You are on page 1of 18

Water Sorption and Diffusion Studies in an

Epoxy Resin System

C. MAGGANA,1 P. PISSIS2
1
National Technical University of Athens, Department of Theoretical and Applied Mechanics, Zografou Campus,
GR 15780 Athens, Greece

2
National Technical University of Athens, Department of Physics, Zografou Campus, GR 15780 Athens, Greece

Received 17 August 1997; revised 6 November 1998; accepted 9 November 1998

ABSTRACT: The kinetics of water absorption in epoxide materials was studied by the
aid of a diglycidyl ether of bisphenol-A–triethylenetetramine (DGEBA–TETA) epoxy–
resin system containing various amounts of the plasticizer THIOCOL (0 – 40 phr). The
presence of plasticizer permits the formation of products with different crosslinking
densities and hydrophilic characters. Dynamic water absorption experiments were
carried out at 15, 40, and 70°C. For the fitting of the experimental results, a new model
was used, based on a model proposed earlier by Jacobs and Jones. This model considers
epoxide product as a two-phase system consisting of a master phase (where the major
part of the water is absorbed), which is homogeneous and nonpolar (phase 1), and of a
second phase with different density and/or hydrophilic character (phase 2). By making
the assumption that water diffusion can take place independently in the different
phases of the material in accordance with Fick’s second law, we can calculate the
diffusion coefficient D and the water content at saturation M ` for each phase sepa-
rately. Equilibrium water sorption measurements were performed at 40°C, and the
data were analyzed and discussed based on the Guggenheim–Anderson– de Boer (GAB)
equation, the results being in support of the two-phase model used in the analysis of
absorption kinetics. The linear expansion coefficient and the glass transition temper-
ature of the materials, employed in the discussion of the results, were measured by
thermomechanical analysis. © 1999 John Wiley & Sons, Inc. J Polym Sci B: Polym Phys 37:
1165–1182, 1999
Keywords: epoxy resins; Fickian diffusion; water sorption; water–polymer interac-
tion; Guggenheim–Anderson– de Boer equation

INTRODUCTION ters that strongly influence water sorption mech-


anisms, leading to deviations from the random-
Polymer-based materials are often exposed to hu- mixing Flory–Huggins theory and from Henry’s
mid environments, where water molecules, as Law.2– 4 In many cases, the second Fick’s Law5,6
well as other low-molecular-weight substances, fails to describe successfully the kinetics of water
can easily migrate in the polymeric matrix, even diffusion, which should be analyzed by consider-
at ambient temperatures, and modify their phys- ing various mechanisms of sorption that can be
ical properties.1 The chemical composition and activated.2,7–11
the microstructure of polymers are two parame- Epoxy resins are widely used as matrices for
structural composite materials, adhesives, and
organic coatings due to their good mechanical
Correspondence to: P. Pissis
Journal of Polymer Science: Part B: Polymer Physics, Vol. 37, 1165–1182 (1999)
properties.12–15 However, it is well known that
© 1999 John Wiley & Sons, Inc. CCC 0887-6266/99/111165-18 these properties are strongly affected by moisture
1165
1166 MAGGANA AND PISSIS

absorption, which causes plasticization,16,17 low- In accordance with Gupta and Drzal, the two–
ering the glass transition temperature,16 –18 and modes sorption involves a Langmuirian mode at
diminishes their mechanical strength.18,19 So the early stages of absorption and at temperatures
transport of moisture in epoxy resin systems is of far below T g since little energy is needed for water
great importance, and the kinetics of absorption molecules to enter into preexisting gaps so that
and distribution of water in the polymeric matrix small differences in free volume can make a con-
is still under investigation. siderable difference to the amount of moisture
The kinetics of water absorption in epoxy res- absorbed. At higher temperatures, Henry’s law
ins is usually assumed to be a concentration-in- mode is more probable as the water molecules
dependent Fickian diffusion process.20 –27 This enter in the material through gaps created by
model, which is usually applicable to the initial segmental motion, a mechanism which requires
stages of the diffusion process, fails to describe more energy.35
the whole phenomenon.28,29 Some authors at- Isothermal equilibrium sorption curves deviate
tempt to explain this by means of a Langmuir- from Henry’s Law. The experimental data of wa-
type diffusion process,30 in which the absorbed ter content versus relative humidity yield para-
material (water) can be divided into mobile and bolic or sigmoidal curves, which are better de-
strongly bound types.2,30 –32 Other authors accept scribed by power law functions.18
the existence of a two-phase structure epoxy ma- Moy and Karasz report that low-temperature
terial composed of areas with different densi- sorption of water by epoxy resins is a multilayer
ties.33,34 Changes in the structure of the material process,16 usually represented by sigmoidal iso-
during the moisture exposure, such as plasticiza- thermal curves of type II in the Brunauer–Em-
tion and formation of cavities, should also not be met–Teller (BET) classification.42 This indicates
that, initially, sorbed water molecules are able to
neglected.27,35
interact with binding sites of the polymer, while
Complex expressions, which take into account
subsequent molecules associate with the primary
parameters, such as the formation of hydrogen
layer in liquid waterlike structures. At more ele-
bonds,2,28,31,32 the heterogeneity of the epoxy
vated temperatures, isotherms approach linear-
resin system,33,34 changes in free volume,36 relax-
ity, revealing simple solution sorption, which is
ation phenomena during hydration,10,37 degrada-
asymptotically approximated by Henry’s Law.16
tion of epoxide material,38 the depression of T g , 11
The negative and positive deviations of the ex-
and the formation of clusters, crazes, and micro-
perimental results from Henry’s Law of isother-
cracks during hydration, have been proposed by mal equilibrium water absorption have been de-
many authors.29,33 scribed by Galin and Galin,43,44 in other poly-
The physical aspects of the transport process meric systems by use of the Guggenheim–
are also of great importance. Adamson39 has pos- Anderson– de Boer (GAB) isothermal absorption
tulated that the transport of moisture below T g is model.45
a three-stage process in which the absorbed mois- Infrared (IR) and nuclear magnetic resonance
ture first occupies the free volume. In the second (NMR) spectroscopy also confirm the formation of
stage, water becomes bound to network sites, hydrogen bonds between water and the polar
causing swelling. Finally, water enters the groups of the resin. Antoon and Koenig demon-
densely crosslinked regions. strated by Fourier transform infrared (FTIR)
Yang et al. suggest that while the driving force spectroscopy that although some of the changes in
for the absorption of water is the electrostatic the spectrum of wet epoxy resin specimens may
attraction between water and the functional be attributed to solvent-induced alteration of
groups of the resin (especially hydroxyls produced the electronic environment or conformational
during curing reactions or amine groups when changes and bond distortion due swelling, others
there is an excess of the curing agent), the equi- imply that water molecules interact preferen-
librium degree of water uptake is determined pri- tially with polar species such as hydroxyl-ether
marily by unoccupied volume of the epoxy resin.40 and phenyl-ether units, even with phenyl rings,
Carfagna et al. also indicate that water sorp- while saturated hydrocarbon units remain prac-
tion in polymers containing potential hydrogen- tically unaffected.46
bonding sites is driven by strong interactions be- In this work, we investigate the characteristics
tween water molecules and particular segments of isothermal equilibrium and dynamic water ab-
or groups present in the polymer.41 sorption of a model epoxy resin system with dif-
WATER SORPTION IN AN EPOXY RESIN SYSTEM 1167

ferent amounts of plasticizer. By changing the tion, as relaxation phenomena and changes in the
percentage of plasticizer, we can modify the flex- free volume influence the diffusion coefficient and
ibility, length, and packing density of macromo- the water content at saturation.48
lecular chains between crosslinks and the distri-
bution of hydroxyl groups (formed during the cur-
ing reaction) in the final product. This procedure EXPERIMENTAL
permits a better approach to the factors that in-
fluence the water absorption and gives explana-
Materials and Reactions
tions about the way water penetrates and diffuses
into polymer networks. The investigation of the The system studied was based on diglycidyl ether
hydration procedure can also give useful informa- of bisphenol-A (DGEBA) epoxy resin (Epon 828,
tion about the microstructure of the polymeric Shell), with a number-average molecular weight
material.33,34 M ne ' 370, cured with 8% per weight of the resin
The experimental results of dynamic water ab- triethylenetetramine (TETA), (Dow Chemical).
sorption exhibit deviations from Fick’s law, espe- The plasticizer used was a polysulfide with the
cially at low temperatures and low plasticizer commercial name THIOCOL LP3 (Thiokol Chem.
content. To describe this behavior, a new ap- Corp., Switzerland), 1000 molecular weight, and
proach is used, which is based on the Fickian 2% crosslinking. The proportion of plasticizer was
mode of absorption but also takes into account the 0, 10, 20, 30, 40, 50, and 60% per weight of the
formation of hydrogen bonds and/or the heteroge- resin.49 The chemical structure of the materials
neous structure of the materials, as firstly pro- used is as follows.
posed by Jacobs and Jones. Based on the exis-
tence of bound water forming hydrogen bonds
with hydrophilic sites in epoxy resin sys-
tems26,27,46 and on the heterogeneous structure of
these systems,27,47 it is expected that Jacobs and
Jones (two-phase) type models will successfully
describe dynamic water absorption in our sys-
tem.31,33,34 From the fitting parameters of the
aforementioned model, we can calculate the dif-
fusion coefficient D and the water content at sat-
uration M ` for each phase (or each mode of ab-
sorption) separately. This permits the extraction
of useful results about the way water penetrates
the epoxide network and how the temperature
and the presence of plasticizer modify the struc-
ture of the material.
The negative deviations of the experimental
results from Henry’s Law at the first stages of
isothermal equilibrium water absorption (espe- The aminohydrogen to epoxy ratio r is calculated
cially for specimens with small amount of plasti- from the following equation:
cizer) are attributed to the Langmuirian mode of

S D S D
absorption due to the formation of hydrogen
6m a 2m e
bonds. The amount of bonded water and the po- r5 4 , (1)
tential of the bonds between the different types of M na M ne
absorbed water layers are estimated by use of the
GAB isothermal absorption model,43,44 which can where m a and m e are the initial weights of amine
also describe the positive deviations from Henry’s and epoxy prepolymer, and M na and M ne are
Law at the last stages of absorption due to forma- the number-average molecular weights, respec-
tion of water clusters.4 tively.50 If r equals 1, the amount of TETA is
Thermomechanical analysis (TMA) experi- stoichiometric. In our specimens, r equals 0.64
ments were also performed to detect the glass , 1, and so some of the epoxide groups do not
transition temperature (T g ) and the expansion react with TETA as its amount is less than the
coefficient (a g ) of the materials under investiga- stoichiometric one.
1168 MAGGANA AND PISSIS

The epoxide groups of DGEBA react with the


hydrogens of ONH2 and ONHO groups accord-
ing to the following reaction:

This reaction leads to the formation of hydroxyl


(OOH) groups, which show strong hydrophilic
character. As the amount of TETA is less than the
stoichiometric one, we accept that the final prod-
ucts, after the curing and post curing procedure,
do not contain amino groups. The plasticizer also
reacts with DGEBA, creating longer polymeric
chains between crosslinks.49
All specimens were prepared by the following
procedure. DGEBA, THIOCOL (in the specimens Figure 1. Typical geometry of the specimens used in
present), and TETA were mixed at 35°C. The diffusion measurements.
mixture was vacuum-degassed for 5 min and cast
between parallel plates covered with silicone. Af-
ter 24 h, specimens were cut from the glassy
epoxy plates prepared by following the above pro- The experimental error in M t was estimated to
cedure. These specimens were postcured for 24 h 0.05% (absolute values). Typical curves obtained
at 100°C in a vacuum (5 3 1022 torr) oven to by dynamic sorption experiments are plotted in
remove water from the final product. The speci- Figure 2, and they will be discussed in the Results
mens after the postcuring process are considered and Discussion section.
as dry.
Equilibrium Water Sorption Experiments
Dynamic Water Absorption Equilibrium water sorption isotherms were mea-
Three series of dynamic water sorption measure- sured at 40 6 1°C. The samples (cylinders of 13
ments were carried out at 15, 40, and 70°C, by mm diameter and 1 mm thickness) were allowed
immersion of dry rectangular samples into deion- to equilibrate to constant weight in various des-
ized water in 50-mL Pyrex glasses. The dimen- sicators with increasing values of relative humid-
sions of the samples were n equal to 10 mm width, ity (RH). The RH was monitored between 6 and
l equal to 10 mm length, and b equal to 1 mm 98% using different saturated salt solutions.51
thickness (Fig. 1). At appropriate intervals, the The percentage of water content h, defined as
samples were removed from water, blotted dry, grams of water per 100 g of dry sample, was
and weighed on an analytical balance (A200S, determined by weighing on an analytical balance
Sartorius, accuracy 1024 g). The water content M t (A200S, Sartorius, accuracy 1024 g). The experi-
was calculated by eq. (2), as follows: mental error in h was estimated to 0.05% (abso-
lute values). After hydration reached full satura-
tion (at RH 5 1), the samples were dehydrated by
wt 2 wd using the reverse procedure. Figure 3 shows the
Mt 5 3 100 (2)
wd variation of the percentage of water content h
with RH during a full absorption– desorption cy-
where w t is the weight of the wet specimen at cle for two specimens with 0 and 40% THIOCOL.
time t, and w d is the weight of the dry specimen. The results will be discussed later.
WATER SORPTION IN AN EPOXY RESIN SYSTEM 1169

of this, we refer to Fujita et al.7 and Vrentas and


Duda10). The one-dimensional case of Fick’s sec-
ond law implies that the material is exposed to
the environment on one side only, or on two sides
with both sides being parallel, as Figure 1 shows.
Using this assumption, the percentage mois-
ture content as a function of time (M t ) can be
calculated from eq. (4), as follows:

M t 5 G~M ` 2 M i! 1 M i (4)

where M i is the initial percentage of moisture


content of the sample, M ` is the percentage of
moisture content at equilibrium, and G is a time-
dependent parameter given by eq. (5), as follows:

G512
8
O
` F exp 2~2j 1 1! 2p S DG
Dt
b2
(5)
Figure 2. Typical results of dynamic absorption mea- p2 j50
~2j 1 1! 2
surements at three different temperatures obtained
with specimens with 20% THIOCOL. Lines are to guide
the eye. G is obtained by integrating the solution of eq. (3)
over the plate thickness b. An analytical approx-
imation for G is given by eq. (6),23 as follows:
TMA Measurements
TMA measurements were carried out on dry sam-
ples using a Du Pont Thermomechanical Ana-
lyzer (model 943) at a heating rate of 10 K min21
to determine the linear expansion coefficient and
the glass transition temperature T g of dry sam-
ples.48

MODELS FOR WATER


SORPTION/DIFFUSION

Fick’s Theory
In most of the cases, the kinetics of moisture
absorption is described by the one-dimensional
case of Fick’s second law, expressed by eq. (3), as
follows:

­c ­ 2c
5D 2 (3)
­t ­x

where c is the concentration of the diffusing sub-


strate, x is the distance through plate thickness, t
is the time, and D is the diffusion coefficient in Figure 3. Water sorption (closed symbols)– desorp-
the direction of absorption ( x-direction normal to tion (open symbols) isotherms at 40°C for two speci-
the plate surfaces).23 D is considered to be inde- mens with 0 (solid lines) and 40% (dotted lines)
pendent of the moisture content (for a discussion THIOCOL. Lines are to guide the eye.
1170 MAGGANA AND PISSIS

F S D G
0.75 the diffusivity of the fibers (D f ), the volume frac-
Dt
G 5 1 2 exp 27.3 (6) tion of the fibers (V f ), and the orientation of the
b2
fibers with respect to the exposed surface.
At the first stages of diffusion, the moisture con-
tent M t varies linearly with the quantity =t/b,
Modified Jacobs–Jones Model
and the diffusion coefficient D can be calculated
from the initial slope s of the diffusion curve in The Jacobs–Jones model accepts that in each
accordance to eq. (7),23 as follows: phase, the diffusion process is controlled only by
the density of that phase. Possible formation of

D5p S D
sb
4M `
2
(7)
hydrogen bonds between water molecules and hy-
droxyl groups of the resin is neglected because of
the lack of such kind of bonds in their system.
For specimens without moisture-impermeable In the following proposed model, the epoxide
coating at their edges, the calculated diffusion product is also considered as a two-phase sys-
coefficient should be corrected in accordance with tem,47,49 consisting of a master phase (phase 1),
eq. (8), proposed by Shen and Springer,23 as fol- where the major part of water is absorbed, and of
lows: a second phase (phase 2), with a different density
or/and hydrophilic character than phase 1.

S D 22 The Jacobs–Jones model can then be trans-


b b
Dx 5 D 1 1 1 (8) formed to eq. (10), as follows:
l n

H F S D GJ
0.75
where D x is the corrected value of diffusion coef- D 1t
M t 5 M 1 1 2 exp 27.3
ficient, and l and n are the length and the width b2
of the specimen, respectively.
H F S D GJ
0.75
D 2t
1 M 2 1 2 exp 27.3 (10)
b2
Jacobs–Jones Model
The analysis of Jacobs and Jones concerns two- where D 1 and D 2 and M 1 and M 2 are the diffusion
phase materials consisting of one dense and one coefficient and the percentage of absorbed water
less-dense phase.33,34 They propose that diffusion at saturation in the two phases, respectively. Ob-
proceeds by parallel Fickian processes in both viously, M ` equals M 1 1 M 2 . The first term of eq.
phases. Then the variation of the normalized wa- (10) is related to the diffusion of water molecules
ter content M(t) with time t can be described by in the homogeneous and nonpolar phase 1 (mode
the morphology-dependent eq. (9), as follows: 1). The second term of eq. (10) expresses the con-
tribution of water absorption in phase 2 (mode 2)

H F S D GJ
0.75 to the whole diffusion procedure. Figure 4 dem-
D dt
M~t! 5 V d 1 2 exp 27.3 onstrates schematically the evolution of both
b2 mechanisms and their contribution to water ab-

H F S D GJ D lt 0.75 sorption by a specimen with 0% THIOCOL at


1 ~1 2 V d! 1 2 exp 27.3 (9) 40°C. This figure is also a good example to dem-
b2 onstrate that the water absorption behavior can-
not be described by only one diffusion coefficient.
Mt The superposition of mode 1 and 2 results in
where M~t! 5 , D d and D l are the diffusion the solid curve (model), which gives a good de-
M`
coefficients of the less-dense and the dense phase, scription of the total absorption (details are in the
respectively, and V d is the volume fraction of the Result and Discussion section).
dense phase. In accordance with the J. J. model, the diffu-
This analysis is based on the expressions de- sion coefficient D 2 refers to the dense phase and
veloped by Shen and Springer23 to describe the must be at least one order of magnitude less than
diffusion of moisture into fiber-reinforced compos- D 1 , as the penetration of water in this region is
ite materials. According to these authors, the dif- hindered by the stiff crosslinking density. This
fusion coefficient of the material (D x ) can be cal- assumption is correct for systems where water
culated from the diffusivity of the matrix (D m ), molecules are not strongly attached to polar
WATER SORPTION IN AN EPOXY RESIN SYSTEM 1171

RESULTS AND DISCUSSION

Overall Behavior of Dynamic Absorption

Figure 2 shows, as already mentioned, the varia-


tion of M t with the quantity =t/b for a specimen
with 20% THIOCOL at 15, 40, and 70°C. Similar
diagrams are obtained for all specimens. In gen-
eral, a slow positive deviation from an equilib-
rium state is observed at longer times; compare,
for example, the results for the specimen with
20% THIOCOL at 40°C in Figure 2. Such contin-
uous slow absorption processes have been re-
ported by numerous authors and can be attrib-
uted to slow relaxation processes of the glassy
epoxy network,52–54 to the filling of hygrother-
mally induced voids,21,22,25,55 and to further cur-
ing reaction in the system.56 These long-term hy-
grothermal aging processes25,56 have not been
studied in this work. The present investigation is
confined to the early stages of hydration up to the
Fickian pseudoequilibrium state.25 The experi-
mental results are fitted by eq. (9). The values of
fitting parameters M 1 , M 2 , and M ` and D 1 and
D 2 are listed in Table I. The values of D 1 and D 2
have been corrected according to eq. (8). Figures 5
Figure 4. Schematic description of the evolution of and 6 show the variation of D 1 and D 2 and M 1 ,
the two modes of diffusion and their contribution to the M 2 , M ` with plasticizer and temperature.
total diffusion process. Lines express the values pre- These results confirm the existence of two dis-
dicted by the proposed model and symbols the experi- tinguished modes of diffusion and that the whole
mental results. procedure should be considered as a superposition
of mode 1 and mode 2.
In all cases, the diffusion coefficient D 1 char-
acterizes the rate of diffusion in the loose regions
groups of the polymer. We will demonstrate that of the material. D 1 is the basic mechanism of
this is not always true, especially in cases where diffusion and concerns the major part of absorbed
hydration takes place at low temperatures or the water (Table I), which is considered as free water.
material contains strong hydrophilic centers. This observation proves that D 1 describes the

Table I. Fitting Values in Accordance with Eq. (10) of the Diffusion Coefficient and the Percentage of Water
Content at Saturation for Both Phases at 15, 40, and 70°C, for Specimens with Various Amounts of Plasticizer

15°C 40°C 70°C


%
Thiocol log D 1 M1 log D 2 M2 M` log D 1 M1 log D 2 M2 M` log D 1 M1 log D 2 M2 M`

0 213.05 1.95 210.40 0.13 2.08 212.55 1.69 211.66 0.51 2.19 211.90 2.48 213.70 0.96 3.44
10 213.13 2.15 210.60 0.14 2.29 212.54 2.02 211.60 0.41 2.43 211.73 2.98 2.98
20 213.14 1.98 210.88 0.19 2.16 212.29 2.22 213.72 0.54 2.76 211.60 2.70 2.70
30 213.08 2.05 210.09 0.14 2.19 212.18 2.32 213.15 0.29 2.61 211.54 2.70 2.70
40 213.05 1.96 210.26 0.15 2.12 212.13 2.38 212.99 0.02 2.39 211.41 2.70 2.70

Note: D 1 , D 2 are the diffusion coefficients (in m2 s21), and M 1 and M 2 are the percentages of water content at saturation for
phases 1 and 2, respectively.
M ` 5 M 1 1 M 2.
1172 MAGGANA AND PISSIS

water molecules and attach them by hydrogen


bonds. This procedure can be characterized as a
Langmuirian mode of absorption and exhibits sat-
uration immediately after the monolayer capacity
is reached. The attachment of the first water mol-
ecule to the nodule results in its disintegration

Figure 5. Variation of the fitting values of the diffu-


sion coefficients (a) D 1 and (b) D 2 of the two modes with
plasticizer content at 15, 40, and 70°C. Lines are to
guide the eye.

same absorption mechanism in all specimens,


which is affected mainly by the temperature. The
influence of temperature and plasticizer content
on D 1 values will be discussed later.
As mentioned before, D 2 is related with the
diffusion process in the dense and/or polar regions
of the material (mode 2), but it does not describe
the same diffusion mechanism in all cases. When
D 2 . D 1 , as Figure 7 demonstrates, the diffusion
is controlled by the formation of hydrogen bonds
between water and hydrophilic nodules. In this
case, water is considered as bound.
These nodules, formed during curing, are re-
gions with high crosslinking density. The curing Figure 6. Variation of the fitting values of the max-
reaction has as a result the formation of hydroxyl imum water content (a) M 1 and (b) M 2 of the two modes
(OOH) groups. The denser the crosslinking is, the and of the total maximum water content (c) M ` with
larger the concentration of OOH groups in this plasticizer content at 15, 40, and 70°C. Lines are to
region of the material is, so the nodules attract guide the eye.
WATER SORPTION IN AN EPOXY RESIN SYSTEM 1173

Evolution of Mode 1 with Plasticizer and


Temperature
With increasing temperature, the fitted values of
D 1 increase for all specimens (Table I and Fig. 5).
The variation of D 1 with temperature is in accor-
dance with the Arrhenius eq. (11), as follows:

S D
D 1 5 D 0exp 2
W
kT
(11)

where W is the activation energy, D 0 is the pre-


exponential factor, and k is Boltzmann’s constant.
Plotting ln D 1 versus 1/T results in straight
lines (Fig. 10). The variation of the fitting values
of W and ln D 0 with the plasticizer content is
demonstrated in Figure 11. The parallel increase
of ln D 0 and W with the plasticizer content indi-
cates that as the amount of plasticizer increases,
the loose phase becomes looser and exhibits
greater segmental mobility. This permits water
molecules larger motions, which require greater
Figure 7. Description of the evolution of the two
modes of diffusion and of their contribution to the total activation energy.24,57 The increase of D 1 values
diffusion process when the dense phase exhibits a with the plasticizer content, especially at higher
strong hydrophilic character (D 1 , D 2 ). Lines express temperatures, verifies this assumption.
the values predicted by the proposed model, and sym-
bols express the experimental results.

because the formation of the bond between OOH


groups and water requires breaking of the in-
tramolecular hydrogen bonds between the polar
groups of the resin, which are responsible for the
formation of the nodule. This hydrophilic proce-
dure can take place only at low temperatures
(Fig. 7 and Table I) when the hydrogen bond
formation is thermodynamically preferable.
At high temperatures, the hydrogen bond for-
mation tends to disappear, and the density of the
nodules mainly controls the diffusion of water in
the nodules. The absorption– diffusion mecha-
nism is similar with mode 1, but it develops at a
slower rate, as the nodule density prevents the
entrance of water molecules in the highly
crosslinked regions. In this case, D 2 , D 1 , and
the total result is much like that already de-
scribed by Jacobs–Jones (Fig. 8).
In experiments on specimens with large
Figure 8. Description of the evolution of the two
amounts of plasticizer at high temperatures, the
modes of diffusion and of their contribution to the total
mode 2 diffusion disappears (Table I), and the diffusion process when the dense phase does not exhibit
kinetics of absorption can be described simply by hydrophilic character (D 1 . D 2 ). Lines express the
the first term of eq. (10). This kind of behavior is values predicted by the proposed model, and symbols
schematically represented in Figure 9. express the experimental results.
1174 MAGGANA AND PISSIS

the plasticizer, forces water molecules to diffuse


faster through the vacancies of the free volume.
So, as the amount of plasticizer content increases,
the diffusion process becomes faster.
The maximum equilibrium water content of
the loose phase (M 1 ) is related to the free volume
of the material. Although all specimens exhibit
the same density at room temperature, their den-
sity decreases at elevated temperatures due to
expansion. The a g values increase as plasticizer
content increases (Fig. 13). So at elevated temper-
atures, specimens with large amount of THIO-
COL expand more than specimens with little or
without plasticizer, and, correspondingly, their
density decreases, and their free volume in-
creases.
For measurements at 15°C, the diffusion tem-
perature is too far from T g , so the relaxation
times for all specimens are much greater than the
time associated with the diffusion mecha-
nism.10,11,59 In addition, the hydrophilic phase is
Figure 9. Description of the evolution of the two well separated from the loose hydrophobic matrix.
modes of diffusion and of their contribution to the total So, the plasticizer content cannot affect the diffu-
diffusion process at elevated temperatures when the sion coefficient, and all specimens exhibit the
materials become homogeneous (Fick’s Law). Lines ex- same behavior represented by Figure 7. The dif-
press the values predicted by the proposed model, and fusion coefficient D 1 and the maximum water con-
symbols express the experimental results.

As suggested by Figure 11, ln D 0 increase lin-


early with W (Fig. 12, where each point corre-
sponds to a plasticizer content). This linear de-
pendence known as the Mayer–Neldel rule or
compensation law has been observed in several
thermally activated processes.58
The variation of the D 1 values with the plasti-
cizer content should be related to the relaxation
rate, the free volume, and the hydrophobic char-
acter of the polymeric matrix. As the amount of
plasticizer increases, the glass transition temper-
ature (T g ) decreases, whereas both the expansion
coefficient (especially at elevated temperatures)
and the hydrophobic character of the material
increase. The variation of the expansion coeffi-
cient a g and of the glass transition temperature
T g with plasticizer content are plotted in Figures
13 and 14, respectively; a g and T g were obtained
from thermomechanical analysis (TMA) measure-
ments.
For diffusion experiments performed in the
glassy state, the relaxation time of the redistribu- Figure 10. Semilogarithmic plot of diffusion coeffi-
tion of the free volume decreases when the tem- cient D 1 of the loose phase versus the reciprocal of
perature of measurement approaches T g . The hy- temperature, 1/T. (Arrhenius plot) for three specimens
drophobic character of the loose phase, induced by with 0, 20, and 40% THIOCOL.
WATER SORPTION IN AN EPOXY RESIN SYSTEM 1175

Figure 12. Variation of the pre-exponential factor D 0


with the activation energy W (each point corresponds to
a plasticizer content).

Fig. 5(a)]. As discussed above, the effect of an


increase in plasticizer content on D 1 is two-fold.
First, T g decreases with increasing plasticizer
content (Fig. 14), so that for experiments at the
same temperature, the relaxation time of the ma-
trix decreases. Second, the hydrophobic character
of the matrix increases with increasing plasticizer
content. M 1 values also exhibit an increase both

Figure 11. Variation of (a) the pre-exponential factor


D 0 and of (b) the activation energy W calculated from
the Arrhenius plots, with plasticizer content.

tent at saturation M 1 , which characterize the


loose phase, are nearly unaffected by the plasti-
cizer, with mean values of log D 1 5 213.1 and of
M 1 5 2%. These results indicate that at low
temperatures, the evolution of diffusion is similar
for all specimens and independent of plasticizer
content. It should be considered as a free diffusion
process without any preferable bond formation
between water and epoxy resin. All specimens
exhibit similar structure of the loose phase with
approximately the same free volume and the
same relaxation time of the matrix. The diffusion
coefficient D 2 is greater than D 1 (Table I) for all
specimens. Its evolution with plasticizer and tem-
perature will be discussed later.
At 40 and 70°C, D 1 values exhibit a systematic Figure 13. Variation of the expansion coefficient a g
increase with the plasticizer content [Table I and at 15 and 40°C with plasticizer content.
1176 MAGGANA AND PISSIS

ter and become unable to further attract water.


So, the mode 2 mechanism reaches saturation
and diffusion proceeds according to mode 1. The
independence of the values of D 2 and M 2 on the
plasticizer content indicates that at 15°C, all ma-
terials have the same structure consisting of hy-
drophilic nodules embedded in a loose hydropho-
bic matrix. The two phases are well separated and
behave in a similar manner independently of the
plasticizer content. So, all materials can be char-
acterized by similar values of D 1 and D 2 and M 1
and M 2 (Table I).
At 40°C, the structure begins to diversify. Both
the loose and the dense phases of the material
become looser. The effect of this change on the
loose phase has already been discussed. The effect
on the dense phase is more complicated. As
Figure 14. Glass transition temperature T g , deter- mentioned before, specimens with 0 and 10%
mined by TMA, versus plasticizer content. THIOCOL keep the same structure observed at
15°C; but the diffusion in the nodules is now
slower, although their density has diminished.
The nodules keep their hydrophilic character,
with plasticizer content and temperature, related which is now, however, less pronounced, as the
to the increase of the free volume of the material decrease of density and the increase in tempera-
[Fig. 6(a)]. ture suppresses the formation of hydrogen bonds.
At 70°C, specimens with more than 10% The result is a decrease of D 2 values [Fig. 5(b)].
THIOCOL exhibit the same M1 equal to 2.7%, The decrease of the density of the nodules leads to
indicating that the free volume of those mate- an increase of their free volume and, thus, to an
rials is almost the same [Fig. 6(a)]. increase of the M 2 values [Fig. 6(b)].
At 40°C, the specimens with 0 and 10% In specimens with 20 and 30% THIOCOL, the
THIOCOL exhibit similar values of D 1 . Also, for nodules begin to disintegrate, loosing, in this way,
both specimens, D 1 , D 2 , indicating that at 40°C, their hydrophilic character. Their structure at
the structure of the material remains practically 40°C becomes similar to that of the loose phase
the same with the structure at 15°C. For speci- but denser. So the diffusion of water in the nod-
mens with a larger amount of THIOCOL, the ules becomes slower without any preferable for-
increase in D 1 is larger. Additionally, D 1 . D 2 , mation of hydrogen bonds. D 2 values are lower
indicating that at 40°C and plasticizer contents than D 1 values (Fig. 5), and the diffusion mecha-
larger than 10%, the hydrophilic character of the nism is similar to that described by the Jacobs
material gradually disappears. and Jones model.33,34 For even higher plasticizer
content (sample with 40% THIOCOL) at 40°C, the
nodules can no longer be detected. The material
Evolution of Mode 2 with Plasticizer and
becomes homogeneous, and the diffusion can be
Temperature
described by the first term of eq. (10) alone or
At 15°C, the hydrophilic diffusion (mode 2) of all simply by Fick’s second law.
specimens is characterized by similar values of At 70°C, all plasticized specimens exhibit ho-
the diffusion coefficient (log D 2 ' 2 10.5) and of mogeneous structure. Water absorption is consid-
the maximum water content (M 2 ' 0.15%) [Ta- ered as a single-phase diffusion into a continuum
ble I and Figs. 5(b) and 6(b)]. Mode 2 is considered without any bond formation between water and
as a very fast and strongly hydrophilic mecha- the polar groups of the resin. The first term of eq.
nism, which reaches saturation at the very early (10) can successfully describe the kinetics of dif-
stages of diffusion (Fig. 7). The small values of M 2 fusion. In this case, M 2 equals 0 for all plasticized
are attributed to the fact that after the absorption specimens [Fig. 6(b)]. Only the specimen with 0%
of water molecules at the primary hydration sites, THIOCOL exhibits a deviation from this behav-
the nodules loose their strong hydrophilic charac- ior, as it is the only one where the nodules can
WATER SORPTION IN AN EPOXY RESIN SYSTEM 1177

still be detected. As it is expected, their hydro- into account the experimental data over the whole
philic character has disappeared because of the range of water activity a(a 3 1), as follows:
high temperature, but they remain denser than
the rest of the material with a diffusion coefficient
kf a
D 2 , D 1 and maximum water content at satura- n 5 nm (12)
tion M 2 equal to 0.97 [Fig. 6(b)]. ~1 2 f a !@1 1 ~k 2 1!f a #

where n is the number of sorbed molecules per


Equilibrium Water Sorption
monomeric unit, n m is the number of sorption
Equilibrium water sorption experiments were sites per monomeric unit (first monolayer sorp-
performed at 40°C. In the following, the results of tion capacity), k is the ratio of the binding con-
these experiments will be compared with those stants of water molecule directly bound to the
obtained from dynamic absorption measurement sorption site in the first layer and of that bonded
at the same temperature. indirectly in the succeeding liquid-like layers, and
As observed in Figure 3, the desorption proce- f is the ratio of the standard chemical potential of
dure exhibits hysteresis compared to the absorp- the indirectly bound water molecule and that of
tion. This phenomenon is more pronounced in the molecule in the bulk liquid state.
specimens with 0 and 10% THIOCOL. As the For k values higher than 1, the site-bound wa-
amount of plasticizer increases, the hysteresis ter molecules may be considered as strongly
loop decreases and, finally, for the sample with bound, while indirectly bound water may be con-
40% THIOCOL almost disappears (Fig. 3). These sidered as loosely bound.43,44
hysteresis phenomena have also been observed in For k values higher than 1, the directly bonded
other polymeric systems.2,60,61 They are usually is considered as strongly bound, while the indi-
attributed to the formation of hydrogen bonds rectly bound water may be considered as loosely
between water and hydrophilic centers of the bound.43,44
material, which lead to permanent changes in The greater the value of k is, the stronger the
the structure of the specimen in the absorption– bond between the water molecule and the hydro-
desorption cycle. When the binding is strong, philic site. Values of f less than unity imply that
the hysteresis loop is large (sample with 0% the outermost hydration layer is formed by water
THIOCOL in Fig. 3). If the binding is loose, ab- molecules of lower activity than bulk water so
sorption– desorption cycles tend to become almost that extra layers of water are required to be as-
identical (sample with 40% THIOCOL in Fig. 3). sociated with the hydrated specimen before it can
The sorption isotherms of specimens with 0 become fully accommodated into normal bulk wa-
and 10% THIOCOL are sigmoid and belong to ter. Low values of f are indicative of the hydro-
type III in the Brunauer classification.3,42 All the phobic character of the polymer.3 When f equals
rest exhibit a parabolic shape and belong to type 1, the GAB equation reduces to the BET.
II.3,42– 45 The negative deviation from the linear- Our experimental data were fitted by eq. (13),
ity (i.e., sorption curves concave to the RH-axis) as follows:
is obvious for the specimen with 0 and 10%
THIOCOL and for water content values up to
kf ~RH!
0.5% and is indicative of the formation of hydro- h 5 hm (13)
gen bonds (the Langmuirian type of absorp- ~1 2 f ~RH!!@1 1 ~k 2 1!f ~RH!#
tion).2,62 This region becomes almost linear in
specimens with more than 10% THIOCOL. For where h is the percentage of water content in the
water content values greater than 1%, sorption specimen, and h m is the percentage of water con-
curves of all specimens exhibit positive deviation tent directly bound to the hydrophilic sites. RH is
from the linear behavior, usually attributed to the relative humidity, which is approximately the
clustering of water molecules.4,62 same as the activity a. Figure 15 shows the ex-
Sorption isotherms of the aforementioned perimental water content values versus RH (sym-
shapes can be analyzed according to the so-called bols) and the fit of eq. (13) to the experimental
Guggenheim–Anderson–De Boer (GAB) equation, data (solid lines).
which is an amended form of the classical The variation of the fitting values with the
Brunauer–Emmet–Teller (BET) equation.42– 45 amount of plasticizer is listed in Table II and
The GAB equation [eq. (12)] allows one to take plotted in Figure 16.
1178 MAGGANA AND PISSIS

Figure 15. Water absorption isotherms at 40°C for


three specimens with 0, 20, and 40% THIOCOL. Lines
are fitting of the GAB equation [eq. (13)].

The results indicate that at 40°C, specimens


with 0 and 10% THIOCOL demonstrate strong
hydrophilic mode of absorption related with the
high values of the k parameter. This behavior
tends to disappear in specimen with 20% or more
THIOCOL (k 3 1). The systematic increase in h m
with plasticizer’s content is followed by a decrease
in k values. This means that the increase of the
amount of plasticizer hinders the aggregation of
the hydrophilic OOH groups and forces them to
disperse in the material. So the strong hydro-
philic centers disintegrate, and although the first
Figure 16. Variation of the fitting parameters of the
GAB equation (h m , k, and f ) with plasticizer content.
Table II. Results of the Analysis Following the
GAB Equation [eq. (13)] of the Sorption Data at 40°C
for Specimens with Various Amounts of THIOCOL
layer of water molecules is bonded to the hydro-
% Thiocol hm k f
philic sites of the polymer, the binding intensity is
almost comparable to that of the indirectly
0 0.57 6.50 0.76 bonded layer.
10 0.72 3.31 0.73 Especially for the specimen with 0% THIOCOL,
20 1.30 1.13 0.64 the amount of water directly bonded to the polar
30 1.32 1.40 0.64 centers is equal to that calculated by the proposed
40 1.36 1.39 0.60 model of dynamic absorption (Table I).
The calculated h m values for the specimens
Note: h m is the percentage of water content directly bound
to the hydrophilic sites; k is the ratio of the binding constants with 0 and 10% THIOCOL should be compared to
of water molecule directly bound to the sorption site in the M 2 values obtained by the proposed model of dy-
first layer and of that bound indirectly in the succeeding namic absorption at 40°C (Table III). For the
liquid-like layers; f is the ratio of the standard chemical po-
tential of the indirectly bound water molecule and that of the specimen with 0% THIOCOL, h m is nearly equal
molecule in the bulk liquid state. to M 2 and k equals 6.5. This is indicative of the
WATER SORPTION IN AN EPOXY RESIN SYSTEM 1179

Table III. Calculated Values of the Percentage of Table IV. Experimentally Determined Values of the
Water Content Directly Bound to the Hydrophilic Percentage of Water Content at Saturation (h max)
Sites {h m in the GAB equation [eq. (13)] Equilibrium from Equilibrium Water Sorption Experiments at
Water Sorption at 40°C} and of the Percentage of 40°C and the Calculated Values of the Maximum
Water Content at Saturation for Phase 2 [M 2 in Eq. Water Content at Saturation (M ` ) from the Fitting
(10); Dynamic Water Absorption at 40°C] of Dynamic Absorption Results at the Same
Temperature
% Thiocol hm M2
% Thiocol h max M`
0 0.57 0.51
10 0.70 0.41 0 2.04 2.19
20 1.30 0.54 10 2.30 2.43
30 1.30 0.29 20 2.39 2.76
40 1.36 0.02 30 2.48 2.61
40 2.31 2.39

strong hydrophilic mode, which characterizes


both equilibrium absorption and diffusion. For fully in agreement with the results obtained by
the specimen with 10% THIOCOL, the fitting the proposed model of dynamic absorption.
value of h m is higher than M 2 . A possible expla- The systematic decrease of f values with plas-
nation for this is that M 2 and h m do not corre- ticizer content indicates the increase of the hydro-
spond exactly the same quantity. M 2 values are phobic character of the specimen with the amount
attributed to the amount of water absorbed by the of plasticizer.3
strong hydrophilic nodules, while h m values are The results obtained by the isothermal equilib-
attributed to the amount of water bonded to the rium absorption measurements prove the validity
primary positions of hydration. In the specimen of the proposed dynamic absorption model.
with 0% THIOCOL, the primary positions of hy- Table IV and Figure 17 show the experimen-
dration are located on the nodules, and, thus, h m tally determined values of the percentage of wa-
is nearly equal to M 2 . In the specimen with 10%
THIOCOL, the increase of the plasticizer content
leads to a decrease of the number of strong hydro-
philic nodules, resulting in a decrease of the
amount of water absorbed by them. The disinte-
gration of some of the nodules, as mentioned be-
fore, increases the number of hydrophilic sites,
but decreases, at the same time, their binding
intensity. So, only part of the water bound di-
rectly to the hydrophilic sites is strongly bonded
to the nodules, while the rest is loosely bonded to
the dispersed hydrophilic sites, so that h m , M 2 .
The parameter k characterizes the binding
strength of the total amount of water directly
bonded to the primary positions of hydration. Its
decrease in the case of the specimen with 10%
THIOCOL (k 5 3.3) can be attributed to the de-
crease of the percentage of the nodules (compared
with the specimen with 0% THIOCOL).
In the specimens with 20% THIOCOL or more,
M 2 expresses the maximum water content of the
dense phase, while h m is related to the amount of
water directly bonded to the hydrophilic sites. So, Figure 17. Variation of the maximum water content
it is not surprising that M 2 and h m get different at 40°C when absorption takes place in a humid envi-
values (Table III). The small values of k (k 3 1) ronment with 100% RH (h max, experimental values)
calculated for all these specimens are indicative and for immersion into the water (M ` , theoretical val-
of the lack of strong hydrophilic nodules and are ues predicted by the proposed model).
1180 MAGGANA AND PISSIS

ter content at saturation (h max) from equilibrium phase system consisting of a master phase (where
water sorption experiments and the calculated the major part of the water is absorbed), which is
values of the maximum water content at satura- homogeneous and nonpolar (phase 1), and a sec-
tion (M ` ) from the fitting of dynamic absorption ond phase with different density and/or hydro-
results at 40°C. The values of M ` ) are, in all philic character (phase 2). Although the overall
cases, greater than h max values. diffusion process exhibits deviation from Fickian
Differences in water content at saturation be- behavior, the proposed model accepts that the
tween specimens hydrated by immersion into the diffusion of water in each phase is Fickian and
water and specimens hydrated by exposure to can be characterized by a diffusion coefficient (D 1
environment with 100% RH are also observed in for phase 1 and D 2 for phase 2) and a maximum
other polymeric systems. Results to be published water content at saturation (M 1 for phase 1 and
show that in rubber-modified epoxy resins, water M 2 for phase 2).
contents determined from immersion experi- The main results can be summarized as fol-
ments are larger than those determined by expo- lows.
sure to water vapor. This difference is usually
attributed to the formation of microcracks and ● The diffusion coefficient D 1 characterizes the
vacancies22 or to further reaction between water rate of diffusion in the loose regions of the
and epoxy resin.63 These phenomena are more material. At 15°C, all specimens exhibit sim-
pronounced when the specimen is immersed into ilar values of D 1 equals approximately 8
the water. In our experiments, the difference be- 3 10214 m2 s21 and M 1 equals approximately
tween M ` and h max values is small, and, consid- 2%, independent of the plasticizer content,
ering that M ` values are theoretical and h max is indicating that, at this temperature, the
experimental, it can almost be neglected (except structure of the loose phase is similar for all
from the specimens with 20% THIOCOL, which materials. At 40 and 70°C, D 1 exhibits an
will be discussed later). These results indicate increase with the plasticizer content, most
that hydration by immersion does not provoke pronounced at 70°C. This phenomenon is at-
more damage in the specimen at 40°C compared tributed to the decrease in the relaxation
with absorption by exposure to humid environ- time and the decrease of hydrophobic char-
ment. For the specimen with 20% THIOCOL, the acter of the loose phase as the amount of
fitting parameters of the proposed dynamic ab- plasticizer increases. Variations in M 1 val-
sorption model indicate the existence of a dense ues at 40 and 70°C are related to variations
phase with log D 2 nearly equal to 13.72 and M ` in the free volume of the specimen.
nearly equal to 0.54. As the calculated diffusion ● The variation of D 1 with temperature follows
coefficient of the dense phase is very small, and Arrhenius behavior. The parallel increase of
the amount of water absorbed in this phase large, ln D 0 (D 0 , the pre-exponential factor) and W
h max probably does not correspond to equilibrium. (W, the activation energy), calculated from
This means that more water would be absorbed in Arrhenius plots, with the plasticizer content
the dense phase if the specimen were exposed for indicates that, as the amount of plasticizer
longer time in the equilibrium water sorption ex- increases, the loose phase becomes looser
periment. and exhibits greater segmental mobility.
● D 2 is related to the diffusion process in the
dense and/or polar regions of the material
CONCLUSIONS (mode 2). When D 2 . D 1 , the diffusion in
phase 2 is controlled by the formation of hy-
Dynamic water absorption measurements and drogen bonds between water and hydrophilic
isothermal equilibrium sorption measurements nodules. The small values of M 2 at 15°C are
performed on a DGEBA/TETA epoxy resin system attributed to the fast saturation of primary
with various amounts of plasticizer THIOCOL hydration sites of the nodules, which loose
were used to investigate the characteristics of their strong hydrophilic character and be-
sorption and diffusion of water in epoxide sys- come unable to further attract water. This
tems. behavior is observed for all specimens at
The experimental results of diffusion were an- 15°C and for the specimens with 0 and 10%
alyzed by using a model described by eq. (10). THIOCOL at 40°C.
This model considers epoxide product as a two- ● When D 2 , D 1 , the hydrophilic character of
WATER SORPTION IN AN EPOXY RESIN SYSTEM 1181

the nodules has disappeared, and the diffu- 11. Vrentas, J. S.; Vrentas, C. M. J Polym Sci, Polym
sion of water in phase 2 is controlled mainly Phys Ed 1992, 30, 1005.
by their density. The absorption– diffusion 12. Lee, H.; Neville, K. Handbook of Epoxy Resins;
mechanism is similar to mode 1, but it devel- McGraw Hill: New York, 1967.
ops at a slower rate in the denser nodules. In 13. Bruins, P. F., Ed., Epoxy Resin Technology; Inter-
this case, observed at 40°C for specimens science Publishers: New York, 1968.
14. Nicodemo, L.; Bellucci, F.; Marcone, A.; Monetta, T.
with plasticizer concentration larger than
J Membr Sci 1990, 52, 393.
10% and at 70°C for the specimen with 0%
15. Schultz, W. J. International Encyclopedia of Com-
THIOCOL, the total result is very similar to
posites, Vol. 2; Lee, S. M., Ed.; VCH Publishers:
that already described by Jacobs–Jones.33,34 New York, 1990; p 77.
● At 70°C, in all specimens with THIOCOL 16. Moy, P.; Karasz, F. E. in Water in Polymers; ACS
concentration larger than 0%, mode 2 disap- Symposium Series 127; American Chemical Soci-
pears, and the materials are characterized as ety: Washington, DC, 1980; p 505.
homogeneous. Fick’s second law can then de- 17. Ellis, T. S.; Karasz, F. E. Polymer 1984, 25, 664.
scribe the diffusion process. 18. McKague, E. L.; Reynolds, J. D.; Halkias, J. E.
● Sorption isotherms obtained at 40°C were J Appl Polym Sci 1978, 22, 1643.
analyzed according to the GAB equation. 19. Theocaris, P. S.; Kontou, E. A.; Papanicolaou, G. C.
They exhibit deviations from linearity and Colloid Polym Sci 1983, 261, 394.
belong to type II or III in the BET classifica- 20. Theocaris, P. S.; Papanicolaou, G. C.; Kontou, E. A.
tion. J Appl Polym Sci 1983, 28, 3145.
● The large values of parameter k in GAB for 21. DeNeve, B.; Shanahan, M. E. R. Polymer 1993, 34,
the specimens with 0 and 10% THIOCOL 5099.
verifies the strong hydrophilic mode of ab- 22. Zheng, Q.; Morgan, R. J. J Compos Mater 1993, 27,
sorption at the first stages of hydration. The 1465.
23. Shen, C.-H.; Springer, G. S. J Compos Mater 1976,
values of k equals approximately 1 for all the
10, 2.
other specimens indicates that the hydro-
24. Nicholson, J. W. J Mater Sci: Mater Med 1997, 8,
philic centers have disappeared. These re-
691.
sults are in accordance with those obtained 25. Chateauminois, A.; Vincent, L.; Chabert, B.; Sou-
by diffusion measurements at the same tem- lier, J. P. Polymer 1994, 35, 4766.
perature. 26. Pethrick, R. A.; Hollins, E. A.; McEwan, I.; MacK-
innon, A. J.; Hayward, D.; Cannon, L. A.; Jenkins,
REFERENCES AND NOTES S. D.; McGrail, P. T. Macromolecules 1996, 29,
5208.
1. Rowland, S. P., Ed., Water in Polymers; ACS Sym- 27. Hayward, D.; Hollins, E.; Johncock, P.; McEwan, I.;
posium Series 127; American Chemical Society: Pethrick, R. A.; Pollock, E. A. Polymer 1997, 38,
Washington, DC, 1980. 1151.
2. Apicella, A. International Encyclopedia of Compos- 28. Karner, H. C.; Schutz, A. in Proceedings of the 4th
ites, Vol. 2; Lee, S. M., Ed.; VCH Publishers: New
International Conference on Properties and Appli-
York, 1990; p 46.
cations of Dielectric Materials, Brisbane, Austra-
3. Pethig, R. Dielectric and Electronic Properties of
lia, June 3– 8, 1994, p 344.
Biological Materials; Wiley: Chichester, 1979; p
29. Illinger, J. L.; Scheider, N. S. in Water in Polymers;
119.
ACS Symposium Series 127; American Chemical
4. Brown, G. L. in Water in Polymers; ACS Sympo-
sium Series 127, American Chemical Society: Society: Washington, DC, 1980; ref. 1, p 571.
Washington, DC, 1980; p 441. 30. Bonniau, P.; Bunsell, A. R. J Compos Mater 1981,
5. Fick, A. Ann Physik Chemie 1855, 94, 59. 15, 272.
6. Crank, J. The Mathematics of Diffusion; Oxford 31. Cotinaud, M.; Bonniau, P.; Bunsell, A. R. J Mater
University Press: Oxford, 1975. Sci 1982, 17, 867.
7. Fujita, H.; Kushimoto, A.; Matsumoto, K. Trans 32. Apicella, A.; Estiziano, L.; Nicolais, L.; Tucci, V. J
Faraday Soc 1960, 56, 424. Mater Sci 1988, 23, 729.
8. Sun, Y.-M. Polymer 1996, 37, 3921. 33. Jacobs, P. M.; Jones, F. R. J Mater Sci 1989, 24,
9. Carter, H. G.; Kibler, K. G. J Comp Mater 1978, 12, 2331.
118. 34. Jacobs, P. M.; Jones, F. R. J Mater Sci 1989, 24,
10. Vrentas, J. S.; Duda, J. L. Encyclopedia of Polymer 2343.
Science and Engineering, 2nd ed., Vol. 5; New 35. Gupta, V. B.; Drzal, L. T. J Appl Polym Sci 1985,
York: Wiley, 1986; p 36. 30, 4467.
1182 MAGGANA AND PISSIS

36. Suzuki, T.; Oki, Y.; Numajiri, M.; Miura, T.; Kondo, 50. Boye, J.; Demont, P.; Lacabanne, C. J Polym Sci,
K.; Shiomi, Y.; Ito, Y. Polymer 1994, 37, 3025. Polym Phys 1994, 32, 1359.
37. Berens, A. R.; Hopfenberg, H. B. Polymer 1978, 19, 51. Greenspan, L. J Res Natl Bur Stand (U.S.) 1977,
489. 81A, 89.
38. Xiao, G. Z.; Shanahan, M. E. R. J Polym Sci, Polym 52. Johncock, P.; Tudgey, G. F. Br Polym J 1983, 15,
Phys Ed 1997, 35, 2659. 14.
39. Adamson, M. J. J Mater Sci 1980, 15, 1736. 53. Wong, T. C.; Broutman, L. J. Polym Eng Sci 1985,
40. Yang, F.; Gilbert, R. D.; Fornes, R. E.; Memory, 25, 521.
J. D. J Polym Sci, Polym Chem Ed 1986, 24, 2609. 54. Wong, T. C.; Broutman, L. J. Polym Eng Sci 1985,
41. Carfagna, C.; Apicella, A.; Nicolais, L. J Appl 25, 529.
Polym Sci 1982, 27, 105. 55. Carfagna, C.; Apicella, A. J Appl Polym Sci 1983,
42. Brunauer, S. The Absorption of Gases and Vapors, 28, 2881.
Vol. 1; Princeton University Press: Princeton, NJ,
56. Johncock, P. J Appl Polym Sci 1990, 41, 613.
1945.
57. Bastioli, C.; Guanella, I.; Romano, G. Polym Comp
43. Galin, J. C.; Galin, M. J Polym Sci, Polym Phys Ed
1990, 11, 1.
1992, 30, 1113.
58. Yelon, A.; Movachar, B.; Branz, H. M. Phys Rev B
44. Galin, J. C.; Galin, M. J Polym Sci, Polym Phys Ed
1995, 33, 2033. 1992, 49, 12244.
45. Timmermann, E. O. J Chem Soc, Faraday Trans I 59. Ebrahimzadeh, P. R.; McQuieen, D. H. J Mater Sci
1989, 85, 1631. 1989, 33, 1201.
46. Antoon, M. K.; Koenig, J. L. J Polym Sci, Polym 60. Schult, K. A.; Paul, D. R. J Polym Sci, Polym Phys
Phys Ed 1981, 19, 1567. Ed 1996, 34, 2805.
47. Kontou, E.; Spathis, G.; Theocaris, P. S. J Polym 61. Das-Gupta, D. K. IEEE Trans El Insul 1992, 27,
Sci, Polym Chem Ed 1985, 23, 1493. 909.
48. Wendland, W. W.; Gallagher, P. K. Thermal Char- 62. Pissis, P.; Apekis, L.; Christodoulides, C.; Niaou-
acterization of Polymeric Materials; Turi, E. A.; nakis, M.; Kyritsis, A.; Nedbal, J. J Polym Sci,
Ed., Academic Press: London, 1981; p 68. Polym Phys Ed 1996, 34, 1529.
49. Spathis, G.; Kontou, E.; Theocaris, P. S. Rheol Acta 63. Netravali, A. N.; Fornes, R. E.; Gilbert, R. D.; Mem-
1985, 24, 63. ory, J. D. J Appl Polym Sci 1985, 30, 1573.

You might also like