A facile production of microporous carbon spheres and their electrochemical performance in EDLC

Article abstract of DOI:10.1016/j.jpcs.2011.10.028

In the absence of activation process, we prepared a series of carbon particles from saccharine, in which hydrothermal carbonization method was used. These particles have spherical or near-spherical morphology, controllable monodisperse particle size from the analyses of SEM. Raman and XRD results show that they are nongraphitizable. The BET surface area of these carbon spherules is around 400-500 m² g^-1 and the microporosity is about 84%, suggesting that the carbon particles are rich in micropores. The electrochemical behaviors were characterized by means of galvanostatic charging/discharging, cycle voltammetry and impedance spectroscopy. The results show that the specific capacitance of sucrose-based carbon spherule reached 164 F g^-1 in 30% KOH electrolyte and a high volumetric capacitance over 170 F cm^-3 was obtained. These carbon spherules could be promising materials for EDLC according to their facile preparation way, low cost and high packing density.

Full text of DOI:10.1016/j.jpcs.2011.10.028

Journal of Physics and Chemistry of Solids 73 (2012) 385–390
Contents lists available at SciVerse ScienceDirect
Journal of Physics and Chemistry of Solids
journal homepage: www.elsevier.com/locate/jpcs
A facile production of microporous carbon spheres and their electrochemical
performance in EDLC
n
XiaoHong Xia, Lei Shi, HongBo Liu , Li Yang, YueDe He
College of Materials Science and Engineering, Hunan University, Changsha, Hunan 410082, China
a r t i c l e i n f o
a b s t r a c t
Article history:
In the absence of activation process, we prepared a series of carbon particles from saccharine, in which
hydrothermal carbonization method was used. These particles have spherical or near-spherical
morphology, controllable monodisperse particle size from the analyses of SEM. Raman and XRD results
show that they are nongraphitizable. The BET surface area of these carbon spherules is around
Received 2 September 2010
Received in revised form
4
October 2011
Accepted 20 October 2011
Available online 29 October 2011
2
ꢀ1
400–500 m g and the microporosity is about 84%, suggesting that the carbon particles are rich in
micropores. The electrochemical behaviors were characterized by means of galvanostatic charging/
Keywords:
discharging, cycle voltammetry and impedance spectroscopy. The results show that the specific
A. Microporous materials
C. Differential scanning calorimetry (DSC)
C. Raman spectroscopy
ꢀ1
capacitance of sucrose-based carbon spherule reached 164 F g
in 30% KOH electrolyte and a high
ꢀ3
volumetric capacitance over 170 F cm
was obtained. These carbon spherules could be promising
materials for EDLC according to their facile preparation way, low cost and high packing density.
2011 Elsevier Ltd. All rights reserved.
D. Electrochemical properties
&
1
. Introduction
low [8] or the procedures are more complicated [5,9]. Compared
to these methods, hydrothermal carbonization [10,11] is one
of the most efficient way to produce carbon spheres and the
procedure is rather simple.
Electric double-layer capacitors (EDLCs) are unique electrical
ꢀ
1
storage devices exhibiting high power density (over 10 kW kg
)
6
and long durability (over 10 cycles), which allowed their appli-
cations for various power and energy requirements, e.g. backup
power sources for electronic devices, engine start or acceleration
for hybrid electric vehicles and electricity storage generated from
solar or wind energy [1–3].
Many kinds of carbons with high specific surface areas and
suitable pore structures have displayed great potential as active
materials for EDLC [4–7]. Among various carbon materials, carbon
with spherical morphology and nanopores has been proved
practically to be competent for EDLC owing to its high packing
density, low surface-to-volume ratio and maximal structural
stability, etc.
Hydrothermal treatment of saccharine was firstly applied
during the first half of the twentieth century to obtain informa-
tion about the mechanism of natural coalification [8], by which a
carbon-rich solid product could be obtained by a thermal treat-
ment of water mixed with organic substances such as saccharine
or simple compounds such as furfural at temperatures in the
150–350 1C range. However, not until recently, this process has
generated widespread interest [10,11].
In our present work, several kinds of saccharine, such as
b-cyclodextrin, sucrose and glucose were employed for fabricat-
ing carbon spherules by hydrothermal carbonization without
activation. The electrochemical performance in EDLC of carbon
spherules produced from those saccharine was firstly investigated
in a 30 wt% KOH aqueous solution.
Different methods have been used to synthesize carbon
spheres [5,8,9]. Lee et al. have made microporous carbon nano-
spheres from an aromatic resin isotropic pitch through the
precipitation [8]; Li et al. have prepared mesoporous carbon
2. Experimental
spheres by
a
polymerization-induced colloid aggregation
method [5]. Ordered nanoporous carbon spheres have been
synthesized through the inverse replication of the highly ordered
MCM-41-type mesoporous silica template [9]. Among these
methods, however, the surface area of the carbon sphere is quite
2
.1. Preparation of carbon samples
The preparation of carbon samples was carried out according
to the following procedure: 80 mL saccharine solution of
8
ꢀ
1
.4 mol L (calculated by carbon atom) was filled into a stainless
n
steel autoclave, heated up to a temperature at 180 1C for 24 h. The
resulting dark brown powder was washed by distilled water and
Corresponding author. Tel.:þ86 18684656260; fax:þ86 073188823554.
E-mail address: xia_spring@163.com (H. Liu).
0022-3697/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jpcs.2011.10.028

3
86
X. Xia et al. / Journal of Physics and Chemistry of Solids 73 (2012) 385–390
dried at 100 1C for 12 h. Then the powder was further pyrolyzed
in a tube furnace in an argon atmosphere at 900 1C at 5 1C min
and maintained the temperature for 1.5 h. Finally the black
carbon powder was washed with distilled water and collected
hydrothermal reaction, since
cular weight among the three saccharide when other reaction
conditions are totally same. The resistance of -cyclodextrin to
decomposition may be a consequence of the fact that the hydro-
xyl groups of the glucose residues present in the structure of the
b-cyclodextrin has the largest mole-
ꢀ
1
b
after drying at 120 1C for 12 h. The samples obtained from
b-
cyclodextrin, sucrose and glucose are denoted by HS-cd, HS-s and
HS-g, and the carbon samples denoted by CS-cd, CS-s and CS-g,
respectively.
b-cyclodextrin form hydrogen bonds that hold the polymeric
chains firmly together and side-by-side [12].
The structure of carbon samples was characterized by Raman
spectra and XRD. From the Raman spectra (Fig. 2), the carbonized
samples exhibit two broad overlapping bands at around
2.2. Characterization of the carbon spherules
ꢀ
1
ꢀ1
1
320 cm
bands are usually assigned to the bands of disordered carbon and
graphitized carbon, respectively. The relative intensity (I /I ) has
been used to determine the graphitization degree of carbon
species. The I /I of CS-cd, CS-s and CS-g was estimated to be
(D-mode) and 1592 cm
(G-mode). The D and G
The morphology of the sample was investigated by a scanning
electron microscope (JSM-6700F, JEOL, Japan). The structure of
the sample was characterized by Raman spectra (laser confocal
Raman spectrometer LABRAM-010, JY Company, France) and X-
ray diffraction (Rigaku B/max-2400 X-ray diffractometer, Rigaku,
Japan). Nitrogen adsorption and desorption isotherms of carbons
were obtained at 77 K with an automatic adsorption apparatus
D G
D
G
about 1.05, 1.15 and 1.12, respectively, which reveal the presence
2
of C sp atoms in benzene or condensed benzene rings of
amorphous (partially hydrogenated) carbon [13,14]. The XRD
pattern was shown in Fig. 3. The diffraction profile of all carbon
samples show two diffraction peaks at the diffraction angle of
carbon (002), a broad peak around 24.81 and a sharp peak at 26.51,
which indicates there are two different carbon structures in them
(ASAP 2020, Micromeritics, USA). The BET SSA (SBET) and total
pore volumes (Vtot) were calculated using the Brunauer–Emmett–
Teller equation and the single point method, respectively. The
average pore diameter (D) was estimated from the equation 4Vtot
/
SBET. Micropore SSA (Smic) was calculated by the t-plot method.
[15]. Broad peak around 43.01 2y has been observed and this may
The corresponding pore size distributions were calculated with
the desorption data based on the BJH method.
be assigned to the (10) (overlapped 100 and 101) diffraction of
disordered stacking of micrographites [16], confirming the Raman
pattern in Fig. 2. During the course of hydrothermal treatment,
saccharine molecules undergo a series of resetting reactions
including intermolecular dehydration and aldol condensation,
and then the aromatization of polymers takes place. A relative
good graphitic net plane would be formed after carbonization.
However, since the transition time from fluid phase to solid phase
is not long that the thickness of the graphitic plane would be small.
Their nitrogen adsorption–desorption isotherms are shown in
Fig. 4a. The isotherms of the three samples correspond to type I
nitrogen adsorption–desorption isotherm according to the IUPAC
classification, which means there are mainly micropores. Sharp
slopes were illustrated in the isotherms of CS-g and CS-cd over
relative pressure 0.8, indicating besides micropores they have
macropores [17]. Fig. 4b shows the BJH pore size distributions of
these carbon spherules. It is clear that all three carbon samples
exhibit predominantly micropores and meso- and macropores
about 30–60 nm were displayed in CS-g and CS-cd samples,
which is in good agreement with Fig. 4a. A majority of pore
volume was associated with pores ranging from pore diameter of
2.3. Electrochemical measurements of the carbon spherules
Carbon samples (active material), carbon black (conductivity
enhancing material) and PTFE (60 wt% solution, binder) were mixed
in a mass ratio of 90:5:5 and dispersed in deionized water (the mass
ratio of carbon to water was set as 1:2). After homogenization in a
mortar, the slurry was rolled into a thin film of uniform thickness
(0.570.1 mm). From this film, 12 mm circular electrodes were
punched out and pressed onto nickel foam (as a current collector).
The carbon coated nickel foam was then immerged in a 30% (wt%)
KOH solution under vacuum for ca. 3 h before testing.
Electrochemical measurements (CV, EIS) were performed on a
electrochemical workstation CHI660A (Chenhua Instrument
Corp., China) at room temperature, using two-electrode electro-
chemical capacitor cells. Cyclic voltammograms were recorded
from 0 to 1 V at various sweep rates and the Nyquist plots were
recorded potentiostatically (0 V) by applying an alternating vol-
tage of 5 mV amplitude in the 100 kHz–1 mHz frequency range.
The constant-current discharge–charge tests were conducted in
the voltage range of 0–1 V with a Battery Tester (Land, Wuhan,
China) at different specific current.
2–5 nm. Detailed characteristics of the pore structure of all
samples are collected in Table 1. The results show that CS-s
2
ꢀ1
illustrates a maximum BET surface area of 510 m g
and total
3
ꢀ1
pore volume of 0.25 cm g
.
The electrode pairs from the three carbon samples were tested
in 30% KOH aqueous solution at room temperature. The charge/
discharge curves of carbon samples between the voltage range of
3
. Results and discussion
ꢀ
2
Fig. 1 shows SEM images of the products of
b
-cyclodextrin,
0.01–1 V at 1 mA cm
are displayed in Fig. 5 and the specific
sucrose and glucose after hydrothermal treatment (Fig. 1a, c and
e) and carbonization (Fig. 1b, d and f). As shown in Fig. 1, after
hydrothermal treatment at 180 1C microspheres with smooth
capacitances at different discharge density were studied as shown
in Table 2. The charge/discharge behavior of the carbon samples
in 30 wt% KOH solution is highly reversible. The discharge curves
are approximately linear and symmetric to their corresponding
charge curves. At the beginning of the discharge, no sharp change
in voltage is observed. That is, the equivalent series resistance
(ESR) of the EDLCs is rather low. As seen in Table 2, the
capacitance decrease with discharge current, which is mainly
because the potential difference between the mouth and the
bottom of the micropores increases with the current due to ohmic
resistance of the electrolyte in the axial direction of micropores
[18]. At high current density, electrolyte ions cannot move fast
enough to the bottom of the micropores, so the capacitance
declined. The maximum specific capacitance value was achieved
surface in a diameter of 0.5–1.5 mm range were obtained in three
samples. Mild shrinkage was detected after carbonization due to
the elimination of noncarbon atoms at high temperature and the
aggregation phenomenon was more obvious in glucose sample.
The aggregated CS-g particles are close to spheric in morphology,
albeit lacking a clear boundary around a particle. In our previous
work, hydrothermal reactions occurred in glucose and sucrose
solution at 150 1C, whereas no change was detected in
b-cyclo-
dextrin until the temperature was increased up to 180 1C. These
phenomena disclose that the larger the saccharide molecule is,
the higher the temperature is needed for the onset of

X. Xia et al. / Journal of Physics and Chemistry of Solids 73 (2012) 385–390
387
Fig. 1. SEM images of the (a) HS-cd, (c) HS-s, (e) HS-g and (b) CS-cd, (d) CS-s, (f) CS-g.
CS-cd
CS-s
CS-cd
CS-s
CS-g
CS-g
50
4
00
800
1200
Raman shift (cm )
1600
2000
1
0
20
30
2θ (Deg.)
40
-
1
Fig. 2. The Raman spectra of the carbon spherule samples.
Fig. 3. The XRD patterns of the carbon spherule samples.

3
88
X. Xia et al. / Journal of Physics and Chemistry of Solids 73 (2012) 385–390
with the sample of CS-s and the area specific capacitance of CS-s
is about 30 mF cm , which is higher than some other carbonac-
agent and the pore structure was not continuous porosity. So the
area specific capacitance is relatively low [21].
ꢀ
2
ꢀ
1
eous materials [19–21]. The high specific capacitance suggests
Fig. 6 shows cyclic voltammograms at the 2 mV s sweep rate
obtained for the capacitors using CS-cd, CS-s and CS-g as electro-
des, respectively. At the identical scan rate, the most ideal
capacitive behavior was observed for CS-s with a steepest current
change at the switching potential (0 and 1.0 V), resulting in a most
rectangular-shaped I–V curve. The faster change at the switching
potentials in the cyclic voltammogram of CS-s than CS-cd and CS-g
stems from the faster re-organization of the double layer in CS-s,
probably owing to faster ionic motions in its micropores. In
addition, higher capacitance was obtained from CS-s than CS-cd
that most of the micropore surface area in this carbon is
ꢀ
2
accessible at low current loads (1 mA cm ) to electrolyte ions
and double-layer formation. The reason may be that carbon
samples prepared by polycondensation and carbonization of
organic carbon precursor have more continuous micropores, like
carbon aerogel [22], so the accessibility of micropores was
enhanced and more efficient charge storage in EDLC was
achieved. However, pores in activated carbons were primarily
formed by the etching process from surface to inside of activating
ꢀ
1
ꢀ1
ꢀ1
and CS-g (173 F g , 156 F g
and 166 F g , respectively), sug-
gesting that more surface area of the carbon was accessed by the
electrolyte ions due to the higher specific area and more
1
1
1
1
1
80
60
40
20
00
1
1
0
0
0
0
0
.2
.0
.8
.6
.4
.2
.0
CS-cd
CS-s
CS-g
CS-cd
CS-s
CS-g
0
.0
0.2
0.4
0.6
0.8
1.0
Relative pressure (P/P0)
5
00
1000 1500 2000 2500 3000 3500
Time (s)
0
0
0
0
0
0
0
0
.07
.06
.05
.04
.03
.02
.01
.00
Fig. 5. The charge/discharge curves of carbon spherules between the voltage
ꢀ
2
range of 0.01–1 V at 1 mA cm
.
CS-cd
CS-s
CS-g
Table 2
The specific capacitances at the CS-s electrodes for electrolyte in 30% KOH aqueous
electrolyte.
Current density Specific capacitance of carbon spherules
ꢀ
2
(mA cm
)
CS-cd
CS-s
CS-g
ꢀ
1a
ꢀ3 b
ꢀ1a
ꢀ3 b
ꢀ1 a
ꢀ3 b
F cm
F g
F cm
F g
F cm
F g
1
3
5
8
145
125
115
105
102
93
147
127
116
106
103
94
164
154
148
144
143
138
134
170
159
153
148
147
143
138
162
151
142
137
132
127
116
173
161
152
146
141
135
124
10
10
100
1
5
0
2
92
93
Pore diameter (nm)
a
Fig. 4. Nitrogen sorption isotherms (a) and pore size distribution (b) of the carbon
Referred to the carbon mass in a single electrode.
Referred to the electrode volume.
b
spherule samples.
Table 1
The specific surface area and pore structures of carbon spherules.
Samples
BET surface
Microsurface
Pore volume
Micropore
Microporosity
(%)
Average pore
2
ꢀ1
2
ꢀ1
3
ꢀ1
3
ꢀ1
g )
area (m
g
)
area (m
g
)
(cm
g
)
volume (cm
diameter (nm)
CS-cd
CS-s
405.8
510.6
426.9
334.7
433.4
359.7
0.22
0.25
0.23
0.17
0.21
0.17
82.7
84.9
84.3
2.14
1.94
2.13
CS-g

X. Xia et al. / Journal of Physics and Chemistry of Solids 73 (2012) 385–390
389
0
0
0
.2
.1
.0
CS-cd
Fig. 8. Equivalent circuit for the impedance spectra of the carbon electrodes.
CS-s
CS-g
Table 3
Simulation results of equivalent circuit elements from the simulation of the
impedance data.
-
-
0.1
0.2
ꢀ
1
2
Electrode samples
R
s
(
O)
C
c
(
mF)
R
ct
(O
)
C
L
(F g
)
w
CS-cd
CS-s
2.8
2.5
2.1
469.1
0.939
0.265
0.504
147.3
173.4
161.4
3.54 ꢁ 10ꢀ
3
ꢀ3
1068.0
637.2
5.79 ꢁ 10
CS-g
8.97 ꢁ 10ꢀ
3
0.0
0.2
0.4
0.6
0.8
1.0
Potential (V)
Table 4
ꢀ
1
Fig. 6. Cyclic voltammograms at the 5 mV s
sweep rate using CS-cd, CS-s and
The error of equivalent circuit elements from the simulation of the impedance
data.
CS-g as electrodes, respectively.
Electrode
samples
Error (%)
9
8
7
6
5
4
3
2
1
0
0
0
0
0
0
0
0
0
0
R
s
W
R
ct
C
L
CS-cd
CS-s
CS-g
CS-cd
CS-s
1.164
1.473
1.905
4.643
7.24
5.299
2.512
2.923
3.698
18.94
11.24
CS-g
7.539
2
1
1
0
0
.0
.5
.0
.5
.0
capacitance, Rct the contact resistance, and a Warburg diffusion
element ( ) attributable to the diffusion of ions. Besides, the
capacitive nature of the carbon-based electrode in the low-
CS-cd
CS-s
CS-g
D
frequency domain could be reasonably presented as a low-
L
frequency capacitance, C . The simulation results of the impe-
dance data for different carbon electrodes using ZSimpWin soft-
ware and the errors of the parameters for corresponding
equivalent circuits from fitting EIS data are illustrated in
Tables 3 and 4, respectively.
The average error (
impedance spectra of the carbon electrode is relatively small,
2
-0.5
w
) of the fits in Table 3 for the different
2
.0
2.5
3.0
Z' (Ohm)
3.5
4.0
ꢀ
3
ꢀ3
ranging from 3.54 ꢁ 10
to 8.97 ꢁ 10
while the error of the
0
10
20
30
40
50
Z' (Ohm)
60
70
80
90
corresponding element is in the range of 1.164–18.94% (shown in
Table 4), which prove that the selected equivalent circuits are
proper for the investigated electrodes. From a comparison of Rct
Fig. 7. Nyquist plots using CS-cd, CS-s and CS-g as electrodes.
and C
L
of different carbon electrodes in Table 3, the ionic charge
is the biggest
transfer resistance of CS-s is the smallest while its C
L
reasonable pore distribution of CS-s, which is in agreement with
the result obtained from the galvanostatic study. (See Table 2).
Fig. 7 shows typical Nyquist plots in a frequency range
of 100 kHz–1 mHz for the capacitors with the three carbon spher-
ules. In the high frequency region, a depressed semicircle occurred,
indicating a parallel combination of resistive and capacitive compo-
nents. Electrolyte resistance can be estimated from the crossover
point of the highest frequency with the real part of impedance. The
high frequency loop is linked to the accessible electrode porosity to
the electrolyte. The loop diameter can be considered as the
resistance from the mass transport (i.e., transport of electrolyte ions
within micropores of porous carbon), and termed charge transport
resistance. CS-s has less semicircle than other samples (the inset in
Fig. 7), which indicates smaller charge transfer resistance.
among the carbon electrodes. That is to say, the CS-s electrode has
the fastest charge transfer within micropores and the biggest
capacitance, which is in good agreement with the results from the
cyclic voltammetric measurements and galvanostatic study.
4. Conclusion
Hydrothermal carbonization method without activation was
adopted to prepare nanoporous carbon microspheres, using
b-cyclodextrin, sucrose and glucose as carbon precursors. The
method was proved to be an effective process to produce carbons
2
ꢀ1
spherules with specific surface area (400–500 m g ) and a poros-
ity accessible to electrolyte ions (average micropore sizes around
1.9–2.1 nm). The electrochemical tests for these carbon spherules
were performed in 30 wt% KOH aqueous electrolyte. Among these
Several electrical circuits were estimated by nonlinear least
squares fitting of the experimental impedance data. The equiva-
lent circuit shown in Fig. 8 was found to obtain excellent fits
down to frequencies including the low-frequency vertical line, in
ꢀ
1
samples, CS-s exhibits an ideal electric capacity of 164 F g
at
ꢀ
1
1 mA cm
and good capacitance maintenance of 82% even at
ꢀ
1
ꢀ3
which
R
s
is the bulk solution resistance,
C
c
the contact
20 mA cm . A higher volumetric capacitance of 170 F cm
was

3
90
X. Xia et al. / Journal of Physics and Chemistry of Solids 73 (2012) 385–390
achieved due to more available surface area and high packing
density of the carbon spherule. This enhances the industrial poten-
tial of the carbon spheres for small electric power sources.
[6] Ch.X. Zhang, D.H. Long, B.L. Xing, et al., Electrochem. Commun. 10 (2008)
809–1811.
7] M.S. Balathanigaimani, W.G. Shima, M.J. Lee, et al., Electrochem. Commun. 10
2008) 868–871.
1
[
(
[
[
8] S.I. Lee, S. Mitani, Ch.W. Park, et al., J. Power Sour. 139 (2005) 379–383.
9] T. Nakamura, Y. Yamada, K. Yano, Microporous Mesoporous Mater 117 (2009)
478–485.
Acknowledgment
[
[
10] Q. Wang, H. Li, L. Chen, et al., Solid State Ionics 152–153 (2002) 43–50.
11] M. Sevilla, A.B. Fuertes, Carbon 47 (2009) 2281–2289.
Financial support by the National Natural Science Foundation
of China (no. 50872032) and the Fundamental Research Funds for
the Central Universities (no.531107040185).
[12] Sh. Saka, T. Ueno, Cellulose 6 (1999) 177–191.
[
[
[
[
13] A.C. Ferrari, J. Robertson, Phys. Rev. B 61 (2000) 14095–14107.
14] J. Schwan, S. Ulrico, V. Batori, H. Ehrhardt, J. Appl. Phys. 80 (1996) 440–447.
15] R. Xue, H. Liu, P.P. Wang, Z. Shen, Carbon 47 (2008) 318–320.
16] Q. Wang, X.Y. Liang, W.M. Qiao, et al., Fuel Process. Technol. 90 (2009)
3
81–387.
References
[
[
17] J.S. Ima, S.J. Parkb, Y.S. Lee, J. Colloid Interface Sci. 314 (2007) 32–37.
18] Ch.T. Hsieh, H. Teng, Carbon 40 (2002) 667–674.
[
1] B.E. Conway, Electrochemical Supercapacitors: Scientific Fundamentals and
Technological Applications, Kluwer Academic/Plenum Publishers, New York,
[19] M. Olivares-Mar ı´ n, J.A. Fern a´ ndez, M.J. L a´ zaro, Mater. Chem. Phys. 114 (2009)
323–327.
1
999.
[20] A. Bagreev, J.A. Menendez, I. Dukhno, Y. Tarasenko, T.J. Bandosz, Carbon 42
(2004) 469–476.
[21] K. Jurewicza, R. Pietrzakb, P. Nowicki, H. Wachowska, Electrochim. Acta. 53
(2006) 5469–5475.
[
[
[
[
2] J. Chmiola, G. Yushin, Y. Gogotsi, et al., Science 313 (2006) 1760–1763.
3] A.S. Arico, P. Bruce, B. Scrosati, et al., Nat. Mater. 4 (2005) 366–377.
4] D.W. Wang, F. Li, M. Liu, et al., Angew. Chem. Int. Ed. 47 (2008) 373–376.
5] W. Li, D. Chen, Z. Li, et al., Electrochem. Commun. 9 (2007) 569–573.
[22] S.W. Hwang, S.H. Hyun, J. Non-Cryst. Solids 347 (2004) 238–245.