A Hybrid Activated Carbon-Manganese Dioxide Capacitor using a Mild Aqueous Electrolyte

Article abstract of DOI:10.1149/1.1650835

A hybrid electrochemical capacitor using MnO₂ and activated carbon (AC) as positive and negative electrodes, respectively, has been designed. The electrodes were individually tested in a mild aqueous electrolyte (0.65 M K₂SO₄) in order to define the adequate balance of active material in the capacitor as well as the working voltage. The hybrid electrochemical capacitor was cycled between 0 and 2.2 V for over 10,000 constant current charge/discharge cycles. A real energy density of 10 Wh/kg was reproducibly measured with a real power density reaching 3600 W/kg. The hydrogen and oxygen evolution reactions on AC and MnO₂ electrodes, respectively, were investigated in 0.65 M K₂SO₄. Despite the good electrochemical performance of the 2.2 V capacitor, gas evolution could be a hindrance for practical use. Subsequently, a 1.5 V capacitor was tested for more than 23,000 cycles and yielded interesting electrochemical performance with negligible gas evolution.

Full text of DOI:10.1149/1.1650835

A614
Journal of The Electrochemical Society, 151 ͑4͒ A614-A622 ͑2004͒
0013-4651/2004/151͑4͒/A614/9/$7.00 © The Electrochemical Society, Inc.
A Hybrid Activated Carbon-Manganese Dioxide Capacitor
using a Mild Aqueous Electrolyte
a
a,z,
Thierry Brousse,a,b, Mathieu Toupin, and Daniel Belanger
´
*
*
a
´
´
´
´
´
`
´
Laboratoire d’Electrochimie Appliquee, Departement de Chimie, Universite du Quebec a Montreal,
´
succursale Centre-Ville, Montreal, Quebec H3C 3P8, Canada
b
´
´
´
Laboratoire de Genie des Materiaux, Ecole Polytechnique de l’Universite de Nantes,
44306 Nantes Cedex 3, France
A hybrid electrochemical capacitor using MnO2 and activated carbon ͑AC͒ as positive and negative electrodes, respectively, has
been designed. The electrodes were individually tested in a mild aqueous electrolyte ͑0.65 M K2SO4) in order to define the
adequate balance of active material in the capacitor as well as the working voltage. The hybrid electrochemical capacitor was
cycled between 0 and 2.2 V for over 10,000 constant current charge/discharge cycles. A real energy density of 10 Wh/kg was
reproducibly measured with a real power density reaching 3600 W/kg. The hydrogen and oxygen evolution reactions on AC and
MnO2 electrodes, respectively, were investigated in 0.65 M K2SO4 . Despite the good electrochemical performance of the 2.2 V
capacitor, gas evolution could be a hindrance for practical use. Subsequently, a 1.5 V capacitor was tested for more than 23,000
cycles and yielded interesting electrochemical performance with negligible gas evolution.
© 2004 The Electrochemical Society. ͓DOI: 10.1149/1.1650835͔ All rights reserved.
Manuscript submitted March 6, 2003; revised manuscript received September 15, 2003. Available electronically March 1, 2004.
F/g were reported in NaCl or Na2SO4 ,16,17 but in these media, the
conductivity of the electrolyte was smaller than for H2SO4 or KOH
solutions.18
Recently, MnO2 has been proposed as an active electrode mate-
rial for electrochemical capacitors.19-24 It was shown that this com-
pound exhibited a pseudocapacitive behavior with specific capaci-
tance as high as 600 F/g for thin films19-21 and 100-200 F/g for
Nowadays, research on electrochemical capacitors aims to in-
crease both power and energy densities as well as lower fabrication
costs while using environmentally friendly materials. These goals
can be achieved by different approaches including the improvement
of each component of the cell, e.g., negative and positive electrodes,
formulation and composition of the electrolyte, and design of cur-
rent collectors.1-2
Two obvious approaches to increase the power density of an
electrochemical capacitor are to use larger capacitance materials and
larger cell voltages. The latter can be achieved with nonaqueous
electrolytes that are characterized by a wider electrochemical stabil-
ity window relative to aqueous electrolytes. Thus, the combination
of a solvent such as acetonitrile3 or propylene carbonate4 with dif-
ferent salts ͑e.g., tetraethylammonium tetrafluoroborate,5 tetraethy-
lammonium trifluoromethylsulfonate,6 and tetraethylammonium
methylsulfonate7͒ enable the increase of the maximum cell potential,
which can reach more than 3 V,3-6 a value three times the maximum
cell voltage of aqueous-based capacitors. However, despite the re-
markable performance of these organic-based systems, they suffer
from the use of highly toxic and/or flammable solvents which can
cause severe safety hazards. The fabrication costs of capacitors us-
ing such electrolytes are also high due to the use of a water free
environment needed to manipulate and assemble the components of
the capacitor.
In aqueous-based systems, the relatively small voltage range can
be partially compensated by the use of high capacitance materials
such as RuO21,8 which can reach 863 F/g over a 1 V potential range
in 1 M H2SO4 aqueous electrolyte.9 The voltage window can be
even remarkably extended to 1.4 V under some conditions,10-11 thus
enabling higher energy and power densities from the capacitor. AC
electrodes have also been used in aqueous H2SO4 ͑3 M and 10 M͒
and KOH ͑6 M͒ electrolytes but their capacitances are in the range
50-250 F/g12,13 over a 1 V potential range and therefore much
smaller than that of ruthenium oxide-based systems. Although con-
centrated acid or alkaline media are currently used in common elec-
trochemical systems such as lead acid batteries14 or alkaline primary
batteries,15 it might be interesting to use milder aqueous systems
based on safe sodium or potassium salts. In these cases, the specific
capacitance of AC was found to be somewhat smaller in neutral
media than in acidic or alkaline media. Values in the range of 50-75
powder-based electrodes22-24 within a potential window of 0.9-1.2 V
in aqueous electrolytes containing KCl, K2SO4 , or Na2SO4 . Unfor-
tunately, the potential window is relatively small compared to
organic-based capacitors and consequently a MnO2-based electro-
chemical capacitor will deliver a limited power density. An attrac-
tive approach to get around this problem is to use a hybrid system
which will rely on two different electrode materials that have a
different potential range of electroactivity.25-26 This was recently
investigated with carbon25 or conducting polymers26 as the positive
electrode and an intercalation oxide such as Li4Ti5O12 as the nega-
tive electrode25,26 in the presence of nonaqueous electrolytes. In
addition, the same approach can be used with aqueous electrolytes
and very recently a hybrid system with a positive MnO2-based elec-
trode and a carbon negative electrode was investigated.27 An oper-
ating voltage of 2 V was demonstrated for this hybrid capacitor in a
neutral KCl aqueous electrolyte. However, the cycling stability was
only shown for 100 cycles and the problems related to gas evolution
at the electrodes were recognized but not addressed in detail.
In this study, we investigate the electrochemical behavior of a
hybrid capacitor using AC as the negative electrode and a composite
MnO2 electrode as the positive electrode in a mild aqueous electro-
lyte ͑0.65 M K2SO4 , pH 6.5͒. First, we report the performance of
this hybrid system upon charge/discharge cycling over a large po-
tential window. Second, the hydrogen evolution reaction ͑HER͒ on
the AC electrode and the oxygen evolution reaction ͑OER͒ on MnO2
are investigated in these experimental conditions. Third, we present
the results of a prototype cell with a smaller but useful potential
window of 1.5 V over which no noticeable gas evolution was ob-
served during 23,000 cycles.
Experimental
Manganese dioxide synthesis and characterization.—The MnO2
powder was synthesized via an adapted sol-gel route.28-29 A 0.2 M
KMnO4 solution ͑Aldrich 99.99%͒ was prepared first. While stir-
ring, fumaric acid (C4H4O4 , Aldrich͒ was added to 100 mL KMnO4
in a 1:3 molar ratio. The color of the solution immediately turned
dark brown. The resulting sol was degased for 30 min under mod-
erate vacuum and after a few hours, a black gel formed. It was then
* Electrochemical Society Active Member.
z E-mail: belanger.daniel@uqam.ca
Downloaded on 2013-11-04 to IP 150.108.161.71 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

Journal of The Electrochemical Society, 151 ͑4͒ A614-A622 ͑2004͒
A615
washed carefully five times with each of the following solutions or
solvents in the following order: 0.1 M H2SO4 , deionized ͑DI͒ water,
acetone, and hexane. The gel was dried in air for another 24 h. The
resulting dense black powder was ground in an agate mortar.
For the quantitative determinations of K and Mn, about 0.1 g of
powder was dissolved in 6 M H2SO4 . Subsequent heating and stir-
ring were performed for 2 h until complete dissolution of the pow-
der was achieved. The solution was then diluted to 1000 mL with DI
water. The total amount of K and Mn in the powder was first deter-
mined by atomic absorption spectroscopy ͑AAS͒. The amount of
Mn͑IV͒ in the sample was determined by dissolving 0.1 g of powder
in 6 M H2SO4 containing a known amount of Fe2ϩ ions. Mn͑IV͒
ions from the sample were reacted with Fe͑II͒. Finally, the excess of
Fe2ϩ ions were titrated with an aqueous KMnO4 solution. Mn3ϩ was
C ϭ Q/(⌬Em) . The specific capacitance of the active material
͓
͔
͑AC or MnO2) was calculated by dividing the specific capacitance
of the composite electrode by the weight percentage of active ma-
terial in this electrode. The influence of acetylene black ͑specific
capacitance: 12 F/g͒ was neglected in this last calculation.
Hybrid capacitor.—For the hybrid capacitor, the weight ratio of
active materials (MnO2 /AC) was 1.15 for cells operating within a
2.2 V potential range and 0.77 for cells operating within a 1.5 V
potential range. The MnO2 /AC ratios were calculated by consider-
ing the specific voltammetric charge and the appropriate potential
windows of each materials. Typical surfaces were 0.3 to 1 cm2 for
the electrodes. The AC composite electrode was separated from the
MnO2 composite electrode by a glass paper fiber wetted with a 0.65
M K2SO4 solution. The three layers were then pressed in between
two PTFE blocks (1 ϫ 2 cm͒ by the mean of plastic clamps. The
force used to press the layers was not measured. The complete cell
was immersed in 30 mL of 0.65 M K2SO4 in a sealed glass vial
without contact to ambient air. The cell was not thermostated, how-
ever the ambient temperature was between 293 and 295 K.
The real power density, Preal , and the real energy, Ereal , were
determined from the constant current charge/discharge cycles as fol-
lows
obtained as the difference between the total Mn amount and Mn4ϩ
.
The remaining Mn3ϩ ions were assigned to MnOOH which then
allowed the determination of the amount of hydrogen in the com-
pound.
The crystallographic structure of the powder was investigated
using a Siemens D-5000 X-ray diffractometer ͑Co K␣ radiation,
0.179026 nm͒ in a ␪/2␪ geometry. The microstructures of the
samples were observed with a Leica Stereoscan 440 scanning elec-
tron microscope ͑SEM͒ coupled with an energy dispersive X-ray
analyzer ͑Oxford Instruments͒ for semiquantitative analysis. Single
point Brunauer-Emmett-Teller ͑BET͒ surface area measurements
were made with a Micromeritic Flowsorb II/2300 surface area ana-
lyzer using N2 gas.
Preal ϭ ⌬E I/m
W/kg͒
͓1͔
͑
where ⌬E ϭ (Emax ϩ Emin)/2 with Emax the potential at the end of
charge and Emin at the end of discharge, I is the applied current ͑A͒,
and m is the weight of active material in the electrode ͑kg͒
AC.—The carbonaceous material ͑IRH-26͒ was prepared at IRH-
UQTR by pyrolysis of a coconut shell precursor followed by physi-
cal activation in a rotary kiln at 950°C under CO2 atmosphere.30
Nitrogen sorption measurements at 77 K were performed using a
Quantachrome Autosorb apparatus to calculate the BET surface area
and analyze the porous structure. The IRH-26 AC has a high surface
area of ϳ2800 m2/g. The microporous ͑2-20 Å͒ volume calculated
by the Dubinin Raduskevitch method is 1.1 cm3/g, and the mesopo-
rous ͑20-500 Å͒ volume calculated by the Barett-Joyner-Halenda
method is 1.5 cm3/g. The AC has a bulk density of 0.22 g/cm3, a pH
of 10-11 and an atomic concentration as evaluated by X-ray photo-
electron spectroscopy of 94-95 atom %C and 5-6 atom % O.
Ereal ϭ Preal
t
Wh/kg͒
͓2͔
͑
where t is the discharge time ͑h͒. Finally, the coulombic efficiency is
given by the discharge time/charge time ratio.
Alternatively, the maximum specific power, Pmax , was calculated
from
Pmax
ϭ
U2͒/ 4Rm͒
W/kg͒
͑
͓3͔
͑
͑
0
where U0 is the potential at the beginning of discharge ͑after the
ohmic drop͒ and R is the internal resistance measured at 5 Hz ͑in ⍀͒.
The maximum specific energy was defined as
Preparation of the electrodes.—Composite electrodes were pre-
pared by mixing the active material with graphite ͑Alfa Aesar, con-
ducting grade, Ϫ200 mesh͒, acetylene black ͑Alfa Aesar, Ͼ99.9%,
surface area ϭ 80 m2/g͒, and polytetrafluoroethylene ͑PTFE͒ dried
powder in the following ratios: 70/12.5/12.5/5 for the MnO2 elec-
trode and 90/5/0/5 for the AC electrode. The mixtures thus prepared
were cold rolled into 100 ␮m thick films. Typical loads of MnO2
and AC active materials were, respectively, 6.2 and 4.5 mg/cm2 of
film. Pieces of film, typically 7 ϫ 6 mm size were then pressed at
900 MPa on a titanium grid or alternatively on a stainless steel ͑SS͒
grid. The grid was connected to the potentiostat via a copper wire.
The connection between the grid and the copper wire was protected
from the electrolyte with epoxy resin.
Electrochemical tests were performed with a Solartron 1470 bat-
tery tester operated under Corrware II software ͑Scribner Associ-
ates͒. A Hg/Hg2SO4 ͑saturated K2SO4) or Ag/AgCl ͑saturated NaCl͒
assembly and a platinum gauze were used as reference and counter
electrodes, respectively. In this study, all electrodes were tested in
0.1 M or 0.65 M K2SO4 solutions that were prepared by dissolving
the appropriate amount of salt (K2SO4 , Anachemia ACS͒ in DI
water without adjusting the pH. The pH of the electrolyte was
measured at the beginning of the experiment and found to be equal
to 6.5. K2SO4 was chosen because of its higher ionic conductivity
compared to Na2SO4 .18 The specific capacitance, C ͑F/g͒ of given
composite electrodes was determined by integrating the cyclic
voltammogram ͑CV͒ curve to obtain the voltammetric charge
͑Q͒, and subsequently dividing this charge by the mass of the com-
posite electrode ͑m͒ and the width of the potential window (⌬E)
2
Emax
ϭ
C U
͑
͒
/2
Wh/kg͒
͑
͓4͔
͓
͔
max
where C is the system capacity ͑F͒ and Umax the potential at the end
of charge.
Gas evolution.—In the MnO2/carbon hybrid capacitor, the OER
occurs on the MnO2 surface while HER occurs on the AC surface
upon charging the hybrid supercapacitor. In this work, the anodic
and cathodic currents related to the OER and HER, respectively,
were measured and the volume of hydrogen and oxygen were esti-
mated from the graphical representation of the Tafel plots. This was
done by testing each electrode individually. The details of the cal-
culations can be found in Appendices A and B. To avoid the influ-
ence of the current collector, the titanium grid was covered on one
side by the active material as described above and on the other side
by a PTFE foil. Therefore, the current collector was not in direct
contact with the electrolyte. Each electrode was tested in a three
electrode cell using platinum gauze as the counter electrode and
Ag/AgCl as the reference electrode. The current was measured by
cyclic voltammetry at sweep rates of 2, 5, 10, 20, and 40 mV/s.
Results and Discussion
Compositional, structural, and microstructural characterization
of the active electrode materials.—The AAS analysis revealed that
the K/Mn atomic ratio was 0.004, indicating that acid washing is
efficient in removing nearly all the potassium ions from the gel. By
neglecting the remaining potassium ions, the resulting formula of

A616
Journal of The Electrochemical Society, 151 ͑4͒ A614-A622 ͑2004͒
Figure 1. X-ray diffraction pattern of the MnO2 xerogel powder prepared by
sol-gel synthesis ͑Co K␣ radiation, 0.179026 nm, ␪/2␪ geometry͒.
the material was found to be: MnO2H0.2•0.4 H2O. EDX analysis
coupled with the SEM confirmed the presence of trace amounts of
potassium as well as sulfur as impurities. Structural investigation
͑Fig. 1͒ indicated that the MnO2 gel was poorly crystallized. The
crystallite size estimated according to Scherrer equation31 is in the
range 4-6 nm. The more intense reflection at 43.34° corresponds to
0.242 nm, e.g., the ͑100͒ peak of the hexagonal Birnessite structure.
Additional reflections at low angles ͑13.9° and 28.6°͒ also suggest a
layered-type structure. Subsequently, it seems that the gel was
poorly organized and was a precursor of the layered Birnessite struc-
ture. However, high resolution electron microscopy will be needed
to confirm this hypothesis. SEM observations showed that the MnO2
powder was made of agglomerated small grains 20 to 500 nm wide
͑Fig. 2a͒. The surface area determined from BET measurements is
200 Ϯ 30 m2/g and is in good agreement with the values reported
previously.28 On the other hand, the AC is characterized by larger
grains which can reach several tens of ␮m ͑Fig. 2b͒ mixed with
nanosized particles agglomerated in micrometer-size clusters.
Figure 2. Scanning electron micrographs of ͑a͒ MnO2 xerogel powder
͑white bar on the bottom left corner is 200 nm͒ and ͑b͒ AC powder ͑white
bar on the bottom left corner is 2 ␮m͒.
Electrochemical characterizations of AC and MnO2 composite
electrodes in 0.65 M K2SO4.—Figure 3 shows a CV of a composite
MnO2 electrode ͑curve a͒. The maximum stability window of
the MnO2 electrode is between 0 and 1.1 V vs. Ag/AgCl in aqueous
0.65 M K2SO4 . Within this potential window, the shape of the CV
is pseudorectangular as expected for material displaying capacitive
behavior. The capacitance value determined from the CV at 5 mV/s
is 105 Ϯ 10 F/g for the composite electrode ͑including graphite,
carbon black, and PTFE͒, within a potential range 0.0/1.0 V vs.
Ag/AgCl. Subsequently, the specific capacitance of MnO2 is 150
Ϯ 10 F/g. This last value compares well with those recently re-
ported for composite electrodes when only the weight of active ma-
terial is considered.22-24,27
A CV was recorded in the same conditions with an AC electrode
͑Fig. 3, curve b͒. The range of the electrochemical window over
which a capacitive behavior is observed is shifted toward more
negative potentials compared to MnO2 , and is limited by the onset
of the HER occurring at a very low potential ͑Ϫ1.3 V vs. Ag/AgCl͒
and OER occurring above 0.5 V on the carbon electrode. The ca-
pacitance of the AC composite electrode ͑containing 90% of AC͒
calculated from the voltammetric charge is 108 Ϯ 10 F/g within a
potential range Ϫ1.1/ϩ0.2 V vs. Ag/AgCl. Therefore, the capaci-
tance of AC ͑at 5 mV/s͒ is 120 Ϯ 10 F/g. Despite a slight difference
in the capacitance of each active material ͑150 F/g for MnO2 vs. 120
F/g for AC͒, the capacitances of the composite electrodes are almost
identical ͑105 and 108 F/g, respectively͒ due to different percent-
ages of active material.
The contribution of the titanium grid ͑current collector͒ to the
CV of the AC and MnO2 electrodes is illustrated on Fig. 3 ͑curve c͒.
On the surface of the current collector, the onset of H2 or O2 evo-
lution is observed at Ϫ1.3 and 1.4 V, respectively. Therefore, it
seems possible to combine AC and MnO2 to obtain a capacitor with
a potential window close to 2.2 V without a significant decomposi-
tion of the aqueous solvent during a charge/discharge cycle. A simi-
lar conclusion was reached by Hong et al. for their MnO2/carbon
capacitor in 1 M KCl but their operating capacitor voltage was re-
stricted to a slightly smaller value of 2 V.27 However, it will be
shown below that cycling over the aforementioned potential ranges
led to the generation of a non-negligible amount of gas, but the
determination of a more operational potential window will be de-
tailed in the second part of this study. It can be noted that the AC
and MnO2 composite electrodes have nearly the same capacitance
and that there is a significant overlap of their CVs between Ϫ0.1 and
0.5 V. This will be helpful to correctly balance the charge of each
electrodes of the hybrid capacitor.

Journal of The Electrochemical Society, 151 ͑4͒ A614-A622 ͑2004͒
A617
Figure 3. CV of ͑a͒ composite MnO2 electrode ͑70 wt % MnO2), ͑b͒ com-
posite AC electrode ͑90 wt % AC͒, and ͑c͒ titanium grid ͑current collector͒.
Scan rate: 5 mV/s, Electrolyte: 0.65 M K2SO4 .
Electrochemical characterization of
a
hybrid AC/MnO2
capacitor in 0.65 M K2SO4.—The hybrid cell was built according to
the procedure described in the Experimental section. The evolution
of the cell potential during charge/discharge cycling at 0.55 A/g
͑taking into account only the total weight of active materials of the
electrodes͒ is shown on Fig. 4. After a few tens of cycles, the po-
tential limits of both electrodes stabilized and the system can be
operated for hundreds of cycles without significant changes in the
MnO2 and AC cutoff voltages. During charging between 0 and 2.2
V, the potential of the MnO2 electrode increased from Ϫ0.1 to 0.92
V vs. Ag/AgCl. At the same time, the potential of the negative elec-
trode decreased from Ϫ0.1 to Ϫ1.28 V vs. Ag/AgCl. For both MnO2
and AC electrodes of the capacitor, faradaic processes such as dis-
solution of Mn2ϩ, OER and HER that occurred at their correspond-
ing potential limits should be avoided. The charge/discharge curves
of the cell ͑Fig. 4͒ is not perfectly linear as for AC capacitors.12-13
Nonetheless, the average specific capacitance can be calculated for
this capacitor from C ϭ I/(dV/dt), where dV/dt is the slope of the
linear part of the discharge curve ͑Fig. 4͒ and I is the absolute value
of the current. The value was found to be 0.160 F which corresponds
to 29 F per gram of active material.
Figure 5. Galvanostatic charge/discharge cycles of the hybrid capacitor in
0.65 M K2SO4 at various constant currents, ͑a͒ 0.18 and 0.55 A/g, and ͑b͒
1.09, 2.18 and 3.27 A/g.
Figures 5a and 5b show the charge/discharge behavior of the
cell between 0 and 2.2 V for various currents ͑0.18 to 3.27 A per g
of active material͒. As mentioned above, the curves are almost linear
and the sudden drop of potential at the beginning of charge and
discharge is associated to the ohmic resistance of the cell. The
average internal resistance of the capacitor was evaluated at 5 Hz
͑0.2 s between two measurements͒ and calculated according to
R ϭ (EϩI Ϫ EϪI)/(2I) where EϩI Ϫ EϪI are, respectively, the cell
voltage at the end of the charge (EϩI ϭ last voltage point recorded
before the charge current is reversed͒ and at the beginning of the
discharge (EϪI ϭ first voltage point after the discharge current is
applied͒, and I is absolute value of the current which was the same
during the charge and the discharge. This gives an average value of
14 ⍀ and by taking into account the surface of the largest electrode,
the internal resistance can be estimated to be 24 ⍀/cm2. This resis-
tance reflects the nonoptimized contacts between the electrodes and
the current collectors, the solution resistance, as well as the presence
of a thick glass paper fiber used as separator. The contribution of the
latter would need to be reduced significantly in a practical device.
The discharge curves of Fig. 5a and b are linear for periods of times
Figure 4. Galvanostatic charge/discharge cycle (I ϭ 0.55 A/g͒ of the hybrid
capacitor in 0.65 M K2SO4 ͑right voltage axis͒ and variation of the potential
of each individual electrode ͑vs. Ag/AgCl͒.

A618
Journal of The Electrochemical Society, 151 ͑4͒ A614-A622 ͑2004͒
provide an estimation of the performance of the hybrid capacitor.
The open dots are related to titanium current collectors and the black
dots to SS current collectors but no significant effect of the current
collector was observed. The small variation in the performance of
the seven capacitors reflects the difference in the fabrication of the
laboratory cells. It is obvious that the reproducibility of the fabrica-
tion process would need to be enhanced for the large-scale produc-
tion of prototypes. The data of Fig. 7 clearly demonstrates that the
MnO2 /AC hybrid system in a mild aqueous electrolyte can easily
1-2
compete with systems using RuO2 or organic electrolyte-based
capacitors.3-7 For example, the real energy density of the MnO2 /AC
capacitor was 19 Wh/kg ͑calculated from Eq. 2͒ at a power density
of 200 W/kg ͑calculated from Eq. 1͒ for cutoff voltages of 0-2.2 V.
In this case, the maximum energy density ͑calculated from Eq. 4͒
was close to the real energy density ͑20 Wh/kg instead of 19 Wh/
kg͒, but the maximum power density was as high as 19 kW/kg
͑calculated from Eq. 3͒. While decreasing the real energy density to
10 Wh/kg ͑maximum energy density of 15 Wh/kg͒, the real power
density increased up to 3.6 kW/kg ͑maximum power density of 30
kW/kg͒.
The cycling stability upon charge/discharge cycling was tested
over several thousand cycles and the variation of the energy density
for a representative hybrid capacitor is illustrated in Fig. 8a. The cell
using electrodes built with titanium current collectors was cycled for
more than 10 000 cycles. It exhibited a 5% loss in energy density
during the first 400 cycles. After this stage, a continuous decrease
was observed over the 10000 cycles and the remaining energy den-
sity was still 60% of the energy density of the 400th cycle. These
results are promising because they show good cycling behavior for
the MnO2/carbon hybrid electrochemical capacitors. On the other
hand, the cycle lifetime of capacitors using SS as current collectors
for the electrodes is limited relative to Ti current collectors because
a drastic failure was observed after 5000 cycles. Postmortem obser-
vations usually revealed corrosion problems that need to be investi-
gated in more detail and are beyond the scope of this work.
Figure 8b shows the variation of the coulombic efficiency ͑time
of discharge/time of charge͒ of the capacitor with the number of
cycles performed at a constant power density of 1.2 kW/kg. The
coulombic efficiency remained close to 100% despite a slight and
continuous fade of the energy density. However, such a high cou-
lombic efficiency is surprising in light of the observation of gas
evolution. Thus, when the capacitor was fully charged, the HER is
likely to occur on the carbon electrode but the OER did not occur
because the positive MnO2 electrode had not reached a sufficient
positive potential ͑vide supra͒. On the other hand, for the discharged
capacitor, a faradaic reaction occurred on the MnO2 electrode ͑dis-
solution, see below͒ but no gas evolution should be observed on the
carbon electrode. Thus, the coulombic efficiency of 100% indicates
that the charge involved into these two faradaic processes is very
similar in our experimental conditions. These observations for the
charged and discharged capacitor suggest that some caution should
be exercised with the coulombic efficiency data and that sometimes,
a value close to 100% might be misleading.
Figure 6. Galvanostatic charge/discharge cycles of the hybrid capacitor in
0.65 M K2SO4 for various upper cutoff voltages ͑2.2 to 2.7 V͒ at a current of
1.09 A/g.
ranging from 10 to 350 s. This range corresponds to that typically
expected for pulse batteries and therefore the hybrid capacitor cell
presented in this study can be useful for such application. The ca-
pacitance calculated from the discharge curves decreased from 30
F/g ͑at 0.18 A/g͒ to 22 F/g ͑at 1.09 A/g͒.
The capacitors were also characterized for various upper limit
cutoff voltages as illustrated in Fig. 6. Interestingly, the cell main-
tained the same linear discharge slope when titanium was used as a
current collector. Additionally, the coulombic efficiency remained
high and only decreased from 100% at 2.2 V to 97% at 2.7 V. The
lower coulombic efficiency is related to the noticeable gas evolution
that is occurring on both electrodes. This phenomenon will be dis-
cussed in detail below.
Figure 7 presents Ragone plots for seven hybrid cells tested with
cutoff voltages 0-2.2 V, using SS and titanium grids as current col-
lectors. Because the packaging is not taken into account and all the
cell components are not yet optimized, these Ragone plots only
The energy loss, mentioned above, during constant current cy-
cling is due in part to a loss of active material during the discharge.
Because AC has demonstrated very high reversibility in aqueous
media during the capacitive charge storage processes, it seems that
the origin of the loss of active material should be attributed to the
reduction of Mn4ϩ to soluble Mn2ϩ species.32 Therefore, less active
material is available for the next charge and the charge stored de-
creased. The loss of MnO2 is demonstrated by the coloration of the
solution ͑brownish color͒ upon cycling a MnO2 electrode in a three
electrode configuration when the negative potential limit was set
more negative than Ϫ0.05 V vs. Ag/AgCl. Such color change was
not observed around the electrodes of the hybrid capacitor soaked in
the electrolytic solution, but when the hybrid cell was disassembled,
brownish spots were clearly visible on the inner side of the Teflon
holder as well as on the separator.
Figure 7. Ragone plot of different hybrid capacitors. Open dots: Titanium
current collector, Full dots: SS current collectors. All the samples were tested
in the same conditions using 0-2.2 V cutoff voltages. The current range from
0.18 to 3.27 A/g.

Journal of The Electrochemical Society, 151 ͑4͒ A614-A622 ͑2004͒
A619
Figure 9. Tafel plots for ͑a͒ the a MnO2 composite electrode ͑scan rate 2
mV/s͒ and ͑b͒ an AC composite electrode ͑scan rate 2 mV/s͒ in 0.65 M
K2SO4 . Curves i ͑full line are the actual data obtained for each electrode͒
and curves ii ͑dashed line͒ represent the contributions of the OER ͑Fig. 9a͒
and the HER ͑Fig. 9b͒ to the anodic and cathodic current.
Figure 8. ͑a͒ Variation of the energy density with the number of cycles for
a hybrid capacitor ͑titanium current collectors͒. Constant cycling power: 1.2
kW/kg, using an upper cutoff voltage of 2.2 V. ͑b͒ Plot of the coulombic
efficiency vs. the number of cycles for the capacitor cycled at 1.2 kW/kg.
The coulombic efficiency is the ratio: time of discharge/time of charge. For
clarity, only the data obtained at constant power density are presented even if
the capacitor was submitted to different power discharge regimes during the
tests.
oxygen were estimated from the graphical representation of the
Tafel plots ͑see experimental section and Appendices A and B͒.
Figure 9a presents a representative log current/potential plot for
the MnO2 electrode that is characterized by two linear regions. The
current for the low potential region between 0 and 1.15 V corre-
sponds predominantly to the charge storage process in MnO2 . When
the potential was increased to values more positive than about 1.25
V a significant increase of the current was observed and is attributed
in part to the OER. The latter can be roughly estimated by substract-
ing the steady-state current related to the charge storage mechanism
͑constant part of curve i on Fig. 9a͒ to curve i on Fig. 9a. The
resulting OER current is given on curve ii on Fig. 9a. At high po-
tential, above 1.4 V vs. Ag/AgCl, curve ii deviates from linearity due
to mass transport limitations. In the kinetically limited region be-
tween 1.2 and 1.3 V vs. Ag/AgCl, curve ii exhibits a linear region
where the current Ia can be expressed by
Gas evolution.—As it was mentioned above, it seems that gas
evolution occurred upon charge/discharge cycling and accordingly it
seems important to estimate the volume of gas being generated.
When the electrode potential becomes more positive or negative
than the equilibrium potentials related to solvent oxidation or reduc-
tion, respectively, then oxygen ͑Reaction 5͒ or hydrogen ͑Reaction
6͒ evolution will take place
2H2O → O2 ϩ 4Hϩ ϩ 4eϪ or 4OHϪ → O2 ϩ 2H2O ϩ 4eϪ
͓5͔
2Hϩ ϩ 2eϪ → H2 or 2H2O ϩ 2eϪ → H2 ϩ 2OHϪ ͓6͔
log Ia ϭ AaE ϩ Ba
͓7͔
In the MnO2/carbon hybrid capacitor, the OER occurs on the MnO2
surface while Reaction 6 occurs at the activated carbon surface. In
this work, the anodic and cathodic currents, respectively, related to
Reactions 5 and 6 were measured and the volume of hydrogen and
where E is the electrode potential and Aa and Ba are two constants
determined from the Tafel lines.

A620
Journal of The Electrochemical Society, 151 ͑4͒ A614-A622 ͑2004͒
The analysis of the CVs recorded at different sweep rates yielded
Aa and Ba constants close to those determined from the slowest
scan rate ͑2 mV/s͒, i.e., when the electrode was near the quasi-
stationary state, and the average values Aa ϭ 7.3 Ϯ 1.6 VϪ1 and
Ba ϭ Ϫ12.3 Ϯ 2.5 were evaluated. Based on these values, the an-
odic part of the Butler-Volmer equation can be determined
j ϭ j exp ␩/␤ ͒
͓8͔
͑
a
0
a
where ja is the current density (Ia divided by the surface of
the electrode͒ and j0 is the exchange current density for Reaction 5
on the MnO2 electrode in 0.65 M K2SO4 ͑pH 6.5͒, ␤a is the anodic
Tafel coefficient, and ␩ is the overpotential defined as ␩ ϭ E
Ϫ Erev where E is the potential of the electrode and Erev is the
equilibrium potential. However, it must be taken into account that
Eq. 8 will overestimate the anodic current due to O2 evolution for
low values of ␩.33 Similarly, the OER current will be overestimated
at high overpotential, because the Tafel lines diverge from linearity
due to mass-transport limitations. Nonetheless, the expression of the
anodic current calculated from the values of A and B and the surface
of the MnO2 electrode is
Figure 10. Galvanostatic charge/discharge cycle (I ϭ 0.53 A/g͒ of the hy-
brid capacitor in 0.65 M K2SO4 ͑right voltage axis͒ and variation of the
potential of each individual electrode ͑vs. Ag/AgCl͒.
ja ϭ 9.06 ϫ 10Ϫ8 exp ␩/0.059͒
A/cmϪ2
͓9͔
͑
͕
͖
where ␤a ϭ 0.059 V. The exchange current density and the Tafel
anodic coefficient for Reaction 5 on our MnO2 electrode are in
very good agreement with that (j0 ϭ 1.6 ϫ 10Ϫ7 A/cm2 and
␤a ϭ 0.065 V͒ estimated from the data of Fujimura et al.34 for
␥-MnO2 electrode in 0.5 M NaCl ͑pH 8͒.
The kinetic parameters determined above can be used to esti-
mate the volume of oxygen generated during cycling of the hybrid
capacitor. The details of the calculation can be found in Appendix A.
For one complete charge/discharge cycle with cutoff voltages of
0-2.2 V, the volume of oxygen generated at the MnO2 electrode
(area ϭ 0.42 cm2͒ is estimated to be 1.1 nL. Even after 10 000
cycles, the amount of oxygen accumulated at the positive electrode
does not exceed 11 ␮L. Although low, this value will increase for
real systems using a larger electrode area.
The same analysis can be made for the AC electrode from the
current/potential curves ͑Fig. 9b͒ and the details can be found in
Appendix B. The curve i of Fig. 9b represents the actual current
measured for the carbon electrode whereas curve ii represents the
sole contribution of the HER at the carbon electrode. The Tafel
cathodic coefficient estimated from curve ii is ␤c ϭ 0.108 V and the
exchange current density is 2.3 ϫ 10Ϫ6 A/cm2. From these values,
the volume of hydrogen generated during one cycle is 0.96 ␮L
͑when A ϭ 0.56 cm2͒. After 10000 cycles, the amount of hydrogen
accumulated at the positive electrode should be 9.6 mL.
from 0.03 to Ϫ0.68 V vs. Ag/AgCl, whereas the MnO2 electrode
reached 0.82 V vs. Ag/AgCl at the end of the charge. Numeric
simulations using the expression of ja and jc found previously for
oxygen and hydrogen evolution, and the new charge/discharge cy-
cling parameters provided an estimation of the hydrogen ͑13 nL/cm2
of AC electrode͒ and oxygen ͑0.75 nL/cm2 of MnO2 electrode͒ vol-
umes generated during one complete cycle ͑one charge and one
discharge͒. Again, the volume of hydrogen is larger than that of
oxygen. However, the total volume of gases was drastically de-
creased compared to the 2.2 V capacitor and accordingly a visual
inspection of the capacitor during cycling did not reveal any gas
bubbles. We can estimate that the amount of hydrogen would not
exceed 0.13 mL per cm2 of AC electrode after 10 000 cycles. Fi-
nally, it should be realized both hydrogen and oxygen volumes were
overestimated ͑vide supra͒ and further work is needed to correlate
the volume calculated with the volumes of gas actually generated in
the capacitor.
Figure 11 presents the variation of the energy density of the
capacitor upon cycling at a power of 0.4 kW/kg and with cutoff
A comparison of the volumes of oxygen and hydrogen evaluated
for the hybrid cell with cutoff voltages of 0-2.2 V shows that the
oxygen evolution on the positive MnO2 electrode is less significant
than hydrogen evolution on the negative AC electrode. First, this
suggests that the weight of each electrode would have to be bal-
anced more accurately in order to increase the minimum potential
reached at the AC electrode upon charging and to generate a lower
amount of hydrogen. Second, for a practical capacitor which will
use electrodes with surfaces of several tens of cm2, it would not be
realistic to set the upper limit cutoff voltage to 2.2 V for this hybrid
capacitor. Subsequently, new hybrid cells were designed by chang-
ing the AC:MnO2 weight ratio and by using a smaller operating
voltage.
Hybrid capacitor with a 1.5 V operating voltage.—The cell was
prepared as previously described except that the MnO2 /AC active
weight ratio was decreased to 0.77. Thus it was expected that the AC
electrode would work in a more limited potential range. The varia-
tion of the potential of each individual electrode and the cell voltage
of the hybrid capacitor is shown in Fig. 10 for cutoff voltages of
0-1.5 V. As can be seen on Fig. 10, the AC electrode potential varied
Figure 11. Variation of the energy density with the number of cycles for a
hybrid capacitor ͑Titanium current collectors͒ using 0-1.5 V cutoff voltages.
Constant cycling power: 0.4 kW/kg.

Journal of The Electrochemical Society, 151 ͑4͒ A614-A622 ͑2004͒
A621
voltages 0-1.5 V. The capacitor was cycled over more than 23000
cycles with a 24% decrease of the initial energy density, which is
significantly smaller than what was observed for the cell cycled up
to 2.2 V ͑vide supra͒. The capacity loss for the cell cycled between
0 and 1.5 V cannot be related to the maximum cell voltage used
during cycling because the lower voltage limit of the MnO2 elec-
trode was only 0.03 V vs. Ag/AgCl ͑Fig. 10͒. Consequently, there
should not be any significant reduction of Mn4ϩ to soluble Mn2ϩ in
the electrolyte solution. Subsequently, we can postulate that the ca-
pacity fade for the 0-1.5 V capacitor could be due to ion trapping in
the MnO2 electrode which blocks active sites and decreases the
charge storage capacity. This point must be investigated in future
work.
The pH of the 0.65 M K2SO4 electrolyte was analyzed after the
20000th cycle and increased from 6.5 to 9.6 during cycling. In con-
trast, the pH of the same solution in a control experiment which
involved leaving the electrolyte in a sealed beaker for the same
period of time, only increased from 6.5 to 6.7. Presumably, the pH
change of the electrolyte of the capacitor is due to the HER that is
occurring at the negative AC electrode and seems to confirm that it
occurs at a more significant rate than the OER. This is confirmed by
the very good agreement between the hydrogen volume expected
from the electrolyte pH change with the volume calculated above
from the Tafel plot.
the opportunity to work in UQAM, and UQAM for welcoming him
as a visiting professor. Dr. Rene Le Gall and Professor Donald M.
Schleich ͑LGM Nantes͒ are thanked for helpful and fruitful discus-
sions about the manuscript. The financial support of the Natural
Science and Engineering Research Council ͑NSERC͒ and the Cana-
dian Foundation for Innovation ͑CFI͒ is also acknowledged. Minis-
´
`
`
`
tere des Affaires Etrangeres Franc¸ais and Ministere des Relations
´
Internationales du Quebec are also greatly acknowledged for sup-
porting this work within the framework of the French-Quebec Per-
´
manent Commission under scientific project no. 59-102.
The Natural Sciences and Engineering Research Council of Canada as-
sisted in meeting the publication costs of this article.
Appendix A
Calculation of the Oxygen Volume
The volume of oxygen generated during charge/discharge cycling can be estimated
from the Tafel parameters by making the following assumptions. First, the anodic cur-
rent density due to oxygen evolution, given by Eq. 9, occurred between Erev and the
potential limit (Emax) reached by the MnO2 electrode during the cycling of the hybrid
capacitor. This can be determined from Fig. 4 (Emax ϭ 0.92 V͒ and thus will overesti-
mate the true current density due to Reaction 5. Below Erev , Reaction 5 does not takes
place and subsequently ja ϭ 0. The equilibrium potential calculated from Erev ϭ 1.23
Ϫ 0.059 pH was 0.85 V vs. normal hydrogen electrode, i.e., 0.65 V vs. Ag/AgCl. Even
if the voltage/time profile of the MnO2 electrode is not perfectly linear during constant
current charge/discharge cycling, it will be assumed as linear. The linear dependence of
the voltage can be calculated from
The capacitor cycled with cutoff voltages 0-1.5 V was obviously
outperformed by the capacitor cycled at 2.2 V. However, even at
these lower voltages, a real energy density of 7.4 Wh/kg of active
material was measured ͑maximum energy density of 8.0 Wh/kg͒ for
a real power density of 400 W/kg ͑maximum power density of 10
kW/kg͒. These values compare well with those using organic elec-
trolytes without the drawbacks and the excessive costs.
QMnO ϭ It
͓A-1͔
2
where QMnO is the charge ͑C͒ stored during the time t (t ϭ 0 at the beginning of the
2
charge͒ by the MnO2 electrode and I is the constant current of charge ͑A͒. Additionally
QMnO ϭ CMnO ͑E Ϫ Emin
͒
͓A-2͔
2
2
where CMnO is the capacitance of the MnO2 electrode, E is the potential of the MnO2
2
Conclusion
electrode during the charge, and Emin is the lower potential limit ͑at the beginning of the
charge͒ of the MnO2 electrode. It follows
In this study, we presented the concept of an hybrid capacitor
with a negative AC electrode and a positive MnO2 electrode work-
ing with a mild aqueous electrolyte, 0.65 M K2SO4 . The system
was studied in detail and the electrochemical performance is re-
ported. The energy density reached 19 Wh/kg at a power density of
200 W/kg for a capacitor with cutoff voltages of 0-2.2 V. It de-
creased to 10 Wh/kg when the power density increased up to 3600
W/kg. However, the cell performance was limited by the gas evolu-
tion occurring at the electrodes upon cycling. The Tafel coefficients
were determined for both electrodes and the subsequent oxygen and
hydrogen evolution were calculated. Hydrogen evolution is the main
hindrance of the hybrid system because it exceeds by 1000 times the
volume of oxygen. The hydrogen evolution is much too high for a
large-scale system and can cause serious safety problems. Subse-
quently the feasibility of a hybrid cell working at a lower voltage
͑1.5 V͒ was demonstrated and was characterized by negligible gas
evolution over a long cycling period. Despite a decrease in the en-
ergy and power density, the 1.5 V capacitor still exhibited interesting
properties, especially if the low cost and low toxicity of the different
components are taken into account. A compromise has to be found
for the working voltage in order to optimize the electrochemical
performance. However, this kind of hybrid capacitor based on AC or
t ϭ CMnO ͑E Ϫ Emin͒/I
͓A-3͔
2
Introducing the overpotential ␩, Eq. 8 becomes
t ϭ CMnO ͑␩ ϩ Erev Ϫ Emin͒/I
͓A-4͔
2
Thus, Eq. A-4 gives the relationship between the charge ͑or discharge͒ time and the
overpotential. The total charge qa corresponding to oxygen evolution on the MnO2
surface ͑expressed as C cmϪ2͒ can be estimated from the anodic current ja between Erev
and Emax ͑during the charge͒ and the anodic current ja between Emax and Erev ͑during
the discharge͒. Since these two quantities are equivalent, it follows
E
max
qa ϭ 2
jadt
͓A-5͔
͓A-6͔
͓A-7͔
͵
E
rev
dt can be calculated from Eq. A-4
and subsequently
dt ϭ CMnO d␩/I
2
␩ϭE ϪE
max
rev
CMnO
␩
2
qa ϭ 2
j0
exp
ͪ
d␩
ͩ
͵
35
I
␤
a
a low cost metal oxide, such as Fe3O4 as the negative electrode
␩ϭ0
and manganese dioxide as the positive electrode, associated with a
nontoxic aqueous electrolyte can be an interesting alternative to the
current systems.
which is
␩ϭE ϪE
max
rev
␤aCMnO
␩
2
qa
ϭ
2j
exp
͓A-8͔
ͫ
ͬ
ͩ ͪ
0
Acknowledgments
I
␤
a
␩ϭ0
The authors would like to thank Professor Richard Chahine and
The calculation was performed in the case of the hybrid capacitor presented in Fig. 8
͑cell on a titanium current collector͒ with j0 ϭ 9.06 ϫ 10Ϫ8 A/cm2 and ␤a ϭ 0.059 V,
´
´
`
Dr. Daniel Cossement from the Universite du Quebec a Trois-
`
Rivieres for providing the AC. Michel Preda and Raymond Mineau
I is the current used for the charge/discharge cycling test in Fig. 8a ͑6 mA͒, CMnO is the
2
´
`
from Departement des Sciences de la Terre et de l’Atmosphere,
UQAM, are greatly acknowledged for their help with the X-ray
diffraction and electron microscopy. Many thanks to Dr. Alexis
Laforgue for taking care of the long term cycling. One of the au-
capacitance of the MnO2 electrode ͑0.294 F͒,
Emax ϭ 0.92 V vs. Ag/AgCl and
E
rev ϭ 0.65 V vs. Ag/AgCl. Then qa ϭ 4.54 ϫ 10Ϫ5 C/cm2 and the number of moles
of oxygen, nO generated from Reaction 5 can be calculated from
2
nO ϭ qa /nF
͓A-9͔
´
thors, T.B., would like to thank Universite de Nantes for giving him
2

A622
Journal of The Electrochemical Society, 151 ͑4͒ A614-A622 ͑2004͒
where n is the number of electrons involved in Reaction 5 and F the Faraday constant.
The volume of oxygen generated during one cycle ͑one charge followed by one dis-
charge͒ can be estimated from
2. A. Burke, J. Power Sources, 91, 37 ͑2000͒.
3. P. Soudan, P. Lucas, H. A. Ho, D. Jobin, L. Breau, and D. Belanger, J. Mater.
Chem., 11, 773 ͑2001͒.
´
4. J. P. Zheng and T. R. Jow, J. Electrochem. Soc., 144, 2417 ͑1997͒.
5. M. Mastragostino, C. Arbizzani, R. Paraventi, and A. Zanelli, J. Electrochem. Soc.,
147, 407 ͑2000͒.
6. A. Laforgue, P. Simon, and J. F. Fauvarque, Synth. Met., 123, 311 ͑2001͒.
7. J. Gamby, P. L. Taberna, P. Simon, J. F. Fauvarque, and M. Chesneau, J. Power
Sources, 101, 109 ͑2001͒.
VO ϭ nO VM
͓A-10͔
2
2
where VM is the molar volume at 20°C ͑22 400 cm3͒.
Appendix B
8. S. Hadzi-Jordanov, H. A. Kozlowska, and B. E. Conway, J. Electroanal. Chem.
Interfacial Electrochem., 60, 359 ͑1975͒.
9. H. Kim and B. N. Popov, J. Power Sources, 104, 52 ͑2002͒.
10. S. Trasatti and G. Buzzanca, J. Electroanal. Chem. Interfacial Electrochem., 29,
Appl. 1 ͑1971͒.
Hydrogen Evolution Reaction
The cathodic current Ic can be expressed as
log Ic ϭ AcE ϩ Bc
͓B-1͔
where E is the electrode potential, and Ac (Ϫ4.0 Ϯ 1.0 VϪ1͒ and Bc (Ϫ8.2 Ϯ 1.7) are
two constants determined experimentally from the Tafel lines ͑Fig. 9b, curve ii͒ that
were used for the calculation of the cathodic current for Reaction 6. From the data of
Fig. 11, we found
11. D. Galizzioli, F. Tantardini, and S. Trasatti, J. Appl. Electrochem., 4, 57 ͑1975͒.
12. F. Beck, M. Dolata, E. Grivei, and N. Probst, J. Appl. Electrochem., 31, 845 ͑2001͒.
´
13. E. Frackowiak and F. Beguin, Carbon, 39, 937 ͑2001͒.
14. A. Kozawa, H. Oho, M. Sano, D. Brodd, and R. Brodd, J. Power Sources, 80, 12
͑1999͒.
jc ϭ Ϫ2.3 ϫ 10Ϫ6 exp͑Ϫ␩/0.108͒ A/cm2
͓B-2͔
´
15. R. Patrice, B. Gerand, J. B. Leriche, L. Seguin, E. Wang, R. Moses, K. Brandt, and
͕
͖
J. M. Tarascon, J. Electrochem. Soc., 148, A448 ͑2001͒.
16. G. Salitra, A. Soffer, L. Eliad, Y. Cohen, and D. Aurbach, J. Electrochem. Soc., 147,
2486 ͑2000͒.
17. N. L. Wu and S. Y. Wang, J. Power Sources, 110, 233 ͑2002͒.
18. V. M. M. Lobo and J. L. Quaresma, Handbook of Electrolyte Solutions, Elsevier,
Amsterdam ͑1989͒.
The total charge corresponding to hydrogen evolution on the AC surface ͑expressed as
C/cm2͒ can be estimated from the anodic current jc between Erev and Emin ͑during the
charge͒ and the anodic current jc between Er min and Erev ͑during the discharge͒. Erev is
the equilibrium potential for Reaction 6 vs. Ag/AgCl and Emin is the lowest potential of
the AC electrode at the end of the charge ͑at the beginning of the discharge͒. It follows
19. S. C. Pang, M. A. Anderson, and T. W. J. Chapman, J. Electrochem. Soc., 147, 444
͑2000͒.
E
min
qc ϭ 2
jcdt
͓B-3͔
20. S. F. Chin, S. C. Pang, and M. A. Anderson, J. Electrochem. Soc., 149, A379
͑2002͒.
21. C. C. Hu and T. W. Tsou, Electrochem. Commun., 4, 105 ͑2002͒.
22. ͑a͒ H. Y. Lee and J. B. Goodenough, J. Solid State Chem., 144, 220 ͑1999͒; ͑b͒ H.
Y. Lee, S. W. Kim, and H. Y. Lee, Electrochem. Solid-State Lett., 4, A19 ͑2001͒.
23. H. Kim and B. N. Popov, J. Electrochem. Soc., 150, D56 ͑2003͒.
͵
E
rev
dt can be calculated in the same manner as reported for the MnO2 electrode, and
subsequently
␩ϭE ϪE
min
rev
␤cCAC
␩
´
24. M. Toupin, T. Brousse, and D. Belanger, Chem. Mater., 14, 3946 ͑2002͒.
qc
ϭ
Ϫ2j
exp Ϫ
͓B-4͔
ͫ
ͩ
ͪ
ͬ
0
25. G. G. Amatucci, F. Badway, A. du Pasquier, and T. Zheng, J. Electrochem. Soc.,
148, A930 ͑2001͒.
26. A. du Pasquier, A. Laforgue, P. Simon, G. G. Amatucci, and J. F. Fauvarque, J.
Electrochem. Soc., 149, A302 ͑2002͒.
27. M. S. Hong, S. H. Lee, and S. W. Kim, Electrochem. Solid-State Lett., 5, A227
͑2002͒.
I
␤
c
␩ϭ0
j0 and ␤c are taken from Eq. B-2, Emin ϭ Ϫ1.28 V vs. Ag/AgCl and Erev ϭ Ϫ0.57 V
vs. Ag/AgCl, I is the constant current during the charge/discharge cycling ͑6 mA͒ and
CAC is the capacitance of the AC electrode ͑0.254 F͒. Then qc ϭ 14.7 mC/cm2 and the
number of mol of hydrogen, nH , generated from Reaction 6 can be calculated from
2
28. J. W. Long, K. E. Swider-Lyons, R. M. Stroud, and D. R. Rolison, Electrochem.
Solid-State Lett., 3, 453 ͑2000͒.
29. S. Franger, S. Bach, J. Farcy, J. P. Pereira-Ramos, and N. Baffier, J. Power Sources,
109, 262 ͑2002͒.
nH ϭ qc /nF
͓B-5͔
2
where n is the number of electrons involved in Reaction 6 and F is the Faraday constant.
The volume of hydrogen generated during one cycle ͑one charge followed by one
discharge͒ can be estimated from
30. D. Cossement, R. Chahine, and T. K. Bose, Final report, Synergy Program ͑March
2000͒.
´
31. A. Guinier, Theorie et Pratique de la Radiocristallographie, A. Guinier, Editor,
Dunod, Paris ͑1964͒.
32. A. Era, Z. Takehara, and S. Yoshizawa, Electrochim. Acta, 12, 1199 ͑1967͒.
VH ϭ nH VM
͓B-6͔
2
2
´
´
´
33. D. Landolt, Traite des Materiaux, Corrosion et Chimie de Surfaces des Metaux,
Vol. 12, Presse Polytechniques et Universitaires Romandes, Lausanne, Switzerland
͑1993͒.
where VM is the molar volume at 20°C.
References
34. K. Fujimura, K. Izumiya, A. Kawashima, E. Akiyama, H. Habazaki, N. Kumagain,
and K. Hashimoto, J. Appl. Electrochem., 29, 765 ͑1999͒.
1. B. E. Conway, Electrochemical Supercapacitors, Scientific Fundamentals and
Technological Applications, Kluwer Academic/Plenum Press, New York ͑1999͒.
´
35. T. Brousse and D. Belanger, Electrochem. Solid-State Lett., 6, A244 ͑2003͒.