POWER

Power Battery Company, Inc.

Power for a World of Applications

TECHNICAL PUBLICATION

VRLA BATTERIES: REFLECTIONS ON AND REALITIES OF

PREVIOUS INTELEC CONFERENCES

By: F.J.Vaccaro, J. Rhoades, B. Malley, K. Marrion, B. Le

PRESENTED AT INTELEC,1999 - COPENHAGEN, DENMARK

 VRLA BATTERIES: REFLECTIONS ON AND REALITIES OF
PREVIOUS INTELEC CONFERENCES
F.J.Vaccaro, J. Rhoades, B. Malley, K. Marion, B. Le

Power Battery Company Co., Inc.

25 McLean Blvd.
Paterson, NJ 07514, USA

ABSTRACT curve to estimate capacity. One such variable we

Previous Intelec conferences have produced
papers that have addressed the unique properties
of this relatively new cell design. In this work we
have attempted to review those articles critical to
successful battery design, operation, testing, and
performance.
This paper’s analysis starts with a topic of the late
eighties, i.e. the internal resistance-capacity
relationshipi, and ends with the present day
interest in noble metal catalystsii. The review
spans approximately ten years and includes the
following subject matters:
• The relationship of internal resistance to
capacity
• Temperature effects on plate polarizations
• Accelerated aging life testing
• Premature end of life
• Survey of Field Failure Modes
• The catalyst
INTERNAL RESISTANCE
The intense interest in internal resistance(R) is in
the ability of this measurement to determine the
capacity(C) of the VRLA cell. To accomplish this
address is the R-C relationship of a multi-cell
battery having only two post terminals
available for measurements.
Intelec conference papers and VRLA standards
suggest a capacity test discharge be performed
when the cell internal resistance increases by
20%iii. For the multi-cell module, with two
terminal posts, this limit may need to be
dramatically decreased to detect one errant cell
in electrical series with cells of good internal
resistance. But, at a lower limit the sensitivity
of the R-C curve may not be adequate to make
this detection.
The change in resistance of the multi-cell module
is obviously dependent on how uniform the aging
is from cell to cell.
FIGURE 1
AG I NG CELL
D I S C H A R G E C U R V ES
2.1
5
CE LL V OLT AG E

goal it is required that: 2.0

5
• The loss in cell capacity be due to only NEW
increased cell internal resistance 1.9 1/4L
5
• Discharge curves (voltage versus time) must I2F/ 3E L
vary with age according to a predictable 1.8 ILFI F
EE
5 END
mathematical relationship
1.7
For example, given a normal-linear relationship
5
between (R) and (C), the family of discharge
1.6
curves associated with the aging cell process must 5 0 1 2 3 4 5
be parallel (figure 1). The accuracy at which (R)
estimates capacity is then dependent on the degree T I M E
of parallelism between the discharge curves.
Obviously the R-C relationship can be other than
normal-linear, e.g. log-linear. The R-C relationship can be employed to
determine the uniformity of cell aging in a multi-
There are a number of variables that can impact cell module. For example, when the resistance of
on the ability and limits of the R-C one cell in six increases by 20% the total battery
resistance rise is only 3.3%, for this worst case
scenario. The sensitivity (slope) of the R-C curve
may not be adequate enough to relate this small FIGURE 2
resistance change to an accurate estimate of
capacity. The best case scenario is when all cells I M P E D A N C E vs 10 MINUTE RATE
0
in the module age at a similar rate. In this case, AGED AT 25 C
the resistance of each cell in a 12-volt module 12
0
increases by 20% as does the total module
11
resistance enabling a more accurate estimate of the 0
capacity. Obviously for the multi-cell module, cell 10

C A P A C I T,Y%
0
aging rate differences can determine the success or
90 BAT 2
failure of resistance to determine capacity.
BAT3
80
3
The mathematics employed to resolve the R-C 70
relationships for the various multi-cell module
60
arrangements are available. Although this analysis
is important, it must be reserved for another paper. 50
The following figures show the results of testing 0 20 40 60 80

12-volt modules that exhibit the worst and best IMPEDANCE CH ANGE, %
case variations of cell aging. The batteries used in
this study are for UPS applications and high rates
of discharge. These battery products have a
positive plate thickness of 0.060 to 0.100 inches.
FIGURE 3
The plot in figure 2 shows the R-C relationship
for two 12V modules aged at 250C, both failed as 12V BATTERY MODULE AGING AT 25 C
0
A
a result of positive plate growth. The discharge
rate was 10-minutes, a rate common to the UPS 110

industry. It is seen that the changes in impedance 100

C AP AC I T Y , %

of both modules closely tracked the capacity loss.

90
Of more significance is when the capacity was
at 80%, the impedance had changed 20 to 80
30%. This result indicates that the 20%
70
impedance limit would detect capacity failure
and that the individual cells have aged 60
uniformly.
50

Figure 3 also shows R-C data for a 12-volt 40

battery aged at 250C, but where the capacity was 0 10 20 30 40

followed by 5-hour rate of discharge. In this IM P E D AN C E , %

module one cell failed capacity prematurely
because of an internal short. At 80% capacity the
battery impedance increased by only 2% which is
an indication of non-uniform cell aging and the
presence of one bad cell. An impedance
measurement may not have detected the
battery 80% capacity due to the small
resistance rise. The capacity ultimately decreased
to 50% where the impedance change was 30%.
This increase in the impedance change normally
results in not detecting the battery module
capacity fall off until it is 50 to 60%. Without
performing a discharge we consider this ability
to be somewhat extraordinary. The ability to
detect failing cells using higher rates of discharge
can obviously improve the technique.

The ability to identify one failing cell in a 12-
volt battery module using the 20% impedance
limit is not likely. Our data indicates that it is quite
possible to make this determination when the
impedance limit is set at 30 to 40%.
To quantify the R-C detection ability, in figure 4
is a calibration curve whose sensitivity is 1-%
capacity/1-% impedance. It shows that when three
cells of six fail the battery capacity drops only
10% while the resistance increase is only 10%.
For the 2-volt module the one cell capacity failure
can obviously be more easily detected and signal
possible future cell failures. FIGURE 5

FIGURE 4
PL AT ES & R E FE R E NC E ELECTRODE
IMPEDANCE vs. CAPAC I T Y OPEN C I R C U I T V S . -TEMPERATURE

2 V a n d 12 V M O D U L E S 35
105
30

VO LT AG E , m V
100
ONE 25
CELL
IN
0.6 7 m OC ( - ) PLATE
SIX 20
95 V/ (+)PL AT E
THREE 15 E
C AP AC I T Y ,%

Hg REF.
CELLS
IN O
SIX
0..40mV C
90 10 /
5
85 0 . 1 7 m V /OC
0
0 20 40 60 80
80 ONE
CELL
OR SIX T EM PER AT UR E , 0 C
CELLS
75
0 5 10 15 20

I M P E D AN C E , %
EFFECTS OF TEMPERATURE ON PLATE
POLARIZATIONS
The paper we presented in Intelec’97iv reported
on the effect of temperature on plate polarization.
The implication of this work was that both the
positive and negative plate could be at or near
their open circuit voltages, due differences in
environment temperatures.
Because of the impact of this finding on battery
state of charge a closer examination of these
effects where made. In this effort the methods of
measurement and a cell design used where
improved. This work included determining the
effect of temperature on the open circuit voltage
of the plates and reference electrode.
Obviously these measurements have been made
before but we believed it was necessary to repeat
them with our electrodes. The temperature
coefficient of the plates and reference electrode
employed were made with a Mercury-Mercurous
Sulfate electrode in 1.300 specific gravity acid at
250C(acid-silica bridge). Figure 5 and the table
that follows it shows our results. The coefficient is
a positive polarity if the hot electrode is the
positive terminal.
Electrode dVo.c./dT, mV/0C
Compare to the
Hg/Hg2SO4 @ 250C
Hg/Hg2SO4 Ref. 0.40
Neg. Plate 0.17
Pos. Plate 0.67
These coefficients were used to determine plate
polarizations for flooded cells controlled at 2.27
volts and various cell temperatures. The results of
the Intelec’97 effort found the positive plate
polarization decreased by 1 to 1.6 mV for each 0C
rise with the negative polarization increasing by a
smaller value.
The coefficients of the flooded cell are of the
same polarity as that of the starved VRLA, but
smaller. The difference in magnitude of the plate
polarization change with temperature, compared
to the 1997 work, appears to be related to the
degree of electrolyte flooding. For the negative
plate this can be the result of decreases in the
oxygen diffusion rate with flooding. The
temperature effect on the positive plate is perhaps
more difficult to explain. It is possible, that the
direct oxidation (solid state corrosion) of lead in
the VRLA cell is dominant over the solution
reaction. Although the coefficients are
somewhat smaller in this work, the concern
that temperature variations can cause
erroneous plate polarizations is still valid. In
figure 6 is plotted the plate polarization variation
with temperature
FIGURE 6
P L AT E P O L AR I Z AT IO N - T E M P E R AT U R E
FLOODED VRLA CELL @ 2.27 V/Cell
140
0.92 m VO/C
120
P O L AR I Z AT IO N , m V
100
80
60
40
O
0.31 mV/C
20
0
20 30 40
T E M P E R AT U R E , C
ACCELERATED LIFE TESTING AND
PREMATURE CELL FAILURE
When performing accelerated life tests on a lead-
acid cell to estimate life at 250C the plate
polarization should be as they are at 250C, i.e.
independent of the test temperature. Normally,
when performing life tests on flooded cells, the
positive plate polarization is held constant by
either adjusting the cell voltage with temperature
changes or more precisely by employing a
potentiostat to control the positive plate voltage .
The negative plate voltage of the flooded cell is
allowed to vary uncontrolled for the following
reasons:
• There is substantial polarization on this plate
so that small changes in its voltage does not
alter its behavior or the test results
• Water loss is not of concern in the flooded
condition and its loss would not effect the
negative plate oxygen recombination rate. In
addition, if needed, water could be replaced
in these tests
Most flooded cell life tests primarily evaluate
positive plate growth and life and in these tests the
positive plate polarization is constant at the
various temperatures.
In accelerated life tests of the VRLA cell only
the cell voltage is held constant or worse, it is
lowered with increases in test temperature. Based
on the above temperature-polarization coefficients
the positive plate voltage should be increased with
rising temperature. For negative plate consistency
the cell voltage should be decreased as the
temperature increases. It is obvious both
conditions can not be satisfied for the in the same
test. It is perhaps apparent that testing for negative
plate life requires a different set of test procedures
than that of the positive plate.
Under the present test conditions of
uncontrolled plate polarizations the prediction
of life at room temperature and the failure
( - ) PLATE mechanism associated with that life will most
( + ) PLATE
likely be in error because of the following:
• Positive plate life at 250C when extrapolated
from Arhennius plots requires that the
polarization of this plate be the same at all
temperatures. Positive plate growth
50 60
appears is reported to be polarization
0 dependentv!
• The negative plate polarization increases with
temperature with increases in the water loss
rate. The experimental data plotted in figure 7
shows that the extrapolated weight loss (for
250C) from the high temperature life tests is
approximately ten times greater than the
actual weight loss at room temperature.
• When the VRLA cell is at 250C the negative
plate may be on open circuit and self-
discharging. By increasing the cell test
temperature the negative plate polarization
rises and the plate is charged. For this reason
tests at elevated temperatures will not
detect those negative plate failures that
will occur at 250C has was shown in our
negative strap fracture investigationvi.
These scenarios, we believe, rightly predict
that the wrong failure mechanism can be
identified and estimates of cell life times at
250C can be grossly in error.
The Power Battery negative plate strap fracture
study did not detect this straps fracture at elevated
temperatures although they were the primary
failure mode at normal room temperatures. As
shown above this ambiguity was due to the
negative strap polarizing (charging) at the elevated
test temperatures when at normal room
temperatures it was a discharge voltages.
At low temperatures self-discharge of the
negative plate is most likely to take place and
cause cell failure. Actual battery surveys report
vii
on premature cell failures and make the point
that the batteries were in temperature controlled
battery rooms. As shown above negative plate
polarization decreases with temperature. It is than
not surprising that battery failures related to this
plate predominately occur in sites where the
ambient temperature is lower or artificially
cooled.
FIGURE 7
W E I G H T L O S S K I N E T I C S
100
R A T E OF L O S S
10
Ae = 1 9 cal/Mole
1 UL
2. 3 3. 3. 3.
9 1 2 3
Ae = 34 cal/M ol e
0.1
0.01
-
1/K x 10 3
At low temperatures self-discharge of the
negative plate is most likely to take place and
cause cell failure. Actual battery surveys report
viii
on premature cell failures and make the point
that the batteries were in temperature controlled
battery rooms. As shown above negative plate
polarization decreases with temperature. It is than
not surprising that battery failures related to this
plate predominately occur in sites where the
ambient temperature is lower or artificially
cooled.
The premature failure of the negative plate
structure can result from a loss of only several
millivolts of this plate’s polarization. A downward
shift of this magnitude is easily attained by small
decreases in ambient temperature. The logical
conclusion is that the variations in cell life
reported on by the various surveys is, in part, due
to minor differences in ambient temperature. It is
quite probable that the negative plate failures
we are experiencing today are the result of our
inability to detect them in accelerated life tests.
This inability put off addressing negative plate
problems until the present time where the
population of product is disturbingly high.
Again, the above temperature effects on cell life
indicate that accelerating cell life by employing
elevated temperatures is at best questionable.
Based on this concern we strongly recommend a
re-evaluation of the test procedures presently
used to determine cell life and failure
mechanisms.
APPROACHES TO PREMATURE FAILURES
Over approximately the past 5 years there have
been a number of surveys to determine the life of
VRLA batteries. In some instances the findings
were contradictory, but most would agree that
premature VRLA cell failure is a reality. These
works report that VRLA life can be as short as two
CALC
years and for the majority of products less than 10
EXP years.
3.
4
Rigorous experimentation and eventual
dissection of cells can only resolve the cause and
cure of premature cell failure. Power Battery
regularly employs personnel who are also
consultants to the general battery industry. Under
strict operating guide lines the employee-consultant
can interact at levels not available to others and
obtain an overview of battery failure modes.
The following is a list of the causes for
premature VRLA cell failure. This work was
accomplished over a ten-year period and includes
battery product that is five years of age and less. A
number of manufacturer’s products are covered and
their applications were diverse. The following
does not assume to identify all premature failures,
but it should provide a window of information that
is not open to all.
Failure Mode Time to Occur
Years
Neg. Plate Lug Corrosion ~2
Neg. Strap Fracture 1 to 5
Inter-cell Connector 3 to 5
Excessive Plate Growth 1 to 5
Neg. Plate Capacity Loss 2 to 4
Cabinet Explosions NA
1 Cells in a box with heating coils
2. High temp. Alarm, Minor Status
3. No room for Plate Growth
4. Shorts to Ground
Thermal Runaway NA
(+)Plate Short to Neg. Strap 1 to 3
Orientation(leaks & fires) NA
High Ambient Temp. NA
Insufficient plate Growth Space NA
Those deficiencies in italics represent failures
somewhat unique to the VRLA cell design.
From the above it is apparent that many instances
of short cell life are the result of negative plate and
structure failures. Negative plate strap and lug
corrosion failures have been well documented in
the literature, but surprisingly still occur. Less is
known about the cause of negative plate capacity
failure. Resolution of this failure will require a
rigorous experimental effort. This problem seems
to be associated with insufficient electrolyte
wetting of all surfaces. Wetting which insures total
cathodic protection to the negative structure.
Maintaining electrolyte coverage of all
functional surfaces of the VRLA throughout its
life we believe is the most important challenge
confronting the battery designer.
The above cases of thermal runaway were caused
by either, a high ambient temperature, shorted cells
(positive to negative strap), not responding to the
high temperature alarm and shorts to ground.
Thermal runaway is a problem with all starved
electrolyte gas recombining cells. Its prevention
in gas recombining NiCd and NiMH cells is
accomplished by careful control of the charging
circuit. This approach should be the same for
the VRLA cell. Intelligence charger circuits and
schemes are available to control current, voltage
or charge the battery off lineix. The challenge to
the circuit designer is to make this technology
available to the battery user.
Heat generation is a by-product of gas
recombination, one can improve its dissipation but
can not alter the basic laws of thermodynamics.
Limiting the charge current is a good way to
prevent thermal excursions and should be
universally apply. The bulk of the remaining
problems are no different than that of the flooded
lead-acid battery. They are associated with the
following:
• Inadequate or no maintenance
• A high rate positive grid corrosion rate and a
poor grid design
• Insufficient room for positive plate growth
• Insufficient hydrogen ventilation
• High ambient temperatures
• Electrolyte shorts to ground
• Poor workmanship
Vented cells are normally in central office where
they are better maintained and typically replaced
when the capacity falls to 80%. Secondary failures
such as positive plate shorts to the negative strap
are unheard of. But, capacity discharge costs in the
highly distributed battery arrays of today, can be
prohibitive. Cell internal resistance
measurements to replace cell discharges can
help in resolving this problem. As an alternate
approach which has found some acceptance is to
replace cells based solely on manufacturer
estimates of cell life times. Since we question the
validity of accelerated life testing, we can not
recommend this approach
While there are inherent problems with the VRLA
battery the solution to them, we believe, is in
confronting the following problem areas:
• Negative plate failures
• Charging equipment and routines
• Battery room environmental control
• Stay with design concepts that are right for
flooded cells
• Education
CATALYST
There are at least two reasons for placing a
catalyst into a VRLA cell. The first is to prevent
negative plate self-discharge for the cell that is at
or near 100% oxygen recombination efficiency.
The second, to slow down the rate of loss of
water, especially for the long life cell (=>20
years).
To understand both of these needs one must
evaluate the Tafel relationship of the VRLA
battery. Figure 8 shows several negative plate
Tafel conditions,
1. All the oxygen is recombined on the negative
plate(100%)
2. All the oxygen is recombined at the catalyst
surface or 0% at the negative plate.
3. Recombination at both surfaces.
As long as the catalyst material does not enter the
electrolyte and deposit on to the negative plate the
only difference between the curves is the oxygen
reduction rate at the negative plate.
 FIGURE 8
NEGATIVE PLATE TAFEL-CATALYST
AFFECTONO2 RECOMBINATION
1000
100% ALLONNEGATIVEPLATE
Data from previous publications shows that for a
particular cell design inserting a catalyst decreased
the float current by approximately a third and
increased the negative plate polarization by 35
mV. One must keep in mind that any polarization
gained by the negative plate is lost from the
positive plate.

From the above current decrease the positive

CURRENT, m A

100 <100% ONCATALYST&PLATE plate polarization (∆η+) decrease can be verified

by calculation as follows:
(∆η+) = m * log(i2/i1) 3.
0% ALLONCATALYST)
10 Given:
1. m = Tafel slope = 70 mV/decade current
2. (i2/i1) = 1/3, from the above references

1
0 50 100 150
POLARIZATION, m V
This point is important because at 100% OR the
negative plate is charged (polarized,η-) only by
the corrosion current (Ic) which must be greater
than the negative plate Tafel intercept
(I-), as seen in this equation:
η- = (Ic) - (I-) 1.
If this is not the case the negative plate self-
discharges. Self-discharge can be overcome by
placing a noble metal catalyst in the cell
environment to decrease the OR efficiency to
some value below 100%(Id) so that the negative
plate charge current takes on the an additional
component as shown below:
η- = (Ic + Id) + (I-) 2.
The curve in figure 8 shows this effect on the
negative plate polarization. The increased
hydrogen evolved is recombined at the catalyst so
that the water loss rate does not change and is still
dependent on the currents of lead corrosion and
impurity oxidation.
To this point we have perhaps reiterated the
writings of Jones and Berndtx. Our aim in this
section is to determine the possible risks, if any,
when employing a catalyst. The following are
some:
• Positive plate self-discharge
• Catalyst contamination
Positive Plate Discharge
Then:
200 250 (∆η+) = 70 * log(1/3) = -33mV 4.
The calculated decrease in this plate’s
polarization closely parallels that experimentally
reported on in the above reference works. The
problem for a positive plate at 100mV
polarization and 250C is decreases in its voltage.
When the catalyst is employed in this cell the η+ is
67 mV Using the above current change). If the cell
temperature is increase to 500C the polarization is
at 42 mV, not a stable charge polarization.
By employing a voltage/temperature
compensation of minus 3 mV/0C, to prevent
thermal runaway, the positive plate would be in
danger of self-discharging. One must also be
aware that the Tafel lines of the negative plate can
vary within a cell type, and between the various
cell types and manufacturers. Because the
polarization can vary, that employed here is only
an example of the cell employed in the previous
studies. A catalyst that limits the negative plate
polarization to 10 mV or less can alleviate
concerns of positive plate discharge.
Catalyst Contamination
If the catalyst metal dissolves into the electrolyte
solution and deposits onto the negative plate this
will increase the value of the negative plate
Tafel intercept (I-). As seen in the above
equation increasing this Tafel parameter raises the
self-discharge rate of the negative plate and the
ability to polarize it. The affect is the same as that
of the familiar antimony poisoning experience and
at its worst can result in negative plate self-
discharge and water loss in the remaining cells.
 vi
As to the quantitative effect of catalyst F.J. Vaccaro, J. Rhoades, et al, INTELEC 95,
poisoning the negative plate. In experiments on The Hague, Netherlands, Page 70.
vii
the addition of platinum to a flooded cell we W. Cantor, et al, INTELEC 98, San Francisco,
found that the negative intercept doubled with CA., Page 369.
viii
the addition of 5.5x10-2 ppb Pt/Ah cell capacity. W. Cantor, et al, INTELEC 98, San Francisco,
For a 125 Ah cell this requires one to detect CA., Page 369.
ix
approximately 7ppb platinum. Because R. Jones, J. McAndrews, F. Vaccaro, INTELEC
platinum or palladium almost immediately plates 98, San Francisco CA., Page 526.
x
on to the negative plate its detection in electrolyte D. Berndt, INTELEDC 98, San Francisico, CA.,
is not attempted. Of course one can use a direct Page443.
measure of the Tafel intercept to determine the
catalyst contamination.

Another important affect of catalyst use is to

prevent self-discharge of the negative plate when
the VRLA cell is at relatively low ambient
temperatures. Increasing the negative plate
polarization at 250C allows operation at lower
temperatures before the onset of negative plate
self-discharge, because of the temperature
coefficient of the negative plate polarization.
From our work and that of Milner and Berndt the
negative plate polarization decreases by
approximately a one mV/0C. For the VRLA cell
with a negative plate 35 mV polarized at 250C this
translates to operation of the cell to -100C before
negative plate self-discharge. Without the catalyst
this plate’s self-discharge begins several degrees
below 250C. The catalyst use is the only
practical way of resolving this problem
associated with cold temperature operation.

The use of a catalyst is recommend, but only

after careful evaluation of the above points. Of
course when batteries are failing in the field the
risks associated with catalyst use may be
secondary.

ACKNOWLEGDEMENTS
The authors take this opportunity to thank the
Power Battery technical staff for their support and
“Big” Bill for sponsoring this work and its
presentation.
i
F. J. Vaccaro and P. Casson, INTELEC 87,
Stockholm, Sweden, Page128.
ii
W. Jones, et al, INTELEC 98, San Francisco,
CA, Page 461.
iii
VRLA Batteries, ANSI T1-330-1997.
iv
F.J. Vaccaro, J. Rhoades, et al, INTELEC 97,
Melbourne, Australia, Page
v
D.O. Feder, et al, INTELEC 90, Orlando , Fl.,
Page 226.