Electrochimica Acta 90 (2013) 542–549

Contents lists available at SciVerse ScienceDirect

Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta

Charging and discharging electrochemical supercapacitors in the presence of

both parallel leakage process and electrochemical decomposition of solvent
Shuai Ban a , Jiujun Zhang a,∗ , Lei Zhang a , Ken Tsay a , Datong Song a , Xinfu Zou b
a
National Research Council of Canada, Vancouver, BC, Canada V6T 1W5
b
Department of Applied Mathematics, University of Western Ontario, London, ON, Canada N6A 5B7

a r t i c l e i n f o a b s t r a c t

Article history: Simple models for electrochemical supercapacitors are developed to describe the charge–discharge
Received 5 November 2012 behaviors in the presence of both voltage-independent parallel leakage process and electrochemical
Received in revised form decomposition of solvent. The models are validated by experimental data collected using a symmetric
14 December 2012
two-electrode test cell with carbon powder as the electrode layer material and stainless steel as the
Accepted 15 December 2012
current collector.
Available online 22 December 2012
© 2013 Published by Elsevier Ltd.
Keywords:
Electrochemical supercapacitor
Charge and discharge
Parallel leakage
Self-discharging
Solvent decomposition

1. Introduction and ionic liquid solutions which have wider electrode potential
windows [4,7–9]. From the energy density equation mentioned
Electrochemical supercapacitors (ESs) are considered important above, it is obvious that increasing cell voltage to enhance energy
energy efficiency devices for rapid energy storage and deliv- density is more effective than increasing the specific capacitance
ery. Among the advantages of ESs are high power density, long because the voltage is squared in the formula. However, if the cell
lifecycle, high efficiency, wide range of operating temperatures, voltage is too high, the electrochemical decomposition of solvent
environmental friendliness, and safety. ESs also serves as a bridg- may occur due to the limited electrode material stability and/or
ing function for the power/energy gap of traditional dielectric solvent’s thermodynamic decomposition windows. This solvent
capacitors. These characteristics have made ESs very competitive electrochemical decomposition could produce gaseous products,
for applications in electric hybrid vehicles, digital communication leading to the pressure build-up inside the cell, causing safety
devices such as mobile phones, digital cameras, electrical tools, concerns, and self-discharge [10–17]. Another challenge of super-
pulse laser technique, uninterruptible power supplies, and storage capacitors is the parallel leakages, causing a fast self-discharge,
of the energy generated by solar cells [1–3]. Unfortunately, there reducing the shelf-life. The main contributors to the parallel leak-
are still some challenges, such as low energy density, high cost, and ages of the supercapacitors can be summarized into four processes
high self-discharge rate, which have limited wider applications of [7]: (1) Faradaic reaction of electrolyte impurities; (2) parasitic
ESs [3,4]. redox reactions involving impurities (oxygen groups and metals);
In order to increase energy density, according to the energy den- (3) non-uniformity of charge acceptance along the surface of elec-
sity expression E = 1/2CV 2 , where E is the ES’s energy density, C is trode material pores; and (4) possible short-circuit of the anode and
the specific capacitance, and V is the cell voltage, respectively, two cathode from improperly sealed bipolar electrodes. Due to these
major approaches have been taken. One is to increase the specific parallel leakages, the shelf-life could be significantly reduced when
capacitance (C) of the electrode material [3–6], and the other is to compared to those of electrochemical batteries [7].
increase the cell voltage (V) using solvents such as non-aqueous To better understand the effects of parallel leakage process
and/or electrochemical decomposition of solvent on supercapa-
citors’ performance, some experimental and theoretical studies
∗ Corresponding author at: 4250 Wesbrook Mall, Vancouver, B.C. V6T 1W5, have been carried out to investigate the charge and discharge
Canada. Tel.: +1 604 221 3087; fax: +1 604 221 3001. behaviors [18–23]. Since the parallel leakage process and the sol-
E-mail address: [email protected] (J. Zhang). vent decomposition are mainly related to the electrochemical

0013-4686/$ – see front matter © 2013 Published by Elsevier Ltd.

http://dx.doi.org/10.1016/j.electacta.2012.12.056
 S. Ban et al. / Electrochimica Acta 90 (2013) 542–549 543

Nomenclature
Vno standard electrode potentials for negative electrode
Symbol, Meaning reaction (V)
˛n electron transfer coefficient in the rate determining Vp positive electrode potential of the solvent electro-
step of negative electrode reaction chemical decomposition (V)
˛p electron transfer coefficient in the rate determining Vpo standard electrode potentials for positive electrode
step of positive electrode reaction reaction (V)
bn Tafel slope of the negative electrode reaction Vsc supercapacitor voltage (V)
bp Tafel slope of the positive electrode reaction max
Vsc designed charging voltage of the supercapacitor (V)
C capacitance (F cm−2 ) o
Vsc maximum supercapacitor voltage (V)
Cdl double-layer capacitance (F cm−2 )
Cm specific capacitance (F g−1 )
−3
Cs mole concentration  of the solvent (mol cm ) processes occurring inside the supercapacitor cells, describing the
Vo = Vpo − Vno − bp ln ipo − bn ln (ino )
charge and discharge behaviors by electrochemical parameters,
e− electron
such as equivalent series resistance, capacitance, and several elec-
E energy density (Wh cm−2 )
trode kinetic constants, seems more appropriate, in particular for
(Em )max maximum energy density of the supercapacitor
electrochemical decomposition of the solvent.
(Wh kg−1 )
In this paper, for fundamental understanding of supercapacitor
F Faraday’s constant (96,487 C mol−1 )
charging and discharging behaviors, through experiment validation
iF current density induced by the solvent electrochem-
we present some simple mathematical models incorporating the
ical decomposition (A cm−2 )
voltage-independent parallel leakage process and electrochemical
in current density of the negative electrode reaction
decomposition to describe the supercapacitor charge and dis-
(A cm−2 )
charge behaviors. Using these models, the experiment data could
ip current density of the positive electrode reaction
be simulated to obtain the desired values of parameters such as
(A cm−2 )
specific capacitance and equivalent series resistance from which
isc charging or discharging current density of the
both the energy and power density can be calculated. The sim-
supercapacitor (A cm−2 )
ulated parameter values related to both parallel leakages and
Icell constant current density for charging or discharging
solvent decomposition could be used to predict supercapacitor’s
the supercapacitor (A cm−2 )
self-discharge and shelf-life, performance/efficiency losses, as well
kn reaction constant of the negative electrode reaction
as the limiting workable cell voltage of the supercapacitors. There-
(cm s−1 )
fore, the proposed models are useful in evaluating and diagnosing
kp reaction constant of the positive electrode reaction
supercapacitor cells. We hope these simple models could be used as
(cm s−1 )
a tool for understanding the charging–discharging behavior of the
m mass of the electrode material (g)
supercapacitors. Furthermore, we believe that these models could
n˛n electron transfer number in the rate determining
also be useful in obtaining the essential parameters of the supercap-
step of negative electrode reaction
acitors such as equivalent series resistance and capacitance based
n˛p electron transfer number in the rate determining
on the recorded curves of charging–discharging.
step of positive electrode reaction
nn overall electron transfer number in the negative
2. Experimental
electrode reaction
np overall electron transfer number in the positive
2.1. Electrode layer preparation
electrode reaction
pn product of the negative electrode reaction
For electrode layer materials, Carbon black BP2000 (Cabot Cor-
pp product of the positive electrode reaction
poration), conducting carbon Super C45 (from TIMCAL), and 60 wt%
(Pm )max maximum power density of the supercapacitor
Polytetrafluoroethylene (PTFE)-water dispersion (Aldrich) were
(W kg−1 )
used as received without further modification.
R universal gas constant (8.314 J K−1 mol−1 )
For electrode layer preparation, the procedure has been
Resr equivalent series resistance of the supercapacitor
described in our previous publication [3], and is repeated as below.
( cm2 )
Carbon and conducting carbon powders were first mixed with a
Rlk equivalent leakage resistance ( cm2 )
Vortex Mixer (Thermo Scientific) for 30 min to form a uniformly-
S reactant of the solvent
mixed powder, which was then transferred into a beaker containing
t time (s)
both PTFE binder and ethanol solution under constant stirring to
tfd time needed for a supercapacitor being fully dis-
form a suspension. After the solvent was evaporated at high tem-
charged (s)
perature, the paste formed was collected on a glass slide. This paste
T temperature (K)
was manipulated by repeatedly folding and pressing using a spatula
Ts time needed when the supercapacitor voltage goes
until a sufficient mechanical strength was achieved. Then this paste
down to 95% of its maximum voltage (s)
was rolled to a thin electrode sheet with the required thickness
V voltage (V)
using a MTI rolling press. Finally, this electrode sheet was placed
Vcell voltage of the supercapacitor cell (V)
max into a vacuum oven at 90 ◦ C under active vacuum for at least 12 h.
Vcell maximum voltage of the supercapacitor cell (V)
The formed dry electrode sheet was cut into two 2 × 2 square cen-
VF voltage induced by the solvent electrochemical
timeters (cm2 ) for electrode layers, which were then sandwiched
decomposition (V)
with a 30 ␮m thick porous PTFE separator (W. L. Gore & Associates,
Vn negative electrode potential of the solvent electro-
Inc.) in the middle, forming an electrode separator assembly (ESA).
chemical decomposition (V)
The assembly was then impregnated with an aqueous electrolyte
containing 0.5 M Na2 SO4 inside a vacuum oven at 60 ◦ C for at least
544 S. Ban et al. / Electrochimica Acta 90 (2013) 542–549

one hour. This electrolyte-impregnated ESA was then assembled

into the two-electrode test cell.

2.2. Two-electrode test cell measurements

In the two-electrode test cell [24], the active electrode sur-

face for both positive and negative electrodes was 4.0 cm2 . The
electrolyte was 0.5 M Na2 SO4 aqueous solution, and the cur-
rent collectors were made of stainless steel. The additional
effect of electrolyte concentration on supercapacitor’s perfor-
mance was carefully studied in pour previous publication [24]. For
electrochemical measurements, both cyclic voltammograms and
galvanic charge–discharge curves were recorded using a Solartron
1287 potentiostat. For cell internal resistance measurements, AC
impedance spectra at a frequency range of 0.1 Hz–10 kHz were
collected using a Solartron FRA analyzer. For a more detailed
description about the measurement and calculation procedures,
please see our previous publication [24].
For recording charging/discharging curves of supercapacitors,
there are mainly two options, one is to control a constant
charging–discharging current, then record the change of cell volt-
age with time, and the other is to control a constant cell voltage,
then record the change of cell current with time. The most popular
option is the former. In this paper, we used the former option to
obtain the charging–discharging curves.
All measurements were made at room temperature and ambient
pressure.

3. Results and discussion

In general, for charging and discharging a supercapacitor, there

are two major options. One is the charging or discharging at a con- Fig. 1. Proposed equivalent circuits of a double-layer supercapacitor at a constant
stant cell voltage to record the cell current change with time, and current charging (a) and discharging (b) in the absence of both parallel leakage
the other is charging or discharging at a constant current to record process and electrochemical decomposition of the solvent. K: electric switcher; Resr :
that cell voltage change with time. In this paper, we will only focus equivalent series resistance; Icell : constant current for charging or discharging the
supercapacitor; Vcell : supercapacitor cell voltage; Cdl : double-layer capacitance; Vsc :
on the charging and discharging at a constant current. voltage across the Cdl ; and isc : current used to charge or discharge Cdl ; respectively.

3.1. Charging at a constant cell current in the absence of parallel

leakage process and electrochemical decomposition of solvent Eq. (3) suggests that the cell voltage is linearly proportional to the
charge time. If the supercapacitor is charged to a designed cell volt-
max , a discharge process at a constant current (I
age of Vcell
Fig. 1(a) depicts a proposed equivalent circuit for a double- cell ) can be
layer supercapacitor with a constant current in the absence of both started immediately, as shown in Fig. 1(b). Note that at this desired
cell voltage (or maximum cell voltage) of Vcellmax , the voltage across
voltage-independent parallel leakage process and electrochemical
o , should be given by:
the double layer capacitance (Cdl ), Vsc
decomposition of the solvent, where K is the electric switcher, Resr
is the equivalent series resistance, Icell is the constant current for o
Vsc max
= Vcell − Icell Resr (4)
charging or discharging the supercapacitor; Vcell is the superca-
o will be the load to discharge
in the case of discharging, this Vsc
pacitor cell voltage; Cdl is the double-layer capacitance; Vsc is the
voltage across the Cdl ; and isc is the current used to charge or dis- the supercapacitor. The cell voltage (Vcell ) can be expressed by the
charge, respectively. Note that in the case shown in Fig. 1(a), the following equation
current used to charge or discharge the double layer capacitance
o Icell
(isc ) is equal to the charging current (Icell ), which has a constant Vcell = Vsc − Icell Resr − t (discharge process) (5)
Cdl
value:
As an example, Fig. 2 shows the charge–discharge curves using a
Icell = isc (1)
two-electrode test cell in which two electrodes are identical (sym-
assuming that, before the charging starts, the supercapacitor is at metric cell). Based on both Eqs. (3) and (5), several parameters
zero-charge state, that is, the voltage across the supercapacitor (Vsc ) such as the double-layer capacitance (Cdl ), the voltage across the
double layer capacitance at the end of charging (Vsc o ), as well as
is equal to zero. When the switch is turned on at t = 0 (t is the charg-
ing time), the charge process is then started and the voltage across equivalent series resistance (Resr ) can be simulated, respectively,
the supercapacitor can be expressed as: using both the charge and discharge curves. In the case shown in
 t=t  t=t Fig. 2, parameters obtained using the simulation are: capacitance
1 1 Icell Cdl = 0.127 F cm−2 or 0.509 F for the entire cell with electrode area
Vsc = isc dt = Icell dt = t (2)
Cdl t=0
Cdl t=0
Cdl of 4 cm2 , or the specific capacitance (Cm ) of 34 F g−1 for the carbon
loading (m) of 0.015 g for each electrode; the voltage across the dou-
therefore, the supercapacitor cell voltage can be expressed as: o = 0.991 V; as well as the equivalent series
ble layer capacitance Vsc
Icell resistance Resr = 2.36  cm2 or 0.589  for the entire cell, respec-
Vcell = Icell Resr + Vsc = Icell Resr + t (charge process) (3)
Cdl tively. Note that the specific capacitance obtained here was also
 S. Ban et al. / Electrochimica Acta 90 (2013) 542–549 545

Fig. 2. Charge–discharge curves recorded using a symmetric supercapacitor cell

with a geometric area of 4.0 cm2 for each electrode, in 0.5 M Na2 SO4 aqueous solu-
tion at ambient conditions. Stainless steel sheets as both the positive and negative
current collectors. Charge and discharge current density: 0.025 A cm−2 , and total BP
2000 carbon loading of electrode layers: 3.8 mg cm−2 . Note that the Resr showing
in the figure should be the sum equivalent series resistances of both positive and
negative electrodes [24]. Note that the curve corresponds to the simulation and the
points to experimental data.

confirmed by using cyclic voltammetry measurements. Using these

values, the maximum energy and power densities ((Em )max and
(Pm )max , respectively) can be calculated according to the equations
Fig. 3. Proposed equivalent circuits of a double-layer supercapacitor by constant
presented in our previous publication [24]: (Em )max = 1/2Cm (Vsc o )2 ,
current charging (a) and discharging (b) in the presence of parallel leakage process.
o 2
and (Pm )max = 1/4m((Vsc ) /Resr ). The obtained (Em )max and (Pm )max K: electric switch; Rlk : equivalent parallel leakage resistance; Resr : equivalent series
are 4.64 Wh kg−1 , and 2.78 × 104 Wh kg−1 , respectively. resistance; Icell : constant current for charging or discharging the supercapacitor;
Vcell : supercapacitor cell voltage; Cdl : double-layer capacitance; Vsc : voltage across
the Cdl ; and isc : current used to charge or discharge Cdl ; respectively.
3.2. Charging at a constant cell current in the presence of a
voltage-independent parallel leakage process o which is equal to I
as Vsc cell Rlk as t tends to infinity. If there is no
leakage current, Rlk → ∞, for small t, Eq. (8) can be approximated by
Fig. 3 shows the scheme of a supercapacitor charging at a con-
stant current density. Compared to Fig. 1, there is an additional t
Vcell = Icell Resr + Icell (charge process) (9)
equivalent leakage resistance (Rlk ) which represents a voltage- Cdl
independent parallel leakage process. Note that for simplicity, we note that if charging an ideal supercapacitor (Resr = 0, and Rlk = ∞)
treat all voltage-insensitive leakage processes as an equivalent par- using a constant current, the cell voltage will become:
allel leakage (Rlk ), suggesting that the magnitude of this Rlk would
t
not significantly change with the charge or discharge process. How- Vcell = Icell (10)
ever, we do need to keep it in mind that this Rlk may change with Cdl
cell voltage in practice. In a later section, we will give a detailed discussion about the
Again, we assume that before the charge process starts, the effect of solvent electrochemical decomposition on the superca-
supercapacitor is at zero-charge state, that is, the voltage across pacitor charge and discharge behaviors.
the supercapacitor is equal to zero. When the switch shown in After the supercapacitor cell is fully charged with a maximum
Fig. 3(a) is closed at t = 0, the charge process is started. When t ≥ 0, voltage of (Vsco ), a discharge process using a constant current (I
cell )
the supercapacitor charging voltage can be expressed by will be started for which the equivalent circuit can be illustrated in
dVsc Vsc Fig. 3(b). The integration equation used to describe this situation is:
Cdl + − Icell = 0 (6)  t=t
dt Rlk 1 o
isc dt + isc Rlk − (Vsc + Icell Rlk ) = 0 (11)
solving Eq. (6) with the initial condition t = 0, Vsc = 0, the following Cdl t=0
solution can be obtained:
   solving this equation, we can get:
Vsc = Icell Rlk 1 − exp −
t
(charge process) (7) o +I
(Vsc
 
cell Rlk ) t
Rlk Cdl isc = exp − (12)
Rlk Rlk Cdl
accordingly the cell voltage can be expressed as:
   and the voltage across the double-layer capacitance can be
t expressed as:
Vcell = Icell Resr + Icell Rlk 1 − exp − (charge process)
Rlk Cdl  t=t
(8) o 1
Vsc = Vsc − isc dt
Cdl t=0
Eq. (8) shows that at t = 0, Vcell = Icell Resr , and when t = ∞,   
Vcell = Icell Resr +Icell Rlk . In this case, the voltage across the superca- o o t
= Vsc − (Vsc + Icell Rlk ) 1 − exp − (13)
pacitor (Vsc ) increasingly approaches its maximum value, defined Rlk Cdl
546 S. Ban et al. / Electrochimica Acta 90 (2013) 542–549

Fig. 4. Theoretically calculated charge–discharge curves, Rlk is changed from 100 to

2
T
10,000  cm2 with fixed Resr = 0.5 .cm2 , Cdl = 0.4 F/cm , and Icell = 0.05 A/cm2 .

the cell voltage (Vcell ) can be expressed by:

o o
Vcell = −Icell Resr + Vsc − (Vsc + Icell Rlk )
  t

× 1 − exp − (discharge process) (14)
Rlk Cdl

if there is no parallel leakage process (Rlk → ∞), for limited t, Eq.

(14) can be approximated by

o t Fig. 5. Proposed equivalent circuits for a double-layer supercapacitor at a constant

Vcell = −Icell Resr + Vsc − Icell (15) current charging (a) and discharging (b) in the presence of electrochemical decom-
Cdl
position of the solvent. K: electric switcher; Resr : equivalent series resistance; Icell :
constant current for charging or discharging; Vcell : supercapacitor cell voltage; iF :
the supercapacitor full discharge time, denoted by tfd , can be current of solvent electrochemical decomposition; Cdl : double-layer capacitance;
derived by setting Vsc = 0 and solving this equation for t, giving isc : current used to charge or discharge the double-layer capacitance; VF : electrode
 Icell Rlk
 potential of the solvent electrochemical decomposition; and Vsc (=VF ): voltage across
the double-layer capacitance, respectively.
tfd = −Rlk Cdl ln o (16)
Vsc + Icell Rlk
Assuming that the solvent (S) decomposition reactions at the
As an example, Fig. 4 shows the theoretically calculated charge
positive and negative electrodes are expressed by the following
and discharge curves according to Eqs. (8) and (14). It can be seen
Reactions (I) and (II), respectively:
that when the parallel leakage resistance becomes smaller (larger
leakage current) from 10,000 to 100  cm2 , both the cell voltage S → np e− + Pp (I)
and the maximum supercapacitor voltage become smaller, suggest-
−
ing that the performance of the supercapacitor cell is reduced. S + nn e → Pn (II)
Furthermore, due to the parallel leakage process, self-discharge here np is the overall electron number of S oxidation to produce
will become significant. For example, if a fully charged supercapac- Pp , and nn is the overall electron number of S reduction to produce
itor with a voltage of Vsc o is put on the shelf (I
cell = 0 in this case), Pn , respectively. Assuming that both reactions are irreversible, the
according to Eq. (13), the voltage will be gradually reduced due to current density for Reaction (I) (ip ) and that for Reaction (II) (in ) can
the self-discharge through Rlk . When the Vsc goes down to 95% of be expressed by Eqs. (18) and (19), respectively:
max , the time (T ) for V to reach this level is given by:
Vsc s sc
˛p n˛p F(Vp − Vpo ) Vp − Vpo
Ts = 2.996Rlk Cdl (17) ip = np Fkp CS exp = ipo exp (18)
RT bp

in the situation of even Rlk = 10,000  cm2 , and Clk = 0.4 F cm−2 ,
 ˛ n F(V o − V )   Vo − V 
n ˛n n n n n
in = nn Fkn CS exp = ino exp (19)
Ts = 3.33 h. Therefore, self-discharge is one of the major challenges RT bn
for supercapacitors.
Here ipo = np Fkp CS and ino = nn Fkn CS can be defined as the
standard current densities, bp = RT/˛p n˛p F and bn = Rt/˛n n˛n F can
3.3. Charging at a constant cell current in the presence of be defined as the Tafel slopes for Reactions (I) and (II); kp and
electrochemical decomposition of solvent kn are the reaction constants, ˛p and ˛n are the charge transfer
coefficients, n˛p and n˛n are the electron transfer numbers in the
Fig. 5(a) shows the proposed equivalent circuit for a double- reaction determining steps, Vpo and Vno are the standard electrode
layer supercapacitor in the case of constant current charging in the potentials for Reactions (I) and (II); Vp and Vn are the electrode
presence of electrochemical decomposition of the solvent. Com- potential of the positive and negative electrodes, respectively. CS is
pared to Fig. 1, there is an additional parallel electrochemical the mole concentration; F is Faraday’s constant; R is the universal
process included with a potential dependent current (iF ) and volt- gas constant; and T is the temperature; respectively.
age (VF ) induced by the solvent electrochemical decomposition. Note that because both Reactions (I) and (II) represent the
From the equivalent circuits in Fig. 5, it can be seen that Vsc = VF . solvent decomposition, the solvent concentration should not
 S. Ban et al. / Electrochimica Acta 90 (2013) 542–549 547

be changed with its consumption by electrochemical reactions. Using Eq. (28), the maximum voltage (Vsc o ) across the double-

Therefore, there is no mass transfer limitation involved in the elec- layer capacitance can also be derived by setting t → ∞ (the moment
trochemical reactions. when the supercapacitor is fully charged):
From Eqs. (18) and (19), their electrode potentials can be derived
as Eqs. (20) and (21), respectively: o V o
Vsc = (bp + bn ) ln Icell exp (30)
bp + bn
Vp = Vpo − bp ln(ipo ) + bp ln(ip ) (20)
The maximum supercapacitor cell charging voltage (Vcell max ) can

Vn = Vpo + bn ln(ino ) − bn ln(in ) (21) be expressed as the sum of the voltage drop across the equivalent
series resistance (Resr ) and the maximum voltage across the double-
from Fig. 5, it can be seen that the voltage drop across the double- o ):
layer capacitance (Vsc
layer capacitance (Vsc ) can be expressed as:
max o
Vcell = Icell Resr + Vsc
Vsc = Vp − Vn = Vpo − Vno − bp ln(ipo ) − bn ln(ino ) + bp ln(ip ) + bn ln(in )
(22)
= V o + bp ln(ip ) + bn ln(in ) V o
= Icell Resr + (bp + bn ) ln Icell exp (31)
bp + bn
where

V o = Vpo − Vno − bp ln(ipo ) − bn ln(ino ) (23) for example, if the solvent is water, one has Vpo = 1.23V
and Vno = 0.00 V vs. NHE. Assuming that the kinetic param-
for a decomposition reaction, ip = in , they should also be equal to iF , eters bp = bn = 0.0514 V/dec., ip = 1.0x10−7 A cm−2 , in = 1.0x10−3
as is shown in Fig. 5, that is, iF = ip = in . Substituting this into Eq. (22), ↑
A.cm- −2 for water decomposition (H2 O → 21 O2 + 2H+ + 2e− ), and
the following equation for iF can be obtained: Cdl = 0.125 F cm−2 , Vsc
o can then be calculated to be 2.034 V at a

charging current density of 0.025 A cm−2 .

Vsc − V o
iF = exp (24) For the discharge case shown in Fig. 5(b), the integration equa-
bp + bn
tion can be written as:
In the case of constant current charging as shown in Fig. 5, we  t=t o − V o
1 Vsc
assume that before the charging starts, the voltage across the super- isc dt + ln(isc − Icell ) − =0 (32)
(bp + bn )Cdl t=0
bp + bn
capacitor (Vsc ) is equal to zero. Then the charging current (isc ) used
to charging double-layer capacitance (Cdl ) can be expressed as: Solving this equation, we can get isc from which the voltage
across the double layer capacitance (Vsc ) can be derived respec-
dVsc
isc = Cdl (25) tively:
dt
o − V o /b + b ) + I
Icell (exp(Vsc
in Fig. 5(a), the constant charging current can be expressed as: p n cell )
isc = o − V o /b + b )(1 − exp(−(I
Icell + exp(Vsc p n cell /(bp + bn )Cdl ))t)
Icell = isc + iF (26) (33)

 o − V o /b + b ) + I
 o o
o
exp(Vsc p n cell exp(Icell /(bp + bn )Cdl )t) − exp(Vsc − V /bp + bn )
Vsc = Vsc − (bp + bn ) ln (34)
Icell

o . The supercapacitor
Eq. (34) indicates that when t = 0, Vsc = Vsc
cell charging voltage (Vcell ) can be expressed as:
Vcell = −Icell Resr + Vsc
  (discharge process) (35)
o
(exp(Vsc − V o /bp + bn ) + Icell ) exp((Icell /(bp + bn )Cdl )t) − exp(Vsc
o
− V o /bp + bn )
= −Icell Resr + Vsc
o
− (bp + bn ) ln
Icell

As an illustration, Fig. 6 shows the charge and discharge curves

calculated according to Eqs. (29) and (35). It can be seen that the sol-
Combining Eqs. (24), (25) and (26), the following differential
vent decomposition plateau voltages (Vo Eq. (23)) are all higher
equation can be obtained:
than the thermodynamic voltages because of the limited kinetic
dVsc −V o Vsc processes. However, for a safe operation of supercapacitor, it is
Cdl + exp exp − Icell = 0 (27) better not to charge the cell to a thermodynamic voltage of sol-
dt bp + bn bp + bn
vent decomposition. For example, if using water as the solvent, the
maximum charging voltage should not be higher than 1.23 V.
With the initial condition Vsc = 0 at t = 0, Eq. (27) can be solved
to obtain:

Icell
Vsc = (bp + bn ) ln (charge process) (28)
Icell − exp(−V o /bp + bn )) exp(−Icell /(bp + bn )Cdl t) + exp(−V o /bp + bn )

The supercapacitor cell charging voltage (Vcell ) should be the

It is worthwhile to point out that the model shown in Fig. 5(a)
sum of the voltage drop across the equivalent series resistance (Resr )
and its associated equations discussed above do not include the
and the voltage across the double-layer capacitance (Vsc ):
reactant mass transfer limitation since there may not be a limited
Vcell = Icell Resr + Vsc

Icell (charge process) (29)

= Icell Resr + (bp + bn )
Icell − exp(−V o /bp + bn )) exp(−Icell /(bp + bn )Cdl t) + exp(−V o /bp + bn )
548 S. Ban et al. / Electrochimica Acta 90 (2013) 542–549

Fig. 6. Calculated charge–discharge curves according to Eqs. (29) and (35) using
the following assumed parameters: bp = bn = 0.0514 V/dec., ip = 1.0 × 10−7 A cm−2 ,
in = 1.0 × 10−3 A cm2 , Cdl = 0.125 F cm−2 , and Icell = 0.025 A cm−2 at three different
thermodynamic cell voltages (Vp − Vn ) (1.23, 2.20, and 3.20 V, respectively).

supply of reactants. However, please also note that the build of

product gasses in a closed cell, such as O2 and H2 if using an aqueous
electrolyte, may lead to mass transport limitations and also pres-
sure building-up, which will have some impact on the accuracy of
Fig. 7. Proposed equivalent circuits for a double-layer supercapacitor at a constant
the model.
current charging (a) and discharging (b) in the presence of electrochemical decom-
position of the solvent. K: electric switcher; Resr : equivalent series resistance; Rlk :
3.4. Charging at a constant cell current in the presence of both parallel leakage resistance; ilk : parallel leakage current; Icell : constant current for
voltage-independent parallel leakage process and electrochemical charging or discharging; Vcell : supercapacitor cell voltage; iF : current of solvent
decomposition of solvent electrochemical decomposition; Cdl : double-layer capacitance; isc : current used to
charge or discharge the double-layer capacitance; VF : electrode potential of the sol-
vent electrochemical decomposition; and Vsc (=VF ): voltage across the double-layer
Fig. 7(a) shows the proposed equivalent circuit of a double- capacitance, respectively.
layer supercapacitor in the case of constant current charging in
the presence of both parallel leakage process and electrochem-
ical decomposition of the solvent. The integration equation at a For the discharge process, the cell voltage in this figure is
constant charging current (Icell ) can be written as: expressed as:
 t=t  t=t
−V o 1 1
isc + exp exp isc dt o
Vcell = −Icell Resr + Vsc = −Icell Resr + Vsc − isc dt (39)
bp + bn (bp + bn )Cdl t=0 Cdl t=0
 t=t
1
+ isc dt − Icell = 0 (36)
Rlk Cdl t=0

and for the discharge process at a constant current (Icell ):

o − V o
 t=t
Vsc 1
isc − exp exp − isc dt
bp + bn (bp + bn )Cdl t=0
 t=t
1
− isc dt − Icell = 0 (37)
Rlk Cdl t=0

Unfortunately, both of these two equations cannot be solved

analytically, and numeric solutions will be sought using the soft-
ware Matlab.
Fig. 8 shows the numeric solutions for both charge and discharge
curves with several sets of assumed parameters in the presence of
Fig. 8. Charge–discharge curves plotted by numeric solutions of Eqs. (36)–(39)
both voltage-independent parallel leakage resistance and solvent
using the following assumed parameter sets: set #1 bp = bn = 0.0514 V/dec.,
decomposition. The cell voltage (Vcell ) for the charge process in this Vo = 2.413 V, Rlk = 10,000  cm2 , Cdl = 0.125 F cm−2 , and Icell = 0.025 A cm−2 ,
figure is expressed as: max
Vsc = 2.034 V, Resr = 2  cm2 ; set #2 bp = bn = 0.0514 V/dec., Vo = 2.413 V,
Rlk = 100  cm2 , Cdl = 0.125 F cm−2 , and Icell = 0.025 A cm−2 , Vscmax
= 2.034 V,

t=t
Resr = 2  cm2 ; set #3 bp = bn = 0.0514 V/dec., Vo = 2.413 V, Rlk = 100  cm2 ,
1
Vcell = Icell Resr + Vsc = Icell Resr + isc dt (38) −2 −2
Cdl = 0.125 F cm , and Icell = 0.1 A cm , Vsc = 2.034 V, Resr = 2  cm ; set #4
max 2

Cdl bp = bn = 0.0514 V/dec., Vo = 2.413 V, Rlk = 100  cm2 , Cdl = 0.125 F cm−2 , and
t=0 Icell = 0.01 A cm−2 , Vsc
max
= 2.034 V, Resr = 2  cm2 , respectively.
 S. Ban et al. / Electrochimica Acta 90 (2013) 542–549 549

Acknowledgements

The authors would like to thank the support from National

Research Council of Canada (NRC) and Transport Canada. Tech-
nical support from and discussion with Dr. Lucie Robitalille, Dr.
Alexis Laforgue, Dr. Dongfang Yang, Dr. Yves Grincourt, Mr. Francois
Girard, and Mr. Ryan Baker are highly appreciated.

References

[1] F. Béguin, E. Raymundo-Piňero, Carbons for Electrochemical Energy Storage

and Conversion Systems, CRC Press, Boca Raton, FL, 2010.
[2] R. Kötz, M. Carlen, Principles and applications of electrochemical capacitors,
Electrochimica Acta 45 (2000) 2483.
[3] G. Wang, L. Zhang, J. Zhang, A review of electrode materials for electrochemical
supercapacitors, Chemical Society Reviews 41 (2012) 797.
[4] B.W. Ricketts, Self-discharge of carbon-based supercapacitors with organic
electrolytes, Journal of Power Sources 89 (2000) 64.
Fig. 9. Model fitting of Galvanic Cycledata with the measured param- [5] C. Largeot, C. Portet, J. Chmiola, P. Taberna, Y. Gogotsi, P. Simon, Supercapacitors
eters: Cdl = 0.1 F cm−2 , (a) Icell = 0.0025 A cm−2 and (b) Icell = 0.00125 A cm−2 based on flexible graphene/polyaniline nanofiber composite films, Journal of
Resr = 0.55  cm2 . The fitting parameters are bp = bn = 0.065 V/dec., Vo = 2.413 V, the American Chemical Society 130 (2008) 2730.
max
Rlk = 500  cm2 , and, Vsc = 2.034 V. [6] J.S. Huang, B.G. Sumpter, V. Meunier, Theoretical model for nanoporous carbon
supercapacitors, Angewandte Chemie International Edition 47 (2008) 520.
[7] B.E. Conway, Electrochemical Supercapacitors, Scientific Fundamentals and
Technological Applications, Kluwer Academic/Plenum Publishing, New York,
To validate the models, the measured charge–discharge data 1999.
[8] A. Lewandowski, A. Olejniczak, M. Galinski, I. Stepniak, Performance of
at two different current densities are shown in Fig. 9 together carbon–carbon supercapacitors based on organic, aqueous and ionic liquid
with those simulated curves according to Eqs. (36)–(39). It can electrolytes, Journal of Power Sources 195 (2010) 5814.
be seen that the model is reasonably fitting the experiment data, [9] A. Lewandowski, M. Galinski, Practical and theoretical limits for electrochem-
ical double-layer capacitors, Journal of Power Sources 173 (2007) 822.
suggesting that the model developed in this paper can be used [10] H. Zhao, Y. Zhang, Self-discharge analysis and characterization of supercap-
to describe the charge and discharge behaviors of electrochemical acitors for environmentally powered wireless sensor network applications,
supercapacitors in the presence of both voltage-independent par- Journal of Power Sources 196 (2011) 8866.
[11] K. Chiba, T. Ueda, Y. Yamaguchi, Y. Oki, F. Shimodate, K. Naoi, Electrolyte sys-
allel leakage process and solvent decomposition. However, there
tems for high withstand voltage and durability I. linear sulfones for electric
is a slightly mismatching, in particular at the discharge curves. double-layer capacitors, Journal of the Electrochemical Society 158 (2011)
The difference between modeling and experiments could be orig- A872.
inated from experimental trivial factors including the effect of [12] P. Kurzweil, M. Chwisetek, Electrochemical stability of organic electrolytes in
supercapacitors: spectroscopy and gas analysis of decomposition products,
gaseous evolution, dissolution of current collectors, electrolyte Journal of Power Sources 176 (2008) 555.
crystallization in separator, assembly quality, and so on. More [13] A.M. Bittner, M. Zhu, Y. Yang, H.F. Waibel, M. Knouma, U. Starke, C.J. Weber,
work is definitely needed to take care of these affecting fac- Ageing of electrochemical double layer capacitors, Journal of Power Sources
203 (2012) 262.
tors. [14] I. Nicotera, G.D. McLachlan, G.D. Bennett, I. Plitz, F. Badway, G.G. Amatucci,
S.G. Greenbaum, Solid-state NMR characterization of electrolyte breakdown
4. Conclusions products in nonaqueous asymmetric hybrid supercapacitors, Electrochemical
and Solid-State Letters 10 (2007) A5.
[15] T. Abdallah, D. Lemordant, B. Claude-Montigny, Are room temperature ionic liq-
Mathematical models for electrochemical supercapacitors are uids able to improve the safety of supercapacitors organic electrolytes without
developed to describe the charge–discharge behaviors in the degrading the performances? Journal of Power Sources 201 (2012) 353.
[16] R. Kotz, M. Hahn, P. Ruch, R. Gallay, Comparison of pressure evolution in
presence of voltage-independent parallel leakage process and supercapacitor devices using different aprotic solvents, Electrochemistry Com-
electrochemical decomposition of solvent. Four charge–discharge munications 10 (2008) 359.
cases are discussed: (1) an ideal double-layer supercapacitor, [17] S. Ishimoto, Y. Asakawa, M. Shinya, K. Naoi, Degradation responses of activated-
carbon-based EDLCs for higher voltage operation and their factors, Journal of
(2) a supercapacitor in the presence of a voltage-independent the Electrochemical Society 156 (2009) A563.
parallel leakage process, (3) a supercapacitor in the presence of [18] E. Tironi, V. Musolino, Supercapacitor characterization in power electronic
an electrochemical decomposition of solvent, and (4) a super- applications: proposal of a new model, 2009 International Conference on Clean
Electrical Power (IEEE) (2009) 376.
capacitor in the presence of both voltage-independent parallel
[19] F. Rafik, H. Gualous, R. Gallay, A. Crausaz, A. Berthon, Frequency, thermal
leakage process and electrochemical decomposition of solvent. and voltage supercapacitor characterization and modeling, Journal of Power
To validate these models, a symmetric two-electrode test cell Sources 165 (2007) 928.
with carbon powder as the electrode layer material and stain- [20] Z. Cheng, W. Chen, Q. Li, Z. Jiang, Z. Yang, Asia Pacific Power and Energy Engi-
neering Conference (IEEE) (2010) 1.
less steel as the current collector was assembled and used for [21] 2008 IEEE Power and Energy Society General Meeting Conversion and Delivery
charge–discharge data collection. A comparison of the simulation of Electrical Energy in the 21st Century (IEEE), 1–7 (2008).
results based on the models and the experimental data seems [22] A.S. Weddell, G.V. Merrett, T.J. Kazmierski, B.M. Al-Hashimi, Accurate superca-
pacitor modeling for energy harvesting wireless sensor nodes, IEEE Transaction
to conclude that these models are valid in describing the charge on Circuits and Systems. II: Express Brief 58 (2011) 911.
and discharge behaviors of the supercapacitors. The analytical [23] P. Bjornbom, Charge/discharge of an electrochemical supercapacitor electrode
results are also helpful in identifying possible concerns of super- pore; non-uniqueness of mathematical models, Electrochemistry Communica-
tions 9 (2007) 211.
capacitors. For example, we have seen in Section 3.3 that the [24] K. Tsay, L. Zhang, J. Zhang, Effects of electrode layer composition/thickness and
model predicts that self-discharge would be a major challenge for electrolyte concentration on both specific capacitance and energy density of
supercapacitor. supercapacitor, Electrochimica Acta 60 (2012) 428.