Composites Science and Technology 63 (2003) 283–293

www.elsevier.com/locate/compscitech

Dynamic mechanical analysis of banana fiber

reinforced polyester composites
Laly A. Pothana, Zachariah Oommenb, Sabu Thomasc,*
a
Department of Chemistry, Bishop Moore College, Mavelikara 690110, Kerala, India
b
Department of Chemistry, CMS College, Kottayam, Kerala, India
c
School of Chemical Sciences, Mahatma Gandhi University, Priyadarshini Hills PO 686560 Kottayam, Kerala, India

Received 13 November 2001; received in revised form 10 September 2002; accepted 24 September 2002

Abstract
The dynamic mechanical analysis of banana fiber reinforced polyester composites was carried out with special reference to the
effect of fiber loading, frequency and temperature. The intrinsic properties of the components, morphology of the system and the
nature of interface between the phases determine the dynamic mechanical properties of the composite. At lower temperatures (in
the glassy region), the E0 values are maximum for the neat polyester whereas at temperatures above Tg, the E0 values are found to be
maximum for composites with 40% fiber loading, indicating that the incorporation of banana fiber in polyester matrix induces
reinforcing effects appreciably at higher temperatures. The loss modulus and damping peaks were found to be lowered by the
incorporation of fiber. The height of the damping peaks depended on the fiber content. When higher fiber content of 40% was used,
an additional peak in the tan
curve, pointing to micro mechanical transitions due to the immobilized polymer layer was evident.
The glass transition temperature associated with the damping peak was lowered up to a fiber content of 30%. The Tg values were
increased with higher fiber content. Cole–Cole analysis was made to understand the phase behavior of the composite samples. A
master curve was constructed based on time–temperature super position principle, which allows the prediction of long-term effects.
Apparent activation energy of the relaxation process of the composites was also analyzed. The value was found to be maximum for
composites with 40% fiber content.
# 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Banana fiber

1. Introduction Dynamic mechanical test methods have been widely

Manufacturing high performance engineering materi-
als from renewable resources is one ambitious goal cur-
rently being pursued by researchers across the world.
The ecological benefits of renewable raw materials are
clear: they save valuable resources, are environmentally
sound and do not cause health problems. Natural fibers
have already established a track record as simple filler
material in automobile parts. Natural fibers like sisal,
jute, coir, oil palm fiber have all been proved to be good
reinforcement in thermoset and thermoplastic matrices
[1–4]. Our earlier studies have proved banana fibers to be
an effective reinforcement in polyester matrix [5].
* Corresponding author. Tel.: +91-481-598303; fax: +91-481-
561190.
E-mail address: sabut@md4.vsnl.net.in (S. Thomas).
0266-3538/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0266-3538(02)00254-3
employed for investigating the structures and viscoelastic
behavior of polymeric materials for determining their
relevant stiffness and damping characteristics for various
applications. The dynamic properties of polymeric
materials are of considerable practical significance when
determined over a range of temperature and frequencies.
Composite damping property results from the inher-
ent damping of the constituents. This can be represented
as [6]
tan
c ¼ Vf tan
f þ ð1  Vf ÞTan
m ð1Þ
where tan
c, tan
f, and tan
m are the damping values
of the composite, the fiber and the polymer, respectively
and Vf and Vm are the volume fraction of the fiber and
the matrix, respectively. In the transition region, the
molecular chains begin to move and every time a frozen
segment begins to move its excess energy is dissipated as
284 L.A. Pothan et al. / Composites Science and Technology 63 (2003) 283–293
heat. A segment, which is completely frozen in or is free
to move, gives rise to a low damping value. The rule of
mixtures, which is used in predicting composite
mechanical properties, can be used in analyzing the
composite damping property as well, [7] According to
Nielsen, the damping of a composite, tan
c is given by
the proportional contribution of the matrix according
to its relative content.
tan
c ¼ tan
m ð1  f Þ
where tan
m is the damping value of the matrix and f
is the volume fraction of the filler. Around Tg, the filler
particles are not deformed and damping is purely due to
the polymer. But, a specific interaction between the filler
and the polymer layer will create a layer of immobilized
polymer in between the filler and the polymer. In such
cases the Eq. (2) can be rewritten as
tan
c ¼ tan
m
Work done on the dynamic mechanical properties of
fibrous composite materials are mainly aimed at two
objectives.
1. To study the chemical or physical modification of
the matrix by the introduction of fiber.
2. To study the behavior as a function of the
properties of each phase.
Gassan and Bledzki [8] carried out dynamic mechan-
ical analysis of jute/epoxy composites and observed an
improvement in dynamic modulus by the incorporation
of treated jute fiber in epoxy. Finegan and Gibson [9] in
their investigation on the enhancement of damping in
polymer composites have suggested analyzing different
fiber matrix combinations. Saha et al. [10] made com-
parative studies on the damping of unmodified and
chemically modified jute- polyester composite samples.
Valea et al. [11] investigated the influence of cure con-
ditions and the exposure to various chemicals on the
dynamic mechanical properties of several vinyl ester and
unsaturated polyester resins containing glass fiber.
Exposure to aromatic solvents was found to modify the
viscoelastic character of these materials. Amash and
Zugenmaier [12] reported the effectiveness of cellulose
fiber in improving the stiffness and reducing the damp-
ing in poly propylene–cellulose composites. The effect of
adsorbed water on the storage modulus and loss tangent
of wood was analysed by Obataya et al. [13] Adsorption
of water was found to increase the modulus value up to
8% moisture content and to decrease the values there-
after. The dissolved water was found to decrease mod-
ulus values by plasticizing matrix molecules.
Extensive research work is being carried out by Tho-
mas and co-workers in this laboratory regarding the
viscoelastic behavior of various polymer composites and
blends [14–18].
In the present communication we report on the influ-
ence of banana fiber on the viscoelastic properties of
polyester. The effect of fiber content, frequency and
temperature on the viscoelastic properties is reported.
The elevation of Tg is taken as a measure of the inter-
facial interaction and the effect of fiber content on the
Tg values is reported. The Tg is usually interpreted as
the peak of the tan
or the loss modulus curve that is
ð2Þ obtained during a dynamic mechanical test conducted
at a low frequency. The Tg values of the different sam-
ples and the shift in Tg were determined from the loss
modulus and the tan
curves to have more insight into
the fiber/matrix adhesion. The effect of interlayer on the
micro mechanical transitions is also reported.
2. Experimental
ð3Þ
2.1. Materials
Banana fiber was obtained from Sheeba Fibers and
Handicrafts, Poovancode, Tamil Nadu. Unsaturated
polyester HSR 8131 based on isophthalic acid was
obtained from Bakelite Hylam, Hyderabad. Methyl
ethyl ketone peroxide and cobalt naphthenate were of
commercial grade and obtained from Sharon Engineer-
ing Enterprises, Cochin. The details of the material are
given in Table 1.
2.2. Preparation of composites
Banana fibers were neatly separated by hand; adher-
ing pith, if any, were removed and the fibers were cut to
uniform size of 30 mm length. The fibers were then
evenly arranged, in a mold measuring 1501503 mm3
and pressed into a mat. Composite sheets were prepared
by impregnating the fiber with the polyester resin to
which 0.9 volume percent cobalt naphthenate and 1%
methyl ethyl ketone peroxide was added and the air
bubbles were removed carefully with a roller. The closed
mold was kept under pressure for 12 h; samples were
cured at room temperature for 12 h, and then taken out
of the mould. The demolded samples were post cured
Table 1
The mechanical properties of banana fiber
Sample Diameter Initial Young’s SDInitial Tensile %
number of fiber modulus (GPa) Young’s strength Strain
(mm) modulus (GPa) (MPa)
1 50 32.703 8.190 779 2.7
2 100 30.463 4.689 711 2.4
3 150 29.748 8.561 773 3.5
4 200 27.698 7.083 789 3.3
5 250 29.904 4.059 766 3.2
 L.A. Pothan et al. / Composites Science and Technology 63 (2003) 283–293
for a further period of 24 h at room temperature and
samples of dimension 100153 mm3 were cut.
A dynamic mechanical thermal analyzer M K I I of
Polymer Laboratories was employed for dynamic
mechanical property evaluation of the composites. The
experiment was performed under tensile mode at a fre-
quency of 0.1, 1 and 10 Hz. The sample dimension was
23 cm2 and the gauge length of the sample 1.5 cm. The
testing temperature ranged from 20 to 160  C and the
heating rate was 2 K/min.
Composites with different fiber loading were exam-
ined using scanning electron microscope (SEM) model
Jeol JSM-35 C and Cambridge 250 MK3 stereo scan.
3. Results and discussion
3.1. Effect of temperature
Fig. 1 shows the effect of temperature on the dynamic
modulus of the neat polyester and various composites
with different fiber content. Variation in modulus occurs
due to the effect of the incorporated fibers. In the case of
neat isophthalic polyester cured with MEK peroxide
and cobalt naphthenate, it is seen that the dynamic
modulus is higher than the fiber filled system in the
glassy region. The lowering of the modulus values
occurs due to dilution of the polymer. However, the
dynamic modulus curve showed an increase in the E0
value above the Tg region in the rubbery plateau. The
increase of E0 in the rubbery plateau is maximum for the
composites with 40% fiber loading. The effectiveness of
fillers on the moduli of the composites can be repre-
sented by a coefficient C such as [19]

Table 2
E’g =E’r comp
C¼
ð4Þ
E’g =E’r resin
Where E0 g and E0 r are the storage modulus values in
the glassy and rubbery region respectively. The higher
the value of the constant C, the lower the effectiveness
of the filler. The measured E0 values at 45 and 130  C
Fig. 1. The effect of temperature on the dynamic modulus of the neat
polyester and various composites with different fiber content.
285
(for polyester) were employed as E0 g and E0 r respec-
tively. The values obtained for the different systems at
frequency 0.1 Hz are given in Table 2. In this case the
lowest value has been obtained for 40% fiber loading
and the highest value for 20% fiber loading. The effec-
tiveness of the filler is the highest at 40% fiber loading.
It is important to mention that modulus in the glassy
state is determined primarily by the strength of the
intermolecular forces and the way the polymer chains is
packed. Fig. 2 shows the variation of log E0 with fiber
volume fraction at different temperatures of 40,60,90
and 145  C. In all the cases, the E0 value decreases up to
a volume fraction of 0.25 and then increases.
There is a large fall in modulus with increasing tem-
perature in the unfilled system, the stiffness at high
temperature being determined by the amorphous
regions, which are very compliant above the relaxation
transition. The drop in the modulus on passing through
the glass transition temperature is comparatively less for
reinforced composites than for un-reinforced resin. In
other words, banana fibers have larger effect on the
modulus above Tg than below it. However the difference
between the moduli of the glassy state and rubbery state
is smaller in the composites than in the neat polyester.
This can be attributed to the combination of the
hydrodynamic effects of the fibers embedded in a vis-
coelastic medium and to the mechanical restraint intro-
duced by the filler at the high concentrations, which
reduce the mobility and deformability of the matrix.
Other authors have also reported similar observations.
[20]. At low temperature, E0 values of matrix and com-
posite are found to be close to each other emphasizing
The value of the constant C
Sample C
10% 0.97
20% 0.97
30% 0.96
40% 0.91
Fig. 2. The variation of log E0 with fiber volume fraction.
286 L.A. Pothan et al. / Composites Science and Technology 63 (2003) 283–293
that at low temperatures fibers do not contribute much
to imparting stiffness to the material. At higher tem-
peratures any water molecules adhering on to the fiber
will get evaporated making the fiber stiffer. Air-dried
cellulose is rather resistant to mechanical impact and a
large amount of mechanical energy has to be spent in
order to destroy the macroscopic and microscopic struc-
ture [21]. This ultimately contributes to the improved
modulus of the composite at high temperatures.
Tan
is a damping term that can be related to the
impact resistance of a material. Since the damping peak
occurs in the region of the glass transition where the
material changes from a rigid to a more elastic state, it
is associated with the movement of small groups and
chains of molecules within the polymer structure, all of
which are initially frozen in. In a composite system,
damping is affected through the incorporation of fibers.
This is due mainly to shear stress concentrations at the
fiber ends in association with the additional viscoelastic
energy dissipation in the matrix material. Another rea-
son could be the elastic nature of the fiber. Fig. 3
delineates the effect of temperature on tan d. Improve-
ment in interfacial bonding in composites occurs as
observed by the lowering in tan d values. The higher
the damping at the interfaces, the poorer the interface
adhesion. SEM photographs of banana polyester com-
posite with fiber volume fraction of 10, 20 and 40% are
shown in Fig. 4a b and c. It is interesting to note that
there is better fiber/matrix bonding in composites with
40% fiber loading. Whereas fiber/matrix de-bonding is
evident in composites with 10 and 20% fiber loading,
composites with 40% fiber loading show no gap
between the fiber and the matrix because of the stron-
ger bonding. When the fiber concentration is lower, the
packing of the fibers will not be efficient in the compo-
site. This leads to matrix rich regions and thereby easier
failure of the bonding at the interfacial region. When
there is closer packing of the fibers crack propagation
will be prevented by the neighboring fibers. The effec-
tive stress transfer occurs in the case of composites with
Fig. 3. Effect of temperature on the tan d value of the composite at
different fiber loading.
40% loading. The variation of tan
with temperature
of the composites has been analyzed with respect to
fiber loading and frequency. Incorporation of fibers
reduces the tan d peak height by restricting the move-
ment of the polymer molecules. Magnitude of the tan

peak is indicative of the nature of the polymer system.
In an unfilled system, the chain segments are free from
restraints. Addition of fibers above 30% show two
peaks making the two phases, fiber and matrix distinct.
The appearance of two peaks can also be explained as
due to a microgel structure cross-linked to a general
matrix structure. The micro gel particles could develop
at ‘hot spots’ where there were initially small excesses
of the cross-linking agent or catalyst reagent. Addition
of fiber decreases the Tg value at low fiber loading,
showing that the addition of fiber below 30% has only
a plasticising effect. However, at 40% fiber loading, the
Tg values show a positive shift, stressing the effective-
ness of the fiber as a reinforcing agent i.e. the high
polymer- filler interaction. The result is consistent with
the E0 values obtained. However, the introduction of
filler reduces the magnitude of the tan
peak more,
than shifting the temperature. The shifting of Tg to
higher temperatures can be associated with the
decreased mobility of the chains by the addition of
fibers. Elevation of Tg is taken as a measure of the
interfacial interaction. In addition, the stress field sur-
rounding the particles induces the shift in Tg. Chua [22]
concluded based on his studies that a composite with
poor interface bonding tends to dissipate more energy
than that with good interface bonding. At high fiber
content when strain is applied to the composite, the
strain is controlled mainly by the fiber in such a way
that the interface, which is assumed to be the more
dissipative component of the composite, is strained to a
lesser degree [23]. The width of the tan d peak also
becomes broader than that of the matrix. The behavior
suggests that there are molecular relaxations in the
composite that are not present in the pure matrix. The
molecular motions at the interfacial region generally
contribute to the damping of the material apart from
those of the constituents [24]. Hence the width of the
tan
peak is indicative of the increased volume of the
interface. Table 3 shows the peak width at half height
of the samples from the damping curve. The peak width
Table 3
Peak height and peak width at half height of the tan
curves
Sample Peak height Peak width
(cm) (cm)
Poly 10 6.4
10% 5.9 6.5
20% 6.3 6.5
30% 5.3 6.4
40% 4 9
 L.A. Pothan et al. / Composites Science and Technology 63 (2003) 283–293 287

Fig. 4. (a) SEM of the composite with 10% fiber loading showing fiber/matrix debonding and fiber pull out. (b) SEM of the composite with 20%
fiber loading showing fiber/matrix debonding and matrix cracking. (c) SEM of the composite with 40% fiber loading showing good fiber/matrix
adhesion.

Table 4
Peak height and peak width at half height of the E00 curves

Sample Peak height Peak width

(cm) (cm)

Gum 7 8.5
10% 5.8 8
20% 5.4 8
30% 5.2 8.3
40% 4.8 12

is found to be maximum for composites with 40% fiber

loading. Increase in the concentration of the filler
increases the interface [25]. Stress induced motions may
also occur in the composite. The hydrogen bonds
between segments may rupture under stress. After such
a bond is broken, a small amount of molecular motion
can occur which makes it difficult for the same bond to
reform. As a result, a new hydrogen bond forms, which
initially carries little, if any stress. The net result is
energy dissipation and mechanical damping [26].
Fig. 5. The variation of loss modulus with temperature of composites
with different fiber loading.
The loss modulus E00 is defined as the stress 90  C out
of phase with the strain divided by the strain; it is a
measure of the energy dissipated or lost as heat per cycle
of sinusoidal deformation, when different systems are
compared at the same strain amplitude. It is in fact the
viscous response of the material. The loss factors are
288 L.A. Pothan et al. / Composites Science and Technology 63 (2003) 283–293
most sensitive to the molecular motions. Fig. 5 shows
the variation of loss modulus with temperature of com-
posites with different fiber loading. It can be seen from
the figure that the loss modulus peak values decrease
with increase of fiber content at temperatures below the
glass transition. The effect of the filler is prominent
above the glass transition temperature in this case also.
The modulus values increase with fiber content above
the glass transition temperature. Another interesting
result that is observed is the broadening of the loss
modulus curve when the fiber content is increased to
40%. Fig. 6 shows the plot of peak height vs. fiber
volume fraction. The peak height shows a regular
decrease with increase of fiber content. At a fiber load-
ing of 40%, the most pronounced effect of the filler has
been the broadening of the transition region as the fiber
concentration increases. The observed broadening may
be explained as due to the difference in the physical state
of the matrix surrounding the fibers to the rest of the
matrix fiber, matrix and immobilised polymer layer
matrix [27]. As reported by other authors, a shell of
immobilised polymer surrounds the fibers [28]. Fig. 7
shows schematic diagram of fiber, matrix and the
immobilised polymer layer. The higher the volume
fraction of matrix, the more restraint at the interfaces.
The different physical state of the matrix surrounding
the fiber hinders the molecular motion. This can be
taken as the inter layer which causes the additional
Fig. 6. The plot of peak height vs. fiber volume fraction.
Fig. 7. Schematic diagram showing the fiber, matrix and the immobi-
lised polymer layer.
transition [29]. The increase in width of the loss mod-
ulus curve is taken to represent the presence of an
increased range of order. The greater constraints on the
amorphous phase could give rise to a higher or broader
glass transition behaviour. The peak width at half
height of the composites with different fiber volume
fraction is given in Table 4. The maximum peak width is
found to be for the composites with maximum fiber
content (40%)
3.2. Effect of frequency
The storage modulus, loss modulus and damping
peaks have been found to be affected by frequency. The
variation of E0 with frequency of neat polyester as a
function of temperature is shown in Fig. 8. Increase of
frequency has been found to increase the modulus
values. Fig. 9 shows the effect of frequency on the
dynamic modulus of samples with 40% fiber loading.
Frequency has a direct impact on the dynamic modulus
especially at high temperatures. The modulus values are
Fig. 8. The variation of E0 with frequency of neat polyester as a
function of temperature.
Fig. 9. The effect of frequency on the dynamic modulus of samples
with 40% fiber loading.
 L.A. Pothan et al. / Composites Science and Technology 63 (2003) 283–293 289

found to drop at a temperature around 60  C. The drop 40% loading. The nature of the tan
curve is affected
in modulus value continues steadily till a temperature by the incorporation of fiber. Eklind and Maurer
of 120  C is reached. The molecular motion can be [30,31] have reported on an interlayer model to simulate
believed to be set in at 60  C. The glass transition tem- the dynamic mechanical properties of filled blends. The
perature of the neat polyester is around 130  C and the filler particles have been reported to be surrounded by
modulus value remains unchanged after that. The an interlayer attached to the filler surface. This phe-
change in dynamic properties is also associated with nomenon could give rise to filler structure in the matrix
crazing and formation of microscopic cracks and voids. able to alter the dynamic mechanical modulus. The
At high temperature breaking up of the fiber agglom- bound polymer results in a layer with properties differ-
erates and breaking up of the bond between the fiber ent from the bulk properties of the pure polymer. This
and polymer phases may also occur [26]. The lowering micro mechanical transition caused by the interlayer
of the modulus peak is maximum for the neat polyester resulted in a new tan
peak occurring at a temperature
due to the development of microscopic cracks in the lower than the Tg of the pure matrix. The tan
curve in
unfilled resins. Frequency is seen to have a direct this system is found to have two peaks at 40% fiber
impact on the tan
values as well. The viscoelastic loading. This can be believed to be due to the micro-
properties of a material are dependent on temperature, mechanical transition arising from the immobilised
time and frequency. If a material is subjected to a con- polymer layer, which acts as the interlayer. The effect of
stant stress, its elastic modulus will decrease over a the interlayer becomes prominent only at high fiber
period of time. This is due to the fact that the material loading. Increase of frequency is found to have a
undergoes molecular rearrangement in an attempt to broadening effect on the tan delta curve. Broadening of
minimize the localized stresses. Modulus measurements
performed over a short time (high frequency) result
thus in higher values whereas measurements taken over
long times (low frequency) result in lower values [26]. In
this system also, the modulus measurements over a
range of frequencies have been studied. Higher values
were observed for measurements made over a short
time. The tan d values measured over a range of fre-
quencies for the neat polyester samples are shown in
Fig. 10. The tan d peak is found to shift to higher tem-
perature with increase of frequency. The damping peak
is associated with the partial loosening of the polymer
structure so that groups and small chain segments can
move. The tan d curve peak, which is indicative of the
glass transition temperature, is also indicative of the
degree of cross-linking of the system. Fig. 11 shows the Fig. 11. Effect of frequency on the tan delta curve of composites with
effect of frequency on the tan
curve of samples with 40% loading.

Fig. 12. Effect of frequency on the loss modulus curve of neat

Fig. 10. Effect of frequency on the tan d curve of neat polyester. polyester.
290 L.A. Pothan et al. / Composites Science and Technology 63 (2003) 283–293

Table 5
Values of tan
maximum E00 maximum and Tg values of neat polye-
ster and banana fiber reinforced polyester composites at different fiber
loading

Fibre loading Tan
Max Tg from tan
Max

Frequency (Hz) Frequency (Hz)

0.1 1 10 0.1 1 10

Poly 0.421 0.423 0.452 104.92 114.38 125.22

10% 0.248 0.241 0.253 99.44 106.25 119.231
20% 0.262 0.257 0.277 102.619 112.21 122.22
30% 0.228 0.207 0.233 93.5 106.95 120.02
40% 0.220 0.214 0.224 118.49 125.27 133.36
E00 Max107(Pa) Tg from E0 ( C)
7.982 7.979 8.005 85.42 95.08 108.70
7.822 7.802 7.825 85.95 95.82 103.62
Fig. 13. Effect of frequency on the loss modulus curve of composites 7.822 7.768 7.762 85.95 97.35 108.48
with 40% fiber loading. 7.751 7.717 7.713 85.95 95.82 101.02
7.753 7.683 7.729 85.95 116.29 122.22

the curve is due to some kind of heterogeneity in the
network structure. This broadening is more prominent
in composites with high fiber content. Addition of fiber
increases the free volume between monomeric units.
The introduction of fibers, which in turn affects the
curing reaction, affects the molecular motions and dif-
fusion, as well.
Fig. 12 shows the effect of frequency on the loss
modulus values of neat polyester samples. The peak of
the loss modulus curve is seen to be shifted to higher
temperatures with increase of frequency. Fig. 13 shows
the effect of frequency on the loss modulus curve of the
samples with 40% fiber loading. The E00 peak of the
composite is broader revealing the morphological rear-
rangement resulting in a highly plasticized amorphous
region and also the improved interaction between the
fiber and matrix. The loss modulus curves in addition
show two peaks. These very well support the micro
mechanical transition observed in the tan
peaks.
Eklind and Maurer [31] have reported on glass transi-
tion peak on the low temperature side for both loss
modulus and tan
curves. It is interesting to note that
the micro mechanical transition becomes prominent
only at relatively high fiber content indicating the cri-
tical fiber volume fraction and the related interlayer
responsible for the micro mechanical transition. Table 5
shows the tan
max and E00 max values and the corre-
sponding Tg values for the different composites. The
values of Tg obtained from the loss modulus peaks are
found to be lower than that obtained from the damping
peaks. The plasticization results from the addition of
fiber within the polyester matrix. The loss modulus peak
has a double humped nature when fibers are incorpo-
rated. At lower frequencies the peak of the curve is
around 120  C. With increase in frequency, the peak of
the loss modulus peak, which corresponds to the glass
transition temperature, is also found to be shifted to
higher temperature.
Since the response of the sample changes with both
temperature and frequency of oscillation, it was decided
to make a three dimensional thermo gram for compo-
sites with different fiber loading. Fig. 14 shows the
three-dimensional thermo gram. The different peaks are
clearly visible in the thermo gram.
3.3. Cole–Cole plots
The magnitude of polarisation within a material is
represented by dielectric constant, which can be repre-
sented by the Debye and Onsager [32] equations. The
single relaxation peaks are inadequate to describe the
viscoelastic response of polymers. Cole–Cole is a parti-
cular treatment of dielectric relaxation data obtained by
plotting E 00 against E00 , each point corresponding to one
frequency [32].
Structural changes taking place in cross-linked poly-
mers after fiber addition to polymeric matrices can be
studied using the Cole–Cole method. The dynamic
mechanical properties when examined as a function of
temperature and frequency are represented on the Cole–
Cole complex plane,
E ¼ fðE0 Þ ð5Þ
Fig. 15 shows the Cole–Cole plot, where the loss
modulus data log E00 are plotted as a function of the
storage modulus log E0 . The nature of the Cole–Cole
plot is reported to be indicative of the nature of the
system. Homogeneous polymeric systems are reported
to show a semi circle diagram [26]. The Cole–Cole dia-
gram presented in the figure are imperfect semi circles
the shape of the curve points towards the relatively
good fiber–matrix adhesion.
 L.A. Pothan et al. / Composites Science and Technology 63 (2003) 283–293
Fig. 14. Three dimensional thermogram of composites with 40% fiber loading.
Fig. 15. Cole–Cole plot of the composites with different fibre loading.
3.4. Master curve
Polymers obey the superposition principle, i.e.;
dynamic modulus is a function of temperature as well as
frequency (time). The storage modulus of the composite
sample with 40% fiber loading as a function of time at
different temperatures was determined. Time t is related
to the frequency f by the following equation
t ¼ 1=2f ð6Þ
Because changes in time appear to be equivalent to
changes in frequency or time, the process of developing
a master curve is called time–temperature superposition.
Curves obtained at different temperatures were
superposed by horizontal shifts along a logarithmic time
scale to give a single curve covering a very large range of
time. Such curves made by superposition, using a refer-
291
Fig. 16. Master curve of the composites with 40% fiber loading.
ence temperature will be able to cover times outside the
range easily accessible by practical experiments. The
curve made by superposition is called a master curve.
On a logarithmic plot, the storage modulus at a refer-
ence temperature, TR and frequency !R equals the
storage modulus at temperature T and frequency !.
E0 ðlog !R ; TR Þ ¼ Gðlog !; TÞ ð7Þ
The difference between log !R and log ! is constant
equal to log aT where aT is called the temperature shift
factor. In the log E0 vs. log t graph, the shift factor was
calculated as follows
log aT ¼ log t  log t0 ð8Þ
where t0 is the reference temperature i.e. in the current
case 100  C.
292 L.A. Pothan et al. / Composites Science and Technology 63 (2003) 283–293
Table 6
Activation energy values of neat polyester and banana polyester com-
posites under different loading
Fibre Activation
loading energy (kJ/mol)
Poly 25.708
10% 21.031
20% 23.836
30% 22.75
40% 41.16
The method of relating the horizontal shifts along the
log time scale to temperature changes as developed by
Williams, Landel and Ferry is known as the W-L-F
method [32]. The amount of horizontal shift of the time
scale is given by ‘aT’. Use of the master curve and the
variations of the shift factors with temperature allow the
prediction of the storage modulus at frequencies and
temperatures that are not experimentally accessible. In
developing the master curve, the reference temperature
was randomly chosen as 100 and ‘aT’ values were found
out. Fig. 16 shows the master curve.
Any increase in vibrational frequency causes the glass
transition temperature to rise and the amplitude of the
damping peak to increase. The shift of the transition
temperature allows one to calculate the apparent acti-
vation energy of the relaxation process for each of the
samples assuming a linear equation of the type [33].
log f ¼ log f0 þ H=2:303RT ð9Þ
Where f0 is an experimental constant, f and T are the
measuring frequency and the temperature for the dis-
persion peak respectively, and R is a gas constant, and
H the activation energy.
By means of extrapolation, the glass transition tem-
perature Tg can be calculated. Activation energy of the
different composite samples was calculated from the
Arrhenius relationships [26]. The activation energy
values are given in Table 6. The activation energy values
of the composites with 40% fiber loading are the max-
imum. The activation energy values for neat polyester
samples are 25.7 kJ/mol. At low fiber loading, as is evi-
dent from the SEM of the samples, the fiber/matrix
adhesion is low and the activation energy is also low.
However, incorporation of the critical fiber volume
fraction brings about high interfacial interaction and
effective stress transfer. This increases the activation
energy value.
4. Conclusions
Dynamic mechanical properties of short banana fiber
reinforced polyester composites are greatly dependent
on the volume fraction of the fiber. The dynamic mod-
ulus shows a decrease with incorporation of fiber below
the glass transition temperature and has a positive effect
on the modulus at temperatures above Tg. The max-
imum improvement in properties is observed for com-
posites with 40% fiber loading, which is chosen as the
critical fiber loading. Increase of frequency shifts the Tg
to higher temperatures supporting the good fiber/matrix
interaction, which is clear from the SEM.
At the maximum possible fiber loading in this study i.e.
40% loading, the loss modulus peak gets broadened
emphasizing the improved fiber/matrix adhesion. More-
over, an additional peak occurs at high fiber loading in
the tan
curves, due to the interlayer effect. Addition of
fiber lowers the tan
peak height, which again points to
the improved fiber/matrix adhesion. The glass transition
temperature is shifted positively on the addition of fiber.
Cole–Cole plots show an imperfect semicircle showing
the heterogeneity of the system as well as the good
interfacial adhesion at high fiber loading. Activation
energies calculated for the different systems show the
highest value for 40% loading. Use of master curve and
the relationship of the shift factors are useful in pre-
dicting the long-term behavior of the composite.
Acknowledgements
One of the authors (L.A. Pothan) would like to thank
the Science engineering and Technology Department of
the Government of Kerala for financial assistance
received for carrying out the project.
References
[1] Joseph K, Thomas S, Pavithran C. Effect of chemical treatment
on the tensile properties of short sisal fiber-reinforced poly-
ethylene composites. Polymer 1996;37:5139–45.
[2] Varma IK, Ananthakrishnan SR, Krishnamoorthi S. Comp of
glass/modified jute fabric and unsaturated polyester. Composites
1989;20:383.
[3] Geethamma VG, Thomas Mathew K, Lakshminarayanan R,
Thomas S. Composite of short coir fibers and natural rubber:
effect of chemical modification, loading and orientation of fiber.
Polymer 1998;39:1483.
[4] Sreekala MS, Kumaran MG, Thomas S. Oil palm fibers: mor-
phology, chemical composition, surface modification and
mechanical properties. J Appl Poly Sci 1997;66:8–821.
[5] Pothan LA, Thomas S, Neelakantan NR. Short banana fiber
reinforced polyester composites: mechanical, failure and aging
characteristics. J Reinf Plast Comp 1997;16:744.
[6] Richardson MOW, Polymer engineering composite. Applied
Science Publishers 1977.
[7] Nielsen LE. Mechanical properties of polymers and composites.
New York: Marcel Dekker; 1974.
[8] Gassan J, Bledzki AK. Possibilities of improving the mechanical
properties of jute/epoxy composites by alkali treatment of fibers.
Compos Sci Technol 1999;59:1303–9.
[9] Finegan IC, Gibson RF. Recent research on enhancement of
damping in polymer composites. Comp Strs 1999;44:89–98.
 L.A. Pothan et al. / Composites Science and Technology 63 (2003) 283–293
[10] Saha AK, Das S, Bhatta D, Mitra BC. Study of jute fiber rein-
forced polyester composites by dynamic mechanical analysis. J
Appl Poly Sci 1999;71:1505–13.
[11] Valea A, Gonzalez ML, Mondragon I. Vinyl ester and unsatu-
rated polyester resins in contact with different chemicals: dynamic
mechanical behaviour. J Appl Poly Sci 1999;71:21–8.
[12] Amash A, Zugenmaier P. Morphology and properties of iso-
tropic and oriented samples of cellulose fiber–polypropylene
composites. Polymer 2000;41:1589–96.
[13] Obataya E, Norimoto M, Gril J. The effects of adsorbed water on
dynamic mechanical properties of wood. Polymer 1998;39:14.
[14] George S, Neelakantan NR, Varghese KT, Thomas S. Dynamic
mechanical properties of isotactic polypropylene/nitrile rubber
blends: effects of blend ratio, reactive compatibilization, and
dynamic vulcanization. J Poly Sci Part B Polymer Physics 1997;
35:2309–27.
[15] Varghese H, Bhagawan SS, Someswara Rao, Thomas S. Mor-
phology, mechanical and dynamic mechanical properties of
blends of nitrile rubber and ethylene vinyl acetate copolymer. Eur
Poly J 1995;31(10):957–67.
[16] George J, Bhagawan SS, Thomas S. Thermogravimetric and
dynamic mechanical thermal analysis of pineapple fiber rein-
forced polyethylene composites. J Ther Analy 1996;47:1121–40.
[17] Varghese S, Kuriakose B, Thomas S. Mechanicaland viscoelastic
properties of short sisal fiber reinforced natural rubber compo-
sites: effects of interfacial adhesion, fiber loading and orientation.
J Adhes Sci Technol 1994;8:234.
[18] Joseph K, Pavithran C, Thomas S. Dynamic mechanical proper-
ties of short sisal fiber reinforced polyethylene composites. J
Reinf Plast Comp 1993;12:139.
[19] Tan JK, Kitano T, Hatakeyama T. J Crystallisation of carbon
fiber reinforced polypropylene. Mater Sci 1990;25:3380–4.
[20] Marcovich NE, Reboredo MM, Aranguren MI. Mechanical
properties of woodflour unsaturated polyester composites. J Appl
Poly Sci 1998;70:2121–31.
293
[21] Klemm D, Philipp B, Heinze T, Heinze U, Wagenknecht W.
Comprehensive cellulose chemistry. Wiley–VCH; 1998.
[22] Chua PS. Dynamic analysis studies of interphase. Poly Comp
1987;8:308.
[23] Ibarra L, Macias M, Palma E. Viscoelastic properties of short
carbon fiber thermoplastic (SBS) elastomer composites. J Appl
Poly Sci 1995;57:831.
[24] Dong S, Gauvin R. Application of dynamic mechanical analysis
for the study of the interfacial region in carbon fiber/epoxy com-
posite materials. Poly Comp 1993;14:414.
[25] Landel RF. Mechanical properties of polymers and composites.
New York: Marcel Dekker, Inc; 1994.
[26] Murayama T. Dynamic mechanical analysis of polymeric mate-
rials. 2nd ed. Amsterdam: Elsevier; 1978.
[27] Joshi M, Maiti SN, Misra A. Studies on the thermal, dynamic
mechanical and rheological behavior of short-glass-fiber-rein-
forced composite based on poly(butylene terephthalate) high
density polyethylene blends. Polymer 1994;35:17.
[28] Thomason JL. Investigation of composite interphase using
dynamic mechanical analysis—artifacts and reality. Poly Comp
1990;11:2–105.
[29] Maurer FHJ. Controlled interphases in composite materials. New
York: Elsevier; 1990. p.491 [Ishida H, editor].
[30] Eklind H, Maurer FHJ. Micromechanical and microdielectric
transitions in P(S-g-EO) modified PPO/PMMA blends. Polymer
1997;38:1047.
[31] Eklind H, Maurer FHJ. Frequency and temperature depen-
dence of molecular and micromechanical transitions in PPO/
PMMA/(S-g-EO) blends. J Poly Sci Part B Phys Ed 1996;34:
1569.
[32] Aklonis J J, MacKnight W J. Introduction to polymer viscoelas-
ticity, (John Wiley, ed.); 1983.
[33] Folkes MJ. Short fiber reinforced thermoplastics. New York:
Wiley; 1982.