Ensr~y Conversion. Vol. 10. pp. 45-49. Pergamon Press,
Printed in Great Britain
On the Use of Third Electrodes in a Secondary
Zinc-Air Battery
JAMES .I, P0STULA-f and RAYMOND THACKER-f
Introduction
At the present time zinc-air batteries are being devel-
oped for possible application in electric vehicles. As a
primary system [l] the zinc-air battery holds considerable
promise since high performance fuel cell air cathodes can
be used and the problems encountered in the charging
of zinc anodes are avoided. However, as a secondary
system, which would be more convenient, the zinc-air
battery is seriously limited. This is because zinc cannot
be electrodeposited in a suitable form at high rates, and
because fuel cell air cathodes would deteriorate during
the charge cycle. For example, any carbon contained in
the cathode would be irreversibly oxidized. In addition a
noble metal catalyst, such as platinum [2], would form
an oxide which would dissolve in the electrolyte. Mass
transfer would occur and platinum would plate out on the
anode where it would lead to a high rate of self-discharge
of the zinc that was deposited.
A possible solution to the “cathode problem” in a
secondary zinc-air battery consists of the use of a third
electrode for charging the zinc anodes [3]. This procedure
was investigated in the work reported here.
Experimental
The zinc anodes were made in the following manner.
A dry mixture containing zinc powder (95.2 per cent),
carboxymethyl cellulose (1 per cent), polyvinyl alcohol
(2.4 per cent), and mercuric oxide (1.4 per cent) was
prepared. A paste was then made by stirring 105-110 g
of the dry mixture with 30 ml of distilled water. The
paste was applied to a copper screen (20 mesh) using a
neoprene mold. A piece of paper towel was placed in
the mold first followed by half of the paste. The copper
screen was then positioned in the mold whereupon the
remainder of the paste was added. The zinc was covered
with another piece of paper towel after which the lid of
the mold was put into place. The composite was pressed
with a IS-lb weight to distribute the paste uniformly on
both sides of the screen. Finally the electrodes were air
dried at 125°C for 15-20min.
The cathodes consisted of a palladium-silver catalyst
supported on a carbon-Teflon layer which in turn was
supported on a porous silver-plated nickel matrix. They
were wetproofed with a Teflon film.
- air gaps contained a corrugated fibre glass sheet which
i ElectrochemistryDepartment, Research Laboratories, General
MO tors Corporation, Warren, Michigan, U.S.A.
1970.
(Received 20 May 1969)
The third electrodes consisted of sheets of nickel mesh
(10.3 x 15.2 cm; 4 meshes/cm) which were made by
electroforming. The electrolyte was 30-35 per cent
potassium hydroxide solution. In some of the work the
electrolyte was saturated with zinc oxide. A mercury-
mercuric oxide reference electrode was used whenever
possible, so that single electrode potentials could be
measured.
A single cell, which is shown in Figs. 1 and 2, was
constructed in the following manner. Two pieces of the
cathode material (10.8 x 15.0 cm) were soldered along
the edges to a silver plated copper bar (0.6 x O-3cm)
which had been preformed into a “U” shape. A silver-tin
solder was used. The bar served as a current collector.
The cathode assembly was then washed with distilled
water to remove the acid ,tlux remaining from the solder-
ing operation. After drying, the edges were sealed with a
heavy duty polyvinyl chloride cement in such a manner
that a continuous film was formed between the Teflon
on the rear surfaces of the two cathodes. Three coats of
cement were applied in all, ahowing 24 hr for each coat
to dry. The area of each cathode exposed to the electrolyte
was 150 ems.
The anode-third electrode package was assembled as
follows. First of all, a double hinged current collector
consisting of a silver-plated copper bar 0.2 x 0.9 cm
was soldered to the top of the anode support. The
anode was then wrapped in Permion 300, grade No. 3420,
separator material (RAI Research Corporation) by
folding the separator around the bottom of the anodes
so that both sides were covered. The side edges of the
separator were then heat sealed. A second sheet of
Permion separator material was wrapped around the
anode in the same manner. The two third electrodes
were first laid flat on a sheet of polypropylene nonwoven
fabric, 15 x 22.5 cm, (Webril grade E-1451) with the
current tabs protruding at the ends. The side edges of
the third electrodes were electrically insulated with vinyl
adhesive tape. The third electrodes were then mounted
on either side of the anode in such a manner that they
were insulated on the outside and at the bottom by the
polypropylene sheet. The anode-third electrode package
was then inserted into the cathode assembly.
The details of the battery construction are shown in
Fig. 2. Fifteen cells were mounted parallel to each other
in a Lucite box with a l-cm air gap between them. The
served to keep the cathode-cathode separation within
a single cell uniform when electrolyte was added. On
45
46 JAMES J. POSTULA and RAYMOND THACKER
each side of the battery top were fifteen knife switches
which were ganged together with a Lucite strip so that
they could be operated simultaneously. The knife
switches on one side of the battery top served to connect
the anode of one cell to the cathode of the next cell
throughout the whole battery stack. On the other side
of the battery top, the knife switches served to connect
the anode of one cell to the third electrodes of the next
cell throughout the whole battery stack. In the discharge
mode the anode-cathode switches were closed while
the anode-third electrode switches were opened. In the
charge mode the anode-cathode switches were opened,
while the anode-third electrode switches were closed.
Terminal posts were mounted at the ends of the battery
for connecting a load and a battery charger.
The single cells and battery were tested by determining
the cycle life under a variety of conditions. This was done
by using a simple discharge circuit consisting of a carbon
pile rheostat and an ammeter, and a charging circuit
consisting of a battery charger with a suitable output
voltage and an ammeter. All of the work reported here
was carried out at ambient temperatures.
Results and Discussion
The cycling procedure used to test the single cells
consisted of discharge at currents up to 15 A followed
by charge at currents up to 3 A. A low charging rate was
used to minimize the growth of zinc dendrites. The cells
were cycled until they failed. One mode of failure that
was ultimately expected was the formation of internal
shorts resulting from the penetration of the separator
by zinc dendrites. Such a mode of failure was indicated
by (a) poor charge acceptance; (b) the anode and third
electrode exhibiting the same potential at open circuit
vs. a mercury-mercuric oxide reference electrode; and
(c) potential oscillations in the cell potential-time trace
during charge.
A typical cell potential-time trace exhibiting oscilla-
tions is shown in Fig. 3. These potential oscillations
resulted from the alternate formation and removal of
zinc dendrites. When a zinc dendrite touched the third
electrode, a low resistance electronic path was produced.
The cell potential dropped and the current correspond-
ingly increased. All of the current tended to flow through
Table 1. Performance of single cells
Cell Third electrode-cathode Anode-third electrode
No. separator separator
1 1 layer Dyne1 1 layer Permin 300
2 1 layer Dyne1 1 layer Permion 300
3 1 layer Dyne1 1 layer Permion 300
4 1 layer Polypropylene 1 layer Permion 300
5 1 layer Polypropylene 1 layer Permion 300
5a 1 layer Polypropylene 1 layer Permion 300
6 1 layer Polypropylene 3 layers Permion 300
7 1 layer Polypropylene 3 layers Permion 300
8 1 layer Polypropylene 4 layers of fibrous
sausage casing
the dendrite which as a counsequence heated up, and
finally melted. The electronic path was thus opened, and
the cell potential increased again. The current, of course,
decreased to its original value. The dendrite then started
to grow again, and the whole process was repeated. In
time as more dendrites were formed, the pattern of the
potential oscillations changed, and finally the potential
dropped to a low value from which it never increased
again.
The cycle life and failure mode of several single cells
is shown in Table 1. Six of these cells, No. l-5a. had
one layer of Permion 300 separator between the anode
and the third electrodes. The first three cells, No. 1-3,
had one layer of Dyne1 separator between the third
electrodes and the air cathodes. In the other cells the
Dyne1 was replaced by one layer of polypropylene separa-
tor. This was because the Dyne1 separator was thin and
therefore hard to work with, and in addition it had a
tendency to melt at high rates of discharge. The poly-
propylene separator did not have these disadvantages.
Cells 5 and 5a had the same cathode assembly but differ-
ent anode-third electrode packages. When cell 5 failed,
the anode-third electrode package was removed, and the
cathode assembly was washed with hot water, distilled
water, boric acid solution for 16 hr, and finally with
distilled water. After drying, a new anode-third electrode
assembly was inserted.
Of the first six cells tested, No. 5 survived for 74
cycles, No. 5a for 4+ cycles, and No. 4 for 34 cycles.
That cell No. 5a had an appreciable cycle life was indeed
surprising since it was charged at currents up to 6 A.
The cycle life of the first three cells was very low. With
the exception of the first cell, the failure mode was a short
between the anode and the third electrode. These shorts,
which were barely visible to the naked eye, occurred
through the separator, at the edges where the separator
was heat sealed, or over the top of the separator. The
poor reproducibility of the performance of the first six
cells is hard to explain. It is possible that differences in
the mode of construction, or a non-uniformity of the
Permion separator could be responsible. Cells No. 6 and
7 each had three layers of Permion separator between the
anode and third electrodes. The cycle life of these cells,
however, was disappointingly and unaccountably low.
One of these cells failed because of a short between
the cathode and the third electrode, and the other failed
Number of
cycles Mode of failure
Cell dried out on standing after
first discharge
Anode shorted to third electrode
Anode shorted to third electrode
Anode shorted to third electrode
Anode shorted to third electrode
Anode shorted to third electrode
Third electrode shorted to cathode
Anode shorted to third electrode
Anode shorted to third electrode
Fig. 1. Secondary zinc-air cell.

[facing p. 46
Fig. 2. Cutaway view of secondary zinc-air cell.
 3 IN.
4 *

Fig. 3. Fifteen-cell secondary zinc-air battery in discharge mode.

 On the Use of Third Electrodes in a Secondary Zinc-Air
because of a short between the anode and the third
electrode. The best performance was exhibited by cell
No. 8 which contained four layers of fibrous sausage
casing between the anode and the third electrodes. This
cell survived for 16; cycles after which it failed because
of a short between the anode and the third electrodes.
The Permion 300 separator showed no evidence of
degradation as a result of coming into contact with
oxygen that was evolved at the third electrode during
charge. The fibrous sausage casing, however, particularly
the outer layers, was embrittled as a result of contact
with oxygen.
The open-circuit cell potentials were 1.3-l-4 V. On
discharge the potential dropped to 1*0-1-l V at a current
drain of 10 A, while on charge the potential (anode vs.
third electrode) was 2.2-2-5 V at a current of 2 A. After
several cycles the cell potential on discharge dropped to
0.7-09 V. This was due to an increase in the polarization
of the zinc electrode.
The capacity of the zinc anodes initially was 84 A-hr.
In the first cycle about half of this capacity was used up.
As the cells were cycled the useful capacity that could
be obtained was observed to decrease. In cell No. 5 the
capacity was 30 A-hr in the fourth cycle, while it was
only 18 A-hr in the seventh cycle. A cutoff voltage of
O-7 V was used. This decrease in capacity with cycling
is not a new phenomenon as it has often been observed
in the development of silver-zinc batteries [4]. A possible
explanation is shape change, i.e. a redistribution of
zinc which occurs on the anode support with repeated
cycling [5].
An attempt was made to determine the limiting current
of the zinc anodes on discharge. For this purpose the
cathode assembly from cell No. 5a was used after washing
in the manner described previously. A zincanode wrapped
in polypropylene separator was inserted into this cathode
assembly. The third electrodes were omitted for this
investigation. The cell was discharged initially at 25 A
for 30 min. During this period the cell potential was
1.0 V. The current was then increased to 45 A for 30 min.
The cell potential fell immediately to 0.93 V and then
slowly dropped to 0.87 V. At this point the current was
increased again. At 50 A the cell potential was 0.85 V,
I I
Fig. 4. Potential oscillations
Battery 47
and at 60 A it was O-7 V. In all about 40 A-hr or half of
the capacity of the anode, was utilized.
Often in the operation of fuel cell air cathodes in a
potassium hydroxide electrolyte, three problems are
encountered [6]. They are (a) “weeping”, i.e. the passage
of electrolyte through the electrode structure to form
droplets on the gas phase side of the electrode; (b)
flooding of the active layer; and (c) carbonate precipation
within the pores of the electrode. All of these effects can
lead to a decrease in the performance of an air electrode.
In the work reported here, none of these problems were
encountered.
During the operation of the cells the electrolyte level
was observed to decrease. While some of this loss was
caused by evaporation and sputtering from the top of the
cells, it is believed that a substantial loss occurred by
transpiration of water vapor through the air cathodes.
The loss of water from the cells led to a decrease in the
cell potential during discharge, and an increase in the
potential (anode vs. third electrode) during charge.
Another possible consequence of the loss of water from
the cells during charge was the promotion of shape change
at the anodes leading to a premature loss of capacity.
The power curve for the fifteen-cell battery is shown in
Fig. 5. The peak power was 366 W at a current drain of
38 A. The corresponding battery voltage was 9.6 V. At
open circuit the battery voltage was 21.1 V. Up to the
maximum power the battery voltage rapidly reached a
steady value for a particular current. However, beyond
the maximum power, steady voltage readings could not
be obtained. Because of this the descending limb of the
power curve is probably in error.
The weight of the battery was 20 lb dry, and 24 lb wet.
The fifteen cells weighed 12$ lb, while the case and
some of the switching hardware weighed 7s lb. The
volume of the battery was 0.4 ft3. Combining these data
with the power data shown in Fig. 5 and the capacity
data reported above gives the energy and peak power
densities shown in Table 2. For comparison, the energy
and peak power densities of a mechanically rechargeable
20~11 zinc-air battery [7] and of the lead-acid battery
[8] are also included in Table 2. It will be seen that the
peak power density of the electrically rechargeable
TIME (MIN) -
during charge.
 JAMES J. POSTULA and RAYMOND THACKER

zinc-air battery, both on a volume and a weight basis,

0 suffer from this limitation since a more permeable poly-
0 10
Fig. 5. Power curve for fifteen-cell secondary zinc-air battery.
Table 2. The energy and power densities of several batteries
Electrically rechargeable
15-cell zine-air batterv
Mechanically rechargeable 50
20-cell zinc-air battery [7]
Lead-acid battery [8]
is less than that of the mechanically rechargeable zinc-
air battery, which in turn is less than that of the lead-
acid battery. The energy density of the electrically
rechargeable zinc-air battery, both on a volume and a
weight basis, is less than that of the mechanically re-
chargeable zinc-air battery. It is interesting to note that
the energy density of the lead-acid battery on a weight
basis is less than that of the two zinc-air batteries. On
a volume basis, though, the energy density of the lead-
acid battery is between those of the two zinc-air batteries.
There are two reasons for the poorer performance of
the electrically rechargeable zine-air battery. First.
the weight and volume of the battery were not optimized.
It is believed that they both could be reduced by a
factor of two. Second, the battery could not be discharged
at a high rate because of a transport limitation caused by
the low permeability of the Permion separator to OH-
ions. The mechanically rechargeable battery did not
20 30 40 50 60
propylene cloth separator was used. Also the third
CURRENT (A) electrodes in the electrically rechargeable battery impeded
the transport of OH- ions across the cells.
The cycle life data of the fifteen-cell battery are shown
in Table 3. For the discharge part of the cycle, the current
that was drawn from the battery for a particular time
Energy density Peak power density
and the corresponding voltages are shown. Also shown
Y-7
W-hr/lb W-hr/ft3 W/lb
w/f; 3 is the total capacity. Similar data are shown for the charge
part of the cycle. It should be noted that the battery
23 1350 15 890
voltage during discharge is that of the anode vs. the
4170 58 4,830 cathode, and the battery voltage during charge is that of
the anode vs. the third electrodes. In all, the batter]
13 2000 78 12,000
survived S+ cycles.

Table 3. Performance of fifteen-cell zinc-air battery

Discharge Charge
~_------h__.-, ---_-7
----

Cycle Time Current Voltage Capacity Time Current Voltage Capacity

No. (hr) (A) (V) (A-hr) (hr) (A) (V) (A-hr)

1 22.5 0.8 18.5 19 2 30.9-36.5 38

0.25 O-60 X.1 peak 36.7
0.45 O-50 9.1 peak
2 1.45 15 14.3-13.4 64 0.5 30-37.3
0.75 10 13.8-13.7 5 2 35.5-35.6 42
0.67 12 12.4-11.6 41.5
0.75 6 12.0-11.6
3 1 5 16.6-15.1 15 1.2 31.0-33.9 18
1 10 14.8-14.2 15
4 0.75 10 10~0-9.0 17 1.3 34.0-39.3 22
0.5 5 13.2-9.0 10
5 0.75 10 10.1-6.7 18.25 1.2 32.0-39.3 21.9
0.5 5 10.3-7.2 10
Three cells that had reversed were replaced
6 1 10 11.8-9.6 16.75 1.3 33.5-36.9 21.8
1 5 11.7-8.0 15
7 1 10 12.6-7.0 10 6 1.9 34.2-35.6 Il.4
Three cells that had reversed were replaced
8 0.65 10 12.0-6.0 6.5 6 3.0 33.0-37.7 18
0.67 15 13.2-6.0
9 0.5 5 15.2-10.2
0.25 10 10.0-3.7 15.0
During the eighth and ninth cycles four cells reversed
 On the Use of Third Electrodes in a Secondary Zinc-Air Battery 49

During the cycling of the battery, the capacity was
observed to decrease from cycle to cycle. Initially, it was
about 40 A-hr, but was only 10 A-hr in the fifth cycle.
Similar behavior was observed in the single cell studies.
In the tith cycle, on discharge, three cells reversed. After
charging the battery these cells were replaced. Again the
battery was cycled. In the sixth discharge the capacity
obtained was 15 A-hr, and in the seventh cycle it was
10 A-hr. During discharge in the seventh cycle three
more cells reversed. These cells were replaced after
charging. In the eighth and ninth cycles four more cells
reversed on discharge.
The cell reversal was completely unexpected. It occurr-
ed, apparently, because of a nonuniform decrease in the
capacity of the anodes from cycle to cycle. When an
anode lost its useful capacity the potential increased to
rhe point where oxygen was evolved. At this point the
anode was positive to the air cathode which was still
functioning normally. This is illustrated by the following
data:
Potential (V)
r-----7
Cell Cathode
Potential VS.
Zn Reference Zn Reference
WI
C ell behaving normally 1.06 1.25
(Discharge at 10 A)
Same cell after reversal -1.10 1.17
! Discharge at 10 A)
There was no evidence during the cycling of the battery
of shorts between the anode and the third electrodes,
or between the third electrodes and the cathode. Battery
failure, apparently, was caused solely by the loss in
capacity of the zinc anodes.
Acknowledgements-The authors wish to thank Marion E. Wheatley
and Richard L. Adams of the Electrochemistry Department,
Research Laboratories, General Motors Corporation, for their
assistance in this work.
References
[l] R. R. Witherspoon, Paper presented at S.A.E. Congress,
Detroit, Mich., U.S.A., Paper No. 690204, 16 January (1969).
121 R. Thacker. Electrochimica Acta. 14. 433 (19691.
[3j R. S. Gum&io, U.S. Patent 3,21$,48ti, 23 November (1965).
Anode [4] T. P. Dirkse, ASTIA Rpt. AD 804559, January (1967).
VS. [5] G. A. DaIin, Paper presented at the Fall Meeting of the Electro-
chemical Society, October, 1968, Extended Abstracts of Battery
Division, 13, Abst. No. 299.
0.20 [6] K. R. Williams (Editor) .4n Introduction to Fuel Cells, Elsevier,
New York (1966).
2.25 [7] R. R. Witherspoon, private communication.
[S] J. J. Lander, private communication.