Upload
hadieu
View
215
Download
0
Embed Size (px)
Citation preview
The mechanical behaviour of ZnO nano-particle modified
styrene acrylonitrile copolymers
K. Zuoa, B.R.K. Blackmana,*, J.G. Williamsa,c, H. Steiningerb
a Imperial College London, Department of Mechanical Engineering, City & Guilds Building,
South Kensington Campus, London SW7 2AZ, UK
bBASF SE, Materials Physics and Analytics, 67056 Ludwigshafen, Germany
cAMME Department, University of Sydney, Australia
Keywords: A. Nano composites; B. Fracture toughness; C. Crack
Abstract
Two amorphous SAN copolymers with acrylonitrile contents of 24% and 34% have been
modified by the incorporation of ZnO nano-particles at volume fractions of up to 2.00 %. For
the SAN with 24% acrylonitrile content, three types of nano-particles were added. The first two
were cylindrical (nano-rods) and the third were spherical. For the SAN with 34% acrylonitrile
content, just one of the cylindrical nano-particle types were used. The dispersion of the nano-
particles was assessed using atomic force microscopy and agglomeration was observed for the
higher volume fractions. Young's modulus and fracture toughness values were measured for
each system as a function of volume fraction. The smaller nano-rods were found to be the most
effective for both stiffening and toughening the polymers and the spherical particles were found
not to contribute to stiffness due to poor particle-matrix adhesion. At room temperature the
maximum toughening contributions for all particles occurred at volume fractions of about 0.1%,
above which a reduction was observed. At 80 °C the toughening contributions continued to rise
slowly past 0.1% volume fraction. The toughening is consistent with the notion that the
particles debond from the matrix and then plastic hole growth around the detached particles is
initiated. The peak toughening observed at 0.1% volume fractions are the result of
agglomeration occurring at higher volume fractions.
1. Introduction
1
In recent years the modification of polymer properties via the incorporation of nano-particles
has been a very popular research topic. Much of the literature has focused on the addition of
spherical nano-particles to thermosetting matrices [1-3], where relatively large quantities of
particles can be well dispersed, without agglomeration occurring. Typically it is more difficult
to disperse nano-particles into thermoplastic matrices. Extrusion mixers and particle surface
treatments are widely used techniques to facilitate this process [4]. However, agglomeration of
nano-particles remains a problem in thermoplastics at relatively low particle volume fractions
and thus efforts continue in the field to improve this situation. The use of different particle
shapes and aspect ratios can have an effect on dispersion, for example spherical particles may be
replaced with cylindrical, rod-like particles of various diameters and aspect ratios, and there has
been much interest also in the use of carbon nano-tubes [5].
Once the nano-particles are dispersed into the polymer matrix, many researchers have reported
that significant stiffening and toughening can occur. The stiffening requires load transfer to take
place between the relatively low Young’s modulus polymer matrix and the usually much higher
Young’s modulus particles. The simple rule of mixtures can then be used to predict the
resulting Young’s modulus of the nano-composite. Although there are complications (e.g.
aspect ratio and orientation effects and the potential agglomeration of particles) the basic
mechanism for stiffening is well understood. However, a universally accepted explanation for
the improved toughness values is currently lacking. Workers have identified a number of
mechanisms that contribute to the toughness of nano-composites and efforts have been made to
determine their individual contributions and also their combined effects. The toughening
mechanisms that have been proposed include particle-matrix debonding, the subsequent plastic
void growth [6], shear yielding (or shear banding) of the matrix between the particles [1, 7], and
in some cases crack tip bridging [5] (e.g. when nano-tubes are employed). Also, more recently
workers have employed a multiscale modelling strategy to describe the toughening of polymers
by nanofillers with different properties being ascribed to the matrix and interphase regions
around particles [8-10]. Such work has considered the combined effects of debonding, plastic
yielding of nano-voids and shear banding in the polymer matrix [10]. The model proposed in
[6] to describe debonding followed by plastic void growth has been used to determine the
energy density of matrix shells around debonding particles using a geometrical description of
particle-particle interactions [11, 12].
In the present work, two thermoplastic copolymers SAN with 24% acrylonitrile and SAN with
34% acrylonitrile were modified with three different nano-particle types - two rod-like,
2
cylindrical particle types and spherical particles. The enhancements in stiffness and toughness
have been measured experimentally in a series of tests at 23 °C and 80 °C.
The model proposed in [6] to describe nano-composite toughening enhancement due to the
addition of spherical particles by (i) particle-matrix debonding followed by (ii) plastic void
growth was extended to accommodate toughening with nano-rods and nano-tubes [13]. Here
experimental data for SAN toughened with nano-spheres is analysed using the spherical model
[6], and also SAN toughened with two sizes of nano-rods are analysed using the analogous
model for rod-like particles [13]. The models define a toughness enhancement factor, X, which
is determined from experiment (for either spheres or rods). The measured values of X (up to the
limiting volume fraction when agglomeration occurs) have been used to determine the model
parameters Ga (the energy for debonding) and the critical interfacial stress, σ c, which controls
the subsequent plastic void growth from the debonded particle.
2. Materials
Two grades of amorphous SAN copolymers were used in this study. The first had a trade name
of Luran VLN and contained 24% acrylonitrile in the copolymer. This is referred to as SAN-24
in the present work. The second had a trade name of Luran VLP and contained 34% acrylonitrile
in the copolymer. This is referred to as SAN-34 in the present work.
Three types of ZnO nano-particles were employed as fillers, denoted here as Types A, B and C.
Types A and B were nano-rods and Type C were nano-spheres. Type-A nano-rods had an
average diameter of 40 nm with an aspect ratio (length/diameter) of 4, Type-B nano-rods had an
average particle diameter of 12 nm with an aspect ratio of 3, and Type-C particles were
spherical with an average diameter of 10 nm. All materials were supplied by BASF SE,
Germany. The dimensions of the particles are summarised in Table 1.
The particles were surface modified with trioxadecanoic acid (TODS) or with one of two
different types of silane before blending in order to improve the dispersion of the particles into
the matrix. The SAN and ZnO blends were compounded in a Werner & Pfleiderer ZSK-30 twin
screw extruder and the melt was extruded through a flat sheet die. Prior to compression
moulding, the raw sheets were dried at 80 °C in vacuum for 36 hours. The resulting sheets were
then compression moulded into 6 mm and 4 mm thick plates using a hot compression moulding
3
machine (DAKE, USA) at 200 °C and 5 MPa. The composite systems studied are listed in Table
2.
Concerning the nomenclature used in the present paper, when nano-particles are added to the
system, the volume fraction of particles is shown in the system designation preceding the
particle type, e.g. system "SAN-24-0.5A" contains 0.5% by volume fraction of Type-A
particles. As will be discussed in the next section, for systems modified with the Type-B
particles, tests were performed at two temperatures; namely 23 °C and 80 ºC.
3. Experimental
3.1 Thermal analysis
The thermal behaviour of the unmodified and particle modified composites was studied via
dynamic mechanical thermal analysis (DMTA). Tests were performed on bars of dimensions 50
× 3 × 2 mm in a ‘PerkinElmer DMA 8000’ apparatus. Dual cantilever geometry was employed,
at two different frequencies of 1 Hz and 10 Hz over a temperature range from 20 °C to 140 °C.
At least two specimens were tested for each formulation, and the was defined to be the
temperature at the peak value of the loss factor, , as specified in the ISO standard [14].
3.2 Microscopy
To investigate the particle dispersion in the various systems, atomic force microscopy (AFM)
analysis was undertaken using a ‘MultiMode’ scanning probe microscope from Veeco (Santa
Barbara, USA) equipped with a ‘J’ scanner and a ‘NanoScope IV’ controller. A smooth surface
was first prepared by cutting the samples on a ‘cryo-ultramicrotome’ at 23 °C. The scans were
performed in tapping mode using etched silicon probes supplied by Veeco and both height and
phase images were recorded simultaneously.
To allow an investigation of the fracture surfaces following fracture testing, a ‘Hitachi S-3400N
VP’ scanning electron microscope was used. Typically, an accelerating voltage of 15 kV was
used, and the working distance was approximately 10 mm. The fracture surfaces were sputter
coated with a very thin layer of gold (about 15 nm thick) to reduce charging effects prior to the
SEM examinations. The SEM images could show the morphology of the specimens, and also
the roughness of the fracture surface and the fracture path.
4
3.3 Tension, compression and fracture testing
Specimens for testing were prepared in both the longitudinal and transverse directions on the
mouldings with respect to the previous direction of extrusion and no effects of orientation were
observed in the results. Tensile tests were performed on an Instron testing machine (model
5584) running Bluehill™ control software. The tests were conducted in accordance to ISO 527
[15] standards to determine the Young’s modulus, E. Dumbbell samples, approximately 4 mm
wide × 4 mm thick were used. At least five specimens were tested for each formulation. All tests
were performed using a contact extensometer with a gauge length of 25 mm, at a crosshead
speed of 1 mm/min.
Uniaxial compression tests were performed to determine the yield stress, σ Y, since testing of
relatively brittle materials such as SAN in tension can lead to brittle failure prior to the yield
point. During uniaxial compression, brittle fracture was avoided and plastic yielding was
observed prior to failure. The compression specimens were cylindrical, 6 mm in height (parallel
to the direction applied force), and 20 mm in diameter. A synthetic grease was used to reduce
friction between the specimens and the compression plates. The specimens were tested at a
constant crosshead speed of 1 mm/min according to ISO 604 [16].
Fracture tests were performed on single edge notch bending (SENB) specimens according to
ISO 13586 [17] to obtain the mode I fracture toughness, , and the fracture energy, , at
fracture initiation. Natural pre-cracks were produced by tapping new razor blades into the notch
tips prior to testing. The SENB specimens were tested at a constant displacement rate of
1 mm/min, and at a test temperature of 23 ºC or 80 ºC. Indentation correction tests were
performed as specified in [17].
Tensile, compression and fracture tests were also conducted at 80 °C. All tests at 80 °C were
carried out in an environmental chamber. The test chamber allowed for an accurate temperature
control (±1 K) by using two thermocouples; one thermocouple was inserted into a reference
sample, while the other was freely suspended in the chamber. The chamber was heated up
gradually to 80 °C, and all specimens were stored in the heated chamber for 1 hour prior to
testing to reach thermal equilibrium. Once a specimen was mounted and the chamber door
closed, a soaking time of at least 10 minutes elapsed, ensuring the temperature in the chamber
had stabilised.
5
In order for the fracture results to be considered valid [17], it is required that the size criteria for
plane strain conditions be satisfied:
B ,a , (W −a )>2.5( KQ
σY)
2
(1)
where B ,W and aare the thickness, width and crack length (pre-notch plus razor crack) of the
specimen, respectively. KQ is the conditional or trial value of K c and σ Y is the yield stress of the
material at the appropriate temperature and loading rate.
4. Results
4.1 Particle dispersion
The morphology study was undertaken by using atomic force microscopy (AFM). Fig. 1 shows
AFM phase images of SAN-24-A, SAN-24-B and SAN-24-C composites at various particle
volume fractions. The nano-particles are seen as ‘lighter regions’ on the phase image as the
stiffness of ZnO particles is significantly greater than that of the surrounding polymer. The
microstructure of unmodified SAN was featureless at this scale. The Type-A nano-rods were
difficult to disperse into the matrix. With the addition of 2.00 vol.% of Type-A nano-rods,
agglomerates as large as 300 nm were observed in the composite (Fig. 1 (b)). The Type-B nano-
rods and Type-C nano-spheres were relatively well dispersed through the matrix up to 0.30 vol.
%. Fig. 1 (c) & (e) show the good dispersion achieved for 0.10 vol.% Type-B and Type-C
particles in SAN-24. However, agglomerates as large as 80 nm and 120 nm were observed in
composites containing 0.50 vol.% Type-B nano-rods (Fig. 1 (d)) and 1.25 vol.% Type-C nano-
spheres (Fig. 1 (f)), respectively. Similar observations were made for the system comprising
the SAN-34 matrix with the Type-B nano-rods (note systems SAN-34-A and SAN-34-C were
not tested).
4.2 Thermal behaviour
The storage modulus, , loss factor, , and glass transition temperature, , of the
unmodified and particle modified systems determined by using DMTA are shown in Fig. 2.
6
Typical traces for the SAN-24-0.5B system tested at 1 Hz and 10 Hz are shown. The value of
measured was frequency dependent reflecting common viscoelastic materials response,
which was also observed by Johnsen et al. [18] and Brooker [14] . The obtained at 10 Hz
was about 6 K higher than that obtained at 1 Hz. The of unmodified SAN was measured to
be 116 ± 2 °C at 1 Hz, and 122 ± 2 °C at 10 Hz. The addition of ZnO nano-particles did not
significantly affect the glass transition temperatures of the modified SAN composites, as is
shown in Table 3.
4.3 Tensile testing results
For the tensile tests carried out at 23 °C, no obvious plastic deformation was observed for the
unmodified and nano-modified SAN composites and all materials fractured in the elastic region.
The values of the tensile fracture stress, σ f , are given in Table 3 and show significant decreases
with volume fraction for the A and C particles. However, the values remain approximately
constant for the B particles. The addition of nano-particles is expected to stiffen the composite
due to the high modulus of the particles (approximately 140 GPa [20]). An average tensile
modulus of 3.87 ± 0.05 GPa was measured for the unmodified SAN-24 at 23 °C, in good
agreement with that reported by Cebon et al. [15] for this thermoplastic. Fig. 3 plots the values,
for the various modified SAN-24 composites, of normalised Young’s modulus,( Ec
Em−1), against
the volume fraction of nano-particles added, where Ec and Em are the Young’s moduli of the
modified composite and the control matrix respectively. The results are presented in this form so
that the enhancement in the property, i.e. Young’s modulus or fracture toughness below, is
emphasised. The lines drawn are the best linear fits to the experimental values. It can be seen
that the Type-B nano-rods were most effective over the range of volume fractions studied,
giving the greatest slope on Fig 3. This will be discussed further in Section 5. Interestingly, the
moduli for the SAN-24-C system fall somewhat below the unmodified control. It is thought that
this is due to the nano-particles being only very weakly bonded to the matrix in this system, and
hence failing to support the particle-matrix load transfer.
7
4.4 Fracture testing results
In all the tests the cracks initiated suddenly from the pre-crack, failure thus occurred in a brittle
manner. The values of K c and Gc were obtained at fracture initiation as defined in ISO 13586.
All tests passed the linearity and plane strain validity crosschecks (Eq. 1). The Gc values were
calculated using the energy method as described in [17].
The fracture results at 23 °C are presented in Table 3. The normalised values of fracture energy
( Gc
Gm−1), where Gc and Gm are the fracture energies of the composite and the matrix
respectively were plotted against the nano-particle volume fraction in Fig. 4 for the SAN-24
matrix. Although there is a limited amount of data at very low volume fractions, it would appear
that Type B nano-rods were the most efficient of the three particle types in toughening the
matrix. Also for each of the three particle types, the maximum toughness was achieved at a
particle volume fraction of about 0.10 %. For each particle type the fracture toughness then
decayed very quickly with increasing volume fraction after the maximum toughness was
achieved. Type-C nano-spheres produced similar toughening at the maximum to Type-A nano-
rods, but the values then decayed more rapidly post maximum. The dramatic decrease in
toughness following the maxima was caused by the formation of nano-particle agglomerates at
the higher volume fractions, as depicted in Fig. 1.
The results for the two different copolymers at 80 ˚C, are summarised in Table 4. The
copolymer with higher acrylonitrile content (SAN-34) possessed the higher toughness. The
values of Gc for the unmodified control specimens for SAN-24-0 were 217 J/m2 compared to
561 J/m2 for the SAN-34-0 control specimens. Thus, the SAN-34 matrix was significantly
tougher the SAN-24 matrix. However, on a normalised basis, the trends in toughness
enhancement were similar in the SAN-24 and SAN-34 matrices when modified with the Type-B
nano-rods although proportionately the greater enhancement was seen in the more brittle system.
The results are shown in Figure 5.
Comparing the results for the SAN-24 and SAN-34 matrices at 80 ˚C, then for both matrices the
higher temperature led to a change in the toughening trends, with the toughness of the systems
now continuing to rise, although slowly, with increasing volume fraction. The results for SAN-
24-B are shown in Figure 6(a) and the results for SAN-34-B are shown in Figure 6(b).
8
The micrographs obtained in the scanning electron microscope are shown in Figures 7 and 8 for
the systems modified with the Type-B nano-rods and tested at 23 °C and 80 °C respectively.
Figure 7(a) shows the featureless fracture surface from the SAN-24-0 control tests and Figure
7(b) shows the fracture surface following testing of the SAN-24-0.05B specimens. Circled in
this figure are sites where evidence exists for the cavitation and then subsequent fibrillation of
the SAN-24 matrix following addition of Type-B nano-rods, i.e. a crazing mechanism. Figures
7(c) and (d) indicate the presence of conic markings on the surfaces of the SAN-34 specimens
and as these are also present in the unmodified control, they are not due to the presence of the
Type-B nano-rods. Figure 7(d) clearly shows the presence of voids on the fracture surface
following the addition of the Type-B nano-rods. These voids were measured to be
around 100 nm to 200 nm in diameter, i.e. much larger than the Type-B nano-rods used to
reinforce the matrices. These voids have most likely resulted from the debonding of the particle-
matrix interface and the subsequent plastic hole growth. Figure 8 shows the resulting fracture
surfaces for SAN-24 and SAN-34 matrices reinforced with Type-B nano-rods at 80 °C. With
the exception of the SAN-24 control specimen, all surfaces showed the presence of voids with
sizes in the range of 100 nm to 500 nm. Some of the voids are indicated by arrows in the figure.
Again, this would suggest that debonding of the particle-matrix interface and then subsequent
plastic hole growth had occurred. This has been observed by several other workers and will be
explored quantitatively in the next section.
5. Discussion
5.1 Tensile modulus results
If particles are well bonded to the matrix then we would expect a simple law of mixtures for low
volume fractions (ϕ≪1):
Ec ≈ Em+Ep ϕ
where Em is the modulus of the matrix i.e. approximately 4 GPa and Ep is the modulus of the
particle i.e. approximately 140 GPa, so that:
( Ec
Em−1)≈ 35 ϕ
This line is drawn in Fig. 3 and shows that at very small ϕ the Type-B particles follow this line,
implying the particles are adhered. However, for the larger volume fractions, the slope decreases
9
and falls to a value of 11.6 when all the points are included. This would be consistent with a
reduction in the particle to matrix adhesion leading to the subsequent significant deviation from
the rule of mixtures for all but the smallest value of . The Type-A particles followed a line
with a slope of 5.2 and the Type-C particles indicate no adhesion (in fact, there is negative
enhancement with Type-C particles, an observation which is repeated for the toughness
measurement and is discussed further in the next section. The slopes suggest that the Type-B
particles were well adhered for low ϕ but at higher values the data suggest that about 2/3 were
agglomerated and were thus not adhered as the slope fell by this proportion. For the Type-A
particles a similar argument suggests that the fraction of particles agglomerated is about 0.84
and for Type-C particles all are agglomerated.
5.2 Fracture toughness results
The enhancement in toughness resulting from the addition of the various nano-particles to the
SAN copolymers is considered to result largely from the plastic void growth which follows the
debonding of the particle from the surrounding matrix. This mechanism was shown to be
important for the toughening of epoxy with ATH and silica particles and also for the toughening
of polypropylene with CaCO3 [21].
The mechanism was modelled [6] as a two stage process by firstly considering rigid spherical
particles bonded to the matrix. On loading the specimen the particles in the crack tip region
undergo an increase in interfacial stress until the particles debond. The number of particles
participating in this process depends upon the particle volume fraction and the size of the crack
tip process zone. This debonding process dissipates the interfacial energy per unit area of
particle, Ga. Although this is a small amount of the total energy dissipated, the debonding step is
an essential prerequisite for stage two- given by the resulting void in the matrix expanding.
This process dissipates plastic energy. The two energy dissipating processes together cause the
toughness of the composite to increase from the value of the matrix (when no particles are
present, i.e. ϕ=0) to Gc (when the particles are present at a volume fraction of ϕ. The
enhancement in toughness described by the model is given by:
( Gc
Gm−1)=Xϕ (2)
10
where the toughening factor, X , is a function of Ga, the particle radius ro, the matrix yield stress
σ Y and Young’s modulus of the matrix Em. This analysis has been extended [13] to rigid, rod-
like particles where the same two stage mechanism pertains. In this case, the appropriate
relationships for rod-like particles with an aspect ratio of 3, as used here for the Type-B
particles, and for ν=1/3 , are given by:
X=0.584( ex−1
x−0.318) (3)
where
x2=3.25( Em Ga
σY2 r o
)=( 32
σ c
σY)
2
(4)
The procedure followed is that X is determined from the experimentally measured values of Gm
and Gc according to eqn. (2). Eqn (3) is then used to determine the critical stress ratio factor, x,
and eqn (4) is then used to determine the interfacial energy, Ga and also the critical stress for
debonding, σ c. Therefore, if the particle size, matrix properties and the interfacial energy are all
known, the model can be used to predict the toughness enhancement of the composite. In the
present work, the toughness enhancements have been determined experimentally, so the model
has been used to output values of x ,Ga and σc
σY . For each of the materials investigated, the
maximum toughness enhancement occurred at about ϕ=0.1 % (Fig. 4) and therefore X has been
determined at this value.
Table 5 gives the values of E, σ Y , ro, the test temperature T, X , x ,Ga and σ c
σY for the SAN-24
and SAN-34 systems. The values of x ,Ga and σc
σY are calculated using eqns. (3) and (4) except
for the Type-C particles which were spherical and used the form appropriate to that case [6].
(Type-A particles had an aspect ratio of 4 but the difference in the equations is negligible).
The x values were determined, being in the range 8.9 to 10 for the three particle types, and the
Ga values were determined to be in the range 0.19 to 1.0 J/m2. The value of Ga for the Type-B
particles at about 0.3 J/m2 are similar to the value obtained for nano-silica particles of the same
size in an epoxy matrix. The value of Ga = 1.0 J/m2 for the composites with Type-A particles is
notably higher than for particles Types B or C. This observation was noted and discussed at
11
length in [6]. It is of interest to note that for spherical particles x=σc
σY whilst for rods x=3
2σc
σY
so that σ c
σY is about 6 for rods and 9 for spheres. This suggests that some of the problem in
toughening with the Type-C particles here (spheres) may arise from these particles requiring
50% higher stress to trigger debonding. For rods the stresses required for debonding are lower
due to their geometry. A ratio of σc
σY≈ 7 was found for silica spheres in epoxy.
The rapid decline in toughness (post maximum) with increasing ϕ may also be explained by this
analysis. It was noted that agglomeration occurred for the higher ϕ values leading to ‘particles’
forming with a radius of up to 60 nm, i.e. effectively increasing the particle size by a factor of
10. Assuming all other parameters remain constant, this decreases the critical stress ratio factor,
x by 10−1 /2=0.32 , i.e. x=3 giving X ≈ 1.2 from eqn. (3). Thus the toughening from the
agglomerates is negligible and the loss of the agglomerated particles from the toughening
process leads to the decline in toughening. The modulus changes also suggest that agglomerated
particles are not adhered, thus preventing toughening.
The analysis of plastic void growth [13] which is controlled by the parameter x also gives an
estimate of the cavity radius, r v, given by:
( rv
ro)
2
≈ 1+32
σY
E c⋅ex−1
Using typical values of x=9 and σY
E≈ 0.02 from Table 5 we have
rv
ro=9.5 . Thus for ro=6 nm,
the average void diameter would be 110 nm. Fig 8. gives a range of sizes from
100 nm – 500 nm and given that the particle size varies, and given also the existence of
agglomerates, the prediction provides some support for the analysis.
An extraordinary feature in Fig. 4 is that for the Type-A and Type-C particles, the values of Gc
Gm
decrease significantly below unity for volume fractions, ϕ>0.5 % . The toughening model
discussed here does not predict such an effect and its source must be sought elsewhere. Some
indication is provided in Table 3 in that the values of tensile fracture stress also decreased with
increasing ϕ which implies a decrease in the crazing stress. It is suggested that the decrease in
Gc arises from the particles disrupting the energy absorbing crazing process in these cases.
12
Finally, it was noteworthy that increasing the test temperature from 23 °C to 80 °C changed the
toughening behaviour of SAN-24 and SAN-34 matrices with the Type-B particles significantly
at the higher volume fractions (Figs 6a and 6b). For both composites, the peak in toughness
which occurred at about a value of =0.1% at 23 °C disappeared, and the toughness continued to
rise further, albeit at a decreasing rate, over the whole range of tested. At temperatures closer
to the glass transition temperature the molecular mobility increases and in its wake, matrix
toughness decreases. This would increase the size of the process zone, 2c, formed ahead of the
crack tip. A larger process zone will contain an increased number of ‘active particles’ i.e. ones
which debond and induce matrix void growth as part of the fracture process. In [13] this was
termed N. In addition, the critical stress, c, for plastic void growth around debonded particles
drops significantly, i.e. by a factor of about 1/3. These effects are summarised in Table 6. As a
result, some aggregates so far excluded from the fracture process now are ‘debonded’, giving
way to subsequent plastic void growth. In the case of aggregates this may not be debonding in
the true sense, but may involve internal fragmentation at stresses below c, from which plastic
void growth can initiate. Hence the density of active filler particles, i.e. those taking part in the
fracture process of the composite by debonding (or fragmentation) and subsequent plastic void
growth, is increased at elevated temperatures. This can be expressed by an increase in the
number of particles in the process zone per unit length, i.e. N/l.
6. Conclusions
Two amorphous SAN copolymers with acrylonitrile contents of 24% and 34% were modified by
incorporating ZnO nano-particles at volume fractions of up to 2.00%. SAN with 24%
acrylonitrile was modified with three types of nano-particles termed Type-A (40 nm diameter
nano-rods with aspect ratio of 4), Type-B (12 nm diameter nano-rods with aspect ratio of 3), and
Type-C (10 nm diameter nano-spheres). SAN with 34% acrylonitrile was just modified with the
Type-B nano-rods. For each composite investigated the dispersion of the nano particles was
assessed using atomic force microscopy and the glass transition temperature was measured
using dynamic mechanical analysis. For each composite, tensile tests and quasi-static fracture
toughness tests were performed at 23 °C. For the composites containing the Type-B nano-rods,
tests were additionally performed at 80 °C. The incorporation of nano-particles was shown to
have no effect on the Tg of the resulting composite and atomic force microscopy showed that
there was significant agglomeration in all the systems for particle volume fractions above 0.1%.
13
It was for this reason that all the composites studied here contain no more than 2% by volume of
added particles.
The tensile modulus results were compared with the prediction from a simple rule of mixtures
and it was shown that there was only agreement for the Type-B nano-rods at volume fractions of
up to 0.1%. For higher volume fractions the modulus fell significantly below that predicted and
this is consistent with the observation that the particles agglomerated in this range. The tensile
modulus results for the composites modified with the larger Type-A nano-rods were all
significantly below the values predicted and that is also consistent with these larger particles
agglomerating at all concentrations (even the smallest in this study at 0.1%). The tensile
modulus results for the composites modified with the Type-C nano-spheres were also
significantly below the values predicted and indeed negative enhancement was observed.
The fracture toughness was measured for each composite at 23 °C and its enhancement was
determined as a function of volume fraction. For all composites there was an initial increase in
toughness with a maximum occurring at about 0.1% by volume of added particles, after which
the enhancements were reversed. For the SAN-24 matrix, Type-B nano-rods were again the
most effective and led to the greatest percentage increases in toughness at +41% relative to the
unmodified matrix (X=410). They were twice as effective as the larger Type-A nano-rods
(which gave a maximum increase in toughness of +19%) , and also were more effective than the
Type-C nano-spheres (which gave a maximum increase in toughness of +27%) . The SAN-34
matrix with a GIc of 936 J/m2 was significantly tougher than the SAN-24 matrix at 316 J/m2 in
the unmodified condition and the addition of Type-B nano-rods gave a maximum enhancement
of +21% (X=210) at 23 °C. When the temperature was increased from 23 °C to 80 °C, the
fracture toughness was shown to rise continuously (albeit at a slower rate) past 0.1% volume
fraction for both matrices when reinforced with the Type-B nano-rods. The fracture results were
further interrogated by the use of the by particle cavitation and void growth toughening model
previously proposed. This model showed that nano-rods were more efficient at toughening than
nano-spheres because the rod-like particles debond at a lower stress than the spheres and can
therefore impart toughness more readily to the composite. This can explain the lower toughness
enhancement achieved by adding the Type-C particles compared to the Type-B particles for an
equivalent particle volume fraction. By inputting the experimentally determined toughness
enhancement into the model, together with the other parameters of yield stress, Young’s
modulus, and particle radius, then the critical stress ratio factor for debonding could be
calculated, together with the parameter Ga, describing the adhesion between the particle and the
14
matrix. The values obtained were sensible and added weight to the validity of the model. The
model was then used to predict how the toughness enhancement would be affected by
agglomeration of particles. The microscopy undertaken here suggested that agglomerates of up
to 60 nm in radius were formed and the model confirmed that this would almost entirely remove
the toughness enhancement. Whilst the toughening model does not predict any negative
enhancement (deterioration) it is suggested that this observation is associated with the
breakdown of the crazing mechanisms which give the higher toughness in thermoplastics.
Finally the difference in toughening behaviour for the SAN matrices modified with B-particles
at 23 and 80 °C for volume fractions greater than 0.1% was rationalised in terms of the greater
number of particles becoming active in the process zone at the higher temperatures.
References
1. Hsieh, T.H., Kinloch, A.J., Masania, K., Sohn Lee, J., Taylor, A.C., Sprenger, S., The
toughness of epoxy polymers and fibre composites modified with rubber microparticles
and silica nanoparticles. Journal of materials science, 2010. 45(5): p. 1193-1210.
2. Liang, Y.L. and R.A. Pearson, Toughening mechanisms in epoxy-silica nanocomposites
(ESNs). Polymer, 2009. 50(20): p. 4895-4905.
3. Wetzel, B., Rosso, P., Haupert, F., Friedrich, K., Epoxy nanocomposites - fracture and
toughening mechanisms. Engineering Fracture Mechanics, 2006. 73(16): p. 2375-2398.
4. Tang, E., Cheng, G., Ma, X.,Zhao, Q., Surface modification of zinc oxide nanoparticle
by PMAA and its dispersion in aqueous system. Applied Surface Science, 2006. 252(14):
p. 5227-5232.
5. Hsieh, T.H., A.J. Kinloch, A.C. Taylor, I.A. Kinloch. The effect of carbon nanotubes on
the fracture toughness and fatigue performance of a thermosetting epoxy polymer.
Journal of Materials Science, 2011. 46: p. 7525-7535.
6. Williams, J.G., Particle toughening by plastic void growth. Composites Science and
Technology, 2010. 70: p. 885-891.
7. Huang, Y. and A.J. Kinloch, Modelling of the toughening mechanisms in rubber-
modified epoxy polymers. Journal of materials science, 1992. 27(10): p. 2763-2769.
8. Zappalorto, M., M. Salviato, and M. Quaresimin. A multiscale model to describe
nanocomposite fracture toughness enhancement by the plastic yielding of nanovoids.
Composites Science and Technology 72 (2012): p.1693-1691.
15
9. Salviato, M., M. Zappalorto, and M. Quaresimin. Plastic shear bands and fracture
toughness improvements of nanoparticle filled polymers: A multiscale analytical model.
Composites Part A 48 (2013): p.144-152.
10. Quaresimin, M., M. Salviato, and M. Zappalorto. A multiscale and multimechanism
approach for the fracture toughness assessment of polymer nanocomposites. Composites
Science and Technology 91 (2014): p.16-21.
11. Lauke, B., Contribution of matrix yielding energy to the crack resistance of particle
reinforced composites. Composites Science and Technology 86 (2013): p.135-141.
12. Lauke, B., and S. Y. Fu, Aspects of fracture toughness modelling of particle filled
polymer composites. Composites Part B 45 (2013): p.1569-1574.
13. Williams, J.G., B.R.K. Blackman, H. Steininger, K. Zuo. Toughening by plastic
cavitation around cylindrical particles and fibres. Composites Science and Technology
103 (2014): p.119-126.
14. ISO 6721-11, (2012) Plastics - Determination of dynamic mechanical properties - Part
11: Glass transition temperature. ISO, Geneva.
15. ISO 527-1, (1993) Plastics-Determination of tensile properties-Part 1: General
principles. ISO, Geneva.
16. ISO 604: (2003) Plastics-Determination of compressive properties. ISO, Geneva.
17. ISO 13586, (2000) Plastics- Standard test method for the determination of fracture
toughness (Gc and Kc)- A linear elastic fracture mechanics approach. ISO, Geneva.
18. Johnsen, B.B., Kinloch, A.J., Mohammed, R.D., Taylor, A.C., Sprenger, S., Toughening
mechanisms of nanoparticle-modified epoxy polymers. Polymer, 2007. 48(2): p. 530-
541.
19. Brooker, R.D., The morphology and toughness of thermoplastic-modified epoxy polymer
with carbon nanotubes or silica nanoparticles. Ph.D. thesis, Department of Mechanical
Engineering, Imperial College London, London, 2008.
20. Cebon, D., M.F. Ashby, L. Lee-Shothaman, CES EduPack 2010 Software. 2010.
21. Stapountzi, O.A., M. Charalambides, and J.G. Williams, Micromechanical models for
stiffness prediction of alumma tri hydrate (ATH) reinforced PMMA: effect of volume
fraction and temperature. Composites Science and Technology, 2009, 69(11-12): p.
2015-2013.
16
Table 1. Details of the ZnO nano-particles used.
Particle Designation Shape Diameter (nm) Aspect ratio (l/d)
A Nano-rod 40 4
B Nano-rod 12 3
C Nano-sphere 10 1
Table 2. The nano-composite systems studied and testing temperatures.
↓Matrix; Particles→ A B C"Composite designation" and (test temperatures)
SAN-24 "SAN-24-A"
(23 ºC)
"SAN-24-B"
(23 ºC & 80 ºC)
"SAN-24-C"
(23 ºC)
SAN-34 - "SAN-34-B"
(23 ºC & 80 ºC)
-
Note: (-) indicates not tested.
17
Table 3. Summary of the values of glass transition temperature, Tg, Young’s modulus, E, yield stress, σy, fracture stress, σf, fracture toughness, Kc, and fracture energy, Gc, for the unmodified and nano-modified SAN composites tested at 23 ºC.
System Nano-content
Tg E σy σf Kc Gc
(vol.%) (°C) (GPa) (MPa) (MPa) (MPa·m1/2) (J/m2)
SAN-24-0 0 116 ± 2 3.87 ± 0.05 88 ± 1 53 1.12 ± 0.08 316 ± 12
SAN-24-A 0.10 117 ± 2 3.89 ± 0.04 88 ± 1 41 1.28 ± 0.10 375 ± 25
SAN-24-A 1.04 116 ± 0 4.06 ± 0.10 88 ± 1 35 1.06 ± 0.03 257 ± 9
SAN-24-A 2.00 115 ± 2 4.28 ± 0.14 88 ± 1 34 0.98 ± 0.04 202 ± 17
SAN-24-B 0.05 117 ± 0 3.95 ± 0.06 88 ± 1 52 1.41 ± 0.02 441 ± 12
SAN-24-B 0.10 115 ± 0 3.98 ± 0.03 88 ± 1 53 1.40 ± 0.03 445 ± 27
SAN-24-B 0.30 116 ± 2 4.03 ± 0.04 88 ± 1 53 1.34 ± 0.02 402 ± 14
SAN-24-B 0.50 116 ± 0 4.06 ± 0.05 88 ± 1 50 1.26 ± 0.05 372 ± 9
SAN-24-C 0.10 116 ± 0 3.85 ± 0.03 89 ± 1 48 1.31 ± 0.03 401 ± 17
SAN-24-C 0.27 117 ± 2 3.75 ± 0.04 89 ± 1 42 1.08 ± 0.06 277 ± 11
SAN-24-C 0.63 116 ± 1 3.74 ± 0.07 90 ± 1 40 0.93 ± 0.03 186 ± 8
SAN-24-C 1.25 116 ± 1 3.77 ± 0.02 90 ± 1 41 0.66 ± 0.04 117 ± 14
SAN-34-0 0 105 4.14 ± 0.03 89 ± 1 75 2.11 ± 0.08 963 ± 25
SAN34-B 0.10 105 4.21 ± 0.07 89 ± 1 77 2.30 ± 0.02 1166 ± 45
SAN34-B 0.50 105 4.24 ± 0.06 89 ± 1 78 2.18 ± 0.07 1043 ± 89
SAN34-B 1.00 105 4.33 ± 0.05 89 ± 1 75 1.93 ± 0.12 814 ± 69
18
Table 4 Summary of the values of Young’s modulus, E, yield stress, σy, fracture stress, σf
fracture toughness, Kc, and fracture energy, Gc, for the unmodified and nano-modified SAN-24
and SAN-34 composites tested at 80 °C.
SystemNano-
contentE σy σf Kc Gc
vol.% (GPa) (MPa) (MPa) (MPa·m1/2) (J/m2)
SAN24-B
0 3.19 ± 0.03 58 ± 1 34 0.89 ± 0.06 217 ± 17
0.05 3.24 ± 0.07 57 ± 1 35 1.07 ± 0.05 307 ± 12
0.10 3.29 ± 0.06 57 ± 1 36 1.09 ± 0.03 315 ± 12
0.30 3.35 ± 0.07 57 ± 1 33 1.12 ± 0.07 345 ± 15
0.50 3.40 ± 0.05 57 ± 1 37 1.14 ± 0.01 346 ± 12
SAN34-B
0 3.43 ± 0.10 59 ± 1 43 1.49 ± 0.04 561 ± 32
0.10 3.47 ± 0.04 58 ± 2 45 1.54 ± 0.05 632 ± 26
0.50 3.52 ± 0.01 59 ± 2 45 1.58 ± 0.08 677 ± 11
1.00 3.57 ± 0.06 59 ± 1 44 1.60 ± 0.03 699 ± 11
Table 5 Summary of the toughening model input parameters and results for the various material
systems investigated where X, x, Ga, c/y denote: toughening factor, critical stress intensity
factor, interfacial energy and normalised critical stress. Following that sequence the model
parameters are calculated from eqns. (2), (3), (4) and again (4). X is determined from the
toughness enhancement at ϕ=0.1 %.
System T
°C
E
(GPa)
Y
(MPa)
ro
(nm)
X x Ga
(J/m2)
σc
σY
SAN-24-A 23 3.9 88 20 190 8.9 1.00 5.9
SAN-24-B 23 3.9 88 6 410 9.9 0.36 6.6
SAN-24-C 23 3.9 88 5 270 9.7 0.33 9.7
SAN-24-B 80 3.3 57 6 450 10.0 0.19 6.7
SAN-34-B 23 4.2 89 6 210 9.0 0.29 6.0
SAN-34-B 80 3.5 59 6 127 8.4 0.13 5.6
19
Table 6 Comparison of the ratios of critical stress for debonding c, and process zone radius, c for the SAN-24 and SAN-34 matrices at 80°C and 23°C=RT
Matrix c, 80 °C/c, RT c80 °C/cRT
SAN-24 0.66 1.32
SAN-34 0.62 1.25
20
List of Figure Captions
Fig. 1 AFM micrographs of SAN-24 containing (a) 0.10 vol.% Type-A, (b) 2.00 vol.% Type-A,
(c) 0.10 vol.% Type-B, (d) 0.50 vol.% Type-B, (e) 0.10 vol.% Type-C and (f) 1.25 vol.% Type-
C nano-particles. The arrows point to selected nano-particles while the circles indicate some
large sized agglomerates. Scale: Images are 2m x 2m.
Fig. 2 Typical DMTA traces showing the storage modulus and loss factor of SAN-24 containing 0.50 vol.% Type-B nano-rods tested at 1 Hz and 10 Hz.
Fig. 3 Normalised Young’s modulus versus particle volume fraction for the SAN-24
composites modified with Types A, B and C particles. Solid lines are best linear fits to the
normalised Young’s modulus and the horizontal dashed line represents zero enhancement of
stiffness. For well bonded particles at low volume fractions the slope is determined by Ep/Em,
the ratio of particle to matrix modulus, and expected to be 35. SAN-24-B follows this line at low
-values. Deviations would indicate interfacial failure (debonding) i.e. either particle-matrix or
particle-particle for agglomerates.
Fig. 4 Normalised fracture energy at 23 °C versus particle volume fraction for the SAN-24
composites modified with Types A, B and C particles. The horizontal dashed line represents
zero enhancement of fracture energy.
Fig. 5 Normalised fracture energy versus particle volume fraction for the SAN-24 and SAN-34 composites modified with Type-B nano-rods tested at 23 °C. The horizontal dashed line represents zero enhancement fracture energy.
Fig. 6 Normalised fracture energy versus particle volume fraction for the (a) SAN-24 and (b) SAN-34 composites modified with Type-B nano-rods tested at 23 °C and 80 °C. Whilst fracture energy exhibits a maximum at quite low particle volume fractions at 23 °C a continuous increase is observed at a temperature closer to, but still below, the glass transition temperature of the matrix. The gain is highest for the more brittle SAN-24 matrix.
21
Fig. 7 SEM micrographs showing the fracture surfaces of: (a) unmodified SAN-24, (b) SAN-24 containing 0.05 vol.% Type-B nano-rods, (c) unmodified SAN-34 and (d) SAN-34 containing 0.10 vol.% Type B nano-rods tested at 23 °C (Crack propagation was from left to right). Arrows point to some voids and the circles indicate fibrillation of the matrix.
Fig. 8 SEM micrographs showing the fracture surfaces of: (a) unmodified SAN-24, (b) SAN-24 containing 0.10 vol.% Type-B nano-rods, (c) SAN-24 containing 0.50 vol.% Type-B nano-rods and (d) SAN-34 containing 1.00 vol.% Type-B nano-rods tested at 80 °C (Crack propagation was from left to right). Selected voids are indicated with arrows.
22
(b)(a)
(d)(c)
C 0.1% C 1.25%
Fig. 4 AFM micrographs of SAN-24 containing (a) 0.10 vol.% Type-A, (b) 2.00 vol.% Type-A,
(c) 0.10 vol.% Type-B, (d) 0.50 vol.% Type-B, (e) 0.10 vol.% Type-C and (f) 1.25 vol.% Type-
C nano-particles. The arrows point to selected nano-particles while the circles indicate some
large sized agglomerates.
24
(f)
500 nm500 nm
(e)
0 20 40 60 80 100 120 140 1600
1
2
3
4
5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
Modulus @ 1 Hz
Modulus @ 10 Hz
tan(δ) @ 1 Hz
tan(δ) @ 10 Hz
Temperature, (°C)
Stor
age
mod
ulus
, E' (
GPa
)
Loss
fact
or, t
an(δ
)Fig. 5 Typical DMTA traces showing the storage modulus and loss factor of SAN-24 containing
0.50 vol.% Type-B nano-rods tested at 1 Hz and 10 Hz.
25
Fig. 6 Normalised Young’s modulus versus particle volume fraction for the SAN-24
composites modified with Types A, B and C particles. Solid lines are best linear fits
to the normalised Young’s modulus and the horizontal dashed line represents zero
enhancement of stiffness. For well bonded particles at low volume fractions the
slope is determined by Ep/Em, the ratio of particle to matrix modulus, and expected
to be 35. SAN-24-B follows this line at low -values. Deviations would indicate
interfacial failure (debonding) i.e. either particle-matrix or particle-particle for
agglomerates.
26
Fig. 4 Normalised fracture energy at 23 °C versus particle volume fraction for the SAN-24
composites modified with Types A, B and C particles. The horizontal dashed line represents
zero enhancement of fracture energy.
27
Fig. 5 Normalised fracture energy versus particle volume fraction for the SAN-24 and SAN-34
composites modified with Type-B nano-rods tested at 23 °C. The horizontal dashed line
represents zero enhancement fracture energy.
28
(b)
Fig. 6 Normalised fracture energy versus particle volume fraction for the (a) SAN-24 and (b) SAN-34 composites modified with Type-B nano-rods tested at 23 °C and 80 °C. Whilst fracture energy exhibits a maximum at quite low particle volume fractions at 23 °C a continuous increase is observed at a temperature closer to, but still below, the glass transition temperature of the matrix. The gain is highest for the more brittle SAN-24 matrix.
30
Fig. 7 SEM micrographs showing the fracture surfaces of: (a) unmodified SAN-24, (b) SAN-24
containing 0.05 vol.% Type-B nano-rods, (c) unmodified SAN-34 and (d) SAN-34 containing
0.10 vol.% Type B nano-rods tested at 23 °C (Crack propagation was from left to right). Arrows
point to some voids and the circles indicate fibrillation of the matrix.
31
(b)(a)
(d)(c)
Fig. 8 SEM micrographs showing the fracture surfaces of: (a) unmodified SAN-24, (b) SAN-24
containing 0.10 vol.% Type-B nano-rods, (c) SAN-24 containing 0.50 vol.% Type-B nano-rods
and (d) SAN-34 containing 1.00 vol.% Type-B nano-rods tested at 80 °C (Crack propagation
was from left to right). Selected voids are indicated with arrows.
32
(b)(a)
(d)(c)