You are on page 1of 8

Thermodynamics of polymer solutions

M. Muthukumar

Citation: The Journal of Chemical Physics 85, 4722 (1986); doi: 10.1063/1.451748
View online: http://dx.doi.org/10.1063/1.451748
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/85/8?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in


Statistical thermodynamics of polymer solutions
J. Chem. Phys. 119, 3996 (2003); 10.1063/1.1592497

Thermodynamics of Heterogeneous Polymers and Their Solutions


J. Chem. Phys. 12, 425 (1944); 10.1063/1.1723887

Thermodynamics of Heterogeneous Polymer Solutions


J. Chem. Phys. 12, 114 (1944); 10.1063/1.1723916

Thermodynamics of High Polymer Solutions


J. Chem. Phys. 10, 51 (1942); 10.1063/1.1723621

Thermodynamics of High Polymer Solutions


J. Chem. Phys. 9, 660 (1941); 10.1063/1.1750971

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
142.157.212.201 On: Mon, 24 Nov 2014 21:41:47
Thermodynamics of polymer solutions
M. Muthukumar
Polymer Science and Engineering Department. University ofMassachusetts. Amherst. Massachusetts 01003
(Received 22 May 1986; accepted 8 July 1986)
The fre~ energy ~f a polmy~r solution is derived by a consideration of the monomer density
fluct~atlons and mcorporat~ng three-body interactions. Explicit interpolation formulas are
obtamed for the conc.entratl~n dep~n~ence of the correlation length for arbitrary strengths of
~wo- an~ three-~ody mteractl~ns wlthm the random phase approximation. When the ternary
mter~ctlons are I~portant, as IS the case under the conditions of phase separation in polymer
solutIOns, the denved free energy leads to new corresponding-states equations for the
spinodals. The critical volume fraction ,pc, and l,p -,p I/,p are found to be proportional to
n -113 a~d n 119,respect'Ive1y, were
h n IS
. t h e degree of polymerization
c c
of the polymer and ,p is
the c~xlstent polymer volume fraction. A comparison is made between the predictions and the
expenmental results reported in the literature.

I. INTRODUCTION where,pc is the critical polymer volume fraction and n is the


The classical Flory-Huggins theory has been extensive- number of Kuhn segments in one polymer chain.
ly employed to understand almost all experimental results E = IT - Tc IITc' where T is the temperature and T is the
on the thermodynamical properties of polymer-solvent, critical temperature. c
polymer-polymer, and other polymer solutions. The success The recent experimental data of Dobashi et al. \3 and
of the Flory-Huggins theory in terms of the qualitative pre- Shinozaki et al. 14 show marked deviations from the predict-
dictions, and quantitative predictions in certain cases, of a ed results ofEq. (1.1). Based on these high-precision coexis-
variety of phenomena is tremendous. 1- 16 Nevertheless, the tence curve data for solutions of cyclohexane and polysty-
mean field theory clearly fails under the conditions where rene with different molecular weights, Sanchez '5 has
the density fluctuations are dominant such as dilute and se- empirically obtained the following universal scaling law:
midilute solutions. It is well known, e.g., that the Flory- l,p -,pc I _nO.102±O.OO2",P
Huggins theory predicts erroneous molecular weight depen- ,pc c .
( 1.2)
dencies for the second virial coefficient 1 and the critical con-
centration for the phase separation in polymer solu- The experimental value of the critical exponent
tions. 13- 16 Furthermore, the density correlations are such {3( = 0.327 ± 0.002) for these polymer solutions is the same
that the failure to account for these correlations leads to the as that for simple binary mixtures '7 and is in agreement with
experimentally observed apparent "composition-dependent the recent theoretical calculations2' •22 for the Ising model.
X parameter." The renormalization group calculations accounting for the
The density fluctuations in polymer solutions are of two composition fluctuations give{3 = 0.3265 ± 0.0011 which is
types. The first is the composition fluctuation which is the Obviously different from the mean field result, {3 = 112, of
same as that encountered in simple (low-molecular weight) Eq. (1.1). Since the value of {3 is the same for polymer solu-
binary mixtures. These fluctuations grow tremendously as tions and simple binary solutions, the phase separation beha-
the system approaches criticality and playa significant role viors of both solutions belong to the same universality class.
in such phenomena as the dynamics of spinodal decomposi- The difference between these solutions, however, appears
tion, etc. 16•17 The second type of density fluctuations is spe- through the n dependence of Eq. (1.2). The experimental
cific to polymers and arises from the monomer density corre- value of the n exponent, 0.102 ± 0.002 ofEq. (1.2), is signif-
lations. The manner in which two monomers belonging to icantly different from the mean field value of 1/4. This is the
the same chain are spatially correlated is described by the problem which is addressed in this paper by a consideration
monomer density correlations. 18.19 Thus the characteristic of the second kind of fluctuations resulting in the monomer
wavelength of the composition fluctuations is typically density correlations.
much larger than that of monomer density fluctuations A general theory has been recently presented by Muthu-
which in tum is larger than the Kuhn length I but shorter kumar and Edwards for polymer solution properties by
than or comparable to the size of the polymer chain. Any treating the monomer density correlations and the screening
refinement of the Flory-Huggins theory should account for of the excluded volume interaction. 23 It was demonstrated
both of these fluctuations. that the conventional field theoretical methods lead to the
For a polymer-solvent mixture the coexistent polymer scaling results and the conditions for their validity. In addi-
volume fraction ,p along either the binodal or spinOdal curve tion, interpolation formulas were presented for the free ener-
is predicted by the Flory-Huggins theory to vary near the gy of the polymer solution and other related quantities as
critical point as9-12 functions of polymer concentration and the strength of the
excluded volume interaction, both of which can range from
l,p -,pc I _n'/4 E '/2 ,p _n-1/2 (1.1 ) very low to very high. The various scaling results and a quan-
tPc 'c' titative description of the crossover regions were also de-

4722 J. Chern. Phys. 85 (8).15 October 1986 0021-9606/86/204722-07$02.10 @ 1986 American Institute of Physics

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
142.157.212.201 On: Mon, 24 Nov 2014 21:41:47
M. Muthukumar: Thermodynamics of polymer solutions 4723

rived. However, this theory leads to significant corrections AG = 1.. In <,6 + (1-<,6)ln(1-<,6)
to the mean field theory of polymer solutions only at tem- kBT n
peratures higher than the theta temperature. At theta tem-
perature, where the excluded volume interaction is taken to
+ x<,6 - ! <,62 - ~ <,63 + 1.872w3/4 <,63 (1.4)
be absent, the mean field free energy of Flory and Huggins is from which we have shown that
recovered. Furthermore, the results of Muthukumar and
( 1.5)
Edwards become inapplicable for temperatures below theta
temperature since negative values ofthe binary cluster inte- 1<,6 - <,6e 1 1/9
-----n .
gral for a pair of segments are not allowed in their formula- <,6e
tion. The corresponding-states equation for the spinodals is ob-
As originally pointed out by de Gennes, for tempera- tained from Eq. (1.4) to be
tures () and below, three-body interactions become impor-
tant. 24-27 It has also been demonstrated that higher order
<,6'3 + 11.23t/?/2r ,3/2<,6,2 + 1 = 0, (1.6)
interactions from four-body onwards lead to corrections of where
O(n- I / 2 ) to the results of calculations of three-body terms
so that these higher order terms can safely be ignored. 24,26 <,6' = <,6nl/3, r' = (T; B) n 2 9
/ ,

We present below a general theory of polymer solutions by


considering both the two- and three-body interactions for a and ¢ is the entropy parameter.
collection of many chains. Explicit formulas for the free en- After presenting the formulation of the problem in Sec.
ergy, screening length and the size of a labeled polymer chain II, AG is derived in Sec. III. The results are discussed in Sec.
are derived for arbitrary polymer concentrations and arbi- IV.
trary strengths of the two- and three-body interaction terms.
The essential results of this paper may be summarized as II. FORMULATION
follows. The free energy of mixing AG for a polymer solution Consider a polymer solution consisting of N chains each
is shown to be oflength L ( = n/) and Ns solvent molecules in a volume V.
Within the mean field treatment of Flory and Huggins, the
AG = 1.. In <,6 + (1-<,6)ln(1-<,6) +x<,6(1-<,6) free energy of mixing for such a solution AG FH is given by
kBT n
+ (w - V<,63 + (2417'5 3 )-1 AGFH =.t In <,6 + (1 - <,6 )In (1 - <,6)
kBT n
_ _9_ (!-X+w<,6)<,6 ( 1.3)
1617' a25 + x<,6(1- <,6) + (w - i)<,63, (2.1 )
where k B is Boltzmann's constant, Tis the absolute tempera- where <,6 is the volume fraction of the polymer NnP IV,
ture, <,6 is the polymer volume fraction, (! - X) and ware, (1 - <,6) is the volume fraction of the solvent, and X is the
respectively, the second and third virial coefficients, 5 is the interaction parameter. (! - X) and ware, respectively, the
concentration dependent correlation length (in units of second and third virial coefficients for the osmotic pressure
Kuhn length) for the monomer-monomer interaction, and of the polymer solution. The correction to the above mean-
a 2 is the mean square end-to-end distance expansion factor field result arising from monomer density fluctuations is giv-
ofa chain. 5 and a 2 are given, in general by Eqs. (3.9) and en by 19.23
(3.12), from which the various scaling laws are recovered
for both poor and good solutions at different concentrations -AG' = - In
kBT
f
9' [Ra].9' ({Ra }), (2.2)
in addition to a description of the crossover behaviors.
For poor solutions fiG turns out to be where the total probability distribution function is

(2.3 )

In Eq. (2.2), the function integral SpjJ [Ra ] corresponds to the sum over all configurations of the chains without any regard
to considerations of packing of polymer chains which are already accounted for in Eq. (2.1). (! - X)/ 3 is some angular
averaged binary cluster integral for a pair of segments. Similarly wi 6 is some averaged ternary cluster integral for three
segments. Analogous to the treatment of X, w is taken to be a temperature dependent parameter. Below () temperature,
(! - X) becomes negative and w is positive and sufficiently large. To avoid double counting, the contribution to Eq. (2.2)
from the zero wave vector excitation 19 will be ignored which already appears in Eq. (2.1).
The probability distribution function can be handled using the conventional field theoretical methods27.28 and the analogy
with the "<,66 field theory." However, the result and final formulas for the free energy, correlation length, etc., are very
complicated29 and need further assumptions to obtain useful formulas. In this paper we avoid the presentation of such

J. Chem.
This article is copyrighted as indicated in the article. Reuse of AIPPhys.,
contentVol.
is 85, No.8,
subject to 15
theOctober 1986
terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
142.157.212.201 On: Mon, 24 Nov 2014 21:41:47
4724 M. Muthukumar: Thermodynamics of polymer solutions

complicated equations and make the assumptions right in the beginning with the hope of making the paper readable by
experimentalists. One such assumption is to perform the random phase approximation of replacing the ternary contact term
of Eq. (2.3) by an effective two-body term,

2:2:2:
a tJ
SaL dSa SaL dStJ SaL ds y 8[Ra (Sa) -
y 0 0 0
Ry (Sy) ]8[Ry(Sy) - RtJ (StJ)]

= 2: LdSa SaL dStJ SaL ds y


atJr So 0 0 0
I 3I d k
--3
(21T)
--3
(21T)
3
d k' exp [ik· Ra (Sa) - ik'· RtJ (StJ) - i(k - k') • Ry(Sy)]

~ 2: >SaL dSa (L dStJ


a "'1 0 Jo
I I d 3\
(21T)
d 3k >xp [ik • Ra (Sa) -
(21T)
l'k' . RtJ (StJ)] (2: SaL ds y e - i(k -
r 0
k'J • Rr(Sr J )

=pi 2:2: (L dSa SaL dStJ 8[Ra (Sa) - RtJ (StJ)], (2.4)
a tJ Jo 0

where p is the monomer density, p = Nn/ V. Therefore, the probability distribution function can be written in the simple form

g; ({Ra}) = exp { - -3 2:N SaL dSa (a R a


(Sa »)2 - I 2:2: SaL dS
..!!.- a
SoL dStJ 8 [Ra (Sa) - RtJ (StJ ) ] } (2.5)
21 a = I 0 aSa 2 a tJ 0 0

with
u==! - X + w¢. (2.6)
Now the same analysis of Ref. 23 can be carried out on Eq. (2.5). Since the philosophy and the details of the calculations are
already given, only the key features are summarized here.
The potential interaction term ofEq. (2.5) describes the coupled interaction between any two chains. This can be exactly
rewritten as a single chain placed in a field due to the other chains. Thus we get

(2.7)

The second term of the curly brackets represents the interaction of the ath chain with the field and the third term is the
Lagrangian of the field. The denominator ofEq. (2.7) is the normalization factor. Therefore the chains are now decoupled
from each other owing to the introduction of the field. Since there is only a single sum over the chain label a in Eq. (2.7), we
obtain

I 9) [¢][G(¢)]Nexp[ - -hI dr¢2(r)]

I 9) [¢ ]exp [ - 2~1 I dr¢2(r) ]

N! ~ dP. exp [ -
J' 21Tl
(N + 1) lnp]
_1_ dr ¢2(r)]
2ul
I 9) [¢ ]exp [PG(¢) - I (2.8 )
9) [¢]exp [ - I
= ---------------------------------------------------
dr¢2(r)] 2~1 I
where

G(¢)=exp { - -3
21
SoL dSa
0
(a R a (sa
aSa
»)2 - i SoL dS a ¢[Ra (Sa)] } .
0
(2.9)

The second equality ofEq. (2.8) is obtained by parametrizing [G(¢) IN.


The evaluation of the integrals of Eq. (2.8) with full completeness is obviously impossible. It is therefore necessary to
propose an effective distribution function for a labeled chain. As shown below, we will choose this effective distribution
function G based on a variational principle. G describes the distribution function of a labeled ~ain with any pair of its
segments undergoing an effective interaction through the field created by all other chains. Hence G can be written as

J. Chem.
This article is copyrighted as indicated in the article. Reuse Phys.,
of AIP Vol.is85,
content No.8,to
subject 15the
October
terms 1986
at: http://scitation.aip.org/termsconditions. Downloaded to IP:
142.157.212.201 On: Mon, 24 Nov 2014 21:41:47
M. Muthukumar: Thermodynamics of polymer solutions 4725

_ I ..@[t,6]G(t,6)exp [,uG(t,6) - 2~1 I drt,62(r)]


G= ~------------------------~- (2.10)
I..@ [t,6 ]exp [,uG(t,6) - 2~1 I dr t,62(r) ]

= exp {- l.- ( ds (aR(S»)2 _ l.. ( ds fL ds' Il [R(s) - R(S')]}' (2.11 )


21 Jo 2 as Jo Jo
where the effective interaction Il is unknown and will be determined in the next section. Equation (2.8) is rearranged by
adding and subtracting a factor of ,uG so that

N J.. d,u exp [ - (N + 1 )In,u +,uG] I..@ [t,6]exp {,u[ G(t,6) - G] - _1_ I drt,62(r)}
j 2~i 2ul
&,({Ra }) = (2.12)
I PP [t,6 ]exp [ - 2~1 I dr t,62(r) ]

I
The term [G(t,6) - G] contains the fluctuations of the distri- The knowledge of Ilk therefore gives the contributions of the
bution function of a chain about its effective chain distribu- monomer density fluctuations to the free energy through
tion and the f>d,u of exp(,uG) gives the contribution of N Eqs. (2.2), (2.19), and (2.16).
effective chains to the free energy. Equation (2.11) is exact
as it stands but approximations will be made below in mak- III. CALCULATIONS
ing a choice for G. First we calculate the effective interaction Ilk' As dis-
Redefining ,uG(t,6) through an unknown quantity II as cussed above, any labeled chain interacts with other chains
given by through the field due to other chains. Any two space points
1
,uG(t,6) = ,uG(O) - 2ul I (2~)3
3
d k t,6kllkt,6k' (2.13)
at rand r' in the field are correlated by the interaction
Il (r - r'), so that the Lagrangian of the field can be de-
scribed by
when t,6k is the Fourier transform of t,6(r):

t,6k = I drt,6(r)exp( -ik·r). (2.14)


-! I dr dr' t,6(r)1l -I(r - r')t,6(r'), (3.1)

where
Equations (2.9) and (2.10) yield
Il(r - r') = (t,6(r)t,6(r'». (3.2)
G = exp { _ l.-fL ds (aR(S»)2
Thus any pair of segments of the labeled chain interact via
2IJo as
Il[R(s) - R(s')] where R(s) - R(s') is the separation
_ ul fL ds fL ds' I d 3k between the segments. Thus the chain statistics of the labeled
2 Jo Jo (2~)3 chain is altered from Gaussian depending on the nature of Il.
Following Edwards, 18,23,30,3 I let us assume that the labeled
x(1 + llk)-Iexp(ik· [R(s) -R(S')])}, chain can in turn be described by an effective Gaussian dis-
(2.15)
tribution with an effective step length II'
Comparison with Eq. (2.11) gives
1
----- = Ilk' (2.16)
I PP [R]exp [ - ~ SoL~: (a~;s) y] . (3.3)
1 +llk
In general, II is a very complicated function of Il and reflects
where Ilk is the Fourier transform of the effective interaction on the non-Markovian nature of the problem. Therefore we
Il(r). Once Il is known, llk then follows. Actually Ilk is the expect the effective probablity distribution function for the
average of t,6L system to be of the form
Ilk = (t,6Z>, (2.17)
where the average is taken using the distribution function for
the field. From Eqs. (2.10), (2.13), and (2.17) we obtain
-
,u[G(t,6) -G] 1
= - 2ul
I d k
3
(2~)3 (t,6k ll kt,6k -Ilkll k )

(2.18 ) (3.4 )
so that Eq. (2.12) becomes
&' ({Ra}) However the actual &' (t,6,{R a }) is given by Eq. (2.7).
3
Therefore we add and subtract Ho in the exponent of
= (G)Nexp{l..f d k [llkllk -In(1+llk)]}' &' (t,6{R a }) appearing in Eq. (2.7) and then perform a per-
2 (2~)3 ul turbation theory using the term ofEq. (3.4) as the bare pro-
(2.19) pagator. Thus we get from Eq. (2.7),

J. Chem.
This article is copyrighted as indicated in the article. Reuse of AIPPhys., Vol.is85,
content No.8,to15the
subject October 1986
terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
142.157.212.201 On: Mon, 24 Nov 2014 21:41:47
4726 M. Muthukumar: Thermodynamics of polymer solutions

9 (t,6,{Ra }) =exp( -Ho-H'), (3.5)


where

+ .Jii (I +
K
2.) _ .Jii (1 _.i. + ~)
~ 2 ~ K4
K

+ i f d3k3 t,6k exp(lk' Ra (Sa »)] X exp(~)erfc(K) ]. ( 3.12)


(21T)
where
J..f d k (J.. -Ak-l)t,6~.
3
+ (3.6)
2 (21T)3 lu
(3.13)
By definition, A is given by
and erfc is an error function.32 The numerical prefactor in
Eq. (3.12) is chosen to give the modified Flory resule 3 in
2 f'@[Ra].@[t,6]t,6~9(t,6,{Ra}) infinitely dilute solutions (K = 1). Some other prefactor
Ak = (t,6k) = . (3.7) may alternatively be chosen. For semidilute solutions,
t,6"i?, t,6*, with n~oo, Eq. (3.12) leads to
f.@ [Ra ].@ [t,6] 9 (t,6,{Ra })
a5 - a 3 = 13.245(! - X + wt,6)S /1Ta. (3.14)
The result of Eq. (3.14) follows also from the variational
calculation on Eq. (2.15) as described in Sec. IV of Ref. 23.
By expanding exp ( - H ') in Eq. (3.7), a perturbation the-
Use ofEqs. (3.8) and (2.16) in Eq. (2.19) gives
ory is constructed. The leading term of this perturbation
series is Ak . By making the contributions from all terms of 9({Ra}) = (G)Nexp[ - (241TS3)-I], (3.15)
the perturbation series except the leading term to vanish, Ak where S is given by Eqs. (3.19) and(3.14). The free energy
is determined. The calculational details of this procedure are of the effective chain is obtained by maximizing
presented in Sec. III of Ref. 23. The only difference between
the present calculation and that of Ref. 23 is that the factor
I(! - X + wt,6) replaces w of Ref. 23. In the limit of k ap-
In f .@[R]G

proaching zero (i.e., very large characteristic length scales), with respect to II' where Gis given by Eq. (2.11). See Ref. 23
Ak is given by for details. Substitution ofEq. (3.15) into Eq. (2.2) yields
2
A - (l/2-X+wt,6) lk2s f2 (3.8) AG' = (241Ts3)-1 _ _9_ (! - X + wtP)t,6. (3.16)
k- l+ k2 2f2 s kB T 161T a2s
Thus the effective interaction is of the familiar screened Thus the complete expression for the free energy of mixing
form, where S is the screening length in units of Kuhn length, for a polymer solution is from Eqs. (2.1) and (3.16),

S-2= 6(l/2-X+wt,6)t,6
[a 2 + 27/81T(l/2 - X + wt,6)sa- 2 ]
,(3.9) AG = .t ln t,6+ (l-t,6)ln(l-t,6)
kBT n
+ xt,6(l - t,6) + (w - i)t,63
( 3.10)
+ (241TS3)-1 _ 9 C! - X + wt,6)t,6 ( 3.17)
As discussed in Ref. 23, Eq. (3.9) is correct only in the limit 161T a2S
of S and II becoming wave vector independent and t,6 greater
IV. RESULTS
than the overlap concentration t,6*.
II and hence a are obtained by calculating the mean Before we discuss the phase diagram represented by Eq.
square end-to-end distance of the labeled chain (3.17) we obtain the limiting forms of the correlation length
S. For t,6>t,6*, but still low enough, Eq. (3.9) gives the
asymptotic formula

(R2)=
f __________________________
~
.@[Ra ].@[t,6][R(L) -R(0)f9(t,6,{Ra })
~
f:" -1 _
- -161T
- a2,,-
'1" (4.1 )

f .@ [Ra ].@ [t,6] 9 (t,6,{Ra })


9
Under the same conditions, a 2 is obtained from Eq. (3.14)
which in combination with Eq. (4.1) yields
(3.11 )
a8 - a 6 = 0.755(! - X + wt,6)/t,6. (4.2)
If three-body interactions are unimportant, w = 0 or
As before, a perturbation theory is constructed by expanding wt,6<! - X, it follows from Eq. (4.2) that the asymptotic
exp( - H') in Eq. (3.11). The leading term is LII and II is result for a 2 is
determined by making all other terms of the perturbation
a 2 = 0.932(! - X) 1/4t,6-1/4. (4.3)
theory vanish. As shown in Ref. 23, the equation which de-
termines I I is Insertion ofEq. (4.3) into Eq. (4.1) gives

J. Chem.
This article is copyrighted as indicated in the article. Reuse of AIPPhys., Vol.is85,
content No.8,to15
subject October
the 1986
terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
142.157.212.201 On: Mon, 24 Nov 2014 21:41:47
M. Muthukumar: Thermodynamics of polymer solutions 4727

S-I = 5.206(! - X)1/4t/J-3/4. (4.4) S_t/J-I to the regime of S_t/J-3/4 as the temperature is in-
creased can readily be obtained.
SubstitutingEqs. (4.3) and (4.4) intoEq. (3.17) we get for
Since the phase separation in polymer solutions takes
w=O,
place under semidilute conditions and at temperatures be-
AG = 1. In t/J + (1 - t/J)ln(1 - t/J) + Xt/J - ~ t/J2 low 0, Eq. (3.17) should be used to obtain the phase dia-
kBT n gram. Due to the interdependence of S and a through Eqs.
+ 1.872(! - X)3/4t/J9/ 4. (4.5) (3.13) and (3.14), the phase diagram can be derived only
numerically. However, we consider the limit of dominance
The asymptotic result for the osmotic pressure n of the semi- by three-body interactions over the two-body interactions so
dilute polymer solution follows from that results can be obtained analytically. In addition, this
n=t/J2~(AGIt/J) (4.6)
regime is the most relevant one in the context of phase sepa-
rations in polymer-solvent systems. The spinodal line is giv-
at/J
to be en by

~ = 1. + 2.34tfJ9/ 4. (4.7)
a
2
(AG)
at/J2
=k T[ (1-t/J)
B
1 + _1_
nt/J
kBT n
Thus n is proportional to t/J9/ 4 for t/J > t/J*. The results ofEqs. -1+ (11.23w 3/4 -1)t/J] =0. (4.12)
(4.3), (4.4), and (4.7) are the familiar scaling laws 20 for the
concentration dependence of the mean square end-to-end The critical points are obtained from Eq. (4.12) and
distance expansion factor, the correlation length, and the
osmotic pressure, respectively, for a semidilute good solu-
a (AG)
3

at/J3
-k
-
T[ (1 - 1t/J )
B 2
__1_
nt/J2
tion. The corrections to these asymptotic results are ob-
tained by using the full Eqs. (3.9) and (3.14).
If three-body interactions dominate, wt/J>! - x' Eq.
+ (11.23w 3/4 - 1)] = O. (4.13)
( 4.2) shows that a is independent of concentration and it Therefore, the critical concentration t/Je is given by
therefore follows from Eq. (4.1) that
~ -I 1617' t/J; - ~ t/J~ + .±. t/Je - ~ = 0 (4.14 )
~
_
- --alf',
2""
(4.8) n n n
9 so that
where a 2 is given by
as - a 6 = 0.755w. (4.9) t/Je = (~r/3 + O(n- 2/3 ). (4.15 )

Thus at the 0 temperature X = 112 and three-body interac- It is to be noted that this is in sharp contrast to the predic-
tions are dominant, the correlation length is inversely pro- tions of the Flory-Huggins theory, viz. t/Je _n- 1/2 . Substi-
portional to the polymer concentration as given by Eq. tuting the result of Eq. (4.15) into Eq. (4.13) gives
(4.5). For polymer solutions at very high polymer concen-
We Q:0.068t/J!/3
trations, the same scaling behavior S- t/J - I is expected. Only
whenw = 0, and! - X>O,s -t/J-1/2 at highconcentrations, Q:0.093n- 4/9 [ I + O(n- 1/3 )]. ( 4.16)
which is the Edwards regime. 19.23
The critical value of the third virial coefficient We is related
gime. 19.23
to the critical temperature Te. Since the temperature depen-
Substituting Eqs. (4.8) and (4.9) into Eq. (3.17) we dence of w and X is not known a priori, we assume that the
obtain third virial coefficient can approximately be taken 33 to be
proportional to the square of the second virial coefficient at
temperatures away from the 0 temperature
(4.10) w~(~ - X)2, (4.17)
Therefore the osmotic pressure of a semidilute polymer solu- where the numerical prefactor has been taken to be unity.
tion where three-bodY interactions dominate over the two- Since Te is related l •12 to Xc as in
body interactions as in poor solutions is o
2v
lI.e
=T'
- ( 4.18)
e
(4.11)
the relation between We and Te is
Hence, n is proportional to t/J3 for concentrations above the
overlap concentration. The laws of Eqs. (4.8) and (4.11) e
1 ( 1 -0- =(,1_)2
W =-
4 T 'P'C'
)2 (4.19)
e
valid for semidilute poor solutions are also well known. 20
Equations (3.17), (3.13), and (3.14) provide the interpola- where", is the "entropic parameter" of Flory. 1.33 Alterna-
tion formulas for AG, S, and a for arbitrary strengths of two- tively 9.33 7'e can be taken to be (Te - 0)10. By expanding
and three-body interactions and for arbitrary concentra- the (1 - t/J) -I term ofEq. (4.12), since t/J is small near t/Jc.
tions. For example, the crossover from the scaling regime of for sufficiently large n, the equation of the spinodals is

J. Chem.
This article is copyrighted as indicated in the article. Reuse Phys
of AIP .• Vol.is85.
content No.8.to15the
subject October 1986
terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
142.157.212.201 On: Mon, 24 Nov 2014 21:41:47
4728 M. Muthukumar: Thermodynamics of polymer solutions

t/J3 + 11.23w/4 t/J2 + l.. = O. (4.20) In general, both the two- and three-body terms should
n be treated together to obtain the phase diagram by utilizing
In view of Eqs. (4.17) and (4.19) and following the usual Eqs. (3.17), (3.9), and (3.12) in comparing with theexperi-
simple approximation l2•19,33 for the temperature depen- mental data of a particular system. The temperature depen-
dence of X, viz., dence of the binary and ternary interaction terms should also
be better understood. Nevertheless the present treatment
()
2v=- (4.21) demonstrates that the density fluctuations lead to significant
/I. T'
corrections to the mean field predictions of the molecular
Eq. (4.20) becomes independent of n when the change of weight dependencies of the critical properties of polymer
variables t/J' = t/Jnl/3 and 1" = m 2/9are introduced: solutions and the new results derived here are in better agree-
t/J'3 + 11.23 + ¢J3/ 21',3/2t/J,2 + 1 = O. (4.22) ment with the experimental data.
This is the new corresponding-states equation for the spino- ACKNOWLEDGMENTS
dals emerging from the incorporation of monomer density
fluctuations. This equation is to be contrasted with that of The author thanks the National Science Foundation
the Flory-Huggins theory, (Grant No. DMR-8420962) for support of this research and
Dr. I. C. Sanchez for sending a preprint of Ref. 15.
t/J"2 + 2f/n'" t/J" + 1 = 0, (4.23)
where t/J" = t/Jnl/2 and 1'" = ml12. It is to be noted that the
form of Eq. (4.22) is a reflection of Eq. (4.17). Different
Ip. J. Flory, Principles of Polymer Chemistry (Cornell University, Ithaca,
relations between w and X will lead to different correspond- 1953).
ing-states equations. 2H. Tompa, Polymer Solutions (Butterworths, London, 1956).
Using a Landau-type expansion of aG ofEq. (4.lO), the 3p. J. Flory, J. Chem. Phys. 10, 51 (1942).
4M. Huggins, J. Phys. Chem. 46, 151 (1942); Ann. N. Y. Acad. Sci. 41, 1
coexistent polymer volume fraction near the critical point (1942); J. Am. Chem. Soc. 64, 1712 (1942).
can be obtained as sl. C. Sanchez, in Polymer Compatibility and Incompatibility: Principles
and Practices, edited by K. Solc (Harwood, Cooper Station, 1982).
It/J -t/Jcl = [6(aG)Hr(Tc - T) ]112, (4.24) 61. C. Sanchez, Annu. Rev. Mater. Sci. 13, 387 (1983).
(aG)HH 7R. Koningsveld, L. A. Kleintjens, and A. R. Schultz, J. Polym. Sci. Part
A-28, 1261 (1970).
where the subscripts on aG denote partial derivatives which 8R. Koningsveld and L. A. Kleintjens, Macromolecules 4,637 (1971).
are evaluated at the critical point. Substitution ofEq. (4.lO) 9J. Dayantis, Macromolecules 15, 1107 (1982).
into Eq. (4.24) and combining the result with Eqs. (4.13), 1<lM. Daoud and G. Jannink, J. Phys. (Paris) 37,973 (1976).
11M. Daoud, J. Polym. Sci. Polym. Symp. 61, 305 (1977).
(4.15), and (4.19) yields 121. C. Sanchez, Macromolecules 17, 967 (1984).
1t/J-t/Jcl-n-2/9. (4.25) I~. Dobashi, M. Nakata, and M. Kaneko, J. Chem. Phys. 72, 6185, 6692
(1980); 80, 948 (1984).
Therefore, we obtain from Eqs. (4.15) and (4.25), 14K. Shinozaki, T. van Tan, Y. Saito, and T. Nose, Polymer 23,728 (1982).
lSI. C. Sanchez, J. Appl. Phys. 58, 2871 (1985).
1t/J-t/Jcl_ nl/9. (4.26) 16y. Izumi and Y. Miyake, J. Chem. Phys. 81,1501 (1984).
17A. Kumar, H. R. Krishnamurthy, and E. S. R. Gopal, Phys. Rep. 98, 57
t/Jc (1983).
Our exponent 119 is different from 114 of the Flory-Hug- 18L. D. Landau and E. M. Lifshitz, Statistical Physics (Pergamon, New
gins theory and is in good agreement with the experimental York, 1959).
values which range from 0.06 to 0.17. 19S. F. Edwards, Proc. Phys. Soc. London 85,613 (1965); 88, 265 (1966).
20p. G. de Gennes, Scaling Concepts in Polymer Physics (Cornell Universi-
Furthermore, our calculations show a weaker molecu- ty, Ithaca, 1979).
lar weight dependence of the critical concentration than that 21J. C. Le Guillou and J. Zinn-Justin, Phys. Rev. B 21,3976 (1980).
of the Flory-Huggins theory. The experimental data of 00- 220. Z. Alpert, Phys. Rev. B 25, 4810 (1982).
23M. Muthukumar and S. F. Edwards, J. Chem. Phys. 76, 2720 (1982).
bashi et al.13 and Shinozaki et al. 14 show that t/Jc is propor-
24p. G. de Gennes, J. Phys. (Paris) 36, L55 (1975); 39, L299 (1978).
tional to n - 0.38 and n - 0.4 , respectively. Although this expo- 2SC. B. Post and B. H. Zimm, Biopolymers 18,1487 (1979).
nent is considerably lower than the Flory-Huggins value of 261. C. Sanchez, Macromolecules 12, 980 (1979).
112, it is not quite 113 as given by Eq. (4.15). However, it 27B. Duplantier, J. Phys. (Paris) 41, L409 (1980); 43, 991 (1982).
280. J. Amit, Field Theory, Renormalization Group, and Critical Phenome-
must be pointed out that our exponent of 113 is valid only
na (McGraw-Hill, New York, 1978).
asymptotically where the two-body interactions are com- 2~. Muthukumar (unpublished).
pletely ignored. As can readily be shown, by simply repeat- 3OS. F. Edwards, J. Phys. A 8, ll7l (1971).
ing the above calculation, retention of two-body terms with 31S. F. Edwards and P. Singh, J. Chem. Soc. Faraday Trans. 2 75, 1001
(1979); S. F. Edwards and E. F. Jeffers, ibid. 75, 1020 (1979).
complete neglect of three-body terms recovers the various 32M. Abramovitz and I. A. Stegun, Handbook of Mathematical Functions
exponents of the Flory-Huggins theory. In this case, the (Dover, New York, 1972).
contribution of the monomer density fluctuations changes 33H. Yamakawa, Modern Theory ofPolymer Solutions (Harper, New York,
only the prefactors. 1971).

J. Chern.
This article is copyrighted as indicated in the article. Reuse of AIPPhys., Vol.is85,
content No.8,to15the
subject October 1986
terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
142.157.212.201 On: Mon, 24 Nov 2014 21:41:47

You might also like