• Aucun résultat trouvé

Fluctuations near the convective instability in a cholesteric liquid crystal

N/A
N/A
Protected

Academic year: 2021

Partager "Fluctuations near the convective instability in a cholesteric liquid crystal"

Copied!
9
0
0

Texte intégral

(1)

HAL Id: jpa-00208384

https://hal.archives-ouvertes.fr/jpa-00208384

Submitted on 1 Jan 1975

HAL

is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire

HAL, est

destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Fluctuations near the convective instability in a cholesteric liquid crystal

J.D. Parsons

To cite this version:

J.D. Parsons. Fluctuations near the convective instability in a cholesteric liquid crystal. Journal de

Physique, 1975, 36 (12), pp.1363-1370. �10.1051/jphys:0197500360120136300�. �jpa-00208384�

(2)

FLUCTUATIONS NEAR THE CONVECTIVE INSTABILITY

IN A CHOLESTERIC LIQUID CRYSTAL

J. D. PARSONS

Department

of

Mathematics, University of Strathclyde, Glasgow G1,

U.K.

(Reçu

le 28 avril

1975,

révisé le

15 juillet 1975, accepté

le

18 juillet 1975)

Résumé. 2014 On analyse le comportement des modes

hydrodynamiques

d’une couche cholestérique

horizontale soumise à un gradient thermique qui induit une instabilité convective pour une certaine valeur critique. A cause du

couplage

de modes induit par la force extérieure, la fréquence d’un des

modes du système (mode de conduction-orientation) s’annule quand le gradient de température

atteint la valeur critique. Si le pas de l’hélice cholestérique

dépend

faiblement de la température,

seule la

partie

imaginaire de la fréquence s’annule au seuil et on trouve des cellules de convection qui

oscillent.

Abstract. 2014 An analysis is presented of the behaviour of the

hydrodynamic

modes in a horizontal

cholesteric layer

subjected

to a temperature gradient which, when reaching a critical value, drives the system into convective instability. It is found that, because of the mode

coupling

induced by the

external force, the

frequency

of one of the modes of the system,

corresponding

to a thermal conduc- tion-orientation mode, goes to zero when the temperature gradient increases up to a critical value.

When the cholesteric

pitch depends

weakly on temperature, only the imaginary part of the

frequency

vanishes at the instability, and one finds

oscillating

convection cells.

Classification

Physics Abstracts

7.130

1. Introduction. - Consider a

simple

newtonian

fluid confined to a horizontal

layer.

If a downward

directed

temperature gradient

is maintained across

the

layer,

it is well known

[1]

that

stationary

convection

sets in when the

temperature gradient

exceeds a

certain critical value. This is one of the

simplest

cases

of a

hydrodynamical instability

and is

usually

referred to as the Bénard

Rayleigh problem

in classi-

cal

physics.

The

instability

occurs because the

liquid,

when heated from

below, develops

a

buoyancy

force

due to the volume

expansion

of the

liquid

near the

bottom. For

sufficiently

strong

heating,

it is pos- sible for the

buoyancy

force to overcome the vis-

cous shear forces and the fluid

undergoes

a transi-

tion from a state where heat is

transported by

thermal

conduction

alone,

to a state of combined heat conduc- tion and convection.

Recently

the

dynamics

of fluctuations near the

instability

threshold has been studied

[2].

In a

simple

fluid there exists a

propagating

acoustic

mode,

two

overdamped

shear

modes,

and an

overdamped

ther-

mal conduction mode. The presence of a

temperature gradient couples

these modes

together. However, provided

that the thermal

conductivity

is not too

large,

it is

possible

to

approximately decouple

the

thermal mode from the others. One can then show that as the

instability

is

approached

from below the thermal modes goes

soft,

i.e. its

frequency

vanishes

as

(Re - R )

where R is the

Rayleigh

number

(the

dimensionless

temperature gradient) and Rc

is its

critical value. This result suggests a formal

analogy

between the behaviour of the fluid near the

instability

and the situation encountered near the critical

point

in structural

phase

transitions

[3].

The

instability

of a cholesteric

liquid crystal

under

a

temperature gradient

has been considered

by

Dubois-

Violette

[4]

for the case where the

temperature gradient

is

applied along

the helical axis and when the

tempe-

rature

dependence

of the cholesteric

pitch

can be

ignored.

The critical value of the

temperature gradient

is found to be much lower than in a

simple

fluid and

depends

on the

anisotropy

of the thermal conducti-

vity

Ka. The mechanism is

briefly

as follows : the

temperature gradient couples

to the orientation

through Ka

and leads to a distortion in the orientation.

The

instability

occurs when the

buoyancy

force of

the

liquid

near thé bottom overcomes the elastic forces associated with the orientation. Because these elastic forces are much smaller than the viscous shear

forces,

a much smaller

buoyancy

force can

balance them and this leads to a smaller value for the critical

temperature gradient.

A rather

striking aspect

of the

problem

is that the direction of the

temperature gradient

necessary for

instability depends

upon the

sign

of ka. Thus it is

possible

to reach an

instability by heating

the

sample

from the

top !

This

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:0197500360120136300

(3)

1364

situation also occurs in the case of a nematic

liquid crystal

oriented in a horizontal direction

[5].

In this paper we discuss the

hydrodynamic

fluctua-

tions for temperature

gradients slightly

below the

critical value. We consider the case where a

magnetic

field is

applied parallel

to the helical axis and the

temperature gradient,

and where the

anisotropy

of

the

magnetic susceptibility,

X. is

negative.

A

simple

set of

hydrodynamical equations

has been

developed by Lubensky

for this case

[6].

These

equations

are

equivalent

to the director formulation for

slowly varying hydrodynamic motions,

but are more conve-

nient to use because of the

spatial dependence

of the

unperturbed

director in cholesterics. The

equations

also

apply

to the case of a

parallel

electric field with

a

negative

dielectric

anisotropy, provided

the effect

of conduction

impurities

can be

neglected.

The critical

temperature gradient

and normal modes

are evaluated from the

hydrodynamical equations

first

assuming

that the cholesteric

pitch

is

indepen-

dent of

temperature,

as in

[4].

In this case one finds

steady

convection

cells,

and both the critical

tempe-

rature

gradient

and size of the convection cells increase with the field H. Like the case of a

simple fluid,

one of the modes goes soft

(its frequency vanishes)

at

the critical

point.

The soft mode is a

coupled

orienta-

tion-thermal conduction mode. The

frequencies

of

the other modes are well behaved at the critical

point.

Next,

the effect of the

temperature dependence

of

the

pitch

is considered. The critical temperature

gradient

is decreased

slightly

and in this case one

obtains

oscillating

convection cells. The

frequency

of

the slow mode

acquires

a real

part

which stays finite

at the

instability,

while the

imaginary

part vanishes

as before. Thus a

propagating

thermal conduction- orientation mode can exist

sufficiently

close to the

instability.

A distorted

equilibrium

helical structure

is crucial for this effect. It could not occur in a nematic

or a cholesteric whose

pitch

is

independent

of

tempe-

rature.

2.

Governing équations.

- 2. 1 BASIC HYDRODYNA-

MICS. - The cholesteric is

imagined

to be confined to a horizontal

layer

of thickness 1 and of infinite extent in the transverse directions. The

magnetic

field is

applied parallel

to the helical axis which is in the vertical direction. At the lower

boundary

the

temperature

is fixed at

To

and at the upper

boundary

it is fixed at

To -

AT. AT" may be

positive

or

negative

but its

magnitude

must be small

enough

to

keep

the

entire

sample

in the cholesteric state.

As usual we introduce a unit vector n

(the director)

which lies in the direction of average molecular orientation at each

point

in space. In the presence of the

temperature gradient

but in the absence of fluctuations we have

where the helical axis is chosen as the z axis.

qJs(z)

will be determined below and

depends

upon the temperature

gradient ATI1.

When

ATIL =

0 we have

qJs(z)

= qo z where qo =

n/P

with P

being

the

pitch.

The

subscript

s refers to the

steady

state.

Let p be a unit vector

parallel

to the helical axis.

Then n can rotate in a

plane perpendicular

to p and also in a

plane containing

p. When

H //

p, the mole- cules will tend to lie

perpendicular

to the field when

the

anisotropy

of the

magnetic susceptibility, xa

is

negative.

The

principal

effect of the field in this case

is to suppress rotations in the

plane containing

p.

The form of the director in the presence of the fluc- tuations is :

where ç =

(p,,(z)

+

qJ’(r, t),

with

ç’

the small fluctua-

tion. The

only

fluctuations in nz that are

slowly varying

and

hydrodynamic

are those due to a

splay

distor-

tion of the

pitch

vector p. Bend and twist distortions of p lead to finite

energies

in the limit of

vanishing spatial gradients,

and therefore fluctuations asso-

ciated with these distortions can be

neglected

in a

hydrodynamic

treatment. This

point

has been discuss- ed in detail elsewhere

[6, 7, 8].

Since the fluctuations in nz are determined

by

those in ç,

only

the

single

variable

ç’

is needed to describe

hydrodynamic

fluctuations in orientation. In the presence of both

a

temperature gradient

and the field

H,

the contant À in

(2)

can be shown to be

given by

where

kil

and

k33

are the Frank

[9]

elastic constants

for

splay

and

bend, respectively.

The

hydrodynamical equations

in the standard

Boussinesq approximation [1]

are,

following [6] :

where v is the fluid

velocity, ps

is the

density

in the

presence of the

temperature gradient, g

is the acce-

leration due to

gravity,

and G is the energy

density.

The fluxes and forces are

given by

(4)

In these

equations

p is the pressure and h is the vector

variationally conjugate

to

Vç.

It can be shown that

where

k22

is the Frank constant for

twist,

and

with X. o..

The

dissipative

parts of

(8), (9), (10)

are

given by

kl

(Kjj)

is the thermal

conductivity perpendicular (parallel)

to the helical axis. We will use the nota-

tion xa = KII

- xl.

Finally

y has units of a

viscosity

and is

usually

called the

director friction coefficient. In these

equations

we

have

ignored

a

possible coupling

between orientation and the temperature

gradient

discussed

by

Leslie

[10] ;

this term is also

ignored

in

[4]

where it is

argued

that it is small.

2.2 STATIC SOLUTION. - A

possible

solution to

these

equations

when the

temperature gradient

is

small

enough

is the

equilibrium

situation where

convection is absent :

We will assume the

following equation

of state

dp = px dp - pp dT (18)

where x is the isothermal

compressibility

and

fl

is

the volume

expansion

coefficient :

For thin

layers

we will have

where po, po are constants.

Finally,

in

equilibrium

we must have

Now

where we

ignore

the pressure

dependence

of the

pitch.

The

following thermodynamic identity applies :

iôh. Ô T a,Z

= - 1 .

(24)

(Ghz) ( OT) (ÔlfJ’J

oT ({J,z

a lfJ,z

hz

ah z

T = - 1 .

(24)

Using (12)

and

noting

that

(olfJ,z/oT)hz

=

oqoloT,

(23)

leads to the

equation

,

and

using (17)

this

integrates

to

, 2.3 LINEARIZED EQUATIONS. - Next we consider small fluctuations from the

equilibrium

state

(17), (20), (21)

and

(26)

and write p = Ps

+ p’,

p = Ps +

p’,

T =

Ts

+

T’,

9 = ({Js +

q’

and linearize the

hydro- dynamical equations

in the

primed

variables and v.

For thin

layers

we may

replace

p. s

by po in

the first

term in

(5).

The second term is

approximately

po

gflT ’

and afl viscosities and thermal conductivities are

assumed

independent

of

temperature.

These assump- tions form the basis of the

Boussinesq approxi-

mation

[1].

In

addition,

the

problem only

seems

tractable for the case where the cholesteric

pitch depends weakly

on

temperature.

Provided

the

quantity dCPsldz

may be

replaced by

qo in

(3), (8), (10)

and

(13),

and the Frank elastic constants, qo, and

oqolor

may all be

regarded

as constants.

Finally,

we will

simplify

the calculation

by assuming

that all of the

fluctuating quantities depend

upon z and one of the transverse coordinates

only,

say x.

Thus our convection cells will be one dimensional rolls. It can be shown that vy = 0 and then the equa- tions reduce to :

(5)

1366

In

(30),

cp is the

specific

heat at constant pressure.

3. Normal modes near the

instability.

- 3.1 DIS-

PERSION RELATION. - We look for solutions of the

equations

of the

following special

form :

q

with similar

expansions

for v,

p’

and

({J’.

ql is

quasi-

continuous in

(32)

but we will limit q3 to the values q3 =

nnll

where n is a nonzero

integer.

These values of q3 can be shown to lead to solutions

satisfying q’

- T’ = Vz =

èvxlèz

= 0 at both boundaries. This is also the

assumption

used in

[4]. Rigid boundary

conditions

where vx

= 0 at both boundaries

requires

numerical solution even for

simple

fluids

[11]

since

the

expansion (32)

is not

appropriate

for this case.

Inserting (32)

into

(27) through (31)

leads to five

linear

homogeneous equations

for the

amplitudes

vq,

Tq,

({Jq, pq. A nontrivial solution can exist

only

if the determinant of the coefficients vanishes. This leads to a

dispersion

relation between the

frequencies

of the modes and their wavevectors :

where the

following

notation has been introduced :

and ql + q23

=

q2.

The rest of this paper will be concemed with

approximate

solutions of the

disper-

sion relation

(33).

First,

suppose that the

temperature gradient

vani-

shes. An

approximate

solution in the case where

dissipative

terms are small is :

where

1- -1

with ql = q sin

0,

q3 = q cos 0. This is the propa-

gating

shear-orientation mode discussed in

[6].

Notice

that the

phase velocity ct(O)

vanishes for

propagation along

the helical axis. In this case the shear and orientation modes are

uncoupled

and both are

overdamped [12].

The solution

(37)

is

only

valid for

small q. Corrections involve

dissipative

terms which

occur at

higher

order in q. If these terms are small the solution to next order is

where

For most cholesterics it is true that

Then the

damping

factor

(40)

is dominated

by

the

viscous term. The other solution to

(33)

when

A T/1

= 0

is the

overdamped

thermal mode

3.2 STEADY CONVECTION.

ôqolôt

= 0. - Now

let us look for solutions to

(33)

when the

temperature gradient

is near the critical

value, ignoring

the

ôqo/ôT

term as in

[4].

We know that near the insta-

bility

one of the solutions for co is very small. At the critical value of

AT/1

it

vanishes,

hence we can find

the critical value

by setting

co = 0 and

solvifig (33)

for

A7Y/.

We make the

assumption

that

in addition to

(41).

Under these conditions one finds for the critical value

(6)

The true threshold is found

by minimizing (44)

with respect to q.

AT/1

will be smallest when q3 assumes its smallest

possible

value

nll

=

Q. Minimizing (44)

with respect to ql, it is found that

AT11

attains its

smallest value when ql = qc where

The minimum value is

where

The critical

temperature gradient

can be

positive

or

negative depending

upon the

sign

of xa. We get an

instability by heating

from below when

and an

instability by heating

from above when xa > 0

(K

jj >

KJ.).

This unusual behaviour also occurs

in a nematic but there the critical

temperature gradient

is

proportional

to

Q4

rather than to

Q 2 q2.

Thus the

instability

threshold is smaller in a nematic and it has been observed in

samples

submitted to a

tempe-

rature difference between boundaries of

only

a few

degrees [13]

for 1 c>i 1 mm. For

cholesterics,

in order

to

keep

AT within the few

degrees required by

the

Boussinesq approximation,

we will need thick

samples

with

relatively large

cholesteric

pitches.

We note that

although

a

negative

critical

temperature gradient

is

unusual,

it is not confined to

liquid crystals.

In

binary

fluids one can get an

instability by heating

from the

top

when the thermal diffusion coefficient is

negative [14].

The above results may be

compared

to those in

[4]

with H = 0

by setting kl(H)

= 0 in

(34).

However

then we must include the next order term in q1. Accord-

ing

to

[4]

this term is

(3 k33/8) q4lq’

and is

negligible compared

to the term

k_L(H) q2 1

for fields which

satisfy

when H = 0 the critical value

of ql

becomes

and since

Àqo

= 1 when H =

0,

this leads to

which is the result in

[4].

The critical

wavevector qc

(the

wavevector of the first mode to become

unstable)

is now much greater than before

(see

eq.

(45)).

Thus

the size of the convection cells increases with the field H. There is a

simple physical explanation

for

this. The field H increases the energy necessary to distort the

layers.

Since the available energy comes from the

bouyancy

forces which are

independent

of

H,

we must have less distortion as H increases so

that the cell size

qc-1

increases with H. From

(46),

it is seen that

(AT11).i.

also increases with

H,

as would be

expected

in a

parallel

field that stabilizes the helical structure.

For

temperature gradients

close to, but

slightly

below the critical

value,

one of the roots of

(33)

will be small. To find that root, linearize

(33)

in w.

With

oqoloT

=

0,

and the

approximations (40), (43)

and

(48),

it is

straightforward

to show that the solu- tion is :

where

when

AT/1 = 0, R

= 0 and

(51)

reduces to

(42),

an

overdamped

thermal mode. However at the cri- tical value we have

R/Rc(q)

= 1 and the

frequency

of the mode vanishes. We say the mode goes

soft

at the

critical value of R. The form

(51)

is similar to what

happens

in an

ordinary

fluid but the critical

tempe-

rature

gradient

is

différent ;

note the

striking

absence

of the

viscosity

v from

(52).

Also near the critical

point (51)

is not a pure thermal mode because the fluctuations in

temperature

are

strongly coupled

to

those in orientation. From

(30)

and

(31)

it is

possible

to

show,

when co - 0 :

where

Tq

is the

amplitude

of the

divergent

thermal

fluctuations. It is also

possible

to show that the other

solutions to

(33)

are well behaved near the

instability ;

the

frequency

of the shear-orientation mode is

changed only slightly

from

(39).

3 . 3 OSCILLATING CONVECTION.

ôqolê T :0

0. - Next

we consider

(33)

when

oqoloT ¥=

0. Since the term

in

AT/1

is

complex

it follows that the

frequency

ce

must be

complex

also. Write ro = (j),

+ io)i

and use

the fact that near the critical

point

coi will be small.

This suggests that a linearization of

(33)

in roi would be

appropriate. Separating

the

equation

into real

and

imaginary parts :

(7)

1368

At the

instability

threshold Wi = 0 and

(54)

and

(55)

become two

coupled equations

for Wc and

i1TII.

If

K a

Vq 2 «

Po c

kq6

one obtains the

following

solution to lowest order in

(ôqo/ôT) :

The

principal

effect of the

temperature dependence

of the

pitch

is to

give

a real

part

to the

frequency

of

the

soft

thermal-orientation mode. To first order in

(ôqo/ôT),

the

imaginary

part

of o)i

is still

given by (51).

It vanishes at R =

Rc(q),

whereas

(56) clearly

remains finite. The convection cells are

oscillating

rather than

steady.

The mechanism

leading

to the

oscillation is shown in

figure

1 for the case where

the

pitch

decreases with

temperature

and

Ka 0 (this

is the usual case in

cholesterics).

The bottom

plate

is hotter than the

top

so that the

spacing

of the cho-

lesteric

layers (the

surfaces of constant

phase)

increases

from bottom to

top.

Consider a distortion like that in

figure

1. It

requires

more elastic energy to distort the

layers

near the bottom because the local value of qo =

(dPs/dz)

is

larger

there

(see

eq.

(26)).

For

xa 0 the heat flux will be deviated

along

the

layers

because the thermal

conductivity

is

largest

in this

FIG. 1. - Horizontal cholesteric subjected to a temperature gra- dient. When xa 0 the heat flux is deviated along the layers and there appear warmer (w) and cooler (c) regions. The gravity forces

induce opposite velocities in regions (w) and (c). For ôqo/ôT > 0 the layers near the top distort more than those near the bottom.

direction. This causes lateral warmer

(w)

and cooler

(c) regions.

The

bouyancy

forces induce

opposite hydrodynamic

velocities in the

regions (w)

and

(c)

and one can see from

figure

1 that this will tend to distort the

layers

more. This is the

origin

of the insta-

bility,

as discussed in reference

[4].

The

instability

first appears when the

bouyancy

forces are

large enough

the overcome the elastic forces of the

layers

near the

top. However,

near the

bottom,

more elastic energy is

required

than is available from the

bouyancy

forces. There will be an

exchange

of energy between the

bouyancy

forces

giving

rise to convection

(kinetic energy)

and the elastic forces of the cholesteric

(potential energy).

This leads to an oscillation with the

period being

determined

by

the

restoring

force

of the bottom

layers,

and

by

the time it takes for

an elastic distortion to

develop

under an

applied temperature gradient.

It is

possible

to

develop

an

analogy

with a

simple

mass

spring

system. From

(26)

the effective value of qo at each

point

in space is :

The différence in elastic

energies

between that at

the bottom

plate

and that at a

point (z)

in thé fluid

is then

This has the same form as the

potential

energy of a

linear

spring.

The

restoring

force

(and

hence the

frequency

of the

oscillation)

will increase with

ATIL

and

(8qo/8T).

Let us estimate the order of

magnitude

of the

oscillating frequency (56). Suppose

that

this

corresponding

to a 1

% change

in

pitch

per

degree.

With

fipo g -- 1, (KI Po Cp) 10-3 (these

values are

(8)

typical for

most

organic liquids), k22 10-6,

qo

103

we get

where all

quantities

are in cgs units and

N(H)

is

given by

For 1 -

10-1

cm,

(60) gives

co, --

10-1 N(H) (rad/s).

For Ka

0

(the

usual case in

cholesterics), N(H)

will be of order

unity except

when H is close to a

cri tic al field He :

For H

Hc, N(H)

becomes

large

and the oscillation

frequency

will reach a maximum. In

principal,

the

oscillations could be observed

by placing

a thermo-

couple

in the system,

provided

that this does not

appreciably

distort the orientation.

Oscillating

tem- perature fluctuations would then be observed at the

frequency

úJr near the critical

point.

The effect of the

temperature dependence

of the

pitch

is to lower the threshold

(57) slightly. ,Using

the same estimates as before we find that the

ôq,IôT

term in

(57) represents

about a 1

%

correction. For H -

Hc

this correction can be

considerably larger,

however.

It is not difficult to show that the shear-orientation mode

(39)

is

only slightly

affected

by

the

instability.

To first order in

ôqo/ôT we

find

that ct

and T are

modified to :

where

co)

is

given by (38).

The

damping

factor is

decreased when

ATIL

and

ôqolôt

have the same

sign,

and increased when

they

have

opposite signs.

4. Discussion. - We have

presented

a treatment

of the

problem

of a horizontal cholesteric fluid

layer subject

to an external

force resulting

from the combi- nation of a vertical temperature

gradient

and

gravity.

We

have

investigated

the

dynamics

of the system which evolves towards

instability

when the extemal force increases up to a critical value.

An

important

feature of the

temperature-indepen-

dent

pitch problem

is that the

frequency

of one of

the normal modes of the system vanishes at the critical

temperature gradient (a

soft

mode).

This means the

relaxation time associated with random fluctuations

of the mode tends towards

infinity

as the

instability

is

approached

from below. Near

enough

to the insta-

bility

this slow mode will dominate the

hydrodyna-

mical response of the fluid because all of the other modes of the system

decay

much faster. In other

words,

if an

arbitrary

extemal

perturbation

is

applied

to the

fluid,

the

dynamical

response function will contain several terms for each of the different

possible

modes of the system. As the critical

point

is

approached (R - Rc)’

one of the terms

(that corresponding

to

the soft

mode)

will

diverge

as

(Rc - R ) -1

while the

other terms will remain finite at R =

Rc.

An infinite response function indicates that the system is unstable

against

fluctuations of the soft

mode,

and will

undergo

a transition to a new stable state - in this case a state

representing steady

convection.

This behaviour is somewhat similar to what occurs near the critical

point

in an

equilibrium phase

tran-

sition. It is well known

[15]

that there is a considerable

slowing

down of certain

thermodynamic

fluctuations

near the critical

temperature Tc.

One or more of the

modes goes soft at a critical

wavevector qc;

i.e.

w(qc)

= 0 at T =

Tc.

The system becomes unstable

against

the excitation of these modes and the static response function

(which

is

usually

the w = 0 limit

of the

dynamical

response

function) diverges.

The

soft mode is the one that is

predominantly

excited

and we say that the system condenses into this soft mode. The

divergent

response function indicates an

instability

and the system evolves into a new stable state with a

qualitatively

different symmetry

(a phase transition).

Of course the nature of the

soft

mode near a convec-

tive

instability

will differ from system to system, as in the case of

phase

transitions. In a

simple

fluid it is

essentially

the thermal conduction mode that is affected. Both of the

overdamped

shear waves are

approximately uncoupled

and well behaved at R =

Rc.

In the case of a nematic lined up

parallel

to the boun-

daries it can be shown

[16]

that the director

mode nz

is the one to go

soft ;

the thermal mode is finite at R =

Rc. Finally,

we have seen in the case of a choleste- ric that a

coupled

thermal-orientation mode is affected while the other modes are finite at R =

R .

In the

appendix

a

general

argument is

given

which

shows that the nature of the soft mode is determined

by

the slowest mode of the

unperturbed

system.

When the

temperature dependence

of the

pitch

is

taken into account

only

the

imaginary part

of the mode

frequency vanishes;

the real part remains finite at the critical

point R

=

Rc.

This

implies

that the convection cells are

oscillating

rather than

steady,

and that a

propagating

thermal conduction mode can exist in a cholesteric under an extemal

temperature gradient. Oscillating

convection cells also occur in certain

binary

fluids

[14],

and in

general

viscoelastic fluids

[17]

under certain conditions.

Finally,

our calculation is

only

valid for a choles- teric whose

pitch

is

weakly dependent

upon tempe-

(9)

1370

rature

(oqoloT small).

The case of the

pitch strongly dependent

upon

temperature

will

require

a numerical

solution and will

probably

lead to very

complex stability diagrams

which will

depend

on the form of

the function

qo(T).

APPENDIX

In the presence of the

temperature gradient

the

mode structure of the system can be written as

The number of modes N

equals

the number of inde-

pendent hydrodynamical

variables of the system.

F is the external force which is

proportional to AT/1.

If F =

0,

the N solutions of

(A. .1)

are w

= iii,

and

these are the normal modes of the system.

They

may be

propagating

or

overdamped depending

on whether

the

Qn

are

real, complex,

or

imaginary.

The threshold is

given by

Suppose

that one of the

Qm

say

Qg

is much smaller than any of the others for the

appropriate q

which

is determined

by

the

geometry

of the system. Since

we know that OJ --+ 0 at the threshold for

steady convection,

we can write

(A. 1)

as

The soft mode is

given by

For Q.

real

(the

usual

case), (A. 4)

describes an over-

damped

mode which goes soft at the threshold when

(A. 2)

is satisfied. If it so

happens

that

Qs

is

imaginary,

the soft mode will be a

propagating

one and since

propagating

modes occur in

pairs (A. 4)

will be

replaced by

so that the critical exponent is

now 1/2

rather than 1 for

the

overdamped

case.

In a

simple

fluid there are five modes : two sound waves, two

overdamped

shear waves, and one over-

damped

thermal conduction mode.

Except

for

liquid

metals it is true that

(x/po cp) « v/po

so we have

Qr,

=

kq2/ po cp,

and the soft mode is a thermal conduction mode. Next consider a nematic

initially

lined up

parallel

to the

plates.

There are in addition

two

overdamped

director modes of the form

Qd kq2/y,

but

only

one of these is

coupled

into

thé

problem

for

parallel alignment.

Since

kly *« x/po

cp

we have

Qs

=

Qd

and the director mode is the one to go

soft;

the thermal conduction mode remains finite at the threshold.

Finally,

consider the case of a

cholesteric. The director mode is

strongly coupled

to the shear mode

corresponding

to a reactive

exchange

of energy between mass motion and the twist elastic forces of the helix

(see

eq.

(37)).

In this case the thermal

mode has the lowest

frequency

so that the soft mode

is

again

the thermal conduction mode.

References [1] See e.g. CHANDRASEKHAR, S., Hydrodynamic and Hydroma-

gnetic Stability (Clarendon, Oxford) 1961, Chap. 2.

[2] LEKKERKERKER, N. and BOON, J. P., Phys. Rev. A 10 (1974)

1355.

[3] THOMAS, H., IEEE Trans. Magn. 5 (1969) 874.

[4] DUBOIS-VIOLETTE, E., J. Physique 34 (1973) 107.

[5] DUBOIS-VIOLETTE, E., C. R. Hebd. Séan. Acad. Sci. 273B

(1971) 923.

[6] LUBENSKY, T. C., Mol. Cryst. Liq. Cryst. 23 (1973) 99.

[7] LUBENSKY, T. C., Phys. Rev. A 6 (1972) 452.

[8] MARTIN, P. C., PARODI, O. and PERSHAN, P. S., Phys. Rev.

A 6 (1972) 2401.

[9] FRANK, F. C., Discuss. Faraday Soc. 25 (1958) 19.

[10] LESLIE, F. M., Proc. R. Soc. A 307 (1968) 359; Arch. Ration.

Mech. Anal. 28 (1968) 265.

[11] See e.g. MONIN, A. S. and YAGLOM, A. M., Statistical Fluid Mechanics (MIT, Cambridge, Mass.,1971) Vol.1, p.179 ff.

[12] FAN, C., KRAMER, L. and STEPHEN, M. J., Phys. Rev. A 2 (1970) 2482.

[13] GUYON, E., PIERANSKI, P., C. R. Hebd. Séan. Acad. Sci. 274B (1972) 656 ;

DUBOIS-VIOLETTE, E., GUYON, E., PIERANSKI, P., Mol. Cryst.

Liq. Cryst. 26 (1974) 193.

[14] SCHECHTER, R. S., PRIGOGINE, I. and HAMM, J. R., Phys. Fluids

15 (1972) 379.

[15] SCHNEIDER, T., SRINIVASAN, C. and ENZ, C. P., Phys. Rev. A 5 (1972) 1528.

[16] PARSONS, J. D., to be published.

[17] SOKOLOV, M. and TANNER, R. I., Phys. Fluids 15 (1972) 534.

Références

Documents relatifs

The main conclusion of this paper is that although the LG equation has been used since several decades to study the LC dynamics close to the FT, it is actually useless, because in

Clusters of planar-aligned short-pitch cholesteric liquid crystal spheres generate dynamic colour- ful patterns due to multiple selective reflections from the radially

RBsumB. - Une thkorie molkculaire statistique est prksentk qui dkcrit le comportement du pas en fonction de la tempkrature dans les cristaux liquides cholestkriques.

Abstract 2014 The temperature variations of the order parameter invariants are measured in the uniaxial and biaxial nematic phases of different mixtures of potassium

Non linearities close to the thermal threshold in a planar nematic liquid

Bychkov et al [7a] and Gor’kov and Dzyaloshinskii [7b] pointed out several years ago that the occurrence of the Peierls instability, in the spirit of the Landau mean

critical dilation 5th of a homeotropic sample (the smectic layers are parallel to the bounding plates) the layers undulate in order to compensate the increase

Hence if director fluctuations are unimportant within the mean field, we can say immediately that the transition to the smectic state will occur at a temperature AT,