HAL Id: jpa-00212487
https://hal.archives-ouvertes.fr/jpa-00212487
Submitted on 1 Jan 1990
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.
Kinetics of interface formation between weakly
incompatible polymer blends
J.L. Harden
To cite this version:
Kinetics of
interface formation between
weakly
incompatible
polymer
blends
J. L. Harden
Materials
Department, College
ofEngineering, University
of California, Santa Barbara, CA 93106 U.S.A.(ReceiB’ed
on March 7, 1990,accepted
on May 11,1990)
Résumé. 2014 Nous étudions l’évolution
temporelle
de lacomposition
de l’interface entre deuxmélanges
depolymères
faiblementincompatibles
à l’aide d’une théorie dechamp
moyen. Ladépendance
en temps del’épaisseur
de l’interfaceprésente
deuxrégimes
en loi depuissance;
d’une loi de croissance initiale ent1/2,
on passe à une loi de croissance ent1/4
avant de relaxer vers leprofil d’équilibre
de l’interface.Abstract. 2014 A
mean field model of interface formation between two
weakly incompatible
polymer
blends is used toinvestigate
the timedevelopment
of interfacialcomposition.
The interface width as a function of time exhibits two power lawgrowth regimes ;
at1/2 growth
law atearly
times crosses over to at1/4 growth
law beforerelaxing
towards theequilibrium
interfacialprofile.
ClassificationPhysics
Abstracts 05.70 - 64.75 - 68.22 1. Introduction.In the last few years, there has been intense and rather successful
study
of interdiffusion ofcompatible polymer couples
bothexperimentally
[1]
andtheoretically [2].
However,
there has been somewhat less effort towards anunderstanding
of interface formation betweenincompatible polymer species.
This case is ofparticular
interest since mostpolymer couples
are immiscible over a wide range of
temperatures,
including
thosetypical
ofpolymer
processing
conditions. Thisimmiscibility
is a direct consequence of theinherently large
size ofpolymer
chains whichstrongly
suppresses theentropy
ofmixing
of molecules but has little effect on thetypically repulsive
and molecularweight independent enthalpic
interactions between unlikepolymer species [3].
When bulkincompatible polymer couples
arebrought
into contact and heated above the
glass
transition limitedinterpenetration
of unlike chainsoccurs,
leading
to an interfacialregion
of finite extent atlong
times. Therelatively large
spatial
extent of suchpolymer
interfaces and the slow kinetics of their formation make them ideal modelsystems
for thestudy
of interface formation in non-criticalbinary
mixtures. Inaddition,
the kinetics of interface formation in thesesystems
may have relevanttechnological
implications
to theengineering
ofpolymer alloys
andpolymer welding techniques.
1778
Very recently,
experimental
techniques
have beendeveloped
that allowhigh
resolution studies of interface formation betweenpartially incompatible polymer couples
[4-6].
One suchtechnique, utilizing
an ion beam method based on nuclear reactionanalysis,
allows directmeasurements of interface
composition profiles
between deuterated andprotonated
polymer
blends with veryhigh spatial
resolution[4].
Mixtures of deuterated andprotonated
polymers
are
usually
veryweakly incompatible [7], allowing
for the formation of interfacialregions
offinite
yet
verylarge
extent. Such atechnique
allows for detailedstudy
of the timedevelopment
of interface structure in thesepartially incompatible
systems.
Recents exper-imentsemploying
this method[8]
indicate that interfacedevelopment
betweenincompatible
species
isquite
unlike that of free interdiffusion betweencompatible couples
in which interface width growsasymptotically
as the square root of time[1].
A theoreticaldescription
of theequilibrium properties
of such interfaces wasproposed long
ago[9, 10].
Until now,however,
theoretical studies of thedynamics
of interface formation have focused onmoderately incompatible
mixtures,
which have rather narrowequilibrium
interfaces[ 11, 12].
This paper considers the case of interface formation between very
weakly incompatible
polymer couples.
In this case theequilibrium
interface is many radii ofgyration
in extent, andexperimental
observation of interfacegrowth
kinetics isfeasible,
allowing
theoreticalpredictions
ofdynamical scaling
laws to be tested.2. Model.
Consider two semi-infinite blocks
composed
of A and Bpolymer
mixtures in theglassy
statebrought
into contact to form aninitially sharp planar
interface. Forsimplicity,
attention willbe restricted to
incompressible
mixtures ofequal
molecularweight
polymers
(NA
=NB
=N). Subsequently,
thesystem
is heated above theglass
transition to aFig.
1. - Initial interfaceconfiguration.
Two semi-infinite blockscomposed
ofpolymer
mixtures~1
and ~2
from the bulk coexistence curve at the intendedannealing
temperatureTo (near
the criticaltemperature
To
near the criticalpoint
ofde-mixing
of the A-Bsystem,
tg
TO:5 T,.
Thecomposition
of the blocks on either side of the interface is chosen from the bulk coexistence curve for mixtures ofpolymers
A and B at the intendedannealing
temperature
To,
as shown infigure
1. Such a choice is made in order to eliminate thecomplication
of bulk convectivetransport
ofpolymer
across the interface.Upon
heating,
the interfaceundergoes
ahealing
process that smears thesharp
interface to a narrow but finiteregion
some fraction of aradius of
gyration
in extent. Thishealing
process iscomplicated by
constraintsimposed by
chain folds near the initial interface and is rather sensitive to the initialdensity
of chain ends atthe
interface,
aquantity
thatdepends strongly
on the method ofsample preparation.
Thehealing
ofsharp
interfaces is atopic
of activeinvestigation
that we will not address here[13,
14]. Rather,
theanalysis
of this paper will consider the timedevelopment
of the interfaceas it
expands
from thepre-healed
initial widthWo
towards the finalequilibrium
widthWeq,
as sketched infigure
2.Fig.
2. - Sketch of therange of interfacial
growth
to be considered. Theanalysis
considers thedevelopment
of interfacial structure from aninitially
broadenedprofile ~0(z)
of widthWo
towards the final mean fieldequilibrium
profile ~ eq (z)
of widthWeq ~
N 1/2 a (X / Xc - 1)- 1/2.
We assume that the free energy per unit area of the
system
can be describedby
the wellknown
Flory-Huggins
free energyexpression,
extended to account for slowspatial
variations inpolymer
concentration normal to the interface[15-17] :
where 0
is the volume fraction of monomers ofspecies
A, a
is the monomersize,
andf[~J(z)]
is the local free energy perkB
T asgiven by
the usualFlory-Huggins expression
[3, 15] :
1780
while the last term represents the
enthalpy
ofmixing
asgauged by
theFlory
interactionparameter
X (T).
One obtains the local chemical
potential g
(z)
of monomers ofspecies
Aby
functional differentiation of the free energyexpression given by equations (1)
and(2)
withrespect
to41
(z) : 1£ (z )
ocd F / d ~ .
This local chemicalpotential
drives thegrowth
of the interface wi thtime ;
thesystem
will redistribute the Aspecies
in order to relax any non-zerogradient
ofJ.L. The
growth
of the interface may be characterizedby
the timedependence
of the Amonomer concentration
profile
across the interface. This may be determined from aone-dimensional
continuity equation :
where
J(z, t )
is the local current ofspecies
A. Within the context of linear response,J(z, t )
may be related to thegradient
of the chemicalpotential
via a non-localtransport
coefficient[ 15] :
where
A (z - z’ )
is ageneralized
non-localOnsager
coefficient and IL(z’ )
is the local chemicalpotential
perkB
T. Thenon-locality
of A is a consequence of thelarge degree
ofconnectivity
of the monomers in a
polymer
medium. Afterintegration by
parts
and omission of thevanishing boundary
terms,equations (3)
and(4) yield
anequation
of motion for~ (z, t ) :
:Equation (5) together
withboundary
conditions, ~ = ~1
1 at z = + ooand l/J = l/J 2
atz = - 00, and a
specified
initialprofile, ~/0(z, 0) = ~o(z),
constitute awell-posed
initial-boundary
valueproblem.
Inprinciple, given u ( ~ )
andA,
we may solveequation (5)
for0 (z, t ).
Inpractice,
theresulting integro-differential equation
isstrongly
non-linear and notamenable to
simple
solution. Chemicalpotential gradients computed
with the full free energy functional asgiven by equations (1)
and(2)
are non-linear in the monomer concentrationvariables.
Furthermore,
recent theoretical worksuggests
thatOnsager
coefficients fortransport
inspatially inhomogeneous polymer
blends are often acomplicated
functional ofthe monomer concentration
profiles [12].
However,
we mayexploit
the weakincompatibility
of theinterpenetrating
bulk mixtures under consideration to make severalsimplifying approximations.
First,
theOnsager
coefficient may be estimated from calculations ofsingle
chaintransport
in a melt ofnon-interacting
Gaussian chains[16, 17].
Second,
the chemicalpotential
may be linearized around the criticalcomposition l/J c
=1/2.
It is convenient to write theresulting simplified equation
ofmotion in terms of the
gradient
of theconcentration, 0
=a ~ / az.
Assuming A (z - z’)
isindependent of 0 (z)
as discussed above andusing
the linearized chemicalpotential,
theequation
of motionfor 0 (z, t) given by equation (5)
can be recast as a linearintegro-differential
equation
of motionfor gi (z,
t ).
The solution to thisequation
must alsosatisfy
boundary
conditions, 03C8 =
0 at z = ± oo, and an initialcondition, 03C8 (z, 0 )
=dl/J
o/dz.
The newequation
of motion andboundary
conditions areeasily
solved in Fourier space. The modes ofwhere 03C8(q,
t ),
03A8o{q)
andA(q)
are thespatial
fourier transforms of03C8 (z, t ),
03C8o(z)
andA (z ), respectively,
and where xc =y (Tc)
=2 /N
within the context ofFlory-Huggins
meanfield
theory [3].
The wavevector
dependent Onsager
coefficient for monomertransport
inweakly
incompat-ible
polymer
blends may be estimated from the response of asingle
chain in a melt ofnon-interacting
Gaussianpolymers
to anapplied spatially periodic
chemicalpotential gradient
[16].
Theresulting expression
forA (q )
has the formwhere
Rg ~ N 1/2 a
is the radius ofgyration
of apolymer
coil in meltconditions,
L oc N is the
length
of the constraint tubeenclosing
atypical polymer
coil as createdby
itsentanglements
withneighboring polymer
chains,
and wherello
is an effective monomerdiffusion constant
proportional
to themicroscopic
monomermobility
go :Ao - kB Tuo.
In thesmall q
limit,
qRg 1,
Aapproaches
theq-independent
tube diffusion constant,Dt ’"
llo/N,
and chainsrespond by ordinary
diffusion. In thelarge
qlimit,
qRg »
1,
ll scalesas A (q) ~ q - 2,
reflecting
theincreasing
inhibition of chain response toperturbations
that vary
rapidly
on the scale of a chain size.Equations (6)
and(7) yield
anexplicit
form forthe interfacial relaxation
modes ;
modesof tb
decay
exponentially
with time constantTq
given by
The direct space solution to the
equation
of motion andboundary
conditions is obtainedby
inverse Fourier transformationof 03C8 (q,t ) :
with
Tq
asgiven
inequation
(8).
The initial interfacialprofile
is asyet
unspecified. Actually,
the
precise
formof CPo(z),
and hence of03C8o(q),
turns out to be of little conséquence ; any initialprofile
thatsmoothly interpolates
between bulk valuesof .0
over a distancecorresponding
to the initial extentWo
of the interface willyield qualitatively
similar results. Forconvenience,
we obtain03C8o (q ) by
choosing 00
of the form[CPo(z) - CPc] ~
tanh(z/ Wo).
3. Results.
Widths of
experimentally
determinedprofiles
are often estimated from the inverse of themaximum rate of
change
of monomer concentration across the interfacialregion ;
in terms of the difference in bulk values of monomer concentration on either side of theinterface,
A’k = 0 2
- ~ 1,
the width isapproximately given by
In the above
analysis,
thisapproximation
isequivalent
toestimating
the widthby
the1782
Although
notintegrable
in closedform,
this iseasily
evaluatednumerically by treating
time as aparameter.
Beforedoing
so, it is instructive to consider thelimiting
behavior ofW(t)
at short andlong
times. To facilitate thisanalysis,
we assume for the moment thatf/lo(q)
isapproximately
constant and that Xc - X isnegligible.
Thiscorresponds
to the idealized case of a verysharp
initial interfaceseparating
two verynearly compatible
bulkphases.
In this case, theintegral
atearly
times is sensitive to thelarge q
behavior of theintegrand
whichasymptotically approaches
the Gaussian formexp ( - x2)
withx2 ~
tq 2.
Consequently,
the interface grows asW(t) ~ t1/2
atearly
times. On the otherhand,
theintegral
at late times is dominatedby
thesmall q
behavior of theintegrand
whichasymptotically
approaches
the formexp ( - x4)
withx4 ~ tq4.
Thus,
the interface grows asW(t) -
t 1/4 at late times. Atsufficiently
latetimes, however,
thegrowth
of the interface mustslow as the
equilibrium
withWeq -
N 1/2 a (X / Xc - 1)- 1/2
isapproached ;
in the latestages
ofinterface
formation, X - X c
is nolonger negligible
and weexpect
thegrowth
toeventually
deviate from the
t 1/4 law.
The linearanalysis
breaks down at thispoint ;
in order tocorrectly
describe the final
approach
toequilibrium,
non-linear terms in the chemicalpotential
must be retained.As mentioned
earlier,
thehealing
of asharp
initial interface is acomplicated
process that isoutside the scope of this
analysis,
and hence we are limited to consideration ofpre-healed
interfaces.However,
weexpect
thegeneral
case of aninitially
diffuse interface to behave in aqualitatively
similar mannerexcept
at veryearly
times. Profilescomputed numerically
confirm thisexpectation.
Consider,
forexample,
the timedevelopment
of an interfacepre-healed to an initial width
Wo
= 0.25Rg
and with anequilibrium
widthWeq
= 10Rg.
The results of numericalintegration
ofequation (11)
in this case are shown infigure
3.Following
aperiod
of internal redistribution of material within theinitially
diffuseinterface,
interfacialgrowth rapidly
falls onto at1/2 growth
law. Growth kineticssubsequently
cross-over to at 1 /4 growth
law beforefinally relaxing
towardsequilibrium.
The cross-overregion separating
the two power law
regimes
occurs at a time t * that scales with the timerequired
for apolymer
chain to diffuse its own diameter in a melt of identical chains :
t * ~- R2g/ Ds
s whereFig.
3. -Log-log plot
of interface widthW(t)
vs. time t for the case of an interface with initial widthJ-fô
= 0.25Rg
andequilibrium
widthWeq
= 10Rg.
Width is measured in units ofRg,
while time is measured in units of t * ~R g2/Ds,
the timerequired
for self-diffusion of apolymer by
oneDs -- kB
T/N 2
is the self diffusion constant of apolymer
chain withdegree
ofpolymerization
N. For a
given
choice of bulkpolymer
mixtures this cross-over time isuniversal,
i.e.independent
of the width and detailed features of the initial interfacial structure. 4. Discussion.The novel feature of these results is the appearance of two power law
growth regimes.
Theearly
timedynamics
inparticular
isunique
to macromolecularsystems,
being
a directconsequence of the
strong
non-locality
oftransport
coefficients on submolecularlength
scales.At
early
times,
interface width is narrow on the scale of apolymer
coil,
leading
torapidly
varying
chemicalpotential
gradients
within the interfacialregion.
As discussedpreviously,
response to such
rapidly varying
chemicalpotential gradients
isstrongly
inhibited,
leading
todifferent
growth
kinetics atearly
times than at late times when the interface is broad andtransport
coefficients areeffectively
local. In contrast, theapplication
of thisanalysis
toordinary
small moleculebinary
mixtures would lead to asingle
exponent
governing growth
kinetics ;
transport
in suchsystems
is notstrongly
non-local on any reasonablelength
scale[18].
It is instructive to contrast the
physics
of interface formation between immisciblemixtures,
which we have considered in this paper, with that of misciblesystems.
Oneexpects
thedifferences in interfacial
growth
kinetics in these two cases to arise from the differences in the freeenergies
that drive these processes. In the case of immisciblesystems,
interfacialgrowth
is
entirely
drivenby composition gradients
across the interface. This manifests itself in thet1/4
growth
behavior at late times. In misciblesystems,
however,
theenthalpy of mixing
favorsgrowth
and both the bulk and interfacial free energy terms drive the process. As such aninterface
expands,
however,
composition gradients
weaken andgrowth
is drivenincreasingly
by
the bulk free energy terms. Hence in misciblesystems,
anordinary
diffusionequation
governs the
growth
of theinterface,
leading
togrowth
that scalesasymptotically
with time ast 1/2
The results
presented
here are inqualitative
agreement
with recentexperimental
data onthe kinetics of interface formation between mixtures of
protonated
and deuteratedpoly-styrenes
[8].
The measuredprofiles
are consistent with the twogrowth regime
behavior described above.However,
further refinement both ofexperimental
resolution and of the theoretical model are necessary before detailedcomparisons
can be made. Currenttheoretical work is
extending
the model toasymmetric
mixtures, NA # N B,
the situation mostrelevant to current
experiments
and most often encountered inpractice. Forthcoming
aredetailed results of interfacial
composition profiles
for mixtures ofasymmetric
chains,
as wellas a proper
analysis
of the kinetics of the finalapproach
to theequilibrium
interfacialstructure.
Acknowledgements.
This work was initiated
during
astay
with the group of Jacob Klein in thepolymer
researchdepartment
of the Weizmann Institute of Science inRehovot,
Israel. 1 amgrateful
to the US-Israel Binational Science Foundation and the USDepartment
ofEnergy
under contract # DE-FG03-87ER45288 for financialsupport,
and to thefaculty,
staff,
and students of thepolymer
researchdepartment
for theirgracious hospitality.
Inaddition,
1 would like to thank the Institute for TheoreticalPhysics
at theUniversity
ofCalifornia,
Santa Barbara for1784
kinetics. 1 also benefitted from very
stimulating
discussions with D.Andelman,
G.Fredrickson,
P.-G. deGennes,
A.Halperin,
K.Kawasaki,
S.Milner,
and P. Pincus.References
[1]
COMPOSTO R. J., KRAMER E. J. and WHITE D. M., Macromolecules 21(1988)
2580 and references therein.[2]
BROCHARD-WYART F., Molecular Conformation andDynamics
of Macromolecules in CondensedSystems,
Ed. M.Nagasawa (Elsevier, Amsterdam)
1988, pp. 249-256 and references therein; BROCHARD-WYART F., C.R. Hebd. Acad. Sci. Paris 305(1987)
657.[3]
DE GENNES P. G.,Scaling Concepts
inPolymer Physics (Cornell University
Press, Ithaca,NY)
1985, Ch. IV.
[4]
CHATURVEDI U. K., STEINER U., ZAK O., KRAUSCH G. and KLEIN J.,Phys.
Rev. Lett. 63(1989)
616 ;
CHATURVEDI U. K., STEINER U., ZAK O., KRAUSCH G., SCHATZ G. and KLEIN J.,
Appl. Phys.
Lett. 56 (1990) 1228.[5]
MILLS P. J., GREEN P. F., PALMSTROM C. J., MAYER J. W. and KRAMER E. J.,Appl. Phys.
Lett. 45(1984)
957 ;JONES R. A. L., KLEIN J. and DONALD A. M., Nature 321
(1986)
161.[6]
RUSSELL T. P., KARIM A., MANSOUR A. and FELCHER G. P., Macromolecules 21(1988)
1890.[7]
BATES F. S., WIGNALL G. D. and KÖHLER W. C.,Phys.
Rev. Lett. 55(1985)
2425 ; BATES F. S., WIGNALL G. D.,Phys.
Rev. Lett. 57 (1986) 1429.[8]
STEINER U., KRAUSCH G., SCHATZ G. and KLEIN J.,Phys.
Rev. Lett. 64(1990)
1119.[9]
HELFAND E. and TAGAMI Y., J. Chem.Phys.
56(1971)
3592.[10]
JOANNY J. F. and LEIBLER L., J.Phys.
France 39(1978)
951 ; LEIBLER L., Macromolecules 15 (1982) 1283.[11]
DE GENNES P. G., C.R. Hebd. Acad. Sci. Paris(II)
308(1989)
13.[12]
KAWASAKI K. and SEKIMOTO K.,Physica
143A(1987)
349 ;Europhys.
Lett. 4(1987)
339 ;Physica
148A