Professional Documents
Culture Documents
1. INTRODUCTION
Materials utilized in barrier and corrosion protection
applications such as organic coatings and linings
are usually designed with low transport properties,
making them almost impermeable to diffusing fluids. The desired impermeability in these materials
is mostly achieved by using reinforcement. These
reinforcing, impervious fillers are hard, brittle, and
are available in different types. Composite materials
can be divided into three classes according to the
type of reinforcement used [1-3]: one-dimensional
(fibres, whiskers); two-dimensional (flakes, platelets, disks, ribbons); and three-dimensional (spheres,
beads). Among the different types of fillers, fibres
and flakes are more efficient reinforcements than
spheres in terms of mechanical and barrier property
enhancement. While fibres give higher moduli and
strength, flakes are advantageous when properties in
a plane are considered due to their capability to reinforce in two dimensions [2,3].
Many researchers have investigated diffusion in
flake-reinforced composites using analytical, numerical and experimental methods. Most analytical
formulations and numerical simulations of diffusion
in these composites are performed based in the assumption that all flakes are oriented perpendicular
to the diffusion direction. At this orientation, flakes
Advanced Composites Letters, Vol. 21, Iss. 6, 2012
alcohol membranes swollen in aqueous ionic solutions, White and Cussler [17] showed composites
with aligned inorganic flakes have permeabilities
that are lower perpendicular to the flakes than parallel to the flakes.
impermeable
Fig. 1: A compositereinforced
consisting of with
a permeable
matrix reinforced
with
elements.
triangular
impermeable
triangular
elements.triangularThe
138
Table 1:
Preparedfor
specimens
for water absorption-deTable 1: Prepared
specimens
water absorption-desorption-reabsorption
tests
sorption-reabsorption
tests
Specimen
VE
EP
A
B
C
D
Exposed side of
Filler
Remarks
triangular bar fillers alignment
pure vinyl ester
pure epoxy
sloped
parallel
composite
base
parallel
composite
sloped
staggered
composite
base
staggered
composite
Fig. 2: Layer-wise assembly of triangular bar fillers and to the applied foil and adhesive
before proceeding to
Fig. 2: Layer-wise assembly of triangular bar fillers and mold setup.
mold setup.
the experiments.
ester specimens were
in a layer-wise fashion as illustrated in Fig. 2. The Cured pure epoxy and vinyl
arrangement of triangular bars per layer was guided also prepared for testing and comparison. Table 1
1: Prepared specimens for water absorption-desorption-reabsorption tests
by Table
a pattern
laid out in 1-mm
grid paper. Two 3- lists the specimens tested in this work. Triangular
Exposed
side
of
Filler
layer structures
bonded to form bar fillers were oriented inside the epoxy matrix
Specimen were assembled and Remarks
triangular bar fillers alignment
VE
- structure
vinyl ester The tria 6-layer symmetrical
aspurepure
shown.
with the fillers base side or sloped side exposed
EP
epoxy
surface. Triangular bar
sloped bonded
parallel at their
compositeends using
angular barAB fillers were
near the composite specimen
base
parallel
composite
in parallel or staggered
C
sloped
staggered
composite
cyanoacrylate
adhesive
(Aron
Alpha
Extra 2000, fillers could also be aligned
D
base
staggered
composite
with 1 mm spacing.
Thewere
fillersealed
assembly
used steel
condition
Table through-the-thickness
1.
The edges
with was
stainless
foil to inensure
moisture diffusion in the
10 mm
to reinforce aspecimens
100 edges
x 100during
xwere
composite
speciThe
sealed
with stainless
steel was
foil to
ensure
through-the-thickness
moisture
in its
the
experiments.
Vinyl
ester
used
as
adhesive
for
the
stainless
steel diffusion
foil due to
specimens
during
experiments.
Vinyl ester
usedthe
as adhesive
adhesive and
for sealing
the stainless
steel the
foil composite
due to its
men.
moisture
resistance
at high temperature.
Afterwas
curing
the edges,
2.2
Water
absorption-desorption-reabsorption
moisture
resistance
at high
After
curing
theoCadhesive
and 100
sealing
thewas
edges,
the composite
was
again
stored
intemperature.
a inside
convective
air oven
at 50
for at least
h and
weighed
again to
The completespecimen
filler assembly
was then
placed
gravimetric
tests
specimen
was
again
stored
in
a
convective
air
oven
at
50
C
for
at
least
100
h
and
was
weighed
again to
themetal
weight
increase
to plates
the applied
foildetermining
and adhesivethe
before
proceeding
theto
experiments.
a 190 x 190 obtain
x 10 mm
mold.
The due
metal
After
weight
increase todue
ap
obtain
the
weight
increase
due
to specimens
the
foil
and
before
proceeding
theedges,
experiments.
Cured
pure
epoxy
and
vinyl
ester
were
alsoadhesive
prepared
for testing
Table 1 lists the
forfillers
each specimen
condition
Table
1. matrix with the fillers
bar
were oriented
insidein the
epoxy
ed
again
in
deionized
water
for
300
h
80ofillers
C. Weight
specimens
was monitored
composite specimen surface. Triangularatbar
couldofalso
be
ater
absorption-desorption-resorption
tests.
Results
are
presented
as
percent
weight
ater
absorption-desorption-reabsorption
gravimetric
tests
t during filler assembly. Triplicate specimens were prepared and
th
the
initial
dry
weight
of
the
specimens:
determining
the
increase
due
to
applied
adhesive
and
stainless
steel
foil
at the edges, the dried
1.
mens were immersed in deionized water insideBryan
glassPajarito,
containers.
Teflon
holders
were Aoki
used to avoid
Masatoshi Kubouchi, Saiko
o
(1)
where
Wi isWthe
specimen
weight
during of
tests,
Wd
cal
among
immersed
specimens.
The
temperature
immersion
was kept at 80 C for 1200
h
ioncontact
gravimetric
tests
i Wd
(%initial,
)stainless
= thedried
100wereweight,
tand
atowater
bath.
After
1200
h,
specimens
removed
from
water
immersion
and
were
placed
inside
isWthe
specimen
and
W
is
the
applied
adhesive
steel
foil
at
the
edges,
the
dried
b
Wd
uum
oven for
drying.
Thespecimen
temperature
of vacuum
drying
kept at 80oAC for 300 h. After 300 h,Bthe
baseline
before
drying,
ater inside
glass
containers.
Teflonweight
holders
wereimmersion,
used towas
avoid
o
o
h at 80
e/loss
Mtemperature
relation
with
the
weight
before
water absorption,
t in
mens
were
immersed
again
in initial
deionized
water
for
300
C.h Weight
of specimens was monitored
and re-immersion.
ns.
The
of immersion
was
keptofatthe
80specimen
C for
1200
ion:
ecorded
during
water
absorption-desorption-resorption
tests.
Results
are
presented as percent weight
mens were removed from water immersion and were placed
inside
o
e
W
in
relation
with
the
initial
dry
weight
of
the
specimens:
e of tvacuum drying
kept at 80
C for 300equilibrium
h. After 300water
h, theup2.3 was
Estimation
of apparent
Wi Wof
o
(2)
b specimens was monitored
ed water for 300take/loss
hMat (80
C.
Weight
100
t %) =
(1)
Wasi percent
Wuptake/loss
rption-resorptionThe
tests.
ResultsW
are
weightM
apparent
equilibrium
water
b presented
d
100
(
)
W
%
=
t
ight of the specimens:
of specimens during absorption,
and
weight, and
Wd desorption,
n weight during tests, Wd is the initial, dried specimen
W is the baseline
reabsorption tests was estimated using
a method b
immersion,
drying,
and re-immersion.
ercent
water
uptake/loss
M
in
relation
with
the
initial
weight
of
the
t
(1) specimen before water absorption,
100
D
C
materials. Using Mt (%) versus
t (h) data
Wd polymeric
ent equilibrium
water uptake/loss
obtained M
from of
experiments,
MW
was approximated
W absorption, desorption, and
specimens
during
ium
uptake/loss
specimen
(2)
i
b
with water
the initial
weightEq.
of 3:
the
before
water
absorption,
100
(
)
M
%
=
using
stimated using a method suggested by tYurev and Lushchik
[18] for thick polymeric
b
versus t (h) data obtained from experiments, M was
approximated
using Eq. 3:
(3)
(3)Wb is the baseline
Wi is theW
specimen
weight
during
tests,
W
is
the
initial,
dried
specimen
weight, and
d
Wb
M1
(2)
) = ibefore
M
100
(% ) = drying,
Mimmersion,
t (%
men
weight
and re-immersion.
2 (M 2 M 1 )
Wb
uptake/loss
where
Mspecimen
and M2 weight,
are theuptake/loss
percent
water
timation
of initial,
apparent
equilibrium
water
1
s,
Wd is the
dried
and Wb is
the
baseline
specimen
time
t1 and tat
=time
2tof
, respectively.
he
waterofuptake/loss
specimen
and
t2 = 2tThe
The desorption,
1, respectively.
during
absorption,
and of composite speciapparent
equilibrium
water atof
uptake/loss
2M
1 t1specimens
andpercent
re-immersion.
Fig.
3: Cross-section
optical images
ng a time
t2 was
suchmethod
that theinvolves
inequality
1 < M2suggested
M1 <t2 2such
was
observed.
the calculated
a /time
that
theand
in- If
Yurev
orption
tests
estimated
using aselecting
method
by
Lushchik
[18]
for
thick
polymeric
mens images
A, B, of
C,composite
and D. specimens A, B, C, and D.
Fig. 3: Cross-section optical
If of
ess
than
theMrecorded
maximum
percent
uptake/loss
another t2 using Eq. 3:
equality
1t (h)
< Mdata
/ Mobtained
< 2water
wasfrom
observed.
thethe
calcuals.uptake/loss
Using
versus
experiments,
Mspecimen,
t (%)
was approximated
2
1
ter
a. TheMprocedure
wasMrepeated
until the t2calculated
M the
is recorded
greater
or equal
the
lated
at the selected
is lessdesorption,
than
(3) images of polymer
the cross-section
optical
and thanshows
/loss
during absorption,
M
of specimens
1
value.
M
%
M
=
t
maximum
percent
water
uptake/loss
of
the
specicomposite specimens A, B, C, and D. In specimens
[18] for thick polymeric
thod suggested by Yurev and Lushchik
M
2 M
men, another
t2 is approximated
selected from2using
the
data.
The
tained from experiments,
M was
Eq.
3:1 proce- A and C, the sloped side of the reinforced triangles is
r diffusion coefficient
dure was repeated until the calculated M is greater
first to be exposed during water absorption-desorp(3)
M
1
%)
obtained
from
experiments
the estimated
values,
M1 versus
and
are data
the percent
uptake/loss
ofand
specimen
timeMt1tion-reabsorption
and t2 =water
2t1, respectively.
The
= Mt2 (h)
than
or equalwater
the
maximum
experimental
Mt at
value.
gravimetric
experiments. Speci %
specimens
were
calculated
by Levenberg-Marquardt
The
2 selecting
M
df involves
a1 time t2 such
that the inequality 1 < nonlinear
M2 / M1 < regression.
2mens
was observed.
If
the
calculated
2 M
B and D, on the other hand, have the base side
by Crank
[19]
for
Fickian
diffusion
in amaximum
square
plate
of thickness
2h: of theofreinforced
the
selected
t2 is
less
than the
recorded
percent
water uptake/loss
the specimen,
another
t2 to diffusion. It is
2.4
Calculation
of water
diffusion
coefficient
triangles
exposed
cted
from
the
data.
The
procedure
was
repeated
until
the
calculated
M
is
greater
than
or
equal
the
ptake/loss of specimen
at time
t1 and
2t1, respectively.
The
Using the
recorded
Mt t(%)
t (h) data obtained
2 = versus
expected
that water diffusion in specimens(a)
A and C
(4)
value.
mum
M
t
at theexperimental
inequality
1
<
M
/
M
<
2
was
observed.
If
the
calculated
8
from experiments
and the estimated
M values,
wa- is faster than in specimens B and D due to the dif2
2
1
+ 1 2t another
M
exp of
Dthe
2nspecimen,
4h 2 t
t = M 1
2 coefficients
2
d maximum
percent
water
uptake/loss
2
ter
diffusion
of
specimens
were
calcuference in exposed matrix area available for mass
n =0 2n + 1
the
alculation
of until
water
diffusion
coefficient
was
repeated
the
calculated
M is greater nonlinear
than or equal
lated
by
Levenberg-Marquardt
regression.
transport. Alignment of triangular fillers is also exthe recorded Mt (%) versus t (h) data obtained from experiments
and the estimated M values, water
[19] for Fick(b)
The
model
function
is
given
by
Crank
pected
to
contribute to filler orientation-dependent
ffusion
coefficient.
script waswere
written
in GNU by
Octave
software utilizing
its
leasqr
ion
coefficients
of Aspecimens
calculated
Levenberg-Marquardt
nonlinear
regression. The
ian diffusion
in a square data.
plate of thickness
2h:
diffusion in composites. The alignment of rein a square
diffusion
from experimental
function iscoefficient
given by Crank
[19] for Fickian
diffusion in
plate of thickness 2h:
nt
forced triangles in the matrix is parallel for speci
a obtained from experiments and the estimated M values,
water
calculated by Levenberg-Marquardt
nonlinear
regression.
The 2 2
8
2
D.
Studies
of
diffusion
in
flake-filled
films(c)suggest
SCUSSION
= M plate
1 of thickness22h:2 exp D 2n + 1 t 4 h
ickian diffusion inMat square
(4)
lower
diffusivity
in
staggered
than
in
parallel flake
2
1
+
n =0
tortuosity [4-11,15-17].
networks
due
to
increased
(4)
8
2 2
Hence,
it alsoutilizing
expecteditsthat
diffusion in specimens
D is the water
diffusion
coefficient.
A2 script
was
written
in
GNU
Octave
software
leasqr
+
exp
D
2
n
1
t
4
h
2 2
C
and
D
is
slower
than
in
specimens
A and B due to
2
1
+
n
on to calculate the diffusion coefficient from
data.
experimental
( )
( )
Fig. 4 shows the percent weight change curves during water absorption, desorption, and reabsorption
as a function of square root time for all specimens
at 80oC. All specimens gained weight during water
absorption and reabsorption, and lose weight during water desorption. Specimen EP has the highest
weight change during exposure to deionized water,
followed by composite specimens, and lastly by
specimen VE. Specimen VE has achieved water
Advanced Composites Letters, Vol. 21, Iss. 6, 2012
(a)
(b)
(c)
VE
EP
Fig. 5: Water uptake/loss of specimens during absorption, desorption, and reabsorption at 80oC.
Fig. 5: Water uptake/loss of specimens during absorption, desorption, and reabsorption at 80oC.
Advanced Composites Letters, Vol. 21, Iss. 6, 2012
141
er water uptake during reabsorption. The specimens loss of specimens during absorption, desorption,
are made up of cured vinyl ester and epoxy com- and reabsorption assumes effective diffusion across
ponents which are glassy polymers that have time- the materials to be Fickian, characterized by an inidependent properties [19]. During absorption, water tial linear curve of water uptake/loss versus square
to additional free volume. The added free volume or equilibrium stage defined by the parameter M.
will still be available in the polymer network during The double time approach of Yurev and Lushchik
since the stressed poly- [18] was applied in this work to indicate a possible
desorption and reabsorption
VE mer chains of glassy polymers
EP time to relax. value for M which simplifies the determination of
take
The added free volume in the specimens could also D from experimental data. However, a close inspecbe due to irreversible damage during water absorp- tion of experimental data in Fig. 5 shows the percent
tion such as cracks, microcavities, and voids from water uptake/loss curves to be somewhat trending
debonded filler-matrix interphases [21]. Debonding beyond the apparent M values. This observation
and degradation of the interphase between epoxy suggests the behaviour of water uptake/loss in the
matrix and vinyl ester fillers due to fluid sorption- specimens to be non-Fickian, for instance two-stage
induced swelling may result to narrow spaces and sorption [22,23]. In almost all specimens, the first
promote overall water diffusion and transport by linear stage lasts until about t = 20 h0.5, followed
capillary effect. These may facilitate higher effec- by a second linear stage of lower slope. The first
and re- stage of sorption is usually attributed to be Fickian
B
A tive diffusion rates during water desorption
absorption than during absorption.
diffusion-controlled, while the second stage of sorption is described as due to polymer matrix relaxation
Rather than use the maximum percent water uptake/ Fig. 6 shows the effect of triangular bar filler ex
loss as the apparent equilibrium value M [20,21], oposed side and alignment on the average diffusion
Fig. 5: Eq.
Water3uptake/loss
of specimens
during absorption, desorption,
and reabsorption
C.
was used
for approximation,
as the method
wasat 80 coefficient
of composite specimens
during water ab
specifically designed for thick polymeric specimens sorption, desorption, and reabsorption at 80oC. As
diffusion
(b)
filler-matrix interphase degradation [21].
Absorption
Specimen
M (%) D (mm2/h)
VE
1.5466 0.0488
EP
3.1929 0.0280
A
2.5242 0.0245
B
2.5806 0.0194
C
2.8166 0.0187
D
2.8776 0.0174
142
Desorption
M (%) D (mm2/h)
1.5983 0.1987
2.8233 0.1673
2.6936 0.2242
2.6297 0.1993
2.6899 0.2144
2.6192 0.1569
Reabsorption
M (%) D (mm2/h)
1.6020 0.1398
2.6074 0.0913
2.5382 0.0978
2.5109 0.0969
2.5238 0.0820
2.3981 0.0917
Fig. 6: Effect of triangular bar filler a) exposed side and b) alignment on average diffusion coefficient of co
shown from the results during absorption and desorption in Fig. 6a, diffusion is slow when base side
of triangular bar fillers is exposed to diffusion, and
fast when sloped side is otherwise exposed. Orienting the base side of triangular bar fillers outward in
the composite reduces the available area of permeable epoxy matrix for water absorption and desorption. On the other hand, results shown in Fig. 6b
describe the effect of filler alignment in water diffusion. In parallel alignment, diffusion resistance
is mainly from the epoxy matrix and presence of
slits between triangular bars. Aligning the bars in
staggered configuration contributes additional resistance to transport by increasing the path length
of diffusing water molecules in the composite. The
resulting effect, known as wiggling in flake barrier
films [4-11,15-17], is due to increased tortuosity of
the composite at staggered alignment of triangular
bar fillers.
4. CONCLUSIONS
Reinforcing a permeable matrix with aligned, impermeable triangular elements could result to a
composite material with single axis-, filler orientation-dependent diffusion behaviour. To test this idea
experimentally, this work used cured epoxy and
vinyl ester to represent the permeable and impermeable component of the composite, respectively.
Water was chosen as the diffusing fluid, and absorption-desorption-reabsorption tests were performed
on specimens at 80oC. Percent weight change curves
of specimens show weight gain during absorption
and reabsorption, and weight loss during desorption. Weight change values of polymer composite
specimens fall between the values of pure epoxy
and vinyl ester specimens. Water uptake of specimens during absorption is lower than during reabsorption. The result is related to the generation of
additional free volume in the specimens due to the
time-dependent properties of glassy polymers and
irreversible damage sustained during water absorption and desorption. The calculated diffusion coefficients of specimens show high water diffusivity during desorption, followed by reabsorption, and then
absorption. Moisture diffusivity of composite specimens are found to be lower than pure matrix resin
during absorption and higher during desorption and
reabsorption, indicating fast water transport at the
local, degraded interphase regions. The calculated
diffusion coefficients also reveal filler orientationdependent transport in composite specimens. Diffusion is slow when base side of triangular elements is
exposed to diffusion, and it is fast when the sloped
side of triangular elements is exposed. Exposing the
Advanced Composites Letters, Vol. 21, Iss. 6, 2012
144