non-equilibrium steady state response and … steady state response and fluctuations of sheared...
TRANSCRIPT
Instructions for use
Title Non-Equilibrium Steady State Response and Fluctuations of Sheared Nematic Liquid Crystals
Author(s) Jaka Fajar Fatriansyah
Issue Date 2015-09-25
DOI 10.14943/doctoral.k12016
Doc URL http://hdl.handle.net/2115/59918
Type theses (doctoral)
File Information Jaka_Fajar_Fatriansyah.pdf
Hokkaido University Collection of Scholarly and Academic Papers : HUSCAP
Doctoral Dissertation
Non-Equilibrium Steady State Response and
Fluctuations of Sheared Nematic Liquid Crystals
Jaka Fajar Fatriansyah
Laboratory of Soft Matter Physics
Division of Applied Physics, Graduate School of Engineering
Hokkaido University,
Sapporo, Japan, August, 2015
Non-Equilibrium Steady State Response and
Fluctuations of Sheared Nematic Liquid Crystals
Jaka Fajar Fatriansyah
A dissertation submitted to the Division of Applied Physics,
Graduate School of Engineering,
Hokkaido University
In partial fulfillment of the requirements for the degree of
Doctor of Engineering
Sapporo, Japan, August, 2015
Supervisor : Professor Hiroshi Orihara
2015 年度 学位論文
Non-Equilibrium Steady State Response and
Fluctuations of Sheared Nematic Liquid Crystals
(せん断を印加したネマチック液晶の非平衡
定常応答とゆらぎ)
北海道大学大学院 工学院
応用物理学専攻 ソフトマター工学研究室
Jaka Fajar Fatriansyah
Non-Equilibrium Steady State Response and
Fluctuations of Sheared Nematic Liquid Crystals
Dissertation for Doctor of Engineering at Hokkaido University
Submitted on August 20, 2015
Advisory Committee
Supervisor : Prof. Hiroshi Orihara
Vice Judge : Prof. Hiroshi Akera
Vice Judge : Prof. Yuichi Murai
v
ACKNOWLEDGEMENTS
AlhamdulillahiRobbilA’lamin
The present studies were carried out in the Division of Applied Physics, Faculty of Engineering,
Hokkaido University from 2012 to 2015. The author is indebted to the many people who have been of
assistance in various forms to make this investigation possible. Foremost, the author would like to
express sincere gratitude to Professor Hiroshi Orihara, who afforded an opportunity of studying in
Hokkaido University and guided to the field of soft matter, for his continuous guidance and advice
during the course of the research. The author is also very grateful to Professors Hiroshi Akera and
Yuichi Murai for their suggestions and advice, Professor Toshifumi Hiraoki and Assistant Professor
Yuji Sasaki for the helpful comments. The author is also sincerely indebted to Soft Matter Laboratory
members for their assistance and help during experiments. The author was accepted as a JICA AUN
SEED scholarship student and would like to express gratitude to the JICA AUN SEED, JICA AUN
SEED officers and Hokkaido International Center (Sapporo ) staffs. In addition the author would like
to express gratitude to Ministry of Education, Culture, Sports, Science, and Technology of Japan
because part of this work was supported through Grant-in-Aids for Scientific Research on Innovative
Areas “Fluctuation & Structure” (No.25103006) and for Scientific Research (B) (No. 26289032).
Finally, the author wishes to express deep gratitude to my parents especially my mother, family
specially my sisters and best friends for their support and kindness.
Jaka Fajar Fatriansyah
August 2015
vi
CONTENTS
ACKNOWLEDGEMENTS v
CONTENTS vi
LIST OF ABBREVIATIONS viii
LIST OF FIGURES ix
LIST OF PHYSICAL VARIABLES xvi
CHAPTER 1: INTRODUCTION
1.1 Basic of liquid crystal 1
1.2 Order parameter and director 2
1.3 Continuum theories of NLCs 4
1.4 Director alignment under shear flow 8
1.5 Non-equilibrium steady state response of NLC under shear flow 10
1.6 Objectives and outline 15
CHAPTER 2: NON-EQUILIBRIUM STEADY STATE DYNAMICS OF NEMATIC LIQUID
CRYSTALS UNDER SHEAR FLOW AND MAGNETIC FIELDS
2.1 Background 17
2.2 Derivation of equations of motion for a monodomain NLC in shear flow and magnetic fields
18
vii
2.3 Steady and unsteady states 21
2.4 Non-equilibrium fluctuations of NLCs 35
2.5 Conclusions 44
CHAPTER 3: NON-EQUILIBRIUM STEADY-STATE RESPONSE OF A NEMATIC LIQUID
CRYSTAL UNDER SIMPLE SHEAR FLOW AND ELECTRIC FIELDS
3.1 Background 45
3.2 Experiments 45
3.3 EL theory 57
3.4 Experimental results and discussion 64
3.5 Conclusions 75
CHAPTER 4: ELECTRIC-FIELD-INDUCED FLOW-ALIGNING STATE IN A NEMATIC
LIQUID CRYSTAL
4.1 Background 77
4.2 EL theory in FAS 78
4.3 Experiments 83
4.4 Experimental results and discussion 87
4.5 Conclusions 97
Chapter 5: SUMMARY AND CONCLUSIONS 98
REFERENCES 101
LIST OF PUBLICATIONS 104
viii
LIST OF ABBREVIATIONS
LC Liquid crystals
NLC Nematic liquid crystals
EL Ericksen-Leslie
NESS Non equilibrium steady state
FDR Fluctuation dissipation theorem
ac alternating current
dc direct current
LMWLC Low molecular weight liquid crystals
FAS Flow aligning state
NFAS Non-flow aligning state
ix
LIST OF FIGURES
Figure Page
FIG. 1.1. The comparison of solid, LC and liquid phase. Solid phase
has orientational and positional order while liquid phase does not have
order at al. On the other hand LC phase has positional order.
2
FIG. 1.2. Schematic relation between and n in NLC. 3
FIG. 1.3. Dependence of order parameter S on the temperature. TNI is
the transition temperature between liquid crystal and isotropic phases.
3
FIG. 1.4. The director reorientation towards the application of the
external field.
6
FIG. 1.5. The geometry of NLC under application of shear flow. 9
FIG. 1.6. Reorientation of the director under shear flow. This figure
only depicts the flow aligning state.
10
FIG. 1.7. Simple shear flow is applied in the x direction with a velocity
gradient parallel to the z axis. A weak ac electric field is applied in the
z direction as perturbation to measure the stress response.
11
FIG. 1.8. The shear stress response 2,2 2 for shear rate 10 s-1. 14
FIG. 1.9. The shear stress response 2,2 2 for shear rate 40 s-1. 14
FIG. 1.10. The shear stress respons 2,2 2 for shear rate 80 s-1. 15
FIG. 2.1. Geometry of the system. 19
FIG. 2.2. Relation between 0n , 1e and 2e . 22
x
FIG. 2.3. Stable and unstable states for 5CB in planes parallel to the Bx
– By plane at Bz = 0, 0.2, 0.5 and 1 T with11 s . The director 0n
(bar) is given for the stable states and the rotation axis 3e (arrow) for
the tumbling states. The short bar indicates the end of the director that
lies above the plane.
25
FIG. 2.4. Relaxation rates for the stable fluctuations as a function of
magnetic field yB along 0, , 0y
B 0y
B . At the critical point Bc,
one of the relaxation rates becomes zero. This critical point also marks
the existence of non-conservative forces.
26
FIG. 2.5. Three-dimensional phase diagram for 5CB with1
1 s
. The
unstable region (the tumbling state) is colored.
27
FIG. 2.6. Stable and unstable states for 8CB in planes parallel to the Bx
– By plane at Bz = 0, 0.1, 0.5 and 1 T with11 s . The director 0n
(bar) is given for the stable states and the rotation axis 3e (arrow) for
the tumbling states. The short bar indicates the end of the director that
lies above the plane. The stable region is divided into two regions
indicated by thin red bars and thick blue ones corresponding,
respectively, to complex and real eigenvalues.
28
FIG. 2.7. Three dimension phase diagram for 8CB with11 s . The
unstable region (the tumbling state) is colored.
29
FIG. 2.8. Configuration of ie and t . 31
FIG. 2.9. Evolution of t for 5CB with 0.2, 0.2, 0.5 [T]B .
The rotation axis is given by 3 (0.0829, 0.8337, 0.5458)e .
32
xi
FIG. 2.10. Trajectory of an unstable state in 5CB with initial condition
0,0,1 n and 0.2,0.2,0.5 [T]B .
33
FIG. 2.11. Dependence on initial condition of the trajectory of an
unstable state in 8CB without a magnetic field. The initial orientations
are (a) 0,0.9, 0.4359n and (b) 0, 0.4359, 0.9n .
34
FIG. 2.12. Over-damped profiles of (a) the correlation function and (b)
the integrated response function for 8CB at 0.3, 0.1, 0.2B [T].
38
FIG. 2.13. Over-damped profiles of (a) the correlation function and (b)
the integrated response function for 8CB at 0.3, 0.1, 0.2B [T].
39
FIG. 2.14. Magnetic field dependence of (a) the correlation function
0ij
C and (b) the integrated response function ijR for 5CB in the
vorticity direction 0, , 0y
B . At 0.4634 Ty
B , the off-diagonal
components of R , 12R and 21
R have similar value.
41
FIG. 2.15. Magnetic field dependence of (a) the correlation function
0ij
C and (b) the integrated response function ijR in 5CB for
sin cos ,sin sin , cosB B with 135 , 5 .
43
FIG. 3.1. Molecular structure of 5CB. 46
FIG. 3.2. Rheometer (Physica MCR300, Anton Paar). 47
FIG. 3.3. (a) A simple shear flow is applied along the x axis with a
velocity gradient parallel to the z axis. An electric field is also applied
along the z direction. (b) We define to be the angle between the
director and the x axis.
49
xii
FIG. 3.4. Experimental setup. Electric fields are applied between the
top and bottom plates. For the electro-optical experiment we add a light
source and a microscope to observe the change of transmitted intensity.
51
FIG. 3.5. Dependence of 1,1 on E (a) and 2,2 on 2E (b) at
-110 s and -1
0 100 V mmE . Linear relations are obtained at low
ac electric fields.
52
FIG. 3.6. Molecular structure of CYTOP. 53
FIG. 3.7. Texture of 5CB under crossed polarizers at a shear rate of 10
s-1. There is no disclination.
54
FIG. 3.8. Texture of 5CB under crossed polarizers at a shear rate of 80
s-1. There appear a few disclination lines. The appearance of
disclinations may be associated with the director deviation from the
shear plane though the details are not clear.
54
FIG. 3.9. The experimental setup. 56
FIG. 3.10. Dependence of 0 on 0E calculated from the EL theory.
The angle 0 increases monotonically and tends to saturate when
1
0 200 V mmE .
60
FIG. 3.11. 1,1 as a function of angular frequency before
correction.
64
FIG. 3.12. 2,2 as a function of angular frequency before
correction.
65
FIG. 3.13. as a function of frequency . 66
FIG. 3.14. as a function of frequency . 66
xiii
FIG. 3.15. Frequency dispersion of 1,1 as a function of angular
frequency . The theoretical curves are obtained from Eq. (3.15). A
characteristic non-zero plateau, which is due to the dc electric field,
appears at high frequencies in (b), (c) and (d).
67
FIG. 3.16. Frequency dispersion of 2,2 as a function of angular
frequency . The theoretical curves are obtained from Eq. (3.23). A
plateau appears under dc electric fields as well as in the first-order
response.
68
Fig. 3.17. 0E dependence of the first-order response at very low
frequency (0.23 rad s-1 for the experiment and exactly zero for the
theory), 1,1Re ( 0) . Also shown is the dependence of plateau
height 1,1Re ( ) . For dc electric fields higher than around
-190 V mm , 1,1Re ( ) becomes larger than
1,1Re ( 0) .
70
FIG. 3.18. 0E dependence of the first-order response at a very low
frequency (0.23 rad s-1 for the experiment and exactly zero for the
theory), 2,2Re ( 0) . Also shown is the dependence of plateau
height 2,2Re ( ) .
73
FIG. 3.19. Frequency dispersion of the optical response, which
corresponds to the director response, as a function of angular
frequency . Lines are the experiment and circles are the theory. No
plateau is observed, which shows that the plateau observed in the shear
74
xiv
stress response should be ascribed to the time derivative of the director
but not to 1,1 itself.
FIG. 4.1. The dc electric field dependence of the flow alignment angle
0 in the FAS at shear rates of 10 s-1 and 40 s-1 numerically obtained
for 8CB from Eq. (4.2).
80
FIG. 4.2. (a) Experimental setup. The sample is sheared by using a
parallel-plate rheometer under dc and ac electric fields. (b) The flow is
applied in the x direction, the velocity gradient is in the z direction, and
the electric field is applied in the z direction. The setup is similar with
the experiment in the previous chapter except that we do not use the
optical apparatus.
85
FIG. 4.3. (a) Dependence of 1,1 on E at 0E =100 V mm-1 and (b)
2,2 on 2E at 0E =100 V mm-1 at a shear rate of 10 s-1 at a
frequency of -10.63 rad s . Linear relationships are obtained at low
electric fields. Both are conducted in FAS state since for NFAS the
theoretical results are not available.
86
FIG. 4.4. Frequency dispersions of the first-order harmonic response
1,1( ) at a shear rate of 10 s-1 for several dc electric fields. The
dispersions are clearly distinguishable between NFAS ((a) and (b)) and
FAS ((c)-(f)). Solid lines for the FAS are calculated on the basis of the
EL theory.
89
FIG. 4.5. 0E dependence of the peak or dip frequency of the imaginary
part, peak , where dots are experimental results, a solid line is obtained
91
xv
from calculated frequency dispersion curves using Eq. (4.11), and a
dashed line is 1/ calculated from Eq. (4.7).
FIG. 4.6. 0E dependencies of the first-order response at very low
frequency (0.63 rad s-1 for the experiment and exactly zero for the
theory), 1,1Re ( 0) , for shear rates of 10 and 40 s-1.
92
FIG. 4.7. Frequency dispersions of the second-order response
2,2 ( ) at a shear rate of 10 s-1 for several dc electric fields. The
dispersions are clearly distinguishable between the NFAS ((a) and (b))
and the FAS ((c)-(f)). Solid lines in the FAS are calculated on the basis
of the EL theory.
94
FIG. 4.8. 0E dependencies of the second-order response at a very low
frequency (0.3 rad s-1 for the experiment and exactly zero for
the thory), 2,2Re ( 0) , for shear rates of 10 and 40 s-1.
95
FIG. 4.9. 0E dependencies of 0 at shear rates of 10 and 40 s-1. It is
difficult to distinguish the FAS and NFAS based on 0 .
96
xvi
LIST OF PHYSICAL VARIABLES
n Director
S Order Parameter
df Free energy density related to the deformation of director field
1k Splay elastic constant
2k Twist elastic constant
3k Bend elastic constant
Dielectric anisotropy ( / / )
/ / Dielectric constant parallel to the director
Dielectric constant perpendicular to the director
Diamagnetic Anisotropy ( / / )
/ / Permeability constant parallel to the director
Permeability constant perpendicular to the director
ef Free energy density under an external electric field
0 Dielectric constant in vacuum
E Electric field
B Magnetic field
v Velocity
(e)
Ericksen stress
(visc)
Viscous stress
h Molecular field
1 Rotational viscosity coefficient
2 Irrotational viscosity coefficient
N Rate of change of the director with respect to the background fluid
A Symmetric part of the velocity gradient
W Antisymmetric part of the velocity gradient
xvii
P Pressure
i Leslie coefficients
f Free energy
Shear Rate
Shear Stress
0 Shear stress with no ac electric field
First order responses with respect to the ac electric field for the first
harmonics of
2,0 Second order responses with respect to the ac electric field for the zeroth
harmonics of
2,2 Second order responses with respect to the ac electric field for the second
harmonics of
Tumbling parameter
G Dynamic stability matrix
g Eigen value
ijM Mobility matrix
( )pf t Perturbation
( )rf t Random force
Bk Boltzmann constant
T Temperature
V Volume
t Time
K Regression matrix
( )tC Correlation matrix
tR Integrated response function
Angular frequency
Relaxation time
1,1
1
CHAPTER 1
INTRODUCTION
1.1. Basic of liquid crystal
Liquid crystals (LCs) are intermediate states between conventional liquids and
solid crystals [1]. They have fluidity like liquids and anisotropy like solids at the same
time. The differences between them are depicted in Fig. 1.1. In the solid phase molecules
exhibit both the positional and the orientational orders, while in the liquid phase neither
of them. As shown in Fig. 1.1, LCs generally consist of rod-like molecules, which are
necessary to generate the anisotropy. The molecules tend to point in one direction, which
is designated by a unit vector n called the director. There are various LC phases, which
are characterized by the type of ordering. Roughly speaking, there are three types of LCs.
One is the nematic shown in Fig. 1.1. The cholesteric phase is quite similar to the nematic
in a short scale, but exhibits a twisting of molecules perpendicular to the director in a long
scale. The last one is the smectic, which possesses layers that can slide over one another
like a soap film. Therefore, the smectics have a one-dimensional positional order. There
are many different types of smectic phases. In this dissertation we focused only on the
nematic liquid crystal (NLC). NLC might be characterized as the simplest LC which is
distinguished from the isotropic/liquid only by the long-range orientational order [2].
2
FIG. 1.1. The comparison of solid, LC and liquid phase. Solid phase has
orientational and positional order while liquid phase does not have order at all. On the
other hand LC phase has positional order.
1.2. Order parameter and director
The nematic phase has the uniaxial symmetry hD , which appears as a result of
spontaneous symmetry breaking of the isotropic symmetry Kh of the isotropic phase. The
direction of up cannot be distinguished from down. To quantify the degree of the
uniaxiality we can define order parameter which is non-zero in the nematic phase and
vanishes in the isotropic phase. For nematic liquid crystals consisting of rod-like
molecules, the order parameter S is expressed as follows:
213cos 1
2S , (1.1)
where is the angle between the director and the long axes of each molecule as shown
in Fig. 1.2. The bracket denotes an average over all molecules.
3
FIG. 1.2. Schematic relation between and n in NLC.
According to Eq. (1.1), in the anisotropic phase the order parameter is zero, while in
perfectly ordered phase it is unity. Figure 1.3 shows the typical value for the order
parameter of LC which has values ranging from about 0.3 to 0.9 as a function of
temperature. It should be noted that the order parameter changes discontinuously at the
transition point from the isotropic to nematic phases [3], indicating that it is of first order
phase transition.
FIG. 1.3. Dependence of order parameter S on the temperature. TNI is the
transition temperature between liquid crystal and isotropic phases.
4
We have defined the director before, but it is not rigorous. Here, we define it
rigorously by introducing the tensor order parameter, which is an extension of the scalar
order parameter given in Eq. (1.1):
1
32
ij i j ijS a a , (1.2)
where a is the unit vector parallel to a rod-like molecule. The director n is defined to be
parallel to the unique eigenvector of ijS . The ground state for nematic liquid crystals are
the uniformly aligned state. However, nematic liquid crystals are soft so that the director
can be deformed easily by application of external forces and boundary conditions. At a
constant temperature in the nematic phase the scalar order parameter S is almost constant
for usual nematic liquid crystals. Therefore, it is enough to consider only the director
change. Typical theories on the statistics and dynamics for the director will be presented
in the next section.
1.3. Continuum theories of NLCs
Many important physical phenomena of NLCs such as reorientation of director
under external fields and flow properties can be studied by regarding a NLC as a
continuous medium. Oseen and Zocher developed a static theory in the late 1920s [4].
The theoretical formulation was re-examined by Frank thirty years later with the theory
of curvature elasticity. The dynamical theories were then developed by Anzelius and
Oseen but the formulation of general conservation laws and constitutive equations which
describe mechanical properties of NLC is formulated by Ericksen and Leslie [5-7]. The
5
Ericksen-Leslie (EL) theory is the first widely accepted dynamic theory for NLC. It was
formulated using balance laws with continuum mechanics as foundation and equilibrium
theory as principle. The EL theory assumes that the system is isothermal and
incompressible in its derivation.
Frank [8], Oseen [9] and Zocher [10] derived the following excess free energy
density related to the deformation of director field:
2 2
d 1 2 3
1 1 1
2 2 2f k n n k n k n n , (1.3)
where 1k , 2k and 3k are the splay, twist and bend elastic constants, respectively. From
this equation many results are derived, which can reproduce experimentally observed
phenomena. In this dissertation, however, we treat only cases of spatially uniform director
fields, and therefore we do not further describe the details.
The couplings between the director and external fields are important for
controlling the orientation: for example the coupling with electric fields is widely utilized
for liquid crystal displays. Basically, the couplings originate from the anisotropy such as
dielectric and magnetic anisotropies. Hereafter, we will briefly explain the dielectric
anisotropy, which is closely related to our studies. The dielectric anisotropy is
defined as / / with / / and being the dielectric constant parallel and
perpendicular to the director, respectively. The free energy density under an external
electric field is given by
2
2
e 0 0
1 1
2 2f E n E , (1.4)
where 0 is the dielectric constant in vacuum and E is the applied external electric field.
6
FIG. 1.4. The director reorientation towards the application of the external fields.
Based on Eq. (1.4) we can observe that if dielectric anisotropy is positive, in
order to minimize the free energy of the system, the director n tends to be parallel to the
applied electric field, while for negative , n tends to be perpendicular to the applied
electric field. Most of NLCs have positive dielectric anisotropy except for some NLCs
such as MBBA (Methoxybenzilidene Butylanaline) and PAA (1-methoxy-4-[(4-
methoxyphenyl)-NNO-azoxy] benzene).
Here, we summarize the EL theory, which can describe the dynamics of both the
director ( )n r and the velocity ( )v r . Here, the incompressibility is assumed:
0v . (1.5)
The conservation of linear momentum and angular momentum [6] give the following
equations of motion:
7
(e) (visc)dv
dt x
, (1.6)
and
1 2n h n N n n A , (1.7)
where (e)
in Eq. (1.6) is the Ericksen stress originating from the Frank elastic energy
df and the pressure p:
(e) d
nfp
xn x
. (1.8a)
On the other hand, (visc)
is the viscous stress given by
(visc)
4 1 5 6 2 3A n n n n A n n A n n A n N n N , (1.8b)
where A is the symmetric part of the velocity gradient: 1/ 2A v v and
N is the rate of change of the director with respect to the background fluid:
dnN W n
dt
, (1.9)
with the antisymmetric part of the velocity gradient 1/ 2W v v . The
viscosity coefficients i are called the Leslie coefficients, and there is the so-called
Parodi’s relation among them derived from the Onsager reciprocal relations:
6 5 2 3 . (1.10)
In Eq. (1.7) h is called the molecular field:
f fh
x nn x
, (1.11)
8
where the free energy density f consists of the Frank elastic energy df and the electric
interaction energy density ef . In Eq. (1.7) 1 is the rotational viscosity coefficient and 2
is the irrotational viscosity coefficient, which are related to the Leslie coefficients as
1 3 2 and 2 6 5 .
1.4. Director alignment under shear flow
In this section, we examine the effect of shear flow on the director orientation on
the basis of the EL theory. Hereafter, we assume a mono-domain NLC, in which the
director is independent of the position and confined in the shear plane. Under this
assumption we can put cos ,0,sinn , where is the angle between the director
and the flow direction, as shown in Fig. 1.5. On the other hand, for the simple shear flow
the velocity field is expressed as ,0,0v z , where is the shear rate. The mono-
domain assumption means that there is no contribution from the Frank elastic energy. For
the simple shear flow, the incompressibility (Eq. (1.5)) is automatically satisfied and the
equation of motion for the velocity field (Eq. (1.6)) is also satisfied. The director obeys
Eq. (1.7), which is reduced to
1 1 1
1( cos 2 )
2t
. (1.12)
9
FIG. 1.5. The geometry of NLC under application of shear flow.
In the steady state ( / 0t ), we have
1
1 2cos / , (1.13)
which is called the flow aligning angle, if the angle exists. This can exist for 1 2/ 1 .
For 1 2/ 1 , on the other hand, the director will tumble or find another stable
orientation with the director field being spatially deformed. Thus, under a simple shear
flow, NLCs take either one of two types of states depending on the ratio of 1 to 2 : the
flow aligning state and the non-flow aligning state. Most low molecular mass NLCs
under shear flow adopt the flow aligning state except for a small number of NLCs such
as 8CB (4-n-octyl-cyanobiphenyl) [11]. The steady state of reorientation angle and the
director n is shown in Fig. 1.6.
10
FIG. 1.6. Reorientation of the director under shear flow. This figure only depicts
the flow aligning state.
1.5. Non-equilibrium steady state response of NLC under shear flow
The linear response is a powerful tool to examine thermal fluctuations. The
complete theory of linear response in equilibrium has been thoroughly developed by
Kubo [12] and is widely used. As long as the applied perturbation force is small, the
macroscopic linear response should be connected to the dynamical fluctuation of the
observed system through its correlation function multiplied by a proportionality factor.
This factor simply consists of the temperature of the equilibrium system surrounded by a
thermal bath and the Boltzmann constant. This relation is called the fluctuation-
dissipation relation (FDR) [13, 14]. This relation is a consequence of microscopic
reversibility of the system and is connected to the symmetry of the response function.
This is a monumental achievement in statistical physics. However, for a system far away
from equilibrium, the FDR is generally invalid [15-20]. However, we can have similar
11
but modified relations for some non-equilibrium steady states (NESSs). For example,
Sakaue et al. have derived a FDR for a dumbbell subjected to a simple shear flow and
discussed the symmetry of the correlation and response functions [21]. In this case, non-
conservative forces brought about by the shear flow play a crucial role, which will be
examined for our case of sheared liquid crystals in the next chapter.
The shear flow has a unique property that it easily breaks the time reversal
symmetry and brings fluids far from equilibrium [22]. In general, application of steady
shear flow can make NESSs. Orihara et al. [23] has applied a steady shear flow to a liquid
crystal as a source of non-conservative forces and measured the shear stress response to
small ac electric fields applied along the velocity gradient, as shown in Fig. 1.7. In their
experiment, they use 5CB (4-n-pentyl-4'-cyanobyphenyl) with a positive dielectric
anisotropy.
FIG. 1.7. Simple shear flow is applied in the x direction with a velocity gradient
parallel to the z axis. A weak ac electric field is applied in the z direction as perturbation
to measure the stress response.
12
For an ac electric field 0 cosE t E t with an amplitude 0E and an angular
frequency , the change of shear stress is assumed to be proportional to the square of
electric field, so the shear stress can be expressed as [7]
2
0 2,0 2,2( ) Re[ ]i tt e , (1.14)
where 0 is the shear stress with no ac electric field, 2,0 and
2,2 are the second
order responses with respect to the ac electric field for the zeroth and second harmonics
of , respectively.
Fig. 1.8 shows 2,2 as a function of 2 at a shear rate of 5 s-1. A Debye-type
relaxation is observed, which has already been explained by the EL theory [23]. On the
other hand, at high shear rates (Figs. 1.9 and 1.10) the response deviates from the Debye
type: the real part becomes negative in a certain frequency range and the imaginary part
becomes narrow. In order to examine the discrepancy, they proposed a two mode
coupling model in which there are two modes considered and they are coupled to each
other. It is obvious that if we have only one mode which obeys the usual equation of
motion the response must be of the Debye type. One mode is the orientation change in
the shear plane (in-plane mode) and the other is the out-of-shear-plane mode. The
corresponding angles are expressed by and ' , respectively. In their paper,
however, ' has not been clearly specified. This will be more clarified in the next
chapter. Under an ac electric field, the equations of motion is written as follows: [23]
13
2
1 11 12 1' Re i tda a K e
dt
, (1.15a)
2
21 22 2
'' ' Re i td
a a K edt
, (1.15b)
where 1 is the usual rotational viscosity, is the viscosity coefficient for ,
ija (i,j = 1,2) denote coefficients depending on the shear rate and iK (i = 1,2) represent
the electric force. At equilibrium, i.e., without shear flow ija components are symmetric
because they are derived from the free energy F as 2
12 21/ 'a F a . At non-
equilibrium it is possible that 12 21a a , meaning that non-conservative forces should
appear. The shear stress can be calculated by assuming that it is a linear combination of
and ' . Orihara et al. have analyzed their data in terms of the above set of equations.
Fitted curves are shown by lines in Fig. 1.9 and 1.10, which agree well with the
experimental data. As a result, they found that the eigenvalues of ija become complex,
indicating that 12 21a a . Thus, they revealed that there exist non-conservative forces in
the sheared nematic liquid crystal. A similar response has been observed in an immiscible
polymer blend where one polymer is dispersed as droplets in the other [24].
14
FIG. 1.8. The shear stress response 2,2 2 for shear rate 10 s-1.
FIG. 1.9. The shear stress response 2,2 2 for shear rate 40 s-1.
15
FIG. 1.10. The shear stress response 2,2 2 for shear rate 80 s-1.
In the next chapter we will derive equations of motion with non-conservative
forces from the EL theory and discuss the dynamics of response and fluctuations in a
sheared nematic liquid crystal.
1.6 Objectives and outline
This research aims to investigate the non-equilibrium steady state response of
NLCs. The NLCs are brought to non-equilibrium steady states by application of shear
flow and the linear response of shear stress to applied ac electric fields is mainly studied.
In chapter 2, the response and fluctuations of NLCs under shear flow and magnetic
fields are theoretically studied on the basis of the EL theory. We derive equations of
motion, from which the origin of the non-conservative forces observed experimentally in
Ref. [23] is clarified, and calculate the phase diagram of flow and non-flow aligning states
in the three-dimensional space of the applied magnetic field. In addition, we derive a
16
modified fluctuation dissipation relation valid for the present non-equilibrium steady state
and discuss the role of the non-conservative forces.
In chapter 3, the influence of applying a dc electric field on the shear stress
response in addition to the shear flow is investigated both experimentally and
theoretically. The non-equilibrium response under shear flow is expected to qualitatively
change because the steady orientation of director is modified by the application of dc
electric field. A special feature of this system unveiled in the experiment is discussed on
the basis of Parodi’s relation derived from the Onsager reciprocal relation.
In chapter 4, the electric field-induced transition from the non-flow aligning state
to the flow aligning state is studied by the shear stress response when the dc electric field
is increased. The NLC used in this chapter is different from that used in chapter 3: the
latter takes the flow aligning state, but the former takes the non-flow aligning state, which
can be transformed to the flow aligning state by application of dc electric fields. We
expect that the shear response is sensitive to the transition and so some phenomena
characteristic to the non-equilibrium transition are observed.
Chapter 5 is devoted to conclusions.
17
CHAPTER 2
NON-EQUILIBRIUM STEADY STATE DYNAMICS OF NEMATIC LIQUID
CRYSTALS UNDER SHEAR FLOW AND MAGNETIC FIELDS
2.1. Background
As described in the previous chapter, Orihara et al. [23] found characteristic
features of non-equilibrium steady state response under shear flow in a nematic liquid
crystal (NLC): the real part of the response takes negative values and the peak of the
imaginary part becomes narrow at high shear rates. To explain them, they developed a
two mode coupling model, where one mode is a director fluctuation in the shear plane
(in-plane mode) but the other is not identified (the out-of-shear-plane fluctuation may be
the most plausible choice). Based on this model it was shown that the eigen-values of the
dynamical matrix for the two mode become complex above a certain shear rate, indicating
the existence of non-conservative forces which are brought about by the application of
shear flow to the system. However, the two mode coupling model is entirely
phenomenological. So, in this chapter we derive the equation of motion from the EL
theory, which has been firmly established for nematic liquid crystals.
The EL theory has been used for investigating basic mechanisms and explaining
various phenomena in NLCs. In this chapter we discuss general dynamical properties of
NLCs under shear flow and magnetic fields and demonstrate the existence of non-
conservative forces [25]. In addition, we also derive the modified fluctuation dissipation
relation (FDR) for our system, which gives the relationship between response and
fluctuations at non-equilibrium under shear flow. Here, we apply magnetic fields in
18
addition to the shear flow because without a magnetic field non-conservative forces never
appear, as will be shown in subsequent sections. In the experiment by Orihara et al. it is
thought that the director deviates from the shear plane, and the deviation is necessary for
non-conservative forces to appear, which is caused by the boundary condition at the top
and bottom surfaces. It is so complicated to take into account the boundary condition that
we adopt a mono-domain sample in which the director orientation is independent of
position. Thus, we need magnetic fields to induce the deviation in the mono-domain NLC.
The boundary effects can be regarded as secondary effects when studying the non-
equilibrium fluctuations in aligned states. The EL theory has already been applied to
NLCs under shear flow and magnetic fields [6, 7]. In these theoretical investigations,
however, the magnetic field direction is restricted in the shear plane or along the vorticity
direction. We examine the stability of aligned states in any arbitrary directions of
magnetic field under shear flow.
2.2. Derivation of equations of motion for a monodomain NLC in shear flow and
magnetic fields.
We theoretically consider an incompressible mono-domain NLC subjected to
shear flow and magnetic fields on the basis of the EL theory. As shown in Fig. 1, the flow
direction and the velocity gradient are, respectively, along the x and z axes, that is, the
velocity is expressed as
19
,0,0v z , (2.1)
where is the shear rate.
FIG. 2.1. Geometry of the system.
For a monodomain NLC, the time evolution of the director n as described by the EL
theory from angular momentum balance [26] is
1 2n h n N n n A , (2.2)
with
f f
hx nn x
, (2.3)
a
dnN W n
dt
. (2.4)
20
Here, d
dt denotes the material derivative,
1
2A v v and
1
2W v v are, respectively, the symmetric and anti-symmetric parts of the
velocity gradient, 1 and 2 are viscosity coefficients. N is the rate of change of the
director with respect to the background fluid, and h is the molecular field. The molecular
field is created by the free energy of the interaction with the magnetic field:
2
1 2 1
0 0
1 1
2 2f B n B
, (2.5)
where 0 is the vacuum permeability and is the diamagnetic anisotropy defined as
/ / with / / and being the permeabilities parallel and perpendicular to the
director, respectively. Substitution of Eqs. (2.3) - (2.5) into Eq. (2.2) yields the following
equation:
1
dnn n n
dt G , (2.6a)
with
1
0 1 2G B B W A . (2.6b)
This is the starting equation in this chapter. We make numerical calculations using the
parameters of 5CB (4-cyano-4′-pentylbiphenyl) and 8CB (4′-octyl-4′-cyanobiphenyl).
5CB is stable (aligned) under shear flow without a magnetic field, while 8CB is unstable
(tumbling), because they have, respectively, positive and negative tumbling parameters,
1 2
2 1
[7].
21
2.3. Steady and unsteady states
Depending on the parameters in Eq. (2.6b), we have some stable (steady) and
unstable states under shear flow. We examine the linear stability and make phase
diagrams of steady states in the space of magnetic field. In the steady states, the properties
of fluctuations such as the magnetic field dependence of relaxation time are investigated
in detail. On the other hand, in the unstable states tumbling motion of director appears,
the trajectory of which is numerically obtained.
Steady States
As mentioned above, for steady states, if they exist, Eq. (2.6a) is reduced to the
form 0n n G , which is equivalent to the eigen-value equation
n gnG , (2.7)
where g is the eigen-value and n is the eigenvector. In a similar way, we can rewrite
Eq. (2.6a) as 1 / 0n dn dt n G and obtain
1
dnn n
dt G . (2.8)
where is a constant. The stability of the system can be examined by inserting
0n n n into Eq. (2.8). For 0n , we have 0 0n n G , implying that g in
Eq. (2.7). On the other hand, for n we have
1
d ng n
dt
G . (2.9)
22
Defining two unit vectors 1e and 2e , which form a right-handed system of 0n - 1e - 2e
together with 0n , where n can be written as 1 1 2 2n n e n e . Substitution of this
equation into Eq. (2.9) yields
FIG. 2.2. Relation between 0n , 1e and 2e .
2
1
1
iij j
j
d nM n
dt
, (2.10)
with
1
0 1 2 .
G
W A
ij ij i j
ij i j i j i j
M g e e
g e B e B e e e e
(2.11)
Now we discuss the appearance of non-conservative forces on the basis of
Eq. (2.11). First, the term 1 i je e W in Eq. (2.10) is anti-symmetric with respect to the
exchange of indices i and j as W is also anti-symmetric, though the other terms are
symmetric. The anti-symmetricity is easily proven as follows:
23
T
T T T
i j i j j i j ie e e e e e e e W W W W . (2.12)
Therefore, the matrix M can be asymmetric, for which there should be non-conservative
forces arising from the rotational flow. By introducing the vorticity vector which can be
defined as , , yz xz xyW W W , the anti-symmetric term can be rewritten as
1 1 2 1 0e e n W , indicating that the non-conservative forces disappear when 0n is
perpendicular to , which is in the y direction in our case. This result is related to the
conjecture by Orihara et al. [23], in which the observed non-conservative forces may
come from the director deviation from the shear plane. The steady state may be stable in
the case that the real parts of both eigenvalues of M are positive. As shown above, M is
not symmetric, so the eigenvalues can be complex numbers.
For both 5CB and 8CB, we calculate 0n from Eq. (2.7) for various values of B
and examine its stability. The parameters used for the calculation are as follows. For 5CB
we use 1 0.07738 Pa s , 2 0.08604 Pa s , 71.1 10 at 25 C [27], and for
8CB we use 1 0.0618 Pa s , 2 0.0556 Pa s , 71.1 10 at 37 C [28]. The
shear rate is set to be 1 s-1. It is worth pointing out that Eq. (2.6) can be expressed in terms
of scaled variables:
2
1
' ' ' ''
dnn n B B n
dt
W A , (2.13a)
with
1
0
1
' , ' , ' / , ' / .t t B B
W W A A (2.13b)
24
Thus results for other shear rates can be obtained from those for = 1 s-1.
First, we plot stable and unstable states for 5CB in planes parallel to the Bx – By
plane for Bz = 0, 0.1, 0.5 and 1 T in Fig. 2.3. For stable states, 0n is designated by a bar,
because 0n and 0n are the same due to the uniaxiality of NLC, and the short bars
attached to these bars indicate that the end of the vector lies above the plane. Unstable
states are represented by arrows which indicate the rotation axes in the tumbling state, as
described in the next subsection.
At Bz = 0, the system is stable over all the Bx – By plane including 0B . However,
there are two singular points, designated by red dots on the By axis, where the director
orientation changes discontinuously by /2. Along the By axis the eigen-values and eigen-
vectors of G are easily obtained as:
2 2
2 12 1 2 11,2 1,2
2 2
,0, for 2 2 2
n g
,
1 2
3 3 00,1,0 for yn g B . (2.14)
For each eigenvector the eigenvalues of M are also easily calculated:
2 2
2 12 2 1 2
2 1 0, 2
yB
for 1,2n ,
2 2 2 2
2 1 2 11 2 1 2
0 0, 2 2
y yB B
for 3n . (2.15)
25
FIG. 2.3. Stable and unstable states for 5CB in planes parallel to the Bx – By plane
at Bz = 0, 0.2, 0.5 and 1 T with11 s . The director 0n (bar) is given for the stable states
and the rotation axis 3e (arrow) for the tumbling states. The short bar indicates the end of
the director that lies above the plane.
Note that 2 1 , 2 1 and 2 are all negative for 5CB and so all of the eigenvalues are
real. It is readily understood that 2n is always unstable while 1n and 3n are stable,
26
respectively, for the case y cB B and
y cB B , where 1 2 2
0 2 1 / 2cB
(about 0.4634 T) and 0, , 0cB are the singular points in director orientation. This
critical point shows the transition from the state where non-conservative forces does not
exist (fory cB B ) to the state where the non-conservative force exists (for
y cB B ) .
We examine the dynamical properties around the singular points by calculating
the relaxation rates given by the eigenvalues of M divided by 1 . We show the relaxation
rates as a function of yB along 0, , 0yB 0yB in Fig. 2.4. As the critical point cB
is approached, one of the relaxation rates goes to zero from both sides, as in an equilibrium
phase transition.
FIG. 2.4. Relaxation rates for the stable fluctuations as a function of magnetic
field yB along 0, , 0y
B 0y
B . At the critical point Bc, one of the relaxation rates
becomes zero. This critical point also marks the existence of non-conservative forces.
As zB increases from zero, unstable regions appear from the singular points
(Bz = 0.1 T in Fig. 2.5), expand, merge into one region (Bz = 0.5 T), which then shrinks
27
(Bz = 1 T) and finally disappears. In the unstable region there cannot be two positive
eigenvalues as there are in the stable region. The three-dimensional shape of the unstable
region (the phase diagram) is shown in Fig. 2.5. The unstable region is divided into two
by the singular (critical) points. At high magnetic fields the system is stabilized
irrespective of the orientation of the magnetic fields. For 5CB all the eigen-values are real
in the stable region.
FIG. 2.5. Three-dimensional phase diagram for 5CB with1
1 s
. The unstable
region (the tumbling state) is colored.
For 8CB, the 0n plot and the phase diagram are shown in Figs. 2.6 and 2.7,
respectively.
28
FIG. 2.6. Stable and unstable states for 8CB in planes parallel to the Bx – By plane
at Bz = 0, 0.1, 0.5 and 1 T with11 s . The director 0n (bar) is given for the stable states
and the rotation axis 3e (arrow) for the tumbling states. The short bar indicates the end of
the director that lies above the plane. The stable region is divided into two regions
indicated by thin red bars and thick blue ones corresponding, respectively, to complex
and real eigenvalues.
29
Without a magnetic field it is unstable (strictly speaking, the stability may be marginal
because one of the eigenvalues is purely imaginary), and the application of an
infinitesimal magnetic field in the y direction can stabilize it (see Fig. 2.7). 8CB can have
complex eigenvalues of M, which arise because 2 2
2 1 0 , unlike 5CB for which
2 2
2 1 0 .
In Fig. 2.7, the two regions are distinguished by the thickness and color of the
bars; they are thin red bars for complex eigenvalues and thick blue bars for real ones. The
real and complex eigenvalues give rise to over-damped oscillation (relaxation) and under-
damped oscillation, respectively, around stable states. The over-damped region lies near
the Bx-Bz plane and it expands when the z component of the magnetic field is increased,
as shown in Fig. 2.6. These two kinds of stable states have also been observed
theoretically in the studies of liquid crystalline polymers based on Doi theory [29-31].
FIG. 2.7. Three dimension phase diagram for 8CB with11 s . The unstable
region (the tumbling state) is colored.
30
Some of the above results are in agreement with analytical ones provided by
Ericksen and Leslie [5, 32]. However, their investigations are restricted only to special
cases where the magnetic field is either in or perpendicular to the shear plane.
Unstable (tumbling) state
In the NLC, under the application of shear flow only, 5CB will undergo steady
aligning state and 8CB undergo unstable (tumbling) state although for the latter, tumbling
is only transient state in the reality. Both 5CB and 8CB can exhibit unstable states under
application of shear flow and magnetic fields. In this subsection we investigate unstable
states in detail. The matrix G in Eq. (2.7) has three eigenvalues and we assume that, for
unstable states, one is real and the other two are conjugate complex numbers. This will
be numerically confirmed for both 5CB and 8CB.
From the complex eigenvectors corresponding to the complex eigenvalues, we
can make two mutually orthogonal real vectors, 1e and 2e , which are perpendicular to the
left eigenvector of G corresponding to the real eigenvalue, 3e . If n is in the plane of 1e
and 2e , nG remains in this plane. Therefore, from Eq. (2.6a), /dn dt is also in the plane
of 1e and 2e , meaning that n is confined to this plane.
Putting 1 2cos sinn t e t e t and inserting it into Eq. (2.6a), we have
2 2
1 3 21 22 11 12
1cos sin 2 sin 0,
2
de g g g g
dt
(2.16)
31
where we have used 3 1 2e e e and defined ij i jg e e G . Therefore, we have the
equation of motion for t :
22 112 2
1 21 12
21 12 21 12 22 11
2 2
21 12 21 12 22 11 0
cos sin 2 sin2
1 1 1cos 2 sin 2
2 2 2
1cos 2 2 ,
2
g gdg g
dt
g g g g g g
g g g g g g
(2.17)
where 22 110
21 12
tan 2g g
g g
. The schematic picture is shown in Fig. 2.8.
FIG. 2.8. Configuration of ie and t .
For the system in order to become unstable it is necessary that the following inequality is
satisfied:
2 2
21 12 21 12 22 11g g g g g g , (2.18a)
32
Then the solution of Eq. (2.17) can be given by
2
1 2
0
1tan tan 1
1
AA Bt
A
, (2.19a)
with
21 12
2 2
21 12 22 11
g gA
g g g g
,
21 12
12
g gB
A
. (2.19b)
A typical evolution of t is shown in Fig. 2.9 for 5CB with 0.2, 0.2, 0.5 [T]B .
t monotonically increases with time though it is modulated, indicating that the
director should rotate or tumble around the 3e axis.
FIG. 2.9. Evolution of t for 5CB with 0.2, 0.2, 0.5 [T]B . The rotation
axis is given by 3 (0.0829, 0.8337, 0.5458)e .
33
Figure 2.10 shows a trajectory for the director, which was obtained by numerically
solving Eq. (2.6a) with an initial condition of 0 0, 1, 0n . The trajectory approaches
a circle obtained analytically in Eq. (2.19a) as time passes, indicating that it should be a
stable limit cycle. The rotation axis is calculated to be 3 (0.0829,0.8337, 0.5458)e .
FIG. 2.10. Trajectory of an unstable state in 5CB with initial condition
0,0,1 n and 0.2,0.2,0.5 [T]B .
The eigenvalues of M for 8CB without a magnetic field are purely imaginary. This means
that the trajectory is not a limit cycle and strongly depends on the initial conditions. In
this case ( 2 1/ 1 ), the solution of Eq. (2.2) is given by
2 2
2 1 2 11 1 2 10
22 1
2 1
1 1 1tan tanh tanh tan ,
1 2 1
tt
(2.20a)
and
34
1 2 1 00
2 1
1 cos 2tan tan
1 cos 2t
t
, (2.20b).
where t and t are the polar and azimuthal angles at time t, respectively, with the
polar axis in the y direction, and 0 and 0 are the initial conditions. Trajectories with
different initial conditions are shown in Fig. 2.11. Both are closed, but the shape differs
depending on the initial conditions. Note that these states are easily changed into other
states by weak perturbations such as external fields and boundary effects.
FIG. 2.11. Dependence on initial condition of the trajectory of an unstable state in
8CB without a magnetic field. The initial orientations are (a) 0,0.9, 0.4359n and (b)
0, 0.4359, 0.9n .
35
2.4. Non-equilibrium fluctuations of NLCs
In this section we discuss non-equilibrium steady states in more detail. Here, we
turn our attention to thermal fluctuations in the steady state. In our system, as has been
mentioned, there are non-conservative forces, which produce a non-vanishing probability
current violating the principle of detailed balance, leading to the breakdown of the
fluctuation-dissipation theorem [14]. At the Markovian level of description [33], we can
obtain the following Langevin equation for the director fluctuation n by adding a weak
perturbation ( )pf t and a random force ( )rf t to Eq. (2.10):
( ) ( )
1
p rd nn f t f t
dt
M . (2.21)
Note that n should be regarded as the spatial average of the director fluctuation over
the system volume V or the director fluctuation for zero wavenumber. The random force
( )rf t is assumed to be Gaussian white noise with zero mean:
( ) ( ) ( ) 11 2 1 20, r r r B
i i j ij
k Tf t f t f t t t
V
, (2.22)
where Bk is the Boltzmann constant and T is the temperature. First, let us calculate the
time correlation function ( ) ( ) (0)ij i jC t n t n . The solution of Eq. (2.21) with
( ) 0pf t is given as follows:
( )
1
1t
t pn t e f d
K
, (2.23)
where we have defined a regression matrix 1/ K M . Thus, the correlation matrix
( ) 0t n t n t t TC becomes
36
T1 2
T1 2
0 T( ) ( )
1 2 1 22
1
01
1 2 1 22
1
1
1
(0),
t t p p
t t B
t
t d d e f f e
k Td d e e
V
e
K K
K K
K
C
C
(2.24a)
with
T0
1
(0) Bk Td e e
V
K K
C . (2.24b)
By using the following relation:
T T T0 0 0
Td e e d e e d e e
K K K K K K
I K K , (2.25)
for the static correlation function (0)C , we can obtain
T
1
0 0Bk T
V I KC C K . (2.26)
From Eq. (2.26) we can calculate the static correlation function:
1 11 22 11 22 12 21
22 11 22 12 12 21 11 12 21 22
11 12 21 22 11 11 22 21 12 21
(0)
.
BC
k T
V K K K K K K
K K K K K K K K K K
K K K K K K K K K K
(2.27)
For 0t , the time-translation symmetry yields ( ) ( )t t TC C . On the other hand, since
we have no time reversal symmetry under shear flow, ( ) ( )t t TC C in general case.
37
In order to probe the system response, we apply a step force0F , which is switched
on at t = 0. Solving Eq. (2.21) after averaging, the average evolution of n t , n t ,
is given as
0 ,n t t F R (2.28a)
with
1 0
1t
tt e d
TKR . (2.28b)
The integrated response function tR can be rewritten as
1 1 1
0 0
1 1
t t
t t t t tt e e d e e d e e e
K K K K K K K
R L K L K L K L , (2.29)
where 1
1
L= I (I being the unit matrix) is the mobility matrix, which is symmetric due to
the Onsager reciprocity. It is easily confirmed that Eqs. (2.24a) and (2.29) satisfy a
modified fluctuation dissipation relation:
0 t t k V BC C R Θ/ , (2.30)
with
1 0k V BΘ L KC . (2.31)
Note that Θ becomes the scalar T in equilibrium, where non-conservative forces vanish,
and Eq. (2.30) reduces to the usual fluctuation dissipation theorem. In our case Θ is given
by the following matrix
12 21 11 221
12 21 11 22
1 /0
/ 1Θ L KC
K K K KT
K K K K
, (2.32)
38
from which the above statement can be easily confirmed, as the regression matrix
becomes symmetric without shear flow.
Typical correlation and integrated response functions for the over-damped case of
8CB at 0.3, 0.1, 0.2B [T] are shown in Fig. 2.12, where 0n is calculated to be
0.9516, 0.0882, 0.2942 and we chose the basis vectors of n as
1 0.0924, 0.996, 0e , 2 0.293, 0.0271,0.9557e .
FIG. 2.12. Over-damped profiles of (a) the correlation function and (b) the
integrated response function for 8CB at 0.3, 0.1, 0.2B [T].
39
Both functions show monotonic time dependence. Their diagonal components are not so
different, while the off-diagonal ones are quite different and asymmetric. For the under-
damped case of 8CB at 0.2, 0.2, 0B [T], similar plots are shown in Fig. 2.13, where
0n is calculated to be 0.0715, 0.9956, 0.0590 and we chose the basis vectors of n
as 1 0.9974, 0.0716, 0e , 2 0.0042, 0.0589, 0.9982e .
FIG. 2.13. Over-damped profiles of (a) the correlation function and (b) the
integrated response function for 8CB at 0.3, 0.1, 0.2B [T].
40
All the components of the correlation and response functions show under-damped
oscillations, and the corresponding components are totally different. This may be due to
the large off-diagonal components of Θ .
Finally, we examine critical behaviors such as the divergence of fluctuations. We
have observed the divergence of the relaxation time at 0, , 0cB B in 5CB,
indicating that the fluctuation intensity may diverge as well. Figure 2.14 shows the
magnetic field dependence of 0ij i jC n n divided by /Bk T V along the By axis.
Note that 0n , 1e and 2e change at the critical point: 0 cos , 0, sinf fn ,
1 0, 1, 0e and 2 sin , 0, cosf fe with flow-aligning angle
1
1 2cos / / 2f from the x axis for y cB B , and 0 0, 1, 0n ,
1 1/ 2, 0, 1/ 2e and 2 1/ 2, 0, 1/ 2e for y cB B . For
y cB B , where
there is no non-conservative force as 0n is in the shear plane, there is no cross correlation
12 210 0 0C C , and the fluctuation tilting toward the y direction, 11 0C , increases
as the critical point is approached and diverges at that point, though the fluctuation in the
shear plane, 22 0C , is constant. In contrast, for y cB B , there are non-conservative
forces and, therefore, the cross correlation always exists for any mutually perpendicular
bases 1e and 2e . All the components diverge at the critical point. The corresponding
response function R is shown in Fig. 2.14(b), which is related to 0C by
0 /k V BC R Θ (see Eq. (2.30) with t ). For y cB B , R is exactly the
41
same as 0 / BV k TC as there is no non-conservative force. While fory cB B , all the
components are different from those of 0 /V k TBC and diverge at the critical point.
FIG. 2.14. Magnetic field dependence of (a) the correlation function 0ij
C and
(b) the integrated response function ijR for 5CB in the vorticity direction 0, , 0y
B .
At 0.4634 Ty
B , the off-diagonal components of R , 12R and 21
R have
similar value.
42
It should be noted that both below and above the critical magnetic field the system
is at non-equilibrium though non-conservative forces exist only above the critical point.
The existence of non-conservative forces guarantee that the system is at non-equilibrium,
but they do not always exist for non-equilibrium steady states, as we show for the case
y cB B .
The divergence of 0C and R also occurs at the transition point from the
aligned state to the tumbling state, as shown in Fig. 2.15 for 5CB, where
sin cos ,sin sin ,cosB B with 135 , 5 . The discrepancy between
0 /V k TBC and R is not large as the direction of magnetic field is nearly in the
shear plane, that is, the director is also nearly in the shear plane, resulting in small non-
conservative forces.
43
FIG. 2.15. Magnetic field dependence of (a) the correlation function 0
ijC and
(b) the integrated response function ijR in 5CB for
sin cos ,sin sin , cosB B with 135 , 5 .
44
2.5 Conclusions
Using the EL theory, phase diagrams in the three-dimensional space of magnetic
field at a constant shear rate were obtained for 5CB and 8CB which have different flow
properties. An analytical solution was given for the tumbling state, in which the director
rotates about an axis determined as a left eigenvector of the matrix G that governs the
motion of the director. It was also shown that non-conservative forces caused by shear
flow appear when magnetic fields are applied, so that the director points out of the shear
plane. The non-conservative forces experimentally observed in the response function of
5CB [23] are thought to originate from a similar mechanism. Using the Langevin equation,
the orientational fluctuations of the director were examined in detail by calculating the
time correlation function and the response function, which are related to each other
through a modified fluctuation dissipation relation. Critical behaviors such as the
divergence of the relaxation time and fluctuations were found at the critical points where
there is a discontinuous change in director orientation and at the boundary between the
stable and unstable states.
45
CHAPTER 3
NON-EQUILIBRIUM STEADY-STATE RESPONSE OF A NEMATIC LIQUID
CRYSTAL UNDER SIMPLE SHEAR FLOW AND ELECTRIC FIELDS
3.1. Background
As discussed in the previous chapter, NLCs can be easily taken to non-equilibrium
states by applying external forces, so they have been widely used to investigate various
types of non-equilibrium phenomena [34]. Director can easily be controlled by external
fields such as electric field [35]. Experimentally, the response function in the sheared
NLC was obtained by applying a small ac electric field to measure the corresponding
shear stress response. A characteristic response, which is originated from the non-
conservative force, was clearly observed. In this chapter, we investigate both theoretically
and experimentally the influence of a dc electric field on the shear stress response in
addition to the shear flow. The application of dc electric field is expected to change the
steady orientation of the director [36-38], bringing about qualitative changes in response
[39].
3.2. Experiments
In the experiment, we used the NLC 5CB (4-n-pentyl-4'-cyanobiphenyl; Tokyo
Chemical Industry) without any further treatment. 5CB is one of the commonly used low
molecular weight liquid crystals (LMWLC) with the chemical formula C18H19N.
46
Molecular structure of 5CB is shown in Fig. 3.1 below. The material is the same with the
one that was used by Orihara et al. [23]
FIG. 3.1. Molecular structure of 5CB.
In general 5CB exhibits transitions from crystal to nematic phases at 18 C and
from nematic to isotropic phases at 35 C. Near the transition temperature physical
quantities such as the viscosity coefficients are quite sensitive to the temperature change.
Therefore, we conducted experiments at 25 C around the center of the nematic phase.
Measurements were carried out with a parallel plate rheometer (Physica MCR300, Anton
Paar) at room temperature (25 C). The rheometer is shown in Fig. 3.2. The diameter of
the rotating plate and the gap between the two parallel plates are 35 and 0.2 mm,
respectively. The rotating plate and the bottom plate were not treated with alignment
surfactant because we thought that the electric field aligned the director and the effect of
the boundaries were limited to the regions just near the boundaries.
47
FIG. 3.2. Rheometer (Physica MCR300, Anton Paar).
For the parallel plate rheometer, the shear rate is defined at the edge of the upper
plate and the shear stress at the corresponding shear rate is calculated from the mechanical
torque by assuming that the sample is a Newtonian fluid. This means that the apparent
shear stress obtained from the experiment is the average over the sample subjected to
different shear rates. Instead we can use a cone plate to avoid the position dependence of
shear rate, but in this case the electric field becomes radially position-dependent. This
situation is undesirable for our purpose.
As described above, the measured stress is averaged one. So, we average the
theoretically obtained stress to compare the experiment and theory. The shear rate at r
48
from the center for the parallel plate geometry is given as rh
, where is the angular
velocity of the rotating plate and h is the gap between the upper and lower plate. The force
exerted on the plate between r r dr is 2dF rdr , and so the corresponding torque
is given by22dM rdF r dr . By integrating this we can obtain the total torque as
R3
2 2
0 0
2 2
Rh
M r r dr d
, (3.1)
where R is the shear rate at the edge. When the sample is Newtonian, i.e., , the
torque becomes 3 34 3/ 2 / / 2 /R R RM h h . By comparing these, we get
the apparent shear stress, which correspond to the experimentally obtained one,
2
3
R 0
4 R
R d
. (3.2)
In Fig. 3.3, we show the relation among the flow direction, the velocity gradient
and the electric field, which is applied to the sample by using a synthesizer
(Model 1940, NF) and a high-voltage amplifier (Model 4005, NF). In our measurements,
dc electric fields are applied in addition to a weak ac electric field which is used with the
purpose to probe the stress response. The total field applied is thus 0 cosE E t .
However, migration of ions in the NLC sometimes becomes a problem and reduces the
electric field inside the cell. To avoid this, we use a high frequency ac electric field
modulated by the sum of the dc and ac electric fields:
02 cos cos cE t E E t t , (3.3)
49
where c is the angular frequency of the carrier signal. The induced shear stress is
proportional to the square of the applied electric field so that the square of E t can be
approximated by 2
0 cosE E t for c . In our measurements, c is chosen to
be 6280 rad s-1, which is much higher than the maximum frequency of =200 rad s-1 in
our measurements.
FIG. 3.3. (a) A simple shear flow is applied along the x axis with a velocity
gradient parallel to the z axis. An electric field is also applied along the z direction. (b)
We define to be the angle between the director and the x axis.
50
Taking into account the fact that 2
0 cosE E t =
2 2 2
0 02 cos 2 2cos 2E E E t E E t , we can put
2
0 1,1 2,0 2,2Re Rei t i tt e e , (3.4)
where 0 is the shear stress in the absence of the perturbation E , and ,i j is the stress
response of the ith order with respect to E and the jth harmonic of . Therefore,
1,1 should be proportional to E , and both 2,0 and 2,2 should be
proportional to the square of E for small values of E . The and 2 response
components of the shear stress were obtained by using a vector signal analyser
(HP89410A, Hewlett-Packard). The output of the rheometer corresponding to the torque
is a voltage proportional to the torque. The proportionality constant is obtained by using
a standard fluid (glycerin). The data from the vector signal analyzer are then acquired by
a computer and then analysed by using the IGOR software.
51
FIG. 3.4. Experimental setup. Electric fields are applied between the top and
bottom plates. For the electro-optical experiment we add a light source and a microscope
to observe the change of transmitted intensity.
The shear stress response to ac electric field under steady shear flow and electric
fields was measured by applying a small ac electric field cos( )E t . We obtained the
first-order response 1,1( ) and the second-order response 2,2 ( ) , which should be
proportional to E and 2E , respectively, for small E . The dc electric field
dependencies of the first and second-order responses are shown in Fig. 3.5, where the
measurements were done at -110 s and -1
0 100 V mmE . Linearity was confirmed to
52
hold at least up to-120 V mmE . All the measurements were performed at
-114.1 V mmE and -110 s .
FIG. 3.5. Dependence of 1,1 on E (a) and 2,2 on 2E (b) at -110 s and
-1
0 100 V mmE . Linear relations are obtained at low ac electric fields.
In addition, we make the optical response experiment to directly observe the
director reorientation to ac electric field under shear flow and dc electric fields. As will
53
be shown in the next section, the EL theory predicts that the shear stress response and the
director response behave differently at high frequencies. The optical measurement is
made for checking the theoretical prediction.
In the optical response experiment we replace the upper plate by a glass disk with
diameter 40 mm so that we can observe the transmitted light through the sample by using
a microscope. The same rheometer (Physica MCR300, Anton Paar) is used to apply shear
stress. To minimize the number of disclinations, we coat the upper plate and bottom plate
with perfluoropolymer (CYTOP) (Asahi Glass Co.Ltd.) to align the director
perpendicular to the surfaces, i.e., to obtain homeotropic alignment. The molecular
structure of CYTOP is given in Fig. 3.6. The texture under a polarizing microscope is
shown in Fig. 3.6 for the shear rate of 10 s-1 and Fig. 3.7 for the shear rate of 80 s-1. At a
low shear rate almost no disclination appears (see Fig. 3.7) while at the high shear rate a
few disclination lines appear (see Fig. 3.8).
FIG. 3.6. Molecular structure of CYTOP.
54
FIG. 3.7. Texture of 5CB under crossed polarizers at a shear rate of 10 s-1. There
is no disclination.
FIG. 3.8. Texture of 5CB under crossed polarizers at a shear rate of 80 s-1. There
appear a few disclination lines. The appearance of disclinations may be associated with
the director deviation from the shear plane though the details are not clear.
55
As a light source we used a halogen lamp (LS-LHA, Sumita Optical Glass). Polarizer and
analyzer are used to observe the transmitted light intensity, which changes due to the
birefringence change under ac electric fields. The light intensity was converted by a photo
sensor into a voltage and it was amplified (C6386, Hamamatsu Photonics). The signal is
then analyzed using a vector signal analyser (HP89410A, Hewlett-Packard).
Correction for Torque sensor of Rheometer
When we first measured the stress responses 1,1( ) and
2,2 ( ) , we noticed
a problem in the torque sensor of our rheometer. Although the EL theory predicts that
both the two responses do not go to zero and keep finite values at high frequencies of
applied ac electric fields, experimental ones become zero. We thought that this
discrepancy might be due to the reduction of sensitivity of the torque sensor at high
frequencies. To confirm this, we actually measured the sensitivity as a function of
frequency as follows.
The experimental setup is shown in Fig. 3.9. We use a parallel plate having a
circular tray above it. This tray is usually used for applying voltages to the rotating plate
by filling it with a conductive liquid. This time we fill this tray with a viscous glycerine
and immerse the tip of an L-shaped metal wire which is attached to a piezo actuator. The
diameter of the wire is 1 mm and the depth immersed is about 1 mm. We don’t mount
any sample between the rotating plate and the bottom stage. We rotate the plate at a speed
of 0.8 rpm, and apply an ac voltage to the piezo actuator so that the tip of the wire
oscillates along the flow direction with an amplitude of 28 m. The oscillation of the tip
56
in the glycerine gives rise to a drag force on the top plate, which is detected as torque. If
we assume the Stokes’ law, the torque may be proportional to the velocity of the tip.
FIG. 3.9. The experimental setup.
When the displacement of the tip is given by
0 sinx x t , (3.5)
the torque change corresponding to the above displacement can be expressed as
cosappl a t (3.6)
with a constant a. It is natural that this torque change should be transmitted to the rotating
plate. On the other hand, the torque measured with the rheometer may generally be
expressed as
cosmeas a t . (3.7)
If the torque sensor is ideal, and are constant independent of the frequency.
If not, we have to correct data by using these. We obtain function and with
57
changing , which are shown in a subsequent chapter. Both will show stronger frequency
dependence than predicted before the measurement.
3.3. EL theory
NLCs are composed of rod-like molecules with the long axes aligned statistically
parallel to each other. The average orientation of molecules is represented by a unit vector
n which is called the director. Ericksen and Leslie have formulated a continuum theory
for the velocity v and the director n of NLCs [1, 6, 7, 32, 40, 41]. Hereafter, we assume
a mono-domain (i.e., the director is independent of position) and we apply a simple shear
flow. Under these assumptions, the EL equations can be simplified and we need only the
following equations for our purpose. The angular momentum balance gives
1 2+n h n N n An , (3.8)
where h is the molecular field, N is the rate of change of the director with respect to the
background fluid, and 1
2A v v is the symmetric part of the velocity gradient.
The parameters 1 and 2 are the rotational and irrotational viscosity coefficients. The
components of the molecular field h are given by
f
hn
, (3.9)
where f is the free energy density. When subjected to an electric field, the free energy
density can be written as
58
2
2
0 0
1 1 .
2 2f E n E , (3.10)
where 0 is the dielectric constant in a vacuum, is the dielectric anisotropy defined
as / / with / / and being the dielectric constants parallel and perpendicular
to the director, respectively ( 11.5 for 5CB [42]). The rate of change of the director
is defined as
dn
N W ndt
, (3.11)
where 1
2W v v is the anti-symmetric part of the velocity gradient. The
constitutive equation for the viscous stress tensor is
(visc)
4 1 5 6 2 3A n n n n A n n A n n A n N n N , (3.12)
where i (i=1,...,6) are Leslie coefficients, in terms of which 1 and 2 are expressed as
1 3 2 and 2 6 5 . For 5CB these coefficients have been determined [27]:
1 0.00767 Pa s , 2 0.08171 Pa s , 3 0.00433 Pa s , 4 0.06642 Pa s ,
5 0.06725 Pa s , 6 0.01879 Pa s at 25oC. These values are used in numerical
calculations later. Note that, here, an NLC is treated as an incompressible fluid.
In reality, the mono-domain and simple shear flow assumptions may not be
exactly satisfied for various reasons. For example, in our experiment, no surface treatment
for aligning molecules is made so that the director and flow are spatially disturbed at least
near the surfaces. Note that our model also does not accommodate defects (e.g.,
disclinations). Furthermore, larger deformations from the flow-aligning state are reported
59
to take place [43]. For simplicity, however, we adopt the assumptions shown in Fig. 3.1(a),
where ( ) cos ,0,sinn t and ,0,0 v z with being the shear rate. The reason
is that we expect the applied dc electric field to some degree will align the director to the
same direction.
Here, we calculate the stress change due to a small ac electric field in the steady
state under constant shear flow and dc electric fields. In this case, the total electric field
in Eq. (3.10) is given as 0( ) cosE t E E t . Using Eqs. (3.9) and (3.11), Eq. (3.8)
becomes
2
1 1 1 0
1 1( cos 2 ) sin 2
2 2E t
t
, (3.13)
where the electric field is applied in the z direction. Without an ac electric field, that is,
in the non-perturbed state, the flow alignment angle 0 (as shown in Fig. 3.10) in the
steady state can be obtained by solving Eq. (3.13) by putting 0t
. Then we have 0 as
a function of 0E as follows:
2 2 22 2 2 2 2 2 2 2 2
0 0 1 2 2 0 0 0 0 1 21
0 22 2 2
0 0 2
cos2
E E E
E
.
(3.14a)
In the special case of 0 0E , 0 reduces to
1 1
0
2
1cos
2
. (3.14b)
60
The angle 0 monotonically increases with increasing electric field, as shown in Fig. 3.10,
where the values of the viscosities and the dielectric anisotropy for 5CB at 25 C,
corresponding to the experimental conditions, are used.
FIG. 3.10. Dependence of 0 on 0E calculated from the EL theory. The angle 0
increases monotonically and tends to saturate when 1
0 200 V mmE .
Expanding up to the second order with respect to E , we obtain the corresponding
change in :
2
2,0 1,1 2,2( ) Re[ ] Re[ ]i t i tt e e , (3.15)
with
0 0 01,1
1
sin 2( )
1
E E
i
, (3.16)
61
2
2,2 1,1 2 0 0 0 0 1,1 0 0 0
1
2
0 0
1( ) 1/ 2 cos 2 sin 2 cos 2
2 1
sin 2 ,
E E Ei
E
(3.17)
and
2,0 2,2 (0) , (3.18)
where the first subscript of indicates the order with respect to E and the second one
the harmonic order. The relaxation time is defined by
1
2
2 0 0 0 0sin 2 cos 2E
. (3.19)
Next, we calculate the shear stress response. The shear stress ( zx in the present
case) can be also expanded with respect to E :
2
0 2,0 1,1 2,2( ) Re[ ] Re[ ]i t i tt e e , (3.20)
where 0 is the shear stress with no ac electric field. It should be noted that the
response 1,1 appears under dc electric fields, as can be seen from Eq. (3.16); this
response vanishes for 0 0E . The stress is independent of the polarity of the electric field
as NLCs are non-polar and, therefore, the stress response depends on
2 2 2
0 0( ) 2 cos / 2 1 cos 2E t E E E t E t as shown in Eq. (3.13), clearly
62
indicating that the response 1,1 should emerge under dc electric fields in
addition to the 2 response 2,2 .
From Eqs. (3.12), (3.14a) and (3.16)–(3.19), the unperturbed shear stress 0 and
the responses 1,1 , 2,0 and 2,2 are obtained:
2 2 2 2
0 1 0 0 4 5 2 0 3 6 0 0
1sin cos sin cos sin 2
2
, (3.21)
2 2
1,1 1 0 2 3 3 0 2 0 1,1
1cos 2 cos sin
2i
, (3.22)
2 2
2,2 1 0 2 3 0 3 0 2 0 2,2
2
1 0 2 3 0 2 3 0 1,1
( ) sin 4 2 sin 2 2 cos sin ( )2
cos 4 cos 2 sin 2 ( ) ,2
i
i
(3.23)
2,0 2,2 (0) . (3.24)
Here, we have obtained one of the most important theoretical results that 1,1 and
2,2 ( ) do not goes to zero at high frequencies unlike the usual Debye-type frequency
dispersion as is seen from Eqs. (3.22) and (3.23), i.e., we have a plateau at high
frequencies. The details will be discussed in the next section. When we compare the above
theoretical result with the experimental one, the parallel-plate geometry of the rheometer
which is used in our experiment should be considered. The correction method will be
shown also in the next section “Experimental Results and Discussion”.
63
In addition, we make optical experiments with a purpose to demonstrate that the
director itself has no plateau as is readily seen from Eqs. (3.16) and (3.17), which are the
usual Debye-type function. These responses can be observed through optical experiments
as follows. When an electric field is applied, the birefringence 0( ) ( ) ( )a en n n ,
where 0n and en are the ordinary and extraordinary refractive indices, respectively, will
change and here the observation is assumed to be made perpendicular to the parallel plates
as shown in Fig. 3.4. In NLCs, 0n is independent of the director orientation . However,
en is dependent on :
/ // /2 2 2 2 1/2
/ /
( )( cos sin )
e
n nn n
n n
, (3.25)
where / /n and n are the refractive indices parallel and perpendicular to the director. On
the other hand, the transmitted light intensity of the NLC under crossed polarizers is given
by
2 2
0 sin (2 )sinan d
I I
, (3.26)
where 0I is the incoming light intensity, is the angle between the polarizer and the x
axis, d is the sample thickness and is the wavelength of the light in a vacuum. For a
small change of , , due to the application of an AC electric field, the corresponding
change of I , I , is obtained from Eqs. (3.25) and (3.26):
2 2
02 // / / 0 00 2 2 2 2 3/2
0 / / 0
( 2 cos )sinsin (2 )sin 2
2( cos sin )
an d n n n nI I
n n
, (3.27)
where 0 is the angle without any AC electric field. The intensity change can be
measured as described in the previous section.
64
3. 4 Experimental results and discussion
First, we demonstrate that corrections are necessary for the torque sensor as
described in section 3.3. We show the first order shear stress response 1,1 and the
second order response 2,2 before correction at different dc voltages and at constant
shear rate of 10 s-1 in Fig. 3.11. The theoretical curves are also shown, which are obtained
by using Eq. (3.2). The experiment and theory show good agreement at low frequencies
of ac electric field. However, at higher frequencies, we have a discrepancy between them:
the theory has a plateau but the experiment becomes zero. This is resolved by the
corrections for the torque sensor.
FIG. 3.11. 1,1 as a function of angular frequency before correction.
65
FIG. 3.12. 2,2 as a function of angular frequency before correction.
The correction curves for and obtained by the method described in the
previous section are shown in Figs. 3.13 and 3.14, respectively. Then, by using
and , we correct 1,1 as 1,1 /i
e
, which is shown in Fig. 3.15
(lines are experimental results and dots are theoretical predictions calculated using EL
theory). The corrected 1,1 has a plateau and is in good agreement with the theory.
66
The second order response 2,2 is also corrected in the similar manner and the
results are shown in Fig. 3.16.
FIG. 3.13. as a function of frequency .
FIG. 3.14. as a function of frequency .
67
FIG. 3.15. Frequency dispersion of 1,1 as a function of angular frequency .
The theoretical curves are obtained from Eq. (3.15). A characteristic non-zero plateau,
which is due to the dc electric field, appears at high frequencies in (b), (c) and (d).
68
FIG. 3.16. Frequency dispersion of 2,2 as a function of angular frequency .
The theoretical curves are obtained from Eq. (3.23). A plateau appears under dc electric
fields as well as in the first-order response.
69
Hereafter we show only corrected data. First, we discuss in detail the first-order
response 1,1( ) shown in Fig. 3.15. The responses are given at 0E =20 V mm-1,
80 V mm-1, 120 V mm-1 and 200 V mm-1. The first-order response appears only when we
apply a dc electric field 0E , which is seen from Eqs. (3.16) and (3.22). When 0E is low
( 0E = 20 V mm-1) (see Fig. 3.15(a)), the experimental stress response resembles Debye
type relaxation. However, as we increase 0E to 80 V mm-1 (see Fig. 3.15(b)), we can see
that the response remains non-zero even at high frequencies, forming the above-
mentioned plateau. The emergence of the plateau is the most remarkable characteristic of
the response to ac electric field under dc electric fields. From the EL-theory, it is easily
seen that the plateau comes from the third term with i in the bracket on the right hand
side of Eq. (3.22) because 1,1( ) in Eq. (3.22) is inversely proportional to i at high
frequencies, as can be seen from Eq. (3.16). The i term appears as a result of
differentiating the director with respect to time in Eq. (3.11). In the limit as , we
have
2 2
3 0 2 0
1,1 0 0 0
1
cos sin( ) sin 2E E
. (3.28)
It should be noted that the director response 1,1( ) vanishes at high frequencies, but
the stress response 1,1( ) remains non-zero. The experimental results for the director
response will be discussed later in the optical experiment, which convinces us that this
plateau is ascribed to the time derivative from Eq. (3.11).
70
Fig. 3.17. 0E dependence of the first-order response at very low frequency
(0.23 rad s-1 for the experiment and exactly zero for the theory), 1,1Re ( 0) . Also
shown is the dependence of plateau height 1,1Re ( ) . For dc electric fields
higher than around-190 V mm , 1,1Re ( ) becomes larger than
1,1Re ( 0) .
At higher values of 0E , the height of the plateau is larger than at low frequencies,
as shown in Figs. 3.15(b) (80 V mm-1) and (c) (160 V mm-1). The agreement between the
experimental and theoretical results for all 0E shown in Fig. 3.15 is fairly good. The
dependence of plateau height 1,1( ) on 0E is shown in Fig. 3.16 for the experiment
71
and the theoretical result in Eq. (3.22). From Fig. 3.16, we observe that when
1
0 40 V mmE the plateau is small. As we increase E0 to more than150 V mm, the
plateau height increases sharply. In the same figure, we also plot the 0E dependencies of
1,1 calculated at zero frequency and experimentally obtained from Fig. 3.17 at 0.23 rad
s-1. The agreement between them is good. The low-frequency and high-frequency
(plateau) values exchange at around E0 = 90 V mm-1. The imaginary part of the response
(Fig. 3.15) goes to zero at both low and high frequencies, and shows good agreement
between theory and experiment. The peak or dip frequency of the imaginary part is seen
to increase with increasing dc electric field, though it changes sign from positive to
negative at around E0 = 90 V mm-1 corresponding to the above-mentioned exchange of
the low and high frequency values of the response.
Next, let us discuss the second-order response 2,2 (Fig. 3.16). This response is
more complicated than the first-order response since there are contributions from the first-
order mode 1,1 as well as 2,2 , as can be seen from Eq. (3.23). It should be noted that
the second-order response appears even without a dc electric field, as shown in
Fig. 3.16(a). The stress response resembles Debye type relaxation for 0 0E . But, one
may notice that the real part becomes slightly negative at around 10 rad s-1. Orihara et al.
have previously observed this negative part and shown that it originates from non-
conservative forces due to the shear flow [16]. This is a remarkable characteristic of non-
equilibrium steady states of systems under shear flow and is observed also in an
immiscible polymer blend in which one polymer is dispersed as droplets in the other [17].
The non-conservative forces also violate the fluctuation dissipation relation (FDR). We
72
have theoretically clarified the mechanism of the appearance of the non-conservative
forces in NLCs and derived a modified FDR [18]. This was discussed in detail in chapter
2. According to the theory, the non-conservative forces can emerge only when the director
is out of the shear plane. However, we assumed that the director is confined in the shear
plane and, therefore, we cannot reproduce the negative part in the present model. The
director may tend to be in the shear plane under a dc electric field. Therefore, it is expected
that our mono domain model works better as the dc electric field is increased.
From Fig. 3.16, it is obvious that there is a plateau at every dc electric field except
for 0 0E (Fig. 3.16(a)). As we increase 0E , the plateau, which is the characteristic of the
system under DC electric fields, appears. The origin of the plateau in the second-order
response is the same as that in the first-order response. The plateau comes from the
term proportional to 2,2 in Eq. (3.23), which includes a factor 2 2
3 0 2 0cos sin .
This factor also appears in the term of the first-order response (see Eq. (3.22)).
Interestingly, this vanishes at 0 0E due to the Parodi relation 6 5 2 3 , which
is easily proved by using Eq. (3.14b). Note that the second-order response does not vanish
for 0 0E unlike the first-order response. It is obvious that the factor is in general not
zero for 0 0E . When we increase 0E to 80 V mm-1, the plateau at high frequencies rises,
as shown in Fig. 3.16(b). The plateau becomes remarkable for 0E = 120 V mm-1
(Fig. 3.16(d)). The agreement between the experimental and theoretical results is good.
The imaginary part of the response also shows good agreement between theory and
experiment.
73
Figure 3.18 shows the 0E dependence of 2,2 calculated at zero frequency and
experimentally obtained from Fig. 3.17 at 0.23 rad s-1. Also in the same figure we show
the 0E dependence of the height of the plateau. Good agreement is obtained between the
theory and experiment. The 0E dependencies are more complicated than those in the first-
order response.
FIG. 3.18. 0E dependence of the first-order response at a very low frequency
(0.23 rad s-1 for the experiment and exactly zero for the theory), 2,2Re ( 0) . Also
shown is the dependence of plateau height 2,2Re ( ) .
74
Finally, we show experimental evidence that the plateau observed in the real part
of the first-order response should be ascribed to the time derivative of the director, but
not to 1,1 itself. To do so, as was described in the previous section, we made an optical
measurement using crossed polarizers, in which the light intensity is shown to be just
proportional to the first-order response mode1,1 . Therefore, if
1,1 is of the Debye
type, there should be no plateau observed in the optical response at high frequencies,
though it is observed in the shear stress response.
FIG. 3.19. Frequency dispersion of the optical response, which corresponds to the
director response, as a function of angular frequency . Lines are the experiment and
circles are the theory. No plateau is observed, which shows that the plateau observed in
the shear stress response should be ascribed to the time derivative of the director but not
to 1,1 itself.
75
The experimental result is shown in Fig. 3.19 for 1
0 200 V mmE . Note that the
vertical axis is adjusted so that the theoretical and experimental results coincide as well
as possible, because it is difficult to determine the coefficient in Eq. (3.27). The agreement
between the experiment and theory is good, though the data are scattered to some degree.
The relaxation frequency experimentally observed is in good agreement with the one
theoretically obtained. There is no plateau in the optical response, in contrast to the stress
response, convincing us that the plateau observed in the stress response should originate
from the time derivative of the director.
3.5 Conclusions
We have investigated the NESS response of an NLC to an ac electric field under
constant dc electric field and steady shear flow. The first and second-order shear stress
responses were theoretically obtained from the EL theory assuming a mono-domain
model. It was found theoretically and experimentally that when we apply a dc electric
field both responses remain constant and non-zero even at high ac electric field
frequencies. That is, there is a plateau, which originates from the time derivative of the
director. This plateau is a remarkable feature in the shear stress response brought about
by the application of a dc electric field. Furthermore, by performing an optical
measurement, the director response is confirmed to vanish at high frequencies, strongly
supporting the above-mentioned mechanism for the appearance of the plateau. It was also
clarified that the plateau in the second-order response disappears in the absence of a dc
76
electric field, due to the Parodi relation, which is derived from the Onsager reciprocal
relation.
77
CHAPTER 4
ELECTRIC-FIELD-INDUCED FLOW-ALIGNING STATE IN A NEMATIC
LIQUID CRYSTAL
4.1. Background
In the previous chapter, the experiments were conducted with the use of 5CB (4-
n-pentyl-4-cyanobiphenyl). While in this chapter we use 8CB (4-n-octyl-cyanobiphenyl).
The difference between 5CB and 8CB is in the viscosity ratio 1 2/ where 1 3 2
(rotational viscosity) and 2 6 5 (irrotational viscosity). For 5CB 1 2/ 1 , while
for 8CB 1 2/ 1 .
According to the EL theory, when we apply shear flow to NLCs, the angle
between the director and the flow direction in a flow-aligning state is given as
1 2cos2 / . From this expression, it is clear that the solution will be only available
when 1 2/ 1 . The consequence of this equation is that for 5CB under shear flow, its
director orient in the shear plane with a flow alignment angle given by the above equation
[44], while for 8CB the director will tumble or find another stable orientation with the
director field being spatially deformed, possibly in accordance with the viscous and
elastic coefficients and the boundary condition [18, 45]. Then, under application of a
simple shear flow, NLCs will take either one of the two types of states depending on the
ratio of 1 and 2 . In this chapter, we call the state in which the flow alignment angle
exists as flow aligning state (FAS) and the state in which the flow alignment angle does
78
not exist as non-flow aligning state (NFAS). Most low molecular mass NLCs under shear
flow adopt the FAS except for small number of NLCs such as 8CB. This NFAS structure
is more common in liquid crystal polymers due to the complexity of structure [11].
NFAS can be easily transformed to FAS by applying electric fields along the
velocity gradient direction in NLCs with positive dielectric anisotropy. This may be
regarded as an electric field-induced phase transition which is of interest from the
viewpoint of non-equilibrium physics [46]. Furthermore, we expect different behaviours
of shear stress response in NFAS from those in FAS shown in chapter 3 for 5CB.
4.2. EL theory in FAS
Generally, the transition from NFAS to FAS is due to the competition between
the shear flow effect and electric field effect, where the latter compensates for the
imbalance of the torque produced by the former and finally stabilizes the director
orientation back to the shear plane in a strong enough electric field. In the FAS, the
director is assumed to stay in the shear plane [47] and to be spatially independent except
near the boundary plates. In this simple case, it is sufficient to use EL theory, in which
the order parameter is assumed to be constant, for the purpose to calculate the shear stress
responses [39] which was derived in previous chapter. Here we just rewrite some useful
equations. On the other hand, in the NFAS, it is necessary to solve EL theory in three-
dimensional space, which will make the solution more complicated. So we do not deal
with this problem here.
We have already derived the equation of motion for the FAS in the previous
chapter. So, we briefly summarize the results and note some additional considerations
79
regarding the electric-field-induced FAS in the following. Under simple shear flow with
an electric field applied in the velocity gradient direction, the angle between the
director and the flow direction obeys the following equation
2
1 1 2 0
1 1( cos 2 ) sin 2
2 2E t
t
, (4.1)
where is the shear rate, 0 the dielectric constant of vacuum and the dielectric
anisotropy. From this equation, it is readily seen that the change of , which brings about
the change of the shear stress, depends on the square of the applied electric field E t , as
described above. When 0E t E , Eq. (4.1) gives [47]
22 2 2 2 2
0 0 0 0 2 11
0
1 2
tanE E
. (4.2)
The calculated dc electric field dependence of 0 is shown in Fig. 4.1. The Leslie
coefficients for 8CB at 25 C are taken from Ref. [8] as: 1 0.0382 Pa s, 2 0.0587
Pa s, 3 0.0031 Pa s, 4 0.0520 Pa s, 5 0.0472 Pa s, and 6 0.00840 Pa s
(calculated from the Parodi relation 6 5 2 3 ) and dielectric anisotropy 6.5
is used [48]. It should be noted that all of these parameters are very sensitive to the change
of temperature. In the special case where 0 0E , 0 is reduced to the flow aligning angle
1
1 2cos / / 2f . For 8CB the viscous coefficients are given by
1 3 2 0.0618 Pa s and 2 6 5 0.0556 Pa s , and therefore we have no f .
But, Eq. (4.2) has a solution above a critical electric field given by the following equation:
80
2 2
1 22
0
cE
. (4.3)
As shown in Fig. 4.1, 0 increases with 0E , and cE increases with .
FIG. 4.1. The dc electric field dependence of the flow alignment angle 0 in the
FAS at shear rates of 10 s-1 and 40 s-1 numerically obtained for 8CB from Eq. (4.2).
The electric-field-induced transition has already been experimentally probed
using various methods such as direct observation of flow patterns and disclination density
[36, 49], rheological measurements [50], and synchrotron X-ray studies [52]. However,
it is not easy to clearly distinguish between the FAS and NFAS states by means of the
usual rheological measurements. Although the electric field dependence of shear stress in
8CB at a constant shear rate, in which an NFAS appears without electric field, was also
81
reported by Negita [47] and Patricio et al. [52], the change in the shear stress was too
small for the transition to be clearly detected. Under these circumstances, we measure the
shear stress response to an ac electric field as a probe under both shear flow and dc electric
fields [23]. The shear stress response is more sensitive to the change in director orientation
and we can demonstrate that different frequency dispersion curves in the stress response
are clearly observed and the transition from NFAS to FAS is detected by changing the dc
electric field at a constant shear rate.
Equation (4.1) with an ac electric field can be approximately solved by using the
perturbation method. Expressing ( )t as
2
0 1,1 2,0 2,2Re Rei t i tt e e , (4.4)
we have
0 0 01,1
1
sin 2( )
1
E E
i
, (4.5)
2
2,2 1,1 2 0 0 0 0
1
2
1,1 0 0 0 0 0
1 1( ) 1/ 2 cos 2 sin 2
2 2 1
cos 2 sin 2 ,
Ei
E E E
(4.6a)
and
2,0 2,2 (0) , (4.6b)
where the relaxation frequency 1 is defined by
2
2 0 0 0 0
1
1 1sin 2 cos 2E
. (4.7)
Near the critical electric field, the relaxation frequency is proportional to 0 cE E , which
is obtained from Eqs. (4.2), (4.3), and (4.7). It should be emphasized that the relaxation
82
frequency becomes zero at the transition point, but the behaviour is singular as a function
of 0E .
The shear stress can be calculated from the following equation
visc
4 1 5 6 2 3A n n n n A n n A n n A n N n N , (4.8)
where 1 2A v v and 1 2W v v are the symmetric and anti-
symmetric parts of velocity gradient respectively, /N dn dt W n is the rate of
change of the director with respect to the background fluid, and i (i=1,...,6) are the
Leslie viscosity coefficients. In the present case, cos ,0,sinn and ,0,0v z ,
where the x and z axes are taken along the flow and velocity gradient directions,
respectively.
Using t in Eq. (4.4), we can calculate the shear stress visc
zx t t
considered here as follows:
2
0 1,1 2,0 2,2Re Rei t i tt e e , (4.9)
with
2 2 2 2
0 1 0 0 4 5 2 0 3 6 0 0
1sin cos sin cos sin 2
2
, (4.10a)
2 2
1,1 1 0 2 3 3 0 2 0 1,1
1cos 2 cos sin
2i
, (4.10b)
2 2
2,2 1 0 2 3 0 3 0 2 0 2,2
2
1 0 2 3 0 2 3 0 1,1
( ) sin 4 2 sin 2 2 cos sin ( )2
cos 4 cos 2 sin 2 ( ) ,2
i
i
(4.10c)
83
2,0 2,2 (0) , (4.10d)
where 0 is the shear stress in the absence of the perturbation E , and ,i j is the stress
response of the ith order with respect to E and the jth harmonic of .
To compare the above theoretical results with the experimental ones, we need to
consider the parallel-plate geometry of the rheometer which is used in our experiment.
For this geometry, the apparent shear stress is given by
( ) 2
, ,3
0
4( ) ,
R
R
i j i j
R
d
, (4.11)
where R is the shear rate at the edge of the rotating disk. The numerically calculated
results are presented in the next section, where the superscript “(R)” on the left-hand-side
of Eq. (4.10) will be omitted for simplicity.
4.3. Experiments
The nematic liquid crystal 8CB was purchased from Wako Pure Chemical
Industries and used without any further purification. Measurements were carried out by
using a parallel plate rotational rheometer (Physica MCR300, Anton Paar). The diameter
of the rotating plate and the gap between two parallel plates were 35 and 0.2 mm,
respectively. Note that in the parallel plate geometry, the shear rate depends on position,
and shear rate is defined at the edge of the upper plate. The shear stress at the
84
corresponding shear rate is calculated from the mechanical torque by assuming that the
sample is Newtonian. In the same way as the previous chapter, we made corrections for
the measured torque. All the measurements were made at 37 C in the nematic phase.
The experimental setup and the geometry are shown in Figs. 4.2(a) and (b),
respectively. Electric fields were applied to the sample by using a synthesizer (Model
1940, NF Electric Instruments) and a high-voltage amplifier (4005, NF). In our
measurements, dc electric fields were applied to induce the FAS, and a weak ac electric
field was also applied to probe the stress response. The total field applied was thus
0 cosE E t . However, migration of ions in the NLC sometimes becomes a problem
and reduces the electric field inside the cell. To avoid this, we used a high frequency ac
electric field modulated by the sum of the dc and ac electric fields:
02 cos cos cE t E E t t , (4.12)
where c is the angular frequency of the carrier signal. As was shown above, the induced
shear stress is proportional to the square of the applied electric field so that the square of
E t can be approximated by 2
0 cosE E t for c . In our measurements, c
is chosen to be 6280 rad s-1, which is much higher than the maximum frequency of
=200 rad s-1 in our measurements. The and 2 components of the shear stress were
obtained by using a vector signal analyser (HP89410A, Hewlett-Packard).
85
FIG. 4.2. (a) Experimental setup. The sample is sheared by using a parallel-plate
rheometer under dc and ac electric fields. (b) The flow is applied in the x direction, the
velocity gradient is in the z direction, and the electric field is applied in the z direction.
The setup is similar with the experiment in the previous chapter except that we do not use
the optical apparatus.
Let us first check the linear relationships between E and1,1 , and between
2E and 2,2 . Figures 4.3(a) and (b) show E vs. 1,1 and 2E vs.
2,2 ,
respectively, at 1 1 1
010 s , 100 V mm and 0.63 rad sE . The linearities are good
and the amplitude of the ac electric field for all the measurements is determined to be
114.1 V mmE .
86
FIG. 4.3. (a) Dependence of 1,1 on E at 0E =100 V mm-1 and (b) 2,2
on 2E at 0E =100 V mm-1 at a shear rate of 10 s-1 at a frequency of-10.63 rad s .
Linear relationships are obtained at low electric fields. Both are conducted in FAS state
since for NFAS the theoretical results are not available.
87
4.4. Experimental results and discussion
The dc electric field dependence of the frequency dispersion of 1,1 at the
shear rate of 10 s-1 is shown in Fig. 4.4. The critical electric field to induce FAS from
NFAS is 168 V mmcE as calculated from Eq. (4.3). From the figure, one can observe
that the frequency dispersions below and above the critical electric field are quite different
qualitatively. In the NFAS below this field (Figs. 4.4(a) and (b)), an anomalous change is
observed at around 5 rad s-1 in both the real and imaginary parts. The origin of this
behaviour has not yet been understood since the model we use is not applicable to NFAS.
To deal with the NFAS it is necessary to introduce out-of-shear-plane configurations in
the EL theory. The results are thought to be complicated.
On the other hand, in the FAS above the critical field (Figs. 4.4(c)-(f)), the change
is smooth. The behaviour of frequency dispersion in the FAS is almost the same as that
measured in 5CB (4-n-pentyl-cyanobiphenyl), which adopts the FAS even in the absence
of a dc electric field [23]. As described in the previous chapter, in 5CB we have already
observed a characteristic behaviour under shear flow and dc electric fields: the real part
of the frequency dispersion has a plateau at high frequencies. The appearance of the
plateau is understood from Eqs. (4.5) and (4.10b) as follows. 1,1 given by Eq. (4.5)
is the Debye type which goes to zero at high frequencies, but 1,1 differs from it
because in Eq. (4.10b) there is an i term multiplied by 1,1 , which gives a non-
zero constant at the limit of as shown in the following equation:
88
2 2
3 0 2 0
1,1 0 0 0
1
cos sin( ) sin 2E E
. (4.13)
The origin of the i term is the time derivative terms in Eq. (2.4):
2 3 2 32 2
x xz zz x x z z x
dn nn dnn N n N n n
dt dt
, (4.14)
where we have put and z x for our case. From this equation it is also understood
that the plateau is not observed in the NFAS. In the NFAS the director is thought to be
out of the shear plane to some degree, indicating that xn and/or zn should be small. The
theoretical curves for the FAS calculated from Eqs. (4.5), (4.10b) and (4.11) are also
shown, and are in good agreement with the experimental results.
89
FIG. 4.4. Frequency dispersions of the first-order harmonic response 1,1( ) at
a shear rate of 10 s-1 for several dc electric fields. The dispersions are clearly
distinguishable between NFAS ((a) and (b)) and FAS ((c)-(f)). Solid lines for the FAS are
calculated on the basis of the EL theory.
90
Figure 4.5 shows the 0E dependence of the peak or dip frequency of the
imaginary part, peak , where dots represent the experimental results, the solid line is
obtained from the above calculated frequency dispersion curves using Eq. (4.11), and the
dashed line is 1/ calculated from Eq. (4.7). A slowing down of the relaxation time is
observed in the FAS, as is expected from Eqs. (4.2) and (4.7). At around a dc electric field
of 100 V mm-1 the solid line disappears because the peak changes to the dip in the
imaginary part there, and so it is difficult to obtain the peak or dip position. At the critical
point, the value of the dashed line becomes zero with the singularity of 0 cE E , but that
of the solid line is finite because it is averaged out by the integration of Eq. (4.11).
On the other hand, in the experimental results the transition point (82 V mm-1)
between NFAS and FAS seems to be located above the calculated critical point
(68 V mm-1). The reasons for this may be as follows: The FAS changes to the NFAS
before reaching the calculated critical point due to the boundary condition, and/or the
parameters used for the numerical calculations are not suitable for our sample. It should
be noted that the parameters used here (the Leslie coefficients and dielectric anisotropy)
are very sensitive to the change of sample temperature. Except for the details, good
agreement is obtained between experiment and theory in the FAS, while in the NFAS, the
relaxation frequency is relatively low and almost constant except near the critical point.
91
FIG. 4.5. 0E dependence of the peak or dip frequency of the imaginary part,
peak , where dots are experimental results, a solid line is obtained from calculated
frequency dispersion curves using Eq. (4.11), and a dashed line is 1/ calculated from
Eq. (4.7).
Next, let us examine in detail the 0E dependence of the real part of 1,1 in
the low frequency region. The experimentally obtained real part at 0.63 rad s-1 and the
theoretically calculated one at zero frequency are shown in Fig. 4.6 for the shear rates of
10 and 40 s-1. In the FAS, the response at low frequencies increases as the transition point
is approached and peaks around the transition point for both of the two shear rates. This
behaviour is also seen in the NFAS, and is attributed to the coexistence of the FAS and
92
NFAS just below the transition point. The transition electric field at the shear rate of 40
s-1 is higher than that at 10 s-1, as is expected from Eq. (4.9).
FIG. 4.6. 0E dependencies of the first-order response at very low frequency
(0.63 rad s-1 for the experiment and exactly zero for the theory), 1,1Re ( 0) , for
shear rates of 10 and 40 s-1.
Figure 4.7 shows the frequency dispersion of the second-order response
2,2 . In the NFAS (Figs. 4.7(a) and (b)) an anomalous change is seen around 3 rad
s-1 as well as in the first-order response. In the FAS (Figs. 4.7(c)-(f)), the experimental
93
behaviour is almost reproduced by the theory. In the second-order response, however, the
frequency dispersion becomes more complicated because Eq. (4.9c) contains2
1,1( ) .
There is a plateau in both the second-order response and the first-order response. In
Fig. 4.8, the 0E dependence of the real part of 2,2 in the low frequency region is
shown, where the experimental data are obtained at 0.3 rad s-1 and the theoretical curves
are calculated at zero frequency for the shear rates of 10 and 40 s-1. In the FAS, the second-
order response is negative, unlike the first-order response, for both the shear rates of 10
and 40 s-1, but the details are different: at 10 s-1 the absolute value of the response
increases near the transition point, which may be related to the slowing down, whereas at
40 s-1 it is almost constant as the measurement is taken only near the transition point.
94
FIG. 4.7. Frequency dispersions of the second-order response 2,2 ( ) at a shear
rate of 10 s-1 for several dc electric fields. The dispersions are clearly distinguishable
between the NFAS ((a) and (b)) and the FAS ((c)-(f)). Solid lines in the FAS are calculated
on the basis of the EL theory.
95
FIG. 4.8. 0E dependencies of the second-order response at a very low frequency
(0.3 rad s-1 for the experiment and exactly zero for the theory), 2,2Re ( 0) , for
shear rates of 10 and 40 s-1.
Finally, we show the shear stress without ac electric fields, 0 , in Eq. (4.9) as a
function of 0E in Fig. 4.9, from which it is difficult to distinguish between FAS and
NFAS. This clearly shows that the stress response is sensitive to the change in the director
96
field. It is worthwhile to point out that Fig. 4.9 is closely related to Figs. 4.6 and 4.8. At
the limit of 0 , Eq. (4.4) becomes
0 0 1,1 2,20 2 0E E E , (4.15)
where we have used Eq. (4.10) and note that 0 0E . Taking into account that
1,1 E and 2
2,2 E , we have 1,1 00 ' E E and
2
2,2 00 " / 4E E . From the latter equation it is easily understood that
2,2 0 should be negative because 0 0E is a convex function in the FAS, as is
seen from Fig. 4.9.
FIG. 4.9. 0E dependencies of 0 at shear rates of 10 and 40 s-1. It is difficult to
distinguish the FAS and NFAS based on 0 .
97
4.5 Conclusions
We have investigated the field-induced transition of 8CB from the NFAS to FAS
by measuring the shear stress response to an ac electric field. Totally different frequency
dispersions in the two states were observed for both the first- and second-order responses.
The slowing down of relaxation time and the increased zero-frequency response were
found near the transition point as well as in equilibrium phase transitions. The
experimental results in the FAS were in good agreement with the calculated results based
on the EL theory, including the plateau observed at high frequencies. As for NFAS, it is
necessary to elucidate the director field to calculate the stress response. Our measurement
method is expected to be useful for studying non-equilibrium phase transitions under
shear flow.
98
CHAPTER 5
SUMMARY AND CONCLUSIONS
The non-equilibrium steady state response and fluctuations of sheared LCs were
investigated both experimentally and theoretically. On the basis of the Ericksen-Leslie
(EL) theory the dynamical properties were examined. The shear stress response to an ac
electric field was measured in shear flow and dc electric fields, and the results were
analyzed in terms of the EL theory.
Using the EL theory, phase diagrams in the three-dimensional space of magnetic field
at a constant shear rate were obtained for two liquid crystals 5CB and 8CB which have
different flow properties. An analytical solution was given for the tumbling state, in which
the director rotates about an axis determined as a left eigenvector of the dynamical matrix
that governs the motion of the director. It was shown that non-conservative forces caused
by shear flow appear when magnetic fields are applied, so that the director points out of
the shear plane. The non-conservative forces experimentally observed in the response
function of 5CB were thought to originate from a similar mechanism. Using the Langevin
equation, the orientational fluctuations of the director were examined in detail by
calculating the time correlation function and the response function, which are related to
each other through a modified fluctuation dissipation relation. Critical behaviors such as
the divergence of the relaxation time and fluctuations were found at the critical points
where there is a discontinuous change in director orientation and at the boundary between
the stable and unstable states.
99
The non-equilibrium steady state response of an LC to an ac electric field under
constant dc electric field and steady shear flow was investigated. The first and second-
order shear stress responses were theoretically obtained from the EL theory assuming a
mono-domain model. It was found theoretically and experimentally that when a dc
electric field is applied both responses remain constant and non-zero even at high
frequencies of ac electric field. That is, there is a plateau, which originates from the time
derivative of the director in the viscous equation. This plateau is a remarkable feature in
the shear stress response brought about by the application of a dc electric field.
Furthermore, by performing an optical measurement, the director response was confirmed
to vanish at high frequencies, strongly supporting the above-mentioned mechanism for
the appearance of the plateau. It was also clarified that the plateau in the second-order
response disappears in the absence of a dc electric field, due to the Parodi relation, which
is derived from the Onsager reciprocal relation.
The field-induced transition from the non-flow-aligning state (NFAS) to the flow-
aligning state (FAS) was investigated by measuring the shear stress response to an ac
electric field. Totally different frequency dispersions in the two states were observed for
both the first- and second-order responses. The slowing down of relaxation time and the
increased zero-frequency response were found near the transition point as well as in
equilibrium phase transitions. The experimental results in the FAS were in good
agreement with the calculated results based on the Ericksen-Leslie theory, including the
plateau observed at high frequencies. As a result, the present measurement method can
be expected to be useful for studying non-equilibrium phase transitions under shear flow.
However, as for NFAS, it is necessary to elucidate the director field to calculate the stress
response.
100
Our current studies were only limited to the mono-domain assumption (boundary
condition is neglected). The calculation of dynamics by taking into account the boundary
condition and spatially inhomogeneous director field is expected to give abundant
features of non-equilibrium steady state response and fluctuations in sheared nematic
liquid crystals.
101
REFERENCES [1] P. G. De Gennes and J. Frost, The Physics Of Liquid Crystals, edited by L. Kramer
and A. Buka (Springer, Berlin, 1996).
[2] G. Pelzl, W. Weissflog, In Thermotropic Liquid Crystals, Ed. A. Ramanoorthy
(Springer Netherlands, Dordrecht, 2007).
[3] P. G. De Gennes, Phys. Lett. A, 30 8 (1969).
[4] F. M. Leslie, Handbook of Liquid Crystals Volume 1, Ed. D. Demus, J. W. Goodby,
G. W. Gray, H. W. Spiess, V. Vill (Wiley-VCH, New York).
[5] S. Chandrasekar, Liquid Crystals (Cambridge University Press, 1992)
[6] J. L. Ericksen, Trans. Soc. Rheol. 5 (1961).
[7] F. M. Leslie, Arch. of Ration. Mech. An. 4, 265 (1968).
[8] F.C. Franck, Discuss. Faraday. Soc., 25 (1958).
[9] C. W. Oseen, Trans. Faraday. Soc., 29 (1933).
[10] H. Zocher, Trans. Faraday Soc., 19 (1933).
[11] I. K. Yang and A. D. Shine, Macromolecules 26, 1529 (1993).
[12] [5] R. Kubo, J. Phys. Soc. Jpn. 12, 570 (1957).
[13] R. Kubo, Reports on Progress in Physics, 29.1 (1966).
[14] R. Kubo, M. Toda, N. Hashitsume, Statistical Physics II: Nonequilibrium Statistical
Mechanics (Springer Science & Business Media, 2012).
[15] E. Frehland, Biophysical Chemistry, 12 1 (1960).
[16] T. Ohtsuki, T. Keyes, Phys. Rev. A 35 (1987).
102
[17] M. Sellitto, The European Physics Journal B 4 2 (1998).
[18] A. Picone, M. Henkel, Journal of Physics A, 35 27 (2002).
[19] S. Fielding, P. Sollich, Phys. Rev. Lett. 88, 050603 (2002).
[20] R. Mauri, D. Leporini, EPL 76 6 (2006).
[21] [6] T. Sakaue, T. Ohta, Phys. Rev. E 77, 050102(R), (2008).
[22] H. Pleiner, H. Brand, J. Physique Lett. 44 1 (1983)
[23] H. Orihara, F. Yang, Y. Takigami, Y. Takikawa, and Y. H. Na, Phys. Rev. E 86, 041701
(2012).
[24] Y. H. Na, K. Aida, R. Sakai, T. Kakuchi, and H. Orihara, Phys. Rev. E 80, 061803
(2009).
[25] J. F. Fatriansyah and H. Orihara, Phys. Rev. E 88, 012510 (2013).
[26] H. Orihara, 液晶の物理, (内田老鶴圃, 2004).
[27] G. Ahlers, In Pattern Formation in Liquid Crystals, Ed. L. Kramer and A. Buka
(Springer, Berlin, 1996).
[28] L. R. P. de Andrade Lima, A. D. Rey, Phys. Rev. E. 70, 011701 (2004).
[29] T. Tsuji, A. D. Rey,. Phys. Rev. E. 57 (1998).
[30] S. Heidenreich, P. Ilg, S. Hess, Phys. Rev. E. 73 (2006).
[31] S. Fu, T. Tsuji, S. Chono, J. Rheol. 52, 451 (2008).
[32] F.M., Leslie, Rheol. Acta. 10, 91 (1971).
[33] H. Pescelli, Fluctuations in Physical Systems (Cambridge University Press, 2000).
[34] S. Hess, J. Non-Equilib. Thermodyn. 11 (1986).
[35] H. Orihara, H. Tazawa, Y. H. Na, Japanese Journal of Applied Physics 47 12 (2008).
[36] E. Sackmann, Chem. Phys. Lett. 3 4 (1969).
103
[37] M. F. Schiekel, K. Fahrenschon, App. Pyhs. Lett. 19 (1971).
[38] F. J. Kahn, App. Phys. Lett. 20 (1972).
[39] J. F. Fatriansyah, Y. Sasaki and H. Orihara, Phys. Rev. E 90, 032504 (2014).
[40] J. L. Ericksen, Trans. Soc. Rheol. 11, 5 (1967).
[41] O. Parodi, J Phys. France 31, 581 (1970).
[42] P. G. Cummins, D. A. Dunmur, and D. A. Laidler, Mol. Cryst. Liq. Cryst. 30, 109
(1975).
[43] T. Tsuji and A. D. Rey, Phys. Rev. E 57, 5609 (1998).
[44] P. T. Mather, D. S. Pearson, R. G. Larson, Liq. Cryst. 20, 527 (1996).
[45] P. T. Mather, D. S. Pearson, R. G. Larson, Liq. Cryst. 20, 539 (1996).
[46] J. F. Fatriansyah, H. Orihara, Phys. Rev. E 91,042508 (2015).
[47] K. Negita, Mol. Cryst. Liq. Cryst. 300, 163 (1997).
[48] B. R. Ratna and R. Shashidar, Pramana 6, 278 (1975).
[49] K. Skarp, T. Carlsson, S. T. Lagerwall, and B. Stebler, Mol. Cryst. Liq. Cryst. 66,
199 (1981).
[50] D. J Ternet, R. G. Larson, and L. G. Leal, Rheol. Acta 38, 183 (1999).
[51] C. R. Safinya, E. B. Sirota, R. Plano, and R. F. Bruinsma, J. Phys.: Condens. Matter
2 (1990).
[52] P. Patricio, Liquid Crystals 39, 25 (2012).
104
LIST OF PUBLICATIONS
1. Jaka Fajar Fatriansyah and Hiroshi Orihara, Dynamical properties of nematic liquid
crystals subjected to shear flow and magnetic fields: Tumbling instability and
nonequilibrium fluctuations, Physical Review E 88, 012510 (2013).
2. Jaka Fajar Fatriansyah, Yuji Sasaki and Hiroshi Orihara, Nonequilibrium steady-state of
a nematic liquid crystal under simple shear flow and electric fields, Physical Review E
90, 032504 (2014).
3. Jaka Fajar Fatriansyah and Hiroshi Orihara, Electric-field-induced flow-aligning state in
a nematic liquid crystal, Physical Review E 91, 042508 (2015).