uctuations could alter the character of the NI transition and
make it very weakly "rst-order. The RG approach has been applied [92] to the description of
#uctuation behaviour near the isolated critical point on the NI transition line. Recently, Wang and
Keyes [93] have calculated the #uctuation of "ve components of the orientational order parameter
of a nematic liquid crystal involving several types of critical and multicritical points. The critical
behaviour of dielectric permittivity in the isotropic phase has been discussed by Rzoska et al.
[77,94]. Thus, the study of pretransitional (#uctuation) phenomena near the "rst-order phase
transition may provide an answer for their closeness to the second-order transition. Light
scattering and magnetic birefringence measurements in the isotropic phase of nematogens present
strong pretransitional e!ects and can be understood within the framework of LDG theory.
The two important ingredients that in#uence the thermodynamic functions near a phase
transition are the order parameter #uctuation amplitude and the spatial correlations between the
#uctuations of the order parameter. The average size of the range of correlations between
148 S. Singh / Physics Reports 324 (2000) 107}269

the #uctuations de"nes the so-called correlation length m. Far away from the critical point m is of
the order of the intermolecular distance. On the other hand, near a second-order phase transition
the correlations decrease very slowly with distance, indicating a divergence of m at the critical
temperature. The Landau description of #uctuation based phenomena can be expected to be valid
if the Ginzburg criterion is satis"ed,
bm2<k ¹ , (4.47)
B
where b is the height of the barrier seperating the ordered and disordered states. Close to ¹ the
NI
important #uctuations involve regions of volume m3. The inequality (4.47) then states that
thermally activated #uctuations leading from ordered to disordered regions or vice versa do not
occur with a signi"cant probability. As a result, the LDG expansion is expected to be valid in case
the #uctuations are small.
The #uctuations usually can be included in the Landau theory by calculating the free-energy
density at each temperature as a function of both the order parameter and its spatial derivatives.
For the low-energy (long-wavelength) #uctuations only the lower-order spatial derivatives of the
order parameter are considered [13,63,64]. The free-energy density can be expanded with respect to
both Q(r) and its derivatives. As there is no way of forming a scalar quantity linear in d Q , where
r ab
d "d/dx , the lowest order spatial derivative invariants of Q have the form
r r ab
(d Q )2, (d Q )2 .
a bc b bc
As a result the LDG expansion that can describe the e!ect of the local #uctuations in the
isotropic phase reads
f [Q(r), R Q(r), ¹, p]"f #1A Tr Q2#1B Tr Q3#1C[Tr Q2]2
i 0 2 3 4
#1D[Tr(Q2)(Tr Q3)]#1E(Tr Q2)3#E@(Tr Q3)2
5 6
1
!1*s HQH! +e EQE#1¸ (R Q )2#1¸ (R Q )2 (4.48)
2 .!9 12p .!9 2 1 a bc 2 2 b bc

where f represents the free-energy density of the isotropic phase without local #uctuations. Only
0
two new expansion parameters ¸ and ¸ , which are closely related to the elastic constants, have
1 2
appeared. Eq. (4.48) is the basic equation of the generalized Landau}de Gennes (GLDG) theory of
the NI transition that includes long-wavelength #uctuations of the order parameter. The GLDG
theory has various applications, of which the most important ones are the pretransitional e!ects,
for example, the magnetically induced birefringence (Cotton}Mouton e!ect) and light scattering.
The important quantities of interest are the correlation functions relevant to the pretransitional
phenomena and if these correlation functions are not too large the Gaussian approximation is
expected to be a good approximation. Concentrating only on pretransitional light scattering and
assuming B"C"2"0, it follows directly from Eq. (4.48) that the local #uctuations in the
isotropic phase give rise to a local free-energy density of the type
f (r)"f #1a(¹!¹H )Q (r)Q (r)#1¸ [R Q (r)]
0 2 NI ab ba 2 1 a bc
][R Q (r)]#1¸ [R Q (r)][R Q (r)] , (4.49)
a bc 2 2 a ac b bc
 S. Singh / Physics Reports 324 (2000) 107}269 149

where Q (r) can be expanded in a Fourier series,

ab

Q (r)"+ Q (k)e*k > r (4.50a)

ab ab
k
with

P
1
Q (k)" dr Q (r)e~*k > r . (4.50b)
ab < ab

Substituting Eq. (4.50a) into Eq. (4.49) and integrating over the volume of the system, one obtains
the following contribution of the #uctuation to the total free energy:
1 1
f " <a(¹!¹H )+ QH (k)Q (k)# <¸ + k2QH (k)Q (k)
& 2 NI ab ab 2 1 ab ab
k k
1
# <¸ + k k QH (k)Q (k) , (4.51)
2 2 a b ac bc
k
where k2"k k . It is obvious that the value of f strongly depends on the values of the amplitude
a a &
Q (k). The validity of LDG theory here is based on the postulate that the appearance of a given set
ab
of amplitudes MQ (k)N is described by a probability distribution of the form
ab
PMQ (k)N"1exp(!bf ) . (4.52)
ab 2 &
For kO0, the amplitudes are complex quantities,
Q (k)"r (k)#is (k) , (4.53)
ab ab ab
where r (k) and s (k) are real variables. These variables satisfy the relations
ab ab
r (k)"r (!k), s (k)"!s (!k) ,
ab ab ab ab
r (k)"r (k), s (k)"s (k) .
ab ba ab ba
Since the trace of the #uctuation tensor Q(k) is zero,
r (k)"s (k)"0 .
aa aa
For obvious reasons the relevant thermal averages based upon the tensor elements Q (0) should be
ab
calculated. These averages are obtained [95] in the form
k ¹
SQ2 (0)T"SQ2 (0)T"SQ2 (0)T" B , (4.54a)
xy xz yz 2<a(¹!¹H )
NI
4 4
SQ2 (0)T"SQ2 (0)T"SQ2 (0)T" SQ2 (0)T" SQ2 (0)T
xx yy zz 3 xy 3 xz
4 2k ¹
" SQ2 (0)T" B . (4.54b)
3 yz 3<a(¹!¹H )
NI
150 S. Singh / Physics Reports 324 (2000) 107}269

The thermal averages Sr2 (k)T and Ss2 (k)T are obtained [95] as
ab ab
Sr2 (k)T"Ss2 (k)T , (4.54c)
ab ab
k ¹
Sr2 (k)T"Sr2 (k)T"Sr2 (k)T" B (4.54d)
xx yy zz 3<[a(¹!¹H )#¸ k2]
NI 1
and

k ¹
Sr2 (k)T"Sr2 (k)T"Sr2 (k)T" B . (4.54e)
xy xz yz 4<[a(¹!¹H )#¸ k2]
NI 1
In these equations the elastic constant ¸ has been taken to be zero. The correlation functions of
2
the kind SQ (0)Q (R)T and SQ (0)Q (R)T, which are of prime interest, can now be calculated in
xx xx xy xy
a straightforward way. The disatance dependence of these functions is described in terms of the
correlation length m, for example,

k ¹
SQ(0)Q(R)T" B exp(!R/m) (4.55)
xx xx 6p¸ R
1
with m given by

C D
¸ 1@2
m" 1 . (4.56)
a(¹!¹H )
NI
The correlation length m is a measure of the distance over which the local #uctuations are
correlated. mP0 at in"nitely high temperature and diverges as ¹P¹H . Near ¹"¹H this
NI NI
divergent behaviour of m is responsible for the so-called pretransitional phenomena, for example,
the strong increase in the light scattering cross-section. With known values of the thermal
averages, the depolarization ratio, i.e., the ratio between the intensities of the scattered light with
polarizations parallel and perpendicular to the incident light at a scattering wave vector q"0, is
given by
I (0) SQ2 (0)T 4
@@ " xx " . (4.57)
I (0) SQ2 (0)T 3
M xy
The calculation becomes more complicated when the ¸ term is also included in evaluating thermal
2
averages like Sr (k)r (k)T. In this case three correlation lengths appear
ab cd

C D
¸ 1@2
m " 1 , (4.58a)
1 a(¹!¹H )
NI

C D
¸ #1¸ 1@2
m " 1 2 2 , (4.58b)
2 a(¹!¹H )
NI

C D
¸ #2¸ 1@2
m " 1 3 2 , (4.58c)
3 a(¹!¹H )
NI
 S. Singh / Physics Reports 324 (2000) 107}269 151

Owing to the positive sign of the correlation lengths, ¸ and ¸ must satisfy the relation
1 2
¸ #2¸ '0 . (4.59)
1 3 2
Thus, only the largest distance makes sense. This means if ¸ (0, m is the correlation length
2 1
whereas m is the correlation length if ¸ '0. It is important to mention that the discussion
3 2
presented here is not expected to present a quantitatively correct description of pretransitional
phenomena. A comparison with experiment and a detailed discussion on the results within and
beyond the Gaussian approximation are presented elsewhere [64].

4.2. The hard particle or Onsager-type theories

The counterpart of phenomenological theory (Section 4.1) for liquid crystals is the molecular
statistical description. A number of papers [96}145] have appeared on the statistical models of the
nematic phase. These are based on the mean-"eld (MF) approximation in which each molecule is
supposed to be subjected to an orienting "eld due to its interaction with all the other molecules in
the medium. As mentioned earlier there are three well-known approaches to tackle the problem.
The "rst one is due to Onsager [97] and ascribes the origin of nematic ordering to steric hindering
between hard rods of anisotropic shape. The second approach, formulated by Maier and Saupe
[117], states that the nematic ordering essentially originates from the anisotropic attractive
interactions. In the third approach, the van der Waals type theories [131}139], both the anisot-
ropic hard core repulsions and angle-dependent attractions are explicitly incorporated in evaluat-
ing the molecular "eld. In this section, we shall concentrate on the "rst approach and the remaining
two will be discussed in Sections 4.3 and 4.4.
About 50 years ago, Onsager [97] showed that a system of long rigid rods exhibits a transition
from an isotropic phase to a denser anisotropic phase. The calculation of Onsager was based on
a cluster expansion for the free energy given as a functional of the distribution in orientation of the
rigid particles. The model was applied later by Zwanzing [98] to a cetain idealized model and the
rigid rod transition was veri"ed to the order of seventh virial. A PadeH analysis of Zwanzing
approach by Runnels and Colvin [100] has shown that the transition observed by Zwanzing is very
stable, and in three dimensions, at least "rst-order. A slightly di!erent mean-"eld calculation for
the hard rod problem based on the well-known lattice model was made by Flory and coworkers
[101,102]. The above treatments [97}102] of hard rod systems are valid only for very long rods
with length to width ratio x 5100 which is typical of polymeric systems. For the shorter rods
0
(x K3}5 or at most x (10) at high densities applying the scaled particle theory (SPT) due to
0 0
Cotter [103,104] has been very convenient for evaluating the excess free energy. Other treatments
of hard rod systems include y-variable expansion [105,106], application of density functional
theory (DFT) [49], the functional scaling appoach [110], Monte Carlo (MC) simulation
[54,147}160], the role of molecular #exibility on the density change at the transition [111}113],
etc. The works based on DFT and MC simulations will be discussed in Sections 4.5 and 9,
respectively.
The Onsager line of approach boils down to a discussion of the thermodynamic properties of
a system of hard rods. Let us consider a #uid of N rods in a volume < at temperature ¹. Assuming
that a rod can take only l discrete orientations, the con"gurational partition function can be
152 S. Singh / Physics Reports 324 (2000) 107}269

written [104] as

P
1 1
Q " + dr exp[!b; ] , (4.60)
N N! lN X N

where ; , the potential energy of interaction, is aproximated as the sum of pair potentials:
N

; " + u . (4.61)
N ij
1:i:j:N
Here N! allows for the indistinguishability of the particles and lN for the number of orientational
states.
For hard particles,

G
R if i and j overlap ,
u "u(r , X , X )" (4.62)
ij ij i j 0 otherwise .
Since the con"gurational space available for each molecule is l

Pdr exp(!b; )"<Nexp[!b/ (X)] ,

N N
(4.63)

where the function / depends only on the orientations. For a given con"guration of the system, if
N
there are N molecules along X , N molecules along X ,2, N molecules along X , / becomes
1 1 2 2 l l N
a function of occupation number and

A B
<N N N N N!
Q " + + 2 + exp[!b/ (N , N ,2, N )] (4.64)
N N!lN <l N ! N 1 2 l
N1/0 N2/0 Nl/0 a/1 a
with +l N "N. Introducing the mole fraction s "N /N of various components and using the
a/1 a i i
maximum term approximation, the con"gurational Helmholtz free energy of a system of hard rods
can be expressed as
bA 1 l b
)3"! ln Q "ln o!1# + s8 ln(s8 )# / (NI , NI ,2, NI ) , (4.65)
N N N i i N N 1 2 l
i/1
where the tilde denotes that the distribution corresponds to the maximum term of / and
N
/ (NI , NI ,2, NI ), as "rst noted by Zwanzing [98], is the excess Helmholtz free-energy relative to
N 1 2 l
an ideal gas of a system of molecules having "xed orientations. Considering the continuous
distribution of angles, replacing s by f (X ) and "nally converting from sums back to integrals, one
i i
obtains [104]

P
bA b
)3"ln o!1# f (X)ln[4pf (X)] dX# / M f (X)N . (4.66)
N N N
Thus, the major work is to derive the excess free-energy function / . The two most common
N
approaches which has been used to evaluate / are the cluster or virial expansion technique of
N
Onsager [97] and the scaled particle theory [103,104].
 S. Singh / Physics Reports 324 (2000) 107}269 153

Onsager [97] made a virial expansion of / and retained terms only upto the second virial
N
coe$cient, i.e., / was approximated as
N
1
/ K o+ + s s < (X , X ) , (4.67)
N 2 i j %9# i j
i j
where < (X , X ) (,< (X )) is the mutual exclusion volume (or covolume) of two rods with
%9# i j %9# ij
orientations X and X , respectively. Converting from sum to integrats, Eq. (4.66) reads
i j

P P
bA 1
)3"ln o!1# f (X)ln[4pf (X)] dX# o f (X ) f (X )< (X ) dX dX . (4.68)
N 2 1 2 %9# 12 1 z

In order to calculate < (X ) the shape of the rods must be speci"ed. For the long rods (¸<d ),
%9# 12 0
where the end e!ect is ignored, one obtains [3]
< (X )"2¸2d Dsin X D , (4.69)
%9# 12 0 12
where (X ) is the angle between the axes of two molecules with orientations X and X . The most
12 1 2
convient shape of rods, used by Onsager [97] and many others [103,104,133], is a spherocylinder,
which is straight circular cylindrical capped on each end by a hemisphere of the same radius. In this
case [98,133].
< (X , X )"8v #4al2 Dsin X D , (4.70)
%9# 1 2 0 12
where a, l and v are, respectively, the radius, cylindrical length and volume of spherocylinder.
0
The singlet orientational distribution function can be determined by minimizing Eq. (4.68)
subject to the condition.

P f (X) dX"1 . (4.71)

Thus, f (X) can be obtained as a solution of the variational equation

1 0 2 2 P
ln[4pf (X )]"j!1!8v o!4al2o f (X ) dX Dsin X D ,
12
(4.72)

where j is a Lagrange multiplier to be determined from the normalization condition (4.71). Eq.
(4.72) is a non-linear integral equation which has to be solved numerically for f (X). However, it is
di$cult to solve Eq. (4.72) exactly. Onsager obtained an approximate variational solution which is
based on a trial function of the form
f "(const) cosh(a cos h) . (4.73)
1
Here a is a variational parameter and h is the angle between the molecular axis and the nematic
axis. It was found [97] that in the region of interest a is large (&20) and the system exhibits an
abrupt "rst-order phase transition from the isotropic (a"20) to the nematic (a518.6) phase
characterized by
/ "4.5d /¸, / "3.3d /¸, PM K0.84 , (4.74)
/%. 0 *40 0 2NI
154 S. Singh / Physics Reports 324 (2000) 107}269

where / ("1po¸d2) represents the volume fraction of the rods. The relative density change
4 0
*o/o , with *o"o !o , at the transition, in the full Onsager calculation, is about 25%. This
/%. /%. *40
value is much too high as compared to the experimental results. The transition predicted by
Onsager's approach was con"rmed later by Zwanzing [98], who calculated higher virial coe$-
cients upto the seventh restricting the molecules to take only three mutually perpendicular
orientations.
Clearly, Onsager's approach only provides a qualitative insight into the e!ect of repulsive
interactions as far as real nematics are concerned. A detailed discussion on the validity of this
approach is given elsewhere [115]. It is claimed that it gives a single and qualitatively correct
picture of the order}disorder transition due to the shape of the molecules. Straley [116] has argued
that the truncation of the cluster expansion series after the second virial coe$cient can be justi"ed
quantitatively only for very long rods (x '100). For shorter rods (x 440) qualitatively reliable
0 0
results are expected. The numerical solutions obtained by Lasher [99] in the limit of very long rods
show that the NI transition is characterized by *o/o K0.21 and PM K0.784.
/%. 2NI
Flory and Ronca [102] considered the application of a lattice model to treat the hard rod
problem and made the calculations, in the limit of long rods, at relatively high densities. The lattice
models have a few attractive features. They can be used with advantage for the evaluation of the
combinatory part of the con"gurational partition function, i.e., the steric factor which compre-
hends the spatial con"gurations that conform to the virtual requirement that the overlap be
avoided. The partition function can be calculated exactly when all the rods are parallel (PM "1).
2
The approximate form of the free energy derived for PM (1 becomes exact for PM "1. Although
2 2
the approach is useful for a dense, highly ordered phase, it treats the angular function f (X) rather
crudely. Thus, the Onsager and Flory results may supplement each other but neither of them can
be taken to be entirely reliable over the whole density range.
In Flory's lattice approach each rod with parameter ¸/d is constructed to consist of
0
x segments, one segment being accommodated by a cell of the lattice. The preferred axis of a given
0
domain is taken along one of the principal axes of the lattice. Disorder in the orientation of the rods
with respect to the preferred axis of the system is expressed by a parameter y which is determined
by the projection of the rod of parameter ¸/d in a plane perpendicular to the preferred axis. The
0
calculations carried out for the phase equilibrium between the anisotropic (nematic) and the
isotropic phases give higher values for the volume fractions at the transition as compared to
Onsager's results
/ K12.5d /¸, / K8d /¸ . (4.75)
/%. 0 *40 0
The value of PM is not meaningful in view of the crude approximations in f (X). It turns out to be
2NI
larger than the Onsager solution. The combined covolume of the solute species in the isotropic
phase at the coexistence was found to be 7.89, whereas the critical value of the axial ratio for the
coexistence of the two phases in the neat liquid is x K6.417.
0#3*5
Warner [108] presented a theory of nematics which derives much from the Flory approach in
calculating the number of ways of placing hard rods into a limited volume. What is new is that the
steric constraints of the other rods are handled in a way that is consistent for both the component
of rods lying parallel and transverse to the director. The use of a lattice in calculating the
con"gurational freedom of the rod assembly was avoided. The most striking feature of this theory
is that the isotropic state is more stable than that estimated by previous theories. Consequently, the
 S. Singh / Physics Reports 324 (2000) 107}269 155

predicted limiting athermal axial ratio, x , is greater (8.99), and the order parameter and
0#3*5
transition entropy are smaller than in earlier approaches.
From the above discussion it follows that the Onsager [97}99] and Flory [101}107] approaches
are not suitable for studying the NI phase transition in real nematics characterized by x K 3}5 (or
0
at most x (10) and high density. However, these approaches are useful for describing the phase
0
transitions in polymer liquid crystals (see Section 10) and their solutions in suitable solvents. For
real nematics composed of spherocylindrical molecules the scaled particle theory (SPT) [103,104]
has been used to calculate the excess free-energy function / (NI , NI ,2, NI ). The principal
N 1 2 l
quantity in SPT is the work function = (a, j, o) de"ned as the reversible work of adding a scaled
i
spherocylinder of radius aa, cylindrical length jl and "xed orientation X . = is related to the
i i
con"gurational Gibbs free energy through the exact relation
bG
"+ s [ln(s o)#b= (1, 1, o)] . (4.76)
N i i i
i
When the scaling parameters aP0 and jP0, that is, when the scaled particle shrinks to a point, it
is shown [103] that

exp[!b= ]K1!o+ s < (a, j) , (4.77)

i i ij
i
where < (a, j) is the volume excluded to an unscaled molecule j with orientation X by the scaled
ij j
particle i with orientation X at some "xed point. At the opposite extreme when a and j both are
i
very large, = (a, j) approaches the reversible PV work required to create a macroscopic sphe-
i
rocylindrical cavity in the #uid,

C D
4p
lim = " p (aa)2jl# (aa)3 P . (4.78)
i 3
a?=
j?=
Also
4p
< (a, j)" a3(1#a)3#pa2l(1#a)2(1#j)#2al2(1#a)jDsin X D . (4.79)
ij 3 ij

= was calculated by interpolating between these two limit, i.e., between very large and very small
i
values of a and j, and the phase transition was located by equating the Gibbs free energy and
pressure of the ordered and disordered phases. The calculated transition properties are in reason-
able agreement with the experimental data of p-azoxyanisole (PAA) when x K2.5 (see Table 6).
0
Further, it was predicted that as x increases, the transition densities decrease, relative density
0
change increases substantially and the order parameter PM increases slightly. In case of very short
2NI
spherocylinders (x (2) the value of relative density change is less than 1%.
0
A comparison of the SPT equation of state with the results of MC and MD simulations
[148}151] in the isotropic phase shows that for x "2 and 3, SPT overestimates the pressure at the
0
high densities. Savithramma and Madhusudana [136] obtained much better results by extending
the method for hard spheres originally proposed by Andrews [146] to spherocylinders. It is based
on the idea that the reciprocal of the thermodynamics `activitya is simply the probability of being
156 S. Singh / Physics Reports 324 (2000) 107}269

Table 6
A comparison of predicted values of mean density o6("[o #o ]/2) and the relative density change *o/o6 at the
/%. *40
transition

Transition Onsager Zwanzig Flory Cotter and Martire [103] Wulf and de Rocco [111]
quantity [97] [98] [101]

(L/do#1) 5 10 L 10 40
oK 3.9 do/L 1.6 do/L 10 do/L &0.35 &0.21 0.31 0.07
*o
29% 21% 520% 21% 40% 2.3% 22%
o6

able to insert a particle into a system without overlapping with other particles. Another method
which improves the agreement with the simulation results is due to Barboy and Gelbart [105]. It is
based on the expansion of free energy in terms of the `y-variablesa instead of the standard one in
densities using virial coe$cients, de"ned as
y"o/(1!v o) . (4.79b)
0
The expansion coe$cients of yn are related to the virial coe$cients. Mulder and Frenkel [106]
applied the y-expansion technique to study the NI transition in a system of ellipsoidal particles by
restricting the expansion to y2.
Parson [107] derived an expression for the free energy of a hard rod system using `decoupling
approximationa. Assuming that the system interacts through a pair potential u(r, X ), the free
12
energy was derived as

P P
bA 1 n Ru
)3"Sln f (X)T! b dr dX dX f (X ) f (X ) dn@r g(r, X ; n@) . (4.80a)
N 6 1 2 1 2 Rr 12
0
For a system composed of molecules whose only anisotropy is in their shape, the pair potential can
be taken of the special form u(r, X )"u(r/p(r( , X ))"u(r/p), where p is an angle-dependent range
12 12
parameter. In the decoupling approximation g(r, X ) scales as g(r/p). This results in a complete
12
separation between the translational and orientational degrees of freedom. The approximation is
accurate at low density, since gKe~bu, but deviates at higher density. Using the decoupling
approximation, Eq. (4.80a) can be written as

P P
bA 1 n
)3"Sln f (X)T# dX dX f (X ) f (X )< (X ) a(n@) dn@ , (4.80b)
N 2 1 2 1 2 %9# 12
0
where

P
= Ru
a(n)"! dy y3 g(y) (4.80c)
Ry
0
and

P
1
< (X )" dr( p3(r( , X ) . (4.80d)
%9# 12 3 12
 S. Singh / Physics Reports 324 (2000) 107}269 157

Here the translational degrees of freedom appear entirely in the coe$cient a(n). When nPR,
obviously a(n)"1 for the hard particle #uid and one "nds the Onsager result. Using the Berne and
Pechukas [109] expression for p(r, X )
12

C D
(r( ) e( )2#(r( ) e( )2!2s(r( ) e( )(r( ) e( )(e( ) e( ) ~1
p2(r( , X )"d2 1!s 1 2 1 2 1 2 . (4.81)
12 0 1!s2(e( ) e( )2
1 2
Eq. (4.80d) for "xed relative orientation e( ) e( "cos h gives [107]
1 2 12
<M "8(1!s2)~1@2(1!s2 cos2h )1@2 . (4.82)
%9# 12
Here e( and e( are unit vectors along the symmetry axes of the two interacting spheroids, d "2a
1 2 0
and
x2!1
s" 0 . (4.83)
x2#1
0
The decoupling approximation transforms the system into a hard sphere #uid. The calculations
were done for a hard rod #uid using the equation of state for the transformed hard sphere #uid. It
was found that a transition to an orientationally ordered state occurs at a critical packing fraction
g which decreases with x . For x 43.5, g is so high that the system tends to crystallize before the
# 0 0 C
ordered liquid state is reached. Using perturbation theory, the calculations were done for a soft-rod
#uid, where the transformed system of spheres interacts with a potential u(y)"e/ym. When mPR,
this reduces to the hard sphere case. The transition temperature was obtained as a function of
density and molecular shape and the order parameter has been found to obey the scaling law
Q"Q[g(e/kT)m@3]. This scaling property suggested that m"12 for a coe$cient C (Eq. (3.2)) equal
to 4.
Lee [110] introduced a functional scaling concept to study the stability of nematic ordering as
a function of molecular shape anisotropy. The free-energy functional for a system of hard
spherocylinder was constructed from a direct generalization of an analytic equation of state for
a hard spheres under a simple functional scaling via the excluded volume of two hard non-spherical
particles
J(g)PJ(g)1S< (e( ,e( )/v T . (4.84)
8 %9# 1 2 0
This is equivalent to the decoupling approximation [107]. The generalized free energy of a system
of hard spherocylinders in a closed form was obtained as

C A B D
bA g(4!3g) 3 (¸/d )2
)4#"bk (¹)#ln o!1#Sln[4pf (X)]T# 1# 0 SDsin X DT .
N 0 (1!g)2 2p 1#3¸/2d 12
0
(4.85)
For long rods, Eq. (4.85) reduces exactly to Onsager's relation in the low-density limit. Numerical
calculations were performed for a variety of length to diameter ratio x of hard spherocylinders. It
0
was found that as the ratio x increases from one to in"nity, g (x ) increases from 0.740 to 0.907.
0 .!9 0
Further, the discontinuity in the packing fraction at the NI transition becomes smaller as the
molecular shape anisotropy decreases. The theory was extended [110](b) to calculate the transition
158 S. Singh / Physics Reports 324 (2000) 107}269

Table 7
Comparison of the NI transition parameters for hard ellipsoids with the axial ratio x "2.75 and 3.0
0
x Quantities MC Baus, et al. Singh and Singh Mulder and Marko Lee
0
[156,157] [174] [142] Frenkel [106] [177] [110]

2.75 g 0.570 0.512 0.347 0.462 0.518 0.552

/%.
g 0.561 0.501 0.329 0.449 0.517 0.544
*40
*g/g 0.016 0.021 0.052 0.028 0.002 0.014
/%.
PM 0.548 0.532 0.552 0.010 0.517
2
*
3.0 g 0.517 0.484 0.330 0.437 0.444 0.517
/%.
g 0.507 0.472 0.309 0.420 0.443 0.508
*40
*g/g 0.019 0.022 0.063 0.039 0.002 0.017
/%.
PM 0.565 0.547 0.568 0.017 0.533
2
*

properties of hard ellipsoids of revolution. The free-energy expression of a system of N hard

ellipsoids reads

bA g(4!3g)
)%3"bk (¹)#ln o!1#Sln[4pf (X)] T# S(1!s2)~1@2(1!s2cos2 h )1@2T .
N 0 (1!g)2 12

(4.86)

Calculations were done for the thermodynamic parameters at the NI transition for hard ellipsoids
with x "2.75 and 3.0 and the results are compared in Table 7 with those of several other
0
approaches. It can be seen that the functional scaling results agree well with the simulation values
[156,157] than other approaches.
From the above discussions, it is clear that all rigid rod calculations predict too low a value for
the mean density and too large a value for the density discontinuity at the transition to be
applicable to real NI transitions. The experimental value for *o/o for the nematic phase transitions
is of the order of 1%. In fact, there is a basic di$culty with all the hard particle theories. Their
properties are `athermala and entirely density dependent. Thus, for a system of hard particles it
follows that C"R, whereas experimentally C"4 for PAA. Clearly, therefore, proper description
of nematic phase requires inclusion of attractive interactions in theoretical treatments.

4.3. Maier}Saupe (MS) type theories

Maier and Saupe [117] assumed that nematic ordering is caused by the anisotropic part of the
dispersion interaction between molecules. The shape anisotropy of the molecules was ignored
entirely. In accordance with the symmetry of the structure, viz. the cylindrical distribution about
the preferred axis and the absence of polarity, the orientational energy of a molecule i can be
approximated as

u "!u6 <~2PM P (cos h ) , (4.87)

i 2 2 2 i
 S. Singh / Physics Reports 324 (2000) 107}269 159

where h is the angle which the long molecular axis makes with the preferred axis and u6 is taken to
i 2
be a constant independent of pressure, volume and temperature. The <~2 dependence is due to the
dispersion interaction. However, it is well accepted that the exact nature of the interactions need
not be speci"ed for the development of the theory. All that is required to obtain the results of MS
theory is an anisotropic potential with a particular dependence on the molecular orientations
[128]. Using the <~2 dependence, the Helmholtz free energy can be expressed as

P C D
bA 1 1 3
" bu6 <~2PM (PM #1)!ln exp bu6 <~2PM cos2 h d(cos h ) . (4.88)
N 2 2 2 2 2 2 2 i i
0
The consistency relation is obtained by the minimization of the free energy,

A B
RA
"0 , (4.89a)
RPM
2 V,T

A B
RScos2 h T
3PM i !3Scos2 h T#1"0 (4.89b)
2 RPM i
2
and
PM "SP (cos h )T . (4.89c)
2 2 i
The minimization of the free energy occurs at that value of PM which satis"es the consistency
2
relation. The calculations lead to a "rst-order NI transition at
u6
2 "4.541 (4.90a)
k ¹ <2
B NI #
and
PM "0.429 , (4.90b)
2NI
where < is the molar volume of nematic phase at ¹ . The following expression for the volume
# NI
change at ¹ is obtained [117,121]:
NI

A B
RA
*<"!2A (4.91)
R<
T/TNI
or

A BC P G H D
*< <2 1 3
" 2 ln exp buN <~2PM cos2 h d(cos h)
< bu6 PM 2 2 2 2NI
2 2NI 0
!bu6 <~2PM (PM #1) . (4.92)
2 2NI 2NI
If *< is known from the experiment, PM can be determined. However, this method gives only
2NI
a 1}2% change in the values of PM . Because theory contains only one unknown, u6 , the order
2NI 2
parameter PM and the entropy change *R are predicted to be the universal properties of the
2NI NI
nematogen. In addition, the calculated values PM K0.44 and *R /RK0.417 are found to be in
2NI NI
160 S. Singh / Physics Reports 324 (2000) 107}269

agreement with the observed value. Experimentally, PM varies in the range of &0.25}0.5 for
2NI
di!erent compounds [161]. The agreement between theory and experiment can be improved
by extending the MS theory in a variety of ways. These include the addition of terms to the
MS potential function to allow for higher rank interactions [122] and a deviation from
molecular cylindrical symmetry [122] as well as using a MF approximation of higher order
[126]. Marcelja [123] and Luckhurst [162] have analysed the in#uence of the #exible end-chain on
the ordering process. In a Flory-type calculation, Marcelja [123] has included the con"gurational
statistics of end chains and could explain the `odd}evena e!ect of both ¹ and PM as a homolog-
NI 2NI
ous series is ascended. Luckhurst [162] re"ned these calculations for compounds with two rigid
cyanobiphenyl moieties linked by #exible spacers in which the odd}even alternation in ¹ is
NI
about 1003C.
Humphries et al. [122] employed a more general form than Eq. (4.87), for the orientational
pseudo-potential, by adding the fourth Legendre polynomial, as follows:
u "!u6 <~j[PM P (cos h )#j@PM P (cos h )] , (4.93)
i 2 2 2 i 4 4 i
where j and j@ are adjustable parameters to be determined from a comparison with the experiment.
They were able to explain the variation of PM and *R/Nk by a single parameter j@ and obtained
2NI
agreement with the experimental temperature dependence of the order parameter when j"4.
Further, taking account of the deviation from spherical symmetry of the pair spatial correlation
function of the molecules, they modi"ed the orientational pseudo-potential as
u "!u6 <~c[1#dPM ]PM P (cos h) . (4.94)
i 2 2 2 2
Here d denotes the deviation from the spherical symmetry of the spatial correlation function. If
d"0 and c"2 this form immediately reduces to Eq. (4.87). It is important to note that unless d is
equal to 0, this potential is theoretically inconsistent because the solution of the self-consistent
equation does not give minima in the free energy of the system.
The volume dependence of u has been analysed by Cotter [119] in the light of Widom's [118]
i
idea. It has been concluded that the MF model will be thermodynamically consistent if and only if
u J<~m, where m is a number, then Eq. (4.90a) becomes
i
u6
2 "4.541 , (4.95)
k ¹ <m
B NI #
but PM remains unchanged. Thus, the value of PM does not depend critically on the volume
2NI 2NI
dependence of u . In the entire nematic range the variation in PM is usually only of the order of
i 2NI
1}2%. However, the value of the exponent m becomes very important while analysing the in#uence
of pressure on the transition parameter. Pressure studies [58,163,164] for the case of PAA have
shown that the value of C (Eq. (3.2)) is 4 and the thermal range of nematic phase at constant volume
is about 2.5 times that at constant pressure. The PM is almost independent of pressure. The studies
2NI
of pressure-induced mesomorphism in PAA suggest that empirically m"4 which has no theoret-
ical justi"cation. In view of Cotter's argument [119], referred to earlier, m"1, irrespective of the
nature of the pair potential.
There are, however, several unsatisfactory features of the MS theory. In general, MF models
overestimate the strength of the transition. The predicted heat of transition from the nematic to the
 S. Singh / Physics Reports 324 (2000) 107}269 161

isotropic phase is given by

H"¹ [(a/b)*<!PM (< ,¹ )] , (4.96)
NI 2 # NI
where a and b are, respectively, the coe$cients of thermal expansion and isothermal compressibil-
ity of the isotropic phase at ¹ . The theoretical values usually are found to be about 2 or 3 times
NI
higher than the experimental values. Large discrepancies are also found in the values of C and b in
V
the nematic phase. Further, the calculations based on the MF models give (¹ !¹H )/¹ K0.9
NI NI NI
which leads to ¹ !¹H K30}403. The reason for these discrepancies is that the MF method
NI NI
neglects completely the e!ect of short-range order. Making use of the Bethe approximation and of
cluster variation methods, the in#uence of short-range order on the NI transition in the MS model
has been studied by several workers [120,124}127]. These approximations are identical in the
isotropic phase, i.e., they give the same value of ¹H , related to the pretransitional phenomena,
NI
magnetically induced birefringence and the scattering of light by orientational #uctuations. These
phenomena are qualitatively well described in these approximations. In these works, each molecule
is assumed to have c nearest neighbours (c 53), no two nearest neighbours are nearest neigh-
n n
bours to each other. While each such central molecule i is subjected to a potential only due to the
c outer molecules of the cluster, an outer molecule j is also subjected to a mean "eld due to the rest
n
of the medium. In the two-site cluster (TSC) approximation the thermal averages are evaluated
with the two-particle distribution function [125]
1
P (e( , e( )" exp[bJP (e( ) e( )#(c !1)bJs6 (P (e )#P (e ))] (4.97a)
2 i j Z 2 1 2 n 2 1Z 2 2z
12
with

12 P P
Z " de( de( exp[bJP (e( ) e( )#(c !1)bJs6 (P (e )#P (e ))] .
1 2 2 1 2 n 2 1z 2 2z
(4.97b)

Here e( and e( are unit vectors pointing in the direction of the long axis of the molecules, s6 is
1 2
a variational parameter and J the interaction coupling constant.
The internal energy per particle is given by
;"!1c Jp (4.98)
2 n 4
where p , the short-range order parameter, measures the correlation between the orientations of the
4
neighbouring molecules
p "SP (e( , e( )T . (4.99)
4 2 1 2
The calculations yield PM K0.40 and the ratio (¹ !¹H )/¹ K0.05; the exact values depend on
2NI NI NI NI
the lattice type, the number of nearest neighbours. Haegen et al. [130] have extended the
TSC approximation to a &4-particle cluster (FSC) approximation and obtained the result
(¹ !¹H )/¹ K0.04; again the exact value depends on the lattice type.
NI NI NI
Another important origin of the discrepancy between the experimental and MS theoretical
values is the assumption that the molecule is cylindrically symmetric so that it is su$cient to de"ne
one order parameter PM . However, in real nematogens most molecules are lath-shaped and have
2
162 S. Singh / Physics Reports 324 (2000) 107}269

a biaxial character. Hence, two order parameters are required to describe the uniaxial nematic
phase composed of biaxial molecules. The second-order parameter is de"ned as
D"3Ssin2 h cos 2WT . (4.100)
2
The mean-"eld theories have been developed [122,165] by including a term in the potential
function proportional to D, in addition to the usual term. It is obtained that the molecular biaxility
decreases the values of PM and gives a better agreement with the experiment.
2NI
Luckhurst and Zannoni have discussed [128] the question `why is the Maier}Saupe theory of
nematic liquid crystals so successfula?. Their arguments can be summarized as follows. The MS
theory is founded on a MF treatment of long-range contributions to the potential function and
ignores the important short-range forces. The theory has been particularly successful in accounting,
not only for the order}disorder transition, but also for the orientational properties of the nematic
liquid crystal. It is founded on the MF approximation applied to a weak anisotropic pair
interaction arising due to dispersion force and predicts the universal properties of the nematogens.
Despite its qualitative and even semi-quantitative success MS theory has been widely criticized
[119,129]. One of the major criticisms is due to its foundation on the London dispersion forces.
A second attack on the theory is concerned with its complete neglect of anisotropic short-range
repulsive forces. This neglect is completely unjusti"ed as it is accepted that the scalar repulsive
forces are mainly responsible for determining the organization in simple #uids [166]. However, the
results of hard rod theories are in marked contrast with the MS theories. On the one hand, theories
based on the short-range repulsive forces, which are expected to make the dominant contribution
to the potential, give results in poor accord with experiment. On the other hand, however, the MS
theory founded on long-range dispersion forces, provides an excellent description of the orienta-
tional properties of real nematogens. A formal solution to this dilemma is provided by the
molecular "eld theories based on a complete general form of the total potential. This is separated
into a scalar component and an anisotropic part which is then written in terms of the Pople
expansion. The orientational part of the single-particle potential is found to be a series whose "rst
term has the same form as the MS result with u6 containing contributions from all the anisotropic
2
forces. In this respect, the MS theory is equivalent in form to one which includes both long- and
short-range contributions to the anisotropic intermolecular potential. There are several unsatisfac-
tory features of this analysis which are discussed elsewhere [128]. As a solution to these di$culties,
it has been suggested [128] that both the short- and long-range forces are important in determining
the molecular organization in a nematic phase but that they operate at quite di!erent levels. The
short-range forces are responsible for the formation of highly ordered groups or clusters of
molecules. The possibility of cluster formation has already been mentioned in the original MS
theory [117] although with di!erent purpose. In view of Luckhurst and Zannoni [128] the net
e!ect of the existence of clusters would be to reduce the anisotropy associated with the short-range
forces while increasing the in#uence of long-range forces. Further, the clusters are not destroyed at
the NI phase transition. The anisotropic forces between clusters are, therefore, responsible for the
orientational properties of the nematic mesophase. The parameter u6 in the e!ective potential then
2
is not determined by the interaction of two molecules, but by two clusters. From this idea the
failure of theories based on the repulsive forces can also be understood. The fundamental unit in
these calculations is a single particle and at the transition the orientational order is completely
destroyed producing a large increase in entropy. Contrary to it for the cluster model the transition
 S. Singh / Physics Reports 324 (2000) 107}269 163

unlocks the ordering of the clusters but not that of the molecules within the cluster; consequently,
the transition entropy is small. In the language of statistical mechanics the cluster can be described
in terms of the spatial and orientational pair distribution function. Then the present idea [128] is
equivalent to the assumption that the distribution function exhibits short-range order, over several
molecular distances, and this order is not destroyed at the NI transition. As a result, the variation in
the orientational properties on passing from one phase to the other will be determined by the
long-range intermolecular forces corresponding to distances over which the distribution function
does change. Consequently, the MS theory, founded on long-range forces, provides a good
description of nematic liquid crystals, whereas theories based on short-range forces are invariable
less successful.
As is evident from the above discussion, both short- and long-range forces are important in
determining the molecular organization in a nematic liquid crystal. Therefore, a realistic theoretical
model of nematics should be constructed using both the repulsive and attractive interactions
between molecules.

4.4. The van der Waals (vdW) type theories

As discussed earlier, the mesophase molecules possess a strong anisotropy in both intermolecu-
lar repulsions and attractions. Hence, a molecular theory should incorporate both short-range
repulsive and long-range attractive forces. In addition, the real complication which one faces in the
construction of a theory is dealing with both the spatial and angular variables of the molecules
[61]. For the calamitic nematics several attempts have been made to develop theories
[131}145,167}171] in which both parts of the interactions are explicitly included. Most of these,
known as van der Waals (vdW) type theories [131}139,167}171], di!er only in evaluating the
properties of reference system interacting via repulsive force and are almost identical as far as
treating the attractive interaction is concerned. Basic to these works is the recognition that the
predominant factor in determining the liquid crystalline stability is geometric and that the role of
the attractive interactions is, to a "rst approximation, merely to provide a negative, spatially
uniform mean "eld in which the molecules move. While Alben [57] used an orientation-indepen-
dent mean-"eld potential, others have used orientation-dependent mean "elds. The theories
[140}145] based on the density functional approach will be discussed in Section 4.5. In this section,
"rst we present a perturbation method [61], within the mean-"eld approximation, to describe the
equilibrium properties of nematics for a model system composed of non-spherical molecules
interacting via a pair potential having both the repulsive and attractive parts. All the vdW type
theories [131}139] can be derived from this scheme by considering only the "rst-order pertubation
term. We have also discussed the salient features of all these works.
In developing a perturbation theory, one begins by writing the pair potential energy of
interaction, u(x , x ), as a sum of two parts * one part is known as reference potential u(0) and the
i j
other perturbation potential, u(1), i.e.

u(x , x )"u(0)(x , x )#ju(1)(x , x ) . (4.101)

i j i j i j
Here u(0) is chosen to include the rapidly varying short-range repulsive interactions, whereas
u(1) represents the more smoothly varying long-range attractions, j is a perturbation parameter.
164 S. Singh / Physics Reports 324 (2000) 107}269

The con"gurational integral of the system is written as

P P
1
Q " drN dXN exp[!b; (x , x ,2, x )] , (4.102)
N N!(4p)N N 1 2 N
where ; is approximated as the sum of pair potentials (Eq. (4.61)), :drN": f dr :dr 2:dr and
N 1 2 N
:dXN":dX 2:dX . The angular integration in Eq. (4.102) can be approximated to arbitrary
1 N
accuracy by dividing the unit sphere into arbitrary small sections of solid angle *X (n"4p/*X is
the number of discrete orientations) and summing over all possible orientational distributions
MN ,2,N ,2,N N where N is the number of molecules having orientations falling in the pth solid
1 p n p
angle and +n N "1. Thus
p/1 p

P
1 N!(*X)N
Q " + 2+ drNexp[!b; (rN, N ,2, N )] (4.103a)
N N!(4p)N N ! N! N 1 n
N1 Nn 1 2 n

A BC D P
*X N n ~1
K < NK ! drNexp[!b; (rN, NK ,2, NK )] . (4.103b)
4p p N 1 n
p/1
Here the tilde denotes the maximum term values.
Assuming the pairwise additivity of interaction potential (Eq. (4.61)) and dividing the total
potential into two parts, one obtains

A BC D P
*X N n ~1
Q " < N! drNexp[!bu(0)(rN, NM ,2, NM )]
N 4p p N 1 n
p/1
]exp[!bu(p)(rN; NK ,2, NK )] (4.104)
N 1 n
and

P
R ln Q 1 n n
N"! b + + NI NI dr u(p)(r, X , X )g(r, X , X ) , (4.105)
Rj 2< p p{ p p{ p p{
p/1 p{/1
where

A BC D P
<2 *X N n ~1
g(r, X , X )" < N! drNexp[!b; (rN, NK ,2, NK , j)] . (4.106)
p p{ Q 4p p N 1 n
N p{/1
Integration of Eq. (4.105) leads to

P P
b j n n
ln Q "ln Q0 ! dj + NMI + NK d r u(p)(r, X , X )g(r, X , X ) . (4.107)
N N 2< p p{ p p{ p p{
0 p/1 p/1
Choosing a continuous function f (X) to describe the orientational distribution such that
NI "Nf (X ) dX (4.108)
p p p
the Helmholtz free energy can be expressed as

P P P
bA bA(0) 1 1
" # ob dj f (X ) dX f (X ) dX
N N 2 p p p{ p{
0

Pdru(p)(r, X , X )g(r, X , X ) .
p p{ p p{
(4.109)
 S. Singh / Physics Reports 324 (2000) 107}269 165

Writing the expansions

g(r, X , X )"g(0)(r, X , X )#jg(1)(r, X , X )#2 , (4.110a)

p p{ p p{ p p{
A"A(0)#jA(1)#j2A(2)#2 (4.110b)

and inserting these expressions into Eq. (4.109), one obtains on equating the coe$cients of j3 from
both sides

P P P
bA(r) 1
" bo f (X ) dX f (X ) d) dr u(p)(r, X , X )g(r~1)(r, X , X ) , (4.111)
N 2r p p p{ p{ p p{ p p{

where r denotes the order of perturbation. All the zeroth-order terms refer to quantities corre-
sponding to the reference system. Now, if we de"ne the e!ective one-body potential as

P P
1
W(r)(X)" o f (X ) dX dr u(p)(r, X , X )g(r~1)(r, X , X ) . (4.112)
2r p{ p{ p p{ p p{

Eq. (4.109) reads

P C D
bA bA(0) 1 =
" # b f (X ) dX + W(r)(X) . (4.113)
N N 2 p p
r/1
All the vdW-type mean-"eld theories [131}139] can be derived from Eq. (4.113) by considering
only the "rst-order perturbation term which is written as

1 P 2 2 P
W(1)(X )"o f (X ) d) dr u(p)(r, X , X ), g(0)(r, X , X ) ,
1 2 1 2
(4.114)

where g(0)(r, X , X ) is the pair correlation function (PCF) for the reference system. The contribu-
1 2
tion of the reference system A(0) is evaluated in di!erent ways in these works [131}139].
In the derivation of free energy as a function of f (X) by Gelbart and Baron [133] the molecular
shape repulsions are treated within the approximation of scaled particle theory and the long-range
attractions enter directly through a mean-"eld average. Basic to the Gelbart and Baron approach is
the formulation of the e!ective one-body potential as an average over the pair attraction which
excludes all relative positions denied to a pair of molecules because of their anisotropic hard cores.
The Helmholtz free-energy equation reads

P
bA bA 1
" )3# b dX f (X )WM (1) (X ) . (4.115)
N N 2 1 1 GB 1

with

P 2 2P
WM (1) (X )"o f (X ) dX dr exp[!bu (r, X , X )]u (r, X , X ) .
GB 1 )3 1 2 !55. 1 2
(4.116)
166 S. Singh / Physics Reports 324 (2000) 107}269

For A the relation derived, for a system of hard spherocylindrical molecules of length ¸ and
)3
diameter 2a by Cotter [103] using scaled particle theory was used by Gelbart and Baron [133]:

P A B
bA bA 2rv o 1
)3K )4#" f (X)ln[4p f (X)] dX# 0 1# qv o
N N (1!v o)2 2 0
0

P P
dX dX f (X ) f (X ) sin h #Mterm independent of f (X)N ,
1 2 1 2 12
(4.117)

where v "pa2¸#4pa3 is the rod volume, r"aL2/v ,q"4pa3/v , and h is the angle between
0 3 0 3 0 12
two rods. It is important to note that the relation (4.117) was revised by Cotter [132](b).
Using both the isotropic- and orientation-dependent parts of the attractive potential, i.e.,

u (r, X , X )";*40(r)#;!/*40(r, X , X ) , (4.118)

!55 1 2 !55 !55 1 2
it has been shown [133](b) that the main contribution to the angular dependence of the attractive
energy arises from the coupling between u*40 and the hard-rod exculsion. Since exp(!b; ) can
!55 )3
assume only two values 0 and 1, Eq. (4.116) can be written as

GB 1 P
WM (1) (X )"o f (X ) dX
2 2 P X X
m( , )
1 2
dr[u*40(r)#u!/*40(r, X , X )] .
!55 !55 1 2
(4.119)

It is obvious, from Eq. (4.119), that both parts, isotropic as well as anisotropic of the attractive pair
potential contribute to the orientation dependence of the mean-potential WM (1). Thus, even in case of
a spherically symmetric attractive interactions, the e!ective potential felt by a single molecule
would be orientation-dependent. Contrary to it, this would not be true in case the hard cores are
replaced by the hard spheres because then the domain of integration m becomes independent of
X and X and only u!/*40 contributes to the orientation dependence of WM (1).
1 2 !55
Taking the form of attractive pair potential

1
u (r, X , X )"! [C #C cos2 X ] , (4.120)
!55 1 2 r6 *40 !/*40 12

where C and C are the interaction constants and expanding WM (1) as

*40 !/*40
WM (1) (o, X)"A o#A oPM P (cos h)#A oPM P (cos h) , (4.121)
GB 0 2 2 2 4 4 4
Gelbart and Gelbart [133] characterized m(r, a, l) analytically, then evaluated the integrals of Eq.
(4.119) numerically and obtained the coe$cient A 's in terms of C for x (length}breadth
L *40 0
ratio)"1,2,3 and 4.2 and taking for each case C /C "8,50 and 250. They also checked the
*40 !/*40
convergence of the series (4.121). Using Onsager's [97] one-parameter representation of f (X),

f (X)"m cosh(m cos h)/4p sin hm ,

the order parameters were calculated; m was chosen to give PM "0.52, PM "0.13, PM "0.017 for
2 4 6
x "3 and C /C "8.
0 *40 !/*40
 S. Singh / Physics Reports 324 (2000) 107}269 167

Cotter [132] considered the application of the van der Waals approach to a model system of
hard spherocylinders (with cylindrical length l and radius a) subjected to a spatially uniform mean
"eld potential.

WM (1)"!e o!e oPM P (cos h) , (4.122)

#05 0 2 2 2
where e and e are the positive energy parameters. Using the rederived [132] scaled particle
0 2
expression for A ,
)3

A B
bA bA o 3v o
)3, )4#"Sln[4pf (X)]T#ln # 0
N N 1!v o (1!v o)
0 0
[(4#q!1q2)v2o2#2rv o[3!(1!q)v o]SSDsin X DTT]
# 2 0 0 0 12 (4.123)
3(1!v o)2
0
with the additional approximation

P
p 5p
SSDsin X DTT, dX dX f (X ) f (X )Dsin X D" ! PM 2 (4.124)
12 1 2 1 2 12 4 32 2

extensive numerical calculations were carried out [132](b) for the NI transition for the parameter
values x "3, v "230As 3, e /v k"25000k and e /v k"2000k. The results were compared with
0 0 0 0 2 0
the experimental data of PAA (see Table 8). A satisfactory qualitative agreement between theory
[132] and experiment was found; for example, the model system exhibits (i) a "rst-order NI phase
transition, (ii) the temperature dependence of the order parameter (at constant P or constant o) of
roughly the correct shape, (iii) nearly linear plots of ln ¹ versus ln o at constant PM with slope 3.9,
2
(iv) increases in ¹ with increasing pressure of the correct order of magnitude, and (v) large
NI
pretransitional increases in the compressibility, expansivity, and speci"c heat as ¹ is approached
NI
from below. The model could not explain the premonitory phenomena observed experimentally in
the isotropic phase above ¹ . As obvious from the Table 8, no satisfactory quantitative agreement
NI
with experiment was found; the predicted values of the relative density discontinuity, entropy of
transition, slope of the P}¹ coexistence curve are too large and the order parameter at the
transition PM are too large whereas the mean reduced density at the transition is too small.
2NI
Baron and Gelbart [133] applied their GvdW theory to systems with spherocylindrical hard
cores and attractive forces described by Eq. (4.120) and made the calculations using Eq. (4.123) for
A and approximating f (X) by the Onsager representation. The use of Onsager's relation for f (X)
)3
introduces substantial errors and so the quantitative agreement between theory and experiment
cannot be expected. However, it is important to note that the Gvdw approach [133] in which
WM (1) is determined from the model pair potential, is clearly superior in comparison to the use of an
essentially phenomenological pseudo-potential (Eq. (4.122)) by Cotter [132]. Further, despite its
quantitative inadequacies, the van der Waals approach indicates that the anisotropy of the
short-range intermolecular repulsions plays a major role in determining nematic order and
stability and cannot be neglected, even to a "rst approximation. For model system with hard cores
plus attractions, anisotropic hard cores are clearly necessary for explaining the behaviour of
nematogens in even a qualitative manner.
 168
Table 8
Comparison of the NI transition in PAA and in the model systems. g is the nematic packing fraction, v "230 A3 and x "1#l/2a and other symbols
/%. 0 0
have their meanings as de"ned in the text

Quantity Cotter Ypma and Vertogen Savithramma and Singh and Singh Convex PAA
[132](b) [134] Mudhusudana [136] [61,137] peg [168]
x "3.0 model

S. Singh / Physics Reports 324 (2000) 107}269

0
Spherical Nonsperical TSC SPT Andrews Ref [61] TSC Aspect
molecules molecules approx. results method x "1.80 approx ratio
0
a "0 a "0.0168 a "0.0167 x "1.75 results [137](c) 3:1.45:1
2 2 2 0
c "4 x "2.05 x "2.0 (biaxial
n 0 0
c "4 ellipsoid)
n
e"737.8 k

e /v k 25000 k 39000 39000 45000 54155 49650 46058 2830

0 0
* *
e /v k 2000 k 3130 1850 2300 1764 1655 1602 283
2 0
* *
¹ 410.4 408 409 409 409 409 409 409 411 409
NI
PM 0.542 0.437 0.473 0.393 0.455 0.454 0.457 0.411 0.36 0.36}0.4
2NI
g 0.454 0.593 0.596 0.621 0.62 0.62 0.62 0.484 0.523 0.62
/%.
*o/o 0.040 0.0038 0.010 0.0039 0.0057 0.0061 0.0069 0.061 0.012 0.0035
/%.
*R/Nk 0.887 0.542 0.590 0.295 * * 0.488 0.651 0.37 0.218
C(¹ ) 3.90 1.00 3.96 3.96 3.95 3.85 4.0 2.30 4.82 4.0
NI

A B
d¹ 175 19.7 48.6 35.8 30.71 37.2 112 43.0
NI
* *
dp
p/-"!3
K/kbar
 S. Singh / Physics Reports 324 (2000) 107}269 169

A number of computer simulation studies [148}153] on systems of hard spherocylinders with

x "2 and 3 have been reported. A comparison of these results in the isotropic phase with the
0
results of SPT shows that while SPT gives reasonably good values at the low densities, it
overestimates the pressure as the density is increased. Savithramma and Madhusudana [136]
extended the Andrews method [146] to derive the thermodynamic properties of an ensemble of
spherocylinders. They used this extended model to study the NI transition properties and made
calculations in the MF approximation for a system of hard spherocylinders, by using the virial
coe$cients of the isotropic phase as well as for hard spherocylinder superimposed by the attractive
potential of the form (4.122). In the calculation they adjusted the potential parameters so as to get
¹ "409 K and g K0.62, where g"ov is the packing fraction. From these calculations it was
NI /%. 0
found that (i) the calculation based on the extended Andrews model can be carried out upto
x "2.9 while those based on SPT could only be made upto x "2.45, (ii) the trend of di!erent
0 0
transition quantities as a function of x are similar to those as given by SPT, (iii) in case of Andrews
0
model, C"4 for x "2.075, which is substantial improvement over x "1.75 of SPT, and (iv) the
0 0
present calculation leads to reasonably good agreement with the experimental data.
Ypma and Vertogen [134] suggested that a nematic can be looked upon as a normal liquid and
derived an equation of state based on the following point of view. The spatial ordering in the liquid
was considered to be determined by the hard-core repulsion which was, to a "rst approximation,
replaced by a hard-core repulsion. The orientational ordering, resulting due to the e!ect of the
eccentricity of the hard core and the superimposed anisotropic attraction interaction, were treated
in two ways. First a molecular-"eld approximation was applied. Second using a two-particle
orientational distribution function the short range orientational order was investigated. The
molecules were assumed to interact through the van der Waals dispersion forces. Following
Bellemans [170] the free-energy of the system was expressed in terms of the properties of an assembly
of hard spheres. The distance of closest approach between two molecules was expressed as
D(X , X )"D [1!a(X , X )] , (4.125)
1 2 0 1 2
where D is the diameter of an equivalent hard sphere of the reference system and a is de"ned such
0
that the average over all orientations

1 2 P
Sa(X , X )T" dX dX f (X ) f (X )a(X , X )
1 2 1 2 1 2
(4.126)

vanishes in the isotropic phase. Expanding a(X , X ) in terms of Legendre polynomial, and
1 2
truncating the series after the "rst term,
a(X , X )"a P (cos h ) (4.127)
1 2 2 2 12
the con"gurational free energy of the reference hard-core system was expressed as
bA bA 12g(1#1g)
)#" )4! 2 a SP (cos h )T#Sln[4p f (X)]T , (4.128)
N N (1!g)2 2 2 12
where

A B
bA o 3 5
)4"ln # ! . (4.129)
N 1!g 2(1!g)2 2
170 S. Singh / Physics Reports 324 (2000) 107}269

In the model calculations [134] two particular choices of the attractive potential were considered:

u (r, X , X )"!r~6[e #e P (cos h )] (4.130a)

!55 1 2 0 2 2 12
and

u (r, X , X )"![e #e P (cos h )]/< . (4.130b)

!55 1 2 0 2 2 12
In Eq. (4.130b) the interaction constants e and e have been divided by the volume < in order to
0 2
get an extensive expression for the total attractive interaction energy.
Now the total Helmholtz free energy can be written as [134]

P
bA bA 12g(1#1g) 1
" )4#Sln[4p f (X)]T! 2 a PM 2# bo dr g (r)[e (r)#e (r)PM 2] . (4.131)
N N (1!g)2 2 2 2 )4 0 2 2

Thus in the approach of Ypma and Vertogen the mean potential was expressed in terms of
hard-sphere radial distribution function g(r)
)4

YV P 1 1 P
WM (1) "o f (X ) dX dr g (r)[e (r)#e (r)PM P (cos h )] .
)4 0 2 2 2 1
(4.132)

As discussed in Section 4.3, in order to account for the nearest-neighbour correlations between
molecular orientations, at least a two-particle orientational distribution function is required. Using
a cluster variation method, such a function was derived [125] and then Eq. (4.131) is modi"ed as
[134]

P
bA bA 1 bA
" )4# bo dr g (r)e (r)# 03*%/5 , (4.133)
N N 2 )4 0 N

where

bA 1
03*%/5"! c ln Z #(c !1) ln Z (4.134)
N 2 / 12 / 1

with

1 P
Z " dX exp[B (g,¹)s6 P (cos h)] ,
2 2
(4.135a)

P C D
B (g, ¹) c !1
Z " dX dX exp 2 P (cos h )# / B (g, ¹)s6 [P (cos h )#P (cos h )] .
12 1 2 c 2 12 c 2 2 1 2 2
/ /
(4.135b)

Here

P
24g(1#1g)
B (g, ¹)" 2 a !bo dr e (r)g (r) . (4.135c)
2 (1!g)2 2 2 )4
 S. Singh / Physics Reports 324 (2000) 107}269 171

Extensive numerical calculations were done for the model potential (4.130) for a "0 (spherical
2
molecules) as well as a O0 non-spherical molecules. The use of Eq. (4.130b) allows an interesting
2
comparison with the results of Cotter [132](b). Similar trends of the variation of NI transition
properties were observed as those of Cotter's work. As can be seen from Table 8, except for C the
overall agreement of the hard sphere model (a "0) is better than that of Cotter's model. Although
2
Cotter's basic assumption of a spherocylindrical molecular shape seems more realistic than
a spherical shape, it is expected that her model may predict much worse results if the excluded
volume of two spherocylinders is accounted more accurately. Similar results were obtained by the
hard sphere model with a distance-dependent interaction potential (4.130a) which corresponds to
the van der Waals dispersion forces.
The in#uence of a slightly non-spherical molecular shape (a O0) on the NI transition properties
2
was studied by taking the model potential (4.130b) and treating the orientational coordinates in
MF approximation. For a given set of interaction strengths the variation of packing fraction, order
parameter, relative density change, etc., were studied as a function of the eccentricity parameter a .
2
It was found that even small values of a have a strong in#uence on the thermodynamic properties
2
at the transition. With increasing a the steric hindering becomes more e!ective, shifting the phase
2
transition to lower densities with increasing density change and jump of the order parameter.
Further, except for C and (d¹ /dp), results for the transition properties have worsened somewhat
NI
as compared to hard-sphere model, but still are in better agreement with experiment than Cotter's
results for spherocylinders (see Table 8). The extensive calculations were also performed to analyse
the e!ects of short-range orientational correlations on the NI transition. It was found that (i) the
transition temperature, the discontinuity in the order parameter, the density and the entropy
decrease if short-range orientational order is included, (ii) the short-range orientational correla-
tions have strong in#uence on the transition properties, (iii) as the number of nearest neighbour
increases the results become worse in comparison with the MF approximation, and (iv) the overall
agreement with experiment is quite satisfactory with smaller number of nearest neighbours c "3
/
or 4.
The "rst-order perturbation theory (Eqs. (4.113) and (4.114)) was applied by Singh and co-
workers [61,137] to study in detail the NI transition in a model system of hard ellipsoidal
molecules with a superposed attractive potential described by the dispersion and quadrupolar
interactions. The following form for the attractive potential was adopted:

u (r, X , X )"!C r~6!(C r~6#C r~5) P (cos h ) . (4.136)

!55 1 2 *$ !$ !2 2 12
The "rst term represents the isotropic component of the dispersion interaction and the second and
third terms describe, respectively, the anisotropic dispersion and quadrupolar interactions. C , C
*$ !$
and C are their respective constants. The potential (4.136) assumes that r'D(r, X , X ), where
!2 1 2
D(r, X , X ) is the distance of closest approach between two ellipsoids, and depends on the
1 2
intermolecular separation but not on the orientation of the intermolecular vector. It also depends
on the angle between the two molecules only and not on their orientations X and X . This is
1 2
certainly a drastic oversimpli"cation of the intermolecular potential of the real liquid crystals.
However, there are two reasons of our choice of such a simple trial potential. First, such a pair
potential is in accord with the MS theory and, secondly, it provides a convenient means to study
the e!ects of attractive interactions.
172 S. Singh / Physics Reports 324 (2000) 107}269

Taking Eq. (4.136) and the Berne and Pechukas expression [109] for D(r, X , X ) (see Eq. (4.81)),
1 2
the following expression for the e!ective one-body potential was obtained [61,137] as

WM (1)"!e !e PM P (cos h )!e PM P (cos h ) , (4.137)

0 2 2 2 1 4 4 4 1
where

e "( 1 px )CH [A(6)#1(CH /CH )A(6)]gI (g)# 1 (1px )2@3CH A(5)gI (5) , (4.138a)
0 12 0 *$ 0 5 !$ *$ 2 6 10 6 0 !2 2 5
e "( 1 px )CH [A(6)#MA(6)#2(A(6)#A(6))N(CH /CH )gI (g)]
2 12 0 *$ 2 0 7 2 4 !$ *$ 6
#1(1px )2@3CH [A(5)#2(A(5)#A(5))]gI (g) , (4.138b)
26 0 !2 0 7 2 4 5
e "( 1 px )CH [A(6)#M18A(6)#20A(6)N(CH /CH )]gI (g)
4 12 0 *$ 4 35 2 77 4 !$ *$ 6
#1(1px )2@3CH [18 A(5)#20 A(5)]gI (g) (4.138c)
26 0 !2 35 2 77 4 5
with
CH "C /v2, CH "C /v2, CH "C /v2, r*"r/D(X ) , (4.138d)
*$ *$ 0 !$ !$ 0 !2 !2 0 12
and A 's are the constants appearing in the integral
L

P
dr(
I (h )" (4.138e)
n 12 Dn~3(r( ,X )
12
1
" [A(n)#A(n)P (cos h )#A(n)P (cos h )#2] . (4.138f)
Dn~3 0 2 2 12 4 4 12
0
The integrals I (g) are de"ned as
n

P
=
I (o,¹)" dr* rH2~ng(0)(rH) . (4.138g)
n )4
0
Starting from the pressure equation for a system of hard ellipsoids of revolution interacting via
a pair potential u satisfying the relation
0
u (r,X ),u (r,X )"u (r/D(X ))
0 12 )%3 12 )%3 12

G
R, rH(1,
" (4.139)
0, rH'1,
the following expression for the Helmholtz free energy of reference system was derived [61]:
bA0 bA
, )%3
N N
g(4!3g)
"(ln o!1)#Sln[4pf (X)]T# [F (s)!F (s)PM 2!F (s)PM 2] , (4.140)
(1!g)2 0 2 2 4 4
 S. Singh / Physics Reports 324 (2000) 107}269 173

where
F (s)"(1!s2)~1@2(1!1s2! 1 s4!2) , (4.141a)
0 6 40
F (s)"1s2(1!s2)~1@2(1# 3 s2# 5 s4#2) (4.141b)
2 3 14 56
and
F (s)" 1 s4(1!s2)~1@2(1#15s2# 525 s4#2) . (4.141c)
4 35 22 1184
Extensive numerical calculations were done for two model potentials; the "rst one assumes
C "0, and the second C "0. These choices enable us to investigate separately the e!ects of
!2 !$
anisotropic dispersion and quadrupole interactions on the transition properties. From these
calculations it was found that (i) with increasing x the phase transition is shifted to higher
0
temperature, lower density with increasing density change and jump of the order parameter, (ii) the
interaction parameters have a strong in#uence on the thermodynamic properties [137] and the
e!ects are more pronounced in case of quadrupolar interaction [137], (iii) for a given x and ratio
0
of the interaction strength the quadrupole interaction predicts a jump in transition temperature,
the smaller values for g and C but higher values of CH /i,*o/o , PM , *R/Ni and (d¹ /dp) in
*$ /%. 2 NI
comparison with the anisotropic dispersion interaction, (iv) when the PM term is included in the
4
calculation, PM , *o/o and *R/Nk increase whereas g decreases slightly, and (v) the pressure
2 /%.
dependence of the transition properties is in accordance with the experiments; the application of
pressure increases the temperature range of the existence of nematic order.
For the potential model (4.136) with C "0, Singh and Singh [61] carried out the calculations
!2
based on the experimental data of PAA. Setting e "0 the criterion adopted in this calculation for
4
selecting the values of e and e was to adjust ¹ and g to 409 K and 0.62, respectively, which
0 2 NI /%.
are the experimental values for PAA. The results are given in Table 8. It may be seen from the table
that the values reported by Savithramma and Madhusudana [136] and ours [61] are almost
identical and are in quite good agreement with the experimental data of PAA. Singh, Lahiri and
Singh [137] also investigated the in#uence of short-range orientational correlation on the
transition properties by using a two-site cluster (TSC) variation method. Though the results
demonstrate (Table 8) that the short-range orientational order has a strong in#uence on the NI
transition properties, the theoretical values are overall in poor agreement with the experiment as
compared to our earlier work.
Tjipto-Margo and Evans [167] have incorporated the convex peg potential [172] into a vdW
theory of NI phase transition. In the convex peg model, the molecules are envisioned to have a hard
(biaxial) core embedded in a spherically symmetric square well,

G H
R for r3< (X , X )
u(r, X , X )" %9# 1 2 . (4.142)
1 2 !e for r4p
Eq. (4.142) represents repulsion if the cores overlap (reside within the excluded volume < ) and
%9#
attraction if the centre-to-centre distance is less than or equal to the sum of the largest semi-major
axes (p). Thus, the anisotropies in the convex peg potential are derived from both its repulsive and
attractive regions. For formal calculations the hard core was represented as a general convex body
whereas, in the numerical work, by a biaxial ellipsoid [173]. In accordance with the GvdW theories
174 S. Singh / Physics Reports 324 (2000) 107}269

the repulsive interactions were treated to all orders using a resummation technique and the
attractive interactions were incorporated to the lowest order. The thermodynamic system was
considered to be a #uid of hard biaxial ellipsoids with semi-axis lengths a, b and c. The hard
ellipsoid #uid can sustain three stable #uid phases, the uniaxial and biaxial nematic phases and the
isotropic #uid. For the formation of biaxial phase, the a, b and c axes lengths of the biaxial ellipsoid
must closely approximate the geometric mean condition a2"bc and the #uid density must reside
in a small density domain. Tjipto-Margo and Evans [167] did not put these restrictions and
consequently, set the two macroscopic biaxial order parameters to zero in the calculations. Global
phase diagrams were determined and the NI transition properties were calculated (see Table 8).
It was found that (i) the model predicts one uniaxial nematic and two isotropic phases (vapour
and liquid), (ii) the nematic}vapour, the nematic}isotropic liquid, the vapour}isotropic liquid
phases as well as a nematic}vapour}isotropic liquid triple point coexist, (iii) due to the di!ering
strengths of the attractive forces for the prolate and oblate bodies, the prolate}oblate symmetry is
broken in this model. However, in the limit of very high temperature and very high densities this
symmetry is exhibited, (iv) the `"rst-ordernessa of the NI transition increases as temperature
decreases and (v) the NI phase transition properties agree satisfactorily with the experimental data
of PAA.
It is important to mention here that while Ypma and Vertogen [134] have used deformed hard
sphere model and the Bellmann [170] type expansion to calculate the properties of reference
system, Cotter [132] and Savithramma and Madhusudana [136] have used SPT. The latter
authors have also extended a method given by Andrews [146] for hard spheres to calculate the
properties of hard spherocylinders. Singh and Singh [61,137] have described the repulsive interac-
tion by the repulsion between hard ellipsoids of revolution. These authors performed the calcu-
lations by assuming that the pair correlation function g(r, X ) for a #uid of hard ellipsoids of
12
revolution scales as g(r/D(X )) [107]. Though this approximation introduces anisotropy in the
12
pair correlation function and is exact at very low density, it cannot be expected to be correct at
liquid density. In fact, the g of nematic phase must depend on f (X) as well as on r/D(X ). It is
12
di$cult to assess the error introduced by this decoupling approximation but it gives compressibil-
ity factor for the isotropic phase which is in very good agreement with the simulation values. On
the other hand, by construction the `convex peg in a round holea potential underestimates the
e!ects of attractive anisotropy because all the anisotropy is obtained as a result of the hard core
interaction. This model has to be selected in such a way as to "t and to understand the temperature
dependences of second virial coe$cient of simple non-spherical molecules [172]. Further, since
convex peg model has attractive forces in its Hamiltonian, this model system supports a thermo-
dynamically stable vapour phase. Owing to the e!ective anisotropic attractive forces, the "rst-
order character of the NI transition is signi"cantly enhanced as compared to that of a hard uniaxial
body. However, this enhancement is compensated by the biaxiality and the resulting NI transition
becomes weakly "rst-order.
Flory and Ronca [138] extended the lattice model treatment [102] to a system of hard rod
molecules subjected to orientation-dependent mutual attractions. They derived the energy rela-
tions, for a system at constant volume, by considering interactions between pair of segments in
contact, rather than in terms of interactions between entire molecules. A characteristic temperature
¹H was de"ned to measure the intensity of these interactions. The orientational energy of the
system as a whole was basically of the MS type. Numerical calculations were reported with
 S. Singh / Physics Reports 324 (2000) 107}269 175

the following "ndings: (i) steric e!ects of molecular shape asymmetry, embodied in the axial
ratio x , are of foremost importance, (ii) the reduced temperature ¹I "¹ /x ¹H at which
0 NI NI 0
the NI transition takes place in the neat liquid decreases with decrease in x below its athermal
0
limit x "6.417 for ¹I ~1"0, and (iii) both the transition entropy and the orientational heat
0#3*5 NI
capacity C are monotonic through the transition; C diverges at a temperature appreciably above
p p
¹ , where the metastable anisotropic state becomes unstable. The application of lattice model has
NI
also been considered by Dowell [138](c) to examine the question-what happens to the relative
stabilities of isotropic liquid, nematic, smectic A and smectic S $ phases formed by rigid cores and
1 A
partially #exible tails? She assumed site}site hard repulsions, then added soft repulsions and
London dispersion attractions using segmental Lennard}Jones (12-6) potentials, and "nally added
dipolar interactions which include dipole}dipole forces and segmental dipole-induced dipole
forces. The results obtained show that it is not necessary to invoke dipolar forces to have stable
liquid crystal phases and that the dipolar forces increase the stability of these phases in systems
with Lennard}Jones interactions. The dipolar forces also shift the temperature ranges of these
phases closer to ambient. In an other work [138](b) the NI transition properties were calculated for
molecules composed of rigid cores having semi#exible tails and interacting through Lennard}Jones
potential. The calculated values of transition properties are in agreement with the experimental
data of PAA. These works [138] elucidate the importance of realistic intermolecular potentials,
particularly the role of soft repulsions in describing an order}disorder transition between two
phases.
An orientation-averaged pair correlation theory was proposed by Woo and coworkers [171]
which is based on the Bogoliubov}Born}Green}Kirkwood}Yvon (BBGKY) theory and takes
account of the pair spatial correlations arising from the intermolecular attractions and repulsions.
In this theory a Lennard}Jones (12-6) potential was employed but the anisotropy of the pair
correlation function was not considered. Secondly, it is somewhat di$cult to solve the second
BBGKY equation. Nakagawa and Akahane [168] introduced the pair spatial correlation function
approximately into the MF theory in terms of an orientation-averaged pair potential and studied
its e!ects on the NI phase transition in the decoupling approximation [107] which decouples the
orientations and positions. The approximated pair spatial correlation function [172] is much
simpler than Woo's and co-workers [171] but involves the deviation from the spherical symmetry.
The numerical calculations were done for a model potential of Lennard}Jones (12-6) type and
retaining the leading coe$cients of the orientation-averaged pair potential. It was found that the
anisotropy of a pairwise intermolecular potential increases both the long-range orientational order
and the NI transition temperature. The order parameter change near the transition point is in good
agreement with the data for PAA but the transition entropy and C do not agree satisfactorily.
v
The molecular-"eld theories have been reformulated by Vertogen and de Jeu [175] in such a way
that they can be used analytically. Based on the so-called spherical version of the models the
equation of state for nematics was derived as a generalization of the theory of simple liquids [176]
and the calculations were performed in the MF approximation. Starting from this equation of state
for nematics the two well-known approaches, Onsager [97] and Maier-Saupe [117], can be
recovered by a vector of variable length, such that its thermally averaged length is always equal to
one. In view of the mathematical di$culties it is extremely hard to arrive at a reliable equation of
state for a simple model of nematics from the "rst principle. Therefore, only a qualitatively correct
equation of state has been discussed by Vertogen and de Jeu [175]. In complete analogy with the
176 S. Singh / Physics Reports 324 (2000) 107}269

theory of simple liquids [176], the partition function of the model system is given by

T C DU
1
Z"Z exp b + u (r l, X , Xl) , (4.143)
)3 2 !55 i i
iEl
0
where Z denotes the partition function of the hard-rod model and the thermal average is taken
)3
with respect to the hard-rod system. Thus, the calculation of the equation of state contains two
parts: the calculation of the partition function of the hard-rod model and the calculation of the
thermal average for an assumed attractive interaction. The problem thus boils down to the
calculation of

P C D
1 N 1 N
Z " < dr dX exp ! b + u , (4.144)
)3 N!(jq)3N i i 2 l )3
i/1 i, /1
where the constant q is given by

h
q" . (4.145)
[(2p)5@3k ¹I2@3I1@3]1@2
B 1 3
Here I and I are the principal moments of inertia of the cylinder with I about the cylinder axis.
1 3 3
The calculation of the partition function is still an unsolved problem. According to the van der
Waals approximation the integral over the position coordinate (Eq. (4.144)) can be approximated
in a qualitative sense as

C D
1 N 1 N
Z(X , X ,2, X )" < <! + < (X ) . (4.146)
1 2 N N! 2 %9# 12
i/1 jEi
Thus

P
1
Z " Z(X , X ,2, X ) dX dX 2 dX (4.147)
)3 (jq)3N 1 2 N 1 2 N

with

G C DH
<N N 1 N
Z(X , X ,2, X )" exp + ln 1! + < (X ) . (4.148)
1 2 N N! 2< %9# 12
i/1 jEi
In a straightforward manner the van der Waals expression for the Helmholtz free energy per
cylinder is given by

bA /N"3 ln j#3 ln q#ln o!1! ln[1!1oS< (X )T]#S ln f (X)T . (4.149)

)3 2 %9# 12
It is extremely di$cult to evaluate the thermal average over the hard-rod system and so has to be
approximated

T C DU C D
1 1
exp b + u (r l, X , Xl) "exp b + Su (r l, X , Xl)T , (4.150)
2 !55 i i 2 !55 i i 0
iEl 0 iEl
 S. Singh / Physics Reports 324 (2000) 107}269 177

where

P
1
Su (r l, X , Xl)T " dr dr dX dX g (r , X , X )u (r , X , X ) . (4.151)
!55 i i 0 <2 i l i l )3 il i l !55 il i l
In the spirit of vdW theory, Eq. (4.150) can be approximated as

T C DU C D
1 1 1
exp b + u (r l, X , Xl) "exp Nbou6 # Nbou6 SP (cos h )T . (4.152)
2 !55 i i 2 0 2 2 2 12
iEl 0
Starting from the approximate expression
< (X )"A !A P (cos h ) (4.153)
%9# 12 0 1 2 12
for the excluded volume of two rods, the Helmholtz free energy in the spherical version of the model
is obtained [175] as

C D
bA 7 1
"3 ln j#3 ln q#ln o! ! ln 1! o(A !A PM 2)
N 4 2 0 1 2

A B
1 1 1 3 2 1@2
! bou6 ! bou6 PM 2! bPM ! B2PM 2! BPM #1
2 0 2 2 2 4 2 4 2 3 2
2 2p 1
#BPM 2! ln ! ln(1!PM )! ln(1#2PM ) , (4.154)
2 3 3 2 2 2
where the coe$cients A and A are determined by calculating the two simplest excluded volume
0 1
corresponding to two parallel and perpendicular rods [175]. For two parallel rods
A !A "4pd3#2p¸d2 ,
0 1 3 0 0
whereas for the two mutually perpendicular rods
A #1A "4pd3#2p¸d2#2¸2 d .
0 2 1 3 0 0 0
The parameter B is given by
oA
B" 1 #obu6 . (4.155)
1!1o[A !A SP (cos h )T] 2
2 0 1 2 12
Using the thermodynamic relation the equation of state is given by
o
bp" !1bo2u6 !1bo2u6 PM 2 . (4.156)
1!1o(A !A PM 2) 2 0 2 2 2
2 0 1 2
As expected the equation of state (4.156) is found to describe the NI transition rather well from the
qualitative point of view. The experimental curve of order parameter and the values of transition
density can be reproduced fairly well by an appropriate choice of the parameters. However, the
required parameter values do not correlate with the molecular structure. Thus, the derived
equation of state is unable to describe the structure}property relationship in a physically satisfac-
tory way. The discrepancies are not at all surprising in view of the simplicity of the model. Some of
178 S. Singh / Physics Reports 324 (2000) 107}269

these can be attributed to rather poor approximation methods, which overestimate a number of
quantities, e.g., the in#uence of the excluded volume. Another most important source of the
di$culties in interpretation is the poor representation of the intermolecular interactions and the
neglect of molecular #exibility.

4.5. Application of density functional theory (DFT) to the NI phase transition

The density functional theory [37,49,140}145,177}195] is developing as a cost-e!ective proced-

ure for studying the physical properties of non-uniform systems. The theory was pioneered by
Kohn and others [179,180] and Mermin [181] for a quantum mechanical many-body system. Its
initial classical version and development are by Lebowitz and Percus [182] and the present form
by Saam and Ebner [183] and Yang et al. [184]. The general formalism of the classical version of
DFT has played useful role in the development of approximate integral equations for the pair
distribution functions of a uniform #uid [184]. The theory has subsequently been used pro"tably in
the study of freezing of a variety of #uids, solid}melt interfaces, nucleation, liquid crystals,
two-dimensional systems, molten salts, binary mixtures, aperiodic crystals or glasses, quasi-
crystals, the elastic properties of solids and liquid crystals, spectroscopic properties, dislocations
and other topological defects, #exoelectricity in liquid crystals, interfacial phenomena and wetting
transitions, etc. In a review, Singh [49] has given an excellent account of DFT, its application to
some of above development and prospects. The application of the theory to study the elastostatics
in liquid crystals has been given in a self-contained manner by me [37]. Few other articles and
monographs [186}194] are also available. Thus the theory has been clearly discussed several times
in literature and the essentials of the approach are well documented. Therefore, we present [49,142]
here a brief account of the theory relevant to the phase transition studies in liquid crystals.
In the absence of external "eld, the singlet distribution function is written as

1
o(x)" exp[bk#C(x)] , (4.157)
K

where K is the cube of the thermal wavelength associated with a molecule, !k ¹C(x) is the
B
solvent-mediated potential "eld acting at x. The single-particle direct correlation function C(x) is
a functional of o(x) and is related to the Ornstein}Zernike (OZ) direct correlation function by the
relation

dC(x )
1 "C(x , x ) . (4.158)
do(x ) 1 2
2
The excess Helmholtz free energy, de"ned as

b*A"b(A!A ) (4.159)
*$
is also related to C(x) through the relation

d(b*A)
"!C(x) . (4.160)
do(x)
 S. Singh / Physics Reports 324 (2000) 107}269 179

Here bA is the ideal gas part of the reduced Helmholtz free-energy

*$

*$ P
bA " dxo(x)Mln[o(x)K]!1N . (4.161)

These are the starting equations of the density functional theory and have been used to develop
a variety of approximate forms for the free-energy functionals. The functional integration of Eq.
(4.158) from some initial density o (of isotropic liquid) to "nal density o(x) (of the ordered phase)
&
gives

1 1 & 1 2 P
C(x ; Mo(x)N)!C(x ; o )" CI (x , x ; o(x))*o(x ) dx ,
2 2
(4.162)

where

P
1
CI (x , x ; o(x))" da C(x , x ; [o #aM*o(x)N]) (4.163)
1 2 1 2 &
0
and
*o(x )"o(x )!o . (4.164)
i i &
In the above equations, the braces M N indicate the functional dependence of the quantities on the
single-particle distribution function. Parameter a characterizes a path in density space along which
the chemical potential remains constant. From a functional Taylor expansion, it follows that
C(x , x ; [o #aM*o(x)N])
1 2 &

1 2 & 3 1 2 3 &P
"C(x , x ; o )#a dx C(x , x , x ; o )*o(x )*o(x )#2 .
2 3
(4.165)

Combining Eqs. (4.162) and (4.165) we obtain

P P
1
ln[o(x )/o ]" dx C(x , x ; o )*o(x )# dx dx *o(x )C(x , x , x ; o )*o(x ) . (4.166)
1 & 2 1 2 & 2 2 2 3 2 1 2 3 & 3
This is a non-linear equation and relates the single-particle density distribution of the ordered
phase to the direct correlation function of the coexisting liquid.
From the functional Taylor expansion of (4.160) from o to o(x) and using Eqs. (4.162)}(4.165),
&
one obtains

P P
1
b(*A!*A )"! dxC(x; o )*o(x)! dx dx *o(x )C(x , x ; o )*o(x )
& & 2 1 2 1 1 2 & 2

P
1
! dx dx dx *o(x )*o(x )C(x , x , x ; o )*o(x ) . (4.167)
6 1 2 3 1 2 1 2 3 & 3
The grand thermodynamic potential which is generally used to locate the transition is de"ned as

!="bA!bk dx o(x) .
P (4.168)
180 S. Singh / Physics Reports 324 (2000) 107}269

Substituting the value of bA, we obtain correct to second order in *o(x):

& P
*="=!= " dx[o(x) lnMo(x)/o N!*o(x)]
&

P
1
! dx dx *o(x )C(x , x ; o )*o(x )
2 1 2 1 1 2 & 2

P
1
! dx dx dx *o(x )*o(x )C(x , x , x ; o )*o(x ) , (4.169a)
6 1 2 3 1 2 1 2 3 & 3

where = is the grand canonical thermodynamic potential of the isotropic liquid. Combining Eqs.
&
(4.166) and (4.169a) we get for *= an expression which is found to be convenient in many
applications,

P P
1
*="! dx *o(x)# dx dx [o(x )#o ]C(x , x ; o )*o(x )
2 1 2 1 & 1 2 & 2

P
1
# dx dx dx [2o(x )#o ]*o(x )C(x , x , x ; o )*o(x ) . (4.169b)
6 1 2 3 1 & 2 1 2 3 & 3

Eqs. (4.166), (4.169a) and (4.169b) are the basic equations of the theory of freezing and of interfaces
of ordered phase and its melt. Of interest is the solutions of o(x) of Eq. (4.166) having symmetry of
ordered phase. These solutions, inserted into Eqs. (4.169a) and (4.169b), give the grand potential
di!erence between the ordered and liquid phases. The phase with the lowest grand potential is
taken as the stable phase. Phase coexistence occurs at the value of o which makes *="0 for the
&
ordered and liquid phases.
The grand thermodynamic potential is considered to be a functional of the singlet distribution
o(r, X). Eqs. (4.169a) and (4.169b) can be written as
*="=!= "*= #*= (4.170)
& 1 2
with

P
*= 1
1" dr dX[o(r, X)ln(o(r, X)/o )!*o(r, X)] (4.171a)
N o< &
&
and

P
*= 1
2"! dr dX dX *o(r , X )C(r, X , X )*o(r , X ) . (4.171b)
N 2o < 1 2 1 1 1 2 2 2
&
The density of the ordered phase can be obtained by minimizing *= with respect to arbitrary
variation in the ordered phase density subject to the constraint that there is one molecule per lattice
site (for perfect crystal) and/or orientational distribution is normalized to unity. Thus

1 1 & L P
ln[o(r , X )/o ]"j # dr dX C(r , X , X ; o )*o(r , X ) ,
2 2 1 1 2 & 2 2
(4.172)
 S. Singh / Physics Reports 324 (2000) 107}269 181

where j is a Lagrange multiplier which appears in the equation because of constraint imposed on
L
the minimization. For locating transition one attempts to "nd the solution of o(r, X) of Eq. (4.172)
which has symmetry of the ordered phase. Below a certain liquid density, say o@, the only solution is
o(r, X)"o . Above o@ a new solution of o(r, X) can be obtained which corresponds to the ordered
&
phase. The phase with lowest grand potential is taken as the stable phase. The transition point is
determined by the condition *="0. Implicit in this approach is an assumption according to
which system is either entirely liquid or entirely ordered phase, no phase coexistence is permitted.
This signi"es the mean-"eld character of the theory.
For axially symmetric molecules, the singlet density of the nematic phase can be expressed as
o(r, X)"o f (X) (4.173)
/
with
o "o (1#*oH) (4.174)
/ &
and

f (X)"1# +@ (2l#1)PM lPl(cos h) , (4.175)

l52

where *oH"(o !o )/o is the relative change in the density at the transition and o be the number
/ & & /
density of the nematic phase. For a uniaxial phase of cylindrically symmetric molecules, f (X)
depends only on the angle h between the director and the molecular symmetry axis. The orienta-
tional singlet distribution is normalized to unity (Eq. (2.7)).
In case of the nematic phase, it is convenient to use the following ansatz for f (X):

0 lC
f (X)"A exp + j Pl(cos h) .
l D (4.176)

A is determined from the normalization condition (2.7). When j P0, f (X)P1 corresponding to
0 l
the isotropic phase.
The entropy term in Eq. (4.171a) can be reduced to the form

GC D H
*= oA
1"!*oH#(o /o ) ln / 0 # +@ j Pl(cos h) . (4.177)
N / & o l
& l
z2
The interaction term *= /N is evaluated using Eq. (4.176) for f (X),
2
*= 1 (1#*oH)2
2"! *oH@CK 0 ! +2 PM l1PM l2CK l(0) (4.178)
N 2 00 2 1l2
1, 2z2
l l

l l is the structural parameter. Expanding C(r, X , X ) in spherical harmonics in

where CK (0)
1, 2 1 2
space-"xed frame

C(r, X , X )" + + Cl1l2l(r)C (l l l; m m m)>l1 1(X )>l2 2(X )>H (r( ) , (4.179)
1 2 ' 1 2 1 2 m 1 m 2 l.
1 2 m1m2m
l l l
182 S. Singh / Physics Reports 324 (2000) 107}269

from Eq. (4.172) one "nds the expression for the order parameter as [142]

P 1
L
fl l
1 2
1C
(1#*oH)dl #PM l" dX Pl(cos h )exp j #o + [(2l #1)(2l #1)]1@2
0 1 2

' 1 2 1 P
]C (l l 0; 000)Pl(cos h )PM l dr r2Cl1l2 (r) .
0 D (4.180)

In order to "nd the solutions of above equations, the values of Cl1l2 e(r) are required. Cl1l2 e(r) can be
obtained by solving the integral equations [142](c}e).

4.5.1. Modixed weighted-density approximation (MWDA)

The theory of weighted density approximation (WDA) was originally proposed by Nordholm
[196] and Tarazona [197], later re"ned by Curtin and Ashcroft [198], and "nally modi"ed
(MWDA) by Denton and Ashcroft [199].
The Helmholtz free energy of an inhomogeneous system can be expressed as the sum of two
contributions: (i) an ideal gas part A (known exactly), and (ii) an excess contribution A due to
*$ %9
interaction between the particles,

A[o(r, X)]"A [o(r, X)]#A [o(r, X)] , (4.181)

*$ %9
where both terms in Eq. (4.181) are unique functional [186] of the one-particle density o(r, X). The
"rst term is given by

A [o(r, X)]"b~1
*$ PV
dr dX o(r, X)Mln[o(r, X)K]!1N . (4.182)

The second term of Eq. (4.181) is the excess Helmholtz free energy of the non-uniform system.
Unlike the ideal gas contribution, this term is not known exactly, and is the focus of attention in the
approximations presented by density functional theories. In the MWDA the excess free energy of
the non-uniform system is approximated by the excess free energy of a uniform system, but
evaluated at a weighted density o( :

A [o(r, X)]"AMWDA[o(r, X)]"Nf (o( ) , (4.183)

%9 %9 0
where f (o( ) is the excess free energy per particle of a uniform system at density o( . The weighted
0
density is de"ned by [199]

P P
1
o( " dr dX o(r , X ) dr dX o(r , X )w(r , r , X , X , o( ) (4.184)
N 1 1 1 1 2 2 2 2 1 2 1 2

with the constraint that the weight function w(r , r , X , X , o( ) must satisfy the normalization
1 2 1 2
condition

Pdr dr dX dX w(r , r , X , X ; o( )"1 .

1 2 1 2 1 2 1 2
(4.185)
 S. Singh / Physics Reports 324 (2000) 107}269 183

For the molecular #uid, the explicit form for w can be written as

C D
1 1
w(r , r , X , X ; o( )"! b~1C(r , r , X , X ; o( )# o( f A (o( ) . (4.186)
1 2 1 2 2f @ (o( ) 1 2 1 2 < 0
0
The primes on f (o( ) indicate derivative with respect to density. For the nematic phase, one obtains
0
[142]

C D
px d3
o( "o 1! 0 0 (PM 2CK (0)#PM 2CK (0)) . (4.187)
/ 12g( bf @ (o( ) 2 22 4 44
0
Having computed o( , the next step is to substitute o( into Eq. (4.183) to calculate AMWDA. Using Eqs.
%9
(4.173) and (4.176), the non-uniform ideal gas contribution A [o(r, X)] becomes
*$

*$ / C
bA [o(r, X)]"<o ln(A o K)!1#+ j PM l .
0 / l
e D (4.188)

The total free energy of the nematic phase is now written as

0 / C
bA"bNf (o( )#<o ln(A o K)!1#+ j PM l .
0 / l
e D (4.189)

4.5.2. Calculations and results

The usual approach to the above theory is to "rst obtain some approximation to evaluate the
DPCF, C(r, X , X ), appearing in Eq. (4.170). Next to choose a parametrization of the one-body
1 2
density that is appropriate for the kind of expected phase transitions. With these two requirements
speci"ed, the functional (4.170) is completely de"ned and the calculation proceeds by the minimiz-
ation of *= with respect to the parameters in o(r, X). When a con"guration other than the
isotropic liquid satis"es *="0, the ordered and isotropic phases coexist and all the transition
properties can be identi"ed.
The ordered phase (translationally and/or orientationally) coexists with the isotropic liquid
when

A B
R
(*=/N)"0 (4.190a)
Rm
i
and
*=/N"0 , (4.190b)
where m are variational parameters appropriate for the phase under investigation. Eqs. (4.190a)
i
and (4.190b) show a stability condition and a phase coexistence condition. The DFT calculations
are done by minimizing Eq. (4.170) with respect to the variational parameters o ,j and j . (Eq.
/
(4.172)). The coexistence point is then located by varying o with *=/N"0. In2 MWDA 4 the
&
e!ective density o( is computed [142](f, g) "rst and then the free energy (Eq. (4.189)) is minimized
with respect to o , j and j . The transition densities of the coexisting phases are determined by
/ 2
equating the pressure and 4chemical potential of the two phases.
184 S. Singh / Physics Reports 324 (2000) 107}269

The structural parameters for the nematic phase are de"ned as [142]

1 2 1 2 & P
l "(2l #1)(2l #1)o dr dX dX C(r, X , X )Pl (cos h )Pl (cos h ) .
CK l(0)
1 2 1 2 1 1 2 2
(4.191)

Using Eq. (4.179) and after simpli"cation, Eq. (4.191) can be written as

C D P
(2l #1)(2l #1) 1@2
CK (0)
l l " 1 2 o C (l l 0,000) dr r2Cl1l2 (r) . (4.192)
1 2 4p & ' 1 2 0

The theory was successfully applied to the study of crystallization of particles interacting via
spherical pair potentials including the case of hard spheres [200}203]. Singh and Singh [142] have
pioneered the study of transitions to the plastic and nematic states, within decoupling approxima-
tion [107], in the hard ellipsoidal system using this theory. The calculation for isotropic}plastic
transition was performed by assuming the crystalline lattice of the plastic phase to be fcc with
lattice parameter a determined self-consistently by the relation

a"(4/o )1@3, o "o (1#*oH) .

0 0 &
It was predicted [142] that the equilibrium positional freezing (plastic) on fcc lattice takes place for
the value of CKM 1 ["1!1/S(DG D), where S(DG D) is the "rst peak in the structure factor of the centre
0.0 . .
of mass] K0.67 or S(G )K3.07. The equilibrium orientational freezing (nematic) takes place when
.
the orientational correlation CMK 0 K4.45. Further, it was found [142] that the plastic phase
2.2
stabilizes "rst for 0.574x 41.75 and the nematic phase for x (0.57 and x '1.75. Though
0 0 0
these values are in reasonable agreement with simulation results [156], this work [142](b) su!ers
with some uncertainties, the most important of which is the lack of information about C(r, X , X ),
1 2
which even in the case of the isotropic state is a function of four independent variables. This forces
the use of what is essentially a guess for C(r, X , X ). Other problems with the initial application of
1 2
the density functional approach are that the oriented crystal phase was not considered and that the
parametrization of the plastic phase was done so that the rather narrow real-space lattice peaks
associated with hard-core crystallization could not be generated. In an attempt to remedy some of
these problems, Marko [177] applied DFT to study the properties of plastic phase at the
isotropic}plastic transition, of oriented solid phase at the isotropic}solid transition, and of nematic
phase at the isotropic}nematic transition in a #uid of prolate hard ellipsoids. In this study [177] the
direct correlations were computed with a variational technique based on the Percus}Yevick (PY)
equation for the correlation functions. Trial correlation functions are chosen and then the optimal
solution is obtained using the PY equation. The calculation procedure involves the direct calcu-
lation of the excess grand potential for the ordered phase, followed by the resulting non-linear
functional. The "rst step is to specify the one-body distribution in terms of as few parameters as
possible which can describe the ordered phase reasonably. For the plastic phase, the trial one-body
density is taken as a sum of Gaussian distributions centred on real-space lattice sites and is
described by three parameters

o(x)"o D(pp2l )~3@2 + exp[!p~2

l (r!ct)2] , (4.193)
l
t|T&
 S. Singh / Physics Reports 324 (2000) 107}269 185

where o is the ordered phase density, pl is the lattice site distribution width, c is the lattice
l
constant and D is the volume per lattice site. In case of fcc lattice *"c3/J2. ¹ is the set of fcc
&
lattice vectors with unit nearest-neighbour spacing.
The trial density for the nematic was taken as

CP D
1 ~1
o(x)"o exp[a (e( ) z( )2#a (e( ) z( )4] dx@ exp(a x@2#a x@4)
, (4.194)
1 2 1 2
0
l

where a and a parametrize the orientational distribution function. The trial DCF and pair
1 2
distribution were assumed to be of the form
C(r, X , X )"C (rH((1#aP (cos h )) (4.195a)
1 2 0 2 12
and
g(r, X , X )"g (rH)(1#aP (cos h )) , (4.195b)
1 2 0 2 12
where C and g are, respectively, the hard-sphere DCF and PY correlation function [204], and
0 0
rH"r /D(X ). The integration of the PY equation squared over the hard-core region yields
12 12
a functional that quanti"es the accuracy of the trial solution, and thus should be minimized with
respect to a to obtain the best solution. The integration over only the core region emphasizes the
DCF rather than the pair distribution, and limits the integration to a "nite region, which is
important numerically. The error functional was de"ned as

P G P H
2
I(a)" dx 1#C(x , 0; a)#o dx C(x , 0; a)[g(x , x ; a)!1] . (4.196)
1 1 2 2 1 2

I(a) was computed numerically for prolate ellipsoids with anisotropies x "1.0, 1.22, 1.53, 2.0 and
0
3.0 and for the packing fractions g"0.1, 0.2,2,0.7 over appropriate ranges of a (g"0.7405 is the
close-packing density). This correction makes the DCF more negative, corresponding to a larger
free-energy cost, for parallel ellipsoidal con"gurations. For perpendicular con"gurations, the
correction makes the DCF less negative, corresponding to a smaller free-energy cost of these
con"gurations.
The DCF thus evaluated were used in free-energy functional (4.170) with the plastic and nematic
distributions (Eqs. (4.193) and (4.194)) and global minima were determined over ranges of liquid
density. The isotropic}plastic, isotropic}solid and isotropic}nematic transition properties were
calculated. It was found that (i) for small anisotropies (x (1.5) the isotropic-#uid to isotropic-
0
crystal densities are una!ected by the small correction. However, the results show an increase in
transition density as the anisotropy increases which is in accordance with MC simulations but
indicates the opposite trend as observed by Singh and Singh [142](b), (ii) for 24x 43, although
0
the isotropic}nematic transition density agrees well with the simulation results [156], the density
discontinuity is too small, and (iii) the results of density functional caculations depend crucially on
the approximations made for the pair DCF. Thus it is feasible to obtain information about the
DCF by the numerical means, even in case of orientation-dependent interactions.
A general method was developed by Perera et al. [205] to solve numerically the hypernetted
chain (HNC) and Percus}Yevick (PY) integral equation theories for #uids of hard non-spherical
 186 S. Singh / Physics Reports 324 (2000) 107}269

Table 9
Comparison of the NI phase transition properties for hard ellipsoids of length-to-breadth ratio x "3.
0
PH"Pv /k ¹, kH"k/k ¹
0 B B
Quantity MC Singh and coworkers Singh and Colet et al. Marko Perera Holyst and Singh
simulation [142](e, f ) Singh [174](b) [177](b) et al. Poniewiers et al.
[156](a) [142](b) [143] ki [210] [145]

DFT/WDA DFT/MWDA

g 0.507 0.466 0.471 0.309 0.72 0.493 0.418 0.454 0.475

*40
g 0.517 0.479 0.490 0.322 0.484 0.494 0.436 0.474 0.494
/%.
*o/o 0.019 0.027 0.041 0.040 0.024 0.002 0.041 0.042 0.038
/%.
PM 0.50 0.745 0.552 0.547 0.560 0.017 0.657 0.485 0.546
2NI
PM 0.474 0.266 0.197 0.358 0.196
4NI
* * * *
PH 9.786 6.429 6.662 * * * * * *
kH 25.150 17.982 18.480 * * * * * *

particles forming nematic phase. The explicit numerical results were given for #uids of hard
ellipsoids of revolution with length-to-width ratio varying from 1.25 to 5.0 as well as for #uids of
hard spherocylinders with length-to-width ratio varying from 2 to 6 [205](a). Theoretical results
were compared with the available MC data [156,148}153] for the equation of state and the pair
correlation function. The DCF for the isotropic phase were obtained. It was shown that the DFT
results are strongly dependent upon the approximation used for the isotropic direct pair correla-
tion function. Using these results the isotropic}nematic transition properties were calculated [143]
for the hard ellipsoids and spherocylinders. For the hard ellipsoids all results were obained using
HNC DCF because no phase changes were found with PY values. It was found that for x "3 the
0
MC transition density is about 20% higher, whereas the MC fractional density change is smaller
than the density functional results [143]. These values are compared in Table 9 with other values.
For spherocylinders with x "6 it was observed that the second-order DFT theory combined with
0
the HNC DCF yields a transition density which is about 14% lower than the MC values. In
addition to hard particles #uid the calculations were also carried out [143] for #uids characterized
by pair potentials of a generalized MS type,
u(1,2)"u (r)#u (r)P (cos h ) , (4.197)
0 2 2 12
with
u (r)"!4s6 e(p/r)6 (model I) (4.198a)
2
and

CA B A B D
p 12 p 6
u (r)"!4s6 e # (model II) . (4.198b)
2 r r
Here u (r) was taken to be the usual Lennard}Jones interaction and s6 is a variable determining the
0
strength of the anisotropic interaction. The transition properties for both the models were
 S. Singh / Physics Reports 324 (2000) 107}269 187

calculated and it was found that in both cases the transition temperatures are considerably higher
than the absolute stability limits given by the HNC theory. Further, the fractional changes in
density found for both models (Eqs. (4.198a) and (4.198b)) are about an order of magnitude larger
than those obtained for hard ellipsoids and spherocylinders.
A second-order density functional theory was used by Ram and Singh [142] to study within
decoupling approximation the isotropic}nematic transition in a HER system. The direct pair
correlation function of coexisting isotropic liquid was obtained by solving the OZ equation using
the PY closure relation. In this work, the PY integral equation was solved numerically at densities
which span beyond the range studies by Perera and co-workers [143,205] and it was shown that
the IN transition takes place for ellipsoids with x 53.0. The IN transition properties were
0
calculated for several values of x and it is found that a system spontaneously transforms to
0
a nematic phase when the structural parameters CK (0) and CK (0) attain, respectively, values close to
22 44
4.40 and 1.12. These numbers vary, though very weakly, with x , as x is increased the value of
0 0
CK (0) is decreased while CK (0) increases. The transition parameters are compared in Table 9 with other
22 44
results. Singh and co-workers [142](d) solved the PY integral equation for two model #uids: HER
#uid represented by a Gaussian overlap model and a #uid the molecules of which interact via
a Gay}Berne model potential. These model systems are expected to capture some of the basic
features of real ordered phases. For example, a system of HER has been found [156,206] to exhibit
four distinct phases, isotropic #uid, nematic #uid, plastic solid and ordered solid. The simulation
results [207,208] show that the Gay}Berne (GB) pair potential is capable of forming nematic,
smectic A, smectic B and an ordered solid in addition to the isotropic liquid.
The GB pair potential model is written as

CA B A BD
p 12 p 6
u(r, X , X )"4e(r( , X , X ) 0 ! 0 ,
1 2 1 2 r!p(r( , X , X )#p r!p(r( , X , X )#p
1 2 0 1 2 0
(4.199)

where e(r( , X , X ) and p(r( , X , X ) are angle-dependent strength and range parameters and are
1 2 1 2
de"ned as

e(r( , X , X )"e (1!s2[e( ) e( ]2)~1@2

1 2 0 1 2

G H
(r( ) e( )2#(r( ) e( )2!2s@(r( ) e( )(r( ) e( )(e( ) e( ) 2
x 1!s@ 1 2 1 2 1 2 (4.199a)
1!s@2(e( ) e( )2
1 2
and

C D
(r( ) e( )2#(r( ) e( )2!2s@(r( ) e( )(r( ) e( )(e( ) e( ) ~1@2
p(r( , X , X )"p 1!s@ 1 2 1 2 1 2 (4.199b)
1 2 0 1!s@2(e( ) e( )2
1 2
with s@"(Jk@!1)/(Jk@#1); k@ is the ratio of the potential well depths for the side-by-side and
end-to-end con"gurations. It is important to note that in this calculation [142](d) the OZ equation
using PY closure was solved numerically considering 30 harmonic coe$cients, whereas only 14
188 S. Singh / Physics Reports 324 (2000) 107}269

Table 10
The NI transition parameters for GB #uid [142](f, g) (x "3.0, k@"5.0) at ¹H"0.95 and 1.25. Here
0
PH"pp3/e , kH"k/e and oH"op3
0 0 0 0
¹H Theory oH oH *oH/oH PM PM PH kH
*40 /%. /%. 2NI 4NI
0.95 MD[207](a) 0.308 0.314 0.019 0.50 3.50 12.70
DFT 0.322 0.328 0.018 0.675 0.372 3.398 11.28
1.25 MD[207](a) 0.323 0.331 0.024 0.50 5.70 20.90
DFT 0.378 0.382 0.010 0.676 0.377 10.903 34.27
MWDA 0.377 0.380 0.008 0.443 0.182 10.675 33.66

coe$cients were included by Ram and Singh [142] in their calculation. Though the results found
for these models, HER and GB, are in qualitative agreement with the simulation results [205,207],
the quantitative agreement is not satisfactory. Secondly, the HNC and PY approximations do not
give thermodynamic consistency for the virial and compressibility routes to calculate pressures. So
in an other work Singh and co-workers [142](e) developed a thermodynamically consistent (TC)
integral equation for the pair correlation functions of molecular #uids which interpolates continu-
ously between the HNC and PY approximations. The TC integral equation is a generalization of
Rogers and Young [209] method devised for atomic #uids to an angle-dependent pair potential.
More importantly, the thermodynamic consistency between the virial and compressibility equation
of state has been achieved through a suitably chosen adjustable parameter. The solutions obtained
by using TC equation have been found to be in accurate agreement with the simulation results
[205}207].
Taking the values of the spherical harmonic coe$cients of the DPCF as obtained from the TC
closure relation, Singh and co-workers [142](f, g) calculated NI phase transition properties for
the HER #uid represented by a Gaussian overlap model and GB #uids using the DFT and
MWDA methods. The results [142] for hard ellipsoids of length-to-width ratio x "3.0
0
are compared in Table 9 with MC simulation and various other calculations. It can be seen
from Table 9 that the work of Singh and Singh [142], using the functional Taylor expansion
and the decoupling approximation, yields too small densities at the transition. Marko's
[177] version gives almost correct density but predicts an extremely small density jump
and order parameter. Further, the transition densities obtained by Singh et al. [142] using
DFT and MWDA approximation agree very well with the MC simulation [156](a) results. The
fractional density change is overestimated as compared to simulation values. The transition points
at two temperatures ¹H"0.95 and 1.25 were determined [142] for the GB #uid. The results
obtained using the harmonics of DPCF from the PY integral equation are compared in Table 10
with the MD results [207]. It can be seen that the NI transition is predicted at higher density
whereas the fractional density changes are small as compared to MD results [207]. Further, the
coexistence densities increase with increasing temperature. At lower temperature the fractional
density change is in better agreement with MD value as compared to its value at higher
temperature. The DFT and MWDA overestimate the pressure and chemical potential as the
temperature is increased.
 S. Singh / Physics Reports 324 (2000) 107}269 189

5. Nematic}smectic A (NS ) phase transition

A
In spite of considerable reserach activity over the last three decades the nematic-to-smectic phase
transitions remain one of the principal unsolved problem in statistical physics of condensed matter.
The uniaxial nematic-to-smectic A (NS ) transition involves a rearrangement of the centres of the
A
molecules; the centres are disordered in nematic state and ordered in the smectic state. The
ordering in the S phase is such that the molecular centres are, on average, arranged in equidistant
A
planes with interplanar spacing m . Within these layers the molecules move at random, with the
0
restriction that the director remains perpendicular to the smectic layers. Consequently, the density
changes its behaviour at the S N transition from periodic to homogeneous. The one-dimensional
A
density wave characterizing the S phase is associated with the Landau}Peierls instability. In the
A
thermodynamic limit there exist no long-range translational order in the S phase. Thus, the theory
A
of S N transition is a theory of one-dimensional melting. Since covering all aspects of the transition
A
is not possible here, we shall summarize some molecular models developed to describe the origin
and details of S ordering and S N transition.
A A
During the past 25 years, many high-resolution heat capacity and X-ray studies have been
devoted to the S N transition [211}225]. Most of the focus has been on the critical exponents in
A
the hope of determining the universality class of this transition. The most extensively measured
critical exponents are d,c,l and l , where the last two correspond to the divergence of the
@@ M
correlation lengths parallel and perpendicular to the director, respectively. In summarizing the
data, it is convenient to specify the McMillan ratio ¹ /¹ . At "rst, DSC seemed to indicate that
NA NI
most of the observed NS transitions were discontinuous, with only those below the tricritical
A
value of 0.87 being continuous. Later work using adiabatic scanning calorimetry and AC
calorimetry showed that pretransitional e!ect has been interpreted as latent heats in the DSC
experiment, and that continuous behaviour occurred upto much higher values of the McMillam
ratio. The critical exponents for the susceptibility and correlation lengths are measured from di!use
X-ray scattering experiments in the nematic phase [3,21]. At "rst, the X-ray structure factor is "t to
the following form:

s
,
1#m2q2#m2 q2 (1#cm2 q2 )
@@ z M M M M
where c is a constant. Next, the temperature dependence of s, m and m are "t to power laws to "nd
@@ M
the critical exponents c, l and l , respectively. For the McMillan ratio below about 0.93, these
@@ M
exponents are close to the 3D X> universality class, a"!0.007, c"1.316, l "l "0.669.
@@ M
Above a ratio of roughly 0.93, they are no longer fairly constant but tend to approach their
tricritical values, a"0.5, c"1.0, l "l , l "0.5, as the McMillan ratio approaches unity. It is
@@ M @@
important to mention that the value of the McMillan ratio at the tricritical point is not universal
and that the crossover from 3D X> values to tricritical values is not the same for di!erent
homologous series, but the trend in the critical exponents is very similar.
The theory of S N transition has received considerable attention [226}260] than any other
A
smectic transtion, partially due to the analogy with the superconducting normal transition. In spite
of all these e!orts, the situation remains very complicated with numerous questions unresolved.
Frenkel and co-workers [156,159] have performed the computer simulation studies on a system of
190 S. Singh / Physics Reports 324 (2000) 107}269

hard spherocylinders to observe the S N transition. A list of the references to much of the
A
theoretical and experimental work is given in an article by Garland and Nounesis [264]. Longa
[264] has discussed the application of the Landau and molecular-"eld theories to the S N
A
transition very well.
At the NS phase transition the continuous translational symmetry of the nematic phase is
A
spontaneously broken by the appearance of one-dimensional density wave in the S phase. Close to
A
the transition the onset of quasismectic features in the nematic phase may lead to a drastic change
in certain important properties (elastic coe$cients, transport properties, cholesteric pitch, etc.)
[3,261]. Original theories due to McMillan [234] and de Gennes [62] suggested that the NS
A
transition could be "rst or second order. The order of transition changes at a tricritical point
(TCP). Alben [262] predicted a 3He} 4He like TCP in binary liquid crystal mixtures. However,
Halperine et al. [263] argued that the NS transition can never be truely second order, which of
A
course rules out the possibility of a TCP. This controversy has spurred experimental studies
[211}225] which have shown that NS transition can indeed be continuous when measured to the
A
dimensionless temperature ((¹!¹ )/¹ )K10~5. In a recent paper, Lelidis and Durand [265]
NA NI
predicted the "eld-induced TCP in NS transition. Thus, in spite of early controversies, it is
A
witnessed now that the NS transition is continuous in the absence of special circumstances
A
[215,266] in accordance with the suggestion of McMillan [234] and de Gennes [62] for a speci"c
models. The salient features of the NS transition have been documented well by several authors
A
[13,267,268].

5.1. Phenomenological description of the NS transition

A
The Landau}de Gennes theory for the NI transition (Section 4.1) can be extended to the NS
A
transition. We start by de"ning an order parameter for the smectic A phase. The order parameter
for this transtion is DtD, the amplitude of the density wave describing the formation of layers in
the smectic A phase. Since the di!erence between a value of !DtD and DtD only amounts to a shift
of one-half layer spacing in the location of all the layers, no change in the free energy per unit
volume results. Therefore, the expansion in terms of the powers of DtD can only contain even-power
terms.
Let us begin with a simple case. The S phase is characterized by a density modulation in
A
a z( direction orthogonal to the layers [3]
o(r)"o #o cos(qz!U) , (5.1)
0 1
where o represents the "rst harmonic of the density modulation and U an arbitrary phase. In the
1
nematics o "0, whereas in S phase o becomes the natural candidate for the order parameter. In
1 A 1
the vicinity of the NS transition the free energy per unit volume may be expanded in powers of o :
A 1
f A"1a o2#1a o4#2 . (5.2)
S 2 2 1 4 4 1
At certain temperature ¹ , the coe$cient a Ka (¹!¹ ) vanishes. Above this temperature it is
0 2 0 0
positive. The coe$cient a is always positive. Only on these considerations, a second-order
4
transition could be obtained ¹ "¹ . However, a number of complications are to be taken into
AN 0
consideration. First, the in#uence of coupling between o and S, the nematic order parameter, is to
1
 S. Singh / Physics Reports 324 (2000) 107}269 191

be accounted for because of this coupling the optimal value of S does not coincide with
S (¹)(,PM ) obtained in the absence of smectic order.
0 2
Let
dS"S!S (¹) . (5.3)
0
The coupling term, to the lowest order, must have the form
f "!Co2dS , (5.4)
AN 1
where C is a positive constant. Further, to the free-energy density must be added the nematic
free-energy density f which is minimum for dS"0:
N
1
f "f (s )# dS2 , (5.5)
N N 0 2s
where s(¹), the response function, is large in the vicinity of NI transition but is small for ¹(¹ .
NI
The total free-energy density
f"f #f A#f (5.6)
N S AN
must be minimized with respect to dS. This leads to
dS"sCo2 (5.7)
1
and
f"f (s )#1a o2#1a@ o4 , (5.8)
N 0 2 2 1 4 4 1
a@ "a !2sC2 . (5.9)
4 4
The sign of a@ critically determines the order of transition. Clearly, the sign of a@ can change
4 4
depending on whether s is large or small, i.e., whether the nematic range is narrow or wide.
(i) If ¹ &¹ , s(¹ ) is large and a becomes negative. The stability requires that o6 term with
0 NI 0 4 1
positive coe$cient must be added to Eq. (5.8). In this case the transition takes place at a higher
temperature ¹ '¹ , and is "rst order.
AN 0
(ii) If ¹ ;¹ , s(¹ ) is small and a &a '0. The resulting transition is of second order and
0 NI 0 4 2
¹ &¹ .
AN 0
(iii) The point at which a "0 is a tricritical point. Thus, the change from a second order to "rst
4
order is induced by the coupling between o and S.
1
It is well known that the layer #uctuations play an important role in smectics. The layer
displacement u(r) appears through
U(r)"!qu(r) . (5.10)
Consequently, the smectic order can be described by introducing the complex order parameter
[62,134]
t(r)"o (r)e~*qu(r) , (5.11)
1
192 S. Singh / Physics Reports 324 (2000) 107}269

where q"2p/m . Now near the transition, the free-energy density may be expanded in powers of t.
0
Since the spatially dependent order parameter has been de"ned, we must add to the free energy the
gradient terms that express the tendency for the smectic to be homogeneous. Keeping all these
features in view, the general expansion for the free energy density can be written as
f"f (¹, Q, +Q)#f A(¹, t, +t)#f (Q, t, +Q, +t)#f (Q, t, +Q, +t, P) , (5.12)
N S AN %95
where f (¹, Q, +Q) is the free-energy density of the nematic phase (see Section 4.1), f A(¹, t, +t)
N S
corresponds to the smectic A phase, f is the contribution from the coupling between Q (nematic
AN
order parameter) and t and f is the free-energy density associated with the coupling of the order
%95
parameters and the external perturbation:

K K AK K K K B
1 1 1 Rt 2 1 Rt 2 Rt 2
f A" a DtD2# a@ DtD4# C # C # , (5.13)
S 2 2 4 4 2 @@ Rz 2 M Rx Ry
where C OC because of the nematic anisotropy. At the lowest order coupling term we can write
@@ M
f "1DQDtD2#1EQ2DtD2 , (5.14)
AN 2 2
where D and E are coupling constants. D is chosen negative to favour S phase when the nematic
A
phase exists and E'0. Generally, this term allows for reentrant e!ects (see Section 6). Because of
(Q, t) coupling the NS phase transition can be of second or "rst order.
A
In writing the expansion (5.13) the implicit assumption has been made that the director n( is "xed
in the Z-direction. In reality, n( #uctuates. So the gradient terms have to be taken in direction
parallel and perpendicular to n( . Owing to this the C term is modi"ed. Using the notation
M

A B
R R
+ " , ,
M Rx Ry
it becomes
C D(+ !iqdn( )tD2 , (5.15)
M M M
where dn( "n( !z( . When the director #uctuation is taken into account the Frank}Oseen elastic
M
contribution [37] has to be added to the total free energy.
A number of conclusions have been reached on this problem. The transition seems always to be
"rst order in four dimensions due to the coupling between the smectic order parameter and the
director #uctuation [263]. In three dimensions, the behaviour on the low and high sides of the
transition can be reversed from the 3D X> behaviour, i.e., an inverted 3D X> model [269,270].
A dislocation-loop melting theory, in which a divergence in the density of dislocation loops
destroys the smectic order, yields anisotropic critical behaviour in the correlation lengths
[271,272]. A non-inverted behaviour has been observed in MC simulations [54]. A self-consistent
one-loop theory employing intrinsically anisotropic coupling between the director #uctuations and
the smectic order parameter predicts a gradual crossover from isotropic behaviour to strongly
anisotropic behaviour [273,274]. Thus, the NS transition is very complicated with many factors
A
in#uencing the behaviour near the transition. Coupling between the smectic order parameter and
the nematic order drives the transition towards tricritical behaviour, while coupling between the
smectic order and nematic director #uctuations drives it into an anisotropic regime. While the
 S. Singh / Physics Reports 324 (2000) 107}269 193

McMillan ratio is a convenient indicator of the strength of both of these couplings, it is quite
imprecise. Yet the general trends with McMillan ratio are clearly evident [264].

5.2. Mean-xeld description of NS transition

A
McMillan [234] presented an elegant description of smectic A phase by extending the Maier
Saupe (MS) theory of the nematic phase to include the one-dimensional translational order of the
S phase. A similar but somewhat more general treatment, based on the theory of melting [275],
A
was proposed independently by Kobayashi [233]. Kobayashi developed a formalism of the
translational and orientational ordering but due to the two particle distribution for the ordered
phase the detailed numerical results for realistic potential were hard to obtain. A number of other
re"nements and extensions [236}242,250,260] were proposed but McMillan's model remains the
simplest one for its computational convenience and comparison with experiments and explains all
the qualitative features of the S N and S I transitions. A more simpler treatment of the S N
A A A
transition as compared to Kobayashi}McMillan approach has been given by Meyer and Lubensky
(ML) [235]. The ML model limits the discussion of the S N transition to the case of the ideal
A
orientational order, i.e. the temperature dependence of the orientation of the molecules is neglected.
We shall "rst describe the essential features of most widely used McMillan model. The physical
idea behind the model is based on the structure of molecules exhibiting S phase. The mesogenic
A
materials forming S phase typically have a central aromatic core and #exible alkyl chains at the
A
two ends. The aromatic moieties have large polarizabilities and thus the dispersion interaction
energy is very strong between the cores. Consequently, this can lead to the formation of a layered
arrangement as in S phase if the alkyl chains are su$ciently long and serve to e$ciently separate
A
the aromatic cores in layers. McMillan assumed that the anisotropic interaction is short-ranged
and can be expressed as

A B
<
; (r , cos h )"! 0 e~(r12@r0)2P (cos h ) , (5.16)
12 12 12 Nr3p3@2 2 12
0
where r is of the order of the length of the rigid section of the molecule, and the exponential term
0
re#ects the short range character of the interaction.
If the layer thickness is m the self-consistent one-particle potential that a test molecule would
0
experience, retaining only the leading term in the Fourier expansion, can be written as
u (z, cos h)"!< [PM #p6a cos(2pz/m )]P (cos h) , (5.17)
1 0 2 0 2
where the McMillan parameter a is given by
a"2 exp[!(pr /m )2] . (5.18)
0 0
PM is the usual orientational order parameter and p6 is an order parameter which couples
2
translational and orientational orders. Eq. (5.17) ensures that the energy is a minimum when the
molecule is in the smectic layer with its axis along Z.
The single-particle distribution function is given by
f (z, cos h)"exp[!bu (z, cos h)] . (5.19)
1 1
194 S. Singh / Physics Reports 324 (2000) 107}269

Using this distribution function and the interaction potential (5.16) the one-body potential is
recalculated

N:d3x dX u f (z , cos h )
u (z , cos h ), 2 2 12 2 2 "!< [P (cos h )SP (cos h )T
1 1 1 :d3x dX f (z , cos h ) 0 2 1 2 2
2 2 2 2
#a cos(2pz /m )P (cos h )Scos(2pz /m )P (cos h )T] . (5.20)
1 0 2 1 2 0 2 2
Self-consistency of Eqs. (5.17) and (5.20) demands that

PM "SP (cos h)T , (5.21a)

2 2
p6"Scos(2pz/m )P (cos h)T . (5.21b)
0 2
Eq. (5.21) must be solved self-consistently for both the order parameters, the order parameter
PM de"nes the orientational order, exactly as in MS theory, and p6 is a measure of the amplitude of
2
the density wave describing the layered structure. These equations exhibit three types of solutions:
(i) PM "p6"0, no order, characteristic of the isotropic liquid phase, (ii) PM O0, p6"0, orientational
2 2
order only, the theory reduces to the MS theory of the nematic phase; and (iii) PM O0, p6O0,
2
orientational and translational order characteristic of the S phase.
A
The free energy of the system is given by

A";!¹R , (5.22)

where the internal energy

;"!1N< (PM 2#ap62) (5.22a)

2 0 2
and the entropy is evaluated through

CP P D
1 m0 1
!TR"N< (PM 2#ap62)!Ni ¹ ln dm d(cos h) f (z, cos h) . (5.22b)
0 2 B m 0 1
0 0 0
At any temperature the phase with the lowest free energy is thermodynamically stable.
As obvious two physical parameters < and a enter the theory. < determines the ¹ and "xes
0 0 NI
the temperature scale of the model. a, a dimensionless parameter, is a measure of the strength of
layering interactions and can vary between 0 and 2. The interplanar distance is determined by the
competition between the anisotropic forces which produce the smectic order and excluded volume
e!ects. The smectic condensation energy is greater for larger values of a, that is for larger m .
0
Experimentally, the layer thickness is of the order of the molecular length. The parameter
a increases with increasing chain length of the alkyl tails. Eq. (5.21) were solved self-consistently and
the order parameters, entropy, speci"c heat as a function of temperature for several values of a were
evaluated and the transition parameters were calculated. The main results of calculations are
shown in Fig. 10. For a(0.70 and ¹ /¹ (0.87 the model predicts the second-order S N
AN NI A
transition. For a'0.98, the S phase transforms directly into the isotropic phase, while for
A
a(0.98 there is a S N transition followed by a NI transition at higher temperature. Hence a"0.7
A
 S. Singh / Physics Reports 324 (2000) 107}269 195

Fig. 10. (a) Phase diagram for theoretical model parameter a. Inset: typical phase diagram for homologous series of
compounds showing transition temperatures versus length of the alkyl end-chains [234]; (b) The variation of entropy
R and speici"c heat C with reduced temperature for a"0.6. A second order S }N transition and a "rst order NI
V A
transition are exhibited by the model [234].
196 S. Singh / Physics Reports 324 (2000) 107}269

and ¹ /¹ "0.87 correspond to a tricritical point at which the "rst-order transition terminates
AN NI
to a second-order transition. Fig. 10(a) shows that the theoretical phase diagram broadly re#ects
the experimental trends in homologous series. However, it was found that the theoretical
S transition entropy is somewhat higher than the experimental values. McMillan made an
AN
attempt, in a later paper [226], to improve the agreement by using a modi"ed pair potential

; (r , cos h )"!(< /Nr3p3@2)exp(!(r /r )2)(P (cos h )#d) . (5.23)

12 12 12 0 0 12 0 2 12
The three parameters were "xed by the criterion that the theory gives transition temperatures and
entropies. The results were essentially the same as those obtained by the two-parameter model
(Eq. (5.16)) but some quantitative improvements were obtained. McMillan [234] himself suggested
that if the excluded volume e!ect is included in Eq. (5.16) it would favour n( to be parallel to z( .
Further, it was suggested by Priest [260] that a generalized rank-2 tensor model would give
rise to additional terms which favour n( to be parallel to q( . In most homologous series, the S N
A
transition point T reaches a maximum value for some chain length and tends to decrease for
AN
higher homologues.
Woo and co-workers [250] have used a pairwise potential of the type

C A BD
r 2
;(1, 2)"< exp ! 12 [!d!P (X ) X )#eMP (X ) r( )#P (X ) r( )N] . (5.24)
0 r 2 1 2 2 1 12 2 2 12
0
Four physical paramaters < , d, r and e appear in the theory; d is a measure of the interaction
0 0
which gives rise to the translational order even in the absence of orientational order and e roughly
accounts for the steric e!ects, which help to keep n( parallel to q( . The four parameters of the model
are adjusted to "t the measured transition temperature curves and approximate triple point. For
each homologous series, < determines one point on the ¹ curve, while r (or a) measures the
0 NI 0
length of the central section of the molecule. Both these parameters are chosen at the outset and
thereafter held "xed. The remaining two parameters d and e are adjusted to "t rest of the
experimental phase diagram. The model predicted the phase diagrams similar to that of experi-
mental one. Further the parameter e is su$cient to generate a reasonably good "t to the
experimental phase diagram. The role of the end chains is merely to cause a larger interplanar
spacing in the S phase and thus does not a!ect the model interaction. The characteristic feature of
A
the S phase that the director prefers to be perpendicular to the smectic layers is incorporated into
A
the model (5.24). The connection between e and the structure of molecules can explain the
di!erences in phase transition properties between homologous series whose molecules are of
similar structure but di!er in the length of rigid section. Sokalski [251] has used a corner potential
with Lennard}Jones type of interaction

;(1,2)"4e[(p/r)m!(p/r)n] , (5.25)

where p depends on the orientations of the molecules. It was shown that the model (5.25) exhibits
the smectic A, nematic and isotropic phases.
Kloczkowski and Stecki [237] considered a model system of hard spherocylinders superimposed
with the attractive 1/r6 potential for the S phase. By using stability analysis these authors have
A
 S. Singh / Physics Reports 324 (2000) 107}269 197

shown that this type of potential allows for the S phase formation. The in#uence of attractive tail
A
on the S phase formation was studied and it was found that the additional centre to centre
A
r~6 potential produces an instability of the isotropic or nematic phases, towards the S phase
A
formation. The calculations were performed for spherocylinders with length to diameter ratio 5 and
10 for di!erent values of additional attractive potential parameters. This type of potential was also
used by Nakagawa and Akahane [276] in their treatment of the S phase formation. These authors
A
[241] also developed a McMillan-type theory for binary mixtures including both intermolecular
repulsions and attractions.
As discussed above, the original version [234] of McMillan's theory contains just two order
parameters, the orientational order paramater PM and a mixed orientational}translational order
2
parameter p6 de"ned as Scos(2pz/m )P (cos h)T. In this theory occurrence of the mixed order
0 2
parameter results in a strong correlation between the orientational and translational coordinates.
It has been argued that this strong correlation may be responsible for the quantitative failure of the
McMillan theory. In a later paper [226], McMillan introduced a purely translational order
parameter q6 ("Scos 2nz/m T) to improve the quantitative agreement. This third term has the e!ect
0
of stabilizing the S phase and so allows a second-order S N transition to occur at a higher
A A
temperature by reducing the relative correlation between orientational and translational coordi-
nates. In fact, the basic structure of the theory remains unchanged. Despite success the introduction
of third-order parameter leads to the complication of the theoretical treatments in obtaining both
analytic and numerical solutions. Katriel and Kventsel [236] have shown that this complication
could be removed by writing the mixed order parameter as a product of the pure orientational and
translational order parameters:
p6"SP (cos h)TScos(2pz/m )T"PM q6 . (5.26)
2 0 2
This decoupling approximation has the additional advantage that the strength of correlation
between the translational and orientational coordinates is readily controlled. The results obtained
using this simpli"cation (Eq. (5.26)) exhibit the main features of the isotropic}nematic}smectic
A transitions in a satisfactory manner. Kventsel et al. [242] made a quantitative comparison of the
two theories, McMillan's three order parameter theory [276] and Katriel and Kventsel [236]
theory. These authors [242] have demostrated the validity of the decoupling approximation for the
mixed order parameter in both the models. For the decoupled model the phase diagram involving
isotropic, nematic, smectic A and plastic phases was constructed and the S N transition properties,
A
which include the second-order transition temperature and tricritical point, the transition entropy
and the change in the order parameters, etc., were evaluated in detail. These properties have been
found to agree well with the predictions of McMillan's theory for a wide range of strength
parameters.
A lattice model for the S phase was proposed by Dowell and Ronis and Rosenblatt [249].
A
Dowell investigated the e!ects of temperature, pressure, tail chain #exibility and tail chain length
on the relative stabilities of the isotropic, nematic, smectic A and reentrant nematic phases (see
Section 6) using a lattice model [112] having only hard repulsions. The method provides a way to
determine on an individual basis which of the molecular features are su$cient and/or necessary for
the existence and relative stabilities of various phases. The thermodynamic and molecular ordering
properties, such as S order parameter, core and tail intermolecular orientational order para-
A
meters, tail intramolecular orientational order parameter, density and entropy, were evaluated, in
198 S. Singh / Physics Reports 324 (2000) 107}269

these phases and at the phase transitions. It has been found that for the formation of S phase in
A
real system the presence of semi#exible tail chains of signi"cant length is essential and that as the
length of tail chain is shortened the S phase disappears. Further, the dipolar phase is not necessary
A
in this steric packing model to form S phase but do e!ect the stability ranges of temperature and
A
pressure of the phases. As seen in Section 4.2, systems of hard particles, interacting through the
excluded-volume e!ect, have played a fundamental role in the understanding of structural phase
transitions in liquid crystals. Onsager made a major contribution to our understanding of nematic
phase in a system of hard spherocylinders. Stroobants et al. [159](a) provided evidence for the
appearance of S order in the system of perfectly parallel hard spherocylinders from simulation
A
studies (see Section 9). Stimulated by the simulation results, Mulder [255] developed a density
functional calculation and Wen and Meyer [244] analysed the origin of the S phase in a parallel
A
hard-rod system. In their later work [244], these authors described how the appearance of S order
A
reduces the excluded volume of the nematic phase and calculated the entropy for a system of
parallel right circular cylinders. The argument can be expressed quantitatively; there are three
contributions to the entropy changes in going from the S order to the nematic order; a term
A
describing directly the changes in long-range order, a term due to the change of packing density
within each layer, and a term due to the change in axial freedom of motion of each rod. The "rst
term has been expressed in terms of the distribution function of the centres of mass of the rods, and
the second term in terms of 2D packing density. The third term is signi"cant as compared to the
other two terms. The calculation demonstrated that a second-order NS phase transition takes
A
place as the packing fraction was raised to 0.202 times the value for the close packing. One of the
limitations of this model is that the results have no dependence on the length to diameter ratio of
the rods. Holyst and Poniewierski [246] studied a system of hard parallel cylinders in the
framework of smooth density approximation. Using a bifurcation analysis, it was shown that apart
from the nematic phase, smectic A, solid (or crystalline B) and columnar phases should also occur
in this system. This approach fails to distinguish between a solid phase and a crystalline smectic
B phase. Hosino et al. [243] had earlier found the persistence of the S N transition when hard
A
cylinders were allowed to have three orthogonal orientations. Taylor et al. [248] constructed an
excluded volume theory for the S and columnar phases of a system of hard spherocylinders using
A
the scaled particle theory to treat dimensions possessing full translational freedom, combined with
a simple cell model to describe the positionally ordered dimensions. These authors obtained
a phase diagram remarkably similar to that obtained from the Monte Carlo calculation (see Fig. 24
below) [159]. The predicted smectic layer spacing at the nematic}smectic cross-over is also almost
identical to the MC results. The model su!ers with the serious limitation that all transitions are
required to be discontinuous. Consequently, the second-order nematic}smectic transitions as
demostrated also by the MC results was found to be "rst order.
A number of experimental works [277}283] show that the molecular biaxility has to be
incorporated in the theoretical treatment in order to describe the structural and thermodynamic
properties of smectics. The "rst attempt of such consideration was made by the Everyanov and
Primak [253] within the framework of McMillan model generalized to the biaxial molecules. These
authors [253](a) investigated the in#uence of smectic layering on the order parameters of solute
molecules as well as on their positional orientational correlations. In the later paper [253](b)
the thermodynamics of S N phase transition in a system of biaxial molecules was studied. The
A
orientational translational ordering of the molecule can be characterized by a set of "ve order
 S. Singh / Physics Reports 324 (2000) 107}269 199

parameters, PM , q6, p6 and the other two de"ned as

2
GM "3Ssin2 h cos 2/T,S 1 1!S 1 1 , (5.27a)
2 xx yy
KM "3Ssin2 h cos 2/ cos(2pz/m )T . (5.27b)
2 0
Taking into account these order parameters within the framework of the McMillan model
generalized to biaxial molecules the molecular pseudopotential was written as [253](a)

C D
u (h, /, z)"!< PM #j GM P (cos h)#a p6#j KM
1 0 2 1 2 1 A B
A B
3
]P (cos h)cos(2pz/m )# j PM #j GM (sin2 h cos 2/)
2 0 1 2 2 2

A B
3
#a j p6#j KM sin2 hcos 2/ cos(2pz/m )#adq6 cos(2pz/m )] , (5.28)
1 2 2 0 0

where j and j are the biaxial parameters

1 2
2u 2u
j " 220 and j " 222 . (5.29)
1 J6u 2 3u
200
200
Here u "u are the expansion coe$cients of the pairwise potential for the e!ective anisotropic
212 221
intermolecular interactions. Thus, in addition to three adjustable parameters in McMillan's theory
there is a supplementary parameter, in the model using the relation j "j 2. For the "rst-order
S N transition these four parameters can be evaluated from independent 2 1 measurements of
A
¹ , ¹ , *R and the correlation GM (PM ) in the nematic phase near ¹ or from the other set of
NI AN AN 2 AN
experimental data. The sign of the parameters GM and KM coincide with the sign of j , so that
changing the sign of j does not in#uence the values of ¹ , ¹ and *R . The extensive1numerical
1 NI AN AN
calculations were performed and it was found that the biaxiality modi"es the phase diagram of the
McMillan model. With increasing biaxiality parameter j("(3/2)1@2j ),¹ increases but the de-
pendence ¹ (j) is weaker and more complex. For each value of d and 1 jNIthere is a critical value
AN
a (d, j) such that the inequality ¹ (d, j)5¹ (d, 0) are valid for a5a . For each value of j there
# AN AN #
is a limited range of changes 0(d4d (j), where a (d, j)4a (d, j). With d and a being constant
1 5#1 #
the entropy change *R decreases with increasing j. For su$ciently high values of d and for
AN
a corresponding to a narrow nematic range increasing j can result in changing the S N transition
A
from "rst to second order. The parameter d strengthens the layering tendency independent of
orientational ordering and has to increase as the alkyl chain is lengthened at "xed length r of the
0
molecular aromatic core. It is expected that the anisotropic steric interactions do make a signi"cant
contribution to the e!ective value of j. The change *GM (¹ ) and the dependence GM (¹) in the
AN
smectic phase are determined by the change of the order parameter PM at ¹ and its dependence
2 AN
on the temperature. The character of the dependence of the derivative dGM /d¹ on temperature
changes qualitatively both for the "rst- and second-order S N transitions. It is important to
A
mention here that the qualitative considerations, which are in agreement with known experiments,
200 S. Singh / Physics Reports 324 (2000) 107}269

show that the maximum manifestation of the pseudopotential biaxiality should be expected for
mesogens having su$ciently long end chains and narrow nematic ranges.

5.3. Application of density functional theory to NS transition

A
Lipkin and Oxtoby [140] were "rst to consider the application of the density functional
approach to the mean-"eld theory of the isotropic}nematic}smectic A transition. This theory
incorporates the steric e!ect through the pair correlation function and gives the result that q is
parallel to n( . These authors constructed a set of self-consistent equation for the order parameters,

P P
2l#1
dl d #k l" dX Pl(cos h) dr exp(!iG ) r)
0 p0 p 4p< q

q
C
]exp + exp(iG ) r) + (2l #1)J(q, l l )k l2Pl1(cos h) ,
q ll
1 2
1 1 2 q D (5.30)

where k are the order parameters of the theory and

ql
J(q, l l )"J((q, l l ))
1 2 2 1

G H
o 2l#1 1@2
" 0 + G(l l l; 00)i l Cl1l2l(i ) (5.31)
(4p)3@2 l 1 2 (2l #1)(2l #1) 2
1 2
with

P
=
Cl1l2l(i )"4p dr r2jl(i r)Cl1l2l(r) , (5.32)
2 2
0
jl is a spherical Bessel function.
The free-energy di!erence between isotropic #uid and the ordered state was written as [140]

C D
1
b*A"o < !k #J(0, 00)k # +@ J(q, l l ) k l1k l2 , (5.33)
0 00 00 2 l l 1 2 q q
q12
where the summation excludes the term with q"l "l "0. Eq. (5.33) is a direct generalization of
1 2
the work of Slucking and Shukla [141] to the molecular system. A self-consistent solutions of Eqs.
(5.32) and (5.33) relate the phase diagram of the isotropic, nematic and S phases to the direct
A
correlation function of the isotropic liquid. Truncating the order parameter expansion at the lowest
term (q"0,$1, l"0, 2) and relating the term k to the density change between the isotropic
00
liquid and liquid crystal a connection with the McMillan theory was established. With these
approximations the excess free energy was reformulated as

A BP P
b*A 1 1 m0 1
" J(0,22)k2 #J(1,00)k2 #J(1,22)k2 #2J(1,02)k k !ln dz d(cos h)
N 2 02 10 12 10 12 m
0 0 0
]exp[5J(0,22)k P (cos h)#cos(2pz/m )M2J(1,00)
02 2 0
k #2J(1,20)k #[10J(1,20)k #10J(1,22)k ]P (cos h)N] . (5.34)
10 12 10 12 2
 S. Singh / Physics Reports 324 (2000) 107}269 201

Eq. (5.34) has the same form as obtained in the McMillan theory if the following identi"cations are
made.

J(0,22)P 1 b< ,
25 0
J(1,22)P 1 b< a ,
50 0
J(1,00)P1bad< ,
2 0
J(1,20)P0 .

Thus, here an extra cross-term between k and k appears which is absent in the McMillan
10 12
theory of smectic A.
The density functional theory has been used to study the S N transition in a system of parallel
A
hard spherocylinders [144,255}257] as well as system with orientational degrees of freedom [258].
Mahato et al. [259] developed a density wave theory which involves the direct correlation function
of ellipsoids of revolution and showed that the S phase is metastable with respect to the bcc crystal
A
in such a system. Mulder [255] made an attempt to locate the NS transition in a system of
A
perfectly aligned hard spherocylinders (PAHSC) by using a bifurcation analysis of the free-energy
functional in the second virial coe$cient approximation and obtained a second-order transition of
mean-"eld type towards a smectic phase. He also studied the e!ect of higher-order terms in the
density expansion on the location of the bifurcation point by considering the in#uence of the third-
and fourth-order terms. The values of critical packing fraction and wavelength are in good
agreement with the simulation values [159] as compared to the results obtained by Hosino et al.
[243] using the method of symmetry breaking potential in the second virial coe$cient approxima-
tion,

gH K0.37, gH K0.36, gHK0.729 ,

M MC H
jH K1.34, jH K1.27, jHK1.414 .
M MC H
A more systematic calculation was reported by Somoza and Tarazona [144,257]. These authors
constructed a free-energy functional for a system of parallel hard spherocylinders [257] as well as
for a system of hard bodies with arbitrary shape and orientational distribution [144]. Somoza and
Tarazona [257] constructed a free-energy functional by generalizing the Lee's functional scaling
method [110] of hard-sphere reference system to the parallel hard ellipsoid (PHE) and then to use
the latter as the reference system for the real hard bodies (HB). The interaction part of the
free-energy functional expressed as

PP
:dr@:dX@ o(r@, X@)M (r!r@, X, X@)
*A[o(r, X)]" dr dX o(r, X)*W [o6(r)] HB (5.35)
PHE :dr@o(r@)M (r!r@)
PHE
was evaluated at the e!ective density o6(r). Here M and M are the respective Mayer func-
PHE HB
tions, which give the second virial coe$cients by integration over all the variables. A criterion
has to be de"ned for choosing the reference PHE. It should re#ect both the molecular shape
and the orientational distribution function. An empirical rule was proposed [257,144] the
202 S. Singh / Physics Reports 324 (2000) 107}269

basis of the tensor of inertia of the HB, SIHB(X)T, averaged over the orientations with the function
o(r, X),

P
SIHBT" dX o(r, X)IHB(X)/o(r) . (5.36)

The length of the PHE along the principal axes are taken so that the eigenvalues of its inertial
tensor are proportional to the corresponding eigenvalues of the hard-body tensor of inertia:

IPHE IPHE IPHE

1 " 2 " 3 . (5.37)
SIHBT SIHBT SIHBT
1 2 3
These equations together with the equal volume condition < "< , fully specify the PHE. Eq.
PHE HB
(5.35) may be regarded as a way to study the general HB system as a perturbation from the PHE
system, for which the direct mapping onto HS can be used. The density pro"les and the free-enery
of S phase were evaluated numerically and a second-order NS phase transition was observed.
A A
The treatment was extended [144] to a system of HB with arbitrary shape and orientational
distribution. The phase diagram was evaluated for a system of parallel hard spherocylinder, and its
dependence only on the ratio ¸/d was observed. The use of di!erent volumes for the PHSC and
0
PHE as to give the same close packing density, together with Eq. (5.37) gave the best results for the
molecules of intermediate elongation (24¸/d 45). The use of equal volume for the reference and
0
real molecules and Eq. (5.37) underestimates the critical density by about 10%. The continuous
character of the NS phase transition for the PHSC provided an easy way to calculate the phase
A
diagram based on the analysis of the stability of the nematic against a density modulation. In order
to achieve this, Somoza and Tarazona [144] calculated the direct correlation function for the
homogeneous phase from the second functional derivative of Eq. (5.35). From the Fourier
transform of the direct correlation function C(q), the structure factor S(q) was obtained which,
contrary to the case of a simple #uid, depends not only on the modulus of the wave vector q but
also on its direction (relative to the nematic director). The numerical results show that the inclusion
of the contribution to the direct correlation function coming from the real HB Mayer function
changes this qualitatively because it induces a non-trivial dependence of C (q) with the direction
HB
of the wave vector. A stable S phase is observed (see Fig. 11) in an interval of densities between the
A
nematic and the crystal phases. This provides an insight into how the slight di!erence in the shape
between the PHE and the PHSC may lead to such dramatic changes in the phase diagram.
Poniewierski and Holyst [258] used density function theory to study the NI and NS phase
A
transitions in systems of hard spherocylinders with full translational and orientational degrees of
freedom. The free-energy functional was constructed in the spirit of the smooth density approxima-
tion (SDA) as developed by Tarazona [197] and also by Curtin and Ashcroft [198] for the
inhomogeneous hard-sphere #uids. Both phase transitions, NI and NS were observed in a wide
A
range of length-to-width ratio ¸/d and the results tend to Onsager's limit of ¸/d PR. In an
0 0
other paper these authors [256] applied the SDA theory to study the NS transition for the system
A
of perfectly aligned hard spherocylinders. The free-energy functional was constructed as

P P
bA 1 2p 1 2p
"ln(oK3)!1# u(m)ln u(m) dm# u(m)*t(ou6 (m)) dm , (5.38)
N 2p 2p
0 0
 S. Singh / Physics Reports 324 (2000) 107}269 203

Fig. 11. Phase diagram for a system of parallel hard oblique cylinders. N, S , S and N phases may appear. The dashed
A C R
lines are continuous transitions and the full lines "rst order transitions. The circle separates "rst and second order phase
transitions between the S and S phases.
A C

where

=
u6 (m)"1# + k6 u(ni)cos(nm) (5.39)
n
n/1
and

C DN
4p 4pd
u(ni)" Msin[ni(d #¸)]!sin(ni¸)N! 0 cos[ni(d #¸)] v . (5.40)
(ni)3 0 (ni)2 0 0

For two parallel spherocylinders v "2p¸d2#4pd3. k6 is the nth-order parameter and k the
0 0 3 0 n
smectic wave vector. The minimization of }A/N with respect to k and k provides the equilibrium
n
solution for the density wave of the S phase. The bifurcation analysis was used to locate the
A
transition and numerical calculations were performed by taking

g(4!3g)
*t(o)"i ¹ . (5.41)
B (1!g)2

Using bifurcation analysis, the density oH at which the nematic solution becomes unstable with
respect to the perturbations of the symmetry of the S phase was obtained and the following
A
asymptotic relation for the di!erence between the free energies of the smectic and nematic phases
was arrived at

*A AN&!(o!oH)2 .
S
204 S. Singh / Physics Reports 324 (2000) 107}269

The variation of the transition density oH/o (o is the close packing density) with ¸/d was
#1 #1 0
studied. It was found that oH/o ranges from 0.31 for ¸/d PR to 0.41 for ¸/d "0.5 which
#1 0 0
agrees with the simulation results [159] within 25%. Further, the NS transition occurs for all the
A
values of ¸/d whereas in simulations NS transition is not observed for ¸/d 40.25.
0 A 0

6. Re-entrant phase transitions (RPT) in liquid crystals

An exception to the sequence rule (3.1) and observed sequences (Table 3) of phase transitions in
liquid crystals was discovered, at atmospheric pressure, in the year 1975 by Cladis [284] in certain
strongly polar materials. In a binary mixture of two cyano compounds, HBAB Mp-[p-hexyloxyben-
zylidene)-amino]benzonitrileN and CBOOA [N-p-cyanobenzylidene-p-n-octyloxyaniline], over
a range of compositions, the following sequence was observed on cooling the mixture from the
isotropic phase,
ILPN PS $PN Psolid , (6.1)
6 A R
where N stands for a second nematic, known as re-entrant nematic, which appears at lower
R
temperature. The smectic-A phase existing between two nematics was identi"ed to be the partially
bilayer phase S $. A similar result was reported by Cladis et al. [285] in many other binary mixtures
A
and in a pure compound COOB (4-cyano-4@-octyloxy biphenyl) at high pressures (see Fig. 12).
They also observed that the pure compounds CBNA (N-p-cyanobenzylidene-p-nonylaniline),
CBOA (N-p-cyanobenzylidene-p-octylaniline) and COB (4-cyano-4@-octylbiphenyl), although pos-
sessing bilayer smectic-A phase, did not show re-entrant behaviour for pressure under 10 kbar.
X-ray and microscopic studies [286] show that the re-entrant nematic phase is quite di!erent
from the classical (higher-temperature) nematic phase. The transition from smectic-A to re-entrant
nematic phase is reversible from the point of view of X-ray and optical results. The possibility that
the re-entrant nematic is actually a smectic-C phase was excluded [289] on the experimental
grounds; N phase is uniaxial whereas S is biaxial. The defect structure of re-entrant nematic
R C
observed in cylindrical geometry was identical to that observed for the classical nematic. For
mixtures forming a N phase the birefringence [287] has been found to be continuous at both the
R
transitions NS and S N . The magnitude of the increase at the NS transition point and the
A A R A
decrease on going from the S phase to the N phase has been found to depend on the length of
A R
smectic A range and tend to zero as this vanishes.
The above example of the phase transition (6.1) shows that the high-symmetry phase may
re-enter at lower temperature than low-symmetry phase in a rather unexpected way. This kind of
phenomenon is termed as `re-entrant phenomenona. A system is said to be exhibiting re-entrant
phase transitions (RPT) if a monotonic variation of any thermodynamic "eld results in two (or
more) phase transitions and attains a state which is macroscopically similar to the initial state or
the system re-enters the original state. The phenomenon of RPT is intrinsically novel and the
continued interest in this problem [288] is underlined by its discovery in amazingly diverse systems
in addition to liquid crystals, e.g., binary gases [289}291], liquid mixtures [292}296], ferroelectrics
[297], organometallic compounds [298], granular superconductors [299], gels [300], aqueous
electrolytes [301], antiferromagnets [302], etc. The subject is reviewed well by Cladis [303] for
liquid crystals and by Narayan and Kumar [288] by multicomponent liquid mixtures.
 S. Singh / Physics Reports 324 (2000) 107}269 205

Fig. 12. P}T phase diagram for COOB [285]. Data taken on the N S transition in the supercooled liquid are shown as
R A
crosses (X).

The discovery of re-entrant behaviour [284] in liquid crystals has resulted in extensive studies
[288,303,304], both experimental and theoretical, of this intricate kinds of phenomenon. Much of
the experimental work has been concerned with the synthesis of the other re-entrant systems, the
establishment of their phase diagrams as a function of temperature, pressure and composition and
the determination of orientational order and its change at the phase transitions. The main concern
of theoretical work has been to understand the microscopic origin and the true nature of RPT.
A few representative examples of RPT in liquid crystals are illustrated below.

6.1. Examples of single re-entrance

There are quite a few studies [286,287,305}312] on the binary mixture of hexyloxy-
cyanobiphenyl (6 OCB) and 4@-n-octyloxy-4-cyanobiphenyl (8 OCB) which exhibit nematic re-
entrance as a function of temperature, pressure and composition. Kortan et al. [310] reported
a high-resolution X-ray study of the smectic A #uctuation in the nematic phases of this mixture
with emphasis on the region of concentration when the smectic phase barely forms or becomes
unstable before the correlation length has truely diverged. A number of unusual phenomena were
observed in this region, e.g., dramatically increased bare lengths for the smectic correlation, critical
206 S. Singh / Physics Reports 324 (2000) 107}269

Fig. 13. The phase diagram for 6OCB/8OCB mixture [310]. The concentration y is the molecular ratio 6OCB : 8OCB.
The open and solid circles represent, respectively, the data obtained by X-ray scattering and light scattering experiments.
The vertical lines indicate the experimental path.

exponents which may be as large as twice those found in pure 80CB crossover e!ects, etc. The
phase transition temperature was measured by observing the temperature at which the minimum in
the width of q scans occurred rather than being treated as adjustable parameter. Fig. 13 shows the
1
phase diagram so obtained. The in#uence of pressure on the S N and NI phase boundaries were
A
studied [308] as a function of mole fraction of 60CB by DTA and optical microscopy. It was found
that the maximum pressure of occurrence of the smectic A phase decreases with increasing mole
fraction (y) of 6 OCB until yK0.3 where no smectic phase exists. Further, the pressure behaviour of
the NI transition gets drastically a!ected by structural changes occurring at lower temperatures.
Although they could not ascertain whether this e!ect is due to the in#uence of the smectic ordering
or due to the presence of re-entrant nematic at farther temperatures or due to a combination of
both, it was tentatively concluded that the classical or high-temperature nematic, in the concentra-
tion range of the existence of the re-entrant nematic, possesses a molecular ordering which is
somewhat di!erent from the ordering of the nematic phase occurring at higher concentrations.
The orientational order and its change at the phase transitions for the 6 OCB/8 OCB has been
studied [287] by measuring its optical birefringence, a quantity which can be determined with
considerable precision. For the mixture exhibiting re-entrant nematic phase the birefringence is
found to be continuous at both the transitions; nematic}smectic A and smectic A-reentrant
nematic. It is important to note that the birefringence in the smectic A phase is found to be greater
than that obtained by an appropriate extrapolation from the results in the nematic phase. Contrary
to it birefringence of the N phase is observed to be less than that determined by extrapolation from
R
 S. Singh / Physics Reports 324 (2000) 107}269 207

the smectic A phase. The magnitude of the increase in case of nematic}smectic A transition and the
decrease on going from smectic A to reentrant nematic phase are found to depend on the range of
stability of smectic A phase. Emsley et al. [312] investigated in detail the orientational behaviour of
this mixture using deuterium NMR spectroscopy, a technique which is able to provide information
about the orientational order of the individual components of a mixture. More importantly, this
technique can determine the order parameters for the rigid subunits in a mesogenic molecules. This
variation in the order parameter has been determined in the mixture 6 OCB/8 OCB as well as in the
pure components as a function of temperature. The order parameters were found to undergo subtle
changes at the transition from the smectic A to the reentrant nematic phase.
One of the important features of the RPT in liquid mixtures is the existence of a closed-loop
phase diagram with upper and lower critical solution temperatures (¹ and ¹ , respectively).
6 L
A double critical point (DCP) results when the ¹ and ¹ are made to coincide. Such systems
6 L
provide richer informations as they permit a multitude of paths by which a critical point can be
approached. A wide variety of phases can be obtained in these systems by mere variation of
temperature, pressure, composition, additional components, etc. Perhaps the "rst closed-loop
existence curve was obtained by Hudson [313] for nicotine/water mixture while trying to crystal-
lize nicotine from its aqueous solution. Since then the closed-loop phase diagrams have been
reported in many systems [288]. Fig. 14 shows a closed-loop nematic}smectic A phase boundary
observed [314] in mixtures of 71 -CBP (4-n-heptyloxy-4@-cyanobiphenyl) and a binary mixture of
CBOOA and 81 .0.5 (4-n-pentylphenyl-4-n-octyloxy-benzoate) as a function of temperature, pressure
and composition. It can be seen that the pressure reduces the area of smectic A phase and the
re-entrance vanishes (i.e. a DCP is attained) at 145 bar.
In principle, the re-entrant phases can occur [249,138,304,315] in non-polar system also. The
experimental evidence in support of this prediction has been unveiled by the Halle group [316,317].

Fig. 14. T}y phase diagram of the system [81 .0.5/CBOOA (y "0.88)]/71 CBP for di!erent pressures [314].
CBOOA
208 S. Singh / Physics Reports 324 (2000) 107}269

A nematic reentrant system of non-polar compounds, with one component a side-chain LC

polymer, has been reported [318] which exhibits this kind of inversion of symmetry. The polymer is
an utactic polymethylsiloxane substituted with 4-undecyloxyphenyl ester of 4-methoxybenzoic
acid (P ) which follows the phase sequence
11,1
603C 1343C
S &" S &" IL . (6.2)
C A

The second component is a symmetrically susbtituted compound, the 4@-heptyloxybenzoate of

6-heptyloxynaphtyl-2, exhibiting the order of phase stability

603C 1303C
S &" N &" IL . (6.3)
C U

Hence S PN transition is monotropic.

# 6
The re-entrant sequence (N }S }N }IL) has been observed in a homologous mixture composed
R A 6
of 60% (weight) of the low molecular weight compound and 40% (weight) of polymer by using
optical polarizing microscopy and X-ray analysis. On cooling the mixture the focal conic S texture
A
transforms at 703C into a Schlieren twinkling nematic, which is similar to the high-temperature
nematic texture above 1293C. This is connected on the X-ray pattern with a change from
a resolution limited Bragg spot to a di!use scattering at small angles indicative of the loss of the
layered structure at long range. It has been found [318] that the layer thickness in the S phase of
A
the mixture is noticeably larger than both the molecular length of the mesogenic groups side-chain
of the polymer (33.5 As ) and the length of the low molecular weight compound. Similar anomalies of
the periodicity has been observed in the pure P for which the layer thickness increases
11,1
continuously with decreasing temperature as for some re-entrant polar compounds [319].
The re-entrant behaviour has also been observed in discotic liquid crystals. In a hexa-n-
alkanoates of truxene for the higher homologues the following sequence is observed [320,321], on
cooling:

ILPD PD PN PD Pcrystal . (6.4)

) 3 D R)
It has been conjectured that the truxene molecules are probably associated in pairs and that
these pairs break up at higher temperatures and might be responsible for this extra ordinary
behaviour.

6.2. Examples of multiple re-entrance

High-resolution X-ray scattering and heat capacity studies of the nematic}smectic A transitions
in 4@-(4@@-n-alkoxybenzyloxy)-4-cyanostilbene with the alkyl chain of length 7 (T7) and 8 (T8) were
reported by Evans-Lutterodt et al. [322]. It was observed that the material T7 exhibits only a single
nematic}smectic A (S 1, monolayer) transition in which the S 1 period is commensurate with the
1 A A
molecular length ¸ whereas T8,

C H O-/-COO-/-CH"CH-/-CN
8 17
 S. Singh / Physics Reports 324 (2000) 107}269 209

exhibits [322,323] with decreasing temperature, the double-re-entrance sequence

ILPN PS $PN PS 1Psolid . (6.5)
6 A R A
Here / represents a benzene ring and S $ period is incommensurate with dK1.2¸. It has been
A
observed that in the re-entrant nematic phase of T8, S $ and S 1 #uctuations are essentially
A A
independent; the S $ #uctuations change over from being S -like to S -like with decreasing
A A c
temperature in the N phase.
R
Probably, the most spectacular case of multiple reentrance has been observed [319,324,325] in
the material 4-nonyloxypheny1-4@-nitrobenzoyloxybenzoalte (DB90NO ) with molecular structure
2
[326],
C H O-/-OOC-/-OOC-/-NO .
9 19 2
This compound exhibits three nematic, four smectic A and two smectic C phases in the following
sequence, on cooling:
ILPN PS $PN PS $PN PS 1PS M PS 2PS 2Psolid . (6.6)
6 A R A R A C A C
Here S 2 is bilayer and S M and S 2 are two di!erent forms of smectic C phase [326]. This unusual
A C C
behaviour was rationalized in terms of the gradual cross-over from dominant S $ to dominant
A
S 1 #uctuations with decreasing temperature.
A

6.3. Theories for the RPT

From the molecular point of view, only approximate qualitative explanations of the re-entrant
behaviour have been possible. As discussed above, the liquid crystal systems exhibiting re-entrance
consist of organic molecules usually with three or four aromatic rings with ester linkages and
having polar cyano or nitro-end groups. Apart from pure compounds, re-entrant polymorphism,
on cooling, has also been shown by binary mixtures of polar}polar, polar}non-polar and non-
polar}non-polor compounds. Further, as a homologous series is ascended, the re-entrant phase
sequence is exhibited by the higher homologs which are neither very short nor very long. These
experimental observations indicate that the dipolar force plays a crucial role in the re-entrant
polymorphism. However, the observation of single re-entrance [316,317] in a binary mixture of
non-polar compounds probably cannot be caused because of dipolar forces. Reviews of theories
and experiments for the re-entrant polymorphism are available [303,327,328] which present an
account of the subject. The "rst tentative model for the nematic re-entrance was proposed by
Cladis et al. [285]. A number of other elaborate models [304,329}336] have been proposed which
emphasize the role of attractive forces and/or hard core repulsions to exhibit the phase sequence
and assume some sort of bimolecular organisation (dimers) or even trimers or n-mers [333] with
antiparallel association that compensates (not always fully) the dipole moments. These theories
essentially show that the high-temperature smectic phase is an induced phase and the re-entrant
nematic phase is brought about by a competition between two incommensurate lengths. This two
length concept is one of the main ingredient of the Landau theory of the re-entrant phase sequence
[230,337,338]. The frustrated spin gas model [333] is able to show the multiple re-entrance and
the sensitive dependence of re-entrance on the molecular chain length and pressure. In case of
210 S. Singh / Physics Reports 324 (2000) 107}269

single re-entrance observed in a binary mixture of non-polar compound, the experimental results
[317] on the layering thickness give no indication in favour of the kind of molecular organization
as found in re-entrant systems with polar compounds. A lattice model was proposed by Dowell
[138,249,315] for the re-entrant phase sequence of non-polar system. The Dowell model holds the
change in chain con"guration responsible for the re-entrance. It was shown by Bose et al. [331]
that the re-entrant phenomenon is built in the McMillan model if one explicitly incorporates the
e!ect of tail chain conformations in the molecular potential instead of treating the chains as an
extension or rigid part. This model is based on a molecular-"eld approach and predicts the
single-re-entrant phase sequence with lowering of temperature in an idealized non-polar system.
The Cladis model [284,285] utilizes the fact that the re-entrance appears only in compounds,
exhibiting layered phases, of amphiphilic molecules having both a polar (aromatic rings with polar
heads) and a non-polar (alkyl or alkoxy chains) part. The two parts of the molecule are assumed to
be immiscible. The molecules move freely within each layer and with di$culty between the layers.
There exist no correlation between molecules in di!erent layers. Usually, in most of the liquid
crystals, the polar segments of the molecules occupy a middle position with hydrocarbon chains
extending outward (Fig. 15). These kinds of con"gurations prefer to form a monolayer smectic
A phase. In case of re-entrance, owing to antiparallel correlations, the molecules form dimers which
are assumed to be somewhat bulgy in the middle (Fig. 15). As a result, bilayer, but incommensurate,
smectic A phases are probable. Once the smectic S $ phase is formed the bulgy parts are lined up in
A
a plane, whereas the alkyl chains cannot "ll the rest of the space. With decreasing temperature (i.e.
promoting the dimer formation) and also possibly with the sti!ening of the end chains, the packing
becomes so unfavourable that the smectic A phase becomes unstable and nematic re-entrance
appears. In order to explain the re-entrance at elevated pressure the Cladis model assumes that the
stability of bilayer smectic A phase is because of polar}polar (long-range electrostatic) and
non-polar non-polar (short-range) interactions and that the layer spacing extends with increasing
pressure. The increase in layer spacing may be due to slight compacting of the #exible non-polar
part. This weakens the e$ciency of the non-polar segments to hold the layer together and with

Fig. 15. Schematic representation of a bilayer smectic-A phase (a) 1 atm. and (b) under pressure.
 S. Singh / Physics Reports 324 (2000) 107}269 211

pressure the repulsive interaction of the aromatic rings increases and pushes the layers apart, if the
pressure is increased further (and temperature decreased) a transition takes place between less
dense smectic and more dense nematic which is the characteristic of the high-pressure side of the
re-entrant phase diagrams (Fig. 12). The forces stabilizing the layers are the short-range attractive
hydrocarbon (non-polar}non-polar) interactions. These forces are proportional to the length of the
hydrocarbon chains. The increasing repulsive interaction of the aromatic rings with increasing
pressure drives the layers apart.
The Luckhurst}Timimi model [329] of re-entrant polymorphism is a simple extension of the
McMillan theory [234] of the smectic-A liquid crystals in which the scaled strength parameters
related with the stability of smectic-A phase are allowed to vary with temperature. This theory has
been successful in explaining the experimental results, as described in Section 6.1 for the
6 OCB/80CB mixture. However, in order to show re-entrance they had to assume that the strength
parameter associated with mixed order parameter is linearly temperature dependent over the
region of interest. Dong [307] presented a detailed comparison of this approach with the
phenomenological Landau-type theory.
As discussed above within both the Landau and microscopic mean-"eld theories, re-entrance
appears only if either the order parameter of interest is coupled to an additional degrees of freedom
or the Hamiltonian is explicitly temperature-dependent. It was shown by Katriel and Kventsel
[236,339] that the re-entrant nematic and smectic phases are exhibited by a fully self-consistent
treatment of a simpli"ed McMillan's type model involving coupled nematic and smectic order
parameters without any explicit temperature-dependent factors. A re-entrant isotropic phase was
shown to be exhibited by explicit temperature-dependent Hamiltonian as well as by a two-state
model. The molecular theory of Luckhurst and Timimi [329] involves three coupled equations in
the orientational, positional and mixed-order parameters. The simpli"cation proposed by Katriel
and Kventsel [236] involves the decoupling of mixed order parameter (Eq (5.26)) as a product of
the pure orientational and positional order parameters. It was shown that the Hamiltonian so
obtained by the decoupled model exhibits reentrance for appropriate choices of model parameters.
The most successful microscopic theory is probably the `frustrated spin gas modela [333}335]
which invokes the possible dipolar frustration of the molecules in re-entrant systems. The dipolar
interactions, under the molecular close packing conditions, may or may not be frustrated,
depending on the positional #uctuations. Thus, the theory incorporates the coupled degrees of
freedom of dipolar orientations and molecular positions. The idea is that when in the plane normal
to the average molecular axis, frustration is (is not) lifted by the positional #uctuations normal to
the plane, the smectic (nematic) phase is realized. These molecular positional #uctuations normal to
the plane are called permeation #uctuations. The calculations with this `frustrated spin gas modela
have been able to reproduce remarkably the single- as well as double- re-entrance sequences and
also the sensitive dependence of re-entrant polymorphism on the molecular chain length and
pressure. The basic idea of microscopic mechanism involved here is shown in Fig. 16. On
a two-dimensional triangular lattice triplets of molecules are considered. The molecules are taken
to have an aliphatic tail on a polar head along the molecular axis. We begin by considering the limit
of complete positional order [Fig. 16a]. The close packing arrangement in two dimensions is
triangular. The interaction between the dipoles is antiferroelectric. Since each elementary triangle
of the array is frustrated an antiferroelectric long-range order cannot be supported. In case the
local distribution is like Fig. 16b, with one weak and two strong bonds, frustration will be lifted,
212 S. Singh / Physics Reports 324 (2000) 107}269

Fig. 16. Positional con"guration of a layer normal to the molecular axes. Molecular dipoles which are frustrated are,
respectively, indicated by crosses, closed circles or question marks. Strong and weak bonds are shown with full and
dotted lines, respectively.

and two dimensional antiferroelectric order can propagate across the unit. If the arrangements is
like in Fig. 16c, with one strong and two weak bonds, frustration will still, persist, which will
destroy the possibility of the existence of anti"erroelectric order. However, in any actual local
distribution, each elementary triangle is between these two cases, one strong, one intermediate, and
one weak bond [Fig. 16d]. The ordering character of the unit can be determined depending on the
distance between the intermediate bond and the strong bond or the weak bond. If there are enough
of the former type of units, they will percolate across the system and form an in"nite network
(`polymera) of positionally disordered, but antiferroelectrically ordered molecules [Fig. 16e]. Each
layer, consecutively along the Z-direction, will have its own network. These networks will not pass
through each other because this would involve disrupting in"nitely many strong bonds. This leads
to the density modulation along the Z-axis, i.e., the smectic phase. Two points are worth
mentioning here. First, even in the presence of the network, many antiferroelectrically ordered but
"nite clusters slide up and down the Z-axis and provide the constant background to the smectic
modulation. Second, although the percolating network must be sustained for the smectic modula-
tion, individual molecules join it and leave it as time progresses. Finally, as the system is further
cooled, it may be possible that local positional order will set in turning on frustration, eventually
destroying the network, and re-entering the nematic phase.
The microscopic pair potential through which the molecules interact is assumed to be of dipolar
type

As( ) s( 3B(s( ) r( )(s( ) r( )

;(r , s( , r , s( )" 1 2! 1 12 2 12 . (6.7)
1 1 2 2 r3 r5
12 12
For the purely dipolar interaction A"B; s( and s( are unit vectors along the Z-direction
1 2
perpendicular to the two-dimensional lattice and r is the position of the dipolar head of the
*
molecule i. When A'B steric hindrance wins and if A(B the van der Waals or amphiphilic
 S. Singh / Physics Reports 324 (2000) 107}269 213

attraction wins. The important positional #uctuations for the re-entrant systems were found to be
permeation #uctuations, occurring along the Z-direction, which is parallel to the average molecu-
lar axis and normal to the smectic layers. Since for a given molecule the molecular architecture may
not be smooth, a nearest-neighbour molecule has n positions of preferred relative permeation.
These `notchesa are taken to be separated by l/n, where l is the e!ective molecular length.
The nematic}smectic A phase boundaries can be approximated in the following way. A special
prefacing transformation is used to evaluate the e!ect of the positional #uctuations on the
dipole}dipole interactions. In a "nite-cluster approximation, a triplet of nearest-neighbour dipoles
is considered. With one dipole "xed, the partition function is summed over the n2 permeational
con"gurations of the two other dipoles. The result is a triplet spin Hamiltonian, into which have
been projected the average strongest, intermediate, and weakest antiferroelectric couplings,
exp (H #G),exp(K s s #K s s #K s s #G)
123 S12 I23 W31
,+ exp [!b;(r , s( , r , s( )!b;(r , s( , r , s( )!b;(r , s( , r , s( ] , (6.8)
1 1 2 2 2 2 3 3 3 3 1 1
r2,3
where the three molecules are labeled 1,2 and 3 such that (12), (23) and (31) always span the
strongest, intermediate, and weakest antiferroelectric couplings, respectively. K , K and K cor-
S I W
respond to the strongest, intermediate, and weakest couplings, speci"c to each positional con"g-
uration. G is the free energy contribution of the degrees of freedom summed over in the prefacing
transformation. The result is a distorted triangle of Ising spins. The nematic}smectic phase
boundaries are located with Houtappel's condition [340]
sinh(2K )sinh(2K )#sinh(2K )sinh(2K )#sinh(2K )sinh(2K )'1 . (6.9)
S S I 8 8 S
This condition corresponds to a two-dimensional distorted triangular Isign model. If both K and
S
K are positive, in-plane ferromagnetic order sets in. This corresponds to Ss( , s( T+1. This phase
I 1 2
was identi"ed [3](b), as S 1. If K ) K (0; Ss( !s( )T(0, local antiferroelectric order sets in, which
A 4 I 1 2
describes the S $ phase. Since condition (6.9) corresponds to a two-dimensional system, the
A
transitions are, of course, strongly displaced towards low temperatures as compared to what they
should be in a three-dimensional system. This approach reproduces satisfactorily the double
re-entrance as well as the multiple re-entrance. In its present form, although, the spin gas model
cannot describe the S S transition and the more ordered phases, but it is able to predict smectics
A C
with di!erent thicknesses. Madhusudana and Rajan [336] have made an interesting remark that
whereas dipolar interactions favour antiparallel short-range order, induction (dipole-induced
dipole) interactions favour parallel short-range order. The interplay between these two tendencies
may also explain the existence of two natural lengths and re-entrances.
The above procedure can be extended [335] to the calculation of speci"c heat. After the
prefacing transformation, the free-energy per molecule can be expressed as
f"G#ln j(K , K , K ) (6.10)
S I 8
with

C A BD
1
ln j(K #K #K )" lim ln + exp + H . (6.11)
S I W N ijk
N?= @S@ WijkX
214 S. Singh / Physics Reports 324 (2000) 107}269

Here the last sum is over the N up-triangle formed in a plane of N-spins. The speci"c heat per
molecule at constant pressure is given by
L L
C" i ¹2 f. (6.12)
L¹ B L¹
This is evaluated with the chain rule, requiring the "rst and second temperature derivatives of
G and K . j can be evaluated by using Houtappel's exact expression
a

P P
2p 2p
8p2ln(j/2)" du du ln[cosh(2K )cosh(2K )cosh(2K )
1 2 4 I W
0 0
#sinh(2K )sinh(2K )sinh(2K )!sinh(2K ) cos u !sinh(2K ) cos u
S I W S 1 I 2
!sinh(2K )cos(u #u )] . (6.13)
W 1 2
The interest here is in the relative magnitude of the speci"c-heat signals at the di!erent phase
transitions, which are strongly in#uenced by the functions G(¹) and K (¹) of the prefacing
a
transformation. In the calculation it has been found [335] that the `transition enthalpya is much
larger for the S 1-to-N transistion than for the N -to-S $ transition.
A R R A
Dowell [138,249,315] proposed a general lattice model for condensed phases with orientational
and/or partial or total positional ordering of the molecules in the system. The positional ordering
can be in one, two, or three dimensions. Each molecule is composed of a rigid core and one or two
semi#exible tails and has site}site hard repulsive intermolecular interactions. Let each molecule
have r rigid core segments, sf semi#exible tail segments, (r!1) rigid bonds, and sf semi#exible tail
bonds. The con"gurational partition function Q for a whole system of molecules in this model is
C
given by

C D
E
Q "X exp ! # ,
C K ¹
B
where X is the total number of distinguishable ways to arrange the molecules in the system and
E is the average intermolecular energetic contribution to the Q . In"nitely hard repulsions are
C C
implicitly included in X as the two molecular segments are not allowed to occupy the same lattice
site. E includes attractions and soft repulsions between intermolecular segments. In case the
C
molecules have hard repulsions only, E "0 and Q is equal to zero. In order to calculate the
C C
general partition function X, the molecules are placed on the lattice, molecule by molecule, segment
by segment. The general relation for the partition function was derived by Dowell [249] and
applied [138](c) to the special cases of smectic-A and nematic liquid crystals and isotropic liquids
in bulk phases. The relative stabilities of the isotropic, nematic, smectic-A and re-entrant nematic
phases were studied as a function of temperature, pressure, tail #exibility, and tail length. The role
of the semi#exible tails in stabilizing the smectic-A and re-entrant nematic phases was explicitly
excluded.
In this model [138,249] the molecules are assumed to interact via site}site (segmental) hard
repulsions. The lengths of the core and the tails can be varied as well as the #exibility of the
intramolecular bonds in the tails. The model holds the change in chain con"guration responsible
 S. Singh / Physics Reports 324 (2000) 107}269 215

for the re-entrance in a single-component non-polar system. It is seen that a segregated packing of
cores besides cores and chains beside chains occurs with a lowering of temperature leading to
a usual smectic phase. With further lowering of temperature, the chain becomes less #exible, and
the packing di!erences between the rigid cores and tail chains decrease. Thus, the need for the
segregated packing of rigid cores with cores (and tail chains with tail chains) is overcome by the
entropy of unsegregated packing, leading to a disappearance of the re-entrant nematic.

7. Uniaxial nematic}biaxial nematic (NuNb) phase transition

As indicated in Section 1.2.1, in the N phase the molecules tend to align along the director n( .
6
A further breaking of rotational symmetry around the director n( may lead to the formation of
a biaxial nematic (N ) phase. The existence of this phase was predicted on a theoretical basis by
"
Freiser [38]. He showed that the simplest generation of interaction employed in the Maier}Saupe
(MS) theory leads to a "rst-order IN transition followed, at lower temperature, by a second-order
transition to a biaxial state. Considering a hard-plate lattice model, Shih and Alben [349] showed
that the system composed of rectangular plates which are neither very square nor very rod-like in
shape may exhibit N phase at high pressure. At some lower pressure the N phase undergoes
" "
a second-order phase transition to a N phase which at a still lower pressure exhibits a "rst-order
6
NI transition. The hard-plate lattice model was solved by Alben [39] within the framework of
mean-"eld approximation and a number of interesting results were obtained. For example, the
introduction of plate-like molecules increases the NI transition temperature of rod-like molecules.
At a lower temperature, the N phase can undergo a second-order transition to a more highly
6
ordered N phase. Between the two regions N` (positive optical anisotropy) and N~ (negative
" 6 6
optical anisotropy), Alben's calculation showed that the two second-order lines between the N and
6
N phases form a sharp cusp separating the rod-like nematic phase N` and plate-like nematic
" 6
phase N~ and that the cusp touches the "rst-order isotropic}uniaxial nematic transition line. The
6
intersection of the two second-order lines and the "rst-order transition line forms a special critical
point. From the Alben's phase diagram it is di"cult to analyse the details of the thermodynamic
and critical behaviour in the region between N` and N~ transitions.
6 6
The experimental discovery of the long-predicted N phase has been by Saupe and co-workers
"
[40,343,344] and others [345,346]. These authors studied the phase diagram and critical properties
of the ternary system potassium laurate-1-decanol-D O over concentration ranges where nematic
2
phases are likely to occur (Fig. 17). It was shown that in the limited concentration range the
observed phase sequences on heating/cooling may be: isotropic, uniaxial nematic (N`), biaxial
6
nematic (N ) and uniaxial nematic (N~). Thus, an intermediate N phase is formed for a certain
" 6 "
concentration range while in other ranges a direct "rst-order N`}N~ transition occurs. The
6 6
N`}N or N~}N transition appears to be second order. Experimentally, transitions have been
6 " 6 "
observed that seem to lead directly from N`}N~ via a "rst-order transition. In certain micellar
6 6
nematics the phase diagrams suggest the existence of a Landau point on the nematic}isotropic
transition line. The nematic phase formed can be positive or negative uniaxial and even biaxial,
depending on the shape of the micelles. The phase diagram of the mixture of rod-like and plate-like
molecules of comparable size and in comparable amounts also indicates the existence of a Landau
point. Evidence has been found that the mixture undergoes a transition to two coaxial uniaxial
216 S. Singh / Physics Reports 324 (2000) 107}269

Fig. 17. Phase diagram of the potassium laurate/1-decanol/D O system [40]. L is the approximate location of the
2
Landau point.

phases, rather than to a single biaxial phase. The suggestion was made by Chandrasekhar [347]
that a thermotropic N phase can be prepared by bridging the gap between rod- and disc-like
"
mesogens, i.e., by synthesizing a mesogen that combines the features of the rod and the disc. This
has proved to be useful and the N phase has been observed in some relatively simple compounds
"
[41,42].
A number of important ideas concerning the N phase have been discussed theoretically
"
[341,342,92,347}360]. These theoretical investigations show that an isolated critical point is
obtained in the phase diagram, where the N }N` and N }N~ phase boundaries meet a "rst-order
" 6 " 6
N`}N~ line. At this isolated critical point the cubic coe$cient of the order parameter in the
6 6
e!ective Hamiltonian becomes zero.
In order to construct within the framework of the LDG theory the simplest model for the both
uniaxial and biaxial phases, one must retain the following terms in the free-energy expansion (4.10):
f"1A Tr (Q2)#1B Tr (Q3)#1C[Tr Q2]2#E@[Tr Q3]2 . (7.1)
2 3 4
For stability it is required that C'0 and E@'0. Instead of considering only A as the controllable
variable, a phase diagram has to be constructed as a function of A and B. Using Eq. (2.19) the
free-energy expansion becomes [64]
f (x, y)"a (x)#a (x)y2#a (x)y4 , (7.2)
0 2 4
 S. Singh / Physics Reports 324 (2000) 107}269 217

where
9 9
a (x)"3Ax2#1Bx3# Cx4# E@x6 , (7.3a)
0 4 4 16 16

a (x)"1A!1Bx#3Cx2!9E@x4 , (7.3b)
2 4 4 8 8
a (x)"1C#9E@x2 . (7.3c)
4 8 8
Minimization with respect to y gives

y"0, uniaxial (7.4a)

y"!a (x)/a (x), biaxial (7.4b)

2 4
f (x)"a (x), uniaxial (7.4c)
6 0
and

f (x)"a (x)!a2(x)/2a (x) . (7.4d)

" 0 2 4
Since always a (x)'0, the biaxial solution (7.4b) is only allowed when a (x)(0. Further,
2 4
whenever the biaxial solution is allowed, f (x)(f (x). The resulting phase diagram is shown in Fig.
" 6
18. It can be seen that apart from the usual "rst-order NI transition (solid line) two second-order
N`}N and N }N~ transition (broken lines) occur. The biaxial phase is sandwiched between
6 " " 6
N` and N~ phases. All four phases meet in a bicritical point, called the Landau point and located
6 6

Fig. 18 . Phase diagram with a biaxial nematic phase obtained from the free-energy (7.1), with C"2.67, E@"3.56 [64].
Solid lines represent phase transition of "rst-order and dashed lines second-order. A is a measure for the temperature, b is
the degree of #atness of the molecules. L is the Landau point.
218 S. Singh / Physics Reports 324 (2000) 107}269

at A"B"0. The NI transition is second-order here. The isotropic}nematic line of phase

transition is given by
A"3Cx2#9E@x4 (7.5a)
4 4
and
B"9Cx#18E@x3 . (7.5b)
2
The uniaxial nematic}biaxial nematic transition de"ned by
A"!3Cx2 , (7.6a)
2
B"!9E@x3 (7.6b)
2
indicates that the width of the biaxial region is proportional to (¹H}¹)3@2.
The phase diagram without E@ term (Fig. 19) shows that the width of the biaxial region shrinks to
zero leaving a "rst-order N`}N~ transition. The topology of the phase diagram in the vicinity of
6 6
the Landau point does not change even if the free-energy expansion is carried out to higher orders.
The critical exponents for the Landau point using renormalization group method have been
calculated by several authors [82,350].
In principle, following possibilities may lead to the spontaneous formation of a one-component
biaxial nematic phase:
(i) A molecular symmetry that is not (e!ectively) uniaxial.
(ii) Strong correlation of molecules leading to aggregates of molecules having no uniaxial sym-
metry.
(iii) Application of an external "eld.

Fig. 19. Phase diagram without biaxial nematic phase, obtained from the free-energy (7.1), with C"2.67, E@"0 [64].
Symbols are the same as in Fig. 18.
 S. Singh / Physics Reports 324 (2000) 107}269 219

Associated with the second possibility a N phase has been observed in lyotropic systems of
"
amphiphilic (soap-like) solutions in water. In these solutions, the molecules tend to cluster into
aggregates so that the hydrophilic groups occupy the surface to optimize contact with water, while
the lipophilic tails occupy the inert part of the aggregates [361]. These so-called micelles behave
somewhat like large-scale molecules such that under a suitable choice of temperature, concentra-
tion and sometimes additional solvents they tend to exhibit smetic or nematic structure. Depending
on the shape of the micelles the nematic phase can be positive or negative uniaxial and even biaxial,
since the shape changes with concentration. The complete phase diagram of the potassium laurate/
1-decanol/D O system is shown in Fig. 17. In case of one-component thermotropic systems,
2
theoretical studies [38,358] indicate that for systems with molecules without axial symmetry,
a N phase at lower temperature is necessary rather than just possible. However, in an experiment
"
the system may form smectic and crystalline phases before this temperature is reached. Hence, to
form the N phase the uniaxial nematic temperature range should be reduced, possibly by lowering
"
the value of DBD. One possible way to circumvent these problems is to consider another system
exhibiting N phase: a mixture of rod-like and plate-like molecules of comparable size and in
"
comparable amount [39,351].
In order to analyse the in#uence of external "elds on the N phase, a magnetic "eld term is
"
included in the free-energy expansion (7.1). With the "eld along the Z-axis, a term-hx is added to
a (x) in Eq. (7.3a). The resulting phase diagram is shown in Fig. 20. Here the biaxiality is partly
0
spontaneous (large) and partially induced. In the dashed area the spontaneous biaxiality is
predominant. Although the two biaxial regions are clearly distinguishable, no sharp boundary

Fig. 20. Phase diagram of Fig. 18 modi"ed by a magnetic "eld h"0.0018 [64]. In the cross-hatched area spontaneous
biaxiality is predominant.
220 S. Singh / Physics Reports 324 (2000) 107}269

exists between them. The second-order N`}N transition is given by [64]

6 "
4h 3
A" ! Cx2 , (7.7a)
9x 2

4 h 9
B" ! E@x3 , (7.7b)
9 x2 2

x4x . (7.7c)
5#1
At the tricritical point x"x with
5#1
729 16 16
CE@x8 #12hE@x5 # hCx3 ! h2"0 (7.8)
4 5#1 5#1 3 5#1 81

when x'x , Eqs. (7.7a), (7.7b) and (7.7c) gives the superheating limit of the N phase. The isolated
5#1 "
critical point corresponds to the equation

4 3 27
A" h/x# Cx2# E@x4 , (7.9a)
3 2 4

4 h
B"! !6Cx!18E@x3 , (7.9b)
3 x2

x"x (7.9c)
#1
with x given by
#1
!h#9Cx3 #81E@x5 "0 . (7.10)
#1 #1
Here x(x corresponds to the supercooling limit of the paranematic phase, whereas x'x to
#1 #1
the superheating limit of the nematic phase.
Very recently, Mukherjee [342] has constructed a free-energy expansion to reestablish the above
predictions and to calculate the temperature dependence of order parameter and thermodynamic
quantities. The accurate measurements of all the order parameters and of the speci"c heat in the
neighbourhood of the biaxial transition temperature are still lacking. Mukherjee's work provides
theoretical calculations for these quantities and the phase diagram obtained [342] is slightly
di!erent from the phase diagram of earlier works [38,39,51]. In Mukherjee's phase diagram the
actual stability limit of all the phases is indicated and the temperature dependence of the order
parameters are calculated in all the phases; this is absent in Alben's work.
It is important to mention here that the experimental observations should resolve whether
transitions such as N`}N , N~}N , I}N , etc., actually occur. At the present time it seems that all
6 " 6 " "
known uniaxial phases are of type N`, while only in few materials the N has been observed. Thus,
6 "
for answering the questions related to the uniaxial nematic}biaxial nematic phase transition the
available data are yet insu$cient and there is a need to generate experimental data on the biaxial
nematic phase.
 S. Singh / Physics Reports 324 (2000) 107}269 221

8. Phase transitions involving smectic phases

This section covers the smectic A to smectic C transition and other transitions involving smectics
and the N}S }S multicritical point. Only a brief summary is presented. Transitions involving
A C
bilayer smectic phases are given elsewhere [3,390].

8.1. Smectic A}smectic C transition

The S }S transition is characterized by the onset of a tilt in a one-dimensional layered matrix.

A C
A number of experiments [3] have been carried out using a variety of techniques (calorimetric,
optical, X-ray scattering, ESR, NMR, neutron scattering, etc.) which con"rmed the earlier predic-
tion by Taylor et al. [362] that the S }S transition is continuous. Many of these experiments,
A C
however, probably do not reliably obtain the asymptotic behaviour near the transition. The critical
#uctuations for the S }S transition are those in the molecular tilt. Most of these transitions show
A C
mean-"eld behaviour for the same reason as superconductors do; the intrinsic coherence lengths
are long and reduce the #uctuations until very close to ¹ . The mean-"eld value of the critical
AC
exponent b characterizing the temperature dependence of the tilt angle in the S phase was clearly
C
shown. A second-order S }S transition in a 3D space has been discussed by Kats and Lebedav
A C
[363] using renormalization group technique.
For a given set of layers, to describe the smectic C order, one must specify the magnitude u of the
tilt angle, and also the azimuthal direction of tilt, speci"ed by an angle / (see Fig. 21). The tilt angle
u of the molecules can point in any azimuthal direction; there are thus two independent components:
u "u cos / and u "u sin /. Thus, due to the azimuthal degeneracy of the tilt angle one can use
x y
a complex order parameter t ("u exp(i/)) to describe the transition. If one neglects the coupling
1
to the layers' undulation mode, a remarkable analogy is seen with super#uid helium. An overall
change of the phase / does not modify the free energy. This leads to the following possibilities:
The S }S transition may be continuous. The speci"c heat should show a singularity [3,382].
C A
Below ¹ the tilt angle should obey the law
CA
u"const DtDb (8.1)

Fig. 21. Smectic C tilt angle: i is the layer normal, u the tilt amplitude, / the azimuthal angle.
222 S. Singh / Physics Reports 324 (2000) 107}269

with bK0.35 and t"(¹}¹ )/¹ . Above ¹ the application of a magnetic "eld that is oblique
CA CA CA
with respect to the optic axis of the smectic phase induces the tilt angle
sHH
u"C a x z t~c , (8.2)
1 k ¹
B CA
where cK1.33 is the susceptibility critical exponent. The predicted tilt is small (about 10~2 rad for
tK10~4 near room temperature) because the diamagnetic energy s H2 is very weak as compared
!
to k ¹ . Starting from S phase and decreasing ¹ towards ¹ , one expects to observe the onset
B CA A CA
of a strong (depolarized) light scattering due to #uctuations in the tilt angle.
The application of Landau theory to the S }S transition was "rst considered by de Gennes
A C
[364], who treated the tilt angle as the relevant order parameter. It has two components: the
magnitude DuD and the azimuthal angle /. This shows an analogy with the super#uid-normal #uid
transition. So as usual the free-energy density of S phase can be expressed as
C
f "ADuD2#1DDuD4#1EDuD6 . (8.3)
C 2 3
The total free energy involves the order parameter due to the density wave, Dt D, and its coupling
1
with the orientational order parameter Q. It has been shown [365], from an analysis of the speci"c
heat anomaly near the second-order S }S transition, that the sixth-order term in Eq. (8.3) is
A C
unusually important.
Several theories for the S phase were developed taking into account the speci"c features of the
C
molecules. Wulf [366] proposed a steric model which considers, a zig-zag shape for the molecules
and assumes that the tilted structure results due to freezing of free rotation around the long axis. It
was shown [366] that the steric e!ects, in particular, the zig-zag gross shape of the smectogenic
molecules, may be able to account for the second-order S }S phase transition. In S phase the
A C C
biaxial order parameters play a primary role and may grow to values of 0(10~1). McMillan [367]
took note of the molecular transverse dipoles which are present in all the smectogenic compounds
and developed a model for the S }S transition in which the molecules rotate freely about their
A C
long axes in the S phase, leading to the dipolar ordering. Incorporating the dipole}dipole
A
interaction of the permanent molecular dipole moments the theory [367] predicts three orienta-
tionally ordered phases; (1) with the physical properties of the S phase (tilted director, optically
C
biaxial, second-order S }S phase transition), (2) a two-dimensional ferroelectric, and (3) a low-
A C
temperature ordered phase which is both tilted and ferroelectric. If there are two oppositely
oriented net dipoles directed away from the geometric centre of the molecules, the medium will not
exhibit ferroelectric properties. The presence of the two dipoles with longitudinal components also
favours the tilting of molecules in the layers to minimize the energy of the oriented dipoles.
However, NMR and other experiments clearly show that the molecules are practically freely
rotating about their long axes in the S phase in disagreement with the assumptions of both Wulf
C
and McMillan. Cabib and Benguigui [368] made an attempt to overcome this problem by
postulating a special type of molecular structure in which only the longitudinal components of two
symmetrically placed outboard dipoles are e!ective. However, most of the mesogenic compounds
do not have such a structure. Based on the argument that many compounds have only one overall
dipole moment, Van der Meer and Vertogen [369] proposed that in such cases the tilting optimizes
the attractive energy due to dipole-induced dipole interaction. This interaction remains e!ective
even if the molecules rotate freely. Priest [370] developed a model for the nematic, smectic A and
 S. Singh / Physics Reports 324 (2000) 107}269 223

smectic C phases in which the intermolecular interactions are characterized by second-rank tensor
quantities and are supposed to produce orientational order of the molecules. The model predicts
that free rotation is possible for uniaxial molecules and that the extent of this rotation being
hindered depends on the degree of biaxiality in certain intrinsic molecular second-rank tensors.
According to the model, the uniaxial molecules can exhibit S state. The model predicts a second-
C
order S }S transition with two independent degrees of freedom having divergent #uctuations.
A C
With the decreasing temperature the tilt angle of S phase is predicted to grow continuously from
C
zero and to asymptotically saturate at 49.13. However, the model fails to identify the physical origin
of the proposed interactions. The model considered by Matsushita [371] incorporates the excluded
volume e!ects due to the hard-rod features of the molecules which actually favour the S phase. If
A
the mesogenic molecules are uniaxial these models do not give rise to a biaxial order parameter in
the S phase. They describe only a tilted S phase rather than the S phase with its intrinsically
C A C
biaxial symmetry. Goossens [372] has shown that none of the above models using attractive
interactions are satisfactory. In these molecular models the tilt angle does not appear as a natural
order parameter. Only in the Wulf and McMillan models the S phase is characterized by new
C
order parameters. But they are unsatisfactory because of the stringent requirement that rotations
about the long axes of the molecules should be frozen out. A detailed calculation of the inter-
molecular interactions between molecules of ellipsoidal shape has been carried out by Goossens
[372]. The attractive potential arising from the anisotropic dispersion energy and permanent
quadrupole moments has been considered. The mean-"eld potential depends strongly on the
anisotropy of the excluded volume. It has been shown that the relative weights of the three terms
proportional to Scos qzT2, Scos(qz)p (cos h)T and Scos qzT Scos qz p (cos h)T depend on the ¸/d
2 2 0
ratio of the ellipsoidal molecules. Further, from the calculations for a tilted director, about which
the orientational order is still considered to be uniaxial, it was shown that the extra mean-"eld
energy contains a term proportional to sin2 u. The angle of tilt is proportional to a new order
parameter which goes to zero at a temperature which can be identi"ed with ¹ . Very recently,
AC
Gie{elmann and Zugenmaier [521] have developed a model for the S }S phase transition which
A C
is, in principle, analogous to a ferromagnetic phase transition with spin number sPR. Assuming
a bilinear mean-"eld potential, the macroscopic tilt angle has been calculated by Boltzmann
statistics as a thermal average of the molecular tilt. The calculation gives an equation of state for
the S phase which is self-consistent "eld equation involving the Langevin function of a reduced tilt
C
and a reduced temperature. An excellent agreement with the experimental results has been
obtained.

8.2. The nematic}smectic A-smectic C(NS S ) multicritical point

A C
The existence of NS S multicritical point was shown independently by Sigaud et al. [373] and
A C
Johnson et al. [374]. It is the point of intersection of the N}S , S }S and N}S phase boundaries
A A C C
in a thermodynamic plane, e.g., in the ¹}P or ¹}y (concentration) diagram. All three phase
transitions are continuous in the vicinity of it and at the point itself the three phases are
indistinguishable [375]. It has been observed in the ¹}y diagram of binary liquid crystal mixtures
[373,374]; for the P}¹ diagram of a single-component mesogenic material [376]. High-
resolution studies have been carried out [3,4] in both ¹}y and P}¹ planes and it is now accepted
that the topology of the phase diagram in the vicinity of NS S multicritical point exhibits
A C
224 S. Singh / Physics Reports 324 (2000) 107}269

Fig. 22. Topology of phase diagrams in the vicinity of the NS S multicritical point. (a) The temperature}concentration
A C
(¹}y) data for four binary liquid crystal systems [377]; (b) The pressure}temperature (P}¹) data for a single component
system [376].

universal behaviour (Fig. 22). It has been found that [376] the analysis of the phase boundaries
gives identical exponents for both the ¹}y and P}¹ diagrams showing the universal behaviour of
NS S point. Further, in the N}S case only two components of the mass density wave exhibit
A C A
large #uctuations near ¹ , whereas at the N}S transition `skeweda cybotactic groups (i.e. S type
C C C
#uctuations) are concentrated on two rings in the reciprocal space. Thus, the natural order
parameter has an in"nite number of components. In the description of the NS S point one should
A C
be free to move from one type of #uctuations to the other.
The phase boundaries obey simple power laws

¹ !¹ "A DX!X D(1#B(X!X ) , (8.4a)

NA NAC NA NAC NAC
¹ !¹ "A DX!X D(2#B(X!X ) , (8.4b)
NC NAC NC NAC NAC
¹ !¹ "A DX!X D(3#B(X!X ) . (8.4c)
AC NAC NC NAC NAC
Here X is either the pressure or the concentration. All experimental data are consistent with
/ "/ K0.57$0.03 and / K1.4}1.7. The N}S and N}S lines have the same slope parallel to
1 2 3 A C
the temperature axis at the NS S point, whereas that of S }S line is orthogonal. An alternative "t
A C A C
has been proposed by Anisimov [377] in which / "/ K0.67$0.03 and / K0.87$0.04. With
1 3 2
this "t, all phase boundaries are tangent and parallel to the temperature axis at the NS S point
A C
which can only be fortuitous and approximate.
 S. Singh / Physics Reports 324 (2000) 107}269 225

There have been several theoretical descriptions [365,378}381,230] on the NS S point. The
A C
framework of the Chen}Lubensky model [379] seems to correspond more closely to the experi-
ment. It is in some way a generalization of de Gennes model [364] of the N}S and N}S
A C
transition. However, the Landau theory is not able to shed any light on the physical origin of the
tilt. A simple model based on the quadrupolar nature of the molecules was proposed [131] which
made use of the idea that a gradient in the scalar orientational order parameter PM implies that of
2
the quadrupole density, and hence leads to an order electric polarization of the medium [380]. The
resulting dielectric self-energy due to associated order electric polarization is given by
f "1a2PM 2Dt D2(cos2 u@!1)2 , (8.5)
0% 2 2 1 3
where a is related to the quadrupole moment of each molecule and the dielectric constant along the
Z-axis, u@ the orientation of the principal axis of the quadrupole tensor of the medium with respect
to the direction of the gradient in the scalar order parameter PM . Eq. (8.5) will be minimized when
2
cos2 u@"1. This mechanism operates only in case of a layered structure, i.e., in the smectic phase.
3
The total free-energy density can be expressed as

A B A B
1 1 1 1 1 1 2
f" a(¹!¹ )Dt D2# DDt D4! C Dt D2 cos2 u@! # a2Dt D2PM 2 cos2 u@! .
2 AN 1 4 1 4 1 1 3 3 1 2 3
(8.6)
The relative stability of the S and S phases is determined by the ratio PM (¹)a2/C . As temperature
C A 2 1
decreases, PM (¹) increases. It has been found that the calculated a2/C versus (1!¹/¹ )1@2 phase
2 NI
diagram looks similar to the experimental one close to the NS S point.
A C
Grinstein and Toner [383] applied the renormalization group technique and considered the
application of a dislocation loop model to the NS S multicritical point and made a striking
A C
prediction: four, rather than three phases meet at the point where S }N and S }S phase
A A C
boundaries cross. A new, biaxial nematic phase was found to intervene between the nematic and
smectic C phases. It exhibits the orientational long-range order of the S phase, whereas the
C
translational properties are those of the nematic. However, the layer has only short-range
positional order. These four phases meet at a decoupled tetracritical point. This prediction is
supported by the #uctuation-corrected mean-"eld theory of Lubensky [357]. The high-resolution
speci"c heat measurements of Wen et al. [384] have shown anomalous variations near the S }N
C
transition very close to the NS S point. They suggested that these anomalous variations may be
A C
due to biaxial #uctuation.

8.3. Transitions involving hexatic smectic phases

With the discovery of hexatic phases a whole class of many transitions related to the hexatic
order became possible, for example,
S %S ; S %S ; S %S ; Cry%S ; S %S ; S %S ; S %S , etc .
BC3: B)%9 A B)%9 GC3: F I F I F C I C
Since there exists no symmetry di!erence between S , S and S phases, one can infer that either
C F I
a "rst-order phase transition or no transition at all between these phases may occur. It might be
226 S. Singh / Physics Reports 324 (2000) 107}269

interesting to look for critical points in these systems. Further, it has been observed [3] that the
S phase di!ers from the S phase by the existence of a sixfold modulation of the X-ray di!use
B)%9 A
scattering ring corresponding to in-plane molecular correlations. If the X-ray beam is incident
orthogonal to the smectic planes, the angular dependence of the maximum of the X-ray scattering
intensity may be expressed as
I(s)"I #I cos 6(s!/)#higher harmonic , (8.7)
0 6
where s is the angle between the in-plane component of the wave vector transfer q and an arbitrary
reference axis X (see "gure on p. 547, Ref. [3](b)). For S phase I "0 but for S phase I O0.
A 6 B)%9 6
Thus, a complex order parameter can be chosen to describe S phase
B)%9
t "I e6*( . (8.8)
6 6
The Landau free-energy expansion exactly similar to that of Eq. (5.13) can be written and it is
expected that the S }S transition may belong to the super#uid helium universality class [385].
A B)%9
Calorimetric measurements on the S }S transition show a critical exponent of the speci"c heat
A B)%9
aK0.6, which are inconsistent with any available theory but not too far from the tricritical value
0.5 [386]. In two dimensions, the S }S transition can be second order in a dislocation
B)%9 B#3:
unbinding picture, whereas in three dimensions, the transitions towards a crystalline phase are of
"rst order. The di!erence between two- and three-dimensional behaviour can be clearly seen in an
a.c. calorimetric study [387]. The existence of a continuous S }S transition is con"rmed by
A B)%9
these experiments [387,388]. Pleiner and Brand [389] observed anisotropic anomalies in the
damping and velocity of ultrasound near the S }S transition. Based on generalized hydrodyn-
A B)%9
amics these authors explained this anisotropy by an anisotropic, reversible dynamical coupling
between the bond orientational order parameter and elongational #ow. It was found that only
in-plane elongational #ow induces bond orientational order at the transition.

9. Computer simulations of phase transitions in liquid crystals

Computer simulations have played a key role in developing the understanding of phase
transitions and critical phenomena in liquid crystals [21,54,147}160,205}208,391}413]. A survey of
the existing numerical studies of N}I, N}S , S }S , S }N transitions and transitions to the
A A B A R
discotic phase is given in this section. The primary issues addressed in these simulation studies and
the main results obtained from them are summarized.
The models studied in numerical simulations of mesophase transitions may be classi"ed into two
categories called as molecular models and "eld models. Three broad classes of molecular models
used in liquid crystal simulations are Lebwohl}Lasher lattice models, hard particle models and
Gay}Berne models. In the molecular models, the description is in terms of the molecules constitut-
ing the system and their interactions. The interactions included in these models must describe both
translational and orientational orders. In reality, the mesogenic molecules consist of rigid cores and
#exible side chains. As a result, it is quite di$cult to construct a model that provides a realistic
description of the interaction between two such molecules. Even if one could construct a realistic
model for the intermolecular interactions, it would be extremely di$cult and time-consuming to
simulate the properties of liquid crystalline system. Owing to these reasons, the simulations are
 S. Singh / Physics Reports 324 (2000) 107}269 227

usually carried out for much simpler models in which the molecules are assumed to be simple
non-spherical rigid objects such as rotational ellipsoids, spherocylinders, parallel plate cylinders,
cut spheres, etc. The interaction between two such molecules is also assumed to have a simple form.
Most of the simulations carried out for such models assume only a hard-core repulsion between
two molecules [391], arising from the excluded volume interactions. A posteriori justi"cation for
the study of such models comes from the observation that these models do exhibit some of the
phases found in real liquid crystals. The principle of universality provides another justi"cation for
such studies which states that the parameters characterizing the critical behaviour near a continu-
ous phase transition do not depend on the microscopic details of the system and are determined by
a few factors such as the dimensionality of space and the symmetry of the order parameter. Few
attempts [392] have also been made to include in the simulation model the van der Waals
interaction and dipolar interactions in an approximate way. The "eld models provide a coarse-
grained description of the system in terms of an order parameter "eld appropriate for the ordered
state under consideration. A model of this kind is de"ned by a Ginzburg}Landau free-energy
expressed as a functional of the order parameter "eld. In some cases, due to symmetry or other
considerations, terms coupling the order parameter to other non-ordering "elds may have to be
included in the free-energy functional. A typical example is that of de Gennes model [62] for the
N}S transition. It is important to note that unlike molecular models, a "eld model is speci"c only
A
to the transition in which long-range order described by the particular order parameter "eld sets in
and it cannot be used to describe other phase transitions which may be exhibited by the same
physical system. The order parameter "eld appearing in models of this kind are continuous
functions of the spatial coordinates. The results obtained from the numerical studies of such models
can be compared directly with those obtained from the analytic calculations.

9.1. Lebwohl}Lasher model

The questions which have to be addressed in simulations of N}I transition are concerned with
the (i) determination of the minimal characteristics of molecules and their interactions exhibiting
nematic order and (ii) estimation of the magnitudes of the discontinuities shown by various
thermodynamic functions (e.g. order parameter, density, internal energy and entropy) at the
"rst-order transition in a 3D system. The "rst model used in the N}I simulation is the Leb-
wohl}Lasher model [147] which is the lattice version of the MS model of a nematic. It assumes that
each site of a simple cubic lattice is occupied by a classical unit vector. The Hamiltonian for the
model is de"ned as
H"!J+P (e( ) e( ) , (9.1)
2 i j
where the sum is over nearest-neighbour pairs of lattice sites and J'0 measures the strength of the
nematic coupling. Because the molecules are "xed on the sites of a lattice, tranlsational motion is
absent. The Monte Carlo (MC) simulation carried out by Lebwohl and Lasher [147] showed that
this model exhibits a strongly "rst-order N}I transition near J/k ¹"0.89. The order parameter at
B
the transition exhibits a jump from zero to about 0.33 while the transition entropy is very close to
1.09J. This model has been intensively investigated using MC technique [152,155,393}395].
Further numerical work on similar models of the I}N transition has been persued by others
[154,155]. The accuracy of Lebwohl and Lasher's results were improved by Jensen et al. [154] by
228 S. Singh / Physics Reports 324 (2000) 107}269

simulating the behaviour of same model on larger lattices. Luckhurst and Romano [155] have
carried out simulations of a system in which the molecules are not con"ned to a lattice. These
authors assumed that the molecules interact via a MS type anisotropic potential with Len-
nard}Jones-type distance dependence. The in#uence of an external "eld on the thermodynamic
behaviour of the Lebwohl}Lasher model was studied by Luckhurst et al. [155]. The results
obtained for the dependence of the order parameter and the internal energy on the value of the
external "eld were in qualitative agreement with the predictions of mean-"eld theory. Zhang and
co-workers [396,397] have simulated a simple cubic lattice with upto 283 sites. The free-energy
function allows the determination of the limits of stability of the nematic and isotropic phases.
Zhang et al. [396] estimated that these temperatures are within 5]10~4 reduced temperatures of
¹ which is in reasonable agreement with the experimental data.
NI
The Lebwohl and Lasher model has also been extended to model nematics in porous media such
as aerogels [403,404]. The Hamiltonian for the extended model is given by

H"!J + P (e( ) e( )!D+ P (e( ) n( ) , (9.2)

2 i j 2 i i
WijX i
where D is the strength of coupling to the random axis which mimics the local pore environment
selecting a preferred direction for the nematic order within the pore. Cleaver and co-workers [403]
studied this model for D/J"1 using MC techniques on lattices upto 643 in size. These authors did
not "nd any evidence for a "rst-order phase transition and their data are consistent with the
absence of nematic long-range order at low temperatures.

9.2. Hard-core models

The "rst computer simulations of anisotropic hard-core models were carried out by Vieillard-
Baron [148] in two dimensions. Frenkel and Mulder [156] carried out a systematic study of the
properties of a three-dimensional system of hard ellipsoids of revolution by using MC simulation.
These authors studied the behaviour of this system for values of x lying in the range between 3 and
0
1/3. The results show a "rst-order N}I transition with a density change of about 2% at the
transition only in the range x '2.5 or x (0.4. The phase diagram obtained is shown in Fig. 23.
0 0
In addition to the isotropic and nematic phases, they also found orientationally ordered and
disordered (plastic) crystalline phases in the high-density region of the phase diagram. An interest-
ing feature of the phase diagram is its approximate symmetry between oblate and prolate ellipsoids.
However, the values found by Frenkel and Mulder [156] appear to change for the larger systems
studied [405]. Allen and Frenkel [158] have observed for x "3 pretransitional nematic #uctu-
0
ations in the isotropic phase. A number of authors [157,406,407] have performed numerical
simulations of molecular models of two-dimensional nematics. Frenkel and Eppenga [157](b) have
carried out simulations to study the thermodynamics of a system of in"nitely thin hard needles in
two dimensions. They observed at high densities a stable nematic with algebraic decay of
orientational correlations. The behaviour of a system of two-dimensional ellipses with aspect ratios
2, 4 and 6 have been simulated by Cuesta and Frenkel [406]. For the aspect ratios 4 and 6, a stable
nematic phase with a power-law decay of orientational correlations was found. While the N}I
transition in the system with the aspect ratio 6 was found to be continuous, the system with aspect
 S. Singh / Physics Reports 324 (2000) 107}269 229

Fig. 23. Phase diagram of a HER system obtained by MC simulation [156](a). The reduced density oH is de"ned such
that the density of regular close packing is equal to J2 for all x . The shaded areas indicate two phase coexistence. The
0
phases shown are: I-isotropic, S-orientationally ordered crystal, PS-orientationally disordered (plastic crystal) and
N-nematic phase.

ratio 4 was found to undergo a "rst-order transition. Denham et al. [407] have studied by MC
simulation a two-dimensional version of the Lebwohl}Lasher model. The results obtained show
a continuous I}N transition with a power-law decay of orientational correlations.
Allen [391] has carried out a rough survey of the phase diagram of a biaxial phase which can be
produced if the molecules are no longer modelled as ellipsoids of revolution, but rather spheroids
with unequal semi-axes a,b and c. It has been found that the biaxial phase is stable only for
a narrow range of values of the semi-axes near the critical value b/aKJ10. Further, an approxim-
ate symmetry is observed under the transformation (a,b,c)%(c,ca/b,a). The role of topological
defects in the three-dimensional I}N transition has been studied by Lammert et al. [408]. These
authors considered a system whose Hamiltonian contains, in addition to the microscopic interac-
tions producing nematic ordering, a new term which is related to the extra core energy for the line
defects. A phase diagram was obtained in which the I}N transition continues to be second order for
large but "nite values of the core energy of the defect line. The "rst-order I}N transition is retained
when the core energy is lower than a critical value.
Simple arguments [160,393] suggest that hard-core systems composed of ellipsoids cannot
exhibit smectic phases. Therefore, molecules of other shapes have to be considered in the studies of
transitions involving smectic phases. Frenkel and co-workers [159,160,156,398] have performed
extensive MC and MD simulations to study the properties of a system of hard spherocylinder
characterized by the di!erent values of the ratio of ¸/d . The "rst simulations were carried out for
0
perfectly aligned systems. The phase diagram is shown in Fig. 24. It was found that a stable smectic
phase appears for ¸/d '0.5. The later work [156,160,398] showed that a system of freely rotating
0
spherocylinders exhibits a smectic phase if length-to-width ratio is higher than 3. The phase
diagram obtained by Veerman and Frenkel [398] is shown in Fig. 25. The N}S transition for
A
230 S. Singh / Physics Reports 324 (2000) 107}269

Fig. 24. Phase diagram of a system of hard parallel spherocylinder obtained from MC simulation studies [159].

Fig. 25. Phase diagram of a system of freely rotating hard spherocylinders [398](b). The shaded area is the two-phase
region separating the densities of the coexisting solid and #uid phases.

¸/d "5 appears to be continuoous in the simulation. However, due to the accuracy of the reults
0
a weakly "rst-order transition cannot be ruled out. The simulation data could not fully resolve the
phase diagram near the point indicated by the question mark (?). In this region, the existence of two
triple points is suggested, an isotropic}smectic}solid triple point at ¸/d slightly higher than 3 and
0
an isotropic}nematic}smectic triple point at a higher values of ¸/d but less than 5. Dasgupta [409]
0
 S. Singh / Physics Reports 324 (2000) 107}269 231

carried out a MC simulation of a discretized version of the de Gennes model, for understanding the
nature of the N}S transition. This model neglects the #uctuation of the magnitude of the smectic
A
order parameter. No sign of a "rst-order transition was found in these simulations. The observed
behaviour is qualitatively consistent with the predictions of "nite-size scaling theory for a continu-
ous phase transition. This simulation produces strong evidence for the de Gennes model.
Using an improved version of the frustrated spin gas model [334], Netz and Berker [392] carried
out MC simulations and obtained phase diagrams which show the expected nematic re-entrance.
With an appropriate choice of the model parameters, the simulations also show the existence of
a S phase which arises due to a lock-in of molecular permeation and rotation. The onset of the
C
S phase is detected by monitoring the tilt angle of the layer relative to the Z-axis. This work
C
suggests a microscopic origin of the S phase. Further, the model exhibits some of the modulated
C
smectic A phases (such as A , AI and A phases) and indicates the reason for their occurrence in
1 1 $
terms of microscopic properties of the molecules and their interactions.
Frenkel [160] has addressed the question whether hard-core models can exhibit columnar
phases by using MC simulations to study the thermodynamic behaviour of a system of cut spheres
with excluded volume interactions. A cut or truncated sphere is a sphere cut-o! at the top and
bottom by two parallel cuts. It is characterized by the ratio ¸/d where ¸ is the distance between
0
the cuts. Veerman and Frenkel [398] have studied the cases ¸/d "0.1, 0.2 and 0.3. For
0
¸/d "0.1, the system forms a nematic phase at the reduced density oHK0.33 (oH"o/o , where
0 #1
o is the density for close packing). At oHK0.5 the nematic undergoes a strong "rst-order
#1
transition to a columnar phase with oHK0.53. At the higher densities oH"0.8, a colum-
nar}crystalline transition occurs. At oH"0.5 a transition to a cubatic phase having cubic sym-
metry but no translational order occurs. The nematic order parameter in this phase is zero. For
¸/d "0.3 both the nematic and cubatic phases are absent and a direct isotropic #uid}solid
0
transition takes place. Frenkel [160] simulated the thermodynamic behaviour for ¸/d "0.1 and
0
0.2. The system with ¸/d "0.1 was found to exhibit and I}N transition at the reduced density
0
oH"0.335 and a strongly "rst-order transition to a columnar phase at oH"0.49. The transition to
crystalline phase occurs for oH'0.8. The system with ¸/d "0.2 did not exhibit any nematic
0
ordering even at higher density of #uid branch. For oH'0.58 strong evidence of cubatic phase was
found.
Birgeneau and Litster [410] suggested that the S phase found in many liquid crystals may be
B
a realization of a stacked hexatic phases which are characterized by short-positional order and
quasi-long-range bond-orientational order. Evidence for the existence of local herring-bone pack-
ing of the molecules near the S }S transition has been seen in the X-ray scattering studies [411].
A B
Jiang et al. [412] have studied the thermodynamic behaviour of a model system de"ned by the
reduced Hamiltonian

!J J J
H/i ¹" 1 + cos(W !W )! 2 + cos(/ !/ )! 3 + cos(W !3/ ) (9.3)
B ¹ i j ¹ i j ¹ i i
WijX WijX i
in two dimensions by MC simulation. In Eq. (9.3) S(ij)T represents nearest-neighbour pairs of sites
on a d-dimensional hyper-cubic lattice, the two angular variables, W and / , are located at each
i i
lattice site i (!p4W , U 4p) and the dimensionless coupling constants J , J and J are all
i i 1 2 3
positive. The values of J and J were kept "xed at 1.0 and 2.1, respectively, and the behaviour of
1 3
232 S. Singh / Physics Reports 324 (2000) 107}269

the system as a function of temperature was simulated for di!erent values of J . For J "0.3 results
2 2
of simulations show two transitions. The transition at the higher temperature exhibits a rounded
heat capacity peak in accordance with the expectation for a two-dimensional XY transition. At
lower temperature the transition exhibits a sharp heat-capacity peak. For J "1.4 a single
2
transition with a broad heat capacity peak is obtained. For intermediate values of J (e.g. J "0.85
2 2
and 0.95) a single continuous transition with a sharp heat capacity anomaly is observed.

9.3. Gay}Berne model

DeMiguel and coworkers [207,399] have performed the most complete simulation of the
Gay}Berne (GB) potential (Eq. (4.199)) with its original parametrization. These authors simulated
256 molecules using MD simulation in the canonical (NVT) ensemble. The phase diagram obtained
is shown in Fig. 26. The N}I transition is found to be "rst order. It has not been possible to acertain
whether the S phase is crystalline or hexatic because of the small system size considered in the
B
simulation. The biaxial S (t) phase is a tilted version of the S phase. These authors [207] also
B B
studied a purely repulsive GB potential, de"ned by subtracting the attractive part of the potential
at a given relative molecular orientation, and observed the NI transition for "xed ¹ at a slightly
higher values of oH. Further, the S and S (t) phases do not appear and the system remains
B B
nematic even at high density. This signi"es the role of the attractive part of the potential in
stabilizing smectic phases. DeMiguel and co-workers [399] have also studied the rotational

Fig. 26. Phase diagram for the Gay}Berne #uid obtained from the MD simulation [207](b). The dashed lines are
extrapolations of the data. ¹H"i ¹/e and oH"op3.
B 0 0
 S. Singh / Physics Reports 324 (2000) 107}269 233

and translational dynamics of the GB #uid in the isotropic and nematic phases. Various autocorre-
lation functions were studied. The behaviour of translational velocity autocorrelation function
indicates that in the nematic the molecules di!use along cylindrical cages whose long axes are
parallel to the director.
Luckhurst et al. [208] performed an extensive MD simulation of a reparametrized GB potential
in which the side by side con"guration is supposed to be more stable relative to the cross and tee
con"gurations: 256 particles were simulated using molecular dynamics. In addition to the isotropic,
nematic, smectic B phases, a smectic A phase also appeared. A more systematic parametrization
was carried out by Luckhurst and Simmonds [400], who constructed a site}site potential for
p-terphenyl. This potential is not uniaxial, but the biaxiality was projectd out and then the result
was mapped onto the GB potential by examining various con"gurations of molecular pairs.
Luckhurst and Simmonds [400] then simulated 256 particles interacting with this new potential
and isotropic, nematic and smectic A phases were found. The nematic phase disappears when the
reduced density is too low. These authors also compared the results of a number of GB simulations
with those of a hard ellipsoid system (see Fig. 27). The good agreement shows that the N}I
transition is dominated by the excluded volume e!ects. On the other hand, the attractive potential
dominates the formation of smectic phases as observed in GB system. Thus, it is concluded that
hard ellipsoids do not exhibit smectic phases and hard spherocylinder do so only if length to width
ratios exceed 4. The reparametrized GB potential has also been used to simulate a system of 256
discotic molecules by Emerson and coworkers [413]. A phase diagram was obtained consisting of
isotropic, discotic}nematic and columnar phases.
Tsykalo [402] was the "rst to modify the Berne Pechukas potential to include chirality, and later
Memmer and co-workers [401] modi"ed the GB potential. The total potential consists of the GB

Fig. 27. Phase diagram of hard ellipsoids (Fig. 23) along with the data points resulting from simulations of the GB
potential due to Luckhurst and simmonds [400]. Vertical lines indicate the approximate density range where a NI
transition was observed at constant density. The symbols j, L and v correspond to the di!erent values of length to
breadth ratios in the GB potential.
234 S. Singh / Physics Reports 324 (2000) 107}269

term plus an additional chiral interaction:

A B
p 7
u "4ge(r( , e( , e( ) 0 (e( ) e( )(e( ]e( ) rL ) (9.4)
#)*3!- i j r!p(r( , e( , e( )#p i j i j
i j 0
The parameter g measures the strength of the chiral coupling. Memmer and co-workers [401]
simulated a system of 256 particles interacting with this new potential. The phase diagram was
determined by adjusting the chiral coupling. A cholesteric phase was observed for 0.6(g(0.7,
while for 1.0(g(1.1 a phase with geometric structure of blue phase appeared.

10. Phase transitions in liquid crystal polymers (LCPs)

As discussed in Section 1, it is precisely the combination of the polymer-speci"c properties

together with the properties speci"c to the liquid crystal phase that gives rise to the speci"cs of
liquid crystal polymers (LCPs) and sustains the basic and applied interests in studying them. This
unique combination of properties has led to a multitude of new perspectives which are not possible
for the conventional materials in the crystalline or amorphous state and opens wide perspectives
toward their scienti"c understanding and technological applications.
The properties of LCPs are far more involved than a simple combination of two areas of science.
As a result, the understanding, at the molecular level, about their behaviour is much more worse as
compared to the liquid crystal systems. Our main interest in the present section is to pinpoint some
of the outstanding problems in LCPs and then to describe, in brief, the salient features of some of
the work done. The key issue centres around the questions * to what extent the ordered liquid
crystal state in#uences the kinetics, polymerization, stereochemistry and rheological behaviour of
the polymers. What is the e!ect of the spacer unit, mesogenic unit, polymer backbone, distribution
of chain sizes, #exibility of di!erent units, etc., on the mesomorphic behaviour. Some of these
questions will be examined with speci"c reference to the phase transition properties.
In liquid crystal polymers ordering of all three known types can occur: nematic, cholesteric and
smectic. However, the identi"cation of the mesophases generated by polymers is usually far more
di$cult than for low molar mass materials [429,436]. Usually, the nematic phase is readily
characterized but smectic phases, especially the highly ordered analogues, are often uncharac-
terized and simply denoted by S or S . Many LCPs, like conventional polymers, exhibit a glass
9
transition temperature (¹ ) which is de"ned as the temperature at which the material becomes less
'
rigid and more rubbery. The main-chain liquid crystal polymers (MCLCPs) consist of chain units
that vary in size. The distribution of chain sizes leads to wide ranges of melting to a LC phase and
wide ranges over which the polymers clears, to the isotropic liquid. Accordingly, biphasic regions
are observed over wide temperatures ranges. In MCLCPs, both nematic and smectic phases have
been found. Rigid polymers of di!erent chain size usually exhibit the nematic phase because the
packing in a layer like manner becomes inconvenient. However, the polymers with #exible units
between the mesogenic moieties can easily arrange in a layer like manner. The #exible spacer tends
to play the same role as the terminal chains in low molar mass mesogens, and so the longer the
spacer units the greater the smectic tendency. For the generation of a side chain liquid crystal
polymers (SCLCPs) the #exible spacer unit, that links the mesogenic unit to the polymer backbone,
 S. Singh / Physics Reports 324 (2000) 107}269 235

normally is essential. In SCLCPs, the increased ordering generated on polymerization means that
smectic phases predominate and the nematic phase is only exhibited by polymers with a short
spacer and a short terminal chain. The variation in spacer length a!ects both the freedom of the
mesogenic unit from the polymer backbone and overall length of the side-chain unit. When the
spacer length is reasonably short, even}odd e!ects are seen in the clearing points of polymers. The
additional ordering on polymerization causes liquid crystal phases to be more ordered than for the
monomeric analogue and transition temperatures and clearing points are higher. As a conse-
quence, a monomer unit exhibiting a nematic phase on polymerization may exhibit to a smectic
phase of higher clearing point. Similarly, relatively small monomeric units may not form liquid
crystal phases but on polymerization nematic phase may result. The most important aspect of
a polymer backbone, with reference to liquid crystallinity, is the #exibility. As the #exibility of
polymer backbone increases, the ¹ is reduced, leading to the formation of a wider mesophase
'
range. However, the clearing points often fall with increased #exibility, but not too signi"cant and
not in all cases. Another important aspect regarding the backbone relates to the average size of the
backbone overall, i.e., the degree of polymerization (DP) and the polydispersity. Mesophase
transitions and ¹ tend to rise with increasing degree of polymerization but achieve a constant
'
value when a certain DP is reached. The degree of polymerization may also decide about the kind
of mesophase to be observed.
All of the rigid rod polymers are thermally intractable, i.e., they chemically degrade at temper-
atures below their melting points. As a result these polymers must be solubilized in order to exhibit
mesophase formation. At present much of the academic activities focuses on thermotropic semif-
lexible polymers. In this section, we shall restrict our attention to the nematic state of solutions and
melts of polymers of varying architecture, interacting semi#exible polymers and main chain
nematic polymers with spacer of varying degree of #exibility [23}27,414,416,417,447].

10.1. Nematic order in polymer solutions

10.1.1. An athermal solution of long rigid rods

Assuming that the liquid crystalline order arises from purely steric causes, Onsager [97]
proposed the "rst molecular theory of nematic ordering for an athermal solution of cylindrical,
long, rigid rods of length ¸ and diameter d (¸<d ). This system can be taken as a model for
0 0
a system of rigid chain macromolecules with so insigni"cant #exibility that it cannot be manifested
in the length ¸. The basic steps of Onsager method consists of the following. Consider a system of
N rods distributed in volume < such that their concentration is c"N/< and the volume fraction of
rods in the solution is u"pc¸d2/4. The free energy of solution of rods is written as
0

C P P D
1
A"N¹ ln c# f (e( ) ln[4pf (X)] dX# c f (e( ) f (e( )B (X )dX dX . (10.1)
2 1 2 2 12 1 2
The "rst term is due to the translational motion of rods, the second describes the losses of
orientational entropy owing to liquid crystalline order and the third term is the free energy of
interaction of the rods in the second virial approximation. When only steric interactions of the rods
are present,
B (X )"2¸2d sin X , (10.2)
2 12 0 12
236 S. Singh / Physics Reports 324 (2000) 107}269

where X is the angle between unit vectors, e( and e( . A simple estimate of virial coe$cients, giving
12 1 2
B &¸2d and the third virial coe$cient B &¸3d3ln (¸/d )8, shows that the second virial
2 0 3 0 0
approximation (cB <c2B ) is valid under the condition c;1/¸d2 or u;1. In the limit ¸<d
2 3 0 0
a liquid crystalline transition in the solution of rods occurs precisely at u;1.
For each concentration the free energy of the system must be a minimum; the resulting minima
correspond to possible phases. Above a certain concentration such a minimum is obtained by
a state in which part of the system is isotropic and the other part nematic (with di!erent
concentrations cH and cH, respectively; cH"(¸/d )u is a dimensionless concentration). It turns out
* ! 0
that for low concentrations of the rods in solution (cH;1), the anisotropic (nematic)}isotropic
(liquid) transition is "rst order. When cH(cH the solution is isotropic, when cH'cH it is
* !
anisotropic, and when cH(c(cH the solution separates into isotropic and anisotropic phases.
* !
The Onsager trial function gives the following results:
cH"3.340, cH"4.486, a"18.58, PM "0.848 . (10.3)
* ! 2NI
Here a is the variational parameter. Using a Gaussian distribution function leads to the following
results [414]:
cH"3.45, cH"5.12, a"33.4, PM "0.91 . (10.4)
* ! 2NI
These results show that the coexisting concentrations and nematic order parameter depend
critically on the exact form of the distribution function employed in the calculation.
Of course, the integral equation arising upon the exact minimization of the free energy can be
solved numerically to a high degree of accuracy. Using this procedure, the following values were
obtained [415]:
cH"3.290, cH"4.191, PM "0.7922 . (10.5)
* ! 2NI
The results (10.3) and (10.5) show that the use of the variational method leads to a very small error
(&5%) in determining the characteristics of the mesophase transitions. The application of the
Onsager method has been considered to the case of high concentrations also. The details are given
elsewhere [416,417]. Here at this stage we would like to mention that the Onsager method can be
generalized for describing solutions of any arbitrary concentration.
Another approach for the above problem was developed by Flory [101]. Using the lattice model
of polymer solutions, Flory [101] replaced #exible chains with rigid rods and demonstrated the
formation of an ordered phase above a critical volume fraction of rods that depend on the rod
aspect ratio (x "¸/d ). The liquid crystal phase transition was studied [102,138] using a modi"ed
0 0
variant of Flory theory and for an athermal solution (x <1) the results are
0
cH"7.89, cH"11.57, PM "0.92 . (10.6)
* ! 2NI
Upon comparing the results (10.3) and (10.6) we can conclude that the results of lattice model
although qualitatively correct di!er quantitatively rather strongly from the exact results (in the
limit x <1) of Onsager.
0
10.1.1.1. Polydispersity.. In an experimental situation, most often the polymer solutions prove to
be polydisperse. That is, they are composed of macromolecules of di!ering masses. This may have
 S. Singh / Physics Reports 324 (2000) 107}269 237

strong in#uence on the isotropic}nematic phase transition. The in#uence of polydispersity on

a nematic transition has been studied by using both the Onsager method [415,418}421] and the
Flory theory [422}425]. The qualitative results of these studies coincide.
Let us consider a simple bidisperse system composed of particles of equal diameter d but
0
di!erent lengths ¸ and ¸ . Let u and u be the volume fractions in the solution of rods of the "rst
1 2 1 2
and second type, respectively, and let
u"u #u and q"¸ /¸ '1 .
1 2 2 1
z is the weight fraction of the long rods
z"u /u
2
and ¸M 8 is the weight-average length of the rods
¸M 8"¸ (1!z)#z¸ . (10.7)
1 2
Just like a monodisperse system, the nematic transition occurs at u&d /¸M 8. Using the Onsager
0
method, a computer calculation [415] was performed for q"2 to obtain the phase diagram.
Numerical calculations [415,426] have also been performed for two systems with length ratios
q"2 and 5. From these results the following features can be noted (see Fig. 28).
(i) The polydispersity leads to a small decrease in the lower (isotropic) boundary of the separation
region and simultaneously to a strong increase in its upper (anisotropic) boundary. The longer
rods preferentially go into the anisotropic phase.

Fig. 28. Volume fraction scaled by the weight-averaged length against mole fraction of the longer rods for a bidisperse
mixture of rods (¸ /¸ "2) [415,426]. The broken lines indicate in which two phases compositions falling in the
2 1
two-phase region will divide.
238 S. Singh / Physics Reports 324 (2000) 107}269

(ii) The relative width of the separation region, i.e., the concentration di!erence between isotropic
and nematic phase may be much larger as compared to the monodisperse case.
(iii) A weak variation in ¸M 8u /d with mole fraction of the longer rods is found. This implies that the
*
molecular weight dependence of the bifurcation density is a good indication for the onset of
phase separation.
(iv) The mean-order parameter PM M ("(1!z)PM #zPM ) is found to be appreciably higher
2NI 2,1 2,2
(K0.92) than in the monodisperse case (K0.79).
For the length ratio q"5 similar, but more pronounced, results were obtained with two
additional features.
(v) For certain compositions a phase sequence of isotropic}nematic}re-entrant isotropic}re-
entrant nematic has been observed as a function of concentration.
(vi) For some compositions the system shows a biphasic region where the system separates into
two di!erent nematic phases, or even a three-phase region in which an isotropic phase coexists
with two nematic phases.

10.1.2. Athermal solutions of partially yexible polymers

The real rigid-chain macromolecules always have a certain "nite #exibility. The macromolecules
may di!er with respect to the mechanism of the #exibility of the polymer chain. The simplest kind
belongs to the freely linked chain, which amounts to a sequence of hinged rigid rods of length ¸ and
diameter d, with ¸<d (Fig. 29a). The orientation of each successive rod, in the equilibrium state, is
random and is independent of the orientation of previous ones. Consequently, the mean-square
distance between the ends of the chain SR2T is given by

SR2T "¸l, l<¸ , (10.8)

0
where l is the total contour length of the chain.
In case the polymer chain has any other mechanism of #exibility, for example, if the orientations
of adjacent links are correlated, Eq. (10.8) is still satis"ed but with a renormalized length ¸. This
renormalized length is known as the e!ective (Kuhn) segment of the polymer chain.
In a persistent mechanism of #exibility, the #exibility arises from the accumulated e!ect of small
oscillations in the valence angles. A persistent macromolecule can be represented in the form of
a homogeneous cylindrical elastic "lament of diameter d (Fig. 29b). The elasticity of the "lament is
such that it can be substantially bent only on scales of the order of ¸. Most of the macromolecules
belong to this class of polymeric objects. On the basis of the relationship between l and ¸, rigid
chain macromolecules can belong to one of the three following fundamental classes: (a) if ¸<l<d
the #exibility of polymer chain can be neglected; this refers to the case of limiting rigid-chain

Fig. 29. The simplest mechanism of #exibility (a) Freely linked chain, (b) persistent chain.
 S. Singh / Physics Reports 324 (2000) 107}269 239

macromolecules (or rigid rods); (b) if l<¸<d, the rigid-chain macromolecule includes many
Kuhn segments; this refers to the case of semi#exible macromolecule which stays in the state of
a random coil, (c) l&¸: in real experiments this kind of macromolecule is found rather often.
Using the continuum approach (the Onsager approach) the transition of an athermal solution of
partially #exible polymer chains to an anisotropic phase has been studied [427,430] extensively. In
full analogy with the Onsager method, these studies [427,428] led to the following conclusions: For
the model shown in Fig. 29, the orientational ordering of the athermal solution has the features of
a "rst-order phase transition and occurs at low concentrations of the polymer in solution. When
u(u , the solution is homogeneous and isotropic; when u'u it is homogeneous and anisot-
* !
ropic; and when u (u(u it separates into isotropic and nematic phases, with
* !
u &u &d/¸;1. For an athermal solution of freely linked semi#exible chains [427],
* !
cH
cH"3.25, cH"4.86, u" ! !1"0.5, PM "0.87, 2 (10.9)
* ! cH 2NI
*
On comparing the results of (10.9) with the results of (10.3) and (10.5) it is found that hinge linking
of the rods in long chains leads only to quite insigni"cant changes in the characteristics of
isotropic}nematic transition. The region of phase separation is somewhat expanded, while the
order parameter of the orientationally ordered phase is slightly increased.
For an athermal solution of persistent semi#exible chains the following result was obtained
[428]:

cH"10.48, cH"11.39, u"0.09, PM "0.49 . (10.10)

* ! 2NI
It is obvious that for the same d/¸ the orientational ordering in a solution of persistent chains
occurs at su$ciently larger concentrations than in a solution of freely linked macromolecule. The
relative concentration jump of the polymer as well as the order parameter at the transition are
considerably smaller.
The phase transition in case of a persistent chain was calculated in the Gaussian approximation
and the following result was obtained:

cH"7.77, cH"9.71, a"12.34, PM "0.759 . (10.11)

* ! 2NI
A somewhat better result was obtained [414] by the use of the Onsager trial function

cH"5.41, cH"6.197, a"6.502, PM "0.61 . (10.12)

* ! 2NI
Solving the non-linear integro-di!erential equation, Vroege and Odijk [431] obtained the result

cH"5.124, cH"5.509, PM "0.4617 . (10.13)

* ! 2NI
The conformations of the semi#exible macromolecules in the liquid crystalline phase also depend
on the mechanism of the #exibility. A very important conformational characteristic is the mean
square of the projection R of the segment joining the ends of the chain on the director direction
z
(Z-axis), SR2T,
z
SR2T"s SR2T . (10.14)
z 0 0
240 S. Singh / Physics Reports 324 (2000) 107}269

Here SR2T "l¸ is the mean square of the distance between the ends of the chain in the isotropic
0
phase and s is the susceptibility of the system to an external orienting "eld. It has been found that
0
in the nematic solution the freely linked chains are somewhat extended in the direction of the axis
of orientational order. However, the magnitude of SR2T increases by a factor of not more than
z
three as compared to its value in the isotropic phase. In case of persistent #exibility, the behaviour
of susceptibility is completely di!erent. For this case the susceptibility s , and hence SR2T, sharply
0 z
increases according to an exponential law upon increasing the concentration of the nematic
solution. In other words, the macromolecules are strongly stretched out along the director. This
e!ect may be referred to as sti!ening of persistent macromolecules in the liquid crystalline state.
The problem of orientational ordering in solutions of polymer chains has been addressed
[432,433] assuming other #exibility mechanism also.

10.1.3. Non-athermal polymer solutions

E!orts were made [434,435] to analyse the role of attractive forces of the links in nematic
ordering of a solution of rigid-chain polymers. Since the treatment in these studies was based to
some degree on the Flory lattice approach, exhaustive solution of the problem could not be
obtained. One of the problems which arise in a system of non-athermal polymer solutions involves
the fact that the separated anisotropic phase can be very concentrated and the second-virial
approximation of Onsager is not applicable.
This problem was "rst studied consistently by Khokhlov and Semenov [436] by using the
continuum approach. The attractive force was approximated as
lNu
u "! (u #u PM 2) . (10.15)
!55 2d 0 ! 2

Here u and u are constants describing, respectively, the isotropic and anisotropic components of
0 !
the attractive forces. Eq. (10.15) was rewritten as

A B
lNuH u
u "! 1# ! PM 2 . (10.16)
!55 d u 2
0
Here H("u /2) is the theta temperature of the polymer solution which is de"ned as the temper-
0
ature at which the osmotic second virial coe$cient vanishes. In the actual calculation it was
assumed that u /u "0.
! 0
For a solution of persistent macromolecules the phase diagrams were calculated for the liquid
crystalline transition in terms of u and H/¹ for several values of l/d. It was found that in the region
of relatively high temperatures, a narrow corridor of phase separation into isotropic and anisot-
ropic phases lying in the dilute solution region exists. Contrary to it, at low temperatures the region
of phase separation is very broad such that an isotropic, practically fully dilute phase, and
a concentrated, strongly anisotropic phase coexist. These two regimes are separated by the interval
between the triple-point temperature ¹ and the critical temperature ¹ (¹ '¹'¹ ), in which
5 # # 5
there are two regions of phase separation between isotropic and anisotropic phases, and between
two anisotropic phases having di!ering degrees of anisotropy. The temperatures ¹ and
5
¹ substantially exceed the H-temperature. The interval between ¹ and ¹ becomes narrower as
C # 5
the ratio ¸/d decreases and drops out when (¸/d) 1"125. When ¸/d(125 there are no critical or
C
 S. Singh / Physics Reports 324 (2000) 107}269 241

triple points on the diagram, and one can refer only to the crossover temperature ¹ between the
#3
narrow high-temperature corridor of phase separation and the very broad low-temperature region
of separation. With decreasing ¸/d, the temperature ¹ decreases. For (¸/d) 2K50, one obtains
#3 C
triple and critical points corresponding to an additional phase transition between two isotropic
phases. The concentration of one of these phases is extremely low. The in#uence of attractive
interaction on the order parameter of a nematic solution of persistent macromolecules at the
transition point was also analysed. It was found that the attractive forces a!ect the value of PM very
2
substantially when ¹/H42.

10.2. Nematic order in polymer melts

The liquid crystalline polymer melts are dense system and often called `thermotropic liquid
crystalsa. The applications of methods discussed above cannot be applied in a straightforward
manner because the large parameter l/d cannot be used. We discuss below the application of
mainly three approaches: Onsager, Maier and Saupe and Flory.

10.2.1. Melts of linear homopolymers

In the thermotropic solutions neither the concentration c nor the volume fraction u can be used
as the external parameter. The role of pressure was studied [437] in detail with the results that the
in#uence of normal atmospheric pressure on the mesophase ordering is negligibly small. A substan-
tial increase in the region of stability of the nematic phase can be expected only at high pressure of
the order of 103 atm. When pressure becomes in"nitely large in melt of any particles that are
anisodiametric to any degree and have a rigid steric core of interaction, a liquid crystalline phase
can be observed.
The phase diagram for a melt of long persistent chain was calculated [437] as a function of ¹/H
and ¸/d for u /u "0.1 at atmospheric pressure. From the results the following features are evident:
! 0
existence of three phases-isotropic and anisotropic melts, and also a gas-like phase (at high
temperatures) was observed. The nematic order can occur only if the asymmetry parameter ¸/d is
smaller than the critical value (¸/d) +50. When (¸/d)'(¸/d) the melt in the equilibrium state is
# #
always a nematic (at any temperatures). However, the critical value (¸/d) depends on the
#
mechanism of #exibility of the chain. For example, for freely linked #exibility it was found [437]
that (¸/d) +7, while for a melt of rigid rods (¸/d) K3.5. In the formal limit, ¸/dP0 it is seen
# #
that the steric interactions are inessential and the mesophase ordering occurs only due to the
anisotropy of the attractive forces. The results obtained for ¸/d"0 are close to those obtained
by Rusakov and Shliomis [438] in the limit ¸/dP0. It is worth mentioning that these results
are exact to any extent only at small values of ¸/d, while in the most interesting region ¸/d<1
the steric interactions always dominate and the anisotropy of attractive forces is always a
secondary one.
Some additional features were observed [438,439] which do not depend qualitatively on whether
steric forces are taken into account or not. These works show that the order parameter at the
transition point depends on the length of the persistent macromolecule l. When l;¸ the order
parameter PM "0.43. As l increases the value of PM decreases and reaches a minimum value
2NI 2NI
0.34. When l4¸, PM slightly increases to the value PM "0.36. Regarding the conformations
2NI 2NI
of the macromolecules it was found that in the nematic phase the macromolecules are stretched
242 S. Singh / Physics Reports 324 (2000) 107}269

along the axis of ordering (the Z-axis). The degree of extension can be characterized by the
parameter
y"SR2T/SR2T .
z x
Here R and R are the projections of the vector joining the ends of the polymer chain. The
x y
magnitude of this parameter at the transition point depends on the length of the macromolecules.
When l;¸, y "3.25. As l increases, y decreases reaching the minimum value y "2.77, and
0 0 0
then increases substantially to the value y "14.4 in the limit of very long persistent chains (l<¸).
0
With decreasing temperature of the nematic melt (for l<¸) the polymer chains unfold further; the
value of y increases exponentially.

10.2.2. Melts of linear copolymers

The nematic ordering in melts of linear copolymers was studied [440}443] by using a generaliz-
ation of the lattice method [441] for an athermal melt as a function of ¸/d and a (a is the volume
fraction of the #exible component) for the case l /d"1.5 (l is the length of the e!ective segement
0 0
of the #exible region). Two set of transitions, isotropic}nematic and weakly anisotropic}strongly
anisotropic, were observed. The second transition between two nematic ends at a critical point with
coordinates (¸/d) "10, a "0.2. Upon decreasing the #exibility of the #exible fragments of the
# #
chain (i.e., decreasing ratio l /d), the region of stability of the weakly anisotropic phase narrows
0
and is shifted towards larger a.

10.2.3. Melts of comb-like polymers

In a mean-"eld approximation of Meier}Saupe type, the nematic ordering in melts of comb-like
polymers with rigid fragments in the side chains was studied [443] by incorporating only the
interaction of the mesogenic fragements. It was found that the orientational properties of the
system substantially depend on the length ¸@ of the fragment of the main chain between adjacent
branches. The length of the spacer linking the mesogenic group with the main chain is assumed to
be very small. When ¸@'l, the mesophase ordering occurs in a similar manner as the low-
molecular-weight system. The order parameter at the transition is PM "0.43. When the density of
2NI
the side branches increases (i.e. l/¸@ increases), the order parameter at the transition decreases
monotonically. As the length of #exible fragments ¸@ decreases PM either increases or decreases
2NI
very slightly. Thus, the order parameter for the comb-like polymer, always stays smaller than for
their linear analogs. This conclusion is in agreement with experimental data [444}446].
When the density of side branches is su$ciently large (l/¸@'(l/¸@) ) the PM becomes negative.
# 2NI
In this case the main chains are oriented along the director, while conversely the mesogenic groups
lie preferentially in the easy plane. The critical value of density for a persistent chain is found to be
[443]

AB
l
"18 .
¸@
C
Fig. 30 shows the phase diagram [443] of a melt of comb-like macromolecules as a function of l/¸@
and ¹. N and N correspond to the di!erent conformations of the macromolecule and N is
` ~ "
a biaxial nematic phase.
 S. Singh / Physics Reports 324 (2000) 107}269 243

Fig. 30. Phase diagram of a melt of complike macromolecules in the variables ¸/l (e!ective density of side branches)-
temperature [443]. IL-isotropic phase, N -phase of `easy-axisa type, N -phase of `easy planea type, N -biaxial phase.
` ~ "

10.3. Nematic polymers with varying degree of yexibility

Within the lattice model, Flory [101,449] initiated the study of transitions in semi#exible
polymers and presented a simple heuristic treatment of the properties of concentrated, self-
avoiding polymers as a function of their stifness. He found a "rst-order transition from a disordered
state to an entirely ordered state. Kim and Pincus [449] gave the mean-"eld treatments of such
transitions. Assuming polymers to be non-chiral, a simple model for the interacting semi#exible
polymer was developed by Petschek [448]. He used Ursell}Mayer perturbation expansion tech-
nique to systemize the calculation of the properties of the system. Using group theoretical
arguments, the details of this expansion were analysed and a relationship with "eld theory could be
established. It was found that the isotropic}nematic transition in systems of long, semi#exible
polymers can be controlled either by the usual tensor order parameter or by a hidden, vector-like
order parameter. This vector like order parameter is expected to control the behaviour only if the
#uctuations are important in determining the behaviour of the material near the transition. Which
of the two order parameters is the controlling order parameter depends on the details of the
con"guration and interaction energies of the polymer system. However, the results of the calcu-
lations on simple models have no immediate consequences for the behaviour of the realistic
polymer system and it is not evident what physical systems, if any, would be expected to have
transitions which are controlled by vector-like order parameters.
Wang and Warner [450] modelled main chain liquid crystal polymers as either worms or jointed
rods. These authors presented a model, which accounts for molecular parameters such as the
lengths of the mesogenic group and the spacer units and the interactions between them, to describe
the non-homogenous nematic polymers. The mesogenic groups have been modelled as rods in
a quadrupolar potential whereas the spacers were treated exploiting the spheroidal approach. The
spacers have been found to have an order di!ering from the mesogenic units. If the spacer is not
very long and thus in e!ect is in#exible, one end of the spacer can retain to some extent the
orientation of the other end, allowing orientational correlation between spacers mediated by the
intermediate mesogenic unit. This provides the chain a global rod-like behaviour as the nematic
"eld becomes strong or the temperature low. The in#uence of the physical linkage and the van der
Waals interaction between the rods and the worms was examined. The nematic}isotropic
transition and some other properties, such as the orientational order of the two components, i.e. the
244 S. Singh / Physics Reports 324 (2000) 107}269

mesogenic units and the spacer, and the latent entropy were calculated as functions of the
molecular parameters. In accordance with the experimental observations it was found that the
reduced transition temperature (the temperature reduced by ¹ of MS model for pure rods)
NI
decreases signi"cantly when the length of the #exible spacer b increases while the latent entropy
increases. The e!ect of interactions, between mesogenic units u , spacer units u and me-
!! ""
sogenic}spacer units u , on the nematic}isotropic transition temperature was studied. The ratio
!"
u /u does not visibly a!ect the properties of the polymer. The order parameter of the mesogenic
"" !!
unit at the transition does not vary, remaining at about 0.434, while there is a signi"cant variation
in that of the #exible spacer.

11. Phase transitions in chiral liquid crystals

11.1. Non-ferroelectric liquid crystals

11.1.1. Transitions in cholesteric (or chiral nematic) liquid crystals

The cholesteric phase is a twisted variant of the nematic phase. Obviously, the interaction energy
between the chiral molecules should contain extra terms giving rise to the helical structure of the
director in addition to the potential energy of the nematic phase. Goossens [451](a) made the "rst
attempt to develop a statistical model of the cholesteric phase by extending the Maier}Saupe
theory to incorporate the chiral nature of the intermolecular coupling. It was shown that the
second-order perturbation energy due to the dipole}quadrupole interaction must be included to
explain the helicity. This type of interaction can be considerably simpli"ed because of the rapid
rotations of the molecules around their long molecular axes. Consequently, in a "rst approxima-
tion, the molecules may be assumed to rotate uncorrelated. Based on the general idea of Goossens,
a model was proposed by van der Meer et al. [452}455] which allows the molecules to rotate freely
about their long principal axes of inertia. The relevant interaction between two molecules 1 and
2 boils down essentially to

u "!J [(e( ) e( )!3(e( ) r( )(e( ) r( )]2!K [(e( ) e( )!2(e( ) r( )(e( ) r )](e( ]e( ) r( ) . (11.1)
ij ij i j i ij j ij ij i j i ij j ij i j ij
The "rst term in Eq. (11.1) is the Maier}Saupe-induced dipole}dipole interaction and arises due to
the anisotropy of the molecules giving rise to the nematic state. The second term is the chiral
interaction arising due to the dipole}quadrupole part of the dispersion interaction. The coupling
constants J and K denote the respective interaction strengths and depend on the separation r .
ij ij 12
The model (11.1) was solved by van der Meer et al. [452] in the molecular-"eld approximation
under the further assumption that the system is locally nematic. One of the serious problems with
these models [451,452,456] is that in their present form they fail to provide a satisfactory
explanation for the fact that in most cholesterics the pitch decreases with rise of temperature.
Experimentally, it is observed that the helical wave number q of a cholesteric system varies with
temperature. For mesogens exhibiting a cholesteric}smectic phase transition at some temperature
¹ , presmectic #uctuations are responsible for the strong temperature dependence of q near ¹ .
#4 #4
van der Meer et al. [452] made an attempt to consider only the approximately linear temperature
dependence of the intrinsic pitch and not the one resulting due to the smectic short-range ordering.
 S. Singh / Physics Reports 324 (2000) 107}269 245

In order to obtain a temperature-dependent q these authors included extra terms

!¸ (e( ) e( )4#M (e( ) e( )3M(e( ]e( ) ) r( N
ij i j ij i j i j ij
in the interaction potential (11.1). The model was solved and it was found that the temperature
dependence of the helical wave number q can be determined by the ratio of the order parameters
PM /PM ("x),
4 2

C D
1 (7K!3M)!4Mx2
q" . (11.2)
r (14J#12¸)#16¸x2
In fact, PM and PM depend on q, i.e., Eq. (11.2) is an implicit equation for q(¹). Thus, it was found
2 4
that the magnitude of the reciprocal pitch varies nearly linearly with temperature in agreement
with the experiment.
Scholte and Vertogen [458] proposed a simpli"ed version of the model (11.1) and solved it
analytically in the mean-"eld approximation. These authors considered the model (11.1) with one
modi"cation. The vector e6 is no longer a unit vector but requires only to satisfy the so-called
spherical constraints
N
+ e2"N . (11.3)
i
i/1
As pointed out by Vertogen and van der Meer [456] this modi"cation boils down, with regards
to the thermodynamics, to a weakening of the constraint e2"1 to the constraints Se2T"1, the
i i
thermodynamic expectation value of its length squared must be one. The solution gives two
independent order parameters, R and S de"ned by
Sa2T"1(1!2S) , (11.4)
x 3
Sa2T"1(1#S!R) , (11.5)
y 3
Sa2T"1(1#S#R) . (11.6)
z 3
The excess free energy per particle reads
1 3S 2 1 1
b*A" bJ(3S2#A(q)R2)! # bJS! ln(1!2S)! ln[(1#S)2!R2] . (11.7)
9 1!2S 3 2 2
Here the helix wave number q is determined by the relation RA(q)/Rq"0. It is clear that the
description of the cholesteric state must be based upon more than one order parameter. In Scholte
and Vertogen's (SV) model two order parameters appear. In case of large values of the pitch only
one order parameter su$ces, because the deviation of local uniaxiality is of the order of (qr)2. In the
present case this deviation is given by
Sa2T!Sa2T"0.222(qr)2 . (11.8)
x y
Thus, the assumption of van der Meer et al. [452] regarding local uniaxiality is fully justi"ed if the
deviation of uniaxial symmetry is of the order of 10~5 and consequently negligible. Further, it is
important to mention that the SV model gives rise to a pitch, which does not in#uence the jump on
the nematic order parameter upto the order (qr)2 and does not depend on temperature.
246 S. Singh / Physics Reports 324 (2000) 107}269

Evans [459] addressed the question whether the short-range repulsive interactions associated
with a chiral molecular backbone can also achieve macroscopically chiral #uids and what are the
typical pitches resulting from the hard-body models. The chiral body was represented by a hard
convex twisted ellipsoidal core, with and without an encircling isotropic square well. A stability
analysis, within the framework of a density functional theory, was employed to calculate the
wavelength of the pitch at the isotropic}chiral nematic transition. The transition densities were
taken from the work of Tjipto-Margo and Evans [173] and the chiral pitch was derived by the
minimization of the bifurcation density with respect to the chiral wave vector. A general expression
for the chiral pitch was derived for hard and soft bodies,
2p SC(1,2)r2P (e( ) e( )T
p" 2 1 2 . (11.9)
3 SC(1,2)r( ) e( e( ) e( e( ) e( T
y 2 2 z x
The sign of p determines the sense of the rotation (left- or right-hand polarization). The chiral pitch
was expressed in terms of a ratio of moments of the direct correlation function and this ratio was
evaluated within the Person}Lee [102}110] framework for the direct correlation function to give
8p :d) d) J(1,2)R2P (e( ) e( )
p" 1 2 2 1 2 , (11.10)
15 :d) d) J(1,2)RK ) e( e( ) e( e( ) e(
1 2 y 2 z 2 x
where J (1,2) is the Jacobian and R is the center-to-center vector on the excluded volume surface of
the full size bodies. In the present model, the chiral pitch was found to be density and temperature
independent with values in the visible region of the spectrum. This "nding is an artefact of the
Parson}Lee model. It is expected that the long range potential softness may account for the
increase of the pitch with decreasing temperature. The role of the critical #uctuations in the vicinity
of a structural transition point has also been investigated [457].

11.1.2. Blue phases

Blue phases (BPs) are mesophases which occur in a narrow temperature range just below the
clearing point, usually between the isotropic liquid phase and cholesterics of su$ciently short
pitches [43,44,460]. They occur in a system of chiral molecules, though not all chiral compounds
exhibit blue phases. In several chiral compounds up to three blue phases are thermodynamically
stable. The two lower temperature phases, blue phase I (BPI) and blue phase II (BP II), possess
cubic symmetry, while the highest temperature phase, blue phase III (BP III), appears to be
amorphous. As early as 1906, Lehmann [461] detected the blue phase as a stable, optically
isotropic modi"cation not identical to the cholesteric phase. The very "rst observation of this phase
was described by Reinitzer [1] in his famous letter to Lehmann [2]. He reported a blue violet light
re#ection just below the clearing point of cholesteryl benzoate which was actually caused by the
BPI (see Ref. [462, p. 172]).
Blue phases are prominent in cholesteric systems of su$ciently small pitches. The BP no longer
exists when the cholesteric pitch exceeds a critical value, P . For comparison of di!erent systems
#
the pitch is measured at the cholesteric blue phase transition temperature. These results show
[43,44] the important in#uence of the pitch on the stability of blue phases. It can be seen from
Table 11 that the critical pitches, P , do not adopt a universal value but depend on both the
#
molecular structure and the composition of the system. Furthermore, even the ratio P /P depends
. #
on the structure of the mixed systems.
 S. Singh / Physics Reports 324 (2000) 107}269 247

Table 11
Critical pitches for monomorphic (P ) and dimorphic blue phases (P ) in cholesteric mixed systems at ¹
# . C1@BP
System P /nm P /nm P /P
# . . #
Cholesteric/nematic mixtures

CN/PCPB 383 333 0.87

CN/CCH7 439 410 0.93
CN/PCH7 410 356 0.87
CN/CB7 398 336 0.84
CC/CCH7 407 330 0.81
CC/PCH7 395 345 0.87
CC/CB7 350 271 0.77
CV/80BE 415 340 0.82
CM/80BE 390 265 0.68
CB/80BE 355 270 0.76
Compensated mixtures
CN/CBAC 647 443 0.53
573 556 0.97
CN/CC 540 472 0.87
490 370 0.76
Induced BPs
CA/PCPB 310 305 0.98
373 309 0.83
CEEC/EBBA 610 462 0.76
490 413 0.84

Abbreviations: CA cholesteryl acetate; CV cholesteryl valerate; CC cholesteryl chloride; CN cholesteryl nonanoate;

CM cholesteryl myristate; CB cholesteryl benzoats; CEEC cholesteryl ethoxyethoxyethycarbonate; EBBA 4-ethoxy
benzylidene-4@-n-butylaniline; CBAC 4- cyanobenzylidene-4@-aminocinnamic acid opt.act. amylester; PCPB p- pentyl-
phenyl -2 -chloro-4-(p-n-pentylbenzoyloxy)- benzoate; 80BE 4-n- octyloxyphenyl -4@-n- octyloxybenzoate.

Bergmann and Stegemeyer [463] showed the existence of two polymorphic forms BP I and BP II
by DSC thermograph as well as by selective re#ection and optical rotation measurements. The
possibility of the existence of a third BP was "rst mentioned in 1980 [462]. A grey texture was
detected between the isotropic liquid and the BP II platelet texture by polarizing microscopy. Such
a texture was also observed by Marcus [464] and Meiboom and Sammond [465] and they called it
a &blue fog' or a &fog phase'. Collings [466] provided convincing evidence for the stability of a BP III
by optical rotatory dispersion measurements. Keiman et al. [467] resolved directly the BP II/BP
III as well as the BP II/isotropic transition by high-precision speci"c heat measurements. From
these results it became quite obvious that the fog phase is thermodynamically stable, and it has
been called BP III while the CI}BP I, BPI}BP II and BP}III transitions are known to be "rst
order, the BP III}IL transition is weakly "rst order.
In general, blue phases occur between the cholesteric and the isotropic liquid state. In polymor-
phic mesogens the phase sequence

S PClPBPPIL
A
248 S. Singh / Physics Reports 324 (2000) 107}269

Fig. 31. Partial phase diagram of CM/C system [468]. A direct transition from the S phase to BP I is observed.
10 A

has been observed with increasing temperature whenever the cholesteric pitch is small enough. The
blue phase stability of cholesteryl myristate (CM) on admixing 4, 4@-di-n-decylazoxybenzene (C ),
10
which exhibits only S phase, is shown [468] in Fig. 31. It can be seen that on increasing the
A
C concentration the cholesteric range decreases "nally to zero at 15 mol% whereas the BP ranges
10
remain nearly constant. In mixture of 15}18 mol% C a direct S -BPI transition occurs. These
10 A
results show that the occurrence of a blue phase does not require the simultaneous existence of
a cholestric phase. Fig. 32 shows the variation of blue phase polymorphism with pitch. It is
observed [466] that the BP III exists only in systems with very short pitches. The BP III span is
extremely small (40.05 K). It decreases with increasing pitch and becomes zero at the so-called
dimorphic pitch, P . The voltage temperature phase diagrams of mixture of the polar compound
a
4-hexyloxy-4- cyanobiphenyl, M18, and its chiral 2-methylbutyl derivative, CB 115, have been
investigated [43] in the range of 47}57 mol% CB 15. As can be seen, besides the well-known
cholesteric nematic transition, a BP IPCl phase transition is indeed induced by the "eld. Also
a "eld-induced BP II-Cl transition in all CB 15/M18 mixture has been found. In this case the
magnitude of the cholesteric pitch does not play an important role during the "eld-induced phase
transition. It is obvious from Fig. 33 that the coexistence line for the BP IPCl transition is inclined
toward the ordinate whereas it runs parallel to the abscissa in case of the BP IIPCl transition, i.e.
the critical "eld strength for BPIIPCl is independent of temperature.
 S. Singh / Physics Reports 324 (2000) 107}269 249

Fig. 32. Schematic phase diagram showing the variation of blue phase polymorphism.

Fig. 33. Phase diagram showing voltage against temperature for a M18/CB 15 mixture (47 mol% CB 15).

Two di!erent classes of theory have been applied to describe the blue phase structures
[43,44,460,469], the Landau theory, and the defect theory. A detailed comparison of both the
theories is given elsewhere [469]. Based on the common theme of a phenomenological free energy,
two groups of workers have considered the application of Landau theory to blue phases from
250 S. Singh / Physics Reports 324 (2000) 107}269

di!erent points of view. The "rst approach [470}478] uses a Landau free energy constructed from
the most general tensorial order parameter for a nematic. The second approach due to Meiboom
and co-workers [479}482] uses a Frank free energy for a chiral nematic and constructs blue phases
from regular arrays of line defects (disclinations). The "rst approach works well when the gradient
terms in Landau energy dominate, which is true when the molecular chirality is strong. The second
approach is appropriate when the chirality is low. In this limit, the non-gradient (bulk) terms in the
Landau energy are dominant.
The Landau free energy density expansion in powers of Q and its gradients can be expressed as
[44,460]
f"f #f (11.11a)
"6-, '3!$
"A Tr(Q2)!J6B Tr(Q3)#C[Tr(Q2)]2#1K [(+]Q) #2q Q ]2#1K [(+ ) Q) ]2 .
4 1 ij ij 4 0 i
(11.11b)
Here the notation of Wright and Mermin [44] has been used in labelling the phenomenological
coupling constants. Since the bulk and gradient free energies favour di!erent structures, the
minimization problem has not been solved. It is convenient to rescale the variables appearing in
Eq. (11.11) to treat the system in terms of small numbers of important dimensionless parameters.
The dimensionless free-energy density f H, e!ective temperature ¹, order parameter QH, and length
scale r@ are de"ned as [478]

A B A B AB
C3 C C
f H" f, q" A, QH" Q, r@"2qr . (11.12)
B4 B2 B
This scaling is singular in the limits BP0 and qP0. It is, therefore, not suitable for describing the
cases in which the exact free-energy minima are known. Except for these limits, however, the
dimensionless variables do provide the natural scales of energy density, temperature, and length for
a discussion of the blue phases. The dimensionless free-energy density now reads
f H"f H #f H (11.13a)
"6-, '3!$
"q Tr(QH2)!J6 Tr(QH3)#[Tr(QH2)]2#i2M[(+]QH) #QH]2#g[(+ ) QH) ]2N .
ij ij i
(11.13b)
Here i is the positive dimensionless chirality parameter

A B
CK 1@2
i"q 1 "qm"2pm/P (11.14a)
B2
and
g"K /K . (11.14b)
0 1
The rescaling shows that the strength of the chirality parameter i determines whether the bulk
(small i) or the gradient (large i) terms will diminate the free energy. The high-chirality behaviour
can be described by Hornreich and Shtrikman [470}478] model in which the blue phase structures
 S. Singh / Physics Reports 324 (2000) 107}269 251

Fig. 34. Theoretical phase diagrams [478] with (a) two blue phases and (b) three blue phases. i ("2pm/p) is the chirality
parameter and the reduced temperature is given as t"(¹!¹H)/(¹ !¹H) with ¹ the clearing temperature and ¹H the
C C
Landau temperature which occurs mPR.

are viewed as expansions about the relatively simple blue phase order parameters derived in the
limit of in"nite chirality. The low chirality behaviour is given by the model proposed by Meiboom
et al. [479] and subsequently developed by Meiboom and co-workers [480}482]. This model is
characterized as a low-chirality theory, with the proviso that `lowa means low enough for the
helical phase to be only weakly biaxial. A detailed description and the "ndings of these models are
summarized well by Wright and Mermin [44]. Typical theoretical BP phase diagrams as predicted
by the works of Grebel et al. [478](b) are given in Fig. 34. The thermodynamic boundaries between
the isotropic and cholesteric phases and the BPs are given as a function of chirality parameter. As
found experimentally, all boundaries correspond to the "rst-order phase transition. None of the
two phase diagrams include a phase with BP III properties of lacking long-range periodic order.
These phase diagrams show a change of blue phase monomorphism to dimorphism and an increase
of the BP temperature ranges on increasing the chirality. Fig. 34(b) predicts the occurrence of
a third blue phase at very high chirality i'1.3. The topology of the theoretical phase diagrams is
very sensitive to parameter variation and depends on the number of higher harmonic taken into
account in the theoretical treatment [478]. Since the free-energy di!erences between the three blue
phases are extremely small, a universal phase diagram is not expected from a theoretical point of
view. Further, the use of only two characteristic parameters m and ¹H also does not seem to allow
for a universality of the temperature versus P~1 plot because the wide variety of systems with
di!erent molecular structure cannot be taken into account in this way.
It was observed that in low to moderate chirality systems, there occurs a "rst-order phase
transition between the isotropic phase and BP III in chiral liquid crystals. However, recent
experiments [483] have shown that high chirality systems exhibit no transition. Lubensky and
Stark [483] introduced a scalar order parameter St T"S(+]Q) ) QT to describe both the phases
2
and developed a Landau}Ginzburg}Wilson Hamiltonian in t and Q. It has been predicted that
2
the IL}BP III transition belongs to the same universality class (Ising) as the liquid}gas transition.

11.2. Transitions involving ferroelectric liquid crystals

The ferroelectricity in smectic phases requires that the molecules are chiral. The questions related
to the concept of chirality, and its in#uences on the tilted smectic phases are being investigated
252 S. Singh / Physics Reports 324 (2000) 107}269

[29,484] actively at present. However, little is known about the basic thermodynamic functions of
the SH phase and SH}SH phase transition, respectively. A chiral smectic phase is composed of
C C A
optically active molecules and the optical activity arises as a result of molecular asymmetry. When
the S phase is composed of optically active molecules, a macroscopic helical arrangement of the
C
molecules is formed. The helix occurs as a result of a precession of the molecular tilt about an axis
perpendicular to the layer planes as shown in Fig. 7b. When the SH phase is cooled from the SH or
C A
cholesteric or isotropic phase, the spontaneous polarization initially rises very rapidly but there-
after more slowly. In case of second-order phase transition, the spontaneous polarization is
coupled to the change in tilt angle which occurs on cooling the mesophase. The tilt angle varies as
h "h (¹ !¹)a (11.15)
T 0 AC
where h is the tilt angle at the temperature ¹, h is a constant, and the exponent a theoretically is
T 0
equal to 0.5. The spontaneous polarization varies in a similar way as
P "P (¹ !¹)b . (11.16)
S 0 AC
Here again theoretically the exponent bK0.5.
When some of the mesogenic molecules are made to be chiral, the centre of symmetry and mirror
planes are lost and only the C -axis remains (see Fig. 35). As a result, an imbalance arises with
2
respect to the molecular dipoles along the C -axis [484]. The time-dependent alignment of the
2
diploes along the C -axis causes the spontaneous polarization (P ) to develop along this direction
2 4
and parallel to the layer planes. Since each layer essentially has a spontaneous polarization and the
layers are stacked on top of each other in a helical arrangement, the layer polarizations conse-
quently are averaged out to zero and the phase is described as helical [485]. However, if the helix is
unwound, then the layer polarizations point in the same direction and the phase becomes
ferroelectric. From symmetry arguments it has been shown that the spontaneous polarization can
point in one of the two directions along the polar C -axis leading to two possible polarization
2
directions, P (#) and P (!) associated with it (see Fig. 36) [486,487]. The magnitude of P is
4 4 4
mainly dependent on the tilt angle (h) of the phase [487], the size of the dipole at the chiral centre
and the degree of freedom that the chiral centre has to rotate about the long molecular axis.

Fig. 35. The symmetry in the chiral smectic CH phase.

 S. Singh / Physics Reports 324 (2000) 107}269 253

Fig. 36. Polarization direction in the smectic CH phase.

The molecular theory of chiral smectic phases can be constructed by extending those of achiral
counterpart by incorporating the additional interaction term arising from the chiral nature of the
molecules [484}492]. For the SH phase, van der Meer and Vertogen [489] used a chiral interaction
C
term of the type
u "!u (e( ) e( ) (e( ]e( ) r( )!u (e( ) r( ) (e( ) r( ) (e( ]e( ) r( ) . (11.17)
#)*3!- 1 i j i j ij 3 i ij j ij i j ij
Using the dipole-induced dipole model of the S phase, these authors [489] calculated the
C
equilibrium pitch of the structure from the minimization of the total free energy. In the SH phase the
C
pitch increases as the temperature is increased, attaining a maximum value within a few degrees of
¹ H and then rapidly decreases to a small value at the transition temperature. Thus, the direct
CA
in#uence of the temperature variation of the tilt angle h on the pitch can be observed only at
temperatures slightly away from the ¹ H . Assuming that the chiral dipolar molecules have
CA
a banana shape, Osipov and Pikin [490] have developed a model to study the temperature
variation of pitch near ¹ H . They considered that the ferroelectricity in SH phase is induced due to
CA C
chirality and molecular shape asymmetry and calculated the #exoelectric part of the polarization
arising from the bend deformation which is present in a helical arrangement of the tilted molecules.
It is supposed that the steric interactions of the banana-shaped molecules produce a strong
temperature variation of the pitch near ¹ H through the #exoelectric coe$cient. Based on the idea
CA
that the biaxility of the layers which resist twisting goes to zero near ¹ H , Goossens [491] made an
CA
attempt to explain the decrease of pitch close to ¹ H .
CA
Within the framework of phenomenological Landau theory considerable e!orts have been made
[29,69,493}500] to describe the transitions involving SH phase and the temperature variation of
C
pitch, polarization, tilt angle and dielectric susceptibility. Nakagawa [492] has related the Landau
coe$cient with the molecular parameters. Based on the generalized Landau model, the expression
for the excess free-energy density in the SH phase can be written as [29,493}500]
C
1
f H"1Ah2#1Ch4#1Dh6# P2#1gP4!fPh
# 2 4 6 2s 4
0
!1dP2h2!K qh2!K qh4#1K q2h2!kqPh . (11.18)
2 2 4 2 33
254 S. Singh / Physics Reports 324 (2000) 107}269

Eq. (11.18) has 11 phenomenological coe$cients. The tilt angle h plays the role of the primary order
parameter for the SH}S transition. The "rst three terms correspond to the usual Landau
C A
coe$cients for the S }S transition. The polarization P can be considered as another order
A C
parameter. The contribution of the polarization terms to the total free energy is much smaller as
compared to the h dependent terms. The remaining terms arise due to the chirality of the system.
K is the Lifshitz term giving the helical structure. The higher order Lifshitz term K has been
2 4
included to account for the increase in pitch with temperature deep in the SH phase. K is the bend
C 33
elastic constant [37], s the dielectric susceptibility, k the #exoelectric and f the piezoelectric
0
coupling constant. The coupling between the transverse quadrupole moment and the tilt angle has
been accounted by the biquadratic term with coe$cient d whereas the g dependent term has been
added for stability. The thermodynamic properties of the system can be derived by minimizing Eq.
(11.18) with respect to h, P and q. Close to ¹ H , the f term is very important whereas h is very small.
CA
At lower temperatures where h assumes appreciable values, the biquadratic d term becomes very
important. Thus, a cross-over between the two regimes in the variation of pitch with temperature is
obtained and this leads to a maximum value at some temperature.
A phenomenological theory of static and dynamic behaviour in SH phase based on the Landau-
C
type free-energy expansion at the SH}SH phase transition was reported initially by Indenbom et al.
C A
[69] and Zeks [493]. Based on the symmetry of the group representation for the SH, Indenbom et
C
al. proposed a Landau free-energy expansion including leading chiral bilinear coupling terms
between polarization and the tilt angle. This expansion serves as a starting point for the description
of SH}SH transition but fails to explain most of the experimental results, characteristics of SH phase.
C A C
The biquadratic coupling term between tilt and polarization in addition to the bilinear coupling
term was added to the free-energy expansion by Zeks [493]. He showed that the biquadratic terms
are large compared to the chiral bilinear terms. The quadrupolar ordering, which is much larger
than the polar ordering, exists in both the achiral and chiral smectic-C phases. The Zeks model
[493] assumes that the three transverse axes, such as transverse dipole moment, axes of polar and
quadrupolar ordering, are coincident. In general, these axes must point in di!erent directions.
Meister et al. [497] extended the Zeks model by considering the three axes in di!erent directions
and calculated the quadrupolar ordering. The experimental data on the temperature and electric
"eld dependence of the tilt angle and the electroclinic e!ect of the SH and SH phases were analysed
C A
by Gie{elmann and Zugenmaier [498,521] according to the generalized Landau expansion and the
results of a microscopic model [497] for the spontaneous polarization. The results obtained by
these authors, [498] support the evidence of quadrupolar ordering and biquadratic coupling for
the understanding of SH}SH transition. They [498] developed a fast, powerful method to study the
C A
dependence of mean-"eld coe$cients with respect to the electroclinic e!ect, the speci"c heat
singularity and the chirality. The model was extended to temperatures below the phase transition.
It is experimentally con"rmed that the SH}SH transition temperature ¹ H H is higher in chiral
C A CA
compounds in comparison to their non-chiral analogue. It has been shown theoretically by Padmni
et al. [499] that the increase of ¹ H H is due to the bilinear coupling between the tilt and
CA
polarization. Roy et al. [500] have developed a theoretical relationship between S }S phase
C A
transition temperature of chiral smectic compound and that of its non-chiral analogue. These
authors have determined the bilinear and biquadratic coupling constants and the quadrupolar
order parameter from the extended microscopic model [497] in the SH phase. It has been found
C
that both the bilinear and biquadratic couplings are responsible for the shift of the transition
 S. Singh / Physics Reports 324 (2000) 107}269 255

temperature of a chiral compound in comparison to its non-chiral analogue and that the shifting of
transition temperature due to the bilinear coupling is enhanced more by the inclusion of the
biquadratic coupling term in the Landau free-energy expansion. The experimental results show
that the bilinear coupling coe$cient is directly proportional to the reduced polarization, whereas
the biquadratic coupling constant depends on the reduced polarization as well as on the molecular
structure. Mukherjee [500] has calculated the anomalous parts of the speci"c heat capacity of
SH}S phase transition using the #uctuation theory of Landau.
C A
Assuming that a transverse quadrupole order exists in the tilted phase, not necessarily due to the
interaction between the dipoles, Zeks et al. [501] have developed a microscopic model which is
consistent with the Landau expression. The single-particle potential for the rotation of a molecule
around its long axis can be expressed as

u(w)"!a h cos w!a h2 cos 2w , (11.19)

1 2
where w describes the transverse orientation of a molecule. Here the "rst term is like the
piezoelectric coupling of the Landau expansion and the second achiral term gives the quadrupolar
ordering.
Considerable e!ort has been made by several workers [502}507] to understand the e!ect of
chirality on the SH phase. Renn and Lubensky [506] predicted that the chiral smectics could exhibit
C
a dislocated S phase called the twist-grain-boundary (TGB) phase. This was experimentally
A
con"rmed [503}505] by X-ray studies. All these observations are consistent with a twisted stack of
two-dimensional S slabs where the smectic order parameter of a slab is a Landau-orbit solution of
A
the linearized Ginzburg}Landau}de Gennes equations. Further, in several of the compounds and
mixtures exhibiting SH phase the existence of a S }SH}SH multicritical point was observed. In view
A A A C
of the existence of such a multicritical point it is important to investigate whether the TGB phase
obtained by entering from the SH side is the same phase obtained by entering the S side. Lubensky
A A
and Renn [507] addressed the question how the Abrikosov #ux-dislocation lattice behaves when
the ratio of the twist penetration depth to the smectic coherence length i ("j /m) diverges. These
authors examined the approach to the S }S boundary within the framework 2 of the chiral
A C
Chen}Lubensky model [379] of the N}S }S point. Using the same model, Renn [502] derived the
A C
mean-"eld phase diagram of the NH (or Cl)}S }SH point.
A C
It has been shown that the SH phase can exhibit transitions to the TGB , TGB or TGBH phases.
C A C C
In addition, it was found that both the NH-TGB and NH-TGB transitions are possible, but that
A C
the NH-TGBH transition does not occur. He also showed that the transition TGB }TGB is
C A C
replaced by the TGB }TGBH when the cholesteric pitch length P increases beyond &2P /3,
A C C
where P is the SH pitch length.
C C
Although an extensive set of experimental studies have been performed [508}516] on the ferro-
and ferrielectric smectic phases, no clearcut conclusion could be drawn about their phase struc-
tures. Only indirect experiments, such as dielectric [517] and electrooptical measurements or
conoscopic "gure observations [509,518,519] could establish the ferro-, antiferro- or ferrielectric
nature of the reported phases. However, these data fail to provide any insight into the molecular
steps of reorganization from one structure into another. Recently, Lorman et al. [520] have
proposed a model for the description of the observed sequences of ferro-, ferri- and antiferroelectric
smectic phases. The model assumes a bilayer smectic ordering and a mechanism consisting of an
256 S. Singh / Physics Reports 324 (2000) 107}269

azimuthal reorientation of the molecules at the transitions. This mechanism assumes that the tilt
angles remain largely unchanged across the phases and that the molecule can turn freely on cones
possessing "xed vertex angles (conical approximation). It has been found that three helicoidal
ferroelectric structures are possibly stable, in which the helical pitch vary monotonically as
a function of temperature.

12. Overview and perspectives

Liquid crystals exhibit a rich variety of phase transitions which have been studied extensively by
both experimental and theoretical workers. These studies have played an important role in
advancing both our understanding of liquid crystals and our ability to use them in applications. In
this article, an attempt has been made to present a comprehensive analysis of the current
understanding of the phase transitions in liquid crystals. Instead of attempting to compile the latest
results and the published works till date, I have preferred to concentrate on and discuss the
questions related to the concept and ideas employed and the methods developed for its study. More
importantly, the "eld of phase transitions in liquid crystals has grown so vast that it is beyond the
scope of this article to cover all the developments made so far. In view of making the article
self-contained, "rst a brief review on the general information about the liquid crystals has been
given. It describes the molecular structure and types of liquid crystals with speci"c emphasis on the
molecular arrangement and criteria for their classi"cation. The phase sequence, re-entrant behav-
iour, blue phases and phases of chiral smectics and discotics are also described. An e!ort has been
made to explain in a systematic way how the order existing in liquid crystals can be understood in
terms of distribution functions and order parameters. This is followed by a brief discussion on the
thermodynamic properties at and in the vicinity of the phase transitions, which are needed to
generate understanding about the molecular structure phase stability relationship, and on the
critical analysis of most widely used experimental methods for measuring the transition properties.
The remaining part of the article, concerns with the review of the current status of the theoretical
understanding with a particular emphasis on the ideas and concepts involved in the formalism and
the methods adopted for numerical calculations. The Landau}de Gennes (LDG) theory of phase
transitions in liquid crystals has been proved to be very rich in making qualitative predictions. As
these do not depend on the actual values of the phenomenological expansion coe$cients, they test
the general assumption of the theory. The basic ideas of Landau theory are discussed. The strengths
of the Landau}de Gennes theory are its simplicity and ability to predict the most important
features of the phase transitions. The application of the LDG theory to investigate the N}I, N}S ,
A
N }N , S }S transitions and N}S }S multicritical point has been reviewed. The predictions of
6 " A C A C
Landau theory in case of transitions involving hexatic smectic phases and chiral liquid crystals are
also summarized. Since the formation of liquid crystals depends on the anisotropy in the inter-
molecular interactions, the questions concerning its role have been the subject of investigation from
the beginning. The use of the molecular models, i.e., hard-particle-, Maier}Saupe-, and van der
Waals types of theories, to the transitions involving mesophases has been discussed. The density
functional approach provides a convenient way for the theoretical study of a large variety of
problems of ordered phases. The salient features of the density functional theory and its application
within the framework of modi"ed weighted-density approximation to the N}I and N}S
A
 S. Singh / Physics Reports 324 (2000) 107}269 257

transitions have been reviewed. The occurrence of re-entrant phase transition in liquid crystals and
of blue phases in a narrow temperature range in chiral liquid crystals are the examples of unusual
phase sequences observed in mesophase systems. The available information about these unusual
phenomenon is summarized.
Computer simulations play a crucial role in relating experimental observations to the theoretical
results obtained for model systems. It has already made a signi"cant contribution to advancing
the understanding of the behaviour observed near transitions between di!erent liquid crystal
phases. The results obtained from the available simulation studies of phase transitions in liquid
crystals are summarized for the hard-particle-, Lebwohl}Lasher lattice-, and Gay}Berne models.
The current status of the physics of liquid crystalline polymers with particular emphasis to the
theory of nematic order in polymer solutions and melts and interacting semi#exible polymers is
reviewed. We have also summarized the work done on the phase transitions in chiral liquid
crystals.
The future directions of work in this area are likely to be the computer simulations for realistic
potential models and the experimental measurements on biaxial nematic and smectics, hexatic
smectics, ferroelectric, polymer liquid crystalline, discotic nematic and columnar, and lyotropic
phases. Reliable experimental data are essential for the understanding of these phases. From
a theoretical point of view, more important is the extension of existing theories to cholesteric,
smectics, chiral smectics, columnar and polymer liquid crystals. Among the classes of theories
which can be used, for this purpose, although the weighted density functional methods rank near
the top as far as the incorporation of the realistic details of the microscopic interactions is
concerned, it is extremely hard to obtain precise information about the correlation functions. This
limits the applicability of the density functional theory. Numerical simulations are the best hope to
provide answers not only to the many interesting and important questions but also to resolve
elementary issues regarding these phases. With the advancement in the area of superfast computers,
e$cient algorithms and methods of analysing numerical data, simulation studies are becoming
possible and more important. In addition, issues related to the simulations of many-body systems,
such as the maximum size of the system that can be tackled in a simulation, the longest time over
which the numerical experiments can be performed and the accuracy of results, could be resolved
satisfactorily in recent years. All these indicate that computer simulations can play a crucial role in
the study of phase transitions and critical phenomena in liquid crystals.

Acknowledgements

I am grateful to Prof. Y. Singh for many stimulating discussions and to Prof. O.N. Srivastava for
his encouragement and support. The "nancial assistance received from Centre of Advanced
Studies, U.G.C., Government of India (New Delhi) is thankfully acknowledged.

References

[1] F. Reinitzer, Zur Kenntnis des cholesterinus, Monatsch Chem. 9 (1888) 421.
[2] O. Lehmann, Z. Physik Chem. 4 (1889) 462.
258 S. Singh / Physics Reports 324 (2000) 107}269

[3] (a) P.G. de Gennes, The Physics of Liquid Crystals, Clarendon Press, Oxford, 1974; (b) P.G. de Gennes, J. Prost,
The Physics of Liquid Crystals, Clarendon Press, Oxford, 1993.
[4] S. Chandrasekhar, Liquid Crystals, Cambridge Univ. Press, Cambridge, 1977, 1992.
[5] D. Demus, L. Richter, Texture of Liquid Crystals, Leipzig, 1978.
[6] G.R. Luckhurst, G.W. Gray (Eds.), The Molecular Physics of Liquid Crystals, Academic Press, New York, 1979.
[7] M. Kleman, Points, Lines and Walls, Wiley, New York, 1983.
[8] G.W. Gray, J.W. Goodby, Smectic Liquid Crystals: Textures and Structures, London, Leonard Hill, 1984.
[9] H. Sackmann, Polymorphism and phase transitions in liquid crystals, Martin-Luther-Universitat Halle Witten-
berg, Wissenschatil, Beit rage 1986/52 N 17.
[10] J.C. Toledano, P. Toledano, The Landau Theory of Phase Transitions, World Scienti"c, Singapore, 1987.
[11] G.W. Gray (Ed.), Thermotropic Liquid Crystals, Wiley, New York, 1987.
[12] P.S. Pershan, Structure of Liquid Crystal Phases, World Scienti"c, 1988.
[13] G. Vertogen, W.H. de Jeu, Thermotropic Liquid Crystals: Fundamentals, Springer, Berlin, 1988.
[14] P.J. Collings, Liquid Crystals: Nature's Delicate Phase of Matter, Princeton Univ. Press, Princeton, 1990.
[15] (a) B. Bahadur (Ed.), Liquid Crystals: Applications and Uses, World Scienti"c, Singapore Vol. 1 (1990); (b) Vol. 2.
(1991); (c) Vol. 3 (1992).
[16] A.L. Tsykalo, Thermophysical Properties of Liquid Crystals, Gordon and Breach, New York, 1991.
[17] I.C. Khoo (Ed.), Physics of Liquid Crystalline Materials, Gordon and Breach, Amsterdam, 1991.
[18] S. Martelucci, A.N. Chester (Eds.), Phase Transitions in Liquid Crystals, Plenum Press, New York, 1992.
[19] S. Kumar (Ed.), Liquid Crystals in Nineties and Beyond, World Scienti"c, NJ, 1995.
[20] P.J. Collings, M. Hird, Introduction to Liquid Crystals: Physics and Chemistry, Taylor and Francis, London,
1997.
[21] P.J. Collings, J.S. Patel (Eds.), Handbook of Liquid Crystal Research, Oxford Univ. Press, Oxford, 1997.
[22] S. Elston, R. Sambles (Eds.), The Optics of Thermotropic Liquid Crystals, Taylor and Francis, London, 1998.
[23] C.B. McArdle, Side Chain Liquid Crystal Polymers, Blackie, Glasgow, 1989.
[24] A.M. White, A.H. Windle, Liquid Crystal Polymers, Cambridge Univ. Press, Cambridge, 1992.
[25] A.A. Collyer, Liquid Crystal Polymers: From Structures to Applications, Elsevier, Oxford, 1993.
[26] A. Blumstein (Ed.), Polymer Liquid Crystals, Plenum, New York, 1985.
[27] C. Carfagna (Ed.), Liquid Crystals Polymers, Pergamon, Oxford, 1994.
[28] L.A. Beresnev, Ferroelectric Liquid Crystals, Gordon and Breach, New York, 1988.
[29] J.W. Goodby, R. Blinc, N.A. Clark, S.T. Lagerwall, M.A. Osipov, S.A. Pikin, T. Sakurai, K. Yoshino, B. Zeks,
Ferroelectric Liquid Crystals: Principles, Properties and Applications, Gordon and Breach, Philadelphia, 1991.
[30] A. Buka (Ed.), Modern Topics in Liquid Crystals: From Neutron Scattering to Ferroelectricity, World Scienti"c,
Singapore, 1993.
[31] S. Chandrasekhar, B.K. Sadashiva, K.A. Suresh, pramana 9 (1977) 471.
[32] S. Chandrasekhar, Adv. Liq. Cryst. 5 (1982) 47; Philos. Trans. Roy. Soc. London A 309 (1983) 93; Contemp. Phys.
29 (1988) 527.
[33] C. Destrade, P. Foucher, H. Gasparoux, N.H. Tinh, A.M. Levelut, J. Malthete, Mol. Cryst. Liq. Cryst. 106 (1984)
121.
[34] H.T. Nguyen, C. Destrade, H. Gasparoux, Phys. Lett. 77 A (1979) 3.
[35] H. Finkelmann, W. Meier, H. Scheuermann, Liquid crystal Polymers, in: B. Bahadur (Ed.), Liquid Crystals:
Applications and Uses, Vol. 3, World Scienti"c, Singapore, 1992.
[36] G. Friedel, Ann. Phys. 18 (1922) 273.
[37] S. Singh, Phys. Rep. 277 (1996) 283.
[38] M.J. Freiser, Phys. Rev. Lett. 24 (1970) 1041; Mol. Cryst. Liq. Cryst. 14 (1971) 165.
[39] R. Alben, Phys. Rev. Lett. 30 (1973) 778; J. Chem. Phys. 59 (1973) 4299.
[40] L.J. Yu, A. Saupe, Phys. Rev. Lett. 45 (1980) 1000.
[41] S. Chandrasekhar, B.K. Sadashiva, S. Ramesha, B.S. Srikanta, pramana- J. Phys. 27 (1986) L713.
[42] J. Malthete, L. Liebert, A.M. Levelut, C.R. Acad. Sci. (Paris) 303 (1986) 1073.
[43] H. Stegemeyer, Th. Blumel, K. Hiltrop, H. Onusseit, F. Porsch, Liq. Cryst. 1 (1986) 3.
[44] D.C. Wright, N.D. Mermin, Rev. Mod. Phys. 61 (1989) 385.
 S. Singh / Physics Reports 324 (2000) 107}269 259

[45] A.D.L. Chandani, Y. Ouchi, H. Takezoe, A. Fukuda, K. Terashima, K. Furukawa, A. Kishi, Jpn. J. Appl. Phys.
Lett. 28 (1989) 1261.
[46] E. Gorecka, A.D.L. Chandani, Y. Ouchi, H. Takezoe, A. Fukuda, Jpn. J. Appl. Phys. 29 (1990) 131.
[47] S. Inui, S. Kawano, M. Saito, H. Iwana, Y. Takanishi, K. Hiraoka, Y. Ouchi, H. Takezoe, A. Fukuda, Jpn. J. Appl.
Phys. Lett. 29 (1990) 987.
[48] C. Zannoni, in: G.R. Luckhurst, G.W. Gray (Eds.), The Molecular Physics of Liquid Crystals, Acad. Press, New
York, 1979 (Chapter 3).
[49] Y. Singh, Phys. Rep. 207 (1991) 351.
[50] S. Goshen, D. Mukamel, S. Shtrikman, Mol. Cryst. Liq. Cryst. 31 (1975) 171.
[51] J.P. Straley, Phys. Rev. A 10 (1974) 1881.
[52] Y. Singh, K. Rajesh, V.J. Menon, S. Singh, Phys. Rev. E49 (1994) 501.
[53] T.W. Stinson, J.D. Litster, Phys. Rev. Lett. 25 (1970) 503; ibid 30 (1973) 688.
[54] C. DasGupta, Int. J. Mod Phys. 9B (1995) 2219.; J. Phys. 48 (1987).
[55] J.R. Dorfman, T.R. Kirkpatrick, J.V. Sengers, Ann. Phys. Chem. 45 (1994) 213.
[56] E.M. Barrall II, in: F.D. Saeva (Ed.), Liquid Crystals, Mercel Dekker, Inc. 1979, p. 193 (Chapter 9).
[57] R. Alben, Mol. Cryst. Liq. Cryst. 13 (1971) 193.
[58] J.R. McColl, C.S. Shih, Phys. Rev. Lett. 29 (1972) 85.
[59] A. Beguin, J. Billard, F. Bonamy, J.M. Buisine, P. Cuvelier, J.C. Dubois, P.L. Barny, Mol. Cryst. Liq. Cryst. 115
(1984) 1.
[60] H. Zink, W.H. de Jeu, Mol. Cryst. Liq. Cryst. 124 (1985) 287.
[61] S. Singh, Y. Singh, Mol. Cryst. Liq. Cryst. 87 (1982) 211.
[62] P.G. de Gennes, (a) Solid State Comm. 10 (1972) 753; (b) Phys. Lett. 30 A (1969) 454.
[63] P. Sheng, E.B. Priestley, The Landau de Gennes theory of liquid crystal phase transition, in: E.B. Priestley, P.J.
Wojtowicz (Eds.), Introduction to Liquid Crystals, Plenum, New York, 1974, p. 143.
[64] E.F. Gramsbergen, L. Longa, W.H. de Jeu, Phys. Rep. 135 (1986) 195.
[65] L.D. Landau, Phys. Z Sowjefunion 11 (1937) 26 (in: D. Ter Haar (Ed.), Collected papers of L.D. Landau, 2nd
Edition, Gordon and Breach, Science Publisher, New York, 1967, pp. 193}216).
[66] L.D. Landau, E.M. Lifshitz, Statistical Physics, Vol. 1, 3rd Edition, Pergamon, Oxford, 1980.
[67] N. Boccara, Symmetries Brisses, Hermann, Paris, 1976.
[68] M.J. Stephen, J.P. Stratey, Rev. Mod. Phys. 46 (1974) 617.
[69] V.L. Indenbom, S.A. Pikin, E.B. Liginov, Sov. Phys. Crystallogr. 21 (1976) 632.
[70] S.A. Pikin, V.L. Indenbom, Sov. Phys. Usp. 21 (1978) 487.
[71] V.L. Indenbom, E.B. Liginov, M.A. Osipov, Sov. Phys. Crystallogr. 26 (1981) 656.
[72] V.L. Indenbom, E.B. Liginov, Sov. Phys. Crystallogr. 26 (1981) 526.
[73] W. Helfrich, Phys. Rev. Lett. 24 (1970) 201.
[74] C. Rosenblatt, Phys. Rev. A 24 (1981) 2236.
[75] A.J. Nacastro, P.H. Keyes, Phys. Rev. A 30 (1984) 3156.
[76] P.K. Mukherjee, T.B. Mukherjee, Phys. Rev. B 52 (1995) 9964.
[77] A.D. Rzoska, S.J. Rzoska, J. Ziolo, Phys. Rev. E 54 (1996) 6452.
[78] P.K. Mukherjee, Private commun.
[79] M.A. Anisimov, Mol. Cryst. Liq. Cryst. 162 (1986) 1.
[80] P.K. Mukherjee, J. Saha, B. Nandi, M. Saha, Phys. Rev. B 50 (1994) 9778.
[81] P.K. Mukherjee, M. Saha, Phys. Rev. 51E (1995) 5745.
[82] R.G. Priest, T.C. Lubensky, Phys. Rev. B 13 (1976) 4159.
[83] P.K. Mukherjee, M. Saha, Mol. Cryst. Liq. Cryst. 307 (1997) 103.
[84] K.G. Wilson, Phys. Rev. B 4 (1971) 3174, 3184.
[85] P.K. Mukherjee, T.R. Bose, D. Ghosh, M. Saha, Phys. Rev. 51 E (1995) 4570.
[86] B. Nandi, P.K. Mukherjee, M. Saha, Mod. Phys. Lett. B 10 (1996) 777.
[87] P.K. Mukherjee, Mod. Phys. Lett. B 11 (1997) 107.
[88] R. Tao, P. Sheng, Z.F. Lin, Phys. Rev. Lett. 70 (1993) 1271.
[89] D.E. Martire, G.A. Oweimreen, G.I. Agren, S.G. Ryam, H.T. Peterson, J. Chem. Phys. 64 (1976) 1456; 72 (1980) 2500.
260 S. Singh / Physics Reports 324 (2000) 107}269

[90] P.H. Keyes, J.R. Shane, Phys. Rev. Lett. 22 (1979) 722.
[91] D.R. Nelson, R.A. Pelcovits, Phys. Rev. B 16 (1977) 2191.
[92] P.B. Vigman, A.I. Larkin, V.M. Filev, Sov. Phys. JETP 41 (1976) 944.
[93] Z.H. Wang, P.H. Keyes, Phys. Rev. 54E (1996) 5249.
[94] P.K. Mukherjee, Ph.D. Thesis, Calcutta University, 1996.
[95] G. Vertogen, W.H. de Jeu, Thermotropic Liquid Crystals: Fundamentals, Springer, Berlin, 1988 (Chapter 12).
[96] T.E. Faber, Proc. Roy Soc. London Ser. A 375 (1982) 579.
[97] L. Onsager, Ann. N.Y. Acad. Sci 51 (1949) 627.
[98] R. Zwanzig, J. Chem. Phys. 39 (1963) 1714.
[99] G. Lasher, J. Chem. Phys. 53 (1970) 4141.
[100] L.K. Runnels, C. Colvin, J. Chem. Phys. 53 (1970) 4219.
[101] P.J. Flory, Proc. Roy. Soc. A 234 (1956) 73.
[102] P.J. Flory, R. Ronca, Mol. Cryst. Liq. Cryst. 54 (1979) 289.
[103] (a) M.A. Cotter, D.E. Martire, J. Chem. Phys. 52 (1970) 1902, 1909, 4500; (b) M.A. Cotter, Phys. Rev. A10 (1974)
625.
[104] M.A. Cotter, Hard particle theories of nematics, in: G.R. Luckhurst, G.W. Gray (Eds.), The Molecular Physics of
Liquid Crystals, Acad. Press, New York 1979, p. 169.
[105] B. Barboy, W.M. Gelbart, J. Chem. Phys. 71 (1979) 3053.
[106] B.M. Mulder, D. Frenkel, Mol. Phys. 55 (1985) 1193.
[107] J.D. Parson, Phys. Rev. A 19 (1979) 1225.
[108] M. Warner, Mol. Cryst. Liq. Cryst. 80 (1982) 79.
[109] B.J. Berne, P. Pechukas, J. Chem. Phys. 56 (1972) 4213.
[110] S.D. Lee, (a) J. Chem. Phys. 87 (1987) 4972; (b) 89 (1988) 7036.
[111] A. Wulf, A.G. de Rocco, J. Chem. Phys. 55 (1971) 12.
[112] (a) F. Dowell, D.E. Martire, J. Chem. Phys. 68 (1978) 1088; (b) F. Dowell, Phys. Rev. A 28 (1983) 3520.
[113] S. Tang, G.T. Evans, J. Chem. Phys. 99 (1993) 5336.
[114] W.M. Gelbart, A. Ben Shaul, J. Chem. Phys. 77 (1982) 916.
[115] A. Saupe, J. de Phys. Colloq. 40 (1979) C3.
[116] J.P. Straley, Mol. Cryst. Liq. Cryst. 24 (1973) 7.
[117] W. Maier, A. Saupe, Z. Naturforsch. 14a (1959) 882; 15a (1960) 287.
[118] B. Widom, J. Chem. Phys. 39 (1963) 2808.
[119] M.A. Cotter, Mol. Cryst. Liq. Cryst. 39 (1977) 173.
[120] T.J. Krieger, H.M. James, J. Chem. Phys. 22 (1954) 796.
[121] S. Chandrasekhar, N.V. Madhusudana, Acta Crystallogr. A 27 (1971) 303.
[122] (a) R.L. Humphries, P.G. James, G.R. Luckhurst, J. Chem. Soc. Faraday Trans. II 68 (1972) 1031; (b) G.R.
Luckhurst, C. Zannoni, P.L. Nordio, U. Segre, Mol. Phys. 30 (1975) 1345.
[123] S. Marcelja, J. Chem. Phys. 60 (1974) 3599.
[124] N.V. Madhusudana, S. Chandrasekhara, Solid State Comm. 13 (1973) 377.
[125] J.G.J. Ypma, G. Vertogen, H.T. Koster, Mol. Cryst. Liq. Cryst. 37 (1976) 57.
[126] P. Sheng, P.J. Wojtowicz, Phys. Rev. A 14 (1976) 1883.
[127] N.V. Madhusudana, K.L. Savithramma, S. Chandrasekhara, Pramana 8 (1977) 22.
[128] (a) G.R. Luckhurst, Molecular "eld theories of nematics, in: G.R. Luckhurst, G.W. Gray (Eds.), The Molecular
Physics of Liquid Crystals, Acad. Press, New York, 1979, p. 85; (b) G.R. Luckhurst, C. Zannoni, Nature 267 (1977)
412.
[129] (a) A. Wulf, J. Chem. Phys. 64 (1976) 104; (b) J.L. Kaplan, E. Drauglis, Chem. Phys. Lett. 9 (1971) 645.
[130] R. Van der Haegen, J. Debruyne, R. Luyckx, H.N.N. Lekkerkerker, J. Chem. Phys. 73 (1980) 2469.
[131] N.V. Madhusudana, Theories of liquid crystals, in: B. Bahadur (Ed.), Liquid Crystals: Applications and Uses, Vol. 1,
World Scienti"c, Singapore, 1990, p. 37 (Chapter 2).
[132] (a) M.A. Cotter, The Van der Waals approach to nematic liquid crystals, in: G.R. Luckhurst, G.W. Gray (Eds.),
The Molecular Physics of Liquid Crystals, Acad. Press, New York, 1979, p. 181 (Chapter 8); (b) M.A. Cotter, J.
Chem. Phys. 66 (1977) 1098, 4710.
 S. Singh / Physics Reports 324 (2000) 107}269 261

[133] (a) W.M. Gelbart, B.A. Baron, J. Chem. Phys. 66 (1977) 207; (b) W.M. Gelbart, A. Gelbart, Mol. Phys. 33 (1977)
1387; (c) B.A. Baron, W.M. Gelbart, J. Chem. Phys. 67 (1977) 5795.
[134] J.G.J. Ypma, G. Vertogen, Phys. Rev. A 17 (1978) 1490.
[135] W. Warner, J. Chem. Phys. 73 (1980) 6327.
[136] K.L. Savithramma, N.V. Madhusudana, (a) Mol. Cryst. Liq. Cryst. 62 (1980) 63; (b) ibid 97 (1983) 407.
[137] (a) S. Singh, K. Singh, Mol. Cryst. Liq. Cryst. 101 (1983) 77; (b) K. Singh, S. Singh, Mol. Cryst. Liq. Cryst. 108
(1984) 133; (c) S. Singh, T.K. Lahiri, K. Singh, Mol. Cryst. Liq. Cryst. 225 (1993) 361.
[138] (a) P.J. Flory, R. Ronca, Mol. Cryst. Liq. Cryst. 54 (1979) 311; (b) F. Dowell, Phys. Rev. A 28 (1983) 1003; (c) F.
Dowell, Phys. Rev. A 31 (1985) 2464, 3214.
[139] P. Pal!y-Muhoray, B. Bergesen, Phys. Rev. A 6 (1987) 2704.
[140] M.D. Lipkin, D.W. Oxtoby, J. Chem. Phys. 79 (1983) 1939.
[141] T.J. Slucking, P. Shukla, J. Phys. A 16 (1983) 1539.
[142] (a) Y. Singh, Phys. Rev. 30A (1984) 583; (b) U.P. Singh, Y. Singh, Phys. Rev. 33A (1986) 2725; (c) J. Ram, Y. Singh,
Phys. Rev. 44A (1991) 3718; (d) J. Ram, R.C. Singh, Y. Singh, Phys. Rev. 49E (1994) 5117; (e) R.C. Singh, J. Ram, Y.
Singh, Phys. Rev. 54E (1996) 977; (f) R.C. Singh, Ph.D. Thesis, Banaras Hindu University, 1998; (g) R.C. Singh, J.
Ram, Y. Singh, in preparation.
[143] A. Perera, G.N. Patey, J.J. Weis, J. Chem. Phys. 89 (1988) 6941.
[144] A.M. Somoza, P. Tarazona, J. Chem. Phys. 91 (1989) 517.
[145] U.P. Singh, U. Mohanty, Y. Singh, Physica 158A (1989) 817.
[146] F.C. Andrews, J. Chem. Phys. 62 (1975) 272.
[147] P.A. Lebwohl, G. Lasher, Phys. Rev. 6A (1972) 426.
[148] J. Vieillard Baron, Mol. Phys. 28 (1974) 809.
[149] D.W. Rebertus, K.M. Sando, J. Chem. Phys. 67 (1977) 2585.
[150] T. Boublik, I. Nezbeda, O. Tynko, Czech. J. Phys. B 26 (1976) 1081.
[151] P.A. Monson, M. Rigby, Mol. Phys. 35 (1978) 1337.
[152] C. Zannoni, (a) Computer simulations, in: G.R. Luckhurst, G.W. Gray (Eds.), The Molecular Physics of Liquid
Crystals, Acad. Press, New York, 1979, p. 191 (Chapter 6); (b) J. Chem. Phys. 84 (1986) 424.
[153] I. Nezbeda, T. Boublik, Czech, J. Phys. B 28 (1978) 353.
[154] H.J.F. Jensen, G. Vertogen, J.G.J. Ypma, Mol. Cryst. Liq. Cryst. 38 (1977) 87.
[155] (a) G.R. Luckhurst, P. Simpson, Mol. Phys. 47 (1982) 251; (b) G.R. Luckhurst, S. Romano, Proc. Roy. Soc. London
A 373 (1980) 111.
[156] (a) D. Frenkel, B.M. Mulder, Mol. Phys. 55 (1985) 1171; (b) D. Frenkel, B.M. Mulder, J.P. Mc Tague, Phys. Rev.
Lett. 52 (1984) 287; (c) D. Frenkel, H.N.W. Lekkerkerker, A. Stroobants, Nature London 322 (1988) 822.
[157] (a) R. Eppenga, D. Frenkel, Mol. Phys. 52 (1984) 1303; (b) D. Frenkel, R. Eppenga, Phys. Rev. 31A (1985) 1776.
[158] M.P. Allen, D. Frenkel, Phys. Rev. Lett. 58 (1987) 1748.
[159] A. Stroobants, H.N.W. Lekkerkerker, D. Frenkel, (a) Phys. Rev. A 36 (1987) 2929; (b) Phys. Rev. Lett. 57 (1986)
1452.
[160] D. Frenkel, (a) Mol. Phys. 60 (1987) 1, (b) J. Phys. Chem. 92 (1988) 3280; (c) Liq. Cryst. 5. (1989) 929.
[161] A. Beguin, J.C. Dubois, P. Le Barny, J. Billard, F. Bonamy, J.M. Busisine, P. Cuvelier, Sources of thermodynamic
data on mesogens, Mol. Cryst. Liq. Cryst. 115 (1984) 1}326.
[162] G.R. Luckhurst, in: L.L. Chapoy (Ed.), Recent Advances in Liquid Crystalline Polymers, Elsevier, London 1986, p.
105.
[163] B. Deloche, B. Cabane, D. Jerome, Mol. Cryst. Liq. Cryst. 15 (1971) 197.
[164] R.G. Horn, J. de Physique 39 (1978) 199, R.G. Horn, T.E. Faber, Proc. Roy. Soc. (London) A 368 (1979) 199.
[165] (a) J.D. Bunning, D.A. Crellin, T.E. Faber, Liq. Cryst. 1 (1986) 37, (b) B. Bergersen, P. Pal!y-Muhoray, D.A.
Dunmur, Liq. Cryst. 3 (1988) 347.
[166] H.C. Andersen, D. Chandler, J.D. Weeks, Adv. Chem. Phys. 34 (1976) 105.
[167] B. Tjipto-Margo, G.T. Evans, Mol. Phys. 74 (1991) 85.
[168] M. Nakagawa, T. Akahane, Mol. Cryst. Liq. Cryst. 90 (1982) 53.
[169] P.J. Flory, R. Ronca, Mol. Cryst. Liq. Cryst. 54 (1979) 311.
[170] A. Bellemans, Phys. Rev. Lett. 21 (1968) 527.
262 S. Singh / Physics Reports 324 (2000) 107}269

[171] V.T. Rajan, C.W. Woo, Phys. Rev. A 17 (1978) 382, L. Feijoo, V.T. Rajan, C.W. Woo, Phys. Rev. A 19 (1979) 1263.
[172] (a) D.R. Evans, G.T. Evans, D.K. Ho!man, J. Chem. Phys. 94 (1991) 8816; (b) G.T. Evans, E.B. Smith, Mol. Phys.
74 (1991) 79.
[173] B. Tjipto-Margo, G.T. Evans, J. Chem. Phys. 94 (1990) 4546.
[174] (a) M. Baus, J.L. Colot, X.G. Wu, H. Xu, Phys. Rev. Lett. 59 (1987) 2184, (b) J.L. Colot, X.G. Wu, H. Xu, M. Baus,
Phys. Rev. A 38 (1988) 2022.
[175] G. Vertogen, W.H, de Jeu, Thermotropic Liquid Crystals, Fundamentals, Springer, Berlin, 1988, p. 245
(Chapter 13).
[176] J.A. Barker, D. Henderson, Rev. Mod. Phys. 48 (1976) 587.
[177] J.F. Marko, (a) Phys. Rev. Lett. 60 (1988) 325; (b) Phys. Rev. 39A (1989) 2050.
[178] C.N. Likos, N.W. Ashcroft, J. Chem. Phys. 99 (1993) 9090.
[179] P. Honenberg, W. Kohn, Phys. Rev. 136B (1964) 864.
[180] W. Kohn, L.J. Sham, Phys. Rev. A 140 (1965) 1133.
[181] D. Mermin, Phys. Rev. 137 (1964) 1441.
[182] J.L. Lebowitz, J.K. Percus, J. Math. Phys. 4 (1963) 116.
[183] W.F. Saam, C. Ebner, Phys. Rev. A 15 (1977) 2566.
[184] A.J.M. Yang, P.D. Fleming, J.H. Gibbs, J. Chem. Phys. 64 (1976) 3732.
[185] J.K. Percus, in: H.L. Frisch, J.L. Lebowitz (Eds.), The Equilibrium Theory of Classical Fluids, Benjamin, New
York, 1964.
[186] R. Evans, Adv. Phys. 28 (1979) 143.
[187] F.F. Abraham, Phys. Rep. 53 (1979) 93.
[188] J.S. Rowlinson, B. Widom, Molecular Theory of Capillarity, Clarendon Press, Oxford, 1982.
[189] A.D.J. Haymet, Ann. Rev. Phys. Chem. 38 (1987) 89.
[190] M. Baus, J. Phys. Condens. Matter 2 (1990) 2111.
[191] R.G. Parr, W. Yang, Density Functional Theory of Atoms and Molecules, Oxford Univ. Press, New York, 1989.
[192] R.O. Jones, O. Gunnarsson, Rev. Mod. Phys. 61 (1989) 689.
[193] T. Ziegler, Chem. Rev. 91 (1991) 651.
[194] J.K. Labanowski, J. Andzelm (Eds.), Density Functional Methods in Chemistry, Springer, New York, 1991.
[195] B.G. Johnson, P.M.W. Gill, J.A. Pople, J. Chem. Phys. 98 (1993) 5612.
[196] S. Nordholm, M. Johnson, B.C. Freasier, Aust. J. Chem. 33 (1980) 2139.
[197] P. Tarazona, (a) Mol. Phys. 52 (1984) 81; (b) Phys. Rev. A 31 (1985) 2672; erratum 32 (1985) 3148.
[198] W.A. Curtin, N.W. Ashcroft, Phys. Rev. A 32 (1985) 2909.
[199] A.R. Denton, N.W. Ashcroft, Phys. Rev. A 39 (1989) 426, 4701.
[200] T.V. Ramakrishnan, M. Yussou!, Phys. Rev. 19B (1979) 2775.
[201] A.D.J. Haymet, J. Chem. Phys. 78 (1983) 4641.
[202] G.L. Jones, U. Mohanty, Mol. Phys. 54 (1985) 1241.
[203] M. Baus, J.L. Colot, Mol. Phys. 55 (1985) 653.
[204] W.R. Smith, D. Henderson, Mol. Phys. 3 (1970) 411.
[205] (a) A. Perera, P.G. Kusalik, G.N. Patey, J. Chem. Phys. 87 (1987) 1295; 89 (1988) 5969; (b) A. Perera, G.N. Patey, J.
Chem. Phys. 89 (1988) 5961; (c) J. Talbot, A. Perera, G.N. Patey, Mol. Phys. 70 (1990) 285.
[206] D. Frenkel, in: J.P. Hansen, D. Levesque, J. Zinn-Justin (Eds.), Liquids, Freezing and Glass Transition II, 1989,
Les Houches Lectures, North-Holland, Amsterdam, 1991.
[207] (a) E. DeMiguel, L.F. Rull, M.K. Chalam, K.E. Gubbins, F.V. Swol, Mol. Phys. 72 (1991) 593; (b) E. DeMiguel,
L.F. Rull, M.K. Chalam, K.E. Gubbins, Mol. Phys. 74 (1991) 405.
[208] G.R. Luckhurst, R.A. Stephens, R.W. Phippen, Liq. Cryst. 8 (1990) 451.
[209] F.J. Rogers, D.A. Young, Phys. Rev. 30A (1984) 999.
[210] R. Holyst, A. Poniewierski, Mol. Phys. 68 (1989) 381.
[211] W.L. McMillan, Phys. Rev. A 7 (1973) 1419.
[212] J. Als-Nielsen, R.J. Birgeneau, M. Kaplan, J.D. Litster, C.R. Sa"nya, Phys. Rev. Lett. 39 (1977) 352.
[213] P. Brisbin, R. De Ho!, T.E. Lockhart, D.L. Johnson, Phys. Rev. Lett. 43 (1979) 1171.
[214] R.J. Birgeneau, C.W. Garland, G.B. Kasting, B.M. Octo, Phys. Rev. 24A (1981) 2624.
 S. Singh / Physics Reports 324 (2000) 107}269 263

[215] H. Morynissen, J. Thoen, W. Van Dael, Mol. Cryst. Liq. Cryst. 97 (1983).
[216] J.M. Viner, C.C. Huang, Solid State Comm. 39 (1981) 789.
[217] B.M. Ocko, R.J. Birgeneau, J.D. Litster, M.E. Neubert, Phys. Rev. Lett. 52 (1984) 208.
[218] K.K. Chan, P.S. Pershan, L.B. Sorensen, F. Hardouin, Phys. Rev. Lett. 54 (1985) 1694, Phys. Rev. 34A (1986) 1420.
[219] K.K. Chan, M. Deutsch, B.M. Ocko, P.S. Pershan, L.B. Sorensen, Phys. Rev. Lett. 54 (1985) 920.
[220] M. Miranda, L.J. Kortan, R.J. Birgeneau, Phys. Rev. Lett. 56 (1986) 2264.
[221] S.B. Rananavare, V.G.K.M. Pisipati, J.H. Freed, Chem. Phys. Lett. 140 (1987) 255.
[222] (a) C.W. Garland, G. Nouneiss, K.J. Stine, G. Heppke, J. Phys. (Paris) 50 (1989) 2291; (b) C.W. Garland, G.
Nouneiss, Phys. Rev. E 49 (1994) 2964.
[223] L. Chen, J.D. Brock, J. Huang, S. Kumar, Phys. Rev. Lett. 67 (1991) 2037.
[224] G. Nouneiss, K.I. Blum, M.J. Young, C.W. Garland, R.J. Birgencau, Phys. Rev. E 47 (1993) 1910.
[225] L. Wu, M.J. Young, Y. Shao, C.W. Garland, R.J. Birgencau, Phys. Rev. Lett. 72 (1994) 376.
[226] W.L. McMillan, Phys. Rev. 6A (1972) 936.
[227] B.I. Halperin, T.C. Lubensky, Solid State Comm. 14 (1974) 997.
[228] J.H. Chen, T.C. Lubensky, Phys. Rev. A 14 (1976) 1202.
[229] P.E. Cladis, W.A. Sarloos, D.A. Huse, J.S. Patel, J.W. Goodby, P.L. Finn, Phys. Rev. Lett. 62 (1989) 1764.
[230] J. Prost, Adv. Phys. 33 (1984) 1.
[231] R.F. Bruinsma, Phys. Rev. A 43 (1991) 5377.
[232] (a) P.K. Mukherjee, Mol. Cryst. Liq. Cryst. 307 (1997) 1; (b) P.K. Mukherjee, K. Mukhopadhyay, Phys. Rev. E, in
press.
[233] K.K. Kobayashi, J. Phys. Soc. Jpn. 29 (1970) 101; Mol. Cryst. Liq. Cryst. 13 (1971) 137.
[234] W.L. McMillan, Phys. Rev. 4A (1971) 1238. Phys. Rev. 23B (1971) 363.
[235] R.B. Meyer, T.C. Lubensky, Phys. Rev. A 14 (1976) 2307.
[236] J. Katriel, G.F. Kventsel, Phys. Rev. A 28 (1983) 3037.
[237] A. Kloczkowki, J. Stecki, Mol. Phys. 55 (1985) 689.
[238] D.A. Badalyan, Sov. Phys. Crystallogr. 27 (1982) 10.
[239] C.D. Mukherjee, B. Bagchi, T.R. Bose, D. Gosh, M.K. Roy, M. Saha, Phys. Lett. 92A (1982) 403.
[240] W. Wagner, Mol. Cryst. Liq. Cryst. 98 (1983) 247.
[241] M. Nakagawa, T. Akahane, J. Phys. Soc. Jpn. 54 (1985) 69.
[242] G.F. Kventsel, G.R. Luckhurst, H.B. Zewdie, Mol. Phys. 56 (1985) 589.
[243] M. Hosino, H. Nakano, H. Kimura, J. Phys. Soc. Jpn. 46 (1979) 740, 1709.
[244] X. Wen, R.B. Meyer, Phys. Rev. Lett. 59 (1987) 1325.
[245] N.V. Madhusudana, B.S. Srikanta, M. Subramanya, Raj Urs, Mol. Cryst. Liq. Cryst. 108 (1984) 19.
[246] R. Holyst, A Poniewierski, Mol. Phys. 71 (1990) 561.
[247] I.I. Ptichkin, Sov. Phys. Crystallogr. 36 (1991) 134.
[248] M.P. Taylor, R. Hentschke, J. Herzfeld, Phys. Rev. Lett. 62 (1989) 800.
[249] (a) F. Dowell, Phys. Rev. 28A (1983) 3520, 3526; (b) C. Rosenblatt, D. Ronis, Phys. Rev. A 23 (1981) 305; D. Ronis,
C. Rosenblatt, Phys. Rev. A 21 (1980) 1687.
[250] (a) F.T. Lee, H.T. Tan, Y.M. Shih, C.W. Woo, Phys. Rev. Lett. 31 (1973) 1117; (b) L. Senbetu, C.W. Woo, Phys.
Rev. A 17 (1978) 1529.
[251] K. Sokalski, Physica 113A (1982) 133.
[252] C.D. Mukherjee, T.R. Bose, D. Ghosh, M.K. Roy, M. Saha, Mol. Cryst. Liq. Cryst. 124 (1985) 139.
[253] E.M. Everyanov, A.N. Primak, (a) Liq. Cryst. 10 (1991) 555; (b) ibid 13 (1993) 139.
[254] H. Wang, M.Y. Jin, R.C. Jarnagin, T.J. Bunning, W. Adams, B. Cull, Y. Shi, S. Kumar, E.T. Samulski, Nature 238
(1996) 244.
[255] B. Mulder, Phys. Rev. A 35 (1987) 3095.
[256] R. Holyst, A. Poniewierski, Phys. Rev. A 39 (1989) 2742.
[257] A.M. Somoza, P. Tarazona, Phys. Rev. Lett. 61 (1988) 2566.
[258] A. Poniewierski, R. Holyst, Phys. Rev. Lett. 61 (1988) 2461.
[259] M.C. Mahato, M. Roy Lakshmi, R. Pandit, H.R. Krishnamurthy, Phys. Rev. A 38 (1988) 1049.
[260] R.G. Priest, Mol. Cryst. Liq. Cryst. 37 (1976) 101.
264 S. Singh / Physics Reports 324 (2000) 107}269

[261] F. Rondelez, Solid State Comm. 12 (1972) 1675.

[262] R. Alben, Solid State Comm. 13 (1973) 1783.
[263] B.I. Halperin, T.C. Lubensky, S.K. Ma, Phys. Rev. Lett. 32 (1974) 292.
[264] (a) C.W. Garland, G. Nounesis, Phys. Rev. 49E (1994) 2964; (b) L. Longa, J. Chem. Phys. 85 (1986) 2974.
[265] I. Lelidis, G. Durand, J. Phys. II (Paris) 6 (1996) 1359.
[266] M.A. Anisimov, V.P. Voronov, A.O. Kulkov, F. Kholmurodov, J. Phys. (Paris) 46 (1985) 2137.
[267] M.A. Anisimov, Critical Phenomena in Liquids and Liquid Crystals, Gordon and Breach, New York, 1991, p. 10.
[268] W.H. de Jeu, NATO School on Phase Transitions in Liquid Crystals, Erice, Italy, May 1991.
[269] C. Das Gupta, B.I. Halperin, Phys. Rev. Lett. 47 (1981) 1556.
[270] T.C. Lubensky, J. Chem. Phys. 80 (1983) 31.
[271] D.R. Nelson, J. Toner, Phys. Rev. B 24 (1981) 363.
[272] J. Toner, Phys. Rev. B 26 (1982) 462.
[273] B.R. Patton, B.S. Andereck, Phys. Rev. Lett. 69 (1992) 1556.
[274] B.S. Andereck, B.R. Patton, Phys. Rev. 49E (1994) 1393.
[275] J.G. Kirkwood, E. Monroe, J. Chem. Phys. 9 (1941) 514.
[276] M. Nakagawa, T. Akahane, J. Phys. Soc. Japan 53 (1984) 1951.
[277] J.W. Emsley, R. Hashim, G.R. Luckhurst, G.N. Rumbles, F.R. Viloria, Mol. Phys. 49 (1983) 1321.
[278] J.W. Emsley, R. Hashim, G.R. Luckhurst, G.N. Shilstone, Liq. Cryst. 1 (1986) 437.
[279] J.W. Emsley, G.R. Luckhurst, G.N. Shilstone, I. Sage, J. Chem. Soc. Faraday Soc. Trans. II 83 (1987) 371.
[280] D. Catalano, C. Forte, C.A. Veracini, J.W. Emsley, G.N. Shilstone, Liq. Cryst. 2 (1987) 357.
[281] E.M. Everyanov, V.A. Gunyakov, Opt. Spectrosc. 66 (1989) 72.
[282] J.W. Emsley, G.R. Luckhurst, H.S. Sachdev, Liq. Cryst. 5 (1989) 953.
[283] M.P. Fontana, B. Rosi, N. Kirov, I. Dozov, Phys. Rev. A 33 (1986) 4132.
[284] P.E. Cladis, (a) Phys. Rev. Lett. 35 (1975) 48; (b) Philos. Mag. 29 (1975) 641; (c) in: S. Chandrasekhar (Ed.),
Proceedings of International Conference on Liquid Crystals, Bangalore, Heyden, London, 1980, p. 105.
[285] P.E. Cladis, R.K. Begardus, W.B. Daniels, G.N. Taylor, Phys. Rev. Lett. 39 (1977) 720.
[286] D. Guillon, P.E. Cladis, J. Stamato!, Phys. Rev. Lett. 41 (1978) 1598.
[287] N.S. Chen, S.K. Hark, J.T. Ho, Phys. Rev. A 24 (1981) 2843.
[288] T. Narayanan, A. Kumar, Phys. Rep. 249 (1994) 135.
[289] A. Deerenberg, J.A. Schoutem, N.J. Trappeniers, Physica 103A (1980) 183.
[290] J.A. Schouton, Phys. Rep. 172 (1989) 33.
[291] R.J. Tufeu, P.H. Keyes, W.B. Daniels, Phys. Rev. Lett. 35 (1975) 1004.
[292] J.D. Cox, J. Chem. Soc. 4 (1952) 606.
[293] T. Narayanan, A. Kumar, E.S.R. Gopal, Phys. Lett. A 155 (1991) 276.
[294] R.G. Johnston, N.A. Clark, P. Wiltzins, D.S. Cannel, Phys. Rev. Lett. 54 (1985) 49.
[295] V.P. Zaitsev, S.V. Krivokhizha, I.L. Fabelinskii, A. Tsitrovskii, L.L. Chaikov, E.V. Shvelts, P. Yani, Sov. Phys.
JETP Lett. 43 (1986) 112.
[296] S.V. Krivokhizha, O.A. Dugovaya, I.L. Fabelinskii, L.L. Chaikov, A. Tsitrovskii, P. Yani, Sov. Phys. JETP 76
(1993) 62.
[297] G.B. Kozlov, E.B. Kryukova, S.P. Lobedev, A.A. Sobyanin, Sov. Phys. JETP 67 (1988) 1689.
[298] M.J. Naughton, R.V. Chamberlin, X. Yan, S.Y. Ysu, L.Y. Chiang, M.Y. Azbel, P.M. Chaikin, Phys. Rev. Lett. 61
(1988) 621.
[299] T.H. Lin. X.Y. Shao, M.K. Wu. P.H. Hoy, X.C. Jin, C.W. Chu, N. Evans, R. Bayuzick, Phys. Rev. B 29 (1984)
1493.
[300] S. Katayama, Y. Hirokawa, T. Kanaka, Macromolecules 17 (1984) 2649.
[301] H. Glusbrenner, H. Weingartner, J. Phys. Chem. 93 (1989) 3378.
[302] R.B. Gri$ths, in: R.E. Mills, E. Ascher, R.I. Ja!ee (Eds.), Critical Phenomenon in Alloys, Magnets and Supercon-
ductors, McGraw-Hill, New York, 1971, p. 377.
[303] P.E. Cladis, Mol. Cryst. Liq. Cryst. 165 (1988) 85.
[304] L. Longa, W.H. de Jeu, Phys. Rev. A 26 (1982) 1632.
[305] F.R. Bouchet, P.E. Cladis, Mol. Cryst. Liq. Cryst. (Lett.) 64 (1980) 97.
 S. Singh / Physics Reports 324 (2000) 107}269 265

[306] K.J. Lushington, G.B. Kasting, C.W. Garland, Phys. Rev. B 22 (1980) 2569.
[307] R.Y. Dong, Mol. Cryst. Liq. Cryst. (Lett.) 64 (1981) 205.
[308] R. Shashidhar. H.D. Kleinhans, G.M. Schneider, Mol. Cryst. Liq. Cryst. (Lett.) 72 (1981) 119.
[309] P.E. Cladis, Phys. Rev. A 23 (1981) 2594.
[310] A.R. Kortan, H.V. Kanel, R.J. Birgeneau, J.D. Lister, Phys. Rev. Lett. 47 (1981) 1206.
[311] N. Ha"z, N.A.P. Vaz, Z. Yaniv, D. A!ender, J.W. Doane, Phys. Lett. A 91 (1982) 411.
[312] J.W. Emsley, G.R. Luckhurst, P.J. Parson, B.A. Timimi, Mol. Phys. 56 (1985) 767.
[313] C.S. Hudson, Z. Phys. Chem. 47 (1904) 113.
[314] G. Illian, H. Kneppe, F. Schneider, Ber. Bunsenges, Phys. Chem. 92 (1988) 776.
[315] F. Dowell, Phys. Rev. 36A (1987) 5046, (b) 38A (1988) 382.
[316] G. Pelzl, S. Diele, I. Latif, W. Weiss!og, D. Demus. Cryst. Res. Technol. 17 (1982) K 78.
[317] S. Diele, G. Pelzl, I. Latif, D. Demus, Mol. Cryst. Liq. Cryst. (Lett.) 92 (1983) 27.
[318] G. Sigaud, F. Hardouin, M. Mauzac, N.H. Tinh, Phys. Rev. A 33 (1986) 789.
[319] F. Hardouin, A.M. Levelut, M.F. Achard, G. Sigaud, J. Chim. Phys. 80 (1983) 53.
[320] C. Destrade, J. Malthete, N.H. Tinh, H. Gasparoux, Phys. Lett. 78A (1980) 82.
[321] N.H. Tinh, J. Malthete, C. Destrade, J. de Physique Lett. 42 (1981) L-417.
[322] K.W. Evans-Lutterodt, J.W. Chung, B.M. Ocko, R.J. Birgeneau, C. Chiang, C.W. Garland, E. Chin, J. Goodby,
N.H. Tinh, Phys. Rev. A 36 (1987) 1387.
[323] F. Hardouin, G. Sigaud, M.F. Achard, H. Gasparoux, Phys. Lett. 71 (1979) 347.
[324] N.H. Tinh, J. Chim. Phys. 80 (1983) 83.
[325] R. Shashidhar, B.R. Ratna, V. Surendranath, V.N. Raja, K.S. Prasad, C. Nagabhushana, J. Phys. Lett. (Paris) 46
(1985) L445.
[326] N.H. Tinh, F. Hardouin, C. Destrade, A.M. Levelut, J. Phys. Lett. (Paris) 43 (1982) L33.
[327] S. Chandrasekhar, Proceedings of 10th International Liquid Crystal Conference, York, 1984; Mol. Cryst. Liq.
Cryst. 124 (1985) 1.
[328] A. Nayeem, J.H. Freed, J. Phys. Chem. 93 (1989) 65.
[329] G.R. Luckhurst, B.A. Timimi, Mol. Cryst. Liq. Cryst. (Lett.) 64 (1981) 253.
[330] W.H. de Jeu, Solid State Comm. 41 (1982) 529.
[331] T.R. Bose, D. Ghosh, C.D. Mukherjee, J. Saha, M.K. Roy, M. Saha, Phys. Rev. A 43 (1991) 4372.
[332] L.V. Miramtser, Mol. Cryst. Liq. Cryst. 133 (1986) 151.
[333] J.O. Indekeu, A.N. Berker, Phys. Rev. A 33 (1986) 1158; J. de Physique 49 (1988) 353.
[334] A.N. Berker, J.S. Walker, Phys. Rev. Lett. 47 (1981) 1469.
[335] J.O. Indekeu, A.N. Berker, C. Chiang, C.W. Garland, Phys. Rev. A 35 (1987) 1371.
[336] N.V. Madhusudana, J. Rajan, Liq. Cryst. 7 (1990) 31.
[337] J. Prost, P. Barois, J. Chim. Phys. 80 (1983) 65.
[338] P.S. Person, J. Prost, J. Phys. Lett. (Paris) 40 (1979) L-27.
[339] J. Katriel, G.F. Kventsel, Mol. Cryst. Liq. Cryst. 124 (1985) 179.
[340] R.F.M. Houtappel, Physica 16 (1950) 425.
[341] D.W. Allender, M.A. Lee, N. Ha"z, Mol. Cryst. Liq. Cryst. 124 (1985) 45.
[342] P.K. Kukherjee, Liq. Cryst. 22 (1997).
[343] A. Saupe, P. Boonbrahm, L.J. Yu, J. Chem. Phys. 80 (1983) 7.
[344] P. Boonbraham, A. Saupe, J. Chem. Phys. 81 (1984) 2076.
[345] F. Biscarini, C. Chiccoli, P. Pasini, F. Semeria, C. Zannoni, Phys. Rev. Lett. 75 (1995) 1803.
[346] P.O. Quist, Liq. Cryst. 18 (1995) 623.
[347] S. Chandrasekhar, Mol. Cryst. Liq. Cryst. 124 (1985) 1.
[348] C.A. Cajas, J.B. Swift, H.R. Brand, Phys. Rev. A 30 (1984) 1579.
[349] C.H. Shih, R. Alben, J. Chem. Phys. 57 (1972) 3055.
[350] C. Vause, J. Sak, Phys. Rev. 18B (1979) 1455; Phys. Lett. 65A (1978) 183.
[351] Y. Rabin, W.E. McMullen, W.M. Gelbart, Mol. Cryst. Liq. Cryst. 89 (1982) 67.
[352] D.L. Johnson, D. Allender, R. De Ho!, C. Maze, E. Oppenheim, R. Reynolds, Phys. Rev. 16B (1977) 470.
[353] R.G. Cal"sh, Z.Y. Chen, A. Berker, J.M. Deutch, Phys. Rev. A 30 (1984) 2562.
266 S. Singh / Physics Reports 324 (2000) 107}269

[354] Z.Y. Chen, J.M. Deutch, J. Chem. Phys. 80 (1984) 2151.

[355] E.A. Jacobsen, J. Swift, Mol. Cryst. Liq. Cryst. 87 (1982) 29.
[356] S.R. Sharma, P. pal!y Muhoray, B. Bergerson, D.A. Dunmur, Phys. Rev. A 23 (1985) 3752.
[357] T.C. Lubensky, Mol. Cryst. Liq. Cryst. 146 (1987) 55.
[358] B.M. Mulder, Th. W. Ruijgrok, Physica A 113 (1982) 145.
[359] I. Dozov, N. Kirov, B. Petro!, Phys. Rev. A 36 (1987) 2870.
[360] R. Holyst, A. Poniewierski, Mol. Phys. 69 (1990) 193.
[361] C. Tanford, The Hydrophobic E!ect, Wiley, New York, 1973.
[362] T.R. Taylor, J.L. Fergason, S.L. Arora, Phys. Rev. Lett. 24 (1970) 359; 25 (1970) 722.
[363] E.I. Kats, V.V. Lebedav, Physica A135 (1986) 601.
[364] P.G. de Gennes, (a) C.R. Acad. Sci 274B (1972) 758; (b) Mol. Cryst. Liq. Cryst. 21 (1973) 49.
[365] C.C. Huang, S.C. Lien, Phys. Rev. Lett. 47 (1981) 1917.
[366] A. Wulf, Phys. Rev. 11A (1975) 365.
[367] W.L. McMillan, Phys. Rev. 8A (1973) 1921.
[368] D. Cabib, L. Benguigui, J. Phys. (Paris) 38 (1977) 419.
[369] B.W. van der Meer, G. Vertogen, J. Phys. (Paris) 40 (1979) C3-222.
[370] R.G. Priest, J. Phys. (Paris) 36 (1975) 437; J. Chem. Phys. 65 (1976) 408.
[371] M. Matsushita, J. Phys. Soc. Jpn. 50 (1981) 1351.
[372] W.J.A. Goossens, (a) J. Phys. (Paris) 46 (1985) 1411; (b) Euro. Phys. Lett. 3 (1987) 341; Mol. Cryst. Liq. Cryst. 150
(1987) 419.
[373] G. Sigaud, F. Hardouin, M.F. Achard, Solid State Comm. 23 (1977) 35.
[374] D. Johnson, D. Allender, D. Deho!, C. Maze, E. Oppenheim, R. Reynolds, Phys. Rev. B 16 (1977) 470.
[375] D.L. Johnson, J. Chim. Physique 80 (1983) 45.
[376] R. Shashidhar, B.R. Ratna, S. Krishna Prasad, Phys. Rev. Lett. 53 (1984) 2141.
[377] M.A. Anisimov, Mol. Cryst. Liq. Cryst. 162A (1988) 1.
[378] K.C. Chu, W.L. McMillan, Phys. Rev. 15A (1977) 1181.
[379] J. Chen, T.C. Lubensky, Phys. Rev. 14A (1976) 1202.
[380] J. Prost, J.P. Marceron, J. Phys. (Paris) 38 (1977) 315.
[381] L. Benguigui, J. Phys. (Paris) Colloq. 40 (1979) C3.
[382] C.C. Huang, J.M. Viner, Phys. Rev. A 25 (1982) 3385.
[383] G. Grinstein, J. Toner, Phys. Rev. Lett. 51 (1983) 2386.
[384] X. Wen, C.W. Garland, M.D. Wand, Phys. Rev. A 42 (1990) 6087.
[385] R. Pindak, D.E. Moncton, S.C. Davey, J.W. Goodby, Phys. Rev. Lett. 46 (1981) 1135.
[386] (a) C.C. Huang, J.M. Viner, R. Pindak, J.W. Goodby, Phys. Rev. Lett. 46 (1981) 1289; (b) T. Pitchford, G. Nounesis,
S. Dumrograttana, J.M. Viner, C.C. Huang, J.W. Goodby, Phys. Rev. A 32 (1985) 1938.
[387] (a) T. Pitchford, C.C. Huang, R. Pindak, J.W. Goodby, Phys. Rev. Lett. 57 (1986) 1239; (b) R. Geer, C.C. Huang, R.
Pindak, J.W. Goodby, Phys. Rev. Lett. 63 (1989) 540.
[388] (a) R. Geer, C.C. Huang, R. Pindak, J.W. Goodby, Phys. Rev. Lett. 63 (1989) 540; (b) R. Geer, T. Stoebe, C.C.
Huang, R. Pindak, S.W. Goodby, M. Cheng, J.T. HO, S.W. Hui, Lett. Nature 355 (1992) 152.
[389] H. Pleiner, H.R. Brand, Phys. Rev. A 39 (1989) 1563.
[390] J. Wang, T.C. Lubensky, Phys. Rev. A 29 (1984) 2210.
[391] M.P. Allen, (a) Liq. Cryst. 8 (1990) 499; (b) Philos. Trans. Roy. Soc. London A 344 (1993) 323.
[392] R.R. Netz, A.N. Berker, (a) Phys. Rev. Lett. 68 (1992) 333; (b) in: S. Martellucci, A.N. Chester (Eds.), Phase
Transitions in Liquid Crystals, Plenum Press, New York, 1992.
[393] D. Frenkel, in: S. Martellucci, A.N. Chester (Eds.), Phase Transitions in Liquid Crystals, New York, Plenum, 1992,
p. 67.
[394] U. Fabbri, C. Zannoni, Mol. Phys. 58 (1986) 763.
[395] F. Biscarini, C. Zannoni, C. Chiccoli, P. Pasini, Mol. Phys. 73 (1991) 439.
[396] Z. Zhang, O.G. Mouritsen, M.J. Zuckermann, Phys. Rev. Lett. 69 (1992) 1803.
[397] Z. Zhang, M.J. Zuckermann, O.G. Mouritsen, Mol. Phys. 80 (1993) 1195.
[398] J.A.C. Veerman, D. Frenkel, (a) Phys. Rev. A 45 (1992) 5632; (b) Phys. Rev. A 41 (1990) 3237.
 S. Singh / Physics Reports 324 (2000) 107}269 267

[399] E. Demiguel, L.F. Rull, K.E. Gubbins, Phys. Rev. A 45 (1992) 3813.
[400] G.R. Luckhurst, P.S.J. Simmonds, Mol. Phys. 80 (1993) 233.
[401] R. Memmer, H.G. Kuball, A. Schonhofer, (a) Liq. Cryst. 15 (1993) 345; (b) Ber. Bunsen Ges. Phys. Chem. 97 (1993)
1193.
[402] A.L. Tsykalo, Mol. Cryst. Liq. Cryst. 129 (1985) 409.
[403] D.J. Cleaver, D. Kralj, T.J. Sluckin, M.P. Allen, in: G.P. Crawford, S. Zumer (Eds.), Liquid Crystals in Complex
Geometries Formed by Polymer and Porus Network, (Taylor and Francis, London, 1995).
[404] Y.Y. Goldschmidt, in: D.H. Ryan (Ed.), Recent Progress in Random Magnets, World Scienti"c, Singapore, 1992, p.
151.
[405] G.J. Zarragoicocoechea, D. Levesque, J.J. Weis, Mol. Phys. 75 (1992) 989.
[406] J.A. Cuesta, D. Frenkel, Phys. Rev. A 42 (1990) 2126.
[407] J.Y. Dunham, G.R. Luckhurst, C. Zannoni, J.W. Lewis, Mol. Cryst. Liq. Cryst. 60 (1980) 185.
[408] P.E. Lammert, D.S. Rokhsar, J. Toner, Phys. Rev. Lett. 70 (1993) 1650.
[409] C. Dasgupta, Phys. Rev. Lett. 55 (1985) 1771; J. Phys. (Paris) 48 (1987) 957.
[410] R.J. Birgeneau, J.D. Litster, J. Phys. Lett. 39 (1978) L399.
[411] R. Pindak, D.E. Moncton, S.C. Davey, J.W. Goodby, Phys. Rev. Lett. 46 (1981) 1135.
[412] I.M. Jiang, S.N. Huang, J.Y. KO, T. Stoebe, A.J. Jin, C.C. Huang, Phys. Rev. E 48 (1993) 3240.
[413] A.P.J. Emerson, G.R. Luckhurst, S.G. Whatling, Mol. Phys. 82 (1994) 113.
[414] T. Odijk, Macromolecules 19 (1986) 2313.
[415] H.N.W. Lekkerkerker, P. Coulon, R. Van der Haegen, R. Deblieck, J. Chem. Phys. 80 (1984) 3427.
[416] A.N. Semenov, A.R. Kokhlov, Sov. Phys. Usp. 31 (1988) 988.
[417] G.J. Vroege, H.N.W. Lekkerkerker, Rep. Prog. Phys. 55 (1992) 1241.
[418] R. Deblieck, H.N.W. Lekkerkerker, J. Phys. Lett. (Paris) 41 (1980) 351.
[419] J.K. Moscicki, G. Williams, Polymer 24 (1983) 85.
[420] T. Odijk, H.N.W. Lekkerkerker, J. Phys. Chem. 89 (1985) 2090.
[421] W.E. McMullen, W.M. Gelbart, A. Ben Shaul, J. Chem. Phys. 82 (1985) 5616.
[422] P.J. Flory, A. Abe, Macromolecule 11 (1978) 1119.
[423] A. Abe, P.J. Flory, Macromolecule 11 (1978) 1122.
[424] P.J. Flory, R.S. Frost, Macromolecule 11 (1978) 1126.
[425] R.S. Frost, P.J. Flory, Macromolecule 11 (1978) 1134.
[426] T.M. Birshtein, B.I. Kolegov, V.A. Pryamitsyn, Polym. Sci. 30 (1988) 316.
[427] A.R. Khokhlov, Phys. Lett. A 68 (1978) 135.
[428] A.R. Khokhlov, A.N. Semenov, Physica 108A (1981) 526; Macromolecules 17 (1984) 2678.
[429] A.R. Khokhlov, A.N. Semenov, Physica 112A (1982) 605.
[430] S.K. Nechaev, A.N. Semenov, A.R. Khokhlov, Polymer Sci. 25 (1983) 1063.
[431] G.J. Vroege, T. odijk, Macromolecules 21 (1988) 2848.
[432] R.R. Matheson, P.J. Flory, Macromolecule 14 (1981) 954.
[433] T.M. Birshetin, A.R. Merkureva, Polym. Sci. USSR 27 (1985) 1208.
[434] A.R. Khokhlov, Polym. Sci USSR 21 (1979) 2185.
[435] M. Warner, P.J. Flory, J. Chem. Phys. 73 (1980) 6327.
[436] A.R. Khokhlov, A.N. Semenov, J. Statist. Phys. 38 (1985) 161.
[437] A.R. Khokhlov, A.N. Semenov, Macromolecules 19 (1986) 373.
[438] V.V. Rusakov, M.I. Shliomis, J. Phys. Lett. (Paris) 46 (1985) 935.
[439] X.J. Wang, M. Warner, J. Phys. A 19 (1986) 2215.
[440] S.V. Vasilenko, V.P. Shibaev, A.R. Khokhlov, Makromol. Chem. Rapid Comm. 3 (1982) 917.
[441] S.V. Vasilenko, A.R. Khokhlov, V.P. Shibaev, (a) Polym. Sci. 26 (1984) 606; (b) Macromolecules 17 (1984) 2270.
[442] P. Corradini, M. Vacatello, Mol. Cryst. Liq. Cryst. 97 (1983) 119.
[443] V.V. Rusakov, M.I. Shliomis, Polym. Sci. USSR 29 (1987)
[444] V.P. Shibaev, N.A. Plate, Adv. Polym. Sci 60/61 (1984) 173.
[445] H. Finkelmann, G. Rehage, Adv. Polym. Sci 60/61 (1984) 99.
[446] N.A. Plate, V.P. Shibaev, Comblike Polymers and Liquid Crystals, Khimiya, Moscow, 1980 (in Russian).
268 S. Singh / Physics Reports 324 (2000) 107}269

[447] P.J. Collings, J.S. Patel (Eds.), Handbook of Liquid Crystal Research, Oxford Univ. Press, Oxford, 1997, p. 71
(Chapter 3); p. 259 (Chapter 8).
[448] R.G. Petschek, Phys. Rev. A 34 (1986) 1338.
[449] Y.H. Kim, P. Pincus, Biopolymers 18 (1979) 2315.
[450] X.J. Wang, M. Warner, Liq. Cryst. 12 (1992) 385.
[451] W.J.A. Goossens, (a) Mol. Cryst. Liq. Cryst. 12 (1971) 231; (b) J. de Physique 40 (1979) C3-158.
[452] B.W. Van der Meer, G. Vertogen, A.J. Dekker, J.G.J. Ypma, J. Chem. Phys. 65 (1976) 3935.
[453] B.W. Van der Meer, G. Vertogen, A molecular model for the cholesteric mesophase, in: G.R. Luckhurst, G.W.
Gray (Eds.), The Molecular Physics of Liquid Crystals, Academic Press, New York, 1979, p. 149 (Chapter 6).
[454] H. Schrodinger, A molecular "eld theory of the cholesteric liquid crystal state in: G.R. Luckhurst, G.W. Gray
(Eds.), The Molecular Physics of Liquid Crystals, Academic Press, New York, 1979, p. 121 (Chapter 5).
[455] J.P. Straley, Phys. Rev. A 14 (1976) 1835.
[456] G. Vertogen, B.W. Van der Meer, Physica 99A (1979) 237.
[457] S.A. Brazovskii, S.G. Dmitriev, Sov. Phys. JETP 42 (1976) 497.
[458] P.M.L.O. Scholte, G. Vertogen, Physica 113A (1982) 587.
[459] G.T. Evans, Mol. Phys. 77 (1992) 969.
[460] R.A. Pelcovits, P.J. Collings, J.S. Patel (Eds.), Handbook of Liquid Crystal Research, Oxford Univ. Press, Oxford,
1997, p. 71 (Chapter 3).
[461] O. Lehmann, Z. Phys. Chem. 56 (1906) 750.
[462] H. Stegemeyer, K. Bergmann, Springer Ser. Phys. Chem., Vol. 11, 1980, p. 161.
[463] (a) K. Bergmann, H. Stegemeyer, Z. Naturforsch. (a) 34 (1979) 251; (b) 34 (1979) 1031; K. Bergmann, P. Pollmann,
G. Scherer, H. Stegemeyer, Z. Naturforsch. (a) 34 (1979) 253.
[464] M. Marcus, J. Phys. (Paris) 42 (1981) 61.
[465] S. Meiboom, M. Sammon, Phys. Rev. A 24 (1981) 468.
[466] P.J. Collings, Phys. Rev. A 30 (1990) 1984; Mol. Cryst. Liq. Cryst. 113 (1984) 277.
[467] R.N. Kleiman, D.J. Bischop, R. Pindak, P. Taborek, Phys. Rev. Lett. 53 (1984) 2137.
[468] H. Onusseit, H. Stegemeyer, Z. Naturforsch. (a) 39 (1984) 658.
[469] P.P. Croopker, Mol. Cryst. Liq. Cryst. 98 (1983) 31.
[470] S.A. Brazovskii, S.G. Dmitriev, Sov. Phys. JETP 42 (1975) 497.
[471] S.A. Brazovskii, V.M. Filev, Sov. Phys. JETP 48 (1979) 573.
[472] R.M. Hornreich, S. Shtrikman, (a) Bull. ISR. Phys. Soc. 25 (1979) 46, (b) in: W. Helfrich, G. Heppke (Eds.), Liquid
Crystals of One and Two dimensional Order, Springer, Berlin, 1980, p. 185; (c) J. Physique 41 (1980) 335; (d) Phys.
Lett. 82A (1981); 345; (e) Phys. Lett. 84A (1981), 20, (f) J. Physique 42 (1981) 367; (g) Mol. Cryst. Liq. Cryst. 165
(1988) 183.
[473] R.M. Hornreich, M. Kugler, S. Shtrikman, Phys. Rev. Lett. 498 (1982) 1404.
[474] S. Alexander, in: N. Boccara (Ed.), Symmetry and brocken Symmetry in Condended Matter Physics, IDSET,
Paris, 1981, p. 141.
[475] S. Alexander, R.M. Hornreich, S. Shtrikman, in: N. Boccara (Ed.), Symmetry and Brocken Symmetry in
Condensed Matter Physics, IDSET, Paris, 1981, p. 379.
[476] H. Kleinert, K. Maki, Fortschr. Phys. 29 (1981) 219.
[477] R.M. Hornreich, M. Kugler, S. Shtrikman, Phys. Rev. Lett. 498 (1982) 1404.
[478] H. Grebel, R.M. Hornreich, S. Shtrikman, (a) Phys. Rev. A 28 (1983) 1114, 3669; (b) Phys. Rev. A 30 (1984)
3264.
[479] S. Meiboom, J.P. Sethna, P.W. Anderson, W.F. Brinkman, Phys. Rev. Lett. 46 (1981) 1216.
[480] S. Meiboom, M. Sammon, D.W. Berreman, Phys. Rev. A 28 (1983) 3553.
[481] S. Meiboom, M. Sammon, B.F. Brinkman, Phys. Rev. A 27 (1983) 438.
[482] J.P. Sethna, Phys. Rev. B 31 (1985) 6278.
[483] (a) Z. Kutnjak, C.W. Garland, J.L. Passmore, P.J. Collings, Phys. Rev. Lett. 74 (1995) 4859; J.B. Becker, P.J.
Collings, Mol. Cryst. Liq. Cryst. 265 (1995) 163. (b) T.C. Lubensky, H. Stark, Phys. Rev. E 53 (1996) 714.
[484] A.W. Hall, J. Hollingshurst, J.W. Goodby, P.J. Collings, J.S. Patel (Eds.), Handbook of Liquid Crystal Research,
Oxford Univ. Press, Oxford, 1997, p. 40 (Chapter 2).
 S. Singh / Physics Reports 324 (2000) 107}269 269

[485] H.R. Brand, P.E. Cladis, P.L. Finn, Phys. Rev. A 31 (1988) 361.
[486] N.A. Clark, S.T. Langerwall, Ferroelectrics 59 (1984) 25.
[487] (a) J.S. Patel, J.W. Goodby, Philos. Mag. Lett. 55 (1987) 283; (b) J.W. Goodby, J. Mater. Chem. 1 (1991) 307.
[488] L.M. Blinov, L.A. Beresnev, Sov. Phys. Usp. 27 (1985) 492.
[489] B.W. Van der Meer, G. Vertogen, Phys. Lett. 74A (1979) 239.
[490] M.A. Osipov, S.A. Pikin, (a) Sov. Phys. Crystallogr. 26 (1981) 147; (b) Sov. Phys. Tech. Phys. 27 (1982) 109; (c) Mol.
Cryst. Liq. Cryst. 103 (1983) 57.
[491] W.J.A. Goossens, Liq. Cryst. 1 (1986) 521.
[492] M. Nakagawa, Liq. Cryst. 3 (1988) 63.
[493] B. Zeks, Mol. Cryst. Liq. Cryst. 114 (1984) 259.
[494] C.C. Huang, S. Dumrongrattana, Phys. Rev. A 34 (1986) 5020.
[495] T. Carlsson, B. Zeks, A. Levstik, C. Filipik, I. Levstik, R. Blinc, Phys. Rev. A 36 (1987) 1484.
[496] T. Carlsson, B. Zeks, C. Filipic, A. Levstik, R. Blinc, Mol. Cryst. Liq. Cryst. 163 (1988) 11.
[497] R. Meister, H. Stegemeyer, Ber. Bunsenges, Phys. Chem. 97 (1993) 1242.
[498] F. Giebelmann, P. Zugenmaier, Phys. Rev. E 52 (1995) 1762.
[499] H.P. Padmani, N.V. Madhusudana, B. Shivkumar, Bull. Mater. Sci. 17 (1994) 1119.
[500] (a) S.S. Roy, S.K. Roy, P.K. Mukherjee, Int. J. Mod. Phys. B 11 (1997) 3491; (b) P.K. Mukherjee, Private commun.
[501] B. Zeks, T. Carlsson, C. Filipic, B. Urbane, Ferroelectrics 84 (1988) 3; B. Urbane, B. Zeks, Liq. Cryst. 5 (1989) 1075.
[502] S.R. Renn, Phys. Rev. A 45 (1992) 953.
[503] J.W. Goodby, M.A. Waugh, S.M. Stein, E. Chin, R. Pindak, J.S. Patel, Nature 337 (1988) 449; J. Am. Chem. Soc
111 (1989) 8119.
[504] G. Srajer, R. Pindak, M.A. Waugh, J.W. Goodby, J.S. Patel, Phys. Rev. Lett. 64 (1990) 1545.
[505] O.D. Lavrentovich, Y.A. Nishtishin, V.I. Kulishov, Y.S. Narkevich, A.S. Tolochko, S.Y. Shiyanovskii, Europhys.
Lett. 13 (1990) 313.
[506] S.R. Renn, T.C. Lubensky, Phys. Rev. A 38 (1988) 2132.
[507] T.C. Lubensky, S.R. Renn, Phys. Rev. A 41 (1990) 4392.
[508] M. Fukui, H. Orihara, Y. Yamada, N. Yamamoto, Y. Ishibashi, Jpn. J. Appl. Phys. 28 (1989) L849.
[509] E. Gorecka, A.D.L. Chandani, Y. Ouchi, H. Takezoe, A. Fukuda, Jpn. J. Appl. Phys. 29 (1990) 131.
[510] A. Suzuki, H. Orihara, Y. Ishibashi, Y. Yamada, N. Yamamoto, K. Mori, K. Nakamura, Y. Suzuki, T. Hagiwara,
Y. Kawamura, M. Fukui, Jpn. J. Appl. Phys. 29 (1990) L336.
[511] M. Johno, Y. Ouchi, H. Takezoe, A. Fukuda, Jpn. J. Appl. Phys. 29 (1990) L111.
[512] S. Inui, S. Kawano, M. Saito, H. Iwane, Y. Takanishi, K. Hiraoka, Y. Ouchi, H. Takezoe, A. Fukuda, Jpn. J. Appl.
Phys. 29 (1990) L987.
[513] H. Orihara, T. Fujikawa, Y. Ishibashi, Y. Yamada, N. Yamamota, K. Mori, K. Nakamura, Y. Suzuki, T.
Hagiwara, I. Kawamura, Jpn. J. Appl. Phys. 29 (1990) L333.
[514] Y. Takanishi, K. Hiraoka, V.K. Agrawal, H. Takezoe, A. Fukuda, M. Matsushita, Jpn. J. Appl. Phys. 30 (1991)
2023.
[515] J. Lee, Y. Ouchi, H. Takezoe, A. Fukuda, J. Watanabe, J. Phys. Cond. Matter 2 (1990) L103.
[516] M. Hara, T. Umemoto, H. Takezie, A.F. Garito, H. Sasabe, Jpn. J. Appl. Phys. 30 (1991) L2052.
[517] K. Hiraoka, A.D.L. Chandani, E. Gorecka, Y. Ouchi, H. Takezoe, A. Fukuda, Jpn. J. Appl. Phys. 29 (1990) L1473.
[518] N. Okabe, Y. Suzuki, I. Kawamura, I. Isozaki, H. Takezoe, A. Fukada, Jpn. J. Appl. Phys. 31 (1992) L793.
[519] T. Isozaki, H. Hiraoka, Y. Takanishi, A. Fukuda, Y. Suzuki, I. Kawamura, Liq. Cryst. 12 (1992) 59.
[520] V.L. Lorman, A.A. Bulbitch, P. Toledano, Phys. Rev. E 49 (1994) 1367.
[521] F. Giebelmann, P. Zugenmaier, Phys. Rev. E 55 (1997 5613.