Epoxy-layered silicate nanocomposites and their gas permeation properties

Article abstract of DOI:10.1021/ma048798k

The synthesis of epoxy-OM (organo-montmorillonite) nanocomposites was presented and their permeability to oxygen and water vapor was measured. The variation in the chemical structure of the organic monolayer ionically bonded to the montmorillonite surface was studied and its influence on the swelling, intercalation and exfoliation behavior of the OM was presented. The correlation of the gas permeation through the composites to the volume fraction of the impermeable inorganic part of OM was also described. It was shown that the transmission rate of water vapor through the composites was more influenced by the permeant-composite interactions and the hence the hydrophobicity of the monolayer cover the inclusions surface.

Full text of DOI:10.1021/ma048798k

7250
Macromolecules 2004, 37, 7250-7257
Epoxy-Layered Silicate Nanocomposites and Their Gas Permeation
Properties
Ma ged A. Osm a n ,† Vik a s Mitta l,† Ma ssim o Mor bid elli,‡ a n d Ulr ich W. Su ter *,†
Department of Materials, Institute of Polymers, and Department of Chemistry and Applied Biosciences,
Institute of Chemical and Bioengineering, ETH Zurich, CH-8093 Zurich, Switzerland
Received J une 17, 2004
ABSTRACT: Epoxy-OM (organo-montmorillonite) nanocomposites have been synthesized, and their
permeability to oxygen and water vapor has been measured. The chemical structure of the organic
monolayer ionically bonded to the montmorillonite surface has been varied, and its influence on the
swelling, intercalation, and exfoliation behavior of the OM has been studied. Exfoliated aluminosilicate
layers build a barrier for the permeating gas molecules, while the polymer intercalated tactoids do not
contribute much to the permeation barrier performance. The gas permeation through the composites
was correlated to the volume fraction of the impermeable inorganic part of the OM. The incorporation of
small volume fractions of the platelike nanoparticles in the polymer matrix decreased its permeability
coefficient when the interface between the two heterogeneous phases was properly designed. Long alkyl
chains enhanced the polymer intercalation but increased the permeability coefficient probably due to
phase separation at the interface between the polymer and the inclusions. Matching the surface energy
of the OM with that of the matrix as well as tethering polymer molecules to the silicate layers surface
enhanced the exfoliation and decreased the permeation coefficient. The exfoliation process is governed
by interplay of entropic and energetic factors. A macroscopic volume average of the aspect ratio of
montmorillonite platelets was deduced from the relative permeability of the nanocomposites by comparing
the measured values to numerical predictions of gas permeation through composites of misaligned disk-
shaped inclusions. The permeability coefficient of the epoxy matrix was reduced to one-fourth at 5 vol %
Bz1OH loading, and the reduction was attributed to the tortuous pathway the gas molecules have to
cover during their random walk to penetrate the composite. The transmission rate of water vapor through
the composites is more influenced by the permeant-composite interactions and hence the hydrophobicity
of the monolayer covering the inclusions surface. At 5 vol % BzC16 loading, the relative vapor transmission
rate was reduced to half.
In tr od u ction
transparency and gas-barrier properties are important
criteria. Often laminates of a polyolefin, serving as
humidity barrier, and PET, serving as oxygen barrier,
are used. The two foils are usually held together by an
adhesive, which does not contribute much to the barrier
performance of the laminate. Epoxy resins are thermo-
set polymeric materials, which received special attention
as adhesives.10,11 Several efforts have been made to
enhance their permeation barrier properties through
molecular design.12,13 Another promising approach to
improve their barrier properties has become feasible by
the progress in nanocomposites. By incorporating im-
permeable transparent platelike nanoparticles in the
polymer matrix, the permeating molecules are forced
to wiggle around them in a random walk, hence diffus-
ing through a tortuous pathway.14-16 The decrease in
transmission rate of the permeant is a function of the
aspect ratio of the inclusions, their volume fraction, and
orientation. The synthesis of epoxy-clay nancomposites
has been extensively studied, and enhanced mechanical
properties were reported.17-24 However, the permeation
properties of epoxy nanocomposites have not yet been
reported. Exfoliated and intercalated morphologies have
been observed and associated with the curing rate inside
and outside the interlayer of the OM, which is often
carried out at high temperatures. The absence of reflec-
tions below 7° 2Θ in the wide-angle X-ray diffraction
(WAXRD) is usually considered as an evidence of
exfoliation although this is not necessarily true.7,25,26
The improvement in mechanical properties was ascribed
to the high aspect ratio of exfoliated silicate layers.
Polymer nanocomposites are nanoscale materials, in
which at least one of the components has a dimension
smaller than 100 nm. They offer an opportunity to
explore new behaviors and functionalities beyond that
of conventional materials. The inclusion of nanoparticles
with huge surface area and often anisotropic geometry
in a polymer matrix leads to increased particle-matrix
interactions and decreased interparticle distances with
consequent changes in morphology and performance.1
Significant improvements in permeability, thermal
stability, flame retardency, and mechanical and dielec-
tric properties have been observed at low filler volume
fraction.2-7 An important attribute of nanocomposites
is that performance enhancement is achieved without
impairing the optical homogeneity and density of the
material. Aluminosilicates and especially montmorillo-
nite, which can be exfoliated to 1 nm thick platelets,
have been frequently used to prepare polymer nano-
composites. To render the hydrophilic surface of mont-
morillonite compatible with polymers, its inorganic
cations are usually exchanged with organic ammonium
ions to give organo-montmorillonite (OM).8,9
Polymers are currently used in packaging applica-
tions, especially in food and drug packaging, where
† Department of Materials, Institute of Polymers.
‡ Department of Chemistry and Applied Biosciences, Institute
of Chemical and Bioengineering.
* Corresponding author: Ph +41 1 632 2039; Fax +41 1 632
1592; e-mail uwsuter@eth.ch.
10.1021/ma048798k CCC: $27.50 © 2004 American Chemical Society
Published on Web 08/21/2004

Macromolecules, Vol. 37, No. 19, 2004
Epoxy-Layered Silicate Nanocomposites 7251
The reaction mixture was stirred overnight at 70 °C, and the
product was filtered and washed halogen free with a hot
water-ethanol mixture (1:1) and then hot ethanol. To remove
all nonreacted or intercalated (local bilayer) molecules, the
product was suspended in a solvent, in which optimal swelling
of the OM is achieved. The suspension was sonicated for 5 min
and stirred overnight at 70 °C, then filtered, washed, and dried
at 70 °C under reduced pressure. The degree of exchange and
the purity of the product were monitored by Hi-Res TGA.9,33
If the mass loss in the TGA did not correspond to full ion
exchange, the exchange reaction was repeated. The OM was
finally suspended in 500 mL of dioxane, sonicated, and freeze-
dried. In the following, the modified clay is given the acronym
of the organic cation attached to its surface.
Th er m ogr a vim etr ic An a lysis. High-resolution (Hi-Res)
thermogravimetric analysis (TGA) of the modified fillers, in
which the heating rate is coupled to the mass loss, that is, the
sample temperature is not raised until the mass loss at a
particular temperature is completed, was performed on a Q500
thermogravimetric analyzer (TA Instruments, New Castle,
DE). All measurements were carried out under an air stream
in the temperature range 50-900 °C. Quantitative analysis
of the organic monolayer on the montmorillonite surface was
achieved by subtracting the mass loss due to physisorbed water
and to dehydroxylation of the mineral from the total mass
loss.9
The objective of the present investigation is to prepare
epoxy-OM nanocomposites and to study their oxygen
and water vapor permeation properties with the per-
spective of using them in laminate production or in
coating polyolefin foils to improve their oxygen-barrier
performance. The epoxy resin and the curing agent were
chosen to provide a polymer matrix that meets the
requirements of the food-and-health regulations and has
low gas permeability. The curing temperature has also
to be kept low in order not to deform the substrate foils.
Since the inorganic part of the OM is that which is
impermeable, the reduction in permeability is correlated
with its volume fraction. The relative oxygen perme-
ability is compared to numerical predictions, and a
macroscopic average aspect ratio of the inclusions is
deduced thereof. The study also includes an investiga-
tion of the influence of the chemical structure of the
organic monolayer ionically bonded to the clay surface
as well as that of filler loading on the barrier properties
of these nanocomposites.
Exp er im en ta l Section
Ma ter ia ls. Purified sodium bentonite (Cloisite Na+) was
purchased from Southern Clay Products (Gonzales, TX). The
epoxy resin, bisphenol A diglycidyl ether (4,4′-isopropylidene-
diphenol diglycidyl ether) with an epoxide equivalent weight
172-176, was supplied by Sigma (Buchs, Switzerland). Dio-
ctadecyldimethylammonium chloride (2C18), triethanolamine,
N-butyldiethanolamine, 2-(dibutylamino)ethanol, 1-bromooc-
tadecane, benzyl chloride, tetraethylenepentamine (TEPA),
tetrahydrofuran (THF), and dimethylformamide (DMF) were
procured from Fluka (Buchs, Switzerland). N-Benzyl-N-me-
thylethanolamine was supplied by Aldrich (Buchs, Switzer-
land), while benzyldimethylhexadecylammonium chloride
(BzC16) was purchased from Acros Organics (Basel, Switzer-
land). Polypropylene (100 µm thick) and polyamide (15 µm
thick) foils, whose surface was corona-treated to enhance their
wetting and adhesion, were kindly supplied by Alcan Packag-
ing (Neuhausen, Switzerland). A surfactant (trade name BYK-
307) was used to achieve better wetting and adherence of the
neat epoxy coating to the substrate foils and was obtained from
Christ Chemie (Reinach, Switzerland).
Syn th esis of th e Am m on iu m Sa lts. Some of the am-
monium salts used to modify the montmorillonite surface are
not commercially available. Benzyldibutyl(2-hydroxyethyl)-
ammonium chloride (Bz1OH), benzylbis(2-hydroxyethyl)buty-
lammonium chloride (Bz2OH), benzyltriethanolammonium
chloride (Bz3OH), and benzyl(2-hydroxyethyl)methyloctade-
cylammonium chloride (BzC18OH) were synthesized by quat-
ernizing the corresponding amines with alkyl halides (benzyl
chloride and 1-bromooctadecane).27-31 Generally, the alkyl
halide (0.105 mol) was slowly added to an alcoholic solution
of the amine (0.1 mol in 50 mL), and the mixture was stirred
overnight under reflux. The product was purified by extracting
it twice with ether after evaporating the ethanol. The obtained
quaternary ammonium salts were recrystallized from ac-
etone: Bz1OH, mp 137 °C; Bz2OH, mp 105 °C; Bz3OH, mp
80 °C; BzC18OH, mp 110 °C.
Na n ocom p osite P r ep a r a tion . The required amounts of
OM and epoxy resin were calculated on the basis of the desired
inorganic volume fraction as follows:
MOM ) MM + (MMCECMOC
MMVEPFEP
)
MEP
)
- (MMCECMOC)
VMFM
where MOM is the mass of the OM, MM is the mass of the
inorganic aluminosilicate, MOC is the molar mass of the
ammonium ion used to modify the montmorillonite surface,
MEP is the mass of the epoxy resin, VM is the inorganic volume
fraction, FM is the density of sodium montmorillonite (2.6
g/cm3), VEP is the epoxy volume fraction, and FEP is its density
(1.18 g/cm3).
The necessary amount of OM was added to 25 g of solvent
(THF or DMF) and allowed to swell for 2 h. The suspension
was cooled in an ice bath and sonicated (ultrasound horn) twice
at 70% amplitude for 5 min each time with 5 min pause in
between. The epoxy resin solution (5.3 g in 5 g of solvent) was
mixed with the OM suspension, and the mixture was allowed
to stand for 2 h, after which the sonication process was
repeated. The curing agent, i.e., TEPA (0.9 g), was then added
and thoroughly mixed with the epoxy-clay mixture. To achieve
a slow curing rate, favorable for exfoliation, the amine to epoxy
mole ratio was maintained at 0.3:1. The suspension was
shortly sonicated to degas it, and a film was drawn on the
corona-treated surface of PP and PA foils with the help of a
bar coater (90 µm gap). The coated films were dried at ambient
conditions for 15 min and under reduced pressure at RT for
another 15 min, then cured at 70 °C overnight, and postcured
at 90 °C for 4 h. A low curing temperature was chosen to
achieve slow curing favorable for exfoliation. In the case of
the neat epoxy, it was necessary to add 2 mg of a surfactant
(BYK-307) to improve the quality of the coating. The thickness
of the dry coated film was ca. 10 µm, and its correct thickness
was determined as described below.
Su r fa ce Tr ea tm en t of Mon tm or illon ite. The cation
exchange capacity (CEC) of Cloisite Na+ was determined by
exchanging its sodium ions with Cu(trien)2+ to be 0.88 mequiv/
g.32,33 The aluminosilicate surface was rendered organophilic
by exchanging its inorganic cations with organic ammonium
ions, viz., Bz1OH, Bz2OH, Bz3OH, BzC18OH, 2C18, and
BzC16. An 8 g sample of Cloisite Na+ was stirred in 500 mL
of deionized water at 70 °C for 2 h, and then 200 mL of ethanol
was added. The aluminosilicate was dispersed by sonication
for 10 min (ultrasonic horn at 70% amplitude) followed by
shear mixing for 10 min (Ultra-Turax T50, IKA, Staufen,
Germany). To this dispersion, a solution of the desired am-
monium salt corresponding to 150% of the clay’s CEC in 100
mL of ethanol was added dropwise (within 2 h) under stirring.
Den sity a n d Th ick n ess. The density of the substrate foils
(PP and PA) and the neat epoxy resin was determined by
weighing samples in air and in ethanol using an analytical
balance (Mettler AE 200) and a homemade device similar to
the Mettler density kit ME-33360 following the equation
FEtM
M - Ml
F )
where F is the sample density, FEt is the density of ethanol at

7252 Osman et al.
Macromolecules, Vol. 37, No. 19, 2004
22 °C, M is the sample mass in air, and Ml is its mass in
ethanol. The thickness of the substrate foils was calculated
from the weight of a piece of defined area and its density. The
average of five measurements was taken.
The thickness of the coated film (tf) was similarly deter-
mined as follows:
tf ) (M1 - M2)/(FcA)
where M1 is the mass of the coated foil, M2 is the mass of an
uncoated foil with the same area, Fc is the density of the
composite, and A is the area of the foil. The density of the
composite was calculated as
1
Fc )
mOM mEP
+
FOM
FEP
where mOM is the mass fraction of the OM, FOM is its density,
mEP is the mass fraction of the cured epoxy resin, and FEP is
its density. The density of the OM was analogously calculated
from that of the organic monolayer (assumed to be equivalent
to that of octadecanol ) 0.81 g/cm3) and that of sodium
montmorillonite. The organic and inorganic mass fractions in
the OM were calculated from the CEC and the molar mass of
the organic cation.
Ga s P er m ea tion . The oxygen transmission rate through
PP foils coated with the neat epoxy and the nanocomposites
was measured using an OX-TRAN 2/20 ML (Mocon, Min-
neapolis, MN) at 23 °C and 0% RH. The water vapor transmis-
sion rate through PA foils coated with neat epoxy or nano-
composites was measured using a PERMATRAN-W 3/31 MG
(Mocon, Minneapolis, MN) at 23 °C and 100% RH. The PP and
PA substrates were selected on the basis of their high
transmission rate for the specific permeant so that they do
not hinder measuring the permeation through the coated film.
The total transmission rate (Ttot) of oxygen or water vapor was
normalized with respect to the total thickness (ttot), and the
transmission rate through the neat epoxy or nanocomposite
films was calculated as follows:
F igu r e 1. Chemical structure of the ammonium ions used to
organophylize the montmorillonite surface as well as that of
the epoxy resin and its curing agent.
matrix, its inorganic cations were exchanged with
organic ammonium ions of different chemical structures.
Dioctadecylammonium (I), benzylhexadecylammonium
(II), and benzylhydroxyethyloctadecylammonium (III)
as well as mono- (IV), di- (V), and triethylhydroxyam-
monium (VI) moieties were chosen for this purpose
(Figure 1). Long alkyl chains increase the d-spacing,
thus facilitating the penetration of guest molecules in
the interlayer and consequently the intercalation and
exfoliation. The benzyl group is expected to lead to
stronger van der Waals attraction forces between the
filler and the epoxy resin that enhances the intercala-
tion and exfoliation of the clay particles. The hydroxy-
ethyl groups were chosen to match the polarity of the
EP matrix and because it may react with the epoxy
groups of the resin, thus tethering polymer chains to
the silicate layers. Tethered polymer molecules expand
the interlayer and cannot be squeezed out if the silicate
layers collapse, hence enhancing intercalation or exfo-
liation of the OM. The formation of local bilayer during
surface treatment of the clay was avoided because the
presence of unreacted molecules intercalated in the
organic monolayer of the OM in a tail-to-tail arrange-
ment, i.e., with their polar headgroups pointing out-
ward, is expected to influence the surface energy and
reduce the thermal stability of the OM.9 Because of the
high CEC of montmorillonite and the large molecular
weight of the organic cations, the density of the OM is
significantly lower than that of sodium montmorillonite
and should be taken into consideration on calculating
the volume fraction. Table 1 gives the densities of the
different OMs prepared and their weight fraction that
correspond to 0.035 inorganic volume fraction as well
as the densities of the nanocomposites.
ttot/Ttot ) t1/T1 + t2/T2
where ti are the thicknesses of the coated film and the
substrate foil and Ti are the corresponding normalized trans-
mission rates. The average of four measurements for each
coating is reported.
Wid e-An gle X-r a y Diffr a ction (WAXRD). Wide-angle
X-ray diffraction patterns were collected on a Scintag XDS
2000 diffractometer (Scintag Inc., Cupertino, CA) using Cu KR
radiation (λ ) 0.154 06 nm) in reflection mode. The instrument
was equipped with a graphite monochromator and an intrinsic
germanium solid-state detector. The coated foils and the
suspensions were step-scanned (step width 0.02° 2Θ, scanning
rate 0.06°/min) at room temperature from 1.5° to 10° 2Θ. The
OMs were step scanned at a higher rate (0.24°/min) in the
range 1.5°-30° 2Θ. The (001) basal-plane reflection of an
internal standard muscovite (very thin platelets, 2Θ ) 8.84°)
was used to calibrate the line position of the OM reflections.
To determine the peak positions, the diffractograms were fitted
with a split Pearson VII function (diffraction management
system software 1.36b).
Tr a n sm ission Electr on Micr oscop y (TEM). The micro-
structure of the nanocomposites was studied by bright-field
TEM using a Zeiss EM 912 Omega microscope. Pieces of the
coated foils were etched with oxygen plasma for 3 min and
embedded in an epoxy matrix (Epon 812 + Durcopan ACM
3:4, Fluka, Buchs, Switzerland). 50-100 nm thick sections
were microtomed with a diamond knife (Reichert J ung Ul-
tracut E). The sections were supported by 100 mesh grids
sputter-coated with a 3 nm thick carbon layer.
Depending on the length of the alkyl chains, the cross-
sectional area of the organic cations, and the available
area per cation on the clay surface, the d-spacing of the
layered silicate is expanded to different extents.33
A
Resu lts a n d Discu ssion
large basal-plane spacing facilitates polymer intercala-
tion and reduces the attraction forces between the
silicate layers, leading to exfoliation of the OM. Table
To reduce the surface energy and hydrophilicity of the
clay, hence rendering it compatible with the epoxy

Macromolecules, Vol. 37, No. 19, 2004
Epoxy-Layered Silicate Nanocomposites 7253
Ta ble 1. Den sity of th e Or ga n o-Mon tm or illon ites a n d
Th eir 3.5 vol % Com p osites
composites compared to the OM powder (0.12 and 0.46
nm, respectively), indicating polymer intercalation. Cur-
ing the BzC16, BzC18OH, and 2C18 composites led to
further reduction in d-spacing, but it remained ap-
preciable (2.9-3.5 nm). In comparison to the OM
powder, there was an increase in d-spacing of 1.06, 1.39,
and 0.95 nm, respectively, suggesting sizable intercala-
tion. In contrast to Bz1OH, increasing the volume
fraction of BzC16 from 0.01 to 0.05 led to gradual
decrease in d-spacing from 3.07 to 2.90 nm, which
suggests squeezing out of the intercalated polymer. This
may support the notion that the epoxy resin reacts with
the OH groups of BZ1OH. Unfortunately, no spectro-
scopic (IR) evidence could be obtained for this reaction.
These results show that all OMs carrying long alkyl
chains (BzC16, BzC18OH, 2C18) were strongly swollen
by the solvent and intercalated by the polymer at the
end, leading to an increase in d-spacing of 1 nm or more.
The OMs with short alkyl chains (Bz1OH, Bz2OH,
Bz3OH) were less swollen by the solvent and less
intercalated by the epoxy polymer.
Since WAXRD does not give any information on the
exfoliated part of the clay, the microstructure of the
composites was investigated by TEM. Figure 2 shows
TEM micrographs of the 3.5 vol % Bz1OH-EP composite.
It can be seen that most of the OM particles were
exfoliated to single layers (dark lines) or thin tactoids
of very few layers that are flexible (bent) and mis-
aligned. A similar observation has been made in PU
nanocomposites.7 It seems that it is not possible to align
the montmorillonite platelets with the coating technique
used. The composite of Bz2OH showed a similar mor-
phology, but Bz3OH had more intercalated tactoids. The
micrographs in Figure 3 represent the microstructure
of the 3.5 vol % BzC16-EP composite, which consisted
mainly of intercalated stacks. Some of these tactoids
were thinner than the others, but all of them were
misaligned and the d-spacing was ca. 3 nm in ac-
cordance with the WAXRD results. Figure 4 shows the
microstructure of the 3.5 vol % composites of BzC18OH
and 2C18, in which only very few single layers were
observed. They practically consisted of intercalated
tactoids, whose d-spacing corresponded to that mea-
sured by WAXRD (Table 2).
modified filler
filler weight composite density
OM
density [g/cm3] fraction [%]
[g/cm3]
Bz1OH
Bz2OH
Bz3OH
BzC18OH
BzC16
2C18
1.93
2.01
2.13
1.63
1.69
1.53
8.96
8.89
8.82
9.85
9.51
10.60
1.22
1.23
1.23
1.21
1.21
1.21
2 shows that the d-spacing of the OM was modestly
influenced by the short alkyl chains of Bz1OH, Bz2OH,
and Bz3OH but was strongly expanded by the long alkyl
chains of BzC16, BzC18OH, and 2C18. The preparation
of the composites included swelling of the OM in an
organic solvent and sonication to disintegrate the clay
particles into its primary layers. X-ray measurements
of these suspensions (Table 2) showed that all clays
including the sodium montmorillonite swelled in DMF
to different extents, leading to an increase in d-spacing,
whereby the ones carrying long alkyl chains swelled
more. Only the diffractogram of the BzC16 suspension
was “silent”, indicating that the clay was exfoliated to
such an extent that no thick enough stacks (tactoids)
in detectable concentration were present. The degree
of swelling seems to depend on the clay solvatization,
i.e., on the solubility parameters of the monolayer and
the solvent.34,35 Addition of the epoxy resin to the OM
suspension did not affect the d-spacing of Bz1OH and
Bz2OH but decreased that of Bz3OH by 0.16 nm,
indicating a better match between the epoxy resin and
Bz1OH or Bz2OH. The d-spacing of BzC18OH and 2C18
was decreased by 0.23 and 0.15 nm, respectively, on
adding the EP resin, suggesting that the difference in
polarity of the organic monolayer and the epoxy resin
led to partial deswelling of the clay. For the same
reason, the exfoliated BzC16 layers collapsed in the
presence of the epoxy resin, leading to the appearance
of a reflection at 2.49° 2Θ. This confirms that the
interplay of entropic and energetic factors governs the
exfoliation and intercalation process.36,37 Driving the
solvent off and curing the epoxy resin led to almost
complete deswelling of the sodium montmorillonite but
did not affect the d-spacing of Bz1OH much, pointing
out the incompatibility of EP with the inorganic silicate
and the compatibility with Bz1OH. Increasing the
Bz1OH filler concentration had no influence on the
d-spacing, indicating that the intercalated polymer was
not squeezed out on increasing the filler volume fraction.
In the case of Bz2OH and Bz3OH, evaporating the
solvent and curing the resin led to partial deswelling
(0.28 nm), but there was an increase in d-spacing of the
The oxygen permeability coefficients of the 3.5 vol %
composites were measured (Table 3), and the relative
permeability was plotted against the d-spacing in Figure
5. In contrast to all expectations, the relative perme-
ability tendentially increased with increasing d-spacing,
but no correlation between the two parameters could
be found. This suggests that the intercalated part of the
filler does not contribute much to the permeability
Ta ble 2. Ba sa l-P la n e Sp a cin g of th e F iller s, Th eir Su sp en sion s, a n d EP Com p osites
d-spacing of filler
powder [nm]
d-spacing of filler
suspended in DMF [nm]
d-spacing of filler suspended
in DMF + epoxy [nm]
d-spacing of EP
composite [nm]
filler
Cloisite Na+ (3.5 vol %)
Bz1OH (1 vol %)
Bz1OH (2 vol %)
Bz1OH (3.5 vol %)
Bz1OH (5 vol %)
Bz2OH (3.5 vol %)
Bz3OH (3.5 vol %)
BzC18OH (3.5 vol %)
BzC16 (1 vol %)
1.25
1.52
1.98
1.92
1.99
1.35
1.89
1.88
1.87
1.87
1.62
1.98
3.45
3.07
2.99
2.93
2.90
3.46
1.90
1.50
1.52
2.06
1.87
1.88
2.41
4.12
1.90
2.25
3.89
BzC16 (2 vol %)
BzC16 (3.5 vol %)
BzC16 (5 vol %)
no reflection
3.69
3.55
3.54
2C18 (3.5 vol %)
2.51

7254 Osman et al.
Macromolecules, Vol. 37, No. 19, 2004
F igu r e 3. TEM micrographs of the 3.5 vol % BzC16-EP
nanocomposite. The dark lines are cross sections of alumino-
silicate layers or tactoids.
F igu r e 2. TEM micrographs of the 3.5 vol % Bz1OH-EP
nanocomposite. The dark lines are cross sections of alumino-
silicate layers or tactoids.
reduction and that the exfoliated layers are those which
build the permeation barrier. This finding can be
rationalized by realizing that the aspect ratio (longest
axis/shortest axis) of the intercalated tactoids is not
higher than that of the pristine OM, while that of the
single layers can be as high as 300-400. It also confirms
that the origin of the barrier performance of the inclu-
sions in nanocomposites lies in the tortuous pathway
the gas molecules have to go through. The permeability
coefficient of the polymer matrix used is 60 times lower
than that of the PP substrate but double that of the PET
foils commonly used as an oxygen barrier. However, it
is much lower than those of most adhesives commonly
used in laminates, which do not contribute much to the
barrier performance. The permeability coefficient of the
BzC16 composite is the same as that of the sodium
montmorillonite, which is 20% less than that of the neat
polymer, while those of the BzC18OH and 2C18 com-
posites are higher than that of the matrix. It seems that
there is a phase separation between the long alkyl
chains and the epoxy matrix, which leads to a decrease
in density at the interface, resulting in higher “free
volume” and consequently higher transmission rate. The
same observation was made in 2C18-PU nanocompos-
ites.7 The presence of a hydroxyl group in BzC18OH did
not help tethering polymer chains to the silicate surface.
Probably, the long alkyl chain masks the OH group,
hence hindering the reaction with the epoxy resin.
Despite the low permeability of the polymer, Bz1OH and
Bz2OH reduced the permeability coefficient two and
half times, leading to a better oxygen barrier than PET.
Bz3OH was less effective but still reduced the perme-
ability coefficient of the polymer to half. These results
are in accordance with the microstructure elicited by
TEM and WAXRD, suggesting a reaction between the
OH and epoxy groups or a good match between the
surface energies of the two phases, leading to exfoliation
of the OM to single layers. It seems that increasing the
number of OH groups beyond a certain limit can be
counterproductive due to increased polarity and hydro-
gen bonding. Interplay of entropic and energetic factors
governs the exfoliation and intercalation process and
is the key to designing the nanocomposite microstruc-
ture. The water vapor transmission rate through the
polymer increased massively (230%) in the sodium
montmorillonite composite but decreased by ca. 40% in
the OM composites (Table 3). This reflects the difference
in hydrophobicity of the fillers. The variety of chemical
structure of the monolayer tested had little effect on the
vapor transmission rate, and the values varied within
a narrow range.

Macromolecules, Vol. 37, No. 19, 2004
Epoxy-Layered Silicate Nanocomposites 7255
F igu r e 5. Oxygen relative permeability of the composites as
a function of the d-spacing.
Ta ble 4. Dep en d en ce of th e Oxygen (0% RH) a n d Wa ter
Va p or Tr a n sm ission Ra tes on F iller Loa d in g a t 23 °Ca
permeability coeff
(oxygen) [cm3 µm/
(m2 d mmHg)]
water vapor
transmission rate
composite
neat epoxy
[g µm/(m2 d mmHg)]
2.0
10
Bz1OH (1 vol %)
Bz1OH (2 vol %)
Bz1OH (3.5 vol %)
Bz1OH (5 vol %)
BzC16 (1 vol %)
BzC16 (2 vol %)
BzC16 (3.5 vol %)
BzC16 (5 vol %)
1.4
8.9
8.0
6.7
5.7
7.9
6.5
5.3
4.9
0.91
0.77
0.55
1.9
1.7
1.6
1.6
a
Relative probable error 5%.
F igu r e 4. TEM micrographs of 3.5 vol % (a) BzC18OH-EP
and (b) 2C18-EP nanocomposites. The dark lines are cross
sections of aluminosilicate layers.
Ta ble 3. Oxygen (0% RH) a n d Wa ter Va p or Tr a n sm ission
Ra tes th r ou gh 3.5 vol % OM-Ep oxy Na n ocom p osites a t
23 °Ca
permeability coefficient
(oxygen)
water vapor
transmission rate
composite
[cm3 µm/(m2 d mmHg)]
[g µm/(m2 d mmHg)]
neat epoxy
Cloisite Na+
Bz1OH
Bz2OH
Bz3OH
BzC18OH
BzC16
2C18
2.0
1.6
0.77
0.78
1.0
2.2
1.6
3.7
10
23
F igu r e 6. Oxygen relative permeability of Bz1OH- and
BzC16-EP nanocomposites as a function of the inorganic
volume fraction. The solid lines are numerical predictions for
disk-shaped inclusions with aspect ratios 50, 100, and 150
(diameter/thickness), while the dotted line is simply a guide
for the eye.
6.7
5.8
7.1
5.7
5.3
6.8
a
ing. Theoretical models have shown that the relative
permeability is a function of the inclusions aspect ratio,
volume fraction, and orientation.14-16 However, these
models assume perfect parallel orientation, which is not
the case here, and it is difficult to determine the aspect
ratio of montmorillonite platelets in nanocomposites,
especially in cases where mixed morphology is present
and the silicate layers are bent or folded.38 In such cases,
a macroscopic average of the aspect ratio can better
describe the geometry of the particles. Recently, the
relative permeability of composites of misaligned disks
has been numerically calculated as a function of the
inclusions aspect ratio (diameter/thickness).39,40 This
Relative probable error 5%.
The effect of filler loading on the oxygen permeability
coefficient and water vapor transmission rate of Bz1OH-
and BzC16-EP nanocomposites, which represent a
reactive and a nonreactive OM, respectively, is shown
in Table 4. In both cases, the oxygen permeability
coefficient decreased with increasing filler loading, but
the reduction in the Bz1OH nanocomposites was more
impressive, viz., factor of 4 at 5 vol % loading. The
relative oxygen permeability is plotted in Figure 6 as a
function of the inorganic volume fraction, showing that
it decreases asymptotically with increasing filler load-

7256 Osman et al.
Macromolecules, Vol. 37, No. 19, 2004
predict precisely. Tethering polymer molecules to the
silicate layers enhances the exfoliation because these
molecules cannot be squeezed out upon collapse of the
layers. The aspect ratio of montmorillonite platelets is
difficult to determine in polymer nanocomposites be-
cause of their flexibility which leads to bending and
folding, aggravating the determination of their lateral
dimension. Besides, a mixed morphology (exfoliated and
intercalated structures) is often present, and the particle
size of the inclusions is polydisperse, leading to a broad
distribution of the aspect ratio. In such cases, a mac-
roscopic volume average is well-suited to describe the
anisometry of the particles. The aspect ratio can be
deduced from the relative permeability of the nanocom-
posites to gas molecules by comparing the measured
values to numerical predictions of the permeability in
composites of misaligned disk-shaped inclusions. The
incorporation of a small volume fraction of platelike
nanoparticles in the epoxy matrix decreases its perme-
ability coefficient to one-fourth. This can be attributed
to the impermeability of the inorganic nanoparticles and
the tortuous pathway the gas molecules have to cover
during their random molecular motion in the composite.
The transmission rate of water vapor through the
composites is more influenced by the permeant-
composite interactions and hence the hydrophobicity of
the inclusions.
F igu r e 7. Water vapor relative transmission rate through
Bz1OH- and BzC16-EP nanocomposites as a function of the
inorganic volume fraction. The dotted line is simply a guide
for the eye.
approach has been used here to estimate a macroscopic
average of the aspect ratio of the inclusions as a function
of filler loading. The solid lines in Figure 6 represent
the numerical predictions for inclusions of aspect ratios
50, 100, and 150 at filler loading up to 5 vol %. For an
aspect ratio of 150, it was not possible to obtain models
of statistically misaligned disks with nonoverlapping
configurations at loadings higher than 3 vol % as
required by this approach. From Figure 6, it seems that
the inclusions in the nonreactive OM composite (BzC16)
have an aspect ratio of 50 but decreases at filler loadings
higher than 3 vol %, probably due the dramatic decrease
in interparticle distance and the consequent agglomera-
tion above this concentration.39,41 The reactive OM
(Bz1OH) has much higher aspect ratio (250-300),
corresponding to a very high degree of exfoliation in
accordance with the TEM micrographs. It should be
noted that the estimated aspect ratios are volume
averages, which is usually higher than number aver-
ages. The relative transmission rate of water vapor
through the BzC16 and Bz1OH composites is plotted
in Figure 7 as a function of filler loading. The transmis-
sion rate decreases in both cases with increasing
inorganic volume fraction, but the performance of
BZC16 (factor of 2 at 5 vol %) is better than that of
Bz1OH (40% at 5 vol %). This is in contrast to the
oxygen permeability coefficient and reflects the hydro-
phobicity of the monolayer of the OMs, indicating that
the permeant-composite interactions play an important
role here. It is noteworthy to remark that all nanocom-
posites retained the optical properties of the polymer
matrix.
Ack n ow led gm en t. We gratefully acknowledge fi-
nancial support from the Swiss National Science Foun-
dation (SNF). We also thank NESTEC, Lausanne, and
ALCAN PACKAGING, Neuhausen, for supporting the
project.
Refer en ces a n d Notes
(1) Bre´chet, Y.; Cavaille´, J . Y.; Chabert, E.; Chazeau, L.; Dendi-
evel, R.; Flandin, L.; Gauthier, C. Adv. Eng. Mater. 2001, 3,
571.
(2) Yano, K.; Usuki, A.; Okada, A.; Kurauchi, T.; Kamigito, O.
J . Polym. Sci., Part A: Polym. Chem. 1993, 31, 2493.
(3) Giannelis, E. P. Adv. Mater. 1996, 8, 29.
(4) Gilman, J . W.; Kashiwagi, T.; Lichtenhan, J . D. SAMPE J .
1997, 33, 40.
(5) LeBaron, P. C.; Wang, Z.; Pinnavaia, T. J . Appl. Clay Sci.
1999, 15, 11.
(6) Alexandre, M.; Dubois, Ph. Mater. Sci. Eng. R 2000, 28, 1.
(7) Osman, M. A.; Mittal, V.; Morbidelli, M.; Suter, U. W.
Macromolecules 2003, 36, 9851.
(8) J asmund, K., Lagaly, G.; Eds. Tonminerale und Tone;
Steinkopff-Verlag: Darmstadt, 1993.
(9) Osman, M. A.; Plo¨tze, M.; Suter, U. W. J . Mater. Chem. 2003,
13, 2359.
(10) Lee, H.; Neville, K. Handbook of Epoxy Resins; McGraw-
Hill: New York, 1967.
(11) Ellis, B. Chemistry and Ttechnology of Epoxy Resins; Blackie
Academic & Professional: London, 1993.
(12) Sivis, H. C. Trends Polym. Sci. 1997, 5, 75.
(13) Brennan, D. J .; Haag, A. P.; White, J . E.; Brown, C. N.
Macromolecules 1998, 31, 2622.
Con clu sion s
Long alkyl chains increase the basal-plane spacing
of organo-montmorillonites and enhance their swelling
and intercalation. However, they do not lead to the
desired improvement in composite properties unless the
exfoliation is enhanced because the aspect ratio of the
intercalated tactoids is not high. Alkyl chains seem to
be incompatible with EP, which is aromatic to a large
extent; therefore, they lead to an increase in d-spacing
and in intercalation but not to exfoliation. Delaminated
silicate layers have high aspect ratios and contribute
most to the properties of nanocomposites. The exfolia-
tion and intercalation process is governed by interplay
of entropic and energetic factors, which is difficult to
(14) Eitzman, D. M.; Melkote, R. R.; Cussler, E. L. AIChE J . 1996,
42, 2.
(15) Fredrickson, G. H.; Bicerano, J . J . Chem. Phys. 1999, 110,
2181.
(16) Gusev, A. A.; Lusti, H. R. Adv. Mater. 2001, 13, 1641.
(17) Messersmith, P. B.; Giannelis, E. P. Chem. Mater. 1994, 6,
1719.
(18) Lan, T.; Kaviratna, P. D.; Pinnavaia, T. J . Chem. Mater. 1995,
7, 2144.
(19) Zilg, C.; Mu¨lhaupt, R.; Finter, J . Macromol. Chem. Phys.
1999, 200, 661.
(20) Brown, J . M.; Curliss, D.; Vaia, R. A. Chem. Mater. 2000,
12, 3376.
(21) Zerda, A. S.; Lesser, A. J . J . Polym. Sci., Part B: Polym. Phys.
2001, 39, 1137.

Macromolecules, Vol. 37, No. 19, 2004
Epoxy-Layered Silicate Nanocomposites 7257
(22) Kornmann, X.; Lindberg, H.; Berglund, L. A. Polymer 2001,
42, 1303.
(32) Osman, M. A.; Suter, U. W. J . Colloid Interface Sci. 2000,
224, 112.
(33) Osman, M. A.; Plo¨tze, M.; Skrabal, P. J . Phys. Chem. B 2004,
(23) Kornmann, X.; Thomann, R.; Mu¨lhaupt, R.; Finter, J .;
Berglund, L. J . Appl. Polym. Sci. 2002, 86, 2643.
(24) Kong, D.; Park, C. E. Chem. Mater. 2003;15, 419.
(25) Vaia, R. A.; Liu, W. J . Polym. Sci., Part B 2002, 40, 1590.
(26) Morgan, A. B.; Gilman, J . W. J . Appl. Polym. Sci. 2003, 87,
1329.
108, 2580.
(34) Balazs, A. C.; Singh, C.; Zhulina, E.; Lyatskaya, Y. Acc. Chem.
Res. 1999, 32, 651.
(35) Ginzburg, V. V.; Singh, C.; Balazs, A. C. Macromolecules
2000, 33, 1089.
(27) Shimauchi, S.; Minemura, I.; Matsui, K.; Shirasawa, K. J pn.
Tokkyo Koho 1973, J P 48006664; CAN 80:49134.
(28) Smirnova, A. T.; Kabanov, V. P.; Kharkov, S. N.; Ivanova, L.
A. J . Appl. Chem. USSR 1987, 60, 534.
(29) Hayashi, Y.; Fujiwara, T.; Shimizu, T.; Nagano, Y.; Teramura,
K. Yukagaku 1987, 36, 409; CAN 108:74771.
(30) Al-Sabbagh, A. M.; Osman, M. M.; Omar, A. M. A.; El-Gamal,
I. M. Anti-Corros. Method Mater. 1996, 43, 11.
(31) Li, Z.; Fu, L.; Chen, Z. Huagong Daxue Xuebao 1999, 26, 61,
CAN 132:266475.
(36) Vaia, R. A.; Giannelis, P. Macromolecules 1997, 30, 7990.
(37) Vaia, R. A.; Giannelis, P. Macromolecules 1997, 30, 8000.
(38) Fornes, T. D.; Paul, D. R. Polymer 2003, 44, 4993.
(39) Osman, M. A.; Mittal, V.; Lusti, H. R. Macromol. Rapid
Commun. 2004, 25, 1145.
(40) Lusti, H. R.; Gusev, A. A.; Guseva, O. Adv. Mater., submitted.
(41) Vaia, R. A. In Polymer-Clay Nanocomposites; Wiley: Chich-
ester, 2000; p 229.
MA048798K