HAL Id: jpa-00247906
https://hal.archives-ouvertes.fr/jpa-00247906
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.
Interface motion in semidilute polymer solutions
Akira Onuki
To cite this version:
Akira Onuki. Interface motion in semidilute polymer solutions. Journal de Physique II, EDP Sciences,
1993, 3 (9), pp.1299-1303. �10.1051/jp2:1993201�. �jpa-00247906�
Classification Physics Abstracts
46.608 64.70J
Short Communication
Interface motion in semidilute polymer solutions
Akira Onuki
Department of Physics, Kyoto University, Kyoto 606, Japan
(Received
17 May 1993, accepted in final form 19 July1993)
Abstract Linear viscoelastic effects are examined in semidilute polymer solutions in macro-
scopic two-phase coexistence. If the temperature T or the pressure p is oscillated in time, the interface motion exhibits gel-like behavior at
high
frequencies and fluid-like behavior at low frequencies. We also study transient relaxation after a step-wise change of T or p. We propose such experiments to check the validity of fundamental hypotheses on viscoelasticity in polymersolutions.
Considerable theoretical efforts have been made on the form of
dynamic equations
in semidi- lutepolymer
solutions in connection with shear-inducedphase separation [1-5].
It is now estab- lished that there is adynamic coupling
between stress and diffusion in suchentangled polymer
systemsincluding polymer
blends [5]. Thatis,
imbalance in the networkstress$i
can induce relative motion between
polymer
andsolvent,
v~ v~ =
(1 ~)(-i(-v«
+ v.m)
,
(1)
where vp and v~ are the
polymer velocity
and the solventvelocity, #
is thepolymer
volume fraction assumed to be in the semidiluteregion,
and r is the osmotic pressure. The(
is the friction constant betweenpolymer
and solvent and is of order~~fp~,
where ~~ is the solventviscosity
andfb
is the blob size. The first term (ccVr)
in(I)
is the usual term in fluidbinary
mixtures withoutentanglements
while the second term (cc V.$i)
haslong
been known ingels
with permanentcrosslinkages
[6]. We areassuming
that#
isconsiderably larger
than theoverlapping
volume fraction#*
andentanglement
effects are crucial indynamics,
so that the solutionviscosity
~ is muchlarger
than j and thedynamic coupling
can beimportant.
In shear flow a relative
velocity proportional
toI(3~/3#)V#
arises in(I)
from V.$i
due tothe strong concentration
dependence
of ~,i being
the shear rate. Itgives
rise to dramatic enhancement of the concentration fluctuations in theta solvent under shear[7-9].
On the otherhand,
thiscoupling
canexplain nonexponential decay
of the time correlation functionS(q, t)
of the concentration fluctuations aroundequilibrium is,
10]. It can besignificant
in theta solvent and wasoriginally predicted
in reference [11].1300 JOURNAL DE PHYSIQUE II N°9
We then expect that the above
dynamic coupling
shouldprofoundly
influencephase
separa-tion processes in
entangled polymer
systems. Toinvestigate
this aspect,however,
we further-more need to
specify
the interfacialboundary
condition in addition to the bulk relation(I).
To examine nucleation inentangled polymer
solutions the present author hasrecently proposed
a modified Gibbs-Thomson relation at the interface[12],
r-x~x-an+'t~=0
,
(2)
where r~x (Gt
0)
is the osmotic pressure on the coexistence curve, 't is the surfacetension,
and~ is the curvature of the interface. The an is the normal component of the network stress at
the
interface,
an=n.$i.n
,
(3)
where n is the normal unit vector. The normal stress in
(2)
serves to decelerate thegrowth
rate of adroplet consisting mostly
of solvent when its radius R is smaller than along
viscoelasticlength fve
definedby
fve
"(((~~il)~/~
~fb(illils)~/~ (4)
This
length
is also of order(Dor)~/~,
where Do is the collective diffusion constant(see (lo) below)
and r is therheological
relaxation time. It has been introduced in various contexts [4, 5,13].
For R «fve,
R-R~changes exponentially
with a smallgrowth
rate r=
~t/2~Rc
+~
r~~A,
where ~t is the surface tension, Rc is the critical
radius,
and A is the supersaturation smaller than I. For R »fve,
Robeys
the usualLifshitz-Slyozov equation.
On the otherhand,
viscoelasticity
inspinodal decomposition
is verycomplicated
and has not yet been well studied.Instead,
the aim of this letter is to examine theimplications
of(I)
and(2)
in arelatively simple situation,
in which a semidiluteregion (#
>#*)
and anearly
pure solventregion (#
lit0)
areseparated macroscopically by
aplanar
interface. Such a state will appear as afinal state after
quenching
the fluid into metastable or unstableregions.
We thenproduce
a smallhomogeneous change
in the temperature T or the pressure p and follow the motion of the interface position x =x;(t)
and the evolution of the concentration deviationb#(x,t)
in the semidiluteregion
x <x;(t).
Weneglect inhomogeneities
of T and p. Note that a pressurequench
method hasusually
been used inphase separation experiments
of near-criticalbinary
mixtures
[14],
in which T and p arechanged adiabatically
tobring
the fluids into metastable or unstableregions.
After such achange
in p and T the interface moves as a result ofabsorption
or
desorption
of solvent across the interface because thepolymer density
is so small in the diluteregion
x >x;(t).
This process isnearly
the same as inswelling
orshrinking
ofgels
immersed in solvent [15]. It is very instructive to compare the process with those in fluids and
gels.
In addition we comment that the excluded volume interaction is weak and the chains arenearly
Gaussian in the temperatureregion
under consideration.We assume that all the deviations
depend
on spaceonly
in the x direction which is takento be
perpendicular
to the interface. Forsimplicity
we assume that the total massdensity
iS a constant and the average
velocity
v =#vp
+(I #)v~
vanishes. It follows that v~=
-(1 #)~~ #vp.
We introduce the dimensionlessdensity
deviationo(x,t)
=
ii(x,t)/i
,
(s)
which
obeys
to linear order in the deviations=
-t~p (6)
where
#
= 30
lat.
The network stress is related to thegradient
of the networkvelocity
vp [3, 5],ai>
(x, t)
=
/~ dsG(t s) (£vp>(x, s)
+£vpi(x, s) (iii
V VP(x, )j
,
(7)
-m I J
where
G(t)
is the stress relaxation function and its one-sided Fourier(Laplace)
transformation is thecomplex viscosity,
~
~*(uJ) =
/ dte~'"~G(t) (8)
o
The relaxation time
r(+~ ~/G(0))
ofG(t)
is verylong
due toentanglement
effects. It ishighly
nontrivial
that$i
is determinedby Vvp,
but thisassumption just
leads to thegel-like
behavior athigh frequencies. Using (1, 6, 7)
and up; =bmvp,
up may beexpressed
in terms of0,
and(6)
becomes a diffusionequation
in theregion
x <z,(t),
fi2
R(X>
t)
" DO~QIX> t) (9)
with
Do
=(i -1)2(-iK
a(-iK
,
(io) Q(x, t)
=
o(x, t)
+(K-1 /~ dsG(t s)o(x, s)
,
(ii)
where K
=
#(3r/3#)Tp
is the osmotic bulk modulus. The Do is the diffusion constant in the absence ofviscoelasticity,
whileDg
=Do (1+ ~~1/K)
is the diffusion constant ingels
[6],~1=
G(0) being
the shear modulus.Next we set up the
boundary
conditions at x =x,(t).
Note that the deviation of r r~x in the semidiluteregion
consists of theinhomogeneous
part K0 and thehomogeneous
part(br)o
G~(3r/3T)pjbT
+(fir /3p)Tjbp.
In experiments, bT andbp
are the control parameters.In
particular,
ifbp
ischanged adiabatically
as in near-critical fluids[14],
we have(br)o
"
(3r/3p)~jbp.
It is convenient to introduceRo(t)
=-(br)o/K.
Then the modified Gibbs- Thomson relation(2)
becomesQ(x;(t), t)
=
00(t) (12)
The curvature term vanishes in our one dimensional case. The interface
velocity
I; =8zi13t
is related to the diffusion current at z
= z; from the conservation of
polymer
mass, so that it =-DOQ/(xi(t),t)
,
(13)
where
Q~(x,t)
=
3Q(z,t) lax.
Here we stress our two basic
assumptions:
I) We areassuming
that the interface is infinites-imally
thin and that(12)
holdsinstantaneously
after achange
of(br)o.
This isjustified
whenwe are interested in space and time variations much slower than the interface thickness (+~ the blob size
fb)
and the characteristic diffusion timef(/Do
within a blob. On time scaleslonger
than
f(/Do
the concentrationprofile
at the interface is almostequal
to that inequilibrium
determinedby
00(t)
and there remainsslowly varying,
concentration variations far from the in- terface(the
distancebeing
muchlarger
thanfb). ii)
If00(t)
> 0, the fluid becomes metastable and00(t)
coincides with the so-calledsupersaturation
[12]. For00(t)
<o,
asharp
interface still exists aslong
as T <T~(p),T~(p) being
the critical temperaturedependent
on p. Weassume that
00(t)
issufficiently
small so that there appears noappreciable
nucleation in the1302 JOURNAL DE
PHYSIQUE
II N°9fluid within the observation time. Under this condition the
following
relations are valid both for positive andnegative
00.However,
forlarge 00,
resultantphenomena
are very differentdepending
on thesign
of00.
Now
(9-13)
constitute a Stefanproblem
with amoving boundary
and their solutions are gen-erally complicated.
In this letter we assume00(t) («
I and solve theequations
inexpansions
with respect to00(t).
First let us consider anoscillatory
case,00(t)
= 00
cos(uJt) (14)
To
leading
order in00, (9)
may be solvedby setting z;(t)
= 0 to
give
°(z, t)
=
°oRe l~~ exP(~z
+
uJt)j
,
(is)
~ ~
3~°~~~~°~~
where
Re[.
denotestaking
the real part, ~*(uJ) is thecomplex viscosity
definedby (8),
and~~
=
iuJDp~K / K
+
~iuJ~*(uJ)
3
% iuJ
/
[Do +iuJf(~~*(uJ) /~]
,
(16)
where
fve
is definedby (4).
We define ~ such that its real part ispositive.
Thelength 1/
~is the thickness of the
perturbed region.
The interface value0(z,, t)
%0(0, t)
tends to(14)
in the lowfrequency
case uJr < I,
while it behaves in a
complicated
manner foruJr Z I. Here
r is the
rheological
relaxation time. Insertion of(15)
into(13) yields
4 1/2
zj(t)
=
-00Re
e'~~DoKliuJ
K +-iuJ~*(uJ)
,
(17)
3
~
which is valid to first order in 00. The behavior of
z,(t)
is alsosingular
for uJr iS I. IfuJ <
K/~
+~
l/r,
the network canfully reorganize
itselfagainst
theperturbation
and the viscoelastic effect becomes weak. The reverse holds for uJ iS IIT.
Secondly
wechange 00(t)
in astep-wise
manner,~°~~~
loo
"
~onst. ~~ ~
~~~~In this case the
equations
can be solvedusing
theLaplace
transformation with respect to time.If
G(t)
is assumed to relaxexponentially [10,
11] asG(t)
=/1e~~l'
,
(19)
The interface value
0(z;,t) approaches
to 00exponentially
as0(z,, t) loo
" I I +)
exp
-t/r
1+~) (20)
l~ 3
~
The interface
position z,(t)
grows asz;(t)
Gt(4Dt/r)~/~00
for t » r andz,(t)
Gt[4Dt/r(1+
4p/3K)]~/~00
for t « r. Inrheological experiments, however,
it is known thatG(t) decays algebraically
as t~~+P for t < r or~*(uJ) Gt
A(iuJ)~P
for uJr > 1,
(21)
where
p
G£ 0.8(as
the so-called Cox-Merz lawsuggests)
[7,16].
This leads to behavior in the short timeregion
t 1~ r of theforms,
0(z,, t) loo
Gt(3K/4r(2 p)A]t~~P
,
(22) z,(t)
Gt-00(3D/4A)~/~r(2 )~~t~~P/~
,
(23)
where A is the coefficient in
(21)
andr(z)
is the Gamma function. Here we have assumed that(21)
holds even for uJ - cc, so that it follows that0(z;,t)
- 0 as t - 0.However,
iuJ~* (uJ) can tend to a well-defined limit pm for uJ greater than a crossoverfrequency
uJo. Then0(z,,t) loo instantaneously
grows up toK/(K
+)~lm)
in a short time of order uJp~ In thesingle
relaxationapproximation (19)
we find ~lm= ~t and uJo
+~ I
IT.
In fact(20)
shows that0(z,, t) loo
-K/(K
+p)
as t - 0. In usual fluids itapproaches
to I in amicroscopic
time far fromcriticality
and in the order parameter relaxation time nearcriticality.
In summary, we have examined the viscoelastic behavior of the interface motion on the basis of the three fundamental
hypotheses (1, 2, 7),
which are those on thedynamic coupling,
theboundary condition,
and the network stress. Futureexperiments especially
in theoscillatory
case should be very informative to
unambiguously
confirm the theories. We also propose thesame kind of
experiments
inpolymer
blends [5], which will be very informative because the counterpart of(2)
remains unknown in blends.References
Ill
Helfand E., Fredrickson G-H-, Phys. Rev. Lett. 62(1989)
2468.[2] Onuki A., J. Phys. Soc. Jpn. 59
(1990)
3423.[3] Doi M., Dynamics and Patterns in Complex Fluids, A. Onuki, K. Kawasaki Eds.
(Springer, 1990)
p.100.[4] MiIner S-T-, Phys. Rev. Lett. 66
(1991)
1477; preprint.[5] Doi M., Onuki A., J. Phys. II trance 2
(1992)
1631.[6] Tanaka T., Hocker L-O-, Benedik G-B-, J. Chem. Phys. 59
(1973)
5151.[7] Wu X. -L., Pine D-J-, Dixon P-K-, Phys. Rev. Lett. 66
(1991)
2408.[8] Hashimoto T., Fujioka K., J. Phys. Soc. Jpn. 60
(1991)
356.[9] Yanase H., Moldenaers P., Mewis J., Avetz V., van Egmond J-W-, Fuller G-G-, Rheol. Acta 30
(1991)
89.[lo] Dixon P-K-, Pine D-J-, Wu X.-L., Phys. Rev. Lett. 68
(1992)
2239.[ll]
Brochard F., de Gennes P-G-, Macromolecules lo(1977)
l157.[12] Onuki A., J. Phys. II France 2
(1992)
1505.[13] Brochard F., de Gennes P-G-, Physicochem. Hydrodyn. 4
(1983)
313.[14] Wong N-C-, Knobler C-K-, J. Chem. Phys. 69
(1978)
725;Joshua M.,
Goldburg
W-I-,Phys.
Rev. A 31(1985)
3857.[15] Tanaka T., Filmore D.J., J. Chew. Phys. 70
(1979)
1214.[16] Bird R-B-,