• Aucun résultat trouvé

End effects in adsorption of homopolymers and triblock copolymers

N/A
N/A
Protected

Academic year: 2021

Partager "End effects in adsorption of homopolymers and triblock copolymers"

Copied!
20
0
0

Texte intégral

(1)

HAL Id: jpa-00247888

https://hal.archives-ouvertes.fr/jpa-00247888

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.

End effects in adsorption of homopolymers and triblock copolymers

F. Aguilera-Granja, Ryoichi Kikuchi

To cite this version:

F. Aguilera-Granja, Ryoichi Kikuchi. End effects in adsorption of homopolymers and triblock copoly-

mers. Journal de Physique II, EDP Sciences, 1993, 3 (7), pp.1141-1159. �10.1051/jp2:1993188�. �jpa-

00247888�

(2)

Classification Physic-s Abstiacts

05.20 05.50 82.65D 61.25H 82.20W

End effects in adsorption of homopolymers and triblock

copolymers

F.

Aguilera-Granja (I)

and

Ryoichi

Kikuchi

(2)

(')

Instituto de Fisica

« Manuel Sandoval Vallarta », Universidad Aut6noma de San Luis Potosi, San Luis Potosi, S-L-P. 78000, Mdxico

(~) Department of Materials Science and Engineering,

University

of Califomia, Los

Angeles,

CA 90024-1595, U-S-A-

(Received 29 Octobei 1992, revised 18

Februaiy

1993, accepted 23 Maich 1993)

Abstract. The effects of end segments on the conformation of

polymers

adsorbed on a flat surface are studied

theoretically

by the use of an

analytical

statistical mechanics (the Cluster

Variational Method, CVM) and a simulation based on the CVM output. Based on the simulation,

we derive information on the thickness of the adsorbed

layer,

the number of

loops,

the area covered

by

a polymers, fractions of the train, tail and

loop portions

of an adsorbed

polymer,

the average number of tails, and related

geometrical quantities.

The conformations of the adsorbed

polymers

show a strong

dependence

on the end segment behavior. The result derived in the paper give us

guidelines

for obtaining some specific conformation when triblock

copolymers

are used in

the

adsorption

studies.

1. Introduction.

The steric stabilization of colloid

suspension

is

becoming increasingly important.

Often the interaction between colloid

particles

in

suspension

is controlled

by coating

the

particles

with

polymers.

It is

generally

believed that the interaction

depends

on the conformations that the

polymers adopt

on the

particle

surface. Therefore the

important question

is how we can control the

configuration

of the adsorbed

polymers.

With the purpose of

controlling

the

configuration

of the adsorbed

polymers,

different types of

homopolymers

have been tried.

However,

it

seems difficult to control the

polymer configurations only by

the use of different

homopoly-

mers. For this reason, the use of

polymers

with some

special

architecture like diblock

copolymers

and triblock

copolymers

has been started

[1-7].

In this paper, we report a

study

of the

configurations

of the triblock

copolymers.

The

study

of triblock behai,ion of

polymers

are classified in intrinsic and extrinsic cases. The intrinsic

case is a

homopolymer

in which the end segments

(the

first and the last

polymer segments)

have different interaction behavior than the intemal segments. The difference may come as a

consequence of the different arrangements in the chemical bonds at the ends of the

polymer.

(3)

1142 JOURNAL DE

PHYSIQUE

II N° 7

Under such

conditions,

it is

expected

that a

homopolymer

behaves like a triblock

copolymers

(ABA), where the ends of the

polymer belongs

to one kind of block

(A)

and the main

body

of the

polymer belongs

to the other type of block

(B).

The way in which the blocks A and B are

adsorbed on the surfaces may be

completely

different. The tiiblock behavio/. of

homopolymers

has been

pointed

out also

by Scheutjens

and coworkers

[8].

The

explicit

case is when the triblock

copolymer

is indeed a

polymer

built

(synthesized)

from three differents blocks

(ABC).

However, in this paper we limit ourselves to the case in which the blocks A and C are

equal

and

they

are much shorter than the central block. The

particular polymer

concentration in the solution studied here is I

fb,

the parameter for the bulk

length

is L

=

39

[9, 10],

and the size of the system used in the simulation is a box of 50 x 50 x 20 lattice constants. The

length

of the

polymers generated by

the simulation process is denoted

by I [10].

The unit of the energy is

kT,

T

being

the temperature.

2. Model and method.

Before we start the

simulation,

the

equilibrium

state of the

polymers

close to the surface is determined

using

the Cluster Variational Method

(CVM) [9, 10].

The

analytical

results are used as the

input

for the simulation

[9].

Since the method and all the details about the way in which the

analytical

calculation and the simulation are done can be found elsewhere

[9, 10],

we

simply highlight

the

important points

of this method.

ii)

ANALYTICAL INPUT. In the present case, we use the

pair approximation

of the CVM in

the

simple

cubic lattice

[9-12].

The CVM is a statistical mechanical

technique

of

calculating

the

equilibrium

state

by minimizing

the free energy F with respect to the variables chosen to describe the state of the system

[13].

It was

applied

for the first time to

polymer problems by

Kurata and coworkers

[14],

which showed

good

results in the entropy evaluation for

polymers

in the bulk. The CVM is a herarchical system of

approximations.

Each

approximation

is

defined

by

the basic cluster

(made

of

neighboring

lattice

points)

and subclusters. The most

rudimentary approach

is the

point approximation

which is constructed

using

the

probability

variables for the

species occupying

one lattice

point.

This

approach

is

usually

called the mean-

field

approximation

or the

Bragg-Williams approximation [15].

We can mention that the

classical

theory by Flory [16]

is

essentially

a mean-field like

approximation.

The

physics

contained in the

pair approximation

of the CVM which is used in this work is

equivalent

to the well known

Quasi-Chemical

Method

[17], although

the details are different. The

advantage

of the CVM

approximation

used here 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.

In this work the

thermodynamic

behavior of the system is described

by

the

point

variables and the

pair

variables. The

point variable,

x~, is the

probability

of

finding

a

species

I on a

point,

and the

pair variable,

y~~, is the

probability

of the I

j pair

on

nearest-neighboring

lattice

points.

The energy E and the entropy

S,

and then the free energy F

=

E

TS,

are written as a function of the x, s and

y~~ s. The

equilibrium

state is derived

by minimizing

F with respect to these variables. In the present

application

Of the

CVM,

we work on a

layer

above a

substrate,

and hence the system is

inhomogeneous.

This makes us use more

subscripts

and

superscripts

to

specify

x s and y s. Also, because of the

length

and the

density constraints,

we minimize the

grand potential

fl rather than the free energy F. We describe the details in the

following.

The lattice

plains parallel

to the substrate are numbered

by

n, n

= I

being

the first

plane

closest to the solid substrate

(surface).

The

configurational probabilities

of the lattice

points (the single

site

probabilities)

in the

plane

n are written as x~.,, for a solvent

(I

= 0

),

for an end

polymer

segment or end monomer

ii

=1,

2, 3)

and for an intemal

polymer

segment or

(4)

internal monomer

ii

=

4,

5,

6, 7).

A

polymer

segment has

connecting bonds,

an end segment

being

bonded to an

adjacent segment

and an internal segment

being

bonded to its two

neighboring

segments. This is illustrated in table

I,

where all

species

and their statistical

weights (w,

in the case of a

simple

cubic lattice are shown

[9].

A

pair

of

adjacent

are either

connecting

or

non-connecting.

The

pair

is called connected when the

segments

on these two

points

are

adjoining

segment of a

polymer

and is called

non-connecting

when both

points

are not

adjoining

in the same

polymer

or when the two segments

belong

to different

polymers.

When

they

are in the same

plane

n, their

probabilities

are written as

yj~~,~

and y)(~~_ for

connecting

and

non-connecting, respectively.

When in the

pair

the segment I is on the

plane

n and

j

on the n +

I,

their

probabilities

are written as

?j~/~

and

?)~/j~.,,~ (with

h

m 1/2, and n + h

being

the middle

point

of the two

planes).

The

probabilities

on the

plane

n = I need

special

attention because a chemical bond cannot

point

downward toward the solid

surface,

the

probabilities

with bonds

pointing

towards the surface are zero. The number of

equivalent configurations

due to the different type of

pairs (connecting

and

non-connecting)

are listed in tables II and III. Those

pairs

not shown in the tables are zero. In the simulation however, each of

equivalent configurations

of the

bonding

arms is treated as

distinguishable [10, 12].

The energy of the system is made of two terms. For the sake of

simplicity

we consider that the interaction between the surface and the

polymer

segment is extended

only

to the segments

on the n

=

plane

as is illustrated in

figure

I. The first contribution is the

surface-polymer

attraction

only

for intemal

polymer

segments. This is a van der Waals type of interaction and is

proportional

to the number of

polymer

segments on the n

=

I

plane

:

7

Ej

=

Ls

jj

w~ a-j

,

(l)

,=5

where ~ is the parameter to describe the

strength

of the

interaction,

.ij,

(i

=

4,

5,

6, 7)

is the

probability

to find an intemal

polymer

segment on the

plane

n

= I, and L is the number of lattice

points

within a

plane parallel

to the solid surface.

Table I.

Definition of

the

subscript

I and the statistical

wieight factor

(w~

) for

the

single

site

probability

used in our model.

D»finition of the subscript I and the weights tU;

Species and Connections I td;

Solvent molPcule @~ 0

End monomer '

@

2 4

3

Internal mono>Twr 4 4

*

5

6

6

7 4

(5)

l144 JOURNAL DE PHYSIQUE II N° 7

Table II.

a)

The statistical

weight factors for

the

connecting pairs

within the same

plane (w)i~,,~ ).

For the

first layer

n

=

I,

the statistical

weight

vanishes when I or

j

is 4,

b)

The statistical

weight factors for

the

connecting pail-s

between

planes (w~~(,

~

).

For the

first layer,

n =

I,

the statistical

weight

vanishes when I is 6. Those

pail-s (I, j

not shown in the tables are

=eio.

la)

I j 2 4,7 5

2 0 I 3

4, 7 I I 3

5 3 3 9

16)

I on n ion

(n+I)

4 6

3 0 4 I

6 I 4 I

7 4 16 4

Table III. In

(a)

the statistic-al

weight factors for

the

non-connecting pairs

within the same

plane

(w>),f(

~

).

For the

first layer

n = I the statistical

weight

vanishes when I or

j

is 1, 4 or 6.

In

b)

the statistical

weight factors for

the

non-connecting pairs

between

planes (w~f( ).

For the

first layer

n = the statistical

weight

vanishes when I is I or 4. Those

pairs (I, j

not shown in the tables are zero.

ja)

I j 0, I, 3, 6 2, 4, 5, 7

0,1, 3, 6 3

2, 4, 5, 7 3 9

jb)

I on n ion (n+1) 0,3 2,7 5

0, 4 6

2, 4 4 16 24

5 6 24 36

n= 3 n=2

n=

/~

. «

-i

' surface

Fig.

I. Illustration of the interaction energies used in this model. e is the energy of interaction felt by the B block (internal segments) on n I, and A is the energy felt by the A block lend segments), on

n = 1.

(6)

The second contribution is for end

polymer

segments

(the

first and the last segments of

polymers)

and may be

attractive, repulsive

or the same as the case of intemal

polymer

segments. This energy is also

proportional

to the number of

polymer

segments on the

n = I

plane:

3

E2

= LA

£

W, xi,,

,

(2)

,=~

where A is a parameter that describes the interaction of the end

polymer

segment and the solid

surface,

and xi,,

(I

=

1, 2, 3)

is the

probability

to find an end

polymer

segments on the

plane

n =

I. The total energy E is the sum of the two contributions. In this model the solvent condition is assumed «

good

» so that we do not consider any interaction between the solvent and the

polymer segments.

Note that xj,~ and xi, are not allowed on the

plane

n = I, and thus are

missing

in

(I)

and

(2).

Also note that our convention for the energy parameters is that

they

represent attraction when

they

are

positive

and

repulsion

when

negative.

In our energy

formulation,

we

prefer

to use the

pair

energy interactions e and A instead of the

respective

Jfs Parameters which are the same

pair

interaction energy divided

by k~

T.

The entropy of the

system

is written in the

pair approximation

of the CVM for the

simple

cubic structure as

[9-13]

S=k~Ljj ~7

2+5

jjw,£(x,,,)-

n ,=o

2

I lW)( (,

j

£(Y)(~,,

j

)

+

W)( (,

j

£

(Y)I

,,

j)1

,,,

i

[W)~~,

j

£(Z~~

h <,j + W~~~,j £(Z~~~ h,

<,j

)1)

,

(3)

<,j

where w,,

w)[(

~,

w)[

), are the statistical

weight

factors

(number

of

equivalent configurations)

of the

single

site

probabilities (x,,

,

), pair probabilities

within the same

plane ~y)1(

and

pair Probabilities

between

plains (z)flh., j)

as shown in tables I, II and III,

respectively.

The

function

£(x)

is defined as x In x x. This

expression

reduces to the case of the bulk formula when the variables do not

depend

on n

[I I].

The

equilibrium

distribution of the

system

is calculated from the minimization of the

grand potential

D with respect to the

pair probabilities

for

given

values of the interaction

energies,

the

temperature T,

the bulk

polymer length

parameter

L,

and the chemical

potential

values. The

grand potential

D can be written as follows :

D=E-TS-L

I

p,p~,,

(4)

,,,,

where E is the intemal energy defined in

equations

and

(2),

S is the entropy as in

equation (3),

p, is the chemical

potential

and p,~

,

is the

density

of the I-th

species

on the

plane

n. After all the

equilibrium

cluster

distributioni ix,,,,), (y)[(,~ ), (z)fl~., ~)

have been solved for the

equilibrium

state, we can

proceed

to the simulation.

(ii)

SIMULATION PROCEDURE. When all the cluster

probabilities

of the system are known

we are

ready

to

begin

the simulation process

using

the

Crystal

Growth

Probability

Method

(CGPM) [18].

The

advantage

of the simulation when it is done as an addition to the

analytical

statistical method is that the simulation

gives

us information that otherwise would be

inaccessible. The

advantages

of the simulation based on the CVM method over the traditional

(7)

l146 JOURNAL DE PHYSIQUE II N° 7

Monte Carlo

(MCS) [19]

simulation are first the

speed,

and second that our simulation is based

on the distributions that

satisfy equilibrium

conditions, so that no further relaxation processes

are needed.

The simulation process is as follows. We construct the system from bottom to top

beginning

next to the

substrate,

from left to

right,

and from back to

front, by placing

a

species

(a

polymer

segment or a solvent

molecule)

at a lattice

point

one at a time. In order to

place

a

probability

in

a

given point,

we use a random number to choose the

probability

in such a way that the

pair probability

distribution

(coming

from the

CVM)

is satisfied

[10, 12, 18].

The

procedure

is

repeated

at each lattice

point,

and when the entire lattice is

filled,

the simulation is done.

However,

with this type of

simulation,

it is

required

to repeat the

procedure

many times to avoid bias in the different statistical patterns

[10,

12,

18].

In the present

polymer

simulation,

we need a

special

constraint to take care of the direction of chemical bonds

[12],

and in order to avoid conflict with the chemical bonds we need to use a

pseudo six-point

cluster

[10, 12].

An

important

feature of this multi-chain simulations is that the

sample

of the

polymers generated by

this method is

polydispersed [10, 12].

In the

analytical

results the

polymer length

in the bulk is

specified by just

one

parameter

L

[9-12].

The parameter L is defined as the total

number of

polymer

segments divided

by

half of the end

polymer

segments.

Although,

the

average bulk

length

L is

given

in the

analytical

results as we

quoted above,

it does not guarantee a

monodispersed polymer

system in the case of the simulation

[10, 12].

For all the results

presented

in this paper we use an average bulk

length

parameter of L

=

39. In this simulation the average

length

in the bulk measured

by

the direct

counting

in the simulation field

(f )

is

expected

to be

roughly

the same as

L,

but

actually

is

f~~~

= 22. The

polydispersity

of the

generated sample

is understood due to the fact that the L parameter that controls the

length

in the bulk is fixed

only

in the bulk for the

analytical

CVM

calculations,

and we let the

polymer length freely

evolve towards the surface. The free evolution of the

polymer length

in the

sample originated

a distribution on the

length (f ).

From the simulation results,

we know

that in the bulk

(far

away from the

surface)

the

length

distribution of the

polymers

is

negative exponentially [12],

with

polydispersity

index N

=

1.82

(the

ratio of the

weight-average

molar

mass

M~

to the number average mass

M~ [12].

Our

polymer

system is an open system which is in

equilibrium

with a bath

(reservoir)

of

polymers

of different

lengths,

and in the

equilibrium

situation the

polymer

system and the reservoir

exchange polymers freely

between them.

The size of the system used in this simulation is 50 x 50 x 20 lattice constants, and the system is truncated in five of the six faces. Because of the truncation of the system, some

polymers

near a truncation surface have

only

one

ending point

inside the simulation field. For the sake of

simplicity,

when the statistical count is done these

polymers

near the

edge

are

disregarded.

Another

anomaly

of this simulation is the existence of closed

loops

of

polymers,

which are also

disregarded

in the statistical

counting

since the main concern of this paper are

the linear

polymers.

The

probability

of closed

loop polymers

in this type of simulation base on

the CVM decreases when

bigger

clusters are used

[12].

In the present paper, we work with the type of triblock

copolymers

in which the central part of the

polymers

are adsorbed with the energy

e(~

0

),

and for the end segments we consider

both attractive A ~ 0

(ATT)

and

repulsive

A

~ 0

(REP)

cases. The main purpose of this work is to

study

how the conformations of the adsorbed

polymers

vary

depending

on the REP and ATT

conditions.

It is

important

to

emphasize

that the CVM calculation describe the

equilibrium

state of the system, and that the simulation is a new extension of the CVM

developed by

one of the authors in 1980

[18].

The simulation is not needed

[9,

11, 13,

18]

if one is content with the

analytical

results calculated

by

the CVM as were

generally accepted

before this new type of simulation had been devised. However, the simulation is a very useful and

powerful

tool that

displays

in a

(8)

visual form the information which is contained in the

analytical

distribution obtained in the CVM.

By doing

the

simulation,

we can get a

physical representation

of the system, from which we can obtain a

large

amount of information about the

shape

of individual

polymers

adsorbed on the surface. We may say that the simulation is the

key

that allows us to show the information that is hidden in the CVM

analytical

results.

3. Results.

Before

going

to the simulation

results,

we first show in

figure

2 some of the

analytical

results obtained

by

the CVM. These results are some of the ones on which our simulation is based. In

(a),

we show the

density profile

as a function of the distance away from the surface. Both the

Density (a)

o.oi

lo zo

% En d

X

loo

owomooo

',,, ~o--°~

"',,,, ~o~° (b)

,~~

o.oi

Distance

Fig. 2. Analytical results used as the input in this simulation : (a) the density profile and (b) the percentage of end segments as functions of the distance away from the surface. The circles (REP) and the squares (ATT) correspond to the calculated points.

(9)

lI48 JOURNAL DE PHYSIQUE II N° 7

ATT

(e

= 1.0 and A

=

3.0),

and the REP

(e

=

1.0 and A

=

1.0)

cases have

practically

the

same curve. The two values used here for the A parameter are chosen

just

to illustrate the

dependence

of the conformation on this parameter. In this paper the ATT case is

represented by

a square, and the REP case

by

a circle unless otherwise stated. The concentration of the end segments is so small in these cases that their

displacement

towards or away from the surface does not make

changes

in the total

density

of

polymers

close to the surface.

However,

even when the

global

property like the

density profile

does not

change,

the conformations

adopted

by

the two cases may be

completely

different. To find the differences of the conformations of

the two cases in detail is the main purpose of this paper.

Although

the

density profile

does not

change,

the percentage of end segments shows some difference in the ATT and the REP cases, as are shown in

(b).

The dashed line

(together

with the

square)

is for the ATT, and the solid line

(together

with the

circle)

is for the REP case.

After the second

plane,

the two curves are almost

equal.

This is understandable since

only

the segments on the n

=

plane

are

interacting

with the surface in this calculation. In the same

figure,

we include the dotted line as the reference which is when all segments feel the same

attraction from the surface

(F

=

1.0 ). We call this case HOM

(homopolymer),

and represent it

by

a

triangle

unless otherwise stated.

For all the results

presented

in this paper the coverage 0

(concentration

of

polymers

on the

surface)

and the adsorbed amount

r~~~

are

kept fixed, being

their values 0.58 and

0.89, respectively.

It is

important

to mention that the definition of the adsorbed amount used here is the as the same one used

by

Roe

[20].

For the case of

homopolymers,

it is well established

theoretically [10, 21, 22]

and

experimentally [23]

that in the

adsorption

of

polydispersed polymer samples,

there is a

preferential adsorption

for

longer

chains than the average in the bulk

(solution). Now,

we are

concemed how the

properties

of the

homopolymers

are modified because of the triblock

copolymer

behavior

(intrinsic case)

of the

homopolymers

or what

properties

are

expected

in the case of extrinsic triblock

copolymers.

One of the results of this simulation is shown in

figure 3,

which shows the

probability

of adsorbed

polymers

as a function of the

polymer length

~

~

25 ~~

o

',,U

~ ,

O

£~

O

8

_

/

'~,

Q ~° ~'~,

~'~b~~

~~ ~

o,,,~

>~~

° ~~

Le/~th

~°° ~~~ ~~°

Fig.

3. Simulation results of the distribution of the adsorbed

polymers,

or

adsorption probability,

on

the surface. The circles are for REP and the squares for ATT.

(10)

(f).

Note that

although

the

analytical

CVM results

are for the

polydispersed

system, the simulation

technique

can select different

lengths separately.

There is a

preference

of

adsorption

for short

polymer

chains in the ATT case, while in the REP case the

preference

is for

long

chains. The average

polymer length

of the adsorbed

polymers

on the surface for the ATT is

f~~~

=

40,

and for the REP is

f~~p

=

80,

to be

compared

with the average bulk

length

in both

cases which is

f~~~~=22.

The

polydisperse

index N

[12]

of the distributions is

N~~~

=

1.63 and

N~~p

=

1.34, respectively.

It is

interesting

to compare the results with the HOM case

[10],

for which the average

length

of the adsorbed

polymers

on the surface is

f~J~~

=

72,

the

f~~~~

=

22,

and the

polydisperse

index is

N~J~~

=

1.41. The

comparison

of

HOM case with the ATT and REP indicates that in the case of the triblock

copolymers

(intrinsic

or

extrinsic)

the distribution of the adsorbed

polymers

as well as the

polydisperse

index can be controlled

by

the behavior of the end segments of the

polymers.

In the case of the intrinsic triblock

copolymers,

this

change

in the distribution of the adsorbed

polymers depends

on the

change

of the chemical and

physical properties

of the

ends,

as a consequence of the different arrangement of the bonds at the end segments of the

polymer,

and therefore we cannot control this type of

properties,

because

they

are intrinsic in this

type

of

polymers.

On the other hand, for the case of extrinsic triblock

copolymers,

the

properties

of the end may be controled

by changing

the type of end

blocks,

and therefore in this way we can control the distribution of the adsorbed

polymers.

We can obtain more information about the conformation

adopted by

the adsorbed

polymers

when we

study

the end-to-end distance. This

quantity

in this kind of simulations for a

polymer

in the solution

obeys

a power law

[12],

as

rj r~ =

cf

"

,

(5)

where c is a constant that

depends

on the model and v is an exponent

depending

on the

dimensionality

of the system and

independent

of the lattice model

[10,

24,

25].

The results for

this simulation are shown in

figure

4. The exponents obtained here are v~~r =

0.65 and

v

i

Length

Fig. 4.

- The

averagend-to-enddistance for

adsorbed polymers on the surface as a

function of the

(11)

l150 JOURNAL DE PHYSIQUE II N° 7

v~~p = 0.66. The two lines in

figure

4 are almost

parallel.

The exponents v are almost the same

as that in the HOM case

(vHoM

~

0.62) [10].

The small difference in the exponent v does not

necessarily

mean that the conformations are similar it

simply

means that the difference in the conformations cannot be differentiated

by

the measurement of the end-to-end distance in the

case of the

type

of triblock

copolymers

studied here.

In order to

study

the difference in the conformation, we examine the

layer

thickness in the

REP and ATT cases for a fixed value of the coverage and the adsorbed amount. The definition

of the

layer

thickness used here is as follows. We record the

largest

n

(plane number)

of each chain adsorbed on the surface. The

largest

n may be at an end of the adsorbed

polymer

or on an

intemal segment. We call this

quantity

Z~~~ and the results are shown in

figure

5a in a

log-log

scale. The Z~~~

approximately

follows a power law in the

polymer length (cfY ).

As

expected,

the Z~~~ is

bigger

in the REP than in the ATT for short

polymers.

This is

understandable,

because for short

polymers

the tails are the main contribution to

Z~~~.

As the

polymer length

increases, the difference in the two

Z~~~'s

decreases. This is because as the central part of the

polymer (B block)

grows, the difference between the REP and ATT decreases. This behavior

makes sense, because the end size

oust

the first and the last segments in the

polymer

or the block

A)

are

proportionally decreasing.

The exponent y is

bigger

in the ATT

(0.61)

than REP

(0.47)

until

approximately

50 segments,

beyond

which Z~~~ follows the exponent of the REP,

which is in agreement with y = 0.5 of the HOM case

[10].

There is a second way of

characterizing

the adsorbed

layer,

which we call the thickness T.

Using

the

polymer length f,

its fraction in

loops fj~~~

and its average number of

loops (Nj~~~),

the thickness T for a

polymers

of

length f

is defined as

~

2

~~fl)

~~~

The factor two in the denominator appears because in every

loop

about a half of the segments

are used to go up and a half to go down. The results are shown in a

log-log

scale in

figure

5b.

The thickness T can be

interpreted

as the average

height

of the

loop.

The thickness T also

obeys

a power law

(cf~ ).

The

loops

in the ATT are

higher

than those in the REP because of the

preference

of the REP to have

longer

tails as a result of the end

repulsion.

The exponents are

p~~r = 0.48 and p~~p =

0.56. The

bigger

p exponent in the REP case

implies

that for very

long polymer

chains the difference in the thickness T will

disappear.

As a

reference,

we may quote that the

exponent

p in the HOM case is 0.51

[10].

The number of

loops

of the adsorbed chains also

gives

us information of the conformations

adopted by

the

polymers.

The average number of

loops (Nj~~~)

for the ATT

(squares)

and the

REP

(circles)

are shown in a

log-log

scale in

figure

5c. The number of

loops

is

larger

in the ATT than in the REP case ; what is

responsible

for this behavior is

basically

the

preference

of

the ends in the ATT to be close to the surface. The difference

persists

for chains of as

long

as

90

segments, showing

that the conformations of the adsorbed

polymers

are

strongly

modified

by

the end

behavior,

even when the size of the ends

(A block)

are

relatively

small. The number of

loops obeys approximately

a power law

(cf~ ).

The exponents K are I. II and 1.89 in the

ATT and the

REP, respectively.

The exponents in both cases are different from the HOM case 1.72

[10].

The fact that the exponent K is modified more

strongly

in the ATT than in the REP indicates that in the HOM case there is a natural

tendency

of the ends to be away from the surface due to the entropy effect ; this effect is shown

by

the

preferential adsorption

for

longer

chains.

Before

continuing

with our simulation

results,

it is worthwhile to comment that in many of the

experimental

data the thickness of the adsorbed

polymer layer depends

on the

technique

used in the measurement as has been

pointed

out

by

many authors

[3, 8, 26, 27].

The methods

(12)

(a)

Z

max

o o

,,6

a

,d

~

i lo loo zoo

(b)

a

Thickness

o

o

lo loo zoo

(c)

Number of loops

,,' a ,,'

,,' a

,,'

,"" o

,,"

Lenqth

~°°

Fig. 5. -Some of the geometrical properties of the adsorbed REP (circles) and ATT

(square)

as

functions of the length (a) Z~~~, (b) thickness T, and (c) the average number of loops

(N

j~~~). Marks are the simulation output.

(13)

1152 JOURNAL DE PHYSIQUE II N° 7

used for the measurement can be

mainly split

in

dynamics

and statics as Cohen Stuart and coworkers

suggested [3],

and it seems that different

techniques give

information about different aspects of the

layer

thickness.

They give hydrodynamic layer

thickness and static

layer

thickness. The

experimental

data indicate that the

hydrodynamic

thickness is

always larger

than the thickness determined

by

the static

techniques

:

ellipsometry

and neutron

scattering [3, 8, 26, 27].

A very

interesting

observation was made

by Cosgrove

et a/.

[27]

who found that the

profile

measured

by

small

angle

neutron

scattering

vanished

completely

at a

distance which is about half of the

hydrodynamics

thickness.

Cosgrove's

observation

strongly

supports the idea that the adsorbed

polymer layer

has two parts a dense part close to the solid surface and a sparse part

(sponge like)

at the top of the dense part. The dense part of the

layer

thickness may be related with the

ellipsometry

and neutron

scattering

measurements while the

outermost

(exterior)

may be related with the

hydrodynamic

measurements. Since in our

simulation results the thickness T is related with the average

height

of the

loops,

we associate this with the dense part of the adsorbed

layer

while the

Z~~,

which is related with the farthest

polymer

segment is

interpreted

as the «

hydrodynamic

»

layer

thickness. For the purpose of

comparing

our results with the

experimental

observation we calculate the

following

ratio :

~max

The results for C are shown in

figure

6. Since most of the

experimental

results are for very

long polymers,

we are

just commenting

on the calculations for

polymers longer

than 50 segments

(although

the entire range of the

length

is shown in the

figure).

For this range the C value increases very

slowly

as a function of the

polymer length.

Its values in the REP and HOM

cases are in the same order of

magnitude

as

Cosgrove's experimental

observation in which he

points

out that the dense part of the adsorbed

layer

thickness is about a half of the

hydrodynamic

thickness

[8, 27].

In the ATT case the dense part of the adsorbed

layer

is

larger

than that in the REP and HOM cases due to the fact that the end segments are

strongly

attracted

to the

surface, encouraging loop

formation and

resulting

in the

larger

values of T and hence the

larger

values of C in the ATT case. A

question

that may arise is whether or not the C ratio

keeps

on

growing

or levels off as the

polymer length

increases. We cannot answer this

question

because our simulation results are [imitated to a finite

length

range.

~

).

~Gl

'Gt

_~

u ; ._ t- ~ ~_~ ~-~ Q -tH a ~

'A.,

~_ ~_

i

0 25 50 75 100 125 150

Length

Fig.

6. Simulation results for the C

=

~ ratio as a function of the

polymer length

for the cases

~max HOM (triangle), ATT

(square)

and REP (circle).

(14)

The area covered

by

the adsorbed

polymers

is also sensitive to the triblock behavior or the end effect. In order to see this, we calculate the area cover

by

an adsorbed

polymer,

which is

the square area that circumscribes those

polymer

segments on the n

=

I

plane

for every adsorbed

polymer

chain. We take the average for all chains with the same

length.

We call this

area the

A~~~.

The results of this calculation in units of square lattice constant are shown on the

log-log

scale in

figure

7. The results show

clearly

that the

A~~~

follows a power law in the ATT

and REP cases

(cf~

). The

exponents

bare 1.2 and 1.6 in the ATT and REP,

respectively.

The

exponents are

approximately equally

deviated from the HOM case for which the exponent is 1.42

[10].

Amax

~b ,6'

~

u

~

,,'

~r#~ ~~

~p

a

~

,u'~

,,d' ,'~

io loo zoo

Length

Fig. 7. Simulation results of the A~~~ or the maximum area covered by the adsorbed polymers for REP

(circle) and ATT

(square)

as a function of the

length.

We can get more information about the conformation and its variation with the

polymer length using

A~~~ and

Z~~~.

In order to find more information about the conformation we

calculate the

following

ratio

~

fi(8)

~max

This R

gives

us information about the lateral view of the adsorbed

polymer

; there are three

cases as is illustrated in

figure

8a. The cases are the

followings, I)

When R

~ l, a small part of the adsorbed

polymer

touches the surface and the rest of the

body

is

mainly

extended into the

solution. The

polymer

can be contained in a vertical

rectangle,

and the entropy effect is more

important

than the energy effect in this case,

ii)

When R

- I the adsorbed

polymer

can be

contained in a square, and in this case there is a balance between the entropy and the energy contribution and both effects are

important. iii)

When R ~ l the adsorbed

polymer

is

mainly

kept

close to the surface and the

polymer

can be contained in a horizontal

rectangle,

so that in this case the energy effect is more

important

than the entropy effect. In

figure

8b we can see

(15)

l154 JOURNAL DE PHYSIQUE II N° 7

la)

( I R=I R>1

~'~ ~

tl~

~

~'~~ ~

mm nT~

~'~

ar ~_ w

o.m w. a.

.a.'~'

o.50

(~)

O.30

O 25 50 75 loo 125 150

Length

Fig. 8. Simulation results for the R

=

~

ratio as a function of the polymer

length.

This ratio

gives

ax

us information about which is the leading factor in the conformation of the adsorbed

polymers

the energy-dominated when R

~ l and the entropy-dominated when R

~ l. The marks correspond to the simulation

points,

the

triangles

are for HOM, the circles are for REP and the squares for ATT.

that in the ATT case the conformation

adopted by

the adsorbed

polymers

are

mainly

dominated

by

the energy process in the entire range of the

length,

while in the REP case there is a

continuous evolution from conformations in which the entropy dominates the

adsorption

for

short

polymers

to conformations in which the energy dominates the

adsorption

for

long

polymers.

In the same

figure

also shown are the HOM data as a reference and we can see that its behavior is more similar to the REP case than to the ATT case. In the case of very

long polymers,

the results are as

expected,

the end effects

becoming

less

important,

and the

ATT,

REP and HOM show the same behavior.

We counted the fractions in the

train,

tail and

loop portions

of the

polymers

adsorbed on the surface. The fractions are shown in

figures

9a and b for the REP and

ATT, respectively.

The tail fraction is enhanced in the

REP,

while it is reduced in the ATT. The train fraction in the REP shows a weak

dependence

on the

length

and later it tends to level

off,

while for the ATT there is a clear

dependence

on the

length.

The

loop

fraction increases in both cases as the

polymer length

increases, but also the

levelling

off should be present for very

long polymers.

The

loop

fraction is

bigger

in the ATT than in the REP since the

preferential adsorption

of the end segments enhances the

loop

formation. In both cases the

loop

fraction becomes the

largest

fraction for very

long polymer

chains. The

loop

fraction is

bigger

in the ATT than in the REP in

agreement

with the

larger layer

thickness in the ATT than in the REP as was shown in

(16)

~

Toii (a)

lJ> _,..6.

..."~"'K"~.&.A,~

~

b ___f.."~'~

~~~

4 ..."""'

I

°

o

T~0i~

~

LL-

o

a

Loop

lo loo zoo

~,~ o

Troin ~~~

#~ O~

~

Toil

b

G b~

O

~o~

~

a~~

~a~

ad

~.A

o~o

,a..~

~ ~

...~.""~

nooa~~~"

a

Loop

..:

~~

' '°

Le n gth

'°° ~°°

Fig.

9. Simulation results of the fractions in the trains, tails and loops of the adsorbed

polymers

: (a) for the REP and (b) for the ATT.

figure

5b. Our calculation for the train, tail and

loop

in the ATT case is in

qualitatively good

agreement with some recent Monte Carlo simulation

(MCS)

results for

amphiphilic

triblock

copolymers [28]

as is illustrated in

figure

10 where the MCS are

presented. However, slight

numerical differences are present because in the MCS the central block is not adsorbed while in

our case the central block is

adsorbed, although

more

weakly

than the ends. This

qualitative

agreement

indicates that both the MCS and the CVM simulations

predict mainly

the same behavior for similar type of

polymers.

Références

Documents relatifs

We prove that arbitrary actions of compact groups on finite dimensional C ∗ -algebras are equivariantly semiprojec- tive, that quasifree actions of compact groups on the Cuntz

In this paper we study a surface which has many intriguing and puzzling aspects: on one hand it is related to the Fano surface of lines of a cubic threefold, and on the other hand it

Par contre, les eleves expliquent moins bien les moyens qu'ils utilisent pour verifier diffe- rents aspects de leur texte (voir les exemples presentes a la page 85 des Modeles

A Hardware Platform is a family of architectures that satisfy A Hardware Platform is a family of architectures that satisfy a set of architectural constraints imposed to allow the re-

The problem list has been further analyzed in an attempt to determine the root causes at the heart of the perceived problems: The result will be used to guide the next stage

The MPLS MIB modules assume the interpretation of the Interfaces Group to be in accordance with [RFC2863], which states that ifTable contains information on the managed

While a CA is typically advised against issuing a certificate with a validity period that spans a greater period of time than the validity period of the CA’s certificate that

It is to be noted that the C-multicast routing procedures will only impact the group join latency of a said multicast stream for the first receiver that is located across