You are on page 1of 7

16136829, 2012, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.201102686 by Cochrane Portugal, Wiley Online Library on [13/12/2022].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
full papers
Nanocomposite Creep

Control of Epoxy Creep Using Graphene


Ardavan Zandiatashbar, Catalin R. Picu,* and Nikhil Koratkar

The creep behavior of epoxy–graphene platelet (GPL) nanocomposites with different


weight fractions of filler is investigated by macroscopic testing and nanoindentation.
No difference is observed at low stress and ambient temperature between neat epoxy
and nanocomposites. At elevated stress and temperature the nanocomposite with the
optimal weight fraction, 0.1 wt% GPLs, creeps significantly less than the unfilled
polymer. This indicates that thermally activated processes controlling the creep rate
are in part inhibited by the presence of GPLs. The phenomenon is qualitatively
similar at the macroscale and in nanoindentation tests. The results are compared
with the creep of epoxy–single-walled (SWNT) and multi-walled carbon nanotube
(MWNT) composites and it is observed that creep in both these systems is similar to
that in pure epoxy, that is, faster than creep in the epoxy–GPL system considered in
this work.

1. Introduction and fatigue response of this material is of significant practical


importance.
Epoxy is a thermosetting polymer which is used in a wide range
Polymer properties can be modified by mixing with a
of applications as an adhesive, matrix material for various
second phase. In particular, the stiffness and strength of
composites, and for components with a structural role. Epoxies
thermosetting polymers can be increased by the addition of
are generally mixtures of a base resin and a crosslinker (hard-
stiff fibers, and their toughness can be enhanced by embed-
ener) and are molecular networks.[1,2] The average length of
ding rubber inclusions. Over the last decade, composites in
the resin molecule segment between crosslinks is on the nano-
which the second phase has nanoscale dimensions have been
meter scale and hence these network segments are rather stiff
studied extensively with the hope that synergies in the nano-
in both bending and axial deformation modes. This provides
scale material behavior can be exploited to lead to exceptional
significant strength and creep resistance to the network.
macroscale properties. In thermoplastics it was shown that
Since epoxies have a structural role in most applications,
strength and ductility can be increased simultaneously or at
their mechanical stability is essential. Their creep and resist-
least, one of these properties may be increased without com-
ance to fracture under repeated loading conditions (fatigue)
promising the other.[5] These effects have been attributed to
often control the equivalent properties of the structure in
the fact that the presence of fillers modifies the structure and
which they are used. For example, the ultimate failure of an
dynamics of polymeric chains, that is, the matrix material is
epoxy-matrix composite is usually controlled by delamination
modified by the presence of the second phase. This is a concept
and fracture of the epoxy matrix.[3] Similarly, the collapse of
fundamentally different from that controlling reinforcement
the ceiling of the Big Dig tunnel in Boston, MA, in 2006 was
in regular composites in which the properties of individual
traced to creep in epoxy.[4] Therefore, improving the creep
phases remain identical to those of the bulk, and reinforce-
ment is provided by the simple presence of inclusions.
Much more modest claims have been made so far
A. Zandiatashbar, Prof. C. R. Picu, Prof. N. Koratkar regarding improvements in nanocomposites with a thermo-
Department of Mechanical, Aerospace, and Nuclear Engineering setting matrix. The strength and ductility appear to be con-
Rensselaer Polytechnic Institute trolled by the molecular network of the matrix, which appears
Troy, NY 12180, USA
not to be modified significantly by the presence of nanoscale
E-mail: picuc@rpi.edu
inclusions. However, when the crosslinking density of the
matrix is modified by nanofillers, the macroscopic properties
DOI: 10.1002/smll.201102686 do change. It was shown, for example, that epoxy filled with

1676 wileyonlinelibrary.com © 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2012, 8, No. 11, 1676–1682
16136829, 2012, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.201102686 by Cochrane Portugal, Wiley Online Library on [13/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Control of Epoxy Creep Using Graphene

pristine carbon nanotubes (CNTs) exhibits a slower crack Neat Epoxy


growth rate under fatigue loading conditions at small applied GPL/Epoxy 0.1%
3 GPL/Epoxy 0.3%
loads (stress intensity factor amplitudes), but loses these
GPL/Epoxy 0.5% 40 MPa
properties at high loads.[6,7] When the nanotubes are function-
alized to obtain a strong interface with the matrix, the fatigue

Creep Strain (%)


crack growth rate is reduced by an order of magnitude over a
broad range of stress intensity factor amplitudes.[8] 2
Graphene, a single layer of carbon atoms sp2-bonded in a
hexagonal array, has attracted attention recently for its inter-
esting mechanical and transport properties.[9–11] Its use as a filler
material (in the form of platelets) in polymer nanocomposites 1
was also studied.[12] At present there is no efficient technique
20 MPa
to produce ideal, single-layer graphene sheets for use in nano-
composite applications.[13] This situation has directed attention
to the use of graphene platelets (GPLs), which are cheap and 0
0 30000 60000 90000 120000
easy to produce in large quantities.[12,14,15] The results discussed
to date indicate that while graphene affects the monotonic Time (sec)
mechanical properties weakly, it has a far more powerful effect Figure 1. Creep strain versus time for neat epoxy and epoxy–GPL of
on the fatigue crack growth behavior of the nanocomposite.[16] different filler contents, at stresses of 20 and 40 MPa and at room
A number of studies have addressed creep in nanocom- temperature.
posites. Zhang et al. studied the effect of CNT nanofillers on
the creep behavior of epoxy[17] and it is shown that single-
stress levels represent 35 and 70% of the yield stress, respec-
tively. The load is increased fast (σ⋅ = 200 MPa s−1) and then it
walled CNTs (SWNTs), unlike multi-walled CNTs (MWNTs)
and fullerene, lead to improved creep behavior of epoxies at
is held constant. The time at the beginning of the hold period
room temperature and low stresses, while they are not effec-
is considered the initial time for creep. The creep strain was
tive at higher stresses. Sheng et al. studied the creep behavior
recorded for 36 h. No necking or significant variation of the
of bisphenol E cyanate ester (BECy)–alumina nanocompos-
specimen cross-sectional area was observed during the test,
ites and observed an increase of the creep compliance rela-
so the effective stress is that prescribed throughout the entire
tive to neat BECy polymer at all temperatures between 190
test. Figure 1 shows the creep strain (total strain minus the
and 320 °C.[18] In another work it was shown that montmo-
elastic strain corresponding to the creep load) versus time for
rillonite-layered silicate (MLS) improves tensile mechanical
the four materials and for the two stress levels tested.
properties as well as the creep recovery of polyethylene.[19]
The creep behavior is essentially identical for filled and
Lam et al. studied creep in epoxy–nanoclay nanocompos-
pristine epoxy at the smaller load (20 MPa). However, at
ites using nanoindentation and concluded that 4 wt% is the
the larger stress the four materials respond differently. The
optimal filling fraction leading to the smallest creep strain.[20]
smallest creep strain is measured in the nanocomposite with
To our knowledge, the effect of graphene on creep in
0.1 wt% GPLs, which deforms significantly less than pristine
epoxies has not been studied to date. This is the objective of
epoxy. In fact, the strain at the end of the hold period (after
the present work. The time-dependent mechanical behavior
36 h) is 15% smaller in the 0.1 wt% GPL composite than in
is studied at the macroscopic scale by performing standard
pristine epoxy. The other composites (0.3 and 0.5 wt% GPLs)
creep tests, and at the nanoscale, by nanoindentation creep.
creep more than the unfilled polymer.
The results indicate significant creep rate reduction in the
The observation that the optimal properties are exhibited
nanocomposite with the optimal weight fraction of nano-
by the composite with 0.1 wt% GPLs is in agreement with
fillers. The effect depends on temperature and stress level,
previous work on this system, in which it was shown that the
being more pronounced at larger stresses and at elevated
composite with this weight fraction has the highest fracture
temperatures. For comparison, creep of epoxy filled with
toughness.[16] Specifically, the fracture toughness increases
SWNTs and MWNTs was also studied in the same range of
with increasing weight fraction of nanofillers up to 0.1 wt%,
stress and temperature and, consistent with previous work,[17]
after which it decreases. This behavior was associated with
it was observed that both epoxy–SWNT and epoxy–MWNT
the quality of filler dispersion. It has been observed that the
composites creep at the same rate as pure epoxy. Hence,
optimal dispersion obtained with this mixing method results
these additives do not improve the creep response relative to
in the 0.1 wt% GPL sample, while agglomeration is seen upon
that of pure epoxy.
increasing the filler fraction. Since GPLs bond to each other
only through van der Waals interactions, a cluster of GPLs
is quite weak when subjected to tension. If the cluster is not
2. Results and Discussion infiltrated by polymer, one may consider it as a nanovoid
from the mechanical point of view. Clearly, these large inclu-
2.1. Macroscopic Creep Behavior at Room Temperature sions should weaken the material rather than strengthen it.
To determine the distribution, the fracture surfaces in the
Creep tests were performed with a nominal stress of 20 and various materials studied were observed by scanning electron
40 MPa, in separate experiments, at room temperature. These microscopy (SEM) and the size of inclusions was estimated

small 2012, 8, No. 11, 1676–1682 © 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.com 1677
16136829, 2012, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.201102686 by Cochrane Portugal, Wiley Online Library on [13/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
full papers A. Zandiatashbar et al.

Figure 2. a) Fracture surface of a sample with 0.1 wt% filling fraction showing individual GPLs and river pattern marks. b) Probability distribution
function of nanofiller size as observed on the fracture surface of nanocomposites with 0.1 and 0.5 wt% GPLs.

using the linear intercept method described in the ASTM by the function ε(t) ∼ t 1/q . The exponent q is 2.49 for epoxy
standard E112. Figure 2a shows a SEM image of the fracture and 4.71 for the nanocomposite. This functional form of the
surface of a sample with 0.1 wt% GPLs. Inclusions having pro- strain–time relationship results from a constitutive equation
jected dimensions of approximately 1 μm are clearly visible, of the form [Eq. (1)]:
along with river pattern marks produced during the fracture  
∂  q σ (t) pq
process. The river pattern features are clearly distinct from ε(t) = (1)
the inclusions. Large domains of this type were inspected. To ∂t μc
quantify clustering, Figure 2b shows a plot of the inclusion size
for the viscoelastic body, where σ is the stress, and μc and
distribution in composites with 0.1 and 0.5 wt% GPLs. The
p are constants. Evolution equations of this type have been
distribution is fairly broad in both cases, and shifts to larger
used to represent the mechanical behavior of complex vis-
inclusion sizes as the GPL fraction increases. Agglomeration
coelastic liquids and solids,[21] especially glasses,[22] and may
is observed at all filling fractions (see the tail of the distribu-
be expressed formally in terms of fractional calculus.[23] They
tion), but it is more pronounced at higher filler loading. The
lead to slow, power law relaxation behavior and are indicative
0.1 wt% composite has the largest fraction of inclusions with
of hierarchical relaxation processes taking place in the mate-
projected dimensions below 1 μm, that is, individual GPLs.
rial. Multiple molecular-scale mechanisms may lead to this
The curves in Figure 1 can be approximated with a power
type of relaxation, but the specific mechanism taking place in
law function of time. Figure 3 shows the 40 MPa curves for
molecular networks is not entirely understood at present.
pristine epoxy and 0.1 wt% GPLs obtained at room tem-
perature in log–log coordinates. The curve is approximated
2.2. Macroscopic Creep Behavior at Elevated Temperatures

As shown in Figure 1, the nanocomposites creep identically


to the pristine epoxy samples at room temperature and low
1 stress. It is interesting to investigate the effect of tempera-
ture on the creep response, while keeping the stress level low
Log (Creep Strain)

(20 MPa). To this end we consider the 0.1 wt% GPL system.
Figure 4 shows the creep strain versus time for both pristine
and nanocomposite samples at room temperature and at T =
40 and 55 °C. At room temperature the curves corresponding
to the two materials overlap (see also Figure 1). However,
0.1
differences are seen as the temperature increases. At 55 °C
the epoxy creeps significantly more than the nanocomposite.
Neat Epoxy The difference is observed mostly in the transient part of
GPL/Epoxy 0.1 wt%
the creep curve, with the creep rates in the steady-state part
being relatively close to each other.
10 100 1000 10000 100000 The data in Figure 1 and Figure 4 indicate that the
Log(Time) molecular mechanisms leading to differences between pris-
Figure 3. Log–log plot of room-temperature creep strain versus time tine and filled epoxy, that is, those producing more deforma-
for neat epoxy and epoxy–0.1 wt% GPL showing a power law creep tion in unfilled epoxy, are activated by temperature (T) and
compliance function. stress (σ). Let us consider that these thermally activated

1678 www.small-journal.com © 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2012, 8, No. 11, 1676–1682
16136829, 2012, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.201102686 by Cochrane Portugal, Wiley Online Library on [13/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Control of Epoxy Creep Using Graphene

Neat Epoxy o Neat Epoxy


T = 55 C
GPL/Epoxy 0.1wt% GPL/Epoxy 0.1wt%
30

25

20 1E-5

Strain Rate (1/s)


Creep Strain (%)

15

10 o
T = 40 C
1E-6

5 o
T = 55 C

o
T = 40 C

o 1E-7
T = 23 C
0 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Strain
0 15000 30000 45000 60000 75000
Time (sec) Figure 5. Strain rate versus strain for neat epoxy and epoxy–0.1 wt% GPLs
at temperatures of 40 and 55 °C. Tests were performed at 20 MPa.
Figure 4. Creep strain versus time for neat epoxy and epoxy–0.1 wt%
GPL at a stress level of 20 MPa and temperatures of 23, 40, and 55 °C.
The strain rate versus strain plots corresponding to the
higher-temperature curves in Figure 4 are shown in Figure 5.
As the temperature increases, the strain rate differential at
processes can be described by an Arrhenius expression,
given strain between the nanocomposite and pristine epoxy
ε⋅ ∝ exp(−E a (σ )) kB T ), where Ea is the stress-dependent
increases. At larger strains, the strain rate of the composite is
activation energy and kB is the Boltzmann constant. At σ =
more than an order of magnitude smaller than that of epoxy.
20 MPa, the activation energy for creep in epoxy is 1.05 eV,
The difference between the two curves at T = 55 °C increases
while in the nanocomposite it becomes 1.12 eV. Given that
even more with time. This is a significant reduction of the
the difference between the creep of pure epoxy and that of
creep rate in conditions in which creep resistance is more
the nanocomposite depends strongly on temperature, the dif-
critical (higher temperatures and stresses).
ference in activation energy reported here is significant and
outside of the measurement noise. The atomistic/molecular
mechanisms leading to this difference are not fully under- 2.3. Comparison with the Creep of Epoxy–Carbon
stood at this time, but it is conjectured that the good bonding Nanotube Composites
between GPLs and epoxy (as shown in Reference [22]) has a
contribution. Another indication that individual GPLs bond The comparison of the results reported in the previous sec-
well with epoxy is provided by the nanoindentation tests dis- tion with the creep behavior of epoxy–CNT composites is of
cussed in Section 2.4. The comparison with data obtained for interest. To this end, samples of epoxy–SWNT and epoxy–
epoxy–SWNT and epoxy–MWNT (Section 2.3) also sheds MWNT were processed using the method employed for
light on this issue. epoxy–GPL. The filling fraction was selected as 0.1 wt% to
It is appropriate to inquire if these differences are due allow direct comparison with the GPL composite exhibiting
to the variation of the glass transition temperature, Tg, in optimal mechanical performance.
the nanocomposite relative to epoxy. Tg was measured using Figure 6 shows the creep strain versus time for neat epoxy
differential scanning calorimetry (DSC) for neat epoxy and and epoxy–GPL, epoxy–SWNT, and epoxy–MWNT compos-
nanocomposites of 0.1, 0.3, and 0.5 wt% GPLs. Samples were ites subjected to 20 MPa load at 55 °C. The curves for neat
heated and cooled in two cycles between room temperature epoxy and epoxy–GPL are identical to those in Figure 4. It
and 180 °C with a heating/cooling rate of 10 °C min−1 and is observed that the curves corresponding to the epoxy–CNT
5-min hold periods between heating and cooling cycles. Tg systems are very close to that for neat epoxy. The epoxy–GPL
was measured in the second heating cycle. The values range composite creeps significantly less than all other composites
between 73 and 79 °C for all these materials. While the mean studied. Differences are largely observed during the tran-
values of Tg seem to increase with the filler content, the sient part of creep, with the creep rates in the steady-state
experimental noise is rather large (±2 °C) and a clear state- regime being relatively close to each other at a given time.
ment cannot be made. However, the difference observed in This behavior is similar to that described previously[17] and
Figure 4 cannot be due to the small difference of Tg between is attributed to the fact that the bonding between the CNTs
the various materials, primarily because this argument pre- and epoxy is weaker than that between GPLs and epoxy.[24]
dicts the opposite trend relative to the experiment. At a given While the CNTs used in this work (and in Reference [17])
temperature, the nanocomposite is closer to its Tg than the are pristine, the GPL surface is functionalized[16] by the pres-
unfilled epoxy, which suggests that more creep should be ence of oxygen and OH groups, which enhance the interac-
observed in epoxy–GPL. tion with the epoxy matrix. The warped configuration of most

small 2012, 8, No. 11, 1676–1682 © 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.com 1679
16136829, 2012, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.201102686 by Cochrane Portugal, Wiley Online Library on [13/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
full papers A. Zandiatashbar et al.

35 o represents the mean of nine indentation measurements and


T = 55 C
the error bars indicate the variability (standard deviation)
30 from indentation site to indentation site. As on the macro-
scopic scale, the nanocomposite creeps less than the unfilled
25 polymer. The inset shows the same data in logarithmic coor-
Creep Strain (%)

dinates indicating that, as in Figure 3, the variation in time


20 of the creep response can be described by a power law, t1/q,
with exponent q = 1.23 ± 0.05 for epoxy and 1.33 ± 0.18 for
15 the nanocomposite. The indicated variability is associated
with the response measured at different points on a given
10 Neat Epoxy sample. The discrepancy between the values of q measured
SWNT / Epoxy 0.1 wt% at the macroscopic and nanoscale is attributed to the fact that
MWNT / Epoxy 0.1 wt%
5 the indentation creep tests can only last for a relatively short
GPL / Epoxy 0.1 wt%
time (1000 s).
0 The indentation displacement, d(t), obtained with a flat
0 15000 30000 45000 60000 75000 punch can be related to the creep compliance of the viscoe-
Time (sec) lastic material, J(t), via the relation [Eq. (2)]:[25]
Figure 6. Creep strain versus time for neat epoxy, epoxy–0.1 wt% GPLs,
4a
epoxy–SWNT, and epoxy–MWNT at a stress level of 20 MPa and 55 °C. J (t) = d(t) (2)
(1 − ν)P0

GPL inclusions is also a factor which may contribute to the where a, ν, and P0 are the contact radius, Poisson’s ratio, and
slowing down of creep in this system. the indentation load, respectively. Hence, this nanoscale test
allows identification of a material property which is usually
accessible only in macroscopic uniaxial tests. Due to this
2.4. Nanoscale Creep Response equivalence, we may interpret the power law fit to J(t) in the
same way as discussed in conjunction with the macroscopic
The macroscopic creep behavior is compared next with the local, test, that is, this indicates that small strain deformation (incip-
nanoscale creep response. This is determined using nanoinden- ient creep) is controlled by hierarchical processes without a
tation. The indenter is brought into contact with the surface and single, dominant relaxation time.
the load is increased at a constant load rate up to 2 mN. The The variability is more pronounced in the nanocomposite,
load is then held constant for 1000 s and the penetration of the as expected; however, the difference in these fluctuations
probe is monitored. The creep time and creep displacement are between the composite and unfilled epoxy is not very large.
measured relative to the beginning of the hold period. Multiple This is due to the fact that the flat punch diameter (50 μm) is
indentations are performed in each material, in arrays of 3 × 3 more than one order of magnitude larger than the inclusion
indentation points with 200 μm distance between them. size (see Figure 2).
Figure 7a shows the indentation displacement versus creep It is interesting to note that significant variability is
time for pristine epoxy and the 0.1 wt% GPL nanocomposite observed in pristine epoxy when probed on this length scale.
obtained using a flat punch (50 μm in diameter). Each curve Generally, one considers epoxy to be homogeneous down to

(a) log(t) (b)


100 1000
300 250
Neat Epoxy
GPL/Epoxy 0.1%
log(d)

250 100
200
Creep Displacement, d (nm)
Creep Displacement, d (nm)

200
150

150
100
100

50
50 Neat Epoxy
GPL/Epoxy 0.1%

0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
Time (sec) Time (sec)

Figure 7. Nanoindentation creep data obtained with a) a 50-μm-diameter flat punch, and b) a Berkovich tip. The creep displacement versus time
curves obtained with the flat punch can be approximated with a power law (inset in (a)).

1680 www.small-journal.com © 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2012, 8, No. 11, 1676–1682
16136829, 2012, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.201102686 by Cochrane Portugal, Wiley Online Library on [13/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Control of Epoxy Creep Using Graphene

length scales of a few tens of nanometers, where the effect of and the tube was placed in a tube furnace (Thermolyne 79300,
the underlying molecular network structure should become Thermo Fisher Scientific Inc., USA) immediately. The furnace was
visible. Studies of material homogeneity have been con- preheated to 1050 °C and the tube was held in the furnace for
ducted using transmission electron microscopy (TEM) and 40 s. The high heating rate of more than 2000 °C min−1 leads to
swelling experiments.[26] Density fluctuations were observed the exfoliation of graphite oxide to GPLs, as demonstrated in Ref-
in TEM,[26] but later those were discarded as artifacts. We erences [12,15]. The GPLs obtained by this method had residual
did not perform TEM studies of the pristine material, but the hydroxyl groups on the surface, which helped their dispersion in
present probing indicates that the mechanical behavior is not polar solvents.[27–30]
uniform throughout the sample. Nanocomposite Sample Preparation and Processing: A ther-
Figure 7b shows similar indentation data obtained with mosetting bisphenol A blend epoxy resin (Epoxy 2000) and a
the Berkovich indenter. The trends are similar, but the dif- curing agent (2120 Epoxy Hardener), both from Fibreglast, USA,
ference between the pristine and filled materials is more pro- were used for preparing the nanocomposites. To disperse the
nounced. The distinct response obtained with the Berkovich graphene fillers in the epoxy resin the following procedure was fol-
tip is due to the tip geometry, which mandates that the lowed.[16,31] A suitable amount of GPLs was dispersed in acetone,
volume of material probed increases continuously during which is recognized as a polar solvent (100 mL acetone per each
indentation, and to the fact that this tip is much smaller than 0.1 g GPLs), using an ultrasonic probe sonicator (Sonics Vibracell
the flat punch. It is interesting to observe that in all indenta- VC 750, Sonics and Materials Inc., USA) at high amplitude for 1 h.
tion experiments, with the flat punch and with the Berkovich The epoxy resin was added to the suspension of GPLs/acetone
tip, the initial creep response of the composite and the pris- and the ultrasonication procedure was repeated for one more
tine epoxy are identical. hour. After this, acetone was removed from the mixture of epoxy
resin and GPLs by heating the mixture while stirring using a Teflon-
coated magnetic bar on a magnetic stir plate for 3.5 h at 70 °C. The
3. Conclusion mixture was then placed in an oven for 8 h at 70 °C to insure that
all the solvent had been removed. Cooling was performed before
A set of epoxy-based nanocomposites with 0.1, 0.3, and
adding the curing agent to avoid any curing at higher than ambient
0.5 wt% GPLs was studied in this work. When tested for
and uncontrolled temperature. The hardener was then added and
creep, the nanocomposite with 0.1 wt% GPLs creeps less
mixing was performed using a high-speed shear mixer (ARE-250,
than all other materials, including the neat epoxy. The differ-
Thinky, Japan) at 2000 rpm for 4 min. The mixture was degassed
ence is small at small stress levels and room temperature, but
in a vacuum chamber for ≈20 min and poured into silicone molds.
increases with increasing stress and temperature. An order of
The resulting samples were cured in the mold under 90 psi pres-
magnitude difference between the creep rates of neat epoxy
sure, at room temperature for 24 h. This was followed by 4 h of
and 0.1 wt% GPLs is observed at 55 °C and 20 MPa. The
post-curing at 90 °C.
creep strain versus time curves can be fitted with a power
Herein, we consider pure epoxy and epoxy–GPL nanocompos-
law function with exponents larger than 1. This indicates that
ites with 0.1, 0.3, and 0.5 wt% GPLs. Unfilled epoxy samples were
hierarchical deformation processes control creep deforma-
prepared for the purpose of these tests as described above, that is,
tion. Furthermore, the increase of the difference between
by using the same procedure as with the nanocomposite samples.
the neat polymer response and that of the nanocomposites
SWNTs (diameter ≈1–2 nm, length ≈1–3 μm) and MWNTs
with increasing temperature and, to a smaller extent with
(diameter ≈10–20 nm, length ≈10–30 μm) were purchased from
increasing stress, indicates that thermally activated processes
Nanostructured & Amorphous Materials Inc. Composites of epoxy–
which are active in neat epoxy are inhibited by the presence
0.1 wt% SWNTs and epoxy–0.1 wt% MWNTs were produced by
of GPLs. The creep of the epoxy–GPL system was compared
using the method described for epoxy–GPL.
with that of epoxy–SWNT and epoxy–MWNT and it was
Mechanical Creep Testing: Macroscopic tensile creep tests
concluded that epoxy–CNT systems creep at the same rate as
were conducted in load control using an MTS-858 (MTS Systems
neat epoxy at elevated temperatures. The epoxy–GPL creeps
Corp., MN, USA) and dog-bone tensile test samples with gauge
significantly less, which is interpreted as being a consequence
length 25 mm and thickness 4 mm. The strain was measured using
of the stronger bond between GPLs and the epoxy.
an extensometer (MTS 632.26E-20). In tests performed at elevated
temperature, an environmental chamber was used and the tem-
perature was controlled with ±1% accuracy.
4. Experimental Section Nanoscale Creep Testing: Nanoscale creep tests were per-
GPL Preparation: GPLs were produced by the thermal reduc- formed with a NanoTest nanoindenter (Micro Materials Ltd.,
tion of graphite oxide. In this method, graphite oxide was first Wrexham, UK) under load control. The load was increased at a
prepared from natural graphite flakes by oxidation in a solu- rate of 0.01 mN s−1 up to 2 mN and then held constant. Two
tion of nitric acid (68%), sulfuric acid (95–98%), and potassium types of probes were used: a flat punch with diameter 50 μm
chlorate (99.5%) for 96 h. For thermal exfoliation, graphite oxide and a Berkovich diamond tip. The nanoindentation tests were
(≈200 mg) was placed in a quartz tube (inner diameter: 200 mm, performed in a temperature-controlled chamber at 27 °C.
length: 1 m). The tube was sealed at one end with an inlet and The samples used for the nanoscale testing were obtained
outlet for argon flow and the other end was closed by a rubber from the same batches as the samples used for macroscopic
stopper. The graphite oxide was flushed with argon for ≈10 min experiments.

small 2012, 8, No. 11, 1676–1682 © 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.com 1681
16136829, 2012, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.201102686 by Cochrane Portugal, Wiley Online Library on [13/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
full papers A. Zandiatashbar et al.

[14] J. R. Potts, D. R. Dreyer, C. W. Bielawski, R. S. Ruoff, Polymer


2011, 52, 5.
[15] H. C. Schniepp, J.-L. Li, M. J. McAllister, H. Sai, M. Herrera-Alonso,
Supporting Information D. H. Adamson, R. K. Prud’homme, R. Car, D. A. Saville, I. A. Aksay,
J. Phys. Chem. B 2006, 110, 8535.
Supporting Information is available from the Wiley Online Library [16] M. A. Rafiee, J. Rafiee, I. Srivastava, Z. Wang, H. Song, Z.-Z. Yu,
or from the author. N. Koratkar, Small 2010, 6, 179.
[17] W. Zhang, A. Joshi, Z. Wang, R. S. Kane, N. Koratkar, Nanotech-
nology 2007, 18, 185703.
[18] X. Sheng, M. Akinc, M. R. Kessler, Mater. Sci. Eng. A 2010, 527,
Acknowledgements 5892.
[19] A. Shaito, D. Fairbrother, J. Sterling, N. A. D’Souza, Polym. Eng.
We gratefully acknowledge the support from the US National Sci- Sci. 2010, 50, 1620.
ence Foundation through grant CMMI-0900188. [20] C. K. Lam, A. K. T. Lau, L. M. Zhou, Key Eng. Mater. 2007, 334-
335, 669.
[21] I. H. Shames, F. A. Cozzarelli, Elastic and Inelastic Stress Analysis:
Revised Printing, 6th ed., Taylor & Francis, Washington, DC 1997.
[1] W. L. Wu, D. L. Hunston, H. Yang, R. S. Stein, Macromolecules [22] R. Hilfer, J. Non-Cryst. Solids 2002, 305, 122.
1988, 21, 756. [23] K. B. Oldham, J. Spanier, The Fractional Calculus, Academic Press,
[2] J. Duchet, J. P. Pascault, J. Polym. Sci., Part B: Polym. Phys. 2003, New York, NY 1974.
41, 2422. [24] I. Srivastava, R. J. Mehta, Z.-Z. Yu, L. Schadler, N. Koratkar, Appl.
[3] E. A. Armanios, Composite Materials: Fatigue and Fracture, Vol. 6, Phys. Lett. 2011, 98, 063102.
ASTM, Philadelphia, PA 1997. [25] C. Bernard, V. Keryvin, J.-C. Sangleboeuf, T. Rouxel, Mech. Mater.
[4] B. Hansen, Civil Eng. 2007, 77, 18. 2010, 42, 196.
[5] B. J. Ash, R. W. Siegel, L. S. Schadler, Macromolecules 2004, 37, [26] K. Dušek, Angew. Makromol. Chem. 1996, 240, 1.
1358. [27] S. Park, J. A. An, I. Jung, R. D. Piner, S. J. An, X. Li, A. Velamakanni,
[6] W. Zhang, C. R. Picu, N. Koratkar, Nanotechnology 2008, 19, R. S. Ruoff, Nano Lett. 2009, 9, 1593.
285709. [28] S. Stankovich, D. A. Dikin, D. Dommett, K. Kohlhaas, E. J. Zimney,
[7] W. Zhang, C. R. Picu, N. Koratkar, Appl. Phys. Lett. 2007, 91, E. A. Stach, R. D. Piner, S. T. Nguyen, R. S. Ruoff, Nature 2006,
193109. 442, 282.
[8] W. Zhang, I. Srivastava, Y.-F. Zhu, C. R. Picu, N. Koratkar, Small [29] S. Stankovich, R. D. Piner, S.-T. Nguyen, R. S. Ruoff, Carbon 2006,
2009, 5, 1403. 44, 3342.
[9] Y. Zhu, S. Murali, W. Cai, X. Li, J. W. Suk, J. R. Potts, R. S. Ruoff, [30] T. Ramanathan, A. A. Abdala, S. Stankovich, D. A. Dikin,
Adv. Mater. 2010, 22, 3906. M. Herrera-Alonso, R. D. Piner, D. H. Adamson, H. C. Schniepp,
[10] A. K. Geim, K. S. Novoselov, Nat. Mater. 2007, 6, 183. X. Chen, R. S. Ruoff, S. T. Nguyen, I. A. Aksay, R. K. Prud’homme,
[11] O. C. Compton, S. T. Nguyen, Small 2010, 6, 711. L. C. Brinson, Nat. Nanotechnol. 2008, 3, 327.
[12] M. J. McAllister, J.-L. Li, D. H. Adamson, H. C. Schniepp, [31] M. A. Rafiee, J. Rafiee, Z. Wang, H. Song, Z.-Z. Yu, N. Koratkar, ACS
A. A. Abdala, J. Liu, M. Herrera-Alonso, D. L. Milius, R. Car, Nano 2009, 3, 3884.
R. K. Prud’homme, I. A. Aksay, Chem. Mater. 2007, 19, 4396.
[13] A. Dato, V. Radmilovic, Z. Lee, J. Phillips, M. Frenklach, Nano Lett. Received: December 20, 2011
2008, 8, 2012. Published online: February 29, 2012

1682 www.small-journal.com © 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2012, 8, No. 11, 1676–1682

You might also like