Processing and failure behavior of carbon-nanotube composites in sheet forms
Transcript of Processing and failure behavior of carbon-nanotube composites in sheet forms
![Page 1: Processing and failure behavior of carbon-nanotube composites in sheet forms](https://reader037.fdocuments.us/reader037/viewer/2022100102/5750aa3d1a28abcf0cd66b35/html5/thumbnails/1.jpg)
Processing and Failure Behavior of Carbon-NanotubeComposites in Sheet Forms
Wen-Shyong Kuo,1 Hung-Chih Lai,2 Ting-Wei Chang,2 Chan-Chao Wen2
1Department of Aerospace and Systems Engineering, Feng-Chia University, Taichung 407, Taiwan, Republicof China
2Ph.D. Program of Mechanical and Aeronautical Engineering, Feng-Chia University, Taichung 407, Taiwan,Republic of China
In their original forms, carbon nanotubes (CNT) are spa-tially entangled. To make CNT papers (CNTPs), the CNTagglomerates must be dispersed and re-entangled intoa planar sheet. The processing characteristics are verydifferent from those of traditional buck-form nanocom-posites. This article examines the processing, micro-structures, and failure behavior of the CNTP compo-sites. The CNTPs were first made by a dispersion andfiltering process. Then, an epoxy resin was added intothe CNTP by using a vacuum bag method. DifferentCNT weights were employed to make the CNTPs withdifferent thicknesses and areal weights. The CNTPallows direct resin impregnation along the thicknessdirection and avoids the difficulty of dispersing CNTs inthe viscous resin. The CNT content can be much higherthan that attainable in traditional bulk CNT composites.Both tensile and tearing tests were conducted, and thefracture behaviors were examined. POLYM. COMPOS.,00:000–000, 2014. VC 2014 Society of Plastics Engineers
INTRODUCTION
Nanocomposites based upon carbon nanotubes (CNTs)
have been the focus of research in the past two decades.
Like many other fibers, CNTs can be employed to make
paper-like preforms, also known as buckypapers, CNT
papers (CNTPs), or CNT sheets [1–5]. CNTs are naturally
entangled in the form of agglomerates. Compared with
their original forms of irregular agglomerates, the CNTPs
are uniform in thickness and can be manipulated easily as
they are integrated. Because most CNTs are oriented in
the in-plane directions, their mechanical performance
[6–10], thermal and electrical conductive behaviors
[11–16] along the in-plane directions are inherently better
than bulk-type CNT composites. Because of the thin
thickness and the entangled CNT network, the CNTP
composites are highly flexible and can be sharply bent
without disintegration. By adding a suitable matrix mate-
rial, they become CNT composites in the sheet form. The
most notable feature of this nanocomposite is the much
higher CNT content achievable because of the different
processing methods.
In making CNT composites in bulk forms, the inher-
ently entangled CNTs must be properly dispersed and
embedded within a resin [17–19]. However, the viscosity
of the CNT/resin mixture increases drastically with the
CNT loading. Beyond a certain CNT fraction, the mixture
will become too sticky to be processed by typical disper-
sion facilities, such as high-speed mixers, sonication, and
three-roller mills [17, 20]. The previous works indicated
that when the CNT loading reached 5 wt%, the mixture
became a visco-elastic material, and dispersion of CNTs
becomes difficult for the three-roller mill [20, 21]. With-
out proper dispersion, many CNTs are unable to be wet-
ted and bonded by the resin, and the resulting mechanical
behavior can be deteriorated.
In comparison, the high-viscosity problem can be
avoided in CNTP composites because of different proc-
essing approaches. CNTPs are made before resin impreg-
nation. CNTs are dispersed within an ethanol solution
rather than within a polymer resin. At this stage, the pur-
pose of CNT dispersion is to break apart the CNT
agglomerates. After filtration, the individual CNTs can be
deposited and re-entangled to form a CNTP. In this work,
we adopted sonication, and the processing time is critical.
If the processing time is insufficient, the agglomerates
remain too large to form a CNTP. On the other hand, if
the CNTs are overly sonicated, the CNTs can be fractured
and shortened, and the resulting composite properties can
be lowered. Then the CNTP can be impregnated by a
resin and cured to make a CNTP composite. Typical resin
impregnation methods, such as resin transfer molding or
Correspondence to: Wen-Shyong Kuo; e-mail: [email protected]
Contract grant sponsor: National Science Council of Taiwan, R.O.C;
contract grant number: 100-2221-E-035-034-MY3.
DOI 10.1002/pc.23327
Published online in Wiley Online Library (wileyonlinelibrary.com).
VC 2014 Society of Plastics Engineers
POLYMER COMPOSITES—2014
![Page 2: Processing and failure behavior of carbon-nanotube composites in sheet forms](https://reader037.fdocuments.us/reader037/viewer/2022100102/5750aa3d1a28abcf0cd66b35/html5/thumbnails/2.jpg)
resin infusion, are unsuitable for making CNT compo-
sites, because the entangled CNTs are too dense for the
resin to penetrate into. A vacuum-bag method is more
suitable for this, and the resin is permeable along the
thickness direction.
From the microstructure point of view, CNTs in a
CNTP composite are preferentially oriented in the in-
plane directions, in contrast to the 3D random orientation
in buck CNT composites. In theory, the in-plane stiffness,
strength, thermal, and electrical conductivities can be
improved. The impregnated CNTPs are prepreg-like and
can be embedded within carbon-fiber prepregs to make
hybrid composites [22]. The inter-laminar strength and
damping capability can be benefited. Because of the
excellent energy-storage properties of CNTs, potential
applications of the CNTP composites include super-
capacitors and the electrodes for lithium batteries [23,
24]. Because of the conductive, acid-resistant gas-permea-
ble characteristics, CNTPs have the potential to be used
in the gas diffusion layers for fuel cells [25, 26].
The CNTs in CNTPs are dense and integrated in a pla-
nar form, which can be manipulated like typical carbon-
fiber prepregs. These features combine to make this mate-
rial attractive. In this work, CNTPs with different thick-
nesses were made, aiming to study the effect of thickness
on the processing and properties. Since the materials are
thin, they are vulnerable to some forms of loads. The ten-
sile and tearing properties and associated fracture behav-
ior are examined.
MATERIAL PREPARATION
Preparation of CNTP
The preparation of the CNTP is briefly described.
Multi-walled CNTs were employed for making the
CNTP. The CNTs were mixed with ethanol solution for
dispersion. To treat 100 mg of CNTs, 300 ml of the solu-
tion was used. They were first mixed by a magnetic stir-
rer, and then a sonicator (1,500 W, 20 kHz) was
employed to disperse the CNTs. The processing lasted for
30 min. The CNT agglomerates were torn apart into
smaller ones or into individual CNT filaments. The CNT
solution was then filtered by using a porous film placed
within a buffer container. A negative pressure was applied
to the container to enhance CNTP deposition. The CNTP
with the filtering film was dried at 50�C for 24 h. Then
the CNTP was carefully peeled from the film. The CNTP
is circular and 9-cm in diameter. Samples with different
CNT weights were made. The nominal weights for the
specimens were 100, 150, 200, 250, and 300 mg. The
weight of each sample was measured and recorded. A
typical CNTP is shown in Fig. 1.
Resin Infiltration
The method of vacuum bag was adopted for resin
impregnation and curing. A controlled amount of the epoxy
resin was directly applied on one surface of the CNTP. The
amount was just sufficient for a saturated resin impregna-
tion. The assembly for resin infiltration consisted of, from
the top to the bottom, a vacuum bag, a non-porous releasing
film, the wetted CNTP, a porous releasing film, a bleed
cloth, a non-porous releasing film, and a flat steel plate.
Vacuum was applied and a roller was used to roll and
evenly press the entire assembly. The resin on the CNTP
was forced to penetrate into the CNTP along the thickness
direction. The excess resin can be absorbed by the bleed
cloth under the CNTP. The resin was cured under the vac-
uum condition. After curing, the CNTP composite was
removed from the assembly. The weight and the thickness
of the CNTP composite were measured. Table 1 lists the
measured data for some specimens. To ensure resin
impregnation, the sample was cut by a knife and the cross-
section was examined by a Scanning Electron Microscope
(SEM). A typical SEM is shown in Fig. 2. The CNTs were
uniformly distributed through the thickness, and the resin
impregnation was generally satisfactory.
Tensile Test
The CNTP composite is thin and can be readily cut by
a knife to make specimens. The tensile specimen was an
end-tab type, as shown in Fig. 3a. The specimens were
5-mm wide and 60-mm long. The test section in the mid-
dle was 20 mm. A thin aluminum plate was used as the
end-tab material. The end-tabs were tightly bonded on the
specimen. A testing machine with a 1-kN load cell was
used. The loading speed was set at 0.5 mm/min. The
FIG. 1. Typical SEM image showing the surface of a CNTP before
resin impregnation.
TABLE 1. Material samples made in this study.
Sample
CNTP
weight (g)
Composite
weight (g)
Thickness
(mm)
CNT
(wt%)
C100 0.102 0.960 0.195 10.62
C150 0.156 1.368 0.262 11.40
C200 0.204 1.787 0.338 11.41
C250 0.252 2.085 0.391 12.08
C300 0.308 2.651 0.452 11.61
2 POLYMER COMPOSITES—2014 DOI 10.1002/pc
![Page 3: Processing and failure behavior of carbon-nanotube composites in sheet forms](https://reader037.fdocuments.us/reader037/viewer/2022100102/5750aa3d1a28abcf0cd66b35/html5/thumbnails/3.jpg)
displacement and the load were recorded at a rate of 10
data per second. At least five specimens were tested for
each type of the CNTP composite. After the test, the frac-
tured sections were examined by a SEM.
Tearing Test
The specimen for the tearing test was a double-cantilever
type, as shown in Fig. 3b. The specimen sizes were 60 mm
3 10 mm, and a 30-mm artificial precut was made to the
specimen by using a sharp knife. Figure 4b shows a speci-
men clamped by the grips. The clamped portion of each
cantilever beam was 10 mm, leaving a 20-mm portion
deflected by 90�. The composite was thin and flexible and
can be clamped without incurring noticeable damage. The
pulling load by the grips became the tearing force at the
crack tip. The loading speed was set at 3 mm/min. The dis-
placement and the load were recorded. After a distance of
grip moving, the crack started to propagate. The test was
terminated after the tearing crack extended 10 mm. After
the test, the fractured sections induced by the tearing were
observed.
RESULTS AND DISCUSSION
Geometry and Processing Characteristics
Figure 1 shows a CNTP surface and reveals the inter-
mingling of CNTs. The CNTPs are generally flat and the
thickness is uniform. Before dispersion, CNTs were mostly
in the form of ball or rope agglomerates. The microscopic
observations indicated that the agglomerates can no longer
exist in the CNTP. The agglomerates must be eliminated
or significantly reduced in size in order to make CNTPs
with a sufficient cohesive strength. The sonication power
and processing time are thus important for this purpose.
FIG. 2. A typical cross-section of the CNT composite after resin impreg-
nation and consolidation. CNT pullouts are popular on the section.
FIG. 3. Specimens specifications. (a) Tensile test. (b) Tearing test. (c)
Illustration of the tearing test.
FIG. 4. Testing specimens. (a) Tensile specimens after test. (b) The
tearing specimen clamped by the grips. [Color figure can be viewed in
the online issue, which is available at wileyonlinelibrary.com.]
DOI 10.1002/pc POLYMER COMPOSITES—2014 3
![Page 4: Processing and failure behavior of carbon-nanotube composites in sheet forms](https://reader037.fdocuments.us/reader037/viewer/2022100102/5750aa3d1a28abcf0cd66b35/html5/thumbnails/4.jpg)
Once the dispersed CNTs are filtered, the CNTs can be
entangled again and form the paper. If the CNTs are not
well entangled or the amount of CNTs is insufficient, the
CNTP can be easily broken during the processing.
Figure 2 shows a typical through-thickness section of a
CNTP composite after resin infiltration and curing. This
SEM sample was prepared by cutting a CNTP composite
by using a sharp knife. The section shows dense CNT
packing, and many CNT pullouts can be observed. Holes
left by the CNT pullouts can be seen too. Under the
vacuum-assisted process, the resin can effectively infiltrate
into the CNTP. However, the resin is still unable to reach
some small voids entrapped among CNTs. When vacuum
was applied, the resin tended to infiltrate into the CNTs
through the larger flowing paths. For a dense CNT pack-
ing, resin becomes difficult to flow, and small voids in
nanoscales can be formed. Because of the direct resin infil-
tration, the CNT contents of the present materials can be
much higher, typically between 10 and 12 wt%, much
higher than the value in typical bulk CNT composites [20].
Unlike bulk CNT composites in which the CNT/resin
ratio can be varied from zero up to its maximum limit,
the thickness of the CNTP reaches a minimum after the
vacuum is applied. Thus, the space for resin is a constant.
If the resin supply is excess and more than the CNTP can
absorb, resin layers can be accumulated on the CNTP
surfaces. In the present work, the resin amount was con-
trolled to maximize the resin impregnation without form-
ing resin layers on the surfaces. This condition is termed
as saturated impregnation. On the other hand, if the resin
supply is insufficient, more voids can be formed. For the
CNTP composites, the amount of resin is a processing
parameter and can be changed for different purposes. The
saturated impregnation can result in the best mechanical
properties. Other potential applications such as the gas
diffusion layers for fuel cells require the presence of
voids [25, 26]. A controlled wetting below the saturated
impregnation is necessary.
Tensile Behavior
Figure 5 shows typical loading curves of the speci-
mens. The curves become nonlinear at high strains. Typi-
cal fractured specimens are shown in Fig. 4a. The crack
cuts the specimen along the width direction. No necking
was observed, despite the non-brittle behavior and high
failure strains. The failure strain increases with the CNT
areal weight. The nonlinearity is in part because of the
pullout of CNTs that hinder crack growth. In comparison,
the neat epoxy is more brittle. The thinnest sample, C100,
is slightly weaker than the pure epoxy. The results of the
composite strength are shown in Fig. 6. The strength is
increased with the CNT areal-weight. The C300 is about
67% stronger than the C100. The major reason is that the
CNT packing is denser for specimens with a higher CNT
weight. The second reason is that the presence of voids
affects more the thinner specimens than the thicker ones.
Figure 7 shows typical fractured surfaces of the tensile
specimens. Unlike Fig. 2, the fractured sections are uneven
and stepped. The stepped surface was possibly because of
cracks located in different positions. When the tensile strain
increases, each crack extends along the normal direction
(the solid arrows in Fig. 7b). When two cracks of different
altitudes approach each other, the induced transverse shear
stress at the crack front can deflect the crack into the axial
direction (the open arrow in Fig. 7b). These two cracks can
be linked and a step is formed. Similar to Fig. 2, CNT pull-
outs are popular on the fractured surface. Extensive pullouts
deter crack growth and extend the failure strains.
Tearing Behavior
The tearing test is stroke-controlled, and the crack grows
in a stable manner. The loading response is illustrated in
Fig. 8. The loading curve is composed of two parts. The
first part is a gradually increased curve due to the deflec-
tion and elastic deformation of the cantilever beams. In
theory, no damage is induced in this part. The last point is
FIG. 5. Typical loading curves of the tensile test. [Color figure can be
viewed in the online issue, which is available at wileyonlinelibrary.
com.]
FIG. 6. Tensile strengths for the specimens with different CNT areal-
weights.
4 POLYMER COMPOSITES—2014 DOI 10.1002/pc
![Page 5: Processing and failure behavior of carbon-nanotube composites in sheet forms](https://reader037.fdocuments.us/reader037/viewer/2022100102/5750aa3d1a28abcf0cd66b35/html5/thumbnails/5.jpg)
the peak of the curve, at which the tearing crack starts to
grow from the tip of the precut. This point refers to the
crack triggering, and the corresponding load is termed trig-
gering load. The second part of the curve is due to crack
growth. Once the tearing is triggered, the crack will grow
with the displacement of the grip. While the grip moves at
a constant speed, the crack grows in a stop-and-go manner,
and the load changes up and down. This process continues
until the crack cuts the entire specimen.
Figure 9 shows some loading curves of the tests. The
slope and the peak increase with the composite thickness.
Nearly all specimens experienced minor drops in the load
before the peak point. The presence of defects and voids
in the material can trigger minor and local damage before
the crack starts to grow. As expected, the load in the sec-
ond part of the curve goes up-and-down within a range.
For most specimens, the crack grows straight along the
center-line of the specimen (the dotted line in Fig. 3c).
However, in some specimens, the crack grew away from
the center-line. The fractured section became inclined to
FIG. 7. Typical fractured sections of the tensile test. (a) The stepped sur-
face. (b) A step on the surface. Two horizontal cracks approach each other
(open arrows). One of the crack is deflected to the vertical direction due to
the transverse shear (open arrow), forming the step. [Color figure can be
viewed in the online issue, which is available at wileyonlinelibrary.com.]
FIG. 8. Illustration of the loading response of the tearing test. The
curve comprises two parts separated by the dotted line. The first part
involves deflection and elastic deformation of the cantilever beams. The
second is the stage of crack growth, and the curve goes up-and-down
with the displacement.
FIG. 9. Typical loading curves of the tearing tests. [Color figure can
be viewed in the online issue, which is available at wileyonlinelibrary.
com.]
FIG. 10. Normalized results of the triggering load for the specimens.
DOI 10.1002/pc POLYMER COMPOSITES—2014 5
![Page 6: Processing and failure behavior of carbon-nanotube composites in sheet forms](https://reader037.fdocuments.us/reader037/viewer/2022100102/5750aa3d1a28abcf0cd66b35/html5/thumbnails/6.jpg)
FIG. 11. Fractured section at the tip of the precut (C250). The dotted line indicates the tip. The crack
started from the dotted line and grew to the right.
FIG. 12. Typical fractured sections of the tearing specimens, showing a divergent pattern of streamlines.
(a) C200. (b) C300. (c) Schematic illustration of crack growth. The crack grows intermittently. The stream-
line is where the crack stops. The numbers indicate the sequence of the crack growth. [Color figure can be
viewed in the online issue, which is available at wileyonlinelibrary.com.]
![Page 7: Processing and failure behavior of carbon-nanotube composites in sheet forms](https://reader037.fdocuments.us/reader037/viewer/2022100102/5750aa3d1a28abcf0cd66b35/html5/thumbnails/7.jpg)
the loading direction, and the crack was no longer a pure
Mode-III. The test was terminated if this happened.
Figure 10 lists the results of the normalized triggering
load, defined as the load divided by the specimen thickness.
The normalized load is increased with the composite thick-
ness. The reason is that the stress concentration at the tip is
less intense in thicker specimens. The growing of the crack is
sensitive to the stress at the crack tip, and the stress concentra-
tion is disproportionately higher in thinner specimens. On the
fractured section, the CNTs were pulled out, rather than cut-
to-break by the tearing stress. The CNT pullouts represent
energy consumption, which hinders the growth of the crack.
Figure 11 shows a fractured section around the precut tip,
revealing the morphological differences between the precut
and the tearing-induced fracture surfaces. The original loca-
tion of the precut tip is indicated. The left region is the precut
surface opened by the cutting knife. The crack started from
the precut tip and grew toward the right. The fractured sur-
face around the tip is generally irregular because of the
geometry of the precut tip. However, once the crack grows
stably in the tearing mode, the fractured surface becomes reg-
ular. Figure 12 shows two typical tearing-induced surfaces.
The most notable feature is the streamlines on the surface.
Figure 12c illustrates how the streamlines were formed. The
crack grows in a stop-and-go manner, depending on the stress
level and the strain energy. The streamline is where the crack
stops. Once the stored strain energy reaches a critical level,
the crack will quickly propagate to the next streamline and
stop. The numbers indicate the sequence of the crack move-
ment. The upper half and the lower half are symmetric in the
pattern. According to the SEM observations, most CNTs on
the fractured surface were pulled out, and very few were frac-
tured because of shear-cut or stretching. A typical SEM
micrograph is shown in Fig. 13. CNT pullouts and holes left
by the pullouts are abundant on the surface.
CONCLUSIONS
This work examines the processing and fracture behav-
ior of the CNT composite. The two-step approach allows
CNT dispersion and resin infiltration to be conducted sepa-
rately, and the high-viscosity problem in the CNT/resin
mixture can be avoided. The nominal CNT content can
reach 10 wt%, which is much higher than that of traditional
bulk-type CNT composites. The resulting nanocomposite is
paper-like and flexible. Both tensile and tearing properties
have been examined. The tensile strength increases with
the CNT weights, and the C300 specimens are about 38%
stronger than the neat resin. In comparison with the linear
response and brittle failure in the pure epoxy specimens,
the CNTP composites show nonlinear responses at higher
strains, and the fractured surfaces are unsmooth. Pullout of
CNTs is responsible for the nonlinear behavior. The tearing
response is composed of two parts. The first involves
deflection and elastic deformation of the double cantilever
beams, although minor damage could also appear. The sec-
ond part is due to the crack growth. The crack grows in a
stop-and-go manner, leaving symmetric streamlines on the
fractured surfaces. The normalized triggering load is
increased with the CNTP thickness. CNT pullout is the
dominant mode in the fracture section.
REFERENCES
1. M. Endo, H. Muramatsu, T. Hayashi, Y.A. Kim, M.
Terrones, and M.S. Dresselhaus, Nature, 433, 476 (2005).
2. P.G. Whitten, G.M. Spinks, and G.G. Wallace, Carbon, 43,
1891 (2005).
3. E. Lahiff, R. Leahy, J.N. Coleman, and W.J. Blau, Carbon,
44, 1525 (2006).
4. L. Berhan, Y.B. Yi, A.M. Sastry, E. Munoz, M. Selvidge,
and R. Baughman, J. Appl. Phys., 95, 4335 (2004).
5. M. Zhang, S. Fang, A.A. Zakhidov, S.B. Lee, A.E. Aliev,
C.D. Williams, K.R. Atkinson, and R.H. Baughman, Sci-ence, 309, 1215 (2005).
6. Q. Liu, M. Li, Z. Wang, Y. Gu, Y. Li, and Z. Zhang, Com-pos. A, 55, 102 (2013).
7. P.E. Lopes, F. van Hattum, C.M.C. Pereira, P.J.R.O N�ovoa.,
S. Forero, F. Hepp, and L. Pambaguian, Compos. Struct.,92, 1291 (2010).
8. M. Chapartegui, J. Barcena, X. Irastorza, C. Elizetxea, M.
Fernandez, and A. Santamaria, Compos. Sci. Technol., 72,
489 (2012).
9. B. Ashrafi, J. Guan, V. Mirjalili, P. Hubert, B. Simard, and
A. Johnston, Compos. A, 41, 1184 (2010).
10. J.N. Coleman, W.J. Blau, A.B. Dalton, E. Mu~noz, S.
Collins, B.G. Kim, J. Razal, M. Selvidge, G. Vieiro, and
R.H. Baughman, Appl. Phys. Lett., 82, 1682 (2003).
11. E.M. Doherty, S. De, P.E. Lyons, A. Shmeliov, P.N.
Nirmalraj, V. Scardaci, J. Joimel, W.J. Blau, J.J. Boland,
and J.N. Coleman, Carbon, 47, 2466 (2009).
12. P. Gonnet, Z. Liang, E.S. Choi, R.S. Kadambala, C. Zhang,
J.S. Brooks, B. Wang, and L. Kramer, Curr. Appl. Phys., 6,
119 (2006).
13. H. Kim, Y. Miura, and C.W. Macosko, Chem. Mater., 22,
3441 (2010).
14. Q. Chen, J. Bao, J.G. Park, Z. Liang, C. Zhang, and B.
Wang, Adv. Funct. Mater., 19, 3219 (2009).
FIG. 13. Fractured section due to tearing, showing CNT pullouts and
pullout holes left on the surface.
DOI 10.1002/pc POLYMER COMPOSITES—2014 7
![Page 8: Processing and failure behavior of carbon-nanotube composites in sheet forms](https://reader037.fdocuments.us/reader037/viewer/2022100102/5750aa3d1a28abcf0cd66b35/html5/thumbnails/8.jpg)
15. G.T. Pham, Y.B. Park, S. Wang, Z. Liang, B. Wang, C.
Zhang, P. Funchess, and L. Kramer, Nanotechnology, 19,
325705 (2008).
16. E. Kymakis and G.A.J. Amaratunga, J. Appl. Phys., 99,
84302 (2006).
17. T.W. Chou, L. Gao, E.T. Thostenson, Z. Zhang, and J.H.
Byun, Compos. Sci. Technol., 70, 1 (2010).
18. T. Kuilla, S. Bhadra, D. Yao, N.H. Kim, S. Bose, and J.H.
Lee, Prog. Polym. Sci., 35, 1350 (2010).
19. L.S. Schadler, L.C. Brinson, and W.G. Sawyer, JOM, 59,
53 (2007).
20. T.L. Wu, T.S. Lo, and W.S. Kuo, Polym. Compos., 31, 292,
(2010).
21. R.B. Yang, W.S. Kuo, H.C. Lai, J. Appl. Polym. Sci., 131,
40963 (2014).
22. H. Qian, E.S. Greenhalgh, M.S.P. Shaffer, A. Bismarck, J.Mater. Chem., 20, 4751 (2010).
23. X.M. Liu, Z.D. Huang, S.W. Oh, B. Zhang, P.C. Ma, M.M.F.
Yuen, and J.K. Kim, Compos. Sci. Technol., 72, 121 (2012).
24. B. Brown, B. Swain, J. Hiltwine, D.B. Brooks, and Z.
Zhou, J. Power Sources, 272, 979 (2014).
25. C.H. Liu, T.H. Ko, W.S. Kuo, H.K. Chou, H.W. Chang,
and Y.K. Liao, J. Power Sources, 186, 450 (2009).
26. C.J. Hung, C.H. Liu, T.H. Ko, W.H. Chen, S.H. Cheng,
W.S. Chen, A. Yu, and A.M. Kannan, J. Power Sources,
221, 134 (2013).
8 POLYMER COMPOSITES—2014 DOI 10.1002/pc