You are on page 1of 8

Composites Science and Technology 70 (2010) 21682175

Contents lists available at ScienceDirect

Composites Science and Technology


journal homepage: www.elsevier.com/locate/compscitech

Evaluation of residual strains in epoxy with different nano/micro-llers using


embedded ber Bragg grating sensor
M. Lai a, K. Friedrich b,c, J. Botsis a,, T. Burkhart b
a

Ecole Polytechnique Federale de Lausanne (EPFL), LMAF-IGM-STI, CH-1015 Lausanne, Switzerland


Institute for Composite Materials (IVW GmbH), Technical University of Kaiserslautern, Erwin-Schrdinger-Str. 58, 67663 Kaiserslautern, Germany
c
CEREM, College of Engineering, King Saud University, Riyadh, Saudi Arabia
b

a r t i c l e

i n f o

Article history:
Received 23 April 2010
Received in revised form 23 August 2010
Accepted 26 August 2010
Available online 15 September 2010
Keywords:
A. Particle-reinforced composites
B. Mechanical properties
B. Thermal properties
C. Residual stress
Fiber Bragg grating

a b s t r a c t
The inuence of SiO2 nanoparticles and rubber micro-llers on the mechanical and thermal responses of
an epoxy based composite is investigated using classical quantitative thermo-mechanical testing (tensile
tests, DMTA, TMA), microstructural analysis (Micro-CT, TEM, SEM microscopy) as well as distributed optical sensing in order to determine different residual strain elds generated during processing. The results
show that the tensile modulus of the compounds increases with the addition of SiO2 and decreases with
the rubber content, following estimates of the HashinShtrikman model. The coefcient of thermal
expansion appears to be insensitive to the particles content in the temperature range investigated.
The residual strains generated during processing are inuenced by the rubber content that introduces
a strong relief, with respect to the one generated by the neat resin, whereas the silica content tends to
increase their level.
2010 Elsevier Ltd. All rights reserved.

1. Introduction
Epoxy resins are versatile because of the large number of potential epoxy resin/curing agent combinations, each one of them giving a different molecular structure of the nal thermoset material.
They have widespread application, one of which is their use as
matrices of reinforced composites materials. Epoxies have in general good mechanical and thermal properties, however, highly
cross linked epoxy systems are usually brittle, which limit their
usefulness in applications requiring high impact and fracture
strengths. As a consequence, the search for toughened epoxy
matrices has become the subject of numerous recent publications,
and several formulations have been developed [14].
Epoxy resins can be substantially toughened by the addition of a
rubbery phase, although the improvement in toughness appears
commonly to be accompanied by a signicant loss in elastic modulus, strength and glass transition temperature [5,6]. Rigid llers
such as alumina [7], titanium dioxide [8] and clay [9] have been
added to the epoxies to increase their stiffness, strength and thermal properties but they have also demonstrated a favourable
toughening effect. It is possible to increase the toughness of an
epoxy resin by up to 100% using selected rigid llers, normally
sized in the range of nanometers. However, this toughening effect
seems to be moderate when being compared with the toughness of
Corresponding author.
E-mail address: john.botsis@ep.ch (J. Botsis).
0266-3538/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compscitech.2010.08.019

epoxy matrices modied with rubber, and normally it is not


enough to cover the requirements of high demanding applications.
Considering the benets reported from the rubber and rigid
ller modications, recent research converges on the formulation
of hybrid modied epoxy matrices, containing rigid llers and a
rubber toughening phase, as a possibility to obtain toughened
epoxy resins without loss of other important properties [1012].
Since an epoxy resin has excellent solvent resistance, mechanical
properties, good adhesion to many substrates and electrical properties, it is also widely used in the elds of coatings, adhesives,
castings, materials for electrical insulators, etc.
It is well known that residual strains/stresses in epoxy resins
cured at high temperature are produced by the shrinkage occurring
in the cooling process from the cure temperature to room temperature [13,14]. These residual stresses often result in a reduction of
the adhesive or cohesive strength of the resin and occasionally
induce cracks in the casted materials [15]. In order to reduce the
residual stresses in a cured epoxy resin, also here submicron
polymer particles were dispersed therein prior to curing [16,17].
Led by the need to understand and reduce the process induced
internal stresses on the nal products, many investigations were
carried out resulting in the development of both destructive and
non-destructive measurement techniques. The rst group is represented by techniques like hole-drilling [18,19], ring core [20], slitting, layer removal and crack compliance methods [21,22],
originally developed for metals. The procedure is generally expensive and time consuming thus providing information on the

2169

M. Lai et al. / Composites Science and Technology 70 (2010) 21682175

residual stress elds only in a limited amount of points. Nondestructive techniques such as based on Moir interferometry
[23], photo-elasticity [24] and other non-contact full-eld deformation measurement techniques have been proposed, however, a
relatively large specimen is required.
The present work is concentrated on investigating easy to process rubber toughened and nanoparticles reinforced epoxy matrices, with possibly enhanced stiffness/toughness balance as well
as good thermal properties, and with low residual stresses. The
microstructure of these composites is observed by electron microscopy and the mechanical properties are evaluated using standard
mechanical testing procedures. Cylindrical specimens with a
centrally located optical ber with a long ber Bragg grating
(FBG) are produced from the composites. The shrinkage as a result
of curing, produces compressive strains on the FBG which is
recorded by means of the optical low coherence reectometry
(OLCR)-based technique [25]. Such strains allow to determine the
residual strains and to evaluate the coefcient of thermal
expansion (CTE) of certain epoxy ller combination.
2. Experimental
2.1. Material components
The epoxy resins used are all Diglycidyl ether of Bisphenol A
(DGEBA) modied with Hexadioldiglycidyl ether. The three
materials used in this study are: (a) CeTePox VP 823-30R, neat
epoxy resin (EP), delivered by Chemicals and Technologies for
Polymers GmbH [26]. (b) CeTePox VP 823-33R: This master batch
epoxy resin is a suspension containing SiO2 nanoparticles
(Chemicals and Technologies for Polymers GmbH). The dispersed
phase consists of surface-modied, spherical nano-SiO2 with
diameters under 50 nm and an extremely narrow particle size
distribution, obtained from a solgel chemical process. The
particles are agglomerate-free and homogeneously distributed in
the epoxy resin. This behaviour permits the resin to maintain a
relatively low viscosity even at a high ller content [27]. (c) Albidur
EP 2240 A: Master batch epoxy resin, modied with reactive
silicone rubber (SR) micro-particles by Nanoresins GmbH. The elastomer fraction is a special silicon rubber addition with a particle
size of 0.13 lm, nely distributed as a separate phase in the uid
resin. Albidur EP 2240-A is totally compatible with other epoxy
resins and also with the most diverse curing agents. The silicone
elastomer particles have an organic shell structure comprising
reactive groups. The reaction curing rules the hydrophobicity of
the nal particles and can be used to adjust the polarity of the
formulation [28,29]. The main properties of these materials are
summarized in Table 1. The curing agent is Cetepox V 823-2H
and is based on cycle aliphatic polyamine, also supplied by

Chemicals and Technologies for Polymers GmbH [30] (Table 1).


All materials are commercial industrial products and were used directly as supplied, without any additional treatment.
2.2. Compounding and preparation of samples
Nanocomposites and hybrid compounds, with variable content
of ceramic and rubber, were produced by dilution or mixing/dilution of the corresponding master batches with neat epoxy. The
compositions of the mixtures are shown in Table 2. As stated
above, in the case of SiO2, the master batch was a commercial
product obtained by a solgel technology [31].
Nanocomposites and hybrid compounds were prepared using a
dissolver (Wilhelm Niemann GmbH, Germany), operated at
200 rpm, 60 C and vacuum during 1 h (Fig. 1a). Once the epoxy
mixtures were nished, the stoichiometric amount of curing agent
was incorporated to the resin maintaining the temperature at 60 C
and stirring at 50 rpm, during 10 min. Afterwards the reactive mixture was placed into preheated steel molds (plate geometry) and
cured in two stages to minimize the residual stresses.
The same mixtures were also casted in a specially designed
vertical mould (Fig. 1b) in which a cylindrical shape cavity was
machined to produce cylindrical specimens of 12 mm in diameter
and 40 mm in height. Optical bers with a diameter of 125 lm
were aligned along the cylinders axis by a ber holder and kept
straight by the help of small magnets. The FBG sensors, inscribed
by Avensys Tech Company, having a total length of 23 mm, were
placed in such a way that part of the sensor, 5 mm, was outside of
the specimen (Fig. 1c). This conguration allows simultaneous
measurements of strains in the cylinder and ambient temperature.
The curing program used in both cases, i.e. for plate coupons and
the cylindrical specimens, is schematically illustrated in Fig. 2. At
the end of the curing process the resulting plates were machined
in dog-bone like specimens for tensile testing and in rods for DMTA
analysis. The cylindrical specimens with the embedded FBG sensor
were used for measuring the residual strains due to curing and the
thermal strains from which the CTE of the epoxy and the composite
mixtures were calculated.
2.3. Characterization of microstructure and general properties
2.3.1. Microscopy and lCT-analysis
Transmission Electron Microscopy (TEM) images were taken by
the use of a LEO 912 microscope, Omega GmbH, Germany, using an
accelerator voltage of 120 kV. Thin sections (app. 100 nm) of the
specimens were cryo-cut with a diamond knife at 120 C and
used without staining. Fracture surfaces of the samples were studied using a scanning electron microscope (SEM) Type Zeiss Supra
40, Germany. The surfaces were sputtered with a Pt/Pd alloy in a

Table 1
Material properties.
Resins
Aspect

CeTePox VP 823-30R
Transparent uid

CeTePox VP 823-33R
Whitish uid

Albidur EP 2240 A
White uid

Equivalent weight (Eq./100 g)


Density (g/cm3)
Particle content (wt.%)
Viscosity 25 C (MPa s)

0.581
1.148
0
1160

0.485
1.249
19.125
1330

0.3180.345
1.081.12
40
30,00045,000

Aspect

Curing agent
Clear colorless liquid

H equivalent weight
Boiling point
Density

Eq./100 g
C
g/cm3

50.7
240
0.931

2170

M. Lai et al. / Composites Science and Technology 70 (2010) 21682175

T3 = 80 C

Table 2
Compositions of the nanocomposites and hybrid compounds in vol.%, and their
designations.
Silicon rubber

0
3
5.5

Nano-SiO2

t3 = 8 h
t1 = 2 h

0_0
3_0
5.5_0

0_3
3_3
5.5_3

0_8
3_8

special chamber, type SCD-050 Balzers, Liechtenstein, over 150 s,


prior to SEM investigation. The interior of the material was examined using a lCT-machine (Computed Micro-Tomography System
3D Brand: Phoenix X-ray systems + services GmbH, Wunstorf,
Germany), that allows to observe cross sections of the samples at
different positions.
2.3.2. Tensile testing
Tensile tests were performed on a 5848 Instron microtester
machine (equipped with a 2 kN load cell) in displacement control
mode at 1 mm/min. The longitudinal elongation was registered
with a SANDNER Messtechnik GmbH clip gage, possessing a
reference length of 14.935 mm. The environmental conditions were
25 C at a relative humidity of 30%. A series of ve dog-bone specimens, having a rectangular narrow section of 6  3.2 mm2 and a
gage length of 40 mm, was tested for each material composition.
2.3.3. Dynamic mechanical thermal analysis (DMTA)
The viscoelastic response of the materials was studied by DMTA
(Eplexor 150 N, Gabo Qualimiter, Germany). Tests were carried out
using rectangular specimens (50  10  4 mm3) under tensile
loading at 10 Hz frequency, a temperature range of 0180 C, and
a heating rate of 2 C/min. The static load amounted to 3 N, superimposed by a dynamic load amplitude of 1.5 N.
2.3.4. Thermo-mechanical analysis (TMA)
The experimental determination of the CTE was carried out by
the use of a TMA 40 from Mettler Toledo. The testing parameters
were: starting temperature 25 C, ending temperature 80 C, heating rate 5 C/min and load 0.1 N.

T4 = 80 C

T0 = 60 C

t2 = 10 min

T1 = 60 C

Gelation

Crosslinking

Fig. 2. Curing program for epoxy and the modied compounds.

2.4. Fiber Bragg grating technique


Optical ber Bragg grating sensors have been used for strain
measurements in several materials [32]. The FBG is the result of
a refractive index modulation of the core of a standard optical ber
which acts as wavelength selector reecting a narrow wavelength
band and transmitting unaltered all other wavelengths when a
broadband light passes though it. The reected signal presents a
peak centred on the Bragg wavelength kB0 that is linked to the
induced refractive index modication by kB0 2neff K0 [32], where
neff is the mean core index of refraction and K0 is the period of the
induced index modulation.
When a uniform grating is subjected to a homogeneous axial
strain ez and/or a uniform temperature change DT, both neff and
K0 are modied resulting in a shift of the Bragg wavelength.
Assuming that the two lateral strains on the ber ex ; ey are related
to the axial one ez by ex ey mf ez (where mf is the Poissons ratio
of the ber), the wavelength difference DkB kB  kB0 obtained
from the peak shift of the spectra, before and after loading, is
DkB =kB0 1  pe ez af nDT, where pe  0.215 [25] is a grating
gage factor measured experimentally, af  8  107 C1 is the CTE
of the glass ber, and n  8.3  106 C1 is the thermo-optic
constant [33]. Note that the last term provides a compensation to
account for the variation of neff and K0 with temperature. If the
strain is constant through out the sensor, the resulting spectrum
appears to be simply shifted to a lower or higher wavelength if
strains are compressive or tensile in nature.
If non-homogeneous strains ez z are applied on the sensor [25],
the reected spectrum is modied and several peaks may appear.
In this case, the Bragg wavelength kB z becomes a function of

Fig. 1. (a) Dissolver, (b) vertical mould, and (c) cylindrical specimen section.

2171

M. Lai et al. / Composites Science and Technology 70 (2010) 21682175

the axial coordinate z and can be retrieved along the sensor by


means of the OLCR-based technique. For a constant DT, the difference between the local Bragg wavelength kB z and the reference
one kB0 is related to the strains through a similar relation shown
earlier,

kB z  kB0
1  pe ez z 1  pe eT af nDT
kB0

For convenience the strain on the FBG is represented by ez z


(i.e., residual strains) and strains due to thermal expansion
mismatch eTz . The optical bers were single mode, acrylate coated,
with Bragg wavelength kB0 = 1300 nm and elastic properties of the
bers were Ef = 72 GPa and mf = 0.19. The coating was removed on
the entire length of the sensor, plus a few mm on both sides, to
avoid an inuence of the coating on the sensors response.
2.5. Heat transfer simulation
To measure the CTE, a uniform temperature across the
specimen is required. As it is explained later in the paper, the
thermal cycle used for the evaluation of the matrices CTEs is
composed of four successive plateaus over which the specimen is
allowed to reach thermal equilibrium before measuring the local
Bragg wavelength. In order to determine the time necessary for
the specimen to achieve the thermal equilibrium with the
surrounding environment, heat transfer simulations on the
specimen geometry were carried out using common thermal properties of the glass ber (f) and epoxy resin (e): thermal conductivity
ke = 0.3 W m1 K1, kf = 1.3 W m1 K1, specic heat capacity
ce = 1000 J kg1 K1, cf = cglass = 810 J kg1 K1 [34]. Natural convection was considered as a boundary condition. The time necessary
to reach at least 95% of the anticipated constant temperature level
was estimated to be in each case lower than 60 min. Although the
properties cited above are for an epoxy, for simplicity the
calculated time is assumed to be the same for all mixtures.

Fig. 3. TEM micrographs of the modied epoxy resins containing: (a) 8 vol.% SiO2
(material 0_8 of Table 2) and (b) 3 vol.% SiO2 + 5.5 vol.% SR (material 5.5_3 of
Table 2).

SiO2

3. Results and discussion


3.1. Microstructure and density of the materials
SR

The hybrid epoxy nanocomposite modied with SiO2 and


silicone rubber is a complex system. Its performance depends
strongly on the quality of the llers distribution. Fig. 3 shows
TEM photographs of the modied epoxy resins. The nanoparticles
are clearly evident in the specimens of both the nanocomposite
and the hybrid compound. They are homogeneously distributed
in the bulk matrix; neither agglomerates nor clusters are present,
and the observed dark dots are the result of overlapping particles
in the depth direction. The presence of rubber particles has no
impact on the SiO2 particles distribution.
Note that it is not possible to see the rubber particles, in Fig. 3b,
because they were taken out of the matrix due to the shear forces
during the specimen preparation (thin lm cutting), however, their
distribution is evident from the location of the micro-holes inside
the epoxy matrix. Rubber particles are also uniformly distributed
in the matrix and they can be described as being spherically
formed, with a broad particle size distribution. This is also visible
on the fracture surface of a broken cylinder of the material 3_8,
showing larger or smaller craters that can be related to the rubber
particles, and much smaller white dots, that represent the nanoSiO2 llers (Fig. 4).
The density is one of the most important properties of the cured
epoxy resins, because it determines the weight of the nal product,
directly related to the cost of the energy consumption during its
operation. The density of the epoxy resins modied with SiO2

1 m

Fig. 4. SEM picture of the fracture surfaces of material 3_8: SR refers to the rubber
particles, and SiO2 represents the nano-llers.

increases with rising nanoparticle content, because silica is denser


than the polymer. Thus, the density of the pure epoxy resin is
increased by more than 7% once it is modied with 8 vol.% SiO2
(Table 3). The addition of silicone rubber to the epoxy formulation,
results in compounds with slightly reduced density. This is because
the added particles are lighter than the polymer matrix. However,
the effect of SR on the matrix density is less pronounced than the
one induced by the hard silica particles. The addition of 5.5 vol.% SR
to EP and the 3 vol.% SiO2 nanocomposite caused only a marginal
reduction on their densities (lower than 2%).
3.2. Mechanical properties
The tensile modulus of all the mixtures investigated herein is
reported in Fig. 5. It is observed that the value of the neat resin
(E0_0 = 2874 MPa) increases almost linearly with the content of

2172

M. Lai et al. / Composites Science and Technology 70 (2010) 21682175

Table 3
Density of silica-rubber modied epoxy resins at 25 C.
Content SiO2 (vol.%)

EP

+3 vol.% SR
3

0
3
8

+5.5 vol.% SR

Density (g/cm )

Std. dev.

Density (g/cm )

Std. dev.

Density (g/cm3)

Std. dev.

1.1524
1.1841
1.2339

0.0003
0.0005
0.0005

1.1479
1.1921
1.2230

0.0010
0.0040
0.0004

1.1432
1.1741

0.0008
0.0006

ceramic nanoparticles and, conversely, decreases with the content


of rubber particles. In fact, the inuence appears to be more
pronounced at 3 vol.% than at 8 vol.% as is shown by the different
slopes in Fig. 5. Similar conclusion can be drawn for what concerns
the rubber particles. The addition of 3 vol.% results in a 6.6%
decrease in tensile modulus whereas a successive addition of
2.5 vol.% to reach the 5.5_0 system is followed by an additional
modulus decrease of 3.7%.
The trend of the data suggests the presence of a limit in
modulus changes in both particle contents however, it was not
examined. A useful comparison of experimental data with tensile
modulus predictions produced by theoretical homogenisation
models presented in the literature can be conducted. In particular,
examining TEM micrographs in Fig. 3, the system can be

Fig. 5. Tensile moduli of SiO2 and SR modied epoxy matrices and comparison with
estimates of HashinShtrikman variational model.

considered macroscopically isotropic and quasi homogeneous, with


reinforcing particles uniformly dispersed in the polymer matrix. All
the latter conditions are necessary for the application of the
HashinShtrikman model [35,36]. Since the system contains also
combinations of two dispersed phases, two successive iterations
are needed. The rst one is for the neat resin and rubber particles,
where the upper bounds are considered, while the second involves
the resulting homogenised parameters and the SiO2 particles,
where the lower bounds is chosen. Fig. 5 reports the comparison
between experimental data and the calculated values obtained
considering mSiO2 = 0.2, ESiO2 = 73 GPa [37], mSR = 0.45, ESR = 0.5 GPa
[38] showing very good agreement among all the values.
An investigation of the elastic properties of the various systems
investigated can be achieved by using DMTA. The resulting curves
show the changes in the storage modulus E0 and the damping factor tan delta as a function of temperature. A representative DMTA
curve of the epoxy system containing 5.5 vol.% SR is illustrated in
Fig. 6. In general, it is observed that the storage modulus reduces
slightly with increasing temperature, until a sharp drop takes place
when the glass transition temperature region is reached. Here, the
damping factor runs through a distinct maximum. Considering the
effect of the particles, the storage modulus varies with the same
tendency as the tensile modulus measured before at room temperature, i.e. both of them increase with rising nanoparticle content,
and they reduce when rubber is added. The rigid particles stiffen
the polymer matrix and their effect prevails in the hybrid compounds, once the ratio of nanoparticle vs. rubber content is higher
than one. Thus, the hybrid compounds with higher amount of
nanoparticles record moduli clearly higher than the neat epoxy.

Fig. 6. Storage modulus and damping factor as function of temperature for a rubber modied epoxy.

2173

M. Lai et al. / Composites Science and Technology 70 (2010) 21682175

Fig. 7. Residual strains evolution as function of the axial coordinate z for different
ller content and type.

Fig. 9. Matrix thermal expansion as function of temperature.

Table 4
Coefcients of thermal expansion for all the materials considered.
CTE (105 C1)

Nano-SiO2
0

Silicon rubber

Fig. 8. Axial strains evolution during thermal cycles for the 0_0 material.

A good balance between the different llers (within the set of


materials tested) could be 3% SiO2 and 3% SR, in which case the
modulus of the epoxy is nearly maintained, but the material has
a better toughness. The effects of the nanoparticles on glass transition are not investigated in this work. Relevant studies, however,
can be found in the literature [2,3941].
3.3. Residual strain conditions at room temperature
At the end of the processing cycle kB z, for each specimen, were
measured and compared with corresponding curves recorded prior
to embedding. Since each of these measurements were conducted
at ambient temperature, (1) reduces to,

DkB z
1  pe ez z
kB0

where ez(z) is the residual strain along the ber direction (cf.
Fig. 1c). The results for some combination of llers are reported in

0
3
5.5

OLCR

TMA

OLCR

TMA

OLCR

TMA

6.74
6.38
6.67

6.12
6.73
6.79

/
6.46
6.43

5.85
6.58
7.45

/
6.46
/

6.00
6.70
/

Fig. 7, where three zones are clearly distinguishable. Zone I, for negative values of z, represents strains measured by the grating outside
the specimens; since these measurements are taken at room temperature no strains are applied to this part. Zone II, presents a steep
gradient of axial deformation, caused by the presence of the specimen edge and zone III a homogeneous strain level on the ber in
most cases.
With respect to the data corresponding to 0_0 (i.e., pure epoxy)
(Fig. 7), it can be seen that the introduction of rubber particles
reduces the residual strains, at z = 20 mm, by about 37% for the
5.5_0 mixture. Conversely the addition of 3 vol.% ceramic nanoparticles to this latter material resulted in a slight increase of this
residual eld (5.5_3). These results can be explained by considering
the difference in the material properties of the constituent llers.
Interestingly, an increase of the SiO2 content does not signicantly
affect the residual strain (3_8). The introduction of llers results
also in a drastic change of the shape of the axial strain along the
ber. This characteristic is highlighted by the comparison of
compositions 3_8, 5.5_3 and 0_0 in zones II and III which appear
to be affected by the llers as compared to pure epoxy (i.e., 0_0).
Note that for the material 0_8, a steep change of the axial strains
is recorder in zone III and is attributed to an interface failure as
discussed later.
3.4. Effect of temperature on residual strains
After the residual strain calculations at room temperature, selected specimens of each mixture were submitted to a series of
two similar heating proles, consisting of four temperature plateaus with: P1 = 31.5 C, P2 = 39 C, P3 = 48 C and P4 = 58 C, with
a maximum oscillation of 0.3 C. The range of these temperatures

2174

M. Lai et al. / Composites Science and Technology 70 (2010) 21682175

(a)

(b)

FD
FD

Fig. 10. lCT-pictures of ber debonding (FB) in the range of the uncoated part of the optical ber, where FBG was supposed to take place: (a) 2D section through the
composite cylinder and (b) 3D section (specimen 0_8).

has been chosen so that the highest temperature was below the
glass transition temperature of the epoxy. The temperature was
recorded, during the whole test, using K-type thermocouples
placed underneath the specimens, in direct contact with them.
Once the temperature in the furnace was stabilized at each plateau,
the specimens were allowed to reach equilibrium for more than
1 h. Afterwards, at least ve OLCR measurements were recorded
at each temperature.
To calculate the thermal strains, (1) is used. Since the residual
strains represented by the rst term of the right hand side of the
equation are known, the second term, representing the thermal
strains developed by the specimen during the thermal cycle, can
be easily calculated knowing the specimen temperature and the
wavelength shifts from the OLCR measurements. Results for the
neat resin conguration are reported in Fig. 8. Considering globally
the reported curves, it becomes clear that, for a given temperature,
the thermal cycles do not induce any modications on the epoxy
and the strain transfer to the ber, especially towards the centre
of the specimen (zone III). Some slight changes are visible in zone
II, indicating a small strain relief. This behaviour is also similar in
all the other samples, independently of the particle content.

SiO2 system in which case a more than 20% difference is seen.


However, the TMA measurements are less accurate because the
strain-temperature data do not fall onto a straight line and is
difcult to calculate a single value for the slope. But otherwise,
for the temperature range examined here, no clear trend in the
CTE of the mixtures can be established, i.e. their CTE can be considered equal to that of the epoxy material used in the mixtures.
To use the FBG sensors, the interface between the FBG sensor
and the composite mixture should be intact. However, this was
not the case for mixtures containing only ceramic particles
(Fig. 7, composition 0_8). Although the trend on the internal strains
induced by these particles can be conrmed (when comparing
curves 0_0 and 0_8), no quantitative analysis could be performed
for this case. This was due to the fact that along a certain portion
of the ber (i.e., 6 6 z 6 17 mm) a reduction of the strains felt by
the FBG-ber was recorded and attributed to ber debonding. This
lack of adhesion was impossible to avoid in all the specimens
prepared with high SiO2-compositions, and it was conrmed by
micro-CT images reported in Fig. 10a and b. The extension of
debonding can be clearly identied around (a) and along (b) the
FBG-ber.

3.5. Coefcient of thermal expansion

4. Summary and conclusions

As pointed out before, the central part of the specimen can be


approximated by an equivalent one dimensional problem since
hoop and radial strains are negligible compared to axial ones, with
the ber and the surrounding material having the same strains.
This has been conrmed in the past by numerical and experimental
investigations [22]. Assuming an intact interface, the strain on the
ber can be calculated using the difference between the thermal
expansion coefcients of the matrix and ber,

The inuence of different ller type (nano-silica; synthetic


rubber) and content on the thermomechanical properties of an
epoxy matrix has been addressed using different investigation
techniques. Using electron microscopy techniques, a uniform
particles distribution was found. The tensile modulus, determined
on dog-bone specimens, was found to increase with increasing
amount of nano-SiO2 particles, whereas the micro-rubber particles
induced the opposite effect. The effect of the particle on the elastic
Youngs modulus was also estimated using the HashinShtrikman
theory, with a maximum difference of 4.5% with the experimental
results. The axial residual strain from the epoxy curing and postcuring was also retrieved using an FBG inscribed on the centrally
located glass ber of the composite cylinder. The results show that
soft rubber particles reduce the residual strains while the addition
of SiO2 has the tendency to increase the residual strains although
the nanoparticles inuence can only be deduced qualitatively
because of a lack of adhesion between the sensors and the matrices
mixed with SiO2 particles only. The evolution of thermal strains
has also been monitored allowing for the evaluation of the CTE of
the epoxy and the composites. The results show that, for the range
of temperatures investigated, the composites behave linearly and
that the inuence of the llers is small. These results are globally
conrmed by both methods used herein. The introduction of SiO2
and or SR particles results in practically no changes in the CTE.
The changes in the TMA results are attributed to the method itself
since the TMA strain-temperature data do not follow a straight
line.

eT z am  af DT

where am is the CTE of the material surrounding the optical ber.


Using (3) and (1) in zone III (z  18 mm) and the evolution of the
local Bragg wavelength, the axial deformation as a function of
temperature is determined and shown in Fig. 9. Since the maximum
imposed temperature during thermal cycles was well below the
glass transition temperature for all the materials, the data exhibit
a linear behaviour in the considered temperature range.
The CTE for each material can be calculated as the slopes of the
straight lines in Fig. 9, or by solving (3) for am. For a better
accuracy, the solution of (3) is used and the results of the calculations are displayed in Table 4 together with TMA data. The data
show that overall both techniques give consistent results with
the maximum observed difference of 13.5% in the case of 5.5_3
mixture. Additionally, the TMA data show a small decreasing trend
on the CTE when adding SiO2 and an increasing one when adding
SR particles. The addition of SR particles does not signicantly
change the CTE except when adding SR particles in the 3 vol.%

M. Lai et al. / Composites Science and Technology 70 (2010) 21682175

Acknowledgements
One of us, K. Friedrich, is grateful to the laboratory of applied
mechanics and reliability analysis which enabled him to stay at
EPFL as a guest professor during the period of April 1 to May 31,
2009. Further thanks are due to the Swiss National Science Foundation under Grant 200020_124397 for the partial nancial support
of the work.

References
[1] Han JT, Cho K. Nanoparticle-induced enhancement in fracture toughness of
highly loaded epoxy composites over a wide temperature range. J Mater Sci
2006;41(13):423945.
[2] R. Medina. Rubber toughened and nanoparticles reinforced epoxy composites.
Technical University Kaiserslautern; 2009.
[3] Liu W, Hoa SV, Pugh M. Organoclay-modied high performance epoxy
nanocomposites. Compos Sci Technol 2005;65(2):30716.
[4] Nigam V, Setua DK, Mathur GN. Failure analysis of rubber toughened epoxy
resin. J Appl Polym Sci 2003;87(5):8618.
[5] Lovell PA. Rubber toughened plastics: strategies for their preparation and
evaluation. Manchester: Manchester Materials Science Center; 1997.
[6] Verchere D, Sautereau H, Pascault JP, Moschiar SM, Riccardi CC, Williams RJJ.
Rubber-modied epoxies. I. Inuence of carboxyl-terminated butadieneacrylonitrile random copolymers (CTBN) on the polymerization and phase
separation processes. J Appl Polym Sci 1990;41(34):46785.
[7] Wetzel B, Rosso P, Haupert F, Friedrich K. Epoxy nanocomposites fracture and
toughening mechanisms. Eng Fract Mech 2006;73(16):237598.
[8] Wetzel B, Haupert F, Zhang MQ. Epoxy nanocomposites with high mechanical
and tribological performance. Compos Sci Technol 2003;63(14):205567.
[9] Zhang KL, Wang LX, Wang F, Wang GJ, Li ZB. Preparation and characterization
of modied-clay-reinforced and toughened epoxyresin nanocomposites. J
Appl Polym Sci 2004;91(4):264952.
[10] Gam KT, Miyamoto M, Nishimura R, Sue HJ. Fracture behavior of coreshell
rubber-modied clayepoxy nanocomposites. Polym Eng Sci 2003;43(10):
163545.
[11] Liu W, Hoa SV, Pugh M. Morphology and performance of epoxy nanocomposites
modied with organoclay and rubber. Polym Eng Sci 2004;44(6):117886.
[12] Walter RM, Haupert F, Schlarb AK. Improved toughness and stiffness of epoxy
resins modied by preformed rubber microparticles and SiO2 nanoparticles for
resin transfer molding applications. In: 2nd international conference micro
and nano-technology, Vienna, Austria; 2007.
[13] Shimbo Masaki OM, Arai Katsumasa. Internal stress of cured epoxide resin
coatings having different network chains. J Coat Technol 1984;56(713):4551.
[14] Shimbo M, Ochi M, Shigeta Y. Shrinkage and internal stress during curing of
epoxide resins. J Appl Polym Sci 1981;26(7):226577.
[15] Croll SG. Residual stress in a solventless amine-cured epoxy coating. J Coat
Technol 1979;51(659):4955.

2175

[16] Nakamura Y, Tabata H, Suzuki H, Iko K, Okubo M, Matsumoto T. Internal-stress


of epoxy-resin modied with acrylic coreshell particles prepared by seeded
emulsion polymerization. J Appl Polym Sci 1986;32(5):486571.
[17] Shimbo M, Yoshida T, Minamoto M. Internal-stress of cured epoxide-resin
lled with glass-beads. Kobunshi Ronbunshu 1983;40(1):18.
[18] Schajer GS. Hole-drilling residual stress proling with automated smoothing. J
Eng Mater Technol 2007;129(3):4405.
[19] Schajer GS. Measurement of non-uniform residual-stresses using the holedrilling method: 1. Stress calculation procedures. J Eng Mater Technol Trans
ASME 1988;110(4):33843.
[20] Lu DJ. Handbook of measurement of residual stresses. The Fairmont Press, Inc.
[21] Prime MB. Residual stress measurement by successive extension of a slot: the
crack compliance method. Appl Mech Rev 1999;52(2):7596.
[22] Colpo F, Humbert L, Botsis J. Characterisation of residual stresses in a single
bre composite with FBG sensor. Compos Sci Technol 2007;67(9):183041.
[23] Han B, Post D, Ifju P. Moir interferometry for engineering mechanics: current
practices and future developments. J Strain Anal Eng Des 2001;36(1):10117.
[24] Pawlak A, Zinck P, Galeski A, Gerard J-F. Photoelastic studies of residual
stresses around llers embedded in an epoxy matrix. Macromol Symp
2001;169(1):197210.
[25] Giaccari P, Dunkel GR, Humbert L, Botsis J, Limberger HG. RPS. On a direct
determination of non-uniform internal strain elds using bre Bragg gratings.
Smart Mater Struct 2005(14):127.
[26] Technical Data Sheet VP 823-30 R. Chemicals and technologies for polymers
GmbH Germany; 2005.
[27] Technical Data Sheet VP 823-33 R. Chemicals and technologies for polymers
GmbH. Germany; 2005.
[28] Technical Data Sheet, Albidur 2240 A. Nanoresins GmbH. Germany; 2004.
[29] Wacker Silicones. Genioperle. Nanoscalige silicone particles; 2007.
[30] Technical Data Sheet VP 823-2H. Chemicals and technologies for polymers
GmbH, Germany; 2005.
[31] European Patent. EP 1 236 765 A1. Silicium dioxide dispersion. European
patent EP 1 236 765 A1; 2002.
[32] Measures RM. Structural monitoring with ber optic technology; 2001.
[33] Shibata N, Shibata S, Edahiro T. Refractive index dispersion of lightguide
glasses at high temperature. Electron Lett 1981;17(8):3101.
[34] <http://www.matweb.com>.
[35] Hashin Z. Analysis of composite materials a survey. J Appl Mech, Trans ASME
1983;50(3):481505.
[36] Hashin Z, Shtrikman S. A variational approach to the theory of the elastic
behaviour of multiphase materials. J Mech Phys Solids 1962;11(2):12740.
[37] Brook RC. Concise encyclopedia of advanced ceramic materials. Oxford:
Pergamon; 1991.
[38] Tomanek A. Silicone & Technik. Mnchen, Germany: Hanser; 1990.
[39] Liu YL, Hsu CY, Wei WL, Jeng RJ. Preparation and thermal properties of epoxysilica nanocomposites from nanoscale colloidal silica. Polymer 2003;44(18):
515967.
[40] Sun YY, Zhang ZQ, Moon KS, Wong CP. Glass transition and relaxation behavior
of epoxy nanocomposites. J Polym Sci Pt B-Polym Phys 2004;42(21):384958.
[41] Zhang H, Zhang Z, Friedrich K, Eger C. Property improvements of in situ epoxy
nanocomposites with reduced interparticle distance at high nanosilica
content. Acta Materialia 2006;54(7):183342.

You might also like