Carbon 41 (2003) 1765–1775

Influence of pore structure and surface chemistry on electric

double layer capacitance in non-aqueous electrolyte
D. Lozano-Castello´ a , D. Cazorla-Amoros
´ a , A. Linares-Solano a , *, S. Shiraishi b ,
H. Kurihara b , A. Oya b
a
´
Departamento de Quımica ´
Inorganica , Universidad de Alicante, E-03080, Alicante, Spain
b
Department of Chemistry, Faculty of Engineering, Gunma University, Kiryu, Gunma, Japan
Received 2 January 2003; accepted 20 March 2003

Abstract

The performance as electric double layer capacitors (EDLC) in non-aqueous electrolyte of a series of alkaline
agent-activated carbons with high surface area is presented in this work. The results obtained show that, in general,
capacitance increases with surface area. However, the results obtained in this study confirm that capacitance not only
depends on surface area, but also on two other parameters: pore size distribution and surface chemistry. It has been shown
that capacitance is higher for a sample with wider micropore size distribution than for a sample with higher surface area but
too narrow micropore size distribution. In addition, it has been observed that the sample with a very high amount of surface
groups presents very high capacitance values. In the present study, a KOH-activated carbon with a capacitance as high as
220 F / g was prepared. Finally, the results obtained with a mesoporous sample have shown that the presence of mesopores in
activated carbons with very high surface area (e.g. .2000 m 2 / g), do not seem to be effective for double layer capacitors.
 2003 Elsevier Science Ltd. All rights reserved.

Keywords: A. Activated carbon; C. Adsorption; Temperature programmed desorption; D. Electrochemical properties

1. Introduction For applications demanding energy for about 1 s, the

devices used are large and expensive. Thus, a new
Electrical energy storage systems can be classified, in a generation of electrochemical components for energy
very general way, according to their energy density storage known as ‘supercapacitors’ or ‘electric double
(energy stored per volume unit or weight unit) and layer capacitors (EDLC)’ are being investigated and
according to their power (speed of access to the energy commercialized by some research groups and companies
stored). It / one extreme of these classifications would be [1–13]. These supercapacitors bridge the gap between
the conventional electrolytic capacitors, which can deliver conventional electrolytic capacitors and batteries because
very high currents, but only for a very short time because they present a unique combination of high energy density
their energy density is very small. These capacitors show and high power density.
the best performance in applications having characteristic The different behavior of the batteries and super-
times in the micro- to millisecond time range. At the other capacitors is due to the different mechanisms of operation
extreme of the classification are batteries. Batteries are of supercapacitors and battery cells: in the batteries, the
capable of storing a lot of energy, but they are unable to essential process is Faradaic, i.e. electron transfer takes
release the stored charge quickly because they have a place across the interface electrolyte / electrode. In super-
limited power output. Batteries are very suitable in appli- capacitors, the charge storage process is non-Faradaic, i.e.
cations where power is needed for several minutes to ideally no electron transfer takes place across the electrode
several hours. interface. In this device, the stored electrical energy is
based on the separation of charged species in an electrical
double layer across the electrode / solution interface.
*Corresponding author. Fax: 134-965-903454. Because the charging and discharging of these double
E-mail address: linares@ua.es (A. Linares-Solano). layer capacitors involves no chemical reactions, such

0008-6223 / 03 / $ – see front matter  2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016 / S0008-6223(03)00141-6
1766 D. Lozano-Castello´ et al. / Carbon 41 (2003) 1765–1775

Table 1
Properties of coals
Precursor Moisture Volatile matter Ash Fixed carbon
(%) (%) (%) (%)
Anthracite (UA) 2 7 6 86
Bituminous (UB) 2 27 19 52
Lignite 6 44 7 43

capacitors have a high degree of reversibility, which is window. Thus, the case in non-aqueous electrolyte has to
extremely important for this type of device. be clarified.
High energy and high power EDLC uses activated The first objective of the present work was to extend the
carbons or glassy carbons as electrodes. The intention of study of EDLC to alkaline agent-activated carbons with
using activated carbon in EDLC manufacture is attributed high surface area, especially in non-aqueous electrolyte.
to the recognized fact that the larger surface area that an The second objective was to try and understand the
EDLC can provide for adsorption of ions on electrodes, the relationship between carbon properties and capacitance. A
more energy can be stored in the EDLC [3,14–17]. series of activated carbons by using KOH as activating
Additionally, activated carbons are also interesting from an agent and different coals as raw material (anthracite,
economical point of view. bituminous coal and lignite) were prepared. For com-
One of the physical characteristics of activated carbons parison purposes, a commercial activated carbon (Max-
is the wide variety of pore size distribution, which is sorb-A) was also included in this study. The effect of
significant for penetration or mobility of ions in micropore surface area, pore size distribution and surface chemistry
structures. Thus, the double layer capacitance of activated on DLC was analyzed.
carbons has been investigated in terms of the correlation
between capacitance and pore structure [3–13]. On the
other hand, the preparation process of the activated carbon
and the resulting surface characteristics (such as surface 2. Experimental
functionality) can affect the performance of the EDLC, and
has rarely been discussed at length in the literature. All activated carbons were prepared by using the same
Recently, Teng et al. measured the amount of surface chemical activation process and the same activating agent
oxygen functional group on PAN-based activated carbon (KOH). Details of the preparation process are available in
electrode by using temperature-programmed desorption the literature [20]. Three different precursors were used for
(TPD) [18,19]. They revealed the enhancement of the the preparation of the activated carbons in the present
capacitance by the surface functionality. However, their study: (i) an anthracite (A); (ii) a bituminous coal (B); and
reports are limited only in the capacitance performance in (iii) a lignite coal (L). Table 1 includes the properties of
acid aqueous electrolyte. It is known that the energy these three coals used as raw materials and Table 2
density of EDLC in organic electrolytes (e.g. propylene displays the preparation conditions for each sample. The
carbonate, acrylonitrile solution) is several times as much activated carbon was a powdered material and the yield of
as that in the aqueous one because of the electrochemical the preparation process was around 60–70%. In addition to

Table 2
Preparation conditions for obtaining the activated carbons for this study
Sample Precursor KOH / carbon ratio Activation temp. N 2 flow rate
(8C) (ml / min)
A1 Anthracite 1/1 700 800
A2 Anthracite 2/1 700 80
A3 Anthracite 2/1 650 800
A4 Anthracite 3/1 700 800
A5 Anthracite 3/1 750 800
A6 Anthracite 3/1 850 800
A7 Anthracite 3/1 950 800
B1 Bituminous 2/1 700 800
B2 Bituminous 4/1 700 800
L Lignite 4/1 800 800
 D. Lozano-Castello´ et al. / Carbon 41 (2003) 1765–1775 1767

these samples, a commercial activated carbon (Maxsorb-A) The specific gravimetric electric double layer capacitance
was also included. of the activated carbons was calculated by the amount of
Surface area characterization of all samples has been electricity passed during the positive process (2 to 4 V vs.
carried out by physical adsorption (N 2 at 77 K and CO 2 at Li / Li 1 ) on the choronopotentiograms. This capacitance
273 K) using an automatic adsorption system (Autosorb-6, corresponds to the average one derived from the cation
Quantachrome). The total micropore volume (pore size desorption and the anion adsorption since the initial
smaller than 2 nm) was calculated from the application of potential of all electrodes is around 3 V vs. Li / Li 1 . The
the Dubinin–Radushkevich (DR) equation to the N 2 details of the electrochemical measurement are described
adsorption at 77 K. The narrow micropore volume (pore in a previous paper [12,13].
size smaller than around 0.7 nm) has been assessed from
CO 2 adsorption at 273 K using the DR equation [21–24].
The adsorbate densities used were 0.808 and 1.023 g / ml, 3. Results and discussion
for N 2 at 77 K and CO 2 at 273 K, respectively [21–24].
The specific surface area was calculated by using the BET 3.1. Pore texture and surface chemistry characterization
equation.
Temperature-programmed desorption (TPD) experi- Fig. 1 displays the N 2 adsorption isotherms corre-
ments were done in a DSC-TGA (TA, SDT 2960 Simulta- sponding to all the samples of the present study: seven
neous) coupled to a mass spectrometer to characterize the activated carbons prepared from an anthracite (A1–A7),
surface chemistry of all the samples. In these experiments, two activated carbons prepared from a bituminous coal
20 mg of sample was heated to 950 8C (heating rate (B1, B2), an activated carbon prepared from a lignite coke
20 8C / min) under a helium flow rate of 100 ml / min. (L), and the commercial activated carbon with the highest
To measure the electric double layer capacitance, surface area (Maxsorb-A).
composite pellet electrodes were prepared from powder All the samples present type I isotherms [25], charac-
activated carbon, acetylene black (Denki Kagaku Kogyo teristic of microporous solids. However, the shape of the
Co., Japan) and binder (PTFE 6J, Du Pont-Mitsui Fluoro- isotherms corresponding to samples A7 and L is quite
chemicals Co. Ltd., Japan), in a ratio of 86:10:4 wt.%, different to the other samples. This type of isotherm is
respectively. The composite electrode was placed on characteristic of samples containing mesoporosity (con-
titanium mesh as current collector. The standard three- firmed by the existence of a hysteresis cycle in the
electrode cell in highly pure argon glove box was utilized adsorption / desorption process). These activated carbons
to measure the electric double layer capacitance of a single were prepared to study the effect of the mesoporosity in
activated carbon electrode. Propylene carbonate containing the double layer capacitance.
1.0 mol / dm 3 LiClO 4 electrolyte (1.0 M LiClO 4 / PC) was Table 3 contains the BET surface areas and the micro-
used as non-aqueous electrolyte. Measurement of the pore volumes calculated from N 2 adsorption data at 77 K
capacitance was performed by the galvanostatic method [VDR (N 2 )] and CO 2 adsorption data at 273 K [VDR (CO 2 )]
(40 mA / g, a potential range from 2 to 4 V vs. Li / Li 1 ). for each sample. The BET surface area values corre-

Fig. 1. Nitrogen adsorption isotherms at 77 K for all the activated carbons.

1768 D. Lozano-Castello´ et al. / Carbon 41 (2003) 1765–1775

Table 3
Porous texture characterization results corresponding to the activated carbons
Sample SBET VDR (N 2 ) VDR (CO 2 ) VDR (N 2 )2VDR (CO 2 )
(m 2 / g) (cm 3 / g) (cm 3 / g) (cm 3 / g)
A1 726 0.33 0.37 20.04
A2 945 0.43 0.43 0
A3 1452 0.62 0.58 0.04
A4 2647 1.19 0.64 0.55
A5 3012 1.33 0.71 0.62
A6 3506 1.51 0.68 0.83
A7 2179 1.01 0.38 0.63
B1 2105 0.94 0.73 0.21
B2 2123 0.93 0.92 0.01
L 884 0.43 0.30 0.13
Maxsorb-A 3100 1.38 0.64 0.74

sponding to all the samples are plotted in Fig. 2. It can be Another way to observe the evolution of the micropore
observed that this series of samples covers a very wide size distribution in this series of activated carbons is by
range of surface areas (BET surface areas from 700 up to plotting the normalized nitrogen adsorption isotherms (see
more than 3000 m 2 / g). Fig. 3). These normalized curves were obtained by divid-
The comparison between the total micropore volume ing the amount of N 2 adsorbed at a given relative pressure
[VDR (N 2 )] and the narrow micropore volume [VDR (CO 2 )] by the amount of N 2 adsorbed at a relative pressure close
gives an idea of the micropore size distribution. It should to saturation pressure (0.99). In this type of graph, the
be noted that in most cases VDR (N 2 ).VDR (CO 2 ), which comparison of the width of the knee for the different
indicates that the mean pore size of these samples is higher samples is much easier than in the isotherms. As concluded
than approximately 0.8 nm [21–24]. However, sample A1 from the comparison of the micropore volumes, the
has a VDR (N 2 ),VDR (CO 2 ) due to the presence of a very narrowest micropore size distributions corresponds to
narrow microporosity (mean pore size lower than approxi- samples A1, A2, A3 and B2. On the other hand, the widest
mately 0.8 nm), where N 2 adsorption at 77 K presents micropore size distributions correspond to the mesoporous
diffusional problems [21–24]. In addition, three samples sample (A7), the most activated samples (A6 and Max-
(A2, A3, B2) with a very homogeneous microporosity [VDR sorb-A) and the sample prepared from lignite (L).
(N 2 )¯VDR (CO 2 )] and mean pore size around 0.8 nm As explained in the Experimental section, DTP experi-
[21–24] have been prepared. ments were carried out to characterize the surface chemis-

Fig. 2. BET surface area corresponding to all the activated carbons prepared.
 D. Lozano-Castello´ et al. / Carbon 41 (2003) 1765–1775 1769

Fig. 3. Normalized nitrogen adsorption isotherms at 77 K for all the activated carbons.

try of the samples. Most of the complexes at the surface of contained a large amount of oxygenated groups. It can be
activated carbon are oxygenated so that the desorption of observed that, depending on the raw material and the
this complex evolved as CO and CO 2 ; this provides experimental conditions used for the preparation of the
information about the chemistry of the carbon surface. Fig. sample, the amount of oxygenated groups in the activated
4 includes, as an example, the DTP curves corresponding carbon is different.
to samples A4 and A7. It is known that CO 2 evolves at Fig. 5 shows the mmol of CO and CO 2 evolved in the
low temperatures as a consequence of the decomposition TPD experiments together with the BET surface area for
of the acidic surface groups, while CO comes from basic each sample. For the series of samples prepared from
or neutral groups [26,27], which are more thermally stable, anthracite, it can be observed that when the BET surface
so they require higher temperatures for decomposition. area increases, the amount of surface groups reaches a
Table 4 contains the quantification of the amount of CO maximum, which corresponds to sample A4. For this
and CO 2 desorbed in these experiments. Except for the series, the lowest amount of surface groups corresponds to
activated carbon prepared from lignite, all the samples sample A7 (see DTP curves in Fig. 4), which is probably

Fig. 4. DTP curves corresponding to two of the activated carbons.

1770 D. Lozano-Castello´ et al. / Carbon 41 (2003) 1765–1775

Table 4 tance [12,13]. Thus, samples A1 and A2 with steep slopes

Quantification of oxygenated surface groups in activated carbons have a smaller capacitance, while A4–6 and Maxsorb have
obtained from DTP experiments a larger one. This is related to the surface area of each
Sample mmol CO / g mmol CO 2 / g sample. In the next section, the correlation between the
capacitance and the BET surface area is discussed in detail.
A1 – –
A2 1364 437 In the case of samples A4 and A5, the potential–time
A3 3349 1288 curves do not present ideal behavior (see Fig. 6a). It can be
A4 4296 1633 observed that a large ohmic-drop at the switch point of the
A5 2390 707 current direction and curvatures at negative process (4 to 2
A6 2200 718 V vs. Li / Li 1 ) exist. The former is derived from a higher
A7 758 1048 resistance in the electrode. The curvature at negative
B1 1903 1084 process indicates the dependency of the capacitance on the
B2 2853 718 potential, which can be due to the difference in the
L 503 382 capacitance between cation-sorption and anion-sorption or
Maxsorb-A 3304 1404
to the influence of Faradaic processes at the carbon
surface.
due to the higher temperature used for the preparation of
this sample (950 8C) compared with the rest of the
activated carbons (from 650 to 850 8C). 3.3. Correlation between double layer capacitance and
According to the characterization results, the series of surface area
samples prepared for this study present different surface
area, pore size distribution and surface chemistry. Thus, In general, it is believed that there is a direct relationship
they are suitable for studying the effect of these properties between the BET surface area and the double layer
of the activated carbons in their double layer capacitors capacitance (DLC) of porous carbons [1,2,5]. Theoret-
performance. ically, the higher the specific surface area of an activated
carbon, the higher the specific capacitance should be
3.2. Chronopotentiogram expected. Practically, the situation is more complicated.
Table 5 displays the capacitance values corresponding to
Fig. 6 shows the chronopotentiograms (potential–time all the samples. These capacitance values are the specific
curves) of all samples in 1.0 M LiClO 4 / PC. All capacitance per activated carbon material weight, and they
chronopotentiograms were triangular, which indicates a were calculated by the amount of electricity passed during
typical behavior of capacitive electrodes. The inverse of the charge process. It can be observed that capacitance
the gradient at the potentiogram corresponds to the capaci- values as high as 220 F / g were obtained, which indicate

Fig. 5. Amount of CO and CO 2 evolved during the DTP experiments and BET surface area for all the activated carbons.
 D. Lozano-Castello´ et al. / Carbon 41 (2003) 1765–1775 1771

Fig. 6. Chronopotentiograms (potential–time curves) for the activated carbons in 1.0 M LiClO 4 / PC (current density: 40 mA / g). (a) A
series, (b) B series, L, and Maxsorb.

that micropores (pore size less than 2 nm) contribute to the

Table 5 DLC.
Double layer capacitance values corresponding to all the activated Fig. 7 shows the relationship between the BET surface
carbons area and the electric double layer capacitance (potential
Sample Capacitance (F / g) range 2–4 V vs. Li / Li 1 ) in 1.0 LiClO 4 / PC. Even though
the use of BET gives an overestimated surface area value,
A1 18
A2 43 this method is acceptable for the comparison of a series of
A3 114 samples, which have been prepared with the same prepara-
A4 220 tion method, just by changing the preparation parameters.
A5 204 Fig. 7 shows that a general trend between capacitance and
A6 215 BET surface area exists for all the samples, although it is
A7 126 not a perfect linear relationship. This trend was also
B1 167 observed in previous studies using other carbon materials
B2 161 with lower surface area [1,2,5]. These results indicate that
L 65 although the BET surface area is a very important parame-
Maxsorb-A 217
ter, the EDLC also seems to depend on other characteris-
1772 D. Lozano-Castello´ et al. / Carbon 41 (2003) 1765–1775

Fig. 7. Capacitance versus BET surface area for the activated carbons.

tics of the porous carbon materials, probably on the pore their pore size distribution. As concluded from Figs. 1 and
size distribution and surface chemistry. 3, sample L shows the presence of mesopores and a wider
micropore size distribution than sample A2, even though
3.4. Deviation by ion-sieving effect both samples were prepared under similar preparation
conditions and with the same activation method. However,
In Fig. 7 it can be observed that the activated carbon the different type of precursor used determines the differ-
with the lowest surface area (A1) has a very low capaci- ent structure of the resulting carbon. The existence of
tance, much lower than the expected value according to the mesopores and a wider micropore size distribution causes
trend obtained with the rest of the activated carbons. There sample L to have a considerably higher capacitance, which
are a few reasons for the absence of proportionality fits the general trend obtained for the other activated
between specific capacitance and surface area. An im- carbons. These results show the importance of the acces-
portant factor must be the accessibility of micropores to sibility of narrow micropores to electrolytic solution.
the electrolytic medium. Thus, the very low value obtained Depending on the electrolytic medium, the convenient
for sample A1 must be due to its very small mean pore porous carbon material should be selected for a capacitor
size, which makes the entry of the electrolyte difficult. The electrode.
very narrow micropores (non-accessible pores) do not Another sample presenting a remarkable deviation from
contribute to the total double-layer capacitance of the the general trend is sample A7. As seen previously, this
material. This ion-sieving effect has also been observed by sample was selected to study the effect of mesoporosity in
other authors using other materials and electrolytes [4,7]. this type of material. In a previous study [12,13] where the
Sample A2 also presents the same type of deviation, EDLC performance of activated carbon fibers with BET
although it is not as important as for the sample A1, surface areas lower than 2000 m 2 / g were measured (using
because sample A2 has a higher mean pore size. the same experimental set-up as the present study), it was
To confirm that the deviations observed for these shown that mesoporosity is beneficial for EDLC. However,
samples are due to the difficulty of the electrolyte to enter according to the results obtained with sample A7, it seems
into the pores, the performance and the characteristics of that the presence of mesopores in materials with high
sample A2 is compared to those corresponding to sample surface area (.2000 m 2 / g) is not very effective for double
L. Both samples (A2 and L) have a similar BET surface layer capacitance.
area (see Table 3) and the amount of oxygenated groups in
sample L is lower than in sample A2. However, as can be 3.5. Deviation by surface functionalities
observed in Fig. 7 and Table 5, the capacitance of sample
L is higher than that of sample A2. This different According to the values presented in Table 5 the
performance of both samples can be explained considering maximum capacitance value corresponds to sample A4. In
 D. Lozano-Castello´ et al. / Carbon 41 (2003) 1765–1775 1773

Fig. 7 it can be observed that this activated carbon has a

capacitance much higher than expected, showing a higher
capacitance than samples with higher area (see, e.g. sample
A5). In this case, the micropore size distribution of these
samples (A4 and A5) cannot explain their different per-
formance because as shown in Fig. 3, the pore size
distribution of both samples is very similar. Thus, the
accessibility of the electrolyte inside the microporosity,
from the point of view of porosity, should be similar for
both samples.
The different performance of these two samples could
be due to their different surface functionality. As men-
tioned previously, the DTP shows that the amount of CO
and CO 2 evolved in the TPD experiments is at a maximum
for sample A4. The presence of this large amount of
oxygenated groups in the surface of the activated carbon
can affect the capacitance of the material mainly in two
different ways: (1) the oxygenated groups can improve the
wettability of the carbon surface, which is very important
to maximize the access of the electrolyte to the surface of
carbon; and (2) the presence of oxygenated groups causes
the double-layer capacitance to arise from quick Faradaic
charge transfer reactions as well as from electrostatic
charging (pseudocapacitance) [5]. It is known that in
double layer capacitors, usually between 1 and 5% of
capacitance corresponds to pseudocapacitance due to the
Faradic reactivity of the oxygenated groups [28]. A
contribution from both factors probably causes this sample
(with the highest content in surface groups) to present the
highest capacitance per unit mass of carbon.

3.6. Difference in capacitance between cation-sorption

and anion-sorption

Fig. 8a,b shows the correlations between BET surface

area and the double layer capacitance derived from the
cation adsorption / desorption or the anion adsorption / de-
sorption, respectively. The evaluation of each capacitance
was performed in the same way as a previous paper [13].
The cation capacitance in the samples (A1–3 and L) with
low surface area is smaller than the anion one. This is due Fig. 8. Capacitance versus BET surface area for activated car-
to the strong ion-sieving effect of cation-sorption because bons. (a) Cation adsorption / desorption capacitance; (b) anion
the solvated ion size of Li 1 (0.82 nm) is larger than that of adsorption / desorption capacitance.
ClO 24 (0.52 nm) [13]. The capacitance superiority of
sample L to sample A2 was more prominent in the cation-
sorption compared to that in the anion. This suggests that
the ion-sieving relaxation by mesopore is more prominent suggested that the surface functionalities enhance the
in cation-sorption. cation sorption or the Faradaic reaction such as C=O1
In the case of the high-activated samples (A4–A6) and Li 1 1e 2 ⇒C–O 2 Li 1 . The latter is one of the most reason-
Maxsorb, it can be seen that the cation capacitance is able reasons for sample A4 because of the largest amount
larger than the anion one, especially in sample A4. It is of CO evolution corresponding to the presence of the
difficult to explain by the ion-sieving effect because the carbonyl group. However, the larger capacitance for Li 1 -
micropore width of the high-activated samples is wider sorption for all samples cannot be explained only by the
than the rest of the sample. As shown in the previous surface functionalities. The influence of other factors such
section, these high-activated samples have a relatively as activation temperature or precursor types should be
large amount of oxygenated surface group. Thus, it is discussed in future.
1774 D. Lozano-Castello´ et al. / Carbon 41 (2003) 1765–1775

4. Conclusions cloth electrodes and an organic electrolyte. Denki Kagaku

1988;56:892–7.
Activated carbons with very high capacitance have been [2] Morimoto T, Hiratsuka K, Sanada Y, Kurihara K. Electric
prepared by a chemical activation process. The results double-layer capacitor using organic electrolyte. J Power
Sources 1996;60:239–47.
obtained with activated carbon of very different surface
[3] Shi H. Activated carbons and double layer capacitance.
areas (from 700 up to .3000 m 2 / g) show that capacitance
Electrochim Acta 1996;41(10):1633–9.
increases with surface area, reaching a maximum of 220
[4] Qu D, Shi H. Studies of activated carbons used in double-
F / g for a surface area of 2647 m 2 / g (sample A4). The layer capacitors. J Power Sources 1998;74:99–107.
results indicate that micropores (pore size ,2 nm) contrib- [5] Lin C, Ritter JA, Popov BN. Correlation of double-layer
ute to the double layer capacitance. capacitance with the pore structure of sol–gel derived carbon
However, the results obtained in this study confirm that xerogels. J Electrochem Soc 1999;146(10):3639–43.
capacitance not only depends on surface area, but also on [6] Nakagawa H, Shudo A, Miura K. High-capacity electric
two other parameters: pore size distribution and surface double-layer capacitor with high-density-activated carbon
chemistry. The effect of pore size distribution has been fiber electrodes. J Electrochem Soc 2000;147(1):38–42.
observed comparing the capacitance of two samples with [7] Salitra G, Soffer A, Eliad L, Cohen Y, Aurbach D. Carbon
different surface area and different pore size distributions: electrodes for double-layer capacitors. I. Relations between
capacitance is higher for a sample with wider micropore ion and pore. J Electrochem Soc 2000;147(7):2486–93.
size distribution than for a sample with higher surface area [8] Yoon S, Lee J, Hyeon T, Oh SM. Electric double-layer
capacitor performance of a new mesoporous carbon. J
but too narrow micropore size distribution. The existence
Electrochem Soc 2000;147(7):2507–12.
of a very small mean pore size makes the entry of the
[9] ¨
Sullivan MG, Schnyder B, Bartsch M, Alliata D, Barbero C,
electrolyte into the pores difficult. Then, the non-accessible ¨ R. Electrochemically modified glassy carbon
Imhof R, Kotz
pores do not contribute to the total double layer capaci- for capacitor electrodes. Characterization of thick anodic
tance of the material (ion-sieving effect). layers by cyclic voltammetry, differential electrochemical
The effect of surface chemistry was clearly observed for mass spectrometry, spectroscopic ellipsometry, X-ray photo-
the sample with the highest amount of surface groups electron spectroscopy, FTIR, and AFM. J Electrochem Soc
(sample A4), which is the sample with the highest capaci- 2000;147(7):2636–43.
tance (220 F / g), even higher than the capacitance corre- [10] Sakata Y, Muto A, Uddin MA, Yamada N, Marumo C,
sponding to activated carbons with higher surface area. Ibaraki S, Kojima K. Preparation of carbon electrodes for
The presence of oxygenated groups on the surface of the electronic double-layer capacitors by carbonization of metal-
ion-exchanged resins. Eletrochem Solid-State Lett
activated carbon most probably affect the capacitance of
2000;3(1):1–3.
the materials mainly in two different ways: (1) by improv-
[11] Shiraishi S, Kurihara H, Tsubota H, Oya A, Soneda Y,
ing the wettability of the material, increasing the capaci- Yamada Y. Electric double layer capacitance of highly
tance; and (2) by producing pseudocapacitance effects, so porous carbon derived from lithium metal and polytetra-
that Faradaic charge transfer reactions take place. fluoroethylene. Electrochem Solid-State Lett 2001;4(1):A5–
Finally, the results obtained with a mesoporous sample 8.
(sample A7) have shown that the presence of mesopores in [12] Shiraishi S, Kurihara H, Oya A. Preparation and electric
activated carbons with very high surface area (e.g. .2000 double layer capacitance of mesoporous carbon. Carbon
m 2 / g), do not seem to be effective for double layer Science 2001;1(3–4):133–7.
capacitors. The important microporosity development in [13] Shiraishi S, Kurihara H, Shi L, Nakayama T, Oya A. Electric
double layer capacitance of meso / macroporous carbon fibers
the chemically activated carbons prepared in the present
prepared by the blending method I. Ni-loaded activated
work probably produces a mean micropore size wide
carbon fibers in propylene carbonate solution containing
enough to allow the electrolyte access inside the porosity. LiClO 4 salt. J Electrochem Soc 2002;149(7):A855–861.
[14] Liu X, Osaka TJ. All-solid-state electric double-layer
capacitor with isotropic high-density graphite electrode and
polyethylene oxide / LiClO 4 polymer electrolyte. J Electro-
Acknowledgements
chem Soc 1996;143(12):3982–6.
[15] Liu X, Osaka TJ. Properties of electric double-layer
The authors thank MCYT (Projects MAT 2000-0621) capacitors with various polymer gel electrolytes. J Electro-
for financial support. D. Lozano-Castello´ thanks M.E.C. chem Soc 1997;144(9):3066–71.
for the thesis grant. [16] Miller JM, Dunn B, Tran TD, Pekala RW. Deposition of
ruthenium nanoparticles on carbon aerogels for high energy
density supercapacitor electrodes. J Electrochem Soc
1997;144(12):L309–311.
References [17] Saliger R, Fischer U, Herta C, Fricke J. High surface area
carbon aerogels for supercapacitors. J Non-Cryst Solids
[1] Tanahashi I, Yoshida A, Nishino A. Properties of the electric 1998;225:81–5.
double layer capacitors composed of activated carbon fiber [18] Hsieh CT, Teng H. Influence of oxygen treatment on electric
 D. Lozano-Castello´ et al. / Carbon 41 (2003) 1765–1775 1775

double-layer capacitance of activated carbon fabrics. Carbon ´

[24] Linares-Solano A, Salinas-Martınez ˜
de Lecea C, Alcaniz-
2002;40:667–74. Monge J, Cazorla-Amoros ´ D. Further advances in the
[19] Nian YR, Teng H. Nitric acid modification of activated characterization of microporous carbons by physical ad-
carbon electrodes for improvement of electrochemical sorption of gases. Tanso 1998;185:316–25.
capacitance. J Electrochem Soc 2002;149(8):A1008–14. [25] Sing KSW, Everett DH, Haul RAW, Moscou L, Pierotti RA,
[20] Lozano-Castello´ D, Lillo-Rodenas
´ MA, Cazorla-Amoros ´ D, Rouquerol J, Siemieniewska T. Reporting physisorption data
Linares-Solano A. Preparation of activated carbons from for gas / solid systems with special reference to the de-
Spanish anthracite. I. Activation by KOH. Carbon termination of surface area and porosity. Pure Appl Chem
2001;39(5):741–9. 1985;57(4):603–19.
[21] ´
Rodrıguez-Reinoso F, Linares-Solano A. Microporous struc- [26] Voll M, Bohem HP. Catalytic hydrogen chloride oxidation on
ture of activated carbons as revealed by adsorption methods. carbon black. Carbon 1970;8(1):99.
In: Thrower PA, editor, Chemistry and physics of carbon, ´
[27] Roman-Martınez´ MC, Cazorla-Amoros´ D, Linares-Solano A,
vol. 21, New York: Marcel Dekker; 1988, pp. 1–146. ´
Salinas-Martınez de Lecea C. TPD and TPR characterization
[22] Cazorla Amoros ˜ Monge J, Linares Solano A.
´ D, Alcaniz of carbonaceous supports and platinum / carbon catalysts.
Characterization of activated carbon fibers by CO 2 adsorp- Carbon 1993;31(6):895–902.
tion. Langmuir 1996;12(11):2820–4. [28] Conway BE. Electrochemical supercapacitors: Scientific
[23] Cazorla Amoros ˜ Monge J, De la Casa Lillo MA,
´ D, Alcaniz fundamentals and technological applications. Kluwer Aca-
Linares Solano A. CO 2 as an adsorptive to characterize demic; 1999.
carbon molecular sieves and activated carbons. Langmuir
1998;14(16):4589–96.