Vous êtes sur la page 1sur 103

1

Contents
1 INTRODUCTION

2 PECULIARITIES OF THERMODYNAMICS OF HETEROPOLYMERS


5
3 QUANTITATIVE
HETEROPOLYMERS

CHARACTERIZATION

OF
8

4 THERMODYNAMIC APPROACHES TO STATISTICAL


COPOLYMERS
4.1 Simplest Criterion of Polymer Blends Miscibility . . . . . . . .
4.2 Account of polydispersity and microstructure of heteropolymers
4.3 Further development of the theory . . . . . . . . . . . . . . . .

15
20
22
24

5 LARGE SCALE MEAN-FIELD THEORY


5.1 Homophase Systems . . . . . . . . . . . . .
5.2 Heterophase Systems . . . . . . . . . . . . .
5.2.1 Free energy . . . . . . . . . . . . . .
5.2.2 Phase diagram . . . . . . . . . . . .
5.3 Incompressible systems . . . . . . . . . . . .
5.4 Multiple Critical Points . . . . . . . . . . . .

.
.
.
.
.
.

27
27
31
31
34
39
43

.
.
.
.
.
.
.

48
48
48
51
52
57
60
62

7 INTERFACE PROBLEM
7.1 Central Ideas of the Self-consistent Field Approximation . . .
7.2 Heteropolymers with strong unit interactions . . . . . . . . . .
7.3 Heteropolymers with weak unit interactions . . . . . . . . . .

64
64
66
67

.
.
.
.
.
.

6 SMALL SCALE MEAN-FIELD THEORY


6.1 General Treatment . . . . . . . . . . . . . . .
6.1.1 Random Field Approach . . . . . . . .
6.1.2 Density Functional Method . . . . . .
6.2 Angular dependence of scattering intensity . .
6.3 Spinodal and the Lifshitz Point . . . . . . . .
6.4 Critical Points . . . . . . . . . . . . . . . . . .
6.5 Appearance of Spatially Inhomogeneous Phase

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

2
8 THEORY OF MICROPHASE SEPARATION
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.2 Weak Segregation Regime . . . . . . . . . . . . . . . . . . .
8.2.1 Introduction to the Landau theory . . . . . . . . . .
8.2.2 Peculiarities of the Landau theory as applied to polymer systems . . . . . . . . . . . . . . . . . . . . . . .
8.2.3 Main results . . . . . . . . . . . . . . . . . . . . . . .
8.3 Strong Segregation Regime . . . . . . . . . . . . . . . . . . .
8.3.1 The Helfand-Edwards Method . . . . . . . . . . . . .
8.3.2 Modication of the Landau theory . . . . . . . . . .
9 References

68
. 68
. 70
. 70
.
.
.
.
.

72
79
84
84
87
90

Statistical Thermodynamics
of Heteropolymers and their Blends
S.I. KUCHANOV
Keldysh Institute of Applied Mathematics
Moscow, Russia
S.V. PANYUKOV
P.N.Lebedev Physics Institute
Moscow, Russia

INTRODUCTION

Unlike homopolymers heteropolymers consist of macromolecules involving


more than one type of elementary units. Any sample of synthetic heteropolymer represents a mixture of enormous (practically innite) number of
dierent individual chemical compounds. Even at the same degree of polymerization l their molecules vary in chemical composition and structure, i.e.
in fractions of units of dierent types as well as in the fashion of their arrangement, respectively. The number of conceivable isomers exponentially
grows with l and for actual polymers essentially exceeds the Avogadro number. That is why it appears to be impossible in principle to characterize
the sample of such a polymer by setting the concentrations of individual
chemical compounds because vast majority of them are represented in this
sample by no more than the sole macromolecule. There is a fundamental difference between this situation and that of homopolymers where the number
of molecules with any l in a sample is large enough to provide its adequate
macroscopic description in terms of concentrations of homologues (l-mers).
Consequently, polydisperse homopolymers as well as their blends are possible to describe resorting to the traditional thermodynamics, although with
nominally innite number of components, which are homologues. Such a

4
consideration, in view of the argument outlined above, fails in the case of
heteropolymers so that when developing rigorous thermodynamic theory of
their solutions, alloys and blends one is supposed to elaborate nontraditional
approaches.
One more inherent peculiarity of heteropolymers, whose macromolecules
comprise rather long blocks of units of the same type, consists in their ability to undergo microphase separation when with the alteration of external
conditions (e.g. the temperature) thermodynamically stable microdomain
structures appear in the system. The morphology of these phase-separated
structures (i.e. their form, size and spatial distribution of microdomains)
depends along with thermodynamic variables (temperature, pressure) also
on molecular parameters of a sample of particular block-copolymer. The establishment of such dependencies being one of the most important problems
of statistical thermodynamics of heteropolymers calls for its correct solution
the incorporation of sophisticated methods of modern theoretical physics.
Present chapter is aimed at familiarizing the reader with the progress
of theoretical interpretation of equilibrium states of heteropolymer systems,
leaving apart the problems of the kinetics of phase separation process. These
problems as well as the ways of blend preparation along with the factors
controlling their compatibility have already been reported in preceding volumes of Comprehensive Polymer Science (volume 2, chapter 5 and volume
7, chapter 4), whereas only slight mention was made there of quantitative
theory of blends involving heteropolymers. The material of present chapter extends rather than duplicates that published earlier in CPS on polymer
blends. Special emphasis in this review will be placed on the analysis of theoretical conceptions underlying statistical thermodynamics of heteropolymers
while the comparison of theoretical conclusions to the results of both computer simulations and experimental investigation can be found in remarkable
review by Binder[1].

PECULIARITIES
OF
THERMODYNAMICS OF HETEROPOLYMERS

One of fundamental problems of thermodynamics of any system is associated


with the construction of its phase diagram. The latter, due to polydispersity

5
of real heteropolymers for their chemical length l and composition, are expected to exhibit a number of peculiarities distinguishing them from phase
diagrams of solutions of low-molecular weight compounds. These peculiarities are most readily elucidated by comparing the incompressible solutions
of monodisperse and polydisperse homopolymers whose molecules dier only
in number l of units.
On well-known phase-diagram (Figure 1a) the areas of one- and twophase solutions are separated by the binodal. It consists of two branches,
one of which is situated to the left while another to the right of the critical
point. A segment of horizontal line, joining the points of its intersection
with the binodal, is called the tie-line. Its terminal points laying on dierent
binodal branches are conjugate in the sense that at given temperature they
correspond to the compositions of coexisting phases just at the moment when
phase separation commences.
Crucially dierent situation takes place when dealing with the solution
of polydisperse homopolymer. Such a system pertains to the category of
quasibinary ones since in parallel with the solvent it contains only one quasicomponent, e.g. monomer unit. In binary heteropolymer or a mixture of two
dierent homopolymers the number of such quasicomponents will equal two
whereas in general case it coincides with the number of types of monomer
units in the sample of interest. Therefore, for example, the blend of two
styrene copolymers with methyl methacrylate and with acrylonitrile have to
be assigned to quasiternary systems. When constructing phase diagrams the
concentrations or fractions of monomer units might act as thermodynamic
variables. However, it should be borne in mind that unlike low-molecular
solvents they are not true thermodynamic components because the notion of
chemical potential, for instance, proves to be for them physically meaningless.
Looking at Figure 1 one will become aware of qualitative distinctions in
phase diagrams arising when polydispersity is accounted for. This factor is
responsible for the split of a binodal into two curves. One of them, where
the second phase is known to nucleate, is usually referred to as a cloudpoint curve. Separating one- and two-phase regions like a binodal does this
curve distinguishes, nevertheless, from it by a number of specic features.
The main of them consists in the fact that in the solvent with specied
fraction of polymer the value of this fraction in the rst droplets of
precipitated phase corresponds in Figure 1b to the point of the intersection
of horizontal straight line with dotted curve rather than with solid one. The

6
former, called the shadow curve, intersects the cloud-point curve at critical
point which divides each of these curves into two branches corresponding to
two dierent phases. These latter become identical, as in the case of binary
solution, only at critical point which, however, is not located now at the
maximum of the curve where the second phase nucleates. At this point,
termed precipitation threshold, the account of polydispersity factor results in
the alteration of the type of phase transition whose order was found to be
the rst instead of the second one. The spinodal where homophase solution
becomes thermodynamically unstable, in quasibinary system (Figure 1b) has
common tangent at the critical point with the cloud-point curve as it has such
a common tangent with binodal in a binary system (Figure 1a).
The region of the metastable states, located in between the cloud-point
curve and the spinodal, is of special signicance for polymer system as compared to low-molecular weight ones. This is conditioned by the fact that
the phase separation in the former proceeds markedly slower than in the
latter. Consequently, the life-time of the metastable states may often far
exceed the duration of a thermodynamic experiment and even the polymer
operation-time. In such cases, typical of concentrated solutions, melts and
blends of polymers, a metastable system is expected to exhibit an equilibrium
behavior.
Apart from outlined peculiarities of phase diagrams, intrinsic to any polydisperse systems, there are other peculiarities inherent to more complicated
diagrams comprising, for instance, regions where more than two phases coexist. The number of these latter, according to the Gibbs phase rule, generally speaking, can be arbitrary even for the above discussed quasibinary
system, inasmuch as the number of its true thermodynamic components is
formally innite. Setting of fractions of these components by means of the
function of molecular weight distribution (MWD) of homopolymer sample
provides an exhaustive description of its polydispersity. The dimensionality of phase diagram will depend on how many independent stoichiometric
parameters the MWD has.
If stoichiometric parameter is the sole, as for exponential the Flory distribution, the phase diagram of a compressible homopolymer melt will be
three-dimensional one. In this case the roles of spinodal, cloud-point curve
and shadow-curve will be performed by corresponding surfaces, the two rst
of which are tangent along the line of critical points. At given value of stoichiometric parameter the two-dimensional sections of these surfaces akin to

7
those, sketched in Figure 1b, are of utmost importance for the investigation
of demixing of solutions of a polymer with given MWD under variations
of thermodynamic variables (temperature, volume, pressure). Fixing of one
of them makes possible the construction of other two-dimensional sections
which permits one to predict the conditions of phase separation in the course
of polymer synthesis accompanied by the alteration of the MWD parameter.
Though this problem is of crucial practical signicance for manufacturing of
polymers it has not been properly elucidated in the literature so far.
With increasing number of stoichiometric parameters, which dene the
polydispersity of the system in hand, the increase of the dimensionality of its
phase diagram is also the case leading to the appearance of multiple critical
points. The question about the number of such points and their stability
has been thoroughly scrutinized by now only for solutions and blends of
homopolymers[2, 3, 4, 5, 6, 7, 8, 9].
Phase diagrams were found to feature a number of additional peculiarities when a polymer contains macromolecules involving rather long blocks
of units of the same type. Then, provided specic conditions are met, there
is a feasibility of the occurrence of such branches of cloud-point curve (in
general case of hypersurface of certain dimension), where inside homogeneous principal phase the droplets of incipient phase (mesophase) possessing
space periodicity will appear. A phase diagram of solution or blend involving block-copolymers with given molecular architecture may be divided into
several areas, corresponding to mesophases diering by the type of periodic
structure. As for the spinodal, the procedure of its nding for the solutions
and blends comprising block-copolymers proves to be far more tedious one
as compared to that much practiced in traditional thermodynamics. In its
terms the sucient condition of absolute stability of homogeneous phase is
its stability with respect to the uctuations of an innite space scale. This
rule can be broken in block-copolymer systems where homogeneous phase
appears to be unstable relative to the uctuations of nite scale only. Thus,
the regular approach to determining conditions of the loss of stability of
homogeneous state with respect to innite uctuations is believed, broadly
speaking, to be insucient to construct the whole spinodal. This would call
for an additional analysis of the stability of this state in reference to the
space inhomogeneous uctuations on all scales.
The construction of phase diagrams of heteropolymer systems whose general specicities have been discussed in the foregoing constitutes a problem of

8
great practical importance. However, along with this problem the thermodynamics of such systems faces a bunch of other essentially more involved ones.
Among these are the derivation of equations describing heterophase solutions
and blends with participation of copolymers. The solution of such equations
allows one to determine compositions and volume fractions of all phases in
equilibrium as well as the surface tension at the boundaries between them
for any values of thermodynamic and stoichiometric parameters of a system.
Its complete thermodynamic characterization implies also the expression to
be derived for experimentally obtainable elastic scattered intensity of light,
neutron or X-ray as a function of magnitude of scattering angle.
Here we attempted at only tracing the range of potential problems relevant to statistical thermodynamics of heteropolymer systems. The material
presented below provides an idea about the progress made in their solution.

QUANTITATIVE
CHARACTERIZATION OF HETEROPOLYMERS

In line with the arguments put forward in the preceding section traditional
specifying of actual heteropolymer sample by a set of the concentrations of
all its constituents makes no sense. Alternatively, it has been suggested[10]
to characterize such a sample by virtue of a probability measure P {}
on a set {} of all kinds of its macromolecules, i.e. stochastic sequences
= (1 , 2 , ..., l ) of monomer units M ( = 1, ..., m). The symbol ,
(where = 1, 2, ..., l ), denotes here the type of unit situated on the -th
position from the beginning of polymer chain, so that, for instance, to the
terpolymer molecule M2 M1 M2 M3 M1 M3 there corresponds the sequence of
integers (2, 1, 2, 3, , 1, 3). Each of them may be thought of as the realization of a random process of conventional movement along macromolecule
during which the transitions between the states 0, 1, 2, , , , m occur
at equal intervals of time . Upon reaching the zeroth state which corresponds to attaining the edge of macromolecule one will remain in such a
state forever. That is why as distinct from all the other ones the latter is
referred to as absorbing one.
The above stochastic process with discrete time and discrete number of
states bears in mathematics a name of a chain, among which of particular

9
importance are the Markov chains. Here the probability of the transition
at any step is specied only by its initial () and nal () states. The matrix
of transition probabilities Q with elements and the vector of initial states

v with components v equal to their probabilities provides an exhaustive


description of a Markov chain. The probability of its arbitrary realization
is dened by a simple formula
P () = v1 1 2 2 3 l 0

(1)

enabling one to nd all statistical characteristics of this chain.


The Markov chains owe their utmost importance for Polymer Science to
the fact that they characterize molecular structure of numerous commercially
used heteropolymers. Among these are the products of many processes of
free-radical copolymerization and polycondensation of an arbitrary number
of monomers. The markovian character of the sequence distribution in their
macromolecules has been established both experimentally (largely, by the
NMR-spectroscopy method) and theoretically, by means of the approaches
of statistical chemistry of polymers[10, 11, 12]. These approaches allow one

to express the elements of the matrix Q and components of vector


v through
kinetic and stoichiometric parameters of the reaction system, in other words,
to relate any statistic characteristics of actual copolymers to the conditions
of their synthesis.
A notorious particular case of the Markov chain is that, when all rows of
its matrix Q are the same. Here the probability of any transition = P
does not depend on the initial state and thus the Markov chain is reduced
to the sequence of independent trials with probabilities of a result P1 , ..., Pm .
Such a stochastic process describes random copolymers which are the simplest
ones from the viewpoint of the characterization of their chemical structure
among all statistical heteropolymers.
In the case of copolymers with sequence distribution,which can not be
described by any Markov chain, they frequently fail to construct probability
measure P () (1), i.e. to nd molecular structure distribution (MSD) oering
an exhaustive statistical description of their chemical structure. Here one has
to conne himself to determining only some of its statistical characteristics
which carry just limited (in comparison with the MSD) information on chemical structure of a polymer sample. Naturally, within the scope of present
review only those of them fall which are relevant to the thermodynamics of
heteropolymers.

10
An individual heteropolymer molecule is most readily characterized either

by the vector l with components l1 , l2 , , lm equal to the numbers of its


monomer units M1 ,M2 , ,M , ,Mm or by chemical size l = l1 +l2 + +lm

and composition vector with components = l /l( ( =


1, ..., m). It

)
is apparent that the distribution of macromolecules f l, for size and
(
)
composition (SCD) is equivalent to f l , but sometimes it appears to be
more suitable to describe those samples of high-molecular polymers where
the function f arguments may be thought of as continuous variables rather
than discrete ones. Several fashions to introduce the SCD function are known
in macromolecular chemistry. With one of them this function has a meaning
of probability to nd in a sample the molecule of given size and composition
whereas with other it corresponds to the probability for the randomly chosen
unit to constitute such a molecule. In the rst and the second case we
are dealing with the Number SCD (fN ) and with the Weight SCD (fW ),
respectively. These distributions in a simple way
(
(
(
(
)
)/
)
) /
fN l = n l
, fW l = n l l M,
(2)
(
(
)
)
where =
n l , M=
n l l

are expressible through numbers n


third type of the SCD
f

(
)

=n

(
)/

M,

(
)

(
)

of molecules with given l . The

Y,
M

lf

(
)

=1

(3)

diering from the distributions (2) only by the normalization is the best
suited when writing down thermodynamic formulae. Of great importance
are the statistical moments of this SCD
X

l f

(
)

l l f

(
)

(4)

the rst of which specify the average composition of the sample whereas the
second dene its polydispersity. With the knowledge of generating function
of the SCD
m
(
)

G( s )
f l
(5)
s l

=1

11
its statistical moments of any order can be found elementary. Thus, quantities Y, X , X , introduced in formulae (3), (4), are expressible through the

generating function G (
s ) and its derivatives

G (
s)

G (
s)
ln s

G (
s)

2 G (
s)
ln s ln s

(6)

taken at point
s = 1 using the expressions
(
)
Y =G 1 ,

X = G

(
)

1 ,

X = G

(
)

(7)

The extension of the expressions (6) to the statistical moments of arbitrary


order k will read
X

l l l f

(
)

= G

(
)

(8)

where G (
s ) stands for the k-th derivative of the function G (
s ) with
respect to variables ln s , ln s , ..., ln s .
Currently available chromatographic methods enable one to separate
some heteropolymers in accordance with both their size and composition in
order to nd experimentally their SCD. On the other hand this distribution
(or its generating function) is also obtainable theoretically from the solution
of corresponding material balance equations describing chemical transformation during copolymer synthesis. Such solutions are presently known for
several commonly accepted in macromolecular chemistry kinetic models of
free radical polymerization, polycondensation and other ways of manufacturing of polymer materials. The adoption of these solutions in thermodynamic
calculations makes possible to extend these latter, formerly applied for hypothetical model heteropolymers, to real ones. Obviously, the SCD does not
supply complete description of a sample of statistical heteropolymer because
this function carries no information on the pattern of alternation of units in
polymer chains. In order to characterize quantitatively their molecular structure in statistical chemistry of polymers they apply two dierent approaches.
According to traditional approach the microstructure of heteropolymers
is specied in terms of a set of probabilities P (Uk ) of all conceivable sequences Uk , containing k 2 successive units. For instance, the structure of
binary copolymer chains may be roughly dened by fractions (probabilities)

12
of dyads M1 M1 ,M1 M2 ,M2 M2 . Increasingly more complete description of its
microstructure is attainable by setting the probabilities of triads
U3 = M1 M1 M1

M1 M1 M2

M1 M2 M1

M1 M2 M2

M2 M1 M2

M2 M2 M2 (9)

tetrads, pentads and so on.


As distinct from the probabilities of undirected sequences P (Uk ) obtainable experimentally, the probabilities of directed sequences P {Uk } normally
appear under theoretical considerations. The probabilities mentioned coincide P (Uk ) = P {Uk } for symmetric sequences Uk (for example M1 M2 M1 )
and dier by a factor of two P (Uk ) = 2P {Uk } for asymmetric ones (such as
M1 M1 M2 ).
To-days level of experimental equipment allows one to determine by the
NMR-spectroscopy with an accuracy of less than 10% the fractions of sequences comprising up to 5-6 units. Theoretical problem of calculation of
such probabilities P (Uk ), i.e., the sequence distribution in copolymer chains,
turns out to be trivial if these latter are markovian.
However, modern approaches of statistical chemistry permit one to have
this problem solved even for a number of nonmarkovian heteropolymers[10,
13] obtained, for example, in the course of co-operative reaction of chemical
modication of homopolymer such as the hydrolysis of polymethyl methacrylate.
The second way to characterize quantitatively chemical structure of heteropolymers is based on the introduction of chemical correlation functions
(correlators). The simplest of them, Y (k), has a meaning of joint probability that the two chosen at random monomer units, divided along macromolecule by an arbitrary sequence Uk , will be M and M . By this denition
the expression for the two-point correlator can be put down as
(2)

Y (k) Y (k) =

Uk

P {M Uk M } = P M Xk M

(10)

where symbol X stands for a unit of an arbitrary type. The extension of


formula (10) to the p-point correlation function is presented by the expression
(p)

Y (k1 , , kp1 ) = P M Xk1 M Xk2 Xkp1 M

(11)

for the probability to nd p 2 monomer units M ,M , ,M divided


along polymer chain by arbitrary sequences consisting of k1 , k2 , ..., kp1 units,
respectively.

13
Statistical description of the heteropolymers in terms of chemical correlators appears to be comprehensive, permitting one to calculate statistical parameters of polymer samples which characterize in parallel with microstructure their polydispersity as well. In fact, it is readily seen that the correlators
(11) at values k1 = k2 = = kp1 = 0 coincide with the probabilities P {Up }
occurring under the traditional way of the description of sequence distribution in copolymers. On the other hand, statistical moments of the SCD may
also be expressed through these correlators. For instance, the quantities X
(4) were found to be related in a simple way to the two-point correlators
(10)[10, 13].
As for the block-copolymers, the samples obtained can have either determined or stochastic architecture depending on the fashion of their synthesis.
In the rst case, unlike in the second one, the number of blocks n as well
as their arrangement are the same for all macromolecules of a given sample although they may dier in length of the blocks involved. Among such
heteropolymers the most extensively studied both theoretically and experimentally are binary ones comprising two or three blocks
M1 M1 M2 M2 Ml11 Ml22
|

{z

}|

l1

{z

{z
l1

}|

{z
l2

(12)

l2

M2 M2 M1 M1 M2 M2
|

a)

}|

{z

Ml21 Ml12 Ml23

b)

l3

Obviously, statistical characterization of block-copolymers of xed architecture appears to be substantially simpler because no sequence distribution
problem arises for them. The only thing to do for a complete description
of such a heteropolymer is to set the function of distribution of blocks for
lengths. This distribution is, naturally, prescribed by the synthesis conditions
and may be either very narrow, if the blocks are being formed in the course
of living polymerization via anionic mechanism, or more wide for free-radical
or polycondensational mechanism.
Among block-copolymers of stochastic architecture the simplest are segmented copolymers consisting of macromolecules with alternating blocks of
units M1 and M2 . The sample of such a copolymer in accordance with the
conditions of its synthesis represents a mixture of macromolecules, the architecture of each of which can be unambiguously characterized by two integers
n = 1, 2, ... and a = 0, 1, 2. The rst of them indicates the overall number of

14
blocks of a particular type contained in a randomly chosen molecule while
the second one species how many of them are located at its boundaries.
The distribution fna of molecules of any sample of segmented copolymer for
stochastic variables n and a gives a complete statistical characterization of
its architecture.
The question arises about the prospects of the solution of the above problem as applied to other types of block-copolymers whose molecules may vary
not only in number of dierent blocks but also in a fashion of their arrangement along polymer chain. It may be envisaged as a stochastic sequence of
blocks comprising units of dierent kinds regardless of their lengths. Under such an architectural consideration of block-copolymers with stochastic
structure mathematical problem of its characterization is formally reducible
to that on microstructure of statistical copolymers discussed in the foregoing.
Present section provides a general idea about the information required to
characterize completely or partially the chemical structure of a heteropolymer sample. This information proves to be indispensable for the calculation
of thermodynamic parameters of a solution or melt of a given sample. Here
the set of necessary statistical characteristics of its molecular structure distribution depends on particular thermodynamic problem to be solved. These
characteristics can be determined either from treating experimental data or
arrived at theoretically. For this purpose the methods of statistical chemistry prove to be notably ecient since they enable one to establish within the
framework of the generally accepted nowadays kinetic models the correlation
between statistical characteristics of samples of actual polymers and stoichiometric and kinetic parameters of the process of their synthesis. The usage of
the information of this sort on chemical structure of heteropolymers for thermodynamic calculations is of great applied value for theoretical prediction of
equilibrium properties of technologically important polymer systems.
Concluding this section it would be pertinent to point out the problem
relevant to the nomenclature of multicomponent polymers of complicated
architecture. Even when their molecules are linear, but a blend of several
heteropolymer samples obtained under dierent condition is under consideration, yet the elaboration of a universal system of notation suitable for practical use will involve serious diculties. The main of them consists in the fact
that each sample constituting a blend is by itself a mixture of enormous set
of individual compounds. Here a lot of the credit must go to Sperling[14, 15]
who called the attention to this problem and devised for its solution New

15
General Nomenclature Scheme for polymer blends, graft copolymers and interpenetrating polymer networks. Although this scheme adopting the Group
Theory concepts furnishes no complete description of polymers it, nevertheless, not only allows one to accurately name their known compositions but
also to discover new ones. Schemes of such a kind are of particular signicance
for developing computer software where their advantages are indisputable.

THERMODYNAMIC APPROACHES TO
STATISTICAL COPOLYMERS

The very circumstance that the macromolecules involve a great many units
is responsible for a number of special features peculiar to thermodynamic behavior of polymer systems when compared to low-molecular ones. Because
such peculiarities have been observed in solutions and blends either of heteropolymers or homopolymers these latter in view of their simplicity appear
to be more preferable to outline briey these peculiarities in order to switch
then to the detailed discussion of special features inherent to the former only.
A retrospective journey into the history of the progression of theoretical ideas
in the eld of thermodynamics of statistical heteropolymers oers one more
convincing argument for such an introduction anticipating the body of this
Section. The evolution of these ideas bears close resemblance to that of the
theory of miscibility of solutions and blends of homopolymers. Here we will
indicate only the milestones referring an interested reader for details and
references to a wealth of reviews and monographs[16]-[44].
Necessary requirement for miscibility is that the Gibbs free energy of
mixing
G = H T S
(13)
be negative. To derive the expressions for the entropy (S) and the enthalpy
(H) of mixing of macromolecular systems dierent models have been employed, the simplest of which has been put forward independently by Flory
and Huggins. Within the framework of this FH model the S is obtainable
from combinatorial consideration of the arrangements of macromolecules on
regular lattice while for the heat of mixing it has been suggested to resort
to the van Laar relationship. As a result, for the entropy and the enthalpy
of mixing (reduced to the net number of monomer units M ) of a blend of

16
two homopolymers Ml11 and Ml22 with volume fractions 1 and 2 familiar
expressions are arrived at
S
Sc
1
2
=
= ln 1
ln 2 ;
M
M
l1
l2

H
= 12 1 2
TM

(14)

Hence forward for simplicity sake the value of the gas constant is set equal
to unity, so that the entropy is dimensionless whereas the temperature T has
the dimensionality of the energy. The sign of the dimensionless binary interaction parameter 12 = (11 + 22 212 ) /T , where ij (ij = 1, 2) stands
for the energy of the formation of a contact between units Mi and Mj , is
of utmost importance. At positive 12 the mixing of polymers is accompanied by the heat evolving while at negative 12 the process is attended with
heat absorption. For nonpolar polymers they frequently use the Scatchard

assumption ij = ii jj , which makes the parameter 12 = (1 2 )2 be


positive and an increase of the dierence between values of solubility parameters 1 and 2 of polymers gives rise to exothermic heat of their mixing. Even
minor discrepancy between 1 and 2 is enough for real homopolymers to be
incompatible since their entropy of mixing (14) is small in comparison with
the enthalpy in view of large values of l1 and l2 . Reciprocal value of the lesser
of them provides an estimate of the order of magnitude of the parameter 12
above which mixing is made impossible. Under this condition an entropic
and enthalpic terms in expression (13) for the free energy become compara(

1/2

)
1/2 2

ble. Akin estimate is obtainable from the expression 212 = l1


+ l2
for the interaction parameter value at critical point.
The narrowness of the theoretically predicted interval of values 12 where
the mixing of high-molecular homopolymers is feasible, enables the explanation for commonly known incompatibility of vast majority of them. This interval was found to markedly expand when going from blends to the solutions
of polymers where the entropy of mixing due to the contribution from the
solvent becomes comparable with the enthalpy at 12 1, as with solutions
of low-molecular compounds. This conclusion can be inferred immediately
from formulae (14) where l1 would be put equal to unity.
The FH theory, being the origin from which the thermodynamics of polymer systems emanated, in its rst version rests on a number of assumptions.
1) Fluctuation eects are excluded from consideration. That is why like
any mean-eld theory this one does not work when dealing with the diluted

17
solutions as well as with the phenomena occurring in close vicinity of the
critical point.
2) The system is supposed to be incompressible. This restriction precludes
one from explaining the experimentally observed volume alteration under
polymer mixing and leaves apart the eects induced by the pressure change.
3) The application of the rigid-lattice model. Within its framework no
account is taken of the distinctions in size and shape of monomer units M1
and M2 .
4) Random mixing approximation. This assumption, whereby the local
composition of quasicomponents coincides with their average composition in
a mixture, would be expected to be violated for some systems where specic
interactions (either hydrogen bond or -complex formation) are of special
importance.
5) Polydispersity is ignored.
The advancement of theoretical concepts in the eld of thermodynamics
of homopolymer solutions and blends was due to the models developed with
the idea of relaxing some of these ve principal assumptions.
One of possible trends is connected with the extension of the applicability
area of this FH theory by virtue of its generalization to polydisperse polymers
whose free energy of mixing may be presented as
G
TV

(l )
l

(l ) ,

ln (l ) +

= 1,

<

(l ) =

(15)

fW

(l )

Here (l ) and are volume fractions of the -th component (i.e.


of homopolymer molecule Ml containing l units) and of the -th quasicomponent (i.e. of the unit M ). In the expression (15) the solvent is nominally regarded as monodisperse polymer with molecular-weight

distribution fW
(l ) = l 1 where l 1 represents the Kronecker deltasymbol (ij = 0, at i = j; ij = 1, at i = j). Proceeding from expressions
(15) which extend relationship (14) to the arbitrary number of polydisperse
polymers and solvents, general formulae for the cloud-point curve, spinodal
and critical points have been derived. Analogous results have been later rederived in terms of continuous thermodynamics[45, 43] where molecular weight
is thought of as continuous variable, so that the operation of summing in ex-

18
pressions for thermodynamic characteristics is replaced by integration over
this variable.
The FH model, being the simplest in thermodynamics of polymers may
not, naturally, claim to be a universal one. This model is, in principle,
incapable of oering an explanation of a number of experimentally observed
phenomena, such, for instance, as the volume alteration under mixing of the
components or negative sign of the entropy of mixing. Besides, this model
does not describe such important peculiarities of the topology of polymer
system phase diagram as the bimodal character of the cloud-point curve or
the co-existence on it of the upper and lower critical solution temperature.
The above mentioned ndings point to the necessity of relaxing the FH
model assumptions, the most vulnerable of which is concerned with the neglect of the volume alteration under the mixing of polymers one with another
or with the solvent. This eect is easy to understand in terms of the free
volume which is a dierence between true experimentally obtainable volume
of a system and intrinsic volume of its molecules. The latter can be calculated as a sum of volumes of the atoms and their groups contained in these
molecules by using the experimental data obtained by X-ray analysis. Unlike the intrinsic volume the free volume, being controlled by the intensity of
interatomic interactions, does not exhibit the additivity property under the
mixing of the liquids. The most essential among the distinctions between
a variety of models admitting the compressibility of polymer mixtures (including solutions and blends) consists in the mode of the account of the free
volume. In terms of the lattice model such an account is commonly taken in
one of three following ways (see Figure 2).
1) Cell Models. Every site of a lattice (termed cell) is occupied by
either the solvent molecule or monomer unit as in the case of the classical
FH model. However, in contrast to the latter, here such quasicomponent does
not occupy the whole volume of a cell but only its core. The remaining parts
of such cells (which for these models are taken to be the same) constitute
free volume of a polymer system.
2) Hole Models. Like with the FH model here the quasicomponent located
in a cell occupies it completely. The distinction, however, lies in the fact that
the hole models admit the existence of vacant cells (holes), total space inside
of which will be identied with the free volume.
3) Cell-Hole Models. These combine two types of the above models and
take simultaneously into consideration the contributions to the free volume

19
from each of these models.
To the contraction of a mixture, induced by the pressure rise or temperature alteration, in the models of the rst type there corresponds identical
reduction of the volume of each cell with no change of their overall number.
In contrast, for the second-type models this phenomenon is dened by the
decrease of the concentration of holes on retention of the volume of each of
them. Within the framework of the third-type model changes of both the
cell size and number of holes in liquids lead to variation of the total volume.
In the course of the renement of statistical thermodynamics of polymer
systems it has been repeatedly attempted to relax the other restrictions of
the FH approach remaining, nevertheless, within the framework of the lattice
model. Its extension has been put forward which consists in the incorporation of suggestion to drop the rigid-lattice condition that each site has the
same number of the nearest neighbors and to assign interacting surface areas to the sites. The various species on the lattice, including the holes, can
be permitted to dier in surface area, in accordance with their molecular
shape. Allowing for this factor yields the dependence of the heat of mixing
H on volume fractions of polymers, dierent from quadratic dependence
inherent to the FH model. This is formally reected in the fact that the
parameter 12 in the expression (14) becomes governed by 1 or 2 . Just
the same result has been arrived at in terms of other models where to take
due regard of the deviations from additivity of pair interaction and random
mixing approximation additional terms have been introduced, involving association parameters, non-randomness parameters and some other. Several
approaches are also known, in which attempts have been undertaken to allow for the orientational ordering when calculating combinatorial entropy of
mixing of homopolymers. The relationships derived lead, naturally, to the
expressions dierent from Sc (14), which comprise empirical chain exibility parameters.
In thermodynamics of polymer systems it is a common practice to express
the formulae for the free energy or the equation of state in terms of reduced
quantities which are dened as dimensionless variables resulting from the division of each of thermodynamic quantities (pressure, volume, temperature,
entropy and so on) by corresponding parameter of the reduction. The latter
is specied solely by chemical nature of a substance and is believed to be identical for homologues of a particular polymer. Being written down in terms of
reduced parameters the equation of state should be of the same appearance

20
for an arbitrary system pertaining to the class in hand, for instance the blend
of two homopolymers. Major problem, encountered for the derivation of an
equation of state written down in terms of reduced quantities, is the absence
of rigorous algorithm capable of expressing the parameters of reduction for a
mixture through those of its components. In order to cope with this diculty
within the framework of the equation-of-state theory (EST) they generally resort to some semiempirical additivity rules, frequently diering for various
models.
Currently available modications of the EST, enabling one to elucidate
many peculiarities of thermodynamics of polymer systems, exhibit, however,
a number of substantial shortcomings. Among these is cumbersome appearance of nal formulae, incorporating often too many empirical tting parameters which hampers general analysis and the revelation of qualitative
features of phase diagrams.

4.1

Simplest Criterion of Polymer Blends Miscibility

Almost forty years ago in paper by Stockmayer et al[46]. it was pointed


out that when considering the solution of binary monodisperse statistical
copolymer in terms of the FH model the expression (14) for the enthalpy of
mixing remains valid upon substitution in it of binary interaction parameter
for its modied value
12 = 13 1 + 23 2 12 1 2

(16)

Here 1 and 2 are the volume fractions of monomer units M1 and M2 in


the copolymer while 13 ,23 and 12 are parameters of the interaction of
these units with the solvent and one with another. Expression (16) remains
true if we are dealing with the blend of copolymer with homopolymer while
its extension to the blend of two monodisperse copolymers with arbitrary
number of types of monomer units has been provided by Krause et al.[47].
For a blend involving arbitrary number n of such copolymers the expression
for the enthalpy of mixing can be presented as
H
=
TM


ij

(17)

i=1 j=1

The value of parameter ij , characterizing volume interactions between


molecules of i-th and j-th copolymers whose volume fractions i and j ,

21
are dened by the expression
m

)(
)
(
1
ij =
i j j i
2 =1 =1

(18)

where i represents a volume fraction of units M in i-th copolymer. Making


use of expressions (17) and (18) provides a possibility to describe in terms of
the FH model any heteropolymer blend.
Expressions for the parameters ij can be found in a number of
papers[48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58] where there have been examined compatibility conditions of binary blends of a homopolymer with
two-component copolymer[48, 49, 50] or mixture of two fractions of: a) one
such copolymer[48, 50], b) two copolymers containing one identical type
of monomer units[50], c) two copolymers with dierent pairs of monomer
units[49]. In subsequent works[51, 52, 53, 54, 55, 56] the analogous consideration was carried out for ternary blends of: a) binary copolymers[51, 52],
b) two homopolymers and copolymer composed of their monomer units[53]
as well as binary blend of terpolymers with dierent composition obtained
from the same triple of monomers[55].
The case of arbitrary blend of copolymers was examined by Paul and coauthors[54, 55, 56] who made use of the algorithm which permitted them to
write down the simple expression for the heat of mixing Hm . The essence
of this algorithm, applied for the calculation of Hm within the mean-eld
theory formalism, consists in the necessity to subtract the certain term from
traditional one (the second item in the right-hand part of formula (15)).
This term equals the heat of the formation of copolymer molecules from
monomer units[54] and takes into account intramolecular contacts between
dierent type units. The same expression for Hm can be put down as
quadratic form (17) with respect to volume fractions of macromolecules of
dierent types. Coecients (18) of this form, having the sense of the binary
interaction parameters for macromolecule interactions, are readily expressed
through analogous parameters of monomer units[56].
If to share the standpoint of Barlow and Paul[54] and when nding compatibility conditions of heteropolymer blends to neglect the entropic contribution to the free energy, the question about the compatibility of particular
blend will be answered as a result of calculation of its parameters ij (18)
and subsequent determination of the sign of H by formula (17) for given

22
values 1 , . . . , i , . . . , n . It is of prime importance to know at which ratio
between mixture components characterized by a set {i } the system will be
homophase. Mathematically this problem reduces to the analysis of quadratic
form (17) which aims at nding regions of parameters {i } where expression
(17) for H is negative.
Another problem arising for thermodynamic examination of blends involving copolymers consists in determining under given temperature those
regions of their composition values where system remains homophasic one.
The size and shape of such miscibility regions of particular polymer sample have been suggested[57] to term its miscibility map (MM). An exhaustive classication has been put forward[57] of all topologically conceivable
kinds of such MMs depending on the Flory-Huggins parameters for a mixture of two monodisperse binary statistical copolymers having the same type
monomer unit. Later the general principles of such a classication have
been formulated[58] for arbitrary blends with any numbers n and m of highmolecular components and types of their monomer units. Complete sets of
topologically dierent MMs for the blend of a terpolymer with a homopolymer as well as for two binary copolymers have been presented. Besides,
simple rules have been established[58] enabling one to determine which of
the kinds particular system with known values of -parameters pertains
to.

4.2

Account of polydispersity and microstructure of


heteropolymers

Scott[59] seems to be the rst who examined the problem of thermodynamic


compatibility of copolymers. He fairly noticed that for such a compatibility
suciently narrow composition distribution of macromolecules in the sample
is indispensable. Later Koningsveld et al[60] proceeding from Scotts expression for the free energy of mixing for random copolymers varying in chemical
composition and in chain length derived the equation of the spinodal. Almost
thirty years after Scott paper[59] approaching the same problem Leibler[61]
managed in the framework of the Flory-Huggins consideration to obtain the
expression for spinodal and critical point for the blend of three fractions of
the copolymer with dierent composition.
Koningsveld and Kleintjens[51] put forth rather general, in terms of the

23
lattice model, expression for free energy of mixing of low-molecular weight
solvent with arbitrary number of types of binary copolymer macromolecules,
diering in the degree of polymerization and the composition. This intriguing
approach enabled them to take account of such important factors, exerting
inuence on conditions of phase separation, as copolymer composition inhomogeneity and the distinction in dimensions of monomer units and molecules
of the solvent.
The comparison of the experimental data on compatibility of binary
copolymers with the Flory-Huggins theory conclusions demonstrates that
despite its simplicity this theory can be applied to estimate the conditions of
demixing of solutions and blends containing heteropolymers [60, 27, 62, 52].
Another approach to the consideration of the liquid - liquid equilibrium in solutions and blends of polydisperse random copolymers has been
proposed by German scientists[63, 64, 65, 66, 67] in the framework of concept named continuous thermodynamics. They have analyzed the eect
of polydispersity of macromolecules on the cloud-point curve, shadow curve,
the spinodal and critical points as well as have discussed the feasibility of the
existence of three-phase equilibrium regions. In spite of impressive volume of
thermodynamic information available in papers[63, 64, 65, 66, 67] its practical value remains vague because of an arbitrary character of the expression
for free energy of mixing for polydisperse copolymers, which authors applied
in both the primary[63, 64] and later[65, 66, 67] version of their theory.
It is noteworthy one more interesting formalism [68, 69, 70, 71, 72, 73]
allowing one, even in the framework of the mean-eld approximation, to
explain the inuence of the character of the sequence distribution in copolymers on the compatibility of their blends. This approach implies the necessity, when calculating a number of binary contacts between monomer units,
to distinguish their pair interaction parameters depending on types of units
neighbouring with given one. In order to make nal thermodynamic formulas
suitable for their practical application authors of Refs.[68, 69, 70, 71, 72, 73]
assumed the majority of these parameters to be equal each other and practically conne themselves to the consideration of copolymers with the Markov
statistics of sequence distribution, which is characterized by the sole parameter . For many actual systems this approximation seems to be good enough
since it permits one varying the value of to embrace alternating, random and
block-copolymers. In terms of such an approximation the enthalpy of mixing
is determined by eective values of molecular parameters of pair interactions

24
which comprise two groups of items. The rst of them takes account of composition distribution of copolymer macromolecules in the blend and looks just
like that of the previous approach[48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58].
The addition of the second group items, which take into consideration the
character of the sequence distribution in monodisperse copolymers, enables
the interpretation of known experimental data.

4.3

Further development of the theory

Despite the fact that the FH model nds rather wide utility in thermodynamic calculations of phase diagrams for many polymer systems, it, nevertheless, proves to be incapable of explaining considerable body of experimental data. This circumstance apparently provides impetus to the elaboration of more general theoretical approaches where some of the restrictions underlaying the FH model prove to be relaxed. Thus, for instance,
to take into account the deviations from random mixing of monomer units
it has been proposed[74] to resort to a model being a modication of that
introduced earlier by Guggenheim[75] to calculate the enthalpy of mixing of
low-molecular weight liquids. In terms of this modied Guggenheim quasichemical model simple expressions for the enthalpy H of multicomponent
polymer systems have been obtained. Further development of thermodynamic theory of solutions, melts and blends containing copolymers is consistent with the allowance for free-volume eects impossible to consider in
terms of the FH theory. For this purpose they normally use (see, for example, papers[76, 77, 78, 79, 80]) dierent extensions of the Equation-of-State
approaches holding much favor under calculations of thermodynamic characteristics of homopolymers[20, 21, 22, 30, 31]. Referring an interested reader
to numerous reviews [81, 82, 83, 84, 85, 86, 87, 88, 89, 90], which provide a
detailed discussion of the peculiarities of particular approaches, further we
will just highlight their main general features and dwell briey on the most
signicant among such approaches.
Just in the same fashion as it has been done in the theory of low-molecular
weight liquids free volume eects in polymer systems may be considered in
terms of either lattice models or o-lattice (continuous space) models. In
the rst case almost all the models slightly diering in details proceed from
one of versions of compressible lattice (b), (c) or (d) depicted in Figure 2.
Along with a number of indisputable advantages these models exhibit one

25
very serious shortcoming consisting in the fact that in terms of the lattice
model the accurate account of the distinction in size and shape of monomer
units involved in a copolymer macromolecule poses rather serious problem.
The above diculties are possible to successfully overcome using the lattice cluster theory (LCT) introduced by Freed and Dudowicz[91, 92, 93, 94].
Their generalization of the standard lattice model allows a single monomer
unit to occupy several neighbouring lattice sites contrary to traditional approaches when such an occupation is forbidden. Within the framework of
this extension of the Hole Model (Figure 2c) it becomes possible to develope a theory free simultaneously from the second and the third among ve
the theory FH restrictions formulated above. Just this purpose is attainable also by means of approaches based on current concepts of the theory
of low-molecular weight liquids. A key role is played here by an o-lattice,
microscopic statistical mechanical approach being developed by Schweizer,
Curro and co-workers (see[95] and references therein) for calculating thermodynamic characteristics of polymer liquids, among them those comprising
copolymers. The general methodology is known as the polymer reference interaction site model (PRISM) theory and is based on the RISM approach for
low-molecular weight liquids[96]. Under such an approach some equations
have been derived for which upon a suitable closure approximation either
analytic[97, 98] or numerical[99] solutions have been found for model binary
polymer blends. Analytic PRISM theory has been recently extended by
Schweizer[100] to the description of thermodynamics of random copolymer
alloys.
Because of the complexity of polymer liquids their rigorous statistical
mechanical treatment always presents serious diculties so that the progress
in elaborating thermodynamic theory of such liquids is related to correct
choice of the approximation employed. When examining theoretically solutions of low-molecular weight compounds it is customary to consider as
a reference system some simple model permitting one to take into account
as exactly as possible the excluded volume of a molecules and then to add
mean-eld perturbation term allowing for the attractive interactions. Such
an approximation rests on a dierence in spatial scales of the short-range
repulsion and long-range attraction between molecules. Often as a reference system they choose an ensemble of hard spheres[101] whose thermodinamic and correlation characteristics are well known. In papers by Prausnitz
and co-workers (see [102] and references therein) when describing quantita-

26
tively equilibrium properties of polymer melts and blends the authors use
as a reference system athermal ensemble of hard-sphere chains, each possible to picture as a series of freely joint tangent hard spheres. This model
has been much studied[103] and, despite it simplicity, takes into account
some signicant features of real polymer liquids, including excluded volume of monomer units and their connectivity. A perturbed hard-sphere-chain
(PHSC) theory appears rather prospective when performing thermodynamic
calculations of real polymer systems inasmuch as this allows for in a natural way distinctions in size of dierent type monomer units[104] and its
input parameters (the eective hard-sphere-diameters and attractive-energy
of pair interactions) are theoretically based functions of temperature. In the
paper[105] an extension of the PHSC equation of state was presented for
two types of copolymer mixtures as well as theoretical cloud-point curves
and
miscibility maps
random copolymer mixtures
(
) were
( computed for binary
)
l1 l2
t1 t2
l1 l2
t1 t3
M1 M2 + M1 M2 and M1 M2 + M1 M3 . The results of calculations for
the rst of them are in good agreement with experimental data obtained for
the investigation of compatibility of blends involving two fractions of binary
copolymer: poly(styrol-co-butadiene), poly(styrene-co-butyl methacrylate)
or poly(butyl methacrylate-co-methyl methacrylate)[105]. The eciency of
the employment of the PHSC theory in thermodynamics of blends with participation of copolymers has been supported in recent papers by Prausnitz
with co-workers[106].
Analysis of voluminous literature involving the recent reviews[81, 82, 83,
84, 85, 86, 87, 88, 89, 90] supports the conclusion that there are no universal models for phase equilibrium calculations for copolymer systems available
today[89]. In the subsequent Sections 5 and 6 most of the material will be
presented in terms of a unied approach which provides a possibility to make
up to a certain extent this deciency. This theory being of rather universal
character by no means refutes and even on the contrary does accumulate
main ideas of generally accepted approaches. Indisputable advantage of such
a theory resides in its ability to describe heteropolymers with any polydispersity and sequence distribution. This provides a possibility within the
framework of a review to discuss theoretical aspects of thermodynamics of
both statistical copolymers and block-copolymers. On the other hand, when
selecting molecular models preference is given to those which could lead to
nal formulae simple enough to be applied for the treatment of experimental
data. Such an approach allows one to follow all approximations used for

27
obtaining nal results as well as to specify the area of their applicability for
the description of particular systems.

5
5.1

LARGE SCALE MEAN-FIELD THEORY


Homophase Systems

In accordance with general ideas of statistical physics[107] the interactions


between units in terms of the mean-eld approximation are taken into ac
count via introduction of a certain external eld
(r). Each its component
(r) acts upon units M in the three-dimensional space point r forming a
density distribution (r) of such units. Standard procedure of statistical

mechanics enables the dependence of


(r) on
(r) to be established. Further by means of the self-consistence condition it becomes possible to nd
another relation between these vectors which leads to closed set of equations
permitting the calculation of distributions (r) ( = 1, ..., m).
In the present Section the situation will be discussed when scales of spatial

change of densities (and, consequently,


(r)) are large enough as compared
to geometrical dimensions of macromolecules. In this case these latter are
possible to consider as 0-dimensional points, presuming that the value (r)
of the eld acting upon all units M of a macromolecule is the same. Under
such a consideration the external eld contribution to the energy of macromolecule located in the point r

l ;r =
(r) l =

(r) l

(19)

=1

is characterized only by numbers l of units M of given macromolecule and


does not depend on the pattern of their arrangement. Here detailed chemical
structure of polymer chains proves to be unessential, so that it is enough

to distinguish them by vector l only. Consequently, within the framework


of the large scale mean-eld theory( (LSMFT)
one may introduce the notion

)
of macromolecule concentration c l ; r and employ the apparatus of the
traditional thermodynamics envisaging macromolecules as individual components.
Advances in statistical thermodynamics of polymers are to a great extent due to the fact that the consideration of a polymer system with volume

28
interactions was found possible to reduce to that of two simpler auxiliary
systems.[108] In the rst of them all inter-unit chemical bonds in macromolecules of the system of interest are retained as distinct from physical
interactions between them which are completely ignored. Conversely, in the
second auxiliary system such interactions are properly accounted for whereas
all chemical bonds between units are taken to be cut, so that these latter are
viewed as ordinary components of low-molecular weight liquid. It turns out
that any thermodynamic potential in terms of mean-eld approximation can
be presented as a sum of contributions of these systems of Chemical Bonds
and Separate Units. So, for instance, free energy of spatially homogeneous
homophase system will be written down within the framework of the LSMFT
as a sum of two items
F = FCB + F ,

where

F FSU FIG = f V

(20)

the rst of which describes an ideal gas of macromolecules placed in a volume


V . Its free energy
FCB = Fin T SC , where SC = V

(
)

ln

(
)

(21)

along with the contribution Fin from internal degrees of freedom of macromolecules responsible for their conformation isomerism, takes also into account the combinatorial entropy SC . The second item in a right-hand part
of relationship (20) equals free energy FSU of the system of Separate Units
minus analogous quantity FIG for the ideal gas at the same values of densities

of units of all types. The concrete form of the function f (


) depends
on the model chosen of the liquid of Separate Units and is not of crucial
importance for the theory of polymers. Here free energy density f as well
as chemical potentials and pressure P
=

f
,

P =

(22)

are functions of temperature T and densities of units


=

(
)

l ,

where

(
)

=n

(
)/

(23)

29
which are believed to be known from the theory of low-molecular liquid.
When calculating thermodynamic parameters of particular system one
can choose as functions f , and P those which have been derived within
the framework of the simplest model of lattice liquid
f v
1
= (1 ) ln (1 ) +
T
T

2
= ln (1 )
T
T

(24)


P v
1
= ln (1 )
T
T

(25)

(26)

Here as in the foregoing stands for the energy of the formation of a


contact between units M and M whose volume fractions are and
whereas the quantity

vM

Vmin
=
= v =
;
V
V
max

= v
max

(27)

has a meaning of fraction of V falling on the minimum volume Vmin = vM ,


which the M molecules of a liquid could occupy. The free volume of a liquid
is considered to represent the dierence between its actual volume V and the
minimum one Vmin . The by denition (27) is non other than the density
of particles reduced to its maximum value max corresponding to the closest
packing of molecules. Here we restrict ourselves to the simplest version of the
hole model with the parameter of reducing max = 1/v where v is a volume
of one cell, which has physical meaning of the average close-packed volume
of a particle in the uid. Under this approximation the empirical parameter
v, having the same meaning as the excluded volume parameter when use is
made of nonlattice continuous model, is controlled only by geometrical size
of molecules.
Conditions for the system comprising r phases to( be)in thermodynamic

equilibrium are the equalities of chemical potentials l of each component


as well as that of pressure P in all phases. To the free energy (20) following
expressions for these thermodynamic variables correspond

(
)

(
)
F
(
)
=
T
ln
C
l +

n l

(
)

) l + in l
(

(28)

30
P

(
)

(
)

f (
)=T

(
)

+ P (
)

(29)

which in parallel with relationship (23) provide a description of equilibrium


states of multiphase polymer solutions and blends.
Applying formula (26) of lattice uid model of low-molecular weight liquid
we will arrive at the equation of state of polymer liquid
vP
T

= ln (1 ) (1 Y )
m

where

ef

ef 2
,
T

(30)

X X

=1 =1

where Y and X have been determined earlier (7). Expression (30) appears
to be quite similar to that derived by Sanchez and Lacombe in the frame
work of the Lattice Fluid Theory[30, 31] provided parameters
and of
their equation are substituted for v and , respectively. Unlike the simplied version of lattice uid model (24)-(26) the Sanchez-Lacombe treatment
enables the correct account of the distinction in values of excluded volume
of dierent type units. Under such a treatment the parameter v is usually
written down as a quadratic form with phenomenological coecients. No
objective diculties emerge if instead of relationships (24)-(26) for the description of the Separate Units system recourse is made to those employed in
the Sanchez-Lacombe theory and thus more accurate account is taken of the
excluded volume eect. However, for simplicity sake further we will conne
our consideration to the simplest version of a lattice model which contains
as few adjusting parameters as possible.
Within the framework of the LSMFT general expressions have been derived for the enthalpy H and the entropy S of mixing[108]
H
= P V ef +
M

(
)

/ (
)
V
=v 11 l
M
(
) [ (
) / (
) ]
S
=

f
l
ln
f
l

l
+
W
M

l]
[ / (
)
[
(
)]

1 l 1 ln 1 l
(1/ 1) ln (1 )

(31)

=1 =1

(32)

(33)

31
where angular brackets denote the procedure
of averaging of the quantity
(
)
standing between them over Weight SCD fW l (2). The reduced density

of the blend with given values of statistical parameters Y and X is obtainable


from the equation of state
(30). For pure component, characterized by vector
(

)
l = l , the value l can be calculated from the same equation (30)

where Y and X should be replaced, respectively, by the reciprocal degree

of polymerization 1/l and by the composition vector of this particular


molecule.

5.2
5.2.1

Heterophase Systems
Free energy

To provide an exhaustive description of a system involving r spatially homogeneous phases it is necessary (to )set the values of their volumes V i

(i = 1, ..., r) and concentrations C i l (i = 1, ..., r) of all components with

specied l in each i-th phase. Given the values of these internal variables,
the free energy is easily found by formula
{

F =

i=1

i T

(
)

ln C

}
(
)/ ]

+ Fin

(34)

(
)

where components i of the vector


i are related to C i l in any i-th
phase by stoichiometric condition

i =

Ci

(
)

(i = 1, ..., r; = 1, ..., m)

(35)

(
)
When in equilibrium, chemical potential l (28) of each macromolecule

with given l has the same value in each phase. This circumstance along
with the stoichiometric condition
r

C
i=1

(
)

Vi =n

(
)

(36)

32
(
)
permits one to exclude the l and to express concentrations of molecules
(
)
in each phase through n l

(
)

=n

(
)

V e

j j

(37)

j=1

where following designations are used

ei exp

1
T

=1

i l

(38)

(
)
Substitution of concentrations C i l (37) in expressions (29) and (31) with
allowance made for the equality of pressure in all phases yields closed set of
(m + 1) r equations with respect to the same number of internal variables V i
and i . Between them there are (m + 1) stoichiometric relationships
r

i=1

i V i

= M X

V = V

( = 1, ..., m)

(39)

i=1

resulting from formulae (35), (36) and (22), so that there will be only
(m + 1) (r 1) independent internal variables. Volume fractions y i = V i /V

and concentrations of units


i in all phases except the r-th one are conve
niently chosen as such variables whereas y r and
r can be determined from
relationships (39).
The expression for free energy (34) in view of (22) and (37) will read
(
)

n l ln y e V y P ( ) +
(
) [ (
)/ ]
+T n l ln n l
eV + F

F = T

i i

i=1

i=1

(40)

in

According
to formula (40) free energy of a multiphase system depends on
(
)
the n l , V and T not only in explicit but also in implicit manner through

the dependence of concentrations


i of units in dierent phases and their
volume fractions y i on aforementioned external variables. However, when
calculating the rst derivatives of free energy proceeding from formula (40)

33
there is no need to take into account this implicit dependence since the
function F reaches an extremum with respect to internal variables y i and i ,
i.e. there are the case the equalities
F
= 0,
y i

F
=0
i

(i = 1, ..., r 1; = 1, ..., m)

(41)

at any xed values of external variables. Above circumstance makes possible


to write out at once general expressions

(
)

= F n

(
)

= T ln V

y e /n
i i

i=1

)
(

+ in

(42)

(
)
F
T

P =
=
+
yiP
i
(43)
V
V
i=1
for the calculation of the values of chemical potential and pressure in multiphase polymer system which are also rederivable immediately from formulae
(28) and (29), provided the recourse is made to the relationship (37).
Thermodynamic description of an arbitrary multiphase polymer blend in
terms of the LSMFT is reducible, as it appears from the above discussion, to

the problem of the calculation of internal parameters y i and


i from the solution of a set of several equations. Their number is dened along with r only
by number of types m of units regardless of how many chemically individual
constituents polymer solution is composed of. That is why such a mixture
is called quasi-m-component. It should be borne in mind that monomer
units unlike solvent molecules are mere quasi-components and hence the thermodynamic consideration in their terms exhibits a number of fundamental
peculiarities. One of them resides in the impossibility to introduce the concept of chemical potential of monomer unit which by denition represents
a work on transferring this unit from the given phase to the innity[107].
This procedure proves to be unfeasible in principle because any monomer
unit can be transferred only together with polymer molecule it enters in.
Another substantial peculiarity intrinsic of thermodynamic consideration in
terms of quasi-components is the fact that free energy (40) of multiphase
system is not divided into items, each being dependent only on the density

i of monomer units in i-th phase and its volume V i . Clearly, traditional


thermodynamic approach is free from above anomalies when the constituents
are polymer molecules.

34
5.2.2

Phase diagram

When constructing phase diagrams of the solution of heteropolymer with


xed SCD, the fractions of molecules of low-molecular weight solvents and total fraction of monomer units in polymer along with temperature and volume
(or pressure) are independent external variables. For the blends of several
polymers (with given SCD each of them) the system is dened by fractions
of their units.
The character of variation of equilibrium state of polymer system with an
arbitrary SCD, resulting from the change of any of its n parameters (thermodynamic or structural), can be predicted within the framework of the
LSMFT. In the course of this change the point characterizing the equilibrium state of a system will drift within n-dimensional hyperspace of aforementioned external parameters. For such a drift this point can intersect some
specic (n 1)-dimensional hypersurfaces.
The most important among them is a hypersurface upon intersecting
which the second liquid phase appears for the rst time inside homophase
system. Equations dening such a hypersurface of cloud-points have been
derived[108]

= G (
s ) , s = exp

{[

(
(
)
)] }
/T

(44)

(
(
(
)
)
)

T G (
s ) + vP = T G 1 + vP = vP
(45)

where the components G (


s ) of the vector G (
s ) represent derivatives

(6) of generating function G ( s ) (5) of the SCD of all constituents including

solvents.

If to turn to equations (44) for vector = v


with components equal
to dimensionless concentrations of dierent type units it will be readily seen
that these equations always have the trivial solution
(
)
= Utr X ,

( = 1, ..., m)

(46)

where the dependence of total density of units of initial single-phase system


on pressure and temperature at given SCD can be found from the equation of
state (30). Within certain region of values of the parameters of the function
SCD, pressure P and temperature T , the equation
(44) along with trivial
(
)

solution can possess a nontrivial one = U which determines the

35

dependence of densities of units in the rst drops of precipitated phase on

their values in the bulk. Analogous dependencies of on in the case


of traditional thermodynamics
are normally
obtained[107] from the equality
(
(
)
)
of chemical potentials = of the constituents in both phases.
These particular equations can be derived from general ones (44) if molecules
(
)
of solvent are considered as units. Substituting every for function U
in the expression (45), which has a meaning of the equality of pressure of
coexisting phases, we will arrive at the equation for hypersurface of cloud

points. To each point located on it there corresponds particulare value

of the densities of units of all types in precipitating phase, so that the

manifold of all such points will be the shadow hypersurface. Its particular
case, namely, the shadow curve, is depicted in Figure 1b.
Another important specic hypersurface is the spinodal where the loss
of the local stability of spatially homogeneous single-phase state comes. In
terms of traditional thermodynamics to this condition there corresponds the
reducing to zero of the determinant of the matrix K with elements K =
2 f / = / , equal to the second derivatives of the density of

free energy f (
) with respect to the concentrations of particles of sort
and . This symmetrical matrix is positively-determined one (to say,
all its eigenvalues are positive) within that region of external parameters
where small deviations = of particle concentrations
from their average values do not disturb the system from thermodynamic
equilibrium. At this region boundary, i.e. on the spinodal, the minimum
eigenvalue in the spectrum of the matrix in hand turns into zero and then
becomes negative upon intersecting spinodal hypersurface as some external
parameter is being changed.
In thermodynamics of polymer mixtures the matrix K will have as its
elements
(
/
(
(

)
)
)
K l , l 2 f C l C l =
m m
(
(47)
)

)/ (
C l l
C l
l , l

where

(

)

l , l

=1 =1

is the Kronecker delta-symbol whereas elements C of the

36
matrix C of direct correlation functions[109] are dened by the expression
C =

1
1 2f
=
T
T

(48)

It was shown[108] that to have the spinodal found there is no necessity to


determine minimum eigenvalue of the matrix K (47) but it will suce to
solve an analogous problem for the matrix M
M = 1 C

(49)

where the elements of matrix are proportional to the second-order statistical moments (4) of SCD (3)

(
)

l l =

X = X
V
v

(50)

Mathematical condition of the spinodal consisting in the equality to zero


of the determinant |M| of the matrix M (49), generally speaking, may be
written down in three equivalent forms

a) 1 C = 0,

b) |E C| = 0,

c) C1 = 0

(51)

where E is identity matrix with elements .


The expression (49) makes sense only provided the matrix is not singular one, that is its determinant || is distinct from zero. This condition
is known to be certainly fullled for all actual polymer samples and can be
expected to be broken only when considering model polymer systems with degenerate SCD. In order to reveal this property of the distribution one should
write down corresponding matrix with elements li , equal to the numbers of
-th units in a molecule of i-th type. If the rank of the matrix turns out to
be less than m, the SCD is degenerate whereas in opposite case the determinant || turns never into zero. To nd the spinodal of model system with
degenerate SCD where || 0 recourse should be made to the condition
(51b).
Proceeding from the particular expression (25) one can reduce the spinodal condition to the form suitable for practical calculation dividing the
matrix M (49) into two items
R
where M

M = MR + MS ,
S
= v /(1 )
= (1 ) 2v / T, M

(52)

37
exhibiting at 1 regular and singular behavior, respectively. The determinant of the matrix M is expressible through the sum DR of all elements
of matrix adjoint for MR and its determinant which enables one to write out
the equation (51a) as follows

|M| = MR +

v
DR = 0
1

(53)

The (n 2)-dimensional manifold (hyperline) of critical points on the


cloud-point hypersurface can be found provided the densities of units
of all types within the incipient phase coincide with their values within
principal phase. This means that the equations (44) in all points of such

a hyperline has a multiple root = . Mathematically speaking this


implies that the determinant of the matrix E C equals zero which agrees
with the spinodal condition (51b). Thus, the manifold of common points of
both cloud-point hypersurface and spinodal one will contain the hyperline of
critical points of a system.
The equation dening such a hyperline may be written down in a variety
of ways, one of which looks as follows
L3

(3)

=0

(54)

Here along with components of the eigenvector of matrix M (49), cor(3)


responding to its zeroth eigenvalue, the components of the third-order
symmetric tensor (3) are also involved. These latter in accordance with
formula
(3)

(3)
ijk

(
i

(
j

(3)

)
k

f
+
T

(55)

are expressed through the components


(3)

(
)

l l l =
/

M
X = X
V
v

(3)

f 3 f (
)

(56)
(57)

of two other third-order symmetric tensors (3) and f (3) . The rst of them
is characterized by statistical moments (8) of the SCD whereas the second

38
one is dened by thermodynamic model chosen of the system of Separate
Units. Condition (54) together with the spinodal condition (51) yields the
equation of the hyperline of critical points of an arbitrary polymer mixture.
Sometimes, when calculating this hyperline it appears to be more convenient

to resort to the expression other than (54) for L3 where the role of will be

played by the eigenvector


of the matrix MC1 = C1 while instead
(3)
of there will appear the tensor G(3) with components
(3)
G

(3)

(3)
fijk ( 1 ) ( 1 ) ( 1 )
C
C
C
i
j
k
T

(58)

With the knowledge of one of the vectors or


another one can be easily

found because is obtainable from via multiplying the latter on the left
by any of matrices 1 or C.
When constructing phase diagrams of solutions of multicomponent heteropolymers it appears more convenient to reformulate the problem in terms
of only macromolecules, having formally reduced the number of quasicomponents at the cost (of an
exclusion of molecules of solvents. Under such

)
an approach function f l (3) stands for the SCD of polymers exclusively
which should be employed in all foregoing formulae of present Section. However, all characteristics of the system of Separate Units appearing in them
now have to be replaced by their modied values. Thus, for instance, to
calculate the spinodal using formulae (51) one should substitute in them the
matrix of direct correlation functions (48) for the matrix
= CP P + CP S K1 CSP
C

(59)

with number of rows equal to that of the types of monomer units. Into
formula (59) there enter submatrices of C (48)
(

C=

CP P CP S
CSP CSS

Kij

1 2 fSU (
)
ij
= CSS
ij +
T i j
i

(60)

which take into account volume interactions of dierent types: (polymer +


polymer), (polymer + solvent) and (solvent + solvent). In conformity with
(59) in reduced system the interactions of such a kind between monomer
units proceed either directly (the rst item) or through solvent molecules
(the second item).

39
When calculating critical point in a reduced system it can be also re(3)
sorted to general formulae (54), (55) or (58) substituting in them f for its
modied value
(3)
(3)
f = f +

K
i

(3)
ijk i j k

(61)

3 fSU (
)
=
T i j k

(62)

where following designations are used


i

CSP ;
ij j

(3)
Kijk

while summing is over all types of solvents.

5.3

Incompressible systems

In the theory of polymer blends and solutions of widespread use is an approximation when for thermodynamic calculations they neglect the compressibility of a system. This approximation in the framework of the lattice model is
characterized by the following conditions
m

=1

=1

(i = 1, ..., r)

(63)

replacing conditions of the equality of pressures P = P i in all coexisting


phases which become physically meaningless if conditions (63) are implemented. It should be pointed out that the formal substitution of relationship (63) into expressions (25) and (26) leads to nonphysical divergencies not
only for P but also for . However, their singular parts S turn out to be
in fact unessential since they are canceled in formulae (37). Consequently,
S
functions = R
+ which they contain should be substituted in terms
of incompressible system approximation for their regular parts
(
R
)

R X,Y =
T

1
2

X X

(64)

resulting from eliminating logarithm term in formulae (25), (26) with due
regard for the expression (30) for pressure. Substantially, that in the framework of the approximation (63) the energies of pairwise contacts of units

40
are present in all formulae only as combinations = T1 [ + 2 ]
which are usually referred to as the binary interaction parameters.
Thus in the approximation under consideration the state of multicomponent multiphase system is dened by the set of internal variables
y i , Xi , Y i (i = 1, ..., r; = 1, ..., m) which are connected between each
other by (m + r + 1) independent stoichiometry conditions
r

i=1

Xi

y = 1

X0

=1

(65)

i=1

Xi

y Y
i

=1

=Y0

(66)

i=1

Therefore among r (m + 2) internal system variables a number of independent ones is (r 1) (m + 1). They can be found from the solution of the set
of equations
Xi =

l f

(
) ei

Yi =

(
) ei

(67)

where following designations are used


[

e exp
i

X ,Y

l ,

y e

i i

(68)

i=1

while components R of vector R are dened by the expression (64). The


very fact that among r (m + 1) equations (67) the number of independent
ones equals (r 1) (m + 1) testies to the correct formulation of mathematical problem of calculation of internal variables. To solve it one is supposed
to know along
with v and interaction parameters only explicit form of
(
)
the SCD f l (3).

Generating function G (
s ) (5) of this distribution prescribes the location
in phase diagram of the cloud-point hypersurface which can be found from

the nontrivial solution X = X 0 , Y = Y 0 of equations

X = s G (
s )/ s ,

Y = G (
s)

(
(
[
)

)]
s = exp R X 0 , Y 0 R X , Y

(69)
(70)

41
Such a solution as distinct from the trivial one
X = X0 = G

(
)

(
)
Y =Y0 =G 1

1 ,

(71)

can exist only on hypersurface of parameters of a system where a new phase


appears for the rst time.
Formulae (67)-(70) can be derived by various means, the most elegant
of which is realized in terms of general approach, set forth in Subsection
6.1. The allowance for the incompressibility condition (63) reduces here to
the
( addition
) to the integrand (96) of functional Delta-function of argument
. Within the mean-eld approximation the functional (97) should
be minimized with regard to the mentioned constraint. Instead of this un
constrained extremum of modied functional Fb [
] is conveniently sought
for (by having)added to the right-hand part of the expression (97) the item

T wdr where the Lagrange factor T w (r) is to be found from the


incompressibility condition (63). As a result it turns out that in the limit of
the compressibility absence in formula (19) for eective eld (r) ought to
be replaced by other one
(r) = v

(r) + w (r)

(72)

For the case of spatially homogeneous phases the components of this eld
will equal R
and the expression (72) goes over into (64).
Turning to (53) it is possible to write out the spinodal condition for
incompressible systems
D
A = 0
(73)

where A are elements of matrix A adjoint for matrix M with elements


(

M = 1

+ v

(74)

No diculties arise also in calculating eigenvector components entering


into the equation (54) which dene critical point since for incompressible
systems simple equations hold[108]
=

=0

(75)

42
As for the equation (54) for determination of critical points, when dealing
with incompressible system in expressions (55) and (58) the second items may
be put equal to zero, taking into consideration (the )second formula (75).

For any incompressible system, since all l as well as the equal


unity, formulae (31)-(33) become essentially simpler. Under such an approximation, V = 0, the expression for the entropy of mixing S contains only
combinatorial contribution whereas the enthalpy of mixing H in a trivial
manner
H
=

(76)
MT

is expressed through elements of the correlation matrix of composition

vector . By denition we have

(77)

where angular brackets denote the averaging of the random value between
them over the Weight SCD (2). If the system does not contain molecules
of heteropolymers the rst item of the expression (77) is equal to zero at
= and the formula for free energy of mixing of incompressible blend of
homopolymers reduces to the form presented in equation (15).
The expression (76) for the enthalpy of mixing of incompressible heteropolymer system coincides up to designations with earlier presented expressions for H (17), distinguishing by more convenient notation for the
calculation of actual polydisperse systems. The sum of elements of each of
m rows and columns of symmetric matrix equals zero, so that among m2
of its elements only m (m 1) /2 will be independent which coincides with
the number of parameters .
For the melt of polydisperse binary copolymer the formula (76) reduces
to the form H/M T = 12 2 which indicates that the enthalpy of mixing is
proportional to the variation 2 of composition distribution and turns into
zero for monodisperse sample. Dealing with terpolymer it appears more
convenient to choose as independent elements of matrix diagonal ones,
= 2 , by virtue of which the expression (76) will become as follows
3

H
(78)
2 , where 2 +
=
MT
=1
Here (, , ) is a triple of indices among which there is no a pair of identical
ones.

43

5.4

Multiple Critical Points

At certain values of external parameters of a system critical points on its


phase diagram upon their conuence become degenerate. Depending upon
the degree of degeneration they distinguish tricritical, tetracritical and so
on points which prescribe the topology of phase diagrams[110],[111]. The
problem of nding of multiple critical points in macromolecular systems features a number of qualitative peculiarities associated with polydispersity of
polymer samples. Unlike for homopolymer systems, where this problem has
been solved earlier[4], for solutions and blends involving heteropolymer until
work[108] appeared only some particular problems have been treated in literature. Key ideas of this study connected with nding of critical points follow
from general approach (see Subsection 6.1) and consist in the following.
Since all critical points are located on cloud-point hypersurface it will
suce to restrict the consideration to the contribution linear in small volume
V of incipient phase to the alteration of free energy F. Proportionality
coecient

F ( )
Lp p
=

(79)

V T
p=2 p!
turns into zero at the hypersurface mentioned. The meaning of this formula
is conveniently claried by considering F as a function of m variables
which have a sense of deviations of -th type unit concentration inside nucleus
of incipient phase from their values within the principal phase. In the region
of stability of spatially homogeneous state this function has a minimum at

the point
= 0 . This stable node of hypersurface F (
) is known
to transform on the spinodal into saddle which the vanishing of minimum
eigenvalue of matrix (49) mathematically corresponds to. The normalized

eigenvector corresponding to indicates the direction along which the loss


of stability of spatially homogeneous state of the system occurs. Because the
remaining eigenvalues of this matrix are positive the shape of hypersurface

F (
) in the vicinity of the minimum will resemble a ravine. For the
movement of the point along the hollow of this ravine (i.e. along a specic
line, the location on which is characterized by parameter ) the height of
the moving point will alter in accordance with formula (79). This specic

line will pass through all the extrema of the function F (


) laying in

the vicinity of its principal minimum at point = 0 . To the minima,

44
appearing among these extrema, there will correspond other (meta)stable
states of a system. As external parameters of the system change the above
minima for their shift can be expected to merge at critical points with the
main one. At ordinary critical point two minima merge, at tricritical point
three do and so on. Mathematical condition for bifurcation inducing the

conuence of n minima of function F (


) will be the vanishing of all
coecients of the expansion into series (79) up to the (2n 1)-th one
L2 = L3 = . . . = L2n1 = 0 ,

L2n = 0

(80)

Equations (80) represent general conditions for nding of multiple critical


points.
Coecient L2 = goes to zero on the spinodal while the problem of
determination of all the remaining coecients Lp of the expansion (79) into
power series was found possible to solve in general form for an arbitrary
heteropolymer system[108]. It appears more convenient to present nal results in a diagram form rather than in analytical one when to each diagram
element appropriate analytical expression is corresponded and general algorithm of the construction of necessary phase diagrams is formulated. This
can be accomplished in two equivalent fashions, the rst of which is based on
the density functional (DF) formalism (see 6.1.2) while the second one incorporates the random eld (RF) formalism (see 6.1.1). Both these approaches
make use of one and the same diagram technique for the calculation of the
sought for values Lp , to which there correspond in Figure 3 grey regular p-gon.
To the same diagram elements depicted in the Figure 3 dierent analytical
expressions correspond in each approach.
In terms of the DF formalism the elements of the diagram technique
are depicted in Figure 3a, where an index i corresponds to the color of
relevant i-th apex. To elements 1 , 2 and 3 there correspond, respectively,
analytical expressions
1

2
/T
+ f(p)
(p)1 ...p = (p)
1 ...p
1 ...p

3
M+1 2

(81)

where are components of the eigenvector of matrix (49) corresponding


to its eigenvalue = 0 while the meaning of the remaining designations will
be elucidated below.

45
In order to write out analytical expression for coecient Lp one should
draw a set of all cactuses with number p of external apices comprising elements in Figure 3a in such a way as to provide for each apex to be adjoined
to either element 1 or to one of element 3 edges. Essentially, that the
junction is admissible of only apices of the same color, over which the summation is made thereupon. This may be conventionally corresponded to the
color-erasing procedure resulting in the graphs with uncolored apices (Figure
3b). Integer appearing in front of each cactus is equal to order of a group
of its automorphisms and can be calculated by means of the graph theory
methods[112] while the sign equals (1)n where n represents the number
of elements 3 in a cactus. The value of interest Lp is obtainable by summing of contributions of all cactuses with the allowance for weight coecients
appearing in front of them in corresponding diagram.
It is easy to verify that above-formulated diagram rules enabling one to
nd the expressions for Lp are immediately conducive at p = 3 to expression
(54) presented in the foregoing while for L4 , for instance, the recourse to this
diagram technique yields following relationship
L4 =

{i }

(4)
1 2 3 4

(3)
1 2

+ (3)
M
3 4

i=1

(82)

where summation is made over all sets of indices.


Having formulated the principle of the construction of the diagrams which
permits the calculation of the unknown coecients Lp of the expansion into
power series (79) let us elucidate now the meaning of quantities (81) un+
dened so far. Among them are the elements M
of matrix M+ which is
pseudoinverse to the matrix M.
To obtain matrix M+ elements
m
+
M

r=2

1 r r
r

(83)

it is sucient to nd complete set of eigenvalues r , and corresponding to

them eigenvectors r of the matrix M (49), with the exception of the prin

cipal eigenvalue 1 and its eigenvector 1 which enter into the


expression for spinodal and critical point.
To have the components of symmetric tensor of p-th order (p) expressed
through statistical moments (8) of the SCD (3) recourse should be made to

46
the diagram technique similar to the above discussed one, whose elements
are
1
(1 )

2
(p)
1 ...p

3
(1 )1 2

(84)

Matrix is dened by the expression (50). The symmetric tensors of pth order (p) as well as f (p) /T entering into the expression for (p) (81),
which characterize the system of Chemical Bonds and that of Separate Units,
respectively, look as follows
(n)
1 ...n

l1 ...ln C

(
)

f(n)
1 n f (
)
1 ...n

T
T 1 ...n

(85)

As distinct from (81) to the element 1 there corresponds not the point but
a line whose ends are marked by colors and . The rst of these colors
should be erased in accordance with the above described the color-erasing
procedure while the second one should be retained. As a result the diagrams
with colored apices are arrived at similar to those depicted in Figure 4a. The
tensor (p) , to which here grey p-gon corresponds, can be found by summing
the contributions of all cactuses with p external colored apices in line with the
rules formulated for the diagram technique (81). Thus, for instance, diagram
expression for (5) is presented in Figure 4b where the symbol means the
summation over all dierent ways of coloring of the cactus.
When calculating coecients Lp in terms of the RF formalism the diagram technique (81) remains the same as that discussed above with the only
dierence that to diagram elements now will correspond other quantities
1

2
G(p)
1 ...p

= (1)

p1

3
(p)
1 ...p

B(p)
1 ...p

G+
1 2

(86)

Here are the components of eigenvector of matrix G = C1 corresponding to the eigenvalue = 0 whereas G+
1 2 represents the elements of
+
the matrix G pseudoinverse to the matrix G. The components of tensor
B(p) with p colored external apices, describing volume interactions in the
system of Separate Units, can be written down immediately using diagram
technique (84) under the following correspondence of diagram and analytical
elements

47

f(p)
/T
1 ...p

(C )

3
(C )1 2

(87)

Which particular among above discussed approaches is more appropriate


to calculate coecients Lp is completely prescribed by a researcher requirements. When constructing phase diagrams of particular polymer sample
investigated under dierent conditions (for instance, in various multicomponent solvents) preference should be given to the formalism (81). The reason
is that the most tedious procedure of nding of explicit form of tensors (p)
should be performed in this case only once. On the other hand, the advantages of the formalism (86) is especially pronounced in the case when
thermodynamic properties of reaction systems are examined in the processes
of polymer synthesis. In such systems physical parameters (temperature,
pressure) generally remain unaltered whereas the SCD of polymer formed
changes continuously in the course of the process. The random-eld formalism proves to be exceptionally convenient when considering thermodynamics
of incompressible systems since all the components of tensors B(p) with p 3
turn into zero in terms of such an approximation (63). The only exclusion is
the tensor B(2) B whose components are determined by simple expression

B =

[
(

(88)

The DF formalism, if applied to limit transition to incompressible system,


is more cumbersome inasmuch as the components of all tensors f (p) at p 2
tend in this limit to innity. However, by virtue of the fact that the sum
of elements of any row or column of matrix M+ equals zero and the second
condition (75) is fullled, coecients of singular terms turn into zero. Thus,
when performing particular calculations for incompressible systems in terms
of the DF formalism, in expression (81) 2 the components of all tensors
f (p) /T but one, f (2) /T , whose components become equal to v under this
approximation, should be set zero.

48

6
6.1
6.1.1

SMALL SCALE MEAN-FIELD THEORY


General Treatment
Random Field Approach

The above introduced polymer model where macromolecules have been envisaged as pointwise particles provides no possibility to describe such important
phenomenon as microphase separation, the structure of interface boundary
as well as to estimate the surface tension and nd angular dependence of
the scattering intensity in polymer systems. Under such a consideration of
prime importance are spatial scales of order of characteristic size RG of a
macromolecule. The elaboration of quantitative theory would call here for
new approaches capable of taking into account not only chemical size and

composition l of macromolecule but also its detailed microstructure .


The most general among known approaches of such a kind to the description of equilibrium properties of amorphous multicomponent polymer
systems[113, 114] provides a possibility within the context of a unied treatment to take into consideration all the above mentioned eects for arbitrary polymer samples involving linear macromolecules of any microstructure. Such a high degree of universality within the framework of a unied
approach becomes possible to gain owing to the incorporation of the functional integration method extensively applied in contemporary theoretical
physics. Nowadays this method is known to be successfully employed also
to tackle various problems of the theory of polymers[115, 116]. Despite the
fact that such current theoretical approaches might appear at rst glance
too sophisticated (to be more precise too unaccustomed) for those who are
not experts in this eld the resultant formulae prove normally to be rather
simple and do contain only measurable parameters. The last circumstance
opens fresh opportunities for experimental verication at the quantitative
level of main theoretical results. These latter will be discussed within the
subsequent Sections while here we will briey formulate key ideas of general
approach[113, 114].
Dealing with this kind of problems one is supposed to characterize the
sample of interest by numbers n () of macromolecules with given microstructure . The synthesis conditions do not x unambiguously the set {n ()}
of these numbers, specifying only the probability distribution of various sets

49
{n ()}. So, to give thermodynamic account of a system it is necessary to perform the averaging of two types: over the Gibbs measure at xed set {n ()}
and then over dierent sets {n ()} with the measure specied by the synthesis conditions. As pointed out in the theory of disordered systems[117], free
energy F of the system rather than its partition function is a self-averaging
quantity. This means that experimentally observed free energy F ({n ()}) of
particular (typical) sample coincides in thermodynamic limit with the quantity F obtained by averaging F ({n ()}) over all sets {n ()}. To describe
the phenomena occurring at space scales R RG , it is essential to choose
appropriate conformational polymer model. For simplicity sake let us discuss
free-joined model characterizing exible chain polymers, which is dened by
conditional probability (r) for the vector connecting by chemical bond a
pair of neighboring units to be equal to r.
In the absence of volume interactions between units partition function
of a macromolecule of given microstructure with units located in external

eld H e (r) will read


( [
])
Q , H e =

drn exp Hen (rn ) /T

(r

rk+1 )

(89)

where the rst product is over monomer units while the second one is over
chemical bonds between neighboring units of a macromolecule. Inasmuch as
free energy F is estimated with an accuracy of the additive arbitrary constant, the quantity FCB for such a system of Chemical Bonds is conveniently

reckoned from its value at H e = 0


(
[
])
FCB {n ()} , H e = T

n () ln

( [
]) }
Q , H e /V

(90)

Due to self-averaging
of the free energy the expression (90) coincides with
[
e ]
the quantity FCB H arrived at by the replacement in (90) of n () by their
averaged (in accordance with the synthesis conditions) values n () = P ()
(see Section 3).
Each macromolecule makes additive contribution to the expression (90)
because we are obviously dealing with their ideal gas. No such an additivity
will be the case, of course, if the volume interactions are accounted for. It is
a common knowledge that the problem of nding of the partition function of
the system of interacting particles can be formally reduced to an analogous

50
one for the system of noninteracting particles but residing in a random eld

(r)[117]
[
(
[
]) ]
exp F {n ()} , H e /T =

{ [
(
[
])
]
}

D
exp FCB {n ()} , H e +
+ [
] /T

(91)

Functional [
] has a meaning of the eld
(r) energy. When density
uctuations in the system of Separate Units are disregarded this functional
is dened by relationships

[
]=

drP (
(r)) ,

(r) = (
(r))

(92)

Due to the self-averaging of the functional


becomes ])
possible to substi( FCB it[
e

tute in formula (91) the quantity FCB {n ()} , H +


for its averaged
over microstructure value
[
]
{ ( [
])
}

FCB H e +
= T
P () ln Q , H e +
/V
(93)

As a result the right-hand part of the expression (91) is governed no longer


by specic set {n ()} . This testies to the fact that the allowance made for
the volume interactions does not violate the self-averaging property of free
energy.
Given the density uctuations are neglected when calculating the integral
(91) recourse can be made to the steepest descent method which yields
[
[
]
]

F H e = FCB H e +
+ [
]

(94)

where the eld


(r) is obtainable from the extremum condition of the righthand part of the expression (94)
]\
[

(r)
(r) = FCB H e +

(95)

The above discussed method appears to be highly convenient when treating eects associated with density distributions inhomogeneous in space.
However, to deal with these eects other approach is known to be normally
invoked in literature[118] consisting in the introduction of the density functional.

51
6.1.2

Density Functional Method

This method rests on the property that the partition function of a system
situated within the external eld is a generating functional of the probability

distribution of various space congurations of densities {


(r)}
(
[
] )
exp F H e /T =

(96)

[ (
)
]

D
exp F [
] + dr H e (r)
(r) /T

Functional F [
] has a meaning of conditional free energy of specied con
guration
(r)

F [
]=

drf (
(r)) T SL [
]

(97)

Function f (
) has been determined earlier (24) while conditional free en
ergy of a system of Chemical Bonds, T SL [
], is expressed through the so

called Lifshitz Entropy[119] SL ( )

exp SL [
]=

D H exp

dr H (r)
(r) FCB H

(98)

Expression (96) for free energy may be reduced to the formula (91) if a
series of consecutive procedures is carried out. Performing such procedures
one is supposed: to substitute expression (97) for (98) and then (96) for

(97); to change the order of the integration over


(r) and H (r); neglecting
density uctuations of the system of Separate Units to calculate via the

steepest descent method internal integral over


(r) and, nally, to replace

the variable of integration H (r) = H (r) + (r).


The calculation of the integral (96) by means of the steepest descent
method can be conducted in two stages. For the rst of them the Lifshitz
Entropy (98) should be found. The steepest descent method is applicable to
have the integral (97) calculated under high densities when the number of

units inside the coil proves to be suciently large. In this case the eld H
can be found from the extremum conditions
[
]
(r) = FCB H /H (r)

(99)

52

T SL [
]=

dr H (r)
(r) FCB H

(100)

Substituting the obtained expression of the Lifshitz Entropy into (96)

and calculating the integral over


by virtue of the steepest descent method
we will get nal formula for the free energy
[
]

F H

= F [
]

dr H e (r)
(r)

(101)

This expression holds in the mean-eld approximation where the dependence

of density
on the eld H e is found from the equation resulting from

relationship (98) upon the replacement in it of H (r) by sum of external

eld H e (r) and self-consistent eld


(
).
It is readily seen that nal formulae for free energy derived via two dierent methods, RF and DF, coincide in the mean-eld approximation if use is
made of formulae (92) and (22). It should be emphasized that this approximation substantially diers from that discussed in Section 5. The distinction
consists in markedly less scale of those spatial inhomogeneities which the
Small Scale Mean -Field Theory (SSMFT) discussed here enables to describe
as distinct from the LSMFT.

6.2

Angular dependence of scattering intensity

Static scattering of light or neutrons from polymer systems is a powerful


tool for the investigation of their equilibrium property[120]. The treatment
of the data obtained in studies by means of this method of heteropolymer
blends and solutions provides an important information on their chemical
structure as well as enables one to determine experimentally the location of
the spinodal on a phase diagram. Because the scattering is known to be
controlled by unit density uctuations their description grows an important
theoretical problem. In terms of the uctuation theory allowance is made
for random deviations (r) = (r) (r) ( = 1, ..., m) of densities
(r) of units of all types from their average values (r) at the point r

of three-dimensional space. The probabilities of such uctuations


(r)

are dened by the Gibbs measure, exp (F [


] /T ), where functional F has
been introduced above (97). Averaging over this probability distribution
the product of density uctuation at two arbitrary points r and r let us

53
determine the two-point correlation function
S (r , r ) (r ) (r ) = T

[
]
2F H e

He (r ) He (r )

(102)

where the second equality follows from expression (96).In the absence of
external eld, when all He = 0 , the system is spatially homogeneous, so
that the elements (102) of correlation matrix S are dependent only upon the
dierence r = r r of three-dimensional vectors r and r . Systems of only

such a kind, with H e = 0 ,will be discussed later on.


When theoretical expressions for the Fourier component Se (q) of elements S (r) of correlation matrix are possible to obtain, the scattering
intensity I at any angle of the incident radiation with the wavelength
can be calculated by known formula[121]
I (q) =

a a Se

(q) ;

q=

sin

(103)

where a stands for scattering length of an -type unit. In view of spatial


isotropy of the system the matrix S and, consequently, scattering intensity
(103) will be dependent on modulus q only of wave vector q.
The calculation of the second derivative (102) of functional (101) leads in
terms of the SSMFT to simple matrix expression
e 1 (q) =
e 1 (q) C
S

(104)

e and that of strucconnecting the Fourier-transform of correlation matrix S


e The latter represents, by denition, the second derivative
tural matrix .
[
]

of the functional FCB H e with respect to the components of the eld H e .


From explicit expression (90) for this functional microscopic representation
e elements follows
of matrix

e (q) =

P () e

iq(rk rl )

(105)

Here angular brackets denote the procedure of averaging over the Gibbs
measure of conformational set of an individual macromolecule with the microstructure while inner summing is over all its units k and l whose types

54
e the right-hand part of the
are k = and l = , respectively. Along with k
expression (104) comprises also the matrix of direct correlation functions C
with elements (48) which characterize volume interactions within the system
of Separate Units.
The elements of structural matrix (105) are expressible through two-point
chemical correlators (10). To this end it proves to be convenient at =
to separate out from inner double sum (105) the items corresponding to the
same units k = l into isolated term
e (q) = + e (q)

(106)

where = X is the density of -type units averaged over space. The


second term in formula (106) is expressible in a trivial way[114]
[

e (q)
e (q) e (q) = W (x) + W (x) , where x =
(2)

(2)

(2)

(107)

(2)

through generating function W (x) of two-point chemical correlators (10)

(2)
W

(x) =

Y
k=1

(2)

(k) xk

(108)

e (q) of the probawhere the argument x is replaced by the Fourier-transform


bility (r) to nd a couple of neighboring monomer units of a macromolecule
e (q) it is well known
at a distance r from one another. As for the function
that regardless of the choice of conformational microscopic model this funce (0) = 1 and
e () = 0,
tion is positive and alter as q grows between values
e (q) satdecaying at characteristic scale 1/a. Inasmuch as any function
isfying these conditions leads to the same nal results, it is convenient to
e (q) = exp (q 2 a2 /6).
choose
Apart from the above demonstrated straightforward derivation of fundamental expression (104) at least three more fashions of its obtaining are available in literature. The rst of them goes back to the papers by de Gennes[122], Leibler[123] and Eruchimovich[124] who proceeded
from the random phase approximation (RPA) method. Benoit and coworkers[125] as well as Curro and Schweizer[126] have extended the OrnsteinZernike equation, known in the theory of low-molecular solvents, to macromolecular systems whereas the third approach makes use of the Edwards

55
Hamiltonian[115, 127, 128]. Brief overview of these three ways of deriving
of fundamental relationship (104) is given in the review[129]. The fact that
several formally dierent approaches are conducive to the same result comes
as no surprise because all these approaches take into account uctuations in
the Gaussian approximation.
For the model of incompressible lattice liquid (see Figure 2) general expression for the Fourier-transform of correlation matrix elements (102) can
be presented as follows[130, 131]
Se

f 1
= M

f 1
M

f 1
M

f 1
M

(109)

f on q is dened by formula
Here the dependence of elements of matrix M
e over index
e (105). Summation of the S
(74) where is substituted for

e
or yields zero and, thus, symmetric matrix S possesses only m (m 1) /2
independent elements. In a particular case, when number m of monomer
unit types equals two, general formula (103) for scattering intensity with due
regard for the expression (21) will look in a quite familiar way[130, 131]

I (q) =

(a1 a2 )2
,
D (q)

D (q) =

e 11 +
e 22 + 2
e 12

2v12
2
e 11
e 22
e 12

(110)

Hence, no problem will present constructing for any polymer system of


theoretical dependence of scattering intensity on the angle , provided it becomes possible to obtain for this system explicit expression for the Fouriere (105). If so, the shape of the curve I (q) for
transform of structural matrix
heteropolymer samples with dierent molecular structure may vary qualitatively. As an illustration Figure 5 depicts two such dependencies for
the melt of binary copolymer with dierent character of sequence distribution in macromolecules. In the case of statistical copolymer the curve I (q)
decays monotonically from maximum value, I (0), up to zero whereas for
block-copolymer on this curve the appearance of the maximum at a certain
q = q = 0 can be expected. At a specic value of a system parameters corresponding to spinodal the value I (q) at one of these maxima will become
innity. According to which of them it occurs, spatially homogeneous state
will loose its stability at the spinodal with respect to density uctuations
either with zeroth wave vector or with nite one q = q .

56
Simple model system where such transitions are possible has been theoretically studied in detail by Leibler[130]. He examined incompressible melt of
diblock copolymer (12a) with xed numbers l1 = lX1 , l2 = lX2 of monomer
units M1 and M2 in each block and derived for this system at l simple
expressions for elements of structural matrix
e 11 = lg (X1 , y) ,

e 22 = lg (X2 , y)

(111)
e 12 = l [g (1, y) g (X1 , y) g (X2 , y)]/ 2

which are expressible through the Debye function


g (X, y) =

)
2 (
Xy
Xy

e
,
y2

2
y = q 2 RG
,

2
RG
=

a2 l
6

(112)

where RG represents a gyration radius of the Gaussian coil consisting of l


monomer units. Under the decrease of the angle of scattering , i.e. at q 0,
scattering power vanishes due to the system incompressibility in accordance
with the law[132] I (q) q 2 while in the region of large q function I (q) tends
to zero proportionally to 1/q 2 . At q = q 1/RG scattering intensity has a
maximum whose height increases as parameter grows and, eventually, at a
certain value = s becomes innity along with the correlation radius of
uctuations in conformity with asymptotical formulae
I (q ) (s )1

al1/2 (s )1/2

(113)

The inuence of compressibility eect in a system of interest on angular dependence of scattering intensity has been analyzed by Freed and coworkers[133, 134], who showed that this remains nite at q = 0 but retains
the same form as in the case of incompressible systems. Within the area of
large q this curve exhibits asymptotical behavior I (q) 1/q 2 only at rather
high value of q. The compressibility eect acts upon the system behavior in
the strongest way in the vicinity of phase transitions.
Of particular interest are studies[135, 136], where the eect of heteropolymer polydispersity is accounted for. For this purpose it has been
suggested[135] to employ formulae (111), characterizing monodisperse diblock copolymer, upon substituting in them the Debye function (112) for its
number-averaged value. Later it has been shown[136] that to this end the
weight-averaged value of this function should be used. This assertion proved

57
by the authors[136] for concrete model ensues in general case directly from
the denition (105) of structural matrix.
The question has been theoretically examined[137, 138] which alterations
in scattering intensity causes the addition to the incompressible melt of
monodisperse diblock-copolymer (12a) of homopolymer with units M1 . It
was found[137] that under low concentration of the latter these alterations
are insignicant. However, as homopolymer concentration enhances such a
system can undergo a transition into micellar phase[138]. Numerical calculation of correlation matrix (104) which dene the scattering intensity (103),
for monodisperse melt of multiblock binary copolymers has been performed
in paper[139] where the rise of critical value of quantity s l has been predicted with the growth of number of blocks in macromolecule.
De Gennes[140] has analyzed the eect of polydispersity on the shape of
curve I (q) for mixture of macromolecules containing dierent length blocks.
He has shown that under random distribution of these blocks for length such
a structural disorder can lead to the situation when at certain values of the
parameters the maximum disappears on curve I (q) at nite value q > 0.
Explicit expressions for the function I (q) for the treatment of experimental data on the light-scattering in some particular compressible systems with three and four unit types have been found by virtue of diagram technique[125]. Analogous results are also obtainable using the RPA
method[141].

6.3

Spinodal and the Lifshitz Point

The loss of stability of spatially homogeneous state happens on a spinodal


where correlation functions (102) and correlation length diverge. The condition for spinodal transition is, conformably to expression (104), the presence
of the sole non-negative root of the following equation

e 1 (q) C = 0
D (q)

(114)

Generally speaking, this root can be of multiplicity either 1 or 2. In the rst


case the stability loss occurs at the zeroth wave vector (q = 0) while in the
second case this takes place at the vector distinct from zero, (q = q = 0). To
two such cases there correspond two dierent regions on spinodal hypersurface which are separated by a hyperline of the Lifshitz points, known from

58
thermodynamics of low-molecular compounds[142]. To these specic points
the root q = 0 of multiplicity 2 of equation (114) corresponds. By the very
denition of such a root along with the function D (q) its derivative must
vanish as well. This condition together with the equation (114) denes the
region of spinodal hypersurface where spatially-homogeneous state, remaining stable with respect to the uctuations with the zeroth wave vector, looses
its stability on the non-zeroth one. In order to nd such spinodal regions in
the case of incompressible systems one should analyze instead of (114) the
equation
Ae = 0
(115)

which ensues from equation (73), provided in the latter A are replaced by
f adjoint for matrix M.
f
elements of matrix A
The above introduced condition for nding of the spinodal can be reformue 1 . Because the determinant of
lated in terms of spectral problem of matrix S
any matrix is equal to the product of its eigenvalues, to the condition (104)
there corresponds vanishing of one of them, i.e. the minimum one, (q).
This is obtainable from the condition of the existence of nontrivial solution
of the following set of homogeneous linear equations

e
e
e 1

C = ,

=1

(116)

Here the eigenvalue as well as components e of corresponding normalized


e
eigenvector are controlled by q because of the dependence on q of matrix .
e
It is readily noticed that and at q = 0 reduce to the quantities and
, introduced in Subsubsection 5.2.2. As for the components of eigenvector
they are expressed directly through cofactors Ae (q) in the same fashion
that in formula (75) when the compressibility is neglected.
Noteworthy, the spinodal condition as well as the eigenvalues problem is
e as it has been done in
possible to formulate in terms of the matrix C1
Section 5. To do so the matrix in all formulae of that Section should be
e (q).
substituted for
Thus, the problem of theoretical determination of the spinodal with allowance for formulae (114), (115) reduces to nding of the Fourier-transform
e of structural matrix which is governed only by microstructure and polydis
persity of a polymer sample. A number of papers are available in literature

59
where this problem has been solved for some simplest models, primarily in
relation to derivation of formulae for the scattering intensity.[125]-[141]
The rst who explicitly identied the Lifshitz point in polymer systems were, apparently, Broseta and Fredrickson[143]. They analyzed the
dependence on q )of the structure factor D (110) of homogeneous melt
(
Al + B l + Al B l comprising two dierent homopolymers and diblockcopolymer. As a result of such an analysis the conditions were found of the
existence of the Lifshitz point as well as its location on phase diagram of the
quasibinary system under consideration characterized by four parameters:
volume fractions of homopolymers; the ratio of their degree of polymerization to that of the diblock-copolymer; incompatibility degree h = l where
is the Flory-Huggins parameter.
Comparatively recent are theoretical works[144, 113, 114], where the possibility has been predicted of the existence of the Lifshitz point on phase
diagram of heteropolymer systems with the markovian statistics of sequence
distribution. In the rst of them[144] the authors proceeded from the assumption of incompressibility of the system and monodispersity of multiblock copolymer. They made a recourse to the Edwards Collective Variables method[115], having taken into account volume interactions by means
of supplementing the free energy by quadratic with respect to density of units
item with nonlocal parameter of their interaction. Within the framework of
such formal approach, where parameter is dependent on the coordinates,
they constructed a spinodal and shown the existence of the Lifshitz point
on it. The very fact of its occurrence in terms of the model in hand is due
to nonlocal character of the interactions inasmuch as when this nonlocality
scale approaches zero the Lifshitz point is known to vanish. However, it can
appear in traditional theory of heteropolymers whose units interact in a local
manner provided the system is compressible. For the rst time this result
has been arrived at in works[113, 114] where simple analytical expressions
have been derived for the spinodal and the Lifshitz point for any markovian
binary copolymer with arbitrary character of physical interactions. These
expressions involve as parameters the elements of two matrices, the rst of
which describes the polymer chain microstructure whereas the second one is
a matrix of direct correlation functions (48).

60

6.4

Critical Points

In melts and blends with participation of block-copolymers there is a possibility of the existence of thermodynamically stable periodic spatial structures
with dierent symmetry of crystal lattice. Free-energy minimum corresponding to these structures can be located in the vicinity of the principal minimum
which corresponds to spatially homogeneous state and even merge with it at
critical point under certain values of external parameters. These values will
lay on hyperline obtainable from two conditions (80), which now should be
supplemented with one more equation /q = 0 for determining the modulus of the wave vector q = q of a superstructure. Expressions for coecients
L2 ,L3 in case under consideration look as follows
3

L2 = (q ) ,

L3 =

{qi } {i }

e (3)

1 2 3

(q1 q2 q3 )

e
i=1

(qi )

(117)

where external summation is over of reciprocal lattice primitive translations


(RLPT) of the superstructure. The modulus of all these vectors is the same
and equals q . This condition substantially restricts the number of potential
types of d-dimensional superstructures which are normally classied by number n of their RLPT vectors. The set of such structures includes at d = 1
lamellar structure (n = 1), at d = 2 does quadratic (n = 2), and triangular
ones (n = 3), as well as at d = 3 comprises the body-centered-cubic structure (n = 6). The expression (117) diers from analogous one (54) in that
e (3) depends on the wave vectors unlike (3) . With the allowance
the tensor
made for this dependence which extends also to the eigenvector with components e , all the expressions for coecients Lp , describing the states with
spatially homogeneous phases are retained with minor modication. This
modication suggests the procedure of summing in all formulae of the Subsection 5.4 to be made not only over indices i of units but also over their
wave vectors qi belonging to the discrete set of n RLPTs of one of the above
mentioned possible superstructures. With due regard for this modication
it is an easy matter to calculate necessary coecients Lp for a system being in any spatially inhomogeneous state provided the components of tensor
e (p) are known. They have a meaning of the Fourier-transform of p-point

61
correlation functions in a system of Chemical Bonds

e (p)

1 ...p (q1 . . . qp ) =
V

P ()

exp i

q r
j j

(118)

j=1

Here as in all the other formulae for components of tensors


e (p) ({q }) of arbitrary order p, the sum of all wave vectors
e (p) ({qi }) and
i
q1 + . . . + qp equals zero. Note, if to put all qj = 0 the expression (118)
e (p) will go into that appearing in (85) for the quantity (p)
for the quantity
then the expansion coecients Lp will reduce to those derived in Subsection
5.4.
As in the case p = 2 (105) it turns out that the components of tensor
(p)
e
can be expressed in a simple manner[114] through generating functions
of chemical correlators (11)
p1

W(p)
1 ...p

(x1 , . . . , xp1 ) =

{ki }

Y(p)1 ...p

(k1 , . . . , kp1 )

x
i=1

ki
i

(119)

where summing begins with ki = 1. To do this let us, by analogy with expression (106), separate out into individual items the contributions of summing
over repeating units.
The simple general algorithm can be put forward enabling one to write
down immediately analytical expression for any component (118) of tensor
e (p) as a sum of several items. To any such item there corresponds particular

partition of p points, each of which being dened by index j and wave vector (momentum) qj , into r groups (1 r p), comprising only the points
with the same indices. This index will characterize the given group while the
momentum attributed to it will be equal to the sum of momenta of all points
entering into this group. If each such a group is conventionally thought of
now as a point, to this particular partition in terms of the algorithm suggested corresponds particular r-point correlator whose tensor indices and
momenta characterize now not individual points but their groups. Having
(r)
made a summation over contributions e from all such partitions we will
e (p) . The use of this algorithm
arrive at the expression for correlation tensor
in a particular case r = 3 leads to the expression
e(2)
e (3)

1 2 3 (q1 q2 q3 ) = 1 1 2 2 3 + 1 2 (q1 q2 ) 2 3 +

e(2)

2 3

(q2 q3 ) 3 1

+ e(2)

3 1

(q3 q1 ) 1 2

+ e(3)

1 2 3

(120)
(q1 q2 q3 )

62
where the Kroneker-symbols take into consideration the similarity of all
points of any group with respect to their type.
The expression for arbitrary component e(p)1 ...p (q1 q2 qp ) of the sym(p)

metric tensor e represents up to


[ the factor the sum of p! items, the rst of]
e (q ) ,
e (q + q ) , . . . ,
e (q + + q

which looks as follows W(p)


1
1
2
1
p1 )
1 ...p
while the remaining items are obtainable from it as a result of various transpositions of indices. Such a symmetrization procedure permits any compo(p)
nent of symmetric tensor e to be expressed through the components of
nonsymmetric one W(p) . When p = 2 this algorithm is conducive to the
expression (107), while for p = 3 it gives
[

e(3)1 2 3 (q1 q2 q3 ) = W(3)


(b1 , b3 ) + W(3)
(b1 , b2 ) +
1 2 3
1 3 2
W(3)
(b2 , b3 ) + W(3)
(b2 , b1 ) +
2 1 3
2 3 1

(121)

W(3)
(b3 , b2 ) + W(3)
(b3 , b1 )
3 1 2
3 2 1

e (q ). The above expression for


where use is made of the designation bi
i
L3 (54) of systems with spatially-homogeneous phases follows from formulae
(118), (120), (121) if all vectors qi in them are put zero.
When calculating critical point, where the conuence happens of spatially
inhomogeneous state with homogeneous one, it is possible for incompressible
systems to apply the above reported relationships, in which all components of

tensor f (3) would be equated to zero while components f


/T are to be put
equal to v . The same rule works for the calculation of all the remaining
coecients Lp with values p > 3.

6.5

Appearance of Spatially Inhomogeneous Phase

In the theoretical construction of the cloud-point curves they used to presume the concentration of units inside the rst drops of an incipient phase
appearing in the system to be the same at all points of these drops. In
this case the problem of nding of cloud-point curve hypersurfaces within
the space of external parameters of a system can be formulated in a trivial

way in terms of generating function G (


s )(5) of the distribution of polymer
molecules for size and composition (3). Given this function, it becomes possible using equations (44), (45) to nd the density of units of all types in the

63
incipient phase at the moment of its appearance as well as the values of the
parameters at which this phase transition occurs.
The information on statistical characteristics of polymer sample contained

in generating function G (
s ) proves to be insucient to nd those regions of
the cloud-point curve hypersurface where the formation of spatially inhomogeneous phase begins. It was demonstrated[114], that to solve this problem
knowledge is required of generating functional

G [
s]=
P () Q (; [T ln
s (r)])
(122)
V
It has a meaning (up to the factor /V ) of the partition function Q of
an individual molecule with chemical structure located in the external

inhomogeneous eld H e (r) = T ln


s (r), which is averaged over molecular
structure distribution P () . The integration in expression (89) for Q, which
should be inserted into the right-hand part of formula (122), is performed
over the incipient phase volume V .

Density distribution
(r) within the nucleus of incipient phase can be
determined by solving a set of equations

(r) = s (r) G [
s ]/ s (r) ,

s (r) = exp

{[

(
)

]/

(
)

(123)

where instead of ordinary derivative entering into equation (44) here variational one appears.
To determine cloud-points the expression should be written down for
the alteration of the free energy F of the system due to the formation of
incipient phase nucleus against the background of its spatially homogeneous
state (principal phase), and then to equate F to zero. As a result the
condition is arrived at

F T G [ s ] dr [P (
(r)) P ] = 0
(124)
V

which in the case when the nucleus size is large as compared with the superstructure period reduces to the condition of the equality of pressure in the
incipient and principal phases.
Equations (123), (124) extend (44), (45) and reduce to these latter in a

particular case when distribution


(r) of densities of quasicomponents does
not depend upon r.

64

7
7.1

INTERFACE PROBLEM
Central Ideas of the Self-consistent Field Approximation

In thermodynamics of polymers of primary importance are the problems of


the description of the region between two incompatible homogeneous phases
as well as that of establishing of quantitative correlations between surface
tension and statistical characteristics of a polymer sample. Interface tension
is dened as excess free energy of a system per unit surface A which is a
functional of distributions of quasicomponent density.
To solve the interface problem they normally resort to various modications of the self-consistent eld method, key ideas of which are easily
understood if illustrated by the simplest example[145, 146, 147] of incompressible melt of two immiscible monodisperse homopolymers with number
of monomer units l1 and l2 . In a spirit of this method they initially consider
macromolecule Ml in the absence of volume interactions, but situated in
the external eld H (r). Position of a unit in polymer chain is conveniently
characterized by continuous variable 0 < s < l . Partition function of such
a chain whose initial and nal unit are at points r0 and r can be presented
as an integral over trajectories r (s)[115]
r

Q (l ; r0 r ; [H ]) =

Dr (s) exp

r0

l [
0

r 2
H (r (s))
ds
+

3a2
T

(125)

This function is obtainable from the solution of the diusion equation


Q
a2
H (r)
= Q
Q ,
l
6
T

Q (0; r 0 , r ; [H ]) = (r0 r)

(126)

The density of monomer units at point r is expressed in a simple manner

1
(r) =
ds q (s ; r ; [H ]) q (l s ; r ; [H ])
Q [H ]
l

(127)

through partition functions q and Q of a molecule with number of xed


ends equal to 1 or 0. These quantities can be found by means of single and

65
double integration of partition function (125) over coordinates of the terminal
units. The procedure of closing of a set of equations consists in choosing of
the expression for self-consistent elds H = T where has been dened
above (72). This closed set of equations in view of the fact that the eld H
depends only upon the sole coordinate x admits an exact analytical solution
for the prole of dimensionless concentrations = v ( = 1, 2)
[

1 (x) =

( )]

1
x
1 tanh
2
d

, 2 (x) =

( )]

1
x
1 + tanh
2
d

(128)

where d stands for the interface width. Its dependence on the Flory parameter as well as analogous dependence on it of surface tension
(

d = 2/61/2 a1/2

= 61/2 aT 1/2

(129)

can be arrived at proceeding from simple estimates[148]. To provide them


let us consider typical uctuation resulting in penetration of a chain segment involving z units M1 from the rst-monomer-rich phase into the secondmonomer-rich phase. The energy loss T z for the formation of typical uctuation is of order T , which yields the assessment z 1/. The width of the
interface is of order of Gaussian size of such a fragment d az 1/2 a1/2 .
When the interface energy is being evaluated it is taken to be a product of
two factors, the rst of which is the number of units of such penetrating segments dA and the second one is the energy of volume interactions of each
of them T . As a result the following estimate aT 1/2 is arrived at.
Analogous results can be also obtained within the framework of the Helfand
lattice model[149]. The solution of these problems turns out to be possible
if niteness of lengths of macromolecules is neglected. In order to elucidate
this approximation let us present the solution of equation (126) for function
Q (l; r0 r) as an expansion into a series
Q (l ; r0 r) =

1
exp (k l) k (r0) k (r)
l k=0

(130)

in terms of eigenfunctions of equation


[

a2
2 + H (r) k (r) = k k (r)
6

(131)

66
normalized to number l of units of a molecule. In the limit l the main
contribution into expansion (130) is made by the term with k = 0 which is the
only to be retained in the approximation of interest. As a result expressions
for partition function of a macromolecule and its monomer unit density will
read
Q exp (0 l) ,
(r) = 02 (r)
(132)
The condition of the applicability of this approximation is the presence of
nite gap between two eigenvalues 0 and 1 of the equation (131).

7.2

Heteropolymers with strong unit interactions

Quantitative description of the interface based on the ideas formulated in the


preceding Subsection has been performed in a series of works by Noolandi et
al.[150, 151, 152]. They considered multicomponent solutions of homo- and
block-copolymers in low-molecular solvent on the assumption of the system
incompressibility. Account of this restriction is carried out by the introduction of external eld w (72) whose value is found from the condition
that overall unit density is constant at any point of the space. Joint numerical solution of the diusion equations for every type of macromolecules
and equations for self-consistent elds (72) acting upon units enabled to the
authors[150, 151, 152] to obtain many interesting results.
The eect has been examined[150] of the addition of diblock-copolymer
(12a) upon the decrease of interfacial tension of immiscible blend of homopolymers Ml11 and Ml22 conditioned by the diminution of the entropy of mixing due to the displacement of molecules of these latter by block-copolymer
macromolecules from the region of interface. Localization in it of copolymer junctions results in decrease of the repulsion energy of dierent type
units. This is responsible for the eect mentioned which is known to be
most strongly pronounced in blends of highly incompatible polymers. Another important factor, conducive to the reinforcement of the exclusion of
homopolymers from the interface region as the molecular weight of the blockcopolymer is increased, is the growth of the enthalpy of mixing which is due
to the penetration of -th type blocks into region occupied by units of -th
homopolymer.
Curious consequence of theoretical consideration is the fact that interfacial tension is predicted to vanish for some concentration of block-copolymer.

67
To explain this eect it has been suggested[150] that only some of the blockcopolymer macromolecules nd it energetically favorable to settle at the interface while the remainder is randomly dispersed in the bulk phases.
Of particular interest is the question about the inuence of a solvent addition upon properties of a system considered in paper[150]. This question has
been answered for the case when the solvent is athermic[151]. In this case
the interface width in the limit of innitely long chains of homopolymers
grows exponentially while the surface tension decreases drastically as molecular weight of a copolymer increases because localization of its molecules at
the interface region. Special attention merits work[152] where general consideration of interface problem was carried out for a blend containing two
homopolymers, arbitrary block-copolymer and solvent.

7.3

Heteropolymers with weak unit interactions

Previous Subsection has been devoted to the discussion of interface problem for blends of strongly incompatible polymers with sharply pronounced
boundary between phases. Under changes of external parameters of a system
as it approaches the critical point this boundary can spread because of the
decay of volume interactions between units resulting also in the diminution of
the distinctions in composition of coexisting phases. Theoretical treatment of
the interface of such relatively compatible polymers can be performed within
the framework of the Landau theory (see Subsubsection 8.2.1).
In paper[153] there has been considered the melt of two monodisperse
homopolymers Ml1 and Ml2 with the same degree of polymerization l and
l/2 l/2
symmetric block-copolymer M1 M2 when volume fractions of these three
components are 01 , 02 and 03 , respectively. In the vicinity of this system
critical point, which is the case at 01 = 02 = 1/l, the Landau free energy
has been presented as the expansion in powers of two independent small parameters of order (1 2 ) and (3 03 ). Then ordinary theoretical analysis has been performed in order to derive expressions for surface tension
. This simple expression looks like a dierence of two terms corresponding to two dierent physical mechanisms which result in the decrease
of interface tension when copolymer is added into a system. The rst term,
, reects corresponding diminution of fraction of units M1 and M2 under
the increase of 3 in M1 -monomer-rich and M2 -monomer-rich phases whereas
the second term, , represents the preferential location of copolymers at the

68
interface. The contribution increases with growth of incompatibility of
polymers, remaining nevertheless small as compared to at the vicinity of
critical point.

8
8.1

THEORY OF MICROPHASE SEPARATION


Introduction

Melts and solutions of block-copolymers as well as blends containing their


macromolecules can be often found in thermodynamically stable mesomorphic states. These latter are characterized by inhomogeneous density distribution of monomer units in space. Basic mesomorphic states of polymer
systems incapable of forming neither regular nor liquid crystals are micellar
solutions and spatially periodic structures. Such superstructures provided
they are not formed in close vicinity to critical points consist of blocks localized within microdomains periodically distributed in space. Usually this
mesophase is one of three possibilities: (i) alternating layers (lamellae), (ii)
cylinders on a two-dimensional lattice, (iii) spheres on a three-dimensional
lattice. Which particular of these structures is the case depends upon statistical characteristics of the polymer sample as well as upon observing conditions. Voluminous literature is devoted to the discussion of microdomain
formation[154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165] both
from theoretical and experimental standpoint.
Considerable body of experimental material is available up to now which
appears to be sucient for the understanding of fundamental peculiarities
of this phenomena. However, the elaboration of general microscopic theory still remains a challenge for statistical thermodynamics. Such a theory,
if available, is believed to be able on the one hand to provide a recipe for
constructing phase diagrams while on the other hand to predict the morphology of the phase-separated superstructure, i.e. the size, shape, and spatial
distribution of microdomains. In the absence of a unied theory two limit
regimes of the formation of spatially periodic structures are generally being
considered.
In the rst regime of strong incompatibility between dierent-type units
of molecules of binary block-copolymer, microdomains are composed mainly

69
of units of the same type being separated by the interfaces very thin in
comparison with their sizes. The area of applicability of this limit, so called
strong segregation regime (SSR), covers polymer systems situated far enough
from critical points of their phase diagrams. Conversely, if phase transition
occurs under conditions close to critical ones, the superstructure formation
takes place under weak segregation regime (WSR). Here the distinctions in
concentrations of monomer units inside dierent type domains are known to
be small and no sharp interface exists between them. Under such a regime the
size of macromolecule is only slightly perturbed by volume interactions unlike
for the SSR where block-copolymer chains are strongly extended, having the
gyration radius RG noticeably exceeding that of unperturbed Gaussian coil.
Theoretical approaches normally employed in statistical thermodynamics of superstructures, formed under two limit regimes, generally speaking
dier markedly. Major success has been achieved here in the case of the
WSR where starting with basic study by Leibler[130] extensive use has been
made of the Landau theory. Its mathematical apparatus has been thoroughly devised in modern theoretical physics, so once a particular problem
of the description of equilibrium properties of a heteropolymer system is
formulated in terms of the Landau theory further solution of this problem
will be a routine procedure. As for the SSR, despite the fact that the attempts to develop here an approach, as universal as the Landau theory in the
case of the WSR, have not met with success so far, nevertheless, substantial
progress has been made in describing of superstructures within the framework of the mean-eld approximation. Key idea of the latter, going back
to pioneer works by Edwards[166] and Helfand[167] which deal with thermodynamics of binary block-copolymers, resides in derivation under certain
assumptions of self-consistent equations enabling one to nd spatial distribution of concentrations of dierent type monomer units. These nonlinear equations remind those describing nonstationary diusion within two-component
gas with molecules capable of entering into chemical reactions. The role of
the time, appearing as independent variable in the latter equations, will be
performed in the former ones by the distance along polymer chain, reckoned
from its end. Mathematical apparatus used in the theory to describe superstructures forming in the SSR resembles that employed when solving the
interface problem.

70

8.2
8.2.1

Weak Segregation Regime


Introduction to the Landau theory

When considering the phase transitions it is pertinent to introduce the parameter of order equal to zero in the principal phase and distinct from it in
the incipient one. For the solutions of low-molecular compounds as one can
chose, for instance, the dierence in densities of molecules in both phases. If
these solutions contain m types of particles, the system will be dened by a
set of m such variables which can be thought of as components of vector

being the m-component parameter of order. The latter can be either the
constant or the function of coordinates depending upon the incipient phase
symmetry. If it is spatially homogeneous, we are dealing with the rst case,
conversely, if periodical structure forms, the second case takes place. The
role of the parameter of order can be performed by other variables as well
which are known to turn into zero within the principal phase and are distinct
from it within the incipient one.
When the phase transition occurs in the vicinity of critical point, the
parameter of order has a small value and thus the system free energy can be
expanded into the power series of this parameter. The latter will be either
ordinary series or functional one depending on whether the incipient phase
is spatially homogeneous or not.
Under thermodynamic equilibrium the free energy, envisaged as a function of the parameter of order, attains its minimum. This condition permits

one to nd the value of parameter , and, consequently, the dependence of


free energy on the coecients of expansion. Finding of explicit expressions
for them is, strictly speaking, beyond of the scope of the traditional Landau
theory which is phenomenological one. In order to specify these coecients
it is necessary to take into account special features of the particular system
within the framework of corresponding microscopic theory. Peculiarities of
this procedure when applied for heteropolymer systems will be discussed in
Subsubsection to follow while here we will just outline briey (supposing for
simplicity sake the parameter of order to be one-component) central ideas
of the Landau approach, indispensable for the understanding of the results
arrived at in terms of the weak segregation theory.
Functional expansion of the free energy F into power series of the parameter of order (r) is normally written down going to its Fourier-transform

71
e (q). Here the expansion coecients which are the[functions
of wave vec]
e
tors become constants, provided the functional F is considered, as is
customary, on the space of functions of particular form
n
e (q) =

[ (q Q ) + (q + Q )]
k

(133)

k=1

where in square brackets there appears the sum of two the Dirac[ delta]
functions. Upon substitution of this expression into functional F e the
latter reduces to the ordinary Taylor series in powers of parameter n with
coecients controlled by n vectors {Qk } which are the reciprocal lattice primitive translations (RLPT) of the superstructure formed. In terms of this approximation all these vectors are generally taken to have the same modulus,
equal to the value of the wave vector q where the loss of equilibrium of
spatially homogeneous state
{
}happens at critical point.
(k)
Among coecients an
of the expansion of free energy F [n ] only
that turns out to be independent on n which appears in front of n2 . This
coecient along with a(3)
n is known to vanish at critical point (not on the
spinodal !). Provided the coecient a(4)
n is positive when expanding the
function F (n ) they normally restrict themselves to the rst three items
Fn
3
(4) 4
= a(2) n2 + a(3)
n n + an n
TM

(134)

omitting the other small terms. Having minimized this expression with respect to n it is possible to nd its value whose substitution into (134) yields
that of free energy Fnmin reckoned from its value in spatially homogeneous
phase. The condition of the transition of a system from this phase to superstructure is Fnmin = 0. Analogous condition, but for the transition between
two mesophases of dierent symmetry is obtainable as a result of equating
of values Fnmin of these phases. In order to nd the condition for the loss
of stability of any mesophase in terms of approximation in question it is necessary to substitute the value n obtained by minimization of Fn into the
equation 2 Fn /n2 = 0.
When the coecient a(4)
n is either comparable in value to dropped terms
of the expansion (134) or negative, the next items should be retained in this
expansion.

72
Elements of the Landau theory introduced here can be extended to the
case when the parameter of order has more than one component. This circumstance appears to be rather promising when dealing with multicomponent systems.
8.2.2

Peculiarities of the Landau theory as applied to polymer


systems

The number of components of the parameter of order coincides with that


of quasicomponents of a polymer system or is unity less if the system is
presumed to be incompressible. The circumstance that monomer units are
linked to form polymer chains and, consequently, are uncapable of moving
independently in space is responsible for the appearance of the fundamental
peculiarity. The latter consists in the impossibility (with the exception of
degenerate model systems (see Subsection 5.2)) to present the free energy
of multiphase polymer system, considered as a function of parameters of
order in dierent phases, as a sum of additive contributions from separate
phases. This results in the appearance in the last item in the right-hand
part of expression (134) as well as in all subsequent terms of the free energy
expansion of the items which are dependent upon the parameters of order
in dierent phases. Such a nonlocality eect, inconceivable in principle in
solutions of low-molecular compounds, has been pointed out for the rst time
in the study[168].
In this Section general algorithm[114] will[ be] formulated allowing one

to nd the coecient of the expansion F into power series of the

component of the parameter of order . As deviations can be chosen


from average values either of the unit densities
(r) (r)
= (r)

dr
(r)
V

(135)

or of the components of the eld H conjugate to these densities


(r) H(r) = H (r) H

H =

dr
H (r)
V

(136)

When the Landau expansion is performed with respect to spatially homogeneous state its coecients are conveniently found in momentum rather than

73
co-ordinate representation where their Fourier-transforms (q) appear instead of (r).
In order to reveal specic features inherent to the solutions and blends
with the participation of macromolecules it will suce to consider only the
system of Chemical
Bonds, i.e. to nd coecients of the expansion of func[
]

tionals FCB H and SL [


] occurring in the random eld formalism (see
Subsubsection (6.1.1)) and the density functional formalism (see Subsubsection (6.1.2)), respectively
[
]

[
]

FCB H = FCB H +

p=2

(1/T )p1
p!

ge
i=1 {qi } {i }

(p)
1 p

f (q ) H
f (q )
(q1 qp ) H
1
1
p
p

(137)
[
]

SL [
] = SL

1
p!


p=2

(138)

p
i=1 {qi } {i }

e(p)
1 p

(q1 qp ) e1 (q1 ) ep (qp )

The summation over momenta {qi } in these expressions is over all vectors of
the RLPT Q1 , , Qn of the superstructure forming in the incipient phase.
Let us discuss rst the expansion (137) whose coecients have a meaning
of the Fourier-transforms of functions of the response of polymer system to

inhomogeneous external eld H (r). The rst two of them coincide with
e (2) and
e (3) while for the components of tensor g
e (4)
components of tensors

74
following expressions take place
ge(4)1 2 3 4 (q1 q2 q3 q4 ) =
e (4)

1 2 3 4 (q1 q2 q3 q4 ) (q1 + q2 + q3 + q4 )
[

(4)

e | (q1 q2 |q3 q4 ) (q1 + q2 ) (q3 + q4 ) +

1 2 3 4

(139)

(4)

e | (q1 q3 |q2 q4 ) (q1 + q3 ) (q2 + q4 ) +

1 3 2 4
(4)

e | (q1 q4 |q2 q3 ) (q1 + q4 ) (q2 + q3 )]

1 4 2 3
e (4) have been determined above (118). In
where the components of tensor
e (4) comprises
parallel with these local functions of the response, the tensor g
also nonlocal terms described by the expressions in square brackets (139)
which broke the locality of the response. The density distribution at given
point, as it follows from formula (139), was found to depend upon the values
of external eld not only in the local vicinity of this point but also in overall
volume of a system.
General algorithm enabling one to write out the expression for the come (p) ({qi }) of any order involves three successive
ponents of symmetric tensor g
stages, each consisting in expressing of the tensor through the other one.
This procedure can be schematically presented as follows
e W
e
e
g
(1)

(2)

(3)

(140)

e (p) , according to
The sum of wave vectors entering as arguments into tensor g
its denition, is zero at any p.
Stage 1. The set of p points, each being characterized by index and
wave vector (momentum) q, is partitioned into groups in such a way
that any -th group comprises the number of points p 2 whose overall
e (p) an
momentum equals zero. Every group makes into the expression for g

75
additive contribution which up to the factor (1)1 will read

(p)

e |

|p

q1 |
| {z }

| qp =
| {z }

p1 terms

=1

P ()

p terms

exp i

(141)

q r
j j

j=1

where the designations of the formula (118) are used. Here the product is over
all groups of the partitioning in hand, so that summing in the exponent is
made only over the points belonging to one and the same group. Vertical bars
appearing in indices and arguments in expression (141) separate the points
which pertain to dierent groups. Inasmuch as all points of any group are
taken to be indistinguishable, the values of tensor (141) components remain
the same under any transposition (i , qi ) and (j , qj ) inside this group.
Stage 2. Let us partition the points of every group mentioned in accore
dance with the algorithm introduced in Subsection 6.4 to relate tensors
e
and . As a result of such a procedure any -th group will contain number
p of points lesser than p . To each such partition there will correspond the
(r)

contribution e (r = p1 + + p - represents overall number of points in


e (p) . Summing of the contributions
a reduced system) to the expression for
from all possible partitions of such a kind yields the unknown value of come (r) the designations used in
e (p) . Retaining for
ponent (141) of this tensor
formula (141) and proceeding from the above formulated algorithm we will
get in a particular case p = 4 the simple expression
e | (q1 q2 | q3 q4 ) = e | (q1 q2 | q3 q4 )

1 2 3 4
1 2 3 4
(4)

(4)

(r)

(142)

Stage 3. To have e expressed through W(r) it can be resorted to the


relationships derivable by virtue of the algorithm of Subsection 5.4 upon
minor modication of these latter. In line with this algorithm tensor-function
e of the sum of momenta of dierent
W(r) involves as arguments the functions
points which belong, generally speaking, to dierent groups. The set of all
items in this sum is conveniently divided into subsets q1 , , q , , each
containing the momenta of points of only one group. The above modication

76
e by the product of factors
e (q )taken
consists in replacing of each function

over all groups of type .


So, having resorted to the above procedure (140) it becomes possible
to express the coecients of the expansion (137) through the generating
functions of chemical correlators (119). The nal expression for coecients
of the expansion of free energy (94), appearing in the Landau theory with
the parameter of order (136), for arbitrary polymer mixture will read
(p)
e (p)
e
G
(q1 qp ) Be (p)
(q1 + + qp )
1 p (q1 qp ) = g
1 p
1 p

(143)

e (p) components have been determined earlier (86). The


where the tensor B
fact that nonlocal terms in coecients (143) appear only beginning with
the forth-order terms with respect to the parameter of order comes as no
surprise. This is concerned with the necessity when nding free energy of
polydisperse polymer sample to average over the distribution P () not the
very partition function of separate macromolecules, but its logarithm. If to

compare the expansion in power series of the eld H of this functional


with that of generating functional of separate molecules (122), which can
be obtained by averaging of their partition function, it will be evident, that
the distinctions in coecients commence just from the forth-order terms.
These distinctions are nothing but nonlocal items (141), which are absent in
formulae generally applied for the construction of phase diagram provided
the transitions between mesophases are not considered.
In vast majority of studies on the weak segregation regime use has been
made of the Landau theory with the parameter of order (135) corresponding
to the density functional formalism (see Subsubsection (6.1.2)). Coecients

of the expansion of this functional into power series of


in a simple fashion
(see (85))
(

e (p)
e(p)

(q1 qp ) + f(p)
T (q1 + + qp )
1 p =
1 p
1 p

(144)

are connected with coecients of analogous expansion (138) of the Lifshitz


entropy dened by formulae (99), (100). Substituting into the rst of them
the expansion (137) and solving the equation obtained by means of iterations

one can determine the dependence of eld H (r) on density


(r) as an expansion into power series of the parameter of order (135) Fourier-transform.
Insertion of this solution into formula (100) leads to the expansion (138)

77
where the tensor e (p) components are determined using the diagram teche (p) are substituted for
nique discussed in Subsection 6.4 provided tensors
e (p) . Apart from this distinction there is another one residing in the fact that
g
summing over momenta of all lines is made now under the condition q = 0.
Having unied the expansion (138) with the contribution of volume interactions one will arrive in accordance with (97) at the expansion of functional

F [
] /T with coecients (144).
Fundamental peculiarity of the Landau expansion of free energy in the
case of polydisperse polymers consists in the presence of nonlocal terms in
e (p) , caused by the nonlocality of the Lifshitz entropy. The role
coecients
of these terms is conveniently illustrated for the derivation of tensor e (4)
via diagram technique generalizing that introduced in the foregoing (Subsections 5.4 and 6.4). To have the nonlocal contribution to e (4) separated it is
pertinent to introduce new diagram elements shown in Figure 6a.
e (4)

1 2 3 4

I
(q1 q2 q3 q4 )

II
III
1
e 1 2 (0) (q)
(q1 q2 |q3 q4 )

(4)
e |

1 2 3 4

(145)

e (4) has a form


In terms of this diagram technique the expression (139) for g
(4)
presented in Figure 6b. As for the expression for e , its diagram representation in Figure 6c diers from the above-presented traditional one (Figure
3) by terms in square brackets[168]. They are mutually cancelled in the case
of low-molecular weight solutions whereas for polymer systems these terms
annihilate each other only under the model assumption on a total absence of
polydispersity. It was shown[114] that in this case such an annihilation takes
place for all coecients e (p) of the Landau expansion.
The above introduced the Landau expansion has been formulated in terms
of the small scale mean-eld theory. When characteristic scale of the change
of monomer unit densities is large as compared to the size of individual
molecule this expansion is noticeably simplied reducing to the form, appropriate for the description in terms of the large scale mean-eld theory of
multiphase polymer systems situated in the vicinity of a critical point. The
coecients of such an expansion can be found in two independent manners.
The rst of them resides in going from free energy F (34) to new thermodynamic potential
m

Fb = F +

=1 i=1

hi

(
)

(146)

78
(
)
minimization of which with respect to C i l leads to the equations whose
( )
( )

solution looks like (37), (38) where is replaced by i + hi . In


i
i
order to determine the dependence of the elds
(
) h on densities one should

substitute the expressions obtained for C i l into stoichiometric condition


(
)
(35). By virtue of such an exclusion of concentrations C i l and elds
hi it becomes possible to nd the
for thermodynamic potential
}
{ expression

Fb as a function of the densities i of monomer units in dierent phases

(i = 1, . . . , r) and their volumes {V i }. This potential is equal to the functional of free energy F taken on stepwise distribution of monomer unit densities describing the state of a system involving several spatially({
homogeneous
)

i }
b
, {V i }
phases. For low-molecular weight systems the function F
(

i i )
b
equals the sum of contributions F , V of all phases provided the surface eects are ignored. This property of additivity, as is easily shown using
the above described procedure of derivation of the expression for Fb , is violated for real polymer systems. Despite the fact that such a procedure is
possible to perform in principle it proves to be inecient for thermodynamic
calculations of real polymer systems unless the latter is in the proximity of
critical point where monomer unit densities in each phase dier slightly from
their average values. In this case expanding of F into the power series of
these deviations will result in the sought for Landau expansion whose
coecients are, generally speaking in a nonlinear way dependent on volume
fractions y i of dierent phases.
Another, more straightforward fashion of derivation of these Landau expansion coecients suggests the consideration of the functional (138) on the
space of stepwise distributions (r) with macroscopic linear sizes Li of
the phases. Because characteristic values of the modulus of the wave vectors
q i 1/Li on such a distribution are small as compared to the inverse size of
a macromolecule, the factors in front of -functions in expressions for tensors
(p) components may be taken outside a symbol of the sums (integrals) provided the values of all momenta are put zero. Hence any sum (integral) over
momenta is divided into the product of sums (integrals) of the individual
groups described above, each -th group containing p terms. As a result
the expressions will be arrived at for the Landau expansion coecients as

79
polynomials in volume fractions y i of dierent phases. To each coecient
Lp the polynomial corresponds whose degree is equal to the integer part of a
number p/2 while the coecients of this polynomial depend on the statistical
moments of the distribution for size and composition (2) of macromolecules
of the sample.
8.2.3

Main results

Phase diagram of the simplest reference system, i.e. the melt of monodisperse
diblock-copolymers (12a) has been constructed in work[130]. In such a system
when the fraction of units M1 is f = 1/2 there is a critical point in the
vicinity of which the Landau expansion of free energy is possible to apply.
Since for the melt the incompressibility condition holds, it is sucient for
such a system to introduce only one dimensionless parameter of order (r)
1 (r) = [f + (r)] ,

2 (r) = [1 f (r)]

(147)

Nearby the spinodal at nite wave vector q = q , when s s , major


contribution to the free energy functional is from Fourier-transform with this
wave vector and thus, the parameter of order (r) can be chosen as
n

2
(r) = n
cos Qk r
n k=1

(148)

The summing in expression (148) is over vectors Qk of the RLPT of the


superstructure.
The substitution of the parameter of order (148) into the expression (147)
for the densities of monomer units permits one to write down free energy
density, reckoned from spatially homogeneous state as follows
Fn l/T V = 2 n2 n n3 + n n4 ,

l (s )

(149)

The expansion coecients n and n are controlled both by the symmetry of


the superstructure and the parameter f ; coecient n vanishes for symmetric
polymers with f = 1/2 while for lamellar structure this is identically equal
to zero. Equilibrium value of the amplitude n is found by minimization of
free energy (149)
F1 l
= 2 /1 ,
TV

)
Fn l
27 n4 (
3
1
+

=
n (1 3n ) ,
TV
(64)2 n3

(150)

80
where at n > 1

n = 1 64n /9n2 l

]1/2

(151)

Conditions of a phase transition are obtainable from the comparison of free


energies of dierent microstructures and that of homogeneous state. As a
result phase diagram has been constructed Figure 7 involving the regions of
dierent microdomain structures for a block-copolymer melt, the transitions
between which at f = 1/2 are the rst-order phase transitions.
In paper[169] the properties have been studied of the above reference system upon the addition in it of the solvent good for both quasicomponents M1
and M2 . To provide the description of such a system along with the parameter of order (148) the second one, s (r), should be introduced, proportional
to the solvent density. In the vicinity of the spinodal it looks as
2s
s (r) =
n

cos (Q

+ Ql ) r

(152)

k=1 l=1

For lamellar structure this parameter of order is dened by the wave vector
2q which corresponds to the displacement of the solvent from the regions
enriched by monomer units to interphase region. The amplitude s of solvent
inhomogeneities is a factor l1 less than the n . Due to the smallness of this
parameter it appears possible to neglect the contribution of the solvent into
free energy of a system.
The main qualitative eect taking place in concentrated solution
(1 << 1) of the asymmetric polymers ( f = 1/2) is a splitting of lines
of phase transitions shown in Figure 7 into the coexistence regions of corresponding phases. The widths of these regions, however, are too small to
be observed experimentally. For symmetric polymers with f = 1/2 the twophase region has been obtained[169] at low temperatures.
The mixture of block-copolymers, homopolymers and solvent has been
studied in works[170, 171]. Numerous phase diagrams constructed in these
papers show the presence of eutectic points and mesophase formation induced
by homopolymer. Unfortunately, the possibility has been taken into consideration of the formation of lamellar phase only, for which the cubic term
in the Landau expansion is absent in principle. That is because the phase
transitions have been found to be of the second-order ones which contradicts
to the above-discussed results.

81
Theoretical analysis[172] of copolymer melts of more complicated architecture, such as asymmetric triblock- copolymer (12b), brought the authors
to the conclusion that the critical point is the case only at particular value
f = fc of a fraction of units M1 . By virtue of the Landau expansion in
the vicinity of critical point the phase diagram of the system has been constructed (see Figure 8). From the comparison of Figure 8 with Figure 7 for
reference system it is evident, that as the parameter growths the phase
diagram becomes progressively more asymmetric as well as the widths of
regions 2 and 3 are found to increase.
Among the assumptions employed in the above-discussed studies on the
theory of weak segregation regime the most vulnerable one is the approximation neglecting high-order harmonics of the parameter of order (r)
(148). Their inuence on the value of superstructure wave vector q has
been elucidated[173]. Amplitudes of high-order harmonics of the Fourier series of the (r) as well as equilibrium value x = q 2 R2 can be found from
the minimum of the free energy functional. For lamellar structure numerical calculations predict linear dependence x 0.37 (s ) which is in a
good agreement with experimental results obtained by Owens et al.[174] for
styrene-isoprene copolymer with f = 0.4. Analogous results have been also
arrived at for the phases of other morphology.
The account of only the rst harmonics (148) provides no possibility to
dierentiate superstructures with the same rst Bragg reections pertaining to dierent spatial symmetry groups. Among them are 3-dimensional
structures bcc (body-centered-cubic) and bdd (bicontinuous-double-diamond)
as well as 2-dimensional triangular structure and 3-dimensional hexagonal
structure. Such alternative structures for the reference system have been
investigated[175] with due regard for the contribution from high-order harmonics. The phase diagram obtained here diers essentially from that presented in Figure 7 only within the region of values 0.47 < f < 0.52, when
three-dimensional hexagonal structure has free energy lesser than that of
lamellar, square, triangular, bcc and a simple cubic structures.
Dobrynin and Yerukhimovich[176] carried out theoretical analysis of microphase separation in melts of binary copolymers of complicated architecture. By the numerical calculation the phase diagrams have been constructed
in the space of parameters describing the chemical structure and temperature
for asymmetric triblock- and polyblock- copolymers.
The allowance made for polydispersity can change materially phase dia-

82
grams of polymer systems. When polydispersity is weak enough each aboveconsidered line of phase transitions on diagrams versus f expands to turn
into the area of nite width where both phases coexist[168]. If polydispersity
is strongly pronounced new qualitative eects appear which are dened by
nonlocal terms of the expansion of the Landau free energy functional. The
disregard of the latter when constructing the phase diagram brought the
authors of the paper[177] to the incorrect results.
The simplest polydisperse system whose behavior was examined[178] is
a melt of statistically-random heteropolymer with fraction f = 1/2 of units
M1 . On the assumption that the Flory parameter is dependent upon wave
vector q, the microdomain structure in such a system was shown to appear
by the third-order continuous phase-transition. Wave vector modulus q
(s )1/2 of such a superstructure in the vicinity of the spinodal at the
zeroth wave vector ( = s ) is strongly controlled by the Flory parameter
. The appearance of this superstructure is conditioned by the presence of
randomly positioned along the chain M1 -rich and M2 -rich fragments which
facilitate segregation.
This work is of prime importance for thermodynamic theory of heteropolymers since here for the rst time the expression has been derived
for the Landau free energy, comprising nonlocal terms owing their existence
to polydispersity of a sample. The account of these terms provides a possibility to describe the superstructure whose period is strongly governed by
temperature. This eect is due to the dependence (q) introduced by hand
and is known to be absent in commonly accepted thermodynamic models
of polymer melts where = const. This sound paper would be even more
valid but for the restrictions, implemented by the incompressibility approximation, which do not permit adequate description of experimental observed
macrophase separation in these systems. In order to describe the superstructures considered in the paper[178] there is no need to use an arbitrary
assumption about the dependence of the Flory-Huggins parameter on the
wave vector q. It has been illustrated by authors[113, 114] who, proceeding
from general approach (see Subsection 6.1) characterizing thermodynamics
of polymer systems with arbitrary chemical structure, examined in detail binary markovian copolymers. The period of such a structure in the proximity
of the point = s of its appearance is large as compared to the character1/2
istic spatial size aNb of one block whose average number of monomer units
is Nb . For simplicity sake the case is considered of large number of blocks

83
which corresponds to the innite length limit of a macromolecule. Substituting expressions (119) for generating functions of chemical correlators and
relationships (147), (148) for the parameter of order (r) into general the
Landau expansion for free energy density we will get
(

Fn Nb /T V = 2Nb s + a2 Q2 n2 n n3 + n n4 + n4 /a2 Nb Q2 (153)


where coecients and of the order of unity depend rather on the parameter f than on the structure symmetry.
The most essential qualitative distinction of expression (153) from that
(149), which refers to monodisperse systems, is the presence of the additional
last item in its right-hand part. The value of this item, representing the
contribution from non-local eects, was found to increase as Q2 with the
growth of the superstructure wave vector Q. This is due to the decrease
of probability to nd the density uctuations with spatial scale R Q1
1/2
which is large enough in comparison with the spatial size of block aNb .
Such uctuations result from local redistribution of macromolecules with the
excess l R2 /a2 of units of given type = 1, 2 at the scale R Q1 from
the interphase areas into internal ones of domains of corresponding type .
Losses in the entropy of mixing induced by such a redistribution of chains are
proportional to the number l of the excess units and are dened by nonlocal
term of the free energy (153).
Obviously, the above discussed physical mechanism, related to polydispersity of the system of interest, is absent in the case of homodisperse polymer.
Equilibrium value of a superstructure wave vector is obtainable by minimization of free
energy (153) with respect to the value of Q which leads
2
2
to a Nb Q = /2 |n |. Substituting of this expression into formula (153)
yields the free energy functional which depends exclusively upon the superstructure amplitude n

Fn Nb /T V = 2Nb (s ) n2 + 2 2 |n |3 n n3 + n n4

(154)

Here the coecient in front of the cubic term is positive until value |n | is
not too large. Since the value n (f 1/2) then, at |f 1/2| < c 1, a
nontrivial solution n = 0 minimizing the free energy (154) exists only under
the condition > s of the loss of positive deniteness by the quadratic form
of this functional and remains continuous at the third-order phase transition
point.

84
Equilibrium value of the superstructure amplitude can be found, in turn,
by minimizing of free energy (154) with respect to n . Neglecting the last
item in expression (154) which is small in comparison with the remaining
terms in the vicinity of the transition point it is possible to nd the dependence of the superstructure parameters on the Flory parameter
|n | Nb ( s ) ,

Q a1 ( s )1/2

(155)

More detailed analysis shows that absolute minimum (154) is attainable on


the solution with bcc symmetry (n = 6). The expression (148) with allowance
for the dependence (155) for the parameter of order (r) becomes asymptotically exact within the limit f 1/2. In the case f = 1/2 this involves
also high-order harmonics whose amplitude decreases as parameter |f 1/2|
does.
When s s 1/Nb in the expansions (153) and (154) account
should be taken for the forth- and higher-order terms with respect to n . The
amplitude of such a superstructure is of the order of overall density n 1
1/2
while its period is comparable with a block spatial size aNb . To describe
such a superstructure the higher-order terms in the Landau expansion of the
free energy should be taken into account.

8.3

Strong Segregation Regime

8.3.1

The Helfand-Edwards Method

Vast majority of theoretical papers, devoted to periodic superstructures in


systems being far from critical point, performed in terms of the self-consistent
eld approximation. Mathematical apparatus employed in these papers is
similar to that which was used for the solution of the interface problem (see
Section 7). However, the distinction exists, since the solution of the diusion equation (126) is searched in periodic self-consistent eld H (r). This
solution can exhibit various symmetry corresponding to the type of the superstructure crystal lattice. Finding of its period D constitutes one of crucial
problems of a theory. The value D of given superstructure is normally determined by minimization of the free energy. Comparing equilibrium value of
the free energy of dierent phases it is possible to determine the values of parameters under which corresponding phase transitions occur. For the solution

85
of this problem they normally calculate periodic proles of quasicomponent
concentrations necessary for realizing the self-consistence procedure.
The approach outlined in Subsection 7.1 is easily extended to the case of
block-copolymers. Thus, for instance, for macromolecules (12a) we have
[
1
]
Q12 H =
dr q (l1 ; r ; [H1 ]) q (l2 ; r ; [H2 ])
V

(156)

q (l ; r ; [H ]) =

dr0 Q (l ; r0 , r ; [H ])

(157)

where the integration in (156) is over coordinate r of the junction point while
in (157) it is over coordinates of the end points of such a macromolecule.
Its monomer unit densities are obtainable by dierentiation of logarithm of
partition function (156) with respect to external elds H (r)
1 (r) =

Q1
12

l1
[
]

ds q (s ; r ; [H1 ])

(158)

dr Q (l1 s ; r, r ; [H1 ]) q (l2 ; r ; [H2 ])

The density 2 (r) of units M2 is of similar form. Expressions (156), (158)


completely dene the free energy of the system in hand within self-consistent
eld approximation. Thus the problem reduces to the solution of diusion
equation (126) for each block inuenced by given external elds H (r) and
to nding of self-consistent values of these latter.
When elaborating a theory of microdomain formation under the SSR
(D >> d) it is customary to assume density proles within boundary region
between two domains to be similar to those found for the solution of the
interface problem. In series of papers by Helfand and Wasserman[179]-[184]
lamellar structure formation in melts of diblock- and triblock- copolymers
(12a) and (12b) has been studied. Period D of such a structure is related to
the widths D1 and D2 of domains of units M1 and M2 of the same volume in
a simple way
D = D1 + D2
(159)
where neglecting the interface width one can calculate values D1 and D2 by
formulae
D2 2
D
D1 1
=
=
(160)
l1
l2
l1 / 1 + l2 / 2

86
The theory of the strong segregation regime (SSR) developed in the abovereferred studies is based on approximation at which copolymer chain partition
function is factorable
Q12 Q1 Q2 d/D
(161)
becoming a product of partition functions Q of each block and the interface
contribution d/D . The system free energy with the allowance for expressions
(160) and (161) can be presented as follows
F
2
=
V T
TD

l1
l2
D
(l1 / 1 ) (l2 / 2 )
+
+ln
ln Q1 ln Q2
(162)
1 2
d
(l1 / 1 ) + (l2 / 2 )

The rst term represents the contribution of the interface, the rst logarithmic term describes the entropy loss due to the junction localization on it. The
sum of the last three items in the right-hand part of equation (162) equals
free energy of mixture of two homopolymers Ml11 and Ml22 . Here logarithmic
terms provide an entropy contribution while the last one takes into account
the contribution of volume interactions.
The dependence of values Q on domain sizes D has been arrived at
by means of numerical solution of the diusion equations (126). For the
diblock-copolymer (12a) this dependence can be successfully interpolated by
formula
(
)2.5
ln Q const D /al1/2
(163)
Substitution of (163) into (162) with due regard for relationships (160)
enables one to nd explicit dependence of free energy F (D) on the superstructure period D which can be obtained by subsequent minimizing of the
function F (D). In particular, at l = 37 the expression D 3.1al1/2 for the
period was found.
This theory has been extended to the description of domains of either
spheric[185] or cylindric[186] form.
Further development of the theory for the SSR[187, 188] was associated
with the improvement of numerical calculation method of solution of the
diusion equations (126). As a result the authors of the paper[187] managed
with no recourse to the approximation of factorization (161) to describe not
only the SSR for the solution of diblock-copolymer in athermic solvent, but
also to follow the transition from the WSR to the SSR. For the superstructure
period in these two limit cases there have been obtained following expression
D a p lq rc

(164)

87
where c is copolymer volume fraction. For the WSR exponents in (164)
coincide with those found in terms of the Landau theory
p 1/3,

q 0.8,

r 0.4

(165)

By contrast, for the SSR they prove to be equal


p 0.2,
8.3.2

q 2/3,

r 0.22

(166)

Modification of the Landau theory

At the rst glance it appears impossible to describe the SSR in terms of the
Landau theory in view of the fact that the distinction in densities (r) is
not small for this regime as compared to the, . Besides the proles (r)
are here very anharmonic, so that all coecients of the expansion of this
periodic function into Fourier series are of the same order.
The above diculties are possible to surmount provided correct account is
taken of normally dropped terms of the expansion of the Landau free energy
of the parameter of order (r)[189, 190, 191]. Central idea of these papers
consists in separating of the short-range F S and long-range F L contributions
into the free energy functional
F [] = F S [] + F L []

(167)

The substantiation of such a separation of contributions lays with the existence of a hierarchy of spatial scales of the domains forming under the SSR.
The rst such a scale, d, dening the interface width proves to be small in
comparison with the Gaussian block size. Functional F S describes all eects,
responsible for a system properties on such small scales. This functional has a
sharp minimum on congurations (r), dening the density prole between
domains and its value on such congurations
F S [] = A

(168)

is expressed through interfacial tension and total area of the interface A.


The value of , entering into theoretical expressions as a phenomenological
parameter, has not been calculated in these works.
The second scale, D, dening characteristic size of domains appear to
be large as compared to the Gaussian block size because of strong extension of chains inside domains. The contribution to the free energy due to

88
the alteration (r) on this scale is described by functional F L . In view of
the existence of hierarchy scales, D >> d, the coarse-grained stepwise function (r) = a3 (r/D) may be substituted into the functional F L . As a
consequence of the Edwards screening of unit interactions in melts[115] suciently long chains are asymptotically Gaussian ones. This result known for
homopolymer chains has been veried by the investigation of the screening
eect rst in nonextended heteropolymer chains[128] and then with due regard for their extension[192]. Hence, to have the density functional F L found
it will suce to make a recourse to the Gaussian model of noninteracting
chains. Scaling analysis [192] resting on the above presented considerations
yields the following expression for the free energy functional

F []

gn (n )
= A +
kn
2
2 k1
V
D
n=2 k1 kn k1 + + kn

(169)

where A = Da/V represents dimensionless surface of the domain walls; k


stand for dimensionless Fourier-transforms of the function (x) and summing is over dimensionless wave vectors k1 , , kn of reciprocal lattice of a
superstructure. The domain characteristic size was found to be describable
by following scaling expression
(

D = a5

)1/3

l2/3

(170)

whereas functions gn (n ) in (169) are controlled solely by relative orientation n of vectors k1 , , kn . In practice these have been calculated only up
to terms with n = 4. As a result for fractions f and f of units M1 under
transitions from bcc into triangular structure and from this into lamellar one
there have been calculated following values f = 0.194 and f = 0.35. The
exponent 2/3 in formula (170) is in excellent agreement with the results of
numerical calculations by (166). It should be pointed out, however, that
the evaluation of higher-order terms of a series (169) changes materially this
result. That is why other method of determination of functional F L scaling
asymptotics has been devised[191] which is not concerned with the Landau
expansion (169). Computer calculations of parameter of the lamellar structure, carried out within the framework of this method, reproduce the results
obtained by Semenov[193] as well as Helfand and Wasserman[181, 182] in the
SSR. This method veries also the scaling dependence D l2/3 (170) which

89
has been suggested to derive phenomenologically using the expression
2L 2
3
F =
L +
nD
2la2

D
2

)2

(171)

for free energy of typical molecule with degree of polymerization l. The


rst term in (171) characterizes the excess of its energy due to the interface
formation while the second item stands for entropic losses by chain under its
extension up to the distance D/2 between its ends. Here n is the number of
copolymer chains in the cube with linear dimension L and the factor 2L/D
represents the number of interfaces. Minimum of function F (171) is attained
when the domain size is
(

D = 2 a2 /3

)1/3

l2/3

(172)

where = nl/L3 is average density of monomer units. Expression (172)


coincides with the result (170) of more sophisticated calculations in the case
of polymer melt.
The dependence of the period D of a superstructure on the fraction f units
M1 in diblock-copolymer molecule (12a) is discussed in terms of the Landau
theory[194] where this dependence was shown to be of importance only under
the WSR. Under the SSR both scales, d and D, reach the asymptotic values
which are not controlled by the parameter f . This inference correlates well
with the result obtained in paper [195].

Acknowledgements - One of the authors (S.I.K) expresses his gratitude to K.Binder, G.H.Fredrickson, T.Hashimoto, L.Leibler, D.R.Paul and
J.M.Prausnitz who provided him with the reprints of their papers. S.I.K. is
very indebted to Professor S.Edwards for fruitful discussions on application
of the Field Theory approaches to the thermodynamics of heteropolymers.

90

References

References
[1] K.Binder, Adv.Polym.Sci., 1994, 112, 181.

[2] K.Solc,
J.Polym.Sci.,Polym.Phys.Ed., 1982, 20, 1947.

[3] K.Solc,
Macromolecules, 1983, 16, 236.

[4] K.Solc,
L.A.Kleintjens and R.Koningsveld, Macromolecules, 1984, 17,
573.

[5] K.Solc
and K.Battjes, Macromolecules, 1985, 18, 220.

[6] K.Solc,
Macromolecules, 1986, 19, 1166.

[7] K.Solc,
Macromolecules, 1987, 20, 2506.

[8] K.Solc
and Y.C.Yang, Macromolecules, 1988, 21, 829.

[9] K.Solc,
Y.-H.Huang and Y.C.Yang, Macromolecules, 1989, 22, 4334.
[10] S.I.Kuchanov, Methods of Kinetic Calculations in Polymer Chemistry
, (in Russian), Khimia Publ., Moscow, 1978.
[11] Markov Chains and Monte Carlo Calculations in Polymer Science,
ed. G.Lawry, Marcel Dekker Inc., New York, 1970.
[12] J.L.Koenig, Chemical Microstructure of Polymer Chains, WileyInterscience Publ., New York, 1980.
[13] S.I.Kuchanov, in Mathematical Methods in Contemporary Chemistry, ed. S.I.Kuchanov, Gordon and Breach Publ.Comp., 1996, ch.5,
p.267.
[14] L.H.Sperling, in Recent Advances in Polymer Blends, Grafts and
Blocks, ed. L.H.Sperling, Plenum Press, 1974, p.93.
[15] L.H.Sperling and E.M.Corwin, A Proposed Generalized Nomenclature Scheme for Multipolymer and Multimonomer Systems, ACS
Adv.Chem.Ser., 1979, 176, 609.

91
[16] P.J.Flory, Principles of Polymer Chemistry, Cornell University Press,
New York,1953.
[17] H.Tompa, Polymer Solutions, Butterworth, London, 1956.
[18] M.L.Higgins, Physical Chemistry of High Polymers, John Wiley, New
York, 1958.
[19] I.Prigogine, Molecular Theory of Solutions,
Publ.Comp., Amsterdam, 1957.

North- Holland

[20] K.Kamide, Thermodynamics of Polymer Solutions,


Sci.Publ., Amsterdam - New York, 1990.

Elsevier

[21] O.Olabisi, L.M.Roveson and M.T.Shaw, Polymer-Polymer Miscibility, Academic Press, New York - London, 1979.
[22] L.A.Utracki, Polymer Alloys and Blends. Thermodynamics and Rheology, Hanser Publ., Munich - Vienna - New York, 1989.
[23] R.Koningsveld, in Polymer Science, ed. A.D.Jenkins, North-Holland
Publ.Comp., 1972, vol. 5, p. 1047.
[24] R.Koningsveld, Brit.Polym.Journ, 1975, 7, 435.
[25] R.Koningsveld and L.A.Kleintjens, J.Polym.Sci., Polym.Symp., Ser.C,
1977, N 61, 221.
[26] R.Koningsveld, M.H.Onclin and L.A.Kleintjens, in Polymer Compat
ibility and Incompatibility, ed. K.Solc,
Harwood Acad.Publ., 1982,
p.25.
[27] R.Koningsveld and L.A.Kleintjens, in Polymer Blends and Mixtures,
eds. D.J.Walsh, J.S.Huggins and A.Maconnachie, Martinus Nijho
Publ., 1985, p.89.
[28] E.J.Beckman, R.S.Porter, R.Koningsveld and L.A.Kleintjens, in Integration of Fundamental Polymer Science and Technology, eds.
P.J.Lemstra and L.A.Kleintjens, Elsevier, 1988, p.197.
[29] R.Koningsveld, Pure Appl.Chem., 1989, 61, 1051.

92
[30] I.C.Sanchez, in Polymer Blends, eds. D.R.Paul and S.Newman, Academic Press, 1978, ch.3.
[31] I.C.Sanchez, in Polymer Compatibility and Incompatibility, ed.

K.Solc,
Harwood Acad.Publ., 1982, p.59.
[32] W.J.MacKnight, F.E.Karasz, in Comprehensive Polymer Science, ed.
G.Allen, Pergamon Press, 1989, vol.7, p.111.
[33] D.R.Paul and J.W.Barlow, in Polymer Compatibility and Incompat
ibility, ed. K.Solk,
Harwood Acad.Publ., 1982, p.1.
[34] D.R.Paul,
in Multicomponent Polymer material, ACS Adv.Chem.Ser., 1986,
211, 3.
[35] D.R.Paul, J.W.Barlow and H.Keskkula, in Encyclopedia of Polymer
Science and Engineering, John Wiley, 1988, vol.12, p.399.
[36] S.Krause, J.Macromol.Sci.,Rev.Macromol.Chem., Ser.C, 1972, 7, 251.
[37] S.Krause, in Polymer Blends, eds. D.R.Paul and S.Newman, Academic Press, 1978, ch.2.
[38] J.G.Curro, J.Macromol.Sci.-Revs.Macromol.Chem., Ser.C, 1974, 11,
321.
[39] T.Kwei and T.Wang, in Polymer Blends, eds. D.R.Paul and
S.Newman, Academic Press, 1978, ch.2.
[40] J.W.Kennedy, in Macromolecular Chemistry, 1980, vol.1, p.296 and
1984, vol.3, p.248.
[41] R.Koningsveld, L.A.Kleintjens and M.H.Onclin, J.Macromol.Sci.Phys., Ser.B, 1980, 18,363.
[42] D.J.Walsh, in Comprehensive Polymer Science, ed. G.Allen, Pergamon Press,1989, vol.2, p.135.
[43] M.T.Ratzsch and H. Kehlen, Progr.Polym.Sci., 1989, 14, 1.

93
[44] S.Rostami, in Multicomponent Polymer Systems, eds. I.S.Miles and
S.Rostami, Longman Scientic and Technical, 1992, p.63.
[45] R.L.Catterman, R.Bender and
Proc.Des.Dev., 1985, 24, 194.

J.M.Prausnitz,

[46] W.H.Stockmayer,
L.D.Moore,
J.Polym.Sci., 1955, 26, 517.

M.Fixman

Ind.Eng.Chem.,

and

B.N.Epstein,

[47] S.Krause, A.L.Smith and M.G.Duden, J.Chem.Phys., 1965, 43, 2144.


[48] R.P.Kambour, J.T.Bendler and R.C.Bopp, Macromolecules, 1983, 16,
753.
[49] G.ten Brinke, F.E.Karasz and W.J.MacKnight, Macromolecules, 1983,
16, 1827.
[50] D.R.Paul and J.W.Barlow, Polymer, 1984, 25, 487.
[51] R.Koningsveld and L.A.Kleintjens, Macromolecules, 1985, 18,243.
[52] J.M.G.Cowie, G.Li, R.Ferguson and I.J.McEwen,
J.Polym.Sci.,Polym.Phys.Ed., Ser.B, 1992, 30, 1351.
[53] D.Rigby, J.L.Lin and R.-J.Roe, Macromolecules, 1985, 18, 2269.
[54] J.W.Barlow and D.R.Paul, Polym.Eng.Sci., 1987, 27, 1482.
[55] M.Nishimoto, H.Keskkula and D.R.Paul, Polymer, 1989, 30, 1279.
[56] G.R.Brannock and D.R.Paul, Macromolecules, 1990, 23, 5240.
[57] T.Schiomi, F.E.Karasz and W.J.MacKnight, Macromolecules, 1986,
19, 2274.
[58] S.I.Kuchanov, Polym.Networks & Blends, 1995, 5, 191.
[59] R.L.Scott, J.Polym.Sci., 1952, 9, 423.
[60] R.Koningsveld, L.A.Kleintjens and G.Markert, Macromolecules, 1977,
10, 1105.

94
[61] L.Leibler, Makromol.Chem., Rapid Commun., 1981, 2, 393.
[62] R.Koningsveld and W.J.MacKnight, Makromol.Chem., 1989, 190, 419.
[63] M.T.Ratzsch, H.Kehlen and D.Browarzik, J.Macromol.Sci., Ser.A,
1986, 23, 1349.
[64] M.T.Ratzsch, D.Browarzik and H.Kehlen, J.Macromol.Sci., Ser.A,
1989, 26, 903.
[65] M.T.Ratzsch, C.Wohlfarth, D.Browarzik and H.Kehlen, J.Macromol.
Sci., Ser.A, 1989, 26, 1497.
[66] M.T.Ratzsch, D.Browarzik and H.Kehlen, J.Macromol. Sci., Ser.A,
1990, 27, 809.
[67] M.T.Ratzsch, S.Enders, L.Tschersich and H.Kehlen, J.Macromol. Sci.,
Ser.A, 1991, 28, 31.
[68] A.V.Balazs,
I.C.Sanchez,
I.R.Epstein, F.E.Karasz and W.J.MacKnight, Macromolecules, 1985,
18, 2188.
[69] A.V.Balazs, F.E.Karasz , W.J.MacKnight, H.Ueda and I.C.Sanchez,
Macromolecules, 1985, 18, 2784.
[70] J.vanHunsel, A.C.Balazs, R.Koningsveld and W.J.MacKnight , Macromolecules, 1988, 21, 1528.
[71] A.C.Balazs and M.T.Meuse, Macromolecules, 1989, 22, 4260.
[72] H.-J.Cantow and O.Schulz, Polym.Bull., 1986, 15, 449.
[73] H.-J.Cantow and O.Schulz, Polym.Bull., 1986, 15, 539.
[74] J.W.Barlow, C.H.Lai and D.R.Paul, in Multiphase Macromolecular
Systems, ed. B.M.Culberton, Plenum Press, 1989, p.505.
[75] E.A.Guggenheim, Mixtures, Clarendon Press, Oxford, 1952.
[76] C.G.Panayiotou, Makromol.Chem., 1987, 188, 2733.

95
[77] P.Dafniotis, E.Karayianni and C.Panayiotou, Makromol.Chem., 1989,
190, 1103.
[78] T.Shiomi, H.Ishimatsu, T.Eguchi and K.Imai, Macromolecules, 1990,
23, 4970.
[79] T.Shiomi, T.Eguchi, H.Ishimatsu and K.Imai, Macromolecules, 1990,
23, 4978.
[80] T.Shiomi and K.Imai, Polymer, 1991, 32, 73.
[81] D.R.Paul, Pure Appl.Chem., 1995, 67, 977.
[82] F.E.Karasz and W.J.MacKnight, in Multicomponent Polymer Materials, ed. D.R.Paul and L.H.Sperling, ACS Adv.Chem.Ser., 1986, 211,
p.67.
[83] R.J.Roe and D.Rigby, Adv.Polym.Sci., 1987, 82, 103.
[84] W.J.MacKnight and F.E.Karasz, in Comprehensive Polymer Science, ed. G.Allen, Pergamon Press, 1989, vol.7, p.111.
[85] D.J.Walsh, in Comprehensive Polymer Science, ed. G.Allen, Pergamon Press, 1989, vol.2, p.135.
[86] M.T.Ratzsch, H.Kehlen, in Frontiers of Macromolecular Science ed.
T.Saegusa, T.Higashimura and A.Abe, Blackwell Sci. Publ., 1989, p.76.
[87] P.Munk, P.Hattam and Q.Du, J.Appl.Polym.Sci., Appl.Polym.Symp.,
1989, 43, 373.
[88] R.Koningsveld, W.H.Stockmayer and E.Nies,
Macromol.Symp., 1990, 39,1.

Makromol.Chem.,

[89] C.Wohlfarth, Makromol.Chem., Theory.Simul., 1993, 2, 605.


[90] H.W.Kammer, J.Kressler and C.Kummerloewe, Adv.Polym.Sci., 1993,
106, 31.
[91] J.Dudowicz and K.F.Freed, Macromolecules, 1991, 24, 5076, 5112.
[92] K.F.Freed and J.Dudowicz, Theor.Chim.Acta, 1992, 82, 357.

96
[93] J.Dudowicz and K.F.Freed, J.Chem.Phys., 1992, 96, 1644, 9147.
[94] K.F.Freed and J.Dudowicz, J.Chem.Phys., 1992, 97, 2105.
[95] K.S.Schweizer, Macromolecules, 1993, 26, 6033.
[96] D.Chandler and H.C.Andersen, J.Chem.Phys., 1972, 57, 1930.
[97] K.S.Schweizer and J.G.Curro, Chem.Phys., 1990, 149, 105.
[98] K.S.Schweizer and J.G.Curro, J.Chem.Phys., 1991, 94, 3986.
[99] K.S.Schweizer and J.G.Curro, Phys.Rev.Lett., 1988, 60, 809.
[100] K.S.Schweizer , Macromolecules, 1993, 26, 6050.
[101] J.A.Barker and D.Henderson, J.Chem.Phys., 1967,47, 4714.
[102] Y.Song, S.M.Lambert and J.M.Prausnitz, Chem.Eng.Sci., 1994, 49,
2765.
[103] Y.C.Chiew, Molec.Phys., 1990, 70, 129.
[104] Y.Song, S.M.Lambert and J.M.Prausnitz, Macromolecules, 1994, 27,
441.
[105] T.Hino, Y.Song and J.M.Prausnitz, Macromolecules, 1994, 27, 5681.
[106] T.Hino, Y.Song and J.M.Prausnitz, Macromolecules, 1995, 28, 5717,
5725.
[107] L.D.Landau and E.M.Lifshits, Statistical Physics, Addison-Wesley
Publ. Comp., Reading-London, 1969.
[108] S.I.Kuchanov and S.V.Panyukov, paper submitted to Macromolecules.
[109] C.A.Croxton, Liquid State Physics - A Statistical Mechanical Introduction, Cambridge University Press, Cambridge-London, 1974.
[110] A.Aharony, in Critical Phenomena, Lecture Notes in Physics, 1983,
186, p.209.

97
[111] J.Sivardiere, in Static Critical Phenomena in Inhomogeneous Systems, Lecture Notes in Physics, 1984, 206, 247.
[112] F.Harary, Graph Theory, Addison-Wesley Publ. Comp., ReadingLondon, 1969.
[113] S.V.Panyukov and S.I.Kuchanov, Sov.Phys.JETP, 1991, 72, 368.
[114] S.V.Panyukov and S.I.Kuchanov, J.Phys.II (France), 1992, 2, 1973.
[115] M.Doi and S.F.Edwards,The Theory of Polymer Dynamics, Clarendon Press, Oxford, 1986.
[116] K.F.Freed, Renormalization Group Theory of Macromolecules, John
Wiley, New York, 1987.
[117] J.M.Ziman, Models of Disorder, Cambridge Univ.Press, CambridgeLondon-New York, 1979.
[118] Density Functional Theory, ed. J.Keller and J.L.Gazquez, Lecture
Notes in Physics, 1983, 187.
[119] I.M.Lifshitz,A.Yu.Grosberg and A.R.Khokhlov, Rev.Mod.Phys., 1978,
50, 683.
[120] I.A.Katime and J.R.Quintana, in Comprehensive Polymer Science,
ed. G.Allen, Pergamon Press, 1989, vol.1, 103.
[121] H.Benoit and D.Froelich, in Light Scattering from Polymer Solutions,
ed. M.B.Huglin, London-New York, 1972, p.467.
[122] P.G.de Gennes,Scaling Conceptions in Polymer Physics, Cornell
Univ.Press., Ithaca-London, 1979.
[123] L.Leibler, Macromolecules, 1982, 15, 1283.
[124] I.Ya. Erukhimovich, Vysokomol.Soedin., A, 1979, 21, 427
[125] H.Benoit, M.Benmouna and W.li.Wu, Macromolecules, 1990, 23, 1511.
[126] J.G.Curro and K.S.Schweizer, Macromolecules, 1987, 20, 1928.

98
[127] M.Olvera de la Cruz and I.C.Sanchez, Macromolecules, 1986, 19, 2051.
[128] T.A.Vilgis and R.Borsali, Macromolecules, 1990, 23, 3172.
[129] T.A.Vilgis, M.Benmouna and H.Benoit, Macromolecules, 1991, 24,
4481.
[130] L.Leibler, Macromolecules, 1980, 13, 1602.
[131] I.Ya.Erukhimovich, Vysokomol.Soedin., A, 1982, 24, 1942.
[132] P.G.de Gennes, J.Phys.(France), 1970, 31, 235.
[133] W.E.McMullen and K.F.Freed, Macromolecules, 1990, 23, 255.
[134] H.Tang and K.F.Freed, Macromolecules, 1991, 24, 958.
[135] L.Leibler and H.Benoit, Polymer, 1981, 22, 195.
[136] K.M.Hong and J.Noolandi, Polymer Commun., 1984, 25, 265.
[137] M.Olvera de la Cruz and I.C.Sanchez, Macromolecules, 1987, 20, 440.
[138] L.Leibler, H.Orland and J.C.Wheeler, J.Chem.Phys., 1983, 79, 3550.
[139] T.A.Kavassalis and M.D.Whitmore, Macromolecules, 1991, 24, 5340.
[140] P.G.de Gennes, Faraday Discussion Chem.Soc., 1979, 68, 96.
[141] Y.Ijichi and T.Hashimoto, Polym.Commun., 1988, 29, 135.
[142] R.M.Hornreich, M.Luban and S.Strikman, Phys.Rev.Letters, 1975, 35,
1678.
[143] D.Broseta and G.H.Fredrickson, J.Chem.Phys., 1990, 93, 2927.
[144] G.H.Fredrickson and S.T.Milner, Phys.Rev.Letters, 1991, 67, 835.
[145] A.Dondos, Makromol.Chem., 1971, 147, 123.
[146] A.Dondos, J.Polym.Sci., Lett.Ed., 1971, 9, 871.
[147] E.Helfand and Y.Tagami, J.Polym.Sci., Ser.B, 1971, 9, 741.

99

[148] E.Helfand, in Polymer Compatibility and Incompatibility, ed.K.Solc,


Harwood Acad.Publ., 1983, p.143.
[149] E.Helfand, J.Chem.Phys., 1975, 63, 2192.
[150] J.Noolandi and K.M.Hong, Macromolecules, 1982, 15, 482.
[151] J.Noolandi and K.M.Hong, Macromolecules, 1984, 17, 1531.
[152] T.A.Vilgis and J.Noolandi, Macromolecules, 1990, 23, 2941.
[153] L.Leibler, Macromolecules, 1982, 15, 1283.
[154] T.Inoue, T.Soen, T.Hashimoto and H.Kawai, in Block Copolymers,
ed. S.L.Aggarwal, Plenum Press, 1970, p.53
[155] S.Krause, in Colloid and Morphological Behaviour of Block and Graft
Copolymers, ed. G.E.Molau, Plenum Press, 1971, p.223.
[156] S.Krause, in Block and Graft Copolymers, ed. J.J.Burke and
V.Weiss, Syracuse Univ.Press, 1973, p.143.
[157] A.E.Skoulous, in Developments in Block
I.Goodman, Academic Press, 1982, vol.1, p.81.

Copolymers,

ed.

[158] E.Gelfand, Z.R.Wasserman, in Developments in Block Copolymers,


ed. I.Goodman, Academic Press, 1982, vol.1, p.99.
[159] T.Hashimoto, M.Shibayama, M.Fujimura and H.Kawai, in Block
Copolymers, ed. D.J.Meier, Harwood Acad.Publ., 1983, p.63.
[160] B.R.M.Gallot, Adv.Polym.Sci., 1978, 29, 85.
[161] R.A.Brown, A.J.Masters, C.Price and X.F.Yuan, in Comprehensive
Polymer Science, ed. G.Allen, Pergamon Press, 1989, vol.2, p.155.
[162] T.Hashimoto, H.Tanaka and N.Iizuka, in Space-Time Organization in
Macromolecular Fluids, ed. F.Tanaka, M.Doi and T.Ohta, Springer
Verlag, 1989, p.2.

100
[163] G.H.Fredrickson, E.Helfand, F.S.Bates and L.Leibler, in Space-Time
Organization in Macromolecular Fluids, ed. F.Tanaka, M.Doi and
T.Ohta, Springer Verlag, 1989, p.13.
[164] M.D.Whitmore, in Space-Time Organization in Macromolecular Fluids, ed. F.Tanaka, M.Doi and T.Ohta, Springer Verlag, 1989, p.30.
[165] Yu.D.Shibanov and Yu.K.Godovski, Uspekhi Khimii, 1988, 57, 1713.
[166] S.F.Edwards, J.Phys., A, 1974, 7, 332.
[167] E.Helfand, Accounts Chem.Res., 1975, 8, 295.
[168] S.V.Panyukov and S.I.Kuchanov, Sov.JETP Lett, 1992, 54, 501.
[169] G.H.Fredrickson and L. Leibler, Macromolecules, 1989, 22, 1238.
[170] J.Noolandi and K.M.Hong, Polym. Bull., 1988, 7, 561.
[171] K.M.Hong and J.Noolandi, Macromolecules, 1983, 16, 1083.
[172] A.M.Mayes and M.Olvera de la Cruz, J.Chem.Phys. 1989, 91, 7228.
[173] A.M.Mayes and M.Olvera de la Cruz, Macromolecules, 1991, 24, 3975.
[174] J.N.Owens, I.S.Gancarz, J.T.Koberstein and T.P.Russell, Macromolecules, 1989, 22, 3380.
[175] M.Olvera de la Cruz, A.M.Mayes and B.W.Swift, Macromolecules,
1992, 25, 944.
[176] A.V.Dobrynin and I.Ya.Erukhimovich, Vysokomol.Soedin., A, 1991,
33, 1100.
[177] C.Burger, W.Ruland and A.N.Semenov, Macromolecules, 1990, 23,
3339.
[178] E.I.Shakhnovich and A.M.Gutin, J.Phys. (France), 1989, 50, 1843.
[179] E.Helfand, in Recent Advances in Polymer Blends, Grafts and
Blocks, ed. L.H.Sperling, Plenum Press, 1974, p.141.

101
[180] E.Helfand, Accounts Chem.Res., 1975, 8, 295.
[181] E.Helfand, Macromolecules, 1975, 8, 552.
[182] E.Helfand and Z.R.Wasserman, Macromolecules, 1976, 9, 879.
[183] E.Helfand, J.Chem.Phys., 1975, 62, 999.
[184] E.Helfand and Z.R.Wasserman, Polym.Eng.Sci., 1977, 17, 582.
[185] E.Helfand and Z.R.Wasserman, Macromolecules, 1978, 11, 960.
[186] E.Helfand and Z.R.Wasserman, Macromolecules, 1980, 13, 994.
[187] M.D.Whitmore and J.Noolandi, J.Chem.Phys., 1990, 93, 2946.
[188] M.D.Whitmore and J.D.Vavasour, Macromolecules, 1992, 25, 2041.
[189] T.Ohta and K.Kawasaki, Macromolecules, 1986, 19, 2621.
[190] K.Kawasaki, T.Ohta and M.Kohrogui, Macromolecules, 1988, 21, 2972.
[191] K.Kawasaki and T.Kawakatzu, Macromolecules, 1990, 23, 4006.
[192] K.Kawasaki, T.Ohta and M.Kohrogui, Macromolecules, 1986, 19, 2621.
[193] A.N.Semenov, Sov.Phys.JETP, 1985, 61, 733.
[194] T.Ohta and K.Kawasaki, Macromolecules, 1990, 23, 2413.
[195] Z.G.Wang and S.A.Safran, J.Phys.(France), 1990, 51, 185.

Figure captions

a)

a2
a2

Fig.
1. The shape of two=
dimensional phase diagram for binary
(a) and quasibinary (b) systems de=
scribing, respectively, the solutions
of monodisperse and polydisperse homopolymer in low-molecular weight
b)
solvent.
+
=
(T - temperature, - volume fraction of polymer in a solvent; full line binodal (a) and cloud-point curve (b),
dotted-dashed line - spinodal, dashed line - shadow curve, solid circle - critical
point, solid square - precipitation threshold). 1
2
3
a)
Fig. 2. Four principal types of
lattice model employed in statistical
thermodynamics of polymers:
-3
=
=
b)
a) The Flory-Huggins model; b)
Cell model; c) Hole model; d) Cell+15
=
-10
hole model.
Fig. 3. Diagram expressions re=
-15
+105
-105
lating coecients of the Landau expansion of free energy in formalisms
of density functional (DF) and of ran=
+210
-35
+280
-21
dom eld (RF).
Fig. 4. Elements of diagram
-945
-1260
technique (a) and diagram expres=
-56
-35
+280
-28
sions for the coecient of expansion
(79)(b).
-3150
+378
-1260
Fig. 5. Characteristic forms
-10395
-17325
of dependences of scattering intensity
upon the modulus of the scattering wave vector q.
Fig. 6. Elements of diagram technique (a) for description of polydisperse
system in terms of the formalism of random eld (b) and density functional
(c).
I(q)
Fig.
7.
Phase diagram for
a
b
melt of monodisperse diblock-copolymer
l(1f )
lf
(12a).[130] Figures denote the
M1 M2
type of phase:
1) lamellar (d=1, n=2);
2) planar triangular lattice (d=2, n=3)
3) body-centred cubic lattice (d=3, n=6)
q
4) spatially homogeneous
0
Fig. 8. Phase diagram for melt of monodisperse triblock-copolymer
l(1f )
lf (1 )
M1lf M2
M1
(12b) for the case = 0.25.[172] Here denotes the
a1

a3

a2

a3

a1

a4

a1

a3

a2

a3

a1

a4

a2

a3

a1

a3

a4

a3

a1

a4

a2

a4

a1

a2

symmetry parameter while the remaining designations are the same as in Figure 8.
1

a2

a3

a2

a3

a1

a4

a1

a4

a)

b)

a1

a2

c)

S
-

Vous aimerez peut-être aussi