Cycle stability of birnessite manganese dioxide for electrochemical capacitors

Article abstract of DOI:10.1016/j.electacta.2010.01.005

A combination of step potential electrochemical spectroscopy (SPECS) and electrochemical impedance spectroscopy (EIS) has been used to examine the electrochemical cycling behaviour of a well-characterized birnessite-phase manganese dioxide sample for use in electrochemical capacitors. The resistance and interfacial properties of the macroscopic electrode were found to be irreversible with cycling. However, the corresponding properties for the individual particles were more reversible, although they did show substantial hysteresis in their behaviour during cycling. This behaviour was discussed in terms of the structural, conductivity and morphological characteristics of the birnessite at different depths of discharge.

Full text of DOI:10.1016/j.electacta.2010.01.005

Electrochimica Acta 55 (2010) 7470–7478
Contents lists available at ScienceDirect
Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta
Cycle stability of birnessite manganese dioxide for electrochemical capacitors
Ariana Cormiea, Andrew Crossa, Anthony. F. Hollenkampb, Scott W. Donnea,∗
a Discipline of Chemistry, University of Newcastle, Callaghan, NSW 2308, Australia
b CSIRO Division of Energy Technology, Box 312, Clayton South, VIC 3169, Australia
a r t i c l e i n f o
a b s t r a c t
Article history:
A combination of step potential electrochemical spectroscopy (SPECS) and electrochemical impedance
Received 29 September 2009
Received in revised form 4 January 2010
Accepted 4 January 2010
spectroscopy (EIS) has been used to examine the electrochemical cycling behaviour of
a well-
characterized birnessite-phase manganese dioxide sample for use in electrochemical capacitors. The
resistance and interfacial properties of the macroscopic electrode were found to be irreversible with
cycling. However, the corresponding properties for the individual particles were more reversible,
although they did show substantial hysteresis in their behaviour during cycling. This behaviour was
discussed in terms of the structural, conductivity and morphological characteristics of the birnessite at
different depths of discharge.
Available online 25 January 2010
Keywords:
Electrochemical capacitors
Supercapacitors
© 2010 Elsevier Ltd. All rights reserved.
Manganese oxides
Electrochemical impedance spectroscopy
1. Introduction
supercapacitors are symmetrical devices (identical electrodes)
employing activated carbon electrodes with either an aqueous
(e.g., H2SO4) or non-aqueous (e.g., tetra alkyl ammonium tetrafluo-
robroate in acetonitrile) based electrolytes [3]. The performance
1.1. General
Undoubtedly the efficient supply of energy is one of the greatest
challenges modern society faces in the immediate future. Consid-
erable effort has been expended recently in the development and
expansion of more renewable forms of energy generation; how-
ever, with current technologies none of these has the realistic
ability to economically substitute for fossil fuel based power [1].
Therefore, energy production in the future is anticipated to be more
distributed and intermittent in nature to capitalize on the smaller
scale, more renewable forms of energy. With energy production
heading in such a direction, the importance of energy storage is
also growing for load-levelling applications, as well as a means to
improve efficiency. Various energy storage options exist, the rela-
tive merits of which can be summarized by a Ragone diagram [2].
Supercapacitors are an emerging technology, meshing well with
modern electronic devices in the sense they have the ability to pro-
vide high power for pulse applications over an extended cycle life.
Their main limitation is poor energy density.
of these devices is quite good, with high power density and
long cycle life; however, as mentioned, their energy density is
limited. A strategy for improving supercapacitor energy den-
sity is to incorporate a pseudo-capacitive electrode [4]; i.e., an
electrode that can store charge in the double layer, as well as
undergo fast reversible surface redox reactions. These types of
electrodes have an enhanced capacity compared to carbon-based
electrodes as a result of the faradaic contribution, as well as a
higher density meaning that their volumetric energy density is
enhanced.
The prototypical supercapacitor electrode material is amor-
phous hydrated ruthenium dioxide (RuO2·xH2O) which has been
reported to have a capacitance exceeding 900 F/g in an aqueous
H2SO4 electrolyte [5]. What primarily limits its commercial imple-
mentation is cost, which has prompted the search for suitable
alternatives with similar behaviour. Alternative systems based on
other metal oxides and conductive polymers have been reported in
the literature to behave similarly; however, of these, metal oxides
are preferred because of their higher energy density [4,5]. Within
the category of metal oxides, systems based on manganese dioxide
are presently receiving considerable attention from the scientific
community [6]. With a myriad of available synthesis methods
[7–14], and as a result wide ranging morphologies, coupled with
good faradaic electrochemical performance, low cost and environ-
mental friendliness, this class of materials represents an interesting
and promising area of research.
1.2. Supercapacitors
The underlying basis of the research reported here is on boost-
ing the energy density of supercapacitors. Commercially available
∗
Corresponding author. Tel.: +61 2 49215477; fax: +61 2 49215472.
E-mail address: scott.donne@newcastle.edu.au (S.W. Donne).
0013-4686/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2010.01.005

A. Cormie et al. / Electrochimica Acta 55 (2010) 7470–7478
7471
1.3. This work
using N2 as the adsorbate at 77 K. Degassing under these condi-
tions may lead to the loss of some structural water in the sample;
however, what is most important is the removal of surface bound
water that would otherwise interfere with the adsorption measure-
ment, and can only be achieved under these conditions. The specific
surface area was extracted from the gas adsorption data using the
BET isotherm [17], while the pore size distribution was determined
using a Density Functional Theory based approach (Micromeritics
DFTPlus V2.00) [18].
What we are reporting on in this communication are our efforts
to understand the cycle stability of one of the preferred phases
used in supercapacitors; namely, the birnessite phase of manganese
dioxide. The popularity of this material has arisen as a result of
its ability to be synthesized with a relatively high surface area in
an attempt to maximize the contributions from both faradaic and
non-faradaic charge storage mechanisms. However, as with many
faradaic charge storage materials, their cycle life can be somewhat
limited by mechanical breakdown of the electrode as a result of
repeated charge insertion/de-insertion cycles, as well as solid state
phase changes. This of course is manifest as a decrease in charge
storage capacitance with cycle number. However, what is unclear in
this process are the changes in electronic and electrochemical phe-
nomena that actually give rise to the performance decrease. Here
we have used electrochemical impedance spectroscopy to system-
atically explore these phenomena as a function of state of charge
under quasi-equilibrium conditions.
2.3. Electrode preparation
The first step in electrode assembly was to disperse 0.500 g of
birnessite sample, 0.067 g of conductive carbon black and 0.100 g of
PVdF binder (75:10:15) into 10 mL of N-methylpyrrolidone (NMP)
using an ultrasonic bath. After 10 min dispersion, 0.050 mL of this
homogeneous suspension was then drop-coated evenly onto a
clean flat 2 cm2 titanium foil (1 cm × 2 cm; 200 ␮m thick) to com-
pletely cover the electrode surface. Previously, the titanium foil
had been cleaned by firstly immersing into acidified (2% H2SO4)
H2O2 (10% solution) to reduce any manganese oxides present,
mechanically polished with 300 grit emery paper, and finally thor-
oughly washed with acetone and wiped dry with a lint-free tissue.
The coated electrode was then transferred to a 50 ◦C oven where
the NMP was allowed to evaporate leaving behind a dry, adher-
ent and coherent thin film of electrode material on the titanium
substrate (positive electrode). A negative electrode was prepared
in a similar fashion except the active birnessite was replaced by
supercapacitor-grade activated carbon (CAP-XX).
2. Experimental
2.1. Material synthesis
The birnessite phase of manganese dioxide was prepared in
this work via the reaction between KMnO4 (Sigma–Aldrich) and
MnSO4·H2O (Sigma–Aldrich); i.e.,
2MnO4− + 3Mn2+ + 2H2O → 5MnO2 + 4H+
(1)
2.00 g of KMnO4 (0.013 mol) was dissolved into 100 mL of Milli-
Q ultra-pure water (resistivity >18 Mꢀ cm) in a 500 mL beaker. To
maintain good temperature control of the precipitation the 500 mL
beaker was immersed into a temperature controlled water bath set
at 22 1 ◦C. A separate solution of Mn2+ was prepared by dissolving
3.209 g of MnSO4·H2O (0.019 mol) into another 100 mL of Milli-Q
ultra-pure water. This solution of Mn2+ was added drop-wise with
a burette (over the course of 1 h) to the MnO4− solution which was
being stirred evenly by a magnetic stirrer. Stirring was continued
for a further 1 h after complete addition of the Mn2+ solution. Any
solids formed were collected by filtration, and then subjected to
a washing sequence designed to retain the pore structure in the
solid by minimizing the surface tension at the solid–liquid inter-
face when drying takes place [15]. This involved firstly washing the
solid thoroughly on the filter paper with Milli-Q ultra-pure water
(surface tension of 72.75 mN/m at 20 ◦C [16]) until the filtrate was
the same pH as the wash water. The solid was then re-suspended
in acetone (surface tension of 23.70 mN/m at 20 ◦C [16]) for 1 h to
extract water from the pores. The suspension was then filtered,
and the re-suspension process repeated this time with liquid hex-
ane (surface tension of 18.43 mN/m at 20 ◦C [16]). After this third
washing the suspension was filtered with the collected solid dried
in air at 50 ◦C.
2.4. Electrochemical cell assembly and testing protocols
A schematic of the electrochemical cell used in this work is
shown in Fig. 1. It is a sandwich-type cell in which the positive
and negative electrodes are separated by two layers of a porous
paper separator, and held in place by Perspex base- and cover-
plates secured using four nylon nuts and bolts tightened to a torque
of 75 cN m to ensure a constant pressure. The entire cell when
assembled is immersed into the cycling electrolyte (0.5 M Na2SO4)
together with a saturated calomel reference electrode (SCE). The
cell was immersed to a level that ensured electrolyte coverage of
the active material on each electrode, but did not cover the contact
points at the top of the electrode. Note that the available area of the
birnessite electrode was adjusted to 1 cm2 by mechanically scrap-
ing off the excess electrode material using a razor blade to ensure
that its performance limits the measured response.
2.2. Physical characterization
Material structure was determined by X-ray diffraction using a
Philips 1710 diffractometer equipped with a Cu K␣ radiation source
(ꢁ = 1.5418 Å) operating at 40 kV and 30 mA. The scan range was
from 10◦ to 90◦ 2ꢂ with a step size of 0.1◦ and a count time of 2.5 s.
Morphology was evaluated using a Philips XL30 scanning
electron microscope (SEM) at a range of magnifications, a JEOL
JEM-1200EXII transmission electron microscope (TEM) again at a
range of magnifications, as well as by gas adsorption to determine
material porosity. In this latter case a Micromeritics ASAP 2020
Surface Area and Porosity Analyzer was used, with the samples
being degassed under vacuum at 110 ◦C for 6 h prior to analysis
Fig. 1. Schematic diagram showing electrochemical test cell construction.

7472
A. Cormie et al. / Electrochimica Acta 55 (2010) 7470–7478
The cycling performance of the birnessite containing cell was
evaluated using a Princeton Applied Research VMP multi-channel
potentiostat/galvanostat. Cycling was conducted at 5 mV/s in the
voltage range 0.0–0.8 V versus SCE, for at least 50 cycles.
A more detailed analysis of the cycle stability was conducted
using a combination of step potential electrochemical spectroscopy
(SPECS) and electrochemical impedance spectroscopy (EIS), carried
out using the combination of a Solartron 1254 Frequency Response
Analyzer and a Solartron 1287 Electrochemical Interface controlled
by ZPlot software. From the open circuit voltage of the birnessite
electrode (0.350 V) the voltage was stepped in the anodic direc-
tion 25 mV, after which it was allowed to equilibrate for a period of
15 min. After this, an impedance spectrum on the birnessite elec-
trode was measured using the frequency range 20 kHz to 0.1 Hz
and a 10 mV excitation signal. This sequence was repeated to the
upper voltage limit (0.8 V versus SCE), down to the lower voltage
limit (0.0 V versus SCE), and then once more back up to the upper
voltage limit. The resultant impedance spectra were then inter-
preted using an appropriate equivalent circuit, which was fitted
to the data using complex non-linear least squares regression, as
used in the ZView software package. While this process does not
encompass extended cycling, it does allow the electrode to achieve
quasi-equilibrium with the electrolyte at its state of charge, mean-
ing that any chemical or electrochemical processes occurring would
be very apparent.
Fig. 3. Scanning electron microscope image of the birnessite manganese dioxide
sample prepared in this study.
spacings within the layers. The almost complete absence of basal
plane reflections is typical for birnessite, indicating the turbostratic
disorder of the layers (also shown in Fig. 2) [22]. The pattern for
birnessite has been indexed using a hexagonal unit cell with dimen-
sions a0 = 0.238 nm and c0 = 0.794 nm.
The morphology of the birnessite is shown in the SEM image of
Fig. 3. It shows clearly that the birnessite sample consists of rather
large agglomerates (10–20 ␮m in size) of much smaller crystal-
lites. From the XRD pattern shown in Fig. 2, and using the Scherrer
equation [23], the primary crystallite size has been estimated to by
3–4 nm. This estimation is supported by TEM images of the material
(Fig. 4) which shows a very finely divided morphology interspersed
amongst some needles of manganese dioxide.
3. Results and discussion
3.1. Physical characterization
An XRD pattern of the birnessite manganese dioxide sample pre-
pared in this work is shown in Fig. 2, together with a schematic of
the birnessite structure [19–21]. The pattern reflects the very dis-
ordered nature of the birnessite with very few peaks available for
characterizing material structure. Nevertheless, there is evidence
for the birnessite structure including the (0 0 2) peak at ∼24◦ 2ꢂ,
representing the interlayer spacing, and the (1 0 0) and (1 1 0) peaks
at ∼37◦ and ∼67◦ 2ꢂ, respectively, representing various atomic
Fig. 2. X-ray diffraction pattern of the as-prepared birnessite sample. Inset: The
birnessite structure showing the layers of [MnO6] octahedra separated by foreign
cations and water molecules, as well as the tubostratic structure.
Fig. 4. Transmission electron microscope image of the birnessite sample prepared
and used in this study.

A. Cormie et al. / Electrochimica Acta 55 (2010) 7470–7478
7473
Fig. 5. Pore size distribution of the birnessite phase.
Other relevant morphological characteristics include the spe-
cific surface area (174 m2/g calculated using the BET isotherm) and
pore size distribution of the birnessite phase (Fig. 5). This figure
shows the material has a very broad pore size distribution, ranging
from micro-porous (which constitutes ∼40% of the available surface
area; cf. Fig. 5), through meso-porous to macro-porous. This surface
is not all available for energy storage, either through direct charge
storage on the surface or via pseudo-capacitance, since transport of
electrolyte ions and their solvent sheaths through these pores may
be restricted due to the pores being too small. A necessary step in
the charge storage process involves the movement of charge along
the pores, and if the rate of this process is inhibited by ion size, then
of course not all the surface area will be available.
3.2. Birnessite voltametric performance
An asymmetric cell constructed according to Fig. 1 was assem-
bled and cycled at 5 mV/s in an electrolyte of 0.5 M Na2SO4
using a voltage window for the positive electrode correspond-
ing to 0.0–0.8 V versus SCE. A typical voltammogram is shown
in Fig. 6(a) (cycle 30), together with the specific capacitance as
a function of cycle life of the birnessite phase (Fig. 6(b)). In the
latter figure the capacitance is seen to decrease initially but then
increase as the cycle number increases. While these results are
comparable to the literature for similar materials [24], the cause
of the increasing capacitance is not known with certainty, how-
ever, it most likely involves mechanical breakdown of the material
opening up the pores to expose a higher surface area to the elec-
trolyte.
Fig. 6. (a) Sample voltammogram of the birnessite electrode (cycle 30); (b) specific
capacitance vs. cycle number.
impedances at lower frequencies, and at first glance it would seem
that the system is reversible given the similarity between the scans
taken at 0.8 V after the first and second charge half cycles. At higher
frequencies, where the impedance is much smaller, the differences
between the spectra are more subtle, as seen in the inset in Fig. 7.
Overall, from the data presented in this figure it would appear that
the electrode impedance is a minimum in the fully charged state,
and at a maximum in the fully discharged state. This observation
is consistent with the previously reported changes in conductivity
associated with manganese oxides used in battery systems [25].
Additionally, assuming that mechanical changes to the electrode
are not too significant, it is a clear indication that the birnessite
is undergoing redox cycling in this supercapacitor configuration
given the semi-circular data.
3.3. EIS spectra
An identical electrode to that used to evaluate voltametric per-
formance was subjected to the combination of SPECS and EIS
electrochemical methods, as described in Section 2. Spectra were
collected from the OCV up to 0.8 V for the first charge, then down to
0.0 V for the first discharge, and then back up to 0.8 V for the second
charge cycle. Sample EIS spectra are shown in Fig. 7 covering the
end points of the complete scan. The spectra are dominated by large

7474
A. Cormie et al. / Electrochimica Acta 55 (2010) 7470–7478
from an ideal capacitor (0 ≤ P ≤ 1) [27]. For instance, if P = 1 then
Eq. (2) breaks down to the expression for an ideal capacitor. Val-
ues of P in the range 0.75 ≤ P ≤ 1 typically describe the non-ideal
behaviour of a capacitor, taking into account surface roughness,
and of most relevance in this case, the porosity of the electrode. In
series with R2 is another parallel combination of a resistance and
constant phase element (R3 and CPE2). In this configuration these
elements are used to represent the solid–electrolyte at the surface
of a birnessite particle in the electrode. Here the charge can either
undergo charge transfer from ionic conduction in the electrolyte to
electronic conduction in the birnessite, or alternatively be stored
as charge in the double layer – hence the parallel arrangement.
The parallel arrangement of R3 and CPE2 essentially represents a
Randles circuit for the electrified interface [28]. A constant phase
element has again been used here for much the same reasons as
described before.
To obtain the optimized values for the equivalent circuit com-
ponents compared to experimental data a complex non-linear
least squares (CNLS) regression technique has been used. This
method was proposed initially by Boukamp [29,30], and involves
simultaneously fitting the real and imaginary components of the
impedance, predicted from the equivalent circuit, using the expres-
sion
ꢀ
Fig. 7. Sample electrochemical impedance spectra for the birnessite-activated car-
bon asymmetric cell.
S =
wi[{Zꢀ − Zꢀ(ω)}2 + {Ziꢀꢀ − Zꢀꢀ(ω)}2]
(3)
i
i
i
where S is indicative of the divergence between experimental and
predicted data, Zi∗ = Ziꢀ + jZꢀꢀ is the measured data set, Zi∗(ω) =
3.4. Equivalent circuit
i
Ziꢀ(ω) + jZꢀꢀ i(ω) is the predicted data set, and wi is a weighting factor
inversely proportional the square of the modulus of the measured
data set; i.e.,
To be able to more fully comprehend the changes the birnessite
electrode is undergoing during cycling an appropriate equivalent
circuit is necessary. From the sample EIS data in Fig. 7 it is apparent
that over the frequency range considered there are two semi-
circular arcs present indicating the presence of at least two different
time constant processes. Given the physical construction of the bir-
nessite electrode in this cell, which is essentially a porous thin film
separating an electronically conducting substrate from the ioni-
cally conducting electrolyte, an appropriate equivalent circuit is
shown in Fig. 8. This circuit is similar to those used in corrosion
studies to characterize the nature of the passive film or coating on
the metal surface [26]. Here R1 represents the equivalent series
resistance between the reference electrode and the current collec-
tor substrate supporting the electrode. The parallel arrangement of
R2 and CPE1 represents the bulk properties of the electrode film. R2
corresponds to the cross-sectional resistance of the coating, while
CPE1 is a constant phase element used to represent the capacitance
of the coating. The impedance of a constant phase element (ZCPE) is
given by
1
|Z∗|2
wi =
(4)
The fitting procedure involved optimizing the equivalent circuit
parameters such that S is minimized. In all cases the fitting between
the model and experimental data was very good, meaning that S
was always small.
3.5. Parameter variation with state of charge
As a result of the combination of EIS and SPECS electrochemical
techniques, and analysis of the resultant EIS data, we are now in the
position to be able to examine how each of the equivalent circuit
parameters, and hence the electrode, changes with voltage or state
of charge. The variation of R1 with voltage is shown in Fig. 9(a).
The first noteworthy point is the magnitude of R1, which starts at
0.60 ꢀ, and at its maximum reaches 0.80 ꢀ. While certainly these
values are larger than that expected for a commercial supercapaci-
tor (20–50 mꢀ [3]), they are nevertheless sufficiently small enough
to conclude that we have produced a satisfactory supercapacitor
for study. Furthermore, the measured values for R1 can be readily
explained in terms of the different substrate used (Ti in an aque-
ous media compared to Al in a non-aqueous media in a commercial
device), as well as the use of more resistive birnessite compared to
activated carbon [31].
1
ZCPE
=
(2)
T(jω)P
where T is the equivalent capacitance of the film, j is the imaginary
√
number (j = −1), ω is the angular frequency (rad/s), and P is an
exponent reflecting the deviation of the constant phase element
What are next apparent are the systematic charges in R1 with
state of charge. From the OCV, R1 progressively increases during the
first anodic sweep from 0.60 ꢀ to 0.70 ꢀ when in the fully charged
state at 0.8 V. It then continues to increase during the subsequent
discharge half cycle, reaching a maximum of 0.80 ꢀ at 0.5 V before
decreasing gradually to 0.73 ꢀ in the fully discharged state at 0.0 V.
During the second anodic charge half cycle R1 goes through a mini-
mum of 0.70 ꢀ at 0.35 V (interestingly this corresponds to the OCV),
after which it increases to 0.79 ꢀ in the fully charged state. It is also
quite interesting to note that the curve shape during both charging
Fig. 8. Equivalent circuit used for electrochemical impedance spectroscopy mod-
elling.

A. Cormie et al. / Electrochimica Acta 55 (2010) 7470–7478
7475
Fig. 9. (a–e) Equivalent circuit parameter variation with respect to voltage for the birnessite electrode. Note that in each graph the larger symbol represents the starting
point for the experiment, and that CPE-T and CPE-P refers to the T and P parameters for the corresponding constant phase element.
half cycles are quite similar indicating similarities in the processes
occurring. Overall it is apparent that R1 undergoes an irreversible
change during cycling. The causes of these changes are most likely a
combination of the mechanical and chemical features of the system.
During electrochemical cycling the manganese dioxide structure
undergoes expansion and contraction upon reduction and oxida-
tion, respectively [32]. Since the birnessite is expected to undergo
redox reactions as a pseudo-capacitive material, its expansion and
contraction is expected to cause some mechanical strain between
particles (cohesion), as well as with the substrate (adhesion). The

7476
A. Cormie et al. / Electrochimica Acta 55 (2010) 7470–7478
effectiveness of the binder in terms of adhesion and cohesion will
determine how closely the birnessite particles return to their orig-
inal assembly (starting impedance), but clearly in this case this
has not been too successful. Another possible explanation for the
changes in R1 revolves around the changes in conductivity the bir-
nessite undergoes during redox cycling. Of course this is dependent
on the discharge mechanism and the phases that are produced,
but one common feature is the fact that the conductivity drops
considerably as reduction of the starting material occurs [25,31].
Birnessite is not a naturally very conductive phase of manganese
dioxide, due mainly to resistive inter-particle phase boundaries, but
upon reduction it becomes dramatically worse [31]. Furthermore,
if the phase changes that do occur are irreversible, then certainly
the resistance of the electrode will continue to increase [33–35].
The changes in CPE1, reflecting the capacitance of the electrode
film, with state of charge, are shown in Fig. 9(b). For a constant
phase element there are two parameters that need to be consid-
ered; namely the T and P parameters, as described in Eq. (2). As
was seen with R1, both CPE1 parameters exhibited irreversible
behaviour during cycling. The T parameter (which reflects the elec-
trode film capacitance) decreased continually from its value at the
OCV to a voltage of 0.70 V, after which it remained relatively con-
stant until the fully charged state. During the first discharge half
cycle CPE1-T overlay the corresponding values measured during
the first anodic half cycle until a voltage of 0.55 V was reached. At
lower voltages its value increased gradually until the lower voltage
limit was reached. Then during the second charge half cycle the
CPE1-T parameter value decreased gradually over the entire volt-
age range, albeit at a higher value than that measured during the
discharge. The behaviour of the CPE1-P parameter was inversely
proportional to the CPE1-T parameter.
These parameter changes for CPE1 are consistent with the
changes in R1, and can be justified using the same explanations. For
instance, mechanical changes in the electrode resulting from vol-
ume changes in the birnessite could certainly explain changes in the
electrode capacitance (T). Certainly changes in the CPE1-P parame-
ter support mechanical changes in the electrode. A P value of unity
would imply that the electrode is behaving like an ideal capacitor.
A decreasing P value means that the capacitive behaviour of the
electrode is losing its ideality, effectively through increased surface
roughness and hence intersecting double layers [27]. This would
be expected to occur if the electrode were undergoing mechanical
changes as a result of particle swelling.
The behaviour of R2 is shown in Fig. 9(c). In this case there
is a significant difference in parameter values between the first
and second charge half cycles at intermediate voltages, but what
is important is that the R2 values are the same at the upper voltage
limit. Furthermore, there is a substantial hysteresis loop between
the first discharge and second charge half cycles, suggesting that
the processes this parameter represents do not behave in the same
fashion during charge and discharge, although ultimately, they end
up in the same state.
will occur at the electrode–electrolyte interface, rather than at the
electrode–substrate interface, simply because of the differences in
electrolyte accessibility [36]. Therefore, the behaviour of R2 dur-
ing the discharge half cycle reflects the progression of discharge
through the electrode from the electrolyte side to the substrate
side. Upon current reversal, charge is moving through the elec-
trode in the opposite direction, and given that the properties of the
electrode are also expected to have changed to reflect the reduced
state, it may now mean that the oxidation processes occurring
begin at the electrode-substrate interface and progress through the
electrode in the opposite direction. Assuming that the electrode
properties are different (worse) than the fresh electrode [25], the
path that R2 adopts should also be different. Ultimately it would
seem that the electrode resistance returns to its original state,
meaning that this aspect of the electrode behaviour is reversible.
The behaviour of CPE2 (as indicated by its T and P parameters)
during cycling is shown in Fig. 9(d). What is clear about CPE2 is its
apparent reversibility during cycling, even to the point that dur-
ing both charging half cycles the data for the CPE2-T and CPE2-P
parameters essentially overlay each other. Another observation is
that the capacitance from this CPE is an approximately an order of
magnitude higher than that of CPE1, indicating that the capacitance
associated with the birnessite–electrolyte interface is substantially
larger than that due to the macroscopic electrode–electrolyte inter-
face. The cause of this is most likely the much higher specific surface
area of the birnessite particles compared to that of the macroscopic
electrode itself. What is also apparent in Fig. 9(d) is the appar-
ent drop in electrode capacitance with discharge. There has been
a report in the literature that the discharge (reduction) of man-
ganese dioxide (specifically ␥-MnO2) results in a decrease in its
specific surface area [37]. The cause of this decrease in surface
area was related back to the structural expansion the material is
undergoing during reduction of Mn4+ to Mn3+ in the structure. This
swelling was shown to result in a proportion of the micro-pore
volume being lost to pore collapse, as well as an increase in the
meso- and macro-pore volumes due to larger scale cracking of the
individual particles. Therefore, in this situation, as electrochemical
cycling proceeds through the discharge half cycle it is likely that
the micro-pores will close down limiting the amount of available
solid–electrolyte interface, thus causing the interfacial capacitance
to drop. This explanation is also supported by the CPE2-P data
which shows a general increase during discharge. Firstly the value
for CPE2-P is relatively low in the fully charged state indicating a
very porous surface. During discharge, with its associated pore col-
lapse, the interface is becoming less porous, and more like an ideal
capacitor (relatively), hence resulting in an increase in the P param-
eter. Finally, as with R2, both CPE2 parameters exhibit a hysteresis
loop during redox cycling indicating that discharge proceeds via a
different path compared to the charge half cycle. While for R2 this
was justified by considering the different directions of current flow
during charge and discharge, for CPE2 the most likely cause is hys-
teresis in pore collapse and re-opening, with pore opening being
easier during charge (higher capacitance) rather than pore closure
during discharge. The effect of reduction on pore closure has been
reported in the literature [37], but their re-opening during charge
is a new phenomenon, certainly one that needs further exploration.
The last equivalent circuit parameter to be considered is R3,
with its associated changes during cycling being shown in Fig.
9(e). From the original description of the equivalent circuit, this
parameter is used to describe the charge transfer resistance at the
birnessite–electrolyte interface. The reaction associated with this
process involves concerted electron and proton (or small alkali
metal ion) insertion into the structure of the manganese dioxide
through the interface; i.e.,
As has been discussed already, the birnessite is undergoing a
range of property changes during redox cycling, including volume
and conductivity changes, and even possibly phase changes. Of
these, conductivity changes in the birnessite would be expected to
have the most significant effect on the resistance of the film; how-
ever, macroscopic changes in the mechanical properties of the elec-
trode could also play a role. The fact that the value of R2 increases
with the depth of discharge is also in agreement with the expected
changes in the material properties. What is difficult to understand
is the origin of the hysteresis loop, which indicates that the charge
and discharge changes in R2 follow different paths. One possibility
stems from the differences in the direction of charge flow through
the electrode during charge and discharge. During discharge it is
reasonable to presume that the redox processes in the electrode
MnOOHx + ꢃH+ + ꢃe− ↔ MnOOHx+ꢃ
(5)

A. Cormie et al. / Electrochimica Acta 55 (2010) 7470–7478
7477
where x represents the mole fraction of protons present in the
structure already, and ꢃ the quantity of protons to be inserted.
Note that Eq. (5) could have quite as easily been written using an
alkali metal cation, although in this case the penetration depth of
the cation into the structure would have been limited to just the
surface, whereas protons and Li+ cations do have a demonstrated
ability to diffuse through the structure. In terms of proton inter-
calation and diffusion, the various manganese dioxide structures
behave quite differently. The preferred structure for aqueous alka-
line applications (␥-MnO2; 1 × 1 and 1 × 2 tunnels) intercalates
protons quite readily, and is the most facile in terms of proton
diffusion [38]. Larger tunnel structures (e.g., ␣-MnO2; 2 × 2 tun-
nels) function in the same way but their mass transport capabilities
are slower due to the larger separation between allowed proton
locations [39]. The smaller tunnel size in ␤-MnO2 (1 × 1 tunnels)
is very poor for proton intercalation, resulting in the reduction of
the surface only [39]. Proton intercalation into layered materials
such as birnessite has been examined previously, in which case it
was shown that the structure collapses to the more thermodynam-
ically stable Mn3O4 after ∼25% of reduction had occurred [32]. This
is significant because this process was shown to be irreversible, at
least in alkaline electrolytes [33–35], and also via chemical reduc-
tion [32]. Furthermore, it also becomes apparent that in previous
studies where birnessite has been used [40–42] as the active mate-
rial, the faradaic contribution to the capacitance is small since if
it were to continue to greater depths of discharge, the irreversible
formation of Mn3O4 would be the result.
the observed behaviour in effect reflects the expected behaviour
of the electrode after extended cycling. In particular, the following
material characteristics were found to be important:
(i) During redox cycling the birnessite unit cell swells and con-
tracts having an effect on the adhesion of the electrode to the
substrate, as well as its inherent mechanical integrity. This
feature is reflected in the irreversible increase in the series
resistance of the electrode. These volume changes were also
found to affect the capacitance and resistance of the macro-
scopic electrode in an irreversible manner.
(ii) Volume changes of the birnessite unit cell also affect the
capacitive performance of the birnessite–electrolyte interface.
Swelling of the structure was deduced to result in the closure
of material pores, effectively limiting the accessibility of elec-
trolyte to the interface, and hence the ability of the birnessite to
store charge. While substantial hysteresis was apparent with
this process, it was nonetheless reversible.
(iii) The redox reactions the birnessite undergoes have also been
shown to decrease in facility as discharge proceeds. Further-
more, the reported structural changes birnessite undergoes
during reduction (to form Mn3O4) also dramatically affect its
electrochemical behaviour, causing hysteresis during cycling
at least during part of the voltage window.
Acknowledgements
The first point to note concerning R3 is the relatively high value
for this resistance (on average ∼4000 ꢀ over the entire voltage
range considered) suggesting that charge transfer across this inter-
face is quite sluggish. Additionally, R3 exhibits some interesting
behaviour as a function of voltage, in particular a hysteresis loop in
the range 0.00–0.45 V, and overlaying charge–discharge data in the
range 0.45–0.80 V. Furthermore, the value of R3 tends to increase
when discharge is occurring, suggesting a progressive decrease in
the material’s ability to undergo efficient reduction. This behaviour
can be justified by considering the redox reactions the birnes-
site undergoes. During the initial stages of discharge (starting at
0.8 V and becoming more cathodic) reduction is as facile as it can
be. However, as reduction continues proton intercalation becomes
more difficult causing an increase in R3 (this is a common feature
on manganese dioxide in general [38]). Then at the state of charge
defined by 0.45 V, perhaps a structural transition occurs (birnes-
site to Mn3O4 [32–35]) causing a dramatic change in electrode
behaviour. During the anodic sweep charging of this new phase
follows a new path given the differences in properties between
birnessite and Mn3O4, thus forming a hysteresis loop, until ulti-
mately the electrode is returned to its original state; i.e., similar
R3 values in the fully charged state. This last point is perhaps con-
tentious given the previous literature which suggests that it is an
irreversible process. However, the depth of discharge into the bir-
nessite particles may not have been so great, meaning that on the
surface this process may have been reversible.
Ariana Cormie acknowledges the financial support provided by
the University of Newcastle in the form of a Summer Scholarship,
while Andrew Cross acknowledges the University of Newcastle and
the CSIRO Division of Energy Technology for a postgraduate schol-
arship.
References
[1] N.S. Lewis, Plenary Lecture 208th Meeting of the Electrochemical Society, Los
Angeles, CA, USA, October, 2005.
[2] D. Ragone, Proc. Soc. Automotive Engineers Conference, Detroit, MI, USA, May,
1968.
[3] www.cap-xx.com.
[4] B.E. Conway, Electrochemical Supercapacitors: Scientific Fundamentals and
Technological Applications, Kluwer-Plenum Publishing Company, New York,
1999.
[5] K. Naoi, P. Simon, Interface 17 (1) (2008) 34.
[6] D. Bélanger, T. Brousse, J.W. Long, Interface 17 (1) (2008) 49.
[7] W. Feitknecht, W. Marti, Helv. Chim. Acta 28 (1945) 129.
[8] W. Feitknecht, W. Marti, Helv. Chim. Acta 28 (1945) 149.
[9] O. Glemser, G. Gattow, H. Meisiek, Z. Anorg. Allg. Chem. 309 (1961) 1.
[10] Y.F. Yao, H.S. Wroblowa, J. Electroanal. Chem. 223 (1987) 107.
[11] R.M. McKenzie, Miner. Mag. 38 (1971) 493.
[12] J.B. Fernandes, B.D. Desai, V.N. Kamat Dalal, J. Power Sources 15 (1985) 209.
[13] E. Narita, T. Okabe, Bull. Chem. Soc. Japan 53 (1980) 525.
[14] M.H. Rossouw, D.C. Liles, M.M. Thackeray, Prog. Batt. Batt. Mater. 15 (1996) 8.
[15] J.W. Long, R.M. Stroud, D.R. Rolison, J. Non-Cryst. Solids 285 (2001) 288.
[16] D.R. Lide (Ed.), Handbook of Chemistry and Physics, 72nd ed., CRC Press Inc.,
1991.
[17] S. Brunauer, P.H. Emmett, E. Teller, J. Am. Chem. Soc. 60 (1938) 309.
[18] P.A. Webb, C. Orr, Analytical Methods in Fine Particle Technology, Micromeritics
Corporation, 1997.
[19] V.M. Burns, R.G. Burns, W.K. Zwicker, in: A. Kozawa, R.J. Brodd (Eds.), Proc.
MnO2 Symp., vol. 1, 1975, p. 288.
[20] R.G. Burns, V.M. Burns, in: A. Kozawa, R.J. Brodd (Eds.), Proc. MnO2 Symp., vol.
1, 1975, p. 306.
4. Conclusions
In this work we have combined the electrochemical tech-
niques of step potential electrochemical spectroscopy (SPECS)
and electrochemical impedance spectroscopy (EIS) to characterize
the quasi-equilibrium electrochemical behaviour of a disordered
birnessite sample in neutral aqueous electrolyte suitable for
supercapacitors. The results obtained reflect the impact that the
structural and morphological changes that occur have on the elec-
trochemical behaviour of the electrode. Furthermore, since the
experiments were carried out under quasi-equilibrium conditions,
[21] R.G. Burns, V.M. Burns, in: B. Schumm, H.M. Joseph, A. Kozawa (Eds.), Proc.
MnO2 Symp., vol. 2, 1980, p. 97.
[22] R. Giovanoli, in: I.M. Varentsov, G. Grasselly (Eds.), Proc. 2nd Inter. Symp., 1980,
p. 159.
[23] A.R. West, Solid State Chemistry and its Applications, John Wiley and Sons,
1984.
[24] L. Athouel, F. Moser, R. Dugas, O. Crosnier, D. Bélanger, T. Brousse, ECS Trans.
16 (2008) 119.
[25] X. Xia, H. Li, Z.H. Chen, J. Electrochem. Soc. 136 (1989) 266.
[26] W.S. Tait, An Introduction to Electrochemical Corrosion Testing for Practicing
Engineers and Scientists, PairODocs Publications, 1994.

7478
A. Cormie et al. / Electrochimica Acta 55 (2010) 7470–7478
[27] For example: S.R. Narayanan, D.H. Shen, S. Surampudi, A.I. Attia, G. Halpert, J.
Electrochem. Soc. 140 (1993) 1854.
[35] S.W. Donne, G.A. Lawrance, D.A.J. Swinkels, J. Electrochem. Soc. 144 (1997)
2961.
[28] J.E.B. Randles, Discuss. Faraday Soc. 1 (1947) 11.
[29] B.A. Boukamp, Solid State Ionics 18 (1986) 136.
[30] B.A. Boukamp, Solid State Ionics 20 (1986) 30.
[31] S.W. Donne, High Capacity Chemically and Physically Modified Manganese
Dioxide, PhD Thesis, University of Newcastle, 1996.
[32] S.W. Donne, R.A. Fredlein, G.A. Lawrance, D.A.J. Swinkels, F.L. Tye, Prog. Batt.
Batt. Mater. 13 (1994) 113.
[36] T.W. Farrell, C.P. Please, D.L.S. McElwain, D.A.J. Swinkels, J. Electrochem. Soc.
147 (2000) 4034.
[37] A.P. Malloy, S.W. Donne, J. Electrochem. Soc. 155 (2008) A817.
[38] G.J. Browning, S.W. Donne, J. Appl. Electrochem. 35 (2005) 871.
[39] S.W. Donne, unpublished work – Discharge characteristics of ␣-/␤-MnO2.
[40] B. Messaoudi, S. Joiret, M. Keddam, H. Takenouti, Electrochim. Acta 46 (2001)
2487.
[33] S.W. Donne, G.A. Lawrance, D.A.J. Swinkels, J. Electrochem. Soc. 144 (1997)
2949.
[34] S.W. Donne, G.A. Lawrance, D.A.J. Swinkels, J. Electrochem. Soc. 144 (1997)
2954.
[41] J.N. Broughton, M.J. Brett, Electrochim. Acta 49 (2004) 4439.
[42] B. Djurfors, J.N. Broughton, M.J. Brett, D.G. Ivey, J. Electrochem. Soc. 153 (2006)
A64.