Composites: Part A 43 (2012) 1007–1015

Contents lists available at SciVerse ScienceDirect

Composites: Part A
journal homepage: www.elsevier.com/locate/compositesa

Poly(phenylene sulphide) and poly(ether ether ketone) composites reinforced

with single-walled carbon nanotube buckypaper: II – Mechanical properties,
electrical and thermal conductivity
Ana M. Díez-Pascual a,⇑, Jingwen Guan b, Benoit Simard b, Marián A. Gómez-Fatou a
a
Institute of Polymer Science and Technology (ICTP-CSIC), Juan de la Cierva 3, 28006 Madrid, Spain
b
Steacie Institute for Molecular Sciences, National Research Council Canada, 100 Sussex Drive, Ottawa, Canada K1A 0R6

a r t i c l e i n f o a b s t r a c t

Article history: The mechanical properties, electrical and thermal conductivity of single-walled carbon nanotube
Received 19 May 2011 (SWCNT) buckypaper (BP) embedded in poly(ether ether ketone) (PEEK) or poly(phenylene sulphide)
Received in revised form 4 August 2011 (PPS) matrices were investigated. Dynamic mechanical analysis demonstrated a significant increase in
Accepted 5 November 2011
the storage modulus and glass transition temperature of the polymers, indicating strong SWCNT–matrix
Available online 13 November 2011
interfacial adhesion. The composites showed improved stiffness and strength, as revealed by tensile and
flexural tests, while their ductility and toughness moderately decreased. Exceptional enhancements in
Keywords:
the electrical and thermal conductivity of PPS and PEEK were found. Their Young’s moduli and thermal
A. Polymer–matrix composites (PMCs)
Carbon nanotube buckypaper
conductivities were compared with the predictions of theoretical models. This investigation indicates
B. Electrical properties that SWCNT-BPs possess great potential to improve the performance of thermoplastics and satisfy a wide
D. Mechanical testing variety of demands in multi-disciplinary technological applications.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction and alignment as well as nanofiller–matrix interfacial adhesion. In

a recent study, Jianwen et al. [8] measured a Young’s modulus of
Buckypapers (BPs), free-standing porous sheets of entangled 30 GPa for epoxy-impregnated buckypaper composites, which
carbon nanotubes (CNTs), can be used to fabricate polymer com- was attributed to an improved quality of impregnation as well as
posites with uniform tube dispersion, controlled nanostructure very high CNT content (45 wt%). Wang et al. [9] reported storage
and high CNT loading (up to 60 wt%) [1,2]. These composite mate- moduli of 45 and 15 GPa for magnetically aligned and non-aligned
rials are expected to have excellent mechanical, thermal and elec- epoxy/BP composites, respectively, with CNT concentrations of
trical properties, characteristics that make them interesting for around 40 wt%. In these works, the improvements in properties ob-
potential applications in the electronics, aeronautic and aerospace tained are close to those calculated through a simple rule of
industries. Several researches have recently investigated the elastic mixtures.
properties of pristine BPs [3–5] and their composites [6–8]. Exper- Polymer/BP composites are also expected to possess excellent
imental Young’s modulus of BPs generally lie between 0.3 and thermal conductivity, which can provide the solution of thermal
2.3 GPa, depending on the type, synthesis method, purification management for advanced electronic devices, heat sinks, connec-
treatment, degree of porosity and characteristics of the CNT bun- tors or printed circuit boards. Gonnet et al. [10] found thermal con-
dles (diameter, length, modulus). For the manufacturing of BP- ductivity values of approximately 4 and 7 W/mK for randomly
reinforced epoxy composites, impregnation processes with liquid oriented and magnetically aligned BP mats impregnated with
resins are preferred, using techniques as vacuum-assisted resin epoxy resin, respectively, which represent increments of more
transfer moulding (VARTM). Melt-compression is the most suitable than one order of magnitude as compared with the conductivity
option for thermoplastics, since the polymer can be heated to reach of the neat thermoset (0.2 W/mK). In addition, BPs can be incorpo-
lower viscosity and allow the infiltration within the CNT network. rated to improve the electrical properties of polymers. Pham et al.
The wide variation in the reported storage and Young’s modulus of [7] investigated the electrical behaviour of polycarbonate/BP com-
BP-reinforced composites can be ascribed to several factors, posites with CNT contents ranging between 46 and 48 wt%. The re-
including degree of impregnation, void content, CNT concentration ported conductivities were around 102 S/cm, about 14 orders of
magnitude higher than that of the pure matrix. The conductive
polymer/BP composites can be used for a wide range of applica-
⇑ Corresponding author. Tel.: +34 915 622900; fax: +34 915 644 853. tions such as electrostatic discharging in electronic components,
E-mail address: [email protected] (A.M. Díez-Pascual). lightweight coating materials for electromagnetic interference

1359-835X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compositesa.2011.11.003
1008 A.M. Díez-Pascual et al. / Composites: Part A 43 (2012) 1007–1015

shielding, lightning strike protection of various structures, includ- equipped with a platinum thermocouple. The electrical contact
ing aircraft, automobile and turbine blades. was made using a silver paste with a ring guard electrode configu-
In our previous work [11], SWCNT-BP composites based on two ration to minimise leakage currents. The voltage applied to the
different high-performance thermoplastic polymers, poly(pheny- samples was cycled five times and was kept low to avoid melting
lene sulphide) (PPS) and poly(ether ether ketone) (PEEK), were or degrading the resin due to Joule heating. The measured cur-
manufactured through hot-press processing. Their thermal stabil- rent–voltage curves were linear (R2 value >0.9), and the resistance
ity, crystallization behaviour and crystalline structure were inves- was obtained as the plot slope. The conductivity r was calculated
tigated; the most interesting finding was the exceptional increase based on r = t/Rv  A, where A, t and Rv are the area, thickness and
in the degradation temperatures of the polymers. The main aim volume resistance of the specimen, respectively. At least three
of this study is to characterise their mechanical behaviour as well specimens of each type of laminate were tested to get an average
as their thermal and electrical conductivity. Different theoretical value.
models have been applied to describe the experimental results,
and their feasibility is discussed.
2.2.4. Thermal conductivity measurements
The thermal diffusivity (a) of the samples, with typical dimen-
2. Experimental sions of 10 mm  10 mm, was measured by laser flash radiometry
technique [12]. A pulsed Nd:YAG laser with wavelength of 1.06 lm
2.1. Materials and composite manufacturing was used to heat the front surface of the sample. The thermal radi-
ation from the rear surface was focused by germanium lens onto a
The materials used, PPS (Fortron 0205B4), PEEK (Victrex 150P) liquid nitrogen cooled Mercury, Cadmium, Tellurium (MCT) infra-
and laser-grown SWCNT-BPs, were the same as in the preceding red detector with response time of 40 ns and cutoff wavelength
paper [11]. The BPs possess an average thickness of 70 lm, density of 12.5 lm. The signals received by the detector were amplified
of 0.60 g/cm3, an average degree of porosity of 54% and 7.2 wt% and collected with a 500 M bandwidth TDS3052B oscillograph.
metal catalyst residues. Measurements were carried out by averaging over 20–100 laser
The buckypaper-reinforced composite laminates were manu- shots. The averaged signals were recorded as a function of time
factured as a sandwich-like assembly of a BP sheet between two and adjusted to fit the Clark–Taylor model. The measurement error
polymer layers via hot-press processing, as described in our previ- of the system was below ±5% after calibration. The thermal con-
ous work [11]. The weight fraction of polymer matrix was about ductivity (K) was calculated according to the expression:
76% and 80% for PEEK and PPS based composites, and their average K = q  Cp  a, where q and Cp are the density and specific heat
void content was calculated to be 3.9% and 2.6%, respectively. The capacity of the samples, respectively. At least three readings for
laminates were cut into test specimens with a cutter knife to be each specimen were taken to ensure reproducibility.
used for the different characterisations.

3. Results and discussion

2.2. Characterisation techniques
3.1. Dynamic mechanical analysis
2.2.1. Scanning electron microscopy
The surface of tensile fractured specimens was analysed with a
The viscoelastic properties (storage modulus, loss modulus and
Philips XL30 scanning electron microscope (SEM) applying an
damping ratio) were measured by DMA, which provides informa-
acceleration voltage of 25 kV and an intensity of 9  109 A. The
tion about the stiffness of the material and the strength of the fil-
samples were coated with a 5 nm Au/Pd overlayer to avoid charg-
ler–matrix interactions. Fig. 1a shows the temperature dependence
ing during electron irradiation.
of the storage modulus (E0 ) for pristine BP, the neat polymers and
the composites. E0 values at 25, 100 and 200 °C are collected in Ta-
2.2.2. Mechanical tests ble 1. At 25 °C, the moduli of neat PPS and PEEK are 2.2 and 3.7 GPa,
The dynamic mechanical performance of the samples was respectively; upon BP impregnation, they increase by 38% and
investigated using a Mettler DMA 861 dynamic mechanical ana- 32%, suggesting the stiffening effect of the SWCNT mat. The slightly
lyser. The measurements were performed on rectangular shaped higher increment attained for PPS based laminate could be related
samples (19.5 mm  4 mm) in the tensile mode at frequencies to the improved impregnation degree of this sample, since PPS has
of 0.1, 1 and 10 Hz, in the temperature range between 100 and lower viscosity than PEEK, hence is able to flow more easily inside
250 °C, at a heating rate of 2 °C/min. A dynamic force of 6 N was the BP pores. Analogous behaviour of E0 enhancement was reported
applied oscillating at fixed frequency and amplitude of 30 lm. for BP-reinforced epoxy composites [9,13]. As the temperature
Tensile and flexural properties were measured with an IN- rises, E0 of all the polymeric samples decreases, showing a strong
STRON 4204 mechanical tester at 23 ± 1 °C and 50 ± 5% relative drop in the vicinity of the glass transition, where the polymer
humidity. Tensile tests were carried out at a constant rate of chains become more mobile and lose their close packing arrange-
1 mm/min using a load cell of 1 kN. The flexural modulus and ment. In the rubbery region, the difference in modulus between
strength of rectangular specimens (10 mm  5 mm) in three- the composites and the neat polymers is amplified; at 200 °C,
point bending mode were measured at a crosshead speed of PPS/SWCNT-BP shows 90% higher E0 than the pure matrix (Ta-
0.5 mm/min. All the samples were conditioned for 24 h before ble 1). This indicates that the stiffening effect is more pronounced
the measurements. A minimum of five specimens of each sample above the softening point of the resins, which could be explained
were tested for statistical evaluation. considering that the BP modulus changes only slightly with tem-
perature, as reported by previous works [14,15]. These results are
2.2.3. Electrical conductivity measurements consistent with DMA studies on PEEK/glass fibre/SWCNT laminates
DC volume resistivity measurements as a function of tempera- [16], which also showed higher relative increase in the matrix stiff-
ture were performed using a Keithley 6221 current source and a ness above than below the Tg upon addition of the SWCNTs.
2182A nanovoltmeter to measure current bias and voltage reading. The evolution of the damping ratio (tan d) as a function of tem-
The sample (10 mm  10 mm) was placed in a temperature con- perature is shown in Fig. 1b. The polymers and their composites
trolled chamber, and the temperature was checked with a sensor exhibit an intense peak (a relaxation), which corresponds to the
 A.M. Díez-Pascual et al. / Composites: Part A 43 (2012) 1007–1015 1009

effective immobilization ability caused by the dense SWCNT mat.

(a)
6 SWCNT-BP On the other hand, PEEK based samples exhibit a small sub-Tg
PPS
PPS/SWCNT-BP
relaxation around 80 °C, associated with local motions of the ke-
PEEK tone groups; this transition temperature is higher in the composite
5 PEEK/SWCNT-BP
as compared to the neat polymer, since the presence of the BP also
hinders the mobility of this functional group.
4 It can also be observed from Fig. 1b that the height of tan d peak
E' (GPa)

for the BP-reinforced composites is considerably lower than that of

3 the neat polymers, indicating higher ratio of the stored-to-loss en-
ergy in the zone of the Tg, hence lower energy loss in the laminates.
This decrease in tan dmax value also points out the reduction in
2
mobility caused by the hindrances of the SWCNT network; the
diminution in height is more pronounced for PEEK/SWCNT-BP lam-
1 inate (34%) as compared to PPS/SWCNT-BP (27%), in agreement
with the trend observed from Tg data. Lopes et al. [13], dealing with
0 epoxy/BP composites, also found a diminution in tan dmax with the
- 100 -50 0 50 100 150 200 250 addition of the CNT entangled mat, ascribed to the fact that the
T (ºC) presence of CNTs enhances hindrance to the physical mobility of
epoxy groups. On the other hand, the composites display wider
(b) 0.18 tan d peak than the neat polymers, which may be due to the con-
SWCNT-BP
0.16 PPS finement of polymer segments within the dense nanotube net-
PPS/SWCNT-BP work; the SWCNTs perturb the relaxation of adjacent polymer
PEEK
0.14 PEEK/SWCNT-BP chains, leading to a longer relaxation time that results in a broad-
ening of the peak. The data for width at half maximum value (b) are
0.12
collected in Table 1. b of pure PPS and PEEK are about 26 and 21 °C,
0.10 and increase remarkably (5 and 9 °C, respectively) with the incor-
poration of the BP. The larger broadening observed for PEEK based
tan δ

0.08 laminate could also be interpreted as enhanced filler–matrix inter-

actions. The aforementioned effect (albeit less intense) has been
0.06
previously reported for other PEEK [17–19] and PPS [20] compos-
0.04 ites, attributed to a more inhomogeneous amorphous phase in
the composites in relation to the pure matrix. Overall, DMA results
0.02 indicate that the SWCNT-BP integrated composites exhibit higher
stiffness and are thermomechanically more stable than the neat
0.00
-100 -50 0 50 100 150 200 250
polymers.
T (ºC)
3.2. Tensile behaviour
Fig. 1. Evolution of the storage modulus E0 (a) and tan d (b) as a function of
temperature, at the frequency of 1 Hz, for pristine SWCNT-BP, PPS, PEEK and their Typical stress–strain curves at 23 °C of the laminates are de-
laminates. (For interpretation of the references to colour in this figure legend, the
picted in Fig. 2, along with those of pristine BP and the neat poly-
reader is referred to the web version of this article.)
mers for comparison. The average Young’s modulus (E), tensile
strength (ry), elongation at break (eb) and toughness (G) obtained
Tg. The entangled SWCNT network of the BP efficiently restricts the from the tensile tests are displayed in Fig. 3. Pristine SWCNT-BP
mobility of the polymer chains, which is reflected in higher Tg val- shows low E, ry and eb, owed to the low energy required to de-
ues (Table 1); these results are in very good agreement with those bridge the CNTs during the tensile deformation [21] as a conse-
obtained from DSC thermograms [11]. Thus, Tg of pure PPS and quence of the weak van der Waal force assembly of the inter-tubes.
PEEK are about 91 and 148 °C, respectively, and increase by 26 The average BP modulus is 1.1 GPa, consistent with the reported
and 31 °C in the corresponding laminates. The higher Tg value ob- data in the range 0.3–2.3 GPa for as-made SWCNT-BPs [3–5]. E of
served for PEEK composite could be ascribed to its stronger filler– neat PPS and PEEK are 2.5 and 3.8 GPa, and increase by 35%
matrix interactions, as revealed by the Raman spectra [11], since and 29%, respectively, upon BP impregnation, in good agreement
this transition temperature is very sensitive to interfacial interac- with the results from DMA tests. These significant increments sug-
tions between the polymer and the reinforcements; it could also gest effective stress transfer ability from the matrices to the
be related with the higher SWCNT content of this laminate. It SWCNT network, indicating good interfacial adhesion between
should be noticed that the increments in Tg attained in this work the two composite phases. Qualitatively similar behaviour has
are considerably larger than those observed for bulk PEEK/SWCNT been reported for other BP/polymer composites [6–8]. As men-
[17] or PPS/SWCNT [18] composites, pointing out the more tioned previously, the extent of modulus improvement is strongly

Table 1
Storage modulus E0 at 25, 100 and 200 °C, glass transition temperature Tg, tan d maximum value and width at half maximum b for the composites and the neat polymers, obtained
from DMA measurements at a frequency of 1 Hz.

Sample E025 C (GPa) E0100 C (GPa) E0200 C (GPa) Tg (°C) Tan dmax (a.u.) b (°C)

PPS 2.22 1.08 0.21 90.8 0.118 26.5

PPS/SWCNT-BP 3.07 2.63 0.40 116.9 0.086 31.7
PEEK 3.68 3.20 0.32 147.7 0.170 21.3
PEEK/SWCNT-BP 4.85 4.09 0.75 178.3 0.112 30.4
1010 A.M. Díez-Pascual et al. / Composites: Part A 43 (2012) 1007–1015

relative increment in modulus as compared to the neat polymer;

120 this could be explained considering the slightly higher void content
of this laminate, since the degree of porosity is known to play an
100 important role on the mechanical performance of reinforced poly-
mer composites [22]. Regarding the tensile strength, the incre-
ments as compared to the neat matrices (23% and 13% for PPS
80
and PEEK laminates, respectively) are considerably lower than
σ (MPa)

those attained for the modulus, which could be attributed to the

60 fact that ry has been reported to be more influenced by the pres-
ence of internal pores [22]. Previous studies on the mechanical per-
40 formance of SWCNT-reinforced PEEK [23] and PPS [18] composites
SWCNT-BP also revealed larger improvements in the stiffness than in the
PPS strength of the polymer.
20 PPS/SWCNT-BP
PEEK Taking into account that the laminates were manufactured as a
PEEK/SWCNT-BP
sandwich-like structure, and essentially consist of three layers of
0
similar thickness (the polymer top layer, the real composite middle
0 2 4 6 8
layer and the polymer bottom layer), it is possible to estimate the
ε (%)
modulus of the real polymer/SWCNT composite through the classi-
Fig. 2. Representative stress–strain curves obtained from tensile tests of pristine cal rule of mixtures assuming perfect adhesion between the layers
SWCNT-BP, the neat polymers and the composite laminates. (For interpretation of as follows [24]:
the references to colour in this figure legend, the reader is referred to the web
version of this article.) Ec ¼ Ep  V p þ V r  Er ð1Þ

where Ec, Ep and Er are the tensile modulus of the whole sandwich-
influenced by the impregnation degree, void content, CNT concen- like composite, the neat polymer and the real polymer/SWCNT
tration as well as nanofiller–matrix interfacial interactions. Despite composite, respectively, and Vp is the volume fraction of the neat
PEEK/SWCNT-BP exhibits stronger nanotube–matrix interactions polymer that can be calculated as:
than PPS laminate, as evidenced by the Raman spectra [11] and
DMA tests, and possesses higher SWCNT content, it shows lower V p ¼ T p =½T r þ T p   100 ð2Þ

5.0
120
4.5
Young's Modulus, E (GPa)

Tensile strength, σ y (MPa)

4.0 100

3.5
80
3.0
2.5 60
2.0
1.5 40

1.0
20
0.5
0.0 0
SWCNT-BP PPS/SWCNT-BP PPS PEEK/SWCNT-BP PEEK SWCNT-BP PPS/SWCNT-BP PPS PEEK/SWCNT-BP PEEK

8
8
7
Elongation at break, εb (%)

Toughness, G (MJ/m3)

6
6
5

4
4
3

2 2

0 0
SWCNT-BP PPS/SWCNT-BP PPS PEEK/SWCNT-BP PEEK SWCNT-BP PPS/SWCNT-BP PPS PEEK/SWCNT-BP PEEK

Fig. 3. Young’s modulus (E), tensile strength (ry), elongation at break (eb) and toughness (G), at 23 °C, for the manufactured composites. For comparison, data of pristine BP
and the neat polymers are also included. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
 A.M. Díez-Pascual et al. / Composites: Part A 43 (2012) 1007–1015 1011

where Tp and Tr are the thickness of the polymer and real composite
layer, respectively; considering that Tr is approximately equal to the
thickness of the buckypaper (70 lm), Tp was determined to be
75 lm. Using these values, the estimated E for PPS and PEEK com-
posites applying Eq. (1) were 7.24 and 5.23 GPa, respectively, about
twice the modulus of the pure matrix. The improvement in E at-
tained is considerably lower than the expected based on the modu-
lus reported for SWCNT bundles (300 GPa) [25]. The gap between
E of the real composites and the theoretical values probably arises
from several factors such as incomplete impregnation of the BP.
Due to the high viscosity of the matrices, the polymer segments
cannot flow easily inside the BP pores even at the high processing
temperatures applied, resulting in small defects (mainly internal
voids), which have detrimental effects on the mechanical
performance. Moreover, the SWCNTs are gathered in bundles, and
shear slippage of individual nanotubes within the bundle may
occur, thereby limiting stress transfer. Furthermore, the presence
of amorphous carbon and metal catalyst residues within the
SWCNT-BP may have caused the reduction of their effective Young’s
modulus. In addition, the models assume perfect filler–matrix
bonding, whilst the real interfacial adhesion might not be strong
enough to allow the nanofillers to elaborate their full capability
for stiffness enhancement. Future work would focus on improving
the SWCNT–matrix interfacial adhesion via functionalization of
the BP and the polymer, since the covalent bonding at the nano-
tube/polymer interface is known to enhance the load transfer be-
tween the two phases [26], being very effective in stiffening the
material.
The elongation at break of neat PPS and PEEK (5.9% and 8.3%,
respectively) fell by 28% and 33% in the corresponding compos-
ites; the larger decrease observed for PEEK based sample should
be attributed to its higher SWCNT content. This drop in ductility
is a common phenomenon in CNT-reinforced composites, since
the nanofillers restrict the plastic deformation of the matrix, and
has been reported for other BP/polymer composites [7]. To obtain
additional information about the failure in tensile tests, the sur-
faces of fractured specimens were analysed by SEM (Fig. 4). As
can be observed, the laminates and the neat polymers exhibit con-
siderably different fractographic features. The surface of the pure
matrices (Fig. 4a) shows a rough morphology, indicating that the
tensile failure was consequence of a ductile deformation; no voids
or cracks were observed in the examined zones. In contrast, the
composites (Fig. 4b and c) display smoother morphology, suggest-
ing a more brittle failure mode. There are three possibilities for the
failure mechanism in polymer/CNT composites [27]: pullout,
immediate fracture and sliding-fracture. Generally, the failure Fig. 4. SEM micrographs at different magnifications from tensile fractured surfaces
behaviour is very complex, and consists in a combination of the of neat PPS (a), PSS/SWCNT-BP (b) and PEEK/SWCNT-BP (c). The solid arrows in (c)
three aforementioned modes. In the micrographs of the compos- indicate SWCNT segments and the circles point out small holes formed due to the
pullout of SWCNT bundles.
ites, small holes and segments of SWCNT bundles can be observed
at the fractured surface (see the arrows and circles marked on
Fig. 4c). The size of the small holes is comparable to that of the The toughness of the composites (Fig. 3), measured as the area
bundles, which should be formed due to the pullout of the under the tensile curve, is lower than that of the pure polymers,
SWCNTs. This suggests that the pullout is the main failure mode being the decrements 13% and 25% for PPS and PEEK laminates,
in PPS and PEEK composites. Additionally, the diameter of the respectively. This diminishing effect of the fillers on the toughness
SWCNTs protruding out of the matrix was on average 52 nm, mod- is due to the significant decrease in the elongation at break; more-
erately higher than the observed from AFM and SEM images of over, stress concentrations could be formed around the filler ends,
pristine SWCNT-BP (40 nm [11]). This confirms that the SWCNTs areas of poorer adhesion with the matrix, leading to a more brittle
pullout of the matrix on fracture, taking simultaneously entangled fracture, as revealed by SEM images. The diminution in toughness
polymer chains. This phenomenon tends to occur in systems where is more pronounced for PEEK/SWCNT-BP laminate, consistent with
the polymer is tightly bonded to the nanofillers [26,28], thereby its lower extent of plastic deformation and its higher void content;
confirming the strong polymer–SWCNT interfacial adhesion, as re- the presence of small voids could nucleate secondary cracks under
vealed from TEM images and Raman spectra [11]. On the other stress, aggravating the brittleness of the material. The toughness
hand, small dimples were formed during the deformation process and dynamic mechanical properties can be correlated in terms of
of the composites (see Fig. 4c), which can act as stress concentra- the area under tan d peaks, since it represents the energy dissi-
tion sites under load, nucleating cracks and leading to premature pated in viscoelastic relaxations [29]; any molecular process that
failure. promotes energy dissipation enhances the impact strength of the
1012 A.M. Díez-Pascual et al. / Composites: Part A 43 (2012) 1007–1015

polymer [30]. The toughness of the neat polymers and the compos- [32]. r rises with increasing temperature, indicating typical semi-
ites shows similar trend to that found for the area under DMA conducting behaviour; this is in agreement with the profile of the
peaks; the results provided by both techniques indicate that the G-band in the Raman spectra [11], characteristic of semiconduct-
dense CNT network decreases the ability of the matrix to absorb ing SWCNTs. Nevertheless, the change in r is relatively small
energy during the deformation process, causing a detriment in (14% within the range 273–373 K), and the conductivity exhibits
the impact resistance of the polymer. a non-linear tendency. Similar temperature dependence was
widely observed in entangled CNT mats [34]. The BP conductivity
is directly proportional to the quantity and quality of the metallic
3.3. Flexural tests
nanotubes present, since the network consists of a random mixture
of semiconducting and metallic tubes. Two models have been pro-
The results of flexural tests are displayed in Fig. 5. The flexural
posed to explain the temperature dependence of r in BP sheets, the
moduli (EF) of pure PPS (3.2 GPa) and PEEK (4.3 GPa) experience
variable-range hopping (VRH) [35] and the heterogeneous mecha-
about 24% and 21% enhancement, respectively, upon BP impregna-
nism [36]. Taking into account the semiconducting behaviour ob-
tion. As mentioned previously, the higher increment observed for
served, the 3D VRH model seems to be the predominant
PPS/SWCNT-BP composite should be owed to its improved impreg-
nation quality and lower void content. On the other hand, the flex-
ural strength (rFM) of the composites is only moderately higher
than that of the neat polymers (11% and 9% for PPS and PEEK lam-
inates, respectively). Interestingly, the increments in EF observed
54 (a) SWCNT-BP
here are similar to those reported for PPS/graphite composites with
10 wt% loading [31]; however, those composites exhibit lower rFM 52
than pure PPS, presumably due to weak nanofiller–matrix interac-

σ (S/cm)
tions and stress concentrations resulting from graphite agglomer-
ates within the polymer. In contrast, the SWCNT-BP and the 50 4.00

matrix chains constitute a nanoscale interpenetrating system, with 3.96

uniform filler dispersion and strong interfacial adhesion between

ln (σ )
48
3.92
the two components, which lead to simultaneous increases in EF
and rFM. Note that the improvements attained in the flexural prop- 3.88 σ = σ 0 exp [-(T 0 /T) ]
1/4

T0 2420 ± 9.3 K
erties are lower than those achieved in the Young’s modulus and 46 σ O 279.78 ± 6.8 S /cm
3.84
tensile strength. This discrepancy in the reinforcing efficiency 0.225 0.230 0.235 0.240 0.245
-1/4 -1/4
should be related to the different deformation modes of the two T (K )
3.4
tests. Moreover, the larger increase in the tensile modulus could
be explained considering that the tensile properties are mainly
3.2
(b) PEEK/SWCNT-BP
filler-dominated, whilst the flexural properties are principally ma-
trix-dominated. Similar observations were reported for SWCNT–
reinforced PEEK [16] and PPS [18] bulk laminates. 3.0
σ (S/cm)

3.4. Electrical conductivity 2.8 3.2

3.0
σ (S/cm)

The temperature dependence of the electrical conductivity (r)

2.6 2.8
for pristine BP and the laminates is depicted in Fig. 6. At room tem- σ = σ o exp[-T 1 /(T-T O )]
perature, r of the randomly oriented SWCNT-BP is 49 S/cm σo 5.52 ± 0.65 S/cm
2.6
T 1 152.6 ± 8.6 K
(Fig. 6a), consistent with [32] or smaller [7,33] than previously re- 2.4 T O 82.5 ± 4.1 K
2.4
ported values. The discrepancies are attributed to the different de- 280 300 320 340 360 380
gree of porosity and thickness of the BP sheets, since thinner T (K)
networks are found to exhibit lower intertube contact resistance 2.4
(c) PPS/SWCNT-BP
2.2
200
ΕF
5 σ FM
2.0
σ (S/cm)

160 1.8
Flexural Modulus, EF (GPa)

Flexural Strenght σ FM (MPa)

2.2
4
σ (S/cm)

1.6 2.0
120
3 1.8 σ = σ o exp [-T 1 /(T-T O )]
1.4 σ o 3.80 ± 0.28 S /cm
T 1 168.1 ± 9.3 K
80 1.6
T O 74.5 ± 6.9 K
2 1.2
280 300 320 340 360 380
T (K )
1.0
1 40
270 285 300 315 330 345 360 375
T (K)
0 0
PPS PPS/SWCNT-BP PEEK PEEK/SWCNT-BP Fig. 6. Temperature dependence of the electrical conductivity r for pristine
SWCNT-BP (a), PEEK/SWCNT-BP (b) and PPS/SWCNT-BP (c) composites. The insets
Fig. 5. Comparison of the flexural modulus (EF) and flexural strength (rFM) of the show the fittings based on the 3D VRH model (a) and the FIT model (b and c). (For
composites and the neat polymers. (For interpretation of the references to colour in interpretation of the references to colour in this figure legend, the reader is referred
this figure legend, the reader is referred to the web version of this article.) to the web version of this article.)
 A.M. Díez-Pascual et al. / Composites: Part A 43 (2012) 1007–1015 1013

conduction mechanism. This approximation considers that the BP 3.5. Thermal conductivity
is a 3D network of dispersed bundle–bundle junctions, which allow
the carriers to move within the sheet. Therefore, the electrical con- Fig. 7a shows the temperature dependence of the thermal con-
duction occurs either tube–tube within the nanotube bundles or/ ductivity (K) for the SWCNT-BP, the neat polymers and the com-
and between neighbour bundles through their contacts. According posites. K of pristine BP at 273 K is 11 W/mK, and increases
to this model, the evolution of r with temperature can be ex- almost linearly with temperature over the range studied; similar
pressed as [35]: quasi-linear behaviour has been reported for BPs and nanotube
bundles prepared from different conditions, ascribed to a gradual
increase in the phonon population [42]. At room temperature, K
r ¼ r0 exp½ðT o =TÞ1=4  ð3Þ is 13.5 W/mK, value that lies far below the theoretical prediction
for a single SWCNT (6600 W/mK). The strong discrepancy could
be attributed to the metallic impurities embedded in the BP, which
where r0 is the conductivity at infinite temperature and To is the
increase scattering during phonon transfer, and the high interfacial
characteristic temperature related to the density of states at the
thermal resistance between nanotubes within a bundle. Moreover,
Fermi level and the localisation length [33]. The inset in Fig. 6a
the BP contains a great number of micropores, as revealed by SEM
demonstrates that r of the SWCNT-BP can be well described by
micrographs [11], filled with air (K 0.026 W/mK). When the BP
the 3D VRH model, in agreement with previous works [32,37]. Hop-
was heated from the front of the film, the heat was transferred
ping conduction is most likely the characteristic electric transport
from the SWCNTs to air, which results in natural convection of
behaviour between nanotubes and bundles.
air in the micropores, reducing the heat transfer to the rear. In
Neat PPS and PEEK are electrically insulating (r  1013 S/cm),
addition, the magnitude of K is strongly dependent on the charac-
and their conductivity improved by about thirteen orders of mag-
teristics of the CNT bundles within the BP mat (diameter, degree of
nitude upon BP impregnation. The composites show r values more
entanglement, alignment, etc.). Previous works [10,33] revealed
than one order of magnitude lower than the BP, since the nanotube
room temperature K data between 50 and 200 W/mK for magnet-
bundles are wrapped by a thin layer of insulating polymer, which
ically aligned BPs, where the nanotubes are tightly packed and ori-
acts as a potential barrier to inter-tube hopping. Thus, at room
temperature, r experiments approximately 18 fold and 25 fold de-
crease for PEEK/SWCNT-BP (Fig. 6b) and PPS/SWCNT-BP (Fig. 6c),
(a)
respectively, in comparison to pristine BP. Qualitatively similar
behaviour has been reported for other BP/polymer composites
10
[7,38]. The lower r found for PPS laminate should be attributed
to its higher resin content, since the electrical resistance SWCNT–
matrix is much more severe than between SWCNTs. r of the com-
posites also follows a non-linear tendency, albeit slightly different
K (W/m·K)

from the observed for the BP. The conductive mechanism in the
SWCNT-BP
composites can be described by quantum tunnelling theory in 1 PPS
which the barrier height diminishes as the temperature rises PPS/SWCNT-BP
PEEK
[39]. This behaviour follows the thermal fluctuation-induced tun- PEEK/SWCNT-BP
nelling (FIT) model proposed by Sheng as [40]:

r ¼ r0 exp½T 1 ðT  T o Þ ð4Þ
0.1
270 285 300 315 330 345 360 375
T (K)
where T1 represents the energy required for an electron to cross the
insulator gap between conductive particles and is related to the 6
activation energy, and To is the temperature above which the ther- (b) PEEK/SWCNT-BP
PEEK/SWCNT-BP PPS/SWCNT-BP
PPS/SWCNT-BP
Rule of Mixtures
mal activated conduction over the barrier begins to occur [38]. As it Nan et al.
can be observed in the inset of Fig. 6b and c, the predictions of the Maxwell-Eucken
5
FIT model correlate very well with the experimental data. This indi-
cates that the polymer wrapping of the nanotubes strongly modifies
the electrical conduction mechanism; r in the composites seems to 4
K (W/m·K)

be hindered by potential barriers between neighbouring nanotube

bundles due to the polymer coating. PPS/SWCNT composite exhibits
slightly lower To than the PEEK counterpart. Taking into account 3
that To has been reported to decrease with increasing degree of dis-
persion and concentration of the fillers [38,41], and that PEEK based
composite has higher CNT loading, the lower To value suggests the 2
more uniformity of the PPS composite, which could be related to
its improved degree of impregnation, ascribed to the lower viscosity
of this polymer in comparison with PEEK. On the other hand, 1
slightly lower T1 is found for PEEK/SWCNT composite in comparison 270 285 300 315 330 345 360 375
to the PPS counterpart, in agreement with its higher CNT content. As T (K)
the concentration increases, the samples become more conductive,
and the conductivity is less sensitive to temperature, hence T1 de- Fig. 7. (a) Thermal conductivity (K) as a function of temperature for randomly
oriented SWCNT-BP, neat PEEK, PPS and their composites. (b) Theoretical predic-
creases [41]. Overall, our results demonstrate the great potential tions for K of PEEK/SWCNT-BP and PPS/SWCNT-BP composites. See the text for the
of the SWCNT-BP to improve the electrical properties of thermo- fitting models. (For interpretation of the references to colour in this figure legend,
plastic polymers for practical applications. the reader is referred to the web version of this article.)
1014 A.M. Díez-Pascual et al. / Composites: Part A 43 (2012) 1007–1015

ented in-plane, whereas values of 20 W/mK have been reported for PEEK and PPS composites are 150% and 132% higher than the
randomly oriented BPs. The lower K obtained in this work is prob- experimental data. The strong deviations should be due to the ef-
ably ascribed to the relatively low density of the BP, since denser fects of the CNT–matrix interactions as well as inter-tube and in-
nanotube networks lead to more tube–tube contacts, hence en- ter-bundle contacts, which are not considered by the model.
hanced conductivity. According to molecular dynamics simulation [47], CNT thermal con-
Both PEEK and PPS are semicrystalline thermoplastics, and ex- ductivity may decrease when the tubes are in contact with the poly-
hibit low K values (0.21 and 0.26 W/mK at room temperature, mer. Moreover, a good interfacial bonding CNT–matrix and the
respectively). It should be noticed that the thermal conductivity existence of intensive p–p stacking, as revealed by the Raman spec-
of this type of polymers is sensitively affected by the chemical tra [11], cause low thermal conductance through the CNT–matrix
components, molecular weight, structural defects, etc. Moreover, interface, reducing K to much lower values than the estimated from
K is reported to increase with crystallinity [43], and due to the pho- the intrinsic thermal conductivity of the nanotubes and their vol-
non scattering at the interface between the amorphous and crys- ume fraction. This overestimation of K using the rule of mixtures
talline phases, the prediction of K vs. temperature is complex. was also reported for bulk PEEK/SWCNT [23] and PPS/SWCNT [18]
The higher K found for PPS in comparison to PEEK is attributed composites.
to its higher degree of crystallinity, as revealed from DSC analysis The Maxwell–Eucken model can also be employed to predict K
[11]. K of PEEK rises slightly with increasing temperature, owed of reinforced composites. In the case that K of the continuous phase
to the increased molecular mobility, being 0.23 W/mK at 373 K, is lower than that of the dispersed phase, it can be written as [48]:
in agreement with previous works [44]. In contrast, PPS exhibits
a maximum K around 350 K and then it decreases slightly at higher
K c ¼ K f ð2K f þ K m  2BÞ=ð2K f þ K m  BÞ ð6Þ
temperatures; the phonon scattering might be associated with where B = (Kf  Km)/(1  Vf). The values obtained by Eq. (6) are also
changes in the amorphous structure, since the Tg lies in this region remarkably higher than the measured data, being the differences at
of lower conductivity. For the BP-reinforced composites, experi- room temperature 76% and 73% for PEEK and PPS composites. A
mental results indicate an average 10 fold K increase in comparison more accurate approximation can be provided by the model pro-
to the neat polymers. This significant enhancement is ascribed to posed by Nan et al. [49]:
the uniform impregnation of the BP with the polymers, and is of
the same order of magnitude to those reported for epoxy compos- K c ¼ ð3K m þ K f V f Þ=ð3  2V f Þ ð7Þ
ites incorporating randomly oriented BPs [10]. The laminates show
Although this model was designed for SWCNTs randomly dis-
the same quasi-linear temperature dependence observed for the
persed within a matrix, it can be applied to BP-reinforced compos-
BP, indicating that the phonon population is the dominating contri-
ites replacing Kf by K of the BP. Eq. (7) predicts a lower thermal
bution. However, a slight deviation is observed at the highest tem-
conductivity than the rule of mixtures or the Maxwell–Eucken
perature tested, particularly for PPS based composite, which could
model, and more closely follows the experimental results. Never-
be related to the fact that this temperature lies close to the glass
theless, the differences between the measured and calculated data
transition region. In fact, previous studies on polystyrene/SWCNT
become larger with increasing temperature; at 273 K the devia-
composites [45] indicated a change in the Kapitza resistance, the
tions are lower than 6%, whilst at 373 K the predictions are 20%
interfacial resistance to heat transfer between the CNTs and the
higher than the experimental results, consistent with the small
surrounding matrix, in the vicinity of the Tg. This resistance origi-
deviations from the quasi-linear relationship observed at higher
nates from a modulus mismatch between the two composite
temperatures. It is confirmed that K of these BP-reinforced lami-
phases; the greater the mismatch, the greater the resistance [46],
nates can be reasonably well predicted by simple models designed
and is very high in composites with CNTs because of their high
for conventional polymer/CNT composites.
stiffness and high surface area. Therefore, the temperature depen-
dence of K changes when the polymer matrix becomes less stiff at
temperatures close to the Tg, indicating an increase in the Kapitza 4. Conclusions
resistance. Room temperature K values of PPS/SWCNT-BP and
PEEK/SWCNT-BP are 1.38 and 1.52 W/mK, and at 373 K increase The mechanical and electrical properties as well as thermal con-
by 78% and 92%, respectively. Note that, albeit neat PEEK pos- ductivity of high-performance PPS and PEEK based laminates
sesses lower thermal conductivity than PPS, upon BP impregnation incorporating randomly oriented SWCNT-BP were analysed. DMA
a significant reversal of their conductivity occurs over the entire studies revealed that these composites exhibit considerably higher
measured range. This is attributed to the higher CNT content of storage modulus and glass transition temperature than the neat
PEEK laminate, as indicated earlier, and further confirms that the polymers. Tensile and flexural tests showed remarkable enhance-
thermal conduction in the composites is SWCNT dominated. ments in stiffness and strength, attributed to strong SWCNT–ma-
Different models have been reported to describe K of polymer trix interfacial adhesion. However, a moderate decrease in
composites [43]. A good prediction can be provided by the simple toughness was observed, related to a reduction in the ductility of
rule of mixtures, which can be expressed as: the matrix. The Young’s moduli of the real polymer/SWCNT com-
posites were estimated based on a simple rule of mixtures. The
K c ¼ K f V f þ K m ð1  V f Þ ð5Þ
incorporation of the BP sheet led to an exceptional increase in
where Kc, Kf and Km are the thermal conductivities of the composite, the electrical conductivity of the polymers. The 3D variable-range
filler and matrix, respectively, and Vf the filler volume fraction. K of hopping model was applied to describe the temperature depen-
isolated and individual SWCNTs would greatly overestimate the dence of the BP electrical conductivity, whilst the fluctuation-in-
conductivity of the composites, due to the large interfacial thermal duced tunnelling mechanism was used to explain the behaviour
resistance between CNTs within a bundle [10]. These discrepancies of the laminates. For the BP-reinforced composites, experimental
may be reduced by instead using the conductivity of the BP; there- results indicated an average 10 fold thermal conductivity improve-
fore, taking K of the BP as Kf, the theoretical values for the two com- ment in comparison to the pure matrices. The temperature depen-
posites were calculated, and the results are shown in Fig. 7b. As it dence of the thermal conductivity for both pristine BP and the
can be observed, the predictions by the rule of mixtures are consid- laminates followed a quasi-linear relationship, ascribed to a pro-
erably above the measured data over the temperature range stud- gressive increment in the phonon population. The thermal conduc-
ied; particularly, at ambient temperature, the estimated values for tivities of the composites were reasonably well predicted using the
 A.M. Díez-Pascual et al. / Composites: Part A 43 (2012) 1007–1015 1015

model proposed by Nan et al.; small deviations were found at high [19] Sandler J, Werner P, Shaffer MSP, Demchuck V, Altstadt V, Windle AH. Carbon-
nanofibre-reinforced poly(ether ether ketone) composites. Composites Part A
temperatures, attributed to a change in the Kapitza resistance. This
2002;33(8):1033–9.
study demonstrates the high potential of SWCNT-BPs as reinforce- [20] Yang J, Xu T, Lu A, Zhang Q, Tan H, Fu Q. Preparation and properties of poly(p-
ment for thermoplastic polymers. The manufactured laminates phenylene sulphide)/multiwall carbon nanotube composites obtained by melt
possess improved electrical, thermal and mechanical performance, compounding. Comp Sci Technol 2009;69:147–53.
[21] Malik S, Rösner H, Hennrich F, Böttcher A, Kappes MM, Beck T, et al. Failure
suitable for a wide variety of ground, air and space applications. mechanism of single-walled carbon nanotubes thin films under tensile loads.
Phys Chem Chem Phys 2004;6:3540–4.
Acknowledgments [22] Ghiorse SR. Effect of void content on the mechanical properties of carbon/
epoxy laminates. SAMPE Quart 1993;24(2):54–9.
[23] Díez-Pascual AM, Naffakh M, Gónzalez-Domínguez JM, Anson A, Martinez-
Financial support from a coordinated project between the Na- Rubi Y, Simard B, et al. High-performance PEEK/carbon nanotube composites
tional Research Council of Canada (NRC) and the Spanish National compatibilized with polysulfones. II – Mechanical and electrical properties.
Carbon 2010;48(12):3500–11.
Research Council (CSIC), and partial support from the Spanish Min- [24] Song L, Zhang H, Zhang Z, Xie S. Processing and performance improvements of
istry of Science and Innovation (National project MAT2010-21070- SWNT paper reinforced PEEK nanocomposites. Composites Part A
C02-01) are gratefully acknowledged. Dr. A. Diez would like to 2007;38(2):388–92.
[25] Yu MF, Files BS, Arepalli S, Ruoff RS. Tensile loading of ropes of single wall
thank to the Spanish Ministry of Science and Innovation (MICINN) carbon nanotubes and their mechanical properties. Phys Rev Lett
for Juan de la Cierva postdoctoral Contract. 2000;84(24):5552–5.
[26] Diez-Pascual AM, Martinez G, Martinez MT, Gomez MA. Novel nanocomposites
reinforced with hydroxylated poly(ether ether ketone)-grafted carbon
References
nanotubes. J Mater Chem 2010;20:8247–56.
[27] Tang L-C, Zhang H, Wu X-P, Zhang Z. A novel failure analysis of multi-walled
[1] Lahiff E, Leahy R, Coleman JN, Blau WJ. Physical properties of novel free- carbon nanotubes in epoxy matrix. Polymer 2011;52:2070–4.
standing polymer-nanotube thin films. Carbon 2006;44(8):1525–9. [28] Blake R, Coleman JN, Byrne MT, McCarthy JE, Perova TS, Blau WJ.
[2] Cheng B, Bao J, Park J, Liang Z, Zhang C, Wang B. High mechanical performance Reinforcement of poly(vinyl chloride) and polystyrene using chlorinated
composite conductor: multi-walled carbon nanotube sheet/bismaleimide polypropylene grafted carbon nanotubes. J Mater Chem 2006;16:4206–13.
nanocomposites. Adv Funct Mater 2009;19(20):3219–25. [29] Jafari SH, Gupta AK. Impact strength and dynamic mechanical properties
[3] Whitten PG, Spinks GM, Wallace GG. Mechanical properties of carbon correlation in elastomer-modified polypropylene. J Appl Polym Sci
nanotube paper in ionic liquid and aqueous electrolytes. Carbon 2000;78(5):962–71.
2005;43(9):1891–6. [30] Grein C, Plummer CJG, Germain Y, Kausch HH, Béguelin P. Essential work of
[4] Zhang X, Sreekumar TV, Liu T, Kumar S. Properties and structure of nitric acid fracture of polypropylene and polypropylene blends over a wide range of test
oxidized single wall carbon nanotube films. J Phys Chem B 2004;108(42): speeds. Polym Eng Sci 2003;43(1):223–33.
16435–40. [31] Zhao YF, Xiaso M, Wang Sj, Ge XC, Meng YZ. Preparation and properties of
[5] Berhan L, Yi YB, Sastry AM, Munoz E, Selvidge M, Baughman R. Mechanical electrically conductive PPS/expanded graphite composites. Comp Sci Technol
properties of nanotube sheets: alterations in joint morphology and achievable 2007;67:2528–34.
moduli in manufacturable materials. J Appl Phys 2004;95(8):4335–45. [32] Park JG, Smithyman J, Lin C-Y, Cooke A, Kismarahardja AW, Li S, et al. Effects of
[6] Frizzell CJ, in Het Panhuis M, Coutinho DH, Balkus KJ, Minett AI, Blau WJ, et al. surfactants and alignment on the physical properties of single walled carbon
Reinforcement of macroscopic carbon nanotube structures by polymer nanotube buckypaper. J Appl Phys 2009;106:104310–6.
intercalation: the role of polymer molecular weight and chain conformation. [33] Fisher JE, Zhou W, Vavro J, Llaguno MC, Guthy C, Haggenmueller R, et al.
Phys Rev B 2005;72(24):245420–8. Magnetically aligned single wall carbon nanotube films: preferred orientation
[7] Pham GT, Park Y-B, Wang S, Liang Z, Wang B, Zhang C. Mechanical and and anisotropic transport properties. J Appl Phys 2003;93:2157–63.
electrical properties of polycarbonate nanotube buckypaper composite sheets. [34] Yosida Y, Oguro IJ. Variable range hopping conduction in bulk samples
Nanotechnology 2008;19(32):325705–12. composed of single-walled carbon nanotubes. J Appl Phys 1999;86:999–1004.
[8] Jianwen BQC, Wang X, Liang R, Zhang C, Wang B, Kramer L et al. [35] Mott NF, Davis EA. Electronic processes in non-crystalline materials. 2nd
Functionalization of carbon nanotube sheets for high-performance ed. Oxford: Clarendon Press; 1979.
composites applications. SAMPE 2009, Baltimore, Maryland; 2009. [36] Kaiser AB, Dusberg G, Roth S. Heterogeneous model for conduction in carbon
[9] Wang Z, Liang ZY, Wang B, Zhang C, Kramer L. Processing and property nanotubes. Phys Rev B 1998;57:1418–21.
investigation of single-walled carbon nanotube (SWNT) buckypaper/epoxy [37] Benoit JM, Corraze B, Chauvet O. Localization, Coulomb interactions, and
resin matrix nanocomposites. Composites Part A 2004;35(10):1225–32. electrical heating in single-wall carbon nanotubes/polymer composites. Phys
[10] Gonnet P, Liang SY, Choi ES, Kadambala RS, Zhang C, Brooks JS, et al. Thermal Rev B 2002;65:241405–9.
conductivity of magnetically aligned carbon nanotube buckypapers and [38] Kymakis E, Amaratunga GAJ. Electrical properties of single-walled carbon
nanocomposites. Curr Appl Phys 2006;6:119–22. nanotube composite films. J Appl Phys 2006;99:84302–9.
[11] Diez-Pascual AM, Jingwen G, Simard B, Gómez-Fatou MA. Poly(phenylene [39] Kim HM, Kim K, Lee CY, Joo J, Cho SJ, Yoon HS, et al. Electrical conductivity and
sulphide) and poly(ether ether ketone) composites reinforced with electromagnetic interference shielding of multiwalled carbon nanotube
single-walled carbon nanotube buckypaper. Part I: Structure and thermal composites containing Fe catalyst. Appl Phys Lett 2004;84(4):589–92.
properties. Composites Part A, doi:10.1016/j.compositesa.2011.11.002. [40] Sheng P. Fluctuation-induced tunneling conduction in disordered materials.
[12] Parker WJ, Jenkins RJ, Butler CP. Flash method of determining thermal Phys Rev B 1980;21(6):2180–95.
diffusivity specific heat and thermal conductivity. J Appl Phys 1961;32(12): [41] Sankapal B, Setyowati K, Chen J, Liu H. Electrical properties of air-stable,
1679–84. iodine-doped carbon nanotube/polymer composites. Appl Phys Lett
[13] Lopes PE, van Hattum F, Pereira CMC, Novoa PJRO, Forero S, Hepp F, et al. High 2007;91:173103–5.
CNT content composites with CNT buckypaper and epoxy resin matrix: [42] Lau KT, Hui D. The revolutionary creation of new advance materials-carbon
impregnation behaviour, composite production and characterization. Comp nanotube composites. Composites Part B 2002;33:263–77.
Struct 2010;92(6):1291–8. [43] Han Z, Fina A. Thermal conductivity of carbon nanotubes and their polymer
[14] Lu JP. Elastic properties of carbon nanotubes and nanoropes. Phys Rev Lett nanocomposites: a review. Prog Polym Sci 2011;36(7):914–44.
1997;79(7):1297–300. [44] Choy CL, Kwok KW, Leung WP, Lau FP. Thermal conductivity of poly(ether
[15] Xu M, Futaba DN, Yamada T, Yumura M, Hata K. Carbon nanotubes with ether ketone) and its short-fiber composites. J Polym Sci Part B: Polym Phys
temperature – invariant viscoeslaticity from 196 °C to 100 °C. Science 1994;32(8):1389–97.
2010;330:1364–8. [45] Peters JE, Papavassiliou DV, Grady BP. Unique thermal conductivity behaviour
[16] Diez-Pascual AM, Ashrafi B, Naffakh M, Gónzalez-Domínguez JM, Johnston A, of single-walled carbon nanotube-polystyrene composites. Macromolecules
Simard B, et al. Influence of carbon nanotubes on the thermal, electrical and 2008;41:7274–7.
mechanical properties of poly(ether ether ketone)/glass fiber laminates. [46] Swartz ET, Pohl RO. Thermal boundary resistance. Rev Mod Phys
Carbon 2011;49:2817–33. 1989;61:605–68.
[17] Díez-Pascual AM, Naffakh M, Gómez MA, Marco C, Ellis G, Gónzalez- [47] Berber S, Kwon Y-K, Tománek D. Unusually high thermal conductivity of
Domínguez JM, et al. Development and characterization of PEEK/carbon carbon nanotubes. Phys Rev Lett 2000;84:4613–6.
nanotube composites. Carbon 2009;47(13):3079–90. [48] Hashin Z, Shtrikman S. A variational approach to the theory of the effective
[18] Diez-Pascual AM, Naffakh M, Gómez-Fatou MA. Mechanical and electrical magnetic permeability of multiphase materials. J Appl Phys 1962;33:3125–31.
properties of novel poly(ether ether ketone)/carbon nanotube/inorganic [49] Nan C-W, Shi Z, Lin Y. A simple model for thermal conductivity of carbon
fullerene-like WS2 hybrid nanocomposites: experimental measurements and nanotube-based composites. Chem Phys Lett 2003;375(5–6):666–9.
theoretical predictions. Mater Chem Phys 2011;130:126–33.