European Polymer Journal 45 (2009) 967984

Contents lists available at ScienceDirect

European Polymer Journal

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

Macromolecular Nanotechnology Review

Permeability of polymer/clay nanocomposites: A review

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 is a review of the existing studies on the permeability of gas molecules in nanocom-
Received 26 June 2008 posite materials that consist of inorganic platelet-shaped llers in polymeric matrices. We

MACROMOLECULAR NANOTECHNOLOGY
Received in revised form 12 January 2009 describe the dominant mechanisms for the transport of small molecules in polymers and
Accepted 14 January 2009
polymer nanocomposites, as well as the procedures for the measurement of the permeabil-
Available online 30 January 2009
ity and the diffusivity. The emphasis is given on the various models that have been pro-
posed for the prediction of permeability in polymerclay nanocomposites. The inuence
of the characteristics of the inorganic particles on the barrier properties of the composite
Keywords:
Gas permeability
membrane is discussed and tested using the model and the available experimental data.
Barrier properties Some aspects on the methods of improving the barrier properties of the nanocomposite
Permeation models are examined and a few applications of these materials as gas barriers are presented.
Nano-clay 2009 Elsevier Ltd. All rights reserved.
Nanocomposite

Contents

1. Introduction: Nanocomposites with phyllosilicate reinforcement and polymeric matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 968

2. The measurement of the permeability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 969
2.1. Direct measurement in a cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 969
2.2. Sorption/desorption experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 971
3. Mass transport in polymers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 971
3.1. Free volume and diffusion in amorphous polymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 971
3.2. Semicrystalline polymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 972
4. The permeability of nanocomposites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 972
4.1. Regular arrangement of parallel nanoplatelets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 973
4.2. Random spatial positioning of parallel plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 975
4.3. Arrangements at angles  6 90 to the main diffusion direction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 977
4.4. Influence of the interfacial regions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 977
4.5. Influence of the particle aggregation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 978
4.6. Experimental and numerical validation of the models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 978
5. Using PALS to study the permeability of polymers and nanocomposites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 979
6. Methods to influence the permeability of the nanocomposite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 981
6.1. Electrically modulated permeability using side-chain liquid crystalline polymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 981
6.2. Improving the barrier properties using reactive additives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 982
7. Low permeability nanocomposite coatings and other applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 982
8. Conclusions and recommendations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 983
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 983
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 983

* Corresponding author.
E-mail address: gchoudalakis@isc.tuc.gr (G. Choudalakis).

0014-3057/$ - see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.eurpolymj.2009.01.027
 968 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984
1. Introduction: Nanocomposites with phyllosilicate
reinforcement and polymeric matrix
Nanocomposites materials are two phase systems that
consist of a polymeric matrix and dispersed inorganic par-
ticles of nanometer scale. The most common inorganic
particles belong to the family of 2:1 phyllosilicates. Their
crystal structure (Fig. 1) [1] consists of an aluminium or
magnesium hydroxide octahedral sheet sandwiched be-
tween two silicon oxide tetrahedral sheets. The layer thick-
ness of each platelet is around 1 nm, and their lateral
dimensions may vary from 30 nm to several microns.
The layers are located on top of each other like the
pages of a book. Van der Waals gaps are created between
the layers, called galleries. The isomorphic substitution of
the tetrahedral or octahedral cations, e.g. the substitution
of Al3+ with Mg2+ or Fe2+ with Li1+, generates negative
charges that are counterbalanced by alkali and alkaline
earth cations located inside the galleries. In the case of
tetrahedral substitution, the negative charge is located on
the surface of the silicate layers and, thus, the polymer
MACROMOLECULAR NANOTECHNOLOGY
matrices can interact more readily with tetrahedral than
with octahedral substituted material.
Two particular characteristics of layered silicates play an
important role in the creation of nanocomposites: the rst
is the ability of silicate sheets to disperse into individual
layers, and the second is the possibility to modify their sur-
face chemistry through ion exchange reactions with organ-
ic and inorganic cations. The simple mixing of polymer and
layered silicates does not always result in the generation of
a nanocomposite, as this usually leads to dispersion of
stacked sheets. This failure is due to the weak interactions
between the polymer and the inorganic component. If these
interactions become stronger, then the inorganic phase can
be dispersed in the organic matrix in nanometer scale.
The layered phyllosilicate particulates usually contain
Na+ or K+ ions and are only compatible with hydrophilic
polymers, such as poly(vinyl alcohol). To render the lay-
ered silicates compatible with non-polar polymers one
must convert the hydrophilic silicate surface to organo-
Fig. 1. The crystal structure of the phyllosilicates [1].
philic. This can be achieved by ion exchange reactions with
cationic surfactants including primary, secondary, tertiary,
and quaternary alkylammonium cations or phosphonium
cations which contain various substituents. At least one
of these substituents must be a long carbon chain of 12
carbon atoms or more, in order to make the clay mineral
compatible with the polymer [24].
The alkylammonium cations lower the surface energy
of the inorganic reinforcement, improve the wetting char-
acteristics of the polymer matrix, and result in larger inter-
layer spacing (swelling). Additionally, the alkylammonium
cations can provide functional groups that react with the
polymer matrix or, in some cases, can initiate the polymer-
isation of monomers to form the polymer in-situ [5].
The phyllosilicate particles present a very high aspect
ratio of width/thickness, in the order of 101000. For very
low concentrations of particles, the total interface between
polymer and layered silicates is much greater than that in
conventional composites. Depending on the strength of the
interfacial interaction, four types of morphology are possi-
ble in nanocomposites (Fig. 2) [1]:
aggregated; intercalated; occulated; and exfoliated.
The main goal for the successful development of clay-
based nanocomposites is to achieve complete exfoliation
of the layered silicate in the polymer matrix. The three
most common methods that are used for the preparation
of a polymer-nanocomposite are:
(1) Intercalation in a suitable monomer and subsequent
in situ polymerisation that leads to exfoliation.
(2) Intercalation of polymer from solution and
exfoliation.
(3) Polymer melt intercalation and exfoliation.
Recent reviews of the methods that are used for the
synthesis of polymerclay nanocomposites can be found
in [6,7].
The nanocomposites have become an area of intensive
research activity. The rst reason for this development is
 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984 969

Fig. 2. Schematic representation of three types of polymer/clay nanocomposites [1].

their high mechanical properties. The complete dispersion tion of the gas in the membrane, cx; t, reaches a steady
of clay nanolayers in a polymer optimises the number of distribution, Jt reaches the steady state ux, J1 . In a at
available reinforcing elements that carry an applied load membrane of thickness d, J 1 depends on the (constant)
and deect the evolving cracks. The coupling between gas pressure, P 0 :
the large surface area of the clay and the polymer matrix
P0
facilitates the stress transfer to the reinforcing phase J1 K : 2:2
d
allowing for the improvement of the tensile stress and

toughness [810].
The second major advantage of the nanocomposites is
their enhanced barrier properties. The impermeable clay
layers force a tortuous pathway for a permeant transvers-
ing the nanocomposite. It is reported that gas permeability
through polymer lms can be reduced by 50500 times
even with small loadings of nanoclays. The relevant re-
search on polymerclay nanocomposite concerns mostly
oxygen, carbon dioxide and nitrogen barrier lms, a.o.,
for packaging food and carbonated drinks. Other applica-
tions include gas tanks and coatings.
2. The measurement of the permeability
The permeation of gases through an organic membrane,
such as a polymer lm, is a complex process that consists
of four processes: the sorption of gas molecules on the sur-
face of the membrane; the dissolution of the gas inside the
membrane; the diffusion through it; and, nally, the
desorption of gas from the other surface of the membrane.
The permeability coefcient, K (in mol Pa1 m1 s1), is the
product of the diffusion coefcient, D (in m2 s1), and the
sorption coefcient or solubility, S (in mol m3 Pa1):
K D  S:
The permeability can be measured directly using a per-
meation cell, or indirectly in sorption/desorption
experiments.
MACROMOLECULAR NANOTECHNOLOGY
The constant of proportionality, K, is the permeability
coefcient. Eq. (2.2) is a special case of the fundamental
law of diffusion:
Jr; t Drcr; t; 2:3
When the ux density and the concentration vary with
time, one has to solve also the time dependent diffusion
(Ficks) equation in a semi-innite body:
ocr; t
Dr2 cr; t : 2:4
ot
The 1D solution applicable in the experiments shown in
Fig. 3 is subject to the following conditions [11]:
t 0 The entire cell is initially under vacuum. The gas is
introduced suddenly, at t 0, into the volume V 0
at pressure P 0 ; thus cx; t 0 0 (Fig. 3).
t > 0 The compartment V 0 is kept at constant pressure, P0 .
The solubility of the gas in the membrane is the ratio
of its concentration to the external gas pressure,
when these two phases are in equilibrium. The
x=0 x=d
2:1
P0 P0
P0 ,V 0 P0 ,V
A c0 cd (t)

2.1. Direct measurement in a cell c0 cd (t)

For the direct measurement of the permeability, one
can use an arrangement such as shown in Fig. 3. A mem-
brane separates the cell into two compartments. The gas
is introduced in the compartment V 0 (left hand side in
Fig. 3) under constant pressure P 0 . The ux density Jt
(in mol m2 s1) of the gas in the other compartment is
monitored at constant temperature. The evolution of Jt
reects the non-steady state process. When the concentra-
c(x,t)
x
d
Fig. 3. Experimental cell for the measurement of the permeability. A
membrane, with thickness d and area A, divides the cell in two chambers.
The gas is introduced in volume V 0 under constant pressure, P 0 . The
concentration of the test gas is c0 on the left surface of the membrane and
cd t on the right surface.The concentration of the gas inside the
membrane is cx; t.
 970 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984

surface of the membrane that is exposed to the gas cd (t)

under pressure P0 will absorb or dissolve the gas,
until the concentration of dissolved gas reaches the
saturation value, c0 , given by Henrys equation; thus
cx 0; t c0 S  P 0 .

The boundary value of c at the surface of the membrane

on the side of volume V is considered zero, so
cx d; t 0.
A solution of (2.4) for a at membrane is [12]:
"  # steady state
x 2X 1
1 mpx m2 p2 Dt
cx; t SP0 1  sin exp  :
d p m1 m d d
2

2:5
At x d Eq. (2.3) for 1D gives the ux of the gas that per-
meates the membrane: t0 t
"  #
DSP0 X
1
m2 p2 Dt Fig. 4. Typical diagram of experimental measurements. The time lag is
Jt 1 2 cos mp  exp  2
: 2:6
d m1 d estimated from the extrapolation of the steady state line.

Eq. (2.6) becomes at steady state (t ! 1):

r 1 " #
MACROMOLECULAR NANOTECHNOLOGY

2
P0 DX d
J1 DS  ; 2:7 Jt 2SP 0  exp  2m 12 : 2:11
d pt m1 4Dt
which gives back Eq. (2.1) when compared to Eq. (2.2). For short times Eq. (2.11) can be approximated by
By integrating Eq. (2.6) from time 0 to t, and since ignoring all but the rst term of the sequence. Multiplying
Jt dc=Adt, with A the area of the membrane, one ob- then with t 1=2 and taking the logarithm on both sides one
tains the concentration, cd t (in mol/m3), at the right hand gets:
side of the experimental cell: "  1=2 # 2
" 2 2 2
# 1=2 D d
ADSP 0 d 2d X
1
expm2 p2 Dt=d lnft  Jtg ln 2SP0  : 2:12
cd t t   1m : p 4Dt
Vd 6D p2 D m1 m2
Plotting the left hand side of Eq. (2.12) against the reci-
2:8
procal time we obtain a straight line. The diffusion coef-
2
For long times Eq. (2.8) becomes: cient can be calculated from the slope of this line (d =4D).
ADSP 0 The permeability measurements are often conducted in
cd t t  t0 2:9 a WickeKallenbach device similar to the cell in Fig. 3 [13].
Vd
2 A mixture of gases (e.g. CO2 and N2) is introduced in the
d
with t0 : 2:10 left chamber, while the right chamber is fed with a pure,
6D reference gas (e.g. N2). Both chambers remain at the same
The characteristic time t0 is a measure of the time that pressure. The outlet stream from the right chamber con-
is required to establish a constant ow. If we plot cd vs. t for sists of the reference gas i.e. N2, and the test gas, i.e. CO2,
sufciently long times to reach a linear response, then we that permeates through the membrane. This stream is fed
can determine the time lag, t 0 , as shown in Fig. 4, and the to a mass spectrometer or analyser in order to determine
diffusion coefcient D from Eq. (2.10). The slope of the line the permeate concentration.
gives the product D  S, i.e. the permeability coefcient. Two types of experiments can be established. First, if
This is a simple and direct way for the determination of the time lag is relatively small (i.e. diffusion coefcient in
the coefcients D, S and K. In practise, it is only necessary the order of 106 cm2 =s, or membrane thickness in the or-
to introduce gas in volume V 0 , and record the mass ux der of 100 lm, or both) one can perform the steady state
in volume V as a function of time, until a straight line ap- experiment using Eq. (2.9). Otherwise a transient experi-
pears. Then the straight line is extrapolated and the time ment should be performed using Eq. (2.12). It has been re-
lag is determined. ported [14] that the diffusion coefcient measured at
In spite of the simplicity of the calculations, the above steady state differs from the one that is extracted from
method has practical difculties because t0 is very large transient experiments because of the existence of dead-
in many cases. Indeed, the diffusion coefcient for small end pathways in the membrane. In some cases the perme-
gas molecules in most polymers is about 106109 cm2/ ation measurements are conducted using pressure differ-
s. For polyethylene or polystyrene the diffusion coefcient ence between the two compartments of the cell in order
is about 108 cm2/s and a membrane of these polymers to accelerate the process. The diffusion coefcient calcu-
with 1 mm thickness requires approximately two days to lated from this kind of experiments does not reect the ac-
reach steady state. An alternative solution for Eq. (2.3) tual diffusion process due to the existence of an external
was proposed by Rogers et al. [11]: driving force.
 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984 971

2.2. Sorption/desorption experiments Applying the above methods, the diffusion coefcient
can be determined in isothermal and isobaric sorption/
Sorption/desorption experiments are used to measure desorption experiments. These methods present, however,
both the diffusion coefcient and the solubility and, there- an uncertainty in the time interval over which they are va-
from, the permeability using Eq. (2.1). In the simplest form lid. A more detailed description of methods that are used to
of such an experiment, the membrane remains initially un- extract the diffusion coefcients from gravimetric data is
der vacuum and the gas is introduced and maintained at presented in [16,17].
constant pressure. The gas is dissolved and diffuses into The solubility coefcient is found by measuring the con-
the membrane, and the weight gain is estimated with a centration of the soluble gas as a function of pressure. In gen-
balance. The fractional mass uptake is reported as a func- eral, the total concentration, C, of gas molecules absorbed in
tion of time. Assuming a at membrane of thickness d a macroscopically non-porous glassy polymer consists of the
and uniform concentration of the gas in the membrane gas molecules in the dissolved state or Henrys population,
there are two possible solutions for Eq. (2.4) [15]: C d , and the Langmuir population, C H , which represents the
s" !# adsorption of gas molecules into the micro-voids or free vol-
Mt Dt 1 X1
md
8 p
2 1 m
ierfc ; 2:13 ume holes that exist in the glass [18]:
2
M1 d p m0 4Dt1=2
" # C 0H bP
Mt 8 X 1
1 D2m 12 p2 t C Cd CH : C d kd P; CH ; 2:17
1 2  exp  ; 2:14 1 bP
M1 p m0 2m 12 d
2
with kd the solubility coefcient, P the partial pressure of
where Mt and M 1 are the mass uptakes at times t and 1, the gas, and C 0H and b the hole saturation and hole afnity
and ierfc is the error function integral. constants.
Henrys law, which describes the sorption behaviour of

MACROMOLECULAR NANOTECHNOLOGY
Eq. (2.13) converges rapidly for short times and can be
approximated then by: C d , is similar to the case of swelling in rubbers, where sol-
 1=2 vent molecules dissolve among the polymer chains and
Mt 8 Dt swell the polymer matrix. The Langmuir holes, on the other
1=2 : 2:15
M1 p d
2
hand, are considered to be non-equilibrium defects and
p their sizes and numbers depend on the thermal history.
By plotting Mt =M1 against t =d we obtain a straight
The coefcients kd , b and C 0H can be determined by tting
line until t1=2 , the half life time when M t =M1 1=2
the experimental isotherms with Eq. (2.17) [19].
(Fig. 5). The diffusion coefcient is calculated from the
slope of this line.
Alternatively, Eq. (2.14) converges rapidly for long
3. Mass transport in polymers
times. Keeping, then, only the rst term and taking the log-
arithms on both sides one obtains a straight line in the
3.1. Free volume and diffusion in amorphous polymers
form:
   
Mt 8 p2 Dt Gas permeation in a membrane is a complex process
ln 1  ln  2 : 2:16 and has two main stages: initially the gas molecules are
M1 p 2
d
adsorbed on the surface of the membrane, and then they
By plotting the right hand side against t we can calcu- diffuse through the membrane. During adsorption, the
late D from the slope of this line. gas molecules are positioned in the free volume holes of
the polymer that are created by Brownian motions of the
chains or by thermal perturbations. The diffusion process
occurs by jumps through neighbouring holes. Thus, it de-
pends on the number and the size of these holes (static free
volume) and on the frequency of the jumps (dynamic free
volume). The static free volume is independent of thermal
motions of the macromolecules and is related to the gas
solubility, while the dynamic free volume originates from
the segmental motions of the polymer chains and is related
to the gas diffusivity. Thus, the diffusion coefcient is a ki-
netic factor that reects the mobility of the gas molecules
in the polymeric phase, while the solubility coefcient, S, is
a thermodynamic factor related to the interactions be-
tween the polymer and the gas molecules. The latter is
associated with the specic free volume, v f [20]:
S
vf  v  v0 ; 3:1
qg

Fig. 5. Typical diagram of sorption/desorption experimental measure-

where v is the specic volume, v 0 is the occupied specic
ments. The curve is linear in the initial stage until t 1=2 which is the half volume and qg is the density of sorbed gas. The fractional
time for sorption. free volume is dened as:
 972 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984

vf S 3.2. Semicrystalline polymers

fv : 3:2
v qg v
In semicrystalline polymers, the size and the shape of
The average radius of the holes, R, can be calculated the crystallites, the crystal structure and the degree of
assuming identical spherical holes arranged in a cubic lat- crystallinity have an important inuence on the perme-
tice with constant a: ation process. The basic assumption is that the crystallites
r
3 3f
consist of a phase impermeable for gas molecules and the
v
Ra : 3:3 diffusion occurs only in the amorphous phase.
4p
The adsorption of gases in amorphous polymers is de-
The fractional free volume can be determined experi- scribed by Henrys law:
mentally, e.g. by Positron Annihilation Lifetime Spectros-
copy (PALS) [21]. Some details on the use of this method c SP: 3:10
to study the permeability in polymers and nanocomposites For semicrystalline polymers:
are given in Section 5.
The diffusion coefcient of a gas in a polymer depends S S0 1  /c ; 3:11
on the free volume in the matrix, the molecular diameter where S0 is the solubility coefcient of the amorphous
of the gas, a , and the velocity of the gas molecules, u, [22]: phase, and /c is the volume fraction of the crystalline
  phase. Obviously, for purely crystalline phase, S 0. Eq.
cv 0
D ga u exp  ; 3:4 (3.11) is applicable to most semicrystalline polymers. It is
vf assumed in this case that the presence of crystallites leaves
the amorphous phase unaffected [24].
where g is a geometric factor and c is a factor related to the
The crystallites have two different effects on the diffu-
MACROMOLECULAR NANOTECHNOLOGY

overlap of free volume, i.e. the degree to which more than

sion process. They act as impermeable barriers to the gas
one molecules may access the same free volume. Regroup-
molecules, forcing them to follow longer paths. This is in-
ing the constants Eq. (3.4) can be written in the form:
  cluded in the tortuosity factor, s. Additionally, the presence
B of the crystals can immobilise side-chains in the amor-
D A exp  : 3:5
fv phous phase and, thus, reduce the free volume there,
resulting in higher activation energies for diffusion. This
The temperature dependence of D and S is of the Arrhe-
inuence can be included in an immobilisation factor, b,
nius type:
[25]:
   
Ed DH s D0
D D0 exp  ; S S0 exp  ; 3:6 D : 3:12
kT kT bs

where Ed is the activation energy for diffusion and DHs is In combination with Eq. (3.5) this gives
the heat of sorption. Therefore, for the permeability: A fB
  D e v 3:13
E d DH s bs
K K 0 exp  : 3:7
kT The chain segment immobility factor b is given by:
h  p i2
where Ed DHs is the total activation energy for perme- b exp k d  /L =2 3:14
ation. Since DHs is often negative, the permeation activa-
tion energy is less than the Ed . where k is independent of the penetrant, d is the diameter
p
The free volume theory neglects the diffusion through of the penetrant and /L =2 is approximately equal to the
the pores. When the pore size is between 0.002 and mean unoccupied distance between two chain segments.
0.05 lm, then the main transport mechanism is Knudsen Increasing the crystallinity leads to an increase in the fac-
ux [23]: tor b because the mean unoccupied distance decreases.

pnr2 DK DP 4. The permeability of nanocomposites

JK ; 3:8
RT s
where n is the number of pores, r is their radius, s is the
tortuosity factor, is the membrane thickness and DK is
the Knudsen diffusion coefcient given by:
r
8RT
DK 0:66r ; 3:9
pM
with M the molecular weight of the gas. From Eq. (3.9) it
can been seen that the diffusion coefcient is inversely
proportional to the molecular weight of the gas. Thus,
membranes where Knudsen ux is dominant are used for
gas separation.
The mass transport mechanism of gasses permeating a
nanoplatelet reinforced polymer is similar to that in a
semicrystalline polymer. In most theoretical treatises the
nanocomposite is considered to consist of a permeable
phase (polymer matrix) in which non-permeable nanoplat-
elets are dispersed. There are three main factors that inu-
ence the permeability of a nanocomposite: the volume
fraction of the nanoplatelets; their orientation relative to
the diffusion direction; and their aspect ratio.
It is generally accepted that the transport mechanism
within the polymer matrix follows Ficks law, and that
the matrix maintains the same properties and characteris-
 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984
Fig. 6. Inuence of the degree of delamination on the tortuosity factor and the aspect ratio of nanoplatelets [26]. W is the thickness of the stacks.
tics as the neat polymer. Therefore, a decrease of the solu-
bility is expected in the nanocomposite due to the reduced
polymer matrix volume, as well as a decrease in diffusion
due to a more tortuous path for the diffusing molecules.
The reduction of the diffusion coefcient is higher than
that of the solubility coefcient. Indeed, the volume frac-
tion of nanoplatelets is low and, thus, the reduction of
the matrix volume is small. The major factor, then, is the
tortuosity, which is connected directly to the shape and
the degree of dispersion of the nanoplatelets.
The degree of dispersion of the nanoplatelets is deter-
mined by the degree of delamination of the clay. The fully
delaminated (exfoliated) nanocomposite presents much
higher values for the tortuosity factor and the aspect ratio
in comparison with the partially delaminated (interca-
lated) nanocomposite (Fig. 6) and it is much more effective
to be used in barrier membranes for gasses.
The terms intercalated and exfoliated layered silicate
nanocomposites are often used to describe two extreme
states of silicate layer organisation in the composite mor-
phology. Intercalation implies the insertion of polymer
chains in the galleries of the initial layered tactoids. This
leads to a longitudinal expansion of the galleries. Exfolia-
tion implies complete breakage of the initial layer stacking
order and homogeneous dispersion of the layers in the
polymer matrix.
Several models have been developed in order to de-
scribe the mass transfer within the nanocomposites. Most
models assume that the platelets have a regular and uni-
form shape (rectangular, sanidic or circular) and form a
regular array in space. They are either parallel to each
other or have a distribution of orientations with the aver-
age orientation at an angle to the main direction of diffu-
Fig. 7. Regular arrangement of orthogonally shaped platelets in a parallel array with their main direction perpendicular to the diffusion direction.
973
sion of the gas molecules. Some of these models are
outlined in the following.
4.1. Regular arrangement of parallel nanoplatelets
One of the rst attempts to describe the permeability of
membranes, where a second phase is dispersed in a regular
arrangement, was made by Barrer and Petropoulos [27].
MACROMOLECULAR NANOTECHNOLOGY
These authors calculated the diffusion through a regular
array of parallelepipeds of a second phase dispersed in a
matrix with a different diffusion coefcient. When this ap-
proach is applied in the case of dispersed impermeable
thin plates, the change of the permeability is found to be
proportional to the fractional cross section (slit) that is
available for the diffusant to move forward and depends
on the tortuosity of the path.
A simple permeability model for a regular arrangement
of platelets has been proposed by Nielsen [28] and is pre-
sented in Fig. 7. The nanoparticles are evenly dispersed
and considered to be rectangular platelets with nite
width, L, and thickness, W. Their orientation is perpendic-
ular to the diffusion direction.
The solubility coefcient of this nanocomposite is:
S S0 1  /; 4:1
where S0 is the diffusion coefcient of the neat polymer,
and / is the volume fraction of the nanoplatelets that are
dispersed in the matrix. In this approximation the solubil-
ity does not depend on the morphological features of the
phases.
The platelets act as impermeable barriers to the diffusing
molecules, forcing them to follow longer and more tortuous
 974 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984

paths in order to diffuse through the nanocomposite. The

diffusion coefcient, D, is inuenced by the tortuosity, s:
D0
D ; 4:2
s
where D0 is the diffusion coefcient of the matrix. The fac-
tor s depends on the aspect ratio, the shape and the orien-
tation of the nanoplatelets, and it is dened as
0
s 4:3

where 0 is the distance that a solute must travel to diffuse
through the membrane when nanoplatelets are present
and is the membrane thickness. From Eqs. (2.1), (4.1)
and (4.2):
K composite 1  /
: 4:4
K matrix s
Fig. 8. Predictions of Nielsens model for the relative permeability as a
The prolonged diffusion length 0 is estimated as fol- function of platelets volume fraction for different particle aspect ratios.
lows: Each nanoplatelet contributes to the enhancement
of the diffusion length by L=2 on the average. If hNi is the
mean number of nanoplatelets that a solute encounters da=d sa b / is the volume fraction of the plate-
MACROMOLECULAR NANOTECHNOLOGY

as it diffuses through the membrane, then: lets, d=a a is the particle aspect ratio and s=a r is the
slit aspect ratio. Cussler et al. [30] neglected the second
L
0 hNi : 4:5 term of Eq. (4.8), since it is relatively much smaller, and
2 suggested for narrow slits (r  1) the following simpler
Since hNi /=W, the factor s becomes: expression for the enhancement of the barrier properties
of the membrane:
L
s1  /: 4:6 !1
2W K composite a2 /2
1 : 4:9
Eq. (4.4), then, gives: K matrix 1/
K composite 1/ This model predicts a rapid reduction of the relative
; 4:7
K matrix 1 a2 / permeability at small values of / [30], in contrast with
the model of Nielsen (Eq. (4.7)) [28], which needs either
where a L=W is the aspect ratio of the nanoplatelets. high volume fraction or high aspect ratio for the same
This equation shows that the permeability of the nanocom- reduction. This is probably attributed to the different re-
posite decreases with the increase of / and a. In practise gions over which the above models are applicable.
however, the limit for its validity is / 6 10%, because the Two variations of Eq. (4.8) were given by Falla et al. [31]
particles have a tendency to aggregate, which increases based on Aris [32] and Wakeham and Mason [29]:
with /. From the most common nanoparticles, such as syn-
 !1
thetic mica, montmorillonite, laponite and vermiculite, the K composite a 2 /2 a / 4 a/ pa2 /
1 ln 4:10
last one is believed to have the highest aspect ratio and is K matrix 1  / r p1  / r1  /
the most efcient for the reduction of K composite . The predic-  !1
tions of this model for different values of the aspect ratio, K composite a 2 /2 a / 1/
1 21  /ln 4:11
a, are shown in Fig. 8. K matrix 1/ r 2r/
The diffusion through a multiperforated single laminae
The fourth term in Eq. (4.10) was derived from arguments
was studied by Wakeham and Mason [29]. These authors
presented in [32]. The same term in Eq. (4.11) was derived
found that the resistance to diffusion through such a mem-
based on [29].
brane had a contribution from the need of the diffusant to
enter the constriction into the pore/slit, and a contribution
due to the length of the pore. This approach was extended
by Cussler et al. [30] to multiple layers of parallel platelets
of innite third dimension, separated by slits. The result
was:
2  !1
K composite da d 2b d
1 ln ;
K matrix sa b ba b l 2s
4:8
where, l is the thickness of the membrane, and the other
dimensions are shown in Fig. 9. In this arrangement, Fig. 9. Ribbon arrangement and characteristic parameters.
 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984 975

The second term of equations (4.10) and (4.11), that in- The distance that the solute molecule diffuses through the
volves a2 , reects the contribution of the tortuous path of nanocomposite, becomes in this case:
the diffusant through the plates. The third term is due to
0 hNihdi: 4:16
the resistance to diffusion of the slits and depends on their
aspect ratio, r. The last part represents the constriction which is the same with Eq. (4.5), except that L=2 is replaced
from the wide space between the plates into the narrow by hdi, the average distance the solute must travel to reach
slits and it should depend on the constriction ratio, a=r, the edge of platelet. Using simple statistical considerations
as it is the case in Eq. (4.10). hdi can be estimated considering that the molecule hits the
Hexagonal akes arranged in regular parallel arrays platelet at a random point along its length, and that the
were examined by Moggridge et al. [33]. The same argu- resistance to mass transfer is proportional to the length
ments as for Eq. (4.9) were used: of the path travelled. Thus 0 becomes:
 
!1 1
K composite 2 a2 /2 0 1 a/ : 4:17
1 4:12 3
K matrix 27 1  /
Eq. (4.17) differs from Eq. (4.6) by a factor 2=3, which is
The difference due to the specic platelet-shape is re- due to the randomness of the positions of the
ected by the coefcient (2/27), which reduces the effec- nanoplatelets.
tiveness of the barrier. The area available for diffusion, Af , can be determined
by dividing the volume available for diffusion by the dis-
4.2. Random spatial positioning of parallel plates tance travelled to cross the membrane:
Af V m  V f =0

MACROMOLECULAR NANOTECHNOLOGY
To describe the actual positioning of parallel platelets in ; 4:18
A0 V m =
the form of ribbons Brydges et al. [34] used the stacking
parameter c x=2d (x and 2d are dened in Fig. 9, and where V m is the total volume of the membrane and V f is
c < 1). This parameter expresses the deviation from the the volume of all platelets. The relative permeability then
periodicity, by dening the horizontal offset of each ribbon becomes:
layer with respect to the layer underneath it. The value K composite 1/
c 1=2 represents the case when the ribbons in one layer
2 : 4:19
K matrix 1 a3 /
are centered over the gap in the layer below. Obviously,
c 1=2 gives the lowest permeability, while for c 0 the The case where the nanoplatelets are circular disks
permeability is maximum. Brydges et al. [34] calculated with radius R, thickness W and aspect ratio a R=W
the different mass uxes around the platelets. In the cases (Fig. 10) was examined by Fredrickson and Bicerano
where the aspect ratio is very large a 2d=a > 100 and [36] using multiple scattering theory. When, the disks
2s a this model gives: were spaced by an average distance exceeding R (dilute
!1 regime), the following expression for the diffusion coef-
K composite a2 /2 cient was derived:
1 c1  c 4:13
K matrix 1/ Dcomposite 1
4:20
This expression has a form analogous to the equation of Dmatrix 1 ja/
Cussler (Eq. (4.9)), except for the factor c1  c. where j p= ln a. This case corresponds to low values of
Cussler et al. [30] also examined the case of slits formed the volume fraction and the aspect ratio of the nanoplat-
between randomly positioned parallel platelets of innite elets (a/  1), and presents a form similar to Nielsens
third dimension and suggested the following expression equation (Eq. (4.7)). For low values of / but high values
for the permeability: of a (a/
1), the disks are overlapping (semi-dilute re-
!1 gime) and the relation becomes:
K composite l0 a2 /2
1 ; 4:14 Dcomposite 1
K matrix 1/ ; 4:21
Dmatrix 1 la2 /2
where l0 is a combined geometrical factor, characterising 2
also the randomness of the porous media. Some ideas on where l p2 =16ln a. This equation has a similar form
the calculation of this parameter are described by Aris [32]. with the equation of Cussler (Eq. (4.9)).
In a similar case, Lape et al. [35] considered rectangular In a different approach, Gusev and Lusti [37] developed
nanoplatelets of the same aspect ratio arranged randomly a 3D computational model for a random array of parallel
but parallel to each other and perpendicular to the diffu- circular disks. This model is based on the solution of the La-
sion direction. The permeability of the nanoplatelet rein- place equation for the local chemical potential:
forced membrane is reduced then from that in the pure rKr  rl 0 4:22
polymer by the product of the reduced area and the in-
creased path length. with a space dependent local permeability coefcient. The
    latter is assumed to have a value of zero inside the platelets
K composite Af and K matrix anywhere else. The proposed expression for the
 0 4:15
K matrix A0 global permeability becomes:
 976 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984

Picard et al. [38] modied the model of Lape et al. [35]

(Eq. (4.26)), considering not only the polydispersity of the
width of the ller but also the polydispersity of its
thickness:

K composite 1/
2 3 ; 4:28
K matrix P wi 2 2
ni t
6 7
41 13 / P n wi 5
i

it
i

where wi and t i are the width and the thicknesses of the

Fig. 10. (a) The dilute regime of concentration in an oriented disk
composite a/  1. The disks are spaced by a mean distance that exceeds
the disks radius R. (b) The semi-dilute regime a/
1. The disks are
strongly overlapping due to their great aspect ratio, creating a tortuous
path.
MACROMOLECULAR NANOTECHNOLOGY
fractions i of the platelets. This model [38] is more appro-
priate for the cases of high loading of the impermeable
phase, where, due to the presence of agglomerates, there
is a distribution in the values of the aspect ratio.
The comparison of the models that have been presented
so far is not straightforward because the denition of the
aspect ratio of the platelets is not common in all the mod-
els. Some authors dene this ratio as the half width to
thickness ratio, while others as the width to thickness ra-
tio. Adopting the latter as the typical denition of the as-

pect ratio, some of the models should be modied

"   # accordingly. This is included in the curves of Fig. 11, which
K composite a/ b compares the predictions of the models. In the model pro-
exp  : 4:23 posed by FredricksonBicerano (Eqs. (4.20) and (4.21)),
K matrix x0
where the platelets are considered to have a circular disk
The parameters b and x0 are usually evaluated by tting shape, another modication of the aspect ratio is needed.
experimental data [37]. The values of b 0:71 and The relation between the disk aspect ratio and the rectan-
x0 3:47 were used by Picard et al. [38] in order to obtain gular platelet aspect ratio can be derived by comparing the
similar results as with the model of Nielsen (Eq. (4.7)) for area of a disk (pR2 ) to the area of a rectangular platelet (L2 ).
p
the diffusion of several gasses in nylon-6/montmorillonite Thus, adisk a= p. The coefcients j and l should also be
composites. modied accordingly.
All the models above assumed nanoplatelets of uniform Fig. 11 compares the predictions of some models as
size. If, instead, there is a continuous distribution of sizes, functions of the platelet volume fraction and for different
gR, e.g. Gaussian of average size R and deviation r0 : aspect ratios. Two important observations can be made
! in this gure: rst that all models require (relatively) large
1 R  R2 volume fractions or large aspect ratios for a large reduction
gR p exp  ; 4:24
2pr0 2r02 of the permeability. Second that the predictions for the
trend of this reduction are different between the several
and constant thickness, W, then the mean number of nano-
models mainly in the area of low volume fractions and as-
platelets of size 2R encountered by the solute becomes
pect ratios. The best way to compare the predictions of the
[35]:
models, then, is to use the product a/ as the signicant
/ gRRdR parameter of the membrane.
hNRi  R1 4:25 Fig. 11 shows that in the dilute regime the models of
W 0 gRRdR
Cussler (Eq. (4.9)), FredricksonBicerano (Eq. (4.21)) and
This gives: Gusev (Eq. (4.23)) present similar behaviour for low parti-
K composite 1/ cle aspect ratios (Fig. 11a, a 10). The models of Nielsen
: 4:26 (Eq. (4.7)) and of FredricksonBicerano (Eq. (4.20)) indicate
K matrix 1 3 WR R2 r02
2 /
a stronger effectiveness of the platelets in reducing the
If one compares this case with the case of nanoplatelets permeability under these conditions. In semi-dilute sys-
with a uniform size R R, then: tems and for intermediate aspect ratios (a 100), how-
" ever, the model of FredricksonBicerano (Eq. (4.21))
 0 2 #2
K uni r seems to underestimate the barrier properties, while Cuss-
 1 : 4:27 lers model overestimates them (Fig. 11b, a 100). For
K polydisp R
higher aspect ratios (Fig. 11c, a 1000) Gusevs model
Increasing the polydispersity of the size distribution, in- predicts almost zero permeability for a/ > 30. As this
creases the standard deviation and reduces the permeabil- product goes to higher values (e.g. a/ > 60 in Fig. 11c)
ity. Thus the barrier properties of the nanoplatelets with the absolute values of the permeability predicted by all
polydispersed sizes are better than these of monodispersed the models become very low, in the range of 13% of the
of the same mean size. value of the unlled polymer.
 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984 977

a 1.0 quate to describe the reduction of the relative permeabil-

ity. An example of the effect of the aspect ratio, a, on the
predictions of Nielsens model is shown in Fig. 8.
0.9

4.3. Arrangements at angles h90 to the main diffusion

0.8
direction
Kcomp / Km

0.7
The tortuosity, s, is the highest when the nanoplatelets
eq. 4.7 are aligned perpendicular to the diffusion direction. For
0.6 other orientation angles Nielsens equation should be mod-
eq. 4.9
eq. 4.20 ied accordingly.
0.5 For non-uniform orientation of the platelets an order
eq. 4.23
eq. 4.21 parameter, S0 , is introduced to quantify the degree of their
0.4 orientation around the diffusion direction [26]:
0.00 0.02 0.04 0.06 0.08 0.10
vol. fraction 1
S0 h3 cos2 h  1i; 4:29
2
b 1.0
eq. 4.7
where h is the angle between the diffusion direction and
eq. 4.9 the unit vector normal to the surface of a platelet, and
0.8 eq. 4.23 the average is taken over all platelets with all possible ori-
eq. 4.21 entations. When all platelets are parallel to the direction of
diffusion (h 0), then the order parameter is S0 1=2.

MACROMOLECULAR NANOTECHNOLOGY
Kcomp / Km

0.6 When h p=2, the orientation of the platelets is perpen-

dicular to the diffusion direction and S0 1. For random
0.4 orientation S0 0 (see Fig. 12).
For non-uniform orientation Nielsens equation takes
the following form [26]:
0.2
K composite 1/

: 4:30
0.0 K matrix 1 a2 23 S0 12 /
0.00 0.02 0.04 0.06 0.08 0.10
Obviously, the case S0 1 corresponds to the maximisa-
vol. fraction
tion of the tortuosity factor and, therefore, to the greatest
reduction of the permeability. The term 23 S0 1=2
c 1.0
( hcos2 hi from Eq. (4.29)) gives the factor 1=3 for S0 0.
eq. 4.21
eq. 4.7 The predictions are then similar to the ones of the models
0.8 eq. 4.23 of Lape et al. [35] and Fredrickson and Bicerano [36] for
eq. 4.9 random spatial positioning of clay layers.
An empirical relation for the prediction of the perme-
Kcomp / Km

0.6
ability in nanocomposites with randomly (3D) oriented
platelets has been proposed by Maksimov et al. [39] in
0.4 the form of:

0.2 K composite K k 2K matrix 1  / ; 4:31

3

0.0 where K k is the permeability of the nanocomposite as gi-

0.00 0.02 0.04 0.06 0.08 0.10 ven by the model of Nielsen for oriented platelets. The sec-
vol. fraction ond term corresponds to the permeability of the
composites with platelets oriented parallel to the diffusion
Fig. 11. Predictions of the relative permeability as a function of platelets direction and it is assumed to be equal to the permeability
volume fraction for different models, for a particle aspect ratio: (a)
of the matrix corrected for its volume fraction.
a 10; (b) a 100; and (c) a 1000.

4.4. Inuence of the interfacial regions

From a practical point of view, complete exfoliation is
difcult to be achieved at high clay fractions and the actual
aspect ratios are much lower than what is needed for ef-
cient enhancement of the barrier properties. For interme-
diate cases, and given the comparisons with the few
available experimental data in the following sections, we
could conclude that the simple model of Nielsen is ade-
In many cases, the permeability of the nanocomposite
can be affected by the existence of interfacial regions be-
tween the matrix and the inorganic particles. Such do-
mains may enhance the diffusion coefcient. The
interfaces are caused either by the surfactant that is used
for the modication of the particles or/and are due to the
formation of voids between the different phases.
 978 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984
Fig. 12. Values of the order parameter for three orientations of the platelets [26].
The relative diffusion coefcient in such cases has been
expressed by Sorrentino et al. [40] as:
Dcomposite a0
; 4:32
MACROMOLECULAR NANOTECHNOLOGY
Dmatrix s the enhancement of the barrier properties.
where a0 is a factor that quanties the effect of the interfa-
cial regions:
a0 1 b0 /: 4:33
The parameter b0 is given by:
V s Ds V s V f
b0   : 4:34
V f D0 Vf
Ds , V s are the diffusion coefcient and the volume of the
interfaces, D0 is the diffusion coefcient for the neat poly-
mer and V f is the volume of the nanoplatelets. If V s =V f is
negligible, then b0 1 and a0 1  /. However, the
parameters Ds , V s are not easily measurable. The value of
the parameter b has a strong effect on the diffusivity for
all concentrations of the impermeable phase.
4.5. Inuence of the particle aggregation
A crucial factor that affects the permeation properties of
the nanocomposite is the aggregation of silicate layers
which leads to a decrease of the aspect ratio of nanoparti-
cles. Manninen et al. [19] indicated that the processing
path taken to prepare the nanocomposites may result in
agglomeration of the organoclay layers. These agglomer-
ates may cause the formation of large scale holes (pores)
in the matrix, which can act as low resistance pathways
for gas transport within the nanocomposite. In such a case
the diffusion becomes mostly of the Knudsen ux type.
For high clay contents, it is difcult to keep a high de-
gree of platelet dispersion and avoid the presence of inter-
calated structures. Nazarenko et al. [41] modied Nielsens
equation to take into account the case of stacks of layers
(aggregates) dispersed homogeneously in the polymer ma-
trix and oriented perpendicular to the diffusion direction.
The proposed expression is:
K composite 1/
a / 4:35
K matrix 1 2N
where N is the number of layers in the layer stack. Obvi-
ously, the value N 1 corresponds to complete layer
delamination (exfoliation). Again, the larger the degree of
stacking, i.e. the higher the value of N, the less efcient is
One observes that Eqs. (4.30) and (4.35) have a similar
form. The former accounts for the orientation of the plate-
lets, while the latter for their degree of stacking. Therefore,
by combining these two equations we can obtain a single
equation that might predict the change in permeability
as a function of the volume fraction, the geometry, the ori-
entation and the degree of stacking of the nano-platelets:
K composite 1/
a S0 1 / : 4:36
K matrix 1 3N 2
4.6. Experimental and numerical validation of the models
A common feature in all the models that were pre-
sented above is the dependence of the permeation proper-
ties of the nanocomposite on the volume fraction and the
aspect ratio of the reinforcing inorganic phase. In fact,
the predictions of all the models are similar. When used
in the reverse way, i.e. to predict the aspect ratio from
the measured permeation properties, however, these mod-
els give different results. Some such predicted results for a
number of the above described methods has been tabu-
lated by Picard et al. [38] as applied in a nylon-6/montmo-
rillonite system.
The average aspect ratio of the particles is difcult to
determine experimentally and, therefore, the comparison
of the models is not straightforward. It has been observed
that during the preparation of the nanocomposites the ob-
tained aspect ratio depended on the volume fraction. In
chlorobutyl/montmorillonite nanocomposites, e.g., the ob-
tained aspect ratio decreased with increasing /, using
either modied or non-modied particles [42]. Obviously,
increasing / results in a reduction of the degree of delam-
ination and, consequently, in the reduction of the nal par-
ticle aspect ratio.
The decrease of the relative permeability is attributed in
all models mainly to the inorganic phase. If their aspect ra-
tio and volume fraction are the same, the particles will in-
 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984
duce the same relative reduction of the permeability in any
matrix they are dispersed, provided that the same degree
of delamination is obtained. Further, if there are no special
interactions between the solute gas and the particles, the
relative enhancement of the barrier properties is not af-
fected by the kind of the gas that is used [43,44],
(Fig. 13). In other words, the improvement of the barrier
properties of the material are determined solely by the dis-
persed inorganic phase.
Fig. 14 gives the predictions of Eq. (4.7) for some pub-
lished data, with the corresponding values of a for best
t. The gure shows that the model of Nielsen (Eq. (4.7))
gives reasonable predictions for most experimental data,
such as butyl rubber/vermiculite composites [14], poly(e-
caprolactone)/MMT [45], siloxane modied epoxy resin/
MMT [43], PLA/MMT [46]. Each tting curve requires a dif-
ferent value of the aspect ratio. In Fig. 14a and c the clay
particles are the same but the matrices in which they are
dispersed are different. The variation of the aspect ratio
seems to be the result of the different degrees of exfolia-
tion of MMT in these matrices. In Fig. 14d a value of
a 150 for the vermiculite particles seems logical as these
particles have greater aspect ratio than montmorillonite.
Fig. 13. Experimental results of relative permeability measurements for
different gases [43,44]. (a) Nitrogen and oxygen in epoxy/MMT; (b)
Oxygen and carbon dioxide in PE/HDPE-g-MA20/MMT.
979
The reduction of the permeability in the platelet-rein-
forced nanocomposites is attributed to the lowering of
the diffusion coefcient due to a more tortuous path for
diffusing molecules, and to the decrease in solubility due
to a reduced polymer matrix volume. An enhancement in
solubility with the addition of inorganic particles has been
reported by Ogasawara et al. [47], with the permeability,
D  S, remaining unchanged. This appeared to be the result
either of gas adsorption at the surface of the inorganic par-
ticles or it might be caused by defects in the nanocompos-
ite. None of these effects is accounted for by the mentioned
models.
An almost linear reduction of the relative permeability
has been reported in some cases, e.g. H2 O vapour or CO2
in nanocomposites of the type PE/HDPE-g-MA20/MMT
[44] and poly(e-caprolactone)/mica [48] (Fig. 15). This lin-
ear reduction cannot be predicted by any of the models
and is probably attributed to the behaviour of the solubil-
ity coefcient. These experimental results indicate that the
tortuous path model may not be enough to cover all per-
meation processes. On the other hand, the permeability
MACROMOLECULAR NANOTECHNOLOGY
of water vapour in montmorillonite/styrene-acrylate
copolymers was found by Maksimov et al. [39] to follow
Nielsens model (Eq. (4.7)) at low weight fractions for
a 210. At / > 0:03 incomplete exfoliation and decrease
of the particle aspect ratio of the montmorillonite lead to
a reduction of the efciency of the barrier properties.
Assuming that Nielsens model holds at / 0:07 (wt.), an
average stacking index of 1.8 can be calculated from the
data of [39] using Eq. (4.35).
Comparisons of the results of Eqs. (4.10) and (4.11) with
Monte Carlo calculations for the diffusion through mem-
branes containing impermeable akes were conducted by
Falla et al. [31] and Swannack et al. [49]. Both papers re-
ported rather good agreement between the theoretical
and the simulation results, verifying the tortuous paths
around the akes and the effects of the constrictions into
the slits and the diffusion through them. Falla et al. [31]
showed that Eq. (4.10) is in general more sensitive to
changes in the ake arrangement that will inuence the
diffusion mechanism, and can predict more accurately
the resulting changes in the permeability of the mem-
brane. Swannack et al. [49] indicated the need for 3D cal-
culations and showed that Eq. (4.10) over-predicts the
barrier effect of the membrane in this case, as the trans-
verse slits are omitted.
5. Using PALS to study the permeability of polymers and
nanocomposites
The molecular transport through a polymer membrane
depends also on the amount of the existing free volume.
The free volume is created by the inefcient chain packing
of the polymer generated by chain segment rearrangement.
The increase of the free volume facilitates the diffusion pro-
cess because of the creation of easier pathways for the sol-
utes. In polymer-nanocomposite membranes the presence
of the inorganic phase can inuence the size and the num-
ber of the free volume holes, especially at the interfaces.
Consequently, the change in the permeability of such
 980 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984
MACROMOLECULAR NANOTECHNOLOGY

Fig. 14. Predictions of Eq. (4.7) for the relative permeability of gasses compared to experimental results taken from [43,45,46,14]: (a) oxygen in epoxy/MMT
(b) water vapour in poly(e-caprolactone/MMT nanocomposite; (c) carbon dioxide in butyl rubber/vermiculite nanocomposite; (d) oxygen in PLA/MMT
nanocomposite. Each tting curve corresponds to a selected value for the aspect ratio L=W.

systems is a balance between the barrier properties of the lation process of o-Ps in the free volume holes. This is the
impermeable nanoplatelets and the possible increase of process that is used to estimate the free volume in amor-
the free volume of the matrix. Therefore, the nanoplatelets phous polymers.
may not always promote efciently the nal barrier proper- The size of the free volume holes, R, in the polymer can
ties of the material. be determined from the measured lifetime, soPs , using the
Positron Annihilation Lifetime Spectroscopy (PALS) is a semi-empirical equation:
nondestructive technique that provides an effective ap-   1
1 R 1 2pR
proach for the study of the free volume in the solid state. soPs 1 sin : 5:1
2 R0 2p R0
It is based on the localisation of the positronium (Ps) in
the free volume holes because of their repulsion by the sur- The relative intensity gives the probability for the for-
rounding atoms. Generally, the annihilation spectra in mation of o-Ps, which is proportional to the number of
polymers consist of three exponentially decaying compo- holes, i.e. the overall free volume [21,50]. Plotting the mea-
nents that correspond to the three main processes of their sured life time of the o-Ps and the intensity as a function of
annihilation [21]. Each of these processes is characterised clay particles loading, one can estimate any variation in the
by the mean life time and the probability of annihilation polymer free volume that may be caused by the presence
(intensity). The longest life time corresponds to the annihi- of the nanoclay particles [51].
 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984 981

The interactions between the polymer matrix and the
inorganic particles may cause the creation of distinct inter-
facial layers. These interfaces provide a favourable area for
positron trapping and annihilation, even in the absence of
extra free volume. Thus, the results of the PALS method
conducted in nanocomposites can be complicated to inter-
pret, because the o-Ps intensity may not reect always di-
rectly the free volume hole concentration [52,53].
Despite these difculties, Garcia et al. [23] have found
that the presence of the nanoplatelets increased the
amount of free volume in Nylon  6=SiO2 nanocomposites
and, therefore, compromised the reduction of the perme-
ability. Thus, in some cases the presence of the inorganic
particles may even act destructively in the barrier proper-
ties of the membrane. On the other hand, in the case of
coatings reinforced with laponite, the free volume was
found not to depend on the volume fraction of the clay
[51]. It seems, therefore, that every system should be stud-
ied separately.
6. Methods to inuence the permeability of the
nanocomposite
6.1. Electrically modulated permeability using side-chain
liquid crystalline polymers
Because the orientation of the impermeable barriers
(crystallites, platelets etc.) is paramount for the efcient
enhancement of the barrier properties of the material, at-
tempts have also been reported of changing this orienta-
tion by means of external elds. An example is the use of
an electric eld to modulate the orientation of liquid crys-
talline (LC) polymers and control, thus, the barrier proper-
ties of the material [54].
Side-chain liquid crystalline polymers (SCLCP) present a
similar picture for their permeability with that of semi-
crystalline polymers. Here the LCP side-chains may form
LC domains by association, which can be impermeable to
the gas and are the equivalent of the crystallites in the
semicrystalline polymer. There is a signicant difference,
however, in that the impermeable LC domains can become

MACROMOLECULAR NANOTECHNOLOGY
aligned by the application of an external electric eld [54].
These domains have strong dielectric anisotropy. Thus, if
a the external electric eld strength exceeds a threshold,
1.0 experimental points H2 O
the domains can become oriented parallel or perpendicular
to its direction, depending on the sign of their dielectric
0.8 anisotropy. Due to their dense structure, the LC domains
can be considered as barriers for the diffusion of other mol-
ecules. Therefore, depending on their orientation, more or
0.6
m

less tortuous paths are created for the gas molecules with-
Kcomp / K

in the LCP and the permeability of the material is affected

0.4 accordingly.
Let us suppose that the LC domains have orthogonal
shape of length L and thickness W, they are aligned parallel
0.2 to each other and at an angle h to the diffusion direction.
Using the schematic model of Fig. 16 Ly and Cheng [55]
0.0 estimated the tortuosity factor as:
0.00 0.01 0.02 0.03 0.04 0.05
clay volume fraction 1
s Fa; r  ; 6:1
sin h
b experimental points CO2 where Fa; r is a function of the aspect ratios of the
1.0 impermeable domains, a L=W, and the diffusion chan-
nels, r d=W. The polarisation, P, induced by the external
electric eld, E, is the dipole moment, l, per unit volume in
the direction of the eld::
m
Kcomp / K

0.8 P e0 e  e1 E ND lhsin hi; 6:2

where e0 is the dielectric permittivity of vacuum, e and e1

are the dielectric and the optical dielectric constants of the
0.6 material, N D is the number of dipoles per unit volume and h
! !
is the angle between l and E . This gives:

e0 e  e1 E
0.00 0.05 0.10 0.15 0.20 hsin hi : 6:3
ND l
clay content (wt%)
From Eqs. (6.1) and (6.3) it seems that s 1= sin h and
Fig. 15. Relative permeability of the nanocomposites versus clay volume sin h E so s 1=E. Therefore, depending on the strength
fraction or clay content: (a) poly(e-caprolactone)/mica [48] for H2O
and the orientation of the applied eld we can adjust the
vapour, and (b) PEmet/HDPE-g-MA20/MMT [44] for CO2. An almost linear
reduction is observed, probably due to the contribution of the solubility tortuosity factor and consequently the permeability of
coefcient. The line does not correspond to any of the described models. the SLCP.
 982 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984
a LC domains
b t0
field direction
diffusion
direction
c and without additives, then:
diffusion
direction
field direction
MACROMOLECULAR NANOTECHNOLOGY
Fig. 16. (a) Uniform and parallel LC domains with an angle h to the
diffusion direction. (b) Alignment of the LC domains due to the external
applied electric eld. This arrangement corresponds to the maximum
tortuosity factor. (c) Minimisation of the tortuosity factor due to the
horizontal electric eld. The arrows indicate the dipole moment of each
domain.
6.2. Improving the barrier properties using reactive additives
The barrier properties of the nanocomposites may fur-
ther be improved by the addition to the membrane of a
substance which either has higher afnity or can react
with the penetrating gas molecules, stopping them from
advancing their diffusion. The additive must preferably
be immobilised inside the membrane and distributed in
channels with low resistance to permeation. In this way
an increase may be accomplished in the needed time lag,
t0 . The process is delayed, thus, while the steady state
properties remain unchanged.
Let us suppose that the reaction between the gas mole-
cules and the additive material is very fast and irreversible.
While the additive molecules are immobilised, the perme-
ating gas reacts and consumes them. The reaction occurs at
a front in depth 0 t, which propagates deeper in the mem-
brane [33]. If the quantity of the additive material is con-
siderable, this front will move slowly.
We assume a quasi-steady state for the diffusion of the
gas molecules from the surface to the front:
2 coefcient. If the latter is reduced, according to any of the
d z
D 2
0;
dz
where z is the distance from the surface of the membrane
where the gas molecules enter. This equation is solved for
the appropriate boundary conditions to give the penetra-
tion depth, 0 t:
s
0 2mDHc1;0 t
; 6:5
c2;0
where c1;0 and c2;0 are the initial concentrations of the gas
and the additive respectively, m is the stoichiometric coef-
cient for the reaction between gas and additive, and H is
the partition coefcient of the gas.
Setting 0 equal to the membrane thickness, we can esti-
mate the time lag:
2 c2;0
 : 6:6
2mDH c1;0
After this time the ux comes to steady state. The model
fails for c2;0 0, as it gives gives t 0 0 then. This is attrib-
uted to the approximation of the pseudo-steady state that
was used.
If one compares the time lag for the membrane with
t0 3c2;0
: 6:7
t0without mHc1;0
Since H is a property of the polymer, the key parameter
that can be adjusted is the concentration of the additive:
the more the better.
7. Low permeability nanocomposite coatings and other
applications
Low permeability nanocomposite materials can be used
in a variety of applications especially in liquid and food
packaging. Generally, polymers present many advantages
in comparison with conventional glass containers, such
as, e.g., low weight and mechanical strength. Polyethylene
(PE), Polypropylene (PP) and poly(ethylene terephthalate)
(PET) are widely used in bottles and containers. However,
their limited barrier properties to oxygen make them inap-
propriate for products requiring long self life. Polyester
nanocomposites offer superior barrier properties to oxygen
and high transparency and are suitable for use in the man-
ufacture of closed containers [56]. Nanoplatelets rein-
forced blends are also used for barrier properties to
oxygen and to organic solvents, as they also offer high
mechanical properties [57,58]. Materials for hydrogen
storage tanks is another area of applications for organoclay
compositions [59]. The use of the microbial barrier proper-
ties of the organoclays in polymers has been studied by
[60].
The working time, tactiv e for such a packaging lm is gi-
ven by Cussler [61]:
2
dfilm
tactive ;
Dfilm
where dfilm is the thickness of the lm and Dfilm its diffusion
6:4 models described in this review, then the working time of
the lm will increase accordingly. e.g. [31]:
 !
2
d a2 /2 a/ 4a/ pa2 /
tactive film 1 ln
D0 1  / r p1  / r1  /
7:1
If, further, use is made of reactive additives (Eq. (6.7)), then
the working time of the composite lm becomes:
 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984 983

3c2;0 barrier properties of a general nanocomposite that will

t active
mHc1;0 lead to the appropriate selection of the nanocomposites
2  ! composition is still wanted.
d a2 /2 a/ 4a/ pa2 /
 film 1 ln
D0 1  / r p1  / r1  /
7:2
Another area of research on this subject is the develop-
ment of coatings incorporating nano-platelets. While this
is mostly done to improve the mechanical properties (e.g.
chipping of automobile paint) [6264], the improvement
of the barrier properties has not gone unnoticed. Polymeric
nanocomposite coatings is an alternative inexpensive oxy-
gen barrier material [65]. Such anticorrosive coatings can
be used on zinc-steel plates for automobiles [66]. Asphalt
nanocomposite based coatings can be useful for materials
for building components such as shingles [67].
One such project that examines the utilisation of nano-
platelets with different aspect ratios for the development
of coatings with exceptional gas barrier properties is cur-
rently under way at the Technical University of Crete.
The research uses this method aiming to reduce the per-
meability of coatings used for the long time protection of
large marble and stone surfaces against polluting gasses.
8. Conclusions and recommendations
The addition of inorganic impermeable nanoplatelets
improves the barrier properties of polymers. This is attrib-
uted mostly to the lengthening of the diffusion path of the
permeating gas molecules due to the increase of the tortu-
osity. Increasing the aspect ratio of the platelets and their
volume fraction improves these properties. The addition
of solute scavengers enhances these properties, at least
by prolonging the induction period for permeation.
Most of the proposed models for the estimation of the
gas permeability of organoclay reinforced polymers deal
with the ideal case of full exfoliation and assume that the
physical characteristics of the polymer remain unchanged
with the addition of the inorganic particles. The common
feature in all the models is the dependence of the relative
permeability on the volume fraction, the aspect ratio and
the orientation of the platelets, something that is veried
by the experiments.
The reduction of the relative permeability is correlated
to and determined by the inorganic phase and seems to
be independent of the nature of the polymer matrix and
the gas species. The experimental validation of the models
and their comparison is not simple, because the actual as-
pect ratio and orientation of the platelets in the composites
is difcult to estimate. When the geometric parameters of
the clay/polymer system are known, it seems that the
model proposed by Nielsen, which is the simplest, does a
reasonable job in predicting the permeability.
In fact, things may be more complicated because the
nanocomposites present a variety of morphological fea-
tures. Different kinds of interactions between the polymer
and inorganic platelets may affect the free volume in the
matrix, the interfacial regions between the two different
phases and the degree of delamination of the silicate lay-
ers. It seems that an accurate enough prediction for the
In some cases the orientation of the platelets can be
measured and controlled using electric or magnetic elds
or mechanical stresses. The barrier properties can be fur-
ther improved by using nanoplatelets that have polydis-
perse size distribution or/and by using reactive additives,
which can obstruct the diffusion process, at least in the ini-
tial stages.
More systematic measurements are required to clarify
the effects of the inorganic particles on the permeation
process. For example, it will be useful to establish perme-
ability measurements for samples containing the same
matrix with different inorganic particles, in various con-
centrations, or/and different matrices with the same dis-
persed particles. PALS measurements may also be useful
to reveal the inuences of the nanoplatelets on the free
volume and the interfacial regions. Further, it may be
worth nding out if the addition of a polar material
through the low resistance diffusion channels (if such ex-
MACROMOLECULAR NANOTECHNOLOGY
ist) can improve the barrier properties, at least in the case
of (also) polar penetrants. Making the diffusion channels
more polar can also be useful for the indirect estimation
of the slit aspect ratio via dielectric measurements.
Acknowledgements
We thank ELKE (the Research Funds Financial Adminis-
tration) of the Technical University of Crete for partially
supporting the research project of G. Choudalakis. We also
thank the unknown (to us) reviewer who suggested the
form of Eq. (4.36).
References
[1] Ray SS, Okamoto M. Polymer/layered silicate nanocomposites: a
review from preparation to processing. Prog Polym Sci
2003;28:1539641.
[2] Usuki A, Kawasumi M, Kojima Y, Okada A, Karauchi T, Kamigaito O.
Swelling behavior of montmorillonite cation exchanged for x-amino
acids by e-caprolactam. J Mat Res 1993;8(5):11748.
[3] Lagaly G. Interaction of alkylamines with different types of layered
compounds. Solid State Ionics 1986;22:4351.
[4] Yoon K, Sung H, Hwang Y, Noh S, Lee D. Modication of
montmorillonite with oligomeric amine derivatives for polymer
nanocomposite preparation. Appl Clay Sci 2007;38(12):18.
[5] Yano A, Usuki A, Okada A, Karauchi T, Kamigaito O. Synthesis and
properties of polyimide-clay hybrid. J Polym Sci Part A: Polym Chem
1993;31:24939.
[6] Usuki A, Hasegawa N, Kato M. Polymerclay nanocomposites. Adv
Polym Sci 2005;179:13595.
[7] Nguyen QT, Baird DG. Preparation of polymerclay nanocomposites
and their properties. Adv Polym Tech 2006;25:27085.
[8] LeBaron PC, Wang Z, Pinnavaia TJ. Polymer-layered silicate
nanocomposites: an overview. Appl Clay Sci 1999;15:1129.
[9] Powell CE, Beall GW. Physical properties of polymer/clay
nanocomposites. Curr Opin Solid State Mater Sci 2006;10:7380.
[10] Jordan J, Jacob K, Tannenbaum R, Sharaf M, Jasiuk I. Experimental
trends in polymer nanocomposites-a review. Mat Sci Eng A
2005;393:111.
[11] Rogers WA, Buritz RS, Alpert D. Diffusion coefcient, solubility and
permeability for helium in glass. J Appl Phys 1954;25:7.
[12] van Amerongen GJ. The permeability of different rubbers to gases
and its relation to diffusivity and solubility. J App Phys 1946;17.
[13] Wicke E, Kallenbach R. Die oberchendiffusion von kohldioxyd in
aktieven kohlen. Kolloid-Z 1941;97:13551.
 984 G. Choudalakis, A.D. Gotsis / European Polymer Journal 45 (2009) 967984
[14] Takashi S, Goldberg HA, Feeney CA, Karim DP, Farrel M, OLeary K,
et al. Gas properties of butyl rubber/vermiculite nanocomposite
coatings. Polymer 2006;47:308393.
[15] Neogi P. Diffusion in polymers. Marcel Dekker Inc.; 1996.
[16] Wong B, Zhang Z, Handa P. High precision gravimetric technique for
determining the solubility and diffusivity of gases in polymers. J
Polym Sci B: Polym Phys 1998;36:202532.
[17] Ballik CM. On the extraction of diffusion coefcients from
gravimetric data for sorption of small molecules by polymer thin
lms. Macromolecules 1996;29:30259.
[18] Kemp DR, Paul DR. Gas sorption in polymer membranes containing
adsorptive llers. J Polym Sci Part B: Polym Phys 1974;12:485500.
[19] Manninen A, Naguib H, Nawaby A, Day M. Co2 sorption and diffusion
in polymethyl methacrylate-clay nanocomposites. Polym Eng Sci
2005;45:90414.
[20] Hiltner A, Liu RYF, Hu YS, Baer E. Oxygen transport as a solid state
structure probe for polymeric materials: a review. J Polym Sci Part B:
Polym Phys 2005;43:104763.
[21] Pethrick R. Positron annihilation a probe for nano-scale voids and
free volume. Prog Polym Sci 1997;22:147.
[22] Cohen MH, Turnbull D. Molecular transport in liquids and glasses. J
Chem Phys 1959;31(5):11649.
[23] Garcia M, Barsema J, Galindo RE, Gangialosi D, Turiel J, Van Zyl WE,
et al. Hybrid organic inorganic Nylon-6/SiO2 nanocomposites:
transport properties. Polym Eng Sci 2004;44:12406.
[24] Mogri P, Paul DR. Gas sorption and transport in side-chain crystalline
and molten poly(octadecyl acrylate. Polymer 2001;42:253142.
[25] Michaels AS, Bixler HJ. Flow of gases through polyethylene. J Polym
MACROMOLECULAR NANOTECHNOLOGY
Sci L 1961:41339.
[26] Bharadwaj RK. Modeling the barrier properties of polymer-layered
silicate nanocomposites. Macromolecules 2001;34:918992.
[27] Barrer RM, Petropoulos JH. Diffusion in heterogeneous media: lattice
and parallelepipeds in a continuous phase. Br J Appl Phys
1961;12:6917.
[28] Nielsen LE. Models for the permeability of lled polymer systems. J
Macromol Sci A 1967;1:92942.
[29] Wakeham WA, Mason EA. Diffusion through multiperforate
lamninae. Ind Eng Chem Fundam 1979;18(4):3015.
[30] Cussler EL, Hughes SE, Ward WJ, Aris R. Barrier membranes. J Membr
Sci 1988;38:16174.
[31] Falla WR, Mulski M, Cussler EL. Estimating diffusion through ake-
lled membranes. J Membr Sci 1996;119:12938.
[32] Aris R. On a problem in hindered diffusion. Arch Ration Mech Anal
1986;12:8391.
[33] Moggridge GD, Lape NK, Yang C, Cussler EL. Barrier lms using akes
and reactive additives. Prog Organ Coat 2003;46:23140.
[34] Brydges WT, Gulati ST, Baum G. Permeability of glass ribbon-
reinforced composites. J Mater Sci 1975;10:20449.
[35] Lape NK, Nuxoll EE, Cussler EL. Polydisperse akes in barrier lms. J
Membr Sci 2004;236:2937.
[36] Fredrickson GH, Bicerano J. Barrier properties of oriented disk
composites. J Chem Phys 1999;110:21818.
[37] Gusev AA, Lusti HR. Rational design of nanocomposites for barrier
applications. Adv Mater 2001;13(21):16413.
[38] Picard E, Vermogen A, Gerard J, Espuche E. Barrier properties of
nylon-6 montmorillonite nanocomposite membranes prepared by
melt blending: inuence of the clay content and dispersion state:
consequences of modelling. J Membr. Sci 2007;292:13344.
[39] Maksimov RD, Gaidukov S, Zicans J, Jansons J. Moisture permeability
of a polymer nanocomposite containing unmodied clay. Mech
Comp Mat 2008;44(5):50514.
[40] Sorrentino A, Tortora M, Vittoria V. Diffusion behavior in polymer
clay nanocomposites. J Polym Sci Part B: Polym Phys 2006;44:26574.
[41] Nazarenko S, Meneghetti P, Julmon B, Qutubuddin S. Gas barrier of
polystyrene montmorillonite clay nanocomposites: effect of mineral
layer aggregation. J Polym Sci Part B: Polym Phys 2007;45:173353.
[42] Sridhar V, Tripathy DK. Barrier properties of chlorobutyl nanoclay
composites. J Appl Polym Sci 2006;101:36307.
[43] Yeh JM, Huang HY, Chen CL, Su WF, Yu YH. Siloxane modied epoxy
resinclay nanocomposite coatings with advanced anticorrosive
properties prepared by a solution dispersion approach. Surf Coat
Technol 2006;200:275363.
[44] Jacquelot E, Espuche E, Gerard JF, Duchet J, Mazabraud P.
Morphology and gas barrier properties of polyethylene-based
nanocomposites. J Polym Sci Part B: Polym Phys 2006;44:43140.
[45] Tortora M, Vittoria V, Galli G, Ritrovati S, Chiellini E. Transport
properties of modied montmorillonite-poly(-caprolactone)
nanocomposites. Macromol Mater Eng 2002;287:2439.
[46] Ray SS, Bousmina M. Biodegradable polymers and their layered
silicate nanocomposites: in greening the 21st century materials
world. Prog Mater Sci 2005;50:9621079.
[47] Ogasawara T, Ishida Y, Ishikawa T, Aoki T, Ogura T. Helium gas
permeability of montmorillonite/epoxy nanocomposites. Comp Part
A 2006;37(12):223640.
[48] Messersmith PB, Giannelis EP. Synthesis and barrier properties of
poly (-caprolactone)-layered silicate nanocomposites. J Polym Sci
Part A: Polym Chem 1995;33:104757.
[49] Swannack C, Cox C, Liakos A, Hirt D. A three-dimensional simulation
of barrier properties of nanocomposite lms. J Membr Sci
2005;263:4756.
[50] Eastmond GC, Daly JH, McKinnon AS, Pethrick RA. Poly(ether
imide)s: correlation of positron annihilation lifetime studies with
polymer structure and gas permeability. Polymer 1999;40:
360510.
[51] Choudalakis G, Gotsis AD, Schut H, Picken SJ. unpublished data,
2008..
[52] Wang ZF, Wang B, Qi N, Zhang HF, Zhang LQ. Inuence of llers on
free volume and gas barrier properties in styrene-butadiene rubber
studied by positrons. Polymer 2005;46:71924.
[53] Wang SJ, Liu LM, Fang PF, Chen Z, Wang HM, Zhang SP.
Microstructure of polymerclay nanocomposites studied by
positrons. Radiat Phys Chem 2007;76:10611.
[54] Ly YP, Cheng YL. Electrically-modulated variable permeability liquid
crystalline polymeric membrane. J Membr Sci 1993;77:99112.
[55] Ly YP, Cheng YL. Diffusion in heterogeneous media containing
impermeable domains arranged in parallel arrays of variable
orientation. J Membr Sci 1997;133:20715.
[56] Barbee R, Matayabas C. Nanocomposites for high barrier 2009
applications 2000. United States Patent WO0078855, 28.12.2000,
European Patent EP 0 670 058 B1, 08.07.1998, Canadian Patent CA
2,149,891, 15.02.2000, Finnish Patent FI 952480, 13.12.2002.
[57] Frounchi M, Dadbin S, Salehpour Z, Noferesti M. Gas barrier
properties of pp/epdm blend nanocomposites. J Membr Sci
2006;282(12):1428.
[58] Myung-Ho Kim, Si Ho Lee, Nanocomposite blend composition
having excellent barrier property 2006. JP2006328426 (07.12.2006).
[59] Jang Bor Z. Nanocomposite compositions for hydrogen storage and
methods for supplying hydrogen to fuel cells 2006. US2006030483
(09.02.2006).
[60] Wang Xiaoying, Du Yumin, Yang Jianhong, Wang Xiaohui, Shi Ying
Hu Xiaowen. Preparation, characterization and antimicrobial activity
of chitosan/layered silicate nanocomposites. Polymer
2006;47(19):673844.
[61] Cussler EL. Mass transfer in uid systems. Cambridge University
Press; 1984.
[62] Nobel ML, Mendes E, Picken SJ. Acrylic-based nanocomposite resins
for coating applications. J Appl Polym Sci 2007;104(4):214656.
[63] Nobel ML, Picken SJ, Mendes E. Waterborne nanocomposite resins
for automotive coating applications. Progr Org Coat 2007;58(2
3):96104.
[64] Nobel ML, Mendes E, Picken SJ. Enhanced properties of innovative
laponite-lled waterborne acrylic resin dispersions. J Appl Polym Sci
2007;103(2):68797.
[65] Kothapalli Venkata Raju, Kattimuttathu Suresh. Oxygen barrier
material for packaging 2005. US2005145138 (07.07.2005).
[66] Young Chul Lee, Myeong Jun Kim. Organo clay containing
anticorrosive coating composition and preparation method 2007.
WO2007055498 (18.05.2007).
[67] Vinay Mehta, Brian Kanze. Asphalt nanocomposite-based roong
products 2007. WO2007109314 (27.09.2007).