IET Nanodielectrics Electrical and Thermal Properties of Surface Passivated Carbon Nanotube/Polyvinylidene Fluoride Composites NDE-2018-0010

IET Nanodielectrics
Electrical and Thermal Properties of Surface
Passivated Carbon Nanotube/Polyvinylidene
Fluoride Composites
NDE-2018-0010.R1 | Research Article
Submitted on: 27-06-2018
Submitted by: Sheng-Guo Lu, Dandan Li, Caihong Lei, Yingxiu Ouyang, Zhihong Cai, Yongmao Deng, Ruijie Xu
Keywords: CARBON NANOTUBES, COMPOSITE MATERIALS, ELECTRICAL INSULATION
ReView by River Valley Technologies IET Nanodielectrics
2018/06/27 15:39:34 IET Review Copy Only 1
Answers to the Reviewer Comments
Comments to the Author
In this paper, the authors have reported the electrical and thermal
Properties of Surface Passivated Carbon Nanotube and Polyvinylidene
Fluoride Composites. The composite material shows the changes of
electrical resistivities, whereas thermal conductivities does not change
significantly. They have studied electrical properties using few existing
methods like Maxwell-Wagner polarization, universal law, and Gerhardt’s
method. The work is experimental and time taking. The paper is well written.
Further improvement is needed.
Thanks!
1. Why author have chosen MWCNT with poly(vinylidene fluoride)? Please
justify.
Answer (Abbreviated as A below)?It is well known that the MWCNT is a good
electrical and thermal conductor, and will enhance the electrical and thermal
conductivity if the polymer matrix in added this kind of material. Here the surface
passivated MWCNT was used because our objective is to procure a composite that
has a lower electrical conductivity, but a higher thermal conductivity. In addition,
the MWCNT is usually cheaper than SWCNT in the commercial market. For
poly(vinylidene fluoride) (PVDF), this is a well-known piezoelectric polymer, and
also a commercially cheap material.
2. Explanation also needed for SWCNT, Mixed CNT and Graphene Nanoribbon
like nanomaterial.
A: SWCNT or mixed CNT and graphene nanoribbon can also be incorporated into
PVDF to form the composite which has the similar properties as MWCNT/PVDF.
Due to the reason that the MWCNT is cheaper, and also more convenient to be
purchased, we chose MWCNT. In principle, they are much similar.
3. What is the electron mean free path of MWCNT they have analyzed in their
experiment?
A: We didn’t investigate the electron mean free path of MWCNT. But according
to Kyriakou et al’s work, the inelastic mean free path of electron in MWCNT with
a diameter larger than 20 nm will be in the range of ~ 1 to 10 nm. The diameter of
the MWCNT used in our work is 45 nm. This means that the electron will be
localized in the MWCNT, to produce the electronic transportation.
4. In Fig.3 the labels are not clear please improve it.
ReView by River Valley Technologies IET Nanodielectrics
2018/06/27 15:39:34 IET Review Copy Only 2
A: Thanks. We have improved the labels in Fig. 3.
5. References are not well organized. Improvement is needed.
A: OK. The references have been organized again. Thanks!
6. In Fig.4 the MWCNT content vs. resistivity plot two different line is
highlighted one is red and other is dotted blue. Please mention the
two types.
A: OK. We have mentioned them in the context.
7. “For simplicity, tau corresponds the maximum in the taul
distribution.” please improve the tau1.
A?OK. The tau1 has been improved.
8. typo mistake z’-z”’ in page-5 1st column.
A: The typo has been corrected.
9. Like equation (3) and (4) the Eqn.(1) and (2) total impedance should
be function of angular frequency.
A: Yes. The total impedance is a function of angular frequency, since the
capacitor is appeared in the equivalent circuit.
10.It will be good if the experimental real and imaginary impedance
presented in a table format including RC value.
A: We have the data of experimental real and imaginary impedance, which are the
same as the curves presented.
ReView by River Valley Technologies IET Nanodielectrics
2018/06/27 15:39:34 IET Review Copy Only 3
1
Electrical and Thermal Properties of Surface Passivated Carbon
Nanotube/Polyvinylidene Fluoride Composites
Yingxiu Ouyang;, Yongmao Deng;, Dandan Li;, Zhihong Cai, Ruijie Xu, Caihong Lei, and Shengguo
Lu*
Guangdong Provincial Key Laboratory of Functional Soft Condensed Matter, School of Materials and Energy,
Guangdong University and Technology, Guangzhou, 510006 China
*[email protected]
;Equally contribute authors
Abstract: The composites reported here were prepared via a melt-rolling mixing process, in which surface passivated
multi-wall carbon nanotube and poly(vinylidene fluoride) (PVDF) were used as the filler and the matrix, respectively. Xray
diffraction (XRD), precision inductance-capacitance-resistance (LCR), and laser flash thermal-diffusivity analyses
were employed to characterize the structural, electrical and thermal properties. Results indicate that through the surface
passivation of carbon nanotubes, the electrical resistivities of the composites were greatly enhanced, while the thermal
conductivities do not change significantly. The electrical properties were discussed in terms of the comparison with the
Maxwell-Wagner polarization, universal law, and Gerhardt’s method.
1. Introduction
Multi-wall carbon nanotube (MWCNT) is a kind of
material having high mechanical strength and sound electrical
and thermal conductivities. MWCNT has been regarded as an
ideal filler for advanced composites due to its excellent
mechanical and electrical properties 1-3. In addition, singlewall
carbon nanotube (SWCNT) 4, mixed CNT 5 and
graphene nanoribbon 6 are also good fillers due to their
good electrical and thermal properties as MWCNT.
Poly(vinylidene fluoride) (PVDF) and its copolymers or
terpolymers are important functional materials, which have
been widely used as ferroelectric memories, dielectric energy
storages, various sensors and mechanical actuators because of
their good dielectric, ferroelectric, piezoelectric and
electromechanical properties 7-9. Thus it is better to form
composites to get multiple functionalities for the versatile
applications. Recently, a large number of investigations have
been carried out in order to enhance the functional properties
of MWCNT filled composites 10-12.
In general, the electrical conductivity has a close
relationship with the thermal conductivity. In some situations,
e.g., thermal management pipes in an electrical vehicle 13,
however, the good thermal conductivity is demanded but in
the meantime the electrical conduction should be avoided, i.e.,
good electrical insulation is needed. Fortunately, the surface
passivation of the MWCNTs may lead to a great reduction of
the electrical conductivity on the surfaces of MWCNTs, thus
the electrical conductivity of the composites using passivated
MWCNT as a filler will be much lower even if they contact
one another. Recently, there are quite a lot of investigations
on the oxidation and functionalization of the surface of
MWCNTs 14-19. The oxidation using acid, e.g. nitric acid
or sulphuric acid, results in the formation of surface
carboxylic groups in terms of rapid formation of carbonyl
groups, and then transformation into phenol, lactones,
anhydrides, or carboxylic groups 15. These groups might be
decomposed by heat treatment, releasing CO and/or CO2.
Partial groups, e.g. carboxylic acids, may be removed when
heat treated at 400 ?C, while almost all the groups will be
cleaned when heat treated at 900 ?C 14. Although the
groups are removed, it was found that the resistivities of the
MWCNT/polymer composites increase quite a lot. This
means that the heat treatment at 900 ?C could not remove all
the groups completely, and the surface structures have been
really changed. In this work, MWCNTs are passivated using
H2SO4/HNO3, then MWCNT/PVDF composites are prepared
by a melt-rolling mixing method. The structural, electrical
and thermal properties were characterized. It is found that the
insulation resistance is greatly enhanced while the thermal
conductivity keeps almost the same.
2. Experimental procedures
2.1. Passivation of MWCNT
For the preparation of oxidized MWCNT, Firstly, the
MWCNT were oxidized by H2SO4 and HNO3 (3:1 in volume)
at 90 ?C for 3 h and 2 h, respectively. Then, the oxidized
CNTs, tetrahydrofuran (KH550) and N,N’-
Dicyclohexylcarbodie (DCC) were mixed in tetrahydrofuran
(THF) at 60 ?C for 1 h to get the silicone grafted CNTs. After
that, the grafted CNTs were hydrolyzed in ethanol. Finally,
the hydrolyzed CNTs, and DCC (ratio 3:1:1) were reacted in
THF at 60 ?C for 2 h. At last the obtained filler was dried in
a vacuum oven at 90 ?C for 6 h 20.
2.2. Preparation of MWCNT/PVDF composites
The MWCNT was prepared via a chemical vapor
deposition method. As shown in Fig. 1, the MWCNT has a
diameter of c.a. 45 nm, a length larger than 1 ?m, and a
specific surface area of 90 – 120 m2/g. For the preparation of
oxidized MWCNT/PVDF composites, the PVDF was firstly
dissolved in acetone, then the passivated MWCNTs were
added and stirred for 5 ? 10 h. After that, the solution was
poured into a glass utensil, stirred at 60 ?C for 5 ? 15 h, and
dried at 80 ?C for 8 h in an oven. Furthermore, the composite
was weighed with the weight ratios of MWCNT/PVDF of
ReView by River Valley Technologies IET Nanodielectrics
2018/06/27 15:39:34 IET Review Copy Only 4
2
1:60, 1:40, 1:20 and 1:15 respectively (specified as Samples1,
2, 3, and 4, respectively) for passivated MWCNT/PVDF
composites, and 1:10, 2:10, 3:10, and 4:10 respectively
(specified as Samples 5, 6, 7, and 8, respectively) for pristine
MWCNT/PVDF composites. Above mentioned
compositions were then melt mixed using a two-roll mixing
machine at 175 ?C for several times, then the rough composite
plates were put into a hot presser and pressed at 180 ?C for 10
min, and then cooled down to room temperature with the
machine, and dwelled for 5 min at room temperature.
2.3. Electrical and Thermal Characterization
The large piece of composite obtained from hotpressing
was cut into pieces with 10?10?0.8 mm3 in size. Au
of 10?10 mm2 in area was sputtered on both surfaces of the
sample (SBC-12, KYKY Technology). The electrical
properties were measured at room temperature using a
precision LCR meter (HP 4284A) with a temperature
chamber (homemade) in which the temperature could be
varied from -150 to 150 ?C.
The same samples were also used to measure their
thermal conductivities at room temperature using a thermal
conductivity apparatus (LFA447 Nanoflash). During the
measurement, a sample with larger size (10 mm in diameter
and 2 mm in thickness) is needed for precise characterization
of the thermal behavior. Three points of each sample were
measured, and the averaged value was calculated and used as
the sample’s thermal conductivity.
3. Results and discussion
3.1. XRD and SEM characterization
Figure 1 shows the scanning electron microscopy
(SEM) image of MWCNT. One can see that the CNTs are
uniform in diameter, and demonstrate multiwall
morphologies. The detailed dimensions and the specific
surface areas are mentioned in the last section.
Figure 2 demonstrates the XRD patterns for
MWCNT/PVDF composites with MWCNT/PVDF ratios of
1:60, 1:40, 1:20 and 1:15 respectively. One can see that with
the increasing MWCNT content, the (002) peak of MWCNT
Fig.1. SEM image of MWCNT
Fig. 2. XRD patterns for MWCNT/PVDF composites of
Samples 1, 2, 3, and 4, respectively
at 26.2? becomes larger, while the peak at 26.8? from ? phase
of PVDF (021) is depressed. In addition, the (020) peak of ?
phase of PVDF is divided into two peaks, i.e., (100) and (020),
after the incorporation of MWCNT, these two peaks become
more distinct with the increase of MWCNT content. The
(110)/(200) peak of ? phase does not change significantly for
four composite samples.
3.2. Electrical properties
The real and imaginary parts of impedance as a
function of frequency for the four composites (Samples 1, 2,
3 and 4) measured at room temperature are shown in Figs. 3
(a), (b), (c) and (d). The figure shown on the right of each
figure shows the conductivity ? as a function of angular
frequency ?. Except for Sample 4, the profiles for other three
samples are similar to Debye relaxations in the real –
imaginary permittivity plot for conventional dielectrics. For
Sample 4, one can find out that the static resistance (~ real
part at the lowest frequency 20 Hz) greatly drops from 143.3
k? (Sample 1), 154.2 k? (Sample 2), to ~ 36.1 k? (Sample
3), and then ~ 5.4 k? (Sample 4). On the contrary, imaginary
part at the lowest frequency 20 Hz rises up from ~ 1.1 k?
(Sample 1), 1.5 k? (Sample 2), to ~ 3.2 k? (Sample 3), and
then 53.2 k? (Sample 4). The large variation of real and
imaginary parts of impedance at the lowest frequency 20 Hz
for 4 samples strongly suggests that there might be a
mechanism that makes contribut ion to the electrical
properties of Sample 4, but it is still unclear right now. For
the right one of each figure, all the samples demonstrate a
power law of the conductivity versus angular frequency,
which is the typical feature of so-called “universal law”
proposed by A. K. Jonscher for large ranges of frequency,
time and temperature, insensitive to the many material
properties 21. The common basis for the universal dielectric
response was regarded as the presence of interactions arising
from the close proximity of atoms and molecules and the
abrupt or discontinuous nature of the dipolar or charge carrier
transitions between their preferred orientations or positions.
The second feature comes from the Maxwell-Wagner (M-W)
polarization 22,23, which is usually referred to in the
ReView by River Valley Technologies IET Nanodielectrics
2018/06/27 15:39:34 IET Review Copy Only 5
3
Fig. 3. (left) Real and imaginary parts of impedance as a function of frequency and (right) conductivity as a function of angular
frequency for MWCNT/PVDF composites of Samples (a) 1,(b) 2, (c) 3, and (d) 4 at room temperature.
inhomogeneous or multiphase systems, for instance, the
composites used here. The mechanism of M-W polarization
is due to the charge accumulation in the interface between two
kinds of materials or phases, which show quite different
permittivities and conductivities. Since two phases are
sterically in series, the M-W polarization could be equivalent
to two series of impedances which consist of a resistor in
parallel to a capacitor, respectively (see inset of Fig. 4).
Hence, the relaxation behavior of the M-W polarization is
actually due to the frequency dependence of resistance and
capacitance of the MWCNT/PVDF composite in response to
the a.c. electrical signal 21.
Based on above discussion, the total impedance can be
expressed in the following form 24,
Z?(?) =
1
?1
?1+???1
+ R2 (1)
However, the electrical response (imaginary – real impedance
plot) of the composites shows a semicircle profile, which
could be illustrated using a modified equation 25,
?(?) = ?2 +
?1
1 + (???)? (2)
Here R2 and R1 are the electrical resistances of MWCNT and
PVDF, respectively, ? is the effective time constant
corresponding to the maximum in the ?
l distribution
( ?
l=R1C1~??0/?, where ?0 is the permittivity of free space, ?
and ? are the local permittivity and the local conductivity of
the PVDF, respectively. ?
l is the Maxwell relaxation time in
the MWCNT-PVDF structure. Owing to the random
distribution of MWCNT throughout the PVDF matrix, the
parameters R1 and C1 are random quantities as well, therefore
?l is also a random distribution. For simplicity, ? corresponds
to the maximum in the ?
l distribution. ?=2?/fmax, here fmax is
the frequency at which Z”(f) shows the maximum.) The
parameter ? illustrates the deviation from the ideal semicircle
profile, i.e., the Z’-Z” semicircle moves downward with
respect to the real resistance axis.
Furthermore, the experimental real and imaginary
impedances could be obtained using the following equations
26,
?? (?)
= ?2 + ?1
1 + (??)? sin (
?
2
(1 ? ?))
1 + 2(??)? sin (
?
2
(1 ? ?)) + (??)2?
(3)
?”(?) = ?1
(??)? cos (
?
2
(1 ? ?))
1 + 2(??)? sin (
?
2
(1 ? ?)) + (??)2?
(4)
Eliminating the ?? term, one can obtain the equation of a
circle 23,
(?? ?
2?2+?1
2
)
2
+ (??? ?
?1
2
tan (
?
2
?))
2
=
(
?1
2
sec (
?
2
?))
2
(5)
Here h=1-?. Then the experimental data can be fitted using
Eq. (5). The results are shown in Fig. 5. The ? values obtained
are 0.57, 0.40, 0.13, and 0.59 for Samples 1 to 4. In the range
of experimental errors, the ? value has a big variance from
sample to sample, indicating that except for the impact of
CNT content, the structure and defects also have impact on
the electrical properties, which make the “polarization time”
distribution (for the M-W polarization) quite different. The
factors affecting the dielectric properties include the main
phase PVDF, filled phase MWCNT, the interface between
the PVDF and MWCNT, and the defects existed during the
sample preparation. For our samples, the MWCNT content is
above the percolation threshold (1.5 wt.% 27), thus the
impact of conductivity would be significant, but for the
surface passivated MWCNT filled composites, the resistivity
has been observed to increase 14. Due to lots of factors
affecting the electrical conductivity of the MWCNT/PVDF
composite, e.g., Bondi et al. did the first-principles densityfunction
theory calculations for the electrical conductivity of
oxygen-deficient tantalum pentoxide and found that the
reduction and oxidation reactions may effectively impact the
ReView by River Valley Technologies IET Nanodielectrics
2018/06/27 15:39:34 IET Review Copy Only 6
4
Fig. 4. Static resistivity as a function of MWCNT content for
different MWCNT/PVDF ratios. The blue dots are
experimental data. The red solid line is guided for eyes.
Fig. 5. Fitted real and imaginary impedances using a
modified Cole-Cole model.
conductivity in terms of the donor activation and deactivation
28 , the semicircles move down the Y-axis, a modified
relationship (Eq. (2)) should be used. The variation of ? or h
indicates that the imhomogeneities, interfaces and defects coexist
in the composites, which affect the real and imaginary
impedances as a function of frequency. In general, the
modified Z’-Z” plot means that the contribution to the
impedance by the two phases, i.e., a conducting phase?an
insulating phase and the interface, in the MWCNT filled
composite, make the Z’ – Z” curve a small departure from the
ideal semicircle, in which the defects or imhomogeneities
might make the composite demonstrate complex “time
constant distribution” (? ~ RC).
For the two-phase system, Gerhardt also developed a
method using complex permittivity or dielectric constant,
complex impedance, complex admittance, complex electric
modulus and dielectric loss or dissipation factor to distinguish
the localized and non-localized conductivities 29,30. The
occurrence of peak of the imaginary part in the real –
imaginary part plot is mathematically associating with a
semicircle (see Fig. 3). The dissipation factor versus
frequency profile at low frequencies is usually associated
with the space charges, or the long-range conductivity, which
has been illustrated in Jonscher’s “university law” of the
dielectric response 21. Thus the semicircle or modified
semicircle that we used to illustrate the impedance behaviors
is actually the same as the description of imaginary part (M”,
tan?, and Z”) as a function of frequency in Gerhardt’s
method.
3.3. Thermal conductivity properties
Figure 6 shows the thermal conductivity of the
composite as a function of MWCNT content in volumetric
fraction. As can be seen from the Fig. 6, the thermal
conductivity exhibits basically a linear relationship with the
MWCNT content. The effective media approximation (EMA)
is usually used to fit the experimental result of thermal
conductivity versus filler content, e.g., Bruggeman 31,
Maxwell Garnett models in dilute ceramic composites 32,
Bruggeman model in binary dispersed composites 33. The
fitted results, however, was quite larger than the experimental
data when using the EMA theory. The reason is probably due
to the interface thermal resistance across the filler and the
matrix. Fortunately, Nan et al. obtained a formula for the high
thermal conductivity MWCNT filled composite for a small
loading of MWCNT (