Professional Documents
Culture Documents
Liquid Crystals
Liquid Crystals
classic book Liquid crystals, first published in 1977. In this second edition
the author has brought the subject completely up to date.
The subject of liquid crystals has now grown to become an exciting
interdisciplinary field of research with important practical applications.
This book presents a systematic and self-contained treatment of the
physics of the different types of thermotropic liquid crystals - the three
classical types, nematic, cholesteric and smectic, composed of rod-shaped
molecules, and the newly discovered discotic type composed of disc-shaped
molecules. The coverage includes a description of the structures of these
four main types and their polymorphic modifications, their thermodynamical, optical and mechanical properties and their behaviour under
external fields. The basic principles underlying the major applications of
liquid crystals in display technology (for example, the twisted and
supertwisted nematic devices, the surface stabilized ferroelectric device,
etc.) and in thermography are also discussed.
This book will be of great value to advanced students and research workers
in condensed matter physics, chemical physics, materials science and
technology with an interest in the physics, chemistry and applications of
liquid crystals.
LIQUID CRYSTALS
LIQUID CRYSTALS
Second edition
S. CHANDRASEKHAR, F.R.S.
Centre for Liquid Crystal Research, Bangalore
CAMBRIDGE
UNIVERSITY PRESS
UP
Contents
xiii
xv
1
1
1
8
10
12
14
17
17
17
22
27
29
38
38
41
43
51
57
60
61
61
63
66
viii
2.5.4
2.5.5
2.6
2.6.1
2.6.2
2.7
3
3.1
3.1.1
3.1.2
3.1.3
3.1.4
3.2
3.3
3.4
3.4.1
3.4.2
3.4.3
3.4.4
3.5
3.5.1
3.5.2
3.5.3
3.5.4
3.5.5
3.5.6
3.5.7
3.5.8
3.5.9
3.6
3.6.1
3.6.2
3.6.3
3.6.4
3.6.5
3.7
3.8
Contents
Flow birefringence
Comparison with the Maier-Saupe theory
Near-neighbour correlations: Bethe's method
The Krieger-James approximation
Antiferroelectric short-range order
The nematic liquid crystal free surface
Continuum theory of the nematic state
The Ericksen-Leslie theory
Conservation laws and the entropy inequality
Constitutive equations
Coefficients of viscosity
Parodi's relation
Curvature elasticity: the Oseen-Zocher-Frank equations
Summary of equations of the continuum theory
Distortions due to magnetic and electric fields: static
theory
The Freedericksz effect
The twisted nematic device
The Freedericksz effect in highly anisotropic nematics:
periodic distortions
The constant k13
Disclinations
Schlieren textures
Interaction between disclinations
Non-singular structures (s = 1): escape in the third
dimension
Twist disclinations
Singular points
Interaction between disclinations and surfaces
Defects in the presence of an external field
Some consequences of elastic anisotropy
The core structure
Flow properties
Miesowicz's experiment
Tsvetkov's experiment
Poiseuille
flow
Shear
flow
Transverse pressure and secondary
flow
Reflexion of shear waves
Dynamics of the Freedericksz effect
69
70
71
71
75
80
85
85
86
88
91
93
94
97
98
98
106
113
115
117
117
120
123
126
129
132
135
139
143
144
144
145
148
152
157
159
161
Contents
3.8.1
3.8.2
3.9
3.9.1
3.9.2
3.9.3
3.10
3.10.1
3.10.2
3.10.3
3.10.4
3.10.5
3.11
3.11.1
3.11.2
3.12
3.13
3.13.1
3.13.2
3.13.3
4
4.1
4.1.1
4.1.2
4.1.3
4.1.4
4.1.5
4.1.6
4.2
4.2.1
4.2.2
4.2.3
4.3
4.4
Twist deformation
Homeotropic to planar transition: backflow and kickback
effects
Light scattering
Orientational
fluctuations
Intensity and angular dependence of the scattering
Eigenmodes and the frequency spectrum of the scattered
light
Electrohydrodynamics
The experimental situation
Helfrich's theory
DC excitation
Square wave excitation
Sinusoidal excitation
Hydrodynamic instabilities
Homogeneous instability in shear
flow
Roll instability in shear
flow
Thermal instability: stationary convection
Flexoelectricity
Determination of the flexoelectric coefficients
Influence of flexoelectricity on electrohy drody namic
instability
Order electricity
Cholesteric liquid crystals
Optical properties
Propagation along the optic axis for wavelengths ^ pitch
Propagation along the optic axis for wavelengths ~pitch:
analogy with Darwin's dynamical theory of X-ray
diffraction
Exact solution of the wave equation for propagation along
the optic axis: the Mauguin-Oseen-de Vries model
Equivalence of the continuum and the dynamical theories
Oblique incidence
Propagation normal to the optic axis
Defects
/-disclinations
Lattice disclinations
Dislocations
Leslie's theory of thermomechanical coupling
The Lehmann rotation phenomenon
ix
161
162
167
167
170
173
177
177
183
186
187
192
19 5
195
197
202
205
205
210
211
213
213
213
222
237
241
245
247
248
248
252
254
258
262
x
4.5
4.5.1
4.5.2
4.6
4.6.1
4.6.2
4.6.3
4.7
4.8
4.9
4.10
5
5.1
5.2
5.3
5.3.1
5.3.2
5.3.3
5.3.4
5.3.5
5.3.6
5.3.7
5.4
5.4.1
5.4.2
5.4.3
5.4.4
5.5
5.5.1
5.5.2
5.6
5.6.1
5.6.2
5.6.3
5.7
5.8
Contents
Flow properties
Flow along the helical axis
Flow normal to the helical axis
Distortions of the structure by external fields
Magnetic field normal to the helical axis: the
cholesteric-nematic transition
Magnetic field along the helical axis: the square grid
pattern
Electric field along the helical axis
Anomalous optical rotation in the isotropic phase
The blue phases
Some factors influencing the pitch
Molecular models
Smectic liquid crystals
Classification of the smectic phases
Extension of the Maier-Saupe theory to smectic A:
McMillan's model
Continuum theory of smectic A
The basic equations
The Peierls-Landau instability
The Helfrich deformation
Fluctuations and Rayleigh scattering
Damping rate of the undulation mode
Ultrasonic propagation and Brillouin scattering
Breakdown of conventional hydrodynamics
Defects in smectic A
Focal conic textures
Edge dislocations
Screw dislocations
Disclinations
The smectic A-nematic transition
Phenomenological theory of the smectic A-nematic
transition
Pretransition effects in the nematic phase
Smectic A polymorphism
Smectic A phases of strongly polar molecules
The phenomenon of re-entrance
Smectic A* or the twist grain boundary phase
The hexatic phase
Smectic C
267
270
274
277
277
281
286
289
292
296
298
300
300
302
310
310
311
313
317
320
323
325
327
327
333
338
338
340
340
341
350
350
355
358
360
362
Contents
5.8.1
5.8.2
5.8.3
5.8.4
5.9
5.10
5.10.1
5.10.2
5.10.3
6
6.1
6.2
xi
362
365
369
370
374
378
378
383
386
388
388
394
398
398
399
400
400
401
403
404
409
410
411
414
418
451
' I am at a loss to give a distinct idea of the nature of this liquid, and
cannot do so without many words.'
The Narrative of Arthur Gordon Pym of Nantucket,
The foundations of the physics of liquid crystals were laid in the 1920s but,
surprisingly, interest in these substances died down almost completely
during the next three decades. The situation was summarized by F. C.
Frank in his opening remarks at a Discussion of the Faraday Society in
1958: 'After the Society's successful Discussion on liquid crystals in 1933,
too many people, perhaps, drew the conclusion that the major puzzles were
eliminated, and too few the equally valid conclusion that quantitative
experimental work on liquid crystals offers powerfully direct information
about molecular interactions in condensed phases.' In the last few years
there has been a resurgence of activity in this field, owing partly to the
realization that liquid crystals have important uses in display technology.
An exposition of the physics of liquid crystals involves many disciplines:
continuum mechanics, optics of anisotropic media, statistical physics,
crystallography etc. In covering such a wide field it is difficult to define
what precisely the reader is expected to know already. An attempt is made
to present as far as possible a self-contained treatment of each of these
different aspects of the subject. Naturally, discussion of some topics has
had to be curtailed for reasons of space. For example, we have not dealt
with lyotropic systems, whose complex structures are only just beginning
to be elucidated; or the special applications of magnetic resonance
techniques, as these have been adequately reviewed elsewhere; or the very
recent results of neutron scattering experiments. The primary aim of this
monograph is to provide an insight into the variety of new phenomena
exhibited by these intermediate states of matter.
Xlll
xiv
This book would probably never have been completed without the
unstinting cooperation of my young colleagues in Bangalore. I am
particularly indebted to N. V. Madhusudana, G. S. Ranganath, R.
Shashidhar, U. D. Kini, R. Nityananda and T. G. Ramesh, whose notes
and critical comments were of inestimable value at every stage of the
writing, and to K. A. Suresh and B. R. Ratna for their assistance in
preparing the diagrams and the list of references. Finally, it is a pleasure to
express my thanks to Michael Woolfson for his encouraging interest and
advice.
Raman Research Institute
August 1975
S. Chandrasekhar
Since the publication of the first edition 15 years ago, the subject of liquid
crystals has grown enormously to become a fascinating interdisciplinary
field of study. A variety of new thermotropic phases have been discovered,
including over a dozen different smectic modifications, discotics, biaxial
nematics, etc., which have opened up a veritable treasure-house for the
theoretical condensed matter physicist. On the technological side, the
advances have been no less spectacular: portable computers and hand-held
TV sets using liquid crystal display devices are being sold in large numbers,
and high-definition LCD-TV would seem to be just round the corner. The
aim of the present edition is to bring the coverage up to date. The chapters
dealing with the classical nematic, cholesteric and smectic types of liquid
crystals have been revised substantially and a new chapter has been
included on discotics. However, mainly for reasons of space, special topics
like the applications of magnetic resonance techniques, non-linear optical
properties, etc., have not been discussed here as these have been
comprehensively reviewed elsewhere.
It is my privilege to express my thanks to my young colleagues who kept
me alive to the subject: in particular to V. N. Raja, S. Krishna Prasad,
D. S. Shankar Rao, S. M. Khened and Geetha Nair for their invaluable
help throughout the preparation of this book, and to Sriram Ramaswamy
and U. D. Kini for their advice on certain theoretical points. I am indebted
to the Council of Scientific and Industrial Research, New Delhi, for a
Bhatnagar Fellowship which made it easier for me to undertake this task.
Centre for Liquid Crystal Research
March 1992
S. Chandrasekhar
xv
1
Introduction
1. Introduction
Crystal
Nematic
(a)
(b)
Fig. 1.1.2. X-ray diffraction patterns from an 'ordinary' nematic liquid crystal:
4-n-propyloxy-benzylidene-4/-n-propylaniline: (a) an unaligned (though not quite
randomly oriented) sample; (b) a well aligned sample. (De Vries.(6))
A biaxial modification of the nematic has been discovered recently (see
6.6).
The cholesteric mesophase is also a nematic type of liquid crystal except
1. Introduction
(a)
(b)
Fig. 1.1.3. X-ray diffraction patterns from a 'cybotactic' nematic: bis-(4'-noctyloxybenzal)-2-chloro-l,4-phenylenediamine: (a) an unaligned (though not
quite randomly oriented) sample; (b) a well aligned sample. (De Vries.(6))
that it is composed of optically active molecules. As a consequence the
structure acquires a spontaneous twist about an axis normal to the
preferred molecular directions (fig. 1.1.4). The twist may be right-handed
or left-handed depending on the molecular conformation. Optically
inactive molecules or racemic mixtures result in a helix of infinite pitch
which corresponds to the true nematic. The energy of twist forms only a
minute part (~10~ 5 ) of the total energy associated with the parallel
alignment of the molecules,(8) so much so that when a small quantity of a
Fig. 1.1.4. The cholesteric liquid crystal: schematic representation of the helical
structure.
cholesteric substance,(4) or even a non-mesomorphic optically active
substance,(9) is added to a nematic the mixture adopts a helical configuration. The spiral arrangement of the molecules in the cholesteric is
responsible for its unique optical properties, viz., selective reflexion of
circularly polarized light and a rotatory power about a thousand times
greater than that of an ordinary optically active substance. Cholesterics of
low pitch, less than about 5000 A, exhibit what are known as blue phases.
1. Introduction
(a)
Illllllllllllllllll
Illlllllllllllllll
Illllllllllllllllll
Fig. 1.1.6. (a) Threads in a nematic liquid crystal. Crossed polarizers. Film
thickness ~ 100 jum. (b) Schlieren texture in a nematic film of thickness ~ 10 jum.
Crossed polarizers. (Sackmann and Demus. (14))
1. Introduction
Fig. 1.1.7. Focal conic textures in smectic A. (a) The polygonal texture. Crossed
polarizers. (Friedel.(4)) (b) Simple fan-shaped texture. Crossed polarizers. (Sackmann and Demus.(14))
systems (see fig. 1.2.2), but a number of variants of this structure have been
identified: hexagonal, rectangular, tilted, etc. The nematic phase has an
orientationally ordered arrangement of the discs without any long-range
translational order (fig. 1.1.8(6)). Unlike the classical nematic of rod-like
molecules, this phase is optically negative. A cholesteric (or twisted
nematic) phase has also been identified. A smectic-like lamellar phase has
been reported(22) but the disposition of the molecules in the layers has not
yet been resolved.
10
1. Introduction
(a)
Fig. 1.1.8. The structure of (a) the columnar phase and (b) the nematic phase of
disc-like molecules.
(b)
Fig. 1.1.9. Polymers that exhibit liquid crystalline phases. The basic monomer units
are low molar mass rod-like or disc-like mesogens which may form part of the main
chain (a) or attached as side groups (b).
1.1.3 Polymer liquid crystals
The structures of polymers that form liquid crystals are illustrated
schematically in fig. 1.1.9. The basic monomer units are low molar mass
mesogens, rod-like or disc-like, which are attached to the polymer
backbone in the main chain itself (fig. 1.1.9 (a)), or as side groups (fig.
1.1.9(6)). The nature of the mesophase depends rather sensitively on the
11
(a)
to
Fig. 1.1.10. Mesophases of discotic polymers: (a) the hexagonal columnar phase.
The diagram illustrates intercolumnar binding as well as intracolumnar backfolding of the main chain (Herrmann-Schonherr, Wendorff and Ringsdorf (24) );
(b) the' sanidic' nematic phase composed of boards stacked parallel to one another
(Herrmann-Schonherr et al.{25)); (c) the columnar nematic phase (Ringsdorf
et al (26) ).
backbone, the mesogenic unit and the spacers. With rod-shaped repeating
units, mesophases similar to the nematic, cholesteric and smectic types of
rod-like molecules are observed(23) whereas with disc-shaped repeating
units, some new kinds of mesophase structures have been found. For
12
1. Introduction
13
14
1. Introduction
15
CH.O
(1.3.1)
OCH,
...
118.2 C
solid -*
29.57 kJ
.
135.3 C
.
nematic -*
isotropic
0.57 kJ
4'-n-Pentyl-4-cyanobiphenyl'I (39)
(1.3.2)
111
solid
22.5 C
17.2 kJ
nematic -*
401 C
35 r
/ /
isotropic
445 C
nematic -*
isotropic
7?-Quinquephenylv
solid -*
o.8 kJ
(1.3.3)
4'-n-Octyl-4-cyanobiphenyl (39)
C8H17-
solid -*
24 C
25.3 kJ
(1.3.4)
t\
34 C
+.
A
smectic A -*
0.13 kJ
.
42.6 C
nematic -*
0.97 kJ
isotropic
16
1. Introduction
Cholesteryl nonanoate (monotropic smectic A) (41>42)
O
CH 3 (CH 2 ) 7
78.6 C
91 C
solid -
cholesteric -*
0.017 kJ
91.3 C
blue phase I -*
0.017 kJ
blue phase II
75.5 C 0.25 kJ
91.45 C 0.53 kJ
smectic A
isotropic
(1.3.5)
C5HnO-
36 C
solid -*
15.2 kJ
-CH =
-C6H13
38 4 C
42.4 C
smectic G -*:
smectic F -*
0.39 kJ
0.15 kJ
smectic B
50 C 25 kJ
isotropic ^ 7 2 - 8 c
0.66 kJ
nematic
0.42 kJ
smectic A ~m smectic C
very
small
(1.3.6)
2
Statistical theories of nematic order
X-ray analyses of the crystal structures of typical nematogenic compounds(13) have established that the long narrow molecules are more or
less parallel and interleave one another to form what was described by
Bernal and Crowfoot(1) as an 'imbricated' arrangement (fig. 2.1.1). The
transformation from the solid to the nematic phase is characterized by the
breakdown of the positional order of the molecules but not of the
orientational order. The mesophase is fluid and at the same time
anisotropic because of the facility with which the molecules slide over one
another while still preserving their parallelism. The degree of orientational
order in the liquid crystal drops gradually on heating, whereas certain
other thermodynamic properties, such as the specific heat, thermal
expansion and isothermal compressibility, increase rapidly as the temperature approaches the nematic-isotropic point T^r At Tm a 'weak' first
order (discontinuous) phase transition takes place, accompanied by the
complete breakdown of the long-range orientational order. The changes of
entropy and volume attending this transition are typically only a few per
cent of the corresponding values for the solid-nematic transition.
In this section, we shall consider an elementary model(4"6) which
illustrates the principal mechanisms responsible for this unusual type of
melting phenomenon.
2.1.1 Order-disorder in positions and orientations
The simplest treatment of the melting of inert gas crystals is that due to
Lennard-Jones and Devonshire(7) (LJD) who regarded the mechanism of
fusion as a positional order-disorder phenomenon. They postulated that
17
18
19
Ne
3.26
Kr
3.36
A
3.35
Xe
3.40
Crystals in which the rotational transition precedes the melting transition
O2
2.0
PH 3
1.92
SiH4
1.8
CD 4
2.42
CF 4
2.0
CH 4
2.47
Crystals in which the two transitions coalesce
CO 2
9.25
SiCl4
9.06
N2O
8.58
HCN
7.73
C2H4
7.70
CS 2
6.51
SO2
8.95
(CN) 2
7.90
2?-sites, Lennard-Jones and Devonshire showed that the melting parameters predicted by the theory agree well with the data for a number of
spherical or nearly spherical molecules.
However, the theory fails for anisotropic molecules where the effects of
orientational disorder become important. The thermodynamic data on
melting suggest that there are two classes of molecular crystals: those
which undergo phase transitions associated with rotational motions at
temperatures below the melting point and those in which the rotational
and melting transitions coalesce. The former have entropies of fusion
lower than the inert gas crystals, while the latter have much higher
entropies of fusion (table 2.1.1).
Pople and Karasz ( 4 ) proposed a simple extension of the L J D model
which at once provides an interpretation of these variations in the melting
properties. They assumed that the molecule can take u p one of two
orientations on any site, so that it now has four possibilities, Ax, A2, B1 and
B2. The state of perfect order (or the solid at zero temperature) may then
be regarded as one in which all molecules occupy sites and orientations of
the same type, say A19 and the state of complete disorder (or the liquid
phase) as one in which all four configurations are equally populated.
Clearly, there can also be states with positional order and no orientational
order and vice versa.
When a molecule is turned into an unfavourable orientation, local
strains are set up and consequently there is an increased tendency for the
molecules in the vicinity to move to interstitial sites. T o allow for this
20
z=f n,
21
where
r(q,s) = \
la
1UJV(1 - ? ) ( 1 +^)]![17V(1 Using Stirling's approximation
n+s
2 )
W(\-q
KT\
4
B
J
kBr{
K1AA)
-(H-FWr)*
<2U)
22
NkBT
vx
kBT
23
-0.2
-0.4
15
zW/kBT
Fig. 2.1.2. (a) Variation of the equilibrium value of the positional order parameter
q and the orientational order parameter s with zW/kB rfor v = 0.3. (b) Theoretical
isotherm for v = 0.3 showing rotational transition in the solid state followed by the
melting transition. (The isotherm is drawn for the solid rotational transition
temperature kB T/e = 0.593.)
L (b)
Fig. 2.1.3. (a) Variation of the equilibrium values of q and s with zW/kB T for
v = 0.8. (b) Theoretical isotherm for v = 0.8 showing a single transition in which
both positional and orientational orders collapse simultaneously (k B T/e = 0.625).
with orientational order but no translational order. (Over a small range of
v, 0.975 < v < 1.047, the theory predicts a second order (continuous)
nematic-isotropic transition, whereas experimentally it is found that this
transition is always discontinuous, though weakly so. This limitation of the
model probably arises from the restriction that the molecule can take up
only two orientations. An extension of the treatment to more than two
discrete orientations has been given by Amzel and Becka (11) for low v.)
24
1.0
q, s
Fig. 2.1.4. (a) Variation of the equilibrium values of q and s with zW/kB T for
v = 1.3. (b) Theoretical isotherm for v = 1.3 showing solid-nematic and nematicisotropic transitions. (The isotherm is drawn for the solid-nematic transition
temperature kB T/e = 0.678.) (After reference 5.)
0.6
0.4
0.2
0.4
0.8
1.2
25
a-1
3 --
\S-N
2 -
S-I
'
1 -
s-s
0 -
0.2
N-I/
i
0.4
0.6
0.8
1.0
1.2
ZT);
\
+ 2 T ( \ ]
)/2T{ aril*
26
Table 2.1.2. Latent heat and volume change data for some nematogenic
compounds
Solid-nematic
transition
Latent
heat
(kJ mole-1)
Compound
4-(4-n-Propylmercapto
benzalamino)azobenzene(12)
/?-Azoxyanisole
/7-Azoxyphenetole
4- Methoxy benzy lidene4/-butylaniline
31.30
29.57(14)
26.87 (14)
18.06 (17)
Nematic-isotropic
transition
Volume
change
AV/V(%)
5.07
11.03 (13)
84d6)
Latent
heat
(kJ mole-1)
Volume
change
AV/V(%)
0.368
0.574(14)
1.366(14)
0.09
0.36(15)
0.60(16)
0.582(17)
0.40(18)
S-N
0.2
0.1
s-s/
n
1
N-I /
^
1
0.4
0.8
1.2
Fig. 2.1.7. Relative change of volume on transition versus v (see legend offig.2.1.5).
27
(b)
0.2
()
Solid /Nematic/Isotropic
pVo
/ iNemat
0.2 -
0.1
"T^~
Solid /
/ Isotropic
/ /
0.1 -
/
/
0.67
0.68
0.69
kBT/e
0.70
0.6
/
i
0.7
kBT/e
0.8
28
500
400
Solid
A
Nematic
/
0/
Isotropic
300
200
100
390
400
410
T(K)
430
Fig. 2.1.10. Experimental phase diagram for PAA. (After reference 22.)
the slope AT/dp for the latter being greater in accordance with the
experimental data (fig. 2.1.10). Of course, such a behaviour is also to be
expected from the Clausius-Clapeyron equation.
Fig. 2.1.9(6) represents a more interesting case. Here v is just below the
critical value for the nematic phase to occur at zero pressure. As the
pressure is raised, there is initially a single transition, viz, the solid-liquid
melting transition, but at higher pressures branching takes place and
there are two transitions. The branching point represents the solidnematic-isotropic triple point.
Experiments were done on the first two members of the p-nalkoxybenzoic acid series to verify this prediction. (22) These two compounds, methoxy- and ethoxybenzoic acids, do not form liquid crystals
at atmospheric pressure, whereas propoxybenzoic acid and the higher
homologues show at least one liquid crystalline phase. As the pressure is
raised, both compounds exhibit mesophases, initially a nematic phase and
200 -
Solid
1/
29
/isotropic
7k
100 -
Triple point tS
^
7
/
/
Triple point
i
205
215
225
Temperature ( C)
then, at higher pressures, a smectic phase as well (fig. 2.1.11). (The smectic
phase is not included in the framework of the theory in its present form. It
would require the introduction of another order parameter, viz, a lamellar
order parameter (see chapter 5).) The transitions were detected by
differential thermal analysis and the mesophases identified by microscopic
observations using a high pressure optical cell (fig. 2.1.12). These studies
also established for the first time the existence of the solid-nematicisotropic and the solid-smectic-nematic triple points in single component systems.
2.2 Phase transition in a fluid of hard rods
30
31
At sufficiently low densities the rods can assume all possible orientations
and the fluid will be isotropic. As the density increases, it becomes
increasingly difficult for the rods to point in random directions and
intuitively one may expect the fluid to undergo a transition to a more
ordered anisotropic phase having uniaxial symmetry. That this does indeed
happen was first proved by Onsager.(23) Other hard rod theories have since
been proposed, which we shall refer to briefly at the end of this section, but
Onsager's approach, based on an exact density expansion for the free
energy, is still probably the most satisfying. However, the mathematical
analysis involved in the theory is rather complex - even in the lowest order
it leads to a non-linear integral equation. Truncation of the series after the
linear term, as was done by Onsager, would appear to be satisfactory only
for very long rods and not for shorter ones, but attempts to evaluate the
coefficients in the expansion beyond that of the linear term have so far been
unsuccessful because of computational difficulties.
Zwanzig(24) proposed a simplified version of Onsager's theory which
enables the virial coefficients to be evaluated to a much higher order. The
simplification consists in the following assumptions: (a) the rods can take
up only discrete orientations along three mutually perpendicular axes,
whereas in the original theory they can point in any direction, and (b) the
length to breadth ratio of the rods is very large (tending to infinity).
Though these assumptions may be somewhat too drastic as far as real
liquid crystal systems are concerned(25) the model does serve to illustrate
the essential principles of Onsager's method and may prove to be useful as
a starting point for possible extensions and applications of the general
theory.
In Zwanzig's model the rods are rectangular parallelopipeds of length /
and square cross-sectional area d2. Assuming that the potential energy of
interaction of TV such hard rods is a sum of pair potentials, we may write
UN = \tuij9
(2.2.1)
where Utj is oo if the molecules / and j intersect and zero if they do not. The
intersection may occur either in a parallel or perpendicular configuration
as shown in fig. 2.2.1. Let the position vector of they'th rod be represented
by Rj and its orientation by up and let the set of all positions be abbreviated
by R and the set of all orientations by u. The configurational integral for
the system consisting of TV rods in the volume V can then be expressed as
32
where TV! allows for the indistinguishability of the particles and 3^ for the
total number of orientational states. The potential energy of the system
depends on the sets R and u. Since the configurational space available for
each molecule is V,
I dR exp ( - UN/kB T) = VN exp [ - <pN(u)/kB T],
where the new function <pN depends only on the orientations. Thus
tT}.
Now for any given configuration of the system, there will be N(\) molecules
pointing in the direction u(l), N(2) in the direction u(2) and N(3) in the
direction u(3). Therefore q>N becomes a function of the occupation numbers
in the three allowed directions, i.e.,
<pN =
<pN[N(l),N(2),N(3)].
The indistinguishability of the molecules requires that for any given set of
V*
J^_
N\lN
{?
N )=a N )=o
N\exp(-<pN/kBT)
N(\)\N{2)\N{?)\
'
) Nil) M3)l
we have
Q=
t[N(l),N(2),N(3)].
(2.2.2)
N-1]nt=l-]np-]n3-txttlnxa-(<pN/NkBT),
(2.2.3)
33
(2.2.4)
where the virial coefficient B[N(l), N(2), N(3)] is defined as
34
= 0,
a =1,2,3.
' 0) =
The first volume integral is evaluated directly so that
35
ni+d)/2
J-(l+d)/2
r(i+d)/2 ra
J-(l+d)/2
d3R2 = -
J -d
~-4l2d(d/l).
Thus, the virial coefficient for parallel configuration is smaller than that for
orthogonal configuration by an order d/l. Higher order virial coefficients
exhibit the same behaviour. To simplify the calculation Zwanzig neglected
all terms of order d/l, i.e., he confined himself to the limit /-> oo, d-^0,
l2d = constant, thereby reducing considerably the number of coefficients to be considered. It is readily verified that only coefficients of the
type B(m,n,0), B(0,m,n) and B(m,0,n) survive, and that by symmetry
B(m, 0, n) = B(n, 0,ra)= B(0,ra,n) = B(0, n, ra).
Let us now choose the z axis (/ = 3) as a preferred direction. By
symmetry the numbers of molecules in the x and y directions are equal, so
that the mole fractions along the three directions can be expressed in terms
of a single order parameter x:
v v
x2 = x,
x3 = 1 2x.
_ sr ^r
D/,^
NkBT
(2.2.6)
The mole fraction x is then determined by
(AT1 In 0 = 0,
ax
i.e.,
2 In
x \
\-2x)
d /
dx\
cpN
NkBT)'
36
37
3.0
2.0
0.2
0.4
0.6
0.8
Volume
= (31nr/ain/>).
(2.2.8)
38
(2.3.1)
where 6 is the angle which the long molecular axis makes with n, and the
angular brackets denote a statistical average. For perfectly parallel
alignment s = \9 while for random orientations s = 0. In the nematic
39
where fj = \{n1 + 2n2) and n the number of molecules per unit volume.
Similarly
Therefore
n2).
(2.3.4)
r/1 and n2 can be obtained from the susceptibilities of the solid crystal if the
structure is known. Thus a measurement of xz>Xz o r t n e anisotropy xz~Xx
in the nematic phase gives the absolute value of s. In a similar manner,
s can be related to the optical anisotropy (birefringence (51) and linear
dichroism(50)) though, as is well known, there is a difficulty in this case as
there is no rigorous way of correcting for the effects of the polarization field
in the medium. Approximate methods have been proposed (50>52) which
appear to yield satisfactory results/ 53 ' 54)
Another important tool for investigating molecular order is nuclear
magnetic resonance (NMR) spectroscopy. (55) The principle of this method
is that if there is a pair of protons in a molecule, each spin is coupled to the
external magnetic field H as well as to the dipole field created by its
neighbour. The latter gives rise to a small perturbation of the eigenstates of
the Zeeman Hamiltonian so that the signal appears as a doublet. The
doublet separation is easily seen to be proportional to (3cos 2 # l)/r 3 .
where 9 is the angle which the interproton vector rtj makes with H. In the
isotropic phase xtj can assume all possible orientations with respect to H
and <3cos 2 # 1> vanishes, but in the nematic phase there is a splitting
which is a measure of the order parameter s. The analysis is, of course, not
so straightforward for complex molecules as it requires precise knowledge
of the interproton distances and angles. The quadrupole splitting in the
magnetic resonance spectrum of 14N(56) or of deuterium(57) has also been
employed for studying molecular order. Other methods involve the use of
40
0.6
a 0.5
O
6
0.4
0.3
-40
-30
-20
-10
r-rNI(Q
Fig. 2.3.1. The temperature variation of the long-range orientational order
parameter s in PAA. Open circles from NMR measurements (McColl and Shih(66)),
filled circles from diamagnetic anisotropy (Gasparoux, Regaya and Prost(67));
triangles from refractive indices.(65)
NMR (58) or electron spin resonance spectroscopy (59) of geometrically
anisotropic molecules aligned in nematic solvents. We shall not discuss any
of these techniques here as a number of excellent reviews on the subject are
available. (605) In fig. 2.3.1 we present experimental values of the order
parameter of PAA determined by three different methods.
NMR and dielectric measurements (see 2.3.5) indicate that there is a
fair degree of rotational freedom of the molecules about their long axes.
Nevertheless the assumption that the molecules are cylindrically symmetric
is not valid in general and can sometimes lead to errors in the determination
of s.{6S) For a molecule of arbitrary shape, (2.3.1) may be written in the
generalized form(69)
where a,/? = x9y9z refer to the space-fixed axes, ij = x\y\z' refer to the
principal axes of the molecule, /a, jp denote the projection of the unit vectors
i, j along a,/?, and 3a/3, S(j are Kronecker deltas. $"/is symmetric in /, j and in
a,/?. It is also a traceless tensor with respect to either pair, i.e.,
where repeated tensor indices imply the usual summation convention. The
generalization of (2.3.2) and (2.3.3) is
41
0
-(Sll
+ s2i.
^oo = i
r>2
Y ( 1 ) jr < 2 )
(2.3.6)
where x$,x$
as
f
_J
4? = f^2)*42)42Vv2)dr(2), etc.,
k J
42
y/^\ y/[2) and EM,EV are respectively the eigenfunctions and eigenvalues of
the unperturbed molecules corresponding to states ju and v, EMV = EM + Ev9
^ 1) ,xj 1) ^ 1) zj 1) are the charges and their coordinates in the first molecule
measured in the space-fixed coordinate system with the origin at the centre
of mass of molecule 1, R = (X, Y,Z) is the position of the centre of mass
of molecule 2, e(*\ x^y^z are the charges and their coordinates in
molecule 2 measured in the coordinate system obtained by translating the
one defined above by XYZ. The second summation in (2.3.6) is over terms
with x9 y and z changed cyclically.
We assume that the rotational motions of the molecules are independent
of their translational motions. Let ^rja)C(1) and f<2y2)f<2) be the two
molecular coordinate systems with their origins at the centres of mass,
coinciding with the long axis of each molecule. We define the Eulerian
angles as follows: 6 = the angle between z and , 0 = angle between and
the normal to the z plane and 0 ' = angle between x and the normal to the
z plane. The components of the transition moments are
(237)
where M is a function of the intermolecular vector Rik only and
is a measure of the anisotropy of the 0-// transition. Maier and Saupe
expressed (2.3.7) in the simple form
A
/3cos2^-l\
[ j
43
I K<exp(-w</fcBr)d(cos0<)
= -\NkBTBs\
- ^
Jo
(2.3.9)
exp(-ui/kBT)d(cos0J
(2.3.10)
= -NkBkBs(2s+I)-In
rexp(ffecos20,)d(cos0jj. (2.3.11)
F=
U-TS
(2.3.12)
IT)
\
or
U l )
-o/V,T
3<cos^)
OS
l=0.
(2.3.13)
44
T>TNJ
0.05 -
yL.
0.0
-0.05
-0.1
i
0.2
0.4
s
0.6
0.8
Fig. 2.3.2. Variation of the free energy with the order parameter calculated from the
Maier-Saupe theory for different values of A/kB TV2. The minima in the curves
occur at values of 5 which fulfil the consistency relation (2.3.13).(71)
The plot of F versus s at different temperatures evaluated from (2.3.12)
is shown in fig. 2.3.2. The minimum of the free energy occurs at that value
of s which satisfies the consistency relation (2.3.13). When T < Tm, there is
only one minimum which corresponds to the stable ordered phase. When
Tis slightly greater than 7 ^ there are two minima, but the one at s = 0, i.e.,
the isotropic phase, represents the absolute minimum. At T = Tm, there
are again two minima, one at s = 0 and the other at s = sc, but the two
states now have equal free energies; at this temperature therefore a
discontinuous transition takes place with no change of volume but an
abrupt change in the order parameter. Putting F = 0 at the transition yields
Vl = 4.541,
(2.3.14)
0.4292,
(2.3.15)
45
(b)
0-7
2 0.6
"2
O
0.5
0.4
-40
-30
-10
-20
T-Tm C
Fig. 2.3.3. Order parameter s versus temperature predicted by the Maier-Saupe
theory, (a) ut oc V~2; (b) ut oc V\
2 In
Bs*X
46
V (cm3 mole- 1 )
224
6.05
6.00
222
220
218
5.38
Fig. 2.3.4. Lines of constant order parameter s versus In Kand In T for PAA. The
slope (aIn T/ein V) = -4.0. (After McColl and Shih.(66))
different from that assumed in (2.3.8). In general, if ut ccVm, where m is a
number, then (2.3.14) gets modified to
TmV = 4.541
(2.3.17)
but sc still has the same value. The rate of variation of s with temperature
does not depend too critically on the exponent m, as the isobaric change of
volume with temperature over the entire nematic range is usually only of
the order of 1-2 per cent (fig. 2.3.3). However, its precise value becomes
rather important in accounting for the effect of pressure on the order
parameter/ 66 ' 76>77) Pressure studies (6676) have established the following
results for PAA:
(i) (8In r / d l n V)s = -4 (fig. 2.3.4).
= -y as defined in (2.2.8).
(2.3.18)
(ii) The thermal range of the nematic phase at constant volume is about
2.5 times that at constant pressure.
(iii) The order parameter sc at the transition point is very nearly
independent of the pressure.
Now, it is clear from (2.3.12) and (2.3.13) that B should have a constant
value if s is to be invariant. If B is to be kept constant while T and V are
varied, as is implied in (2.3.18), TVm should be constant, or
(d\nT/d\nV)s = -
(2.3.19)
Empirically, therefore, m = 4 for PAA. It turns out that with this value of
rn, result (ii) follows at once from the theory; on the other hand with m = 2,
47
the thermal range at constant volume is only 0.74 times the experimental
value.(78) The pressure invariance of sc is, of course, a direct consequence of
(2.3.14) or (2.3.17). Thus, the experimental data can be fitted satisfactorily
with m = 4. However, it is doubtful if this has any theoretical significance,
in view of Cotter's result,(70) referred to earlier, that m should be 1
irrespective of the nature of the intermolecular pair potential.
A drawback of the theory becomes apparent when we calculate the
latent heat of transition from the nematic to the isotropic phase. The heat
of transition is easily shown to be given by(48>72)
H = Txl[(a/P)AV-S(Ve,
Tm)],
(2.3.20)
(2.3.21)
where y/ is the Eulerean angle made by with the line of intersection of the
n plane and the xy plane (i.e., the plane normal to the director). A finite
value of D indicates that there is a difference in the tendency of the two
transverse molecular axes to project on the z axis. Theories have been
developed(79) by incorporating an additional term proportional to D in the
Maier-Saupe potential function. It emerges that molecular biaxiality
48
decreases the value of s at the transition, and leads to better agreement with
experiment. Further, since different properties have, in general, different
biaxial anisotropies, the order parameters determined by different techniques (optical anisotropy, diamagnetic anisotropy, etc.) need not
necessarily agree with one another if the biaxial order parameter (2.3.21) is
included in the calculations.
Until the 1970s only s (P2(cos9)} (where P2 is the second order
Legendre polynomial) was accessible experimentally, but a technique is
now available for measuring both </>2) and <i^> (where P4 is the fourth
order Legendre polynomial). It involves polarized Raman scattering
measurements on aligned samples. Jen et al.m) who developed this
technique, made use of the C = N stretch vibration of a cyano-compound
dissolved in the nematic phase as a probe for the Raman measurements. A
number of pure nematogenic cyano-compounds have since been investigated by this method/81'82) It turns out that in every case studied so far,
<P4> is negative for at least part of the nematic range. While the observed
<P2> can be fitted quantitatively by including higher order terms in the
Maier-Saupe potential, the negative <P4> cannot be accounted for
theoretically. Using the experimental value of <P2> and <P4> one can
calculate an orientational distribution function for the molecule :(80)
f(cos6) =
I (even)
j <i>(cos0)>i>cos0
(2.3.22)
truncated after the third term 1 = 4. The results show that the molecule has
a strong tendency to be tipped away from the nematic axis - very much
more than predicted by the Maier-Saupe theory (fig. 2.3.5). In the three
pure cyanobiphenyl compounds studied/81'82) it is found that the greater
the chain length, the greater is the magnitude of the negative <,P4> (fig.
2.3.6). This may possibly provide a clue to the origin of the observed
angular distribution function. However, further studies need to be carried
out before drawing any definite conclusions.
The odd-even effect
From chemical studies it has long been known that the end-chains play a
significant part in the stability of the mesophase. The transition temperature r m (83) and a number of other properties (e.g., the order parameter,
the excess specific heat, the transition entropy, etc.) show a pronounced
alternation as the homologous series is ascended, i.e., as the number of
carbon atoms in the end-chain is increased. This is often referred to as the
49
T-T=2C
3.0
2.4
0.6
-0.3
1.0
0.6
0.8
0.4
0.2
,OQ 6
0.0
cos 6
50
-o.i
-0.2
Fig. 2.3.6. Orientational order parameter <P 4(cos#)> in three nematic liquid
crystals from Raman measurements: (a) 4 /-n-pentyl-4-cyanobiophenyl (5CB); (81)
(b) 4/-n-heptyl-4-cyanobiphenyl (7CB);(82) (c) 4/-n-octyloxy-4-cyanobiphenyl
(80CB).(82) (After reference 82.)
Fig. 2.3.7. Structure of 4,4/-di-n-alkoxyazoxybenzenes. The addition of an evennumbered carbon atom in the preferred trans conformation is along the major
molecular axis. This is not the case for an odd-numbered carbon atom.
51
52
260 r
220
200
160
120
12
Fig. 2.3.8. The variation of the nematic-isotropic transition 7 ^ with the number of
methylene groups in the flexible spacer for a,o>bis(4,4'-cyanobiphenyloxy)alkanes,
the molecular structure of which is shown at the top of the diagram. The dashed
lines represent the theoretically calculated values. (After Luckhurst.(75))
larger contribution for the measuring field parallel to the long molecular
axis) and (ii) the dipole orientation effect. The sign of the latter contribution
is positive if the net permanent dipole moment of the molecule makes only
a small angle with its long axis and is negative if the angle is large. In the
last case the sign of the net anisotropy depends on the relative magnitudes
of the two contributions. Thus different nematic materials can exhibit
widely different dielectric properties (fig. 2.3.9).
In contrast to diamagnetism where the magnetic interactions between
molecules can be neglected, the polarization field in the medium becomes
important when discussing dielectric anisotropy (see, e.g., Bottcher(92)).
Maier and Meier(89) took this into account by applying the Onsager
theory.(93) The effective induced dipole moments per molecule along and
53
5.9 ,- (a)
tn
5.8
I 5.7
s
5.6
140
130
120
7TQ
30
20
10
1
100
120
140
7TQ
Fig. 2.3.9. Principal dielectric constants of (a) PAA (after Maier and Meier(90)) and
(b) n(4/-ethoxybenzylidene) 4-aminobenzonitrile (after Schadt (91)). s1 and e2 refer to
the values along and perpendicular to the optic axis of the nematic medium and is
is the isotropic value.
perpendicular to the unique axis of the nematic liquid crystal are then given
by expressions essentially similar to (2.3.2) and (2.3.3) but with appropriate
correction factors for the polarization field:
(2.3.23)
(2.3.24)
54
constant, F = 1/(1 af) the reaction field factor, a the mean polarizability,
aa the polarizability anisotropy (the direction of maximum polarizability is
assumed to be the long molecular axis),/= 4nNp(2e 2)/3M(2s+ 1), p the
density, M the molecular weight, s the order parameter defined by (2.3.1),
and E the applied field. Strictly speaking F and h have to be corrected for
the anisotropy of the dielectric constant but we shall ignore these
corrections.
To calculate the effective permanent dipole moment, we choose XYZ as
the space-fixed coordinate system, Z being parallel to the unique axis of the
medium, and rj( as the molecule-fixed coordinate system, the axis
coinciding with the long axis of the molecule. Let v be the Eulerian angle
between the axis and the line of intersection of the XYand rj planes, and
V the angle between this line and X. Suppose that the permanent dipole
moment ju is inclined at an angle ft with respect to the long molecular axis.
In dielectric measurements, low fields are usually employed so that the
potential energy of the dipoles due to the external field is small. We can
therefore write the effective dipole moments along the field directions as
,T)]juzy/(0)sin0d0dvdv/
o Jo Jo
[l+(M,hE1/kBT)]yf(ff)sin9d0dvdv'
hE2/kB T)] jux y/{0) sin 0 d6 dv dv'
> (2.3.26)
Jo Jo
Jo Jo
Jo
Jo
[l+(MxhE2/kBT)](S)sm8dvdv'
where
fiz = ///[cos P cos 6 + sinfi sin v sin #],
jux = Fju[cos P sin V sin 6 + sin /?(cos v cos V sin v sin V cos 9)]
and y/(0) d6 is the probability of a molecule having an orientation between 6 and 6 + d6. If y/(0) is given by the Maier-Saupe distribution,
exp( ut/kB T) where ut is defined by (2.3.8), the integrals reduce to
^
ft = ^T"^[l+i(l-3cos a i8)5]AF 2 2 .
(2.3.27)
(2.3.28)
55
Since
^f[\-(I-3
cos2 p)s
(2.3.29)
and similarly
a = e1-e2 = ^^hFW~^-f(l
-3cos 2 )L.
(2.3.31)
Clearly if /? is small, the two terms in the square brackets of (2.3.31) add to
give rise to a strong positive ea, whereas if /? is sufficiently large a may be
negative. In PAA, for example, ju and /? are estimated to be 2.2 debyes and
62.5 respectively from Kerr constant measurements in dilute solutions
and dielectric measurements in the isotropic phase. Substituting for the
other known parameters for this compound and using s from theory, it
turns out that ea should be weakly negative. The absolute value of ea as well
as its temperature variation are in fair agreement with the experimental
data.(90)
The dielectric constants are, of course, frequency dependent.(94) The
dipole orientation part of the polarization parallel to the preferred
direction (1-direction) may be expected to be characterized by a relatively
long relaxation time. This arises because of the strong hindering of the
rotation of the longitudinal component of the dipole moment about a
transverse axis. On the other hand the orientational polarization along the
2-direction will have a much faster relaxation, comparable to the Debye
relaxation in normal liquids, as this involves rotation about the long axis
of the molecule. If there are additional dipoles in parts of the molecule,
with their own internal degrees of rotation, the corresponding relaxation
times will again be similar to that in a liquid. The expected form of the
dispersion curves for a compound like /7-azoxyanisole is illustrated in fig.
2.3.10. The trends in the curves have been confirmed experimentally.(95>96)
Meier and Saupe(97) have discussed the mechanism of dipole orientation in
PAA and have shown that the relaxation time for polarization parallel to
56
e 3
10 12
Fig. 2.3.10. Expected form of the dispersion of the principal dielectric constants of
4,4/-di-n-alkoxyazoxybenzenes. The suffix 0 refers to the static values and the suffix
oo to the optical values. ex shows the low frequency relaxation and both x and e2
show the normal Debye high frequency relaxation. (After Maier and Meier.(95))
A \\ \l/7 ///
II.
I
Splay
Twist
Bend
Fig. 2.3.11. The three principal types of deformation in a nematic liquid crystal.
57
the 1-direction should increase from its value in the isotropic phase by a
'retardation factor' which may amount to several orders of magnitude
depending on the strength of the nematic potential. However, the
Maier-Saupe potential yields a retardation factor smaller than the
experimental value. This is not surprising since short-range order may be
expected to play a dominant part in the relaxation process and the mean
field theory neglects this completely. No theory of the relaxation
mechanism has yet been proposed taking into account near-neighbour
correlations.
2.3.5 Relationship between elasticity and orientational order
As remarked in chapter 1, a uniformly oriented film of nematic liquid
crystal may be prepared by prior treatment of the surfaces with which it is
in contact. If the preferred orientation imposed by the surfaces is perturbed,
let us say by a magnetic field, a curvature strain will be introduced in the
medium. The theory of such a deformation will be discussed at length in
3.2; for the present it will suffice to state some of the important results.
The free energy per unit volume of the deformed medium relative to the
state of uniform orientation is
*Kik
^
juv* 00
1
J
Y2
o f j f c i \ ay (fc)
l
D2
}XOtiXOv
+ 3^
58
The internal energy per mole due to orientational order in the mean field
approximation is then
U = \NAV~2s\
where V is the molar volume. Now let us suppose that the director at the
site of the fcth molecule is turned through an angle ak with respect to that
at /. If we define a new system of axes X Y'Z' at the site of the /:th molecule,
Z ' making an angle (xk with respect to the Z axis,
*oT = ^OM
v'(fc) _
v(k)
C 0 S
<** ~
^O
Since we are dealing with small deformations, we assume that the director
continues to have cylindrical symmetry about the Z ' axis at the site of the
fcth molecule and that the order parameter s is unchanged in magnitude.
Substituting for x($ etc. in (2.3.33) in terms of x'0{*\ defining the Eulerian
angles between the molecular coordinate system <fcy*><*> and the X'Y'Z'
system and taking appropriate averages, the extra internal energy per mole
of the system in the deformed state turns out to be
1
JLsin,flJ
^ik
r/ z 2
3
L\
\2
z 2 Y21
^ _ ! - 9 ^ 4 (2.3.34)
*^ik
^ik
Now for a pure bend distortion da/dX = da/d Y = 0 and we may set
sin ock ak 7^ifc 3a/6Z. Assuming a spherically symmetric distribution,
Xik/Rik etc. may be taken to be constant (independent of temperature),
and therefore the second sum in (2.3.34) is proportional to F~4/3. The free
energy of deformation per unit volume can then be expressed as
SF
where C is a constant. But by definition
so that
k3S = CV~7/3s2.
(2.3.35)
Similar expressions are obtained for the splay and twist elastic constants.
The temperature dependence of the elastic constants of' simple' nematics
is represented well by (2.3.35).(99'100)
To calculate the absolute values of the elastic constants, one has to
evaluate the second summation in the right-hand side of (2.3.34). Saupe(98)
evaluated the mean value of this sum for an isotropic distribution of
59
da
[ns\2kBT
60
3 1.2
'S
Q
o
A
% io
0.9
0.8
290
300
310
320
T(K)
Fig. 2.3.12. The temperature dependence of the intensity (/) of light scattering from
4-methoxybenzylidene-4/-butylaniline (MBBA) in the nematic phase. The values
(plotted as kB T/I) are normalized to unity at 286 K. The squares, triangles and
circles correspond to three different experimental configurations. For one of the
configurations (squares) the data were not fully corrected for the effect of the
temperature dependence of the refractive indices on the variation of one of
the wave-vector components, which probably explains the slight increase near the
transition temperatures in this case. (After Haller and Litster.(100))
(2.4.1)
The resulting distribution function is similar to that in the MaierSaupe theory, except that the coefficient of the potential has the form
[(v2p/kBT) + A(p)], i.e., a temperature dependent attractive part and an
'athermal' part as given by the scaled particle theory. A similar result can
be obtained using the Andrews model as well.(35) These last two approaches
appear to be promising; for example, calculations show that y 4 for
l/b ~ 2 without violating Cotter's thermodynamic consistency condition
that the mean field potential should be proportional to /?. Further the
transition parameters and the properties of the nematic phase are in
reasonably good agreement with the experimental values for PAA. Gen-
61
100 r-
80
60
40
20
120
TNI 140
160
180
Temperature (C)
Fig. 2.5.1. Magnetic birefringence in the isotropic phase of PAA: the horizontal
dashed line gives the value for nitrobenzene at 22.5 C. (After Zadoc-Kahn.(119))
eralized van der Waals models have also been developed(116 18) which lead
to results that are essentially the same as those predicted by the scaled
particle theory with a superposed attractive potential.
62
+ JCJ 4 + ...,
(2.5.1)
(2.5.4)
For a' weak' first order transition, B is small and 2B2/9aC may be expected
to be a very small quantity.
In principle, a free energy expansion of this type should be valid for
nematic liquid crystals, with s denoting the usual orientational order
parameter defined by (2.3.1). The term of order ss is not precluded by
symmetry, for the states s and s represent two entirely different kinds of
molecular arrangement which are not symmetry related and do not have
equal free energies. In the former case, the molecules are more nearly
parallel to the unique axis, while in the latter they are more nearly
perpendicular to it. However, in the nematic phase s is usually quite large
(greater than about 0.4) so that very many more terms have to be included
in the expansion in order to draw any valid conclusions. Consequently, the
63
(2.5.5)
(2.5.9)
64
I 4
0
44
40
48
52
56
60
Temperature (C)
T)(3cos2p-
\)s,
(2.5.11)
F=a(T-T*)s2-NFh2E2[(xgL-(Fju2/2kBT)(l-3cos2l3)]s.
(2.5.12)
Proceeding as before
s
NFh2E2[ota-(Fju2/2kB
3a(T- T*)
T)(\-3cos2p)]
(2.5.13)
and
21nVa(T-T*)
(2.5.14)
65
20
5 -
15
10
<l
-10
-15 r
405 Ni
415
425
Temperature (K)
435
445
Fig. 2.5.3. The electric birefringence (Tsvetkov and Ryumtsev,(122) open circles) and
the reciprocal of the magnetic birefringence (Zadoc-Kahn,(119) full circles) in the
isotropic phase of PA A versus temperature. The lines represent the theoretical
variations.(129)
positive for the long molecules under consideration, but the sign of the
dipole contribution depends on the angle /?. If /? is small, AE will be
strongly positive, whereas if /3 is sufficiently large AnK may be negative.
Moreover, the second term in the square brackets of (2.5.14) is proportional to T~x, so that, in principle, there can occur a reversal of sign of
AE with temperature. Using the values of a and T* derived from the
magnetic birefringence of PAA and substituting for //,/?, etc. (see 2.3.4) it
is found(129) that there is in fact a reversal of sign of AE, though it occurs
at I NI + 9 K (fig. 2.5.3). Since there is a competition between the
polarizability and the permanent dipole contributions, even a small error
in any of the parameters will cause an appreciable shift in the temperature
at which AE = 0. Taking this into consideration, the agreement may be
regarded as satisfactory.
If p = 0, AE given by (2.5.14) varies essentially as (T 77*)"1. In such
materials, the electric and magnetic birefringence may be expected to
exhibit the same type of behaviour over a wide temperature range.
Measurements(130) on pure samples of hexylcyanobiphenyl, a nematogen
66
- 3.2
3.2 -
2.4
2.4
Magnetic
s
<l
1.6
0.8
0.8
T* Tm 30
35
40
45
Temperature (C)
50
55
Fig. 2.5.4. Reciprocals of the magnetic and electric birefringence in the isotropic
phase of 4-hexyl-4/-cyanobiphenyl versus temperature. Both give the same value of
T*(T* = 28 C, r N I - T* = 1.1 C). (After reference 130.)
= F0 +
\a(T-
(2.5.16)
67
16
t 14
12
10
i
I
44
48
52
56
60
Temperature (C)
Fig. 2.5.5. Reciprocal of the intensity of light scattering in the isotropic phase of
MBBA versus temperature. (After Stinson and Litster. (128))
molecules are all exactly parallel to one another (see (2.3.2)). Both e and Ae
refer, of course, to optical frequencies. If the incident light is linearly
polarized along z and the scattered light polarized along x
(2.5.19)
From the equipartition theorem
so that
An2
(Ae)2kBT
dQ
(2.5.20)
(where we have made use of the condition sxx 4- syy + szz = 0) and
da^
dQ
_ 8TT2
(Ae)2kBT
(2.5.21)
68
+ \L^as^pspr
(2.5.22)
(J]
(2.5.23)
where
{
a(T-T*)\
may be called the coherence length. The spatial correlation function has
the form
<>(0) s(R)} = const. kB Tcxp ( - R/Q
(R > Q,
69
For an incompressible fluid daa = 0. All four tensors are symmetric and
traceless. Further from Onsager's relations, ju = ju'.
Now consider shear flow along x with a velocity gradient d*;/dz. The
flow induces a birefringence proportional to the velocity gradient with
the principal axes of the index ellipsoid inclined at 45 to the x, z axes. In
the steady state R^ = 0, cpxz = \pi dv/dz. Therefore
Sxz =
ju
dv
~2a(T-T*)~dz'
(2.5.27)
0
0
0
0
0
-sv
(2.5.28)
70
Therefore x'yz represent the principal axes of the order parameter tensor.
The difference between the dielectric constants (at optical frequencies) for
polarizations along the x' and z' axes is
where An is the birefringence when the molecules are all perfectly parallel.
The flow birefringence may therefore be expected to show an anomalous
increase as the temperature approaches the transition. This was observed
by Tolstoi and Fedotov(121) many years ago in PAA. The experiments of
Martinoty, Candau and Debeauvais(134) on MBBA have confirmed the
temperature dependence predicted by de Gennes's equation (2.5.29) (see
fig. 2.5.6).
V2
-kBT\n
(T-T*)-
0.0762-
71
-i 5
45 7\
NI
47
49
51
53
55
57
Temperature (C)
Fig. 2.5.6. Flow birefringence in the isotropic phase of MBBA. Crosses represent
the birefringence (3n) per unit shear rate (G) and circles the reciprocal of this
quantity. (After Martinoty, Candau and Debeauvais.(134))
energy expansion (2.5.1) of the Landau model. However, it does not yield
a satisfactory value of 7*. Since (A/kBTNI V2C) = 4.54, T*/Tm = 0.908.
For PAA, TNI = 408K so that 7NI - T* ~ 40 K, whereas empirically
r NI -r* - 1 K.
72
flf(00j)g(0^
(2.6.1)
3=1
where
J f Jfl/(
J-J
(2.6.2)
while that for an outer shell molecule, say 1, to assume an orientation 015
^i is
JJ/(#oi) S(0i) d(cos 0o) d^ o J... J f l / ( ^ ) g(03) d(cos 0,) d^. (2.6.3)
If we postulate that these two probabilities are identical when 0 and 1 have
the same orientation, we obtain the following consistency relation due to
Chang :(139)
V(009tp0;Ol9q>1)=fL001)g(eei)ll
i)ll
ff f/(0
f o ,)^,)d(cos0,)d^,. (2.6.5)
3=2 J JJ
73
Krieger and James postulated that this probability should be the same
irrespective of which molecule is regarded as the central one, i.e.,
; 0 1 5 cp^ = ^ ( 0 1 ? q>x; 00, (p0),
so that
g(00)
TZI = p (constant),
(2.6.6)
(2.6.7)
(2.6.8)
The assumption here is that the energy is independent of cp, i.e., that the
distribution is cylindrically symmetric. We also ignore the volume
dependence of the potential function. For a given value of z, the values of
B2, i?4, etc., can be derived in terms of B* at every temperature such that the
consistency condition (and hence the thermodynamic equilibrium of the
system) is satisfied accurately. All the properties of the system can therefore
be deduced in terms of a single parameter B*. By retaining terms up to
P4(cos0) in the mean field potential (2.6.8), it is found (137141) that the
maximum error in fulfilling Chang's relation (2.6.4) is less than 0.08 per
cent for z = 8. Further it has been verified that when terms up to P12(cos 8)
are included, the error becomes negligibly small (~ 10"9).
The long-range order parameter s is given by
...
(2-6-9)
...
where
eoj)},
(2.6.10)
</>2(cos0o,)>
... P2(cos 60j) y/(60, <pQ; 6p ^ ) d(cos 0O) dcp0 d(cos 0^) d ^
. (2.6.11)
... y/(0o, <p0; 9p I*,) d(cos 0O) d(p0 d(cos 0) &q>}
J
74
B*/kBT
0.60
0.62
0.15
0.20
0.25 -
1
1
phase
" Disordered
/
S
0.66
0.64
'i
__
2nd order
transition
- 1st order
transition
Ordered phase
<
0.30
Fig. 2.6.1. Plot of <P2(cos 0OJ)> (= -2U/NzB*) versus B*/kB Tfor z = 8.(137) At the
first order transition temperature the shaded areas are equal so that the Helmholtz
free energy of the ordered and disordered phases are the same. At the second order
transition temperature <P2(cos6Oj)} = c2 = l/(z1).
The plot of U versus \/T (at constant volume) shows the characteristic
sigmoid shape (fig. 2.6.1). The curves for the ordered and disordered
phases meet at the apparent second order transition point T*, which is also
the temperature at which the short-range order parameter (P 2(cos60j)) in
the isotropic phase = l/(z1). The first order transition point 7 ^ is the
temperature at which the shaded areas are equal, i.e., when the Helmholtz
free energies of the ordered and disordered phases are the same. The
calculation gives (Tm-T*)/TNI
= 0.062 for z = 8, 0.04 for z = 4 and 0.03
for z = 3. This is an improvement over the Maier-Saupe value of 0.092,
though still much higher than the experimental value of 0.003.
The curves for s, <P4(cos#)> and (P2(cos60j)} as functions of temperature are shown in fig. 2.6.2. All three parameters change discontinuously at Tm; the long-range order drops to zero at the transition, while
the short-range order persists even in the isotropic phase. At the transition,
s = 0.400 (for z = 8) which is slightly lower than the Maier-Saupe value of
0.4292.
Methods involving less stringent assumptions, which have the advantage
that the relevant equations are somewhat easier to solve, have since been
put forward.(142"4) These are based on approximations used earlier in
theories of magnetism. (145146) However, when higher order terms (up to
P4(cos0) or higher) are used in (2.6.8), all these methods yield results that
are practically identical to those given by the Krieger-James approxi-
2.6 Near-neighbour
correlations: Bethe's method
0.6
Order parameter
0.5
Nematic
- - -
Isotropic
=-
1
0.90
1
0.95
1.0
T/Tm
Fig. 2.6.2. Short-range order parameter <P2(cos #0;.)> and the long-range order
parameters s and <P4(cos#)> versus T/T^ calculated in the Krieger-James
approximation.(137) The long-range order disappears at the transition but the
short-range order persists even in the isotropic phase.
mation. (137141) Lekkerkerker et al.a*7) and Van der Haegen et al.am have
used three-particle and four-particle cluster approximations in analogy
with a method followed in magnetism. (149) This leads to a slight
improvement in (TNI r*)/7^ I ? the precise value depending on the type of
lattice considered. A Monte Carlo simulation study/ 150 ' 151) of a lattice
version of the Maier-Saupe theory gives (Tm-T*)/Tm
0.01.
2.6.2 Antiferroelectric short-range order
The vast majority of nematogens are polar compounds but the absence of
ferroelectricity in the nematic phase shows that there is equal probability
of the dipoles pointing in either direction. Because of this it is generally
assumed that the permanent dipolar contribution to the orientational
order is negligibly small. However, a simple calculation shows that the
interaction between neighbouring dipoles is by no means trivial compared
with dispersion forces, particularly in strongly polar materials. We shall
now consider a model which takes into account the influence of permanent
dipoles and is at the same time consistent with the non-polar character of
the medium. (136137)
76
o
Fig. 2.6.3. Preferred orientation of neighbouring dipoles in the end-on and
broadside-on positions. However, because of the anisotropic shape of the
molecules, situation I is much more important than II in the nematic structure and
there results a net antiparallel correlation between neighbouring dipoles.
0.6
"
Isotropic
Nematic
0.4
0.2
0 0
-0.2
0.90
0.95
1.0
Fig. 2.6.4. Short-range order parameters (P^cos^.)), <P2(cos^0j)> and the longrange order parameter s versus T/Tm. The negative value of (P^cos^.)) signifies
antiparallel ordering. The curves are calculated for A*/B* = 0.5.(137)
Suppose as before that the molecules are cylindrically symmetric so that
the dipole moment is along the major molecular axis. Now if a dipole is
fixed at O as shown in fig. 2.6.3,1 and II represent situations of minimum
energy for a neighbouring dipole in the broadside-on and end-on positions
11
(2.6.12)
KB 1
Fju2 , . _
(2.6.13)
J
and
1
4nNphF\
(2.6.14)
where the symbols F9 h, etc., have the same meanings as in 2.3.5. Using the
theoretically derived s of fig. 2.6.4, e1, e2 as well as s = l(e + 2e2) calculated
from these equations are presented in fig. 2.6.5 for a strongly polar (nitrile)
compound of the type studied by Schadt (91) (see fig. 2.3.9(b)). The
parameters used in the calculations are: ju = 5 debyes along the major
molecular axis, a = 28 x 10~24 cm3 and aa = 15 x 10~24 cm 3 (evaluated
approximately by assuming addivity of bond polarizabilities extrapolated
78
25 -
Nematic
Isotropic
20 -
I
15 -
-^^^_
10 -
5 -
0.85
0.90
0.95
1.0
Fig. 2.6.5. Theoretical variation of the dielectric constants e19 s2 and e = |(e1
2
with T/Tm.a31) The theory predicts that in the nematic phase should be slightly
lower than the extrapolated value of els, the dielectric constant in the isotropic
phase. This is found to be the case experimentally (see fig.2.6.6).
to low frequency). Since thermal expansion is ignored, the rate of variation
with temperature is reduced especially near the transition, but apart from
that it is clear from a comparison with fig. 2.3.9(b) that the dielectric
anisotropy a is of the right order of magnitude.
An interesting consequence of the theory is that the mean dielectric
constant e should increase by a few per cent on going from the nematic to
the isotropic phase because of the diminution in (P^cos 00j)}. This is found
to be the case experimentally in a number of strongly positive materials^ 1 ' 130152) (fig. 2.6.6). (A similar increase is seen in some negatively
anisotropic materials also, e.g., PAA (90) (see fig. 2.3.9 (a)); this can probably
be explained as due to an antiparallel correlation between the longitudinal
components of the dipole moments. As far as the transverse components
are concerned there will not be on the average any orientational correlation
for position I of figure 2.6.3 because of the cylindrically symmetric
distribution about the optic axis, but there will be an antiparallel
correlation for position II which is, however, likely to be so weak that it can
probably be neglected.)
Direct X-ray evidence for such antiparallel local ordering in the nematic
and isotropic phases of 4/-n-pentyl- and 4/-n-heptyl-4-cyanobiphenyls
79
18 -
16
J4
a
I
12
5
10
25
30
35
Temperature (C)
80
25.7 A
We now turn our attention very briefly to the nematic liquid crystal
surface. A variety of experimental studies have established conclusively
that orientationally ordered states, and in certain materials even density
modulations, develop in the vicinity of the free surface. We describe below
the salient features of these observations.
81
37
35
130
120
140
T(C)
38r
36
34
160
190
180
170
r(c)
29
28
-16
-12
-S
12
16
20
r-r NI (c)
Fig. 2.7.1. Experimental variation of the surface tension with temperature for three
nematic liquid crystals, (a) PAA,(156) (b) /?-anisaldazine(156) and (c) 5CB. (158159)
Open circles represent measurements using the pendant drop method. Filled circles
in (c) are values determined independently by Gannon and Faber(159) using the
Wilhelmy plate method.
82
Fig. 2.7.2. Schematic form of the density profile p(z) and its gradient \dp{z)/dz\
across the nematic liquid crystal-vapour transition zone.
It is a well known result that the gradient of the surface tension (y) versus
temperature is directly related to the surface excess entropy per unit area as
follows:
%- = -(So-SX
(2.7.1)
where the suffixes o and /? refer to the surface and bulk states. Thus if there
is surface ordering SG may be less than Sfi, and y may actually show a
positive slope.(154) Such a trend was discovered by Ferguson and Kennedy(155) in PAA and some other compounds many years ago and has been
confirmed by precise measurements on a number of liquid crystals using
different techniques. (1569) Some experimental curves are shown in fig.
2.7.1. The suggestion has been made (154) that the orientational order near
the surface is determined by two competing effects: (1) the disordering
effect of the spatial delocalization across the liquid-vapour transition zone,
and (2) the ordering effect of a surface torque field. The former may be
assumed to be proportional to the density profile across the interface and
the latter to the gradient of this profile (fig. 2.7.2). The different trends
observed in the different materials (fig. 2.7.1) can then be interpreted as
arising from the relative strengths of these two opposing effects and their
variations with temperature. (160)
Light scattering and optical reflectivity studies (161) again reveal the
existence of orientational ordering at the surface. For MBBA it is found
83
io 4 -
101 -
io1
1.0
l.i
QJQo
Fig. 2.7.3. High resolution specular reflectivity data for 80CB near the peak due to
the formation of smectic-like layers near the free surface of the nematic phase. The
open circles refer to the scale on the left and the filled circles to the scale on the right.
The temperatures T- TNA are (a) 0.10 C, (b) 0.21 C, (c) 0.40 C and (d) 1.80 C.
It is seen that the peak becomes significantly sharper as the temperature decreases,
showing that the number of surface induced layers increases on approaching the
nematic-smectic A transition point 7^A. (After Pershan et al.a62))
that the molecular long axis is inclined at about 75 with respect to the free
surface, the angle being temperature dependent, while for PAA the angle is
zero and temperature independent.
Smectic and cybotactic nematic systems may be expected to show an
oscillatory density profile near the surface. That this is the case has been
strikingly demonstrated by the very fine X-ray reflectivity measurements
of Pershan et a/.(162163) (figs. 2.7.3 and 2.7.4). Surface-induced smectic
layering is observed in the nematic or isotropic phase that extends into the
bulk, the number of layers increasing as the temperature approaches the
transition to the smectic phase.
84
10"
10"
= (T-T1A)/TlA
Fig. 2.7.4. The measured X-ray reflectivity from the free surface of the isotropic
phase of 12CB at temperatures very close to the isotropic-smectic A transition
point. The step-like form of the intensity curve reveals the quantized nature of the
layer growth at the surface. (After Ocko et /.(163))
Not much progress has yet been made in giving a quantitative molecular
statistical description of these remarkable surface phenomena.
3
Continuum theory of the nematic state
85
86
vitj = dvJdXj
and
T=dT/dt.
We shall consider the medium to be incompressible (vt i = 0, where vt is
the linear velocity) and at constant temperature (t = T t = 0). We shall
assume further that the director is of constant magnitude. This implies that
the external forces and fields responsible for elastic deformation, viscous
flow, etc., are very much weaker than the molecular interactions giving rise
to the spontaneous alignment of the neighbouring molecules. It is indeed
a valid assumption in all the static and dynamic phenomena discussed in
this chapter. We may therefore conveniently choose n to be a dimensionless
unit vector (ntnt = 1 ) .
Let the material volume be Vbounded by a surface A. The conservation
laws take the following form:
Conservation of mass
/?dF=0,
(3.1.1)
f pv( d V = f f( d V+ f tn dA,,
(3.1.2)
ai
Jv
Jv
JA
where ft is the body force per unit volume and tn the stress tensor.
Conservation of energy
d
fi
dt)v
**
2 X
f
Jv
l %
f
JA
(3.1.3)
where px is a material constant having the dimensions of moment of inertia
per unit volume (ML'1), U the internal energy per unit volume, G{ the
87
external director body force (which has the dimensions of torque per unit
volume since nt has been chosen to be dimensionless), ti = tn Vj the surface
force per unit area acting across the plane whose unit normal is vp and
st = nn Vj the director surface force (which has the dimensions of torque
per unit area). We assume here that there are no heat sources or sinks.
Conservation of angular momentum
jA
=
{^mxJjc + e^n^dV+X
JV
{e^
(3.1.4)
JA
or in vector notation,
dtjy
[p(rxy)+Pl(nxn)]dV
JA
f
Jv
Pl nt d V
(3.1.5)
JA
where gt is the intrinsic director body force, which has the dimensions of
torque per unit volume and whose existence is independent of Gt.
Converting surface integrals into volume integrals and simplifying,
(3-1.1)(3.1.5) lead to the following differential equations:
p = 0,
(3.1.6)
PVi=fi + hi,P
U = tn dtj + nn NtjgiNt,
(3.1-7)
(3.1.8)
Pifi^Gi + gt + njij,
(3.1.9)
where
hi - nkj "i, k + gj ni = hi - Xjci "j, ic + gi nP
(3.1.10)
88
i I sdv^o,
where S is the entropy per unit volume. Defining the Helmholtz free energy
function per unit volume
F=U- TS,
we obtain for a system in isothermal equilibrium
hi dij + nn Na-gi Nt-F^0.
(3.1.11)
niJ9
nt and vu.
(3.1.12)
i p
Ni9 a n d dtj.
dF d
dnt dt
l 3
'
dF dF d
87Vt- x ddt- dt ij
But
nt = Ni + Wyii^
Nt-wjtnj9
d
(nt,) = Ntj - wki nk j - (dkj + wkj) nt k
and
dF d ,
dF
dF
dF
dF
dF Ar
~ A
dF
*^ ^
dF
i
V ^ ^
dF .
^ Pi AT
dF .
dtj
i ^ 7\rl
dd " 3v
tj
dF
i> k
^
dn
tk
J*
dF
>
(3
-U3)
In view of the constitutive assumptions, it is clear that wji9 Nip Nt and dtj
can be varied arbitrarily and independently of all other quantities and
hence their coefficients must vanish, i.e.,
or
F=F(ni,niJ);
dF
dF
dn^
' 8^>Jfc
Wyhn, *.-
hw*. y -
dF\
' dnk J
(3.1.14)
dF
,-h,
dn
dF
fr-
^M
Yn
(3.1.15)
nji-^L
= 0;
(3.1.16)
?\F\
,^0.
(3.1.17)
Let us write the stress and the intrinsic director body force as
90
Since dtj and Nt can be chosen arbitrarily and independently of the static
parts fn and gf,
tfg,
(3.1.20)
t'pdv-g'tN^O.
(3.1.21)
(3-1.22)
+ s?
and
ni} by p^ + n^
(3.1.8) continues to be satisfied because
n{nt = niNi = n.N^ + n.^N, = 0.
Thus (3.1.19), (3.1.20) and (3.1.16) become
(3.1.23)
n
k,j
t j
nn=P}nt + ^ - ,
(3.1.24)
(3.1.25)
- + ^ + ^ = 0.
(3.1.27)
91
/o
AT,
where A and B are functions of ni (at constant T and p). They can therefore
be expanded as
nt + a4 Sik n, + a5 /if ^ /i^,
Sn 3
km
= du = 0,
(3.1.29)
where
a15 = a9 + a10,
a16 = a7 + a n ,
and
^ = (y0 + 7e dkj nk n,) n{ + y9 dik nk + yx Nt
(3.1.30)
where
But in view of (3.1.22)
y9 = a 1 6 -a 1 5 ,
y1 = a 4 - a 3 .
(3.1.31)
km
92
i.e.,
ax nt nj dtj + (quadratic in dip Nt) ^ 0.
As di:j can be chosen arbitrarily, ax = 0. Also the coefficient of Sjf(= p
say) in tn and that of nt ( = y say) in g[ are arbitrary since du = 0 and
ntNt = 0. Putting:
a
Ml =
i4
M* = 13
//2 = a 4
/z5 = a 1 6
we obtain
t'a = Mi nk nm dkm nt H, + // 2 w, A^, + // 3 /i4 ^ + // 4rfi4+ ju5 n, nk dki + n% nt nk dkp
(3.1.32)
g't-^Nt + ^dfl,
(3.1.33)
/l2 = // 5 -// 6 .
(3.1.34)
PSji
On
k,j
+ ju2 ^ Nt + ju3 nt N, + // 4 dn
+ M5njnkdki
+ /i6ninkdkj9
(3.1.36)
goi+gt
yn.-p^-^
+ X^
+^ n ^ .
(3.1.37)
93
The jus represent the six coefficients of viscosity of a nematic liquid crystal.
However, the number of independent coefficients reduces to five if we
assume Onsager's reciprocal relations.
3.1.4 ParodVs relation
From (3.1.21) we observe that the rate of entropy production per unit
volume
TS=t'ndij-g'iNi
where t'n is given by (3.1.32) and g{ by (3.1.33). Since t'n is an asymmetric
tensor it can be resolved into a symmetric component Ytj and an
antisymmetric component Zip where
Y
^ 4 = 0,
TS-Yfidv-g'tNt.
Thus the entropy production can be separated into two parts, one due to
the linear motion of the fluid and the other due to the orientational motion
of the director. Now h = H x n, where O is the angular velocity of the
director. It is then easily shown that
is)-
94
(3.1.39)
(11)
We may therefore expand the free energy per unit volume of a deformed
liquid crystal relative to that in the state of uniform orientation as
Twist:
Bend:
9w
aw,
6x'
9z'
dz
dy
dz
driy dn^
dy
dz
(say).
0
^21
l_/c 31
k12
K
k
k 13 1
AC
k \
^22
^ 2 3 I >
/c 3 2
A:33J
(3.2.1)
K K
. 0
K
K
0
0
0
0
k + 0,
polar
non-polar
x = 0,
and polar
and non-polar kx = 0,
k2 + 0
k2 ==
| 0
k2 = 0
k2 = 0.
The tensor kijlm (or ktj in the abbreviated notation of (3.2.1)) has 81
95
96
components in general, but as a79 a8 and a9 are zero there remain only 36.
The presence of cylindrical symmetry reduces this number to 18 with only
5 independent constants:
kl2
k22
0
0
0
k15
k12
0
k22
~^12
^24
/Cqq
^24
Lo
k12
k12
0
CO
0
0
0
-k12
0
0
0
0
0
0
^33-
*2 ==o.o.
(15)
(3.2.2)
by standard
^22
+ kM)[nijnji-(niif],
(3.2.3)
ijk nj\np
= 0.
(3-2.4)
It is seen that &24 plays no role in (3.2.4) and can be omitted as far as
equilibrium situations are concerned (Ericksen (17)). When external body
torques are absent {Gi =0), the solutions of (3.2.4) are
n2 = sin (qz+y/\)
97
where
^ + ? S = 0.
dx* dy*
(3.2.6)
(3.3.1)
px h'i Gi + gi + nn p
(3.3.2)
(3.3.3)
where
$i = -P*ii-^t.i>
(3-3.4)
(3.3.5)
98
(3.3.6)
,
(3.3.7)
N^rit-w^
(3.3.8)
2dtj = vu + v^
(3.3.9)
IWt^Vij-v^
(3.3.10)
where
(3.3.11)
gi=yni-PjniJ-dF/dnt9
(3.3.12)
g'^KNt + ^dw
(3.3.13)
JU2+JU3 = JU6-JU5.
= ^jni + dF/dnij.
(3.3.16)
(3.3.17)
fields
99
equilibrium (3.3.17) where Gt is the external director body force due to the
magnetic field H. If/y and/ are the principal diamagnetic susceptibilities
per unit volume along and perpendicular to the director axis respectively,
G,=/a//,,#
(3.4.1)
where / a = / , , - / .
Let us consider first the case of a nematic film in which the initial
undisturbed orientation of the director is throughout parallel to the glass
plates. The magnetic field H is now applied perpendicular to the director
and to the plates (fig. 3.4.1 (a)). For this geometry, n = (cos #,0, sin 6),
H = (0,0,//) and G = (O,O,/ a // 2 sin0). The free energy of elastic
deformation (3.3.6) reduces in this case to
Jo
cu.
'~AtV
(3A2)
In other words, deformation occurs only above a certain critical field Hc.
This is referred to as the Freedericksz effect. The threshold condition can be
used for a direct determination of the splay modulus klv
100
H<H
H<Hn
ZZZ H>HC
1 1 ) 1 ) 1 }
/ / / /
H<HC
>Hc/
/
(c)
/ / /
I I I I I I II
Fig. 3.4.1. The experimental Freedericksz geometries for the determination of the
(a) splay, (b) twist and (c) bend elastic constants of a nematic liquid crystal.
For H > Hc, the deformation at any arbitrary point can be computed
from the expressions*19'20*
J 0
Isin 2 ^_-sin 2 0
= I arc sin
where q =
..., (3A3)
-,
(3.4.4)
(k^-k^/k^.
fields
101
and for n = (sin 0,0, cos 0), H = (//, 0,0) (fig. 3.4.1 (c)),
/XJ-
(3A6)
(21)
De Gennes
introduced a parameter which he called the magnetic
coherence length to define the thickness of the transition layer near the
boundary. Consider, for example, a nematic liquid crystal occupying the
half space z > 0. Let the wall, the xy plane at z = 0, impose an orientation
along x and let the magnetic field be applied along y (analogous to the
geometry of fig. 3.4.1 (&)). If cp ( = \n 6) is the angle made by the director
with the field, the equilibrium condition is easily shown to be
102
121.3 Cu
119.6 C
117.5 C
~liAAAA/\
IWIAA/W
JlMAAA/\
-jmmw
,,-jmm/v
115.5 C
106 C
101.3 C
1
Magnetic field (kG)
Fig. 3.4.2. Raw recorder traces of interference oscillations due to the change in the
sample birefringence with deformation for hexyloxyazoxybenzene at various
temperatures. Polarizer and crossed analyser are inclined at 45 to the principal
axes of the specimen. The sudden onset of oscillations occurs at the threshold field.
The increase in the threshold for the successive traces illustrates the rapid
temperature variation of the elastic constant. Sample thickness 45 /zm. 7^ =
128.5 C. (After Gruler, Scheffer and Meier.(20))
untwisted nematic. For this reason Freedericksz and Tsvetkov(18) used a
total internal reflexion technique by letting the light beam fall at an
appropriate angle on the specimen contained between a convex lens and a
prism. A simple and more direct method has been proposed.(22) If the
ellipsoid of refractive index is viewed obliquely, say at 5 or 10 to the
director (fig. 3.4.3), the effective Sn is reduced to a low value and the
deformed medium can be shown to be optically equivalent to a rotator and
103
If !:
1
f
- i!
,:i I
IS
Ill'll1
!>: I
'iliii!1
(a)
07/
Fig. 3.4.3. (a) The usual experimental configuration for the optical observation of
the Freedericksz effect. Light is incident normal to the film. However, for reasons
discussed in the text, this arrangement is unsuitable for observing a twist
deformation, (b)l Oblique' configuration which enables the optical detection of a
twist deformation.(22) The magnetic field is perpendicular to the plane of the paper
in both cases.
104
-30
-20
-10
r-r NI (c)
Fig. 3.4.4. Twist elastic constant (k22) versus temperature for PAA and PAP.
Squares, circles and triangles represent independent measurements on different
samples. (Karat. (22))
105
0.7
0.6
^
0.5
0.4
0.3
0.2
0.1
0
i
(a)
(b)
Fig. 3.4.6. The orienting effect of grooves. Extra elastic energy is required for the
nematic director to lie across the grooves on the solid surface as in (a) rather than
to lie parallel to them as in (b).
direction than for them to lie parallel to it (fig. 3.4.6). A simple calculation
shows that this energy difference for the material near the surface is quite
appreciable and almost impossible to overcome by means of a magnetic
field. The anchoring is therefore firm. Deposition of silicon monoxide and
certain other materials on the glass surface at oblique incidence has been
shown to have essentially the same effect as rubbing.(25)
106
107
H < Hc
Hc
and
(3.4.8)
where
f(9) = kxl co
g(9) = (k22 cos2 9 + 33 sin2 9) cos2 9.
After integration, (3.4.7) and (3.4.8) yield
(3.4.9)
and
(3.4.10)
where A and B are constants. Equation (3.4.9) may be rewritten as
(3.4.11)
108
A =
Substituting in (3.4.11)
or
Aw)
i//, say.
Similarly from (3.4.8)
7r
Therefore
=2
(3.4.12)
JJ 00
and
(3.4.13)
Transforming to a new variable given by sin X = sin ^/sin 0 max, (3.4.12)
and (3.4.13) become
,=2 r
r__j
Jo L ^ -
and
Jo
where
M{)
T 1
(si
= [fc33 - 2k22 - (kS3 - k22) (sin 2 W + sin 2 emj]/g(6)
g(0m!lx).
109
6 h
Fig. 3.4.8. Computed relative capacitance change AC/C 0 and optical transmission
T between parallel polarizers (both parallel to the director at one of the boundaries)
of a twisted nematic film as functions of H/Hc. The total twist angle is n/2. Film
thicknesses 13 and 54 jum. The threshold for optical transmission increases with the
thickness of the film. (After Van Doorn.(27))
Taking the limit 0 m a x ^ ' ', we have
2<po/d,B^2k22<po/d,M(k33-2k22)/k'
=2n
ll9
22
Jo \x.H*- (4tpl/d*){k^-2k22)
or
g(0)^k22, dcp/dz
and
I
(3.4.14)
which is the critical field for the deformation to occur. (28) Thus a
measurement of Hc in this geometry gives k22, if kxl and kZ3 are known.
However, for reasons discussed in 4.1.1, the deformation cannot be
detected optically by observation in polarized light at normal incidence
until the field is well above the threshold value Hc. A more convenient way
of detecting Hc would appear to be by measuring the change in the
capacitance (27) (fig. 3.4.8).
A noteworthy feature of the twisted nematic (TN) is that the intensity of
110
for 2q>0 = n/2, where ea is the dielectric anisotropy. For voltages of the
order of 2 Vc the twist is lost over most of the sample (fig. 3.4.7, bottom), the
polarization is no longer rotated by 90 and the transmitted light is
extinguished by the second polarizer. With parallel polarizers, one can get
the opposite effect, namely bright field in the ON state and extinction in the
OFF state. LCDs are often operated in the ' reflexion' mode: this is done
by having a diffuse reflector on the rear side of the cell. Unlike the dynamic
scattering mode (see 3.10.1) in which conductivity plays a crucial role, the
TN LCD is a field effect device. Therefore, it is advantageous to have high
purity (low conductivity) materials. A major advance in the materials
development was the discovery by Gray and others(31) of highly stable
mesogens of strong positive dielectric anisotropy, like 5CB, 7CB and
related compounds. A wide nematic range, from about 10 C to 70 C,
that is necessary for most practical devices, is obtained by making suitable
mixtures of these compounds. With a 90 twist there is equal probability of
the medium acquiring a right-handed or left-handed twist, which results in
the formation of disclination walls. To avoid this, a small quantity of a
cholesteric dopant is added to the nematic mixture in order to favour one
sense of twist throughout the sample. Typically, with such materials, the
threshold voltage is about 1 V, the switching time (for a cell of thickness
fields
111
8 //m) a few tens of milliseconds and the power consumption a //W cm" 2 of
display area. A disadvantage with the TN cell is that the viewing angle is
limited to about 45 from the normal, but the advantages far outweigh
this disadvantage. By patterning the electrodes appropriately, it is possible
to display the required information, whether it is digits or letters or any
other symbols.
For higher information content it is convenient to use the dot matrix
configuration, with electrodes as horizontal rows on one glass plate and as
vertical columns on the other. The intersection of a row and a column is a
picture element (pixel). Thus a matrix display with TV rows and M columns
has TV x M pixels but only N+M connections. When the number of pixels
becomes large one resorts to an electronic addressing technique known as
multiplexing: each row in the matrix is selected sequentially, while
appropriate data waveforms coded with the information are applied to the
columns.(32) Because of the slow response times of LCDs (~50 ms), each
pixel responds only to the rms of the resulting waveforms. As the number
of lines TV in the matrix increases, the fraction (I/TV) of the total time that
the selected pixels see the full select pulse decreases, thereby reducing the
ratio J^ ms (sel)/^ ms (unsel). Alt and Pleshko(33) showed that
For TV = 100, this ratio is just about 1.1. If the electrooptic response of the
display were very steep (like a step function), a small change in the voltage
would produce a large change in the director orientation, and it would be
possible to activate a given pixel without altering the state of the other
pixels, but since the response is rather broad (fig. 3.4.8), TV is restricted to
about 100 to get a reasonable contrast ratio. To overcome this limitation,
in the pocket TV, for example, an active addressing technique is used in
which each pixel is backed by a thin film transitor (TFT) which enables one
to apply any desired voltage to the ON pixels and zero voltage to the OFF
pixels. With each column replaced by three which are coated with different
pigments, one can get full colour display. Further, the broad electrooptic
response of the material is actually helpful for producing grey scales.
An advance that has extended the application of LCDs to full page
computer terminals and other high information content displays without
having to resort to TFTs is the supertwisted nematic device/ 34 ' 35) It makes
use of the fact that the electrooptic response of the nematic gets
progressively steeper as the twist angle ^ is increased, until at a certain
112
1.0
1.5
2.0
Reduced voltage (V)
2.5
3.0
Fig. 3.4.9. Computed values of the tilt angle 0max in the midplane of the sample
versus voltage for various twist angles <fi for a standard TN mixture. Cell
thickness/pitch = (j>/2n in all cases and the director tilt at the surface 6S= 1.
(After Scheffer.(32))
value of ^ it becomes infinitely steep. This is illustrated in fig. 3.4.9, which
gives plots of the computed values of the maximum director tilt angle #max
in the midplane of the sample versus voltage for various values of <j> for a
standard TN mixture. In all cases the cell thickness/pitch ratio d/P =
<f>/2n, and the director tilt at the surface 6S= 1, which is the value
generally obtained with standard nematic materials on rubbed polymer
coatings on glass. As remarked earlier, the steeper the response the higher
the multiplexability, and the infinite slope for <j> = 270 represents the best
condition for rms multiplexing. For higher twist angles the slope is
negative, and in practice the device becomes bistable.(36) The precise value
of </> at which the slope becomes infinite depends on combinations of the
relevant material and device parameters,(37) but for most known nematic
materials it lies in the range 180 < <f> < 360. The high twist is stabilized in
the cell by the addition of the appropriate quantity of a cholesteric dopant.
Modified forms of these displays have been developed - e.g., the 'dye'
display which has a pleochroic dye dissolved in the nematic material and
requires the use of just one polarizer - but we shall not be discussing them
here. Analytical expressions have been derived(37) which simplify the
computational effort involved in optimizing the material and device
parameters, but one has to rely on numerical modelling to give a complete
description of the dynamical characteristics of these devices.(38) Certain
unusual dynamical effects observed in the TN device, e.g., the 'reverse-
113
twist' and 'optical bounce' effects, will be explained very briefly in a later
section while discussing the dynamics of the Freedericksz transition (see
3.8).
where rj = 2z/d, d being the sample thickness, and the subscripts denote
partial derivatives, e.g.,
etc.
(3.4.16)
114
^ tanh q 2 tan qx = 0,
(3.4.17)
where
115
0.8
RH
Fig. 3.4.11. Threshold plots of RH (the ratio of the periodic distortion threshold
and the normal (homogeneous) Freedericksz threshold) and of Qyc (the dimensionless wavevector of periodicity at threshold) versus k1/k22. Periodic
distortion is possible only when kxjk22 ^ 3.3; for kxjk22 ^ 3.3 the normal
Freedericksz deformation is favourable. (After reference 42.)
from Hz(0) as Qy increases, reaching a minimum Hz(Qyc) at Qy = Qyc. On
increasing Qy beyond Qyc, Hz(Qy) increases again. Hz(Qyc) is regarded as
the threshold for periodic distortion, and Qyc (and kyc = nd/Qyc) the
wavevector (and wavelength) of periodicity at threshold.
Plots of the ratio RH = Hz(Qyc)/Hz(0) and Qyc as functions of r =
fcn/fc22 are shown in fig. 3.4.11. It is seen that as r decreases RH increases
and Qye decreases. When r tends to a lower limiting value rc ~ 3.3, RH^\,
Qyc-+0, showing that for r < rc, the periodic distortion is no longer
favourable. The transition to the periodically distorted state can be treated
as a second order phase transition with Qy as the order parameter. <43"5)
Nehring and Saupe put forward the view that linear terms of the second
derivatives of n in the nematic free energy expansion can, in principle,
make contributions of the same order as the quadratic terms of the first
116
(3.4.18)
3.5 Disclinations
117
118
Fig. 3.5.1. Director orientation (indicated by arrows) along a polar line making an
angle a. Incident light that is linearly polarized at angle ^ or (j>n/2 will be
extinguished by a crossed analyser and will give rise to a dark brush.
Consider a planar structure in which the director is confined to the xy
plane (the z axis being normal to the film). Taking the components of the
director to be nx = cos(/>, ny = sin (/>, nz = 0, and making the simplifying assumption that the medium is elastically isotropic, i.e., k1 = k22 = 33 = k,
(3.3.7) and (3.3.17) reduce to
(3.5.1)
(3.5.2)
= 0
(apolar)
119
5.5 Disclinations
* = *
5=1, C= 0
S=l,C=7T/2
s=\,c=n/4
5=3/2
5=2
120
x axis, it is seen from fig. 3.5.1 that the polarization will be unchanged at
all points on the polar line a and hence will not be transmitted by the
analyser. This will result in a black brush at an angle a. A similar situation
will arise when <j> changes by n/2. The angle between two successive dark
brushes is therefore Aa = A<p/s = n/2s. Thus the number of dark brushes
per singularity is 2n/\Aa\ = 4|s|. Also if the polarizers are turned through
angle co the brushes rotate by an angle co/s. The rate of rotation of the
brushes of the two-brush disclination (s = ) is therefore twice as fast as
that of the four-brush variety (s = 1). If the polarizers are kept fixed and
the microscope stage is rotated by co the brushes turn by co{s \)/s.
Observations in polarized light therefore enable one to determine s both in
sign and magnitude. The existence of |.s| = \ in nematic liquid crystals
establishes the absence of polarity in this phase.
If the structure is one in which the director is not normal but inclined
with respect to the z axis (as in the case of smectic C, see 5.8), half integral
values of s are not possible even if the molecular order is apolar.
The molecular orientation in the neighbourhood of a disclination as
given by (3.5.3) is shown in fig. 3.5.2 for a few values of s. For s + 1, a
change Ac in the constant c merely causes a rotation of the figure by
Ac/(1 s), while for s = 1, the pattern itself is changed.
When the strengths of two neighbouring disclinations are equal and
opposite, the brushes connecting them are circular. By superposition of
solutions of the type (3.5.3)
<f> = </>i + 4>2 = ^i <h + s2 a 2 + c o n s t .
Putting sx = s2 = s,
(j> = sfi + const.,
\FAxAy.
(3.5.4)
3.5 Disclinations
121
Fig. 3.5.3. Curves of equal alignment around a pair of singularities of equal and
opposite strengths. The orientations marked on the circles refer to the case 5 = 1 ,
The disclination is supposed to have a core whose energy is not known. To
allow for this, we postulate a cut-off radius rc around the disclination and
integrate for distances greater than rc to obtain
W= Wc +
nks2\n(R/rc),
(3.5.5)
+ const.
(3.5.6)
(3.5.7)
122
{a)
3.5 Disclinations
123
(a)
(b)
SO/mi
Fig. 3.5.5. (a) s = 1 disclinations and (b) s = \ disclination in a nematic film; (left)
between crossed linear polarizers; (right) between crossed circular polarizers.
(Meyer.(76))
124
I
\
\
(a)
(3.5.8)
(3.5.9)
We look for a solution with (j> = ^(a) and 9 = 6{r). Therefore, (3.5.9)
reduce to
(3.5.10)
- l - f r ? \ - sin 0 cos 9(V</>)2 = 0,
8V = 0.
(3.5.11)
= 0.
5.5 Disclinations
^
-
125
S*
T
'
_^ _\^
** **
(
^ ^
^_
^
-
\ \ I//
(a)
(b)
Fig. 3.5.7. Escaped configurations of (a) s = I, c = n/2 and (b) s = 1 disclinations. Nails signify that the director is tilted with respect to the plane of the paper.
Assuming the boundary condition
0
at r = 0,
= n/2 at r = rQ,
we get
(3.5.12)
\nk
for s = 1.
Interestingly, the energy is independent of r0. On the other hand, the planar
structure gives
W=nk\n(rjrc)+Wc
for s = 1, where rc is the core radius and Ec the core energy. As rc is
expected to be of molecular dimensions, the planar solution has much
higher energy than the continuous structure if r0 is large enough for optical
observations. On the other hand if the capillary radius r0 is extremely small,
or the elastic constant very large (as can happen in the vicinity of a
nematic-smectic A transition) the planar solution may be more favourable
energetically.
The escaped configurations for s = 1, c = n/2 (involving bend and
twist), and for s = 1 (involving twist, bend and splay) are shown in fig.
3.5.7. Structures with the nails pointing in the reverse direction (i.e., with
nz being replaced by n z) are equally probable.
126
z>0
z>0
z=0
-\//'---
z<0
z>0
= 1
z<0
z>0
\ ^--*
\ -v - ^
z=0
\ w \\\
\\\ \\ \
z<0
z<0
z>0
z>0
\ .-//
\
1 1
1 1
z=0
z<0
'
1 1
z<0
(a)
(ft)
Fig. 3.5.8. Twist disclinations: the director patterns for (<z) s = J, c = 0, n/4 and
7r/2; and (Z?) s = 1, c = 0, rc/4 and n/2. In each case, the disclination line is shown
as a full line in the middle of the z = 0 plane. The patterns for (s, c) and ( s, c)
are mirror images of each other.
5.5 Disclinations
127
each other. It is seen that all three types of elastic distortions, splay, twist
and bend, are now present whereas planar solutions for the wedge
disclination involve only splay and bend.
In the elastically isotropic medium, the expressions for the energies and
interactions derived for wedge disclinations are exactly applicable to the
present case. Similarly, the non-singular solution (3.5.11) for s = 1 is valid
in this case, except that the energy for the escaped configuration turns out
to be 2nk\s\ per unit length.(76) The structure of the escaped configuration
is, of course, rather more complicated than that for the wedge disclination.
From the nature of the director patterns it is clear that dark brushes of
the schlieren type will not be seen under the polarizing microscope for light
propagating normal to the film (see 4.1.1). Twist disclinations may
therefore be expected to be less conspicuous than wedge disclinations, and
few observations have been reported of their existence in ordinary
nematics. They do, however, reveal themselves under favourable circumstances in twisted nematics, often as loops separating regions of
different twist.(79)
The Volterra process for creating a loop, i.e., a closed disclination line,
in a nematic is as follows. Let be the surface enclosed by the loop L. Call
the two sides of the surface S + and S~. Rotate the molecules in contact with
I + by an angle sn and those in contact with Z~ by sn about an axis
normal to the unperturbed orientation of the director, where s = |, 1,
etc., is the strength of the disclination line. At finite distances from S the
director will adjust itself and the resulting configuration will be continuous
everywhere except on I .
We now consider a twist disclination loop in a twisted nematic. The
nematic is supposed to have a planar structure with the director parallel to
the xy plane and an imposed twist of q per unit length about the z axis, and
the disclination loop of radius R is supposed to be in the xy plane. The
director distortions are planar, nx = cos^, ny = sin^, nz = 0. On going
once round the disclination line at any point on the loop, the director
orientation <f> changes by 2ns, the sign of which may be either the same as
that of q or opposite.
The net energy of such a structure is(78)
W = n2ks[2sR In (R/rc) - qR2].
When s and q are of the same sign, W has a maximum value when
128
Fig. 3.5.9. Shrinking of twist disclination loops: (a) thin thread |s| = |, (b) thick
thread \s\ = 1 with an escaped (coreless) structure. Each thread was photographed
twice several seconds apart. (Nehring.(80))
The energy decreases for higher and lower values of R. Thus large loops
with R> Ro may be expected to occur. Smaller loops shrink (fig. 3.5.9) and
disappear. When s and q are of opposite signs, loops may not be expected
to occur at all.
A stability analysis(81) has shown that twist disclinations are less
favourable than wedge disclinations in elastically anisotropic media. This
may explain why the former are so rarely seen in ordinary nematics.
3.5 Disclinations
129
(a)
Fig. 3.5.10. Singular points in droplets: (a) spherically symmetric radial (hedgehog)
configuration with the director normal to the surface; (b) bipolar structure with the
director tangential to the surface; (c) singular points in a capillary.
3.5.5 Singular points
Point singularities of strength s = 1 occur in droplets and can also be
seen in thin capillaries (fig. 3.5.10). To obtain their solutions (68) let us set
nx cos (/> sin 9, ny = sin <j> sin 9, nz = cos 9, and use spherical coordinates
x = p sin S cos a, y = /?sin^sina, z = pcosS. If we assume that <j> = ^(a)
and 9 = 0(8)9 we find that
:
do
= 0.
D-M9I
For \s\ = 1,9 = S. On the other hand, if 9 -> n S as p -> 0, then c becomes
c + n.
Fig. 3.5.11 gives the director configurations for some typical cases. The
sections through the (x, y), and the (x,z) or (y,z) planes are identical with
the patterns for the +1 and 1 wedge disclinations in two dimensions.
Any pattern on the left-hand side of fig. 3.5.11 may be combined with any
one on the right to give a possible point singularity.
130
s = 1,
c = n
Fig. 3.5.11. Director field around singular points. Any pattern on the left may be
combined with any one on the right to give the field around a possible point
singularity. (Saupe.(68))
The total energy for a spherically symmetric radial configuration is
E = %nkR.
where R is the radius of the sphere.(82) This configuration (sometimes
referred to as the ' hedgehog' point defect) is realized in droplets with the
director normal to the surface (fig. 3.5.10(a)).
If the boundary condition is tangential, point defects are formed at the
two poles (fig. 3.5.10(b)). The director components in cylindrical polars
may be taken as nr = sin 6, na = 0, nz = cos 0, and assuming
0 = tan" 1
rz
R2-r
E 5nkR.
Elastic anisotropy modifies the idealized configurations shown in fig.
3.5.10.(83>84) More complex structures with an oblique orientation of the
director at the surface have also been reported/ 85 ' 86)
Point singularities of equal and opposite strengths attract one another
and are annihilated(68'87) (see fig. 3.5.12). As the total energy of elastic
deformation around a point defect increases linearly with the radius of the
3.5 Disclinations
131
132
volume enclosed, it has been suggested that the interaction energy of two
defects grows linearly with separation, analogous to quarks interacting
through a gluon field.(65)
3.5.6 Interaction between disclinations and surfaces
Interaction with a plane surface
From the superposition principle (3.5.6) we know that the director pattern
around a pair of like disclinations located at x = d and d is given by
X l
-DJ
\x-D
3.5 Disclinations
133
(a)
(b)
Fig. 3.5.13. Stable dipole pairs formed between 1 point disclinations and air
bubbles in a nematic between (a) crossed linear polarizers and (b) crossed circular
polarizers. (Meyer.(87))
134
<x,y)
s2
DD
r0 being a unit vector directed away from the centre towards the singularity.
At large distances from the cylinder (D $> R)
5.5 Disclinations
(a)
135
(b)
(c)
Fig. 3.5.15. Helfrich walls: (a) a twist wall parallel to thefield,(b) a bend-splay wall
parallel to the field, and (c) a splay-bend wall perpendicular to the field.
On the other hand, a positive disclination will be repelled by the cavity
at all distances.
136
H<HC
H>He
H> Hc
(a)
(b)
(c)
d2
W---1
Umbilics
Consider a nematic film of negative dielectric anisotropy (ea < 0) aligned
homeotropically between glass plates. If an electric field is applied along
the director axis (z axis) a distortion will set in when the field exceeds the
critical Freedericksz value given by
Ec =
(n/d)(k3S/Ej.
137
5.5 Disclinations
I I I I I lllll
11111 T T T T T
lllll
TT
TTT
iilll T T T T T
lllll l l l l l
T 1
[1
\
Fig. 3.5.17. The structure of umbilics of strength 5 = 1 . Nails signify that the
director is inclined with respect to the plane of the paper.
(The experiment can also be performed with a magnetic field and a
negative diamagnetic anisotropy material like a discotic nematic, see 6.5.)
Two possible types of distortions are depicted in fig. 3.5.17. In the
distorted state we have a n component which is degenerate in the xy plane.
Therefore, there can be defects in the n field, and because of the symmetry
in the xy plane, only defects of integral strength can occur. Such defects
have been observed*87>68) (fig. 3.5.18) and are called umbilics.m) They are
somewhat similar to the s = + 1 of the schlieren texture but differ in detail.
Over a distance
from the centre of the defect, the director gradually tilts towards the z axis,
the tilt angle becoming exactly zero at r = 0. In this sense, umbilics have
a collapsed core, but the structure is not exactly the same as that of the
s = 1 defects discussed in 3.5.3.
Planar and linear solitons
When a magnetic field is applied normal to a half-integral disclination line
in a nematic, there results a domain wall terminating in a singular line. The
wall thickness is of the order of . Figs. 3.5.19(#) and (b) illustrate the
director patterns for s = \ and \. Such walls have been referred to as
planar solitons.m) On the other hand, a magnetic field acting on a point
defect with a radial configuration will give rise to a cylindrical domain
138
(a)
Fig. 3.5.19. Planar soliton produced by a field acting on (a) s = \ and (b) s = \
disclination lines.
3.5 Disclinations
139
ccccc
ccccc
Fig. 3.5.20. Linear soliton produced by afieldacting on a point defect of strength
^= 1.
ending in a singular point (fig. 3.5.20). This has been called a linear soliton.
The occurrence of such solitons in the phases of superfluid 3 He,
antiferromagnetics, etc. has been discussed by Mineev(93).
Other interesting cases have been investigated by Sunil Kumar and
Ranganath. (94)
3.5.8 Some consequences of elastic anisotropy
Real nematics are, of course, elastically anisotropic. In certain situations,
as for example at temperatures close to the nematic-smectic transition, the
anisotropy becomes very large and certainly cannot be ignored. We shall
now investigate some of the consequences of elastic anisotropy on the
properties of disclinations.
Solutions for isolated disclinations
As before, we shall begin by considering a planar sample in which the
director is confined to the xy plane. In such a case, a wedge disclination
involves only splay and bend distortions and we need to take into account
only the splay-bend anisotropy (fcn ==
j fc33).
The free energy density (3.3.7) may be written as
F=^[l+cos2(^-a)]/r2,
where (px = d</>/doc, a = tan~ x (y/x), k = (k11 + k33) and e =
(3.5.13)
(k11-k33)/
(3.5.14)
+....
140
(a)
(b)
TT/2,
141
3.5 Disclinations
0.2 -
1.0
Fig. 3.5.22. Dependence of the energy on e = (k1k zz)/(kxl + A;33) for 5 = \ and |
disclinations. Squares are from the Nehring-Saupe approximation and circles from
the perturbation calculation in the neighbourhood of s = 1. (After reference 97.)
energy than bend when e = 1. The solution for s = |, e = 1 is also shown
in fig. 3.5.21(6).
For a general value of e, the corrections can be evaluated by numerical
methods. Fig. 3.5.22 gives the energies of s = + | and s = \ defects in
units of k\n(R/rc). The results of the Nehring-Saupe formula are plotted
as squares, and those calculated from a perturbation expansion in the
neighbourhood of s = 1 are marked as circles.
Interaction between disclinations
The radial force of interaction between disclination will, of course, be
modified because of anisotropy. (97) In addition there will now be an
angular component of the force.(98) The physical basis for the angular force
can be understood by referring to fig. 3.5.23, which shows the director
patterns for two pairs of unlike defects, ( + f, |) and ( + 1 , - 1 ) , each in
two different situations. It is seen that there are significant differences in the
142
(-L+!)
(-1,+
c= 0
Fig. 3.5.23. The director patterns for (, ) and (1, 1) defect pairs in two
situations, c = 0 and c = n/2. The double-headed arrow at the centre indicates the
director orientation far away from the defect pairs. (After reference 98.)
\[(O2x-Ol
with
= s tan" 1
x-d
6X = and
ox
tan"1|
(3.5.19)
0 y = .
dy
The positive sign in (3.5.19) is taken for like singularities and the negative
sign for unlike singularities.
3.5 Disclinations
143
144
E=2nrT+nks2
In (R/r)
shear stress
velocity gradient
Lt
2dtj'
145
/?-Azoxyanisole
122 C
/7-Azoxyphenetole
144.4 C
Molecules
parallel
to flow
direction
Molecules
parallel to
velocity
gradient
Molecules
perpendicular
to flow direction
and to velocity
gradient
2.4 0.05
9.2 0.4
3.40.3
1.3 0.05
8.30.4
2.50.3
and
xy
yx
2 x, y
Since director gradients are neglected, the elastic part of the stress tensor
t% = 0. Thus, for n = (1,0,0) we have from (3.3.5)
or
Similarly
and
vz = 0.
Since the director lies in the xy plane n = (cos cp, sin (p, 0). If the diameter of
146
the tube is large enough, wall effects and director gradients may be
neglected. Therefore (3.3.2) may be written as
For Q > Q c ,
^j)
( ^ ^ | Q
) ^ - g ,
(3.6.4)
(3.6.5)
where
(3.6.6)
147
from (3.6.4), where V = nR2L is the volume of the cylinder. When Q < Q,c,
T = -Vkxa.
(3.6.7)
The torque increases linearly with the angular velocity and offers a direct
method of determining Ar When Q = Qc,
T = \VX*H\
(3.6.8)
2 tan w
l+tan>
2{QC/Q + (1 1 + {Qc/Q + (1 - Q2/Q2)^ tan [(Q2 - Q2)i/ -10]}2'
(3.6.10)
Above the critical angular velocity, the torque decreases with increasing
Q. The predictions are generally in agreement with observations(105) (fig.
3.6.1). However, the shape of the experimental curve at higher angular
velocities appears to be rather sensitive to the nature of the solid surface in
contact with the liquid crystal, showing that a complete theory has to take
into account boundary effects and the production and migration of
disclination walls at the surface.(106)
Another method of determining A1 is by studying the damped torsional
oscillations of the nematic suspended in a static magnetic field.(107) For
i 1, where D is the torsion constant of the wire and V the
148
1.0
0.5
Fig. 3.6.1. Variation of the torque as a function of the angular velocity of the
rotating magnetic field. Open circles: theoretical values; filled squares: experimental values for PAA, T= 112C, / / = 2 9 0 0 G (Tsvetkov(104)); crosses: experimental values for MBBA, T=24C, 77 = 2230 G; triangles: experimental
values for the same compound, T = 22 C, H = 2230 G, but using a solid teflon
cylinder immersed in the fluid to measure the torque. (After Gasparoux and
Prost.(105))
volume of the sample, the decay in the angular amplitude of the oscillations
can be expressed as
149
making an angle 6(r) with z. Thus we seek for the components of the
director and velocity fields the solutions
nr = sin 0(r),
vr = 0,
n9 = 0,
v(p = 0,
nz = cos 9{r).
vz = v{r).
If a magnetic field with components (Hr9 0, Hz) is applied, the external body
force
Gz = x^Hr sin 6 + Hz cos 6) Hz,
where / a is the diamagnetic anisotropy per unit volume. Substituting in
(3.3.1) and (3.3.2) we have in the steady state
2/d0
k
HzsinO) = 0,
ar b\
(3.6.11)
where
(3.6.13)
(3.6.14)
a = -dp/dz,
and ft is a constant. Equations (3.6.11) and (3.6.12) are also applicable to
flow through the annular space between two coaxial cylinders. For flow
through a capillary we put ft = 0 to avoid the singularity in (3.6.12) at
r = 0. The two equations can then be used to obtain the orientation and
velocity profiles. At high flow rates, the contributions of the elastic terms
tend to become small and
Xx + X2 cos 2 0 ^ 0
in the absence of a magnetic field. The director orientation then approaches
an asymptotic value given by
V ^
(3.6.15)
150
> 4.0
i
(c)
3.0
a,
2.0
0.001
1.0
0.1
AQInR (cm2 s1)
Fig. 3.6.2. Apparent viscosity rj for Poiseuille flow of PAA at 122 C (homeotropic
wall orientation) plotted against the ratio of the flow rate to the radius of the tube.
Open circles are experimental data of Fischer and Fredrickson,(110) (a) values
obtained from table I, (b) values obtained from table II with //1 = 0, (c) values
obtained from table II with ju = 0.038. (After Tseng, Silver and Finlayson.(111))
0.01
Table I
^ = 0.043 (g cm- 1 s-1)
H2 = -0.069
jus = -0.002
ju, = 0.068
AiB = 0.047
/z6 = -0.023
^ - ^ = ^ = -0.067
H - ^ = ; = 0.0705
Table II
ju1 = 0or -0.038 (g cm"1 s"1)
H% = -0.068
//3 = 0.000
/i4 = 0.068
Ai5 = 0.048
//6 = - 0 . 0 2 0
/ o -#i, = A, = - 0 . 0 6 8
attained in practice in relatively thick samples at high flow rates so that the
aligning effect of the walls has negligible influence.
The amount of fluid flowing per second
Jo
v(r)rdr
0.001
151
0.01
152
-0.2
-0.4
-0.6
-0.8
(a)
-1.0
-1.2
-1.4
-a/2
0
-0.2
-0.4
-0.6
-0.8
-1.0
-1.2
-1.4
-ir/2
0.4
0.5
0.6
0.7
r/R
0.8
0.9
1.0
Fig. 3.6.4. Orientation profile for Poiseuille flow for different shear rates. Wall
alignment homeotropic. The values of z a // 2 are 24.2 for (a) and 1.23 for (b). The
values of AQ/nR in (a) are (i) 0.003045, (ii) 0.03354, (iii) 0.1345, and in (b) they are
(i) 0.001245, (ii) 0.004167, (iii) 0.032438, and (iv) 0.1372. (After reference 112.)
9(R), the director orientation at the boundary. Thus imperfect alignment
or weak anchoring can be a serious source of experimental error in the
determination of rjam.
3.6.4 Shear flow
Consider now the steady laminar flow of a nematic fluid between two
parallel plates. If the flow is along x and the velocity gradient along y the
components of the velocity and the director are
v
x =
nx = cos 0(y),
ny = sin 0(y\
nz = 0,
153
l.O
0.9
0.8
0.7
g 0.5
0.4
0.3.
0.2
0.1
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
y/R
Fig. 3.6.5. Velocity profiles for Poiseuille flow; the curves (a) and (b) are for
different values oiAQ/nR but for the samefield,while for curves (b) and (c) 4 Q/nR
is nearly the same but the fields are different. The values of / a H2 and 4Q/nR
are respectively (a) 1.23, 0.001245, (b) 1.23, 0.004167, (c) 24.2, 0.003045 and
(d) xaH2 = oo or 4Q/nR = oo (truly parabolic). (After reference 112.)
where 6 is the angle made by the director with x. Proceeding as before, the
differential equations for the velocity and the director orientation in the
presence of a magnetic field (Hx9 Hy, 0) turn out to be(7)
-n+Tn\*7.) +(ay +
^1 + A2cos2g
W)
sin 0) = 0,
(3.6.17)
(3.6.18)
where f[6) and g(9) are defined by (3.6.13) and (3.6.14) respectively and
a = dp/dx is the pressure gradient and c the constant shear stress applied
to the fluid. If both plates are stationary a ==
| 0, and taking y = 0 half-way
between the plates, c = 0. On the other hand, if there is no pressure
154
gradient, and the flow is caused by one of the plates moving at a uniform
velocity in its own plane, a = 0 and c + 0. Equations (3.6.17) and (3.6.18)
can be solved to yield the apparent viscosity, velocity and orientation
profiles under different boundary conditions.(113)
If the plate separation is large enough, boundary effects and elastic terms
can justifiably be neglected in (3.6.17). For large shear rates and zero
magnetic field the director orientation approaches the value 0O, defined by
(3.6.15). But if a magnetic field of moderate strength is applied, the
orientation profile is modified slightly. Gahwiller(114) has studied this
behaviour by measuring the change in birefringence. He used capillaries
(5 cm long) of rectangular cross-section (4 mm x 0.3 mm) and measured
the rate offlowdue to a pressure gradient. If H is along the flow direction
and shear rates are large, we obtain from (3.6.17) and (3.6.18)
(/I1 + /l 2 cos2fl)^ = z a // 2 sin2fl.
(3.6.19)
(3.6.20)
where v0 is the velocity half-way between the plates and d the plate
separation (though, as emphasized by O'Neill(115), this assumption may
not be strictly valid in practical situations). From (3.6.19) and (3.6.20)
tan0*8^oj;//a//2</2.
(3.6.21)
The phase difference between the two perpendicularly polarized components when light is incident normal to the plates is then
fd/2
\
-d/2
where
[n(0)-no]dy,
where n0 and ne are the ordinary and extraordinary refractive indices. If the
magnetic field is applied along the velocity gradient,
2 2
(3.6.22)
JU3//I2, JJLJXB.
an(
155
1.5
1.0
0.7
IT
s 0.4
0.2
0.1
20
30
40
Temperature (C)
50
60
Fig. 3.6.6. The viscosity coefficients rj19 rj2 and rjs of MBBA as functions of
temperature. The temperature scale is linear in T~x. (After Gahwiller.(114))
At high magnetic fields the experiment reduces in effect to Miesowicz's
method except that Gahwiller extended it to arbitrary orientations of the
magnetic field. If 9 is the angle between the director and the flow direction
and cp that between the projection of the director on the yz plane and the
velocity gradient, then, neglecting secondary flow (see 3.6.5), one may
write approximately
rj(9, (p) = jux sin2 9 cos2 9 cos2 cp 1// 2 sin2 9 cos2 cp
+ IJUS COS2 6 +1// 4 4-1//5 sin2 6 cos2 cp +1// 6 cos2 6
= rjx cos2 6 + (rj2 + jux cos2 9) sin2 8 cos2 cp
By choosing 9 and (p appropriately, one can determine rj19 rj2, rjs and juv
Using these two sets of data, Gahwiller was able to determine all five
independent viscosity coefficients as well as / a . Some of his results are
presented in figs. 3.6.6 and 3.6.7.
156
0.1
0.06 -
0.03
80
100
Temperature (C)
60
120
Fig. 3.6.7. The viscosity coefficients rjx, rj2 and //3 of /?-n-hexyloxybenzylidene-//aminobenzonitrile (HBAB) as functions of temperature. The temperature scale is
linear in T~\ (After Gahwiller. (114))
+3 -
80
100
60
Temperature (C)
Fig. 3.6.8. The ratio // 3 /// 2 versus temperature for MBBA (crosses), HBAB (open
circles) and 1:1:1 molar mixture of HBAB, /?-n-butoxybenzylidene-//-aminobenzonitrile and ^-n-octanoyloxybenzylidene-^'-aminobenzonitrile (filled circles).
(After Gahwiller.(114))
40
157
158
90
Fig. 3.6.9. (a) Ratio of transverse to longitudinal pressure gradient versus director
orientation <j> in plane Poiseuille flow of nematic MBBA. Sample thickness
d = 200 jum. Length of cell L = 4 cm and lateral width / = 4 cm. (b) Deflection
y/ of flow lines with respect to the y axis versus <f> in wide cells (L/l ~ ^ ) . Full line
represents theoretical variation. (From Pieranski and Guyon.(120))
where
l
Oil ~ Vz) S m ^ C 0 S ^
159
Fig. 3.7.1. Experimental arrangement for studying the reflexion of ultrasonic shear
waves at a solid-nematic interface.
the central region of the cell the flow lines were deflected by an angle y/(<f>)
with respect to y, which again showed the expected dependence on cp (fig.
3.6.9(6)).
Though this experiment deals with a particularly simple situation, the
basic principle that it illustrates is relevant to the discussions that follow on
hydrodynamic instability (3.11).
160
Now the velocities associated with the incident, reflected and transmitted
waves may be written as
vt = Ai exp ( F s z) exp (icot),
vr = Ar exp (F s z) exp (icot),
vt = At exp ( - Tn z) exp (icot),
respectively, where F may be complex and the subscripts s and n stand for
solid and nematic. Therefore, in the nematic medium
tzx = -ZnAt
exp (-Tnz)
exp (icot),
F n = (1 -\-i)(pco/ju^
(3.7.3)
Z n = |(1 + i) (pcoju^y.
(3.7.4)
A
r
_ 111 _ ^ s
Ax
^n
Z s + Z n'
^s
l-Hexp(ifl)
l + |r|exp(i0)
i
. .12
, OII
/3-
(3.7.5)
Since according to (3.7.4) the real and imaginary parts of Z n are of equal
magnitude, it follows that
-^t
(3.7.6)
161
d28
8 ? *a
.
Sm
C0S
W
dt ~
'
where
The most general solution satisfying the boundary conditions 8 = 0 at
z
162
or
- 1] exp {- (2
Thus 0max(O attains the value 0(oo) with a time constant T(H) given by
H2).
(3.8.2)
If the field is now reduced to a value less than i/ c , the decay rate is still given
by the same expression only with a negative sign.
If the field is switched off from H > Hc to zero,
T-\0) = (kJX1)(n*/d*\
The twist viscosity can be determined from a measurement of T. (123)
Typically, T(0) for a film of 25 jum is about 10"1 s. This gives an idea of
the order of magnitude of the relaxation time for most nematic liquid
crystal devices.
3.8.2 Homeotropic to planar transition: backflow and kickback effects
The other two geometries used in the Freedericksz experiment are more
interesting as they result in a new effect, namely, hydrodynamic flow
induced by orientational deformation. This is the inverse of the more
familiar property of flow alignment that has been discussed at length in
previous sections.
Let us consider the homeotropic to planar transition (fig. 3.4.1 (c)). For
this geometry, n = (sin 0,0, cos 0), 0 = 0(z), v = vx(z), and vz(z) = 0. Setting
klx = fc33 = k, cos 0 1 and sin 0 0, we get from (3.3.2)
^
^
dz
^
dt
= 0
dz
where
and r = (l2-
(3.8.3)
163
d6
hx = K/^ + ^ - ^ - ^ + ^ Q p
neglecting squares and higher powers of 9. Therefore
(3.8.5)
(3.8.6)
i=
(3-8-7)
and
/ H\2
(HJ
le
p-tan, '
,.
o o
(18 8)
(3.8.9)
164
Fig. 3.8.1. Velocity profile in the homeotropic to planar transition (seefig.3.4.1 (c)).
The velocity has two components, one linear in z and the other oscillatory, the
wavelength A of the latter diminishing with increasing value of H.
Fig. 3.8.2. Torques acting on an elementary volume of the fluid when the molecules
are rotating with angular velocity D: (a) homeotropic to planar transition,
F a = |(2 1 A 2)Q; (b) planar to homeotropic transition, Tb = ^(<X1 + X2)Q. As a
rule, |FJ > | r j .
and
where Q is the angular velocity (fig. 3.8.2). In many nematic liquids, Ax and
X2 are of opposite signs and of comparable magnitude (see, e.g., legend of
165
30
25
20
r(0)
r(h)
15
10
0 a
10
15
20
fig. 3.6.2) so that F is small. In other words, A is small and the solutions of
(3.8.7) and (3.8.8) reduce to y/ = n/2 and
166
Fig. 3.8.4. (a) Velocity profiles at (i) 1.09 s, (ii) 1.744 s, (iii) 2.397 s, (iv) 2.943 s, (v)
4.469 s, (vi) 4.905 s and (vii) 8.61 s after the magnetic field is switched off from
H ^> Hc to zero in the Freedericksz experiment of fig. 3.4.1 (a). The values are
computed for a sample of MBBA of thickness 200 jum using the equations of Clark
and Leslie.(125) (b) Director orientation profiles at (i) 0 s, (ii) 0.109 s, (iii) 0.654 s, (iv)
3.161 s, (v) 6.648 s after switch off. (U. D. Kini, unpublished.)
167
-d/2
(a)
(b)
168
dfidcu
where X = 2nc/co0 is the vacuum wavelength and
(3.9.2)
where e = (e||+2e 1 )/3 is the mean dielectric constant (at optical frequencies), a = || e_L is the dielectric anisotropy assumed to be <^,
n{: = n i and nf = nf. An electric vector polarized along i induces a
displacement Df = eifEt along f. The director at r
n(r) = n0 + dn,
where, to a first approximation, <5n n0 = 0, since we assume the fluctuations
to be of small amplitude. Thus, neglecting density fluctuations (i.e.,
assuming s and a to be constant) we have from (3.9.2)
169
(c)
Fig. 3.9.1. (a) The two uncoupled modes Snx and Sn2; (b) components of the
deformation in the SnY mode, bend and splay; (c) components of the deformation
in the 3n2 mode, bend and twist.
where i0 = n0 i and / 0 = n0 f. Therefore
.(r,/)>.
(3.9.3)
ex = e2 x n0,
where qL is the component of q perpendicular to n0. Defining ia = e a i and
/a = e a f ( a = l , 2 ) ,
170
Hence
<Se2} = el [ (ijo +/aio)\6nJL09 0) SnJLr, t))].
(3.9.4)
(3.9.5)
where
Too
aq,w
j ^
na
q,
and
<Sa(q, /) = (snJLr, t) exp (iq r) dr.
(3.9.7)
(3.9.8)
where
where
171
Optic axis Q
Fig. 3.9.2. A typical experimental configuration used by Chatelain in his lightscattering studies: k and i are the wavevector and unit polarization vector of the
incident light, k' and f the corresponding quantities for the scattered light, the
suffixes e and o denote the extraordinary and ordinary polarizations with respect
to the optic axis of the nematic medium and qx is the wavevector change on
scattering.
From the equipartition theorem (which is certainly valid in the present
problem)
dQ~VlV ~W'
(3.9.10)
(3.9.11)
where /? is the isothermal compressibility. Taking p(ds/dp) ~ ea,
da'/da ~ 0kq2.
Typically P ~ 10"11 cm2 dyn"1, k ~ 10"6 dyn, q ~ 104 cm"1, so that
da'/da ~ 10"8.
172
150 r-
100
I
50
1
0
^
50
100
150
200
cot 2 (<p/2)
Fig. 3.9.3. Angular dependence of the intensity of scattering for PAA. Circles give
the experimental values of Chatelain and the line represents the theoretical
variation. (After de Gennes.(129))
Therefore
dcr
dQ
(3.9.12)
173
Fig. 3.9.3 compares this relation with the experimental data of Chatelain
and as can be seen the agreement is good. In principle a measurement of the
intensity of the scattering and its angular variation offers a method of
determining the elastic constants.
3.9.3 Eigenmodes and the frequency spectrum of the scattered light
For a given mode of vibration of angular frequency co and wavevector q,
we can write
<5n(r, t) = Sn exp [i(q r cot)],
Substituting in the basic equations of motion (3.3.1) and (3.3.2) we obtain
icopvk = &k-qk(qj^/q2\
(3.9.13)
0.
(3.9.14)
K) q\+(K - K
' '
'
174
Is =
) = 0,
6
(3.9.17)
which has two roots. Typically, k ~ 10~ dyn, p ~ 1 g cm" ,77 ~ // ^ O.li5,
2
C(q)
(3919)
cot -
ir/q2/p,
cojcot ~ pk/rj2 ^ 1.
175
the forces responsible for these tilts in the director, then in a first order
theory dn1=%1G19 ^ 2 = / 2 G 2 , where / 1 ? / 2 are susceptibilities; more
generally
We have seen that due to the thermal agitation, there are spontaneous
fluctuations in n whose mean square value is defined by
7a(q, co) = (SnJt - q, 0) Sna(q, co)}.
According to the fluctuation-dissipation theorem,(132) the relation between
h
and
X, is
(3.9.20)
+ k^q)] = G19
iC2(q)v2-\-Sn2[icol1-\-k2(q)] = G2.
Along with (3.9.16), this can be simplified to obtain
AaVM
' '
[pco-iPM][ttiO)+K(<i)]-c(q)QMv
or
7a(q, co) = y^
YT~T\
2\
>
(3.9.21)
where
um = -ico s o c
a n d uta = -icof(X.
(3.9.22)
176
200
150
100
50
10
20
30
40
Fig. 3.9.4. Angular dependence of the width of the Lorentzian spectral density for
mode 2 in PAA at 125 C. Open circles denote experimental values in the [ke,k^]
configuration and open squares the values in the [ko, k^] case. The curve is obtained
by a least squares fit with the theory. (After the Orsay Liquid Crystals Group.(131))
The light scattering is therefore determined entirely by the slow mode. The
integrated intensity
3.10 Electrohydrodynamics
177
(3.9.25)
The integrated intensity gives the elastic constants ku while the half width
yields uS(X. It is therefore possible to measure the viscosity coefficients from
an analysis of the scattered light using appropriate geometries.
As an example, we present in fig. 3.9.4 a convenient geometry for
isolating mode 2. The director is aligned parallel to the walls of the glass
plates. The incident beam is polarized parallel to the director and the
scattered beam perpendicular to it. If ne and no are the extraordinary and
ordinary indices of the liquid crystal, and cp the scattering angle,
ke = Innjk,
k'o =
Innjk,
178
Glass
LIQUID
CR VYSTAL
^> *
Spacers
Glass
Light beam
3.10 Electrohydrodynamics
179
180
Fig. 3.10.3. (a) Flow and (b) orientation patterns of Williams domains. The
periodic orientation pattern and the consequent refractive index variation has a
focussing action for light polarized in the plane of the paper. This gives rise to the
bright domain lines as indicated by the stars above and below the sample. (After
Penz.(143))
surfaces which are coated with a transparent conducting material. When a
DC or low frequency AC field is applied between the transparent
electrodes, there appears above a threshold voltage a regular set of parallel
striations perpendicular to the initial unperturbed orientation of the
director (fig. 3A0.2(a); see, however, 3.13.2). Dust particles are seen to
undergo periodic motion in the field of view proving that the domains are
due to hydrodynamic motion. The distortion of the director orientation
caused by this motion results in a focussing action for light polarized
parallel to the director(143) (fig. 3.10.3). This is responsible for the
appearance of a set of bright lines with a spacing approximately equal to
the film thickness when the microscope is focussed at the top surface. The
lines are shifted by about half the spacing when the focal plane is moved
down to the bottom surface. The pattern disappears when the light is
polarized perpendicular to the director.
The threshold voltage is usually a few volts and is practically independent
of the sample thickness. It is, however, strongly dependent on the
frequency(142) (fig. 3.10.4). There is a cut-off frequency coc above which the
domains do not appear, the value of coc increasing with the conductivity of
181
3.10 Electrohydrodynamics
400 -
2
"o
100 -
100 (Hz)
400
Frequency (Hz)
600
Fig. 3.10.4. Threshold voltage of the AC instabilities versus frequency for MBBA.
Sample thickness 100 jum. Region I: conducting regime (stationary Williams
domains); region II: dielectric regime ('chevrons')- Full line is the theoretical
curve. The cut-off frequency / c = 89 Hz. (After the Orsay Liquid Crystals
Group.(142))
the sample. Below coc, i.e., in the so-called conduction regime, the regular
Williams pattern becomes unstable at about twice the threshold voltage
and the medium goes over to the dynamic scattering mode. Above coc, in
the dielectric regime, another type of domain pattern is observed. Parallel
striations, again perpendicular to the initial orientation of the director but
with a much shorter spacing (a few microns), are formed in the midplane
of the sample. When the field is increased very slightly above the threshold,
the striations bend and move to form a chevron pattern (fig. 3.10.2(b)). In
this regime, the threshold is determined by a critical field strength rather
than a critical voltage. Both the threshold field strength and the spatial
182
150
Oscillating domains
100
B
vi
-a
Region of stability
50
_#
50
100
200
150
Frequency (Hz)
Fig. 3.10.5. Threshold voltage versus frequency for MBBA. Sample thickness
50/urn. Open circles: sinusoidal excitation; triangles: square wave excitation.
(After the Orsay Liquid Crystals Group. (142))
"o
o " o
No domains
/C7H15
H 15 C 7
0
1
10
100
1000
Frequency (Hz)
10000
Fig. 3.10.6. Frequency dependence of the threshold voltage in /?,//-di-n-heptoxyazobenzene, a nematic of positive dielectric anisotropy. (After Gruler and
Meier.(144))
3.10 Electrohydrodynamics
183
184
3.10 Electrohydrodynamics
185
EEX)/4n,
the positive and negative signs standing for the two sides of the plane under
consideration. The applied field Ez acting on these charges causes a flow
along the z axis; the resulting shear stress is evidently
4 = (e^cos2 (p + eLsm2 cp)(Ex- EEX) EJ An.
(3.10.1)
(3.10.2)
where g' is given by (3.3.13). This is the frictional torque exerted by the
molecules on the hydrodynamic motion. Clearly in the present geometry
(3.10.2) reduces to
^^t)
(3.10.3)
186
(3.10.5)
The elastic, dielectric and magnetic torques can be evaluated from the
functional derivative
where
JV
sir
F mag
\\
\\
JEZ)
3.10.3 DC excitation
In the DC case, we may set <p = 0 and the analysis becomes simple. Using
(3.10.1), (3.10.3) and (3.10.4),
3.10 Electrohydrodynamics
187
+ k3Sk2x)
(3.10.8)
(3.10.10)
where
888
"
'
We have treated the distortion as a pure bend but this is not exactly true.
Since the thickness of the sample is of the same order as the periodicity of
the distortion, the orientation of the director will vary in the z direction
also. There will therefore be a non-negligible splay component. To allow
for this, it has been suggested(147) that the bend elastic constant k33 should
be replaced by
where the wavevector in the z direction kz ~ n/d. With this correction, the
DC threshold given by (3.10.10) is in good agreement with the experimental
value. For MBBA it is about 8 V.
3.10.4 Square wave excitation
In the AC case, the time dependence of cp cannot be neglected. Using
(3.10.3), (3.10.5) and (3.10.6)
F
l
visc,y
A
H7l
\/p
\\
b
L\
188
(3.10.12)
where q is the excess charge per unit volume and Jx the electric current given
by
Jx = <7|| Ex + o^Ez(p
retaining only the first order terms in cp and Ex. We suppose that diffusion
currents make a negligible contribution. From the relation
= 0,
(3.10.14)
\ fin
(3.10.15)
where/, is the body force per unit volume, which in this case is equal to qEz.
We are interested only in the z component of this equation. Since the
director orientation q> is assumed to be a function of x only, txz vanishes,
and (3.10.15) reduces to
(3.10.17)
3.10 Electrohydrodynamics
189
Krj, I
ix1
- E2
Therefore, from (3.10.17) we obtain the following equation for the
curvature
(3.10.18)
y + yy/ + ?-Ez = 0,
where
(3.10.19)
T is the decay time for curvature, and rj is an effective viscosity coefficient
given by
Equations (3.10.14) and (3.10.18) are two coupled equations which cannot
be solved analytically for any general value oiEz(i) since y itself depends on
Ez and t. However for a square wave
J + ,
A )
\-E,
0<t<n/co
n/(D<t<2n/oj.
In this case y remains constant in any half-period and the solutions are
simpler and enable a physical interpretation of many of the observed
phenomena. If the solutions are taken in the form (147)
y/ = Cw exp (At/z\
q = Cq exp (At/x),
where
(3.10.20)
190
-<?Jiy/Ex/(\+A1),
a change of sign of E implies that q or y/ changes sign but not both, i.e., if
(q, y/) is a solution for a half-period, it will be (q, y/) or ( q, y/) for the
other half-period. Subject to these conditions we obtain two sets of
solutions given by (3.10.22) and (3.10.23) below:
-expH;
(3.10.22)
1exp
Eo^x
exp
-l-Ao
sx
-exp
A2t
2vx
exp
-\-Ax
where v = co/2n is the frequency of the voltage. Here q changes sign with
E, but y/ does not, which is the situation in the conduction regime.
+ expl ^
\
2 - ^ 1
2VT
(3.10.23)
sx Ax
1+exp
2vx
1
+ exp
^T. n
A2-Ax
2vx
exp
In this case y/ changes sign with E, but q does not. This corresponds to the
dielectric regime in which the director oscillates.
In the above equations, s is a real number which determines whether the
system is stable or not, and which takes the value 0 at the threshold. Setting
s = 0 and defining
3JO Electrohydrodynamics
191
\
\
\
\
\
\
j1
(a)
t
I
^
Fig. 3.10.8. Time dependence of the charge q and the curvature y/ over one period
of the square wave excitation, (a) Conduction regime (coz <4 1, T> z). The charges
oscillate but the domains are stationary, (b) Dielectric regime (coz ^> 1, T < r). The
charges are stationary and the domains oscillate. (After Smith et /.(147))
it can be easily shown that in the conduction regime
<3l0 24)
= U-l
sinh
TTA
'2COT'
(3.10.25)
These equations show clearly that the problem falls naturally into two
distinct parts. For T > z (3.10.25) has no solution and consequently there
is no dielectric regime, while for T < z (3.10.24) has no solution and there
is no conduction regime. For T = T, neither equation has a solution (except
when co = 0).
The theoretical variation of q and y/ over a full period of the exciting
wave is illustrated in fig. 3.10.8 for two values of co, one at low frequency
in the conduction regime (coz < 1, T > z) and the other at high frequency
192
in the dielectric regime (coz > \,T 4, z). In the former case one obtains
essentially stationary domains and oscillating charges and in the latter case
vice versa. The transition from the conduction to the dielectric regime
occurs at a critical frequency coc such that coc z ~ 1.
Certain other interesting conclusions can be drawn from the above
equations. For example, even at frequencies less than coc, there can be
quenching of the conductive instability at high fields. This is because
T = (AEl + AQ)" 1 decreases with increasing Ez9 and when it becomes equal
to T there can be restabilization. At higher fields, T < z and the system
goes over to the dielectric regime. This gives a physical insight into the
origin of the sigmoid shape of the experimental threshold curve. Indeed,
theoretically the threshold field as a function of frequency for conductive
instability may be expected to form a closed loop.
Another result that follows from the theory is that the conduction
regime can be suppressed altogether by using very thin samples. We have
seen that at low frequencies, kx ~ n/d. Now a decrease in sample thickness
increases Ao, where from (3.10.19)
This, in turn, reduces the curvature relaxation T and when T < z the
conductivity instability is eliminated and only a dielectric instability is
possible. A similar quenching can be achieved by applying a stabilizing
magnetic field.
Greubel and Wolff(150) and Vistin(151) observed that for thin and pure
samples the spatial periodicity of the pattern in DC excitation decreases
with increasing voltage above the threshold. Smith et <z/.(147) have suggested
that this variable grating mode(150) is due to non-linear (saturation) effects
at higher voltages in samples that are so thin and pure that the conductivity
instability has been quenched.
3.10.5 Sinusoidal excitation
The behaviour is qualitatively similar when the exciting field is sinusoidal.
Putting Ez = EM cos cot, the coupled equations for the charge and the
curvature become(146)
q + --\-aIly/EMcoscot
= 0,
coscot = 0,
(3.10.26)
(3.10.27)
3.10 Electrohydrodynamics
193
where
we have
i^ M 2
(cos ^ +C0T s m
0J
so that
(3.10.29)
Thus the threshold field increases with co. Though the analysis is not
strictly valid when coT ~ 1, calculations show that (3.10.29) holds good
quite well over the range 0 to coc where
COC = (C2-\)*/T,
(3.10.30)
194
where is the magnetic coherence length (see 3.4.1). For low magnetic
fields, > d, we get a voltage threshold independent of the thickness of the
sample:
__v_w_fv
(3.10.31)
where
47TII
j
V
33 '
HV>
O.10.32)
195
196
rv<o
(a)
Fig. 3.11.1. The mechanism for homogeneous instability: the flow is along y and
the velocity gradient along z. (a) An angular fluctuation 6 = nz > 0 results in a
viscous torque Yz > 0 such that a small twist (/> = ny > 0 is produced, (b) a
deformation <j> > 0 results in a torque Yy < 0 such that the initial deformation 0 is
enhanced.
The rigorous theory of HI threshold has been developed by Manneville
and Dubois-Violette(160) and by Leslie(161), but for the present purpose
it is enough to discuss the approximate treatment given by Pieranski
and Guyon(158) neglecting secondary flow. Taking n = [l,^(z),0(z)] 5
v = (0, Sz, 0), and balancing the viscous torque against the elastic and
magnetic torques, one obtains the coupled equations
(3.11.1)
(3.11.2)
Putting 6 = 61 cos (nz/d) and <j> = <t>x cos (nz/d), we get from the condition
for the compatibility of (3.11.1) and (3.11.2) the threshold shear rate for
H=0:
5 =
(3.11.3)
197
100
80
60
40
20
2
Frequency (Hz)
Fig. 3.11.2. Experimental roll instability threshold curves for MBBA as functions
of the applied voltage and frequency of shear for different values of the effective
plate velocity veU; d = 240 //m, Hx = 3200 G. (From Pieranski and Guyon.(159))
It is possible to define a dimensionless quantity called the Ericksen
number
Er = (Sd*/4) (fi2 jus / * u k22)i
which from (3.11.3) is seen to be n2/4 at the threshold.(162) The more
rigorous calculation, taking into account secondary flow, yields a critical
value of Er = 2.309 for MBBA at the threshold.(160) It follows from the
exact solutions that ET is not a universal number but varies from one
substance to another.
198
i
100
^ ^ ^
v'3
Y
>
| Stable
TR
.'
0.6
v3
Y
0.8
Frequency (Hz)
Fig. 3.11.3. Roll instability threshold curve for vett = 0.148 cm s 1 from fig. 3.11.2
showing the regimes Y,Z and TR (see text). (From Pieranski and Guyon.(159))
diffraction pattern vanishes when the light is polarized perpendicular to the
long axes of the molecules.
Another way of producing this pattern is to apply an AC shear. In this
case the RI appears even in the absence of a magnetic field (except, of
course, at very low frequencies, in which case it reduces, in effect, to DC
shear and HI is regained). Two distinct regimes can be identified by optical
observations. These are referred to as the Fand Z regimes. In the Y regime
the y component of the director ny changes sign at each half-period of the
AC shear but the z component nz does not, and vice versa in the Z regime.
Nematic MBBA was used in the experiment with a stabilizing magnetic
field Hx as well as a high frequency stabilizing electric field Ez (the latter is
stabilizing because a < 0 for MBBA). Sinusoidal shears of up to about
2 Hz were applied. The threshold curves for different values of the effective
velocity vett of the upper plate ( = total plate displacement per half-period)
as functions of the frequency of shear and of the applied voltage Vz are
shown in fig. 3.11.2. To the left of each curve is the domain of existence of
the RI. At larger frequencies rolls do not develop.
199
200
c = 45.70
o)Ty/2n
Fig. 3.11.5. Calculated roll instability curves fitted to the results of fig. 3.11.3 by
adjusting the values at zero frequency, at the cusp and at a point P to define the
frequency scale. (From Pieranski and Guyon.(159))
For a given <p there is the torque ju3 S</> due to the primary velocity (as in HI)
as well as a torque due to the secondary velocity gradient vZt x and together
the total viscous torque Yy brings about a distortion 6. In turn, a 6
distortion gives rise to a torque Tz that increases <f>. (We have already seen
in the case of HI that Tz = ju2 S6, but this value is now modified slightly
by a torque due to vy z.)
The complete calculation of the threshold for RI involves a solution
of the form exp \{qx x + qz z) with boundary conditions for 6, </>, vx9 vy,
and ^.<159'160'163> We present only the salient results of the theory.
Assuming fcn = k22 = fc33 = k, and ignoring material derivatives in (3.3.1),
3.11 Hydrodynamic
instabilities
201
09
0,
(3.11.5)
(3.11.6)
n*j
\J.Y
n n f i
Y.O)
(3.11.9)
which is the threshold condition for the Y regime. On the other hand, with
xy > TZ, Xy> T one gets in a similar fashion
ABS2cxyxz/2(l+co2x2)
= 1
(3.11.10)
= c/(l+x2),
xpy2,
y>\
(3.11.11)
(3.11.12)
202
(3.12.1)
203
(a)
T+AT
(b)
Fig. 3.12.1. (a) Thermal instability in a nematic heated from below. Initial
orientation of the director is horizontal (along y). A bend fluctuation causes 'heat
focussing' because of the anisotropy of thermal conductivity (K^ > K) and gives
rise to warmer ( + ) and colder ( ) regions. The warmer regions move up and the
colder regions down due to buoyancy effects and this, in turn, results in a velocity
vy (the fluid being assumed to be incompressible). The transverse velocity gradient
vz y induces a major destabilizing viscous torque, while the vertical gradient vy z
induces only a very weak stabilizing torque. The resulting torques, shown by the
curved arrows, are destabilizing. Here the long straight arrows represent the
translational velocities vz and vy9 the short straight arrows the heat fluxes, (b) The
same geometry as in (a) but with the top plate at a higher temperature. The system
is stable.
204
\
T+AT
(a)
T+AT
Fig. 3.12.2. (a) The initial orientation of the director is along z. A splay fluctuation
gives rise to warmer and colder regions. In this case, vyz causes a dominant
stabilizing viscous torque while vz y causes only a very weak destabilizing torque.
The system is therefore stable against stationary convection, (b) The same geometry
as in (a) but with the top plate at a higher temperature. The system is unstable (see
legend of fig. 3.12.1
regions. This is referred to as heat focussing. Because of the buoyancy
effect, the warmer regions move up and the cooler regions move down
creating a velocity fluctuation vz9 which in turn gives rise to a torque Tx
that stabilizes or destabilizes the orientation depending on the sign of
Ka. We seek solutions of the form n = [0, l,0(y,i)]9 \ = [0,0,vz(y,t)],
T=-pz+T'(y,t),
and using (3.1.6)(3.1.9) obtain for a variation
Qxp(iqyy) of the fluctuation(152)
exp (iqyy) [vz0 + (vjzv)
f
- ocgT'o] - (jujp) y/ = 0,
= 0,
= 0,
(3.12.2)
(3.12.3)
(3.12.4)
where y/ = d6/dy is the director curvature and 019 vz0, T'o are amplitudes of
fluctuations. In (3.12.4) T = ^x/kzzq^ is the relaxation time for the
director in the absence of any coupling. Ignoring vz0 in (3.12.2) and
3.13 Flexoelectricity
205
eliminating vz0 in (3.12.3) and (3.12.4) one finds the condition for threshold
to be
where r a = (K^q\XJii^f1. In an isotropic liquid Ka = 0,r a = oo, and we
recover (3.12.1). For a nematic film of thickness d ~ 1 mm, r ~ 103 s,
i a ^ 1 s, and ATe reduces to 10~3 times the isotropic value. It is the large
ratio T /r a that leads to a very low threshold. A magnetic field Hx
(stabilizing) or Hz (destabilizing) can be used to decrease or increase T^, or,
equivalently, increase or decrease A Tc; the critical temperature difference
varies linearly as H\ (or # z 2 ). (166)
Pieranski et al.ae7) then showed that when a homeotropically aligned
film of MBBA is heated from above, the orientation becomes unstable
above a critical value of AT of the same order as that for planar
orientation. They also verified that the film is stable when heated from
below. A model of the destabilizing mechanism is no longer purely onedimensional involving only vz (fig. 3.12.2). A thermal fluctuation causes a
fluctuation vz as before, but because of the incompressibility of the fluid
this, in turn, causes a velocity fluctuation vy that contributes the major
destabilizing torque Fx~ju2vyz
for Ka > 0. Again with a stabilizing
magnetic field Hz there is a linear relationship between A Tc and H\. A field
Hy favours rolls with axes normal to y, but in the field-free case the rolls
degenerate into a square pattern that may be regarded as a linear
superposition of crossed convection rolls. When AT is increased well
beyond A Tc a complex hexagonal structure is found with a nematicisotropic interface if the temperature of the upper plate is large enough.
The studies outlined here are the most important ones that established
the fundamental of principles of thermal instability in nematics. A number
of theoretical and experimental investigations on these and other geometries have since been reported. (155) A particularly interesting study is
that of Lekkerkerker(168) who predicted that a homeotropic nematic heated
from below (which, it will be recalled, is stable against stationary
convection) should become unstable with respect to oscillatory convection.
The phenomenon was demonstrated experimentally by Guyon et al.a69t 170)
3.13 Flexoelectricity
3.13.1 Determination of the flexoelectric coefficients
If the molecule possesses shape polarity in addition to a permanent electric
dipole moment then the possibility exists that a splay or bend deformation
206
(c)
id)
Fig. 3.13.1. Meyer's model of curvature electricity. The nematic medium composed
of polar molecules is non-polar in the undeformed state {{a) and (c)) but polar
under splay (b) or bend (d). (After Meyer.(171))
(b)
3.13 Flexoelectricity
207
Fig. 3.13.3. A hybrid aligned cell for the determination of the anisotropy of the
flexoelectric coefficients. In this geometry, the director has a splay-bend distortion
which gives rise to a flexoelectric polarization Px. On applying an electric field Ey,
the director is twisted by an angle $ oc (e1 e3) which can be measured optically.
(Dozov, Martinot-Lagarde and Durand. (174) )
208
<
-30
-50
Fig. 3.13.4. ^(0) versus is for two twin domains in the sample. The linearity of the
slope confirms the validity of (3.13.2). (Dozov, Martinot-Lagarde and Durand. (174) )
small, since the dielectric free energy is proportional to E2.) The twist angle
is maximum close to the bottom plate (fig. 3.13.3) where the curvature is
maximum and is given by
Ed
(3.13.2)
3.13 Flexoelectricity
209
(a)
(b)
210
d = 35 /^m
/ c = 120 Hz
Normal
rolls
50
Frequency (Hz)
90
3.13 Flexoelectricity
211
212
4
Cholesteric liquid crystals
214
effects may be neglected. The problem was investigated by Mauguin (1) with
a view to explaining the optical rotation produced by twisting a nematic
about an axis perpendicular to the preferred direction of the molecules. He
used the Poincare sphere(4) and ' rolling cone' method, but we shall adopt
an identically equivalent formalism, viz, the Jones calculus/ 5 ' 6)
The basic principle underlying the Jones method is that any elliptic
vibration can be represented by the column vector
215
Jones retardation matrix for the first layer referred to its own principal axes
is given by
0
G[exp(-iy)
[
0
exp(iy)
where y is half the phase difference between the waves linearly polarized
along the principal axes after passing through a single layer of thickness /?,
i.e., y = ndnp/k, Sn = na nb being the layer birefringence and I the
vacuum wavelength. The retardation matrix with respect to x9y is then
Jx = SGS"1,
where
and S"1 is the inverse of S so that SS"1 = S-1S = E, the unit matrix.
If Do is the complex column vector with respect to x, y representing the
incident light, the emergent light after passing through the first layer is
where, as we are interested at present only in the state of polarization
of the emergent beam, we neglect the phase factor exp( ir/), where
rj = n(na + nb)p/A. (Throughout, we follow the convention of representing
the phase factor at any point +z by exp( \2nnz/X).) Let Dx be now
incident on a second birefringent layer whose principal axes are inclined
at 2/? with respect to x,y. The Jones matrix for this layer is
and the emergent vector is
D2 = S2GS 2DX = S2GS 2SGS
= Sa(GS-1)aD0 = J 2 D 0 ,
where J2 = S2(GS~1)2 is the appropriate Jones matrix for this system of two
layers. In general, if we have a pile of m layers, where the principal axis of
the sth layer is inclined at sfi with respect to x, y (s = 1,2,..., m), the Jones
matrix for the pile is evidently
J m = Sm(GS~T = [l *] say.
(4.1.1)
(4.1.2)
216
Now, as stated earlier, the layer thickness is assumed to be very small, say
a few A, while the pitch P is taken to be at least a few wavelengths of light,
so that both ft ( = 2np/P) and y are small quantities. Therefore
(4.1.3)
From (4.1.1) and (4.1.2)
a = cos rap cos
_ tan/? .
_.
_ . sinra# .
,
rat/H -sin rap sinrat/ i sinycos(ra + l)p,
tan 6
sin 6
(4.1.4)
b=
tan/?
_.
_ .
_
_ .sinra# .
. ,
-cosmpsmmu smmpcosmu I- sinysin(ra + l)p,
tan 6
sin 6
(4.1.5)
c = b* and
d=a*,
fcos y/
[sm^
siny/ifcos/?
c o s ^ J Lsin/?
sin/?l
cos/? J
ip)
L
0 1 _ r
exp (i#>) J [ sin y/
_ r
cos^J
c = -6*
and
rf=fl*.
(4.1.9)
Equating the real and imaginary parts of (4.1.4) and (4.1.7) and of (4.1.5)
and (4.1.8), we obtain after simplification(10)
p = m(J3- 6') radians,
1
(4.1.10)
(4.1.11)
(4.1.12)
217
I/A2 (Mm-2)
Fig. 4.1.1. Rotatory dispersion of a cholesteric liquid crystal for wavelengths <^
pitch. Solution of poly-y-benzyl-L-glutamate (PBLG) in chloroform (18 g/100 g).
(After Robinson.(12)).
In these equations m represents the total number of layers in the system.
Since the layer thickness is taken to be a few A, it turns out in actual
practice that even with the thinnest specimens employed, m is usually a
very large number. We shall therefore assume m to be large throughout this
discussion.
Optical rotatory power
When /? > y, 0' 6. This condition is satisfied when \PSn <^ A, i.e., when
the pitch is not too large. The optical rotation produced by m layers is then
p = m(p-0) = m[fi
and the phase retardation 2(p 0. Thus the system behaves in effect as a
pure rotator. If m is the number of layers per turn of the helix, rnfi = 2n and
mp = P, so that the rotatory power in radians per unit length
p = -n(dn)2P/4l\
(4.1.13)
the negative sign indicating that the sense of the rotation is opposite to that
of the helical twist of the structure. (11) Typically, dn 0.05 for a cholesteric;
218
0.1
0.2
0.3
0.4
0.5
Fig. 4.1.2. Optical rotation per unit length in a twisted nematic film versus inverse
pitch for light of wavelength 0.5 jum. The incident linear polarization is parallel to
the director on the entrance side of the film. Layer birefringence Sn = 0.1. Film
thickness (a) 1.0 /urn, (b) 1.25 jum and (c) 1.5 jum.
taking P = 5 jum and X = 0.5 jum, p ~ 2000 mm" 1. This equation has been
verified experimentally in considerable detail by Robinson (12) in lyotropic
systems and by Cano and Chatelain (13) in thermotropic systems.
Robinson discovered that solutions of some polypeptides in organic
solvents, for example, poly-y-benzyl-L-glutamate (PBLG) in dioxan,
methylene chloride, chloroform etc., spontaneously adopted the cholesteric mesophase above a certain concentration. Under suitable conditions
the solutions exhibited equi- spaced alternate bright and dark lines when
observed through a microscope (see fig. 4.2.7). The lines may be interpreted
as a view of the structure at right angles to the screw axis, so that the
periodicity of the lines is equal to half the pitch. Robinson confirmed this
interpretation by observations between crossed polaroids and also by the
use of a quartz wedge; the retardation plotted against distance in a
direction perpendicular to the lines had an oscillating value, as is indeed to
be expected from the structure. The pitch for any given polypeptide
depended on factors such as concentration, solvent, temperature etc. When
viewed along the screw axis no lines were seen, but a very high optical
rotatory power was present. The rotation in every solution, with a very
wide range of values of P was found to be proportional to \/k2 (fig. 4.1.1).
Robinson substituted the observed values of p and P in (4.1.13), and
219
+ sin m6
sin ra/TI
cosra//?
Texp(-im(9)
L
0
\sin 9
Tcosra/?
~ [sinm/?
(4.1.14)
cosmfi J[
0
1
exp (im6)\
exp (imy)J'
(4.1.15)
Equation (4.1.15) implies that at any point in the medium there are two
linear vibrations polarized along the local principal axes. The polarization
directions of these two vibrations rotate with the principal axes as they
travel along the axis of twist and the phase difference between them is the
same as that in the untwisted medium. This result was first derived by
Mauguin(1) and is sometimes referred to as the adiabatic approximation. It
is this property that is made use of in the twisted nematic device discussed
in 3.4.2.
Fig. 4.1.2 illustrates the theoretical variation of the optical rotation with
220
pitch and sample thickness derived from (4.1.14) and (4.1.15) for very thin
cholesteric or twisted nematic films (the sample thickness being smaller
than the pitch). (1415) Here the rotation is defined as the quantity that is
most conveniently measured, viz, the angle between the incident plane of
polarization, assumed to be parallel to the director axis on the entrance
side of the film, and the major axis of the emergent ellipse. Unlike p given
by (4.1.10), this rotation is not independent of the azimuth of the plane of
polarization of the incident light/ 16 ' 17) but the general conclusions that we
may draw from these curves are still valid. We observe firstly that in
contrast to an ordinary optically active substance the optical rotation per
unit length in the twisted nematic (or cholesteric) is a function of the
thickness. A noteworthy feature is the reversal of the sign of rotation as the
pitch is varied.(17) In the neighbourhood of the Mauguin limit, the rotation
has the same sign as that of the helical structure. With decrease of P the
rotation drops to zero and then reverses sign; thus for lower values of P the
sense of the rotation is opposite to that of the helix, in conformity with
(4.1.13). With increase of sample thickness the peaks in the rotation
increase in height and become sharper. The predicted trends have been
confirmed experimentally.(18)
Absorbing systems: circular dichroism
When linearly dichroic dye molecules are dissolved in a cholesteric liquid
crystal the medium exhibits circular dichroism because of the helical
arrangement of the solute molecules in the structure.(19) The theory
developed above can be extended to take into account the effect of
absorption by treating the layers as both linearly birefringent and linearly
dichroic.(19'20) Assuming that the principal axes of linear birefringence and
linear dichroism are the same, the Jones matrix of any layer with reference
to its principal axes is
G
i *-v WJ
0 l[exp(-A: a /?)
p(iy)J|_
0
exp(iy)J
exp(-iy)
0
0
1
exp(-kbp)\
where ka, kb are the principal absorption coefficients of the layer, and
y=
y-lKK-
221
(4.1.16)
r = ^f-GS-'-AxA/ 1 .
7 E,
(4.1.17)
where
^ = exp (-<![) exp (i0),
A2 = exp(-)exp(-i/9),
and
cos^ = cos 7 cos )8.
Using (4.1.16) and (4.1.17)
J^exp(-^S-[^GS-^Sin(m-1)'El.
L sin 6
sin 9
J
(4.1.18)
where
_ [cos ^
[sin y/
- sin y/1
cos y/ J '
_ [cos/?
[sin/7
sin/7
cos/7
[ cosh a/2
i sinh a/2l
[ i sinh a/2 cosh a/2 J'
[
K =
[exp(-/c)
exp(fc)J'
Here p is the rotation, a the circular dichroism, 2#? the linear phase
retardation, 2K the linear dichroism, / the attenuation coefficient and y/ the
222
with
_x/sec2^
COS"1
r m0/
Vsec
(4.1.20)
X=
ICLL
tan<9
and
Hence
and
a = -myju/p.
Therefore, the linear dichroism of the layers not only results in circular
dichroism but also makes a small contribution to the optical rotation
which is opposite in sign to that due to linear birefringence. However, this
contribution is usually negligibly small. Another consequence of the theory
is that o changes sign whenever /? or ju changes sign. Further, the parameter
o exhibits a marked dependence on pitch and sample thickness. These
predictions are in qualitative agreement with observations.(18"20)
4.1.2 Propagation along the optic axis for wavelengths ~ pitch: analogy
with Darwin's dynamical theory of X-ray diffraction
When the wavelength is comparable to the pitch, the optical properties are
modified profoundly. Before discussing the rigorous electromagnetic
treatment of the problem it is instructive to examine it first from the
standpoint of X-ray diffraction theory/ 7 ' 21) Since the dynamical theory of
X-ray diffraction from perfect crystals and its applications are now quite
thoroughly understood, this approach may be useful in elucidating the
optical behaviour of cholesterics and in looking for new optical analogues
of certain well established X-ray effects. An example of a new phenomenon
is the Borrmann effect in cholesterics.(22)
223
(4.1.22)
where cps+1 = 2nnn(s+ \)p/X, p being the thickness of each layer. At the
boundary, the vibration emerges from a medium of refractive index
na and the rj vibration from a medium of refractive index nh. Qualitatively,
it is obvious that since the principal axes of the (s + 2)th layer are rotated
slightly with respect to those of the (s+ l)th layer, one of the components
of (4.1.22) will on emerging from the (s+ l)th layer meet a 'rarer' medium
while the other will meet a ' denser' medium. One component therefore gets
reflected without any change of phase and the other with a phase change of
n. Thus, in contrast to reflexion from a normal dielectric, the sense of
circular polarization remains the same after reflexion. Applying the
f Right- and left-circular polarizations are denned from the point of view of an observer
looking at the source of light. If the electric vector rotates clockwise with progress of time,
then it is right-circular. Thus, at any instant of time, the tip of the electric vector forms a
right-handed screw in space for right-circular polarization.
224
rf
where q = fidn/2n. We make the approximation here that sin/? /?, since ft
is assumed to very small (~ 10" 2 rad). On reflexion a very slight ellipticity
is introduced in the transmitted beam but we shall ignore it in the present
discussion. Transforming back to x,y the reflected wave on reaching the
surface of the liquid crystal will be
GI--UJ
(4.1.23)
,., | f
225
Sr+1
r+1
Fig. 4.1.3. Notation for the primary (T) and reflected (S) waves in the dynamical
theory.
(4.1.23). We can then write the difference equations in a simple manner
because, as stated earlier, circularly polarized waves travel practically
without change of form, so that the interference of the multiply reflected
waves with one another and with the primary wave can be evaluated
directly.
We shall suppose, as before, that the structure is right-handed and that
right-circular light is incident normally. Let Tr and Sr be the complex
amplitudes of the primary and reflected waves at a point just above the rth
plane, the topmost plane being designated by the serial number zero (fig.
4.1.3). Neglecting absorption, the difference equations may be written as
Sr = - i f i r r + e x p ( - i p ) S r + 1 ,
Tr+1 = exp ( - iq>) Tr - xQ exp ( - 2ip) S r+1 ,
(4.1.24)
(4.1.25)
where
Tr^ + Tr_x=yTr,
(4.1.26)
Sr+1 + Sr_1=ySr,
(4.1.27)
(4.1.28)
and
-i
v-2.._, ,(v-4)(v-3)
_
(4.1.29)
226
and
Tv.
(4.1.30)
(4.1.32)
(4.1.33)
(4.1.34)
227
1.0
0.8
0.6
:J
(a)
9S 0.4 -
0.2 -
8 o.o
o
'%
<S
^
1.0 -
ib)
0.8 0.6 _A 1
0.4 0.2 00
0.45
0.49
0.47
0.51
0.53
0.55
Wavelength (jum)
Fig. 4.1.4. Reflexion coefficient 0t at normal incidence versus wavelength for a nonabsorbing cholesteric: (a) semi-infinite medium, (b) film of thickness 25P, where P
is the pitch. Curves are derived from the dynamical theory; circles represent values
computed from the exact theory (4.1.3) assuming that the medium external to the
cholesteric (e.g., glass) has a refractive index 1.5. The parameters used in the
calculations are n = 1.5, Sn = 0.07, X0 = nP = 0.5 jum. (After reference 21.)
Substituting in (4.1.31) and simplifying
So_
-iQexpjie)
(4.1.36)
or
+ cf cothV
For the semi-infinite medium, v = oo in (4.1.36) and
So
ei(Q2-ey
To
When - Q < e < Q, is real and
01= = 1.
(4.1.37)
228
0.75
0.50
0.25
0
0.70
0.68
0.66
0.64
0.62
0.60
0.58
0.56
Wavelength (jum)
The reflexion is total within this range. The spectral width of total reflexion
is therefore A/I = Qkjn. Using (4.1.23),
A/I = Pdn,
(3)
(4.1.38)
229
xT
where
x = exp(-)exp(-i<p 0 ),
Inside the totally reflecting range, is real and the wave is strongly
attenuated. In X-ray diffraction theory, this phenomenon is referred to as
primary extinction. The extinction length, defined as the distance over
which the amplitude of the incident wave decreases to l/e of its value, is
P/Q at the centre of the reflexion band.
Outside the range of total reflexion, is imaginary and primary
extinction vanishes. Fig. 4.1.6 gives a plot of the wavevectors Kn
( = 2nnn/X) and KL ( = innJX) for right- and left-circular polarizations
respectively, as functions of the wavelength. The real part of Kn shows a
gap within the reflexion band - analogous to the familiar band gap in solid
state physics - while the imaginary part grows rapidly in the same region.
When e2 > Q2, ^ i(e2 Q2)* and may be positive or negative. The
optical rotation per pitch is clearly
n(5n)2P
niX X
7TT- +
5-
230
18.0 -
19.0 -
&
20.0 -
21.0?0.45
0.47
0.53
0.49
0.51
Wavelength (jim)
0.55
When Q2 > e2, i.e., within the region of total reflexion, p given by
(4.1.39) becomes complex, showing that the medium is now circularly
dichroic. The real part which represents the rotatory power is
n{Sn)2P
U2
'
Pk
(4.1.40)
'
*r
cosech v^
(4.1.41)
k 4- coth v
and
- 1.
Consequently the oscillations which appear in the reflexion curve should
0.45
0.49
0.51
Wavelength (jim)
0.47
231
0.53
0.55
Fig. 4.1.7. Rotatory power versus wavelength for a non-absorbing cholesteric: (a)
semi-infinite medium, (b) film of thickness 25P. Curves are derived from the
dynamical theory; circles represent values computed from the exact theory. , Sn
and Xo same as in fig. 4.1.4.(21)
also be seen in transmission and in circular dichroism. Equation (4.1.41)
may be expressed as
where
tan vy/ =
The optical rotation per thickness P is therefore
and
P
n{8nfP
yj-e
W~ ~^p~'
232
-30
-20
0.8
0.6
-10
0.4
0.2
0.0
10
20
0.4
0.5
0.6
0.7
0.8
Wavelength (//m)
0.9
1.0
Fig. 4.1.8. Experimental circular dichroism (open circles) and rotatory dispersion
(closed circles) of cholesteric cinnamate at 177 C. Sample thickness ~ 3 jum.
(After reference 26.)
already in 4.1.1 the rotatory power of a cholesteric liquid crystal, unlike
that of an ordinary optically active substance, is a function of the sample
thickness. Some measurements(26) of the optical rotation right through the
reflexion band using thin films are represented in fig. 4.1.8. The oscillations
in the theoretical curve for p appear to be smeared out, probably owing to
slight imperfections in the sample, but the trends are in agreement with
theory. There is also some evidence of subsidiary maxima in the circular
dichroism which again is to be expected theoretically.
Absorbing systems: the Borrmann effect
The Borrmann effect is the anomalous increase in the transmitted X-ray
intensity when a crystal is set for Bragg reflexion.(27) An analogous optical
effect in absorbing cholesteric media in the vicinity of the reflexion band
has been predicted and confirmed experimentally/ 22 ' 2829) The origin of the
effect can be readily understood by extending the dynamical theory to
include absorption. However, in contrast to the X-ray case, the polarization of the wave field and the linear dichroism play an essential part.
Suppose that the birefringent layers are also linearly dichroic and that
the principal axes of linear birefringence and linear dichroism are the same.
All the equations obtained for the non-absorbing medium hold good in
233
-1.0
0.45
0.47
0.49
0.53
0.51
0.55
Wavelength (jim)
Fig. 4.1.9. (<z) Reflexion coefficient 0t at normal incidence, (Z?) imaginary parts of Kn
and ATL, (c) rotatory power /? (the real part of />), plotted against the wavelength for
an absorbing semi-infinite medium calculated from the dynamical theory, K = 0.02,
SK = 0.028 and other parameters same as in fig. 4.1.4. (After reference 21.)
Q = nSn/n,
<Pn =
with
2nnnP
X
KR = innJX,
yi
yi
a na
2nnP
X
n(Sn)2P2
4/1 2
2nnP
n(Sn)2P2
4X2
KL = Innjk, etc.,
\rr
1K
a>
yt
b ~
yi
i -is
lK
b>
where Ka,Kb are the principal absorption coefficients. Fig. 4.1.9 gives the
reflexion coefficient 0t and the dependence of the real part of p and the
234
0.00 -
0.45
0.47
0.49
0.51
0.53
0.55
Wavelength (jjm)
Fig. 4.1.10. Transmission coefficients, 5"R and ^ L for right- and left-circular waves
calculated for a film of thickness 25P; (a) non-absorbing and (b) absorbing.
Parameters same as in fig. 4.1.9. The enhanced transmission for the right-circular
component in (b) is the analogue of the Borrmann effect. (After reference 21.)
imaginary parts of Kn and K^ on wavelength. Here Sn, n and P are taken
to be the same as for the non-absorbing case (see fig. 4.1.4) and in addition
it is assumed that K = \{Ka + Kb) = 0.02 and SK = Ka Kb = 0.028. The
interesting result is obtained that on the shorter wavelength side Im (K n) is
less than Im(^ L ), i.e., the right-circular component is less attenuated than
the left component, while on the longer wavelength side the opposite is
true. To observe this effect thin films have to be used.
For an absorbing film of thickness vP,
cosech v
235
-0.8
0.31
0.33
0.35
0.37
Wavelength (//m)
0.39
0.41
Fig. 4.1.11. Circular dichroism versus wavelength computed for different K and SK.
Sample thickness 25/>, n = 1.5, Sn = 0.07 and X0 = 036jiim. The absorption
coefficients were assumed to be Gaussian curves having a maximum at 0.36 jum and
of halfwidth 0.06 jum. The maximum values of K and SK are, respectively, as
follows: (a) 0.0125, 0.0157; (b) 0.0250, 0.0314; (c) 0.0375, 0.0471; (d) 0.0500,
0.0628; (e) 0.0625, 0.0785. (After reference 28.)
absorbing case 3T^ shows an enhanced value on the short wavelength side
of the reflexion band, which is the analogue of the Borrmann effect.
The phenomenon is shown up even more convincingly in the circular
dichroism curves (fig. 4.1.11). For a left-handed structure (i.e., negative)?),
yL exhibits an anomalous increase and for SK < 0, the peak transmission
occurs on the long wavelength side of the reflexion.
236
-0.4 0.2 -
0.1
0.0
-0.1 -0.2 -0.3
-0.4
0.32
0.34
0.36
Wavelength (//m)
0.38
Fig. 4.1.12. Experimental circular dichroism curves versus wavelength, (a) Pure
cholesteryl nonanoate (CN), (b) CN + 0.98 per cent by weight of PAA. Sample
thickness in both cases 6.5 jum. (After reference 28.)
The first experiments demonstrating the effect were conducted on
cholesteryl nonanoate in which was dissolved small quantities of PAA or
n-/7-methoxybenzylidene-/?-phenylazoaniline.(22'28) The temperature of the
system was adjusted so that the reflexion band overlapped with the
strongly linearly dichroic absorption band of the solute molecule. Under
these circumstances, the circular dichroism exhibits the features predicted
by theory (fig. 4.1.12). Similar measurements were reported subsequently
by Aronishidze et a/.(29) (fig. 4.1.13).
237
4.1.3 Exact solution of the wave equation for propagation along the optic
axis: Mauguin-Oseen-de Vries model
We next consider the exact solution of the wave equation for propagation
along the optic axis. The complete theory is contained in the papers of
Mauguin,(1) Oseen(2) and de Vries,(3) and has since been presented in
various forms by other authors. (30~3) We shall discuss an elegant treatment
of the theory developed by Kats, (30) and by Nityananda. (31)
We represent the dielectric tensor by a 'spiralling ellipsoid' whose
principal axis Oc is always parallel to z; the other two principal axes Oa
and Ob (with principal values sa and eb) spiral around z with a twist angle
q = 2n/P per unit length. If Oa, Ob are taken to be along x, y at the origin,
the tensor s at any point z may be expressed with respect to x, y as
fcos^z
[sin#z
sin#zir a
cos qz J L 0
[_ asin2#z
01 f cos^z
e j [ singz
singzl
cosgzj
ocsin2qz ,
e acoslqz
(AAA2)
e2E
=--jeE.
8?2
c
(4.1.43)
E" =
2~\Ex-xEy\
E' is right-circular and E" left-circular for propagation along -fz and vice
versa for propagation along z. Substituting in (4.1.43) we obtain
co2[
e
aexp(2ig.
,22 l ^ ^ /
-:^_\
c [aexp( 2\qz)
e
Ilr7//I'
(4.1.44)
[A'cxp[i(k-q)z]\'
which is a superposition of two waves of opposite circular polarizations
238
-0.5
\JLJLJLJ
I
500
600
A (nm)
700
Fig. 4.1.13. Circular dichroism in a mixture of 91.5% nematic ROTN 103 (of
Hoffman-La Roche) +7.5 % optically active L-menthol + 1 % dye/?-nitrobenzenebis(benzalazo)-/?'-dimethylaniline. Sample thickness 11.3/mi. Crosses represent
experimental values and the solid curve gives the theoretical variation. (After
Aronishidze et al.{29))
(k-qf-t
o?K* = 0,
kt, k, =
(4.1.47)
239
A'
A"
and
}
A"
A'
(4.1.48)
-qf-Kl')
When a is small, (4.1.47) and (4.1.48) give for the k2 solution A' I A" ~ a,
and for the k solution A" I A' ~ a. In other words, each normal wave is
made up of two oppositely polarized circular components with one of the
components generally dominating. The mixing of these two components
with wavevectors differing by 2q is a consequence of the Bragg reflexion.
Equation (4.1.45) may conveniently be rewritten as
U_ T
U
I
where
Kx = k1+q
exp(i^z)
exp(iK2z)
]'
= q + [K2m + q2-(4Klq*
2
+ a*K^]K
2
K2 = K ~ q = - q + [Kl + q + ( 4 ^ q + a K*)i\\
(4.1.51)
(4.1.52)
The fact that the wavevectors Kx and K2 are different is responsible for the
optical activity of the medium, the optical rotation per unit length being
p = K ^ ^g) rad. However, as emphasized in previous sections the phenomenon is not identical with natural optical activity because the normal
waves are not pure circular waves.
The de Vries equation
If now we make the approximation that (Kx K2)/q <^ 1, or that the
rotation per pitch is small compared with n, which is certainly valid in most
cholesterics,
P=
where
240
When x2 < a2K*, p becomes a complex quantity. The real part gives the
rotatory power and the imaginary part the circular dichroism. Since no
dissipative mechanism is built into the model it follows that the imaginary
part of p is associated with the reflexion of one of the components. The
reflexion band is centred at x = 0, i.e., at Km = q or Xo = nP where Xo is the
wavelength in vacuo. The range of reflexion extends from x = +aK2 to
x = OLK2, i.e., from + q\dn/ri) to -q\dn/ri). Since
3x = S(KJ2 = 2Km(SKm) ~ 2q(SKJ,
the spectral width of total reflexion is given by
AX = PSn
(4.1.54)
n{dnfP
(4.1.55)
This is known as the de Vries equation. The sign of the rotatory power
reverses on crossing the reflexion band (Ao). When X <^ Ao, (4.1.55) reduces
to (4.1.21), and when X > 20, p tends asymptotically to 0. The behaviour on
either side of the reflexion band has been confirmed experimentally(26) (fig.
4.1.8).
Thin films
(16)
Nityananda and Kini have applied the theory to obtain exact solutions
for reflexion and transmission by a plane parallel film bounded on either
side by an isotropic medium. The treatment allows for the contribution due
to reflexion at the cholesteric-isotropic interface. In general, for each
circular polarization at normal incidence the reflected and transmitted
waves consist of both circular polarizations. Four coefficients, two for
reflexion and two for transmission, are required to describe the problem
fully and the solution consists of matching the incoming and reflected
waves on one side of the slab with four waves within the slab (two in the
forward direction and two in the backward direction) and the transmitted
wave on the other side. An extension of the treatment to absorbing media
yields the theory of the Borrmann effect.(22)
241
Some calculations for the semi-infinite medium and for the thin film are
shown in figs. 4.1.4, 4.1.6 and 4.1.7. In these calculations, the isotropic
medium external to the liquid crystal is assumed to have a refractive index
equal to n = \{na + nb) so that the contribution of the ordinary Fresnel
reflexion coefficient at the cholesteric-isotropic interface is eliminated. It is
clear from these figures that the results of the exact theory differ only
slightly from those of the dynamical model, indicating that the latter is
probably adequate for most practical calculations. However, the simple
formulation of the dynamical model presented in 4.1.2 does have some
inherent limitations: (i) it is valid only for integral values of the pitch, (ii)
it is developed for small e (= 2U{X XQ)/X) and therefore does not give
exact values for wavelengths away from the reflexion band, and (iii) it fails
when the film thickness is very small, or when the extinction length is of the
order of a pitch. These limitations arise primarily because of the
kinematical approximation made for the reflexion from the m layers per
turn of the helix. We shall now show that when multiple reflexions within
the m layers are also included, the simple difference equations become
matrix difference equations and the resulting solutions turn out to be fully
equivalent to the exact electromagnetic treatment. Proofs of this result
have been given by Joly(32) and by Nityananda. (3134) We shall follow the
latter's treatment(34) which is simpler.
(+!)*-(-1)* T J'
where xa exp (ina Kp), K is the wavevector in vacuo and p the layer
thickness.(35) Exactly similar expressions may be written for the other
polarization also.
We define Is and Js as the amplitudes incident on the sih layer in + z and
242
s+\
Fig. 4.1.14. Notation for the incident and reflected waves. Is and Js are the
amplitudes of the waves incident on the 5th layer in the positive and negative
directions respectively, and Es and Ss the amplitudes emerging from the 5th layer in
these two directions.
z directions respectively, and similarly Es and Ss as the amplitudes
emerging from the sth layer in the + z and z directions respectively (fig.
4.1.14). Therefore, for the sth layer we have
a _
s ~
f
l l
ja
s
(4.1.56)
The second two lines can also be similarly combined. However, in what
follows we shall write them in the form
(4.1.57)
with the understanding that E, I, S and J* are each column vectors ( 2 x 1
matrices), and that r and t are each 2 x 2 diagonal matrices.
Now, the emergent wave E s in the + z direction from the ^sth layer is
physically the same as I s+1 , the wave incident on the (s+l)th layer, but
(4.1.57) will apply to the (s+ l)th layer only if I s+1 is referred to its own
243
principal axes which are rotated through /? with respect to those of the sth
layer. We therefore write
IS+1 = SES
(4.1.58)
and similarly
J^S-1^,
(4.1.59)
where
cosy?
sin/?l
sin/? cos/?J
Strictly these equations should include a phase factor allowing for the air
gaps between the layers, but the gap may be taken to be infinitesimally
small compared with the thickness of the layer, which itself tends to zero.
Using (4.1.58), (4.1.59) and (4.1.57)
(4.1.60)
(4.1.61)
M -3EMS-*
244
Wa + rb)\
[Sr/2 f \
i/0
0 "I
exp(iy9)J'
Since r and t are functions of the thickness p of the layer, we expand them
in powers of/?. It is sufficient to retain the first power in each case:
F=l(s-l)iKp,
dr/2 = \xaKp,
Uk + qf-eK*
-*K* 1
2
1 -a*
(k-q)*-eK'\-
6 )
Uk + qf[ -<xK2
(k-q)2-sK2\[F_\
'
which is precisely the same as (4.1.46). The dynamical theory applied in this
245
manner to a twisted pile of birefringent layers is then exactly valid for any
arbitrary thickness of the sample and for the entire range of wavelengths.
4.1.5 Oblique incidence
The theory of propagation inclined to the optic axis is, of course, very
much more complicated, and analytical solutions have not so far been
found.<36) The first attempt at solving the problem numerically was by
Taupin,(37) but the most complete calculations are those of Berreman and
Scheffer<38) who also carried out a precise experimental study of reflexion
from monodomain samples at oblique incidence. Fig. 4.1.15 presents their
observed reflexion spectra for two polarizations.
Berreman used a 4 x 4 matrix multiplication method. Assuming the
incident and reflected wavevectors to be in the xz plane, z being along the
helical axis of the cholesteric, the dependence on the y coordinate may be
ignored altogether. Writing Ex = Exexp [i(cot kx)] etc., it is easily verified
that Maxwell's equations can be reduced to the matrix form
.kceT
1-
\HV
CO
CO
-e
kcY
CO
CO 1
Ex
iHy
Ey
or
QZ
246
([
0.2 -
00
0.2
0.6 _
jV
r
I
First
order
1.0
0.6
ji^
Second
order
Observed spectra
I
1.4
Computed spectra
Fig. 4.1.15. First and second order reflexion spectra of a cholesteric liquid crystal
film (0.45:0.55 mole fraction mixture of 4/-bis(2-methylbutoxy)-azoxybenzene and
4,4'-di-n-hexoxyazoxybenzene) 15 pitch lengths or 11.47 jum thick. Angle of
incidence 45. Polarizer and analyser are parallel to the plane of reflexion for 0tn and
normal to it for 0to measurements. The small oscillations are interference fringes
from the two cholesteric-glass interfaces. (After Berreman and Scheffer.(38))
247
0>-1(z,-dz).
The results of their computations are shown in fig. 4.1.15. The agreement
with the experimental spectra can be seen to be good. There is a difference
in the intensities, which may conceivably be due to thin regions near the
surface with anomalous dielectric properties or due to the neglect of
absorption.
An important fact that emerged from this study is that the observed
features were best reproduced when the local dielectric ellipsoid was taken
to be a prolate spheroid, with the principal axis Oc parallel to z and
sc = s a. Thus the assumption that is generally made that the local
dielectric ellipsoid is uniaxial would appear to be valid to a very good
approximation as far as optical calculations are concerned (see, however,
4.10).
4.1.6 Propagation normal to the optic axis
When light is incident normal to the optic axis polarized diffraction
maxima are seen in transmission.(39) For the electric vector polarized
parallel to z the refractive index is independent of the z coordinate, whereas
for the vector polarized perpendicular to it the refractive index varies
periodically from na to nb. The wavefront having the latter polarization
therefore suffers changes of phase which vary along z with a periodicity
equal to half the pitch. There can also be changes of amplitude as parallel
rays tend to acquire a slight curvature when travelling in a medium in
which the gradient of refractive index is normal to the direction of
propagation. (40) This periodicity in the phase and amplitude gives rise to
polarized diffraction effects. Approximate expressions for the intensity of
the maxima may be derived by applying the Raman-Nath theory of the
diffraction of light by ultrasonic waves.(41)
248
0.3
0.2
0.1
-0.1
-0.2
-0.3
30
34
38
42
46
50
54
Temperature (C)
Fig. 4.1.16. Variation of inverse pitch with temperature in a 1.75:1 weight mixture
of right-handed cholesteryl chloride and left-handed cholesteryl myristate as
determined by laser diffraction. The mixture becomes nematic at 42 C. (Sackmann
et
Sackmann et al.(39) have investigated the temperature variation of the
pitch of a mixture of right-handed cholesteryl chloride and left-handed
cholesteryl myristate by this method. At a certain temperature (7^) there is
an exact compensation of the two opposite helical structures and the
sample becomes nematic. At this temperature only the central spot (zero
order) is observed, while at the other temperatures, polarized diffraction
maxima of higher order make their appearance. The inverse pitch varies
almost linearly with temperature passing through zero at 7^ (fig. 4.1.16).
4.2 Defects
4.2.1 %-disclinations
We now consider defect structures in the cholesteric liquid crystal. Treating
the cholesteric as a spontaneously twisted nematic,
nx = cos 0, ny = sin #,
6 = qz,
nz = 0,
q = 2n/P,
4.2 Defects
249
and V26> = 0.
disclinations
Such disclinations are closely analogous to nematic wedge disclinations
(3.5.1). The singular line is along the z axis (parallel to the twist axis) and
the director pattern is given by
nx = cos 9, ny = sin 6, nz =
0 = s tan" 1 (y/x) + qz
(4.2.3)
s = N/2,
where N is an integer. The presence of the disclination does not alter the
pitch. Fig. 4.2.1 illustrates the director patterns for disclinations of strength
s = 1 and 1. Many of the conclusions arrived at in 3.5.1 are valid in this
case as well, and in the one constant approximation the expressions for the
energies and interactions are the same as for nematic wedge disclinations.
As already indicated briefly in 3.5.8 the effect of elastic anisotropy has
some interesting implications for cholesterics, especially for long-pitched
structures. We have seen that disclination pairs in nematics have angular
forces in the presence of elastic anisotropy. For all practical purposes, the
solutions that were obtained for nematics will hold good for each nematiclike cholesteric layer, except that the layers now twist continuously in the
250
(a)
Fig. 4.2.2. A pair of like /-screw disclinations forming a stable double helix in a
cholesteric: (a) a pair of s = \ disclinations (after Cladis, White and Brinkmann(42)),
(b) a pair of s = 1 disclinations (after Rault(43)); see fig. 3.5.24.
medium. Therefore, for a pair of disclinations, angular forces will lock the
line joining the disclinations at the same orientation with respect to the
local director n (of the defect-free sample) in every layer. Hence while single
disclinations may be straight, pairs of disclinations in a cholesteric may be
expected to have a helical configuration (see fig. 3.5.24). For a pair of like
disclinations, mutual repulsion increases the separation between them, but
in the helical state this will be opposed by the line tension in each
disclination. In the end the two opposing processes should balance to result
in a stable double helix. This is indeed found to be the case experimentally
4.2 Defects
251
&
Fig. 4.2.3. A helical %(s = 1) disclination wound round a straight %(s = +1)
disclination in a cholesteric liquid crystal (Rault(44)); see fig. 3.5.24.
(fig. 4.2.2). On the other hand, if the individual disclinations of the pair
have different line tensions (as is the case for 1 and 1, both of which are
escaped structures) then the disclination with a lower tension should wind
around the one with the higher tension. This again appears to be in
conformity with experimental observations (fig. 4.2.3). The escaped
configurations for 1 and 1 in the cholesteric are rather complex,(45) which
may perhaps account for the fact that the two do not annihilate one
another.
252
-I
h- I- h- I-
h-
Fig. 4.2.4. The director pattern for s = \ /-edge disclination in a cholesteric. Dots
signify that the director is normal to the plane of the diagram, dashes that it is
parallel to and nails that it is tilted.
X~edge disclinations
In this case the singular line is perpendicular to the twist axis. On going
round this line, one gains or loses an integral number of half-pitches. The
director pattern around the /-edge disclination was first worked out by de
Gennes(46) who proposed a nematic twist disclination type of solution:
nx = cos 0,
ny = sin 0,
nz = 0,>
(4.2.4)
s = N/2.
The cholesteric pitch is altered around the singular line where TV is an
integer. The pattern for s = \ is shown in fig. 4.2.4. Again, the energies and
interactions in the one-constant approximation are the same as for nematic
twist disclinations. A somewhat more elaborate treatment of this model
has been presented by Scheffer(47) and the effect of elastic anisotropy has
been investigated by Caroli and Dubois-Violette.(48)
The Volterra process for creating these disclinations is the same as for
nematic disclinations. For the screw disclination the plane of cut is parallel
to the cholesteric twist axis while for the edge disclination it is perpendicular
to it.
4.2.2 Lattice disclinations
The cholesteric may also be regarded as having a layered structure with a
periodicity of P/2 along the z axis. This lattice can have disclinations just
as in smectic A and the Volterra process for creating them is also essentially
the same (see 5.4.4). If the cut is such that the line L is along the local
molecular axis, the disclinations so created are designated as X+ and A~,
{X standing for ' longitudinal' and the plus (or minus) sign indicating that
4.2 Defects
253
(b)
(a)
\ N
id)
+
Fig. 4.2.5. The configurations for (a) A", (6) A , (c) T" and (d) T+ disclinations in a
cholesteric. Dots, dashes and nails have the same significance as in fig. 4.2.4.
(a)
(d)
Fig. 4.2.6. Examples of pairing of X and z disclinations of opposite signs in a
cholesteric. Edge disclinations composed of (a) X' and A+, (b) T~ and T+, (C) T~ and
A+, (d) X~ and T+ and (e) pincement composed of z+ and T~.
254
where B = k22 q2 and K = |fc33 (see (4.6.22)). This expression first derived
by de Gennes is exactly of the same form as (5.3.3) for smectic A. We shall
now consider some applications of this model.(53)
Screw dislocations
The deformation around a screw dislocation may be written as
NP
u = -^Uur1(y/x)9
(4.2.5)
where PQ is the pitch and TV is an integer. The singular line is along the twist
axis. A circuit around the singular line results in a displacement of the
layer along the helical axis by an integral number of half-pitches. Equation
(4.2.5) can be recast to give the director orientation, and it is easily verified
that it becomes identical to (4.2.3) for the /-screw disclination and leads to
same results for the energies and interactions.
Edge dislocations
Let the layers be in the xy plane and the line of singularity along y.
Following the theory for smectic A (5.4.2), one may write,
p Y If 00 H/r
1
u(x,z) = - 1 + T-exp(-Aic a |z|+iK*)
where Po is the pitch of the undistorted
cholesteric and
X=
4.2 Defects
255
(a)
(b)
(iv)
256
4.2 Defects
257
(P0/4n)(3kS3/2k22)i. From this one gets the layer tilt 6 (with respect to the
z axis) and the layer dilatation S = du/dz as
)\
<4 2 6)
- -
0.6nkP
where Ec is the energy of the core and C the core radius, which in this model
can probably be assumed to be of the order of Po. Hence the energy turns
out to be finite and does not diverge logarithmically with sample size as
does a nematic-lilce solution of the form (4.2.4).
If there are two like dislocations, one at (0,0) and the other at (x0, z0) the
forces Fx and Fz (along and perpendicular to the layers) acting between
them are
kPl
258
The fact that the properties of the cholesteric liquid crystal are not
invariant with respect to reflexion introduces some additional complexity
into the equations of the continuum theory. The possibility now exists of
259
(4.3.1)
(4.3.3)
where Q is the heat supply per unit volume, qt the heat flux vector per unit
area per unit time and the other symbols have the usual meanings (see
3.3). If /?^ is the entropy flux vector per unit area per unit time, it is
convenient to introduce a vector cpi such that
+ nttNv-gtNt-PtTt-F-St-v^
> 0,
(4.3.5)
where F = U TS is the free energy function. Making use of the fact that
Ti9 dtj and nu can be chosen arbitrarily and independently, we obtain the
following equations:
with
<Pt = ^m^(Nk
+ dkpnp),
260
we have
~~^nk,p
(4.3.8)
Pt = 0 ,
(4.3.9)
t'iidii-g',N<-(p'i+%j!) T{ > 0.
(4.3.10)
(4.3.11)
Also
(4.3.12)
+ M7etpqTQnjnp
+ /i8eJpqTqntnp9
(4.3.13)
(4.3.14)
(4.3.15)
(4.3.16)
where // 15 ...,// 6 are the six viscosity coefficients already defined in the
continuum theory of the nematic state; ju7 and ju8 are two additional
viscosity coefficients coupling thermal and mechanical effects; A^1?...,AT4
are the coefficients of thermal conductivity.
The free energy of elastic deformation per unit volume is given by (3.2.7).
Martin, Parodi and Pershan(57) and Lubensky(58) developed a general
261
262
A _ 87*
+.
^ M* *)
_ 60
8/
80\
(4A)
^)-{x^-x^k>
(4A2)
where
Af -k
dz\
giving
CT
Jo
de
\-)idT-o
22
dz)
6z
263
and
^ Jo
22 \ Jo
Jo ^22
+ oc)
= [k22-k 2
+ oc)
=0.
One can then work out a simple solution if the material constants are
assumed to be independent of temperature. Let 9 = cot +/(z) and T = g(z);
(4.4.1) and (4.4.2) then reduce to
^ 2 2 0 - ^ +^
a2?
a?
oz
oz
= 0,
(4.4.3)
(4.4.5)
e = eo + ^ + [(*a-a)z/fc22],
(4.4.6)
(4.4.7)
Thus the director rotates about Oz with an angular velocity co, which
explains Lehmann's observations.
In the absence of a temperature gradient (4.4.6) reduces to
0 = 60 +
[(k2-a)z/k22]
264
Fig. 4.4.2. A spherical drop of a long-pitched cholesteric liquid crystal showing the
characteristic/-line (from Robinson(62)). The structure of this defect was explained
by Frank and Pryce(61).
and it is not known whether its contribution is of practical significance.
According to the Oseen-Zocher-Frank elasticity equations a = 0, and in
the absence of any evidence to the contrary it is generally neglected in most
discussions.
The Lehmann rotation phenomenon has never been reproduced since its
discovery. However, the experiment has been successfully repeated in the
author's laboratory (60) using a DC electric field instead of a thermal
gradient. Difficulties arising from anchoring effects at the boundaries were
eliminated by forming cholesteric drops suspended in the isotropic phase.
This was achieved in the following manner. The material was carefully
chosen to avoid other electrical effects on the orientation of the director.
A binary mixture of alkoxyphenyl-rras-alkylcyclohexyl carboxylates
(supplied by Merck) was prepared to give a room temperature nematic
with dielectric anisotropy 1. The mixture did not exhibit any
electrohydrodynamic instabilities up to DC voltages of ~ 8 V. Addition of
~ 5 wt % of cholesteryl chloride resulted in a left-handed cholesteric liquid
crystal with pitch P % 5//m, while ~ 10% of methylbutylbenzoyloxyheptyloxycinnamite yielded a right-handed cholesteric with the same P.
265
266
had an essentially planar structure with the helical axis normal to the flat
region and the /-line was confined to the periphery. Since the anchoring
energy at the cholesteric-isotropic interface may be expected to be
negligible, these flat drops proved to be most suitable for studying the
phenomenon. In principle, any transport current can produce a crosscoupling effect and hence a DC electricfieldwas used for convenience. At
2 V DC, the dark brushes emanating from the /-line became curved, and
then the whole structure started to rotate, apparently without any further
distortion. Fig. 4.4.3 shows the photographs of the rotating drops, which
can be seen to be closely similar to Lehmann's diagrams reproduced in fig.
4.4.1. Systematic observations on a number of drops established the
following results: (a) all the drops rotate in the same direction for a given
sense of the field: the right-handed helix has an anticlockwise rotation
when viewed along the field direction. When the voltage is reversed, the
curvature of the dark brushes and the sense of rotation of the structure are
reversed; (b) the angular velocity increases linearly with applied voltage up
to ~ 3.5 V, beyond which the structure of the drop changes and the
rotational velocity becomes a non-linear function of the applied voltage
(fig. 4.4.4); (c) nematic drops do not rotate under the action of E; (d) when
the handedness of the helix is reversed, the angular velocity also reverses
sign for any given sense of the field E; (e) the angular velocity does not
depend on the radius of the drop, showing that it is a rotation of the
structure rather than a rigid body rotation of the drop as a whole. All these
observations are in conformity with the theory. Though the angular
velocity was approximately the same for all drops, some drops which had
dust particles attached to them rotated with a lower velocity. For the sake
of completeness, even these values have been plotted in fig. 4.4.4. The
extrapolated angular velocity becomes zero for V 1.9 V (fig. 4.4.4). The
last point indicates that the DCfieldis totally screened up to 1.9 V and
that the redox potential of at least one of the components in the mixture is
about 1.9 V.
For a defect-free planar structure, the angular velocity in the presence of
an electric field E is in analogy with (4.4.7),
267
0.6
OX
0.5 A
J^
3
Voltage (V)
Fig. 4.4.4. The rotational velocity against applied voltage. The different symbols
denote measurements on different drops. The angular velocity was noticeably less
for drops which accidentally had dust particles attached to them (see the circles in
thefigure),but only the data for the fastest rotating drops were considered in the
calculations. Between 3.5 and 5 V there was visible disturbance within the drop and
measurements were not possible. At 5 V and above, the drop regained a uniform
texture. (After reference 60.)
by de Gennes.(64) Extending this theory to the case of the /-line of strength
2, it turns out that in the presence of the line defect rotating with the
structure (60)
Using the slope of the linear part of the |o>| against |E| curve (fig. 4.4.4),
and putting Xx = 0.7 P, it was found that |vE| = 0.28 cgs, and vE/q =
2 x 10~5 cgs for the material used in the experiment.
4.5 Flow properties
The flow properties of a cholesteric liquid crystal are surprisingly different
from those of a nematic. Its viscosity increases by about a million times as
the shear rate drops to a very low value(65) (fig. 4.5.1). One of the difficulties
in interpreting this highly non-Newtonian behaviour is the uncertainty in
the wall orientation which cannot be controlled as easily as in the nematic
case. Some careful measurements of the apparent viscosity ^ app in Poiseuille
flow have been made by Candau, Martinoty and Debeauvais (66) of a
268
105
Is
1O4
103
102
Cholesteric
Isotropic
lO"1
I
io- 2
100
110
i l l
120
^
1
130
140
Temperature (C)
269
0.30 -
0.28
0.26
0.24
fe)
0.22
0.25
0.50
0.75
Shear rate (103 s"1)
1.00
1.25
Fig. 4.5.2. Apparent viscosity in Poiseuille flow as a function of shear rate. Flow
normal to the helical axis (see text). Pitch (//m) = (a) 1.9, (b) 2.6, (c) 3, (d) 3.9, (e)
6, (/) 9.1 and (g) oo. (After Candau et a/.<66
1.5 V-
1.0
0.5
10
15
20
Fig. 4.5.3. Threshold of shear rate above which the fluid becomes non-Newtonian
plotted against 1/P2, where P is the undistorted value of the pitch. The arrows
represent the upper and lower limits of the shear rate (see fig. 4.5.2). (After Candau
et
66
270
where dp/dz is the pressure gradient. The quantity of fluid flowing per
second is
Q = _nR\dp/dz)
where R is the radius of the capillary. Applying Poiseuille's law,
/
/app =
(4.5.1)
271
We shall now show that the essential features of Helfrich's model can be
derived on the basis of the Ericksen-Leslie theory. (68)
Flow between parallel plates
We shall consider flow between two parallel plates, caused by a pressure
gradient. Choosing a right-handed cartesian system such that the plates
occupy x = h, we seek solutions of the form
nx = cos (qz + (p) cos 0, ny = sin (qz + q>) cos #,
nz = sin 0,
vx = 0,
vy = 0,
vz = w,
ny S+cpC,
nz = 6,
where C = cos qz and S = sin qz. Neglecting director inertia and product
terms involving w6, w X6 i9 6 XX9, w X6 x etc., (4.3.4) reduces to
ezx(k11-k22s2)-(pxx(k11s+k33sc2)-lp^k22s-d^(k!>3+^2)qsc
-2<p ^k^qC-^wqS+yiC-Sg?)
= 0, (4.5.2)
2S2)q]
= 0,
(4.5.3)
0.
(4.5.4)
P,z =
(4.5.5)
(4.5.6)
S^ik^-k^qS-cp^ik^S^k^C^q-y^k^q
-^, I (A; 22 + fc33)^C + > ^
+ ^6-/u2)C2].
(4.5.7)
2 2 + k33)qC-A,
^ - p . , = 0.
Wq
= 0,
(4.5.8)
(4.5.9)
272
(4.5.10)
w(x)dx = 2
w(x)dx
3Q
3[1
Taking 2h = 100 jum and P= 1 jum, the velocity attains 0.99 of the
maximum value within a thickness of 0.5 //m of the boundary. Thus in all
practical situations the velocity profile is flat over most of the region
between the plates and
*
a p p
- ^
(4.5.14)
v = 0,
vz = w(r).
273
(4.5.15)
IWr
0.
(4.5.16)
~ffIQ(KRY
and
where R is the radius of the capillary. The quantity of liquid crystal flowing
per second
KI0{KR)\
where Ir is the modified first order Bessel function of the first kind.
Hence
/aPP
8{ x
[2ii(KR)/KRI0(KR)]}'
Again, in practical situations the velocity profile is flat except very near the
boundary and
which is identical with (4.5.1). Thus Helfrich's idea of permeation along the
helical axis of a cholesteric can be justified in terms of the Ericksen-Leslie
equations.
274
The general theory of shear flow normal to the helical axis has been
discussed by Leslie.(69) An interesting feature that comes out of this
analysis is that a shear in the xz plane can give rise to secondary flow along
y. (See 3.6.5; in principle secondary flow should occur in the Helfrich
configuration also, but only in a negligibly thin layer very near the
boundary where the velocity profile is not flat.) We shall now present a
simplified version of Leslie's theory ignoring thermomechanical coupling.
Consider a cholesteric film between two plane parallel plates, one of
which is moving with constant velocity V in its own plane. The plates
occupy the planes z = h. We examine solutions of the form
nz = cosOcosp,
vx = u(z),
vz = 0,
(4.5.17)
(4.5.18)
where 6 = 6{z) and (p = cp{z). Then, tzx = a (constant shear), tzy = 0 and
tzz = p (an arbitrary constant). Using (4.3.7) and (4.3.13), we get
(H1 + H2 cos 2 cp) + H2 rj sin cp cos cp = a,
(4.5.19)
(4.5.20)
where
2 = dw/dz,
and
2r] = dv/dz,
XdFJddY
2 dO \dzj
\dF2(d(p\2
2 d6 \dz)
Ay
dz
(4.5.21)
and
i ^ ^ 2 uifl 1/COS (/~
(4.5.22)
From the symmetry of the problem it is clear that 6 and v should be even
275
(4.5.23)
+ HJ,
(4.5.24)
P <Jdz, r = - 2
77 dz.
(4.5.25)
1 Uij
I KXIS 1
1 VJ.J.
I \J.$
\dz/
d^?
2
dz
= 0 (4.5.26)
and
2
dz2
d# dz dz
2 sin#cos#-;al
dz
= 0,
(4.5.27)
where
Q = ( ^ _|_ ^ 2 cos20)/(/f 1 + /f2)
and
P = (A 2 -.^ s i n ^ c o s ^ / ^ .
Leslie assumed the following boundary conditions:
fd<p\
dz)h
/ d ^ fc2 }
\dz)_h
(4.5.28)
k22
Now
V=4\
Jo
_ a
~ K/2A
2
(4.5.29)
276
1.2
l.l
1000
100
Fig. 4.5.5. Theoretical variation of the apparent viscosity //app with pitch P = 2n/q
for flow normal to the helical axis of a cholesteric (or twisted nematic) at low shear
rates. Plot ofrjapp(q)/rjapp(0) versus P for twisted PAA. The separation between the
plates = 100 jum. The horizontal dashed line corresponds to f] (co)/rfapp(0). (After
ref. 70.)
3.7
3.3
2.9
T
2.5
'50
51
52
T(C)
53
54
fields
277
(7o,7i)
i s
It is seen that for pitch values of the order of the sample thickness, ?/app
should exhibit oscillatory behaviour with varying pitch because of the term
sin qh/qh in (4.5.30). A representative theoretical curve is presented in Fig.
4.5.5. This prediction has been verified qualitatively. (72) Measurements of
the apparent viscosity on the cholesteryl chloride-cholesteryl myristate
mixture, whose pitch, as we have seen earlier, is sensitive to temperature
(fig. 4.1.16), showed evidence of oscillations as the temperature was varied
(fig. 4.5.6).
Often, in practical cases, qh is so large that sin qh/qh is negligibly small
and ?/app approaches a maximum limiting value which is independent of the
pitch or gap width. This value is several orders of magnitude less than that
for flow along the helical axis and is comparable to that for a nematic.
When a is sufficiently large
q> = 0
and
0 = 00 = i c o s - 1 ^ / ^ ) ,
which is again independent of the pitch or gap width. This lower limit is
reached when a & k22 q2 or more. All these predictions are in qualitative
agreement with the observations of Candau et #/.(66) (see fig. 4.5.2).
4.6 Distortions of the structure by external fields
4.6.1 Magnetic field normal to the helical axis: the cholesteric-nematic
transition
When a magnetic field is applied at right angles to the helical axis of an
unbounded cholesteric liquid crystal composed of molecules of positive
diamagnetic anisotropy (xa = X\\~X > 0) the structure gets distorted as
illustrated schematically in fig. 4.6.1. As the field strength approaches a
certain critical value Hc the pitch increases logarithmically; for H > Hc the
helix is destroyed completely and the structure becomes nematic. (73) The
dependence of the pitch on the field strength was calculated by de
Gennes(74) and by Meyer.(75)
278
>H
>H
H=0
0<H<Hc
H>HC
Fig. 4.6.1. Schematic representation of the effect of a magnetic fieldapplied normal
to the helical axis of a cholesteric liquid crystal composed of molecules of positive
diamagnetic anisotropy. For H > Hc the cholesteric is transformed into a nematic.
Taking the helical axis to be along z, H = (H, 0,0) and n = (cos cp, sin q>,
0), the free energy of the system is
= ((Fdz = 1 J[(^f-
-/aH2
sin2 J d p + constant,
* *
which yields
COS 9 =
z = 2AZK(A),
where
22/
and
72
K(A) =
is the complete elliptic integral of the first kind; A is a constant which can
279
1.5
1.0
H/Hc
Fig. 4.6.2. Dependence of the pitch P on the magnetic field strength H in PAA
mixed with a small quantity of cholesteryl acetate. Curve represents the theoretical
variation predicted by de Gennes's equation (4.6.1). (After Meyer.(76))
where
=2
rnr.
Jo
4
?)2 dcp =
A
and
E(A) =
is the elliptic integral of the second kind. The condition
to the relations
= 0 leads
280
V///////////////////////A
W7/7///////////////////.
2*
Fig. 4.6.3. Deformation of a planar structure due to a magnetic field acting along
the helical axis of cholesteric liquid crystal composed of molecules of positive
diamagnetic anisotropy. A similar deformation superposed in an orthogonal
direction results in the square-grid pattern (see fig. 4.6.4). (Helfrich. (79))
2E(A)'
When
(4.6.2)
Hc = Inqo(k22/XJK
(4.6.3)
fields
281
4.6.2 Magnetic field along the helical axis: the square grid pattern
We next examine the effect of a magnetic field acting along the helical axis
of a cholesteric film having a planar texture. If / a > 0 and boundary
constraints are absent, there is a possibility of a 90 rotation of the helical
axis because \{X\\+X) > X- I o n t n e other hand, boundary effects are
such as to maintain the orientation of the helix, an expected type of
deformation is for the director at every point to be tilted towards the field,
i.e., a conical distortion/ 75 ' 78) However, it was pointed out by Helfrich(79)
that yet another type of deformation can set in, viz, a corrugation of the
layers (fig. 4.6.3). This has since been confirmed experimentally using both
magnetic(80) and electric fields/81'82) and takes place at a much lower
threshold. It results in the so-called square grid pattern (fig. 4.6.4), the
theory of which was first proposed by Helfrich(79) and subsequently
elaborated by Hurault. (84) We shall discuss first the magnetic field case.
For the unperturbed cholesteric
n = (cos# 0 z,sin# 0 z,0),
where we take z normal to the film. For small perturbations we may put
nx = cos(q0z + (p) cosq 0 z cpsinq0z,
ny = sin (q0 z + (p) sin q0 z + q> cos q0 z,
(4.6.4)
nz = 9 cos qoz.
Substituting in (4.2.1) the local energy density
M\
?>d(
d<p\
^ + smqozcosqozJ
(4.6.5)
\FdV.
Now
-^-=
\AdV
and
^=
\BdV,
(4.6.6)
282
Fig. 4.6.4. The square grid pattern in a cholesteric liquid crystal induced by (a) a
magnetic field and (b) an electric field. (Rondelez.(83))
(ffm
fields
283
dQ
dx
(4A7)
and
8>
Aoz
c)2Q
rvn wS 2 ^ 0 Zr.
oz
(4.6.8)
(4.6.9)
(4.6.10)
/c2 ^ = o,
(4.6.12)
where we make a 'coarse-grained' approximation, i.e., take into consideration only the slowly varying parts of A. The minimum energy density
O 00 ~ K <Po) (*11 + ^
( 4 - 6 - 13 )
284
(4.6.14)
and
ffo ^o ~ \i(ic
+L. \ + k ^\(k
+k } -
(k
k ^*1* ~ 1
(4.6.16)
where P is the local value of the pitch. We know that the twist energy of
deformation of the cholesteric structure can be written in terms of the pitch
as \k22(q q0)2 where q = 2n/P and q0 = 2n/P0. Comparing this with the
first term on the right-hand side of (4.6.18), it is clear that
B*k22q*.
(4.6.19)
In order to evaluate K, one may use the following argument. Suppose that
fields
285
the film is rolled up into a cylinder, i.e., the screw axes are oriented radially
about the cylinder axis and the layers form coaxial cylinders. In cylindrical
coordinates, the components of the director are now
nr = 0,
and
nz = sin 6{r)
1f
cos 4 0
^-^^-
(4-6.20)
The optimum value of 6{r) compatible with the periodicity 6{r) = 6{r + Po)
corresponds to
d0
1 . .
.
- = # n + - s m 0 c o s 0 .
dr
r
Averaging over cos4 0,
K = \k33,
(4-6.21)
so that
(4.6.23)
()V
(4-6.24)
(4.6.25)
286
excites the bend mode while kx -> 0 excites the twist mode whose amplitude
diverges in the limit. The optimum wavevector corresponds to an
admixture of both modes and is given by
8
2 / 2
tA<LK\
kx = ^ q\ k\,
7 4
^ 2 2
(4.6.26)
or
kxK(Pod)-K
(4.6.27)
H^ = (6k 22 k 33 ) 2 q 0 k z ,
(4.6.28)
77HocOP0 </)-".
(4.6.29)
/a
or
It is interesting to note that this threshold field is lower than that for a
conical distortion or that for cholesteric-nematic (unwinding) transition.
For a conical distortion, the theory is closely similar to that discussed in
3.4.2 and has been treated by Leslie (78); the critical field is given by
HI = (k, J
/a
^ 2 2 V/0>
Aa
fields
287
(4.6.30)
which should be zero (V-J = 0). Here Eo is the applied field and Ex that
caused by the Carr-Helfrich mechanism. Therefore
OX
(Tl
, c 4v
( E) =
An
(4.6.32)
An e.
(4.6.33)
U^
An -\E0
<+n alh ox
/c ' /diel-
~~
_87T 3 <7 | | +a 1
^2^22 ^ 3 3 /
V^0u/
)-\
(4.6.34)
where crn and aL are the conductivities parallel and perpendicular to the
288
preferred molecular direction and el{ and e are similarly defined. It may be
noted here that we have neglected the contribution of the viscous torque
altogether. This is because, as remarked earlier, fluid motion takes place
only by permeation, and, moreover, the distortions are infinitesimal and of
such long wavelength {kx <4 q0) that the effect of shear flow will be very
small indeed. The dependence of the spatial periodicity of the pattern and
the threshold field on d and Po is similar to the magnetic field case, and is
borne out by experiments/ 81 ' 8285)
Hurault(84) has extended the treatment to AC fields. The method is
somewhat analogous to the one discussed in 3.10 for nematic instabilities
and leads, in the conduction regime, to the following threshold for
distortion:
_8n3e]]+s
1+Q) 2 T 2
^(PdY1
(4 6 35)
where
289
where P(x) = 7Va(x) F(x) is the polarization (or the dipole moment per
unit volume) at x, a is the polarizability of the molecule at x, G(x x', k0)
a tensor representing the field at x due to a dipole at x', k0 the wavevector
of the incident radiation, v the volume of the Lorentz cavity which is not
supposed to contribute to the effective field and V the total volume.
290
35
40
35
30
25
I 20
o
"S3
60.5
15
60
65
70
75
80
85
90
95
100
105
63.5
110
115
Temperature (C)
Fig. 4.7.1. Anomalous optical rotation in the isotropic phase of a cholesteric liquid
crystal. Open and closed circles are measurements on two different samples
appropriately normalized. Cholesteric-isotropic transition temperature 60.57 C.
X = 0.6328 /an. (After Cheng and Meyer.(88))
Writing
) + <SP(x).
f dV<<5a(x)-G-<Ja(x')>-F(x'),
J V
J.
+ q) <s*(q) s(q)>
291
8 =
-iA;;
L-iAL
S+A;,
-iA;' 2
iA;; .
e + A^
or an optical rotation
The rotation exists because of the correlations (s%asyfiy which are nonvanishing in the case of a cholesteric. The averages may be evaluated on
the basis of de Gennes's model(89) (see 2.5). To allow for the noncentrosymmetric ordering in the cholesteric, we include an additional term
of the form s V x s in the free energy of the isotropic phase
/? c = /5 N + 2 t f 0 L ' ^ ^ - i f e ,
oxy
(4.7.1)
where FIN is the free energy per unit volume in the isotropic phase of a
nematic given by (2.5.15) and (2.5.22), q0 is a pseudo-scalar and L is a
constant. The average can then be worked out. For example
292
BPII
Disordered
-
BPIII
* = = --
BPI
Helicoidal
Inverse pitch
Fig. 4.8.1. Schematic phase diagram, showing the three experimentally observed
blue phases (BP).
where
A = a(T-T*),
l\ = LJA,
The blue phases occur in cholesteric systems of sufficiently low pitch, less
than about 5000 A. They exist over a narrow temperature range, usually
~ 1 C, between the cholesteric liquid crystal phase and the isotropic liquid
phase (see (1.3.5)). The first observation of a blue phase was described by
Reinitzer(90) himself in his historic letter to Lehmann as follows: 'On
cooling (the liquid phase of cholesteryl benzoate) a violet and blue
phenomenon appears, which then quickly disappears leaving the substance
cloudy but still liquid.' Although Lehmann (91) recognized it as a stable
phase, not until the 1970s was it generally accepted that the blue phases are
thermodynamically distinct phases. The nature of these phases has now
become a subject of considerable interest to condensed matter physicists.
Fig. 4.8.2. (a) Faceted single crystals of BP I in equilibrium with the isotropic phase
(58 wt% mixture of cyano-4-methylbutylbiphenyl (CB15) in nematic ZLI 1840 of
Merck). (From Cladis, Pieranskio and Joanicot(95).) (b) Optical Kossel diagram of
BP I, {110} direction, X = 5290 A (42.5 wt% mixture of CB15 in nematic E9 of
BDH). (From Cladis, Garel and Pieranski.(96))
293
(a)
294
Fig. 4.8.3. Unit cells of BP disclination lattices. O2 is simple cubic, O5, O 8 + and
O8 are body-centred cubic. The tubes represent disclination lines whose cores are
supposed to be isotropic (liquid) material. (From Berreman.(98))
Three distinct blue phases have been identified: BP I, BP II and BP III,
occurring in that order with increasing temperature (fig. 4.8.1). All of them
are optically active but isotropic. (They may have colours other than blue,
but are still referred to as blue phases). From observations of optical Bragg
reflexions(92) and other studies, it is found that BP I is a body-centred cubic
lattice (crystallographic space group I4X32 or O8), BP II a simple cubic
lattice (P4232 or O2) and BP III probably amorphous. The suggestion has
been made that BP III, called the blue fog, may be quasi-crystalline.(93)
295
Fig. 4.8.4. A representative phase diagram calculated from the Landau theory.(110)
The normalized ordinate t is proportional to T T* and the normalized abscissa K
is proportional to 1/P, where P is the cholesteric pitch. The diagram illustrates how
different BPs can occur by changing the chirality. (From Crooker.(104))
Striking confirmation of the cubic structures of BP I and BP II was
obtained by Onusseit and Stegemeyer(94) and others, who succeeded in
growing beautiful single crystals of up to a few hundred microns in size(95)
(fig. 4.8.2(#)). Optical Kossel diagrams, analogous to the Kossel lines
observed in X-ray diffraction from crystals, have confirmed their symmetry (96) ( f i g 4.8<2(6)).
Topologically, it turns out that the helical structure of the cholesteric
cannot be deformed continuously to produce a cubic lattice without
creating defects. Thus BP I and BP II are unique examples in nature of a
regular three-dimensional lattice composed of disclination lines. (97) Possible unit cells of such a disclination network, arrived at by minimizing the
Oseen-Frank free energy, are shown in fig. 4.8.3.(98) The tubes in the
diagram represent disclination lines, whose cores are supposed to consist
of isotropic (liquid) material. Precisely which of these configurations
represents the true situation is a matter for further study.
The occurrence of the BPs can also be described in terms of the Landau
theory. (99103) The free energy expansion contains the usual nematic terms
296
and an additional term of the form s - V x s to allow for the noncentrosymmetric ordering (see (4.7.1)). The order parameter is expressed as
a Fourier series, and by choosing the appropriate (hkl) reciprocal lattice
vectors for the Fourier components and minimizing the free energy, one
can generate phase diagrams. A representative phase diagram illustrating
how the different phases can occur by changing the chirality is given in fig.
4.8.4. Calculations show that there can be stable body-centred cubic and
simple cubic structures, and in the presence of an electric field an hexagonal
phase as well. All these predictions are in broad agreement with the
experimental facts. There has been a surge of activity on various aspects of
the BP problem - phase stability and its dependence on pitch, phase
diagrams, electric field effects, single crystal morphology, growth rate,
mechanical properties, etc.(104)
where A is the coefficient of the cubic anharmonicity term, (106) a>0 the
angular frequency and / the moment of inertia of the molecule. Thus the
pitch P (oc 1/(6}) may be expected to decrease slightly with temperature.
However, in many substances the rate of variation is extremely high. It is
now established that if the cholesteric phase is preceded by a smectic phase
at a lower temperature, the pitch increases very rapidly as the sample is
cooled to the smectic-cholesteric transition point (see fig. 5.5.3).
The strong temperature dependence of the pitch has practical applications in thermography, as was first demonstrated by Fergason. (107108)
The material has to be so chosen that the pitch is of the order of the
wavelength of visible light in the temperature range of interest. This is
achieved by preparing suitable mixtures. Small variations of temperature
297
are shown up as changes in the colour of the scattered light and can be used
for visual display of surface temperatures, (109) imaging of infrared(110) and
microwave(111) patterns, etc.
Dependence of pitch on pressure
Pollmann and Stegemeyer(112) investigated the effect of pressure on the
pitch of cholesteryl oleyl carbonate (COC) mixed with cholesteryl chloride
and found that the pitch increases very rapidly with pressure, the effect
being more pronounced the greater the concentration of COC. This
appears at first quite surprising, but, in fact, the explanation is straightforward.(113) We have emphasized that the pitch diverges as the temperature
approaches the cholesteric-smectic transition point. Now pure COC
exhibits a smectic A phase below 14 C at atmospheric pressure. This
temperature may, of course, be somewhat lower in the case of the mixture.
However, as the pressure is raised the temperature of transition goes
Updi4,ii5) s o ^ a ^- t k e p^ch at room temperature may be expected to rise
accordingly.
Mixtures: dependence of pitch on composition
We have seen in 4.1.6 that a mixture of right- and left-handed cholesterics
adopts a helical structure whose pitch is sensitive to temperature and
composition. This result was first described by Friedel. (116) For a given
composition, there is an inversion of the rotatory power as the temperature
is varied, indicating a change of handedness of the helix. The inverse pitch
exhibits a linear dependence on temperature, passing through zero at the
nematic point where there is an exact compensation of the right- and lefthanded forms (fig. 4.1.16).
A similar reversal of handedness takes place as the composition is
varied.(117) The inverse pitch shows a nearly linear relationship with
composition around the nematic point, but there are significant departures
when one of the components has a smectic phase at a lower temperature. (118) The anomaly may again be attributed to smectic-like shortrange order.
It is well known that a nematic liquid crystal readily adopts a helical
configuration if a small amount of a cholesteric is added to it. For low
concentrations of the cholesteric, the inverse pitch is a linear function of
the concentration, but at higher concentrations the linear law is not
obeyed.(119) The curve attains a maximum at a certain concentration,
beyond which it decreases. It turns out, for example, that the twist per unit
length of pure cholesteryl propionate is actually less than that of its
298
mixture with a small amount of MBBA. Saeva and Wysocki (120) have
observed that a compensation occurs even in an MBBA + cholesteryl
chloride mixture when the MBBA concentration is about 30 per cent by
weight. They suggest that certain optically inactive molecules may become
chiral in the helical environment of a cholesteric liquid crystal. In this case,
the MBBA in the mixture would appear to have a chirality opposite to that
of the cholesteryl chloride.
A small quantity of a non-mesomorphic optically active compound may
also transform a nematic into a cholesteric. (121) However, the handedness
of the helix does not seem to be directly related to the absolute
configuration of the solute molecule, as has been shown by Saeva. (122)
For example, (S)-s-amyl-/?-aminocinnamate and (S)-2-(octyl)-/?-aminocinnamate, both of which have the same absolute configuration, result
in helices of opposite senses when dissolved in MBBA. Impurities, indeed
even the vapour of an organic liquid coming into contact with a
cholesteric,(108) have a profound influence on the pitch.
4.10 Molecular models
The first attempt to develop a statistical model of the cholesteric phase was
by Goossens(123) who extended the Maier-Saupe theory to take into
account the chiral nature of the intermolecular coupling and showed that
the second order perturbation energy due to the dipole-quadrupolar
interaction must be included to explain the helicity. However, a difficulty
with this and some of the other models that have since been proposed <124)
is that in their present form they do not give a satisfactory explanation of
the fact that in most cholesterics the pitch decreases with rise of
temperature.
The local order in a cholesteric may be expected to be very weakly
biaxial.(125) The director fluctuations in a plane containing the helical axis
are necessarily different from those in an orthogonal plane and result in a
'phase biaxiality'. (126) Further, there will be a contribution due to the
molecular biaxiality as well. It turns out that the phase biaxiality plays a
significant role in determining the temperature dependence of the pitch.
Goossens(127) has developed a general model taking this into account. The
theory now involves four order parameters; the pitch depends on all four
of them and is temperature dependent. However, a comparison of the
theory with experiment is possible only if the order parameters can be
measured.
Interesting evidence in regard to the factors responsible for the helical
299
II
N = C Q C H = N C H = C H C O C CH 2 CH 2 CH 3
X
X = H, nematic; X = D, cholesteric.
The CH and CD bond lengths are the same, 1.085 A,(129) when one
takes into account the anharmonicity of the vibrations. Thus it would seem
that steric effects are not essential for the helical arrangement.
5
Smectic liquid crystals
300
301
302
__,.
(5.2.1)
(5.2.2)
where
(7rr0A02].
(5.2.3)
(5.2.4)
303
1.0
s
tic
<
3
'20
opi
0.5
is - 15
on
>
- 10
1 -
0.8
0.9
1.0
Reduced temperature
Fig. 5.2.1. Order parameters s and a, entropy S and specific heat cv versus reduced
temperature kB T/0.2202 Vo predicted by the model for a = 1.1 showing afirstorder
smectic A-isotropic transition. S and cv are expressed in terms of Ro, the gas
constant. (After McMillan.(7))
3 cos2 0 - 1
o = /cos(27iz/d)f :
(5.2.6)
where the angular brackets denote statistical averages over the distribution
fv The parameter s defines the orientational order, exactly as in the
Maier-Saupe theory, while a is a new order parameter which is a measure
of the amplitude of the density wave describing the layered structure. The
last two equations can be solved numerically to obtain the following types
of solutions:
(i) o = s = 0
(isotropic phase)
(ii) a = 0, s =1= 0 (nematic phase)
(iii) a H= 0, s ^F 0 (smectic phase).
304
0.8
0.9
Reduced temperature
1.0
Fig. 5.2.2. Order parameters s and a, entropy S and specific heat cv versus reduced
temperature for a = 0.85 showing first order smectic A-nematic and nematicisotropic transitions. (After McMillan.(7))
The free energy of the system can be calculated in the usual manner:
F=U-
where
TS,
(5.2.7)
and
-TS
= NV0(s2
Jo
J
(5.2.8)
The two parameters characterizing the material are Vo, which determines
305
1.0
0.5
)I
4
o?
<
- 11 //
.a
s
20
a,
o
- 15
- 10
-
1 -
1.0
0.9
Reduced temperature
Fig. 5.2.3. Order parameters s and cr, entropy S and specific heat cv versus reduced
temperature for a = 0.6 showing a second order smectic A-nematic transition and
a first order nematic-isotropic transition. (After McMillan. (7))
0.8
306
l.l
Isotropic
1.0
Nematic
Smectic A
0.8
Transition temperature
0.9 -
0.7
\.
Isotropic
Nematic\.
or
\ ^ ^
cholesteric^^"^^^^-^^
/
Smectic A
1.0
1.2
1.1
0.7 0.8 0.9
Model parameter a
Fig. 5.2.4. 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. (After McMillan.(7))
0.5
0.6
(5.2.9)
307
1.0
0.5
0.85
0.90
0.95
1.00
Fig. 5.2.5. Smectic A-nematic (or cholesteric) transition entropy versus ratio of
transition temperatures TAN/Tm. Solid line is theoretical curve taken from fig.
5.2.4; open circles are experimental values of Davis and Porter (MoL Cryst. Liquid
Cryst., 10, 1 (1970)); open triangles are data of Arnold (Z. Physik. Chem. (Leipzig),
239, 283 (1968); ibid., 240, 185 (1969)). (After McMillan.(7))
200
(a)
S 100
Isotropic
Smectic A
I
60
70
80
90
Temperature (C)
Fig. 5.2.6. Measured intensity of X-ray scattering at the Bragg angle versus
temperature for cholesteryl myristate. The dashed line is the calculated diffuse
scattering and fluctuation scattering contribution. The full lines represent the
theoretical curves for the total intensity due to Bragg, diffuse and fluctuation
scattering derived from (a) the simpler model potential and (b) the refined one. The
theoretical intensity has been adjusted to be equal to the experimental value at the
lowest temperature. (After McMillan. (11))
308
0.8
0.6
(a)
0.4
0.8
Z. 0.6
0.4
0.8
0.6
0.4
0.8
(c)
0.9
Reduced temperature
1.0
309
N==C
O-C 8 H 17
O
1.0
0.5
1.2 -
90 C
J
100
110
80
90
Temperature T (C)
Fig. 5.2.8. Temperature dependence of the orientational order parameter s
determined by 14 N quadrupolar splitting measurements for CBOOA. The inset
shows the discontinuity in slope at the smectic A-nematic transition. (After
Cabane and Clark. (23) )
0
50
60
70
310
(5.3.1)
Vxn = 0
n-Vxn = 0 I
and
\
(5.3.2)
n x V x n = 0.J
In other words, both twist and bend distortions are absent, leaving only
the splay term in the Oseen-Frank free energy expression (3.3.7). It is seen
from fig. 5.3.1, that by merely bending or corrugating the layers a splay
deformation can be readily achieved without affecting the layer thickness.
A more complete description of smectic A needs to take into account the
compressibility of the layers, though, of course, the elastic constant for
compression may be expected to be quite large. The basic ideas of this
model were put forward by de Gennes.(24) We consider an idealized
structure which has negligible positional correlation within each smectic
layer and which is optically uniaxial and non-ferroelectric. For small
displacements u of the layers normal to their planes, the free energy density
in the presence of a magnetic field along z, the layer normal, takes the form
/6M
8w\
(a?+#
.....
(5 3 3)
'-
where the first term is the elastic energy for the compression of the layers
311
nTTTTTTT
Fig. 5.3.1. Flexibility of smectic A layers: only such deformations as preserve the
interlayer spacing take place readily.
(5.3.4)
(5.3.5)
kBT
(5.3.6)
(5.3.7)
312
Fig. 5.3.2. A diagram depicting the molecular arrangement in (a) the nematic and
(b) the smectic A phases. In the nematic phase the molecules are randomly
distributed so that any horizontal line intersects the same number of molecules. In
the smectic A phase the number of molecules intersected by a line varies sinusoidally
with the position of the line, being 50 per cent more than the average at the
positions indicated by the arrows and 50 per cent less than the average half way
between. This represents a density wave whose amplitude is 50 per cent of the mean
density, which is far greater than what actually occurs near the A-N transition in
a real system. (After Schaetzing and Litster.(26))
(5.3.8)
313
that the density wave is, in fact, very well described by a sinusoidal
function; the higher order diffraction maxima are either absent or, when
they do occur, extremely weak, about 10~3 or 10~4 times weaker than the
intensity of the principal peak. (27)
In the harmonic approximation, the displacement-displacement correlation function may be defined as
G(r) = exp{-[^ 2 <lW-t/(0)| 2 >]}.
(5.3.9)
For the crystalline lattice, this is the familiar Debye-Waller factor. In the
case of smectic A, one gets from (5.3.4)
(5.3.10)
This was evaluated by Caille (28): for d <^ \r\ < L, the correlation functions
for the two principal directions vary as
and
G(r)ocz-
r = ( * + , ) * - 0,1
2
G(r) oc rL \
z~0,
( 5 3 U )
where
n = ~~zi,
(5-3.12)
q = 0,
qz = 0,
(5.3.13)
(5.3.14)
where
q0 = In Id.
The sharp Bragg peak characteristic of the long-range ordered crystal
lattice is now replaced by strong thermal diffuse scattering with a ' powerlaw' singularity.(29) The excellent high resolution X-ray measurements of
Als-Neilsen et al.m) have confirmed this prediction (fig. 5.3.3).
533
314
: / \
6
10"1
\\ v
M
\ \
10"2
\
\
A
\\ V
\\
\
\ 1
ry
= 0.38
O
= 0.17
-1
Fig. 5.3.3. X-ray scattering intensity profile from the smectic A phase of 80CB at
two reduced temperatures, ( r A N - T)/TAN = 9 x 10"4 (filled circles) and 10"6 (open
circles). The dashed line is the experimental resolution function as would be seen if
smectic A had true long-range order. The full lines are the best fits of the theoretical
line shape \qz qo\~2+n folded with the resolution function. The values of rj so
obtained agree with those calculated from (5.3.12) using experimentally determined
values of q0, B and k1. (Als-Nielson et al.(m)
F =where
X=
(5.3.16)
(5.3.17)
315
ir = i^r^_I/ r ^yi 2 _ h fc n ^Y
(5.3.19)
(5.3.20)
316
V/////////////////7/A
Y////X
Y/////. 777/77//A
277/*,
50
100
150
317
(5.3.22)
k T
(5.3.23)
(5.3.24)
is quite small; consequently, in certain geometries, e.g., when a homeotropically aligned sample is held against an extended source of light and
viewed normally, the medium will not appear very turbid.
Strictly speaking, one should take into account the contributions of k 22
and k33, since the layers are assumed to be compressible. Following the
procedure outlined in 3.9, the fluctuations may be decomposed into two
modes, and choosing the wavevector q in the xz plane, one gets the general
expressions(34)
,,* , , s =
xl?
k q i + k^
kBT[l+(B/D)(qJqLn
+ iB/DHD
+k q l +k ^ D i q J q r
318
Fig. 5.3.6. Distortion caused by an irregularity on the glass surface. The bending of
the layers can be relaxed only by compression, which requires considerable energy,
so that the surface distortion extends to appreciable depths inside the specimen.
In a perfect sample of smectic A, the most important contribution to
light scattering comes from the undulation mode with the wavevector
parallel to the layers (qz = 0). However, unless very special precautions are
taken, it is difficult to observe this scattering, for even small irregularities
on the surfaces of the glass plates cause static distortions of the layers
which extend to appreciable depths inside the specimen (3532) (fig. 5.3.6).
The reason for the high penetration length of a surface distortion is easily
understood. A corrugation of the layer which involves a nematic-like splay
elastic constant and requires very little energy can be relaxed only by
compression, which needs considerable energy and therefore occurs over a
large distance. If a static undulation of this type is written as
u = u0 exp ( - z / 0 cos qx,
where C is the 'attenuation' length, substitution in the free energy
expression (5.3.18) gives
or
C = (V)
(whereas in nematics, ~ q x ). These long-range static undulations give
rise to an intense scattering of light which completely swamps that due to
the thermal fluctuations. Indeed in the first experiments*36'37) it was the
static effects that were observed as was confirmed by a temporal analysis of
319
r-
104
103
77 C
102
79 C
10
0.1
74 C
I
10"
io
101
102
320
B = B0-(2C20/A0)
(5.3.29)
(5.3.30)
where gt is the force on the layers and t'n the viscous stress tensor which, in
contrast to the nematic case, is assumed to be symmetrical and dependent
on the velocity gradients only. In the problem under consideration we have
(5.3.31)
and
g = -SF/Su.
(5.3.32)
The force g normal to the layers will be associated with permeation effects.
The idea of permeation was put forward originally by Helfrich(40) to
explain the very high viscosity coefficients of cholesteric and smectic liquid
crystals at low shear rates (see figs. 4.5.1 and 5.3.7). In cholesterics,
permeation falls conceptually within the framework of the Ericksen-Leslie
theory(42) (see 4.5.1), but in the case of smectics, it invokes an entirely new
mechanism reminiscent of the drift of charge carriers in the hopping model
for electrical conduction (fig. 5.3.8).
The rate of entropy production may be written as (see 3.1.4)
TS=t'ndij+g(u-vz),
(5.3.33)
where dtj = \{v{ j + Vj t), and (u vz) describes the permeation. Treating t'n
321
Fig. 5.3.8. Helfrich's model of permeation in smectic liquid crystals. The flow takes
place normal to the layers in a manner similar to the drift of charge carriers in the
hopping model for electrical conduction.
10 --
A
\
<\
\
L
5 T
10
15
q2 (108 cm- 2 )
1
20
25
Fig. 5.3.9. Damping rate of the undulation mode in the smectic A phase of CBOOA
determined by laser beat spectroscopy for two different sample thicknesses, 200
and 800 /zm. Solid lines represent the theoretical curves calculated from (5.3.40)
and (5.3.41). (After Ribotta, Salin and Durand.(38))
and g as fluxes and dtj and (u vz) as forces, and expanding t'n one obtains
the following relations:
^ = Mi sn
+ M2
(5.3.34)
u-vz
= v p g,
(5.3.35)
where //15 ...,// 5 are five viscosity coefficients and vp the permeation
coefficient.
322
(5.3.36)
where
g = -kliq*u
= (u-vz)/vv
(5.3.37)
and
* = SG<s + / O .
(5.3.38)
For small q, the last term may be neglected (except very near the boundary,
but we shall ignore the boundary layer). This results in a purely damped
mode whose relaxation rate is
1
L=
T
n2
h3_.
rj
(5.3.39)
More generally, taking into account the thickness L of the sample one may
write
t]lL
2 ^
rj
where
1 = qc = (n/lL)i
(5.3.42)
323
(5.3.43)
y and 6 being the independent variables. This leads to the secular equation
co{\pco2 - (Bo + Co) q% (po? - Ao q\) -(Au + Co) q\{paf + Co q\)} = 0.
(5.3.45)
We are interested here in propagating modes and ignore the case co = 0.
Solving for the velocities c = co/q, we get
c\ + c\ = p-\AQ cos2 <p + (Ao + Bo + 2C0) sin2 p],
2
cos ^,
(5.3.46)
(5.3.47)
^33
Their estimates show that Ao is much larger than Bo or Co (fig. 5.3.10). This
enables us to give a simple physical interpretation of the two branches
324
1.6
1.5
1.4
110
115
120
Temperature (C)
0.3 r
0.2
X
0.1
8- -8-
no
120
115
Temperature (C)
Fig. 5.3.10. Temperature dependence of the elastic constants of the smectic A phase
of diethyl 4,4/-azoxydibenzoate determined by ultrasonic velocity measurements:
open circles, 2 MHz; filled circles, 5 MHz; triangles, 12 MHz; crosses, 20 MHz.
(After Miyano and Ketterson.(44))
described by (5.3.46) and (5.3.47): the density and layer oscillations are, in
effect, uncoupled. Hence one of the branches corresponds to the normal
longitudinal wave whose velocity can be seen from (5.3.46) to be
Cl
(AJp)\
(5.3.48)
(5.3.49)
325
u = vpg
(5.3.50)
u = vpBd2u/dz2.
(5.3.51)
(5.3.52)
Furthermore, second sound is a critical mode, i.e., its velocity goes to zero
as r - TAN.
A direct confirmation of the existence of these two branches has been
found by Liao, Clark and Pershan(47) from their Brillouin scattering
experiments on a monodomain sample of ^-methyl butyl /?((/?-methoxybenzylidene)amino) cinnamate. This compound shows the nematic,
smectic A and smectic B phases. Choosing both the incident and the
scattered light to be polarized either as ordinary or extraordinary waves,
they observed two peaks corresponding to the two modes, the angular
dependence of which is in excellent agreement with the theory (fig. 5.3.11).
5.3.7 Breakdown of conventional hydrodynamics
Theoretical developments*48'49) in the early 1980s showed that the nonlinear
interaction of thermally excited layer undulations, which as we have seen
have large amplitudes because of the Peierls-Landau instability, leads to
interesting new effects in the hydrodynamics of smectic A at small
wavevector and frequency. We present below a very brief outline of the
physical arguments involved.(50)
In writing the expression for the free energy density (5.3.18) we neglected
terms involving (du/dx)2 and (du/dy)2 since a uniform rotation of the layers
does not cost energy (assuming, of course, that there is no externally
applied magnetic field). Consequently in the harmonic approximation,
<|wJ2> takes the form (5.3.22). However, while discussing the undulation
instability in smectic A (5.3.3), we observed that a bending of the layers
alters the effective layer spacing along z and therefore makes a second
326
2.0 H
1.5
1.0
(a)
0.5
1.0
2.0
2.0
1.5
1.0
(b)
0.5
1.0
2.0
Fig. 5.3.11. Dependence of the sound velocities on the polar angle q> from Brillouin
scattering experiments on ^-methyl butyl /?((/>-methoxy-benzylidene)amino) cinnamate. (a) Smectic A (T = 60.7 C), (b) smectic B (T = 48.1 C). The dashed lines
are calculated from theory. The presence of a third component in (b) indicates that
the shear modulus does not vanish in smectic B at these very high frequencies.
Circles, triangles and squares represent measurements at different scattering
angles. (After Liao, Clark and Pershan.(47))
order contribution to the layer dilatation in that direction. The corrected
form of the free energy density is given by (5.3.19), which we write here as
follows:
(say),
(5.3.53)
where the subscripts h and a stand for the harmonic and anharmonic parts
327
328
(a)
(b)
Circle L(.
Hyperbola Lh
to
Ellipse Le
Fig. 5.4.1. (a) Smectic layers in concentric cylinders to form a myelin sheath with
a singular line L along the axis; (b) the cylinders are closed to form tori: there are
two singular lines, a circle Lc and a straight line Ls; (c) the general case when the
smectic layers form Dupin cyclides: the circle becomes an ellipse L e and the straight
line a hyperbola Lh.
rise to beautiful optical patterns known as focal conic textures. These were
studied in considerable detail by Friedel, (55) to whom we owe the
explanation of their origin. We shall now examine the structures of these
focal conic domains. (56)
(a) The simplest case is shown in fig. 5.4.1 (a). The layers are wrapped
round in concentric cylinders to form a myelin sheath and there is a
singular line L along the axis, (b) When the myelin sheath is bent and closed
to form tori, the singular line L becomes a circle Lc (fig. 5.4.1 (b)). In
addition, the structure now has another singular line Ls - a straight line
through the centre of the circle and perpendicular to it. Such a structure is
{a)
329
(b)
Fig. 5.4.2. A rare example of a pair of singular lines, one almost straight and the
other almost circular in a toric domain in the (a) smectic A and (b) smectic C
phases. Additional disclination lines develop near the centre in the smectic C phase
for reasons discussed in 5.8.3. (From A. Perez, M. Brunet and O. Parodi, J. de
Physique Lettres, 39, 353 (1978)).
observed only very rarely (fig. 5.4.2). (c) In the most general case, the
smectic layers lie on Dupin cyclides and not on tori. The circle Lc is then
transformed to an ellipse Le and the straight line into a hyperbola Lh (fig.
5.4.1 (c)). Le and Lh are conic sections in perpendicular planes, one going
through the focus of the other (figs. 5.4.3 and 5.4.4). It is these structures
that are most commonly seen under the polarizing microscope (fig.
1.1.7 (a)).
Based on the fact that the layers are bent into Dupin cyclides, one can
derive expressions for the principal curvatures ox and cr2, and thus work out
the energy,
330
(a)
Fig. 5.4.3. (a) Geometry of a pair of focal conies. (After Friedel.(55)). (b) A section
of (a) in the plane of the hyperbola, showing parts of Dupin cyclides. (After
Bragg.(56))
For the general focal conic structure, Kleman (57) has shown that
331
(a)
Fig. 5.4.4. (a) Arrangement of a set of cones within a pyramid. The hyperbolas
belonging to the ellipses meet at the vertex of the pyramid, (b) The base of the
pyramid and the axial directions radiating from the foci at the base of the
hyperbolas. (After Bragg.(56))
In principle, the focal lines can both be parabolas. Such parabolic focal
conies have been observed by Rosenblatt et al.{58) but they are not
common. Kleman(57) has shown that in this case
4/rJ'
where/is the focal length of the parabola and R the sample radius. The
energy increases with R and diverges as f-+ 0. Thus, in general, such
structures are not energetically favourable. However, it turns out that
when R < 3/, they can have smaller energies and Kleman argues that this
may explain the observations of Rosenblatt et al
A remarkable property of the smectic liquid crystal is that an open
332
333
dj
where
A. = (k11/B)* and u = u(z, x).
We now proceed to work out the singular solutions of (5.4.1). For an edge
dislocation with the extra half-plane on the side x > 0, the boundary
conditions are
\0
for i < 0
m
"(0, x) = i
[b/2 for x > 0 .
334
(a)
(b)
Fig. 5.4.6. (a) Positive and (b) negative edge dislocations of unit strength.
Also, u(z,x) = u( z,x). We look for a solution of the form<62)
b
b f , x fexp (i<7.x)l ,
( 5 A 2 )
We then have
(5.4.3)
The tilt in the layer normal 6 = du/dx, and the dilatation in the layer
spacing S = Ad/d = Xffl/dx; 6 and 5 can be derived explicitly from the
form of u:
|=
b__ex
J
x
i__x'
(5.4.4)
Xb
(5.4.5)
From (5.4.5) it is seen that there is a compression of the layers for x > 0 and
an extension for x < 0.
The stresses associated with the dislocation cannot be derived uniquely.
However, one acceptable set of values can be obtained from the equation
of motion. (63) They are
Bb
-x
17-3 e x P
2
XBb
-x
/15Z? z I
(5.4.6)
-:
72
2 AX\z exp
x
2X\z exp
- AX\z\
- 4A|zj
(5.4.7)
(5.4.8)
Apart from the exponentially decaying term, a33 varies as l/n, GX1 and a13
as \/n.
335
where v is Poisson's ratio. Again the form of this expression can be seen to
be quite different from (5.4.2) for u in smectic A.
The energy of a single edge dislocation
The total energy is given by
E= I \Fdxdz.
(5.4.10)
(5.4.12)
Aq2(z0-z)]}[exp(iqx)]jt,
(5.4.13)
336
Here e is 1 for like dislocations and 1 for unlike ones. The interaction
energy
Ex= J \Fdxdz-E0,
(5.4.15)
(5.4.16)
^ ( ^ f f ^ )
(5.4.17)
From this we get two forces Fx and Fz along and normal to the smectic
planes:
Since k.JX2 = b,
and
Fx = -ebaS3
(5.4.20)
Fz = ebals.
(5.4.21)
Here cr33 and <713 are the stresses at the origin, where the first dislocation is
situated, due to the presence of the other dislocation at (x o,zo). This is
known as the Peach-Koehler force(65) on a dislocation arising from the
stress field of the other. This can be generalized to mean that under
application of a stress oip a dislocation experiences a force Fi whose exact
relationship is given by the above expressions.
It is seen from (5.4.18) that Fx is always repulsive for a pair of like
dislocations and always attractive for an unlike pair. On the other hand,
Fz can be attractive or repulsive depending on the relative positions of the
two dislocations. For example, Fz is repulsive for like dislocations when
xl < 2Xz and attractive when x\ > 2Az0. This provides a mechanism for the
337
Fig. 5.4.7. Clustering of like edge dislocations in smectic A to form a domain wall
or 'grain boundary'. (After Pershan.(63))
338
(a)
vllllIMIIIII
^11111111111
dmmmii
1 III 1
11111
r 1HIM
1,111,1.1,1,1,1.1.1
II
(c)
(d)
Fig. 5.4.8. Disclinations in smectic A: (a) la(n); (b) Qb(n); (c) Qo( - n); (</) Qb( - n).
u =
tan"1 r-
2n
\x
where u is the layer displacement along z, and b the Burgers vector is equal
to an integral multiple of the layer spacing d. For a positive screw
dislocation one gains a step of height b on going once round the line L in
the anticlockwise direction, while for a negative dislocation one loses a
step.
A noteworthy aspect of the above solution is that it does not involve
either lattice dilatation du/dz or layer undulation V-n. Therefore, within
the approximations of the linear theory considered here, screw dislocations
in smectic A have no self energy (apart from the core), nor do they interact
amongst themselves. In this respect they are entirely different from screw
dislocations in crystals.
5.4.4 Disclinations
The geometrical process for creating a disclination in smectic A is as
follows.(67) Cut the material by a semi-infinite plane that runs parallel to the
layers, the limit of the cut being the disclination line L. Rotate the two faces
rr
IT
1,
^\\\\ II
11II
^X//lll
II
1111
i i II
I I I II 1 IN
n1
1111 1 II MI
11 111 i i n i
Y l III I I I I
II i
sttt
III -UrrU
41 El III! I I I II
(b)
()
339
i IIII
II ,1,1,1
II
i 1111
HIM II
III
111
{d)
(c)
(a)
(b)
340
(5.5.1)
where p0 is the mean density, \y/\ the amplitude and qs = 2n/d the
wavevector of the density wave, d the interlayer spacing and (p a phase
factor which gives the position of the layers. Thus the smectic order can be
fully specified by the complex parameter
y/ = Hexp(i<p).
(5.5.2)
Near the transition, the free energy density may be expanded in powers of
y/ as follows
....
(5.5.3)
From symmetry considerations it is clear that only even powers of y/ may
be included. The coefficient /? is always positive. At a certain temperature
7"*, which is the second order transition point, a = 0. Accordingly, as
explained in 2.5.1, we set in the mean field approximation,
a = a(T-T*).
To allow for the coupling between \y/\ and the orientational order
parameter s, the free energy is expressed as(69)
F= \\*+P\y\* + [(dsy/2x\-C\w\>ds,
(5.5.4)
341
term in the free energy to ensure stability. The tricritical point occurs when
/?' = 0, i.e., x(TAN) = 2/3/C2. This is in agreement with the prediction of the
microscopic theory, according to which the tricritical point occurs at
7; N /r NI = 0.87 (see 5.2). /(7^ N ) may be altered by varying the length of
the end-chain, or by preparing mixtures, or by the application of
pressure.(70)
5.5.2 Pretransition effects in the nematic phase
In 5.2, we referred briefly to McMillan's X-ray evidence(11) for the growth
of smectic-like short-range order in the nematic (or cholesteric) phase near
the A-N transition (see fig. 5.2.6). This pretransition effect manifests itself
more strikingly in the temperature dependence of the elastic properties of
the nematic phase, as was first demonstrated by Gruler.(71) The twist and
bend distortions, which are normally disallowed in the smectic phase,
become increasingly difficult as the smectic clusters build up and these two
elastic constants rise much more rapidly than expected from the simple s2
law discussed in 2.3.5. To investigate these properties, we expand the free
energy in powers of y/ and its gradients. We shall suppose for the present
that the A-N transition is of second order. For a fixed orientation of the
director,
In the smectic phase (T < T*), the amplitude of the density wave may be
taken to be constant, so that only q> varies. The gradient terms of F
therefore become
Comparing this with (5.3.3) it is at once clear that cp is related to the layer
displacement u:\q>\2 = q2\u\2 and B =\y/\2 ql/Mv.
dcp/dy represent the tilt of the layers with respect to the director. If the
director orientation is not fixed, it is the relative tilt between the layers and
the director that should be considered and therefore (5.5.6) takes the
generalized form
^(^)\
2Mv\dz)
2M T '
(5.5.7)
where VT is the gradient operator in the plane of the layers. This equation
342
(5.5.9)
(5.5.10)
(5.5.11)
+ Fs{yy\
(5.5.12)
<f(Vn)2,
(5.5.13)
343
(5.5.14)
Thus from (5.5.12), (5.5.13) and (5.5.14), the effective elastic constant for
twist or bend will be k + Sk, where
Skoc^t
(5.5.15)
Since the coherence lengths diverge rapidly near the A-N transition, the
elastic constants for twist and bend should also show critical behaviour. In
the mean field approximation
oc ( r - r * ) ~ 1 / 2 .
(5.5.16)
(5.5.17)
(5.5.19)
(5.5.20)
Dccl/r
(5.5.21)
The theory continues to be valid even if the A-N transition is weakly first
order, except that T* represents a hypothetical second order transition
temperature slightly below 7^N.
The first X-ray scattering experiments of McMillan(74) on /?-n-octyloxybenzylidene-/?-toluidine, which exhibits a first order A-N transition,
agreed with the mean field theory. On the other hand, CBOOA showed
appreciable anisotropy in the temperature variation of the longitudinal
and transverse coherence lengths. Later studies on a number of materials
have confirmed that anisotropy is a general feature exhibited by all of
them.
The increases in k22 and fc33 in the nematic phase near the A-N transition
over and above that given by the usual s2 law are now well established. Fig.
5.5.1 presents the data of Cheung, Meyer and Gruler(75) for CBOOA; &33
344
70 r-
60
50
40
Ax
30
20
10
,, 1
0.1
1.0
r-r A
10
30
Fig. 5.5.1. The temperature dependence of the splay and bend elastic constants, kxl
(crosses) and k33 (circles) respectively, in the nematic phase of CBOOA prior to the
smectic A-nematic transition. The values are plotted as &../A/ where A/ is the
anisotropy of the diamagnetic susceptibility of the nematic. The full line shows the
order parameter s, normalized to fit fcn/A/ at higher temperatures. The splay
constant klx deviates only very slightly from ordinary nematic behaviour while the
bend constant kZ2> exhibitsa critical increase near 7^N due(75)to pretransition
fluctuations. (After Cheung, Meyer and Gruler. )
diverges rapidly while fcn exhibits normal behaviour. Similarly, it has been
verified that the layer compressibility constant B shows a critical variation
in the vicinity of the transition.
Critical divergence of the viscosity coefficients
Certain nematic viscosity coefficients also exhibit critical behaviour. The
origin of this effect may be explained physically as follows. Take, for
example, the frictional torque associated with the twist viscosity coefficient
Xx defined by (3.3.14). The formation of smectic clusters results in an extra
torque due to the flow of the liquid normal to the smectic planes. This extra
torque increases as the clusters become larger and longer lived, and in
consequence there is a net enhancement of the effective Xx. To estimate this
enhancement/ 73 ' 76) consider a slowly rotating magnetic field having an
angular velocity Q. We have seen in 3.6.2 that the director follows the field
at a constant inclination, the torque being given by 2XQ. Now, the
layered structure of the smectic-like regions makes an additional con-
345
tribution to the torque, say F s . If the angle between the layer normal and
the director is 0, then
6= QV
(5.5.22)
where TV is the relaxation time of \y/\. The torque Ts may be derived from
(5.5.13) with 0 = <jVn:
M
< w 2 >
(5 5 23)
- -
(5.5.24)
33
(mean
field)
(dynamical scaling).
(5.5.25)
(5.5.26)
346
weakly first order. The shift in the transition point due to this effect,
relative to the second order transition point T*, was estimated to be of
the order of 10 mK. Whether or not the A-N transition can, in principle,
be of second order has been the subject of some discussion. (7981) From
high resolution experimental studies it appears that the second order
nature of the transition has been more or less established in at least a
few cases(82) (though Anisimov et al.m) have claimed to have observed
the 'Halperin-Lubensky-Ma effect' in three different systems). A
number of theories have been proposed, which have been reviewed in
authoritative articles,(84~6) but a precise description of the A-N transition
still remains elusive. We shall give a very brief summary of the current
situation.
347
0.07
0.05
-0.007
0
0.15
0.15
0.2
0.22
0.31
0.45
0.50
-0.007
0.5
0.70
0.69
0.78
0.83
0.70
0.70
0.71
0.71
0.67
0.61
0.57
0.669
0.5
0.65
0.61
0.65
0.68
0.62
0.57
0.58
0.57
0.51
0.51
0.39
0.669
0.5
1.22
1.22
1.46
1.53
]1.30
1.31
1.32
1.31
11.26
]1.10
11.10
] .316
11.0
Compound
T8
T7
40.7
8S5
CBOOA
40.8
80CB
9S5
8CB
10S5
9CB
XY model
Tricritical
0.66
0.71
0.926
0.936
0.94
0.958
0.963
0.967
0.977
0.983
0.994
a + 2v + v|| Reference
1.93 + 0.15
2.06 + 0.15
2.07 + 0.1
2.19 + 0.16
2.09 + 0.14
1.99 + 0.18
2.07 + 0.17
2.01 0.18
2.00 + 0.13
2.08 + 0.18
1.85 + 0.18
2.0
2.0
92
92
89,93
94,95
96,97
98
96,99
95, 100
101,82
95, 100
100, 82
102, 103
Tn = 4-alkoxybenzoyloxy-4/-cyanostilbene n = 7,8
40.m = butyloxybenzylidene alkylaniline m = 7,8
nS5 = 4-n-pentylphenylthiol-4/-alkyloxybenzoate n = 8,9,10
Reduced temperature: t =
(T-TAN)/TAN.
Specific heat at constant pressure: Cv = A\t\~
Correlation length,
longitudinal (parallel to n): ^ = ^\t\'vw
transverse (perpendicular to n):
(5.5.27)
= \
Susceptibility: x = Xo\*\~
(5.5.28)
= 0.67
and
y =1.32.
348
v = lvxy = 0.53.
We note from Table 5.5.1 that v is anisotropic for all the materials, v|( v
being about 0.13. In no case is v(| = 2v. Also, all the materials satisfy the
anisotropic hyperscaling relation (5.5.28) to within experimental limits.
For two compounds, 407 and 8S5, a agrees with the predicted value of
-0.007, but the vs do not. Brisbin et al.{95) have suggested that the A-N
transition for some of the compounds, e.g., 10S5 and 9CB, is near a
tricritical point as the values of TAN/TNl are high for these compounds. This
may well be true, for y, a and (v)( + 2v )/3 for both 10S5 and 9CB are close
to the tricritical values of 1.0, 0.5 and 0.5 respectively. But this still leaves
the question of anisotropy unexplained.
From (5.5.20) and (5.5.21) it is seen that B oc t*9 where ^ = 2 v - v,, and
D oc fw. If V|| = 2v1? B should be finite at the transition temperature.
However, experimentally/ 107'108) it appears that B at the transition is
almost vanishingly small within experimental limits. Few measurements
are available on D to draw any definite conclusions. In any case, as pointed
out earlier, the exponents are neither universal nor do they agree with the
predictions of any of the theoretical models. Vithana et al.ao9) have
suggested that the widely differing values of the exponents for the different
compounds may be a consequence of the fact that one is measuring
effective values associated with crossover effects between the XY class and
a tricritical point. A further complication is that the experiments of
Evans-Lutterodt et al.(92) appear to indicate that the occurrence of different
forms of smectic A (see 5.6) may have a bearing on the nature of the A-N
transition.
In summary, therefore, it would be fair to conclude that no theory
predicts all the exponents correctly for any of the systems. Thus the A-N
transition, which is probably the most extensively studied transition in
liquid crystals, still remains a major unsolved problem in condensed matter
physics.
The smectic A-cholesteric transition
Lubensky(110) has shown that, in principle, the smectic A-cholesteric
transition should always be of first order, in analogy with the behaviour of
a superconductor in an external magnetic field. The relative shift of the
349
0.40
0.35
0.30
0.25
72
76
80
84
Temperature (C)
Fig. 5.5.3. Pitch versus temperature in cholesteryl nonanoate prior to the smectic
A-cholesteric transition (74 C); the crosses are values obtained from observations
of the Grandjean-Cano walls and the circles from the wavelengths of maximum
reflexion. (After Kassubek and Meier.(111))
transition temperature AT/T* due to this effect is estimated to be ~ 10~3,
but no experimental study of the effect appears to have been reported.
Near the transition the free energy of the cholesteric may be written as
F =\k\2q2 k\2qQq + FJ<y/)
(5.5.29)
omitting the ' background' term, where q0 is the equilibrium value of the
twist per unit length in the absence of smectic-like short range order and q
the actual value. Therefore
(5.5.30)
Minimizing with respect to q, we get
/Coo I* Ofcn
(5.5.31)
which decreases rapidly as the temperature drops to the smectic A-cholesteric transition point, or, in other words, the pitch P = 2n/q increases.
Fig. 5.5.3 presents the temperature dependence of the pitch of cholesteryl
nonanoate, (111) a compound which shows the smectic A phase below 74 C.
350
Pindak, Huang and Ho (112) have reported that dk22 oc r 0 6 7 for this
compound. The high temperature sensitivity of the pitch of such materials
has applications in thermography, as we have already seen in 4.9.
(5.6.3)
(5.6.4)
351
A^
in,
1
1
Ad
1
I
1
ri
i, - .
i, , , ,
ii
!!
!
I
I I P IIII
ii
ji
II
K vi
II
if
'!
>;
\i
II
I
i
ii
" "h i
.
l> '
i;
ii
11
i
i
352
Soliton region
D3I
II'7T-1I
Locked phase
region
JIT"
nx.
Wi Wn'fi
Soliton region (a)
Fig. 5.6.2. Schematic representation of the molecular arrangement in (a) the weakly
coupled incommensurate smectic A, and (b) the strongly coupled incommensurate
smectic A. (After Prost and Barois. (115))
of the two waves are modulated (fig. 5.6.2(b)) exactly in accordance with
the theory proposed originally by Frank and Van der Merwe. (116) In the
latter case, the structure consists of large regions of A 2 separated
periodically by defect walls or phase solitons. Experimentally, however,
the occurrence of incommensurate smectic A phases has yet to be
established conclusively. (117~119)
Barois, Prost and Lubensky(120) have used the phenomenological model,
within the framework of the mean field theory, to construct phase diagrams
involving the polymorphic forms of the A phase. They have predicted three
kinds of critical points, the A d -A 2 critical point, the A x -A 2 tricritical point
and the A 1 -N-A d bicritical point, the salient features of which are
summarized below.
Since Ad and A2 have the same symmetry, it is to be expected that a first
order A d -A 2 phase boundary should terminate at a critical point. This has
been confirmed(121) in a binary liquid crystal system. Plots of wavevectors
against temperature for different concentrations (X) in the vicinity of the
critical point are shown in fig. 5.6.3. It is seen that the first order A d -A 2
transition manifests itself as a jump in the wavevector, accompanied by a
two-phase coexistence region. On increasing X the magnitude of the jump
as well as the width of the two-phase region decrease, until finally only a
119
120
121
353
124
125
126
127
122
123
124
125
Temperature (C)
Fig. 5.6.3. Plots of the magnitudes of wavevectors q0 and q'Q (corresponding to A2
and Ad periodicities respectively) against temperature (7") for mixtures of 4-nundecyloxyphenyl-4/-(4//-cyanobenzyloxy)benzoate (11OPCBOB) and 4-n-nonyloxybiphenyl-4'-cyanobenzoate near the A d -A 2 critical point (CP). The mole
fractions of 11OPCBOB in the mixtures are (a) 0.550, (b) 0.571, (c) 0.597, (d) 0.619,
(e) 0.642, (/) 0.715, (g) 0.80 and (h) 1.0. The first order A d -A 2 transition is identified
for plots (a)-(d) by a two-phase region consisting of both Ad and A2 periodicities
(shown as closed circles) and also by a jump in the wavevector. As CP is
approached the two-phase region shrinks with an accompanying decrease in the
difference between q0 and q'o. At CP (plot (e)) only a vertical inflexion is seen.
Beyond CP, Ad evolves continuously into the A2 phase (plots (/)-(//)). The
behaviour is similar to that seen in the vicinity of the gas-liquid critical point.
354
120
I
5
10
15
Concentration (molar % TBBA)
Fig. 5.6.4. Binary phase diagram for mixtures of 4-n-hexylphenyl-4/-cyanobenzoyloxybenzoate and terephthal-to-butylaniline (TBBA) showing the A^-Ag
tricritical point (TCP). The solid line represents the first order phase boundary
and the dashed lines represent the second order phase boundaries. (After Chan
et
tf/.(123))
52
53
54
Concentration, X (mol %)
355
56
A 1 -N R -A d point show that the bicritical point has, in fact, split into a
tricritical point (for the A X -N R boundary) and a critical end point (for
the A d -N R boundary, fig. 5.6.5). This new result, though predicted
theoretically for magnetic systems,(128) has not been envisaged in any of
the theories of frustrated smectics.
5,6.2 The phenomenon of re-entrance
When a compound exhibits both nematic and smectic phases, then, as a
rule, the nematic occurs at a higher temperature. Exceptions to the rule
were discovered by Cladis(129) in certain strongly polar materials. The first
observations were on binary mixtures of two cyano compounds: over a
range of composition the sequence of transitions on cooling was as follows :
iso -> N -> SA -> N R -> crystal,
where N stands for the usual nematic and N R for a second nematic, called
356
1.4
1.0
0.6
Isotropic
Smectic A
0.2
50
70
90
110
Temperature (C)
Fig. 5.6.6. Experimental pressure-temperature diagram for 80CB showing the reentrant nematic phase. (After Cladis et al.a30))
C 9 H 19 O
OOC
OOC
Fig. 5.6.7. The structural formula of 4-nonyloxyphenyl-4 /-nitrobenzoyloxybenzoate, a compound that shows multiple re-entrant behaviour.
(a)
357
(b)
358
359
-COOC*H(CH 3 )C 6 H 13
C 13 H 27 O
81.6 C
88.3 C
94.1 C
*>C*
* A*
33.7 C
s4 <
s,
43.7 C
Fig. 5.6.9. The structural formula of (R)- and (S)-l-methylheptyl 4'-(((4"-nalkoxyphenyl)propioloyl)oxy)biphenyl-4-carboxylates, which show the smectic A*
phase.
Fig. 5.6.10. The structure of smectic A* or the twist grain boundary phase.
A-N transition is of second order, or nearly so, the material may be
expected to be of type I.(69) Measurements(161) on CBOOA, which has a
nearly second order A-N transition, have confirmed that the variation of
7^N with the twist per unit length does indeed reproduce the theoretically
predicted trend for A/f < \/y/2.
For a type II material, de Gennes envisaged the possibility of a regular
network of dislocations being formed in smectic A under an imposed twist
or bend distortion. A model for such a network, similar to the Abrikosov
flux lattice phase(162) in a type II superconductor subjected to an external
magnetic field, was discussed by Renn and Lubensky.(163) Soon afterwards,
a phase consisting of a regular array of dislocations was discovered by
Goodby et al.am who designated it as smectic A*.
Smectic A* is composed of optically active molecules. The molecular
structure of one of the compounds studied by Goodby et #/.(164) and the
phase transitions occurring in it are given in fig. 5.6.9. The structure of
smectic A* is depicted schematically in fig. 5.6.10. Unlike smectic C* whose
helical axis is normal to the layers (see fig. 5.10.1) smectic A* has its helical
axis parallel to them. Hence the structure may be looked upon as a series
of smectic A blocks or grains separated by twist grain boundaries. The
director (which in this case is the layer normal) is rotated by a constant
angle on going from one grain to the next. From observations of the
360
Six) = I0\c0+ t
*ncos6,2(90-/)l + /bg,
(5.7.1)
361
(a)
0'
(b)
Fig. 5.7.1. (a) The structure of the hexatic phase of a two-dimensional lattice. The
orientation of the lattice vectors a and b is preserved over a long range, but the
molecular positions are correlated only over a short distance fp. (b) The
orientational order as well as the positional order are short range. (After Litster
and Birgeneau.(88))
C6w measure the degree of 6-fold ordering. An application of (5.7.1) to
study the bond-orientational order across the Sj-S c transition(172) will be
described in 5.8.1.
The hexatic B-S A transition is expected to belong to the XY universality
class,(171) but high resolution calorimetric investigations(173) appear to
indicate that this may not be the case. Tilted analogues of hexatic B,
namely SF and SI5 have also been identified.(174)
362
4 -
7=68.1 C
I o
15
(c)
10
r=63.9C
ir/6
Rotation angle, x
Fig. 5.7.2. A 60 segment of a/-scan about an axis normal to the smectic layers of
the intensity of X-ray scattering from n-hexyl-4'-n-pentyloxybiphenyl-4-carboxylate exhibiting the smectic A-hexatic B phase transition, (a) The scattering in
the A phase is a diffuse ring independent of/. For temperatures below the smectic
A-hexatic B transition the ring develops a sixfold modulation (Z>), which
eventually breaks up into six diffuse spots (c). (From Pindak et al.ai0))
5.8 Smectic C
5.8.1 Description of the structure
In S c the molecules are disordered within the layers, as in SA, but inclined
with respect to the layer normal. The structure has biaxial symmetry of the
monoclinic class, with a plane of symmetry as indicated in fig. 5.8.1. The
origin of the biaxiality may be explained as follows. The tilt angle 6 is
directly coupled with the layer thickness whereas the azimuthal angle <p is
not. Therefore, at any given temperature, the amplitudes of the 6
oscillations of the director are small compared with those of the (/>
5.8 Smectic C
363
^ Symmetry plane
' Molecular axis
'///////////////////A
V///////////////////,
V////////////////A
364
"2
o
75
77
Temperature (C)
79
81
that thermodynamically the two phases are not distinct. Tilt angle
measurements (179) (using X-ray methods) have confirmed that there is
indeed a continuous evolution of S c into SI5 as the temperature is lowered.
Generally speaking, compounds exhibiting the S c phase have transverse
components of permanent electric dipole moments. A number of molecular
statistical models (including hard rod theories for systems composed of
oblique cylinders) have been developed. (1808) Goossens(189) has proposed a
model composed of ellipsoidal molecules with attractive interactions
arising from anisotropic dispersion forces as well as permanent quadrupole
moments. His calculations show that the interaction between the permanent quadrupole moments can produce a tilting of the molecules, but a
detailed comparison of the predictions with experimental data has yet to be
made.
When the molecule has a strong longitudinal dipole moment, several
modifications of the S c structure have been identified. They are shown in
fig. 5.8.3, and as can be seen they are closely analogous to the SA phases of
polar molecules (fig. 5.6.1). However, the theoretical situation for S c is not
as clear as for SA. Another modification of S c is the chiral S c or the Sc*
phase which will be discussed in 5.10.
5.8 Smectic C
365
366
(arbitrary units)
Fig. 5.8.4. Plot of qxl* versus qzl^ from light scattering measurements in the
smectic C phase of di(4-n-decyloxybenzal)-2-chloro-l-4-phenylene diamine. The
wavevector q relative to the layers is indicated in the inset at the top left-hand
corner. The experimental points lie on an ellipse as expected theoretically. However,
the minor axis of the ellipse does not coincide exactly with the molecular axis,
which is assumed to be inclined at 45 to the layer normal. (After Galerne et a/.(193))
coordinate system such that the projection of the mean molecular direction
on the basal (xy) plane is along x. If the layer displacement along z is
represented by w, we observe that
du/dy = lx,
5.8 Smectic C
367
Here the A terms describe curvature distortions of the smectic planes, the
B terms the distortions of the director when the smectic planes are
unperturbed, and the C terms the coupling between these two types of
distortions. All the coefficients are approximately of the same order of
magnitude as the nematic elastic constants. A term of the type \B(<du/dz)2
may also be included to allow for the compression of the layers, but we
shall neglect it in the present discussion.
The fluctuations of the director are evidently related to fluctuations in Q.
From (5.8.2) and the equipartition theorem, we obtain for a general
wavevector q
const,
(5.8.4)
0)
(ii)
Fig. 5.8.5. Disclinations in the odirector field of smectic C. (a) 5 = 1 wedge
disclination with a radial configuration: (i) sink, (ii) source and (iii) meridian
section of (i). (b) s = 1 wedge disclination with a circular configuration: (i) vortex,
(ii) antivortex and (iii) meridian section of (i); the nails signifying that the director
is tilted with respect to the plane of the paper, (c) (i) s = 1 wedge disclination and
(ii) s = 1 twist disclination.
Parodi and Pershan (MPP), (39) and by others.(194) Leslie, Stewart and
Nakagawa (195) have formulated a general non-linear theory, which in the
static case becomes identical with that proposed by the Orsay group. The
full implications of the dynamical aspects of this theory and its comparison
with the MPP theory have yet to be worked out.
5.8 Smectic C
369
370
(a)
(b)
Fig. 5.8.6. Disposition of the molecules and layers in elliptical domains in (a)
smectic A and {b) smectic C. The mismatch of the molecular orientations in smectic
C gives rise to additional singular lines. (Bourdon, Sommeria and Kleman. (198))
Fig. 5.8.7. Edge dislocations in smectic C at a temperature very close to the smectic
C-smectic A transition. The sample, prepared in the form of a wedge, consists of
two domains of opposite molecular tilts. (Lagerwall and Stebler. (199))
5.8 Smectic C
371
tilt fluctuations is so large that the true critical region, as defined by the
Ginsburg criterion, should be very small and only a mean field behaviour
should be observable. Huang and Viner(202) and Birgeneau et al.(20S) have
demonstrated that this is, in fact, the case, and that the transition is well
described by a mean field expression but with an unusually large sixth
order term.
We write the free energy as
F= Fo + atO2 + b6* + c6% + . . . .
(5.8.6)
0 = 0,
where
T>TCA
3A1/2
1--J
-11/2
- l j , T<TCA,
(5.8.7)
R = (b/3cr\
t0 = b*/ac.
Substituting (5.8.7) in (5.8.6) and making use of the standard relation
we get
fC
C,=
T>T
CA
"
l.C 0 H-^ii(i m 7 J
(5.8.8)
i < i CA
where
A = ^3/2 /[2(3c) 1/2 r3/A2],
x = |^,
r>rCA.
(5.8.9)
The parameter /0 represents the full width at half maximum of the excess
specific heat curve. It also represents the cross-over temperature from
mean field to tricritical behaviour, for
9 - t1/2, t < t0 (mean field)
6 ~ J1/4,
t > t0 (tricritical).
Fig. 5.8.8 presents the experimental data for the tilt angle, specific heat,
and the susceptibility (the last being measured from the magnetoclinic
effect, the magnetic analogue of the electroclinic effect, see 5.10.1) along
with the curves fitted to (5.8.7), (5.8.8) and (5.8.9). It is seen that the
agreement is very satisfactory.
372
1.0
0.6
-0.2
0.2
1.0
T-TC{K)
Fig. 5.8.8. (a) Tilt angle 6 versus temperature (open circles and filled circles
representing two different determinations) in butyloxybenzylidene heptylaniline
(40.7). The solid line is a fit to (5.8.7) with t0 = 1.3 x 10"3. The triangles are the
reciprocal of the susceptibility / in arbitrary units measured for two different
samples, (b) Heat capacity near the C-A transition in 40.7. The dashed curve is the
background scaled from 40.8 and the solid line is the fit to (5.8.8). (After Birgeneau
et a/.(203))
(5.8.10)
5.8 Smectic C
373
the molecular axial ratio. For a second order C-A transition b > 0, and in
the high temperature phase a > 0, and 6 = 0. Clearly for a large enough
compression (i.e., large negative s), there is a finite tilt angle
0 = [-(a + asB)/2b]*
(5.8.11)
(5.8.12)
(5.8.13)
where a = oc0(T TCA), k119 k22 and A:33 are the elastic constants for in-plane
distortions, nx = 0cos(f>9 ny = 0sin(/> and nz&\. Assuming the oneconstant approximation and that the distortions are such that there is no
z dependence of 0 and <fi, the equations of equilibrium become
\k[V20-0(V(f>)2]-(x0-2p03
=0
(5.8.14)
and
= 0.
(5.8.15)
'
(5 8 i6)
- -
where / = 0/0o, = r/0, r = (x +y )*, 0O is the tilt angle at r-> oo, and
o = { k/2(xf is the coherence length. Equation (5.8.16) is of the same
form as the Ginsburg-Pitaevskii equation for a quantized vortex filament
in a superfluid or a magnetic fluxoid in a type II superconductor. The tilt
angle drops to zero at the centre of the singularity, most of the variation
taking place over a distance of the order of <J0 from the centre (fig. 5.8.9).
374
This result appears to account for the experimental observation (206) that
there is an appreciable region near the centre of the + 1 singularity over
which 6 is very nearly zero, i.e., the material is smectic A-like.
5.9 The nematic-smectic A-smectic C multicritical point
The nematic-smectic A-smectic C (NAC) point is the point of intersection
of the N-A, A-C and N - C phase boundaries in a thermodynamic plane say, e.g. in the temperature (T) against pressure (P) or concentration (X)
diagram. It is termed a multicritical point because close to it all three phase
transitions are continuous, and at the point itself the three phases are
indistinguishable. (207) It has been observed in the T-X diagram of binary
liquid crystal mixtures(208) as well as in the P-T diagram of a singlecomponent liquid crystal system.(209) High resolution studies have been
made in both T-X(210) and P-T(209) planes, and it is now well established
that the topology of the phase diagram in the vicinity of this point exhibits
universal behaviour (fig. 5.9.1).
Another type of multicritical point, namely the re-entrant nematicsmectic A-smectic C (RNCA) point has also been observed. In the first
studies,(211~13) it was noted that the singularities in the phase boundaries of
the kind seen near the NAC phase were conspicuously absent near the
point. However, it was demonstrated subsequently (214) that this was
because the RNAC point was far away from the N-I transition, so much
so that the Brazovskii fluctuations had a negligible influence near the
multicritical point and hence the critical region was unobservably small. By
investigating different binary systems and 'tuning' the experimentally
observable critical region it was shown that singularities do appear in the
immediate vicinity of the RNCA point as well. Analysis of the topology of
these phase diagrams confirmed that the RNCA point also exhibits the
same universal behaviour.
Earlier, Hornreich, Luban and Shtrikman (215) had proposed a new type
of higher order critical point, which they called the Lifshitz point, by
considering a Landau-Ginsburg free energy expansion in terms of a scalar
order parameter M for a magnetic system
F = ^ 2 M 2 + 4 M 4 + 6 M 6 + ... + c1(VM)2 + c2(V2M)2.
(5.9.1)
The phase diagram involves the ferromagnetic, paramagnetic and helicoidal phases, the last being the modulated form of the ordered phase
described by the gradient terms. The Lifshitz point occurs when the
coefficients a2 and cx vanish, and is characterized by the fact that the
375
Fig. 5.8.9. Variation of the superfluid order parameter with distance from the
centre of a quantized vortex filament as given by the Ginsburg-Pitaevskii equation
(5.8.16). The same curve describes the variation of the tilt angle in the core region
of the smectic C disclination.(206)
wavevector of the modulated phase increases from zero as the system
moves away from this point. Soon afterwards Chen and Lubensky(216)
predicted that the point at which the nematic, smectic A and smectic C
phases meet and become identical is a possible realization of a Lifshitz
point in a liquid crystal system. A noteworthy feature of their model is that
there is a single order parameter defined as
.8exp(ikT)/>(k)d8fc,
(5.9.2)
-/.
where /?(k) is the Fourier transform of the centre of mass density p(r). The
domain of integration D takes into account the fact that the X-ray
scattering in the S c phase will be a ring defined by (qp qcos <fi, qsin (/>). The
general form of the free energy is written as
87T
C2
+ D[V2y/-n
(n
3
+ Fel d r,
(5.9.3)
where Fel is the Frank elastic energy for distortions of the nematic director.
The coefficients C , D and D are positive and a = ao(T 7^ A )/7^ A . The
type of fluctuations in the nematic phase is thus determined by the
coefficient C , i.e.,
if C > 0, SA-like fluctuations
if CL < 0, Sc-like fluctuations.
376
D
4 3 -
<>
1-
0 /DvM
MX
-1
-2
X-
X*A*
(4 data sets)
l
(a)
_2
*\
1
x\
-1
151.4
150.6
149.8
(b)
275
295
305
Pressure (bar)
315
325
311
and
q\ = \C\/2D9
a = a'0(T-T
378
The transverse mass density fluctuations then become purely quartic. Thus
the model makes the important prediction that there will be a change from
Lorentzian to quartic in the scattering profile near the Lifshitz point.
All these predictions are generally borne out by high resolution X(208 22 2)
' - and light scattering studies. (223"5)
ray( 2i7-i9) calorimetric
Near the A-N transition, the present theory leads to the same
conclusions as de Gennes's model discussed in 5.5.2; the divergent
contributions to k22 and k33 in the nematic phase are given by (5.5.18) and
(5.5.19). Near the C-N transition, all three Frank constants diverge:
T
(5.9.7)
k T
At the Lifshitz point itself, there is a drastic change and Sk33 oc , while dklx
and Sk22 oc In .
Grinstein and Toner(226) applied the renormalization group technique
and presented a dislocation loop model for the NAC point. The most
striking conclusion of this theory is that a biaxial nematic phase N b should
intervene between the N and S c phases. Thus four phases, N, N b , SA and
S c , meet at a point giving rise to a tetracritical point topology. A
fluctuation corrected mean field theory developed by Lubensky(227)
supports this prediction, but points out that it may be difficult to detect the
presence of the N b phase through X-ray and calorimetric measurements.
All experimental studies but one have so far failed to detect the N b phase.
The one study is by Wen, Garland and Wand(228) whose high resolution
specific heat measurements have revealed anomalous variations near the
C-N transition very close to the NAC point, which they suggest may be
related to biaxial fluctuations. An attempt has been made to explain the
universal topology of the phase diagram in the vicinity of this multicritical
point.(258)
5.10 Ferroelectric liquid crystals
5.70.7 The properties of smectic C*
An interesting modification of smectic C is the Sc* phase, which has a twist
axis normal to the layers (fig. 5.10.1 (a)). The possibility of such a structure
being formed by the addition of optically active molecules to the ordinary
\\L
urn ii II
11/111/111]
WWWY
\
379
Helical
pitch, p
1111111111
miiiiui
inn IIui
mmmi
IIIIIIIIII
im/yini
iiii/wmr
mil! mi'
{a)
(b)
Fig. 5.10.1. (a) Helicoidal structure of the ferroelectric smectic C* phase, (b) a
' poled' sample with the helix unwound by an electric field applied normal to the
helical axis.
smectic C was envisaged by Saupe,(190) but the first identification of this
chiral phase in pure optically active compounds was by Helfrich and
Ok (229) J^Q optical properties of this structure are similar to those of a
cholesteric, though there are some obvious differences. In the cholesteric,
the local dielectric properties can be represented by an ellipsoid of
revolution with ea 4= sb = sc, the principal axis Oc being parallel to the
helical axis Oz. In Sc* the local dielectric ellipsoid is triaxial, ea =t= eb 4= sc,
with Oa making an angle 6 with Oz. As far as light propagation along the
twist axis is concerned, Sc* is identical to the cholesteric, but at oblique
incidence some additional features may be expected/ 230 ' 231)
Meyer et al.(232) demonstrated that the Sc* phase has ferroelectric
properties. The first studies were carried out on DOBAMBC (pdecyloxybenzylidene-/7/-amino-2-methylbutylcinnamate) which shows the
following transitions
crystal -
isotropic
Both Sc* and Sj* (the chiral forms of S c and Sx) were shown to be
ferroelectric.
380
where p is the pitch of the undistorted helix and k the 'twist' elastic
constant (which may be expected to vary rapidly with the tilt angle 6), the
helix is completely unwound and the sample is poled (fig. 5.10.1 ()). The
molecules are then aligned in a plane perpendicular to E with a tilt 6 with
respect to the layer normal. When the field is reversed, the polarization P
reverses direction, and because of the coupling between polarization and
tilt, the molecular orientation switches from 6 to 6. As we shall see later,
this effect finds important applications in electro-optic display devices.
The value of the spontaneous polarization in these materials is quite
small, usually between 10 and 1000 nC cm" 2, i.e, about one or two orders
of magnitude less than that for a solid ferroelectric like KH 2 PO 4 .
The coupling between P and 6 manifests itself even above the C*-A
transition: an electric field induces a tilt in the A phase as well. This is called
the electro clinic effect, and was first demonstrated by Garoff and Meyer.(233)
Induced tilt angles as high as 10 have been observed in high polarization
materials. Due to its submicrosecond response and its linear dependence
381
15
-r(K)
Fig. 5.10.2. Variation of the polarization P with relative temperature in the smectic
C* phase of DOBAMBC. The curve represents thefitobtained with the generalized
Landau theory. (After Dumrongrattana and Huang.(236))
15
Fig. 5.10.3. Variation of the tilt angle 6 and the ratio P/0 with relative temperature
in the smectic C* phase of DOBAMBC. The curves represent thefitsobtained with
the generalized Landau theory. (After Dumrongrattana and Huang.(236))
on the applied field, the electroclinic effect has proved to be useful in
making optical modulator devices.
The C*-A transition may be of first or second order/ 234 ' 235) The
polarization, tilt and pitch are, of course, temperature dependent and drop
to zero at the transition. The experimental curves for these three parameters
for DOBAMBC. which exhibits a second order C*-A transition, are
382
10
Is
f 5
15
10
TC.A-T(K)
Fig. 5.10.4. Variation of the pitch with relative temperature in the smectic C* phase
of DOBAMBC. (Ostrovskii et al.(237)). The curve represents the fit obtained with
the generalized Landau theory. (After Dumrongrattana and Huang. (236))
20
T(C)
Fig. 5.10.5. The temperature variation of the field-induced tilt (or the electroclinic
effect) in the smectic A phase of 4-(3-methyl-2-chlorobutanoyloxy)-4 /-heptyloxy
biphenyl. (After Bahr and Heppke. (238))
shown in figs. 5.10.2, 5.10.3 and 5.10.4 and the variation of the electroclinic
effect with temperature in a high polarization material is shown in fig.
5.10.5. The static dielectric constant (for measuring field parallel to the
layers) increases slightly with temperature initially, but drops rapidly close
to 7^*A (fig. 5.10.6). From the dielectric relaxation one can study the two
important director modes, viz, the symmetry-recovering Goldstone mode
and the soft mode, in the vicinity of the C*-A transition. The Goldstone
mode is associated with the fluctuations of the azimuthal angle (<f>) of the
director, and the soft mode with the tilt (9) fluctuations. The former owes
its origin to the helical structure, and hence its frequency and strength
300
0.3 kHz
240 -
180
~"
0.4 kHz
120
0.5 kHz
60 10 kHz
0
1
-
i
2
i
1
i ,
1
r-r c . A (K)
Fig. 5.10.6. The dielectric constant for measuring field parallel to the layers as a
function of temperature in the smectic C* phase of [S]-4'-(2-chloro-4-methylpentanoyloxy)phenylftYtfw-4//-n-decyloxycinnamate.The variation of the dielectric constant with frequency is a consequence of the Goldstone mode relaxation.
(After reference 239.)
vanish in the SA phase (fig. 5.10.7(<z)). As for the soft mode, its frequency
decreases and strength increases on approaching the transition point from
either side (fig. 5.10.7 (b)). The coefficients of viscosity, y^ and y0, associated
with these modes are also strongly temperature dependent (fig. 5.10.8).
By virtue of their symmetry, ferroelectric smectics are piezoelectric.
Polarization can be induced by mechanical shear (2413) and, conversely, an
electric field can produce shear flow. They also possess pyroelectric
properties/ 244 ' 245)
An antiferroelectric smectic phase has been identified recently. (246) The
structure is expected to have an interlayer herringbone arrangement.
5.10,2 The smectic C*-smectic A transition
As the difference in the transition temperatures between the chiral and
racemic forms of the same compound is found to be only of the order of
1 C, it is clear that the transition is driven by intermolecular forces
responsible for the tilted smectic phase and not by the dipole-dipole
384
640
o o o 0 O Oo
oo
#
|320
OOQQ
#
#
0.5
DC
160 -
(a) 0
1.0
1
%
1
-6
10
-4
-2
T- TC*A (K)
0.16
200
0.12
w
<
< 0.08 -
0.04
150
o
100
kHz)
480
o
o
iI
50
o ooc, 6>
b)
-0.5
0.5
1.0
T--T
15
A(K)
Fig. 5.10.7. The variation of the frequency (open circles) and dielectric strength
(filled circles) of (a) the Goldstone mode and (b) the soft mode in the vicinity of the
C*A transition. Material same as in Fig. 5.10.6. (After reference 239.)
10
fe, 1.0
0.1
90
80
70
T(C)
Fig. 5.10.8. The temperature variation of the Goldstone mode and soft mode
viscosity coefficients, y^ and y0, in the smectic C* phase of DOBAMBC. (After
Pozhidayev et a/.(240))
385
^eErj0P,
(5.10.2)
oil
where A = a(T TCA). The first two terms are the leading terms of the usual
Landau expansion, the next three terms describe the electrostatic free
energy and the last term expresses the coupling between P and 6.
Minimizing with respect to P and 0,
and hence,
P = X(E+r,6),
e = t,XE/A,
(5.10.3)
(5.10.4)
^!iy
(5.10.5)
We may draw the following conclusions: (a) from (5.10.5), we see that
the PB coupling produces a divergent component in the dielectric
constant, (b) from (5.10.3), that polarization can be induced by shear stress
in the absence of an electric field; this is the piezoelectric effect discussed
earlier, and (c) from (5.10.4), that the field-induced tilt, or the electroclinic
effect in the SA phase should diverge as the temperature approaches 7^*A.
However, this treatment is inadequate in many respects. For example,
fig. 5.10.5 shows that the dielectric constant does not exhibit a divergence
at all on approaching TC*A. In order to give a better description of the
properties of the Sc* phase, Pikin and Indenbom (248) proposed the following
free energy expression:
F = \a92 + \b6* + \kq292 - A02q + ^- P2-juP9q - CP6.
(5.10.6)
Here a = ao(T To), To being the transition temperature for the corresponding racemate, b > 0 and is temperature independent, k the elastic
constant, s the high temperature dielectric constant (i.e., in the absence of
ferroelectricity), ju and C are the constants describing respectively the
386
k-sju
(5.10.7)
The main predictions of the model are: (i) the transition temperature of the
chiral compound is higher than that of the racemate, (ii) the tilt angle 0 and
hence P (which is proportional to 6) exhibit a power law variation with
temperature with the mean field exponent of 0.5, and (iii) the pitch and the
ratio P/0 are independent of temperature.
While this theory is an improvement over the previous one, its
predictions are still not quite in agreement with observations. The exponent
governing the 6 variation is generally found to be smaller than the mean
field value. Further, the theory is unable to account for the non-monotonic
temperature variation of the pitch and the anomalous behaviour of P/0.
More elaborate models have been proposed(249236) involving additional
phenomenological coefficients, including a sixth order term in 0 and a term
involving P292. A 'generalized' expansion of the form
F = \aO2 + \b6* + \c6 - AqO2 + \Kzq202
+ ^P2-HqP6-CP6-\ClP262
+ \riP*-dq0*
(5.10.8)
wwwwwwwwwww
\\\\\\\\\\\\\\\\\\\\\
387
11 ii mi mini mi
wwwwwwwwwwww
Fig. 5.10.9. The SSFLC cell: The bookshelf geometry of a thin film of smectic C*
sandwiched between two glass plates (a) field ' up' (normal to the plane of the
diagram) and (b) field 'down'.
helix is unwound. The molecules are then arranged in the so-called
'bookshelf geometry, all inclined one way at an angle of about 20 with
respect to the layer normal as illustrated in fig. 5.10.9 (a). The spontaneous
polarization of every layer will therefore point in the same direction (say
upwards). If now a reverse electrical pulse is applied across the film, the
polarization will switch over to the opposite direction (downwards) and
the molecular tilt will change by 26 ~ 40 (fig. 5.10.9 (b)). By using a pair
of polarizers and adjusting the cell thickness so that, in effect, the liquid
crystal film becomes a half-wave plate, high optical contrast can be
achieved. The remarkable feature of the device - which is known as the
SSFLC (surface stabilized ferroelectric liquid crystal) device - is that the
switch on and switch off times are just a few tens of microseconds, i.e.,
about 1000 times faster than the twisted nematic device. For a given cell
thickness, the material parameters that determine the switching speed are
the polarization and viscosity. The system is bistable and has memory as
well.
However, in practice, it is found that the molecules adopt a chevron type
of arrangement(252) rather than the bookshelf geometry and this, of course,
affects the contrast ratio slightly. The mechanism of reorientation from
one stable state to the other on reversing the voltage is a complex one
involving the formation of defects.(252) Though the SSFLC cell is more
difficult to fabricate than the twisted nematic or supertwisted nematic cell,
its very fast switching time makes it extremely useful for large area, high
information content displays.(253~6)
Another phenomenon that has potential applications (257) is the fieldinduced tilt or the electroclinic effect. Unlike the SSFLC device, this effect
does not possess bistability but it has a faster (submicrosecond) response.
By using the same bookshelf geometry and a suitable polarizer and
retarder arrangement, the electroclinic effect can be used for modulating a
light signal with a transmitted intensity linearly proportional to the applied
voltage or as a tunable colour filter.
6
Discotic liquid crystals
389
RO
RO
I = Cu
(g)
M = Cu
(h)
390
Space group
C5HUO
C 7 H 15 O
C 8 H 17 O
C U H 23 COO
P6 2/m 2/m
P6 2/m 2/m
P6 2/m 2/m
C 7 H 15 COO
QH^-O-C.H.-COOf
PVa
P6 2/m 2/m
P^/a
P 2x/a
C2/m
Lattice
parameters (A)
a=
a=
a=
a=
a=
a=
a=
a=
18.95
22.2
23.3
44.9, b
26.3
37.8,/?
51.8, 6
30.7, 6
= 26.4
= 22.2
= 32.6
= 28.4
Temperature
(C)
^80
^80
^80
117
105
100
165
185
(b)
id)
(e)
Fig. 6.1.2. (a)-(e) Columnar phases of disc-shaped molecules.(6) Plan views of the
two-dimensional lattice; ellipses denote discs that are tilted with respect to the
column axis: (a) hexagonal (P6 2/m 2/m); (b) rectangular (P2 1 /a); (c) oblique
(P^; (d) rectangular (P 2 /a); (e) rectangular face-centred, tilted columns (C2/m).
(/) The nematic phase.
long, with the column axis parallel to the axis of the strand/ 2 2 6) The results
of these studies are summarized below.
(a) The correlation length of the two-dimensional lattice of the
hexagonal columnar phase of (C 1 3 H 2 7 COO) 6 -truxene (fig. 6.1.1 ( / ) ) is
greater than 4000 A (or approximately 200 columns), the lower limit being
391
set by the instrumental resolution. (25) Long-range order of the twodimensional lattice is, in fact, to be expected theoretically (see 6.3.1).
Within each column, on the other hand, the flat molecular cores form an
oriented, one-dimensional liquid, i.e., they are orientationally ordered but
translationally disordered. In contrast the flexible hydrocarbon chains are
highly disordered and produce a nearly isotropic scattering pattern. This
striking difference between the ordering of the cores and the tails is also
seen in the X-ray scattering from the columnar mesophases of triphenylene
compounds (fig. 6.1.1 (d)), but is not so apparent in hexa-n-octanoate of
benzene (fig. 6.1.1 (a)), which has a much smaller core. (2324)
Deuterium NMR spectroscopy of the disco tic phase of hexa-n-hexyloxy
triphenylene has led to similar conclusions. (27) Spectra of two selectively
deuterated isotopic species, one in which all aromatic positions are
substituted and the other in which only the a-carbon side chains are
substituted, bring out the difference between the order parameters of the
cores and the tails. Fig. 6.1.3 gives the quadrupole splittings of the
aromatic and the a-aliphatic deuterons versus temperature in the mesophase region. It is seen that the rigid core is highly ordered, the
orientational order parameter s ranging from 0.95 to 0.90, whereas the
a-aliphatic chains are in a disordered state.
These studies emphasize the fact that any realistic theory of the statistical
mechanics of discotic phases cannot treat the molecules as rigid discs, but
has to take into account the conformational degrees of freedom of the
hydrocarbon chains.
(b) Triphenylene hexa-n-dodecanoate exhibits two columnar phases,
designated as D h and D r , the subscripts h and r standing for hexagonal and
rectangular respectively. The D h -D r transition, which is weakly first order,
is associated with a small distortion of the lattice, consistent with a
herringbone arrangement in the rectangular structure with only the core of
the molecule tilted with respect to the column axis. However, high resolution synchrotron X-ray studies on monodomain discotic strands (2324)
have established that the tilt of the molecular core persists in the D h
phase as well, except that the tilts in neighbouring columns are now
rotationally uncorrelated, i.e., they are free to assume different azimuthal
angles with reference to the column axis (neglecting possible orientational
short-range order). Thus the D h -D r transition may be looked upon as an
orientational order-disorder transition.
(c) Hexa-hexyl thiotriphenylene (fig. 6.1.1 (d) with R = SC 6H13) shows
a transition from an hexagonal 'ordered' phase to an hexagonal 'disordered' one. The former is a phase in which there is regularity in the
392
0.40
10
rc-r(c)
Fig. 6.1.3. Quadrupole splittings for the aromatic v and a-aliphatic VQ deuterons
of deuterated hexa-n-hexyloxytriphenylene (THE6) as functions of temperature
(T Tc) in the mesophase region, where Tc is the mesophase-isotropic transition
point. The open circles correspond to measurements on neat THE6-ard 6 and
THE6-ad 12 separately, while the filled circles correspond to a 2:1 mixture of the
two isotopic species. The scale on the upper right-hand side gives the orientational
order parameter of the aromatic part. The curve at the bottom gives the ratio of the
quadrupole splittings for the a-aliphatic and aromatic deuterons (Goldfarb, Luz
and Zimmermann(27)).
stacking of the triphenylene cores in each column, and the latter one in
which the column is liquid-like. X-ray studies using freely suspended
strands,(26) have shown that in the ordered phase there is a helicoidal
stacking of the triphenylene cores within each column, the helical period
being incommensurate with the intermolecular spacing.(28) In addition, a
three-column superlattice develops as a result of the frustration caused by
molecular interdigitation in triangular symmetry. It may be argued that if
there is no intercolumn interaction true long-range translational order
393
RO
OC
R0-
R =C9H19COOR
R
=C12H25-
-C-N
0
.CO
RO
-OR
OR
-coo-
(b)
(a)
Fig. 6.1.4. (a) Cyclotricatechylene hexaesters, the cores of which are cone-shaped
(Malthete and Collet(29)), (b) hexa-(/?-n-dodecyloxybenzoyl) derivative of macrocyclic polyamines which is hollow at the centre (Lehn, Malthete and Levelut(33)).
Both (a) and (b) show columnar mesophases; the latter mesophase has been
described as 'tubular'.
cannot exist within a column because of the Peierls-Landau instability.
The existence of a regular periodicity in the stacking in each column
therefore implies that neighbouring columns must be in register. Thus,
ordered columnar phases, of which several have been reported, can
probably be compared with the highly ordered smectic phases of rod-like
molecules, e.g., smectics B, E, G, H, etc., which possess three-dimensional
positional order. However, more high resolution studies are necessary
before general conclusions can be drawn about the true nature of these
phases.
Columnar mesophases are also formed when the flat core of the
molecule is replaced by a conical one (2932) as in the cyclotricatechylene
hexaesters (fig. 6.1.4(a)). With macrocyclic molecules, which are hollow at
the centre (fig. 6.1.4(6)), the columns are in the form of tubes; these
mesophases have been described as ' tubular \ (33)
The nematic phase (N D) is exhibited by relatively few compounds;
examples are hexakis((4-octylphenyl)ethynyl)benzene (fig. 6.1.1(6)) and
the hexa-n-alkyl and alkoxybenzoates of triphenylene (fig. 6.1.1 (e)). The
N D phase has an orientationally ordered arrangement of the discs with no
long-range translational order (fig. 6.1.2(/)). Unlike the usual nematic of
rod-like molecules, N D is optically negative, the director n now representing
the preferred axis of orientation of the disc normal. The properties of this
phase will be discussed in greater detail in 6.5. A twisted nematic (or
cholesteric) phase, with the helical axis normal to the director, has also
been identified.(34)
394
395
12H25
V 3d
R
V3
C = A + B,
where d is the lattice constant.
The appropriate single particle potential, which depends on the
orientation of the short axis of the molecule as well as the position r of its
centre of mass, may be written in the mean field approximation as
V^x, y9 cos 0) = -Vo P2(cos 8) {s + a<r[cos (A r) + cos (B r) + cos (C r)]},
(6.2.1)
retaining only the leading terms in the Fourier expansion. Here Vo is the
396
(6.2.2)
r0 being the range of interaction which is of the order of the size of the
aromatic core,
= </(cos0)>
(6.2.3)
(6.2 A)
(6.2.5)
Numerical calculations yield three possible solutions to the equations:
(1) s = a = 0
(isotropic phase),
(2) s 4= 0, a = 0 (nematic phase),
(3) s 4= 0,(7 41 0 (hexagonal columnar phase).
For any given values of a and T, the stable phase is the one which
minimizes the free energy (6.2.5). Interpreting a as a measure of the chain
length as in McMillan's model, the phase diagram as a function of a (fig.
6.2.1) reflects the observed trends for a homologous series of compounds.
The temperature range of the nematic phase decreases with increasing a,
and for a > 0.64 the D phase transforms directly to the isotropic phase. It
should be noted that since the wavevectors of the hexagonal lattice satisfy
the relation A + B C = 0, the columnar-nematic transition is always of
first order: a Landau expansion of the free energy contains a nonvanishing cubic term in the order parameter o.
The theory has been extended using a variational principle to solve the
problem with the full potential rather than the one truncated to the first
Fourier component.(44) This method, which is closely analogous to the
extension of McMillan's model by Lee ettf/.,(45)leads to some qualitative
397
Isotropic
1.0
I0-8
Nematic
Columnar
I
0.4
I
0
0.2
0.4
0.6
Model parameter (a)
J
0.8
Fig. 6.2.1. Theoretical plot of the reduced transition temperature against the model
parameter a showing the hexagonal, nematic and isotropic phase boundaries. All
the transitions are of first order.
improvements in the phase diagram for a homologous series. A simpler
mean field version of this theory has been presented which yields essentially
similar results.(46)
For the rectangular lattice, the nature of the phase diagram depends on
the axial ratio b/a.m) When b/a is only slightly different from y/3, the
phase diagram is similar to that for the hexagonal structure. As the
asymmetry of the lattice is increased, one of the density waves disappears
before the other to give rise to a smectic (or lamellar) phase, which, in turn,
transforms to the N D phase at a higher temperature. Evidence of a smecticlike disco tic phase has been reported, (357) but whether this corresponds to
the lamellar phase predicted by theory is a matter for further study.
As mentioned in 6.1, transitions are observed between different types of
columnar phases. In particular, the transition from the rectangular (D r ) to
the hexagonal (Dh) structures has been established experimentally to be an
orientational order-disorder transition. The tilts of the molecular cores
with respect to the column axis in neighbouring columns are ordered or
correlated to give rise to a rectangular packing with a herringbone
structure in the D r phase, whereas they are free to assume different
398
2 \ 9x
dy )
2 \_\ dx
dy
where B and D are the elastic constants for the deformation of the twodimensional lattice in its own plane; ux and uy are the displacements along
x and y at any lattice point, and /r33 is the Frank constant for the curvature
deformation (bending) of the columns. (In the notation of the standard
crystal elasticity theory, B = | ( c n + c12) and D = \{clx c12).) We neglect
here the splay and twist deformations because they give rise to a distortion
of the lattice and involve considerable energy. (51) We also neglect any
contributions from the surface of the sample. Writing the displacement u
in terms of its Fourier components,
H(r) = X>(q)exp(iq-r);
Q
(6.3.2)
399
where L is the length of the columns, L the linear dimension of the lattice
in the xy plane and d its periodicity. Assuming that U > L
<W2> = [kB T/4BXd$\ [l-(d/L%
(6.3.3)
(5>52>53)
The structure is
where X = (ko/Bo)* is a characteristic length.
therefore stable as L -> oo. As is well known from the classical work of
Peierls(54) and Landau (55) , the two-dimensional lattice itself is an unstable
system with <w2> diverging as lnL. ( 5 6 5 7 ) Hence the curvature elasticity of
the liquid-like columns stabilizes the two-dimensional order in the
columnar liquid crystal. This result was first proved by Landau, who
observed: ' Thus bodies having such a structure could in theory exist, but
it is not known whether they do in fact exist in Nature.' (58)
6.3.2 X-ray scattering
The ' structure factor' for the intensity of X-ray scattering may be written
as
5(K)=
The second exponential term on the right-hand side is the familiar
Debye-Waller factor exp( W). Now
400
In Z/
for larger L.
If the surfaces are free <V> will have additional terms, which may be
analogous to those for a two-dimensional lattice. In any case, it is clear that
the mean square fluctuations of the lattice as well as the Debye-Waller
factor may be expected to show a certain dependence on the linear
dimensions of the sample.
6.3.3 Light scattering
The mean square amplitude of the orientational fluctuations of the director
and the corresponding intensity of light scattering may be worked out in a
similar fashion. For example, when the incident and scattered beams are
both polarized in the plane perpendicular to the optic axis (or column axis)
the scattered intensity is given by
Tco*
where q = (ql + qffi and co is the angular frequency of the light.(41) The
scattering is similar to that from smectic A in certain geometries (see
5.3.4), but is, of course, small compared to that from a nematic.
6.3.4 Mechanical instabilities
If the elasticity theory of the columnar phase is developed to the next
higher order, one finds that ux x, uxy and uz z of the linear theory have to
be replaced as follows:
U
x,x^Ux,x
\Ul,x~\Ul,z~Ux,zUz,x
+ Ux z + Uy z)
401
kS3
the columns buckle. This has actually been observed, (60) but there is a
significant discrepancy between theory and experiment which has not been
explained satisfactorily.
6.3.5 Acoustic wave propagation
The basic hydrodynamic equations for the hexagonal columnar phase
(
l T+ 8^
402
where
8
^2
^2
A
^2
+
^2
8
where y/ is the angle between the direction of propagation and the column
axis. We have ignored the coupling between the lattice and the curvature
deformation of the columns. The third mode is a transverse wave whose
polarization is orthogonal to that of second sound. This wave propagates
403
Fig. 6.3.1. Dependence of the sound velocities on the polar angle y/ in the columnar
liquid crystal.
because the two-dimensional lattice can sustain a shear. Its velocity for any
arbitrary direction is given by
K3 = (/)//>)* sin <p.
Fig. 6.3.1 depicts the variation of the velocities of the three waves with
polar angle. This angular dependence can be studied by Brillouin
scattering, as has been demonstrated for smectic A (see fig. 5.3.11). In the
present case, there should be one, two or three pairs of Brillouin
components depending on the orientation, but no experiments have yet
been reported.
404
Fig. 6.4.1. The hexagonal columnar phase with the symmetry elements in the
structure.
Fig. 6.4.2 gives photographs of the columnar liquid crystal appearing in
the isotropic phase as the sample is cooled slowly. The growth pattern
reflects the hexagonal symmetry of the columnar mesophase.
6.4.1 Dislocations
The Burgers vector b of a dislocation in the columnar structure is normal
to the columnar axis n. In general
b = /a + ma',
where / and m are positive or negative integers. For edge dislocations, b is
perpendicular to the dislocation line L. Two types of edge dislocations are
possible: longitudinal edge dislocations (L parallel to n, figs. 6.4.3(a) and
(b)) and transverse edge dislocations (L perpendicular to n, figs. 6.4.3 (c)
and (d)). For screw dislocations b is parallel to L (figs. 6.4.3 (e) and (/)). A
hybrid composed of screw and edge dislocations is shown in fig. 6.4.3 (g).
Longitudinal edge dislocations
In this case, the dislocation line L is along n, the z axis. Let the Burgers
vector be along y. Minimization of the free energy (6.3.1) gives
{B + D)
9 (%
ox\ax
oy
-DA(%*-%*)
=0
oy\ox
oy J
405
Fig. 6.4.2. The discotic liquid crystal appearing in the isotropic phase when a
sample is cooled very slowly. The growth pattern is diagnostic of the hexagonal
symmetry of the columnar structure: (a) Queguiner, Zann and Dubois (62), (b)
Bouligand(61).
406
,M
Fig. 6.4.3. Dislocations in the columnar phase: (a) and (Z?) longitudinal edge
dislocations; (c) and (d) transverse edge dislocations; (e) and (/) screw dislocations ; (g) a hybrid of screw and edge dislocations. It should be noted that the
Burgers vector b = a for (a), (c), (e) and (g), and b = a a' for (b), (d) and (/).
(Bouligand.(61))
and
These are exactly like the equations for edge dislocations in crystals.(63) The
solutions are
B y2\
407
where Ec and rc are the energy and radius of the core. The radial and
angular components of the force of interaction between two edge
dislocations are
Db1b2
fr = 2n(\-v)r'
/ TTrV s i n 2 a '
2n(\ v)r
where v is Poisson's ratio.
Transverse edge dislocations
The dislocation line is now normal to n. Let it be along the x axis, so that
the only component of the displacement that needs to be considered is u y.
Minimization of the free energy yields
dy2
33
dz* '
which is exactly of the same form as that for edge dislocations in smectic A
(see 5.4.2). The solution is
b
8/ ~
1^
33
8z4 '
408
(a)
(b)
Oo>o O nO
oo
o o oo oo
o oo ooo
o o o oo
o
0
o o S o o
o
o o oo
o ooo
o o oo
(c)
Fig. 6.4.4. Disclinations in the columnar phase: (a) and (b) n/3 and n/3
longitudinal wedge disclinations about the sixfold axis L6; (c) and (d) n transverse
wedge disclinations about the binary axes T2 and 62 respectively; (e) n transverse
wedge disclination leading to the formation of walls; (/) two n disclinations at right
angles to each other, one about T2 and the other about 92. (Bouligand.(61))
which is again of the same form as the equation for an edge dislocation in
smectic A. The solution is
b b
uxx = 7 +
4
An
409
Fig. 6.4.5. Photograph indicating the presence of two n disclinations at right angles
to each other as depicted in fig. 6.4.4 (/). (Oswald.(64))
410
(a)
Fig. 6.4.6. Developable domains in the columnar phase, (a) The developable
surface is degenerated into a straight line S common to the planes P. The columnar
axes C form coaxial circles about S. (b) The developable surface D is a cylinder. The
columns (or the layers in the case of smectic A) are a set of parallel and equispaced
involutes of a circle, (c) A Reimann surface generated by half-tangents to a helix.
The columns are normal to the half-tangents. (Bouligand.(61))
parallel to the glass slides and at right angles to each other as depicted in
fig. 6.4.4 (/).
The symmetry of the columnar phase also permits the occurrence of
twist disclinations in the hexagonal lattice and of hybrids consisting of a
twist disclination in the hexagonal lattice and a wedge disclination in the
director field. According to Bouligand these defects are not likely to exist.
411
412
(a)
413
(b)
Fig. 6.5.2. Flow alignment of the director in nematic liquid crystals, (a) For rodshaped molecules the alignment angle 9 with respect to the flow direction lies
between 0 and 45, while (b) for disc-shaped molecules it lies between 90 and
-45 (or equivalent^ between 90 and 135).
414
Fig. 6.6.1. Two examples of molecules which exhibit the biaxial nematic phase:
(a) (bis-1 -(/?-n-decylbiphenyl)-3-(/?-ethoxyphenyl) propane-1,3-dionatocopper
(H))<8i,82) (hereafter referred to as complex A); (b) l,12-te(pentakis((4-pentylphenyl)ethynyl)pheny loxy)dodecane.(83)
The biaxial nematic (Nb) phase was first identified by Yu and Saupe(79) in
a ternary amphiphilic system composed of potassium laurate, 1-decanol
and D 2 O. In such systems the constituent units are molecular aggregates,
called micelles, whose size and shape are sensitive to the temperature and
concentration; the N b phase was found to occur over a range of
temperature/concentration. There are obvious advantages in having a
single-component, low-molar-mass thermotropic N b phase. The sugges-
415
Fig. 6.6.2. Conoscopic figure demonstrating the biaxiality of the nematic phase of
complex A. (8284)
416
(i)
-12 - 6
(a)
(ii)
12 - 2 4 - 1 8 - 1 2 - 6
26 (deg)
-6
12
(ii)
(i)
-12
12
18
24
-6
12
18
24
Fig. 6.6.3. Raw microdensitometer scans of the X-ray intensity plotted against
diffracting angle (26) for magnetically aligned nematic samples: (a) the uniaxial
nematic phase of 80CB at 77 C; (b) the biaxial nematic phase of complex A at
168.5 C, (i) meridional scan (parallel to / / ) , (ii) equatorial scan (perpendicular to
H). M represents the diffraction peaks from the mylar film which covered the
windows of the sample holder and heater assembly. (8586)
(a)
417
(b)
Fig. 6.6.4. Schematic diagram of the molecular order in (a) the uniaxial nematic
phase and (b) the biaxial nematic phase.
Toulouse(118) describes a s ' topological rigidity'. These and other ideas have
yet to be investigated experimentally. The availability of the N b phase in
simple thermotropic systems is likely to make it conveniently possible to
test some of these predictions.
References
Chapter 1
1 F. Reinitzer, Monatsch Chem., 9, 421 (1888).
2 O. Lehmann, Z. Physikal. Chem., 4, 462 (1889). For a brief historical
account of the early work see, H. Kelker and R. Hatz, Handbook of Liquid
Crystals, Verlag Chemie, Weinheim, Chapter 1 (1980).
3 D. Demus, H. Demus and H. Zaschke, Flussige Kristalle in Tabellen, VEB
Deutscher Verlag fur Grundstoffindustrie, Leipzig (1974).
D. Demus and H. Zaschke, Flussige Kristalle in Tabellen, Vol. II, VEB
Deutscher Verlag fur Grundstoffindustrie, Leipzig (1984).
4 G. Friedel, Ann. Physique, 18, 273 (1922).
5 I. G. Chistyakov and W. M. Chaikowsky, Mol. Cryst. Liq. Cryst., 7, 269
(1969).
6 A. de Vries, Mol. Cryst. Liq. Cryst., 10, 219 (1970); in Proceedings of the
International Liquid Crystal Conference, Bangalore, December 1973,
Pramana Supplement-1, p. 93.
7 G. W. Stewart, Discussions of Faraday Soc, p. 982 (1933).
8 A. Saupe, Angew. Chem. Intnl. Edn., 1, 97 (1968).
9 A. D. Buckingham, G. P. Ceasar and M. B. Dunn, Chem. Phys. Letters, 3,
540 (1969).
10 A. J. Leadbetter, in Thermotropic Liquid Crystals (ed. G. W. Gray), p. 1,
Wiley, Chichester (1987).
11 P. S. Pershan, Structure of Liquid Crystal Phases, World Scientific,
Singapore (1988).
12 R. J. Birgeneau and J. D. Litster / . de Physique Lettres, 39, 399 (1978).
R. Pindak, D. E. Moncton, S. C. Davey and J. W. Goodby, Phys. Rev. Lett.
46, 1135(1981).
13 S. Diele, P. Brand and H. Sackmann, Mol. Cryst. Liq. Cryst., 17, 163 (1972)
A. Tardieu and J. Billard, / . Phys. (Paris) Colloq. 37, C3-79 (1976); G.
Etherington, A. J. Leadbetter, X. J. Wang, G. W. Gray and A. Tajbaksh,
Liquid Crystals, 1, 209 (1986).
14 H. Sackmann and D. Demus, Mol. Cryst. Liq. Cryst., 21, 239 (1973).
15 D. Demus and L. Richter, Textures of Liquid Crystals, Verlag Chemie, New
York (1978).
16 G. W. Gray and J. W. Goodby, Smectic Liquid Crystals, Leonard Hill,
London (1984).
418
References: Chapter 1
419
420
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
1
2
3
4
5
6
7
8
9
10
References: Chapter 2
P. A. Winsor, Mol. Cryst. Liquid Cryst., 12, 141 (1971).
G. H. Brown, J. W. Doane and V. D. Neff, A Review of the Structure and
Physical Properties of Liquid Crystals, Butterworths, London (1971).
C. R. Safinya, Lecture Notes for Nato ASI on 'Phase Transitions in Soft
Condensed Matter', Geil, Norway, April 4-14, 1989 (Plenum) (to be
published).
V. Luzzati, A. Tardieu, T. Gulik-Krzywickyi, E. Rivas and F. Reiss-Husson
Nature, 220, 485 (1968).
K. D. Lawson and T. J. Flautt, / . Am. Chem. Soc, 89, 5489 (1967).
C. L. Khetrapal, A. C. Kunwar, A. S. Tracey and P. Diehl, in NMR-Basic,
Principles and Progress, Vol. 9 (eds. P. Diehl, E. Fluck and R. Kosfild),
Springer-Verlag, Berlin (1975).
P. A. Winsor, Chem. Rev., 68, 1 (1968).
J. M. Seddon, Biochim. et Biophys. Ada., 1031, Pl-69 (1990).
C. Robinson, Trans. Faraday Soc, 52, 571 (1956).
D. Chapman, Faraday Soc. Symp., No. 5, 163 (1971).
A. Beguin, J. Billard, F. Bonamy, J. M. Buisine, P. Cuvelier, J. C. Dubois
and P. Le Barny, Mol. Cryst. Liquid Cryst., 115, 9 (1984).
G. W. Gray (ed.) Thermotropic Liquid Crystals, Critical Reports on Applied
Chemistry Vol. 22, Wiley, Chichester (1987), and references therein.
H. Arnold, Z. Phys. Chem. (Leipzig), 226, 146 (1964).
G. W. Gray, in Liquid Crystals and Ordered Fluids, Vol. 2 (eds. J. F.
Johnson and R. S. Porter), p. 617, Plenum (1974).
G. W. Gray, K. J. Harrison and J. A. Nash, Proceedings of the International
Liquid Crystal Conference, Bangalore, December 1973, Pramana Supplement
I, p. 381.
D. Vorlander, Z. Phys. Chem., A126, 449 (1927).
W. L. Mcmillan, Phys. Rev., A6, 936 (1972).
K. Bergmann and H. Stegemeyer, Z. Naturforsch., 34a, 251 (1979).
J. W. Goodby, G. W. Gray, A. J. Leadbetter and M. A. Mazid, Proceedings
of the Conference on Liquid Crystals of One- and Two-dimensional Order,
Garmisch-Partenkirchen, January 1980 (eds. W. Helfrich and G. Heppke),
Springer-Verlag, p. 3 (1980).
Chapter 2
J. D. Bernal and D. Crowfoot, Trans. Faraday Soc, 29, 1032 (1933).
W. R. Krigbaum, Y. Chatani and P. G. Barber, Ada Cryst., B26, 97 (1970).
J. L. Galigne and J. Falgueirettes, Ada Cryst., B24, 1523 (1968).
J. A. Pople and F. E. Karasz, J. Phys. Chem. Solids, 18, 28 (1961).
F. E. Karasz and J. A. Pople, / . Phys. Chem. Solids, 20, 294 (1961).
S. Chandrasekhar, R. Shashidhar and N. Tara, Mol. Cryst. Liquid Cryst.,
10, 337 (1970).
S. Chandrasekhar, R. Shashidhar and N. Tara, Mol. Cryst. Liquid Cryst.,
12, 245 (1972).
J. E. Lennard-Jones and A. F. Devonshire, Proc Roy. Soc, A169, 317
(1939); ibid., A170, 464 (1939).
See, for example, A. R. Ubbelohde, Melting and Crystal and Structure,
Clarendon, Oxford (1965).
See, for example, R. Fowler and E. A. Guggenheim, Statistical
Thermodynamics, Cambridge University Press, Cambridge (1965).
R. H. Wentorf, R. J. Buehler, J. O. Hirschfelder and C. F. Curtiss, / . Chem.
Phys., 18, 1484 (1950).
References: Chapter 2
421
422
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
References: Chapter 2
D. Frenkel, B. M. Mulder and J. P. McTague, Phys. Rev. Letters., 52, 287
(1984).
The approximate symmetry between the phase diagrams for rod-like and
disc-like particles was also noted by K. L. Savithramma and N. V.
Madhusudana, Mol. Cryst. Liquid Cryst., 74, 243 (1981); 90, 35 (1982), who
used the scaled particle theory.
A. Wulf and A. G. De Rocco, / . Chem. Phys., 55, 12 (1971).
G. I. Agren and D. E. Martire, / . Chem. Phys., 61, 3959 (1974).
R. Alben, / . Chem. Phys., 59, 4299 (1973).
G. I. Agren and D. E. Martire, J. de Physique, 36, Cl-141 (1975).
J. P. Straley, Phys. Rev., A10, 1881 (1974).
W. M. Gelbart and B. Barbory, Mol. Cryst. Liquid Cryst., 55, 209 (1979).
M. Born, Sitz d. Phys. Math., 25, 614 (1916).
W. Maier and A. Saupe, Z. Naturforsch., 13a, 564 (1958); ibid., U*, 882
(1959); ibid., 15a, 287 (1960).
For an introduction to intermolecular forces, see J. O. Hirschfelder, C. F.
Curtiss and R. B. Bird, Molecular Theory of Gases and Liquids, Part 3,
Wiley, NY, Chapman and Hall, London (1954).
V. Tsvetkov, Acta Physicochim. (USSR), 16, 132 (1942).
A. Saupe and W. Maier, Z. Naturforsch., 16a, 816 (1961).
P. Chatelain, Bull. Soc. Franc. Miner. Crist., 78, 262 (1955).
S. Chandrasekhar and N. V. Madhusudana, / . de Physique, 30, C4-24
(1969).
N. V. Madhusudana, 'A statistical theory of the nematic phase', thesis,
Mysore University, India (1970).
H. S. Subramhanyam and D. Krishnamurti, Mol. Cryst. Liquid Cryst., 22,
239 (1973).
I. Haller, H. A. Huggins, H. R. Lilienthal and T. R. Mcguire, / . Phys.
Chem., 11, 950 (1973).
Y. Poggi, G. Labrunie and J. Robert, C. R. Acad. Sci., B277, 561 (1973).
C. C. Huang, R. S. Pindak and J. T. Ho, / . de Physique Lettres, 35, L-185
(1974).
J. Cheng and R. B. Meyer, Phys. Rev., A9, 2744 (1974).
S. A. Shaya and H. Yu, / . de Physique, 36, Cl-59 (1975).
J. W. Emsley (ed.) Nuclear Magnetic Resonance of Liquid Crystals, NATO
ASI Series C-141 (1985).
B. Cabane and W. G. Clark, Phys. Rev. Letters, 25, 91 (1970); Sol. State
Commun., 13, 129 (1973).
J. C. Rowell, W. D. Phillips, L. R. Melby and M. Panar, J. Chem. Phys., 43,
3442 (1965).
W. D. Phillips, J. C. Rowell and L. R. Melby, / . Chem. Phys., 41, 2551
(1964).
A. Saupe and G. Englert, Phys. Rev. Letters, 11, 462 (1963).
A. Carrington and G. R. Luckhurst, Mol. Phys., 8, 401 (1964).
A. Saupe, Angew. Chem. Intnl. Edn., 1, 97 (1968); Angew. Chemie, 80, 99
(1968).
A. Saupe, in Magnetic Resonance (eds. C. K. Coogan, N. S. Ham, S. N.
Stuart, J. R. Pitbrow and G. V. H. Wilson), Plenum, New York (1970).
L. C. Snyder, / . Chem. Phys., 43, 4041 (1965).
A. D. Buckingham and K. A. McLauchlan, in Progress in NMR
Spectroscopy, Vol. 2, (eds. J. M. Emsley, J. Feeney and L. H. Sutcliffe),
Pergamon, Oxford (1967).
References: Chapter 2
423
62 G. R. Luckhurst, Quart. Rev., 22, 179 (1968); Mol. Cryst. Liquid Cryst., 21,
125 (1973).
63 G. H. Brown, J. W. Doane and V. D. Neff, A Review of the Structure and
Physical Properties of Liquid Crystals, Butterworths, London (1971).
64 P. Diehl and C. L. Khetrapal, in NMR - Basic Principles and Progress, Vol.
1 (eds. P. Diehl, E. Fluck and R. Kosfeld), Springer-Verlag, Berlin (1969).
65 S. Chandrasekhar and N. V. Madhusudana, Applied Spectroscopy Reviews,
6, 189(1972).
66 J. R. McColl and C. S. Shih, Phys. Rev. Letters, 29, 85 (1972).
67 H. Gasparoux, B. Regaya and J. Prost, C R. Acad. ScL, 272B, 1168 (1971).
68 R. Alben, J. R. McColl and C. S. Shih, Sol. State Commun., 11, 1081 (1972).
69 A. Saupe, Z. Naturforsch., 19a, 161 (1964).
70 M. A. Cotter, Mol. Cryst. Liquid Cryst., 39, 173 (1977).
See also B. Widom, J. Chem. Phys., 39, 2808 (1963).
71 S. Chandrasekhar and N. V. Madhusudana, Mol. Cryst. Liquid Cryst., 17, 37
(1972).
72 S. Chandrasekhar and N. V. Madhusudana, Acta Cryst., A27 303 (1971).
S. Chandrasekhar, N. V. Madhusudana and K. Shubha, Faraday Soc.
Syrnp., no. 5, 26 (1971).
73 R. L. Humphries, P. G. James and G. R. Luckhurst, / . Chem. Soc, Faraday
Trans. 2,68, 1031 (1972).
R. L. Humphries and G. R. Luckhurst, Mol. Cryst. Liquid Cryst., 26, 269
(1974).
74 S. Marcelja, / . Chem. Phys., 60, 3599 (1974).
75 G. R. Luckhurst, in Recent Advances in Liquid Crystalline Polymers (ed.
L. L. Chapoy), Elsevier, London, p. 105 (1986).
76 B. Deloche, B. Cabane and D. Jerome, Mol. Cryst. Liquid Cryst., 15, 197
(1971).
77 R. G. Horn, / . de Physique, 39, 199 (1978).
R. G. Horn and T. E. Faber, Proc. Roy. Soc. {London), A368, 199 (1979).
78 S. Chandrasekhar and N. V. Madhusudana, Mol. Cryst. Liquid Cryst., 24,
179 (1973).
79 G. R. Luckhurst, C. Zannoni, P. G. Nordio and C. Segre, Mol. Phys., 30,
1345 (1975).
J. D. Bunning, D. A. Crellin and T. E. Faber, Liquid Crystal, 1, 37 (1986).
B. Bergersen, P. Palffy-Muhoray and D. A. Dunmur, Liquid Crystal, 3, 347
(1988).
80 S. Jen, N. A. Clark, P. S. Pershan and E. B. Priestley, Phys. Rev. Letters, 31,
1552(1973).
S. Jen, N. A. Clark, P. B. Pershan and E. B. Priestley, / . Chem. Phys., 66,
4635 (1977).
81 K. Miyano, / . Chem. Phys., 69, 4807 (1978).
82 S. N. Prasad and S. Venugopalan, / . Chem. Phys., 75, 3035 (1981).
83 G. W. Gray, Molecular Structure and the Properties of Liquid Crystals,
Academic, London (1962).
84 P. J. Flory, Statistical Mechanics of Chain Molecules, Interscience, New
York (1969).
85 R. L. Humphries, P. G. James and G. R. Luckhurst, Symp. Faraday Soc,
no. 5, 107 (1971).
86 P. Palffy-Muhoray, D. A. Dunmur, W. H. Muller and D. A. Balzarini, in
Liquid Crystals and Ordered Fluids, Vol. 4 (ed. A. C. Griffin and J. F.
Johnson), Plenum, New York, p. 615 (1984).
424
References: Chapter 2
References: Chapter 2
425
426
References: Chapter 3
Chapter 3
C. W. Oseen, Trans. Faraday Soc, 29, 883 (1933).
H. Zocher, Trans. Faraday Soc, 29, 945 (1933).
F. C. Frank, Disc. Faraday Soc, 25, 19 (1958).
A. Anzelius, Uppsala Univ. Arsskr., Mat. Och Naturvet, 1, (1931).
J. L. Ericksen, Arch. Rational Mech. Anal, 4, 231 (I960).
J. L. Ericksen, Trans. Soc. Rheol, 5, 23 (1961).
F. M. Leslie, Quart. J. Mech. Appl Math., 19, 357 (1966); Arch. Rational
Mech. Anal, 28, 265 (1968); Advances in Liquid Crystals, Vol. 4 (ed. G. H.
Brown), p. 1, Academic, New York (1979).
P. C. Martin, O. Parodi and P. S. Pershan, Phys. Rev., A6, 2401 (1972).
J. D. Lee and A. C. Eringen, / . Chem. Phys., 54, 5027 (1971); ibid., 55, 4504
4509 (1971); ibid., 58, 4203 (1973).
H. Schmidt and J. Jahnig, Annals of Physics, 71, 129 (1972).
For a full list of references to papers on continuum theories, see review by
References: Chapter 3
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
427
428
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
References: Chapter 3
Crystals Conference, Bangalore, December 1973, Pramana Supplement I (ed.
S. Chandrasekhar), p. 520.
G. W. Gray, K. J. Harrison and J. A. Nash, Electron. Lett., 9, 130 (1973); in
Proceedings of International Liquid Crystal Conference, Bangalore, December
1973, Pramana Supplement I (ed. S. Chandrasekhar) p. 381.
See also G. W. Gray (ed.) Thermotropic Liquid Crystals, Wiley, Chichester
(1987).
T. J. Scheffer, SID Seminar Lecture Notes, 7.1 (1986).
P. M. Alt and P. Pleshko, IEEE Trans. Electron Devices, ED-21, 146 (1974).
C. M. Waters, V. Brimmell and E. P. Raynes, in Proceedings of the Third
International Display Research Conference, Kobe, Japan (1983), p. 396; Proc.
SID 25/4, 261 (1984).
C. M. Waters, E. P. Raynes and V. Brimmell, Mol. Cryst. Liquid Cryst., 123,
303 (1985).
E. P. Raynes and C. M. Waters, Displays, April 1987, p, 59.
T. J. Scheffer and J. Nehring, Appl. Phys. Lett., 45, 1021 (1984).
W. R. Hoffner and D. W. Berremann, J. Appl. Phys., 53, 8599 (1982).
E. P. Raynes, Mol. Cryst. Liquid Cryst. Lett., 4, 1 (1986); 4, 69 (1987).
D. W. Berremann, Phil. Trans. Roy. Soc, A309, 203 (1983).
V. G. Taratula, A. J. Hurd and R. B. Meyer, Phys. Rev. Lett., 55, 246
(1985).
For a concise review of Freedericksz transition in polymer nematics, see U.
D. Kini, Mol. Cryst. Liquid Cryst., 153, 1 (1987).
F. Lonberg and R. B. Meyer, Phys. Rev. Lett., 55, 718 (1985).
U. D. Kini, / . de Physique, 47, 693 (1986).
C. Oldano, Phys. Rev. Lett., 56, 1098 (1986).
W. Zimmerman and L. Kramer, Phys. Rev. Lett., 56, 2655 (1986).
E. Miraldi, C. Oldano and A. Strigazzi, Phys. Rev. A 34, 4348 (1986).
J. Nehring and A. Saupe, / . Chem. Phys., 54, 337 (1971).
H. P. Hinov and A. I. Derzhanski, / . de Physique Colloq., France 40, C3-308
(1979).
T. V. Korkishko, V. G. Chigrinov, R. V. Galiulin, A. A. Sonin and N. A.
Tikhnomirova, Sov. Phys. Cristallogr., 32, 263 (1987).
C. Oldano and G. Barbero, Phys. Lett., 110A, 213 (1985); / . de Physique
Lett., 46,451 (1985).
G. Barbero and A. Strigazzi, Liquid Crystals, 5, 693 (1989).
C. Oldano, presented at the Int. Topical Meeting on Optics of Liquid
Crystals, Torino, Italy, 14-20, Oct. 1988.
H. P. Hinov, Mol. Cryst. Liquid Cryst., 148, 197 (1987).
A. I. Derzhanski and H. P. Hinov, Phys. Lett., A56, 465 (1976).
M. I. Barnik, L. M. Blinov, T. V. Korkishko, B. A. Umansky and V. G.
Chigrinov, Mol. Cryst. Liquid Cryst., 99, 53 (1983).
S. Faetti and V. Palleschi, / . de Physique, 46, 415 (1985).
N. V. Madhusudana and R. Pratibha, Mol. Cryst. Liquid Cryst., 179, 207
(1990).
S. Faetti, Mol. Cryst. Liquid Cryst., 179, 217 (1990).
G. Barbero, N. V. Madhusudana and C. Oldano, / . de Physique, 50, 2263
(1989).
O. Lehmann, Ann. Physik, 2, 649 (1900).
G. Friedel, Ann. Physique, 18, 273 (1922).
J. Nehring and A. Saupe, / . Chem. Soc, Faraday Trans. II, 68, 1 (1972).
M. Kleman, Points, Lines and Walls, Wiley, Chichester (1983).
References: Chapter 3
63
64
65
66
67
429
430
References: Chapter 3
101 For a review of viscometric studies on the different mesophase types see
R. S. Porter, E. M. Barrall and J. F. Johnson, / . Chem. Phys., 45, 1452 (1966).
102 M. Miesowicz, Nature, 158, 27 (1946).
103 F. M. Leslie, G. R. Luckhurst and H. J. Smith, Chem. Phys. Lett., 13, 368
(1972).
104 V. Tsvetkov, Acta Physicochim. (USSR), 10, 557 (1939).
105 H. Gasparoux and J. Prost, / . de Physique, 32, 65 (1971).
106 P. G. de Gennes, / . de Physique, 32, 789 (1971).
107 F. J. Bock, H. Kneppe and F. Schneider, Liquid Crystals, 3, 217 (1988).
108 R. J. Atkin, Arch. Rational Mech. Anal, 38, 224 (1970).
109 J. L. Ericksen, Trans. Soc. Rheol, 13, 9 (1969).
110 J. Fisher and A. G. Frederickson, Mol. Cryst. Liquid Cryst., 6, 255 (1969).
111 H. C. Tseng, D. K. Silver and B. A. Finlayson, Phys. Fluids, 15, 1213
(1972).
112 U. D. Kini and G. S. Ranganath, Pramana, 4, 19 (1975).
113 B. A. Finlayson, in Liquid Crystals and Ordered Fluids, Vol. 2 (eds. J. F.
Johnson and R. S. Porter), p. 211, Plenum, New York (1974).
114 Ch. Gahwiller, Mol. Cryst. Liquid Cryst., 20, 301 (1973).
115 F. J. O'Neill, Liquid Crystals, 1, 271 (1986).
116 K. Skarp, T. Carlsson, S. T. Lagerwall and B. Stebler, Mol. Cryst. Liquid
Cryst., 66, 199(1981).
T. Carlsson and K. Skarp, ibid, 78, 157 (1981).
117 T. Carlsson, 'Theoretical investigations of the flow properties of nematic
liquid crystals', thesis, Chalmers University of Technology, Goteborg
(1984).
118 P. G. de Gennes, Phys. Letters, 41A, 479 (1972).
119 F. M. Leslie, Mol. Cryst. Liquid Cryst., 1, 407 (1969).
120 P. Pieranski and E. Guyon, Phys. Letters, A 49, 237 (1974).
121 P. Martinoty and S. Candau, Mol. Cryst. Liquid Cryst., 14, 243 (1971).
122 P. J. Barratt and F. M. Leslie, / . de Physique, 40, C3-73 (1979).
123 P. Pieranski, F. Brochard and E. Guyon, / . de Physique, 34, 35 (1973).
See also F. Brochard, Mol Cryst. Liquid Cryst., 23, 51 (1973).
124 L. Leger, Sol. State Comm., 10, 697 (1972); Mol. Cryst. Liquid Cryst.,
24, 33 (1973).
125 M. G. Clark and F. M. Leslie, Proc. Roy. Soc, A361, 463 (1978).
126 C. Z. Van Doom, / . de Physique, 36, CI-261 (1975).
127 C. Z. Van Doom, / . Appl Phys., 46, 3738 (1975); C. Z. Van Doom, C. J.
Gerritsma and J. J. de Klerk in The Physics and Chemistry of Liquid
Crystal Devices (ed. G. Sprokel), p. 95, Plenum Press, New York (1980).
D. W. Berreman, / . Appl. Phys., 8, 101 (1975).
128 P. Chatelain, Acta Cryst., 4, 453 (1951).
129 P. G. de Gennes, C R. Acad. Scl, 266, 15 (1968).
130 Orsay Liquid Crystals Group, / . Chem. Phys., 51, 816 (1969).
131 Orsay Liquid Crystals Group, Phys. Rev. Letters, 22, 1361 (1969).
L. Leger-Quercy, ' Etude experimentale des fluctuations thermiques
d'orientation dans un cristal liquide nematique par diffusion inelastique de
la lumiere', thesis, University of Paris, France (1970).
132 L. D. Landau and E. M. Lifshitz, Statistical Physics, Part I, 3rd edn.,
Pergamon, Oxford (1980).
See also, I. L. Fabelinskii, Molecular Scattering of Light, Plenum, New
York (1968).
133 W. Kast, Angew. Chemie, 67, 592 (1955).
References:
Chapter 3
431
134 E. F. Carr, / . Chem. Phys., 38, 1536 (1963); ibid., 39, 1979 (1963); ibid., 42,
738 (1965); Advances in Chemistry Series, 63, 76 (1967); Mol. Cryst. Liquid
Cryst., 1, 253 (1969).
135 V. N. Tsvetkov and G. M. Mikhailov, Acta Physicochim. (USSR), 8, 77
(1938).
136 Y. Bjornstahl, Z. Phys. Chem., A175, 17 (1933).
137 V. Naggiar, Ann. Physique, 18, 5 (1943).
138 R. Williams, / . Chem. Phys., 39, 384 (1963).
139 G. H. Heilmeier, L. A. Zanoni and L. A. Barton, Proc. IEEE, 56, 1162
(1968).
140 A. P. Kapustin and L. K. Vistin, Kristallografiya, 10, 118 (1965).
G. Elliot and J. G. Gibson, Nature, 205, 995 (1965).
141 P. A. Penz, Phys. Rev. Lett., 24, 1405 (1970).
142 Orsay Liquid Crystals Group, Mol. Cryst. Liquid Cryst., 12, 251 (1971).
143 P. A. Penz, Mol. Cryst. Liquid Cryst., 15, 141 (1971).
144 H. Gruler and G. Meier, Mol. Cryst. Liquid Cryst., 16, 299 (1972).
145 W. Helfrich, / . Chem. Phys. 51, 4092 (1969).
For a concise review of electric field effects see W. Helfrich, Mol. Cryst.
Liquid Cryst., 21, 187 (1973).
146 E. Dubois-Violette, P. G. de Gennes and O. Parodi, / . de Physique, 32, 305
(1971).
147 I. W. Smith, Y. Galerne, S. T. Lagerwall, E. Dubois-Violette and G.
Durand, / . de Physique, 36, Cl-237 (1975).
148 See, for example, L. D. Landau and E. M. Lifshitz, Fluid Mechanics,
Pergamon, Oxford (1966).
149 P. A. Penz and G. W. Ford, Phys. Rev., A6, 414 (1972).
150 W. Greubel and U. Wolff, Appl. Phys. Lett., 19, 213 (1971).
151 L. K. Vistin, Sov. Phys. - Crystallogr., 15, 514 (1970).
152 E. Dubois-Violotte, G. Durand, E. Guyon, P. Manneville and P. Pieranski,
Solid State Phys., Suppl. 41, 147 (1978).
153 W. J. A. Goossens, Adv. in Liquid Crystals, Vol. 3 (ed. G. H. Brown), p. 1,
Academic, New York (1978).
154 See L. M. Blinov, Electro and Magneto-Optics of Liquid Crystals, Nauka,
Moscow (1978).
155 For a concise review of instabilities in nematics and cholesterics, see S.
Chandrasekhar and U. D. Kini in Polymer Liquid Crystals, (eds. A. Cferri,
W. R. Krigbaum and R. B. Meyer), Chapter 8, Academic, New York
(1982).
156 E. Dubois-Violette, C. R. Hebd. Seanc Acad. Sci., Ser. B 273, 923-6
(1971).
157 E. Guyon and P. Pieranski, C R. Hebd. Seanc Acad. Sci., Ser. B 274, 656
(1972).
158 P. Pieranski and E. Guyon, Solid State Commun., 13, 435 (1973).
159 P. Pieranski and E. Guyon, Phys. Rev., A 9, 404 (1974).
160 P. Manneville and E. Dubois-Violette, J. de Physique, 37, 285 (1976).
161 F. M. Leslie, / . de Physique, D 9, 925 (1976).
162 P. G. de Gennes, in Molecular Fluids (eds. R. Balian and G. Weill),
Gordon & Breach, New York p. 373 (1976).
163 E. Dubois-Violette, E. Guyon, I. Janossy, P. Pieranski and P. Manneville,
/ . Mec, 16, 733 (1977).
164 S. Chandrasekhar, Hydrodynamic & Hydromagnetic Stability, Clarendon
Press, Oxford (1961).
432
References: Chapter 4
1
2
3
4
References: Chapter 4
433
434
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
References: Chapter 4
Ondes electromagnetiques dans les Piles de Reusch', thesis, University of
Science and Technology, Lille, France (1972).
M. Aihara and H. Inaba, Optics Commun., 3, 77 (1971).
A. S. Marathay, / . Opt. Soc. Am., 61, 1363 (1971).
R. Nityananda, ' Study of Some Novel Optical Effects in Periodic
Structures', thesis, Bangalore University, India (1975).
See, for example, M. Born and E. Wolf, Principles of Optics, Pergamon,
Oxford (1959).
An analytical solution in the two-wave approximation has been given by
V. A. Belyakov and V. E. Dmitrienko, Sov. Phys. - Solid State, 15, 1811
(1974).
V. E. Dmitrienko and V. A. Belyakov, Sov. Phys. - Solid State, 15, 2365
(1974).
D. Taupin, / . de Physique, 30, C4-32 (1969).
D. W. Berreman and T. J. Scheffer, Phys. Rev. Lett., 25, 577 (1970).
E. Sackmann, S. Meiboom, L. C. Snyder, A. E. Meixner and R. E. Dietz,
/ . Am. Chem. Soc, 90, 3567 (1968).
See, for example, R. W. Wood, Physical Optics, 3rd edn., Macmillan, New
York (1934).
S. Chandrasekhar and J. Shashidhara Prasad, in Physics of the Solid State
(eds. S. Balakrishna, M. Krishnamurthi and B. R. Rao), p. 77, Academic,
London (1969).
P. E. Cladis, A. E. White and W. F. Brinkmann, / . de Physique, 40, 325
(1979).
J. Rault, Phil. Mag., 28, 11 (1973).
J. Rault, Sol State Commun., 9, 1965 (1971).
J. Rault, Phil Mag., 30, 621 (1974).
P. G. de Gennes, C. R. Hebd Seanc. Acad. Sci., Paris, 266B, 571 (1968).
T. J. Scheffer, Phys. Rev., A 5, 1327 (1972).
C. Caroli and E. Dubois-Violette, Sol State Commun., 7, 799 (1969).
Y. Bouligand and M. Kleman, / . de Physique, 31, 1041 (1970).
C. Robinson, J. C. Ward and R. B. Beevers, Disc. Faraday Soc, 25, 29
(1958).
Y. Bouligand, Defects and Textures in Liquid Crystals in Dislocations in
Solids, Vol. 5 (ed. F. R. N. Nabarro) p. 299, North Holland, Amsterdam
(1980).
D. Demus and L. Richter, Textures of Liquid Crystals, Verlag Chemie, New
York, 1978.
G. W. Gray and J. W. Goodby, Smectic Liquid Crystals, Leonard Hill,
Glasgow, 1984.
G. S. Ranganath, Mol. Cryst. Liquid Cryst., 94, 285 (1983).
Orsay Liquid Crystal Group, / . de Physique, 30, C4-38 (1969).
M. Kleman and J. Friedel, / . de Physique, 30, C4-43 (1969).
F. M. Leslie, Proc Roy. Soc, A307, 359 (1968).
P. C. Martin, O. Parodi and P. S. Pershan, Phys. Rev., A 6, 2401 (1972).
T. C. Lubensky, Mol Cryst. Liquid Cryst., 23, 99 (1973).
O. Lehmann, Ann. Physik, 4, 649 (1900).
N. V. Madhusudana and R. Pratibha, Mol. Cryst. Liquid Cryst. Lett., 5, 43
(1987).
F. C. Frank and M. H. L. Pryce, 1958, quoted by C. Robinson, J. C. Ward
and R. Beevers (reference 50).
C. Robinson, Trans. Faraday Soc, 52, 571 (1956).
References: Chapter 4
435
436
References: Chapter 4
References: Chapter 5
437
438
References:
Chapter 5
References:
Chapter 5
439
440
References: Chapter 5
References: Chapter 5
441
442
References: Chapter 5
References: Chapter 5
443
444
References: Chapter 5
References: Chapter 5
445
446
247
248
249
250
251
252
253
254
255
256
257
258
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
References: Chapter 6
Furukawa, A. Kishi, Jap. J. AppL Phys., 28, LI261 (1989), and references
therein.
R. B. Meyer, Mol. Cryst. Liquid Cryst., 40, 33 (1977).
S. A. Pikin and V. L. Indenbom, Sov. Phys. USP, 21, 487 (1978).
B. Zeks, Mol. Cryst. Liquid Cryst., 114, 259 (1984).
T. Carlsson, B. Zeks, A. Levstik, C. Filipic, I. Levstik and R. Blinc, Phys.
Rev., A 36, 1484 (1987).
R. Blinc, B. Zeks, M. Copic, A. Levstik, I. Musevic, I. Drevensek,
Ferroelectrics, 104, 159 (1990).
N. A. Clark and S. T. Lagerwall, AppL Phys. Lett., 36, 899 (1980).
T. P. Rieker, N. A. Clark, G. S. Smith, D. S. Parmar, E. B. Sirota and
C. R. Safinya, Phys. Rev. Lett., 59, 2658 (1987).
K. Skarp and M. Handschy, Mol. Cryst. Liquid Cryst., 165, 439 (1988).
D. Armitage, J. I. Thackara and W. D. Eades, Ferroelectrics, 85, 291
(1988).
S. Matsumoto, H. Hatoh and A. Murayama, Liquid Crystals, 5, 1345
(1989).
J. Dijon in Liquid Crystal Applications and Uses (ed. B. Bahadur), World
Scientific, Singapore (1990).
G. Andersson, I. Dahl, W. Kuczynski, S. T. Lagerwall, K. Skarp and B.
Stebler, Ferroelectrics, 84, 285 (1988).
E. E. Gorodetskii and V. E. Podneks, presented at the 13th International
Liquid Crystal Conference, Vancouver, 22-27 July, 1990.
Chapter 6
O. Lehmann, Z. Physikal. Chem., 4, 462 (1889).
L. Gattermann and A. Ritschke, Ber. Deut. Chem. Ges., 23, 1738 (1890).
D. Vorlander, Ber. Deut. Chem. Ges., 39, 803 (1906).
S. Chandrasekhar, B. K. Sadashiva and K. A. Suresh, Pramana, 9, 471
(1977).
For reviews on discotics, see refs. 4-9 below.
S. Chandrasekhar, Advances in Liquid Crystals, Vol. 5 (ed. G. H. Brown), p.
47, Academic Press, New York (1982).
S. Chandrasekhar, Phil. Trans. Roy. Soc, London, A309, 93 (1983).
A. M. Levelut, / . Chim. Phys., 88, 149 (1983).
C. Destrade, P. Foucher, H. Gasparoux, N. H. Tinh, A. M. Levelut and J.
Malthete, Mol. Cryst. Liquid Cryst., 106, 121 (1984).
A. J. Leadbetter, in Thermotropic Liquid Crystals, Critical Reports on
Applied Chem., Vol. 22 (ed. G. W. Gray), p. 1, John Wiley & Sons,
Chichester (1987); P. S. Pershan, Structure of Liquid Crystal Phases, World
Scientific, Singapore (1988).
S. Chandrasekhar and G. S. Ranganath, Rep. Prog, in Physics, 53, 57 (1990)
A more or less complete bibliography on the subject up to 1984 has been
compiled by H. P. Hinov, Mol. Cryst. Liquid Cryst., 136, 221 (1986).
B. Kohne and K. Praefcke, Chimia, 41, 196 (1987).
B. Kohne and K. Praefcke, Angew. Chem., 96, 70 (1984).
J. C. Dubois, Annals Phys., 3, 131 (1978).
C. Destrade, M. C. Mondon, J. Malthete, / . de Physique, 40, C3-17 (1979).
C. Destrade, M. C. Bernaud, H. Gasparoux, A. M. Levelut and N. H. Tinh,
Proceedings of the International Conference on Liquid Crystals, Bangalore,
December 1979 (ed. S. Chandrasekhar), p. 29, Heyden, London (1980).
References: Chapter 6
447
448
References:
Chapter 6
References: Chapter 6
449
450
References: Chapter 6
Index
anisaldazine
crystal structure, 18
orientational order parameter, 27
surface tension, 81
antiferroelectric short range order in
nematics, 75-80
antiparallel correlations in smectic A,
350-8
see also near neighbour correlations,
smectic A polymorphism
applications
liquid crystal display devices (LCD):
active matrices, 111; colour displays,
111; dye displays, 51, 112; dynamic
scattering mode, 178, 288; multiplexed
displays, 111; storage mode, 288;
supertwisted nematic (STN), 111-12;
surface-stabilized ferroelectric LCD,
380, 386-7; twisted nematic (TN),
106-11; twisted nematic, dynamical
characteristics of, 161-2; twisted
nematic, optical bounce and reverse
twist effects in, 112-13, 167
optical modulator, 381, 387
thermography, 296-7
tunable colour filter, 387
p-azoxyanisole, see PAA
p-azoxyphenetole, see PAP
backflow, 162-7
Bethe approximation, 71
biaxial nematics, 3, 378, 414-17, 449
(reference 87)
biaxiality:
in nematics, see biaxial nematics
in smectic C, 362-3
in smectic C*, 379
molecular, 37, 47-8, 298
bicritical point, 354
binary nematic mixtures, 51
nematic-nematic coexistence, 51
order parameter of individual
components, 51
blue phases, 5-6, 16
Landau theory of, 295-6
optical Kossel diagrams, 292
single crystals, 292
unit cells of blue phases I and II as cubic
networks of disclinations, 294
bond orientational order, 6, 360-2, 3 6 3 ^ ,
see also hexatic phases
Borrmann effect in cholesterics, 232-8
Bragg-Williams approximation, 20
Brillouin scattering, 325-6, 402, 404
Carr-Helfrich instability, 181-2, 286-7
CBOOA
correlation lengths, 343, 347
critical exponents associated with the
smectic A-nematic transition, 347
damping rate of undulation mode, 321-2
Helfrich deformation, 316
influence of twist distortion on the
smectic A-nematic transition
temperature, 305
layer compressibility, 315
orientational order parameter, 309
penetration depth, 315-16, 322
splay and bend constants, 344
chevron patterns, 178, 180, 387
cholesteric pitch
compensated mixtures of right- and lefthanded cholesterics, 218, 248, 276-7
critical divergence of, 349
dependence on
composition (in mixtures) 297-8;
magnetic field strength normal to the
helical axis, 277-80; pressure, 297;
temperature, 296-7
twist per unit length, 97
451
452
Index
cholesteryl nonanoate:
blue phases, 16
Borrmann effect, 236
critical divergence of pitch, 349
molecular structure, 16
transitions, 16
Clausius-Clapeyron equation, 27
conductivity:
electrical, 177, 181
thermal, 202-5
consistency relation
Chang, 72
Krieger-James, 72
Maier-Saupe, 44
continuum theory
of cholesterics:
coefficients of thermal conductivity,
260; distortion by external fields,
277-88; elastic free energy density, 97,
284-5; energetics of defects, 248-58;
flow along the helical axis
(permeation), 270-3; flow normal to
the helical axis, 274-6;
thermomechanical coupling, 258-67;
torque induced by electric field
(electromechanical coupling), 264-7;
torque induced by heat flux (the
Lehmann rotation phenomenon),
262-3; viscosity coefficients, 260
of columnar liquid crystals:
acoustic wave propagation, 401-2;
elastic free energy density, 397;
energetics of defects, 402-8;
fluctuations, 398; light scattering, 400;
mechanical instabilities, 400-1;
Peierls-Landau instability, 398; X-ray
scattering, 399
of nematics (including discotic
nematics):
backflow and kickback effects, 162-7;
distortions due to external fields,
98-117; elastic free energy density, 57,
97, 116; electrohydrodynamic
instabilities, 177-95; energetics of
defects, 117-44; Ericksen-Leslie
equations, 85-94; flexoelectricity,
205-12; flow properties, 144-59,
413-14; Freedericksz effect, static
theory, 98-106; Freedericksz effect,
dynamics of, 161-7; hydrodynamic
instabilities, 155-7, 195-201,413-14;
light scattering, 162-77;
Oseen-Zocher-Frank elasticity
equations, 94-7; Parodi's relation, 94;
periodic distortions in highly
anisotropic media, 113-15; reflexion
of shear waves, 159-61; summary of
Index
cybotactic groups, 2, 4, 83
Darwin's theory of X-ray diffraction, see
optical properties of cholesterics
defects
in cholesterics:
X (chi)-edge disclinations, 252; / (chi)screw disclinations, 249-51;
disclination pairs, 250-1; dislocations,
254-8; edge dislocations, 254-7;
fingerprint textures, 254-6;
Grandjean-Cano pattern, 257-8;
lattice disclinations, 252^4;
pincements, 254-6; screw dislocations,
254
in columnar liquid crystals, 402-11
developable domains, 40911;
disclinations, 408-9; dislocations,
403-8; longitudinal dislocations,
406-7; longitudinal wedge
disclinations, 408; screw dislocations,
408; transverse dislocations, 407;
transverse wedge disclinations, 408
in nematics:
angular forces between disclinations,
140-3; Brochard-Leger walls, 136;
consequences of elastic anisotropy,
13943; core structure, 1434; director
field in the neighbourhood of, 119-20;
disclination loops, 127-8; energies and
interactions, 120-2; Helfrich walls,
135; non-singular structures, 123-6;
schlieren textures in discotic nematics,
412; singular points, 129-30; solitons,
137-9; threads and schlieren textures,
7, 117; twist disclinations, 126-8;
umbilics, 136-7; wedge disclinations,
117-28
in smectic A:
disclinations, 328-9; edge dislocations,
333-7; energies and interactions,
335-7; focal conic textures, 327-33;
screw dislocations, 338
in smectic C:
core structure 3 7 3 ^ ; disclinations in
the odirector field, 368-9; edge
dislocations, 369-70; toric domains,
329
dielectric anisotropy, theory of, 51-7, 76-8
dielectric dispersion, 56
4,4'-di-n-alkoxyazoxybenzenes
dielectric dispersion, 56
odd-even effect, 48-51
structure, 50
see also PAA, PAP
director, definition of, 85-6
disclinations, see defects
453
454
Index
nematics:
Freedericksz transition, 106; principle
of the twisted and supertwisted
nematic devices, 110-13, see also
electrohydrodynamic instabilities,
flexoelectricity
smectic C*:
electroclinic effect, 380, 387; the
surface-stabilized ferroelectric display
device, 386-7; unwinding of helix,
380
electroclinic effect, 380, 382
application as optical modulators and
tunable colour filters, 387
electrohydrodynamic instabilities in
nematics
Carr-Helfrich mechanism, 181^1
chevron pattern, 178, 180
conduction and dielectric regimes, 178,
183
dynamic scattering, 178
flow and orientation patterns, 182
Helfrich's theory, 181-6
influence of flexoelectricity, 210-11
threshold for DC and AC excitation,
186-7, 187-94
Williams domains, 178-82
electromechanical coupling coefficient in
cholesterics, 266-7
electron spin resonance, 39, 146
enantiotropic transition, definition of, 14
end chains, their contribution to
mesophase stability
hard rods with semiflexible tails, 37
influence of chain length on
the columnar-nematic-isotropic phase
sequence, 394-7; the smectic
A-nematic-isotropic phase sequence,
306
odd-even effect in homologous series,
48-51
ordering of the cores and chains in the
columnar mesophase, 391-2
stiffening of the chains as a mechanism
for reentrance, 357-8
Ericksen-Leslie theory, 85-94
equal areas principle, 22-4, 37, 73
ferroelectric liquid crystals, 378-87
applications, 386-7
experimental studies on the polarization,
tilt angle, pitch, dielectric properties,
381^
generalized Landau theory, 385-6
liquid crystalline ferroelectrics as
improper ferroelectrics, 385
rotational viscosities, 384
Index
Freedericksz effect
in nematics
dynamical theory, 161-7; static
theory, 98-101
in smectic C, 365
free energy of elastic deformation, see
continuum theory
free standing monodomain smectic films,
studies on, 360-4
freely suspended monodomain discotic
strands, studies on, 389-93
Ginsburg criterion, 371
Ginsburg-Pitaevskii equation, 373
Goldstone mode in ferroelectric (smectic
C*) liquid crystals, 382-4
Grandjean-Cano pattern, 219, 257-8, 349
hard particle theories
Andrew's method, 37
Flory's lattice model, 36
hard rods with semiflexible tails, 37
Onsager's theory, 31
scaled particle theory, 36-7
stiffening of the tails as a mechanism for
reentrance, 357-8
summary of theories, 37, 411
Zwanzig's model, 31-6
Helfrich deformation in
cholesterics (square grid pattern), 281-9
columnar liquid crystals, 400
smectic A, 314-15
Helfrich's model of permeation, 270-1,
321
Helfrich's theory of electrohydrodynamic
instabilities in nematics, 181-7
Helfrich walls, 135
hexatic phases, 6
normal and tilted types, 301
X-ray studies of the degree of bondorientational ordering, 360^4
homeotropic orientation, 6, 106
homogeneous (or planar) orientation, 6,
104-5
orienting effect of grooves, 104-5
hybrid models: hard rods with a
superposed attractive potential, 60-1
hydrodynamics of liquid crystals
general formulation by Martin, Parodi
andPershan, 319,367-8
hydrodynamics of nematics, cholesterics,
smectics and discotics, see continuum
theory, flow properties
induced discotic phases, 12
induced smectic phases, 358
isotherms. 2 2 ^ . 37. 73
455
456
Index
Index
propagation along the helical axis,
213^5
propagation normal to the helical axis,
247-8
rotatory power, reflectivity and circular
dichroism as functions of wavelength,
pitch and sample thickness, 217-20,
227-32, 239^1
optical textures
cholesterics:
droplets with a /-line of strength 2,
442; fingerprint texture, 255-6;
Grandjean-Cano pattern, 258; helical
configurations of disclination pairs,
143, 249-51
columnar liquid crystals:
texture showing two orthogonal n
disclinations, 410
nematics:
disclination loops, 127-8; schlieren
textures {structures a noyaux), 6, 30,
117-20, 365, 412; singular points,
129-32; threads, 7-8; umbilics,
136-8
smectic A:
batonnets, 30, 333; fan-shaped and
polygonal textures, 9, 331; focal conic
textures, 9, 30, 327-33; stepped drop
(goutte a gradins), 332, toric domains,
329
smectic C:
edge dislocations, 370; toric domains,
329
order parameter
complex order parameter
smectic A, 340; smectic C, 370
orientational order parameter
curves, 40, 45, 75, 77, 303-5, 308-9;
definition, 3841; experimental
determination of <P2>, 39-^0, 308-9;
<.P4> from polarized Raman studies,
48-9; short range order parameter,
745, 77; variation near an interface,
82,211
positional order parameter in melting
theory, 23-4
translational order parameter
amplitude of one-dimensional density
wave (smectic A), 303-5, 307, 340;
density wave in two dimensions
(columnar phase), 395
orientation distribution function in MBBA
derived from Raman studies, 49
Oseen's equation of motion for the
director, 87
Oseen-Zocker-Frank elasticity equations,
94-7
457
PAA
comparison of properties with hard rod
and hybrid model calculations, 37,
60
dielectric constants, 53
dielectric dispersion, 56
dipole moment, 55
electrohydrodynamic distortions, 178
flow birefringence, 69-71
heat of transition and volume change,
26
Leslie coefficients, 150
light scattering studies, 59, 66-8, 83, 172,
176
lines of constant order parameter vs. In V
and In T, 46
magnetic and electric birefringence, 62-5
Miesowicz coefficients, 146
molar volume at the transition, 44
molecular orientation at the free surface,
83
molecular structure, 15
orientational order parameter, 27, 40
PAA -I- cholesteryl nonanoate, Borrmann
effect in, 236;
Poiseuille flow, 150-3
P-T diagram and other pressure studies,
28,46
surface tension, 81
torque in a rotating magnetic field
(Tsvetkov's experiment), 148
transitions, 15
twist elastic constant, 104
see also 4,4/-di-n-alkoxyazoxybenzenes
PAP
heat of transition and volume change,
26
light scattering studies, 59
Miesowicz coefficients, 146,
orientational order parameter, 27
twist elastic constant, 104
see also 4,4/-di-n-alkoxyazoxybenzenes
Parodi's relation, 93, 98
penetration depth, 314, 322, 342, 398
phase diagrams, 28-9, 291, 295, 306, 353-6,
376, 397
planar structure, see homogeneous
orientation
plastic crystals, 22
Poiseuille flow, 148-52, 157-9, 268-9,
270-3
polymorphism in
smectic A, 350-2
smectic C, 364-5
thermotropics, 14-16
pressure studies, 27-9, 46-7, 297, 356, 372,
374-6
458
Index
Index
electroclinic effect, 380
measurements of the tilt angle,
spontaneous polarization, pitch,
dielectric constant and dispersion,
381-3
optical properties, 379
rotational viscosity measurements,
382-3
smectic C*-smectic A transition,
generalized Landau theory, 386; see
also under pretransition effects
smectic liquid crystals: structural
classification of smectics A-K, and the
chiral modifications of some of them,
301
stepped drop (goutte a gradins), 332
supertwisted nematic device, 111-12, see
also under applications, continuum
theory
surface-stabilized ferroelectric liquid crystal
device, 380, 3867, see also under
applications
surface studies:
interferometric study of the free surface
of a smectic, the stepped drop, 332-3
light scattering and optical reflectivity
from free surface 82-3
surface-induced orientational order, 82
surface-induced smectic order 83^4
surface tension measurements, 81-2
variation of the order parameter in the
liquid-vapour transition zone, 82, 211
X-ray reflectivity from free surface, 83^4
tetracritical point, 378
thermography, 2967
threshold
cholesteric-nematic transition, 280
electrohydrodynamic instabilities, 187,
194,211
Freedericksz transition, 99-101
Helfrich deformation in smectics, 315
hydrodynamic instabilities, 196, 201
mechanical instabilities in columnar
liquid crystals, 400-1
smectic C-smectic A transition through
layer compression, 373
square grid pattern in cholesterics, 281
thermal instability, 204
twisted nematic cell, 109-10
translational order parameter, see under
order parameter
tricritical point, 309, 341, 347-8, 371
triple point, 28-9
Tsevetkov's experiment, 145-8
twisted nematic device, 106-15, see also
applications, and under continuum
theory
459
460
location of
critical end point, 355; critical point
353; tricritical point, 354
longitudinal and transverse correlation
lengths and susceptibility near the
smectic A-nematic transition, 343,
346-7
ordering of cores and tails in columnar
phase, 391
reflectivity from free surface showing
surface induced smectic order.
Index
structure factor and Debye-Waller
factor for the columnar structure, 399
structures of columnar phases, 390-3
studies in the vicinity of the NAC
multicritical point, 377-8
tilt angle measurements in smectics C
and C*, 372, 381
XY model, 347, 361, 370
Zwanzig's model, 31-6