Current Opinion in Colloid & Interface Science 17 (2012) 132–140

Contents lists available at SciVerse ScienceDirect

Current Opinion in Colloid & Interface Science

journal homepage: www.elsevier.com/locate/cocis

Free volume and mass transport in polymer nanocomposites

G. Choudalakis ⁎, A.D. Gotsis
Department of Sciences, Technical University of Crete, GR-73100, Hania, Greece

a r t i c l e i n f o a b s t r a c t

Article history: This review relates the free volume properties and the morphology to the mass transport coefficients of poly-
Received 14 December 2011 mer nanocomposites. Direct, utilising the method of Positron Annihilation Life-time Spectroscopy (PALS), or
Received in revised form 13 January 2012 indirect measurements of the free volume in the nanocomposites are discussed and the influence of spherical
Accepted 16 January 2012
or anisometric nanoparticles on its properties is examined.
Available online 26 January 2012
© 2012 Elsevier Ltd. All rights reserved.
Keywords:
Free volume
Positron annihilation
Gas transport
Diffusion
Solubility
Permeability
Nanocomposites

1. Introduction: polymer nanocomposites
Polymer nanocomposites materials are hybrid systems that consist of
a polymeric matrix and dispersed inorganic particles of nanometer scale.
When the dispersed inorganic phase consists of spherical nanoparticles
with diameter of the order of 50 nm or less, e.g. fumed silica (FS), carbon
black (CB), TiO2, the materials are referred to as Polymer Nanocompo-
sites (PN). If the dispersed particles originate from the family of phyllosi-
licates and have flat shape with high surface area, the composites are
called Polymer Clay Nanocomposites (PCN). Some common particles
used for PCNs are montmorillonite (MMT), rectorite, hectorite, vermicu-
lite etc.
The creation of a PN or PCN is not a simple issue. In powder form the
inorganic phase consists of aggregates or stacks of nanoparticles held to-
gether by electrostatic forces. The formation of the nanocomposite
requires the deagglomeration of the primary nano-particles or nano-
layers (Fig. 1). This can be done if the electrostatic forces between the
particles are weakened and the interactions with the organic polymer
are enhanced. To exfoliate the clay, therefore, modification of the sur-
faces of the mineral particulates is needed. The objectives of the modifi-
cation process are (i) the reduction of the electrostatic forces (swelling)
and (ii) the compatibilisation of the filler with the matrix. The latter de-
termines the degree of interaction between the two phases, which is cru-
cial for the final nanocomposite morphology.
The nanocomposites can have excellent mechanical properties. The
nano-dimensions maximise the number of available reinforcing
⁎ Corresponding author.
E-mail address: gchoudalakis@isc.tuc.gr (G. Choudalakis).
elements that carry the applied load in the polymer, while the particles
also deflect the evolving cracks. The stress transfer to the reinforcement
phase results in significant tensile and toughening improvement. An-
other major advantage of the nanocomposites is their diverse barrier
properties. Spherical nanoparticles can enhance the permeation of
gases through the PN, the rate depending on the size of the gas mole-
cule. These materials are good candidates for gas separation applica-
tions. On the other hand, PCNs reduce the gas permeation rate
because the impermeable clay layers, with their large surface area,
force a tortuous pathway for a permeant. These materials are ideal as
barrier layers in packaging against the penetration of unwanted gases,
e.g. O2, or to prevent gasses like CO2 or H2 to escape. Recent reviews of
the methods used for synthesis of PN and PCN and for their various
properties can be found in the literature [1, e.g.].
2. Free volume
The term “free volume” (f.v.) in a polymer refers to the volume of the
total mass that is not actually occupied by the molecules themselves. This
is the room that the parts of the chains have to move around. The concept
of the free volume is not uniquely defined. Various definitions are often
used, such as hole, configurational, fluctuation and excess free volume:
Hole free volume: even if the polymer chains were perfectly
aligned, there is free space (holes) between them.
Configurational free volume: additional free space is created, due to
the insufficient chain packing.
Fluctuation (dynamic) free volume: the polymer main- or side-
chains are not fixed but can move (rotate, vibrate etc.) randomly,

1359-0294/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cocis.2012.01.004
 G. Choudalakis, A.D. Gotsis / Current Opinion in Colloid & Interface Science 17 (2012) 132–140
Fig. 1. Delamination of the primary mineral particles into individual nano-layers. The
layers are located on top of each other like the pages of a book. The Van der Waals
gaps between the layers, are called galleries.
due to thermal activation. These motions generate transient gaps,
which create extra free volume.
Excess free volume: at temperatures below Tg, the polymer usually
has an actual volume greater than what it would have had, if it had
been allowed to cool under equilibrium conditions. The difference
between equilibrium free volume and the actual volume under
quenched conditions is the “excess free volume” (Fig. 2).
In time, the polymer molecules will try to rearrange themselves, to
approach the equilibrium volume of the material, thus eliminating
the excess free volume. This time dependent slow decrease in volume
is known as “physical ageing” and affects all material properties
which change drastically around Tg.
The sum of hole and configurational f.v. is the “static” free volume.
The excess f.v. may be static or dynamic, depending on the time scale
of the relaxation processes. The ratio of static to dynamic excess f.v. is
thought to be greater than 1 and depends on temperature.
The static f.v., Vf = V − V0, with V the specific volume of the poly-
mer and V0 the volume occupied by the atoms of the polymer chains
at 0 K, is often expressed in terms of the Van der Waals volume, Vw,
calculated using the group contribution method [2] as V0 = 1.3Vw.
The coefficient 1.3 is arbitrary and may sometimes result in faulty in-
terpretations. The fractional free volume is: FFV = Vf/V.
The size, shape and position of the polymer f.v. holes vary with
time. Around the glass transition temperature, Tg, the polymer un-
dergoes a dramatic transformation, as the degrees of freedom of the
id
Polymer Volume
qu
Li
rim
lib
ui
Eq
m Glass
uilibriu
Non Eq
Excess Free
Volume
Glass
ibrium
Equil
Tg
Temperature
Fig. 2. Definition of the excess free volume.
133
molecular chains and the resulting mobility change considerably
and abruptly.
High Tg polymers consist of inflexible chains and are limited in
their chain mobility. Low Tg polymers have, in general, more flexible
chains and present higher segmental mobility. The higher the chain
stiffness, the higher the Tg and the broader the distribution of f.v.
hole sizes, since the available variety of chain conformations is
restricted.
At T b Tg the segmental motion is frozen and thermal expansion of
free volume is almost impossible. In this case, the size, shape and
position of the f.v. holes remain constant (very long relaxation
times) and a definite, time-averaged Gaussian size distribution of
holes, F(Vh), is established. Assuming N spherical holes:
" #
2
N ðV −V Þ
F ðV h Þ ¼ pffiffiffiffiffiffi exp − h 2 0 : ð2:1Þ
σ 2π 2σ
The width of this distribution is:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffi
V h RT g
σ¼ ; ð2:2Þ
B
with B the bulk modulus at Tg and Vh the average hole size [3 •].
Amorphous polymers in the rubbery state have high chain mobil-
ity and short relaxation times. Motions of much longer segments of
the backbone chain can take place. These motions are directly related
to the increase in free volume caused by the thermal expansion of the
system.
 
FFV ¼ ðFFV ÞT g þ T−T g α T ð2:3Þ
with αT the thermal expansion coefficient for free volume. The WLF
theory gives (FFV)Tg = 0.025, αT = 4.8 × 10 − 4 K − 1. The positions of
the free volume holes change with time following the motions of
the polymer chains. That is, the holes move freely inside the polymer,
as no energy change is required for their redistribution.
At a given temperature the number of the holes remains constant,
while their size, shape and position vary with time. The number of
holes increases with temperature, i.e. additional free volume sites
are created. The increase is induced by thermal activation (Brownian
segmental motion) and not by an orientational effect. The size of the
free volume holes is sensitive to the applied stresses during mechan-
ical deformation but the number of the holes is not affected [4].
Crosslinking does not contribute to changing the free volume size in
the glassy state, since the macromolecular motions are frozen (vitri-
fied). In the rubbery state, crosslinking can decrease the free volume,
as it confines the mobility of chain segments. The effect is stronger
when starting with a looser, mobile polymer network (e.g. crosslinking
flexible linear chains). In rigid network structures (e.g. thermosetting
polymers) the influence of crosslinking is expected to be smaller.
3. Mass transfer coefficients
Gas transport through a polymer membrane is a complex process
that includes the sorption of gas molecules on the surface of the mem-
brane; the diffusion through it; and, finally, the desorption of gas from
the other surface of the membrane. The permeability coefficient, P
(mol Pa − 1m− 1s− 1), of the gas molecules through the membrane is
the product of the diffusion coefficient, D (m2s− 1), and the sorption co-
efficient or solubility, S (mol m 3Pa− 1):
P ¼ D⋅S: ð3:1Þ
These three transport coefficients are discussed separately and
their connection with the polymer free volume is analysed in the fol-
lowing paragraphs.
134 G. Choudalakis, A.D. Gotsis / Current Opinion in Colloid & Interface Science 17 (2012) 132–140
3.1. Solubility coefficient
In the rubbery state of the polymer the solubility coefficient of the
gas is determined by Henry's law, C = kdp, with C the concentration of
the sorbed gas; kd Henry's constant (mol m 3 Pa − 1); and p the pres-
sure. In the glassy state we cannot neglect Langmuir's contribution
of the concentration of the sorbed gas:
CH b
C¼ p: ð3:2Þ
1 þ bp
The parameter CH characterises the amount of unrelaxed free vol-
ume in the glassy state (excess free volume):
!
V g −V l 22400
CH ¼ : ð3:3Þ
Vg Vp
Vg and Vl are the specific volumes in the glassy and the liquid (rubbery)
states and Vp is the molar volume of the sorbent gas (ml/mol). The pa-
rameter b characterises the tendency of a given penetrant to sorb into
the excess unrelaxed volume in the non equilibrium matrix. Taking
into account both contributions, the solubility coefficient in the glassy
state becomes:
CH b
S ¼ kd þ : ð3:4Þ
1 þ bp
The relations between solubility and free volume are not always
obvious, since the way penetrants dissolve in the polymer depends
on the total pressure, the affinity between gas and polymer and the
temperature. For a specific polymer nanocomposite system, however,
the solubility coefficient will change only if the f.v. changes due to the
presence of the particles. Henry's constant depends on the facility of
forming holes in the space among chains [5] and cannot be associated
with these changes. For f.v. holes that are all large enough to accom-
modate gas molecules, the solubility coefficient should reflect directly
their number, remaining unaffected by changes in their sizes. Other-
wise, kd has a tendency to increase with increasing average hole
size as well.
The connection between the solubility coefficient and the number of
free volume holes has been verified experimentally in acrylic resin/
laponite nanocomposites [6•]. The total f.v. fraction, measured by Posi-
tron Annihilation Lifetime Spectroscopy, has an identical trend, as a
function of laponite particles loading, with that of the solubility coeffi-
cient, which is measured separately in a permeation cell. On the other
hand, Langmuir's sorption is principally governed by the available free
volume. Consequently, the variation in solubility is attributed mostly
to CH, which is determined by the difference Vg − Vl (Eq. (3.3)). This,
in turn, is influenced by the Tg. Therefore, higher Tg polymers generally
exhibit higher solubility coefficients. It can be stated that the FFV is re-
lated to the solubility because it corresponds to the free space in the
polymer matrix where gas molecules can locate [5].
3.2. Diffusion coefficient
In contrast to the solubility, the diffusion is a kinetic process which
depends on the mobility of gas molecules in the polymer matrix. At
temperatures below Tg, the polymer backbone is considered to be in
a frozen state, segmental chain motions are drastically reduced, the
number of f.v. holes is fixed and no hole redistribution is likely. Gas
transport is, therefore, assumed to take place via fixed (pre-existing)
holes. A gas molecule must “find its way” from hole to hole along
pathways involving only minor segmental rearrangements (Fig. 3).
This means that the diffusivity depends largely on the number of
the holes with an appropriate size, able to accommodate the diffusing
gas molecule. This mechanism is expected when the diffusion rates
Fig. 3. Schematic representation of the diffusion process of a gas molecule through free
volume in a polymer.
are much slower than those of polymer relaxations (Fickian diffu-
sion). When diffusion and relaxation rates are comparable the diffu-
sion is non-Fickian.
A molecule can jump into a hole if the hole is larger than a critical
size υ*. The probability for such a process is proportional to
exp(− γυ*/υf), where υf is the molar mean hole free volume and γ is
an overlap factor (0 b γ b 1) indicating that the same hole may be
available to more than one molecules. The diffusion coefficient is [2]:
!
 
−E −γξυ
D ¼ D~0 exp exp ; ð3:5Þ
kT υf
where ξ = υs*/υ*; and υs* is the critical volume of the penetrant. The ac-
tivation energy, E*, for the diffusion of trapped molecules from one hole
to another is related to the molecular “jumping length”. The proportion-
ality coefficient D~0 is related to the size and shape of the permeant:
RTσ 2
D~0 ¼ 1=2 : ð3:6Þ
M
where σ is the Lennard–Jones size parameter and M is the molecular
weight of the permeant [2].
The amount of free volume alone does not provide information
about the possible connectivity of the holes. When f.v. elements are
added randomly to a system, the percolation threshold will be reached
at some point and a continuous pathway will exist for the permeating
gas in the system. This is commonly identified with a pore-flow diffu-
sion process.
Diffusivity should be correlated to the accessible (for a given gas)
free volume fraction, AFV. For glassy polymers the percolation thresh-
old is in the range of 2–4% AFV. This is equivalent to 25–31% FFV [7 •].
In the rubbery state the picture of fixed holes needs to be replaced,
since the polymer chains are mobile and the f.v. holes show a dynamic
variation about their size, shape and position. The gas molecules diffuse
within the fluctuating interstitial f.v. with much greater mobility than in
the glassy state. That is to say, the jumping step is larger in polymers
with a stiff backbone chain than in flexible chains [2]. The factor
exp(− E*/kT) can now be neglected, since less energy is required for dif-
fusional jumping, and Eq. (3.5) is often written in the following form:
D ¼ ART expð−B=FFV Þ; ð3:7Þ
where B is equal to γυs*/υ* and A is related to the size and shape of the
permeant and replaces the term σ2/M1/2 of Eq. (3.6). It seems that B de-
pends not only on the type of gas but also on the polymer, since υ* is the
minimal free volume hole size required for the penetrant diffusion. The
size of the diffusing molecule has a smaller influence if the polymer is in
 G. Choudalakis, A.D. Gotsis / Current Opinion in Colloid & Interface Science 17 (2012) 132–140
the rubbery state than on a rigid glassy membrane, where the hole size
distribution is frozen. Glassy polymers seem to be more suitable for gas
separation membranes.
While correlations between lnD and 1/FFV of the type of Eq. (3.7)
have been verified for various glassy polymers the values of parame-
ters A and B show a significant variance. A linear fit of data for the dif-
fusion of CO2 in four polymers (PS, PMMA, PC, PP) gives A = 4⋅ 10 − 6
cm 2/s and B = 0.97 [2]. On the other hand, values of A = 3.9⋅ 10 − 7
cm 2/s, B = 0.297 were reported for diffusion of CO2 in polycarbonates
[8], and A = 7.4⋅ 10 − 5 cm 2/s, B = 0.1336 in polyurethanes [9]. This
variance indicates that the parameters depend also on the polymer
and not only on the type of gas.
There is also the case where the free volume in the material consists
of a network of large holes connected by channel-like holes. Diffusing
molecules may now move with little friction within the network.
Large variations in the size or in the number of larger holes have small
effect on the diffusion through the free volume network, whereas a
small decrease in the size of the smaller holes may destroy the connec-
tivity of the channel network. In this case the mobility of diffusing mol-
ecules is strongly hindered [10]. This means that a broader distribution
of free volume holes can result in higher values of the gas diffusion co-
efficient than a narrow distribution with the same average size. Both the
amount of the free volume and its distribution over the hole sizes are
important factors that influence the diffusion coefficient.
4. Direct measurement of the free volume using PALS
Positron Annihilation Lifetime Spectroscopy (PALS) is an accept-
able method for the direct measurement of polymer free volume.
The positrons are used for bulk measurements as they can penetrate
a few tenths of a mm inside the sample. When a positron meets an
electron, they either annihilate generating photons or they form a
temporary pair, a positronium, Ps. There are two possible Ps atoms:
p-Ps corresponding to the antiparallel spin configuration of the two
particles, and o-Ps corresponding to the parallel. Ps formation can be
achieved only in low electron density regions such as in voids or in
the free volume of the polymer (Fig. 4) and its annihilation is accom-
panied by the emission of 511 keV γ-photons. If no such areas are
present, then the positrons will be annihilated directly.
To estimate the lifetimes, the spectrum of the emissions due to
positron annihilations is resolved into N Debye decaying processes:
 
X
N
Ii t
F ðt Þ ¼ exp − : ð4:1Þ
i¼1
τi τi
Thermalization
511 keV
e where Ra = 8 Å and b = 0.55 are fitted parameters.
e
+
+
e mation and results in a decrease of I3, is often observed upon addition
Diffusion
511 keV
Trapping and Annihilation
Fig. 4. The process from the time of positron implantation to the time of Ps annihilation.
135
Here Ii is the relative intensity and τi is the lifetime of the i-th Debye
process. In polymers the lifetime spectrum (Eq. (4.1)) is usually re-
solved into three terms. The first term (I1, τ1) has a relaxation time (life-
time) τ1 ∼ 0.125 ns and is related to the fast annihilation of the p-Ps and
the free positrons. The second process (I2, τ2) with a relaxation time
τ2 ∼ 0.4 ns corresponds to the direct annihilation of the free positrons.
The third process (I3, τ3) is the “pick-off” annihilation process of the o-
Ps and has a relaxation time τ3 in the range of 1 to 5 ns. In some rare
cases of high free volume polymers there is a fourth significant lifetime
component, (I4, τ4), with τ4 greater than 10 ns [11 etc.] Lifetime, τ4, is
ascribed to the bimodal distribution of possibly non spherical free vol-
ume elements.
To o-Ps lifetime is related to the size, R, of the free volume hole
where it is trapped (Tao-Eldrup model):
" #
2πR −1
1 R sin RþΔR
τ3 ¼ 1− þ ; ð4:2Þ
2 R þ ΔR 2π
where ΔR ∼ 0.1656 nm is the electron layer thickness around the hole,
usually assumed to be uniform. This model is sufficient for characteris-
ing small voids with cavity diameter smaller than the thermal de Broglie
wavelength of Ps (∼ 6 nm). There is also a low threshold for detecting
f.v. holes by PALS, equal to the minimum for o-Ps localisation Rc:
!1=2
π ℏ2
Rc ¼ : ð4:3Þ
2 2mV~ 0
ℏ is Planck's constant, m the electron mass and V~ 0 the potential barrier
(∼ 1.4 eV). Typically, Rc ≈ 2 Å.
The intensity I3 is related to the formation probability of o-Ps in the
f.v. holes and is connected to their number. The fractional free volume
in polymers is FFV∼ VhI3, where Vh = (4/3)πR3, in Å3, is the volume of
the average spherical hole. The total fractional free volume cannot be
estimated, since I3 originates only from the holes, where the o-Ps can
be formed, which are not the total free volume. Thus, a proportionality
parameter, C, is used:
FFV ¼ CV h I3 ; ð4:4Þ
where C depends on the polymer and can be deduced by comparing the
values of FFV from PALS measurements with values of FFV from P–V–T
data. Indicative values are 0.325 nm− 3 for poly(vinyl acetate) [12]
and 1.8 nm− 3 for amine-cured epoxy polymers [13].
When a fourth mode (I4, τ4) is found, the size can be estimated by
an extended Tao-Eldrup model [14]:
 "  #−1
R−Ra b 1
τ4 ¼ 0:0626 1− þ ; ð4:5Þ
R þ ΔR 140
In molecular solids, a decrease in I3 can be due either to a decrease in
the number of o-Ps annihilation sites or to a decrease in the probability
of Ps formation. Negatively charged polar groups, M− (nitro, halogen
and hydroxyl), can suppress Ps formation considerably by arresting
the positrons. This competing mechanism, which reduces the o-Ps for-
of salts. The latter increase the dipole moment of the polymer molecules
enhancing localisation of the negative charges. This decrease in I3 is ap-
proximately exponential with anion concentration, approaching a satu-
ration value [15].
Besides its effect on I3, the chemistry of the material also affects τ3. A
stronger polarity effectively quenches the Ps atoms and, thus, shortens
the Ps lifetimes. The size of the cavities may remain constant, but the
ortho-Ps lifetimes can be quenched by various oxidisers, including
136 G. Choudalakis, A.D. Gotsis / Current Opinion in Colloid & Interface Science 17 (2012) 132–140

paramagnetic ions and chlorides. Oxygen is particularly efficient in pro-
moting the conversion of o-Ps to p-Ps [10]:
 particles perturb the polymer chain packing and can reduce the
o−Ps þ O2 → ½Ps−O2  → p−Ps þ O2 ;
At high temperatures, well above Tg, the o-Ps lifetime may reflect the
formation of so-called o-Ps bubbles. When the surface tension is small
enough, the repulsive forces between the Ps and the surrounding poly-
mer chains are sufficient to form a “bubble” around the o-Ps atom. This
bubble formation is only possible if the segmental relaxation times are
in the order of, or shorter than the o-Ps lifetime [16]. Obviously, when
bubbles are present, the o-Ps lifetime does not reflect the actual size of
a f.v. hole.
5. Effects of nanoparticles on polymer free volume
The incorporation of inorganic nano-fillers into the polymer matrix,
will inevitably change the morphological features of the matrix and,
consequently, the free volume properties. The following effects are
expected in the nano-composite:
Changes of the crystallinity of the matrix: the incorporation of nano-
particles often changes the crystallinity of the matrix polymer. The
ð4:6Þ
crystallinity in the polymer, alleviating, thus, some of the decrease
in chain mobility that accompanies crystallisation. Inevitably, this
enhances the free volume. On the other hand, the nanofillers may
also act as nucleating agents, increasing the degree of crystallinity
and reducing the free volume. The net result will depend on the
crystallisation characteristics of the matrix and its interaction with
nucleation agents.
Changes of the cross-linking density of the matrix: Adding nano-
particles in a thermosetting polymer can have two opposite re-
sults. The particles may serve as crosslinkers, which increases
the cross-link density and decreases the polymer mobility and
the free volume. Or they may hinder the cross-linking reaction
and enhance the free volume. The net effect depends on the chem-
istry of the crosslinking reaction, the possible catalytic action of
the nano-particles and the interfacial properties of the system.

Interfacial regions: because of their large surface area the inorganic
particles create interfacial regions in the organic polymer. These
regions are numerous and have finite thickness because of the
weak interaction/miscibility between the inorganic reinforcement
and the organic polymer and the need for compatibilisation. The
interfacial layers help the stress transfer between the nanoparti-
cles and the matrix, but contribute to the enhancement of the
overall free volume. Since clays create more interfacial area than
spherical particles, the interfacial effect on free volume is expected
to be stronger in PCN than in PN.
Interstitial cavities in the filler agglomerates: if exfoliation in a PCN is
not complete, stacks of nano-layers (intercalated morphology)
exist inside the polymer matrix. The interlayer spacing of these
stacks varies from 2 to 5 nm. If polymer chain penetration in the gal-
leries is incomplete, the volume of the galleries contributes to the
overall free volume of the composite. This extra f.v. is in layers rather
than spherical holes. However, experimental results of Na-MMT
powder showed that the FFV was only 1.2% [17]. There is some un-
certainty in these results, since the interlayer distances in this clay
are very small: (i) Ps formation may have been hindered by the neg-
ative charges on the clay particle surfaces; and (ii) the penetration of
the positrons may be limited by the presence of the cations in the
galleries. In organomodified clays the FFV is expected to be much
greater.
In the case of spherical nanoparticles, the weak interaction can
lead to the formation of filler aggregates enclosing interstitial vol-
ume inaccessible to polymer segments. These enclosed small voids
can be in the order of 10 Å [11] and create extra free volume.
Chain segmental motion immobilisation: the randomly distributed
filler particles may restrict the chain segmental motion and reduce
the chain mobility and the amount of free volume. Such effects are
common to both PN and PCN. Due to their high surface area, nano-
clays are expected to be more effective in immobilisation because
they act as bridges over polymer chains, limiting their capability to
change their conformation.
Insufficient chain packing: the filler particles may confine the poly-
mer chains and disrupt their packing. This effect can enhance the
free volume due to this inefficient packing or due to the increase
of the distances between polymer segments.
Changes of the free volume hole size distribution: the particles may
alter the f.v. hole sizes distribution in the system [6 •].
All these effects on the polymer free volume coexist in any nano-
composite system. Which of them become dominant depends pri-
marily on the degree of interaction between the two components,
the volume fraction of the filler and the geometrical features of the
particles.
6. Free volume measurements in polymer nanocomposites
6.1. Spherical nano-particles
The changes in free volume that nanoparticles induce in the ma-
trix polymer have been verified experimentally in several reports.
However, because of the complexity of the studied systems, some un-
certainties remain on what actually is measured.
The incorporation of particles has small effect on the polymer free
volume. In poly(dimethyl siloxane)/polysilicate nanocomposites, 30%
nanoparticles content reduces the hole radius from 4.35 Å to 4.31 Å
[18], while the intensity I3 remains almost constant. Crosslinking has
no effect on the f.v. properties. All this casts some doubt on the suitabil-
ity of PALS to reveal differences in dynamic free volume, which is in
principle affected by temperature and chain mobility. The o-Ps may
“see” these sites as “occupied” if the frequency of the molecular motion
through the sites is faster than the annihilation rate.
The longer lifetime, τ4, that has been observed in some polymers,
is ascribed to large, possibly interconnected, free volume elements. In
poly(1-trimethylsilyl-1-propyne) the average radius of these cavities
was about 5.5 Å and increased to 11.5 Å with the addition of fumed
silica (FS) nanoparticles [11]. The enhancement of the longer o-Ps life-
time (τ4) was a consequence of o-Ps annihilating in the interstitial
cavities of the FS agglomerates. If the interstitial voids in the FS pow-
der are filled with polymer chains, then τ4 is not present.
In general, the o-Ps intensity, I3, decreases with particles volume
fraction due to the interfacial layers between the two components.
In HDPE/CaCO3 nanocomposites τ3 remains practically constant with
particles content, but I3 decreases [19 •]. The number of interfacial
layers between CaCO3 and HDPE increases with adding more CaCO3
and the probability for Ps formation decreases, due to the increasing
fraction of positrons annihilating directly in the interfacial layers:
The reduction in o-Ps intensity is accompanied by the enhancement
in I2. Under these circumstances the interpretation of I3 as being pro-
portional to the number of f.v. holes may be inaccurate.
Highly electronegative atoms or radicals on the particle surfaces
may also induce a reduction in o-Ps intensity, which does not always
reflect changes in the free volume. Decrease in I3 was observed in PP/
CB and LDPE/CB nanocomposites [20]. This reduction was ascribed to
 G. Choudalakis, A.D. Gotsis / Current Opinion in Colloid & Interface Science 17 (2012) 132–140
the existence on the particle surfaces of radicals associated with the
aromatic structures of the carbon sheets and with functional groups
containing oxygen. A significant reduction on both lifetime and inten-
sity of o-Ps was observed in PMMA/TiO2 [15], presumably caused by
the − OH groups on the surface of the titania particles. An in-
crease of τ3 has been reported for polystyrene/SiO2 [21] and in
poly-(ethylene-alt-propylene)/SiO2 [22 •] nanocomposites. In all
these cases the o-Ps intensity remains unchanged or decreases slightly.
It seems that silica particles cause an enhancement on free volume in
any matrix they are dispersed. This is due either to chain packing dis-
ruption or to the interstitial voids in the silica agglomerates.
6.2. Clay nano-particles
The influence of the clay on the polymer free volume is different
from that of the isometric nanoparticles. Most experimental results
show a reduction of the amount of the f.v. as a function of clay load-
ing. This reduction concerns mainly the o-Ps intensity, I3, and not
the lifetime, τ3, which remains mostly unaffected: While the mean
hole size remains constant, the number of the f.v. holes seems to de-
crease. This is a clear evidence that the clays inhibit the polymer chain
motions resulting in a reduction of the dynamic f.v. relative to the
neat polymer. However, if the interaction between the two phases
is very weak, then the contribution of interfacial layers on f.v. can
overwhelm the immobilisation effect resulting in an increased free
volume.
In elastomers, such as styrene-butadiene rubber/MMT [17], SBR/
Rectorite [23] etc., a decrease in FFV is observed as a function of clay
loading, mainly due to the o-Ps intensity reduction. The hole radius
in these elastomers is about 3 Å and does not vary with the addition
of filler. The reduction of the FFV is stronger (27%) in nanocomposites
of SBR with 40 phr rectorite, while the other materials present a slight
reduction of the order of 5 − 6% for filler contents of 2.5–10 wt.%.
Similar trends for I3 have been observed for rectorite in epoxy [24]
and ethylene-co-(vinyl acetate) [25] nanocomposites. The o-Ps life-
time is unaffected or slightly reduced, due to the increasing fraction
of positrons that annihilate in the interfacial layers between silicate
platelets and polymer matrix. This reduction is accompanied by a
small increase in I2. In EVA/rectorite systems the 1% reduction in
FFV is due to the chain immobilisation effect. Adding MMT to an
epoxy/polyurethane (EP/PU) blend has no significant effect on FFV
for clay loadings in the area 0–5 wt.% [26]; for laponite particles
added to an acrylic resin [6 •]; and in HDPE/PA6 blend/clay nanocom-
posites [27].
These results indicate that the incorporation of clay in the polymer
matrix does not alter the free volume properties significantly, espe-
cially at relatively low clay volume fractions. Higher clay loadings
have been found to have a stronger effect on f.v. The exfoliation of
the particles and the successful formation of the nanocomposite be-
come particularly difficult in these cases. In polyamide-6/MMT an en-
hancement in o-Ps lifetime has been reported [28], especially when
Fig. 5. Schematic representation of the diffusion process of a gas molecule through a polymer/clay nanocomposite.
137
the clay loading exceeds 19 wt.%. At these loadings the samples are
rather microcomposites with large particle aggregates. Since the
clay on its own shows an o-Ps lifetime in the order of 2.17 ns and
the neat polyamide shows a lifetime of 1.67 ns, it is expected that ad-
dition of clay above 10 wt.% will lead to an increase in τ3, simply by
applying the rule of mixtures.
There are some reports that clays may enhance significantly the
polymer free volume also for clay loadings lower than 10%. In Ref.
[29] the mean volume of the holes increased by 26% when 5 wt.% of
MMT was added in PVA-PES hydrogels. This behaviour is attributed
to the f.v. in the interfacial layers between the organic and inorganic
components. As in the case of epoxy/rectorite, a reduction in I3 was
accompanied by an enhancement in I2. Both the o-Ps lifetime and
the intensity increased in cyanate/ bentonite systems [30]. The en-
hancement of f.v. was about 4.5% for the sample containing 2.5 wt.%
of particles. This increasing trend is ascribed to the insufficient
chain packing that the particles cause in the polymer.
Summarising, the free volume in the polymer matrix remains
mostly unaffected by the addition of clay at low volume fractions or
even decreases slightly. This small reduction is seen mainly as a de-
crease of the number of the f.v. holes, while their average size remains
the same. Exceptions to this involve systems with large interfacial re-
gions, where extra f.v. may develop. Further, at high clay volume frac-
tions exfoliation/deagglomeration is incomplete. The free volume,
then, is enhanced by the presence of interstitial empty space inside
the aggregates, which is inaccessible to the polymer.
7. The permeability of polymer nanocomposites
In polymer nanocomposite systems there is an additional feature
that strongly affects the gas diffusion process. The essentially imper-
meable nanoparticles act as obstacles to the gas molecules, forcing
them to follow longer and complicated routes to diffuse through the
material. This extension/complication of the path is called tortuosity
and is schematically represented in Fig. 5. The tortuosity factor is de-
fined as τ ¼ ‘′=‘, where ‘′ is the diffusion path length when particles
are present and ‘ is the length in the neat polymer. The effect of tor-
tuosity on the diffusion process depends on the geometrical features
of the dispersed nanoparticles, their arrangement in space, their ori-
entation compared to the diffusion direction and their volume frac-
tion [31]. It is obvious that nanoclays will be more efficient in
reducing diffusion, since their surface area is much greater than the
spherical nanoparticles. In the case of nanocomposites, therefore,
Eq. (3.6) should be modified to account for the tortuosity effect.
A′
D¼ expð−B=FFV Þ: ð7:1Þ
τ
The diffusion coefficient in a polymer nanocomposite depends
both on the fractional free volume and the tortuosity factor. These
two parameters can be either in accordance or in competition. The
138 G. Choudalakis, A.D. Gotsis / Current Opinion in Colloid & Interface Science 17 (2012) 132–140
final increase or decrease of the diffusion coefficient upon adding
nanoparticles depends on which of the two becomes dominant.
The curve of lnD vs. 1/FFV of some PVA-PA6/MMT nanocomposites
[32] is linear in the range of 1/FFV b 15. That is, relations of the type of
Eq. (7.1) are valid in this case, only when the fractional free volume is
higher than 6.7%. For smaller FFV the curve is not linear, showing a
saturation with decreasing FFV. Similar results were obtained in
SBR/rectorite nanocomposites, where Eq. (7.1) holds mainly for
FFV > 3.33% [23]. This suggests that when the FFV is relatively high,
it overcomes the influence of the tortuosity and the diffusion coeffi-
cient is determined by the free volume variations.
The incorporation of nanoparticles in a polymeric matrix can
cause redistribution of the free volume hole sizes. Even when the
total FFV is identical in two polymers, differences in their hole size
distribution may be adequate to induce different diffusivities, both
in glassy and rubbery polymers. This has been measured in acrylic
resin/laponite nanocomposites, where the particles do not influence
the FFV but, nevertheless, the diffusion coefficient increases as a func-
tion of laponite loading [6 •]. This enhancement is ascribed to changes
in the hole size distribution.
The permeability coefficient is the product of the solubility and
diffusion coefficients. For polymer nanocomposites:
A′ S
P¼ expð−B=FFV Þ: ð7:2Þ
τ
Therefore, P depends on the diffusion coefficient, the solubility
and the tortuosity. As described previously, the first two depend on
the free volume in the system. However, the permeability coefficient
seems to be less sensitive to the FFV than each of these two coeffi-
cients is. I.e., the tortuosity of the system seems to play a more impor-
tant role in determining the value of P in nanocomposites.
Assuming an ideal periodical arrangement of nano-layers, the tor-
tuosity factor is:
αφw
τ ¼1þ ; ð7:3Þ
2 higher in 1,2-polybutadiene/TiO2 nanocomposites than in the unfilled
where α, = L/W (Fig. 5). For example, the reduction of the permeabil-
ity due to the tortuosity, that 3 vol.% spherical nanoparticles induce, is
in the order of 15%. For the same vol. fraction of clay nanoparticles
and for α = 100 the permeability decreases by 60%, all other parame-
ters remaining the same.
In polymer clay nanocomposites, where the tortuosity dominates
the fractional free volume variations, the percentage reduction on
gas permeability is 20% – 90%. In polyimide/MMT nanocomposites a
reduction of 83% in oxygen permeabilityis observed for only 1 wt.%
MMT [33 •], while for polyamide 6/MMT a reduction in the order of
77% was reported [34] for 2 vol.% of MMT. A 76% reduction of the He
permeability was measured for 8 wt.% MMT in polyurethane/MMT
nanocomposites [35 •], while the presence of 5 wt.% MMT in polyacry-
late decreased the permeability by up to 85% [36]. The addition of
4 wt.% of MMT in Nylon-6 reduced the oxygen permeability by 50%
[37] and in Epoxy/MMT the reduction is of the order of 86% for
7 wt.% MMT [38]. An appreciable reduction in gas permeability com-
pared to the unfilled polymer film was reported in PP/MMT nano-
composites [39], where both the diffusivity and the solubility were
reduced by the presence of the filler.
There are also cases, where the permeability is affected apprecia-
bly by free volume changes. The nitrogen permeability of SBR con-
taining 40 phr rectorite decreased by 69% compared to the pure
rubber [23]. This reduction, however, cannot be attributed solely to
the tortuosity, since the fractional f.v. of the nanocomposite was
also reduced by 27%. The two factors that affect f.v. and permeation
properties may be competitive and the gas permeability can in
some cases show a small decrease or remain constant. Such a case
was reported in PE/HDPE-g-MA20/MMT nanocomposite [40], where
the diffusion coefficient was reduced by 44% when 18.42 wt.% of
MMT was added, while the solubility increased by 36% and the per-
meability decreased by 24%. The poor compatibility between the
polymer and the inorganic particles creates interfacial regions,
which enhance the f.v. This was the case in polyurethane/vermiculite
nanocomposites [41 •], where the permeability of CO2 and N2, was re-
duced only by 20 and 30%, although the high aspect ratio particles
were expected to be more effective barriers.
The overall transport properties of the nanocomposites are affected
by the presence of interfacial regions. In their multi-scale hierarchical
numerical model, Manke et al. [42•] defined a polymer-nanoparticle in-
teraction strength parameter to quantify this effect. For high values of
this parameter, they showed that the polymer density next to the inter-
face increases, indicating that the enhancement of the barrier properties
are due also to the decrease of the free volume there. For large gas mol-
ecules, however, Pryamitsyn et al. [43•] showed that the polymer-
particle interactions are less important than the tortuosity effect. The
mass transport properties, therefore, can be determined by the segmen-
tal dynamics of the matrix.
While the total amount of free volume affects the permeability of
a material as described above, the distribution of the f.v. hole sizes
may also influence the permeation process. This distribution can
strongly affect the diffusion process because the very small holes,
where the gas molecules cannot enter, do not contribute significant-
ly to the diffusion. The solubility, which is connected to the overall
f.v., remains unaffected by the hole size distribution. An example is
the PMMA/MMT system [44]: While the solubility of CO2 in the
PMMA/MMT is approximately the same as that for pure PMMA, the
diffusion coefficient is much greater. This has been attributed to
changes in the f.v. size distribution towards a relatively greater num-
ber of larger holes.
In polymer nanocomposites filled with spherical nanoparticles the
gas permeability has an opposite trend compared to PCN. This type of
fillers often enhances the free volume and is expected to reduce the
barrier properties of the matrix. The gas permeability is 3–4 times
polymer. This is due mainly to the increase in gas solubility. Diffusion
initially decreases with increasing particle loading, due to the in-
crease in tortuosity, and then it increases at higher loadings, as the
space in the interstitial cavities in the agglomerates becomes more
prominent [45].
These interstitial cavities seem to have a significant positive effect
on the permeability of gases in the nanocomposites. For more than
10 wt.% silica nanoparticles in PTMSP, this effect dominates over the
negative effect of the reduced diffusivity presented by Eq. (7.3) with
α = 1. For lower filler content, the higher tortuosity seems to be the
dominant factor. The CO2 permeability in PTMSP containing 40 vol.%
MgO particles is 4.5 times higher than that of the unfilled polymer.
Up to 75% of this increase is due to the increase in the CO2 diffusion
coefficient [46]. In Poly(4-methyl-2-pentene)/FS the increased per-
meability is ascribed entirely to the diffusion coefficient, since
fumed silica has little impact on gas solubility [47]. This is evidence
for the effect of the free volume hole size distribution on gas perme-
ability. Broader distribution results in higher gas diffusion coefficient.
Enhanced gas permeability is found in PMP/FS [48] and in ethylene
vinyl acetate/silica [49] nanocomposites a.o.
A reduction in gas permeability due to the incorporation of spher-
ical nanoparticles has also been reported. In PVC/SiO2 nanocompo-
sites the oxygen and water permeability rates decreased with 3 wt.%
of filler [50]. As it was mentioned above, for low particle loading the
tortuosity becomes the dominant factor that determines the perme-
ation properties, even though this effect is much weaker for nano-
spheres than for clay nanolayers. Similar results are observed in
polyurethane/silica systems, where CO2 permeability is reduced by
about 35% for samples containing 12 vol.% nanoparticles [51].
 G. Choudalakis, A.D. Gotsis / Current Opinion in Colloid & Interface Science 17 (2012) 132–140
Summarising the effect of nanoparticles on the mass transport
properties of the nanocomposites, one could state that when spheri-
cal nanoplatelets are added to a polymer, then the gas permeability
will most likely increase, since the tortuosity that the spheres impose
is small, while the dominant factor is the increase of the free volume
at the interfaces and the interstitial regions within the agglomerates.
On the other hand, when exfoliated nano-clay is added to the poly-
mer, then the permeability is expected to be reduced significantly.
The resulting tortuosity of the diffusion path is the major cause for
this enhancement. The free volume may increase somewhat, due to
the presence of the expansive interfacial regions, but this is a minor
effect and, in most cases, cannot overrule that of the tortuosity. Of
course, this is the case when the clay particles are well exfoliated
and properly oriented. At lower degrees of exfoliation the situation
can point towards that of the isometric nano-fillers.
8. Conclusions
The free volume characteristics of polymers are altered signifi-
cantly by the incorporation of nanoparticles. The changes have been
verified experimentally by Positron Annihilation Lifetime Spectrosco-
py for various systems and types of fillers. The creation of interfacial
layers between the organic and inorganic components, the immobili-
sation of chain segments, the formation of interstitial cavities, and the
changes of hole size distribution are the main effects nanoparticles
have on the polymer, that can alter the free volume properties of
the material.
The transport properties of polymer nanocomposites are deter-
mined by the free volume changes and the tortuosity of the diffusion
path that the presence of the particles causes. These factors can follow
the same trend or be competitive. The final effect on the transport co-
efficients depends on which of them dominates. In polymer clay
nanocomposites the tortuosity effect overcomes the free volume var-
iations and finally determines the transport coefficients. In the case of
spherical nanoparticles the free volume properties dominate over the
tortuosity, controlling, thus, the gas permeation rates.
An important role for the determination of the transport properties
of polymer nanocomposites seems to be played also by the free volume
hole size distribution. The total amount of the free volume itself does
not provide enough information about the free volume network
which determines the diffusion process. The higher the relative number
of smaller holes, the lower the mobility of gas molecules. Therefore, the
changes in the free volume hole size distribution can strongly influence
the diffusion coefficient, even if the total free volume of the system re-
mains unaffected. They may control, thus, the permeability of the sys-
tem, since they have little effect on the solubility.
The search of high/low gas permeability nanocomposite materials
requires more than information on the chemical compatibility between
the components. While the later may be adequate for the improvement
of their mechanical performance, their transport properties are more
complicated and require the study of the physicochemical conse-
quences that the nanoparticles cause on the polymer matrix, where
they are dispersed.
• [28] Winberg P, Eldrup M, Pedersen N, Van Es M, Maurer F. Free volume sizes in inter-
References
[1] Robeson L, Paul D. Polymer nanotechnology: nanocomposites. Polymer 2008;49:
3187–204.
[2] Thran A, Faupel G. Correlation between fractional free volume and diffusivity of
gas molecules in glassy polymers. J Polym Sci, Part B: Polym Phys 1999;37:
3344–58.
[3]
•
Arnold J. A free-volume hole-filling model for the solubility of liquid molecules in
glassy polymers 1: Model derivation. Eur Polym J 2010;46:1131–40.The model is
based on the principal of successive filling of pre-existing holes in a polymer using
a Gaussian distribution of hole sizes. The contributing terms include elastic expan-
sion of small holes and bonding between liquid and polymer in larger holes.
• •
of special interest
139
[4] Xie L, Gidley D, Hristov H, Yee A. Evolution of nanometer voids in polycarbonate
under mechanical stress and thermal expansion using spectroscopy. J Polym Sci,
Part B: Polym Phys 1995;33:77–84.
[5] Garcia A, Iriarte M, Uriarte C, Etxeberria A. Study of the relationship between transport
properties and free volume based in polyamide blends. J Membr Sci 2006;284:173–9.
[6] Choudalakis G, Gotsis A, Schut H, Picken S. The free volume in acrylic
•
resin/laponite nanocomposite coatings. Eur Polym J 2011;47:264–72.PALS mea-
surements of free volume in polymer/clay nanocomposties and analysis of the influ-
ence of its hole size distribution on gas transport.
[7] Thornton A, Nairn K, Hill A, Hill J. New relation between diffusion and free
•
volume: predicting gas diffusion. J Membr Sci 2009;338:29–37.An empirically re-
lation between gas diffusion and free volume derived using a large database of
105 polymers with additional high free volume polymers.
[8] Jean Y, Yuan J, Liu J, Deng Q, Yang H. Correlations between gas permeation and
free-volume hole properties probed by positron annihilation spectroscopy. J
Polym Sci, Part B: Polym Phys 1995;33:2365–71.
[9] Wang Z, Wang B, Yang Y, Hu C. Correlations between gas permeation and free-
volume hole properties of pu membranes. Eur Polym J 2003;39:2345–9.
[10] Consolati G, Genco I, Pegoraro M, Zanderighi L. Positron annihilation lifetime in
poly[1-(trimethyl-silyl)propine]: free volume determination and time depen-
dence of permeability. J Polym Sci, Part B: Polym Phys 1996;34:357–67.
[11] Winberg P, DeSitter K, Dotremont C, Mullens S, Vankelecom I, Maurer F. Free vol-
ume and interstitial mesopores in silica filled poly(1-trimethylsilyl-1-propyne)
nanocomposites. Macromolecules 2005;38:3776–82.
[12] Kobayashi Y, Zheng W, Meyer E, McGervey J, Jamieson A, Simha R. Free volume
and physical aging of poly(vinyl acetate) studied by positron annihilation. Macro-
molecules 1989;22:2302–6.
[13] Wang Y, Nakanishi H, Jean Y. Positron annihilation in amine-cured epoxy polymers-
pressure dependence. J Polym Sci, Part B: Polym Phys 1990;28:1431–41.
[14] Ito K, Nakahishi H, Ujihira Y. Extension of the equation for the annihilation life-
time of ortho-positronium at a cavity larger than 1 nm in radius. J Phys Chem B
1999;103:4555–8.
[15]
•
Zhang J, Yang M, Maurer F. Effect of TiO2 formation on the free volume properties
of electrospun PMMA nanohybrids. Macromolecules 2011;44:5711–21.The influ-
ence of the hydroxyl groups present in titania due to the strong inhibition effect
that they cause on the o-Ps formation.
[16] Winberg P, Eldrup M, Maurer F. Nanoscopic properties of silica filled polydi-
methylsiloxane by means of positron annihilation lifetime spectroscopy. Polymer
2004;45:8253–64.
[17] Wang Y, Wu Y, Zhang H, Wang B, Wang Z. Free volume of montmorillonite/styrene-
butadiene rubber nanocomposites estimated by positron annihilation lifetime spec-
troscopy. Macromol Rapid Commun 2004;25:1973–8.
[18] Dreiss C, Cosgrove T, Benton N, Kilburn D, Alam M, Schmidt R, et al. Effect of cross-
linking on the mobility of PDMS filled with polysilicate nanoparticles: positron
lifetime, rheology and NMR relaxation studies. Polymer 2007;48:4419–28.
[19] Zhang M, Fang P, Zhang S, Wang B, Wang S. Study of structural characteristics of
HDPE/CaCO3 nanocomosites by positrons. Radiat Phys Chem 2003;68:565–7.
[20] Debowska M, Girulska JR, Jezierski A, Pasternak A, Pozniak R. Positron annihila-
tion in carbon black-polymer composites. Radiat Phys Chem 2000;58:575–9.
[21] Chen H, Jean Y, Lee L, Yang J, Huang J. Positron annihilation study in inorganic-
polymer nano-composites. Phys Status Solidi C 2009;6:2397–400.
[22] Harms S, Ratzke K, Faupel F, Schneider G, Willner L, Richter D. Free volume of in-
•
terphases in model nanocomposites studied by positron annihilation lifetime
spectroscopy. Macromolecules 2010;43:10505–11.The enhancement of the o-Ps
lifetime as function of filler loading is ascribed to out-diffusion of the positrons
from the nanoparticle surface. While the findings are interesting this interpreta-
tion is under debate.
[23] Wang Z, Wang B, Qi N, Zhang H, Zhang L. Influence of fillers on free volume and
gas barrier properties in styrene-butadiene rubber studied by positrons. Polymer
2005;46:719–24.
[24] Wang S, Liu L, Fang P, Chen Z, Wang H, Zhang S. Microstructure of polymer-clay
nanocomposites studied by positrons. Radiat Phys Chem 2007;76:106–11.
[25] Fang P, Chen Z, Zhang S, Wang S, Wang L, Feng J. Microstructure and thermal
properties of ethylene-(vinyl acetate) copolymer/rectorite nanocomposites.
Polym Int 2006;55:312–8.
[26] Jia Q, Zheng M, Zhu Y, Li J, Xu C. Effects of organophilic montmorillonite on hydrogen
bonding, free volume and glass transition temperature of epoxy resin/polyurethane
interpenetrating polymer networks. Eur Polym J 2007;43:35–42.
[27] Fang Z, Xu Y, Tong L. Free-volume hole properties of two kinds thermoplastic
nanocomposites based on polymer blends probed by positron annihilation life-
time spectroscopy. J Appl Polym Sci 2006;102:2463–9.
calated polyamide 6/clay nanocomposites. Polymer 2005;46:8239–49.
[29] Paranchos C, Soares B, Machado J, Windmoller D, Pessan L. Microstructure and
free volume evaluation of poly(vinyl alcohol) nanocomposite hydrogels. Eur
Polym J 2007;43:4882–90.
[30] Yu D, Wang B, Feng Y, Fang Z. Investigation of free volume, interfacial, and toughening
behavior for cyanate ester/bentonite nanocomposites by positron annihilation. J Appl
Polym Sci 2006;102:1509–15.
[31] Choudalakis G, Gotsis A. Permeability of polymer/clay nanocomposites: a review.
Eur Polym J 2009;45:967–84.
[32] Cui L, Yeh JT, Wang K, Tsai FC, Fu Q. Relation of free volume and barrier properties
in the miscible blends of poly(vinyl alcohol) and nylon 6-clay nanocomposites
film. J Membr Sci 2009;327:226–33.
[33] Min U, Kim J, Chang J. Transparent polyimide nanocomposite films: thermo-
optical properties, morphology, and gas permeability. Polym Eng Sci 2011;51:
140 G. Choudalakis, A.D. Gotsis / Current Opinion in Colloid & Interface Science 17 (2012) 132–140
2143–150.A remarkable reduction of the oxygen permeability by addition of only
1wt % of nanoparticles.
[34] Fasihi M, Abolghasemi M. Oxygen barrier and mechanical properties of
masterbatch-based PA-6/nanoclay composite films. J Appl Polym Sci in press;
doi:10.1002/app.35647.
[35]
•
Maji P, Das N, Bhowmick A. Preparation and properties of polyurethane nanocom-
posites of novel architecture as advanced barrier materials. Polymer 2010;51:
1100–10.The effect of crosslinking density on the helium barrier properties of
polyurethane has been demonstrated.
[36] Herrera-Alonso J, Sedlakova Z, Marand E. Gas transport properties of polyacrylate/clay
nanocomposites prepared via emulsion polymerization. J Membr Sci 2010;363:48–56.
[37] Sadeghi F, Fereydoon M, Ajji A. Rheological, mechanical and barrier properties of
multilayer nylon/clay nanocomposite film. Adv Polym Technol in press; doi:10.
1002/adv.20270.
[38] Dai C, Li P, Yeh J. Comparative studies for the effect of intercalating agent on the
physical properties of epoxy resin-clay based nanocomposite materials. Eur
Polym J 2008;44:2439–47.
[39] Villaluenga J, Khayet M, Lopez-Manchado M, Valentin J, Seoane B, Mengual J. Gas
transport properties of polypropylene/clay composite membrane. Eur Polym J
2007;43:1132–43.
[40] Jacquelot E, Espuche E, Gerard J, Duchet J, Mazabraud P. Morphology and gas bar-
rier properties of polyethylene-based nanocomposites. J Polym Sci, Part B: Polym
Phys 2006;44:431–40.
[41]
•
Qian Y, Lindsay C, Macosko C, Stein A. Synthesis and properties of vermiculite-
reinforced polyurethane nanocomposites. Appl Mater Interfaces 2011;3:
3709–17.While the tensile modulus and tensile strength showed a remarkable
enhancement, the improvement of the gas barrier properties was weaker, due
to the free volume in the interfacial regions.
[42]
•
Xiao J, Huang Y, Manke C. Computational design of polymer nanocomposite coatings:
a multiscale hierarchical approach for barrier property prediction. Ind Eng Chem Res
2010;49:7718–27.A numerical multiscale modeling and simulation method is devel-
oped in order to predict the relative permeability in a polymer nanocomposite. The
model takes the interfacial layer effect into account by incorporating an interaction
strength parameter.
[43]
•
Pryamitsyn V, Hanson B, Ganesan V. Coarse-grained simulations of penetrant trans-
port in polymer nanocomposites. Macromolecules 2011;44:9839–51.A simulation
study that examines the influence of the interfacial layers and the polymer matrix
dynamics on the transport properties of polymer nanocomposites.
[44] Manninen A, Naguib H, Nawaby A, Day M. CO2 sorption and diffusion in poly-
methyl methacrylate-clay nanocomposites. Polym Eng Sci 2005;45:904–14.
[45] Matteucci S, Kusuma V, Swinnea S, Freeman B. Gas permeability, solubility and
diffusivity in 1,2-polybutadiene containing brookite nanoparticles. Polymer
2008;49:757–73.
[46] Matteucci S, Kusuma V, Swinnea S, Freeman B. Gas transport properties of MgO
filled poly(1-trimethylsilyl-1-propyne) nanocomposites. Polymer 2008;49:
1659–75.
[47] Merkel T, Freeman B, Spontak R, He Z, Pinnau I, Meakin P, et al. Sorption, trans-
port, and structural evidence for enhanced free volume in poly(4-methyl-2-
pentyne)/fumed silica nanocomposite membranes. Chem Mater 2003;15:
109–23.
[48] Shao L, Samseth J, Hagg M. Crosslinking and stabilization of nanoparticle filled
pmp nanocomposite membranes for gas separation. J Membr Sci 2009;326:
285–92.
[49] Sadeghi M, Khanbabaei G, Dehaghani A, Sadeghi M, Aravand M, Akbarzade M,
et al. Gas permeation properties of ethylene vinyl acetate-silica nanocomposite
membranes. J Membr Sci 2008;322:423–8.
[50] Zhu A, Cai A, Zhang J, Jia H, Wang J. PMMA-grafted-silica/PVC nanocomposites:
mechanical performance and barrier properties. J Appl Polym Sci 2008;108:
2189–96.
[51] Sadeghi M, Semsarzadeh M, Barikani M, Chenar M. Gas separation properties of
polyether-based polyurethane-silica nanocomposite membrane. J Membr Sci
2011;376:188–95.