Role of bismuth and factors influencing the formation of Mn3O4 in rechargeable alkaline batteries based on bismuth-containing manganese oxides

Article abstract of DOI:10.1149/1.1524611

The two-electron reduction/oxidation reactions of manganese oxide in alkaline electrolytes have been investigated in the presence and absence of bismuth oxide by both galvanostatic discharge/charge and linear sweep voltammetry followed by examining the products with X-ray diffraction. Although the presence of bismuth helps the formation of birnessite MnO₂, it could not completely prevent the formation of Mn₃O₄. However, the formation of Mn₃O₄ could be suppressed by controlling the scan rate during linear sweep voltammetry and/or the microstructure of the electrode. Mn₃O₄ appears to be formed by a chemical reaction between dissolved Mn(II) and Mn(III) complexes (intermediates) that have diffused away from the conductive path (graphite). Both a faster scan rate (discharge/charge rate) and an intimate mixing of the manganese oxide with graphite achieved by ballmilling suppress the formation of Mn₃O₄ independent of the presence of bismuth.

Full text of DOI:10.1149/1.1524611

A68
Journal of The Electrochemical Society, 150 ͑1͒ A68-A73 ͑2003͒
0013-4651/2002/150͑1͒/A68/6/$7.00 © The Electrochemical Society, Inc.
Role of Bismuth and Factors Influencing the Formation
of Mn3O4 in Rechargeable Alkaline Batteries Based
on Bismuth-Containing Manganese Oxides
*
**,z
D. Im and A. Manthiram
Materials Science and Engineering Program, The University of Texas at Austin, Austin, Texas 78712, USA
The two-electron reduction/oxidation reactions of manganese oxide in alkaline electrolytes have been investigated in the presence
and absence of bismuth oxide by both galvanostatic discharge/charge and linear sweep voltammetry followed by examining the
products with X-ray diffraction. Although the presence of bismuth helps the formation of birnessite MnO2 , it could not completely
prevent the formation of Mn3O4 . However, the formation of Mn3O4 could be suppressed by controlling the scan rate during linear
sweep voltammetry and/or the microstructure of the electrode. Mn3O4 appears to be formed by a chemical reaction between
dissolved Mn͑II͒ and Mn͑III͒ complexes ͑intermediates͒ that have diffused away from the conductive path ͑graphite͒. Both a faster
scan rate ͑discharge/charge rate͒ and an intimate mixing of the manganese oxide with graphite achieved by ballmilling suppress
the formation of Mn3O4 independent of the presence of bismuth.
© 2002 The Electrochemical Society. ͓DOI: 10.1149/1.1524611͔ All rights reserved.
Manuscript submitted April 24, 2002; revised manuscript received July 5, 2002. Available electronicallyNovember 21, 2002.
Mn͑OH͒ and/or birnessite, are formed from the dissolved Mn3ϩ
Rechargeable batteries based on manganese oxides are attractive
for applications such as electric vehicles that require large cells be-
cause manganese is abundant, inexpensive, and environmentally be-
nign. Particularly, rechargeable alkaline cells involving two elec-
trons per manganese between Mn4ϩ and Mn2ϩ are appealing as they
can provide high energy densities exceeding that of currently avail-
able lithium-ion cells. Additionally, with the use of aqueous electro-
lytes, the alkaline cells offer better safety characteristics compared
to lithium-ion cells. However, the poor rechargeability of manganese
oxides in alkaline cells has limited their use to mainly primary bat-
teries. During the two-electron reduction in concentrated KOH elec-
trolytes, ␥-MnO2 ͑EMD͒ is discharged to Mn͑OH͒2 , which is then
recharged to ␦-MnO2 ͑layered birnessite structure͒.1,2 The reduction
of birnessite during the subsequent cycles results in the formation of
Mn3O4 ͑spinel structure͒, which has been regarded as a reason for
the poor rechargeability.
Many efforts have been made to achieve better rechargeability
with the alkaline manganese oxide cells. For example, Kordesch
et al.3 investigated limited depth of discharge of ␥-MnO2 by using a
controlled amount of Zn at the anode. Although the imposition of
such a limit provided better cyclability, the discharge depth was only
0.35 electron per Mn, which corresponds to less than 20% of the
theoretical two-electron capacity. An important breakthrough was
made by Wroblowa et al.4-6 in the mid-1980s by adding Bi3ϩ either
chemically or physically to manganese oxide cathodes. They dem-
onstrated that the cells made with bismuth-modified manganese di-
oxides ͑BMD͒ could be cycled several hundred times with approxi-
mately 1.7 electrons per Mn, regardless of the way of bismuth
incorporation, but with low active material loading.
Although some problems such as the formation of ZnMn2O4 on
coupling with Zn anode remained to be solved, the success of Wrob-
lowa et al.4-6 provoked a burst of research activity to understand the
role of bismuth. Initially, Wroblowa et al.4-6 suggested that the Bi3ϩ
ions play a critical role in maintaining the open-layered structures of
manganese oxides during the discharge-charge process. They ex-
plained that the Bi3ϩ ions help to maintain the layered structures of
both the charge product, birnessite MnO2 , and the discharge prod-
uct, Mn͑OH͒2 , consisting of edge-shared MnO6 octahedra. Later,
Conway et al.7-9 detected dissolved Mn3ϩ species from spectro-
scopic and rotating ring-disk electrode ͑RRDE͒ experiments during
the discharge-charge process. They suggested that the solid phases,
2
intermediates and Bi3ϩ might promote the nucleation and growth of
those crystal phases. This idea of solution process, which was in fact
originally suggested by Ruetschi10 with optical microscopy, was re-
cently validated by Patrice et al.11 with scanning electron micro-
scopic ͑SEM͒ studies. They showed that the particles are reshaped
after the charge and discharge processes, indicating that the redox
reaction is not simply an all solid-state process.
While the above groups discussed the promotional or catalytic
role of bismuth in rendering rechargeability, some other groups con-
sidered a preventive role played by bismuth in the formation of the
unwanted phase Mn3O4 . For example, Donne et al.12 found from
X-ray diffraction ͑XRD͒ studies the formation of Mn3O4 instead of
Mn͑OH͒ as the final reduction product with bismuth-free birnessite
2
cathodes and Mn͑OH͒ as the major product with bismuth-
2
containing cathodes. Yu et al.13,14 believed that the reaction between
dissolved Mn2ϩ and Mn3ϩ complexes leads to the formation of
Mn3O4 , and the formation of bismuth-manganese complexes in the
case of bismuth-containing cathodes as intermediate compounds
prevents such a reaction. Abou-El-Sherbini et al.15 investigated the
electrodes by elemental analysis in SEM and concluded that bismuth
complexes adsorbed on the sides of the layers of manganese oxides
protect them from being attacked by Mn2ϩ complexes such as
4Ϫ
Mn͑OH͒
.
͓
͔
6
Recently, our group and RBC Technologies16 systematically in-
vestigated the phases formed at various depths of discharge and
charge and showed that the two-electron discharge-charge process
in bismuth-modified manganese oxides involves
a reversible
dissolution/insertion of Kϩ ions from/into the cathode lattice. The
insertion of Kϩ ions was found to assist the formation of a well-
crystallized birnessite phase during charge. With an objective to de-
termine the role of bismuth more clearly and the factors that influ-
ence the formation of unwanted Mn3O4 , we present here a
systematic analysis of the products formed during both galvanostatic
cycling and linear sweep voltammetry. The products formed under
various conditions are analyzed by slow scan XRD.
Experimental
A layered manganese dioxide supplied by Carus Chemical Com-
pany was employed as active material. XRD indicated the sample to
have the layered birnessite structure with low crystallinity. Atomic
absorption spectroscopic analysis showed the sample to consist of
only 39.7 wt % manganese, suggesting the presence of plenty of
interlayer water. The average oxidation state of manganese in the
sample was determined to be ϩ3.90 by a redox titration employing
oxalate. Bi2O3 was synthesized by dissolving Bi(NO3)3 in a weakly
* Electrochemical Society Student Member.
** Electrochemical Society Active Member.
z E-mail: rmanth@mail.utexas.edu
Downloaded on 2015-03-17 to IP 140.182.176.13 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 150 ͑1͒ A68-A73 ͑2003͒
acidic solution ͑obtained by adding several drops of nitric acid to
A69
100 mL of deionized water͒ and then adding ammonium hydroxide
until the pH reached 12. The solid which formed was filtered,
washed with deionized water, and fired at 600°C for 8 h. TIMREX
SFG-15 graphite was used as the conductive agent in the fabrication
of electrodes.
Bismuth-containing electrodes were made by mixing 45 wt %
manganese dioxide, 5 wt % bismuth oxide, 45 wt % graphite, and 5
wt % polytetrafluoroethylene ͑PTFE͒. Bismuth-free electrodes were
made with 50 wt % manganese dioxide, 45 wt % graphite, and 5 wt
% PTFE. The components were mixed in a mortar and pestle or in
some cases by ballmilling for 15 min in a SPEX CertiPrep model
8000M mixer/mill, and formed into thin-film type electrodes con-
sisting of approximately 5 mg of active material. The electrochemi-
cal performances were evaluated with a typical three-electrode sys-
tem constructed with a Hg/HgO ͑in 9 M KOH solution͒ reference
electrode and a porous nickel counter electrode consisting of
Ni͑OH͒2 . A nickel mesh current collector was used for cathodes and
9 M KOH aqueous solution was used as electrolyte. Galvanostatic
experiments were carried out with an Arbin model BT2000 battery
cycler. Linear sweep voltammetry was controlled with a VoltaLab
PGZ402 potentiostat ͑Radiometer Analytical S.A.͒. All voltages pre-
sented in this paper are relative to the Hg/HgO reference electrode.
Structural characterizations of the cathode materials before and
after the electrochemical tests were carried out with XRD. In the
case of discharged/charged cathodes, the electrodes were removed
from the cell, washed with water, and dried in vacuum at room
temperature. The diffractograms were collected in the 2␪ range of 10
to 70° with a step scan rate of 5 s at each 0.02° step.
Results and Discussion
Galvanostatic discharge/charge experiments.—The galvanostatic
discharge/charge experiments were carried out by discharging the
cells at a C/2 rate to Ϫ1.1 V and charging at a C/4 rate to 0.35 V.
The first cycle discharge-charge voltage profiles of the bismuth-
containing and bismuth-free electrodes are compared in Fig. 1. The
XRD patterns of the electrodes before and after the discharge-charge
processes are given in Fig. 2 and 3. The bismuth-containing sample
͑Fig. 1a͒ shows slightly higher discharge capacity than the bismuth-
free sample ͑Fig. 1b͒, but the difference is not significant. The ob-
served difference could be due to the reduction of bismuth oxide to
metallic bismuth in the former. The differences seen in the discharge
profile below Ϫ0.6 V are also due to the reduction of bismuth oxide.
The degree of active material utilization is quite high for both the
electrodes. Even the Bi-free electrode shows capacity corresponding
to more than 1.5 electrons per Mn.
Despite the close similarities in the discharge profiles, the two
samples show significant differences in the charge profiles. The
bismuth-containing sample shows a three-step charging behavior
(Ϫ0.4, Ϫ0.2, and ϩ0.2 V). The first one at Ϫ0.4 V appears to be
due to the oxidation of bismuth, while the other two correspond to
the oxidation of manganese oxide.13 The recharge capacity of the
bismuth-containing sample is slightly higher than the discharge ca-
pacity, which could be due to (i) the initial manganese oxidation
state of less than ϩ4 as mentioned in the experimental section and
(ii) some possible overcharge resulting from the oxidation of water.
In contrast, the recharge capacity of the bismuth-free sample is
much lower than the discharge capacity. The charge capacity corre-
sponds to only 0.75 electrons per Mn, and there is no clear evidence
of a two-phase reaction.
Figure 1. Galvanostatic discharge ͑C/2͒ and charge ͑C/4͒ profiles of ͑a͒
bismuth-containing and ͑b͒ bismuth-free electrodes.
muth helps to suppress the formation of Mn3O4 . The formation of
Mn3O4 is in agreement with our previous work,16 but is in contrast
to the single-phase Mn͑OH͒ reported by Donne et al.12 in the pres-
2
ence of bismuth. The difference could be due to the differences in
mixing with graphite ͑see below͒.
It has been suggested in the literature that Mn3O4 could be
formed by an oxidation of the discharge product Mn͑OH͒ on expo-
2
sure to air rather than during the discharge process.10,11 In order to
clarify this issue, we investigated by XRD the oxidation product of
Mn͑OH͒ in air. We found that the oxidation product is ␤-MnOOH
2
and not Mn3O4 , even after exposing Mn͑OH͒ to air for an ex-
2
tended period of time ͑2 months͒ either at room temperature or at
elevated temperatures ͑60°C͒. These experiments demonstrate that
the formation of Mn3O4 occurs during the discharge-charge process
rather than due to air oxidation.
As expected from the differences in the voltage profiles of
bismuth-containing and bismuth-free electrodes, the phases formed
after recharge are completely different for the two electrodes ͑Fig.
2c and 3c͒. While well-developed crystalline birnessite MnO2 is
formed in the case of bismuth-containing electrode ͑Fig. 2c͒, only
very weak broad reflections corresponding to birnessite MnO2 ,
␤-MnOOH, and Mn3O4 are observed in the case of bismuth-free
electrode ͑Fig. 3c͒, suggesting the presence of some amorphous or
poorly crystalline phases in the latter case. The results clearly dem-
onstrate that the presence of bismuth helps the formation of birnes-
The XRD data shown in Fig. 2 and 3 are consistent with the
voltage profiles in Fig. 1. For the fully discharged electrodes ͑Fig.
2b and 3b͒, Mn͑OH͒ appears to be the major product regardless of
2
the presence of bismuth as one would expect from the high dis-
charge capacity observed for both the electrodes. Also, weak reflec-
tions corresponding to Mn3O4 can be noticed in the case of both the
electrodes, but the amount of Mn3O4 is slightly less in the case of
bismuth-containing electrode suggesting that the presence of bis-
Downloaded on 2015-03-17 to IP 140.182.176.13 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A70
Journal of The Electrochemical Society, 150 ͑1͒ A68-A73 ͑2003͒
Figure 2. XRD patterns of bismuth-containing electrodes before and after
galvanostatic experiments: ͑a͒ fresh electrode, ͑b͒ discharged electrode
(Ϫ1.1 V), ͑c͒ subsequently charged electrode ͑0.35 V͒, and ͑d͒ charged elec-
trode ͑0.35 V͒ without prior discharge.
Figure 3. XRD patterns of bismuth-free electrodes before and after galvano-
static experiments: ͑a͒ fresh electrode, ͑b͒ discharged electrode (Ϫ1.1 V),
͑c͒ subsequently charged electrode ͑0.35 V͒, and ͑d͒ charged electrode ͑0.35
V͒ without prior discharge.
site with an incorporation of Kϩ ions from the electrolyte into the
manganese oxide lattice,16 and the realization of a large reversible
capacity. Additionally, the vanishing of Mn3O4 on going from Fig.
2b to 2c indicates that the presence of bismuth renders rechargeabil-
ity to any Mn3O4 that was formed during the discharge process,
which is in accordance with our previously reported data on a physi-
cal mixture consisting of Mn3O4 and Bi2O3 .16
Another interesting observation is found on charging the elec-
trodes without prior discharge ͑Fig. 2d and 3d͒. Since the average
oxidation state of manganese in the original material is ϩ3.90 ͑close
to ϩ4), only a very small amount of charging was possible. Never-
theless, a small, but sharp reflection corresponding to birnessite can
be noticed in the case of bismuth-containing electrode ͑Fig. 2d͒,
while no significant change is found in the case of bismuth-free
electrode ͑Fig. 3d͒. The difference between Fig. 2d and 3d again
demonstrates that the presence of bismuth helps the formation of a
well-crystallized birnessite phase.
Linear sweep voltammetry combined with X-ray diffrac-
tion.—The linear voltage sweep method can provide more informa-
tion about the kinetic effect compared to the galvanostatic method.
In the galvanostatic mode, the system would spend more time in the
voltage regions where electrochemical oxidation or reduction oc-
curs. In contrast, the time spent by the system would be more evenly
distributed through out the entire voltage region in the linear voltage
sweep method. For example, if an electrochemical process consists
of two steps, the intermediate produced from the first step would
undergo the reaction after the first step is finished in the case of
galvanostatic mode. On the other hand, in the case of a linear volt-
age sweep method, the intermediate will not undergo an electro-
chemical reaction while the voltage is being swept between the two
potentials representing each step. We have investigated the reduction
reaction of bismuth-containing manganese dioxide with linear
sweep voltammetry with an aim to find out if a deliberate change in
kinetics can affect the formation of Mn3O4 .
Returning to the debate about the promotional vs. preventive role
of bismuth as discussed in the Introduction, a comparison of the data
in Fig. 2d and 3d clearly supports the former. In other words, bis-
muth promotes the birnessite formation reaction independent of the
prevention of Mn3O4 formation. Nonetheless, we also noticed ear-
lier that the presence of bismuth reduces the amount of Mn3O4 at
lower voltages ͑Fig. 2b, 2c, 3b, and 3d͒. From these results, we can
consider two possibilities. The first possibility is that bismuth not
only promotes birnessite formation, but also prevents the crystalli-
zation of Mn3O4 . The second possibility is that bismuth does not
prevent the formation of Mn3O4 , but it appears to do so because it
Figure 4 shows the voltammogram of the bismuth-containing
electrode in which voltage was scanned from open-circuit potential
(ϳ0 V) to Ϫ1.1 V at a scan rate of 0.2 mV/s. A couple of broad
peaks are observed above Ϫ0.6 V, and a sharper cathodic peak is
located between Ϫ0.6 and Ϫ0.8 V. The broadness of the peak sug-
gests that the reduction processes are non-Nernstian. The exponen-
tial increase of current below Ϫ1.0 V could be due to the decompo-
sition of water.
To develop a better understanding of each process, a series of
linear sweep experiments were carried out by terminating the sweep
at various voltages (Ϫ0.6, Ϫ0.8, and Ϫ1.1 V) indicated by arrows
in Fig. 4 and the product obtained at each voltage level was exam-
ined by XRD after washing and drying the product in vacuum. The
catalyzes the formation of birnessite or Mn͑OH͒ and, thereby di-
2
minishes the chances of formation of Mn3O4 .
Downloaded on 2015-03-17 to IP 140.182.176.13 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 150 ͑1͒ A68-A73 ͑2003͒
A71
Figure 4. A linear sweep voltammogram of a bimuth-containing electrode.
Voltage was swept cathodically from open-circuit potential to Ϫ1.1 V at a
scan rate of 0.2 mV/s.
XRD patterns of the products obtained at various voltages are shown
in Fig. 5. When the potential reaches Ϫ0.6 V ͑Fig. 5b͒, the reflection
corresponding to birnessite at around 2␪ Ϸ 12°, though very weak
initially in Fig. 5a, vanishes completely and a sharp reflection cor-
responding to Mn͑OH͒ appears. Also, bismuth oxide does not ap-
2
pear to be reduced yet. As the electrode goes through a sharp volta-
mmetric peak and the voltage reaches Ϫ0.8 V, the reflections
corresponding to Mn͑OH͒ become more prominent and metallic
2
Figure 6. XRD patterns of bismuth-containing electrodes that were cathodi-
cally scanned to Ϫ0.8 V at various sweep rates: ͑a͒ 0.5, ͑b͒ 0.2, and ͑c͒ 0.01
mV/s.
bismuth instead of bismuth oxide could be seen ͑Fig. 5c͒. Also, the
reflections corresponding to Mn3O4 become more pronounced. Al-
though the reflections corresponding to Mn3O4 are still weak, it can
be recognized that they are slightly more intense compared to that
observed in the case of the product obtained by a galvanostatic
discharge of the bismuth-containing electrode ͑Fig. 2b͒. Further re-
duction down to Ϫ1.1 V does not cause much change excepting a
slight increase in the intensity of the Mn͑OH͒ reflections ͑Fig. 5d͒.
2
In order to see the influence of scan rate, further experiments
were carried out with various sweep rates for the bismuth-containing
electrodes. Since we found from the previous experiment in Fig. 5
that Mn3O4 has already formed at Ϫ0.8 V, the experiments with
different sweep rates were carried out down to a fixed end potential
of Ϫ0.8 V. Figure 6 shows the XRD patterns of the products ob-
tained at Ϫ0.8 V for various sweep rates. While the reflections cor-
responding to Mn3O4 are negligible after a sweep rate of 0.5 mV/s,
they are conspicuous for 0.01 mV/s.
The linear sweep voltammetry results indicate at least the follow-
ing three points. First, bismuth is not completely able to prevent the
formation of Mn3O4 . However, this could be due to the poor mixing
of bismuth oxide with the manganese oxide in the electrode, which
is discussed later. Second, the amount of Mn3O4 is governed by the
residual time of Mn͑III͒ intermediates since a slower sweep rate
͑discharge rate͒ gives more Mn3O4 . It suggests that Mn3O4 could
be formed through a chemical reaction between intermediates rather
than a direct reduction from manganese dioxide. Third, once Mn3O4
is formed, it is difficult to reduce it even after the potential reaches
Ϫ1.1 V ͑Fig. 5d͒ although it is known that it can be reduced to
Mn͑OH͒2 .1,16 The difference could be due to the location of Mn3O4
away from the conduction path provided by the graphite network
͑see later͒. Thus it may be concluded that some intermediates diffuse
Figure 5. XRD patterns of bismuth-containing electrodes that were col-
lected at various voltages ͑points marked with a, b, c, and d in Fig. 4͒ during
a linear sweep voltammetry with 0.2 mV/s: ͑a͒ fresh electrode ͑point a in
Fig. 4͒, ͑b͒ at Ϫ0.6 V ͑point b in Fig. 4͒, ͑c͒ at Ϫ0.8 V ͑point c in Fig. 4͒,
and ͑d͒ at Ϫ1.1 V ͑point d in Fig. 4͒.
Downloaded on 2015-03-17 to IP 140.182.176.13 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A72
Journal of The Electrochemical Society, 150 ͑1͒ A68-A73 ͑2003͒
Figure 8. XRD patterns of bismuth-free electrodes that were cathodically
scanned to Ϫ0.8 V at 0.01 mV/s. Electrodes were made by ͑a͒ simply mixing
graphite, manganese oxide, and PTFE in a mortar and pestle without any
ballmilling, ͑b͒ ballmilling graphite, followed by mixing the ballmilled
graphite with manganese oxide, and PTFE in a mortar and pestle, and ͑c͒
ballmilling graphite and manganese oxide, followed by mixing the ballmilled
composite with PTFE in a mortar and pestle.
Figure 7. XRD patterns of bismuth-containing electrodes that were cathodi-
cally scanned to Ϫ0.8 V at 0.01 mV/s. Electrodes were made by ͑a͒ simply
mixing graphite, manganese oxide, bismuth oxide, and PTFE in a mortar and
pestle without any ballmilling, ͑b͒ ballmilling graphite, followed by mixing
the ballmilled graphite with manganese oxide, bismuth oxide, and PTFE in a
mortar and pestle, and ͑c͒ ballmilling graphite, manganese oxide, and bis-
muth oxide, followed by mixing the ballmilled composite with PTFE in a
mortar and pestle.
scanning electron microscopic ͑SEM͒ examination showed that ball-
milling has reduced the grain size of manganese oxide from ϳ10 to
Ͻ1 ␮m. The electrodes were then subjected to linear sweep voltam-
metry at a sweep rate of 0.01 mV/s down to Ϫ0.8 V similar to that
for the sample in Fig. 6c. The XRD patterns of the products obtained
after the sweep are shown in Fig. 7b and 7c along with the XRD
pattern for a product obtained from an electrode without any ball-
milling ͑Fig. 7a͒.
The data show that the amount of Mn3O4 decreases significantly
on ballmilling the graphite ͑Fig. 7b͒. More importantly, Mn3O4 is
completely eliminated on ballmilling both graphite and the manga-
nese oxide together ͑Fig. 7c͒. The experiments clearly demonstrate
that an intimate contact of the manganese oxide particles with the
conduction path ͑graphite͒ is effective in suppressing the formation
of Mn3O4 . Furthermore, the data question the full preventive role of
bismuth since Mn3O4 formation is retarded even without enhanced
mixing ͑without ballmilling͒ with bismuth oxide ͑Fig. 7b͒. It thus
appears that the microstructure of carbon is more important than
mixing with Bi2O3 with respect to the suppression of the formation
of Mn3O4 .
from the place of their formation and meet other intermediates to
form Mn3O4 . This is in agreement with the proposal of Yu14 that
Mn3O4 is formed through a polymerization reaction between dis-
3Ϫ
4Ϫ
solved Mn͑OH͒
and Mn͑OH͒
͔
6
.
͓
͔
͓
6
Role of electrode microstructure.—If Mn3O4 is formed by a
chemical reaction between dissolved Mn͑II͒ and Mn͑III͒ complexes,
we can anticipate that its formation can be suppressed by shortening
the distance between the conductive paths. In other words, a denser
network of conduction pathways could reduce the chance of Mn3O4
formation. Even if Mn3O4 is formed, it could be reduced to
Mn͑OH͒ if it is in contact with the conduction path. With an aim to
2
provide a better conduction network, we decided to utilize ballmill-
ing for the production of electrodes.
Initially, only graphite was ballmilled and then mixed with other
components including bismuth oxide in a mortar and pestle. This
procedure does not affect the degree of mixing of bismuth oxide
significantly, but it would shorten the average distance between the
carbon particles. In addition, the surface area of carbon is expected
to be increased to offer better kinetics for the formation of Mn͑OH͒
2
Similar sweep voltammetry experiments with a scan rate of 0.01
mV/s were also conducted with bismuth-free electrodes with and
without ballmilling. The XRD patterns of the products obtained after
sweeping down to Ϫ0.8 V are shown in Fig. 8. A comparison of Fig.
8a and Fig. 7a indicates that the presence of bismuth suppresses the
formation of Mn3O4 even without ballmilling. Bismuth appears to
be somewhat effective in preventing the formation of Mn3O4 as we
or birnessite since both the charging and discharging involve
dissolution-precipitation processes.7-11 In another experiment,
graphite and a mixture of manganese oxide and bismuth oxide were
ballmilled together before mixing with PTFE. This procedure is an-
ticipated to breakdown the large agglomerates of the manganese
oxide and provide an intimate mixing and better distribution of the
conduction path ͑graphite͒ with the manganese oxide. In fact, a
Downloaded on 2015-03-17 to IP 140.182.176.13 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 150 ͑1͒ A68-A73 ͑2003͒
A73
have seen earlier in galvanostatic discharge experiments ͑Fig. 2b
and 3b͒. However, as evident from Fig. 8b and 8c, which involved,
respectively, the ballmilling of graphite alone and graphite and man-
ganese oxide together, a shortening of the average distance between
the carbon particles and/or an intimate mixing and better distribution
of the conduction path ͑graphite͒ with the manganese oxide are still
very effective in suppressing the Mn3O4 formation even in the ab-
sence of bismuth. It is interesting to see that no Mn3O4 phase could
be detected on ballmilling graphite and manganese oxide even in the
absence of bismuth ͑Fig. 8c͒.
The presence of bismuth is found to promote the formation of
birnessite during recharge and suppress slightly the formation of
Mn3O4 during discharge. However, the discharge/charge rate and
the microstructure of the electrode are more important factors in
controlling the formation of Mn3O4 than the presence of bismuth.
Bismuth appears to catalyze the reduction of Mn͑III͒ intermediates
to Mn͑OH͒ and thereby decreases the amount of Mn3O4 slightly
2
rather than directly preventing the formation of Mn3O4 .
Although bismuth undoubtedly helps to suppress the formation
of Mn3O4 , the electrode kinetics appears to be more influential than
the presence of bismuth. If Mn͑III͒ intermediates are present and a
reduction potential is applied, the electrochemical reduction forming
Acknowledgments
Financial support by RBC Technologies, College Station, TX, is
gratefully acknowledged.
Mn͑OH͒ and the chemical reaction forming Mn3O4 will compete
2
with each other. As we pointed out earlier from the shape of the
voltammogram, the electrochemical reactions here are very slow
͑non-Nernstian͒, and therefore some overpotential may exist. The
observation that faster discharge results in less Mn3O4 can be un-
derstood as follows. An earlier escape from the overpotential region
can accelerate the reduction reaction without allowing enough time
for the other chemical reaction to give Mn3O4 . Similarly, if bismuth
catalyzes the formation of Mn͑OH͒2 , then it can also help to retard
the formation of Mn3O4 . Additionally, the formation of Bi-Mn com-
plexes suggested in the literature13 may also lower the overpotential
for the Mn͑III͒ to Mn͑II͒ reaction. Although we do not have direct
evidence to understand the exact role of bismuth, it appears that
The University of Texas at Austin assisted in meeting the publication
costs of this article.
References
1. J. McBreen, in Power Sources 5: Proceedings 9th International Power Sources
Symposium, D. H. Collins, Editor, p. 525, Academic Press, New York ͑1975͒.
2. J. McBreen, Electrochim. Acta, 20, 221 ͑1975͒.
3. K. Kordesch, J. Gsellmann, M. Peri, K. Tomantschger, and R. Chemelli, Electro-
chim. Acta, 26, 1495 ͑1981͒.
4. Y. F. Yao, N. Gupta, and H. S. Wroblowa, J. Electroanal. Chem. Interfacial Elec-
trochem., 223, 107 ͑1987͒.
5. H. S. Wroblowa and N. Gupta, J. Electroanal. Chem., 238, 98 ͑1987͒.
6. M. A. Dzieciuch, N. Gupta, and H. S. Wroblowa, J. Electrochem. Soc., 135, 2415
͑1988͒.
bismuth promotes the formation of Mn͑OH͒ during reduction and
2
7. D. Y. Qu, B. E. Conway, L. Bai, Y. H. Zhou, and W. A. Adams, J. Appl. Electro-
chem., 23, 693 ͑1993͒.
facilitates the formation of birnessite during oxidation.
8. C. G. Castledine and B. E. Conway, J. Appl. Electrochem., 25, 707 ͑1995͒.
9. B. E. Conway, D. Qu, and J. McBreen, in Synchrotron Techniques in Interfacial
Electrochemistry, C. A. Melendres and A. Tadjeddine, Editors, p. 331, Kluwer
Academic Publishers, Dordrecht ͑1994͒.
10. P. Ruetschi, J. Electrochem. Soc., 123, 495 ͑1976͒.
11. R. Patrice, B. Gerand, J. B. Leriche, L. Seguin, E. Wang, R. Moses, K. Brandt, and
J. N. Tarascon, J. Electrochem. Soc., 148, A448 ͑2001͒.
12. S. W. Donne, G. A. Lawrance, and D. A. J. Swinkels, J. Electrochem. Soc., 144,
2961 ͑1997͒.
13. M. Bode, C. Cachet, S. Bach, J.-P. Pereira-Ramos, J. C. Ginoux, and L. T. Yu, J.
Electrochem. Soc., 144, 792 ͑1997͒.
14. L. T. Yu, J. Electrochem. Soc., 144, 802 ͑1997͒.
15. Kh. S. Abou-El-Sherbini, M. H. Askar, and R. Schollhorn, Solid State Ionics, 139,
121 ͑2001͒.
16. A. M. Kannan, S. Bhavaraju, F. Prado, M. Manivel Raja, and A. Manthiram, J.
Electrochem. Soc., 149, A483 ͑2002͒.
Conclusions
The factors that influence the formation of Mn3O4 during the
two-electron reduction-oxidation process of a layered birnessite
MnO2 in the presence and absence of bismuth oxide have been
investigated systematically. The formation of Mn3O4 is found to be
controlled by both the reaction kinetics and the microstructure of the
electrode. The amount of Mn3O4 decreases with increasing scan rate
during linear sweep voltammetry ͑faster discharge/charge rate͒ and a
better distribution of the conductive network ͑graphite͒ by ballmill-
ing. The results suggest that Mn3O4 is formed by a chemical reac-
tion between dissolved Mn͑II͒ and Mn͑III͒ complexes in the elec-
trolyte as proposed by Yu14 and not by a direct electrochemical
reaction.
Downloaded on 2015-03-17 to IP 140.182.176.13 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).