• Aucun résultat trouvé

On the Validity of the Maxwell Model and Related Constitutive Equations; a Study Based on the First Normal Stress Coefficient

N/A
N/A
Protected

Academic year: 2021

Partager "On the Validity of the Maxwell Model and Related Constitutive Equations; a Study Based on the First Normal Stress Coefficient"

Copied!
12
0
0

Texte intégral

(1)

HAL Id: jpa-00247469

https://hal.archives-ouvertes.fr/jpa-00247469

Submitted on 1 Jan 1997

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.

On the Validity of the Maxwell Model and Related Constitutive Equations; a Study Based on the First

Normal Stress Coefficient

H.-P. Wittmann

To cite this version:

H.-P. Wittmann. On the Validity of the Maxwell Model and Related Constitutive Equations; a Study

Based on the First Normal Stress Coefficient. Journal de Physique I, EDP Sciences, 1997, 7 (12),

pp.1523-1533. �10.1051/jp1:1997153�. �jpa-00247469�

(2)

On the Validity of the Maxwell Model and Related Constitutive

Equations;

a

Study Based

on

the First Normal Stress Coefficient

H.-P. Wittmann

(*)

Max-Planck-Institut ffir

Polymerforschung, Ackermannweg

10, 55128

Mainzi Germany

(Received

4

April

1997, revised 21

August

1997,

accepted

2

September 1997)

PACS.47.15.-x Laminar flows

PACS.47.27.-I Turbulent flows, convection, and heat transfer PACS.47.50.+d Non-Nevionian fluid flows

Abstract. In a previous

publication (J.

Phys. f £hance 4

(1994) 1791)

the semi-microscopical Rouse model was

projected

onto macroscopic

Langevin equations

for concentration and stress

tensor fluctuations.

Adding

afterwards appropriate convective terms due to a simple shear

flow the

resulting

equations were shown to be a

generalization

of the upper convected Maxwell model. Based on the projected equations the first normal stress coefficient flfi,o is now calcu- lated in this publication: flfi,o =

2~]/kBTco

where ~o and co denote the zero-shear

viscosity

and

the monomer concentration

respectively.

The

resulting steady

state

compliance

Js is found to

be Js

=

1/kBTco.

This turns out to be in

disagreement

with

experimental findings.

Based on the derivation of the

Lodge

equation one can conclude the

following:

In order to get the right first normal stress coefficient it is necessary to

incorporate

the shear flow at first into the Rouse

model and to project afterwards the modified Rouse model.

Introduction

A lot of constitutive

equations

for

polymer

melts and solutions were accumulated over the

course of time.

They

are reviewed

by

Larson

ill.

Most of these constitutive

equations

refer to

a stress tensor

averaged

over the

sample.

This

corresponds

from an theoretical

point

of view

to the restriction of the

hydrodynamic

limit.

Moreover,

most of these constitutive

equations

known

today rely

upon

phenomenological

elements like stress relaxation rates.

On the other hand shear enhanced concentration fluctuations in

polymer

solutions have attracted a lot of attention in recent years

[2-llj.

From an

experimental point

of view one

was interested for

example

in

measuring

the

non-equilibrium steady

state structure factor.

As demonstrated

by

Wu et al. [4j this

quantity

turned out to become

anisotropic

due to the presence of an external shear flow.

In order to be able to calculate the structure factor under shear at first a reliable

equation

of motion for the concentration fluctuations is needed. As shown

by

Helfand and Fredrickson [2j elastic forces characteristic for a

polymeric system

lead to a

type

of elastic stress-induced diffusion term that has to be added to the

equation

of motion for the concentration fluctuations.

In this way the concentration fluctuations and the fluctuations of the stress tensor field become

* e-mail:

wittmann@th21.mpip-mainz.mpg.

de

@

Les

#ditions

de

Physique

1997

(3)

coupled.

In order to close the

equations

of motion a constitutive

equation

is needed. This fact has revitalized the interest in constitutive

equations.

To be more

specific,

Helfand and Fredrickson [2j formulated for sheared

polymer

solutions a

phenomenological theory

where

coupled equations

of motion for monomer

concentration,

the

velocity

field of the

solvent,

and the deviatoric

polymeric

stress tensor field were

presented.

For the

equation

of niotion for the stress tensor variable these authors utilized the so-called second order fluid

[lj.

In

modeling

the viscometric coefficient as functions linear in the concentration fluctuations

they

were able to

explain

the

anisotropic growth

of concentration fluctuations in the presence of shear flow.

Simultaneously

Onuki [3j

published

a

theory

that could be used to describe

polymer dy-

namics in

solution, assuming

either Rouse or

reptation dynamics depending

on the

polymer

concentration. In this

context,

he used Marrucci's constitutive model

[12j

in which a

long

lived strain variable is used as an alternative to stress in the second order fluid model. Later on MiIner

[6,9j presented

a

dynamical theory

for

entangled polymer

solutions under shear based

on the two-fluid model introduced

by

Brochard and de Gennes

[13j

and discussed

by

Onuki

[14j

and Doi and Onuki

[lsj. Finally

Ji and Helfand

ill]

used for their

theory

also the two-fluid

model to obtain

Langevin equations

for

concentration, velocity

and a defined

polymer

strain.

Since the fluctuations of the concentration and the strain tensor are slow

compared

to the

velocity variables,

these authors eliminated the

velocity

variables from their set of slow vari- ables. Both MiIner [9j as well as Ji and Helfand

ill]

were able to calculate

mimerically

the

non-equilibrium steady

state structure factor in excellent

agreement

with results obtained

by

Wu et al. [4j ~ia

light scattering experiments.

As

already

mentioned above the constitutive

equations

as reviewed

by

Larson

[lj

are usu-

ally

based on

phenomenological arguments.

In order to overcome this

phenomenology

it was

speculated early

on [2,

6j

that it should be

possible

to

project

the

semi-microscopical

Rouse niodel

[16,17j

onto

macroscopic equations

of motion for

polymers

under shear. The set of re-

sulting coupled Langevin equations

for collective variables like concentration and stress tensor fluctuations can then be considered as a

generalized

constitutive relation. For a

polymeric

sys-

tem under shear there are in

principle

two

possibilities

to

proceed:

The first one is to

project

at first the Rouse model where the shear flow is turned off onto

macroscopic equations

and to add in a second step

appropriate

convective terms. This

approach

was

adapted by

Wittmann and Fredrickson

[18j.

In order to deduce

macroscopic equations

of motion from the Rouse model

one has to know how to calculate the matrix of

decay

rates or memory kernels and the matrix of

Onsager

coefficients in a

systematic

fashion. Based on the

projection operator technique

of

Zwanzig

and Mori

[19-22j

Wittmann and Fredrickson

[18j

found the

following:

The matrix of

decay

rates is

given by

the matrix of the

dynamical

response functions

multiplied by

the

inverse of the matrix of

dynamical

correlation functions.

Furthermore,

since the

projection

op-

erator

technique

assumes

implicitely

a free energy that is bilinear in the collective variables the matrix of

Onsager

coefficients is

given by

the matrix of

decay

rates times the matrix of static

correlation functions. Due to the Gaussian nature of the Rouse model

[16,17j

the

required

correlation and response functions can be determined

[18j

at least in the

hydrodynamic

limit.

The

Onsager

coefficients calculated in this way can be used to formulate

Langevin equations

for concentration and stress tensor fluctuations. These

equations

contain as the

hydrodynaniic

limit the upper convected Maxwell model

[lj.

Therefore these

equations

shall be called the

generalized

Maxwell model.

On the other hand there is a second way to

project appropriate equations:

At first convective

terms are added in the

equation

of motion for individual Rouse modes. These so modified Rouse

triodes [17] are then used for a

projection

onto

macroscopic equations

of motion.

Up

to now

such a

procedure

was not

persued. However,

there is a constitutive

equation

known as the

Lodge equation

or the rubberlike

liquid

model

[I,17, 23]

which -relies on the Rouse modes under

(4)

shear.

Starting point

for this constitutive

equation

is the thermal average of a

microscopic expression

for the stress tensor in the

hydrodynamic

limit where this

quantity

is

essentially

a

sum over second moments of individual Rouse modes. If these moments are calculated for the

case that a shear field is

present

one finds a stress tensor which can be

expressed

in terms of the shear modulus and the

finger

strain tensor

[17].

The

deficiency

of the

Lodge equation

however is that these constitutive

equations

start from the very

beginning

with the

hydrodynamic

limit.

As described

above,

constitutive

equations

are needed as an

input

for the calculation of the

non-equilibrium steady

state structure factor under shear.

However,

constitutive

equations

can also be used to

predict

normal stress differences

[1,17,23]. By comparing

these

predictions

with

experimental

results the normal stress differences can be used to decide whether a constitutive

equation

is reasonable or not. This

approach

will be

adapted

in the

present

paper. In Section the

generalized

Maxwell model introduced in reference

[18]

shall be recalled. In Section 2 the

viscometric coefficients are calculated in the framework of the

generalized

Maxwell model. As a

byproduct

the

Lodge

Meissner

relationship [24-27]

is rediscovered. Section 3 is used to review the

Lodge equation including

the associated viscometric functions

following

reference

[17].

As it will be discussed in Section 4 the

generalized

Maxwell model and the

Lodge equation yield

the

same result for the

viscosity

but differ in the

prediction

with respect to the first nornial stress coefficient. Based on the

viscosity

and the first normal stress coefficient the so-called

steady

state

compliance

can be defined. From

experiments

done

by Onogi

et al.

[28]

on melts of narrow- distribution

polystyrenes

it is known that the

compliance

is for

sufficiently

short

polymers (Rouse polymers) proportional

to the

degree

of

polymerization,

while the

compliance

beconies

independent

of the

degree

of

polymerization

for

sufficiently long polymers (reptation regime).

As it is well known the

compliance

as

predicted by

the

siniple Lodge equation

is in

agreement

with the

experimental findings.

The more elaborate

generalized

Maxwell model however

yields

for Rouse

polymers

a

conipliance

which is

independent

of the

degree

of

polymerization.

This is in clear

disagreement

with the

experimental

result. Based on these

findings

one can conclude the

following:

If one is interested in a

projection

of the Rouse model onto

macroscopic equations

of motion one has at first to

incorporate

the effect of a shear field into the

semi-microscopical

Rouse model. The Rouse modes under shear can then be used for a

projection

onto

macroscopic Langevin equations

for collective variables like concentration and stress tensor fluctuations.

1. Generalized Maxwell Model

In reference

[18]

for a melt of Rouse

polymers coupled Langevin equations

were formulated for collective variables

c(q, w)

and

Q(q,

++ w defined as follows:

c(q, w)

=

/ ~ dte~~"~

/

ds

e~~~'~~~'° coda l

N

~

In this context

r(s, t)

denotes the

position

of a Kuhnian

segment

at contour

position

s and time t whereas b, N and co are the

segment size,

the number of Kuhnian

segments

per chain and the

averaged

mononier concentration

respectively.

Therefore

c(q, w)

is the fluctuation of

++

the nionomer concentration about its average value co and

Q(q,w)

is the fluctuation of the

deviatoric stress tensor field measured in units of

kBT

times the

polymer

concentration

co/N.

As

pointed

out in reference

[18j

the

equation

of niotion for the concentration fluctuation con- tains a term

proportional

to the

divergence

of the elastic force

density

field which is

essentially

(5)

++

qq

kBT Q(q, w).

In reference

[18] advantage

was taken from this

specific

sort of

coupling by decomposing

++

Q(q, w)

into

longitudinal

and transverse

components Q"(q, w)

=

#~#po~p(q, w)

and

Q~(q,w)

=

(d~p (~(p)Q~p(q, t),

with

(~

= q~

/q.

This leads to

(c, Q", Q~)

as a set of collective variables. Based on the

projection operator technique

of

Zwanzig

and Mori

[19-22j

the

Langevin equation

for a collective variable A e

(c, Q", Q~)

in the presence of an external shear flow reads:

iw iqx )lAiq w)

-

iQ14 rt~ i~ ~d)iBi~

~°~ +

H~i~

~°~. i~~

For

siniple

shear flow characterized

by

a flow field

u(x)

=

§vex

with shear rate

§

the differential

operator §qxb/bqy

describes the convection of Kuhnian

segments by

the flow field whereas the

frequency

matrix

in((

with the

following

non

vanishing

niatrix elements

jajjc

1

=

jai(Q"

=

-iQ)$~

~

~~~~~

~~

~'~~~~~

appears as a consequence of the reorientation of Kuhnian

segments by

the flow field. The

quantity El~(q, w)

is a randoni force with zero mean,

(B~(q, w))

=

0,

and covariance

ie~lq, w)e~ I-q, -w))

=

2Rel©~~'"~l

where the

right

hand side denotes up to a factor of 2 the real

part

of the

Onsager

coefficient

A(~(q,w)

with

A,

B e

(c,Q",Q~).

Both the

Onsager

coefficient

A(~(q,w)

as well as the

decay

rate

r)~ (q, w)

which appears in the

equation

of motion were calculated in reference

[18j

in the framework of the

semi-microscopical

Rouse model.

However,

in order to have access to

++

the normal stress coefficients it proves convenient to undo the

decomposition

of

Q

into Q" and

Q~

This leads to the

following equations

of motion

iw iqx)lclq,

uJ) =

-rl~lq, uJ)clqiuJ) rl~lq, uJ)=iilq,

uJ) +

e~lq,uJ) 12)

»

~

l~

H ~ H H ~T ~

iW

iqx~) Qlq,

W) ~

Qlq, uJ)- Qlq,

uJ) ~ 2 e

clq,

uJ)

=

y

=

-ri~(q, w): ~(q,

w) ri~(q, w)c(q, w)

+

El§q, w). (3)

The left hand side of the last

equation

denotes the upper convected time derivative. The term

2$c(q,w)

appears in order to compensate the fact that not the stress itself but the deviatoric stress is used

[18j.

The

quantity I

=

(i7u)~

with ~~p =

§d~xdpy

is in this context the

velocity gradient

tensor whereas

$= (1/2)(I

+

i~)

with e~p =

(§/2)(d~~dpy

+

d~ydpx)

is known as the rate of strain tensor. Due to

equation (20)

of reference

[18]

the

decay

rates are in

leading

order

given by

r[~(q~w)

=

+Dq~ r[~(q~ w)

=

-(kBT/()qq

,

rf~(q, w)

=

-2Dqq (4)

where

(

is the friction constant and D the center of mass diffusion constant. In the

hy-

drodynamic

--

_ _

no(q,w):Q (q,w) might

(6)

an

appropriate

scalar. Furthermore it proves convenient to use instead of

no(q, w)

the Green's

function

~~~~~'"~

iuJ +

~~(q,

w) ~~~

~

~~~~~

~~~

In the last

relationship

which follows also from

equation (20)

of reference

[18] ~*(w)

denotes the

complex viscosity

which is related to the shear modulus

Gjt)

=

~~~~° f~-2t/rp j~)

N p=1

where Tp denotes the relaxation time of Rouse mode p ~ia a

Laplace

transform:

~*(w) =

pling

of

the equation

of

niotion

for

concentration and

stress

tensor

luctuations. Since both

quantities are of

order

O(qq) the quations of motion decouple in the strict ydrodynamic

limit.

Furthermore,

if equation

(5)

is

not

used but

if

nstead the quantity

no(q,w)

placed

by

ome

phenomenological tress ecay rate I/TQ,

equation (3) yields in

real space

in

the

rodynamic limit

the

following

relationship:

~ ++

_ H ++ _T i H _

bikBT Q +q i7kBT Q

~

kBT Q -kBT Q

~

=

-kBT Q

+2 e

G(0) (7)

TQ where the noise was

suppressed

and where

G(0)

=

kBTc

is

interpreted

as the shear modulus for t

= 0. The second term on the left hand side of the last

equation

describes the convection of Kuhnian segnients due to the flow field whereas the third and fourth ternis niimic the

reorientation of Kuhnian

segments

due to the flow field.

However,

since

equation (7)

is

nothing

else than the upper convected Maxwell model

ill,

the set of the

coupled equations (2, 3)

can

be considered as a

generalized

Maxwell model.

2. Viscometric Functions

In the

following

the considerations of reference [18] shall be extended

by calculating

the vis-

-- _ _

cometric functions. For this purpose the

replacement no(q, w):Q(q, w)

~

no(q, w) Q(q, w)

shall be used

again.

In terms of

ili~(q, w)

=

I/GiQ(q, w)

the

equation

of niotion for the stress

variable I.e.

equation (3)

niay be rewritten as

§q

~

Q jq ~)

=

_~yQQjq ~) Q jq ~)

+

gojq ~) j8)

~bqy

~

' ° '

~

'

~

'

where the

quantity

°~lq, w)

=

t. llq, w)+ 41q,

w) t~

+

12? -ri~lq,w)iclq,w)

+

6ilq, w)

has the

nieaning

of an

inhomogeneity.

In order to see this more

clearly

it is instructive to write

_ ++ ++ _T ++

down the elements of the tensors ~.

Q

+

Q

.~ and e

explicitly:

_ _ ~

ilo~»+Q»x) iQ»» iQ»z

o

I

o

~. Q

+

Q .~

=

§Qyy

0 0 and 2

$= §

0 0

iozy

0 0 0 0 0

(7)

Equation (8)

formulated for the tensor element

Qxx(q,w)

for

example

contains the tensor element

0~~~(q,w).

This

quantity

is a functional of

Qxy(q,w)

and

Qyx(Q, w)

but it is inde-

pendent

of

Q~~(q, w).

Therefore

0~~~(q, w)

can be considered as an

inhomogeneity.

A similar

++

behavior is also true for all the other tensor elenients of

Q(q,w).

As a result the methods of characteristics

[2,29]

allow one to rewrite

equation (8)

as

llq, w)

=

/~ dt'exPl- /)dt" 4~i~lqlt"), w)1°~lqlt'), w) 19)

where the time

dependent

wave vector

q(t')

= q

t'§q~ey

was introduced. In the

following

the noise shall be

neglected again.

Then one finds for

off

e

(yy,

yz,

zy,zz)

QaPlq, w)

=

/~ dt' exPl- /)dt" i~i~ lqlt"), w)in°~~ lqlt'), w)Clqlt'), w).

This

expression

can be

expanded

in a power series with

respect

to the shear rate

§.

The

zero order contribution is obtained

by

the

replacements q(t')

~ q and

q(t")

~ q. Then the

integrations

with

respect

to t' and t" can be done:

/~~dt' ~XPI

)~~"~i~~~'~°~ Yi~)q,

°~)

~~~~~~ ~~ ~~~~

~

~~~~

where

~*(w)

was for convenience

replaced by

its low

frequency

limit ~o =

limw~o ~*(w). Taking equation (4)

into account it can be seen that for

off

e

(yy,yz,zy,zz)

the deviatoric stress

tensor in the

hydrodynamic

limit is

given by:

QaPlq,1°)

"

2fiDqnqpClq,1°)

+

Oli) 11°)

At next the shear stress

Q~y(q, w)

shall be considered. At first

equation (9) implies:

Qxyiq,w)

=

f~dt'expj- f)dt"ilyojqjt"),w)j

X

i)Qyyiqit'), ))

+

ii Fi~~~ iqit'), ~°)iciqit'),

L°)1

However,

since

Qyy(q,w)

=

O(qyqy)

due to

equation (10)

and

ri~~~(q,w)

=

O(q~qy)

due to

equation (4)

both

quantities

shall be

neglected

in the

following

for the

hydrodynamic

limit.

Analogous

to the

reasoning given

above a power series

expansion

with

respect

to the shear rate

I yields immediately

kBTQxy(~,td)

"

kBTQyx(~,td) ~'f~0~~~~~~

+

°l'f~) (11)

Co

since the tensor

~j

is

symmetric. Finally,

if ternis

containing ri~~~(q,w)

and

ri"~(q,w)

are

again neglected

because

they

are of order

O(q~)

one

obtains,

due to

equation (9),

kBTjoxxlq, w) Qyylq,

w)1 =

2i ~j)~~ kBToxylq, w)

+

°li~) (12)

Such an

equation

is known as the

Lodge

Meissner

relationship [24-27j.

Note that T~ =

~o/kBTco

is a characteristic time and that the

product §T~

can be considered as

(8)

the Deborah number

[lj. Furthermore,

in

combining equations (11, 12),

one finds for the first normal stress difference

[lj

kBT(Qxx(~> td)

QyY(~>

td)j ~'f~ /~~~ ~~j~~

~~~~

Finally

it shall be nientioned that in the framework of the

approxiniations

made above the second normal stress difference

[lj

vanishes:

kBTioyylq, w) Qz=lq,

w)1

= °.

l14)

The viscometric functions which are the

viscosity dqo

(cj and the first and second normal stress coefficients

dill,o[cj

and

d1l2,o[cj respectively [lj

are in the context with the deviatoric stress

++

tensor field

kBT Q

defined as

d~o(c]

=

lint(kBToxyli) (15)

i~o

di~l.0(cj

#

11r$(kBT(Qxx QyYjl'l~) (16)

~~

~~y~~~jcj =

j~IkBTIQYY Q~ll'~

~ ~~~~

From

equation (12)

one can

identify: d~o[c]

=

~oc/co

where ~o

=

ffdtG(t)

denotes the

viscosity.

In a similar fashion one can write:

dili,o(c]

=

iii,oc/co

and

d1l2,o(c]

=

1l2,oc/co,

where

iii,o

and

1l2,o

are known as the zero shear limit normal stress coefficients. In addition

a

quantity Js

known as the

steady

state

compliance

is defined via

Js

=

iii,o/2~(, Equations (12-14) yield:

~° ~~ ~~~~

'

~~'°

~

k~ico

'

~~'°

~

'

~ B~co

~~~~

3.

Lodge Equation

The

perspective

of this section is to review the calculation of the viscometric coefficients on the basis of the

Lodge equation

also known as the rubberlike

liquid

model

following

reference

[17].

For a melt of Rouse

polymers

the

fluctuating

stress tensor field

Six, t)

at

position

x and time

t reads:

~ ~~'~~ ~~~

~~~

~~~'~~~

~~~ ~~~

~~

~~~

~~ ~~~~

Note that the stress tensor

$ (x, t)

is related to the deviatoric stress variable

lx, t) through

_ ++ ++

a

(x, t)

=

kBT[Q (x, t)

+

c(x, t)

d]. For the constitutive

equation

known as the

Lodge equation

one considers

usually

not

$ (x, t) itself,

but the stress tensor

IS it))

=

/ d~xi?ix t)i

=

j ~(i~ j~dsi~~l'~~ ~~l'~~1 120)

From an

experimental point

of view this means to average over the

sample

which

corresponds

from a theoretical

point

of view to consider the

hydrodynanlic

limit.

Representing

the

position r(s, t)

of a Kuhnian

segment

in terms of the set

(xp(t))p

of Rouse modes [17] as

cc

r(s, t)

= xo

It)

+ 2

~j

xp

(t)cos(~~~

N P=1

(9)

the stress tensor as

given by equation (22)

can be rewritten [17] in the

following

manner

IS it))

=

) f

kpixplt)xpit)) 121)

where

kp

=

(67r2kBT/Nb~)p~.

In order to derive from here an

expression

for the stress tensor in the presence of

simple

shear

flow,

the

Langevin equation

for an individual Rouse mode

xp(t)

is modified

by adding

an

appropriate

term due to the shear flow

[17].

This

implies

a first order differential

equation

for

lip it)

=

(co/N)kp(xp(t)xp(t))

which is

given explicitly by [17]

bi imp It)) t imp lt)) imp it)) t~

=

-( (i?p It)) ~~(~° ij 122)

~ ~T

While Tp denotes the relaxation time of Rouse mode p, ~ and ~ are the

velocity gradient

tensor and his

transposed counterpart.

As

pointed

out in reference

[17j

the differential

equation (24)

can be solved without

difficulty.

In ternis of the shear modulus

G(t)

as defined

by equation (6)

and the

finger

strain tensor ++

Bit) [23]

the solution of

equation (22) yields

in combination

with

equation (21) immediately:

ii it)

~

f~ dt'°Gij~j t') tit t')

123)

For

simple

shear flow which will -be considered here

only

the tensor elements of the

finger

strain tensor are

given explicitly by [17j

~

B~p(t)

=

d~p

+

§t(d~~dpy

+

d~ydp~)

+

i~t~da~dp~

Equation (23)

is known as the

Lodge equation

or the rubberlike

liquid

model

[lj

and re- sults also from the

Green-Tobolsky

model

[1,17, 23, 30-33j.

The viscometric coefficients which

are the

viscosity

~o

"

limi~o((a~y) Ii)

and the first and second normal stress coefficient

iii,o

=

limi~o (a~~ ayy) li~)

and

1l2,o

"

linii~o (ayy a~~) li~) respectively

are

given

in the framework of the

Lodge equation (25)

as

~o "

/

dt

G(t)

,

iii,o

" 2

/

dt

tG(t)

,

1l2,o

" 0

,

Js

=

/

dt

tG(t) (24)

~ ~

~o

~

where the

steady

state

compliance Js

=

iii,o/2~(

was added

again.

4. Discussion

At first the

conceptual

differences of the

generalized

Maxwell model and the

Lodge equation

shall be discussed. In order to obtain

equations (2,3)

which

represent

the

generalized

Maxwell model at first the

decay

rates

r)~(q,w)

as well as the

Onsager

coefficients

A(~(q,w)

with

A,

B e

(c, Q", Q~)

were calculated for the case that no external shear field is

applied.

As shown in reference

[18j

both the

decay

rates as well as the

Onsager

coefficients can be related to correlation and response functions. These

quantities

can be calculated

explicitly

in the

framework of the

semi-microscopical

Rouse model

[16,17j.

In a second

step

convective terms that mimic convection and reorientation of Kuhnian

segments

due to the shear flow were added

yielding equation (8). By

the method of characteristics this

equation

could be transformed into

(10)

equation (9).

From a mathematical

point

of view the convection of Kuhnian

segments

is now

accounted for

by

the tinie

dependent

wavevector

q(t)

= q

t§q~ey.

In

setting q(t)

to q in the

subsequent equations

the effect of convection is turned off. The occurrence of the shear rate

§

in

equation (11)

to

equation (13)

is thus

entirely

due to the effect that a flow field is able to

reorient Kuhnian segments.

The

strategy persued

in context with the

Lodge equation

differs. There one starts out front the very

beginning

with the stress variable

Ii (t))

which is the

hydrodynaniic

liniit of

the stress variable

Ii lx, t)).

While

I lx, t) depends

both on the

position

as well as on the

orientation of Kuhnian

segments, Ii (t)) depends only

on the orientation of Kuhnian

segments.

Due to

equation (21)

the term

lip It))

=

(co/N)kp(xp(t)xp(t)) represents

the contribution of Rouse mode p to the stress tensor

Ii (t)).

The

corresponding quantity

for the stress variable

++ ++ ++ ++

(kBT Q (t))

=

(kBT Q (q

=

o, t))

is

given by (kBT Q~ (t))

=

(co IN) [kp(xp(t)xp(t)) kBT dj.

In terms of this stress variable

equation(22)

can be rewritten as follows:

bi ikBT (~ it) t jkBT Iv it)j jkBT (~ jt)j t~

=

=

-?jkBT 4p itjj

+ 2

t Gpj0) 125)

where

Gp(0)

=

kBTco/N

is considered to be the contribution of Rouse mode p to the shear modulus at time t

= 0. The last

equation

can now be

compared

with the upper convected Maxwell niodel as

given by equation (7).

The

right

hand sides of both

equations correspond

to one another,

However,

the second term on the left hand side of

equation (7) mimiking

the convection of Kuhnian

segments by

the flow is

missing

on the left hand side of

equation (25).

This behaviour is in agreement with the fact that the considerations

leading

to

equation (23)

started out from the

hydrodynamic

limit

Ii (t))

of

Ii lx, t)).

In reference

[18]

it was

explained

how the

transport

coefficients of the

niacroscopic Langevin equations (2, 3)

can be calculated in the franiework of the

semi-niicroscopical

Rouse niodel. In the

hydrodynamic

limit the

macroscopic transport

coefficients could be

represented

in a

siniple

way in ternis of the center of mass diffusion constant and the shear modulus of the Rouse

model. In

particular

it was shown that the Green's function

G)Q

can be related either to the

complex viscosity ~*(w)

or to the shear modulus

G(t) respectively:

Gi~lq,1°)

~

)(j~~

+

°lq~)

Or

Gi~lq,t)

~

/jj~

+

°lq~)

This result

originally

obtained in the framework of the Rouse model can be transfered to other models on the level of a one

polymer approximation

like the Zimm model

[34]

or the

reptation

model [35]

by exchanging

the shear modulus. As it is well known the shear modulus of the Zimm model has the same structure as that of the Rouse model

given by equation (6).

The

relaxation time Tp,

however,

scales now like Tp

mJ

N~p~~

where N denotes the number of Kuhnian

segments

and where

/L =

3/2

for a 0 solvent and /L

= 3v for a

good

solvent. In the

following

the mean field value v

=

3/5

of the self

avoiding

walk

exponent

v valid in 3 dimensions shall be used. On the other hand for the

reptation

model the shear modulus reads

[17]

G(t)

=

kBTco

~

~

~ ( exp(-p~t/Td)

~~

p,odd l~ ~

where a and b denote the

step length

of the

priniitive

chain and the size of a Kuhnian seg- ment

respectively.

The

disentanglement

time Td scales as Td

~ N~ where the theoretical value

(11)

of the

exponent

is z

= 3 whereas the

experiments yield

x m 3.4. After

specifying

the ex-

pressions

for the shear modulus it is easy to derive how the viscometric coefficients scale with

N,

the

degree

of

polymerization.

In order to

distinguish

the first normal stress coefficient as

given by equation (18, 24)

the

following

notation shall be introduced:

ilf~

= 2Y/(

/kBTco

and

ill

~ = 2

ffdt tG(t)

where the

superscripts

M and L indicate whether the

generalized

Maxwell

midel

or the

Lodge equation

was used to calculate the first normal stress coefficient. Analo-

gously

one has

Jfl

=

I/kBTco

and

J)

=

(1/~() flidt tG(t).

For the Rouse model one finds:

w~N, il(~mJN~, il[,~mJN~, JfImJN°, J)mJN.

For the Zimm model one has to

distinguish

between the case of the 0 solvent with

~~ rW

~/~

~

, il)~Q rW

~/

,

il)Q

rW

~/~

,

~~

rW

~/~ ~~

rW ~/

and the case of a

good

solvent which results for v

= 0.6 in

~QrW~/~~, ilforw~/~~, il)QrW~/~~, ~~rW~/~, ~~rW~/.

Finally

the results of the

reptation

model shall be

given

in terms of the exponent x introduced above:

w ~ N~

,

ilf~

mJ

N~~

,

ill

~ mJ

N~~

,

Jf

mJ

N°, J)

mJ

N°.

Note

that, by accident,

the

scaling

behavior of

ilf~

and

ill

~

coincide for the

reptation model,

whereas

they

differ for the other niodels.

Froni'the expiriments

done

by Onogi

et al.

[28]

on melts of narrow-distribution

polystyrenes

it is known that the

compliance

for

sufficiently

short

polymers (Rouse polymers)

is

proportional

to the

degree

of

polymerization,

while the

compliance

becomes

independent

of the

degree

of

polymerization

for

sufficiently long polymers (reptation regime).

This well known

experimental

result rules out the

generalized

Maxwell

model. In other

words,

in order to

get

the

right

first normal stress coefficient or the

steady

state

compliance respectively

it is necessary to

incorporate

the shear flow at first into the Rouse model and to

project

afterwards the modified Rouse niodel.

Acknowledgments

thank Glenn Fredrickson for a critical

reading

of the

manuscript.

References

ill

Larson

R.G.,

Constitutive

Equations

for

Polymer

Melts and Solutions

(Butterworth, Boston, 1988).

[2] Helfand E. and Fredrickson

G.H., Phys.

Rev. Lett. 62

(1989)

2468.

[3] Onuki

A., Phys.

Rev. Lett. 62

(1989)

2472.

[4j Wu

X.-L.,

Pine D-J- and Dixon

P-K-, Phys.

Rev. Lett. 66

(1991)

2408.

[5] Hashimoto T. and

Fujioka K.,

J.

Phys.

Soc.

Jpn

60

(1991)

356.

[6] MiIner

S.T., Phys.

Rev. Lett. 66

(1991)

1477.

[7] Hashimoto T. and Kume

T.,

J.

Phys.

Soc.

Jpn

61

(1992)1839.

[8] van

Egmond J.W.,

Werner D.E. and Fuller

G-G-,

J. Chem.

Phys.

96

(1992)

7742.

(12)

[9] MiIner

S.T., Phys.

Rev. E 48

(1993)

3674.

[10] van

Egmond

J.W. and Fuller

G.G.,

Macromolecules 26

(1993)

7182.

ill]

Ji H. and Helfand

E.,

Macromolecules 28

(1995)

3869.

[12]

Marrucci

G.,

7kans. Soc.

Rheology16 (1972)

321.

[13] Brochard F. and de Gennes

P-G-,

Macromolecules io

(1977)

l157.

[14j

Onuki

A., J.Phys.

Soc.

Jpn

59

(1990)

3423.

[15]

Doi M. and Onuki

A.,

J.

Phys.

II £hance 2

(1992)

1631.

[16]

Rouse

P.E.,

J. Chem.

Phys.

21

(1953)1272.

[17]

Doi M. and Edwards

S.F.,

The

Theory

of

Polynier Dynamics (Clarendon Press, Oxford, 1986).

[18] Wittmann H.-P. and Fredrickson

G.H.,

J.

Phys.

I France 4

(1994)

1791.

[19]

Zwanzig R.,

J. Chem.

Phys.

33

(1960)

1338.

[20] Zwanzig R., Phys.

Re~. 124

(1961)

983.

[21] Mori

H., Frog.

Theor.

Phys.

33

(1965)

425.

[22] Fick E. and Sauermann

G.,

The

Quantum

Statistics of

Dynamical

Processes

(Springer,

Berlin

Heidelberg

New

York, 1990).

[23] Dealy

J.M. and Wissbrunn

K-F.,

Melt

Rheology

and its Role in Plastic

Processing: Theory

and

Applications (van

Nostrand

Rheinhold,

New

York, 1990).

[24]

Rivlin

R.S.,

Proc. Second Rubber

Technology

Conf.

London,

June

23,

1948.

[25]

Lodge

A.S. and Meissner

J.,

Rheol. Acta11

(1972)

351.

[26]

Lodge A-S-,

Rheol. Acta14

(1975)

664.

[27] Lodge A.S.,

J. Non-Newt. Fl. 14

(1948)

67.

[28] Onogi S.,

Masuda T. and

Kitagawa K.,

Macromolecules 3

(1970)

109.

[29]

Binney J.J.,

Dowrick

N.J.,

Fisher A.J. and Newnian

M.E.J.,

The

Theory

of Critical Phe-

nomena

(Clarendon Press, Oxford, 1992).

[30]

Green M.S. and

Tobolsky A.V.,

J. Chem.

Phys.

14

(1946)

80.

[31] Lodge A.S.,

7kans.

Faraday

Soc. 52

(1956) 120;

Rheol. Acta 7

(1968)

379.

[32]

Yamamoto

M.,

J.

Phys.

Soc.

Jpn11 (1956) 413;

12

(1957) l148;13 (1958)

1200.

[33] Lodge A.S.,

Elastic

Liquids (Academic Press,

New

York, 1964).

[34] Zimm

B.H.,

J. Chem.

Phys.

24

(1956)

269.

[35]

de Gennes

P.G.,

J. Chem.

Phys.

55

(1971)

572.

Références

Documents relatifs

We use limit of sums in a specific form to define the definite integral of a continuous function over a closed and bounded interval.. This is to make the definition easier to

Among quantum Langevin equations describing the unitary time evolution of a quantum system in contact with a quantum bath, we completely characterize those equations which are

We have argued that several such episodes of star formation, due to AGN feedback activity, may account for the observed size evolution of massive galaxies (Ishibashi et al. King 2010

Figure 2, the average SFR and dust-derived gas mass of HRS ETGs is indicated by a filled magenta ellipse, with error bars showing the median absolute deviation of the sample. B): SFR

cyclic tests on a completely aged material, the constitutive equations are deduced by assuming a direct coupling between aging and isotropic hardening, and a

arnong collective polymer variables con be expressed in terras of correlation and response func- tions For a collection of noninteracting Rouse polymers, and with no externat

However, compared to the simulation presented in the previ- ous section, for sufficiently long integration times, we observe a better energy conservation for the Hamiltonian

The idea underlying this schemes relies (see [14]) on a time splitting scheme between configuration space and velocity space for the Vlasov equation and on the computation of