Power loss and energy density of the asymmetric ultracapacitor loaded with molybdenum doped manganese oxide

Article abstract of DOI:10.1016/j.electacta.2012.02.038

Ultracapacitors of asymmetric configuration have been prepared with activated carbon (AC) and undoped or Mo-doped manganese oxide (MnO₂) in 1.0 M Na₂SO₄ electrolyte. Phase analysis shows the AC powder, 1-15 μm in size, contains both disordered and graphitic structures, and the undoped and Mo-doped oxide powder, 0.05-0.20 μm in particle size, mainly involves amorphous MnO₂ and MoO₂. CV results indicate the single electrode of AC plus 10 wt% Mo-doped MnO₂ (A9O_M1) is superior to the electrode with undoped MnO₂ or high content of doped MnO₂, exhibiting features of double layer capacitance at high scan rate and pseudocapacitance characteristics at low scan rate. When assembled with a negative electrode of AC, the capacitor of positive A9O_M1 electrode demonstrates the least power loss among three asymmetric capacitors. This asymmetric capacitor also shows a higher capacitance than the symmetric AC capacitor when the current density is less than 8.0 A g^-1 in 1.8 V potential window. But a higher electrode resistance of A9O_M1, in contrast with AC, compromises its capacitance plus. When the energy density of A9O_M1 asymmetric capacitor is compared with that of symmetric AC capacitor at the same power level, the capacitance benefit on energy density is restricted to current density ≤ 3.0 A g^-1.

Full text of DOI:10.1016/j.electacta.2012.02.038

Electrochimica Acta 68 (2012) 95–102
Contents lists available at SciVerse ScienceDirect
Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta
Power loss and energy density of the asymmetric ultracapacitor loaded with
molybdenum doped manganese oxide
Yue-Sheng Wanga, Dah-Shyang Tsaia,∗, Wen-Hung Chunga, Yong-Sin Syua, Ying-Sheng Huangb
a Department of Chemical Engineering, National Taiwan University of Science and Technology, Taipei 10607, Taiwan
b Department of Electronic Engineering, National Taiwan University of Science and Technology, Taipei 10607, Taiwan
a r t i c l e i n f o
a b s t r a c t
Article history:
Ultracapacitors of asymmetric configuration have been prepared with activated carbon (AC) and undoped
or Mo-doped manganese oxide (MnO2) in 1.0 M Na2SO4 electrolyte. Phase analysis shows the AC powder,
1–15 ␮m in size, contains both disordered and graphitic structures, and the undoped and Mo-doped oxide
powder, 0.05–0.20 ␮m in particle size, mainly involves amorphous MnO2 and MoO2. CV results indicate
the single electrode of AC plus 10 wt% Mo-doped MnO2 (A9OM1) is superior to the electrode with undoped
MnO2 or high content of doped MnO2, exhibiting features of double layer capacitance at high scan rate
and pseudocapacitance characteristics at low scan rate. When assembled with a negative electrode of AC,
the capacitor of positive A9OM1 electrode demonstrates the least power loss among three asymmetric
capacitors. This asymmetric capacitor also shows a higher capacitance than the symmetric AC capacitor
when the current density is less than 8.0 A g−1 in 1.8 V potential window. But a higher electrode resistance
Received 12 October 2011
Received in revised form
29 December 2011
Accepted 9 February 2012
Available online 28 February 2012
Keywords:
Asymmetric ultracapacitor
Activated carbon
Manganese oxide
Mo-doping
of A9OM1, in contrast with AC, compromises its capacitance plus. When the energy density of A9OM
1
Polarization loss
asymmetric capacitor is compared with that of symmetric AC capacitor at the same power level, the
capacitance benefit on energy density is restricted to current density ≤ 3.0 A g−1
.
© 2012 Elsevier Ltd. All rights reserved.
1. Introduction
A larger tunnel size renders more interfacial area to be acces-
sible. Accordingly, in parallel with the decreasing tunnel size of
the crystallographic form, the mass-specific value of MnO2 pseu-
docapacitance descends in this following order, ˛ ∼ ı > ꢀ > ꢁ > ˇ
[4,5]. The reported capacitance value of manganese oxide powder
varies considerably 150–440 F g−1 [6–8], while that of its thin film
is estimated much higher 1380 F g−1 [9]. As a positive electrode,
manganese oxide can be polarized in a broad potential window and
repeatedly cycled with long-term stability [10,11]. The major draw-
back of MnO2 is its poor electrical conductivity which restricts its
faradaic process. Mixing with carbon nanotubes, carbon aerogel, or
activated carbon (AC) improves the capacitor performance [12–17].
A less explored approach of modification is doping MnO2 with alio-
valent cations. Nakayama and coworkers have demonstrated that
doping ∼15 mol% Mo improves the capacitance and reversibility in
cyclic voltammetry (CV) results. They attributed the better perfor-
mance to an improved electrical conductivity of MnO2 through Mo
doping [18].
Electrochemical capacitors (ultracapacitors or supercapacitors)
have emerged as an auxiliary energy device to handle peak current
for the primary energy source, such as battery or fuel cell, because
of their high power capability and long cycle life. The energy stor-
age mechanism of ultracapacitor mainly relies on the double layer
capacitance which segregates charge physically at the interface of
electrolyte and electrode. Yet, an active material of pseudocapaci-
tance is often added to augment the relatively small double layer
capacitance and energy capacity. Rooted in the faradaic process
of surface redox reaction, pseudocapacitance is generally poten-
tial dependent and large in magnitude [1–3]. Among the transition
metal oxides of sizable pseudocapacitance, manganese oxide has
attracted a great deal of research attention. Powders of manganese
oxide, being widely used in batteries, have proven to be a nontoxic,
cost-effective, and environmentally friendly material of electro-
chemistry.
Hydrous manganese oxide is known to crystallize in several
structural forms with different tunnel sizes to accommodate ions.
However, the upgraded capacitive performance of Mo-doped
manganese oxide is only shown in a thin film form on platinum
substrate [18]. There is no report, to the best of our knowledge, on
the influences of Mo-doped MnO2 at the level of capacitor power
and energy. In this work, we prepare three asymmetric capaci-
tors, assembled with a negative AC electrode and a positive oxide
electrode containing 10 or 90 wt% Mo-doped MnO2, or undoped
MnO2, along with a symmetric capacitor of AC for reference. CV
∗
Corresponding author at: Department of Chemical Engineering, National Taiwan
University of Science and Technology, 43, Keelung Road, Section 4, Taipei 10607,
Taiwan. Tel.: +886 2 27376618; fax: +886 2 27376644.
E-mail address: dstsai@mail.ntust.edu.tw (D.-S. Tsai).
0013-4686/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2012.02.038

96
Y.-S. Wang et al. / Electrochimica Acta 68 (2012) 95–102
measurements are performed to study the behavior of single elec-
trodes. Power losses and polarization losses of the capacitors are
analyzed and compared so that the effects of Mo-doping can be
understood.
paste of A9O1 was prepared similarly, without Na2MoO4 in the
recipe.
2.2. Electrochemical measurement and calculation
Capacitance of the single electrode was measured using cyclic
voltammetry, with a thermostated setup of three electrodes in
1.0 M Na2SO4 at 27 ◦C. The working electrode was one of the
four electrodes including AC, A9OM1, A1OM9, A9O1; the reference
electrode was an Ag/AgCl (sat. KCl) reference electrode (XR820,
Radiometer Analytical); the counter electrode was a 2 cm2 plat-
inum plate. CV capacitance values are estimated as a half of anodic
and cathodic charge sum divided by the active mass loading of
working electrode and the scanned potential range. Galvanostatic
charge discharge data and impedance data of the capacitor cell
were collected in a two-electrode configuration. Data of CV and
charge discharge experiments were controlled using a multichan-
nel potentiostat (Solartron 1470E). The impedance spectra were
scanned in the frequency range 0.1 Hz–1.0 MHz at open circuit
potential, by imposing an alternating current of magnitude 50 mV
with a Solartron 1260 frequency analyzer. Morphology of the mix-
tures of activated carbon and manganese oxide was analyzed with
a field-emission scanning electron microscope (SEM, JSM-6500F,
JEOL), which was equipped with an energy dispersive X-ray spec-
trometer (EDS) for elemental analysis (INCA, Oxford UK). Powder
diffraction patterns of these mixtures were scanned using an X-ray
diffractometer (D2 Phaser, Bruker) with a CuK␣ radiation source.
XPS spectra were collected in a Thermo VG Scientific Theta Probe
system under high vacuum condition ∼7 × 10−9 mbar. The Al K␣
1486.6 eV line was the X-ray source and the spectral line was cali-
brated with the Ag 3d5/2 line at 368.26 eV. The peak positions were
determined through curve fitting using the software Unifit Version
2006, which employs the convoluting Gaussian and Lorentzian (or
Doniach–Sunjic type) functions.
2. Experimental details
2.1. Preparation of electrodes and capacitors
Three asymmetric capacitors and one symmetric capacitor were
fabricated in this work. These electrochemical capacitors were
labeled according to the electrode content of active materials.
The symmetric capacitor, AC–AC, had its electrodes loaded with
a mixture of activated charcoal (C4386, Sigma–Aldrich) and Vul-
can (XC-72, Carbot) in a 5:1 weight ratio. The activated charcoal
imparted high interfacial surface and capacitance (BET surface
area, 1100 m2 g−1), while Vulcan (BET surface area, 223 m2 g−1
)
enhanced the electrode conductivity. The powder mixture was
blended in a mixed solvent of isopropyl alcohol and deionized water
(1:1 in weight ratio) with 10 wt% PTFE added as binder to formu-
late a paste [19]. The paste was added with a small amount of
deflocculant and ball milled overnight, then vacuumed to remove
trapped bubbles. The paste was drawn with a micropipette and
spread out on the glass substrate (Eagle 2000, Corning), which was
metalized with a sputtered Au/Ti thin film as current collector.
After drying at 80 ◦C, the mass of activated charcoal and Vulcan
was weighed using a precision balance; 0.3–0.4 mg active material
on the 0.6 cm2 current collector. The two wired AC electrodes, sep-
arated by 0.14 mm thick polypropylene separator (PPA-14S), were
firmly pressed against each other under pressure 8.2 MPa, face-
to-face, and immersed in 1.0 M Na2SO4 to complete the capacitor
preparation.
The three asymmetric capacitors were also prepared in the
parallel-plate configuration, labeled as AC–A9OM1, AC–A1OM9, and
AC–A9O1. The AC electrode was designated as negative electrode
in the asymmetric capacitor. The positive electrode was one of
A9OM1, A9O1, and A1OM9 electrodes, whose numbers denote their
compositions. For instance, the A9OM1 electrode was loaded with
activated carbon (described earlier) and Mo-doped manganese
oxide in a 9:1 weight ratio. The oxide with a subscript M meant
doping with 15 mol% molybdenum, without the subscript M meant
undoped manganese oxide. In a preliminary test, we also prepared
AC–A7OM3 and AC–A5OM5 capacitors, whose positive electrodes
contained 30 and 50 wt% Mo-doped MnO2. But their internal resis-
tances estimated by impedance spectroscopy are higher than that
of AC–A9OM1. And their pseudocapacitances were less evident than
AC–A1OM9. Hence we chose AC–A9OM1 to represent the influences
of Mo-doped manganese oxide. Preliminary data of AC–A7OM3 and
AC–A5OM5 can be found in Supplementary data.
Powder synthesis of A9OM1 and A1OM9 was done via the dispro-
portionation reaction between potassium permanganate KMnO4
and manganese(II) sulfate MnSO4 [12,20]. The powder mixture of
activated carbon, KMnO4, Na2MoO4 was first dissolved and sus-
pended in de-ionized water to make a slurry containing 0.05 M
solution with Mn and Mo in a 1.8:1 molar ratio. Then 0.05 M
MnSO4 was added dropwisely with vigorously stirring, to formu-
late the final molar ratio KMnO4:MnSO4:Na2MoO4 = 1.8:4:1. This
molar ratio ensured that we had slightly more MnSO4 in reducing
KMnO4 and Na2MoO4. The stirring continued for 12 h at 60 ◦C to
make sure a full redox reaction. Excess MnSO4 would be washed
away after the precipitation completed. The mixture of precipi-
tates and activated carbon were filtered, washed, and dried at 80 ◦C
overnight. The dried mixture was then dispersed with the mixed
solvent of 1:1 ratio isopropanol and water, followed by adding
the PTFE binder to prepare the paste for positive electrode. The
Based on the galvanostatic discharge curves, we calculated the
specific capacitance (CD), energy output density (E) and power
output density (P) of four electrochemical capacitors. Specific
capacitance value was evaluated as the integral of current (I) over
the elapsed time (ꢂt) during discharge divided by the potential
window (ꢂV) and the mass of active materials on two electrodes
(ma). Energy density was evaluated as the integral of the current
and cell voltage (V) product divided by the activated materials mass
of two electrodes.
ꢀ
1
ma
ꢂt Idt
0
CD
=
(1)
(2)
maꢂV
ꢁ
ꢂt
E =
IV(t)dt
0
Power density (P) is calculated as the energy density divided by
the elapsed time ꢂt of discharge.
E
P =
(3)
ꢂt
3. Results and discussion
3.1. Particle size and phase analysis of manganese oxide and
activated carbon
Morphology of the A9OM1 powder is predominated by the pres-
ence of coarse carbon grains with rectangular plate geometry.
The carbon particle size ranges from 1 to 15 ␮m, as illustrated in
Fig. 1(a). This particle size is somewhat less than the as-received
activated charcoal (C4386), since the A9OM1 powder has under-
gone ball milling in preparation. Also present in Fig. 1(a) are less

Y.-S. Wang et al. / Electrochimica Acta 68 (2012) 95–102
97
Fig. 2. X-ray diffraction patterns of the AC, A9OM1, and A1OM9 powder dried at
80 ◦C overnight, showing oxide is amorphous.
Fig. 1. SEM images of the powders used in the (a) A9OM1 electrode and (b) A1OM
9
electrode. Note the scale bars of two micrographs are different in size. The image (b)
has a higher magnification than the image (b), since the MnO2 particles are much
smaller than the carbon particles.
populated and much smaller manganese oxide particles, dispersed
around the carbon. The granular shape of manganese oxide is more
evident in Fig. 1(b), which shows the SEM image of A1OM9 pow-
der containing 90 wt% MnO2. The particle size range of Mo-doped
MnO2 is relatively narrow, 50–200 nm. Elemental analysis indi-
cates the Mo and Mn atomic percentages of these particles are 3.1
and 16.5; respectively. More details of EDS analysis are provided in
Supplementary data. Hidden underneath the numerous manganese
oxide particles are a few large carbon particles. The micrograph of
A9O1 powder is very similar to that of A9OM1, since doped and
undoped manganese oxide particles look alike.
Fig. 2 presents the X-ray powder diffraction results of the active
materials used in formulating the electrodes. The diffraction pat-
tern of AC shows two very broad features at ∼25◦ and 44◦, which
are characteristic of disordered carbon. Meanwhile, there is a sharp
peak at 2ꢃ = 26.6◦, superimposed on the broad feature of disordered
carbon. The 26.6◦ peak is attributed to the (0 0 2) basal planes of
hexagonal graphite [21]. Hence, the carbon powder of AC electrode
exhibits both disordered and graphitic structures. Also plotted in
Fig. 2 are the diffraction patterns of 80 ◦C dried A9OM1 and A1OM
9
powder. The features of A9OM1 powder are nearly identical to those
of AC, since A9OM1 contains 90 wt% activated carbon. On the other
hand, the diffraction pattern of A1OM9 of 90 wt% oxide is feature-
less in the 2ꢃ range of 10–80◦. The diffraction pattern of A9O1 is
similar to that of A9OM1. Judging from the patterns of A9OM1 and
A1OM9, we infer the 80 ◦C dried oxide powder is amorphous and
its reflection contributions vanish in the background noise.
Fig. 3. XPS spectra of 80 ◦C dried A1OM9 powder, showing (a) Mn 2p line and (b)
Mo 3d line.
Fig. 3 shows XPS spectra of Mn 2p and Mo 3d electrons
in the 80 ◦C dried A1OM9 powder. According to data in the

98
Y.-S. Wang et al. / Electrochimica Acta 68 (2012) 95–102
Fig. 4. CV diagrams of four single electrodes, including the (a) AC electrode, (b) A9OM1 electrode, (c) A1OM9 electrode, (d) A9O1 electrode. Note the potential range of AC
electrode is between −0.8 and 0.2 V, intended to be a negative electrode; while those of A9OM1, A1OM9, A9O1 electrodes are between −0.2 and 0.8 V, intended to be a positive
electrode.
handbook [22], binding energy for the strong 2p3/2 line of man-
ganese increases with its oxidation number; 641.0 (Mn2+), 641.5
(Mn3+), and 642.1 eV (Mn4+). These data slightly differ from the
average values summarized by Nesbitt and Banerjee on Mn oxides
[23], 640.8 (Mn2+), 641.8 (Mn3+), and 642.3 eV (Mn4+). The fitted
2p3/2 line 642.4 eV of Fig. 3(a) indicates the Mn oxidation state
in A1OM9 ought to be 4+. The energy separation between 2p3/2
and 2p1/2 photoelectrons is 11.3 eV, also in agreement with 11.7 eV
of the handbook [22]. On the other hand, Fig. 3(b) shows the Mo
exhibits nearly rectangular voltammograms between −0.2 and
0.8 V (vs. Ag/AgCl), indicating a mixture of 90 wt% activated carbon
and 10 wt% Mo-doped manganese oxide has the potential being
an excellent positive electrode. The voltage of AC electrode was
scanned between −0.8 and 0.2 V (vs. Ag/AgCl) since it is used as
negative electrode in the asymmetric cell.
Quantitative differences between the four single electrodes are
shown in terms of CV capacitance, listed in Table 1. The tabu-
lated value and its variation with scan rate reveal the intrinsic
features of each electrode. For instance, the AC electrode is least
capacitive at 2 mV s−1 (89.5 F g−1), but becomes most capacitive at
500 mV s−1 (56.8 F g−1). Its moderately varying capacitance sug-
gests the AC electrode is of low impedance and fast response.
In contrast to AC, the A1OM9 electrode exhibits a widely vary-
ing capacitance. It is the most capacitive electrode at 2 mV s−1
(166 F g−1), and the least capacitive one at 500 mV s−1 (23.4 F g−1).
(3d5/2, 3/2) spin–orbit components are 229.2 and 232.3 eV. Follow-
ing the XPS study on partially reduced MoO3 catalysts [24,25],
Mo in A1OM9 is assigned an oxidation number of 4+, based on
the reported binding energy values: 232.7 and 235.9 eV for MoO3;
231.7 and 234.9 eV for Mo2O5; 229.1, 232.3 eV for MoO2; 227.7
and 230.9 eV for Mo metal. We conclude the oxides exist as either
two well-mixed dioxides or a dioxide solid solution of amorphous
structure.
Table 1
3.2. CV results of single electrodes
Values of CV capacitance versus scan rate for the single electrode AC, A9OM1, A1OM9,
and A9O1. CV diagrams were plotted in Fig. 4, measured in 1.0 M Na2SO4 using a
typical 3-electrode setup.
Fig. 4 presents four sets of voltammogram for single electrode
AC, A9OM1, A1OM9, A9O1 at scan rate 2–500 mV s−1. Among these
four electrodes, the voltammograms of AC, A9OM1 are nearly rect-
angular in shape, but A1OM9 and A9O1 deviates significantly from
Scan rate (mV s−1
)
AC (F g−1
)
A9OM1 (F g−1
)
A1OM9 (F g−1
)
A9O1 (F g−1
)
2
10
50
200
500
89.5
88.8
82.1
70.9
56.8
143
104
83.0
66.8
51.2
166
133
79.5
39.2
23.4
124
99.8
73.0
48.6
30.6
this model shape at 50, 200, 500 mV s−1. CV currents of A1OM
9
and A9O1 tilt considerably at high scan rate, suggesting substantial
internal resistance in these two electrodes. The A9OM1 electrode

Y.-S. Wang et al. / Electrochimica Acta 68 (2012) 95–102
99
Fig. 5. Nyquist plots of four electrochemical capacitors; (a) the AC–AC and
AC–A9OM1 capacitors, (b) the AC–A1OM9 and AC–A9O1 capacitors. The ESR values,
defined at 1 kHz, are marked on the graphs.
Fig. 6. Variation of power output with respect to current density in the poten-
tial window, (a) 1.8 V and (b) 1.0 V. The dashed line denotes the power of an ideal
capacitor. The power losses of four capacitors are marked in (a) for 1.8 V.
Hence, A1OM9 displays the sluggish characteristics of pseudoca-
pacitance. CV capacitance of A9OM1 electrode is interesting, near
the value of AC at 500 mV s−1, but next to that of A1OM9 at the
responsive than AC–A1OM9, and AC–A9O1 capacitors of Fig. 5(b),
because they differ in the positive electrode. The internal resis-
tance of a capacitor can be symbolized by its equivalent series
resistance (ESR) value, which is defined as the real component of
impedance at 1 kHz. Fig. 5 shows the ESR values of AC–AC and
AC–A9OM1 are 1.2 ꢄ (0.72 ꢄ cm2) and 1.4 ꢄ (0.84 ꢄ cm2), substan-
tially less than those of AC–A1OM9 3.6 ꢄ (2.2 ꢄ cm2) and AC–A9O1
3.1 ꢄ (1.9 ꢄ cm2). These impedance data are fitted to two models
of series/parallel ladder electrical circuit. Impedance data of sym-
metric AC–AC capacitor are fitted to a circuit of two RC in series
with a Warburg element. While those of the asymmetric capacitors
are fitted to a circuit of one RC in series with a Warburg element,
and a pseudocapacitor pore circuit, because the positive electrode
is pseudocapacitive. Fitting parameters along with the electrical
circuit are detailed in Supplementary data.
The power loss of a capacitor is intimately associated with
its internal resistance. Fig. 6 shows the power output devi-
ates further away from the ideal power (dashed line) as the
current density increases. That deviation is the power loss. Con-
sistent with the impedance results, the power losses of four
capacitors are divided into two groups. Fig. 6(a) shows that
the power output losses of AC–AC and AC–A9OM1 are signifi-
cantly less than those of AC–A1OM9 and AC–A9O1 in the 1.8 V
potential window. For instance, at current density 8 A g−1, the
cell power output decreases in the following order; 6.0 kW kg−1
other extreme, 2 mV s−1. Judging from the AC, A1OM9, and A9OM
1
capacitance values, we infer that a small fraction of Mo-doped
MnO2 provides a proper mix of pseudocapacitance and double layer
capacitance. When 10 wt% Mo-doped MnO2 is mixed with activated
carbon, the electrode response slows down a little, its pseudocapac-
itance is operative at low scan rate, its double layer capacitance
emerges at high scan rate. On the other hand, the electrode of
90 wt% Mo-doped MnO2 responds sluggishly, since pseudocapaci-
tance dominates at 2 and 500 mV s−1 both.
It is of interest to find out whether the Mo-doping is beneficial to
electrode behavior. We contrast the CV capacitances of A9OM1 and
A9O1 electrodes, which only differ in their 10 wt% MnO2 being Mo-
doped or undoped. The CV capacitance value of A9OM1 is generally
higher than that of A9O1, especially at high scan rate. Better values
of A9OM1 at high scan rate signify it is a more conductive electrode.
Therefore, Mo-doping appears to promote both capacitance and
conductivity of the oxide-containing electrode, but not by a large
margin.
3.3. Impedance and power loss of capacitors
Impedance results of the four capacitors can be divided into two
groups. The AC–AC and AC–A9OM1 capacitors of Fig. 5(a) are more

100
Y.-S. Wang et al. / Electrochimica Acta 68 (2012) 95–102
(AC–AC), 4.8 kW kg−1 (AC–A9OM1), 2.8 kW kg−1 (AC–A1OM9), and
2.6 kW kg−1 (AC–A9O1), in contrast to the ideal value 7.2 kW kg−1
cell resistances. Table 2 lists the Reff values of AC–AC, ∼15 ꢄ, and
those of AC–A9OM1, ∼75 ꢄ. Therefore, the AC–A9OM1 asymmet-
ric capacitor exhibits a larger ohmic polarization loss IReff than the
AC–AC symmetric capacitor at the same current.
Table 2 also shows the Veff value is always less than the max-
imum cell voltage, since the capacitor must pay the price of IR
drop to turn on the energy flow. In general, the deviation of Veff
increases with increasing potential window. For AC–AC, the devia-
tion is small, increasing from 0.09 V (1.0 V window) to 0.18 V (1.8 V
window). For AC–A9OM1, the IR drop is relatively higher, 0.17 V in
1.0 V window and 0.24 V in 1.8 V window. Thus, adding 10 wt% Mo-
doped MnO2 in the positive electrode diminishes the cell driving
potential (Veff) and increases the cell resistance (Reff).
.
Fig. 6(b) shows a similar trend in the 1.0 V window, the power
output is ranked as follows, 3.0 kW kg−1 (ideal), 2.5 kW kg−1
(AC–AC), 2.0 kW kg−1 (AC–A9OM1), 1.2 kW kg−1 (AC–A1OM9), and
0.84 kW kg−1 (AC–A9O1) at 6 A g−1. Since the potentiostat con-
trols the discharge current, the power loss is attributed to the
overpotential penalty on cell voltage. The capacitor overpo-
tential most probably arises from the ohmic polarization loss,
considering the evident connection between power loss and
ESR.
3.4. Polarization modeling and Ragone plot
For the two sluggish capacitors, Fig. 8 shows that data of (2P/I)
versus I have to be correlated in two segments, instead of one. For
AC–A1OM9, when the current is less than 1.7 mA (current density
∼1.7 A g−1), the (2P/I) value decreases sharply with increasing cur-
rent, are shown in Fig. 8(a). When the current is over 1.7 mA, the
decreasing trend slows down as if the discharge enters a differ-
Hence, assuming the polarization loss is ohmic in nature, we
apply a simple model to correlate the power output (P) with respect
to the discharge current (I).
I
P = (Veff − IReff
)
(4)
2
ent polarization loss mode. Thus, energy drainage of AC–A1OM
9
has two different polarization losses in one potential window.
The stored energy of AC–A9O1 is also drained in two modes, but
their differences are less, compared with AC–A1OM9. One plausi-
ble reason is that the positive electrode of AC–A9O1 contains more
activated carbon, not 10 wt%. Fig. 8(b) presents the (2P/I) value
of AC–A9O1, when less than 2.4 mA (current density ∼2 A g−1),
where Veff is the effective voltage driving the energy discharge and
Reff the effective value of cell resistance. Fig. 7(a) and (b) show that,
for AC–AC and AC–A9OM1, their data of (2P/I) versus I can be well
fitted by straight lines in potential window 1.0–1.8 V. The correlated
lines of AC–AC are parallel and flat, meaning the effective resis-
tances are small and similar in magnitude. The correlated lines of
AC–A9OM1, Fig. 7(b), are also parallel but steeper, suggesting larger
Fig. 8. Correlation lines of (2P/I) versus current I for the (a) AC–A1OM9 and (b)
AC–A9O1 capacitors. Note these data are more adequately fitted in two segments
of high and low current.
Fig. 7. Correlation lines of (2P/I) versus current I for the (a) AC–AC and (b) AC–A9OM
capacitors.
1

Y.-S. Wang et al. / Electrochimica Acta 68 (2012) 95–102
101
Table 2
Values of Veff and Reff correlated from the discharge power results of electrochemical capacitors, AC–AC, AC–A9OM1, AC–A1OM9, and AC–A9O1 in potential window 1.0–1.8 V.
For AC–A1OM9 and AC–A9O1, there are two sets of correlated Veff and Reff for each potential window. One upper set is in the low current region, while the lower one is in the
high current region.
Potential window (V)
AC–AC
AC–A9OM
1
AC–A1OM
9
AC–A9O1
Veff (V)
Reff (ꢄ)
Veff (V)
Reff (ꢄ)
Veff (V)
Reff (ꢄ)
Veff (V)
Reff (ꢄ)
1.8
1.6
1.4
1.2
1.0
1.62
16
1.56
78
1.53
1.24
1.32
1.02
1.13
0.82
0.90
0.67
0.69
0.52
220
70
209
53
206
36
163
30
1.67
1.54
1.47
1.32
1.26
1.11
1.05
0.92
0.85
0.72
144
92
143
89
135
80
120
72
1.44
1.26
1.08
0.91
15
14
13
14
1.41
1.22
1.04
0.83
80
79
75
66
119
19
111
64
decreases rapidly with increasing current. And the (2P/I) value of
AC–A9O1 decreases not so fast when the current is higher than
2.4 mA.
are compared at the same power, and AC–A9OM1 suffers more
polarization loss than AC–AC. To maintain at the same power den-
sity, AC–A9OM1 is operated at higher current density, losing its
energy storage edge. The more polarization loss can be traced to the
higher resistance of Mo-doped MnO2, in contrast to AC. Concern-
ing AC–A1OM9 and AC–A9O1 that exhibit even more polarization
losses, their energy benefits are probably located in the region of
very small current densities. Hence their significance is much less
The correlated Veff and Reff values of AC–A1OM9 and AC–A9O1
are also listed in Table 2. Reff of AC–A1OM9 is higher than that of
AC–A9O1 in the low current region, consistent with the impedance
results. But in the high current region, Reff of AC–A1OM9 is less than
that of AC–A9O1, indicating that the current region is a significant
factor of polarization loss. Veff of AC–A9O1 is consistently higher
than that of AC–A1OM9, regardless the current region, again indi-
cating that addition of pseduocapacitive MnO2 diminishes Veff. It is
also of interest to contrast the Veff and Reff values of AC–A9OM1 and
AC–A9O1, because the only difference between the two capacitors
is the Mo doping on manganese oxide. The effective resistance of
AC–A9OM1 is less than that of AC–A9O1, especially in the low cur-
rent region. We attribute the lower resistance of AC–A9OM1 to Mo
doping.
The CD values of AC–AC and AC–A9OM1 in potential window
1.8 V are plotted in Fig. 9(a) to emphasize the capacitance bene-
fits of 10 wt% Mo-doped MnO2. When the current density is less
than 8.0 A g−1, the capacitance of AC–A9OM1 is higher than that of
AC–AC. The CD value of AC–A9OM1 is expected to become less than
that of AC–AC at current density >9.0 A g−1, since the capacitance of
AC–A9OM1 decreases faster with increasing current density, com-
pared with AC–AC. The capacitance of AC–A1OM9 varies even more
drastically, because the positive electrode contains 90 wt% Mo-
doped MnO2. Fig. 9(a) shows that the capacitance of AC–A1OM
9
can be higher than that of AC–AC, only when the current density
is less than 1.0 A g−1. Such a limitation arises from the fact that the
pseudocapacitance of Mo-doped MnO2 accompanies with an ohmic
resistance higher than AC.
Fig. 9(b) shows the Ragone plots of AC–AC and AC–A9OM1 dif-
fer considerably from those of AC–A1OM9 and AC–A9O1. And their
differences mainly reside in the high power region. AC–A9OM1,
similar to AC–AC, exhibits better power in a narrow energy density
range. On the other hand, the flat inverted L curves of AC–A1OM
9
and AC–A9O1 indicate the two capacitors have a broader range
of energy density at the similar power level. We further use
the power-energy data of symmetric AC–AC capacitor as bench-
mark, and discuss the influences of Mo-doping on the asymmetric
capacitor. Addition of 10 wt% Mo-doped MnO2 imparts AC–A9OM
1
more energy density, at the expense of AC power density. At
0.80 kW kg−1, the energy density of AC–A9OM1 is 5.2 Wh kg−1, 15%
higher than that of AC–AC, 4.5 Wh kg−1. Nevertheless, when the
power density is over 2.3 kW kg−1 (corresponding to current den-
sity 3.1 A g−1), the energy density of AC–A9OM1 becomes less than
that of AC–AC. In other words, the energy benefit of Mo-doping
MnO2 is restricted to current density less than 3.0 A g−1, which is
less than the capacitance benefit limitation 8.0 A g−1. This further
restriction arises because the energy densities of two capacitors
Fig. 9. (a) Variations of specific capacitance (CD) with respect to current density
of four capacitors; and (b) Ragone plots of AC–AC, AC–A9OM1, AC–A1OM9, and
AC–A9O1 capacitors in potential window 1.8 V.

102
Y.-S. Wang et al. / Electrochimica Acta 68 (2012) 95–102
than AC–A9OM1, considering the importance of high power capa-
bility to an electrochemical capacitor.
References
[1] P.J. Hall, M. Mirzaeian, S.I. Fletcher, F.B. Sillars, A.J.R. Rennie, G.O. Shitta-Bey, G.
Wilson, A. Cruden, R. Carter, Energy Environ. Sci. 3 (2010) 1238.
[2] M. Inagaki, H. Konno, O. Tanaike, J. Power Sources 195 (2010) 7880.
[3] P. Simon, Y. Gogotsi, Nat. Mater. 7 (2008) 845.
4. Conclusions
[4] S. Devaraj, N. Munichandraiah, J. Phys. Chem. C 112 (2008) 4406.
[5] C. Xu, F. Kang, B. Li, H. Du, J. Mater. Res. 25 (2010) 1421.
[6] X.H. Yang, Y.G. Wang, H.M. Xiong, Y.Y. Xia, Electrochim. Acta 53 (2007) 752.
[7] C. Ye, Z.M. Lin, S.Z. Hui, J. Electrochem. Soc. 152 (2005) A1272.
[8] M. Toupin, T. Brousse, D. Belanger, Chem. Mater. 14 (2002) 3946.
[9] M. Toupin, T. Brousse, D. Belanger, Chem. Mater. 16 (2004) 3184.
[10] V. Khomenko, E. Raymundo-Pinero, F. Beguin, J. Power Sources 153 (2006)
183.
[11] M. Xu, L. Kong, W. Zhou, H. Li, J. Phys. Chem. C 111 (2007) 19141.
[12] E. Raymundo-Pinero, V. Khomenko, E. Frackowiak, F. Beguin, J. Electrochem.
Soc. 152 (2005) A229.
[13] G.R. Li, Z.P. Feng, Y.N. Ou, D. Wu, R. Fu, Y.X. Tong, Langmuir 26 (2010)
2209.
[14] G.M. Jacob, Q.M. Yang, I. Zhitomirsky, J. Appl. Electrochem. 39 (2009) 2579.
[15] R. Amade, E. Jover, B. Caglar, T. Mutlu, E. Bertran, J. Power Sources 196 (2011)
5779.
[16] Y.J. Kang, B. Kim, H. Chung, W. Kim, Synth. Met. 160 (2010) 2510.
[17] F. Lufrano, P. Staiti, E.G. Calvo, E.J. Juarez-Perez, J.A. Menendez, A. Arenillas, Int.
J. Electrochem. Sci. 6 (2011) 596.
Three asymmetric capacitors have been prepared with a pos-
itive electrode of AC and MnO2. The influences of 10 or 90 wt%
of Mo-doped and undoped MnO2 are evaluated at the capacitor
level to explore the potential benefits of manganese oxide. Doping
of 15 mol% Mo does diminish the resistance of manganese oxide,
whose pseudocapacitance is higher than the double layer capac-
itance of activated carbon at low scan rate. But the conductivity
of doped manganese oxide does not match up to that of activated
carbon. Hence a high content of Mo-doped MnO2 in the positive
electrode, such as AC–A1OM9, or a low content of undoped man-
ganese oxide in the positive electrode, such as AC–A9O1, gives rise
to sizable ohmic polarization loss, which reduces the power perfor-
mance at high currents. Only a small fraction of 15 mol% Mo-doped
manganese oxide benefits the energy density of electrochemical
capacitor at a small penalty on the power side.
[18] M. Nakayama, A. Tanaka, Y. Sato, T. Tonosaki, K. Ogura, Langmuir 21 (2005)
5907.
[19] O. Barbieri, M. Hahn, R. Kotz, Carbon 43 (2005) 1303.
[20] T. Brousse, P.L. Taberna, O. Crosnier, R. Dugas, P. Guillemet, Y. Scudeller, Y. Zhou,
F. Favier, D. Belanger, P. Simon, J. Power Sources 173 (2007) 633.
[21] Z.Q. Li, C.J. Lu, Z.P. Xia, Y. Zhou, Z. Luo, Carbon 45 (2007) 1686.
[22] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, Handbook of X-ray Photo-
electron Spectroscopy, Physical Electronics Inc., Minnesota, 1995, p. 79.
[23] H.W. Nesbitt, D. Banerjee, Am. Mineral. 83 (1998) 305.
[24] H. Al-Kandari, A.M. Mohamed, F. Al-Kharafi, A. Katrib, J. Electron Spectrosc.
Relat. Phenom. 184 (2011) 472.
Acknowledgement
This work is financially supported by the National Science Coun-
cil of Taiwan through the project NSC100-2221-E011-079-MY3.
Appendix A. Supplementary data
[25] S. Al-Kandari, H. Al-Kandari, F. Al-Kharafi, A. Katrib, Appl. Catal. A 341 (2008)
160.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.electacta.2012.02.038.