= —E2w€ 2 o where
is the rate of power absorption, E0 is the
electric field intensity, w is the frequency of radiation
and e" is the loss factor. Differences in dielectric
properties between fiber and matrix may give rise to energy
localization effects in the interphase region. Further the
reflection of microwave radiation from the conducting carbon
fiber surface in laminated composites could cause different
unidentified phenomena in the interphase region. Sharp
temperature gradients at the interface may cause the
processing of the interface to be different from the bulk.
All these could be the source for variations in the local
morphology of polymers in the interphase region.
6
All the different possible phenomena occuring in the
interphase region as discussed above could be a beneficial
or detrimental development depending on the polymer
chemistry. The creation of an excessively crosslinked region
could increase the efficiency of stress transfer from fiber
to matrix while reducing the composite fracture toughness.
The promotion of fracture toughness with a reduction in
mechanical properties could be a result if matrix crosslink
density is reduced.
Microwave heating can also be utilized for the
processing of high temperature thermoplastics and
thermoplastic matrix composites. In order to use microwave
radiation in this area, interactions of microwave radiation
with thermoplastic matrices need to be explored. Microwave
radiation could alter the conformation of the matrix and
could also cause morphological changes in the matrix and in
the fiber-matrix interphase which would change the
structure-property behaviour of the system. In the case of
semi crystalline polymers, rate of nucleation, crystal
growth rates, and degree of crystallinity could all be
affected by the microwave heating mechanisms. Again, all
these effects could be beneficial or detrimental and need to
be investigated.
BACKGROUND AND LITERATURE REVIEW
2.1 BACKGROUND
A dielectric material may be defined as one in which it
is possible to store electric energy by the application of
an electric field and recover the energy when the field is
removed. Dielectric materials are usually very poor heat
conductors. Heating such substances throughout their volume
is very difficult with conventional processes that apply
heat to the surface only. Electromagnetic energy in the
radio-frequency (RF) range, on the other hand, can act below
the surface of a material and heat all parts of the volume
simultaneously with substantially greater speed and
uniformity of heating than conventional methods. This
heating process is termed "dielectric heating".
For dielectric heating, two ranges of radio-frequencies
are used: a frequency somewhere in 1-200 MHz range, usually
known as high-frequency or radio-frequency heating;
frequency in 1-5000 GHz range, known as microwave heating.
In high-frequency heating, the material to be heated is
usually placed between two electrodes where as in microwave
8
heating, energy is applied by specially designed microwave
applicators and wave guides.
Dielectric heating at any frequency is the result of the
interaction of electromagnetic energy with the atomic and
molecular structure of materials. Many polymer systems
contain various polar functional groups with permanent
dipole moments. These dipole moments are randomly oriented
due to the thermal brownian motion. The resultant
polarization, which is defined as the net dipole moment per
unit volume, is zero because of the random orientation of
all the dipoles. The incident electromagnetic radiation
creates an electric field which interacts with the dipole of
the molecule. The molecule rotates, thereby aligning itself
with the field and thus increasing the polarization.
Polarization would be maximum if all the dipoles were in
alignment, but the random molecular motion continuously
knocks dipoles out of alignment, keeping the polarization
below the maximum level.
Molecules can’t rotate instantaneously into alignment
with the electric field as they have mass spread over a
certain volume and there are retarding forces exerted by the
surrounding molecules. The response time of a system can be
considered in terms of the decay of its polarization if an
electric field is suddenly turned off. The response time of
a system determines whether the dipole moments can keep up
9
with the oscillating electric field in an electromagnetic
wave .
At low frequencies, the time taken by the electric field
to change direction is longer than the response time of the
dipoles and polarization keeps in phase with the electric
field. The field provides energy to make the molecules
rotate into alignment. Some energy is transferred to the
random motion each time a dipole is knocked out of the
alignment and realigned. The transfer of energy is so small,
however, that the temperature hardly rises. At high
frequency, the electric field oscillates so rapidly that it
changes direction faster than the response time of the
dipoles. Since the dipoles do not rotate, no energy is
absorbed and the system does not heat up.
In RF and microwave range of frequencies, the time in
which the field changes is about the same as the response
time of the dipoles. Dipoles rotate because of the resulting
torque they experience but the resulting polarization lags
behind the changes in.the direction of the electric field.
When the field is at its maximum strength, the polarization
may still be low and keeps rising as the field weakens. The.
lag indicates that the system absorbs energy from the field.
Dielectric heating has been in use for quite some time
in medical, food, textile, and paper industries but has
10
recently been applied in the field of composites.
Measurements of dielectric properties have been used to
monitor chemical reactions in organic materials for more
than fifty years but the use of electromagnetic processing
of composites is still under investigation. Work in this
area has been greatly stimulated in the past decade due to
the increasing importance of thermosets and thermoplastics
as matrix resins in fiber reinforced composites.
2.2 LITERATURE REVIEW
2.2.1 MICROWAVE EFFECTS ON POLYMER COMPOSITE :
Microwave processing of epoxy resins at 2.45 GHz has
been studied in conventional microwave (multimode) ovens
[2,3], in TE01 wave guides [4] and in a TE10 wave guides
[7]. Karmazsin [7] has used continuous and pulsed microwave
radiation at 2.45 GHz in TE wave guides to cure epoxy.
10
Springer [8,9] has tried to model the interactions of
electromagnetic radiation with polymeric composite and
electromagnetic processing of composite materials. A kinetic
model of epoxy with DDS has also been proposed by Sheppard
[11-13]. The following section will summarize and discuss
the significant recent research work in this area.
Wilson and Salerno [2] have used a conventional
microwave oven at 2.45 GHz to study curing behaviour of
ll
epoxy with different curing agents at power levels of 245
and 700 watts. Time-temperature profiles of two systems: 100
parts of Epon 828 with 49 parts of T-403 (Jefferson Chemical
Company) and 100 parts of Epon 828 with 20 parts of curing
agent Z (Shell Chemical Co.) were investigated under the
microwave environment. They have identified three different
regions in the time-temperature profiles. In the lower
temperature zone, the temperature slope increases with time
due to the rapid microwave heating. The central portions of
the curves are of approximately constant slope or slightly
concave downward as most of the epoxy is reacted. In the
third zone, the slope increases with time after the
microwave source is turned off and this is due to the heat
produced by exothermic reactions.
They have also measured the dielectric properties of
these two gelled, uncured epoxy systems at 23, 60 and 70 0C
at frequencies ranging from 1.0 to 2.45 GHz. using a slotted
coaxial transmission line. Their experiments indicate that
the dielectric properties increase with increasing
temperature and are not strong functions of frequency in the
desired range. The rate of increase of dielectric properties
with increasing temperature is faster in 828-z than 828-T403
.due to the characteristics of the curing agent.
The proposed model for the absorption of microwave
energy as heat is based on the assumption that the curing
12
epoxies exhibit a bulk loss tangent close to that of pure
water, rather than the value measured for the gelled epoxy,
and the dielectric constant is in the same range as that of
gelled material. They have not considered the problem of
microwave reflection by laminate structure or the dependence
of dielectric properties on the extent of conversion.
Strand [3] has studied the feasibility of microwave
processing in the plastic molding industry. He has compared
the temperature-time profiles of thermally cured epoxies,
polyesters, and polyurethanes at different temperatures with
the ones under microwave environment at various power
levels. The cure time of epoxy by the 6 KW microwave energy
is found to be 30 times less than that of thermal heating at
177 oC. The experiments indicate that fast cure and high
efficiency of energy utilization can be obtained by the
microwave curing compared to the thermal curing.
Gourdenne, et al. [4] have described the thermal
interactions between the microwaves and some organic
materials to be polymerized and have measured the transfer
of energy from the electromagnetic beam to the irradiated
samples. They have used a TE wave guide operated at 2.45
01
GHz to study the temperature-time profile of a DGEBA type
epoxy cured with DDM (diamino diphenyl methane) at different
initial power inputs of 40, 60, 80, and 100 watts. They have
also theoretically modeled two microwave heating profiles:
13
before and after polymerization. In the experimental
temperature-time profiles the temperature increases
regularly in the beginning, then more rapidly until a
maximum from which it decreases slowly afterwards. The shape
of the peak and the intensity and position in time of its
maximum depend on the input power level. The peak is
broadened and translated towards the high values of time,
whereas the intensity of its maximum is lowered, with
decreasing input power. The uncured and cured portions of
the temperature-time profiles match well with the simulated
curves .
Gourdenne and Van [5] have used the above mentioned
epoxy-resin curing-agent and wave-guide systems to study the
microwave curing of glass fiber filled epoxy systems. Two
types of experiments are reported: 1. microwave treatment at
given epoxy/glass composition and at variable power levels,
2. microwave treatment at given power values and variable
glass composition. The curves of temperature at different
power inputs are found to be similar to the curves without
glass fiber. Also the temperature profiles at fixed
microwave power but at increasing fiber contents are found
to be similar to those temperature profiles of epoxy without
fiber at decreasing power inputs. They have concluded that
the absorbed energy in epoxy is reduced by increasing fiber
content the same as by lowering the input power to epoxy
without fiber.
14
Lee and Springer [8] have developed a model of
electromagnetic waves interactions with organic matrix
composites in a wave guide to calculate the complex
dielectric constants by measuring the reflectance and
transmittance. They have also developed a thermochemical
model [9] in order to predict the temperature distribution,
degree of cure, resin viscosity, resin content and void
sizes during microwave processing of continuous fiber
reinforced organic matrix composites. The model is verified
by testing Fiberite 82/91348 glass epoxy and Hercules
AS/3501-6 graphite epoxy composites using a 2.45 GHz , 700
watt microwave oven (Litton model 1290).
Karmazsin and Satre [6] have studied the continuous and
pulsed microwave processing of epoxy system and also have
compared its thermomechanical properties with that of
conventional heat polymerization samples. They have used a
2.45 GHz rectangular wave guide along with a 75 watt
continuous or an equivalent (150 watt with 50% cyclic ratio)
pulsed microwave power.supply. The epoxy system studied is
resin AY 103 with HY 991 as hardener and the conventional
heat polymerization cycle used is 1 hr. at 373 K. They have
reported that the thermomechanical behaviour of the samples
depend on the frequency of the pulsed microwave energy. Tg
values of the various microwave cured and conventional heat
cured samples are reported to be of the same order. The
15
Young’s modulus of the samples polymerized under microwave
environment of 75 watt for 600 seconds are reported to be
10% greater than the Young’s modulus of the sample cured
thermally. The authors have not reported the various degrees
of polymerization in different microwave and conventional
heat polymerization processes.
Sheppard and Senturia [11] have demonstrated a
relationship between the decrease in the relaxed
permittivity and the consumption of polar reactive groups
during the curing of DGEBA (Epon 825) with DDS. Isothermal
extent of conversion versus time results from DSC are fit
to a kinetic model, which is then used to predict reactive
group concentrations. Microdielectrometry is used to
measure the dielectric constants at frequencies between 0.1
Hz to 10 MHz. They have proposed a combined
kinetic/dielectric model which has shown good agreement
between experiments and model values up to 70% conversion.
They have also determined dipole moments for the reactive
groups by fitting the data to an empirically modified
Onsager relation for the permittivity.
2.2.2 IRIRRFAQE EEEEQIS URDER MICROWAVE RADIATIO :
Composite properties are dependent on the fiber and
matrix properties and the specific interactions at the
fiber-matrix interface. It is well established that the
16
fiber-matrix interface gives fiber composites their
structural integrity and strength. Load is transferred from
matrix to fiber through the interface and the interphase
region determines the type and level of adhesion between
fiber and matrix. The interface in a fiber-matrix composite
is common to both fiber and matrix. The interphase exists
from some point in the fiber where the local properties
begin to change from fiber bulk properties, through the
actual interface into the matrix where the local properties
again equal the bulk properties. Thus it has physical and
mechanical properties which are neither those of the fiber
nor those of the matrix. Within this region, various
components of known and unknown effect on the interphase can
be identified. The fiber may have morphological variations
near the fiber surface which are not present in the bulk of
the fiber. The surface area of the fiber can be much greater
than its geometrical values because of pores and cracks
present on the fiber surface. The atomic and molecular
composition of the fiber surface can be quite different from
the bulk of the fiber. Surface treatments can add or remove
surface chemical groups giving rise to a chemically and
structurally different region. Fiber coatings to improve the
compatibility of the fiber and the matrix is a major source
for causing variations in the interphase region. Both
chemical and physical bonds exist at the interface and
causes the structure of the matrix in the interphase region
to be different from the bulk. Further unreacted matrix
17
components and impurities can diffuse to the interphase
region altering the local structure. The various
characteristics of this interphase region are best
illustrated by a schematic model shown in fig. 1 [16].
Characterization of interface region can be done on the
basis of single fiber tests and multifiber or composite
tests [44]. Several methods has been used to obtain a
measure of the stress state and the strength of the bond at
the interface. These methods can be divided into two groups.
One group deals with direct measurements and involves model
studies with a single fiber in a matrix casting. The other
group involves indirect measure of the bond strength at the
interface and involves testing of actual composites. single
fiber tests have the advantage that they are easy to perform
and require small quantities of materials. Sample
preparation is generally inexpensive. A disadvantage is that
the single fiber test is possibly not indicative of the
performance of an actual composite. Composite tests on the
other hand will give a good indication of the expected
composite performanceobut these tests are expensive, time
consuming and require large quantities of materials.
Two common single fiber tests are the fiber pull out
test and the fiber critical length test. A third category of
single fiber tests are compression tests.
18
thormot , mocMnIcat
and chemical
onvlronmonts
- matrix
morpho'oov
- unreacted
epochs
- lmpurltlos
- voids
- wrtaco
chuntstry
- topography
- tlbor
morphology
Figure 1. Schematic model of the epoxy -
reinforcement interphase highlighting the
possible components, size and imposed
envrronments acting in this region. [16]
19
Fiber Pull Out:
In a fiber pull out test, a fiber is embedded in a very
thin film (normally about 0.5 mm) [17] or head [18] or a
disc [19] of polymer. In the bead test, the bead is held
between two shearing plates (fig. 2) and the fiber is
pulled out. In the disc test, the force is applied on the
disc (fig. 3) to cause debonding between fiber and matrix.
The force required to remove the fiber is measured with a
load cell. The bonding strength is calculated by dividing
the measured load by the area of contact of the fiber with
the polymer. The area of contact is normally measured in a
scanning electron microscope.
Fiber Critical Length Test:
In the fiber critical length experiment, a single fiber
is embedded in a polymeric matrix [20-28]. The specimen is
then subjected to increasing strain. Tensile stress applied
to the specimen is transmitted to the fiber through shear
stresses at the fiber-matrix interface. When the shear
stress exceeds the local tensile strength of the fiber, the
fiber breaks into fragments inside the specimen (fig. 4).
The fiber breakage continues with the increasing strain
until the fiber fragments become so small that the matrix
can no longer transfer stresses over the fragments to
further fracture the fiber. The length of the broken fiber
20
Figure 2. Arrangement of shear debonding (top) and
enlarged schematic of a resin droplet on a
fiber under the shearing plates (bottom).
[18]
LOAD
I
FORCI /80N0 MK .
0N SIAIt FRKZUON
DISK /
amnauwmwum
Figure 3.
Typical load-displacement curve for button
type test. [21]
21
Ic
2 to
Figure 4. Diagram of fiber critical length
experiments. [22]
22
fragment is referred to as the fiber critical length (1c).
The fiber critical length is an indication of the ability of
the polymeric matrix to transfer stress to the fiber .
Determination of fiber fragment length to diameter ratio
allows an interfacial shear stress r to be determined
according to [28]
where of is the fiber tensile strength at the fragment
length 1C for the fiber diameter d.
Due to random defects in the fiber, fiber fragments in
an experiment will have a range of lengths. The lengths of
the broken fragments will range from one half of the
critical length to the critical length. Ohsawa, et at. [23]
have used a simple average to calculate lc from the average
fiber length la as follows
Drzal, et a1. [24] have taken a statistical approach and
have used a Weibull distribution to describe the fragment
lengths.
In order to calculate the interfacial shear stress in
23
the fiber critical length experiment, the tensile strength
of the fiber at the critical length must be known. However,
the strength of the fiber depends on the flaw distribution
of the fiber [31,32]. The longer the length of the fiber,
the more defects it will have, and the defects will lower
its strength. Rich and Drzal [25] have measured the strength
of carbon fibers at their critical lengths (zO.5 mm).
Single Fiber Compression Test:
Single fiber compression test methods are used to
determine the shear strength and the tensile strength at the
interface. Two typical compression specimens are illustrated
in Fig. 5 [20]. The parallel sided specimen in Fig. 5(a)
represents a block of resin containing a short fiber along
the central axis. When this specimen is compressed along the
axis parallel to the fiber, shear stresses are created at
the ends of the fiber because of the different elastic
properties of the fiber and the matrix. The shear strength
rs of the interface bond can be determined as
where "C is the applied compressive stress causing
debonding at the ends of the fiber.
The neck-down specimen in Fig. 5(b) is used to measure
hi1
\
\
\
¥------o-—-4 no
Figure 5.
(a)
2i:
Single fiber compression test specimen to
measure (a) shear strength (b) tensile
strength of interface. [20]
(I))
Figure 6.
L— r
L...—
1......1:
Tests on unidirectional laminae to measure
(a) interlaminar shear strength and (b)
transverse. [37]
25
the the tensile strength of the interface. When this
specimen is compressed the difference in the Poisson’s ratio
of the fiber and the matrix results in a tensile stress in
the center of the neck perpendicular to the fiber matrix
interface which is given by:
ac ("m-Vf) Ef
2
(1+vm) Ef + (1-uf-2uf ) Em
where um and uf are the Poisson’s ratios of the matrix
and the fiber respectively, 0C is the neck section
compressive stress at which debonding occurs and E is
Young’s modulus. Broutman [21] used these models to measure
the interfacial bond strength. in boron fiber-epoxy
composites. He found that the tensile bond strength was
approximately 800 psi and the shear about 8000 psi.
According to these findings, the bond shear strength is ten
times greater than the tensile.
Indirect Methods:
An indirect approach to measure the interface bond
strength is to test unidirectional laminae in such a way
that failure occurs in a shear mode parallel to the fibers
or in a tensile mode normal to the fibers. Two such tests,
the interlaminar shear strength test and the transverse
strength test, are illustrated in Fig. 6 [37]. The
26
interlaminar shear test is a three point bending test and
requires careful selection of span to depth ratio S/a (Fig.
6a) to insure shear failure along xx’ rather than tensile or
flexural failure at point 0. The transverse tensile test
shown in Fig. 6b is simple but requires great care for
reliable data. Both the interlaminar shear strength and the
transverse tensile strength depend on the fiber volume
fraction. Results obtained using the interlaminar shear
strength test are reported by Goan and Prosen [33] and
Daniels, et al. [34]. Daniels, et al. found that the fiber
microstructure, as reflected by its modulus and the type of
fiber surface treatment, affects interlaminar shear strength
and therefore influences the interfacial bond strength. An
indication of the interlaminar shear strength dependence on
fiber modulus for composites with untreated fibers is
illustrated by Goan and Prosen [33].
There is no single method universally accepted to
characterize the interface. Depending on the properties of
the materials and the situation, one can choose a particular
method or combinationoof methods to evaluate the interface.
In the microwave field, most of the experiments are carried
out from the electrical point of view to determine the
dielectric properties of polymeric systems. There is no
archival literature on the interaction of microwave
radiation with fiber-matrix interface or with the individual
fibers itself.
EXPERIMENTAL
3 . 1 APPROACH
Adhesion between fiber and matrix was evaluated by
single fiber critical length tests [22]. When a single fiber
specimen is subjected to an increasing tensile load, the
fiber starts fragmenting until the remaining fiber fragments
are shorter than the crirical length. Determination of the
fiber critical length to diameter ratio allows the
interfacial shear strength to be determined according to:
where is the fiber tensile strength at the fragment
”f
length 1C for the fiber diameter d.
Due to random defects in the fiber, fiber fragments have
a range of lengths. It has been shown that the critical
length data fit the two parameter Weibull distribution [24]
27
28
F(x) = 1-exp - - : x>0
where a and B are the two parameters of the distribution
and can be evaluated by maximum likelihood methods (see
attached computer code). Using these two parameters, a mean
value of interfacial shear strength can be calculated as
follows:
a
Var (r) = f F (1- —) - r2 (1- —)
419 a a
Valuable information regarding the stress transfer
between the fiber and the matrix can be obtained by
examining the fractured ends of a fiber fragment under
polarized light. The axial normal stress and the interfacial
shear stress distribution in a broken fiber fragment is
shown in Fig. 7 [26]. Failure modes at the fiber tip have
been reviewed by Mullin and co-workers (27, 28). If the
fiber-matrix adhesion is poor, failure will occur at the
interface as shown in Fig. 8(b). If the matrix is brittle
and the adhesion between fiber and matrix is high, matrix
cracking will occur with fiber failure as shown in Fig.
8(a). If a ductile matrix is used and the fiber-matrix
adhesion is high, the matrix will fail by shear as shown in
29
Hem. .ucmemmuu
Honwu coxoun m c“ cowusnfluumfio mmouum
Hmonm Hmflommumu:w 0cm mmouum Hmfiuoc Hmfixd
.b ousaflm
J mm
.mm
If. i w...
m. m"
two N
5.». H
~f\\\\\\||lllliy .J1\\\\|IIIIIIIVHHH
8.3182063 6335.
30
Fig. 8(c).
Evaluation of the interface in microwave cured specimens
as compared with thermally cured specimens requires that the
bulk matrix properties be the same in both cases. In order
to achieve this the following steps were followed:
1.Establish a database of mechanical properties and extent
of cure for the matrix cured thermally under a standard cure
cycle.
2.Develop a microwave cure cycle which would produce the
same matrix properties as the baseline data.
3.Evaluate and compare the interfacial shear strength and
failure mode of various fibers in thermally cured and
microwave cured (under the above determined cure cycle)
specimens.
3.2 MATERIALS
The materials used in this study were chosen very
carefully to facilitate the detection of different
mechanisms of microwave interactions at the interphase.
31
A
a iL
7
High energy radial
crack normal to fiber
[W—
Interface unbonding due to high
shgar stress at newly formed
on S
é). 166%?
c EN -
Low energy resolved shear
'stress induced tensile
cracks in the matrix
Figure 8. Diagram of possible failure modes in the
fiber critical length experiment (a)
matrix cracking, (b) frictional stress
transfer, (c) shear stress transfer. [27]
32
3.2.1 THERMOSET RESIN SYSTEM:
The matrix material used was an epoxy resin based on
diglycidyl ether of Bisphenol—A made by Dow Chemicals (DER
331). This resin cured with the stoichiometric amount of
metaphenylenediamine (mPDA) was shown to be very suitable
for single fiber critical length tests [24]. The chemical
structure of the resin is:
43..-...{.QEQS_C.._I._C.JI00:04....1m2...
The curing agent has the following chemical structure:
NH,
l
0
Physical properties of DER 331 cured with the
\
NH,
stoichoimetric amount of mPDA are listed in table 1. The
dielectric constant and the dissipation factor of thermally
cured DGEBA with mPDA at various frequencies and
temperatures are shown in fig. 9 and fig. 10 respectively
whereas the values for other diamines are listed in table 2
[36]. A typical value of dielectric constant of cured epoxy
resin is 3.8 and that of loss factor is between 0.01-0.08.
Jow et al. [35] have reported the on-line dielectric
33
Table 1. Material properties of DER 331 cured with
stoichiometric amount of mPDA. [26]
PROPERTY DER 33]
E], GPa (M51) 3.8 (0.55)
£2, GPa (M51) 3.8 (0.55)
012 0.35
023, GPa (M51) 1.4 (0.20)
a], 10'5/°c (10'5/°r) 68 (32)
a2, 10’6/°c (10’5/°F) 68 (32)
5.50 10’
C05
/ .1
5.00 V 104
y//// ,0.
E 3.12“" ‘1 g/ 10.
:3 4.50 \\ . -— —-———- . r<—— 10'
U
.2! /
o L / ‘ _______________——- 10“
/
3.50 /
L ___——-0-/
3.00 1 ‘ 1 I I 1 1
20 40 60 80 100 120 140 160
Temperature. ’C
Figure 9. Dielectric constant vs. temperature of
DGEBA cured with mPDA.
[36]
35
3 1o5
‘6
.13
C
.9
E
.9
i3
0004
0002
0001
20 40 60 80 100 120 140 160
Temperature. ‘C
Figure 10. Dissipation factor vs. temperature of
DGEBA cured with mPDA. [36]
36
Table 2. Electrical properties of DGEBA cured with
aromatic diamines. [36]
Eutectic
Property M I’DA M DA DA DPS blend
Arc resistance, seconds ............. 98 97 ...... 100
Dielectric strength. volts/mil ........ 400 - 420 410 430
Volume resistivity, ohm-em ......... 1 x 10“ 0.18 x 10" 2 x 10" >2 X 10'
Surface resistivity. ohms ........... >4 x 10'” >4 x 10" 4 x 10' >4 X 10'
Dielectric constant at 25"C :
At 60 cps ...................... 4.42 4.35 4.22 4.20
At 10’ cps ...................... 4.34 4.27 4.15 4.13
At 10' cps ...................... 3.80 3.72 3.94 3.58
Dielectric constant at 80“C :
At 60 cps ...................... 4.74 4.61 4.71 4.47
At 10’ cps ...................... 4.70 4.58 4.68 4.43
At 10‘ cps ...................... 4.33 4.22 4.35 3.86
Dielectric constant at l00"C :
At 60 cps ...................... 4.70 4.58 4.71 4.72
At 10' cps ...................... 4.68 4.56 4.62 4.66
At 10‘ cps ...................... 4.42 4.35 4.38 4.12
Dissipation factor at 25"C:
At 60 cps ...................... 0.0068 0.0066 0.0044 0.0055
At 10' cps ...................... 0.0183 0.0172 0.0128 0.0133
At 10‘ cps ...................... 0.0346 0.0352 0.0268 0.0318
Dissipation factor at 80°C :
At 60 cps ...................... 0.0038 0.0029 0.0084 0.0335
At 10' cps ...................... 0.0035 0.0025 0.0059 0.0523
At 10‘ cps ...................... 0.0353 0.0358 0.0273 0.0352
Dissipation factor. 100°C °
At 60 cps ...................... 0.0030 0.0021 0.0177 0.0407
At 10’ cps ...................... 0.0032 0.0022 0.0076 0.0627
At 10‘ cps ...................... 0.0268 0.0260 0.0214 0.0422
37
properties of the reacting DGEBA and DDS system under
microwave environment. Some of their results are shown in
fig. 11.
3.2.2 FIBERS:
Based on dielectric properties and surface chemistry
three different fibers were chosen for reinforcement.
3.2.2.1 Glass Fiber:
Unsized E-glass fibers made by Owens-Corning Fiberglas
Corp. (lot no. 456, BG-740) were used in this work. Physical
properties of this fiber are listed in table 3. E-glass
fiber has a surface rich in hydroxyl groups and absorbs
water. The surface of the glass fiber contains randomly
distributed groups of oxides [37]. Some of the oxides, such
2, Fe203, and A1203 are non-hygroscopic and adsorb
water both as hydroxyl groups and as molecular water which
as $10
is held to the hydroxyl groups by hydrogen bonding. Other
oxides are hygroscopic and when water is adsorbed at the
surface they become hydrated. Thus, glass picks up water
very rapidly to form a well-bonded surface layer which may
be many molecules thick. The presence of moisture on the
fiber surface along with the fiber’s low dielectric
dissipation factor could produce some interesting effects in
the interphase region under microwave environment. Glass
5.51%.?
9'
1
‘1 3131331910
. ‘0.
.33
(.9) 1110151103 3131391310
38
“mm“ .nmo mm4.~ am noon; ms.m «o H0>0H
um3oa wanna amaua>m ca um man one man
mmo now was» mcfluao o>azouafis .m> ucaumcoa
ofluuaaaofio xmamsoo can annuaummemu mafiaico .HH magmas
AmmSEEV mE:
.Ov .mm .0” .ON . N .0— .OF .0 .O
. s s _L p p p — h p p s b s u p p _ s p . p p . h p b L s p p s — h p n p .
. OAuN
OLNI 2% r
. dem
0.? ..
1 .%
IO.OO_.
i meagoumdrcap
(go) elnialadwei
39
TABLE 3
Properties of A84, E-glass and Kevlar-49 Fibers Reported in Literature
Property Units A84 E-glass Kevlar-49
Diameter pm 8 10 11.9
Density 103Kg/m3 1.80 2.54 1.45
Young's modulus GPa 235 76 125
Modulus (perpend- GPa 21 76 _
ieular to fiber axis)
Tensile strength GPa 3.58 3.45 3.6
Elongation to frac. % 1.53 4.8 2.6
Coeff. of thermal 10‘6/C(10'6/r) -o.11(-0.5) (2.8) -2
expansion parallel parallel parallel
8.54(4) _ 59
radial radial radial
Thermal conductivity me'C-1 105 1.04 0.04
(parallel to axis)
Specific heat Cal/g C 0.22 0.197 _
Btu/1b F
Poisson's ratio 0.25 0.22 0.33
40
fiber is non conductive and its dissipation factor ranges
between 0.002-0.005 at 1 MHz.
3.2.2.2 Kevlar Fiber:
An aramid fiber Kevlar-49 manufactured by DuPont (lot
no. 123481, 7200 denier) was used as the second
reinforcement. Kevlar-49 fiber comes supplied with a surface
finish. The fiber was washed with absolute ethanol according
to the manufacturer’s recommendations prior to use in order
to remove the finish. The washing procedure involved three
successive washes in absolute ethanol with six hours of
soaking in between each wash and then drying in a vented
oven for 2 hrs. 0 125 °C.
Kevlar is a non conducting fiber. Its dielectric
constant is 3.8 and the dissipation factor lies between
0.0002-0.001. The fiber composed of 500 to 700 nm fibrils
[38], has a surface that is relatively inert because the
surface atoms are covalently bonded into the polymeric
structure. Kevlar being very hygroscopic, can absorb upto 3
wt% water [40]. Some of the physical properties of Kevlar-49
are listed in table 3.
41
3.2.2.3 Carbon Fiber:
Unsized carbon fiber AS4-12K made by Hercules (lot no.
638-4B) was used as the third reinforcing fiber. Physical
properties of this fiber are listed in table 3. Carbon fiber
is a very good conductor of electricity and has a very high
value of loss factor.
Carbon fiber has a highly active surface and readily
. absorbs gases which affect the surface properties [37]. A
range of active functional groups like carboxylic, lactone,
carbonyl, phenolic,and hydroxyl (fig. 12, [36]) are present
on the surface. Additional species besides oxygen are also
present on the fiber surface [41]. Nitrogen in the form of
amine or cyano groups is almost always present on the low
heat treatment temperature fiber surface. Traces of elements
such as silicon and iron can also be present. Drzal [42] has
identified sodium, trapped in the lower modulus fibers from
the earlier polymer fiber spinning steps, as being present
on the fiber surface and has shown that the sodium has the
ability to diffuse to the fiber surface from the bulk of the
fiber under moderate elevated temperature conditions. Thus,
there are a large number of sites for chemical bonding and
there is a large area of contact with the resin.
42
Figure 12. Schematic illustration of the potential
surface chemical groups which have been
found on the surface of carbon fibers.
[41]
43
3.3 THERMAL CURING
Single fiber specimens of epoxy were prepared with the
help of a silicone RTV-664 eight cavity mold. Standard ASTM
63.5 mm (2.5") dogbone specimen cavities with a 3.175 mm
(1/8") wide x 1.59 mm (1/16") deep x 25.4 mm (1") long gage
section were molded into a 7.62 cm (3") x 20.32 cm (8") x
1.27 cm (1/2") thick silicone piece. 0.8 mm (1/32") deep
sprue slots were molded at both the ends of each dogbone
cavity for aligning the fiber axially in the dogbone. Single
filaments of desired fiber were taken out of a fiber bundle
and placed tightly in the sprues of a dogbone cavity with
rubber cement such that the fiber runs across the cavity
axially.
Epoxy resin (DER 331) and the stoichiometric amount
(14.5 phr) of mPDA were weighed in separate beakers. These
beakers were then transferred to an oven and heated at 65 0C
until the mPDA melted. Then the epoxy resin and the curing
agent were mixed thoroughly and degassed in a vacuum oven
along with the silicone mold at 65 oC and 29" Hg vacuum. The
degassed resin and curing agent mixture was then poured in
all the cavities of the mold with the help of a drip rod and
to a level just above the height of the mold surface. The
assembly was then transferred to an electronically
temperature controlled air circulating oven which was
44
preprogrammed for a curing cycle of 2 hrs @ 75 °C followed
by 2 hrs 0 125 0C. At the end of the cure cycle, the oven
temperature was reduced to room temperature at the given
rate. The cured specimens were then taken out of the
silicone mold and the extra material was sanded off the
surface of the specimens in order to get smooth and uniform
specimens. The samples were then stored in a dessicator
until ready for analysis.
3.4 MICROWAVE CURING
Coupling of microwave energy to polymer specimens was
achieved by interactions between a sustained electric field
in a coupling applicator and the dipoles contained in the
polymer system. The schematic of the experimental setup is
shown in fig. 13. There are four basic components in this
setup:
(1) a microwave power source
(2) transmission lines
(3) microwave applicator
(4) temperature measuring device
A cylindrical brass cavity [43] of 7" inner diameter
covered by two transverse brass shorting planes, constructed
in M80 machine shop, was used as a microwave applicator
(fig. 14 (a,b)). The cavity length could be adjusted by
45
.eauuafio
uwaaufio Eoumam a>mzou0fis Hmucaewuomxm .mH shaman
8.02 830..
88.. 588.5 28. .3880
8.8.8.2
38368.2
.225 032...;
8.950 38.825
3. I
8.980 38.825
83 .82.:
8.8.3.5.
8.9.5.2 8328......
03308.2 .833...
882 .28.. 8.0.). 830d .
52.8% 2... one...
0....quth 2.83.3...
Figure 14(a) Cylindrical brass cavity for microwave
curing.
(U
'4'
r4
a.)
Li
:1
U"
"-1
in
47
Hun. .Amnoud coaumuaoxo
any nusousu mommaa ocean o u m an». >ua>aa
H80wupcfiaxo on» ma 30fl> Hac0fluaamimmouo An.va madman
L .l J
_ m 7 m
_
i _ N/i
F
_ P
F
M _ 4 WK
11.1 _ _i
~11 7.4
i _
j_ _ n l i.
r... e. .r
_
_
_ F.
.80.. u
48
moving the top plane of the cavity through gears while the
bottom plane was .fixed but could easily be removed. The
cavity wall and the bottom had several diagnostic holes.
Microwave power was coupled into the cavity through an
adjustable excitation probe. The coaxial probe had an outer
conductor of 1.27 cm. diameter and an inner conductor of
0.442 cm. diameter and was located 3.81 cm. above the bottom
plate (fig. 15).
A magnetron based fixed frequency (2.45 GHz) generator
(Opthos MPG-4) with adjustable duty cycles from 5% to 85%,
and power output from 0 to 120 watts, was used as the
microwave source. For stability considerations, the power
source was operated at higher power levels and the desired
power was channelled through a 10 db directional coupler. A
circulator was used to protect the source from the reflected
power. 50 ohm impedence coaxial cables were used to transmit
the power from the source to the cavity. Two 20 db
directional couplers were used to decouple the incident and
reflected signals. Both incident and reflected signals were
attenuated and measured directly by power meters.
Silicone dogbone cavities filled with the epoxy mixture
were placed in the center of the bottom of the cavity and in
alignment with the excitation probe. Due to non-uniformity
of the fields inside the cavity, only one dogbone specimen
was cured at a time. A fluoroptic temperature measuring
Figure 15. Adjustable microwave excitation probe.
50
system (Luxtron Model 750) equipped with four channel
measurements was used to continuously monitor the
temperature of the reacting mixture at different locations.
The cavity was tuned initially to critically couple the
microwave energy to the loaded material in the lowest order
made TE by moving the the top plate of the cavity and
111’
the excitation probe. Afterwards the cavity was tuned
continuously throughout the experiment as the dielectric
properties of the reacting mixture inside the cavity were
changing continuously.
3.5 MATERIAL PROPERTIES
Specimens cured thermally or under the microwave
environment were tested for thermal and mechanical
properties. The Extent of conversion (a) and glass
transition temperature (T9) were determined using
Differential Scanning Calorimeter (DSC DuPont 9900). DSC
scans of cured and uncured matrices were made at 5 oC/min
under nitrogen purge using open pans. In a DSC plot of heat
flow vs. temperature, the area underneath the curve of a
partially cured material is the heat of reaction required
for complete conversion. The extent of conversion can be
calculated as follows:
Heat of react. for part. cured
Extent of conversion = 1 -
Heat of react. for uncured matl.
51
Mechanical properties, including tensile strength and
tensile modulus were determined using a servohydraulic
tensile tester (Material Testing System or MTS 880). ASTM
Dogbone specimens were loaded in the hydraulic grips using
specially designed supplementary grips to protect the
specimens. The load-elongation curves were recorded on a
chart recorder at a strain rate of 0.02"/min (about 2%/min).
In calculating the strain from crosshead displacement, the
gage length of the sample was taken as the distance between
the supplementary grips (25.8 mm.). Young’s modulus was
calculated from the initial slope of the load-elongation
curve whereas the tensile strength to failure was recorded
from the machine itself.
Dynamic mechanical properties such as flexural modulus
(E’) and loss tangent (tan 6) of cured specimens were
determined using a Dynamic Mechanical Analyser (DMA, DuPont
9900). Fixed frequency (1 Hz) DMA scans were made at 10
oC/min and 0.2 mm. amplitude. Glass transition data were
also obtained from the DMA scans.
I
3.6 SINGLE FIBER CRITICAL LENGTH TEST
Single fiber dogbone specimens (fig. 16) of the desired
fiber were prepared and cured thermally or with microwaves
as discussed above. Once fabricated, these specimens were
52
$2R
5 /
/ 5
SECTION A-A
Figure 16. Schematic diagram of a single fiber
interfacial shear strength specimen.
53
then tested for fiber critical lengths. Prior to testing,
each specimen was examined visually and under the microscope
for voids, fiber breaks, additional fibers and other foreign
materials, and other fabrication defects which would affect
the testing and the results. The testing procedure involved
the careful straining of the specimen in a hand operated
tensile testing fixture [24], designed and built in the
Composite Materials and Structures Center laboratory and
Michigan State University machine shop. The fixture was
capable of applying enough load to the specimen to cause
fracture and incorporated a dial gage to give an indication
of the amount of strain. The fixture was then mounted on the
stage of a microscope (Olympus BH-2). The clarity of the
image was improved by using two cover glasses along with the
proper refractive index fluid, one on the top surface of the
specimen and one on the bottom surface, to eliminate the
effect of specimen roughness.
The specimen was then subjected to an increasing tensile
load and fiber breaks were monitored in-situ through the
microscope. When the. specimen reached a state of constant
number of fiber breaks irrespective of the increasing load,
the length of each fiber segment was measured using a
A calibrated filar eyepiece. Fiber diameter was also measured
using the same filar eyepiece. The fiber fractures were then
examined under transmitted polarized light and
photomicrographs were taken using the attached camera
assembly to the microscope.
54
55
RESULTS AND DISCUSSION
4.1 MATRIX PROPERTIES
Baseline mechanical properties of the thermally cured
specimens under the curing cycle of 2 hrs @ 75 °C followed
by 2 hrs. @ 125 0C are listed in table 4. For microwave
processing, specimens are cured at various power levels and
durations and their temperature-time profiles are shown in
fig. 17(A,B,C). In the beginning of the process, the
temperature increases very rapidly as the dipoles are free
to rotate and thus the coupling of microwave energy is very
efficient. As the reaction progresses, the network starts
to form and as a result, dipoles are no longer free to
rotate. Power absorption decreases and temperature decreases
slowly.
Orientation of dogbone specimens either parallel or
perpendicular to the excitation probe is crucial in
microwave curing as the fields inside the cavity are not
uniform. Experiments are performed by aligning the specimens
56
.mufi>aa o>asonowe on» mowmcfi
casuaucofluo ucauouuflo spa; mcosfiaaam some
can Hmm man no moaauoum ofifluiouzumuamaoa Aa.ha anamfim
2:3 m2:
mm om m: 0.. m
h p p p F b b
— b P 1— » h b b p _ p p p
QUE OH 4wj