Electric double-layer capacitance of meso/macroporous activated carbon fibers prepared by the blending method: I. Nickel-loaded activated carbon fibers in propylene carbonate solution containing LiClO4 salt

Article abstract of DOI:10.1149/1.1481525

Meso/macroporous activated carbon fibers (ACFs), containing mesopores and macropores in addition to micropores, were prepared from carbonization and steam-activation of the phenolic resin fibers blended with a small amount (0.1 wt %) of an organic nickel complex. Conventional (microporous) ACF as reference sample, mainly composed of micropores, was also prepared from pure phenolic resin fibers without any agent. The electric double-layer capacitance of these ACFs was measured in propylene carbonate containing 1.0 mol dm^-3 LiClO₄ (1.0 M LiClO₄/PC). The correlation between the capacitance and the BET (Brunauer-Emmett-Teller) specific surface area of the microporous ACFs showed the nonlinearity due to strong ion-sieving of micropores. The double-layer capacitance of the meso/macroporous ACFs was higher than the microporous ACF, because the ion-sieving effect of micropores was relaxed by the presence of meso/macropores. The relaxation by the meso/macropores was enhanced for the cation adsorption or the high current density measurement. These results confirm that the presence of many meso/macropores promotes the formation of an effective double layer or fast transfer of ions in the microporous structure.

Full text of DOI:10.1149/1.1481525

Journal of The Electrochemical Society, 149 ͑7͒ A855-A861 ͑2002͒
A855
0013-4651/2002/149͑7͒/A855/7/$7.00 © The Electrochemical Society, Inc.
Electric Double-Layer Capacitance of MesoÕMacroporous
Activated Carbon Fibers Prepared by the Blending Method
I. Nickel-Loaded Activated Carbon Fibers in Propylene Carbonate
Solution Containing LiClO4 Salt
Soshi Shiraishi, ,z Hideyuki Kurihara, Lan Shi, Takuya Nakayama, and Asao Oya
*
Department of Chemistry, Faculty of Engineering, Gunma University, Kiryu, Gunma 376-8515, Japan
Meso/macroporous activated carbon fibers ͑ACFs͒, containing mesopores and macropores in addition to micropores, were pre-
pared from carbonization and steam-activation of the phenolic resin fibers blended with a small amount ͑0.1 wt %͒ of an organic
nickel complex. Conventional ͑microporous͒ ACF as reference sample, mainly composed of micropores, was also prepared from
pure phenolic resin fibers without any agent. The electric double-layer capacitance of these ACFs was measured in propylene
carbonate containing 1.0 mol dmϪ3 LiClO4 (1.0 M LiClO4 /PC). The correlation between the capacitance and the BET ͑Brunauer-
Emmett-Teller͒ specific surface area of the microporous ACFs showed the nonlinearity due to strong ion-sieving of micropores.
The double-layer capacitance of the meso/macroporous ACFs was higher than the microporous ACF, because the ion-sieving
effect of micropores was relaxed by the presence of meso/macropores. The relaxation by the meso/macropores was enhanced for
the cation adsorption or the high current density measurement. These results confirm that the presence of many meso/macropores
promotes the formation of an effective double layer or fast transfer of ions in the microporous structure.
© 2002 The Electrochemical Society. ͓DOI: 10.1149/1.1481525͔ All rights reserved.
Manuscript submitted September 21, 2001; revised manuscript received January 25, 2002. Available electronically May 14, 2002.
1,2
Recently, Electric double-layer capacitors ͑EDLCs͒ have been
proposed as the subpower source for hybrid electric vehicles
͑HEVs͒ because of their higher power density ͑larger than 1000 W
kgϪ1͒, high cycle-efficiency, and long lifetime. However, since the
energy density of the EDLC is only several Wh kgϪ1 and therefore
much lower than that of rechargeable batteries, the capacitance of
EDLCs has to be increased. The practical electrode material for the
EDLC is porous carbon such as activated carbons. The high capaci-
tance ͑100-200 F gϪ1͒ is due to high specific surface area ͑Ͼ1000
the large pores on the double-layer capacitance, because the conven-
tional ACF prepared without an activation catalyst has few meso/
macropores.
In our previous work,19,20 the double-layer capacitance of the
ACFs with meso/macropores ͑referred to as meso/macroporous
ACF͒ was compared with that of the conventional ACFs ͑referred to
as microporous ACF͒. As a result, it was revealed that the double-
layer capacitance of ACFs was enhanced by the presence of meso/
macropores. However, some significant properties of the meso/
macroporous ACFs ͑e.g., such as the influence of activation catalysts
on the capacitance, the difference in the capacitance between cation
and anion, and the rate property͒ have not been reported yet. In this
paper, the properties of the double-layer capacitance for the meso/
macroporous ACFs are discussed in detail in order to confirm the
effect of meso/macropores on the double-layer capacitance.
3
m2 gϪ1͒ due to many micropores (2 nm
pore width ).
Ͼ
In general, it is believed that there is a proportional correlation
between the specific surface area and the EDLC of activated car-
bons, as based on the following equation1,4,5
␧ ␧
0
r
C ϭ
dS
͓1͔
͵
␦
Experimental
Preparation of meso/macroporous ACFs.—It is well known that
activated carbons containing mesopores or macropores can be ob-
tained through the activation process of carbon materials under the
presence of an activation metal catalyst.17,18 There are two ways for
loading the catalyst in the pure carbon matrix: one is the impregna-
tion method, in which the activation catalyst is impregnated by the
immersion of carbon materials in the solution containing the catalyst
chemicals. Another one is known as the blending method,17,21 indi-
cating the mixing of carbon precursor and catalyst precursor before
carbonization. The latter has the advantage that micropores in acti-
vated carbons are not closed with the loaded catalyst. In the present
study, the characteristic ACFs with meso/macropores were prepared
by a blending method using nickel as the activation catalyst. The
detailed procedure is described in previous papers.19,20 Novolak-
type phenolic resin ͑obtained from Gun-Ei Chemical Industry Co.,
Ltd., Japan͒ and bis͑2.4-pentanedionate͒nickel͑II͒ dihydrate ͑nickel
acetylacetonate dihydrate, (CH3COCHCOCH3)2Ni•2H2O, Wako
Chemical, Japan͒ were used as the carbon precursor and the catalyst
precursor for steam activation, respectively. The percentage of
nickel in the raw materials was 0.1 wt %. The samples containing Ni
species or without an agent were referred to as 0.1 Ni-ACF ͓activa-
tion duration ͑min͔͒ or Ref-ACF ͓activation duration ͑min͔͒, respec-
tively.
where C is specific capacitance, ␧ is permittivity in vacuum, ␧ is
0
r
relative permittivity of the double layer, ␦ is thickness of the double
layer, and S is surface area. This equation is derived from an ideal
capacitor consisting of a solid dielectric layer between two parallel-
plate electrodes such as practical ceramic or film capacitors ͑con-
denser͒. However, some researchers have reported nonlinearity of
the double-layer capacitance on surface area.6-14 This may be due
to the dimension of ion or solvent in electrolyte and the pore
size distribution ͑PSD͒ of activated carbons. In particular, meso-
3
Ͻ
Ͻ
pores (2 nm
pore width
50 nm ) or macropores (pore width
3
Ͼ 50 nm ) should not be ignored because they can influence the
permeation of electrolyte to microporous structure or the mobility of
ions in the pore.
In the meantime, activated carbon fiber ͑ACF͒ is known to have
a sharp PSD of micropores without larger pores such as mesopores
or macropores.15,16 Generally, ACF is manufactured by carboniza-
tion and gas activation of carbon precursor such as phenolic resin
fiber, pitch fiber, and polyacrylonitrile fiber. However, uncommon
ACFs containing many mesopores ͑or macropores͒ were recently
prepared by blending an activation catalyst with phenolic resin17 or
isotropic pitch.18 These characteristic ACFs with mesopores and
macropores are a suitable material for investigating the influence of
An acid treatment of 0.1 Ni-ACFs was conducted to discuss the
influence of nickel species in 0.1 Ni-ACFs on carbon matrix, pore
structure, or double-layer capacitance. In this paper, 1.0 mol dmϪ3
H2SO4 aqueous solution was used as the acid. The 0.1 Ni-ACFs
* Electrochemical Society Active Member.
z E-mail: ssiraisi@chem.gunma-u.ac.jp
Downloaded on 2014-10-29 to IP 128.248.155.225 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A856
Journal of The Electrochemical Society, 149 ͑7͒ A855-A861 ͑2002͒
were immersed in 1.0 M H2SO4 aq for 12 h. The samples treated
with the acid were washed twice in pure water to eliminate residual
H2SO4 and dried at 60°C for 24 h.
Characterization of ACFs.—X-ray diffraction ͑XRD, RINT
2100V/PC, Rigaku, Japan͒ analysis was used to investigate the car-
bon matrix structure of the ACFs and the chemical state of nickel
in 0.1 Ni-ACFs. The N2 adsorption/desorption measurement at 77
K was conducted using an N2 adsorption/desorption system
͑BELSORP28SA, BEL Japan, Inc.͒ to characterize the pore struc-
ture of the ACFs. The Brunauer-Emmett-Teller specific surface area
͑BET-SSA͒ was calculated from the BET plot of the N2 adsorption
isotherm in the low relative pressure between 0 and 0.05 to prevent
the overestimation of surface area due to micropore filling effect.22
The PSD of mesopores was obtained by the Dolimore-Heal ͑DH͒
method using the adsorption branch.23 In the DH method, morphol-
ogy of mesopore is considered to be cylindrical shape. The average
micropore width was estimated by the Dubinin-Radushevich ͑DR͒
equation.24,25 The surface morphology or the surface composition of
ACFs was observed with scanning electron microscopy ͑SEM, JSM-
5300, JEOL, Japan͒ or X-ray photoelectron spectroscopy ͑XPS,
ESCA5600, Ulvac-Phi, Inc., Japan͒, respectively.
Measurement of electric double-layer capacitance of ACFs.—A
composite pellet electrode was prepared from ground ACF ͑0.05 g͒,
acetylene black ͑BET-SSA: 53 m2 gϪ1, Denki Kagaku Kogyo Co.,
Japan͒, and polytetrafluoroethylene binder ͑PTFE 6J, DuPont-Mitsui
Fluorochemicals Co., Ltd., Japan͒. The ratio of ACF ͑dried at 200°C
for 2 h under ϳ1 Pa͒, acetylene black, and binder in the electrode
was 86:10:4 wt %, respectively. Ti mesh was used as the current
collector for the composite electrode. A standard three-electrode cell
was used to measure the double-layer capacitance of a single ACF
electrode. The counter electrode was lithium metal foil. The refer-
ence electrode was lithium metal connected with nickel wire. The
measurement of the double-layer capacitance was conducted under
galvanostatic conditions in propylene carbonate ͑PC͒ containing 1.0
mol dmϪ3 LiClO4 electrolyte ͑1.0 M LiClO4 /PC, Kishida Chemical
Co., Ltd., Japan͒. The upper limited potential for the measurement
was 4.0 V vs. Li/Liϩ, and the lower was 2.0 V vs. Li/Liϩ. The
concentration of water as an impurity in the electrolyte was esti-
mated to be Ͻ30 ppm by a Karl-Fischer moisture titrator ͑MKC-
510N, Kyoto Denshi Kogyo, Japan͒. In this paper, the process of
electrons passing from ͑to͒ the carbon electrode is referred to as a
positive ͑negative͒ process. The double-layer capacitance was cal-
culated by the amount of electricity passed during a positive or
negative process according to the following equation
Figure 1. XRDs of ͑a͒ Ref-ACFs and ͑b͒ 0.1 Ni-ACFs. The sample
‘‘0.1 Ni-ACF(60)acid’’ or ‘‘0.1 Ni-ACF(120)acid’’ is the 0.1 Ni-ACF͑60͒ or
0.1 Ni-ACF͑120͒ after immersion in 1.0 M H2SO4aq for 12 h.
observed for these 0.1 Ni-ACFs. When the activation duration was
longer, the peak intensity of metallic nickel became larger and the
full width at half maximum ͑fwhm͒ of metallic nickel became
shaper. Applying the Scherrer equation to the 111 line of metallic
nickel in 0.1 Ni-ACF͑10͒, 0.1 Ni-ACF͑120͒, or 0.1 Ni-ACF͑180͒,
the crystallite size was estimated to be about 39, 51, or 62 nm,
respectively. These suggested that the metallic nickel particles grew
during the activation process. The two XRD patterns in the top of
Fig. 1b were obtained from the 0.1 Ni-ACF͑60͒ and the 0.1 Ni-
ACF͑120͒ after the immersion process in 1 M H2SO4 aq ͑the sample
names are 0.1 Ni-ACF(60)acid and 0.1 Ni-ACF(120)acid͒. No nickel
peaks were observed in these XRD patterns. Therefore, it can be
said that the loaded nickel particles in 0.1 Ni-ACFs are exposed on
the surface of the carbon matrix.
C ϭ it/m⌬V
͓2͔
where C is the electric double-layer capacitance ͑F gϪ1͒, i current
͑A͒, t the positive ͑or negative͒ duration ͑s͒, m the weight of ACF in
the electrode ͑g͒, and ⌬V is the potential change during the positive
͑or negative͒ process ͑V͒ ͑2 V in this study͒. The capacitance cal-
culated by Eq. 2 corresponds to the specific gravimetric electric
double-layer capacitance of the ACFs.
Results and Discussion
Matrix condition of ACFs.—Figure 1a and b shows the XRD
patterns of Ref-ACFs and 0.1 Ni-ACFs, respectively. In the case of
Ref-ACFs, all patterns were typical of amorphous carbon, composed
of a broad 002 line and 10 line attributed to micrographite structure.
The intensive background around 2␪ ϭ 10-15° was concerned with
the small-angle X-ray scattering derived from the presence of many
micropores in the carbon matrix. The X-ray diffraction ͑XRD͒ pat-
terns of 0.1 Ni-ACFs showed three metallic nickel lines ͑111, 200,
and 220͒, in addition to the same broadened lines of amorphous
carbon as Ref-ACF. This result indicates that the elemental nickel in
0.1 Ni-ACFs was present as metallic nickel particles in the amor-
phous carbon structure. Catalytic graphitization by nickel was not
Pore structure and surface of ACF.—The N2 adsorption/
desorption isotherms of Ref-ACFs and 0.1 Ni-ACFs are shown in
Fig. 2. In all isotherms of the Ref-ACFs, a large amount of N2 was
remarkably adsorbed at low relative pressure ͑Ͻ0.1 P/Ps͒, while the
N2 adsorption was saturated at middle and high P/Ps. Therefore, the
adsorption behavior of the Ref-ACFs is regarded as a typical Type I
isotherm for microporous carbon. The drastic increment in adsorp-
tion at the low relative pressure is due to the micropore-filling
effect.3,22 The Ref-ACF prepared with longer activation has a larger
amount of N2 adsorption. This shows that the microporous structure
in ACF is developed with the activation duration. In the case of 0.1
Downloaded on 2014-10-29 to IP 128.248.155.225 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, 149 ͑7͒ A855-A861 ͑2002͒
A857
Figure 2. N2 adsorption and desorption
isotherms at 77 K for ͑a͒ Ref-ACFs and
͑b͒ 0.1 Ni-ACFs: ͑᭹, ᭡, ࡗ, ᭿, ᭹͒ ad-
sorption isotherms, ͑᭺, ᭝, छ, ᮀ, ᭺͒ de-
sorption isotherms.
mesopore volume compared to Ref-ACF with almost the same BET-
SSA. The micropore volume (Vmicro) and the average micropore
width (wmicro) estimated by the DR equation22,23 are also summa-
rized in Table I. Vmicro and Wmicro became larger with longer activa-
tion duration, regardless of the Ref-ACF series and 0.1 Ni ACF
series. There was little difference between Vmicro and wmicro between
Ref-ACF and 0.1 Ni-ACF with almost identical BET surface area
͓for example, Ref-ACF͑120͒ vs. 0.1 Ni-ACF͑120͒, and Ref-
ACF͑480͒ vs. 0.1 Ni-ACF͑180͔͒. These results suggested that the
surface area of mesopores in 0.1 Ni-ACFs does not contribute sig-
nificantly to the total surface area, although the mesopore volume
of 0.1 Ni-ACFs is significant in the total pore volume (Vmeso
ϩ Vmicro). Consequently, 0.1 Ni-ACFs can be considered to be
characteristic ACFs containing both numerous mesopores and mi-
cropores, and the large specific surface area of 0.1 Ni-ACFs is due
to the developed micropore structure as well as Ref-ACFs.
The BET surface area, the pore volume, and the micropore width
of 0.1 Ni-ACF(60)acid or 0.1 Ni-ACF(120)acid are also shown in
Table I. The mesopore volume of these 0.1 Ni-ACFs was increased
slightly by the immersion in the acid. This increment in the meso-
pore volume can be due to the removal of the metallic nickel par-
ticles in the 0.1 Ni-ACFs that contribute to blocking the entrance of
some pores. In contrast, the BET surface area, Vmicro , and wmicro
were slightly smaller than before the acid treatment. These suggest
some contraction of micropore structure by the acid treatment. How-
Ni-ACFs, drastic adsorption of N2 was observed at low relative
pressure as well as the case in Ref-ACFs. However, the amount of
adsorbed N2 of 0.1 Ni-ACFs gradually increased even in the region
of middle P/Ps and markedly increased again in the region of high
P/Ps, such as Ͼ0.8 P/Ps. The adsorption increasing in the region of
middle and high P/Ps was observed more distinctly in the 0.1 Ni-
ACFs prepared with longer activation duration. This adsorption be-
havior can be attributed to the capillary condensation of N2 in the
mesopores ͑or macropores͒ and multilayer adsorption on the meso-
pores or the macropores. The PSD of the mesopores for 0.1 Ni-
ACFs and Ref-ACFs calculated by the DH method are shown in Fig.
3. The former has few mesopores except for small pore size ͑ϳ2
nm͒ closed to micropores, while the latter has great pore volume in
the whole mesopore region. The amount of mesopore volume in the
range 10-40 nm pores became much larger in the Ni-ACF prepared
with longer activation. This indicates that the nickel catalyst is ef-
fective in the formation of mesopores with relatively large pore size.
The BET-SSA ͑corresponding to the total SSA of ACFs͒ and the
DH pore volume ͑Vmeso , corresponding to the total pore volume of
mesopores in the ACFs͒ are summarized in Table I. In both Ref-
ACFs and 0.1 Ni-ACFs, the BET-SSA was higher as the activation
duration was extended. This means that the pore structure in the
ACFs is well developed with longer activation time. The mesopore
volume in Ref-ACFs was increased slightly through the activation,
while the mesopore volume in 0.1 Ni-ACFs was markedly changed
by the activation from 0.10 to 0.86 mL gϪ1. 0.1 Ni-ACF had larger
Figure 3. PSDs for the mesopore region
of ͑a͒ Ref-ACFs and ͑b͒ 0.1 Ni-
ACFs͑180͒, calculated by the DH
method. Rp is pore radius, Vp pore vol-
ume.
Downloaded on 2014-10-29 to IP 128.248.155.225 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A858
Journal of The Electrochemical Society, 149 ͑7͒ A855-A861 ͑2002͒
Table I. BET-SSA, mesopore volume, average micropore width,
and micropore volume of various ACFs.a
Table II. Yield of various ACFs, bulk density of composite elec-
trode, and surface oxygenÕcarbon atomic ratio of various ACFs.
Sampleb
BET-SSA
Sample
Yield ͑%͒
d a ͑g cmϪ3
͒
O/Cb
Vmeso
Vmicro
wmicro ͑nm͒
͑m2 gϪ1
͒
͑mL gϪ1
͒
͑mL gϪ1
͒
Ref-ACF͑10͒
Ref-ACF͑60͒
52
45
41
35
23
47
29
16
15
5
1.00
0.87
0.85
0.74
0.68
0.85
0.62
0.54
0.43
0.38
0.07
0.08
0.06
0.06
0.06
0.07
0.05
0.04
0.04
0.05
Ref-ACF͑10͒
Ref-ACF͑60͒
650
930
1150
1480
1780
710
960
920
1050
1000
1290
1660
0.06
0.03
0.08
0.09
0.16
0.10
0.21
0.23
0.30
0.32
0.35
0.86
0.28
0.37
0.47
0.60
0.73
0.28
0.38
0.36
0.42
0.39
0.52
0.69
0.64
0.73
0.85
0.94
1.07
0.69
0.74
0.72
0.83
0.79
0.88
1.06
Ref-ACF͑120͒
Ref-ACF͑240͒
Ref-ACF͑480͒
0.1 Ni-ACF͑10͒
0.1 Ni-ACF͑60͒
0.1 Ni-ACF͑120͒
0.1 Ni-ACF͑140͒
0.1 Ni-ACF͑180͒
Ref-ACF͑120͒
Ref-ACF͑240͒
Ref-ACF͑480͒
0.1 Ni-ACF͑10͒
0.1 Ni-ACF͑60͒
0.1 Ni-ACF(60)acid
0.1 Ni-ACF͑120͒
0.1 Ni-ACF(120)acid
0.1 Ni-ACF͑140͒
0.1 Ni-ACF͑180͒
c
c
a d: Bulk density of the composite electrode composed of ACF, acety-
lene black, and binder ͑86:10:4 wt %͒.
b O/C: Surface oxygen/carbon atomic ratio estimated by XPS.
a BET-SSA calculated by BET plot in the region of 0-0.05 P/Ps.
w
V
V
micro : Average micropore width calculated from DR method.
micro : Pore volume of micropore calculated from DR method.
meso : Pore volume of mesopore ͑1.6-40 nm͒ calculated from DH
0.1 Ni-ACF were lower than those of Ref-ACF with almost the
same BET-SSA. The lower yield and the lower bulk density are due
to the catalytic activation ͑gasification͒ of nickel. This is in a good
agreement with the results of pore analysis by N2 adsorption and
SEM observation.
method of adsorption branch.
b Number in parenthesis indicates the activation duration ͑min͒.
c The 0.1 Ni-ACFs after the immersion in 1.0 M H2SO4 aq for 12 h.
The surface oxygen/carbon atomic ratios of the ACFs, estimated
by the XPS analysis, are also shown in Table II. These results re-
vealed little difference in surface oxygen content between Ref-ACF
and 0.1 Ni-ACF. The XPS O 1s spectra and the C 1s spectra in
Ref-ACF were similar to those in 0.1 Ni-ACFs ͑these spectra are not
shown in this paper͒, which also indicates little difference in the
kinds of surface functionalities between Ref-ACFs and 0.1 Ni-
ACFs. Therefore, these facts suggest that the effect of surface func-
tionalities on double-layer capacitance does not need to be consid-
ered in this paper.
ever, since these changes in the pore structure before and after the
acid treatment are essentially small, it can be said that the removal
process of metallic nickel particles essentially does not influence the
pore structure of the 0.1 Ni-ACFs.
Figure 4 is the SEM images of Ref-ACF͑480͒, 0.1 Ni-ACF͑60͒,
and 0.1 Ni-ACF͑180͒. Many macropores with pore size smaller than
1 ␮m and many mesopores with relatively large pore size were
observed in these 0.1 Ni-ACFs, especially 0.1 Ni-ACF͑180͒. No
macropores or no mesopores were observed in the Ref-ACF͑480͒.
Therefore, it can be seen that the steam activation under nickel
species also produces many macropores in addition to mesopores in
ACF.
Electric double-layer capacitance of ACF.—Figure 5a and b
shows the potential-time curves ͑chronopotentiograms͒ for Ref-
ACFs and 0.1 Ni-ACFs in 1.0 M LiClO4 /PC electrolyte. The elec-
trode potential of Ref-ACF͑240͒, Ref-ACF͑480͒, 0.1 Ni-ACF͑120͒,
0.1 Ni-ACF͑140͒, and 0.1 Ni-ACF͑180͒ was changed linearly by the
amount of electricity passed for the positive or negative process.
Therefore, these ACF electrodes are regarded as a polarizable
electrode with a constant capacitance. Any plateaus derived from
electrochemical reactions such as nickel dissolution or oxidation
were not observed in the curves of these 0.1 Ni-ACFs. Moreover,
the chronopotentiograms of 0.1Ni-ACF(60)acid and 0.1
Ni-ACF(120)acid in Fig. 5c were similar to those of 0.1 Ni-ACF͑60͒
and 0.1 Ni-ACF͑120͒, so it can be seen that the metallic nickel
particles loaded on 0.1 Ni-ACFs do not affect the double-layer ca-
pacitance. On the other hand, the profiles of Ref-ACF͑60͒, Ref-
ACF͑120͒, and 0.1 Ni-ACF͑60͒ were distorted triangles with a vari-
able gradient. Since the slope of the curve corresponds to the
capacitance, it follows that the double-layer capacitance of these
ill-activated ACF electrodes depends on the electrode potential dur-
ing the positive or negative process. The dependence of the capaci-
tance on the electrode potential may be due to the difference in the
capacitance between the anion adsorption/desorption process and
the cation adsorption/desorption process. The slopes of the potential-
time curves of Ref-ACF͑10͒ and 0.1 Ni-ACF͑10͒ were very steep,
so the capacitance of these ACFs is very small. The capacitance was
estimated by Eq. 2 to determine the relationship between the BET-
SSA and the double-layer capacitance in detail.
The yield and the bulk density of electrodes for Ref-ACFs and
0.1 Ni-ACFs are shown in Table II. The yield and the electrode bulk
density became smaller with longer activation duration in both Ref-
ACFs and 0.1 Ni-ACFs. The yield and the electrode bulk density of
Figure 6a is the correlation between the BET-SSA and the elec-
tric double-layer capacitance in 1.0 M LiClO4 /PC, obtained from
the positive process. This correlation was very similar to that ob-
tained from the same ACFs in PC solution containing tetraethyl
ammonium salt.24,25 The capacitance of each ACF electrode during
the positive process was nearly the same as that obtained from the
negative process, which means that the coulombic efficiency was
Figure 4. SEMs of ͑a͒ 0.1 Ni-ACF͑10͒, ͑b͒ 0.1 Ni-ACF͑60͒, ͑c͒ 0.1 Ni-
ACF͑180͒, or ͑d͒ Ref-ACF͑480͒.
Downloaded on 2014-10-29 to IP 128.248.155.225 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, 149 ͑7͒ A855-A861 ͑2002͒
A859
limiting molar conductivity ͑S cm2 molϪ1͒, ␩ is solvent viscosity
͑cP͒, and x is 4 in the case of ClOϪ4 , 5.7 in the case of Liϩ.27
With Eq. 2 and the single ion limiting molar conductivities of
Liϩ and ClO4Ϫ in PC,27 the Stokes radius of Liϩ cation or ClO4Ϫ
anion in PC solution was estimated to be 0.41 or 0.26 nm. Thus the
solvated Liϩ cation size ͑0.82 nm͒ or ClOϪ4 anion size ͑0.52 nm͒ is
very close to the micropore width of Ref-ACF͑10͒ or 0.1 Ni-
ACF͑10͒. Therefore, if the Stokes radius can be applied to the dis-
cussion about the ions in micropores, ‘‘ion-sieving effect’’ of narrow
micropores, something like ‘‘molecular sieving,’’ is a good explana-
tion of very small capacitance of the ill-activated ACFs ͑the nonlin-
earity of capacitance to surface area͒. Soffer’s group also reported
the ion-sieve phenomenon of micropores in ACF.14
In the case of the better-activated ACFs with high surface area,
the double-layer capacitance of 0.1 Ni-ACFs was higher than that of
Ref-ACFs with almost the same BET-SSA. Particularly, the higher
capacitance of 0.1 Ni-ACFs was prominent in the region of 900-
1200 m2gϪ1 BET-SSA ͓e.g., Ref-ACF͑60͒ vs. 0.1 Ni-ACF͑60͒, or
Ref-ACF͑120͒ vs. 0.1 Ni-ACF͑120͔͒. In the previous section, it has
already been revealed that the mesopores in the 0.1 Ni-ACF series
were developed better than those in the Ref-ACFs, although there
was little difference in the micropore structure for both series of
samples from the viewpoint of the average micropore width and the
micropore volume. Consequently, this advantage of the capacitance
of 0.1 N-ACFs can be attributed to the presence of many mesopores
͑or macropore͒, which assists the transfer of ions through the mi-
croporous structure or the permeation of nonaqueous electrolyte into
the pores to form the electric double layer on the pore surface.
Furthermore, these mesopores effects can be expected to be weaker
in ACFs with sufficiently large micropores to ions. In fact, Fig. 6a
indicates that the advantage of the mesopores for the capacitance
was not so effective in the ACFs with large micropore width ͓e.g.,
Ref-ACF͑480͒ and 0.1 Ni-ACF͑180͔͒ as that of the ACFs with 900-
1200 m2 gϪ1 BET-SSA. This result also confirms the ion sieving of
micropores and the mesopore effect.
The double-layer capacitance seems to be saturated in the high
surface area such as higher than 1200 m2 gϪ1 for both Ref-ACFs
and 0.1 Ni-ACFs. It cannot be explained with the ion-sieving effect
because the micropore width is large enough for the ions to move in
the ACFs. One of the reasons for the saturation may be the overes-
timation of SSA with the BET model for activated carbons with high
surface area.22,28 However, the possibility of the overestimation
should be avoided using the BET plots at low relative pressure,22 as
described in the experimental section. Therefore, the capacitance
saturation of the ACFs with high surface area is caused by another
factor. The variation of the surface carbon structures by the
activation29 ͑e.g., such as the ratio of the edge plane and the basal
plane of micrographite͒ may also affect the correlation between the
capacitance and the surface area. Randin and Yeager already re-
ported the difference in the specific capacitance per surface area
between the edge plane and the basal plane of highly oriented py-
rolytic graphite ͑HOPG͒.30
Figure 5. Potential-time curves ͑chronopotentiograms͒ for ͑a͒ Ref-ACFs,
͑b͒ 0.1 Ni-ACFs, and ͑c͒ 0.1 Ni-ACFsacid in 1.0 mol dmϪ3 LiClO4 /PC.
Galvanostatic method ͑40 mA gϪ1͒, potential range 2-4 V vs. Li/Liϩ.
almost 100%. The correlation for 0.1 Ni-ACFsacid was also plotted
in Fig. 6a. These data confirm that the loaded nickel particles in 0.1
Ni-ACFs do not influence the double-layer capacitance.
Cation capacitance and anion capacitance of ACF.—In the pre-
vious section, it was seen that the pore size distribution of ACFs
effectively affects the double-layer capacitance in 1.0 M LiClO4 /PC
electrolyte. Based on the comparison between the pore size and the
solvated ion size, the difference in capacitance between cation and
anion should be also observed if the cation size is different from the
anion. The solvated Liϩ cation ͑0.82 nm͒ is larger than the solvated
ClOϪ4 ͑0.52 nm͒, so it can be expected that the advantage of 0.1
Ni-ACFs for the anion adsorption/desorption capacitance is weaker
than that for cation adsorption/desorption. The evaluation of each
adsorption capacitance can be conducted by separation of chronopo-
tentiograms at the initial potential before the measurement ͑namely,
pzc: the immersion potential of electrode31͒ as shown in Fig. 7. The
initial potential of each ACF electrode was between 3.0-3.1 V vs.
Li/Liϩ. The chronopotentiogram in the upper potential region to the
initial potential reflects the capacitance of the anion adsorption/
Both the correlations for Ref-ACFs and 0.1 Ni-ACFs revealed
that the double-layer capacitance was not linearly proportional to the
BET-SSA. Since these ill-activated ACFs ͓e.g., Ref-ACF͑10͒ or 0.1
Ni-ACF͑10͔͒ have narrower micropores, Ͻ0.7 nm pore width, the
very small capacitance of these ACFs can be attributed to the diffi-
culty in forming an effective double layer or the low mobility of the
ion in the narrow micropores. The Stokes radius is well known as an
index of the solvated ion size. The Stokes radius of Liϩ cation or
ClOϪ4 anion in PC can be estimated with the following equation26,27
z2i F2
0.492 ϫ 102z2i
rs ϭ
ϭ
͓3͔
x␲NA␭ ␩
x␭ ␩
0
0
where rs is the Stokes radius ͑nm͒, zi is the charge number of ion, F
the Faraday constant, NA Avogadro’s constant, ␭ the single ion
0
Downloaded on 2014-10-29 to IP 128.248.155.225 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A860
Journal of The Electrochemical Society, 149 ͑7͒ A855-A861 ͑2002͒
Figure 6. Correlation between BET-SSA and electric double-layer capacitance during positive process for Ref-ACFs and 0.1 Ni-ACFs in 1.0 mol dmϪ3
LiClO4 /PC. Galvanostatic method ͑40 mA gϪ1͒, positive process 2 V → 4 V vs. Li/Liϩ. ͑ᮀ͒ Obtained from 0.1Ni-ACFsacid
.
desorption, since the ACF electrode must compensate the lack of
electrons with negative charge of anion in electrolyte. Similarly, the
lower potential region to the initial potential corresponds to the ca-
pacitance of the cation adsorption/desorption because of the excess
electrons. These suggest that each capacitance shown in Fig. 6a is an
average capacitance of cation desorption and anion adsorption.
Thus, the cation adsorption capacitance was calculated from the
potential-time curve at potentials lower than 3 V vs. Li/Liϩ during
the negative process. The anion adsorption capacitance was also
evaluated from the curve at potentials higher than 3 V vs. Li/Liϩ
during the positive process. These capacitances are shown in Fig. 6b
and c, respectively. These results revealed that the cation capaci-
tance of Ref-ACFs was smaller than the anion capacitance, and that
the advantage of 0.1 Ni-ACF in the capacitance was more prominent
for the cation adsorption capacitance than the anion adsorption.
These results suggest the following two points. First, the solvated
Liϩ cation cannot go easily through microporous structure compared
with the ClOϪ4 anion. Second, the mesopores can assist Liϩ cation
Rate property of electric double-layer capacitance of ACF.—As
mentioned in the introduction, one of the merits of EDLC is high
power density, so the rate property of the double-layer capacitance is
significant. Figure 8 is the dependence of the double-layer capaci-
tance on the current density for Ref-ACFs and 0.1 Ni-ACFs. Since
higher current density causes more serious IR drop derived from
electrode resistance, each capacitance in Fig. 8 was estimated from
the chronopotentiograms only in the region of 2.25-3.75 V vs.
Li/Liϩ to eliminate the influence of the IR drop on the capacitance
evaluation. The double-layer capacitance of 0.1 Ni-ACFs during the
negative process decreased slightly with high current density, while
that of Ref-ACFs reduced markedly at higher current density except
for Ref-ACF ͑240͒. These results suggest that the ions in the mi-
croporous structure with meso/macropores can follow the large elec-
tric perturbation such as high current density. According to Morita’s
results,32 one kind of mesoporous ACF also showed a good rate
property in EC/DMC ͑ethylene carbonate-dimethyl carbonate͒ mix-
ture solution containing LiBF4 . Thus, meso/macroporous ACFs can
be considered as a promising material for high-power EDLCs. How-
ever, the bulk electrode density of 0.1 Ni-ACF is considerably
smaller than that of Ref-ACFs, and the yield of 0.1 Ni-ACFs is also
lower. Therefore, the advantage of 0.1 Ni-ACFs should be consid-
better to move in microporous structure than ClOϪ4 anion. Therefore,
the ion sieving of micropores or the assistance of mesopores to the
ion moving in the microporous structure is more effective for the
Liϩ cation as larger ions.
Figure 7. Schematic illustration for
evaluation of anion adsorption capaci-
tance and cation adsorption capacitance
from chronopotentiograms of ACF elec-
trode. The chronopotentiogram in the
upper ͑lower͒ potential region than the
initial potential reflects the anion ͑cat-
ion͒ adsorption/desorption.
Downloaded on 2014-10-29 to IP 128.248.155.225 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, 149 ͑7͒ A855-A861 ͑2002͒
A861
These results show how significant PSD in activated carbons and ion
size in electrolyte are for the improvement of the double-layer ca-
pacitance. Therefore, the correlation of the double-layer capacitance
with the pore structure of activated carbons and the physicochemical
properties of electrolyte must be investigated in more detail using
other samples in the future ͑for example, ACFs with various pore
size distributions prepared from other activation catalysts, other
electrolytes with various quaternary onium salts, etc.͒.
Acknowledgment
A part of this study was financially supported by Industrial Tech-
nology Research Grant Program in 2001 from the New Energy and
Industrial Technology Development Organization ͑NEDO͒ of Japan.
Gunma University assisted in meeting the publication costs of this ar-
ticle.
References
1. B. E. Conway, Electrochemical Super Capacitors, Kluwer Academic/Plenum Pub-
lishers, New York ͑1999͒.
2. A. Nishino, J. Power Sources, 60, 137 ͑1996͒.
3. S. J. Gregg and K. S. W. Sing, Adsorption, Surface Area and Porosity, Academic
Press, London ͑1982͒.
4. I. Takahashi, A. Yoshida, and A. Nishino, Electrochemistry, 56, 892 ͑1988͒.
5. T. Morimoto, K. Hiratsuka, Y. Sanada, and K. Kurihara, J. Power Sources, 60, 239
͑1996͒.
Figure 8. Dependence of electric double-layer capacitance ͑negative pro-
cess͒ on current density for various Ref-ACFs and 0.1 Ni-ACFs in 1.0 mol
dmϪ3 LiClO4 /PC. Galvanostatic method ͑10, 40, 80, 160 mA gϪ1͒, negative
process 4 V → 2 V vs. Li/Liϩ. Each capacitance was calculated from each
chronopotentiogram in the region of 2.25-3.75 V vs. Li/Liϩ to eliminate the
influence of IR drop.
6. H. Shi, Electrochim. Acta, 41, 1633 ͑1996͒.
7. D. Qu and H. Shi, J. Power Sources, 74, 99 ͑1998͒.
8. S. Iwasaki, Y. Okusako, M. Miyahara, and M. Okazaki, Kagaku Kogaku Ronbun-
shu, 23, 512 ͑1997͒ ͑in Japanese͒.
9. C. Lin, J. A. Ritter, and B. N. Popov, J. Electrochem. Soc., 146, 3639 ͑1999͒.
10. H. Nakagawa, A. Shudo, and K. Miura, J. Electrochem. Soc., 147, 38 ͑2000͒.
11. Y. Sakata, A. Muto, M. A. Uddin, N. Yamada, C. Marumo, S. Ibaraki, and K.
Kojima, Electrochem. Solid-State Lett., 3, 1 ͑2000͒.
12. S. Yoon, J. Lee, T. Hyeon, and S. M. Oh, J. Electrochem. Soc., 147, 2507 ͑2000͒.
13. S. Shiraishi, H. Kurihara, H. Tsubota, A. Oya, S. Soneda, and Y. Yamada, Electro-
chem. Solid-State Lett., 4, A5 ͑2001͒.
14. G. Salitra, A. Soffer, L. Eliad, Y. Cohen, and D. Aurbach, J. Electrochem. Soc., 147,
2486 ͑2000͒.
ered on the viewpoint of volumetric capacitance and the cost when
applying these mesoporous ACFs to a practical EDLCs.
Conclusion
Meso/macroporous ACFs that contain many meso/macropores in
addition to micropores were prepared from carbonization and steam
activation of a phenolic resin fiber blended with activation catalyst
of nickel. The gravimetric electric double-layer capacitance of the
meso/macroporous ACFs and the conventional microporous ACFs
was measured in 1.0 M LiClO4 /PC using a three-electrode cell. As
a result, the following points were clarified about the correlation
between the double-layer capacitance and the pore structure.
1. In both cases of the meso/macroporous ACFs and the mi-
croporous ACFs, the double-layer capacitance could not be ex-
plained simply with the linearity to the specific surface area. This
nonlinearity is due to the low permeability of nonaqueous electro-
lyte to the microporous structure or the low mobility of the ion in
the narrow micropores of ACFs.
15. M. Suzuki, Carbon, 32, 577 ͑1994͒.
16. S. K. Ryu, High Temp.-High Press., 22, 345 ͑1990͒.
17. A. Oya, S. Yoshida, J. Alcaniz-Monge, and A. Linares-Solano, Carbon, 33, 1085
͑1995͒.
18. H. Tamai, K. Kakii, Y. Hirota, T. Kumamoto, and H. Yasuda, Chem. Mater., 8, 454
͑1996͒.
19. S. Shiraishi, H. Kurihara, and A. Oya, Carbon Science, 1, 133 ͑2001͒.
20. S. Shiraishi, H. Kurihara, and A. Oya, Electrochemistry, 69, 440 ͑2001͒.
21. J. Przepiorski and A. Oya, J. Mater. Sci. Lett., 17, 679 ͑1998͒.
22. K. Kaneko, C. Ishii, M. Ruike, and H. Kuwabara, Carbon, 30, 1075 ͑1992͒.
23. D. Dollimore and G. R. Heal, J. Appl. Chem., 14, 109 ͑1964͒.
24. M. M. Dubinin and H. F. Stoeckli, J. Colloid Interface Sci., 75, 34 ͑1980͒.
25. M. E. Merraoui, H. Tamai, H. Yasuda, T. Kanata, J. Mondori, K. Nadai, and K.
Kaneko, Carbon, 36, 1769 ͑1998͒.
2. The meso/macroporous ACFs showed higher double-layer ca-
pacitance than the microporous ACFs. This result suggests that the
presence of many mesopores ͑or macropores͒ promotes the forma-
tion of an effective double layer or the transfer of ions in the mi-
croporous structure. This effect is more prominant when a cation
with large size in the electrolyte, such as Liϩ cation, is adsorbed/
desorbed.
3. The capacitance of the meso/macroporous ACFs was higher
even at high current density measurement compared with that of the
microporous ACFs.
26. Y. Matsuda, H. Nakashima, M. Morita, and Y. Takasu, J. Electrochem. Soc., 128,
2552 ͑1981͒.
27. M. Ue, J. Electrochem. Soc., 141, 3336 ͑1994͒.
28. H. Marsh, Activated Carbon Compendium, H. Marsh, Editor, Elsevier, Oxford
͑2001͒.
29. P. L. Walker, Jr., F. Rusinko, Jr., and L. G. Austin, Advances in Catalysis, Vol. XI,
D. D. Eley, P. W. Selwood, and P. B. Weisz, Editors, p. 136, Academic Press, Inc.,
New York ͑1959͒.
30. J. P. Randin and E. Yeager, J. Electrochem. Soc., 118, 711 ͑1971͒.
31. H. Tobias and A. Soffer, J. Electroanal. Chem., 148, 221 ͑1983͒.
32. M. Morita, S. Watanabe, M. Ishikawa, H. Tamai, and H. Yasuda, Electrochemistry,
69, 462 ͑2001͒.
Downloaded on 2014-10-29 to IP 128.248.155.225 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).