Electrochimica Acta 50 (2005) 3262–3272

Analysis of ac impedance of AgO-Zn cells: effects of state-of-charge,

temperature and cycle-life
P. Suresha , D.H. Nagarajub , A.K. Shuklaa , N. Munichandraiahb,∗
a Solid State and Structural Chemistry Unit, Indian Institute of Science, Bangalore 560012, India
b Department of Inorganic and Physical Chemistry, Indian Institute of Science, Bangalore 560012, India

Received 30 July 2004; received in revised form 4 November 2004; accepted 27 November 2004

Abstract

The ac impedance of commercial silver oxide-zinc cells has been studied in the temperature range −20 to 40 ◦ C at various state-of-charge
(SOC) values in conjunction with the effect of charge–discharge cycling. The impedance data in the Nyquist form comprise an inductive part
in the frequency region 100–10 kHz, and a capacitive semicircle in the frequency region 10 kHz–10 mHz in the SOC range between 0 and 0.4,
or two capacitive semicircles at SOC ≥ 0.4. The impedance parameters have been evaluated by analysis of the data using an equivalent circuit
and a non-linear least squares fitting procedure. The inductance has been found to be independent of both temperature and SOC of the cell.
The resistance values corresponding to the semicircles of the Nyquist impedance plot, which are attributed to the positive electrode, show a
strong dependence on the SOC of the cell. The energy of activation for the charge-transfer reaction has been calculated from the temperature
dependence studies. There is a decline in the capacity values of the cells on cycling, while the constant phase elements are found to increase
with increasing cycle number. The scanning electron micrographs reveal an increase in particle size of the negative electrode on cycling,
which causes the performance of AgO-Zn cells to deteriorate gradually.
© 2004 Elsevier Ltd. All rights reserved.

Keywords: Silver oxide-zinc cells; ac impedance; Charge-transfer resistance; State-of-charge; Cycle-life

1. Introduction the negative electrode. Several studies have been reported to

Alkaline silver oxide-zinc batteries belong to a unique
class of high power batteries. During the initial stages of
charge–discharge cycling they offer the highest power den-
sity, up to about 600 W kg−1 , among commercial recharge-
able batteries. Also, an energy density value as high as
300 Wh kg−1 is another important property of this battery
[1]. By contrast, the best available lithium-ion batteries, of-
fer only about 120–150 Wh kg−1 of energy density and less
than 200 W kg−1 of power density [2]. Although AgO-Zn
battery is expensive due to its silver ingredient, it is a pre-
ferred power system for several critical applications in space
programmes, underwater applications, etc. But, short wet-life
and low cycle-life are its serious limitations [3]. These limita-
tions arise essentially from the inter-electrode separator and
∗ Corresponding author. Tel.: +91 80 2293 2240; fax: +91 80 2360 0683.
E-mail address: muni@ipc.iisc.ernet.in (N. Munichandraiah).
improve the characteristics of these two cell components [4].
Although there is a potential demand for AgO-Zn batteries
for specific applications, studies on characterization of these
batteries are rarely reported [5,6].
In the literature electrochemical impedance spectroscopy
has been employed as a versatile experimental technique for
non-destructive characterization of electrochemical cells [7].
Since this technique allows separation of resistive and capaci-
tive components (sometimes, inductive components as well),
several electrochemical parameters can be evaluated [8]. The
ac impedance measurements of several battery systems have
been reported and reviewed [9–11]. A large number of AgO-
Zn cells have been subjected to a prolonged period of storage
under various experimental conditions and impedance mea-
surements have been carried out at 40 Hz square-wave current
[5]. The measured impedance has been treated as a lumped
parameter, and it has not been separated into resistive and
capacitive components in this study. Prediction of state-of-

0013-4686/$ – see front matter © 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2004.11.055
 P. Suresh et al. / Electrochimica Acta 50 (2005) 3262–3272 3263

charge of Ag2 O-Zn button cells from several manufacturers 40 ◦ C, a refrigerator cum-heater (Julabo Model F 25) with
has been investigated using ac impedance over a frequency ethylene glycol and water mixture as the thermal medium was
range from 10 kHz to 1 mHz [12]. These cells differ from employed. The cell was charged to the required SOC value at
AgO-Zn cells in the oxidation state of the positive electrode ambient temperature (24 ± 2 ◦ C) at C/5 rate, thermostatted
material. The impedance characteristics of AgO-Zn cells are for 4–5 h at the required temperature and the impedance
anticipated to be interesting as the composition of the positive spectrum was recorded. The impedance data were recorded
electrode transforms from the semi-conducting AgO through in the frequency range between 100 kHz and 10 mHz using
the insulating Ag2 O to the metallic Ag during the course of an electrochemical impedance analyzer (EG&G PARC
discharge. To our knowledge, there is little information on Model 6310). The cell was excited potentiostatically by
ac impedance behaviour of commercial AgO-Zn cells over an ac signal of 5 mV at open-circuit condition. Some cells
a wide frequency range as well as on the effects of state-of- were subjected to continuous charge–discharge cycling, and
charge (SOC), temperature and cycle-life on impedance. impedance was measured at certain intervals of cycles.
In the present study, ac impedance of AgO-Zn cells are
measured in the frequency range from 100 kHz to 10 mHz,
and the data are analyzed using non-linear least squares 3. Results and discussion
(NLLS) fitting programme and appropriate equivalent cir-
cuits. The influence of SOC, temperature and cycle-life on A charged AgO-Zn cell consists of a positive electrode
the impedance parameters are studied. made of AgO and a negative electrode made of Zn in 6 M
KOH as the electrolyte. The electrode reactions are reversible
in nature, even though the cycle-life of the AgO-Zn battery
2. Experimental is generally reported to be rather limited. However, longer
cycle life has also been reported by suitable modifications of
AgO-Zn cells of capacity ≈8 Ah were obtained from the cell design [13]. The reversible reactions with standard
High-Energy Batteries (India) Limited, Mathur, Tamilnadu, electrode potentials are given below [14].
in reserve condition. The cell dimensions were 63.9 mm At the negative electrode,
(height) × 52.7 mm (width) × 20.3 mm (thickness). The total discharge
cell weight was found to be 132.6 g in dry state. The cells Zn + 2OH−
Zn(OH)2 + 2e− , E0 = −1.25 V (1)
were positive limited with six positive and seven negative charge
electrodes in each cell. Each electrode was a rectangular Thus, Zn/Zn(OH)2 constitutes the redox couple at the nega-
plate of 25 cm2 area with a weight of 4.6 and 4.3 g, respec- tive electrode. The reaction mechanism involves the forma-
tively, for positive and negative electrodes. The 16 ml of tion of super-saturated zincate solution through the following
6 M KOH electrolyte was added and the electrodes were electron-transfer reaction [15]:
allowed to soak for 5 days. The cells were conditioned by
discharge
conducting a few charge–discharge formation cycles. During Zn + 4OH−
Zn(OH)4 2− + 2e− (2)
the course of formation, the cells were charged at C/20 charge
rate up to 2.05 V followed by discharge to 1 V at C/10 rate.
Formation cycles were continued until the cells reached their From Zn(OH)4 2− , the formation of porous ZnO on the elec-
maximum attainable capacity. Usually, this required about trode surface takes place by the following chemical reaction.
three charge–discharge cycles. The capacity (C) obtained discharge
was considered as the cell capacity for computation of SOC. Zn(OH)4 2−
ZnO + 2OH− + H2 O (3)
charge
Thereafter, C/5 rate was used during the experimental stud-
ies. The charge–discharge circuit consisted of a regulated dc At the positive electrode,
power source in constant voltage mode, an ammeter, and a
2AgO + H2 O + 2e−
high resistance in series with the cell. Prior to connecting the
cell to the circuit, the voltage of the power source, and the discharge
resistance were adjusted to get the required current to flow.
Ag2 O + 2OH− , E0 = 0.60 V (4)
charge
The current was maintained constant due to presence of high
resistance in the circuit during cell cycling. Cell voltage was
measured using a high input impedance digital voltmeter. For Ag2 O + H2 O + 2e−
the purpose of recording impedance at different SOC values, discharge
the cells were charged galvanostastically from SOC ≈ 0
2Ag + 2OH− , E0 = 0.34 V (5)
charge
to the required SOC (for e.g., SOC = 0.2) at C/5 rate and
allowed to equilibrate for 2 h at open circuit prior to obtaining The redox couples that exist at the positive electrode are
the impedance spectrum. After completing the experiments, AgO/Ag2 O and Ag2 O/Ag. Thus, the reduction of AgO to
the cells were charged to the next higher value of SOC. Ag occurs in two stages. Where as the redox reaction be-
For experiments in the temperature range between −20 and tween AgO and Ag2 O (Reaction (4)) occurs via a solid-state
3264 P. Suresh et al. / Electrochimica Acta 50 (2005) 3262–3272
mechanism resulting in the formation of a solid solution of
AgO–Ag2 O, the reaction between Ag2 O and Ag (Reaction
(5)) involves a dissolution–precipitation mechanism as fol-
lows,
discharge
Ag2 O + 2OH− + H2 O
2Ag(OH)2 −
charge
The dissolved Ag(OH)2 − species undergo reduction to metal-
lic Ag as follows,
discharge
Ag(OH)2 − + e−
Ag + 2OH−
charge
The standard electrode potentials of these stages are distinctly
different and the reactions also proceed with different mecha-
nisms. The voltage of a AgO-Zn cell varies over a wide range,
which depends largely on the positive electrode.
A AgO-Zn cell was subjected to galvanostatic
charge–discharge cycling in the voltage limits between
2.05 and 1.0 V. The variation of cell voltage during a
cycle is shown in Fig. 1. During charge, there are two
plateau regions at 1.65 and 1.95 V corresponding to the two
stages of oxidation of the positive electrode (Reactions (5)
and (4), respectively). The voltage spike observed at the
transition of the first and the second plateau is due to highly
resistive Ag2 O completely surrounding the electrode, as
reported by Frank et al. [5]. The sharp rise corresponds to
the time during charge when the layer of high resistance
Ag2 O is completely formed and the decline corresponds
to the onset of production of lower resistance AgO. The
sharp increase in voltage at the end of charge corresponds
to the onset of electrolysis of water. During discharge,
Fig. 1. Charge–discharge curves for AgO-Zn cell.
there is a gradual decrease of voltage from 1.8 to 1.5 V,
and, subsequently, it remains fairly constant at 1.5 V till
the end of discharge. It decreases rapidly at the end. The
potential of the negative electrode, namely Zn/Zn(OH)2 is
expected to be constant throughout the discharge as it is
(6) governed by the constant concentration of supersaturated
Zn(OH)4 2− ions. The variation of AgO-Zn cell voltage
during discharge is mainly due to a variation of the positive
electrode potential. The gradual decrease of voltage from
1.8 to 1.5 V is due to the first-stage reduction of AgO to
Ag2 O (Reaction (2)), wherein these two compounds form a
(7) solid solution. The decrease in voltage is due to an increase
in concentration of Ag2 O in AgO–Ag2 O solid solution. The
second discharge-plateau at 1.5 V is due to the constant
concentration of Ag(OH)2 − (Reaction (6)). During charging,
processes opposite to the above take place between 1.6
and 2.0 V.
3.1. Impedance of AgO-Zn cells
A typical impedance spectrum of AgO-Zn cell at
SOC ≈ 0.4 is shown in Fig. 2 as Nyquist and Bode plots. It is
seen from the Nyquist plot (Fig. 2(a)) that the high-frequency
data between 100 and 10 kHz are inductive in nature; there
is a large semicircle in the mid-frequency range and a small
semicircle at low frequencies. The phase angle (φ) is positive
at high frequencies corresponding to the inductive behaviour
(Fig. 2(b)). As the frequency decreases, the φ gradually be-
comes negative and reaches −70◦ , reflecting the capacitive
behaviour in the frequency range between 10 kHz and 1 Hz.
Furthermore, a decrease in modulus of impedance with fre-
quency also supports the capacitive behaviour of the cell in
this frequency range (Fig. 2(c)).
The impedance parameters were obtained by fitting the
data to an equivalent circuit using NLLS fitting procedure
due to Boukamp [16]. The equivalent circuit, which was
found to fit the impedance spectrum satisfactorily, is shown
in Fig. 3. The circuit elements are explained as follows.
The high-frequency impedance data are inductive in nature
hence the inductance (L) and the resistance (RL ) are taken
in parallel. The element RΩ represents the ohmic resistance
of the cell, which includes the resistance of the electrolyte,
separator, current collectors, electrical leads, etc. R1 and
Q1 , respectively, are the resistance and constant phase
element (CPE) for the large semicircle in mid-frequency
range; and R2 and Q2 are resistance and CPE for the second
small-semicircle in low-frequency range. The CPE arises
due to distribution of microscopic material properties [7].
The interfaces between the electrode/electrolyte of the
AgO-Zn cell are not smooth and uniform, as the electrodes
are made using fine particles of the active materials. Reaction
resistance and capacitance contribution may differ with the
electrode position, non-uniform thickness of the electrode
materials, etc. Owing to these distributions, the semicircles
of the Nyquist plots are usually depressed. The admittance
 P. Suresh et al. / Electrochimica Acta 50 (2005) 3262–3272
. The inductance behaviour has been attributed
Fig. 2. Impedance spectrum of AgO-Zn cell shown as (a) Nyquist, (b) Bode
φ and (c) Bode /Z/ plots at SOC = 0.4 at 20 ◦ C. Frequency values (in Hz) are
given for some data points of the Nyquist diagram.
representation of CPE is given as [16]:
Q∗ = Q0 (jω)n
where Q0 is an adjustable parameter, ω = 2πf and f is the
frequency. For n = 0, CPE represents a resistance, R (=Q0 −1 );
for n = 1, a capacitance, C (=Q0 ); for n = 0.5, a Warburg
impedance; and for n = −1, an inductance, L (=Q0 −1 ). The
NLLS fit technique due to Boukamp [16] was employed
using the equivalent circuit with the circuit description code:
(RL)R(RQ)(RQ). The elements within a set of parentheses
3265
denote their parallel combination, while the elements out
of parentheses denote a series combination. The starting
values were obtained from the data cruncher sub-program
prior to using the values in the NLLS-fit program. A similar
procedure has been successfully employed for other batteries
[17–19]. The impedance parameters obtained for the data in
Fig. 2 are listed in Table 1.
The inductance behaviour of electrochemical power
sources has been reported and reviewed in the literature
[20,21]. Gutmann, from his studies on Zn-MnO2 dry cells,
has reported that these systems exhibit lagging currents [20].
In order to explain this behaviour, a pseudo inductance, de-
fined as the capability of a space to store electrical energy re-
versibly and kinetically, has been proposed. The kinetic pro-
cess has been ascribed to visco-elastic behaviour exhibited at
high frequencies and/or in higher viscous systems. Keddam et
al. [21] have reported inductive behaviour of Pb-acid batteries
at high frequencies. The value of the cell-inductance has been
nearly invariant with SOC of the battery, which is equal to
about 1 ␮H and schematically shunted to a parallel resistance
of about 0.6
to the geometrical nature of conductors and electrodes, and
not to the faradaic processes in the battery. Therefore, it fol-
lows that the shunting resistance (RL ) across the inductance
was only schematic, and is an artifact of the fitting procedure.
The values of the inductance (L) and the shunting resistance
(RL ) are, therefore, considered irrelevant for characterizing
the cell and these parameters are omitted during subsequent
discussion. The fitting procedure was repeated after exclud-
ing the inductive dispersion of the impedance data, and L
and RL in the equivalent circuit. Values of the rest of the pa-
rameters remained unaltered. It was thus confirmed that the
inductance data did not influence the rest of the impedance
spectra.
The appearance of two capacitive semicircles (Fig. 2) sug-
gests the presence of two relaxation processes each repre-
sented by (RQ) (Fig. 3). For the purpose of identifying the
appropriate electrochemical reactions, which correspond to
these relaxation processes, the electrodes of a AgO-Zn cell
were separated out of the cell container. The impedance of a
positive electrode was measured in 6 M KOH by mounting
the electrode between two large area Ni electrodes. Similarly,
impedance of a negative electrode, and impedance across the
two Ni electrodes without any test electrode were also mea-
sured. Since the impedance spectra were recorded at open
circuit conditions of the electrodes, a reference electrode was
not introduced to assess the relative impedance’s of the pos-
(8)
Fig. 3. Equivalent circuit used for the NLLS fitting of the impedance spectra.
3266 P. Suresh et al. / Electrochimica Acta 50 (2005) 3262–3272

Table 1
Impedance parameters at different SOCs at 20 ◦ C
Parameters SOC
1 0.8 0.6 0.4 0.2 0
L (H) 2.70 × 10−6 2.74 × 10−6 2.59 × 10−6 2.63 × 10−6 2.75 × 10−6 2.66 × 10−6
RL (
) 9.87 9.99 8.88 8.10 7.65 6.99
RΩ (
) 0.12 0.12 0.12 0.11 0.10 0.10
R1 (
) 0.61 1.24 4.86 3.80 – –
Q1 1.72 × 10−4 3.21 × 10−4 4.39 × 10−4 6.81 × 10−4 – –
n1 0.99 0.96 0.96 0.97 – –
R2 (
) 0.36 0.21 0.74 0.86 0.17 0.20
Q2 63.11 4.41 2.59 1.97 12.94 26.16
n2 0.26 0.37 0.63 0.63 0.65 0.88

itive and negative electrodes. The large area Ni electrodes 3.2. Effect of state-of-charge on impedance
on either side of the test electrode were expected to have low
impedance, and the contribution of these electrodes was com- The impedance of a AgO-Zn cell was measured at sev-
mon for both positive and negative electrodes when measured eral SOC values and the data measured at 20 ◦ C are shown
separately by placing them in between the Ni electrodes. As in Fig. 5. At SOC between 1 and 0.4, the Nyquist plots con-
expected the impedance of Ni electrodes was found to be sisted of a pair of capacitive semicircles similar to the spec-
negligibly smaller than the impedance of both positive and trum shown in Fig. 2. However, at SOC between 0.4 and 0,
negative electrodes. Among the positive and negative elec- the data consisted of a single semicircle. For the purpose of
trodes, the impedance of the negative electrode was found NLLS fitting of a spectrum consisting of a single semicir-
to be smaller than that of the positive electrode by about cle, the equivalent circuit shown in Fig. 3 was modified by
100 times as assessed from the diameter of the semicircles. deleting the R1 and Q1 branch. The impedance parameters
This was found to be true at all SOC values of the cell at were obtained excluding these parameters. The values of the
20 ◦ C, and it was anticipated that the same trend would fol- parameters are listed in Table 1. It is seen that the value of
low at all temperatures used in the study. The AgO-Zn cells ohmic resistance (RΩ ) is nearly constant at 0.11 ± 0.01
.
were also designed with an excess of Zn electrode. Using It is interesting to note that the value of R1 at SOC = 1 is
these observations, it is considered that the impedance mea- low at 0.61
. With a decrease of SOC, R1 increases and
sured for AgO-Zn cell is essentially the impedance of posi- reaches a maximum of 4.86
at SOC = 0.6. The value of
tive electrodes. As the composition of the positive electrode
changes from AgO at high SOCs to Ag2 O at mid-SOCs and
further to Ag at low SOCs, it is very likely that these three
components are present in the electrode over the entire SOC
range, although their ratios vary with SOC. A rough estima-
tion of the capacitance corresponding to the high frequency
semicircle and the low frequency semicircle provided val-
ues in the ranges of ␮F cm−2 and mF cm−2 , respectively.
Since the values of capacitance in ␮F cm−2 are reasonable
for a double layer, R1 is attributed to charge transfer re-
sistance of AgO/Ag2 O couple. High values of capacitance
observed for the low frequency semicircle are likely to be
due to coverage rate of Ag2 O on the electrode surface. Ac-
cordingly R2 is attributed to the surface coverage resistance.
The XRD patterns of the positive electrode, during the single
electrode study, are shown in Fig. 4(a)–(c), at SOC values
of 0, 0.6 and 1, respectively. It can be seen that at SOC = 0,
the pattern shows the diffraction lines corresponding to pure
Ag, where-as at SOC = 0.6 and 1, the diffraction lines cor-
responding to all the three components namely, Ag, Ag2 O
and AgO exists. Hence, the three components Ag, Ag2 O and
AgO exist at all SOCs but in different proportions. Thus the
XRD patterns further support the above assignment, for the
semicircles. Fig. 4. XRD patterns of positive electrode at SOC values of (a) 0, (b) 0.6
and (c) 1.
 P. Suresh et al. / Electrochimica Acta 50 (2005) 3262–3272 3267

Fig. 5. Nyquist impedance spectra at 20 ◦ C of AgO-Zn cell at SOC values of (a) 0, (b) 0.2, (c) 0.4, (d) 0.6, (e) 0.8 and (f) 1.0. Frequency values (in Hz) are
given for some data points of the Nyquist diagram.

Q1 increases from 1.72 × 10−4 at SOC ≈ 1 to 6.81 × 10−4
at SOC ≈ 0.4 with n values close to unity. Thus, Q1 repre-
sents double-layer capacitance (Cdl ) of the positive electrode.
There is an increase in R2 from 0.36
at SOC ≈ 1 to 0.86
than Ag2 O and AgO.
at SOC ≈ 0.4 due to increase in coverage of Ag2 O (Reaction
(4)). As Ag2 O is reduced to Ag (Reaction (5)) during second
stage of discharge, R2 decrease from 0.86
at SOC ≈ 0.4 to rechargeable cells. For this purpose, a linear variation of the
0.20
at SOC ≈ 0. Only the semicircle corresponding to the
coverage of Ag2 O is observed at low SOC range because of
fast kinetics of Ag2 O/Ag couple (Reaction (5)). The values
of Q2 remain higher than Q1 during the entire range of SOC
and there is an increase of n2 from 0.26 at SOC ≈ 1 to 0.88
at SOC ≈ 0. It is interesting to note that the total cell resis-
tance, Rt (=RΩ + R1 + R2 ) is smaller in the discharged state
than in the charged state. This is because of the faster kinetics
of Ag2 O/Ag couple and also the greater conductivity of Ag
Prediction of SOC by monitoring appropriate impedance
parameters is important for non-destructive evaluation of
impedance parameters with appreciable magnitudes over the
SOC range from 1 to 0 was considered to be a general re-
quirement [22], although this may not be essential with the
possible use of microprocessors in recent years. A linear vari-
ation in the values of the parameters is expected if the corre-
sponding electrode reaction occurs with a single mechanism
3268 P. Suresh et al. / Electrochimica Acta 50 (2005) 3262–3272

throughout the cell discharge. In the case of AgO-Zn cell,
however, the positive electrode reaction, which decides the
cell impedance, occurs in two consecutive stages (Reactions
(4) and (5)) with different mechanisms. The kinetics of these
reactions differ appreciably. As a result, there is no linear
variation of any impedance parameters. Nevertheless, it is
noted that the cell impedance (or, Rt ) is higher at SOC values
between 1 and 0.4, corresponding to the first stage of posi-
tive electrode discharge (Reaction (4)), and it is substantially
lower at SOC values between 0.4 and 0, corresponding to the
second stage of positive electrode reaction (Table 1). Hence,
by measuring the cell impedance, it is possible to predict
qualitatively the SOC as higher or lower than 0.4.
3.3. Effect of temperature on impedance
The impedance of AgO-Zn cell was recorded in the tem-
perature range between −20 and 40 ◦ C at several SOC val-
ues. Nyquist plots at SOC = 0.6 are shown in Fig. 6. The
high-frequency semicircle is present at all temperatures, and
increases in size with a decrease in temperature. The low-
frequency semicircle is observed between 10 and 40 ◦ C. At

Fig. 6. Nyquist impedance spectra of AgO-Zn cell at (a) −20, (b) −10, (c) 0, (d) 10, (e) 20 and (f) 40 ◦ C at SOC = 0.6. Frequency values (in Hz) are given for
some data points of the Nyquist diagram.
 P. Suresh et al. / Electrochimica Acta 50 (2005) 3262–3272 3269

Table 2
Impedance parameters at different temperatures at SOC = 0.6
Parameters Temperature (K)

253 263 273 283 293 303 313

L (H) 2.71 × 10−6 2.82 × 10−6 2.79 × 10−6 2.74 × 10−6 2.59 × 10−6 2.69 × 10−6 2.69 × 10−6
RL (
) 8.91 9.39 9.94 8.79 8.88 9.10 10.19
RΩ (
) 0.13 0.10 0.10 0.11 0.12 0.14 0.12
R1 (
) 43.89 27.05 15.55 8.83 4.86 2.52 1.14
Q1 3.15 × 10−4 3.39 × 10−4 3.69 × 10−4 4.09 × 10−4 4.39 × 10−4 4.69 × 10−4 4.98 × 10−4
n1 0.99 0.98 0.98 0.97 0.96 0.96 0.97
R2 (
) – – – 0.99 0.74 0.49 0.31
Q2 – – – 3.84 2.59 2.58 2.12
n2 – – – 0.65 0.63 0.58 0.51

low temperatures, however, the data points corresponding to is perhaps due to the impedance of Zn electrode becoming
this semicircle fall on the real axis inside the larger semi- dominant.
circle (Fig. 6(a)–(c)). This type of impedance behaviour is For a reaction of the type: O + ne− = R, where O and R
attributed to adsorption effects at the electrode/electrolyte are respectively oxidized and reduced species present in the
interface [23]. In the case of Ag2 O/Ag couple (Reaction (5)), electrolyte, the Rct is related to exchange current (I0 ) from
the reaction occurs through the formation of Ag(OH)2 − (Re- Butler–Volmer relationship as;
action (6)) ions. Adsorption of these ions is likely to be strong
RT
at low temperatures. Another possible explanation is that it Rct = (9)
requires a long time for impedance measurements at low fre- FI0
quencies and the system would not be stationary due to drift- The relationship between I0 and concentrations of O and R
ing of current because of poor reaction kinetics at low tem- (CO and CR , respectively), is given as,
peratures. It is likely that the Kramers Kronig relationships
1−α α
are not satisfied. At present, the exact reason for the low I0 = FAks CO CR (10)
frequency behavior present in Fig. 6(a)–(c) is not clearly un-
where α is the transfer coefficient, A the electrode area and ks
derstood. Nevertheless, The data of impedance spectra were
is the standard rate constant. In the case of Reaction (4) for
subjected to NLLS fitting and the parameters are listed in
AgO/Ag2 O couple, where the kinetics essentially depends on
Table 2. For the data recorded between 0 and −20 ◦ C, only
the relative quantities of AgO and Ag2 O in AgO–Ag2 O solid
the high-frequency semicircle was used for fitting. The value
solution, I0 is approximated as,
of R1 increases from 1.14
at 40 ◦ C to 43.89
at −20 ◦ C.


Interestingly, there is only a marginal increase in R2 while CAgO β
decreasing the temperature from 40 ◦ C to 10 ◦ C. The vari- I0 ∝ ks (11)
CAg2 O
ation of Rt obtained at different temperatures with SOC is
shown in Fig. 7. At all temperatures, the cell resistance is where CAgO and CAg2 O are concentrations of AgO and Ag2 O,
maximum at SOC ≈ 0.4, and is minimum at SOC ≈ 0 except respectively, in the solid solution and β is a constant. Since
at −20 ◦ C. A high value of Rt obtained at −20 ◦ C for SOC ≈ 0 the SOC of AgO-Zn cell between 1 and 0.4 depends on the
concentration ratio, we can write Eq. (11) as,

I0 = kks S β (12)

where k is a constant and S is the abbreviation for SOC.

From Eqs. (9) and (12), we get,
RT
Rct = (13)
Fkks S β
The standard rate constant (ks ) depends on T, and the rela-
tionship follows Arrehenius equation,


−Ea
ks = A0 exp (14)
RT
Combining Eqs. (13) and (14), we get,
RT
Fig. 7. Variation of total resistance (Rt ) with SOC at (a) −20, (b) −10, (c) Rct = (15)
0, (d) 10, (e) 20, (f) 30 and (g) 40 ◦ C. FkS β A0 exp(−Ea /RT )
3270 P. Suresh et al. / Electrochimica Acta 50 (2005) 3262–3272

Fig. 8. Plots of ln(Rct /T) vs. 1000/T at SOC values of (i) 0.4, (ii) 0.6, (iii)
0.8 and (iv) 1.

or,
Rct k
= β exp (Ea /RT ) (16)
T S
Fig. 9. Cycle-life data of a AgO-Zn cell.
where k (=R/FkA0 ) is a constant, which is independent of T
and SOC. We can rewrite Eq. (16) as in Eq. (17), Q2 at SOC ≈ 0.5 with cycle number are shown in Fig. 10. Al-


 though there is an increase in R1 during the initial ten cycles,
Rct  Ea
ln = ln k + − β ln S (17) there is a decrease there after. There is no particular trend of
T RT
R1 during the cycling. In the case of R2 , the values are nearly
On differentiation, we get for a certain SOC constant with some scatter (Fig. 10(b)). On the other hand,
Q1 and Q2 show an increasing trend with some scatter of data
d ln(Rct /T ) Ea on subjecting the AgO-Zn cell to charge–discharge cycling
= (18)
d(1/T ) R (Fig. 10(c) and (d)).
Thus, it is possible to evaluate apparent activation energy (Ea ) After completing 25 charge–discharge cycles of a AgO-
of the AgO/Ag2 O couple. Zn cell, the cell in discharged state was opened and small
The variations of ln(Rct /T) with 103 /T of the AgO-Zn cell portions of positive and negative electrodes were sectioned
is shown in Fig. 8 for different SOC values between 1 and 0.4. out for XRD and SEM studies. The diffractograms and mi-
It is seen that the plots are linear as expected from Eq. (17). crographs of the positive electrode before cycling and after
The values of Ea as obtained for SOC ≈ 1, 0.8, 0.6 and 0.4 25 cycles are shown in Fig. 11. The corresponding data for
are 0.38, 0.43, 0.44 and 0.51 eV, respectively. Thus, there is the negative electrode are shown in Fig. 12. As the electrodes
an increase in energy barrier for the AgO/Ag2 O couple with are in discharged state, Fig. 11 corresponds to essentially
increase in the concentration of Ag2 O. Ag present in the positive and Fig. 12 corresponds to essen-

3.4. Effect of charge–discharge cycling

A AgO-Zn cell was subjected to galvanostatic

charge–discharge cycling at C/5 rate, and impedance
of the cell was recorded at some intervals of cycle-life. The
discharge capacity is shown as a function of cycle number in
Fig. 9. The capacity increased from 8 Ah of the first cycle to
10 Ah of the third cycle. Thereafter, it decreased rapidly on
cycling, and reached a value of about 3.5 Ah at the end of 27
cycles. This rapid fall of discharge capacity makes AgO-Zn
battery less attractive as a rechargeable battery.
For the purpose of identifying appropriate impedance pa-
rameters that reflect the decrease in capacity on cycling, the
impedance of cell was measured at SOC = 0, 0.5 and 1.0, at
certain intervals of cycling. The variations in R1 , R2 , Q1 and Fig. 10. Dependence of cycle number on (a) R1 , (b) R2 , (c) Q1 and (d) Q2 .
 P. Suresh et al. / Electrochimica Acta 50 (2005) 3262–3272 3271

Fig. 11. XRD patterns and SEM images of (a, c) fresh and (b, d) cycled positive electrode.

Fig. 12. XRD patterns and SEM images of (a, c) fresh and (b, d) cycled negative electrode.
3272 P. Suresh et al. / Electrochimica Acta 50 (2005) 3262–3272
tially ZnO in the negative electrode. The XRD patterns of Ag
(Fig. 11(a) and (b)) remain almost unaltered even after cy-
cling. Both patterns are indexed in cubic space-group Fm3m
[24], and the respective lattice parameter values obtained are:
a = 4.11 Å and 4.12 Å. Accordingly, there is only 0.75% in-
crease in unit cell volume. The XRD pattern in Fig. 11(b)
shows weak lines between 2θ = 30◦ and 34◦ , which corre-
spond, to AgO. Thus, after cycling, even in the discharged
state there is some AgO present indicating incomplete uti-
lization of the positive electrode material, even though the
cell is positive limited.
The scanning electron micrographs (Fig. 11(c) and (d))
reveal a decrease in particle size and formation of large voids
on cycling. The increasing tendency of Q1 and Q2 (Fig. 10(c)
and (d)) is probably due to an increase in surface area ow-
ing to decreasing particle size and higher porous morphology
obtained upon cycling. In the case of negative electrode also,
the crystallographic form of ZnO remains unaltered during
charge–discharge cycling (Fig. 12(a) and (b)). Both the XRD
patterns are indexed in the hexagonal space-group; P63mc
[25] and the lattice parameter values obtained for the fresh
sample are: a = 3.24 Å and c = 5.20 Å. The corresponding val-
ues obtained for the cycled sample are 3.27 Å and 5.28 Å,
accordingly there is 3.2% increase in the unit cell volume.
However, the scanning electron micrographs (Fig. 12(c) and
(d)) reveal a conversion of spherical particles of Zn before
cycling to rod-like structure after cycling. Thus, both XRD
and SEM suggest the shape change [1] in ZnO electrode,
which is one of the reasons for deterioration of AgO-Zn cell
performance on repeated cycling.
4. Conclusions
Commercial AgO-Zn cells studied for ac impedance char-
acteristics with SOC, temperature and charge–discharge cy-
cling as variables. The impedance parameters are evaluated
using a NLLS fitting procedure with the help of appropriate
equal circuits. The cell impedance is attributed essentially
due to the positive electrode. Since the two reactions of the
positive electrode are different in their mechanisms and kinet-
ics, impedance parameters that vary linearly over the entire
SOC range are not identified. The temperature dependence
on the nature of impedance spectra as well as the magnitude
of impedance parameters is markedly found. The activation
energy for AgO/Ag2 O couple is calculated and it is found
to increase with a decrease of SOC. The effect of cell de-
terioration due to charge–discharge cycling on impedance
parameters is not clearly noticed. Nevertheless, there is an
increasing trend in the constant phase elements, Q1 and Q2
with cycle number.
Acknowledgements
Authors thank High Energy Batteries Ltd., India for sup-
plying the AgO-Zn cells.
References
[1] A.P. Karpinski, B. Makovetski, S.J. Russell, J.R. Serenyi, D.C.
Williams, J. Power Sources 80 (1999) 53.
[2] D. Linden, T.B. Reddy, Handbook of Batteries, second ed., McGraw-
Hill, New York, USA, 1994, p. 14.1.
[3] J. Skelton, R. Serenyi, J. Power Sources 65 (1997) 39.
[4] H.L. Lewis, T. Danko, A. Himy, W. Johnson, J. Power Sources 80
(1999) 61.
[5] H.A. Frank, W.L. Long, A.A. Uchiyama, J. Electrochem. Soc. 123
(1976) 1.
[6] B. Hariprakash, S.K. Martha, A.K. Shukla, J. Power Sources 117
(2003) 242.
[7] J.R. Macdonald, Impedance Spectroscopy, John Wiley & Sons, New
York, USA, 1987, p. 84.
[8] S.M. Park, J.S. Yoo, Anal. Chem. 75 (2003) 455A.
[9] S. Rodrigues, N. Munichandraiah, A.K. Shukla, J. Power Sources
87 (2000) 12.
[10] F. Huet, J. Power Sources 70 (1998) 59.
[11] N.A. Hampson, S.A.G.R. Karunathilaka, R. Leek, J. Appl. Elec-
trochem. 10 (1980) 3.
[12] J.P. Randin, J. Appl. Electrochem. 15 (1985) 365.
[13] U. Falk, A.J. Salkind, Alkaline Storage Batteries, Wiley, New York,
USA, 1971.
[14] J.P. Randin, J. Appl. Electrochem. 15 (1985) 293.
[15] T.D. Dirkse, D. Dewit, R. Shoemaker, J. Electrochem. Soc. 114
(1967) 1196.
[16] B.A. Boukamp, Equivalent Circuit, Users Manual, University of
Twente, Enschede, The Netherlands, 1989, p. 12.
[17] S. Rodrigues, N. Munichandraiah, A.K. Shukla, J. Solid State Elec-
trochem. 3 (1999) 397.
[18] P. Suresh, A.K. Shukla, N. Munichandraiah, J. Appl. Electrochem.
32 (2002) 267.
[19] S. Sathyanarayana, S. Venugopalan, M.L. Gopikanth, J. Appl. Elec-
trochem. 9 (1979) 125.
[20] F. Gutmann, J. Electrochem. Soc. 112 (1965) 94.
[21] M. Keddam, Z. Stoynov, H. Takenouti, J. Appl. Electrochem. 7
(1977) 539.
[22] S.A.G.R. Karunathilaka, N.A. Hampson, R. Leek, T.J. Sinclair, J.
Appl. Electrochem. 10 (1980) 603.
[23] I. Epelboin, C. Gabrielli, M. Keddam, H. Takenouti, Electrochim.
Acta 16 (1966) 913.
[24] JCPDS file No.: 04-0783.
[25] JCPDS file No.: 36-1451.