lull11111
ILLLI
THE RELATIVEPERFORMANCE
OF FeS_ ANDCoS_,IN LONG-LIFE
THERMAL-BATTERY
APPL]CATIONS
Ronald A. Guidotti and FrederickW. Reinhardt
Battery DevelopmentDepartment
Sandia National Laboratories
P.O. Box 5800
Albuquerque,NM 87185-0614
ABSTRACT
The relative performance of FeS2 and CoS_ was measured in
single cells based on the LiBr-_Br-LiF,eute_tic electrolyte
over a temperaturerange of 400 to 550vC using both standard
(dry) and flooded anodes . This electrolyteis designed to be
used in long-life(>] h) thermal6batteryapplicationsbecause
it has a lower meltingpoint (313 C) and a higher electrical
conductivitythan the standard LiCI-KCIeutectic (m.p.=352C).
The cells were continuouslypulsed for I ms every 10 ms from
a backgroundcurrentdensity of 190 mA/sq cm to 1,130 mA/sq cm
until end of life.
INTROOUCTION
The Li(Si)/FeS_electrochemicalsystem has been used extensivelyby
Sandia National Labbratoriesin the design of thermallyactivated
("thermal")batteriesfor nuclear-weaponsapplications The batteries_
function only when the electrolytebecomes molten,which occurs at 352°C
for the LiCI-KCI eutectictypicallyused. The first-generationthermal
batteriesutilized the Ca/CaCrO4 couple. This system suffered from a
number of shortcomings,however, such as lot-to-lotvariability in Ca (I)
or the cathode mixes (2,3) and liquid Ca-Li alloy formationwhich tended to
result in shortingwhen not properly controlled. In contrast, the Li-
alloy/FeS_couple has functionedquite well for almost all of its intended
applicatibnsand, from a design perspective,is a much easier system to
engineerthan the Ca/CaCrO4 system.
Thermal batteriesutilize a molten salt immobilizedin a MgO matrix as
the separatorbetween the Li-alloy anode and FeS2 cathode. Normal
operatingtemperaturesfor a thermal battery are generallybetween 550°C
and the melting point of the electrolyte. By use of the LiBr-KBr-LiF
eutectic,a much greater liquid range is possible. This makes this
OI_BUTION OFTHISgOCUMEN_
IS UNLfMITE_
MASTER
electrolyteideally suited for thermal-batteryapplicationswhere lifetimes
in excess of one hour are needed.
The main limitationof the use of FeS2 as the cathode in long-life
thermalbatteries is its thermal stability. The FeS_ thermallydecomRoses
to FeS and elemental sulfur vapor when held at temperaturesabove 550vC for
prolongedperiods. The evolved sulfur can then migrate to the anode and
chemicallyreact with it generatingconsiderableheat which in turn causes
even more thermal decompositionof the FeS2.
In contrast,CoS_ is much more thermallystable than FeS2 and has
excellentkinetics for high-temperaturesecondary-battery applications,as
reportedby researchersat Westinghouse(4,5). These results indicate that
CoSp shouldperform equally as well in primary (thermal)batteries. The
gre_terthermal stabilityof CoS_ relative to FeS2 makes this material
attractivefor long-life(>I h) thermal batteries,as it allows a much
higher initialtemperatureto be sustainedwithout degradationof the
cathode.
This report describesthe relative electrochemicalperformanceof
ti(Si)/FeS_and Li(Si)/CoS_thermal cells under high-rateconditionswhen
formulated_withthe low-me_tingLiBr-KBr-LiFeutectic,which was adapted at
Sandiafor thermal-batteryapplications(6).n Singleocellswere discharged
isothermallyover a temperaturerange of 400v to 550 C using standard (dry
or unflooded)Li(Si) anodes as well as flooded anodes--anodescontaining
free electrolyte.
The cells were subjectedto a continuouscycle of pulsingduring
discharge. The performanceparametersfor characterizationincludedthe
maximumtime for sustainingthe pulse current,the steady-statevoltage
prior to pulsing, the minimum voltage during pulsing,and the polarization
(voltageloss) that occurred during pulsing.
EXPERIMENTALPROCEDURES _: _....
_ !::
_'J
FEB 0 7 193
E_quipment 0 _ T
Li(Si)/FeS_and Li(Si)/CoSpsingle cells (3.2 cm dia.) were discharged
betweenheated _latens in a glo_e box under high purityargon. The
moistureand oxygen contents were maintainedat <10 ppm each. An HP6060B
programmableelectronicload was used to step the backgroundcurrent from
].5 A (- 190 mA/sq cm) to 9 A (-1,130mA/sq cm) during a ] ms pulse. A
duty cycle of 10% was used for pulsing, i.e.,a pulse duration of ] ms
followedby 9 ms at the steady state or backgroundcurrent. The cell was
subjectedto this cycle continuouslyfor five minutesuntil the cathode
capacitywas exhausted . The current throughthe cell and the voltage
across the cell during the a pulse was periodicallymonitoredduring
dischargeusing IIP3458Ahigh-speedDVMs. The experimentwas under the
control of a HP9000 Series 200 computer.
Materialsand Processing
The catholytescontained73.5% FeS2 or CoS_, 25% electrolyte-binder
(EB) mix, and 1.5% LifO as a lithiationagent (In the case of FeS_) or as a
wettingagent (in the"caseof CoS_). (Unlessotherwisestated, a_l
compositionsare in weight percent.) The CoS_ was obtained from C_rac
(Milwaukee,WI) and was used as receivedafter vacuum drying at 80 to
100 C overnight. The -325 mesh FeS_ was obtained from American Mineral and
was purified by leachingwith 1:1 vyv HCf and then concentratedHF. After
treatment,the nominal FeS2 purity was 98.6% or better.
The catholytemixes were fused under argon at 400°C for 16 hours and
were then granulatedand cold pressed into pellets using a graphite paper
backing. The FeS_-basedcathodewas 0.581 g and the CoS2-basedcathode was
0.636 g, so as to'have comparablecapacities.
The low-melting(313°C)LiBr-KBr-LiFeutectic electrolyteused in the
EB was preparedby fusing the appropriateamountsof reagent-gradehalides
57._3% LiBr, 42% KBr, and 0.67% LiF together in a fused-quartzcrucible at
600 C for three hours in a dry room where the relativehumiditywas
maintainedat <3%. The EB used in the catholytewas also used as the
separatorfor the cell. It was prepared by blendingthe electrolytewith
25% Maglite S MgO (Calgon,Pittsburgh,PA) that ha_ been calcinedat 600vC
for four hours. The mixturewas then fused at 400 C for 16 hours in a dry
room. The calcinedEB mix was then granulatedand cold pressed into 0.38-
mm thick separatorpellets or blended into the catholytemix.
The Li(Si) anode materialwas -100+325mesh in size and contained44%
Li by weight. This materialwas cold pressed as is (0.3 g pellet) for
unfloodedanodes and 25% electrolytewas added before pressing for the
floodedanodes (0.41 g pellet).
The single cells consistedof a separatorpellet sandwichedbetween an
anode and cathode pellet. The cells were assembledbetween 0.25 mm
stainlesssteel currentcollectorsand held togetherby staplingbetween
two mica sheets.
RESULTS
Pulse Current
The CoS_ was able to sustain the pulse currentfor a longerperiod of
time relativ_to FeS2. As shown in Figure I for a temperatureof 400 C, a
currentdensity of 1,130mA/sq cm was maintainedfor -156 s for CoS_
comparedto only 78 s for FeSp. (The run time has been correctedfbr the
time requiredfor data acquisltionand data transfer.) At higher
temperatures,this time increasedto 234 s for both cathodes after which
I
l
o
the peak current densitydropped. The drop was greaterfor the FeS2
cathode in all cases.
When flooded anodes w_re used, the pulse currentdensity could be
sustainedfor 234 s at 400 C for the CoS_ cells. For the corresponding
FeS_ cells, the pulse currentdensity cobld be sustainedfor about the same
tim_ as for cells with unfloodedanodes (about 78 s). However, the
relativerate of decrease in the maximum sustainedpulse current density
after 78 s was reduced.
Cell Resistance
The apparent cell resistancewas determinedby dividingthe voltage
drop by the current change during pulsing. It includesbot_ ohmic as well
as concentrationpolarization. The cell resistancesat 400 C are compared
in Figure2 for unfloodedanodes. The resistancedid not increasevery
much with depth of dischargefor the CoS_ cells. In contrast,the cell
resistanceof the FeS_ cells rose dramaticallynear the end of life. The
same trendswere observed at the higher temperatures,except that the
absolutemagnitudeof the resistancewas reduced. The use of flooded
anodes reducedthe differences in resistancebetween the CoS_ and FeS2
cells at all temperaturesas well as the absolutevalues of _ell
resistance.
Minimum Pulse Voltage
The open-circuitvoltagefor the Li(Si)/FeS_cell at 500°C is 1.99 V
which is higher than that of 1.85 V for Li(Si)/CbS_. Thus, at the
beginningof discharge,the CoS_ cells showed a slightly lower minimum
voltageduring pulsing. HoweveR, since the CoS_ cells had a lower
resistance,they outperformedthe FeS_ cells late_ in life. As shown in
Figure 3, this crossoveroccurred at _25 s at 400 C, where polarization
losseswere greatest. As the temperaturewas increased,the time for
crossoverlengthened. Similar trends occurred when floodedanodes were _
used, except that the time for crossoverwas markedly increased. At 500°C,
for example,crossovertook place at 75 s for cells with unfloodedanodes
and ]75 s for cells with flooded anodes.
DISCUSSION
The relative differencesin performanceof CoS_ and FeS_ are related
to differencesin dischargemechanism. In the case_ofFeS2,_thefirst
dischargestep is"
FeS2 + 3/2 Li+ + 3/2 e- ---> ]/2 Li3Fe2S4 [I]
Q
The Li3Fe2S4 is furtherreducedto Li2FeS2 accordingto Eq. 2:
Li3Fe2S4 + Li+ + e- ---> 2 Li2FeS2 [2]
In the case of CoS2, the first discharge step is:
CoS2 + 4/3 e- ---> 1/3 Co3S4 + 2/3 S-2 [3]
This material can be furtherdischargedaccordingto Eq. 4:
Co3S4 + 8/3 e- ---> 1/3 Co8S9 + 4/3 S"2 [4]
The reactionsof Eqns. I and 3 were used to calculatethe capacitiesof the
cathodes in this study. (On a weight basis, this correspondsto 1,206
coulombs/gFeS2 and 1,046 coulombs/gCoS_; on a volume basis, this
correspondsto 6,030 and 4,465'coulombs/_c,respectively.)
Compared to FeS_, the,Li_Fe_S_phase is more resistiveby several
orders of magnitudeSt 400vC,_whllBLi_FeS_ is only four times more
resistive (7). This accounts for the _hap_ of the resistance-discharge
time curves for the FeS_ cells. While comparabledata do not exist for the
dischargephases for th_ Li(alloy)/CoS_system, the resistancedata for the
CoS_ cells (Fig. 3) suggestthat they _re as good or better conductorsthan
the_parentCoS2.
_ote that the dischargephases o_ CoS_ do not depend upon the presence
of Li as do those of FeS_. Thus, Li concentrationgradientsat the
cathode in the case of Fe%_ can severely limit the dischargerate. CoS_
also has negligiblesolubilityin the molten electrolytewhen comparedto
FeS2 •
The presenceof electrolytein the anode improvesthe performanceby
providinga largerreservoir(relativeto dry anodes) for assimilationof
Li generatedduring discharge. It also preventswicking of electrolyte
from the separatorwhich results in a higher separatorresistance. The use
of flooded anodes results in a lower cell resistancebecause it reducesthe
anodic contributionto the overall cell polarization. As a consequence,
the time that the maximumpulse current density can be sustainedis
increased.
For a long-lifethermalbattery, the performanceat the end of life
becomes critical,relativeto short-duration(e.g., 5 min) applications.
Since the temperaturewill be at its lowest at this time, the relative
performanceof FeS_ and CoSp at the lower temperBturesbecomes increasingly
important. The relativepePformancedata at 400 C indicatesthat CoSp
should oLltperform
FeS2 for such applications. To test this theory, slngle
cells were subjectedto a ramped-temperature profile comparableto that of
a long-lifebattery during discharge. The cells were dischargedunder a
resistivebackgroundload of 15.7 ohms and a pulse load of I ohm for 5 s at
BOO s and 3,580 s into discharge. The test results are shown in Figure 4.
The FeS_ cell performedbetter initiallybut was surpassedby the CoS2
cell after -30 min. The dischargecurve for the CoS_ cell was also much
flatter. The minimum pulse voltage at the first pulse was slightlyhigher
for the FeS_ cell but was much lower than the CoS_ cell at the second pulse
near the end of life. The substantialdrop in voltageafter 55 min for the
FeS_ cell is a consequenceof thermaldecompositionof the FeS_ to sulfur
vapbr and FeS which has a much loweropen-circuitpotential. _learly, CoS2
is the preferredcathodechoice for long-lifethermal batteries. These
data have since been corroboratedin tests with full-sizedthermal
batteries (8).
CONCLUSIONS
CoS_ shows superiorperformanceotOFeS_,in single-celltests with
Li(Si) abodes at temperaturesof 400 to 55_vC with the low-meltingLiBr-
KBr-LiF eutecticelectrolyte. The CoS9 cells show a higher rate capability
and a lower voltage loss during heavy-Currentpulses as a result of their
lower internalresistance,even though the FeSp cells have a higher open-
circuit voltage. The use of flooded anodes improvesthe performanceof
both types of cells relativeto the use of dry (unflooded)anodes. Because
of its superiordischargecharacteristicsand higher thermal stability,
CoS is preferredto FeS2 for use in thermalbatterieswith lifetimesof 60
min2or more.
ACKNOWLEDGEMENT
This work was supportedby the U. S. Departmentof Energy under
contractDE-AC04-94AL85000.
REFERENCES
I. R. A. Guidotti,F. W. Reinhardt,and G. C. Nelson, "Characterization of
Sheet Calcium for Ca/CaCrO_Thermal Batteries,"SAND83-2269(1985).
2. R. A. Guidotti and F. W. R_inhardt,"Characterization of DEB Powdersand
Pelletsfor Ca/CaCrO4ThermalBatteries,"SAND83-2270(1985)
3. R. A. Guidotti,F. W. Reinhardt,D. R. Tallant,and K. L. Higgins,
"Dissolutionof CaCrO4 in Molten LiCI-KCI Eutectic,"SAND83-2272
(June, 1984).
4. H. N. Sieger, "Lithium/CobaltSulfide Pulse Power Battery,"Proc. 34th
PoweF Sources Symp., pp. 334-338 (1990).
, 5. N. Papadakis, "Development of a Bipolar Lithium/Cobalt Disulfide
Battery for Pulse Power", Ibid., pp. 339-342 (1990).
6. R. A. Guidotti and F. W. Reinhardt,"Evaluationof Alternate
Electrolytesfor Use in Li(Si)/FeS_ThermalBatteries,"Proc. 33rd
Power Sources Symp., pp. 369-376 (_988).
7. S. P. S. Badwal and R. J. Thorn, d. Sol. St. Chem., 43, 163 (1982).
8. R. A. Guidotti and A. R. Baldwin, "Developmentof a Two-Hour Hour
Thermal Battery,"to be presented at 36th Power Sources Symposium,
June 6-9, 1994, Cherry Hill, NJ.
1.4
m_
i_ 1.0 "-... -
"_ O_
_ 0.6 -
_ .
0.4 -
,__0.2 "
I_0.0 .... I .., ., .I ....... I .... I , ..
0 100 200 3O0 400 500
Actual Run Time, s
Figure I. Maximum Pulse CurrentDensity as a Function of
Time of Li(Si)/FeS
o and L,_(Si)/CoS
2 Thermal Cells During
ContinuousPulsing_at400vC.
0.4
0 l , _ l , , . I . , . I
o I(_ 2o0 _ _o 5(x)
True Run Time, s
Figure 2. Apparent Cell Resistanceas a Function of Time
of Li(Si)/FeS
2 and Li(Si)_CoS2 Thermal Cells During
ContinuousPu_singat 400_C.
i !
1.0
_i_ 0"8 .....
_" ...... ,__•...............
O.6 .... -
0o2
0.0 , , I ,,,, , I.... I , I ........
0 100 200 300 400 500
True Run Time, s
.........
Figure 3. MinimumPulse Voltage as a Function of Time of
Li(Si)/FeS 2 and Li(Si)/Co_i _ Thermal Cells During
ContinuousPulsingat 400vC.
2.5
2.O
I
;>i_ - " -.
Pulse Pulse I
-'"" I
1.0 @ 500 s @ 3580 s
1.73 V 1.86 CoS z
1.81 V 1.21 FeS:
0.5
0.0 I I I 1
0 20 40 60 80 100
Time, min
CoS2l_x_2
Figure 4. Voltage as a Function of Time of Li(Si)/FeS
2
and Li(Si)/CoS
9 Thermal Cells Under a 15.7 Ohms Steady-
State/1Ohm Pu_se Load While Subjectedto a Temperature
Profileof a Long-LifeThermal Battery.
DISCLAIMER
This report was prepared as an account of work sponsored by an agency of the United States
Government. Neither the United States Government nor any agency thereof, nor any of their
employees, makes any warranty, express or implied, or assumes any legal liability or responsi-
bility for the accuracy, completeness, or usefulness of any information, apparatus, product, or
process disclosed, or represents that its use would not infringe privately owned rights. Refer-
enee herein to any specific commercial product, process, or service by trade name, trademark,
manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recom-
mendation, or favoring by the United States Government or any agency thereof. The views
and opinions of authors expressed herein do not necessarily state or reflect those of the
United States Government or any agency thereof.
j I
I
