• Aucun résultat trouvé

Study of the phase separation in polymer solutions using the pair approximation of the CVM: analytical solution of the problem

N/A
N/A
Protected

Academic year: 2021

Partager "Study of the phase separation in polymer solutions using the pair approximation of the CVM: analytical solution of the problem"

Copied!
24
0
0

Texte intégral

(1)

HAL Id: jpa-00247985

https://hal.archives-ouvertes.fr/jpa-00247985

Submitted on 1 Jan 1994

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.

the pair approximation of the CVM: analytical solution of the problem

F. de J. Guevara-Rodríguez, F. Aguilera-Granja, Ryoichi Kikuchi

To cite this version:

F. de J. Guevara-Rodríguez, F. Aguilera-Granja, Ryoichi Kikuchi. Study of the phase separation in polymer solutions using the pair approximation of the CVM: analytical solution of the problem.

Journal de Physique II, EDP Sciences, 1994, 4 (4), pp.589-611. �10.1051/jp2:1994150�. �jpa-00247985�

(2)

Classification Physics Abstracts

05.20 05.50 61.25H

Study of the phase separation in polymer solutions using the

pair approximation of the CVM: analytical solution of the

problem

F. de J.

Guevara-Rodriguez (~),

F.

Aguilera-Granja

(~) and

Ryoichi

Kikuchi (~)

(~ Instituto de Fisica "Manuel Sandoval Vallarta", Universidad Aut6noma de San Luis Potosi,

San Luis Potosi, S-L-P. 78000, Mexico

(~) Department of Materials Science and Engineering, University of California, Los Angeles, CA 90024-1595, U-S-A-

(Received

2 August 1993, revised 13 December 1993, accepted 14

January1994)

Abstract We study the phase diagrams of polymer solutions using the pair approximation

of the Cluster Variational Method

(CVM).

In this model the solvent

(oligomers)

and the solute

(polymers)

are chains with different lengths. In particular, we study the dependence of the phase diagram on the lengths of the components of the mixture, on the lattice, and

on the relative size of the components of the mixture. We also study the critical temperature and the critical concentration. Some comparison of our results with the Flory and Guggenheim approximations

are done for the symmetric case

(when

oligomer chains and polymer chains have the same

length)

and the asymmetric case

(when

oligomer chains and polymer chains have different

length).

We present a novel technique to solve the set of equations derived from the CVM, and an analytical

solution of the problem is given.

1 Introduction.

Since

Meyer iii proposed

the

original

idea to work out the

polymer

solutions in a lattice

model,

a

large

number of works based on this idea have been

published.

One of the

major breakthroughs

is the mean-field treatment

by Flory

[2] and

Huggins

[3] which is

probably

the most

widely

used

theory

of

thermodynamics

of

polymer

solutions in the last

fifty

years.

Flory's

approximation has some

limitations,

and the most criticized points are the

following:

I) the correlation in the

occupational probabilities

are

neglected; it)

the

counting

of the number of the effective units of other chains which are nearest

neighbors

of a considered

(inner)

unit of a considered chain is

neglected

so that the two

neighboring

lattice sites must be taken

by

neighbors belonging

to the same

chain; iii)

for the evaluation of the entropy of the chains

(3)

most of the excluded volume effects are

neglected. Huggins' approximation develops

in an

independent

way and at the same time, different from

Flory's approximation,

corrects the points I) and

it).

After the

original

formulation

by Flory (which

is

usally

called the

Flory- Huggins approximation),

there have been many successful attempts to

improve Flory's

entropy

exprbssion

in such a way that

Flory's approximation

is recovered in some limit. Some of them

use the

original approach

of

Flory

[4, 6] and others use

sophisticated techniques including

the lattice

spin

field theories with an

expansion

in

1/z,

z

being

the lattice coordination

number,

in order to find corrections for finite z [7].

Nowadays,

many new and

powerful analytical

as well as numerical

techniques

have been

developed

which

depart

from the

original

lattice model

approach

of

Flory

and

Huggins

[8, 11].

However,

the lattice models still

play

an

important

role in

understanding

and

comprehension

of some basic

phenomena

in the

polymer

solutions. These new

techniques

and the traditional lattice model have

complementary

functions due to the limitations in any one of these

theories,

as consequences of

approximations,

lack of

mobility

of the chains in

simulations,

and numerical difficulties or

large

amount of CPU-time

required

in the case of very

long

chains in some methods.

In this paper, we present an

approach

for

studying polymer

solutions which is different from that used

by Flory

and

Huggins. Although

it is also based on a lattice

model,

the present paper

applies

the Cluster Variation Method

(CVM)

[12] which was

originally designed

for

alloys

and the

Ising

model. Our method

partially

corrects the limitations I) and

it) quoted above,

and

moreover the numerical output can lead to a simulation that takes into account the volume

exclusion effects [13]. In this paper we use the

pair approximation

[12] with

neighboring

lattice

points

as the basic cluster to

study

the

equilibrium

structure of

polymer

solutions. There are in the literature

only

a few attempts to

apply

the CVM to

polymer

solutions

[13, 14], probably

the most elaborate of this kind of work is

by

Kurata et al. [14]. The pair

approximation

we formulate in the present paper is somewhat different from Kurata's pair

approximation [14],

and is of the kind used

by

the authors elsewhere [13].

In this paper, we present a novel

technique

to solve the set of equations derived from the CVM. This

technique

has the

advantage

that it leads to an

analytical

solution of the

problem.

We also describe the results of a systematic calculation of

phase

separation

diagrams using

the CVM. We

study

in

particular

the

dependence

of the

phase diagrams

on the

lengths

of the components of the

mixture,

on the lattice used in this

model,

and on the relative size of the components of the mixture. We derive the critical temperature, the critical

concentration,

and

briefly

comment on the comparison of our results with others available in the literature.

In section 2 to section 4 we present the model and the method of solution. The results and conclusions are

presented

in section 5.

2. The model and the method.

In order to make it

possible

to

apply

the established entropy

expression

of lattice statistics,

we assume an

underlying crystal

lattice in which the

oligomer

chains and the

polymer

chains

are

placed.

The

oligomer

chains as well as the

polymer

chains are made of segments, and each segment

occupies

a lattice

point.

In this

approximation

the coordination number z is the

only

parameter needed to

specify

the lattice. A twc-dimensional

representation

of the model is

given

in

figure

1, the white part

being

the

oligomer

chain and the black the

polymer

chain.

From here to the end of this paper, we will call the shorter chain an

oligomer

or a solvent and the

larger

one a

polymer, although

the difference in size may be small.

In order to define the

length

of the

oligomers,

or the

polymers,

we

distinguish

the "end"

(4)

+ ~~ ~

~/~Q

Fig. i. Two-dimensional representation of the model. The white chains represent the oligomers

while the black chains represent the polymers.

segment and the "internal" segment for both

oligomers

and

polymers. Thus,

at each lattice site we find either an end segment

ii

=

1),

or an internal segment

ii

=

2)

of the

oligomers

and the

polymers.

To

distinguish

between the two types of

chains,

we use a

capital

letter

(A)

in the

probabilities

and in the statistical

weight factors,

with values A

= for

oligomers

and A = 2 for

polymers.

An end segment in this model is bonded to an

adjacent

segment in one of the

z(a wiA) directions,

and an internal segment is bonded to its two

neighboring

segments in one of

z(z 1)/2(e

w2 A) ways. These numbers are called the statistical

weight

factors for the

single

site

probabilities

and are listed in table I.

Including

the

bonding directions,

there are wi

A different

"end

segments"

and w2

A different "internal

segments". Therefore,

our mathematical

problem

is to find the

equilibrium

distribution of wi1+w21+ wi

2 +w2 2 different

species

over the lattice points, with the requirement that chemical

bondings

be

consistently

distributed.

Table I. Statistical

weight

factors w,

A for the

single

site

probabilities.

Since both compc- nents in this mixture are chains w, A and w, B are the same.

~

l z

2

z(z-1)/2

After the model is thus

defined,

our

approach

is to write the Helmholtz free energy F in terms of the state

variables,

which describe the state of the system, and then to minimize F to find the

equilibrium

distribution of the state variables. The difference between the present work and our

previous polymer

papers [13] is that in this paper both components of the mixture

are chains while in the

previous

work a solvent molecule

occupies

a

single

lattice site. The pair

approximation

of the CVM used here considers two sets of state variables: one is the set of

probability

variables for the

point configurations

z,A, and the other is for a

pair

of

nearest-neighboring

points,

y)j~~

~. The energy E and the entropy S of a system are written

as functions of these variables.

Then

the number of

species

can be

varied,

it is more convenient to use the

grand potential

Q rather than the Helmholtz free energy F. The former is defined

(5)

~

Qllx, Al, lYli,j Bl)

=

EllYli~,j

B

I) Tsllx, Al, lYli~,j Bl) N~PAPA

,

Ii)

where T is the absolute temperature, pA is the chemical

potential

of the A species and pA is the molar fraction or concentration of the A~~

species.

The number of lattice

points

in a system is written as N

throughout

this paper. Note that the definition of F does not contain the last term in

equation ii).

The

equilibrium

state is derived

by minimizing

Q with respect to

(y)j~~ ~) keeping (pA)

fixed,

rather than the

composition (pA)

fixed. '

3. Variables.

3.1 DEFINITIONS. We let xi A denote the

probability

of

finding,

on a certain lattice

point,

an end segment of the A chain

(A

= 1

being

for

oligomer

chains and A

= 2 for

polymer chains)

with its

bonding pointing

toward one of the wiA

Possible

directions. This is

equivalent

to

saying

that the number of such end segments of the

oligomer (A

= 1) or the

polymer (A

=

2)

in a system is Nwi Axi A. For an internal segment of the

oligomer

or the

polymer,

we

similarly

define z2A with the

weight

factor w2

A. It is

important

to

point

out that due to the fact that both components in the mixture are chains the statistical

weigth

factors

satisfy

w, A

" w, B for

A

#

B. This means that the statistical

weight

factors for the solvent

(oligomers)

monomers

are same as those for the solute

(polymers).

We next consider a pair of

nearest-neighboring

lattice

points,

that is, the

probabilities

of

finding

the

configuration ii j)

inside the system. There are two types of

pairs.

I) When the segments on these two

points

are

adjoining

segments of an

oligomer

or of a

polymer,

we

call them the "connected"

pair

and the

probabilities

are denoted

by y)j~~

~ where the Greek

superscript

o

(in

the

probabilities

and in the statistical

weight factori)

takes the value 1.

There are

only

connected

pairs

between segments of the same

kind,

that is

oligomer-oligomer

and

polymer-polymer.

This means that

w)j

~ ~ = o for A

#

B. The

weight

factors

w)j

~ ~

(A

= 1,

2)

for these

configurations

are

countid

from the allowed directions of chemical

boids

and are listed in table II.

ii)

When the segments on these two

points

are not

directly

bonded within the same

oligomer

or

polymer,

we call them "non-connected"

pair

and denote the

probability by y)j~~

~ where a takes the value 2. Their

weight

factors

w)j

~ ~ are listed in table III. In

figure'2

we illustrate the different types of variables used in this

~nodel

as well as the nomenclature.

Table II. Statistical

weight

factors

w)j,~~ (A

= 1,

2)

for the

connecting pairs.

The con- nected

pairs

are

only

for

oligomer-oligomer

and

polymer-polymer

cases, and

w)j,~

~ = o for A

#

B.

~(i)

~A,jA

I

( j

1 2

1 o z 1

2

(z-1) (z-1)~

(6)

Table III. Statistical

weight

factors

w)I

~ ~ for

non-connecting pairs. They satisfy

w)(~~

~(2) ~(2) ~(2)

' '

12,j2~ il,j2~ 12,jl'

~iA,jB

(2)

I 2

lz

1)~

lz i)~lz 2)/2

2

lz i)~lz 2)/2 lz i)~lz 2)~/4

2 2,2 2,1

~ ~ M W-

W + p©

a b

,

2,2 1,2 2,1 1,1

1,1

Q' Q. Q0 QQ

~~~ ,~'~ 0' 0~ 00

~'~

" "

Fig. 2. Illustation of different types of variables. There are four types of basic monomers; terminal

monomers and internal monomers for oligomers and polymers as is shown in

a).

In the case of a pair

there are four basic types of connected pairs

(oligomer-oligomer

and polymer-polymer) as illustrated in 16), and ten different types of nonconnected pairs as is shown in (c).

Inspection

of tables II and III suggests that the

weight

factors in them cal~ be written in

a condensed form when we define the

"semi-weights", w)j

with

o = 1 for sites associated with connected

pairs

and o

= 2 for sites related with non-connected pairs, I

= 1 for an

end segment and I = 2 for an internal segment, and A

= 1 for

oligomers

and A

= 2 for

(7)

Table IV.

Semi-weight

factors

w)(~.

The

w)(~

factors are

independent

of the A index.

I

~(l) ~(2)

IA IA

1

(z-1)

2

(z

1)

(z 1)(z 2) /2

polymers.

Due to the fact that both components in the mixture are chains, the

semi-weight

factors

satisfy w)j~

=

w)#

for A

#

B. This means that the

semi-weight

factors

depend only

on the

geometrical properties

of the lattice and on the type of segment. The

w)j~

are listed in table IV. Then

using

the

semi-weight

factors we can write:

w)j~~

~ =

w)j~ wjj

except

w)~j

~ = o and w)(~~~ = w)(~~~ = o ,

(2)

We also see that we can write w,A in table I as

~A

W)I

~

~~~

~~~

3.2 REDUCTION RELATIONS. As is the standard

procedure

in the CVM

formulation,

we

use

geometrical

relations to write cluster

probability

variables as linear combinations of a

larger

cluster variables. Thus z's are written in terms of

y's

as

GA

~ Wj~

Y~~,j

B' ~~~

j B

or

GA

fi ~

W)~,j

B Y~~,j B'

(5)

1 ~ ~~

3.3 CONSTRAINT RELATIONS. When we choose the

pair

variables as the basis of formu-

lation,

these variables are

subject

to several constraints.

3.3.1 Normalization. Since the x's and

y's

are

probability variables, they

are normalized to

unity

~ ~j ~~i,

j

BY~i,j

B

(6)

iA jBa

When we write the

grand potential

in the next section, we have to add the normalization constraint

using

a

Lagrange multiplier

I in the terms

(8)

3.3.2

Consistency

constraint. A

single

site

probability

can be written in two different ways,

depending

on whether a bond is connected

(a

= 1) or not

(a

=

2)

as we see in

equation (4).

Since the system is

isotropic,

a = and 2 expressions of x~ A in

(4)

are

equivalent:

~iA

~ ~~~i Y~i,j

B

~ ~~~i Y~i,j

B'

(8)

J B j B

or

~ ~~~

Y~~,jB

(d~ d~)

°

,

(9)

j B a

where

b[

is Kronecker's delta. In the minimization

procedure,

the

consistency

constraint that expresses

(8)

can be written as

~A

"

~ Al

A

~ Wj~

Y~~,j B

(d~ d~)

,

(1°)

1A j B a

where

A,A Ii

= 1,2 and A =1,

2)

are the

Lagrange multipliers.

3.3.3 Chain

lengths.

In

minimizing

the

grand potential,

we

specify

the average

length

of the

oligomer

chains in the system and that of the

polymers by

means of

Lagrange multipliers.

The average

length

LA of the chains of the species A is defined as the total number of segments,

N(wi

Axi A +w2 Ax2 A

),

divided

by

half the number of end segments, Nwi Axi

A/2. Defining

the

following

parameter RA

la

w2 Ax2

A/wi

Axi

A),

the average

lengths

can be written as follows

LA =

2(RA

+

1). (11)

By controlling

the ratio

RA,

we can control the chain

lengths.

In terms of the

single

site

probabilities

and

using

the definiton of the parameter

RA,

we can write an equation

equivalent

to the one for the

length LA

as follows

~j

w, Ax, A

RAb) b)

= o.

(12)

1

Using Lagrange multipliers

rA and

equation is),

the

length

constraints can be introduced in the

following

way in terms of the

pair probabilities

Cr

+

~j rA l~j ~j w)j~~ ~y)(~~ ~)(RAb) b)) (13)

, ,

A i jB~

Note that in

equation (13)

the

subscript

A

= 1 is for

oligomers

and A

= 2 for

polymers

and that I

= 1 is for an end segment and I

= 2 for an internal segment.

3.3.4

Composition

constraint. The concentration of

oligomers

and

polymers

are fixed

using

the chemical

potential

terms:

C~

=

~

PAPA,

(~~)

A

where pA is the molar fraction of the

species

A and pA is the chemical

potential

of the species A. When there are no vacancies in the system, the two

p's

are not

independent,

but their

linear combination is a constant. Thus we may choose without loss of

generality

~j

pA

"

0, (IS)

A

and hence we write p e p2

= -pi and call p

simply

the "chemical

potential"

in the rest of the paper. In

analogy

with 3.3.1, 3.3.2 and 3.3.3, the chemical

potential

can be

interpreted

as

a

Lagrange multiplier

for the

composition.

(9)

4. Grand

potential

and its minimum.

4.I GRAND POTENTIAL

IQ).

In view of

experiments

that show that

phase

separation

occurs in the

oligomer-polymer

solution, we assume that

oligomer

segments and

polymer

seg-

ments

repel

each other when

they

sit next to each other

(as

the nearest

neighbors)

in the lattice. When segments of A and B

species

sit on

adjacent

lattice

points,

we define the energy

parameters as

~~~

#

~~~~

where J > o. When we assume there are no vacant sites in the

lattice,

eA B are the

only

energy parameters we need. The energy for the total system per lattice

point

is written as a sum of the

nearest-neighbor energies

for the entire system as

Et ~

~ EABW)ij

~

Y)ij

B, (~~~

~

~AjBa ~ ~

where the

1/2

factor is to avoid double

counting.

Two

things

are to be noted in

equation (17).

(a)

the energy expression is exact based on the

given

model and the

variables;

16)

only y)j

~ ~

appears because we do not consider intra-chain

bonding

interactions. '

Different from the energy, the entropy per lattice point is written

only approximately

in terms of the

pair

variables. The CVM formula is [12, 13]

S = kB

I)

i +

iz

1)

~ W,A£ixiA) ~ Witj ~£iyli~,j ~)

,

i18)

iA iAjBw

where kB is the Boltzmann constant and

£(v)

represents the function v In v v. A reader who is familiar with the

quasi-chemical approximation

[4] or the CVM will

recognize

the coefficient

in this expression. The

advantage

of the CVM is that the entropy expression can be

improved systematically

when we choose a

larger

cluster as the basis of the

formulation,

and the CVM expression is the most efficient

(for practical purposes)

for the chosen cluster.

When the

particle density

is a variable parameter and is not

required

to be

fixed,

as was

pointed

out in section 2, it is convenient to mininize the

grand potential

Q defined in

equation Ii)

while

keeping

the chemical

potential

p

fixed,

rather than

minimizing

the Helmholtz free

energy. Q is written

explicitly

as

Q=E-TS-C~+Cr+CA+C,. (19)

Note that the constraint terms C of

(7), (lo), (13),

and

(14)

are included.

4.2 MINIMIZATION OF fl. The

equilibrium

state of the system is found as a minimum of

the

grand potential

Q for

given

values of the interaction

energies (eAB),

the temperature

(T),

the

length

parameters

(RA),

and a

particularly

value of the chemical

potential (p)

as follows

l~(

= o

(20)

dy~(~~

~~~,T,R~,~

The differentiations with respect to

y)j~~

~ lead to the

following

set of basic

equations:

y)j~~

~ =

y)j~exp(-eAB/kBT)yjj, (21)

(10)

where

y)j~

are to be called the

semi-pair probabilities

and are defined as follows:

y)j~

a (x~A)~~~~~~~exp

Iii

+ pA

rA(RAb) b))

+

(b( b() ~fA ~~~ /zkBTj, (22)

It is to be noted that

g)(~

are the

key

variables in our formulation since

they

allow us to formulate the

analytical

solution of the

problem.

The way to write

equation (21) coming

from the minimization of Q is different from the common way to write the CVM

equilibrium equations [12, 13],

except for a few

applications

[15].

Since z's on the

right-hand

side of

(22)

are functions of the

pair

variables y)j~,~ ~ in

(21),

the next step is to solve

y)j~~

~ from

(21), (22)

combined with the reduction

equations

in

(8)

and other constraint

relatiois. Although

the set of

equations

is not

simple,

it

cal~ be

proved

that the

semi-pair probabilities y)j~

can be solved

analytically

in terms of

T,

p,

J,

L, and the

statistical

weights.

The solution for the semi-pairs is

given

as follows

Y)I

" ~~~

,

(23.a)

/~(l)~~

~

~(l)~

~

2A IA I

Yji /w(1)

~~~

~(l)

2A~2A W~~Xi A

~~

'

(23,b)

Y~i

"

~fi,

j~~_~)

PA where

p(~

=

~j w)jx,

A

(24)

~

and

pi(

~~ ~~~~~ ~~

~~~

jp(~

+

pi~)H Q

PAB " ~~~~

2[1

HI

'

where H a

exp(-2J/kBT)

and

Q

e

/4p(~p(~H

+

(p(~ pi~)~H2.

Details of the derivation

are

given

in

Appendix.

Equations

(22)

to

(25)

represent the

analytical

solution of the

problem.

When we use the

equilibrium condition,

we can

simplify

Q in

(19)

and we can show that the

Lagrange multiplier

for the normalization is the

equilibrium

value of the

grand potential

1 =

Q~~. (26)

5. Results.

5. I PHASE-SEPARATION DIAGRAM. The

phase diagram

calculation can be done

by using

the intersection of the two branches of the Q us. p curve [13), or

by using

the Maxwell construction [16) in the

graph

of p us. p.

Considering

that the solution for the equilibrium state has been solved

analytically,

numerical

computation

is

greatly simplified.

Since the coexistence of the two

phases

p(~)

(the gas-like)

and p(~)

(the liquid-like)

is a standard

technique

for the

calculation of the

phase diagrams

we consider that no

exp1al~ation

is necessary [13).

(11)

5. 2 EFFECT OF THE POLYMER LENGTH IN THE PHASE SEPARATION DIAGRAM. To

StUdy

this we fix the

oligomer length

to Li = 5 and

change

the

polymer length

L2. The results are

presented

in

figure

3 for L2 =

5, So,

soo in the case of a

simple

cubic lattice z

= 6. Let us call

the maximum of the

phase separation diagram

the critical

point.

The critical concentration p~ and the critical temperature T~ are defined for this

point.

The

phase separation diagrams

are more

asymmetric

as the

polymer length

increases. We can see that p~ shifts to the lower

polymer

concentration as the

polymer

becomes

longer,

p~

approaching

zero for the infinite

length

[2, 17]. It is clear that the asymmetry in the

phase diagrams

is due to the

large

difference

in the molecular size of the two components. Our results are in

qualitative

agreement with

Flory's

calculations [2, 17] as well as with other calculations

using

the CVM [13], as

expected;

in all

previous calculations, however,

the shorter

polymer

is treated as a

single

segment

occupying

one lattice

point.

The

(kBT~/zJLi)

values

corresponding

to

figure

3 are o.35, o.79 and 1.13 for

(L2/Li)

=1, lo, and loo

respectively.

o-o

Fig. 3. Phase separation diagrams calculated by the present CVM theory based on the simple cubic lattice

(z

= 6) and for different polymer lengths. The short dashed line is for L2 = 5, the long dashed line is for L2

" 50 and the solid line is for L2

" 500. All of them

are for the fixed oligomer length

(Li

" 5).

5. 3 EFFECT OF THE COORDINATION NUMBER IN THE PHASE SEPARATION DIAGRAM. In

this

model,

in order to make it

possible

to

apply

the established entropy

expression

of lattice statistics, we have to assume an

underlying crystal

lattice in which

oligomer

segments and

polymer

segments are

placed.

Since the lattice is

hypothetical,

it will be desirable that the results be

independent

of the coordination number

(z),

however in some of the cases the lattice has some influence on the results [3,

4,

7, 13]. To

study

the influence of the coordination

number,

we calculate the

phase diagrams

for different z values as shown in

figure

4, in which

(a)

is for the

simple

cubic lattice

(z

=

6), (b)

is for the

body

center cubic

(z

=

8)

and

(c)

is for

face center cubic

(z

=

12).

The results in

figure

4 are for a mixture with

(L2/Li)

" lo with Li = 5. From

figure

4, we can say that the

shape

of the

phase diagram

in our

ajproximation

(12)

is

practically independent

of z when the temperature axis is scaled

by

T~. We can say that when the

phase separation diagrams

are scaled to T~

Flory's approximation

[2, 17] which is

independent

of z is a

fairly good approximation.

The

(kBT~/zJLi)

values obtained with our model for a mixture with

(L2/Li)

= lo and

Li

" 5 are

0.79,

0.88 and 0.97

respectively,

for

z = 6, 8 and12.

IQ) 16)

ic)

oo lo

fi

Fig. 4. Dependence of the phase separation diagram on the coordination number z for a mixture with the oligomer length Li " 5 and polymer length L2 = 50, with the T axis normalized by the

critical temperature Tc. The simple cubic is presented by

la),

the bcc

by16)

and fcc by

(c).

5.4 EFFECT OF THE SOLVENT SIZE IN THE PHASE SEPARATION DIAGRAM. Now we fix

the coordination number z = 6 and the ratio

(L2/Li

" 10 for three systems with different

solvent

(oligomer)

sizes Li = 5, lo and 20. The results are

given

in

figure

5 for these three

cases. The results for the

phase diagrams

scaled to T~ indicate that, as the solvent

(oligomer)

size is

increased,

the

phase diagrams

become

slightly

narrower. The scaled

phase diagrams

are

almost

independent

of the solvent size and the

important

parameter seems to be the relative size between the

polymer

and the solvent

(L2/Li).

The

(kBT~/zJLi)

values obtained are

0.79,

0.78 and 0.77 for Li

" 5,

10,

and 20

respectively.

Since the

(kBT~/zJLi)

ratio is almost

constant the results suggest that

(kBT~/zJ)

is

approximately

a linear function of

Li

when we fix the

(L2/Li)

ratio as in the present case.

5.5 CRITICAL TEMPERATURE BEHAVIOR. In order to examine the critical temperatUre

as a funtion of the

polymer length,

we consider two cases:

a)

we fix the

oligomer length

and

change

the coordination number;

b)

we fix the coordination number and

change

the

oligomer length.

The results for these two cases are shown in

figure

6. The results for

a)

indicate that as the ratio

(L2/Li)

increases the critical temperature

(kBT~/J)

increases,

being

very

fast for low values of

(L2/Li)

and

levelling-off

for

large

values of the ratio

(L2/Li). They

also indicate that as the coordination number z increases the critical temperature increases.

Since the calculation for an infinite

polymer length

is time

consuming,

we can

extrapolate

in

(13)

2

la) ib) ic)

o-o i-o

p

Fig. 5. Phase separation diagrams for three systems with different component lengths on a simple

cubic lattice. The ratio

(L2/Li

has been fixed in the three cases to 10.

la)

is for the polymer length

L2 " 50, (b) for 100, and (c) for 200.

order to calculate the saturation value. For the

exptrapolation

of the temperature we can take

advantage

of

Flory's analytical

formula in the case of the

point

solvent for the temperature

(kT~/zJ)

=

2L/(1+ li)~

[2, 17]. The temperature for very

long polymers

can be written

approximately

as

(kT~/zJ)

m

(kT~/zJ)~ 411i,

where

(kT~/zJ)~

means the temperature for an infinite

length polymer.

In our case, in

analogy

with

Flory's analytical expression,

we

assume the

following

form for the

extrapolation

of the temperatures

(kBT~/zJLi)

"

(kBT~/zJLi)

~'~

(27)

" "

fi

'

where the

subscript

n means the value of the ratio

(L2/Li)

and A is a parameter to be calculated

together

with (kBT~

/JLI )m

The temperature

T~(oc)

for an

infinitely long polymer

is

usally

called the 8 temperature. The results of

(kBT~/JLI)m

for an

oligomer length

of Li = 4 are 4.24, 8.25, 12.25 and 20.25 for z

= 4, 6, 8 and 12,

respectively.

For the case

b),

in which we fix z and examine the temperature

dependences

for two

oligomer lengths

Li = 4 and 5. The results are shown in

figure

6b. In a similar way as in the

a)

case,

the critical temperature

(kBT~/J)

increases, very fast for low values of

(L2/Li)

and levels off for

large

values. The results also indicate that as the

oligomer length (Li

increases the critical

temperature increases. It is worth

noting

that the results scale

by

a factor of

5/4,

that is, the ratio of the two

oligomer lengths.

The critical temperatures

(kBT~/JLI) extrapolated

to the infinite

length polymer using equation (27) give

the value 8.22 for both

oligomer lengths.

In

Flory's theory

the 8 temperature

plays

an important role. The 8 temperature has the

following

two characteristics:

ii)

at

8,

the second virial coefficient vanishes and therefore the effect of the excluded volume

disappears, making

the

polymer

behave as a random

walk; iii)

8

corresponds

to the critical temperature of the

phase separation diagram

for an

infinitely long polymer. Using

the

extrapolated

value of

(kBT~/JLI)m,

we can calculate

Flory's

parameter

x~(e JzLi /kB8).

In his

theory

x~ is

always 1/2

for any coordination number. Our results

presented

in

figure

7

(square marks)

indicate that x~ deviate from

1/2

for any finite coordination

(14)

Z=12

<

Z=8

@

m

~ Z=6

(a)

0.0

O loo 200

L2/Li

Li = 5

~

'

f ~l

" '

r~

(b)

O.0

O 250 SOD

L2/Li

Fig. 6. The critical temperature as a function of

(L2/Li)

for three different lattices in la) and for two different oligomer lengths in 16).

number. In the same

figure,

we alto present results obtained when the solvent is a

point particle (circle marks)

[13] as well as the results obtained

by

Saleur

(triangle)

[18] in the case of a square

lattice. When we compare our present calculations in which we consider that the solvent has

a structure with the calculation where the solvent is a point

particle

[13], we find that x~ is

larger

when the solvent molecule has a structure.

5.6 CRITICAL CONCENTRATION. The behavior of the critical concentration p~ is illus-

trated in

figure

8. In the same

figure Flory's

calculations are

presented

for comparison. A mean-field

theory

like

Flory's

has the

advantage

of

simplicity,

and even the

possibility

to de-

(15)

, I

I I

',

'

O

~K

~i

0. A ", Q

',,

'O~,

""'-4X

___

2 4 6 8 lo 12 14 16 18 20

Coordination Number

Fig. 7. The critical xc

(+

JzLi

/kBB)

parameter for different coordination numbers and for different

approximations. The results of this work are in squares. Circles are the results when the solvent is a

point particle [13], and a triangle is for a square lattice by Saleur [18]. The continuous curves are aids for the eye.

°.

Flory

~

cvm Ct

o-o

i io ioo

L2/Ll

Fig. 8. The critical concentration pc as a function of the normalized length

(L2/Li).

The solid line presents the results of this work and the dashed line Flory's results.

rive an

analytical

expression for the p~ as is

given

in references [2] and [17]

where the solvent is considered as a

single point

and L is the

polymer length.

The results show that p~

begins

at the value 0.5 for the

symmetric

case

(Li

"

L2).

As the difference

in size between the

polymer

and the

oligomer increases,

p~ decreases and goes to zero in the

(16)

limit when the ratio

(L2/Li)

goes to

infinity.

In

general

our results for p~ are in

qualitative

agreement with

Flory's

calculations. It may be noted that the results

presented

here show that the

phase diagrams

calculated

by Flory's approximation

are

slightly

more

asymmetric

than

our

phase diagrams.

5. 7 PHASE-SEPARATION DIAGRAM FOR POINT SOLVENT. Now we compare the

phase

dia-

grams calculated here

using

the structural solvent

(CVM-II)

with the other calculation where the solvent is a

point particle (CVM-1)

[13]. Both calculations were

performed using

the CVM

pair approximation

on a

simple

cubic lattice. The results are

presented

in

figure

9 for two

different

lengths

cases

(L2/Li)

" 12 and 102. The dashed curves are for Li

= 1 and solid

curve for Li = 4. The

comparison

indicates that the solvent structure increases asymmetry

of the

phase diagram

and that it is

slightly

narrower in the case of

long polymers.

For the comparison of the critical temperature, we calculate (kBT~

/JLI ),

which are 4.975 and 6.825 for

(L2/Li )=

12 and 102,

respectively.

The

correponding

results for the

point

solvent are 5.4 and 7.375. The temperature decreases to about 92% in both cases. This decrease is in agreement with the

generally acepted

property that the critical temperature decreases as the statistical

description

of the system is

improved.

.2

(a) (b)

L~/Li

= 12

L~/Li

= l02

o-o i-o o-o i-o

p p

Fig. 9. Comparison between the two different CVM approaches. The solid

curves represent the

result of this work, and the dashed curves are when the solvent is considered a point particle [13].

5.8 SYMMETRIC OLIGOMER-POLYMER MIXTURE. This case is the

simplest

one to

Study

among mixtures. The

phase diagrams

are

symmetric

around the concentration 0.5. In order to compare our results with those

by Flory

[2, 17] and

Guggenheim

[4], we calculate the ratio

(zJL/kBT~),

where L denotes the

polymer length,

or the

oligomer length.

This ratio in

Flory's

approximation is 2 and is

independent

of the lattice. In the case of the

Guggenheim approximation

this ratio is

given by

(zJL/kBT~)

= z L

In(1

~

,

(29)

JOURNAL DE PHhS'QUE '< -T 4 N'4 APR<L 1994

Références

Documents relatifs

If the « s-d » exchange Hamiltonian is arbitrary and sufficiently large in comparison with the exchange integral in the local spin Heisenberg Hamiltonian, one

2014 High resolution measurements of the temperature dependence of the optical rotatory power 03C1 (T) in the SmC* phase of the ferroelectric liquid crystal DOBAMBC

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

[r]

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

l’utilisation d’un remède autre que le médicament, le mélange de miel et citron était le remède le plus utilisé, ce remède était efficace dans 75% des cas, le

Belouchrani, A., Abed Meraim, K., Cardoso, J.-F., Moulines, E.: A blind source separation technique based on second order statistics. on Signal Pro-

Figure 25 59 : montrant a angio IRM une malformation vasculaire complexe à prédominance veineuse A) veine marginale externe B) localisation des malformations