Poly(vinylidene chloride)-Based Carbon as an Electrode Material for High Power Capacitors with an Aqueous Electrolyte

Article abstract of DOI:10.1149/1.1401084

The specific capacitance (F/g) of a poly(vinylidene chloride) (PVDC)-based electric double-layer capacitor (EDLC) carbon electrode prepared by heat-treatment at only 700°C shows a capacitance as high as 64 F/g at 1 mA output current density, although it s

Full text of DOI:10.1149/1.1401084

Journal of The Electrochemical Society, 148 ͑10͒ A1135-A1140 ͑2001͒
A1135
0013-4651/2001/148͑10͒/A1135/6/$7.00 © The Electrochemical Society, Inc.
Poly„vinylidene chloride…-Based Carbon as an Electrode
Material for High Power Capacitors with an Aqueous Electrolyte
M. Endo,a,z Y. J. Kim,a T. Takeda,a T. Maeda,a T. Hayashi,a K. Koshiba,a H. Hara,a
and M. S. Dresselhausb
aFaculty of Engineering, Shinshu University, Nagano 380-8553, Japan
bDepartment of Physics and Department of Electrical Engineering and Computer Science, Massachusetts
Institute of Technology, Cambridge, Massachusetts 02139, USA
The specific capacitance ͑F/g͒ of a poly͑vinylidene chloride͒ ͑PVDC͒-based electric double-layer capacitor ͑EDLC͒ carbon
electrode prepared by heat-treatment at only 700°C shows a capacitance as high as 64 F/g at 1 mA output current density, although
it shows a specific surface area of only 700 m2/g, smaller than those of other conventional activated carbon fibers and activated
carbons. To elucidate the relationship between the specific capacitance and the pore structure for PVDC-based carbons, from an
EDLC practical applications viewpoint, structural characterization was performed using various techniques. PVDC-based carbon
had sufficient porosity during the carbonization process without any additional activation process for use as an EDLC electrode.
It has been shown that the strongest peak in the pore size distribution is at a diameter around 9 Å. The most probable electrolyte
ion sizes, solvated as the hydrated SO24Ϫ ions in aqueous solution, have been obtained by computer simulation as 9 Å. This range
of electrolyte ion sizes is highly convenient for entering the pores of the present PVDC-based carbons to establish the electric
double layer at the pore wall. Furthermore the evolution of Cl2 and HCl during the carbonization of PVDC creates rich open pores
with nanometer dimensions connecting to the free surface of the particles in the samples. It is concluded that the higher specific
capacitance of PVDC-based carbons, compared to conventional activated carbons in aqueous electronic double-layer capacitors, is
due to their optimal pore size distribution. The present precursor has been shown to be a promising material for high power EDLC
applications.
© 2001 The Electrochemical Society. ͓DOI: 10.1149/1.1401084͔ All rights reserved.
Manuscript submitted November 21, 2000; revised manuscript received June 22, 2001. Available electronically September 4,
2001.
Electric double-layer capacitors ͑EDLCs͒ have been used as a
small scale energy storage device in electronic stationary applica-
tions, such as memory back-up devices and solar batteries with a
semipermanent charge-discharge life cycle.1-5 The EDLC device
consists of polarized anode and cathode electrodes made of acti-
vated carbon, having a large specific surface area, with an aqueous
or organic electrolyte. The capacitance ͑F͒ stored in this system is
described by Eq. 1
commercialized AC activated carbon fiber ͑ACF͒ electrodes have a
lower capacitance than that of carbonized PVDC electrodes, even
though they have much higher SSA. In the present paper, the spe-
cific EDLC properties of PVDC-based carbons are presented and the
nanopore structure is discussed. This result could contribute to em-
phasizing the performance of specific EDLCs with aqueous as well
as organic electrolytes from the viewpoint of the pore design for
practical EDLCs.
␧
0
Experimental
C ϭ
dS
͓1͔
͵
4␲␦
The samples were prepared from PVDC ͑CH2CCl2͒, using vari-
ous heat-treatment temperatures ͑HTT͒ in the range 300-1000°C
with an inert atmosphere of N2 and using an electric furnace for the
heat-treatment. These samples were used for structural analysis and
EDLC measurements. The process of electrode preparation for the
EDLCs was as follows: the PVDC heat-treated at 400°C and 40 wt
% of pristine PVDC as a binder were mixed and solubilized in
tetrahydrofuran ͑THF͒ and then the solvent was removed by drying
at room temperature in air. The samples were subsequently pressed
into the shape of a disk, 12.4 mm in radius and 0.7 mm in thickness,
using a steel-mold-type pressure apparatus operating at 3000
kgf/cm2. In order to get the fully carbonized EDLC electrode, the
pellets were again heat-treated also at 700°C for 1 h.
where ␧ is the dielectric constant of the electrolyte, ␦ is the thick-
0
ness of an electric double layer and, S is the surface area of the
interface, respectively. Materials with granular or fibrous forms of
carbons, having high specific surface area ͑SSA, m2/g͒, have been
used as polarizable electrodes because of their larger working sur-
face area as well as their relatively high electric conductivity. Ex-
tensive investigations to increase the specific capacitance ͑per
weight and per volume͒ have been tried in various ways,6 in order to
expand EDLC applications in the field of hybrid electric vehicles
͑HEV͒, load leveling of electric power, and as promising high power
energy storage devices.
In the present paper, a new type of carbon electrode for an aque-
ous EDLC, that is able to store a high power density, has been
demonstrated. Generally, activated carbon materials have been pre-
pared by an activation process, which includes an oxidizing process
using steam or CO2 and KOH, after the stabilization or carboniza-
tion processing steps. We have used a saran resin of poly͑vinylidene
chloride͒ ͑PVDC, uCH2CCl2Ϫ͒ as the starting material, because it
can be obtained with a porous structure without any additional acti-
vation process, by using a heat-treatment that only yields carboniza-
tion. This material has been applied for use in aqueous EDLCs with
1 M H2SO4 solution. Conventional activated carbons, which have a
SSA as large as 1500-2000 m2/g or more, have been used to form an
EDLC because of the relatively high conductivity of carbon as well
as the high SSA of activated carbons ͑ACs͒. But the conventional
The following measurements were performed to characterize the
structure of the original PVDC-based carbon: thermogravimetric
analysis ͑TGA͒, differential scanning calorimetry ͑DSC͒, X-ray dif-
fraction ͑XRD͒, transmission electron microscopy ͑TEM, JEM-
2010, JEOL͒, SSA measurements, the specific capacitance ͑F/g,
F/cm3͒, and equivalent series resistance measurements. In particular,
the specific capacitance ͑F/g͒ of various kinds of conventional acti-
vated carbons was also measured as a function of the specific sur-
face area in order to compare with the present PVDC-based carbons.
Figure 1a illustrates the experimental cell used for the aqueous
electrochemical double layer capacitor, in which identical electrodes
for the cathode and anode were adopted for the cell configuration.
PVDC-based carbon electrodes were set up in a glass beaker as an
anode and cathode, with 30 wt % sulfuric acid aqueous solution.
Before the capacitance measurement, each electrode was treated un-
der primary vacuum conditions at room temperature for more than
14 h to remove the residual air remaining in the electrode and in the
z E-mail: endo@endomoribu.shinshu-u.ac.jp
Downloaded on 2014-07-11 to IP 132.203.227.63 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A1136
Journal of The Electrochemical Society, 148 ͑10͒ A1135-A1140 ͑2001͒
Figure 1. Cell construction and the method of capacitance measurement; ͑a, left͒ configuration of measurement cell, ͑b right͒ schematic discharge curve of the
electric double-layer capacitor for capacitance measurement.
impregnated cell containing the electrolyte. Platinum plates were
attached mechanically to the electrode samples for the output termi-
nal. Glass paper ͑Oribest͒ was used as a separator. By using a simi-
lar system, various kinds of conventional AC and ACFs have been
measured for their specific capacitance as a function of SSA at 0.9 V
charge voltage and 1 mA discharge current. Also, other various
types of ACs have been evaluated regarding their specific capaci-
tance. To prepare a typical sample, 40 mg of AC/ACF material was
mounted on a Pt plate and then pressed to make enough electric
contact to measure the capacitance. The capacitance measurements
were carried out using a constant voltage charge-constant current
discharge method.
ing HTT, reaching a maximum surface area ͑771.92 m2/g͒ at HTT
800°C, and then decreases again. Lower SSA for samples heat-
treated above 800°C might originate from a more completely devel-
oped carbon structure. That is, the pore formation caused by the
release of the chlorine and hydrogen gas gradually decreases with
increasing heat-treatment temperature. Thus, the SSA also de-
creases, but the average pore diameters show no variation with HTT
and are more or less insensitive to heat-treatment after the gas is
released.
As illustrated in Fig. 4a, the adsorption isotherms show type I
behavior except for the sample heat-treated at 300°C. All the other
samples studied here show isotherm behavior that is well identified
as a microporous-based characteristic, the meso- and macroporosity
being negligible. On the other hand, the sample heat-treated at
Results and Discussion
Figure 2 shows the results of thermogravimetric analysis ͑TGA͒
of PVDC under flowing nitrogen gas. The weight variation of the
sample occurred at the two distinct temperatures of 236.2 and 476°C
͑indicated by arrows͒, and the carbon yield obtained at 1000°C HTT
was about 14.86%. The reasons for the two separate and distinct
regions are expected to be related to the different binding energies of
the C-Cl bond ͑328 kJ/mol͒, which breaks first, and to the subse-
quent breakage of the C-H bond ͑413 kJ/mol͒.7
In the low temperature region of HTT ͑200-300°C͒, the weight
of the carbon residue decreases markedly ͑70% of the total weight
loss͒, due to the release of chlorine (Cl2) and hydrogen chloride
͑HCl͒ molecules. It is thought that the extensive gas release in the
sample induces the formation of an open microporous structure in
the resultant carbonaceous materials. Another weight loss, observed
at relatively high HTT at 476°C, might originate from the loss of
hydrogen, which presumably creates the nanopores, which are ob-
served in the pore characterization studies described later.
Figure 3 shows the XRD profiles of PVDC-based carbons as a
function of HTT from 700 to 1000°C. All samples show a broad
XRD line at d002, indicating a typical nongraphitizable carbon. Two
broad lines are seen in the patterns, near 2␪ Х 24 and 44°, corre-
sponding to the 002 and 10 peaks, respectively. These results show
that the carbons have no three-dimensional graphitic structure, and
the computed apparent crystallite thickness Lc and d002 ͑d-spacing͒
are about 12 and 3.95 Å, respectively. These show that PVDC-based
carbons are recognized as a material of very low crystallinity.
Table I shows a summary of the pore analysis results obtained
from a Brunauer-Emmett-Teller ͑BET͒ study of the samples. The
table demonstrates a tendency that the SSA increases with increas-
Figure 2. The TGA, DTA, and DSC profiles of weight percent vs. tempera-
ture obtained for PVDC.
Downloaded on 2014-07-11 to IP 132.203.227.63 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 148 ͑10͒ A1135-A1140 ͑2001͒
A1137
Figure 3. XRD profiles of PVDC-based carbons heat-treated at 700, 900,
and 1000°C.
300°C shows a somewhat different profile. The point, called the B
point indicating the relative pressure where the adsorbed species
reaches a critical magnitude, shows a difference in the adsorption
rate where it starts to absorb. Figure 4b shows the isotherms con-
verted to a percentage ͑%͒ of pore volume against the maximum
partial pressure. This plot clarifies the difference between the behav-
ior of 300°C HTT samples and the others, where the former evi-
dently shows the isotherm for the 300°C to be closer to type II.
These differences might originate from a porous structure of the
HTT 300°C sample that is not developed enough. The isotherms for
PVDC-based carbons show a remarkably large gas uptake at relative
pressures in the range 0.01-0.3, possibly due to the presence of a
large volume of micropores. It is thought that this adsorption occurs
by a cooperative pore filling mechanism by molecules in a mono-
layer coating of the pore walls.8 The sample heat-treated at 300°C
shows a lower slope for the adsorption isotherm in this pressure
range as compared with those of the other samples.
Figure 5 presents the pore-size distribution ͑PSD͒ of the samples,
which was determined by application of the Cranston-Inkley ͑CI͒
equation.9 These results indicate that the sample heat-treated at
700°C has its highest intensity peak around 9 Å, and another broad
peak is observed around 16 Å in pore diameter, but with a lower
intensity, less than that of any of the other samples. In other words,
the HTT 700°C sample has a microporous structure dominated by
Figure 4. Adsorption isotherms of PVDC-based carbons heat-treated at vari-
ous temperatures: ͑a, top͒ amount of adsorption vs. relative pressure (P/P0),
͑b, bottom͒ pore volume converted into percent vs. relative pressure (P/P0).
pores of around 10 Å, as compared with the other samples. Mea-
surements of the PSD were also tried by means of image analysis of
Table I. The results of BET analysis for SSA, total pore volume, average pore diameter of PVDC-based carbon materials heat-treated to
various temperatures.
HTTa
BET
characteristics
300°C
500°C
700°C
800°C
900°C
1000°C
SSAb ͑cm2/g͒
442.63
0.25
640.66
0.34
617.01
0.33
771.92
0.41
692.64
0.37
659.90
0.35
Total pore volume
͑cm3/g͒
Average pore diameter ͑Å͒
15.62
14.62
14.44
14.69
14.48
14.48
a Heat-treated temperature.
b Specific surface area.
Downloaded on 2014-07-11 to IP 132.203.227.63 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A1138
Journal of The Electrochemical Society, 148 ͑10͒ A1135-A1140 ͑2001͒
Figure 5. Pore size distribution ͑PSD͒ of PVDC-based carbons obtained by
the CI method.
TEM micrographs. The TEM image obtained at 200,000 times mag-
nification (⌬f ϭ 512 nm͒ was converted to a 512 ϫ 512 pixel, 8
bits grayscale image in order to perform the 2D-FFT. The 2D power
spectrum thus obtained showed no sign of anisotropy of the speci-
men. Thus, we can integrate the strength of the 2D power spectrum
around the concentric circles that are centered at the center of the
image to get the 1D integrated spectrum. The 1D integrated spec-
trum has the pore diameter preferentially aligned along the x axis,
and the intensity of each value is evaluated along the y axis, to
obtain the PSD.
Figure 6 shows high-resolution TEM images and the correspond-
ing pore size distributions of PVDC-base carbon samples in com-
parison to the PSD obtained from the adsorption isotherms. The
results show the existence of large numbers of nanopores with di-
ameters less than 1 nm. Figures 6b, d, and f show the PSDs obtained
by image analysis of the TEM pictures, using the corresponding FFT
patterns for each picture.10,11 The x axis denotes the spatial wave-
length, and the y axis denotes the intensity of the power spectrum. In
this figure, it is shown that the pore size is shifted toward larger
sizes with increasing HTT, and that the dominant pore sizes are less
than 1.5 nm for all three cases.
The pore-size distributions obtained by image analysis are quite
consistent with those of the N2 adsorption isotherms. As mentioned
above, it is certain that PVDC-based carbon has a porous structure
with a pore-size distribution that is rather smaller than that of acti-
vated carbons which use only a carbonization step, and which could
be more suitable for EDLC applications.
The EDLC test was discharged at 10 mA of constant current, and
charged at 0.9 V of constant voltage for 5 min. Figure 1b shows a
schematic discharge curve for the experimental cell with an aqueous
electrolyte. The capacitance of the cell was calculated from the time
period, ⌬t, which corresponds to a voltage change, ⌬V, taken from
0.54 to 0.45 V ͑from 60 to 50% of the initial voltage͒ under constant
current discharge conditions. When the discharge starts using con-
stant current, an initial voltage drop ͑IR drop͒ originating from the
internal resistance ͑called the equivalent series resistance, ESR͒ of
the cell was evaluated. To avoid any effect of the internal resistance
on the capacitance, the capacitance measurement was carried out
during the voltage drop ranging from 0.54 to 0.45 V by using the
following equation
Figure 6. TEM photographs and PSD curves obtained by image analysis; ͑a͒
and ͑b͒ PVDC-700°C; ͑c͒ and ͑d͒ PVDC-800°C; ͑e͒ and ͑f͒ PVDC-1000°C.
where C is the capacitance ͑F͒, I is the constant output current ͑A͒,
⌬t is the discharge time ͑s͒, and ⌬V is potential change ͑V͒, respec-
tively.
Figure 7 shows the specific capacitance per unit weight ͑F/g͒ as
a function of the output current density for various PVDC-based
carbon bulk electrodes. The specific capacitance for each sample is
observed to decrease with increasing current density. It is worth
while to note, in particular, that the PVDC-based carbon 1000°C
sample shows an extremely high efficiency for the variation of the
current density. The figure shows that the PVDC-based carbon
700°C showed the largest capacitance of 64 F/g at a low current
density of 1 mA, although the PVDC-800°C sample had a larger
specific surface area than that of PVDC 700°C. The other samples
Ͼ
Ͼ
500
have capacitance values in the order of PVDC-800
900
Ͼ1000. The higher the heat-treatment temperature, the better is the
maintenance of the specific capacitance with increasing current den-
sity. For example, the sample heat-treated at 1000°C shows the best
maintenance of current density at 1 A/cm2, with a value of 34%.
This current density stability seems to be related to the equivalent
series resistance ͑ESR͒ and to the increase of the pore volume, with
a mean pore size of 15 Å in the electrode.
Figure 8 shows the results of ESR after the measurement of the
capacitance. It is shown that the ESR decreases with increasing
heat-treatment temperature.
These PVDC-based carbon electrode samples have a much
smaller specific surface area than those of typical ACF/AC ͑ϳ2000
m2/g͒ ͑Fig. 9͒, but nevertheless show a two times higher capaci-
C ϭ I ϫ ⌬t͒/⌬V
͓2͔
͑
Downloaded on 2014-07-11 to IP 132.203.227.63 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 148 ͑10͒ A1135-A1140 ͑2001͒
A1139
Figure 9. Specific capacitance as a function of specific surface area using
various ACF/AC and PVDC-based carbons.
model based on the fractal theory to obtain the relation between ions
for double-layer capacitance and pore size. The theory in the con-
ventional parallel plate condenser has been applied, and the structure
of the pore which exists in the interior of activated carbon was also
assumed ideally to resemble slits and tubes, etc. However, real pores
are quite different from the ideal structures according to the TEM
observation. Therefore, we tried to clarify the real morphology of
the pores by TEM. From the results of the TEM observation, it was
confirmed that the real pores have a complicated contact surface that
can be expressed with a fractal.10–12 In the conventional model for
the capacitance in the pore, the ideal structures is very applicable,
but the real structure of ACF/AC is very complicated according to
the result of TEM observation. Therefore, we proposed a new model
on the basis of the real structure revealed by TEM.10,12 This model
provides information on the behavior of solvated ions in the pore.
When the solvated ion size is large, the molecule cannot enter the
pore. When the solvated ion size is small, it can go to the deep
section of the continuous pore. The pores exist in the activated car-
bon material in a structure which can be expressed with a fractal,
because the starting material of the activated carbon material is com-
posed of randomly oriented microscopic carbon layers. These partial
imperfections in the carbon layers are fractal and appear in the TEM
image. Especially, PVDC-based carbon material has a shallow pit
and/or pore. According to the experimental results, the size of 9-10
Å was found to give the optimum specific capacitance as electrodes.
Pores of this size are also found by TEM image analysis. Conse-
quently, we propose the size of 9-10 Å as a proper pore size or the
diameter of the curved carbon surface as an electrode for EDLC
based on the results of TEM image and computer simulation. In
other words there are many pores and/or imperfections 9-10 Å in
size and these play a role like an active site for ion storage and
adsorption of solvated ion molecules.
Figure 7. Specific capacitance ͑F/g͒ as a function of output current density
for various PVDC-based carbon samples.
tance. This result can be related to the relative size difference be-
tween the pore and the electrolyte ion. As observed in Fig. 5 and 6,
the capacitance for this sample is dominated by ultramicropores of
less than 10 Å size.
The H2SO4 electrolyte separates into a Hϩ ion and a SO42Ϫ ion in
aqueous solution. The Hϩ ion, with a very small size can easily
participate in a repeated insertion and departure process. For the
case of the SO24Ϫ ion, it was considered as a possible structure of a
hydrated sulfate. There could be various kinds of solvated sulfate
species, such as 6͑SO42Ϫ͑H O͒ ͒ and 12 SO2Ϫ͑H2O͒ . According
͓
͔
12
2
6
4
to a simulation by a semiexperimental calculation, these species
have ion sizes of around 10 Å, indicating that these sulfate ions can
fit into a pore having a typical size of around 10 Å. Such a relative
range in size between the pore and the inserted ion into the pore
seems to be best for getting a high capacitance.
A new consideration of pore structure is required for the problem
about ion size, that is, we start from the idea of a different pore
Figure 9 shows a summary of results for capacitance measure-
ments as a function of SSA for various AC/ACFs and carbonized
PVDCs that were measured under 1 mA charge and discharge cur-
rent. At this point, it is concluded that the PVDC-based electrodes
show a higher capacitance than those of commercial AC/ACFs ma-
terials, which is reflected in the suitable pore size for the H2SO4
aqueous electrolyte, by investigating the relation between the ca-
pacitance and the PSD of the host material.13
Conclusions
In the present paper, a carbonized PVDC electrode, with an op-
timum porous structure, dominated by ultramicropores less than 10
Å in size without using any activation process, was used as the
electrode of an EDLC in a sulfuric acid aqueous solution. This
Figure 8. ESR vs. HTT for PVDC-based carbon.
Downloaded on 2014-07-11 to IP 132.203.227.63 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A1140
Journal of The Electrochemical Society, 148 ͑10͒ A1135-A1140 ͑2001͒
PVDC-based carbon electrode performance showed a high capaci-
tance of 64 F/g, even though the material has a smaller specific
surface area than those of conventional ACF/AC materials. This
result can be explained by the relation between the pore size distri-
bution of the sample and the electrolyte ion size. PVDC-based car-
bons have a favorable pore size distribution for this application,
based on various characterization methods. Besides the lower
equivalent series resistance, a higher efficiency electrode can be
obtained with this material. The present PVDC-based carbons ex-
hibit an especially suitable pore size and pore size distribution with
a higher bulk density that induces a high specific capacitance, which
also can provide various technical merits in device applications. It is
further possible to develop large-scale capacitor production using
the present PVDC-based polarizable carbons.
References
1. M. Endo, in Extended Abstracts of the 24th Biennial Conference on Carbon, p.
522, Charleston, SC, USA, American Carbon Society ͑1999͒.
2. B. E. Conway, J. Electrochem. Soc., 138, 1539 ͑1991͒.
3. S. Trasatti and P. Kurzweil, Platinum Met. Rev., 38, 46 ͑1994͒.
4. M. Endo, Y. Okada, and H. Nakamura, Synth. Met., 34, 739 ͑1989͒.
5. S. Sarangapani, B. V. Tilak, and C.-P. Chen, J. Electrochem. Soc., 143, 3791
͑1996͒.
6. S. T. Mayer, R. W. Pekala, and J. L. Kaschmiter, J. Electrochem. Soc., 140, 446
͑1993͒.
7. T. W. G. Solomons, Organic Chemistry, 5th ed., Chap. 7, John Wiley & Sons, Inc.,
New York ͑1992͒.
8. S. J. Gregg and K. S. W. Sing, Adsorption, Surface Area and Porosity, 2nd ed.,
Chap. 4, Academic Press, London ͑1982͒.
9. R. W. Cranston and F. A. Inkley, Adv. Catal., 9, 143 ͑1957͒.
10. K. Oshida, K. Kogiso, K. Matsubayahi, K. Takeuchi, S. Kobayashi, M. Endo, M. S.
Dresselhaus, and G. Dresselhaus, J. Mater. Res., 10, 2507 ͑1995͒.
11. M. Endo, T. Furuta, F. Minoura, C. Kim, K. Ohsida, G. Dresselhaus, and M. S.
Dresselhaus, Supramol. Sci., 5, 261 ͑1998͒.
Acknowledgments
12. M. Endo, K. Takeuchi, Y. Sasuda, K. Matsubayashi, K. Oshida, and M. S. Dressel-
haus, Electron. Commun. Jpn., Part 2: Electron., 77, 98 ͑1994͒.
13. M. Endo, K. Oshida, K. Kogiso, K. Matsubayashi, K. Takeuchi, S. Kobayashi, and
M. S. Dresselhaus, Mater. Res. Soc. Symp. Proc., 371, 511 ͑1995͒.
Part of the work by M. Endo was supported by a grant-in-aid for
‘‘Research for the Future Program,’’ Nano-carbon, from the Japan
Society for the Promotion of Science.
Downloaded on 2014-07-11 to IP 132.203.227.63 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).