• Aucun résultat trouvé

Properties of uniaxial nematic liquid crystal of semiflexible even and odd dimers

N/A
N/A
Protected

Academic year: 2021

Partager "Properties of uniaxial nematic liquid crystal of semiflexible even and odd dimers"

Copied!
21
0
0

Texte intégral

(1)

HAL Id: jpa-00247863

https://hal.archives-ouvertes.fr/jpa-00247863

Submitted on 1 Jan 1993

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.

Properties of uniaxial nematic liquid crystal of semiflexible even and odd dimers

Eugene Terentjev, Rolfe Petschek

To cite this version:

Eugene Terentjev, Rolfe Petschek. Properties of uniaxial nematic liquid crystal of semiflexible even and odd dimers. Journal de Physique II, EDP Sciences, 1993, 3 (5), pp.661-680. �10.1051/jp2:1993159�.

�jpa-00247863�

(2)

Classification Physics Abstracts

61.308 36.20

Properties of uniaxial nematic liquid crystal of semiflexible

even

and odd dimers

Eugene Terentjev (~)

and Rolfe G. Petschek

(~)

(~) Physics Department, Case Western Reserve University~ Cleveland~ OH 44106, U.S.A.

(~) Cavendish Laboratory, Madingley Road, Cambridge CB3 OHE, Great-Britain

(Received

2 November 1992, accepted in final form 8

February1993)

Abstract. A microscopic model is developed to describe statistical properties of a nematic

phase consisting of semiflexible dimers. The effect of the spacer chain bonding the two mesogens is described by the local conditional probability in terms of a bare stiffness parameter Q +J EB

/kT

and a bare equilibrium angle bo between mesogenic monomers. In the framework of a molecular field approximation, taking into account long-range attraction and steric repulsion between

monomers, we obtain a complete statistical description of the nematic phase with expressions for order parameter, mean-field potential, phase transition parameters and Frank elastic constants

depending

on details of spe&tic molecular structure. Comparison of results for dimers that are straight and bent in equilibrium and for corresponding monomers reveals the effect of molecular biaxiality and biaxial fluctuations.

1. Introduction.

Thermotropic

main-chain

liquid crystalline polymers

and

oligomers

have been the

objects

of extensive

study

in recent years. In the nematic

phase

there is a

preferred

axis of orientation of mesogens which are bound to the flexible spacer chain. We

develop

here a consistent molecular-

statistical

theory

of

a

liquid crystal composed

from such molecules. This

approach

is

designed

to

predict macroscopic properties

of the material in terms of

specific microscopic

parameters

of the

mesogenic

monomers and the characteristics of the

bonding

chain. We treat this chain in a continuous way,

assigning

two parameters to it: the bare

rigidity

Q and the bare

angle

00 of the bend between its terminal groups.

If,

for

example,

the

aliphatic

chain

[-CH2-]n

in all-trans

configuration

is

considered,

one expects 00 = 0° in the

ground

state for all even

n. For an odd number of

CH2

groups the

angle

00 is around 70° for the

ground

state of the free

chain;

the presence of

bulky

and

strongly interacting

mesogens on both ends

presumably

reduces this

angle, depending

on the

length

of the chain and its

rigidity

Q.

This latter parameter is

especially important

to

investigate

when

dealing

with chains of ne- matic mesogens, because it determines the

specific regimes

of the chain behaviour of

longer

(3)

polymers.

When Q «

I,

one arrives at the case of

freely jointed

monomers,

becoming

com-

pletely free,

at Q

= 0. As Q - oo the spacer

approaches

the limit of a

rigid rod, straight

or

bent, dependent

on the value of 00. The intermediate

region

of Q

(

I defines the most com-

mon case of semiflexible

bonding,

when the

tendency

of

subsequent

monomers to

align along

their

equilibrium

direction competes with the nematic mean

field,

which acts on each monomer

separately

and tends to

depress

fluctuations of a molecular

shape.

This

gives

rise to

compli-

cated

dependences

of

macroscopic,

observable

properties

on the parameters of intermonomer interactions and the spacer internal

rigidity

and bend.

A favourable

object

to

investigate

these

properties

is a dimer

composed

of two

nematogenic

monomers, where the influence of the spacer is not obscured

by

the entropy effects

imposed by

the

long

chain

[Ii.

The purpose of the present paper is to

apply

a

general

statistical

approach

to the case of a

thermotropic

nematic

liquid crystal composed

of such dimers and to obtain observable

macroscopic

characteristics of the system.

Before

proceeding,

we note that over the years countless observations of so-called "odd- even" effects have been made

(see,

for

example

[2,

3]).

The reason for the observed behaviour is clear, since the

ground

state bend between

subsequent

mesogens makes the

corresponding

dimer

effectively

biaxial. A detailed theoretical

study

of this

question

has been made

recently by

Heaton and Luckhurst [4]. That

work, along

with other related

publications

[5],

employs

a

Flory description

of the conformational statistics of a spacer chain based on the torsion rotations of the carbon-carbon bonds. These authors

calculated,

in the limit of small order parameter, the N-I transition temperature and the bifurcation

point,

where the non-zero solution for order

parameter

appears. Both these

quantities

show

significant

"even-odd" effect. It has to be noted that this version of the

Flory approach

to the chain

statistics,

while

suiting

the chemical view and often

giving

a

good

agreement with

experiment,

is inconsistent from the

point

of view of statistical mechanics. In this

approach

the

probability

distribution is written

arbitrarily

from symmetry arguments and does not,

generally,

minimize the free energy. As a

result, properties

based on

single particle

averages, such as order parameter, bifurcation temperature, etc.,

usually

are determined

correctly

in this

model,

whereas characteristics based on the free

energy

(mean-field coupling

constant,

elasticity, etc.)

are often

non-adequate.

An

alternative, Onsager's approach

based on the virial

expansion,

is free from this

disadvantage

and has been

preferably

used over the years in molecular theories of

polymers

and

liquid crystals.

This

approach

offers a different set of

approximations

from that in the model of a torsion chain and is

specifically

suitable for mean field calculations of the free energy.

In this paper we present an alternative

(to

the Ref. [4]) way of

calculating properties

of

semiflexible dimers that is based on the virial

expansion

of the free energy. This

approach

for dimers is

essentially

the

opposite limiting

case of the model that we used to calculate curvature elastic constants of a nematic of

long polymer

chains

iii.

The difference between these two

limiting

cases is that for the

long

semiflexible chain one is allowed to use the

(approximate)

differential

equation

for the chain propagator,

neglecting

end effects.

Very

short

chains,

dimers in

particular, require

the discrete treatment of

subsequent

monomers. Our model

employs

the concept of a local conditional

probability

for the relative orientation of bonded monomers that describes the spacer

flexibility by

means of an

integrated

parameter Q. In this way both conformational

(trans-gauche) changes

and C C bonds fluctuations of the spacer are accounted for as thermal excitations around some

ground-state configuration parallel

for even dimers and bent to the

angle

00 for odd dimers. We do not take into account biaxial

ordering and, therefore,

our results are

quantitatively applicable only

to the case of a

regular

uniaxial

nematic.

Nevertheless,

since the effective

biaxiality

of the odd dimer is included in this model in the

form, averaged

over the molecular rotation about its

long axis,

many of our results will

show a considerable effect of the

shape

on the observable

macroscopic properties.

(4)

The purpose of this work is to describe in a framework of the

relatively simple theory broad, qualitative effects,

caused

by

the

flexibility

of the spacer and the

equilibrium

bend between the bounded mesogens. We shall use the mean-field

approximation

in order to obtain some of the results in

analytical

form. We

emphasize,

that this is a first necessary step in

describing

such

a

complex

system; some

phenomena (such

as biaxial

ordering,

mentioned

above)

will

require

a

more detailed treatment.

2.

Equifibriurn

statistics.

We consider a dimer as a sequence of two rod-like mesogens each of

length

and width

d, figure

I. It is necessary to

give

a consistent

sign

to the orientation u of each of these

rods,

because the

probable position

of the

adjacent

monomer

depends

on this

sign.

We define

P~(r r',

u,

u')

to be the

probability

that on an isolated dimer the mesogen with its center of symmetry

position

r and orientation u is followed

by

another mesogen, which has the

position

r' and orientation u'. Relative coordinates of the two monomers are

always

bounded

by

the constraint

&(r-r'- )I[u+u'])

if we assume a constant

length

for the spacer. We do not consider here the limit of

freely jointed

mesogens for their

properties

are not very much different from those for free monomers. In the

opposite limit,

when the mesogen orientation is

strongly

determined

by

the

adjacent

mesogen, a reasonable

approximation

for the

bond-probability P~

is the Gaussian. For the even spacer

chain,

I-e-, for the dimer that is

straight

in its

equilibrium

all-trans

configuration,

we write:

P~(u, u')

m

(

~ exp

[Q(u u')] &(r

r'

)I[u

+

u']), (I)

sin

where the

coupling

Q is determined

by

an effective energy of

bending

of the spacer of constant

length,

Q

r-

EB/kT

> I and

essentially incorporates

all mechanisms of the chain

flexibility.

In the case when the spacer chain

imposes

a finite

equilibrium

bend 00 on the dimer the bond-

probability

has its

ground

state maximum around

(u. u')

= cos 00, the width of this distribution is still determined

by

a bare

rigidity

parameter Q. We will use a similar Gaussian

P~ (u, u')

m

~

exp

~~

(u

u')

+

2V5sin( fij

&(r

r'

jl[u

+

u']), (2)

+ C°S

o

~~

i

~~ (~

So

Fig-I.

Mesogenic monomer and its corresponding dimer with inherited bend.

(5)

where

Z~

is the normalization.

Equation (I) implies

that if a

long polymer

chain of these

monomers was

considered,

its

isotropic persistence length (the

so-called Kuhn

segment)

would

be

equal

to al.

Considering equations (1,2),

it is

important

to

emphasize

that

P~

is

a condi- tional

probability

of mutual orientation of the two bound mesogens. It does not include the effect of external fields

or of other

particles,

which will come into the

theory later,

when

writing

down the

one-particle probability

of the

given

monomer orientation. For this reason we refer to the parameters Q and 00 as "bare" characteristics of the spacer; in the

interacting

system their

effective

(observable)

values will be

different,

as we shall see below. Note that

P~ depends

on the mutual

position

of the centers of mass of bound monomers

only through

the exact delta-functional constraint. However, the

symmetric

Gaussian form of equations

(I)

and

(2)

is an

approximation.

This

approximation

is

certainly

valid in the case of the weak

long-range

interaction between mesogens. If the two

mesogenic

monomers

strongly

attract each

other,

in order for the

P~

distribution to

keep

the

symmetric

form of

equation (2)

it is necessary to restrict ourselves from consideration of

large

bend

angles and/or

very flexible spacer chains.

As a

practical

matter this

requires

(00 +

/fl)

< 90°. One can see that the

approximation

is not very

demanding and, therefore, equation (2)

is

general.

In the nematic

phase

the orientational

probability

distribution function

P(u)

for each

given

mesogen may be determined

by integration

of the dimer

pair probability P(u,u')

over the variables of the

adjacent

monomer u'. We use

P~(u,u')

as a correlator and account for the

mean orientational

field, acting separately

on each

mesogenic

monomer:

P(u,u')

=

je~P~l(~'~)lP~(u,u')e~P~l(~"~ll

P(U)

=

/ P(U> U')dU' (3)

Here Z is the normalization

factor, fl

=

(kT)~~, U((u n))

is the mean-field

potential acting

on a

given

monomer, and n is the nematic director.

Generally,

this

potential U((u n))

has

an

axially symmetric

form on the unit

sphere

with two

deep

wells at the

poles (n, -n)

and a barrier of

height

J at the equator

(uIn).

To make the molecular-statistical

description

of our system self-consistent we now need the free energy functional in the mean field

approximation,

which

depends

on the

single particle probability P(u), equation (3),

and for which the mean field

potential

is a

minimizing

function.

This free energy takes the form

(compare

with

[Ii):

F = p~

/ P(ui)P(u~ )U~~'(1~ 2) dridr2dui

du2

jNkT

In Z N

/ U(u, n)P(u)du (4)

where the

partition

function of a semiflexible dimer is

Z =

e~P~(~'~lP~(u,u')e~P~(~"~ldu'du (5)

and N

=

pV

is the total number of mesogens in the system. The last two terms of

the free energy

equation (4)

reduce to the usual orientational entropy contribution Fjoc =

pkT J

PlnPdudr in both

limiting

cases

(Q

-

0)

and

(Q

- oo; 00 -

0).

These

correspond

respectively

to

freely jointed

or unbounded mesogens, and to

rigidly

bounded mesogens which form a

straight rigid

rod.

(6)

Direct

varying

of

equation (4)

shows that in order for the

probability P(u)

in the form

equation (3)

to be the minimum of the free energy

functional,

the mean-field

potential

must be

exactly

the

one-particle

average of an effective

pair potential:

U(ui, n)

m p

/ P(u2)U~~'(1, 2)dr12du2

" p

/ U~~'(1, 2)P(u2 u[)dr12du2du[. (6)

Here p is the monomer number

density

and the effective

pair potential

[6] is a sum

(see Eq.(7)

below)

of a

long-range

attractive part, which is modulated

by

steric

repulsion,

and the

packing

entropy term, which is non-zero

only

when the two momoners are in contact. Note that these

interacting

monomers

belong

to different

dimers,

because all

pair

interactions between the bounded mesogens are assumed to be accounted for in the

bond-probability P~

The effective

interaction

potential

is

given by

U~~.

(1, 2)

re U~~~.

(1, 2) 8(1, 2)

+

~~

[l 8(1, 2)] (7)

VOP where

U~~~'(1,

2)

G3

-( ((Ul 'U2)~ ((Ul 'U2)(Ul 'r12)(U2 'r12)

+..

(8)

12 12 12

The factor

(I-vop)

in

equation (7),

where vo is the monomer

volume,

accounts for the correction to the

Onsager approximation

for the dense

packing,

which was discussed

by

Gelbart and Baron

iii.

The steric

repulsion

is accounted for

by

a step function

8(1, 2)

=

8((12

r12)> which is

equal

to

unity

when the distance between the two monomers centers of mass r12 is greater than the

corresponding

contact form-factor

(12 (closest approach distance).

In the case of unbound

rod-like monomers this form-factor in the absence of steric

dipoles

is

~12 * d +

~j~

[(Ul r12)~

+ (U2

'r12)~j (~)

With an accuracy

r-

d/I (cf.

[8], for

example)

the

corresponding

steric

integral

for dimers will

give

Ill 8(f12 r12)jdr12

*

2dl~

ui x u2 + ui x

u[

+

u[

x u2 +

u[

x

u[ (10)

where the

prime

denotes the monomer

adjacent

to the

given

one on its dimer. In the

limiting

case 00 " 0 and Q = oo

(rigid straight dimer)

this

integral

will be four times

bigger

than that

for unbound monomers. It is

important

to

emphasize

that the attractive

potential

contribution to the mean field is not so sensitive to the difference between dimers and unbound monomers.

Indeed,

since the attraction

potential equation (8)

is a

rapidly decreasing

function of r12> the main contribution to the

integral J

U~~~(1,

2)8(1, 2)dr12

is determined

by

the distances r12

'~

d

(side

to side

packing

of

monomers). Therefore,

with an accuracy

r-

d/I,

effects of attraction to

adjacent

monomers are

insignificant

in

equation (6).

One has to carry out the

angular integration

in the

equation (6)

with the

two-particle prob-

ability

distribution

P(u, u')

in order to account for the

flexibility

of the spacer. The

simplest possible

form of mean field allowed

by

the symmetry is U

= Uo

J(u n)~,

an

approximation

(7)

that works

fairly

well if the

coupling

constant J is

sufficiently large.

An

approximate analytical expression

for

J,

which is

proportional

to the nematic order parameter, takes the form [9]:

J re

)p ($G

~ + (12 +

)lo)

+ ~~~

Tj

S e

)HS (II)

VOP

where

G,

lo and 12 are the coefficients

appearing

in

equation (8),

the nematic order parameter is S

= -0.5

+1.5( (u n)~)

and where

equation (II)

defines the material parameter H with

dimensionality

of energy.

3. N-I

phase

transition.

It is

interesting

to

investigate

how the

properties

of the N-I

phase

transition in a system of semiflexible dimers

depend

on the spacer

rigidity

Q and bend

angle

00. Since we are

using

the

simple

mean field

approach

in this paper, the limit Q = 0

exactly corresponds

to a Maier-

Saupe

case, with the

only important

difference that we estimate steric

(entropic)

and

energetic

contributions to the

coupling

constant J

independently. Thus,

the

properties

of monomer nematic will serve us as a reference basis with the transition at

kTjjl

m 0.15H

(I.e. J/kT

m

4.55, see Eq.

(II)),

and order parameter

jump

at the transition AS m 0.44. We define the

point

of the first order N-I transition as the temperature at which the free energy of the

non-symmetric phase equation (4)

becomes lower than that of the

isotropic phase.

There are two other

important (observable)

characteristic temperatures,

describing

this transition the ultimate

superheating point

Ti where the bifurcation of the non-trivial

(S # 0)

local minimum of the free energy occurs, and the

supercooling point

T* at which the

isotropic phase

becomes

unstable.

So,

we determine

Ti

as a solution of

equation

S

= -0.5

+1.5((u n)~)

=

l.5J/H,

and T*

by deriving

the first term of the free energy

equation (4) expansion

in powers of small S: F Fo '~

ao(T T*)S~

+ The

equilibrium

transition temperature TNT is deternfined

by

a direct

integration

of the free energy

equation (4).

For small Q « I we can

expand

the

complicated bond-probability exponential

and calculate

corresponding integrals analytically. Keeping

in nfind that the

approximation

for P~

equation (2)

will not be

good

at very small Q when there is a strong

long-range

attraction in the system,

we write down the

following

estimate for the transition temperature

(determined by

the Landau free energy

expansion

[9]):

~

~(oj ~

~ ~

~Q sin(00/2)

~~

Q~

~~ ~

~~ ~

l + cos 00 ~

(l

+ cos 00)~

~

Q~[0.44sin(00/2)

+ 1.76

sin~(00/2)]

+

(12) (1+

C°S

°o)

One can

clearly

observe the

competing

action of the spacer

rigidity

Q and the bend

angle

00.

Straight (even)

dimers have 00 " 0. TNT increases

together

with Q until it reaches saturation at very

rigid,

Q »

I,

dimers. Nonzero

equilibrium

bend between mesogens

composing

the flexible dimer reduces its

mesogenic

power

and, consequently,

decreases the N-I transition temperature when the spacer

rigidity

increases.

When Q is not small

(and

we expect

typical

values of Q to be between 3 and 20 for

[CH2]n

spacers with n

varying

from 16 down to

2,

[9]) one has to

perform

calculations

numerically.

Figure

2 shows the evolution of the transition temperature with

increasing

spacer

rigidity

for several fixed values of 00. We can see that in

spite

of considerable

bend,

if the spacer is

rigid

(8)

2

a

~~

i.

~~

~_

0.

O lo 20 30 40 50

Q

Fig.2.

Reduced transition temperature

TNT/7f~,

vs. the spacer stiffness Q: even dimers

(a),

odd dimers vith equilibrium bend 20°

(b),

40°

(c),

60°

(d)

respectively.

Table I. Estimated spacer

rigidity Q,

and the

equilibrium

bent

angle,

00, for

aliphatic

spacer chains with

increasing

number of

carbons,

n.

n 2 3 4 5 6 7 8 9 10 II 12

Q 33.3 24.4 17.2 13.9 11.8 10.2 8.7 7.9 7.0 6A 5.8

bo(°)

0 47 0 40 0 35 0 32 0 30 0

enough,

the dimer remains

mesogenic, though

with

significantly

decreased transition tempera-

ture. For the even,

straight

in

equilibrium,

dimers we have obtained [9] from comparison with

experiment

that Q has a universal

dependence

on n, the number of carbons in an

aliphatic

spacer chain: Q

r-

70/n.

If we assume on this basis that the whole chain

rigidity

is determined

by

the

independent bending

of each C- C

bond,

it is

possible

to estimate Q for odd-numbered

spacers

by interpolation.

Then the

only

unknown parameter for the

given

odd chain is the

bare

angle

00. For each

given

odd-numbered

dimer,

we can estimate the actual bend an-

gle

of a spacer 00

by fitting

the

appropriate

curve in

figure

2 to the

experimental

transition temperature for its

particular

value of Q. For the series of dimers of mesogen

4-n-alkyl-N-[4- n-alkyloxy-benzyliden]-aniline

[10] we obtain in this way the

expected

values for the dimer's

spacer

rigidity

and its average

equilibrium

bend

(see

the Tab.

I).

The numbers

presented

in this table cannot be considered as very accurate, since

they

involve a considerable

experimental

error, our

simplifying approximations

and the

averaging

over all

possible

trans and

gauche

con-

formations in the real system. However, the effect of

straightening

a bent dimer with

decreasing

spacer stiffness under the influence of interaction between monomers (00 '- 70° x n~°.~~ at the transition

point)

is

clearly

described

by

this model.

Note,

that this is a different effect from the

(also present)

additional extension of a dimer affected

by

the nematic mean field.

There are several characteristic parameters of the first order N-I

phase transition,

which can be measured

experimentally.

We consider two of

them,

the width of the transition

hysteresis

AT =

(Ti

T* and the

jump

of the order parameter AS at the transition

point. Again, using

the

expansion

at small Q we obtain the

following expressions,

which

help

to understand the

(9)

- 7

~y

~

b

O

fl

ig.3.

60°

equflibrium bend.

trends of AT and AS behaviour:

~~

~~(~j ~

~

~~Q sin(00/2)

~ ~~

Q~

~ l + cos 00 ~

(l

+ cos 00)~

(l

+

~s

00)~~~~~ ~~ ~~~~~°~~~ ~'~~ ~~~~~~°~~~~ ~ ~~~~

~ ~ ~~~~j

~

~ ~~ Q

sin(00/2)

~ ~~

Q~

~ l + cos 00

(1

+ cos 00)~

Q~

[2.60 sin(00/2)

1.96

sin~(00/2)]

+

(14) (1+

C°S

°o)~

For an

arbitrary

Q these parameters have to be calculated

numerically: they

are determined

by isotropic (at

J

=

0)

averages of

corresponding

mean-field

expressions

if we assume a

weakly

first-order N-I transition. In

figure

3 we

plot

the reduced

hysteresis

width TNT T*. This result then can be

compared

with

experimental

data for dimers of

4,4' dialkoxyphenylbenzoate

"5005"

monomer

ill,

12], where the

supercooling

temperature T* was determined

by extrapolation

of the inverse Cotton-Mouton coefficient

C~~

r-

(T T*).

Since we are unable to estimate

microscopic

attraction

potential

constants

G,

lo and 12 in

equation (8)

with any reasonable accuracy, we take J

=

)HS

from the data for

corresponding

monomers

ill]. Although

our

data for even (00 "

0)

and

small-angle

odd dimers exhibit some increase in the N-I transition

hysteresis

at

relatively

small

Q,

this increase is not so dramatic as has been observed

by

Rosenblatt and Griffin

[I Ii

(TNT T* was observed to increase for the dimer

by

a factor of 7 with respect to its

monomer).

The similar weak

dependence

on Q

(which

describes the

gradual change

from monomers to even

dimers)

is

presented

in

figure

4

(curve a)

the

plot

of the order parameter

jump

AS at the transition. When the

rigidity

of the even dimer

increases,

the characteristic behaviour of the

phase

transition is

exactly

that of the monomers

(except

for the transition temperature

itself, Fig.

2) as is

expected

in a self-consistent mean-field

theory.

(10)

O.

0.

c w

-n

lo 20 30 40 50

Q

FigA.

Jump of the order parameter at the N-I transition vs. Q for even

(a)

and odd dimers lvith 20°

(b),

40°

(c),

60°

(d)

equilibrium bend.

When odd dimers are

considered,

the

equilibrium biaxiality

of a molecule causes noticeable

changes

in the transition behaviour. Both AT and AS are related to the

enthalpy

of the first order

phase

transition which is a measure of

discontinuity

of this transition.

Therefore,

a

significant

decrease of AT and AS for

considerably

bent dimers indicates the

narrowing

of the

region

of two

phase

coexistence when the molecule becomes more biaxial. This

phenomenon

has been observed

by

several authors

(see

[14], for

example)

and it is caused

by

an

approach

to the isolated Lifshits

point

for biaxial nematic on the

corresponding phase diagram.

4. Uniaxial nematic

phase.

In

spite

of a certain success of our

theory

to describe the N-I

phase

transition in the system of semiflexible dimers we note that this

simple

mean-field model is not

designed

for this purpose.

Its weak

point

is its omission of fluctuation effects. It is obvious that biaxial fluctuations are

important

at the

transition, especially

for the

significantly

bent and

rigid

dimers. Far below the transition

point, however,

we expect this model to work

quite

well.

One of the

questions

which has to be clarified at this

point,

is the

configuration

of bent semiflexible dimers in the nematic

phase.

One may expect that such

dimers, being

affected

by

a strong uniaxial mean

field,

will be

effectively extended,

or

straightened

with respect to their

equilibrium shape

in the

isotropic phase,

PO- It

depends, apparently,

on the

magnitude

of Q:

for a very

rigid

spacer the dimer will remain in its

V-shape (Fig.I)

and even as T - 0 the uniaxial order parameter will never reach

unity. Figure

5 shows the order parameter, calculated at the same scaled temperature (TNT

T)/TNT

for dimers with different bare bend

angle

00 and

several fixed values of

Q,

which represent different characteristic

regimes.

One can observe two

phenomena in the

figure

5 an effective

straightening

of the dimers and an increase of effective

stiffness,

as a

secondary

effect.

First,

consider the

points

where all

plots

intersect the ordinate axis these

correspond

to the

even (00 "

0)

dimer. When the spacer is rather

flexible,

the order parameter

slightly decreases,

then it returns back to the hard-rod value as Q becomes very

large (compare

with

Fig.4).

In

spite

of the

increasing

bare bend

angle

Ro> dimers with an

odd,

but flexible spacer do not

(11)

O.

monomer

w O.

lo 20 30 40 50 60 70

6

Fig-S-

Order parameter at r

(=

I T/TNT " 0.05 vs. equilibrium bend angle bo for odd dimers lvith different spacer stiffness: Q

= 2

(a);

Q

= 4,

(b);

Q

= 10

(c);

Q

= 18

(d);

Q

= 45

(e).

show any further decrease in the order parameter; we can even observe a

slight

raise of S. This shows that in the nematic

phase

bent dimers with flexible

(Q

<

5)

spacers are almost

completely

extended. In this

straightened configuration,

relative fluctuations of the bounded monomers

are

depressed

and the

degree

of

ordering

in the system becomes closer to the hard-rod limit.

Bent dimers with a considerable spacer stiffness

(Q

r-

10)

also tend to extend their

isotropic equilibrium configuration,

when affected

by

the nematic mean field. Its

strength, however,

is not sufficient to make the dimer

completely straight,

when the initial

angle

PO is

large enough.

Flatter

regions

on each of

corresponding

curves indicates the limit of initial bend 00> after which the total

straightening

cannot occur.

Finally,

when the spacer is very stiff

(Q

>

20),

the shape of the dimer is not

significantly

affected

by

an external

ordering

field

acting

on its

monomers

and, therefore,

the uniaxial nematic

ordering

decreases

monotonically,

as cos~ 00,

when the molecular asymmetry increases.

Our model allows

investigation

of the

properties

of the spacer and the molecular

configura-

tion in more detail since

equation (3)

is the

joint probability

for the two

adjacent

monomers

orientations. After some

geometric simplifications,

which do not affect the

qualitative

be- haviour of the

dimer,

we can estimate an

increasing

effective stiffness of the spacer and a

decreasing

effective

bending angle

in a nematic

phase (which

we define as a location of the

joint probability

maximum with respect to the

angle

between u and

u'):

Q m Q +

flJ;

~"~ ~

"~~~~°~~~

ii

+

(flJ/2Q)(1

+ cos 00)]~

~~~~

(compare

with the Tab. I in the

previous Sect.).

When the order parameter is close to

unity (which

means that

J/kT

>

I),

dimers may be considered

effectively straightened

since the

average

cos00

- 1. There is a

competition

between

ordering

and spacer stiffness in

equation (IS)

very

rigid

dimers

require

very low temperatures to achieve the same

degree

of

ordering.

In the

isotropic phase,

when J =

0,

the effective and the bare

bending angle

coincide: 00 = 00.

(12)

5. Elastic constants.

The Frank elastic constants are

important

parameters of a nematic

liquid crystal

because

they

influence almost every observable property,

including

those used in

applications.

Therefore we derive

microscopic expressions

for

Ki,

K2 and K3> which

correspond

to

splay (i7 n)~,

twist

(n

i7 x

n)~

and bend

in

x i7 x

n]~

deformations. The molecular-statistical

theory

of curvature

elasticity

for an

ordinary

rod-like nematic has been

developed by

Gelbart [6] for a

thermotropic

system

and, earlier, by Straley

[8] and Priest [15] for a system with

only

steric interactions.

The basic assertion of this

approach

is that

(I)

the response of the

liquid crystal

to an

ordering field,

which varies

slowly

in space, is

just

that the

preferred

orientation

aligns everywhere along

that field and

(it)

the relative

probability

of a molecule at the

point

r

having

the orientation u is

P(u, n(r)),

where P is the same distribution function as in the uniform nematic. Then the free energy

equation (4),

which describes the

coupling

of the two molecules

(u,r)

and

(u', r')

must be

expanded

in powers of the small director

gradients: n(r')

=

n(r)

+

(r

r' i7

n(r)

+...

up to second-order terms.

By averaging

of the

corresponding

correction to F one obtains the

phenomenological expression

for the Frank free energy with curvature constants

expressed

in

terms of molecular parameters. The orientational entropy term r- PIn P does not contribute

to the elastic free energy because of its local nature.

In our case the

single-particle

distribution

P(u,n(r))

includes characteristics of the other mesogen at the

point

r'

(where

r r'

=

) flu

+

u']

is the distance between centers of mass of the two bound

monomers). So,

in contrast to the established theories of nematics for immo- bile

molecules,

there is an additional functional correction to the

single-particle

distribution

equation (3),

which is due to the different director orientation at the

points

r and r'. The variables of the

P(u, n(r))

must include

only positional

coordinates r of the

monomer under

consideration,

thus we have to transform

exp[-flU(u')]

in

equation (3)

to refer to this

point.

Expanding n(r')

in

equation (3)

as indicated above and

keeping only

terms up to the second

order,

we obtain:

~~~~ " ~~~~jiiiil,~l'l lll'lllllll~ fl m i i~~U'~~~' +

~~

?~~~~' ~~~~

here

we

the derivative of

the ean

field

/~" dU/d(u . n)

and Z =

J P(u)du.

Since the

mean

field

potential

U(u, n)

is

determined by a free energy minimum

no

steric dipoles or

chirality

in monomers which would generate on-zero averages with linear

i7n),

there will be no additional change in P(u) due

to the

U(u,n), which would effect theecond-order elastic free energy. The

microscopic expression

for this

F = p~

/ P(ui)U~~'(1, 2) I

+

(r12 i7)

+

)(r12 7)~j

P(u2)drdr12dui

du2

jNkT

In Z N

/ U(u, n)P(u)du (Ii)

where

equation (16)

must be used for the

single-particle probability density P(u).

After sub-

(13)

stituting

the mean field

potential

we obtain the elastic free energy of

a dimer nematic:

Y-Yom

/(Fi+F2]dr,

where the

leading

contribution to the free energy

density

can be written as follows:

~l

~2P~(fl~)~~i

~k Vullj

VP

ill

~ / II ~~~ )~(2~~2~~12j (18)

I I k

~-i-I

k ~ i I-k-I

-I-I-k-1)

~2~2~l~l ~2~2~l~l ~2~2~l~l + ~2~2~l~l

e~P~(~'l P~ (ui

fir

)e~P~(°il e~P~(~21P~(u2, fi2)e~P~(~?l duidu2dfii dfi2

where the

subscript

indexes

(I

and

2)

characterize the

given

monomer of the

interacting pair

under

consideration,

and the

corresponding

fi

gives

the orientation of the

adjacent

monomer

on a

given

dimer. Other terms, that did not contribute to the elastic free energy in the casd of unbound monomers, now take the form:

F2 =

~Pl~ninki7anji7pni (U(u, n)[2(flJ)~fi;fik flJ&;k](fijfiiuoup

+

fijfiifinfip))

8

((U(u, n))

+

kT) ([2(flJ)~uiuk flJ&;k](ujuiunup

+

ujuifiofip )) (19)

Integration

over r12 in the

equation (18)

can be carried out

analytically,

if one uses

approximate

methods with small parameter

(d/I).

The

result, however,

is very

complicated.

In this paper

we are

dealing

with

qualitative

effects which describe the difference between monomeric and dimeric

liquid crystals.

Let us make a

rough

estimate of this

integral

which will preserve all effects of

bonding,

but which will make

equations

much more

simple

and clear. This is

possible

if we consider

separately

the

integration

of

U~~.(r12)

and

r(~r(~

in equation

(18).

This will

cause an error of the order

(d/1)

r- 0.2 in

typical

materials. This error,

however,

does not exceed the accuracy of other

approximations

we have made, for

example,

in

equations (8)

and

(9)

for the effective

pair potential

U~~. Denote

li

the

average

pair potential

and v12 the excluded volume for the two monomers

li(1, 2)

=

/ U~~'(1, 2)dr12

G3 A +

B(ui u2)~

+ V12

v12 =

Ill 8(1, 2)]dr12

G3

2dl~

ui X u2

The

averaged pair

interaction energy

depends only

on the

product (ui

.u2 and can be

expanded

in

Legendre polynomials

of even index. This

expansion

has a small parameter

(d/I) (in

addition to

independent anisotropic

contributions see

equation (8))

and is well

justified. Comparing

with the mean-field

potential gives readily:

J =

-pBS.

Now we rewrite the second virial part of the elastic free energy

density equation (18)

note that the term with A vanishes

exactly

for symmetry reasons:

~

p2j3

Fi m

jpl~~nknpi7nn1i7pnq (u[u[u(u[)(u[u(u(u(u[u~)+

~

(20)

+(~~~~~~~~)(~~~~~~~~~~~~)+(~~~~~i~~~~~~)(~~~~~~~~)+

(~~~~~~~~)(~~~~~(~~~~~~))

(14)

where the

averaging

must be carried out with the

two-particle probability

distribution

P(u, fi) given by equation (3).

In the case Q - 0

only

the first average survives in the

equation (20)

and it

gives

us Frank elastic constants:

ICI =

3K2

=

~~

Pl~

fI~H~S~ I

+

~S ~~

4j

(1

~~P4 +

~~

P6)

,

(21)

42525 7 7 II II

1(~ = ~~

pl~fl~H~s~ 1

+

~

S ~~

4j

II

+

~

S ~~

P4 ~

P6) (22)

42525 7 7 3 385 231

Here P4 and P6 are

equilibrium

averages of the fourth and sixth order

Legendre polynomials (which

are

normally

much smaller than the main order parameter

S,

I-e-, than the average of

the second

Legendre polynomial)

and the energy parameter H is determined

by equation II):

J =

2HS/3.

The exact ratio

Iti/1(2

" 3 is a consequence of

accounting only

for the first term

B(u .u')~

in

the average

pair potential li(1, 2);

we have

already acknowledged

that corrections of order

(d/I)

and

(I/G) (see equation (8))

are

expected

in a more exact evaluation of the free energy

density

Fi

Keeping

in mind this level of

approximation

of the Frank elastic constants

estimates,

we

still can observe several

important

features of their behaviour.

First,

at small order parameter

all constants increase as

S~,

which is consistent with observations and other models [6]. The

ratio

K3/1(1

'~ 1+ 0.71S +

O(P4)

is

increasing

when the temperature decreases from the N-I transition.

And, finally,

the order of

magnitude

of Frank elastic constants is close to the

observed values

It m I-S X

l0~~pl~S~(flH)~kT

r-

~kTS~

r- 3 X

10~~S~ dyn

Other terms in the elastic free energy,

given by equation (19),

are caused

by

the non-local correction to the

probability

distribution

equation (Ii).

It is not not so easy to estimate these

terms

analytically. However,

there are two

limiting

cases: Q - 0 and Q - oo, that

correspond

(at

bo "

0)

to the

ordinary

rod-like nematic with molecules of

single (I)

and double (21)

lengths,

in which such estimation can be carried out. It is very

important

to

emphasize,

that these two cases do not

simply correspond

to a - 21substitution in results of other models [6,

8],

because such

change

must be

accornpanied by

a

corresponding change

in parameters of

pair

interaction: the

strength

and

anisotropy

of

dispersion

forces for the

effectively

doubled

particles.

This effect is

naturally

accounted for in our

approach,

which considers each monomer

separately. So,

both for Q - 0 and Q - oo the

single-particle

corrections in the free energy transform into

J

P In Pdu. For Q

= 0 the correction to

P(u)

vanishes

exactly,

while for Q >

flJ

we obtain the additional contribution to the free energy

density:

F2 m

pl~J 2flJ[(uiujukuiupuq

In

P(u)) (uiujukuiupuq)(In P(u))]np

nq +

+(u;ujukui

In

P(u)) (uiujukui)(In P(u)) i7;nji7kni (23)

where P

r-

exp[-2flU].

This contribution is

small, compared

to the

pair

interaction free energy

Fi which becomes for Q

- oo

Fl *

)P~~~~~ l'~~'~~'~~'~~)l'~~'~~'~~'~~'~~'~~)~kllpi7alll~flllq (~~)

Références

Documents relatifs

described in reference [2].. The dots and circles are calculated with viscoelastic ratios obtained with the help of computer fitting procedures, and using elastic

compounds : open circles, electrical measurements; filled circles, magnetic measurements. measurements of density or susceptibility have so far extended; we used

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

amplitude modes in the biaxial phase mix the components of biaxiality and birefringence of the order parameter, the critical (quasi-biaxial) mode presenting

2014 The influence of permanent dipoles on the nematic isotropic transition is studied by adding a simple polar interaction to the Maier-Saupe model of nematic

Precise measurements of the Kerr effect [6, 9] and magnetic birefringence [6, 7] in the isotropic phase of. some nematic substances have been

Knowing these masses, which were determined by weighing before and after each run, and knowing the densities p(T) of the nematic and of water as functions of

surface density of carriers, determined by the initial concentration of carriers and the dark current of termogeneration of semiconductor NSPh = GPh L, AtPh is