J Mol Model (2011) 17:607619

DOI 10.1007/s00894-010-0755-z

ORIGINAL PAPER

Modeling and analysis of the compatibility

of polystyrene/poly(methyl methacrylate) blends
with four inducing effects
Dan Mu & Jian-Quan Li & Yi-Han Zhou

Received: 8 February 2010 / Accepted: 1 April 2010 / Published online: 5 June 2010
# Springer-Verlag 2010

Abstract The compatibility of polystyrene (PS) and poly

(methyl methacrylate) (PMMA) blends was studied over a
wide range of compositions at 383, 413 and 443 K,
respectively, by atomistic and mesoscopic modeling. All
the calculated FloryHuggins interaction parameters
showed positive values; furthermore, they were all above
the critical FloryHuggins interaction parameter value,
which means that the PS/PMMA blends were immiscible.
Both the addition of a block copolymer and the introduction
of a shear field influenced the phase morphologies of the
blends, while the degree of influence depended on the
compositions of the blends. The study of PS/PMMA blends
doped with nanoparticles showed that the mesoscopic phase
was influenced by not only the properties of the nanoparticles, such as their size, number and number density,
but also the compositions of the blends. The effect of the
surface roughness of the planes on the phase separation of
the blends was also studied.
Keywords PS/PMMA blend . Inducing effect . Phase
morphology . Nanoparticle

D. Mu (*)
Department of Chemistry, Zaozhuang University,
Shandong 277160, China
e-mail: [email protected]
J.-Q. Li
Physics and Electronic Engineering Department,
Zaozhuang University,
Shandong 277160, China
Y.-H. Zhou
National Analytical Research Center of Electrochemistry
and Spectroscopy, Changchun Institute of Applied Chemistry,
Chinese Academy of Sciences,
Changchun, China

Introduction
Polymer blend systems play an important role in the plastic
industry because they can be tuned to have a better
combination of physical properties than the corresponding
homopolymers. The process of physical blending is much
easier to perform than chemical blending (e.g., grafting),
and the former is also more economical. As a general rule,
the aim of blending is to obtain a product with new or better
properties. Because of the different physical properties of
candidate polymers, it is not always easy to obtain
completely miscible polymer mixtures. When the components of polymers can form energetically favorable interactions, such as hydrogen bonding, dipole interactions, etc.,
it is possible to obtain a miscible blend. For immiscible
blends, adding interferents is a common method of
changing the microscopic morphology. This improves the
blend miscibility and thus enhances its properties.
Polystyrene (PS) and poly(methyl methacrylate)
(PMMA) are conventional model systems in polymer
science. PS/PMMA blends are a well-known immiscible
combination [1-8], and their bulk and surface phase
separation has been observed [9, 10]. Immiscible blends
are known to have properties combining those of their
component polymers, and also to have segregated structures
with domains that are predominantly formed from the
individual homopolymers. It has been shown that changing
the relative homopolymer proportions in such blends can
change the domain structure, surface morphology, and even
the phase morphology [11-13].
Some contradictory results can be found in the literature
regarding the miscibilities of certain polymers and the
possibility of predicting miscibility from surface tension or
from parameters derived from it. These contradictions can
be attributed to several reasons. Firstly, the structures and

608

the properties of partially miscible polymer blends depend

on the processing conditions used, which may influence
conclusions about polymer compatibility when they are
deduced from macroscopic properties or from the structure
of the blend. Secondly, methods that can be used to predict
miscibility are limited [14, 15]. For example, various
experimental difficulties can be encountered during the
determination of the FloryHuggins interaction parameter,
, which can be used to characterize the strength of the
interaction between the polymers. can be determined by
various methods, such as inverse gas chromatography,
solvent diffusion, or the dependence of composition on
mechanical properties, but all of these methods have their
drawbacks. Finally, when blend systems have two calorimetric Tg values, it does not necessarily indicate that their
components are immiscible [16]. Fortunately, computer
calculations and simulations have become available as
efficient methods of solving these experimental problems.
This paper is divided into two main parts. The first part
is about the calculation of typical thermodynamic parameters, such as the solubility parameter, of the studied
systems via a microscopic calculation method, and comparing those calculated parameters with experimental data
in order to maintain a high level of accuracy in further
calculations and simulations; the second part describes
mesoscopic dynamics simulations of several kinds of
inducing effects that can change the phase morphologies
of PS/PMMA blends. Mesoscale structures are of the
utmost importance during the production processes of
many materials, such as polymer blends, block copolymer
systems, surfactant aggregates in detergent materials, latex
particles, and drug delivery systems.
Mesoscopic dynamics models have been receiving increasing attention recently, as they form a bridge between
microscale and macroscale properties [17-20]. Mesoscopic
dynamics (MesoDyn) [21-23] and dissipative particle dynamics (DPD) [24, 25] methods both treat the polymer chain
at a mesoscopic level by grouping atoms together and then
coarse-graining them to be persistent-length polymer chains.
These two mesoscopic modeling methods increase the scale
of the simulation by several orders of magnitude compared
with atomistic simulation methods. MesoDyn utilizes dynamic mean-field density functional theory (DFT), in which
the dynamics of phase separation can be described by
Langevin-type equations to investigate polymer diffusion.
Thermodynamic forces are found via mean-field DFT, using
the Gaussian chain as a model. The most important
molecular parameter is the incompatibility parameter N,
where is the FloryHuggins interaction parameter and N is
the degree of polymerization.
However, owing to the soft interaction potential of the
DPD method, it is not suitable to apply the DPD method in our
work, especially when dealing with the interactions between

J Mol Model (2011) 17:607619

blending materials and planes or doped nanoparticles.

Furthermore, in the study of the compatibility of PEO/PMMA
blends in our former work, the MesoDyn simulation method
was successfully applied to detect the phase morphologies of
PEO/PMMA blends. From a theoretical point of view, these
results clarified several conflicting conclusions of different
experiments [26]. Therefore, the MesoDyn method has been
shown to be reliable when dealing with PS/PMMA blends.
The mesoscale simulation process used in this paper was
carried out with the MesoDyn module in the Material Studio
commercial software provided by Accelrys.

Simulation details, results and discussions

This paper consists of two parts: calculations performed via
(1) molecular dynamic (MD) simulations and (2) mesoscopic
simulations. The intermediate parameter connecting the
microscale and the mesoscale is the FloryHuggins parameter, . MD simulation is used to accurately study bulk
polymer properties when this is difficult to do experimentally. Although MD simulation is widely used nowadays, it is
difficult to detect the phase separation of blends via a
microscopic method, such as MD simulation. Therefore, it is
necessary to combine it with a mesoscopic method, such as
MesoDyn simulation, to perform a full study of polymer
blends from the microscale to the mesoscale.
Atomistic molecular simulation
Model and simulation
The selection of the representative chain length (Nmon) was
our primary task. This simulation process was carried out at
room temperature (298 K) on an SGI workstation by
following the MD procedure in the commercial Cerius2
package program provided by Accelrys.
The simulation processes (including the construction of the
models) were the same for both PS and PMMA, so we will use
an atomistic chain length N of 10 for PS as an example here
to describe the simulation process in detail. Initially, the
proposed polymer chain structure was generated using the
rotational isomeric state (RIS) model of Flory [27], which
describes the conformations of the unperturbed chains. Three
PS polymer chains were then generated following the same
principle, each with ten repeating units; they were then put
into a simulation box with a density of 1.05, which is the
density of bulk PS [28]. Minimization was then done using
the conjugate gradient method (CGM) with a convergence
level of 0.1 kcal mol1 1 until the energy reached a
minimum. This minimized configuration was refined by MD
simulation for 30 ns under constant temperature and density
(NVT ensemble). A time step of 1 fs was used to ensure the

J Mol Model (2011) 17:607619

stability of simulation. The NosHoover [30] thermostat in

a canonical ensemble was used to create the canonical
molecular dynamics trajectories, and the COMPASS force
field was chosen to deal with such a condensed state system
[31-33]. The spline treatment of van der Waals (VDW)
interactions was chosen, with a spline window of between 11
and 12 . The Ewald summation [34], an accurate method of
calculating long-range interactions, was applied in order to
calculate long-range electrostatic interactions. Next, the
refined polymer was annealed in the temperature range from
383 K (above 373 K [28], the Tg of PS) to 443 K (below
513 K [28], the melting temperature Tm of PS) in steps of
10 K for 50 cycles. The simulation time for each annealing
cycle was set to 2 ns, and the system was minimized at the
beginning of each annealing cycle. Because such oligomers
always have strong endend interactions, a system consisting
of oligomers is different from a real polymeric system. Thus,
an MD simulation was executed with the NPT (with constant
temperature and pressure) ensemble for 20 ns; furthermore,
MD simulations with a repeating ensemble cycle from NVT
to NPT were carried out three times with a simulation time of
20 ns for each cycle to reduce the endend interactions
sufficiently. Finally, an MD simulation was executed with
the NVT ensemble for 30 ns at 298 K to collect data. In
general, about 200 ns was needed in total for each simulated
case. During the whole simulation process, the selection of
the appropriate force field was a crucial task for obtaining
accurate information about the structures and dynamics of
the blends. The COMPASS force field has been shown to be
a suitable choice for calculating solubility parameters,
especially for polymeric systems [35].
Results and discussion
The cohesive energy density (CED) of each pure component was used to calculate the solubility parameter, ,
according to the equation below:
p
d CED

609

Fig. 1 The solubility parameter for different PS chain lengths

numbers of repeating units of PMMA. These become

constant from N=60 onwards, and they are 6.40% smaller
than the experimental results. This may result from the
presence of a bulky CH3 group in PMMA, which also
restricts free rotation around the CC bond of its backbone.
The fact that the calculated relaxation times were
shorter than the experimental results presented a rather
difficult obstacle to overcome in the simulations. However, taking N=60 to represent the whole polymer chain
for PMMA appeared a reasonable approach. Hence, the
minimum representative polymer chain lengths of PS and
PMMA, Nmon(PS) and Nmon(PMMA), were chosen to be
30 and 60, respectively. Figure 3 shows a flowchart that
illustrates the process described above.
The blend systems were binary, with PS (density =
1.05 gcm3) and PMMA (density = 1.188 g cm3) as the
components. The glass transition temperature of PS is

r
Ecoh
;
V

To determine the minimum representative polymer chain

length of the blend components, Nmon, cases with different
chain lengths, N=10, 20, 30, 40, 50, 60, 70 and 80 for both
PS and PMMA, were selected in order to calculate the
value of each component. Figure 1 displays the dependence
of of PS on the number of repeating units of PS. It can be
seen that a nearly constant value is approached when the
number of repeating units reaches 30, and the values are
8.32% smaller than the experimental data. This may result
from the presence of a bulky C6H5 group in PS, which
restricts free rotation around the CC bond of its backbone.
Figure 2 shows the calculated values for different

Fig. 2 The solubility parameter for different PMMA chain lengths

610

J Mol Model (2011) 17:607619

chains were randomly distributed in the simulation box, the

model was assumed to be a well-mixed model and accepted
for further simulations; otherwise, it was discarded, and a
new blending model was built until the component chains
contacted each other sufficiently. Figure 4 shows a typical
1/1 blend. Because of the endend interactions of short
polymer chains, the initial amorphous structure of the blend
must first be minimized. The subsequent simulation process
was the same as that used to calculate values for the pure
components, and the simulation temperatures were also
383, 413 and 443 K.
The tendency for binary polymer chains to mix depends on
the cohesive energies of the blends and their pure components.
Both are applied to calculate the parameters of blends with
different compositions. The parameter is defined as


Emix

Vmon ;
RT
where Vmon is the unit volume of one monomer. Here, Vmon
equals 165.54 cm3 mol1 for blends of PS/PMMA. Emix for
a binary mixture is defined as






Ecoh
Ecoh
Ecoh
Emix f1
f2

;
V pure1
V pure2
V mix

Fig. 3 The procedure used to select Nmon for PS

373 K, so PS is prone to crystallize at a low temperature.

PMMA starts to soften at the glass transition temperature
(Tg =378 K), and does not substantially depolymerize until
453 K [36]. Therefore, the blending simulation temperatures adopted were 383, 413 and 443 K.
PS/PMMA blends with different weight percentages of
PS were put in a three-dimensional (3D) periodic boundary
simulation box with the corresponding density. Table 1
describes the blending models in detail, including the molar
ratios of PS to PMMA, the chain number per unit cell, the
compositions by weight, and the densities. The blend
properties calculated in subsequent studies depend on how
well the components are blended. Therefore, when the
blending model was generated, if the PS and the PMMA

where f is the volume fraction, and the subscripts pure

and mix denote the cohesive energy densities of the pure
components and the binary mixture, respectively. is shown
as a function of the PS volume fraction in Fig. 5. Several
features of these data are impressive. Firstly, the three
curves vary in almost the same way at different temperatures,
which means that temperature was not an efficient factor for
changing the degree of phase separation. This is a totally
different conclusion from that obtained for PEO/PMMA
blends, which are miscible [26]. Secondly, the value shows
a decreasing trend as the volume fraction of PS increases,
except for the 1/4 blend case, where it shows a small rising
trend. Then the 4/1, 6/1 and 8/1 blend cases also show another
decreasing trend in . These two decreasing trends are termed
T1 and T2, respectively. Considering the absolute data on
slope variation for T1 and T2, where T1 < T2, it is supposed
that the degree of phase separation would intensify with
increasing PS content when the blends have a higher content
of PS. In addition, the 1/4 and 4/1 blend cases produced
outstandingly high values for both the T1 and the T2 lines,
which means that it is possible that extremely distinct phase
separation was present in these two blend cases.
A positive value alone cannot prove that the blends are
immiscible; it is necessary to calculate the critical value of
via the equation below:
AB critical

1
2

1
1
p p
mA
mB

2
;

J Mol Model (2011) 17:607619

Table 1 Details of the
PS/PMMA blend cases carried
out via MD simulation

611
Number

System

Content

wt% of PMMA

p(gcm3)

1
2
3
4
5
6
7
8
9
10
11
12

Pure PMMA
1/6
1/4
1/3
1/2
1/1
2/1
3/1
4/1
6/1
8/1
Pure PS

1PMMA chain
1PS, 6PMMA
1PS, 4PMMA
1PS, 3PMMA
1PS, 2PMMA
1PS, 1PMMA
2PS, 1PMMA
3PS, 1PMMA
4PS, 1PMMA
6PS, 1PMMA
8PS, 1PMMA
1PS chain

100
92.02
88.49
85.22
79.36
65.78
49.01
39.06
32.46
24.27
19.38
0

1.1880
1.1772
1.1724
1.1681
1.1604
1.1429
1.1220
1.1099
1.1021
1.0925
1.0868
1.0500

where mA and mB represent the actual number of repeating

units of PS and PMMA, respectively. When the calculated
and the critical values of are based on the same reference
volume, AB)critical can be used to estimate the compatibility of the blends. The blends are miscible if (AB)calculated <
(AB)critical. In this case, the Gibbs free energy, Gm, is
negative. In contrast, when (AB)calculated (AB)critical, the
blends are totally immiscible; the phases will even separate.
When the difference between AB)calculated and AB)critical is
small or even zero, the blends will be partially or even
totally miscible. The value of AB)critical in our studied PS/
PMMA blends was 0.0486, which was much smaller than
the calculated values. Therefore, we can see that the PS/
PMMA blends were immiscible at 383, 413 and 443 K.
As long as the calculated AB and the corresponding
critical values are based on the same reference volume,
comparing the calculated AB with critical values will
provide useful information about the degree of miscibility
of the blends. Therefore, AB, rather than the enthalpy of
Fig. 4 Amorphous cell from the
1/1 PS/PMMA blend case.
Red, O; gray, C; white, H

mixing Hm, could be used to characterize the miscibility

of polymer blends [37].
However, it is also necessary to calculate thermodynamic
parameters such as entropy, enthalpy and Gibbs free energy
(G) to illustrate the phenomenon of polymer blending. The
entropy change upon mixing long polymers is so small that
the mixing enthalpy becomes a key influence on the degree
of mixing. The entropy change upon mixing, Sm, must be
considered in addition to the enthalpy of mixing. Hence, it
was important to compute Hm in order to study the
miscibilities of the PS/PMMA blends using the following
equation [37]:
Hm Vm

hn

0:5
CED0:5
PS CEDPMMA

oi2
fPSfPMMA;

where Vm is the total volume of the blend. The CED values

used to compute Hm were obtained from the data in the
MD simulations.

612

J Mol Model (2011) 17:607619

they exhibit the strongest attractive or repulsive interactions.

Because of the low entropy of mixing, polymeric blends
are mostly incompatible, and their phases will separate
under the appropriate conditions. Therefore, it is necessary
to carry out further research from a dynamic point of
view at the mesoscopic level to study how the phase
separations of PS/PMMA blends are influenced by certain
inducing effects.
Mesoscopic simulation
PS/PMMA blends with a block copolymer and a shear field

Fig. 5 FloryHuggins interaction parameter as a function of the PS

volume fraction for binary blends of PS/PMMA at 383, 413 and
443 K, respectively

To map the representative polymer chains onto Gaussian

chains, the bead number (Nmeso) and the unit chain length
of the Gaussian chain (ameso) are estimated by
N mon
;
C1
amon C1;

Nmeso
To calculate Gm, the change in entropy during mixing,
Scomb
was needed, and this was calculated using the
m
combinatorial entropy term for binary systems via Flory
Huggins theory [38] for two-component systems, as given by
Scomb
R# A ln # A # B ln # B :
m
Thus, by using the thermodynamic formula
;
Gm Hm  T Scomb
m
it was possible to calculate the Gibbs free energy of mixing.
Both Gm and Hm exhibited similar trends at 413 K over
the entire composition range, as shown in Fig. 6. The
extreme values of these two curves both appeared in the 1/1
blend case. It is presumed that when the molar proportions
of PS and PMMA are the same, such as in the 1/1 blend,

Fig. 6 Changes in enthalpy Hmix (solid curve) and Gibbs free

energy Gmix (dotted curve) with volume fraction of PS

ameso

where Nmon is the number of repeating units in the polymer

chain. amon is the reference unit monomer length, and C is
the characteristic ratio of a chain of infinite length. Nmon
was determined by the representative chain lengths of PS
and PMMA, i.e., Nmon =30 for PS (MW =3200) and Nmon =
60 for PMMA (MW =6000), while C(PS)=9.85 and C
(PMMA)=8.65 were found in the Polymer Handbook [29].
The connection between the microscale and the mesoscale
is as follows:
v1 "ij # AB RT ;
where the parameter AB is calculated by atomistic
simulation for each blend composition at different temperatures. R is the molar gas constant, 8.314 J mol1 K1, and
T is the simulation temperature. The calculated values for
various compositions were all positive at 383, 413 and
443 K. Nmeso and 1ij are the input parameters used in
MesoDyn simulations to describe the number of Gaussian
beads and the interaction between beads. In the 1/1 blend
case, the Nmeso values for PS and PMMA were 3 and 6,
respectively.
In order to stabilize the numerical calculation, the time
step for MesoDyn calculation, , was chosen to be 100 ns,
and the total simulation time was 1000 s for each blend
case. A default noise parameter value of 75.002 was used
for numerical speed and stability. The grid dimensions
adopted were 323232 nm. Ten mesoscopic simulations
were carried out with different PS/PMMA blend compositions, 1/1, 1/2, 1/3, 1/4, 1/6, 2/1, 3/1, 4/1, 6/1 and 8/1 in
molar ratio, which roughly covered the whole composition
range.

J Mol Model (2011) 17:607619

613

Results and discussion

The order parameter, denoted P, is defined as the average
difference between the local density squared and the overall
density squared, as given by the equation
Pi

1
V

Z
V


h2i r  h2i dr;

where i is the dimensionless density (volume fraction) of

species i. The larger the value of the order parameter, the
stronger the phase separation. A decrease in P indicates
better compatibility or miscibility, and the polymer phases
mix more randomly.
It should be pointed out that the P data calculated by
mesoscopic simulation can only be compared for the same
blend composition. Therefore, plain PS/PMMA blends
were used as reference cases, and their order parameter
data were labeled A; in addition, the order parameter data
of PS/PMMA blends of the same compositions as for the
A blends but with inducing effects, such as being doped
with a block copolymer or nanoparticles, or having planes
introduced, were labeled B. The value of (B A)/A is
termed the variation rate of order parameters (V ROP for
short). By comparing the V ROP values, we can detect
which factor is the most effective at changing the phase
morphology.
In order to study the influence of the addition of a block
copolymer on the phase morphology, a block copolymer
made of PS and PMMA with a volume fraction of 10% was
added to each blend. Generally speaking, when block
copolymers of the same composition are added to an
immiscible blend, the compatibility of the blend will be
enhanced. However, in our study, two opposite results were
obtained when the block copolymer was added: an
enhancement of or a deterioration in compatibility were
both observed.
Several striking features of these data can be seen in
Fig. 7. First, the V ROP values of blends with less PS (the
1/1, 1/2, 1/3, 1/4, 1/6, 2/1 and 3/1 blend cases) fluctuated
around zero, and this trend was most obvious for the 1/1, 1/
3 and 1/6 blend cases, which means that adding a block
copolymer to the blends with less PS did not result in an
obvious inducing effect. However, obvious enhancing
effects were seen for the blends with more PS (the 4/1, 6/
1 and 8/1 blend cases). This result can be attributed to the
difference in the self-diffusion coefficients between PS and
PMMA and the difference in their self-compatibility
properties. Second, the V ROP values of the 1/2, 1/1 and
2/1 blend cases were negative at 383 K; when the
temperature was increased to 413 K, the V ROP value of
the 1/3 blend case turned from positive to negative. When
the temperature was raised again, to 443 K, the V ROP

Fig. 7 V ROP value versus volume fraction of PS for ten blend cases
with 26 block copolymers added. The zero point on the Y axis was set
as the reference line. When the data are above this line, it means that
the block copolymer has an enhancing effect on the PS/PMMA blend;
when the data are below this line, it has a deteriorating effect on the
blend. Isodensity surface pictures of a 6/1 (PS/PMMA) blend at 443,
413 and 383 K are displayed from top to bottom in the chart,
respectively. Red, PS; green, PMMA

value of the 3/1 blend case also turned from positive to

negative. Based on this information, and using the 1/1
blend case as a reference, the V ROP values of pairs of
blend cases (such as the 1/3 and 3/1 blend cases) turned
from positive to negative as the temperature was gradually
increased. Following this principle, when the temperature
was raised above 443 K, the V ROP values of the 1/4 and 4/
1 blend cases would be expected to become negative. The
three pictures inserted into Fig. 7 are the phase morphologies of the 6/1 blend case with block copolymers added at
443, 413 and 383 K (top to bottom, respectively).
In order to investigate the influence of the shear field on
the microscopic phase morphology, the shear rate was set to
0.001 ns1 and 0.002 ns1 as examples of low and high
shear effects. Simulations were then performed on the basis
of plain PS/PMMA blends with P data used as a reference.
It can be seen from Fig. 8 that low shear stress can have a
significant inducing effect on PS-rich blends, such as in the
4/1 and 6/1 blend cases. Phase separation can clearly be
seen in the inserted pictures. However, shear stress had no
obvious effect on the 8/1 blend case. When the shear rate
was increased to 0.002 ns1, all of the V ROP values for the
ten different blend cases were high, as shown in Fig. 9,
which means that high shear stress plays an important role
in changing the microscopic phase morphology, especially
in PS-rich blends. In addition, the V ROP value of the 8/1
blend case was relatively high. Furthermore, the phase
morphologies of all the blends exhibited obvious phase
separation. In a word, the higher the shear rate, the more
obvious the phase separation.

614

Fig. 8 V ROP data based on plain blends versus the volume fraction
of PS. Isodensity surface pictures of a 4/1 (PS/PMMA) blend induced
by a shear rate of 0.001 ns1 are also shown, as are isodensity surface
pictures of a 4/1 (PS/PMMA) blend at 383, 443 and 413 K (top to
bottom in the chart, respectively). Red, PS; green, PMMA

PS/PMMA blends doped with nanoparticles

The pairs of blend cases, such as the 1/3, 3/1 and the 1/2, 2/
1 blend pairs, were then used as examples to study the
inducing effect of doping with nanoparticles. In addition,
the 1/3 (with a PS volume fraction of 16.21%) and 1/2
(with a PS volume fraction of 22.49%) blend cases were
used to represent compositions rich in PMMA, while the 2/
1 (with a PS volume fraction of 53.72%) blend case was
used to represent compositions that contain roughly equal
amounts of PS and PMMA, and the 3/1 (PS volume
fraction of 63.52%) blend case was chosen to represent
compositions rich in PS.

J Mol Model (2011) 17:607619

Column-shaped nanoparticles were applied as an example. In order to determine the factor that exerts the most
influence on phase separation, seven different kinds of
nanoparticles were made for the doping study. In Table 2,
the number of nanoparticles in each layer (Np), the radius of
each nanoparticle (rp), the height of each nanoparticle (hp),
the number of layers (NL) and the total number of
nanoparticles doped (Ntp) are listed in detail. Among these
cases, the 4-3-4-2 case (with four nanoparticles in each of
two layers, a nanoparticle radius of 3 nm, and a nanoparticle height of 4 nm) was used as a reference, and all the
other cases were derived from this. For example, adding
only one more nanoparticle in the center resulted in the 4-34-3 case; increasing the number of layers to four but not
changing any other settings yielded the 4-3-4-4 case;
doubling the nanoparticle density of each layer produced
the 8-3-4-2 case; doubling the nanoparticle height yielded
the 4-3-8-2 case; adding one more nanoparticle to the middle
of the simulation box and increasing the number of layers to
three on the basis of the 4-3-8-2 case produced the 4-3-8-3
case; and doubling the nanoparticle radius on the basis of the
4-3-8-2 case gave the 4-6-8-2 case. The parameters chosen
were the same as those used in the previous mesoscopic
simulations; the value was that obtained at 400 K, and
the total simulation time was 10 ms in each case. The simulation models were the plain PS/PMMA blends shown in
Table 1.
Result and discussion
Figure 10 shows the V ROP values obtained for various
kinds of nanoparticle doping situations. A few trends can be
seen in these cases. First, the trends seen in these four
pictures are almost the same. Second, since V ROP1/3 > V
ROP1/2 > V ROP2/1 > V ROP3/1, it can be deduced that the
degree of phase separation intensifies with increasing
PMMA content when the blend is rich in PMMA. Third,
since V ROP4-3-4-3 > V ROP4-3-4-2 and V ROP4-3-8-3 > V
ROP4-3-8-2, the order of the PS/PMMA blend increases
when the number of nanoparticles of the same size is
Table 2 Mesoscopic simulations of PS/PMMA blends doped with
nanoparticles

Fig. 9 V ROP data with the shear effect versus the volume fraction of
PS. Isodensity surface pictures of an 8/1 (PS/PMMA) blend induced
by a shear rate of 0.002 ns1 are also shown, as are isodensity surface
pictures of an 8/1 (PS/PMMA) blend at 443, 413 and 383 K (top to
bottom in the chart, respectively). Red, PS; green, PMMA

Simulation

Case

Np

rp (nm)

hp (nm)

NL

Ntp

1
2
3
4
5
6
7

4-3-4-2
4-3-4-3
4-3-4-4
4-3-8-2
4-3-8-3
4-6-8-2
8-3-4-2

4
4
4
4
4
4
8

3
3
3
3
3
6
3

4
4
4
8
8
8
4

2
3
4
2
3
2
2

8
9
16
8
9
8
16

J Mol Model (2011) 17:607619

615

Fig. 10 V ROP data for the 1/2, 2/1 and the 1/3, 3/1 blend pairs doped with nanoparticles at 383, 413 and 443 K

increased. Fourth, since V ROP8-3-4-2 > V ROP4-3-4-4, the

order of the PS/PMMA blend can be increased by
increasing the density of the nanoparticles of the same
size. Fifth, it can be seen from the extremely high value of
V ROP4-6-8-2 that the nanoparticle size is the factor that
exerts the most influence on the phase morphology.
Table 3 Mesoscopic simulations
of PS/PMMA blends induced
by ci and co planes

g
This is the abbreviation for
effective/total row number.

ci type
ci-444
ci-882
co type
co-444
co-448
co-4412
co-4432
co-884
co-888
co-8812
co-8832

From the third, fourth and fifth results discussed above

(V ROP4-6-8-2 V ROP8-3-4-2 > V ROP4-3-4-4 > V ROP4-3-8-3
> V ROP4-3-8-2 > VROP4-3-4-3 > V ROP4-3-4-2), we can see
that the factors investigated had the following order of
influence on the phase separation: nanoparticle size >
nanoparticle density > number of doped nanoparticles.

Row number (X-axis)

4
8
E/Tg (X-axis)
2/4
2/4
2/4
2/4
4/8
4/8
4/8
4/8

Row number (Y-axis)

4
8
E/Tg (Y-axis)
2/4
2/4
2/4
2/4
4/8
4/8
4/8
4/8

Radius (nm)
4
2
Height (nm)
4
8
12
32
4
8
12
32

616
Table 4 Mesoscopic simulations
of PS/PMMA blends induced
by gra and rg planes

This is the abbreviation for

symmetry; N means no and
Y means yes

J Mol Model (2011) 17:607619

gra type

Row number (X-axis)

Height (nm)

Sfin X-axis

gra-444
gra-888
gra-2(444)
gra-2(448)
gra-2(888)
rg type
rg-442
rg-884
rg-16168

4
8
8
8
16
Layer number
2
4
8

4,3,2,1
8,7,6,5,4,3,2,1
4,3,2,1,1,2,3,4
8,4,2,1,1,2,4,8
8,7,6,5,4,3,2,1,1,2,3,4,5,6,7,8
Height (nm)
2,4
8,4,2,1
16,14,12,10,8,6,4,2

N
N
Y
Y
Y
Sf in the X- and Y-axes
Y
Y
Y

Furthermore, all of the PS/PMMA blend cases showed the

same trends as those described above with nanoparticle
doping.
PS/PMMA blends induced by various planes
To investigate which planes induce the most intense phase
separation in PS/PMMA blends, four different types of
inducing planes were designed as substrates using our own
program. These four types were denoted ci, co, gra,
and rg. The ci series of planes used half-spheres with
different radii as a mask that simulated different degrees of
surface roughness. The co series had equally spaced
cubic columns as a mask. The columns had different sizes
and heights to simulate different degrees of surface
roughness. The gra series were planes with different
widths to simulate different degrees of surface roughness.
The mask was generated by gradually increasing the
column height across the plane, so that it resembled stairs
viewed side on. In addition, monodirectional asymmetric
planes, such as gra-444 and gra-888, and monodirectional
symmetric planes, such as gra-2(444), gra-2(448) and gra-2
(888), were considered. The rg series were bidirectional
symmetric planes originating from monodirectional symmetric planes, as used in the gra series. Tables 3 and 4
describe these four types of planes used in the simulations,
and Fig. 11 shows the eighteen planes pictorially.
Results and discussion
Figure 12 shows the V ROP values of PS/PMMA blends
induced by the four types of planes mentioned above;
eighteen isodensity surface pictures are shown for the 1/3
blend case at 413 K as examples at the top of each chart.
The following trends can be discerned from these figures:
(1) For the co planes, V ROPco-8832 > V ROPco-8812 > V
ROPco-888 > V ROPco-884 and V ROPco-4432 > V
ROPco-4412 > V ROPco-448 > V ROPco-444, which
shows that the PS/PMMA blends underwent more

vigorous phase separation on rougher planes. The

values of V ROPco-884 and V ROPco-888 were almost
the same at 383, 413 and 443 K. The values of
V ROPco-8812 at 383 and 413 K were also the same, but
the value of V ROPco-8812 at 443 K jumped to 13.3,
which was one order of magnitude higher than its
values at 383 and 413 K. On a much rougher surface,
such as a co-8832 type of plane, V ROPco-8832,443K >
V ROPco-8832,413K > V ROPco-8832,383K, which reveals
that PS/PMMA blends undergo more intense phase
separation at higher temperatures. The surfaces of the
co-4xx series of planes were rougher than those of the co8xx series of planes. The V ROP values of the co-4xx
series were higher than those of the co-8xx series at
443 K, which is also consistent with the conclusion that
more intense phase separation occurs on rougher planes.
(2) For the ci planes, V ROPci-882 > V ROPci-444 at the
same temperature. There was no obvious difference
between the V ROPci-882 values at 383 and 443 K, and
there was no obvious difference between the V ROPci444 values at 383 and 443 K either. Furthermore, the V
ROPci-882 and V ROPci-444 values at 413 K were both
higher than their corresponding values at lower
(383 K) or higher (443 K) temperatures. This shows
that temperature is the most influential factor on the
phase separation of PS/PMMA blends induced by ci
planes, but also that the V ROP values vary irregularly
with temperature. Both the temperature and the degree
of surface roughness contribute to the results.
(3) For the rg planes, the same order of V ROPrg-16168 >
V ROPrg-884 > V ROPrg-442 was seen at three temperatures, which shows that the degree of surface roughness
is an important factor for inducing phase separation,
especially for rougher planes with high step gradients,
such as rg-16168. Nearly the same value of V ROPrg-442
was observed at 383 and 443 Kthe same situation as
seen for V ROPrg-884so it can be presumed that for
the smoother planes, such as rg-884 and rg-442,
temperature is more important than the degree of
surface roughness in inducing phase separation.

J Mol Model (2011) 17:607619

617

Fig. 11 The four types of plane used in the simulations: ci, rg, gra, and co

(4) For the gr planes, the V ROP values for gr-444 were
lower than those for other gr planes at 383, 413 and
443 K, and all of these planes presented similar values at
383 and 443 K; in addition, the other gr planes except
for gr-444 presented nearly the same V ROP values at
383, 413 and 443 K. This shows that the degree of

surface roughness plays a small role in inducing phase

separation for gr planes. In contrast, it can be deduced
from the extremely high value of V ROP at 413 K that
temperature is the main inducing factor for these planes.
(5) For all of the ci, rg, and gr planes, the V ROP
values at 413 K were much higher than those at 383

618

J Mol Model (2011) 17:607619

Fig. 12 V ROP data for the 1/3 blend case at 383, 413 and 443 K, respectively. Isodensity surface pictures are shown for PS/PMMA blends
induced by different types of planes at 413 K. Red, PS; green, PMMA

and 443 K; furthermore, the V ROP values at 383 and

443 K were nearly the same for each plane, which
reveals that using a higher (443 K) or lower (383 K)
temperature had nearly the same effect on phase
separation, and that there was a critical temperature
between 443 and 383 K (such as 413 K) that could
induce the most vigorous phase separation. The
influence of each ci, rg, or gr plane on phase
separation mainly depended on this critical temperature, not on the degree of surface roughness, in
contrast to what was seen for the co planes.
The ci, rg, and gr planes were different from the
co planes in that the first three types of plane had
continuous mask distributions on their surfaces, which
allowed the PS/PMMA blends to move and spread more
freely until they could adjust themselves to reach stable
phase separation. In contrast, the co planes did not have
continuous mask distributions on their surfaces, which
meant that the polymer did not have enough room to move

and spread; they could only adjust themselves in the small

cubic spaces formed by neighboring column masks.
Furthermore, the big difference between the diffusion
abilities of PS and PMMA was also a barrier to movement
and adjustment. Because of the two reasons mentioned
above, the degrees of phase separation differed widely
between the first three types of plane and the co planes.

Conclusion
The two most important processes during the MD simulation are the annealing process after minimization and the
MD simulations with three NVT and NPT ensemble cycles.
The strong polymer endend interactions, which evidently
influence the simulation results, can only be reduced by
performing the above two processes. Based on the
calculated values for different numbers of repeating units
of PS and PMMA, their representative chain lengths were
found to be 30 and 60, respectively. By comparing the

J Mol Model (2011) 17:607619

calculated with critical, it was found that the PS/PMMA

blends were immiscible in our simulations. Furthermore,
the data did not depend greatly on the simulation
temperatures.
The input parameters used in MesoDyn were derived from
the MD simulations, and mesoscopic simulations were carried
out at 383, 413 and 443 K. After calculating the V ROP values
of ten different representative composition cases with block
copolymer added, two outstanding features were found.
First, the PS-rich blend cases at higher simulation temperatures, such as 413 and 443 K, underwent a more obvious
phase separation in microscopic phase morphology. Second,
when the 1/1 blend case was used as the reference, the
symmetric blend pairs, such as 1/2, 2/1 and 1/3, 3/1, etc.,
gradually showed miscibility as the temperature was increased.
Mesoscopic simulations were then carried out on PS/
PMMA blends with shear stress factored in. The results of
those simulations showed that this shearing was a good way of
inducing phase separation, especially in PS-rich blends. Here,
the temperature was not found to be an important factor in
phase separation, especially for the PMMA-rich blends.
Next, mesoscopic simulations were carried out on plain
PS/PMMA blends doped with various numbers and densities
of nanoparticles of various sizes. The simulation results
showed that nanoparticle doping was an effective way of
improving the degree of order in the microscopic phases,
especially in the PMMA-rich blends. No matter which PS/
PMMA blend composition was considered, the nanoparticle
size had a stronger influence on the phase morphology than
the density of nanoparticles, and the density of nanoparticles
had a stronger influence than the number of nanoparticles.
Finally, in order to research the effects of various planes on
the phase separation of PS/PMMA blends, four types (ci,
rg, gr and co) of planes were constructed. For the first
three types of planes, the degree of surface roughness played
only a small role in inducing phase separation: the temperature
was the most important factor in this respect. Furthermore,
there was a critical temperature between 443 and 383 K
(such as 413 K) at which the phase separation occurred most
intensely. However, rougher planes and higher temperatures
both helped to induce phase separation on co planes.

References
1. Paul DR, Newman S (1978) Polymer blends. Academic, New
York

619
2. Abd-El-Messieh SL (2003) Polym Plast Technol Eng 42(1):153169
3. Schneider IA, Calugaru EM (1976) Eur Polym J 12(12):879881
4. Bada R, Perez Jubindo MA, De LA, Fuente MR (1987) Mater
Chem Phys 18(4):359373
5. Zhu PP (1997) Eur Polym J 33(3):411413
6. Lee JK, Han CD (1999) Polymer 40:62776296
7. Lee CF (2000) Polymer 41(4):13371344
8. Li X, Han YC, An LJ (2003) Polymer 44(26):81558165
9. Walheim S, Boltau M, Mlynek J, Krausch G, Steiner U (1997)
Macromolecules 30:49955003
10. Schmidt JJ, Gardella JA Jr, Salvati L Jr (1989) Macromolecules
22:44894495
11. Chiou JS, Barlow JW, Paul DR (1987) J Polym Sci Part B Polym
Phys 25:14591471
12. Lhoest JB, Bertrand P, Weng LT, Dewez JL (1995) Macromolecules 28:46314637
13. Davies MC, Shakesheff KM, Shard A, Domb A, Roberts CJ,
Tendler SJB, Williams PM (2000) Macromolecules 29:22052212
14. Schreiber HP (1993) Aspects of component interactions in
polymer systems: the interfacial interactions in polymeric composites. Kluwer, Boston
15. Buki L, Gonczy E, Fekete E, Hellmann GP, Pukanszky B (2001)
Macromol Symp 170:920
16. Lodge TP, Wood ER, Haley JC (2006) J Polym Sci Part B Polym
Phys 44:756763
17. Valls OT, Farrell JE (1993) Phys Rev E 47:R36R39
18. Ramirez-Piscina L, Hernandz-Machado A, Sancho JM (1993)
Phys Rev B 48:125131
19. Kawakatsu T, Kawasaki K, Furusaka M, Okabayashi H, Kanaya T
(1993) J Chem Phys 99(10):82008217
20. Shinozaki A, Oono Y (1993) Phys Rev E 48:26222654
21. Fraaije JEM (1993) J Chem Phys 99:92029212
22. Fraaije JGEM, van Vlimmeren BAC, Maurits NM, Postma M,
Evers OA, Hoff-man C, Altevogt P, Goldbeck-Wood G (1997)
J Chem Phys 106:42604269
23. Jawalkar SS, Adoor SG, Sairam M, Nadagouda MN, Aminabhavi TM
(2005) J Phys Chem B 109(32):1561115620
24. Groot RD, Warren PB (1997) J Chem Phys 107:44234435
25. Groot RD, Madden TJ (1998) J Chem Phys 108:87138724
26. Mu D, Huang XR, Lu ZY, Sun CC (2008) Chem Phys 348:122129
27. Flory PJ (1989) Statistical mechanics of chain molecules. Munich,
Germany
28. Mark JE (1999) Polymer data handbook. Oxford University Press,
New York
29. Brandrup J, Immergut EH, Grulke EA (1999) Polymer handbook,
4th edn. Wiley, New York
30. Hoover WG (1985) Phys Rev A 31:16951697
31. Rigby D, Sun H, Eichinger BE (1997) Polym Int 44:311330
32. Sun H (1998) J Phys Chem B 102:73387364
33. Sun H, Ren P, Fried JR (1998) Comput Theor Polym Sci 8:229246
34. Ewald PP (1921) Ann Phys 64:253287
35. Grulke EA (1999) Solubility parameter values. In: Polymer
handbook, 4th edn. Wiley, New York
36. Calahorra E, Cortazar M, Guzman GM (1985) J Polym Sci Polym
Lett Ed 23:257260
37. Krause S, Paul DR, Newman S (1978) Polymerpolymer
compatibility in polymer blends, vol 1. Academic, New York
38. Fan CF, Cagin T, Chen ZM, Smith KA (1994) Macromolecules
27:23832391