Electrochimica Acta, Vol. 38, No. 9, pp. 1221-1231, 1993
`Printed in Great Britain.
`
`0013-4686/93 $6.00 + 0.00
`© 1993. Pergamon Press Ltd.
`
`THELi, ,,Mn,0,/C ROCKING-CHAIR SYSTEM: A REVIEW
`
`J. M. TARASCON and D. GUYOMARD
`Bellcore, Red Bank, NJ 07701, U.S.A.
`
`(Received 15 October 1992)
`
`Abstract—The new emerging rechargeable battery technology, called “rocking-chair” or “Li-ion”, that
`uses an intercalation compound for both the positive and negative electrodes is safer than the battery
`technology using pure Li metal. In this paper we review our study of the Li-ion type battery based on the
`spinel Li, ..Mn,O, positive electrode and the negative carbon electrode. First we give a brief history of
`rocking-chair batteries, followed by a description of how to select and optimizeintercalation materials for
`such a type of cell and then we present the main findings which allowed usto bring this battery system
`from concept to realization. Among the main findings are: (1) the use of the second Li intercalation
`plateau of the spinel as a Li reservoir; (2) the discovery of a new electrolyte composition resistant against
`oxidation up to 5 V; and (3) the ability to reversibly intercalate 0.9 Li ions into graphite at high rate with
`a small reversible loss. Based on these findings, AA research prototype cells with enhanced safety charac-
`teristics were constructed. We report on the performance ofthese ceils, compare our results with existing
`or emerging battery technologies such as NiCd or Ni metal hydrides, respectively, and discuss problems
`inherent to the Li-ion technology.
`
`Key words:intercalation, battery, Li-ion.
`
`INTRODUCTION
`
`For the last two decades an increasing interest in Li
`batteries (eg a battery that uses Li metal as the nega-
`tive electrode) has developed amongst battery manu-
`facturers. One reason for this is the high specific
`capacity (3800 Ahkg~' of lithium compared to only
`260 Ah kg™' for lead). In addition, Liis also a highly
`electropositive metal, leading to high voltage cells.
`With this combination of properties, Li-based bat-
`teries are expected to have energy densities much
`larger than other systems.
`To date, only primary lithium cells (among them
`CuO/Li, MnO,/Li and FeS,/Li systems) have been
`commercialized. In contrast, the developmentofsec-
`ondary Li cells (eg cells that consist of an inter-
`calation compound as the positive electrode) has
`been slow because of safety problems associated with
`the use of Li metal as the negative electrode[1—4]
`and due to problems
`stemming from dendritic
`regrowth of Li upon cycling that might short-circuit
`thecell.
`A more advanced andinherently safe approach to
`rechargeable Li batteries, proposed a decade ago[5—
`9],
`is to replace Li metal with a material able to
`reversibly intercalate Li
`ions. This has been called
`the “rocking-chair” battery because Li
`ions rock
`back and forth between the intercalation compounds
`during charge/discharge cycles (Fig. 1). The output
`voltage of such a cell is determined by the difference
`between the electrochemical potential of Li within
`the two Liintercalating electrodes. It is important to
`have as the positive and negative electrodes, cheap
`compounds which can reversibly intercalate Li at
`high and low voltages, respectively. The Li
`inter-
`calation voltage for some intercalation compoundsis
`shown in Fig. 2. Among the materials proposed for
`1221
`
`replacement of Li metal are LiAl, WO,, MoO,[10],
`Mo,Se,[11] or carbon[{12-14], with the latter pro-
`viding the best compromise between large specific
`capacity and reversible cyling behavior. However, in
`rocking chaircells a price is paid in terms of average
`
`Li-ion battery
`
`Cathode
`(Li,.,MingOq)
`
`Charge
`
`Anode
`(carbon)
`
`
`
`Cathode
`
`Discharge
`
`Anode
`(carbon)
`
`© Li*
`
`Li+
`
`©
`
`Fig. 1. The charge and discharge mechanism for a Li-ion
`battery is shown (both the cathode and anode materials are
`intercalation compounds).
`
`APPLE 1024
`
`APPLE 1024
`
`1
`
`

`

`1222
`
`J. M. TARASCON and D. GUYOMARD
`
`output voltage and energy density when compared
`to a lithium metal cell; thus strongly oxidizing com-
`pounds (ie compounds which reversibly intercalate
`lithium above 4 V) must be used as the positive elec-
`trode. So far, as indicated in Fig. 2, there are only
`three Li-based compounds, LiNiO,, LiCoO, and
`LiMn,O,
`that satisfy this requirement. These Li-
`bearing positive electrode materials are not moisture
`sensitive and can be handled in ambient atmospheres
`as can Li free carbon negative electrode materials.
`The Li-ion cell is assembled in its discharged state,
`where the output voltage is close to zero. As with the
`Ni-Cdcells, they need to be charged prior to use.
`Recently, Sony Energytec[15]
`announced the
`commercialization of a battery based on this
`concept. The Li-ion rechargeable battery developed
`by Sony used the layered LiCoO, intercalation com-
`poundasthe positive electrode and a form of carbon
`(petroleum coke) as the negative intercalation elec-
`trode. Following up on this success, Moli Energy
`Ltd. (1990) is developing a battery using the same
`concept[16, 17], but using the layered compound
`LiNiO, as the positive electrode instead of LiCoO,.
`Herein we will review Bellcore’s work on rocking-
`chair type batteries using the Li,,,Mn,O, spinel
`phase as the positive electrode and a form of carbon
`as the negative electrode[18—20].
`system shows
`Our LiMn,0O.,/C rocking-chair
`similar performances, to the Sony and Moli systems,
`with the exception that the specific capacity is about
`5% smaller. However,
`the use of the spinel man-
`ganese oxide material offers the following advan-
`tages: (1) a lower overall electrode cost coming from
`an easier synthesis (only one low temperature step)
`and the use of Mn which is naturally more abundant
`and cheaper than Co or Ni; and (2) the low toxicity,
`a well known characteristics of the widely used
`manganese-based oxide materials. A large number of
`procedures have already been developed for
`the
`recovery of MnO, from alkaline-zinc batteries[21]
`and there has been extensive work on the recovery of
`lithium metal from minerals[22].
`Another advantage of the LiMn,O, system[23-
`26] over the LiCoO, and LiNiO, systems is that
`
`Voltage vs Li
`
`Voltage vs Li
`
`5
`
`5
`
`3
`
`LIA)
`
`Li,Coke
`
`I
`{
`—_—
`Li,Graphite
`Lifetal
`Fig. 2. Electrochemical potential range of Li intercalation
`(as measured with respect to Li metal) for a wide variety of
`compoundsare shown.
`
`LiMn,O, can intercalate a second lithium atom at
`an average voltage of 3V (ie a potential at which
`Li-based materials should be barely moisture sensi-
`tive if synthesized under mild conditions). We will
`show how this second plateau can be used as a
`lithium reservoir to improve the specific capacity of
`these batteries by 10% (a property unique to the
`spinel system) and detail a new way to synthesize
`Li,,,Mn,0, powders of well defined composition,
`x. An problem inherent to rocking-chair technology,
`is the risk of electrolyte oxidation at high operating
`voltages (>4V). For instance,
`the voltage of the
`Li,,,Mn,0,/Li couple is about 4.1V and one
`would need to be able to charge the cell up to 4.5V
`to take full advantage of this redox system. This
`meansthat theelectrolyte of choice for this rocking-
`chair system must be stable over a voltage window
`ranging from about 0 to 5V. We have found a new
`electrolyte composition with such a range of stabil-
`ity.
`In this paper, we will illustrate and comment on
`the different problems faced in going from the idea of
`using the spinel LiMn,O, as the positive electrode
`to the realization of rocking-chair prototypes based
`on this compound.
`
`EXPERIMENTAL
`
`The LiMn,O, powders were prepared by reacting
`a mixture of Li,CO, and MnO, powdersin air, in
`an alumina crucible for three days, followed by two
`successive grinding and annealing sequences. An
`alternative route is the synthesis of LiMn,O, at
`lower temperatures (< 600°C) via a sol gel route that
`consists
`in using LiOH and Mn acetates
`as
`precursors[27]
`leading
`to small
`particle
`sizes
`(<2 pm). We haveinitially used petroleum coke and
`then graphite as intercalation materials for our nega-
`tive electrode. A one molar solution of LiClO, salt
`dissolved in a 50/50 weight ratio of ethylene carbon-
`ate (EC) and diethoxyethane (DEE) wasinitially
`used aselectrolyte.
`The above intercalation compounds are mixed
`with binders and highly conducting powders prior to
`use in electrochemical test cells. The composite elec-
`trodes are made by dissolving a small amount of
`binder, either ehtylene propylene diene monomer
`(EPDM)in a solution of cyclohexane, or polyvinyl-
`idene fluoride (PVDF) in a solution of cylopenta-
`none, and by adding the appropriate amounts of
`intercalation compound and black carbon to this
`solution. The resulting slurry is deposited onto an
`aluminum disk for the positive electrode, and onto a
`copperdisk for the negative electrode and then dried
`at 150°C for 1-2h.
`All the electrochemical measurements were carried
`out using Swagelock laboratorytest cells, Depending
`upon the type of measurements performed, two or
`three electrode cells were used. A “Mac Pile”
`system[28] that can operate ether in a galvanostatic
`or potentiostatic mode was used to perform the mea-
`surements. In the galvanostatic mode,
`the output
`voltage of the cell was monitored while the cell
`delivered a constant current. From the elapsed time
`and the amount of current one can deduce the
`
`4
`3
`2
`;
`
`0
`
`heal
`4
`[ees
`3 i3
`2
`fr "2
`;
`|"
`
`Liymng04
`
`Li,N10,
`
`Lit
`
`LiMo,
`
`[ete
`[ree
`|
`
`2
`
`

`

`The Li, ,,.Mn,0,/C rocking-chair system
`
`1223
`
`amountofintercalated species x in Li,Mn,O, or the
`capacity of the cell (a more meaningful number for
`battery manufacturers). In the potentiostatic mode,
`the current was monitored while the voltage was
`scanned at very low rate (10-50 mV h7!).
`
`RESULTS
`
`Thefirst step in the process of assembling a Li-ion
`cell consists in optimizing the electrochemical behav-
`ior (cycling performance and rate capability) of each
`composite intercalation electrode with respect
`to
`lithium over a wide temperature range. A survey of
`several (s, t, u) mixtures, where s, t and u represent,
`respectively, the weights in % of intercalation com-
`pound,binder and black carbon wascarried out. We
`found that a percentage weight ratio of [89:10:1]
`and [94:5:1] gives the best result for the positive
`and negative electrodes when EPDM is used and
`[90:5:5] and [80:5:5] when PVDFis used.
`Figure 3a illustrates
`the cycling behavior of
`LiMn,O,/LiClO, based electrolyte/Li cells both at
`room temperature and at 55°C. At
`room tem-
`perature the cell was cycled between 3.4 and 4.5V,
`but at 55°C the charge cut-off voltage was limited to
`4.25 V because ofrisks of electrolyte oxidation. The
`spinel phase LiMn,O, can rapidly and reversibly
`intercalate 0.8 Li per formula unit both at room tem-
`perature and at 55°C. However, note that only 0.9 Li
`per formula unit can be removed from LiMn,O,
`during the first charge, while the nominal composi-
`tion is 1.
`The cycling behavior of a petroleum coke/LiClO,
`based electrolyte/Li cell using an optimized compos-
`ite electrode is shown in Fig. 3b. After a subsequent
`loss of reversibility during the first cycle (25%), the
`
`found to rapidly and
`carbon electrodes were
`reversibly intercalate 0.5 and 0.55 Li per six carbons
`(Lig 55Cg)
`(at
`room temperature and at 55°C,
`respectively). The range of reversibility of our pet-
`roleum cokeelectrode was increased to x = 0.7 when
`the discharge cut-off voltage was decreased to 2mV
`vs. Li, but
`this voltage is undesirable for safety
`reasons, because it is too close to those required for
`Li plating. Thus, we have fixed 0.02 V vs. Li as the
`lowest voltage limit that our negative electrode can
`reachin order to rule out anyrisk of Li plating.
`The delithiation of the spinel occurring during the
`charge of a LiMn,O,/Electrolyte/Li electrochemical
`cell to 4.3 V vs. Li results in a A-MnO, phase. This
`phase was reported to be metastable and to decom-
`pose upon heating in air into a non Li-intercalable
`B-MnO, phase[29]. Temperatures ranging from 60
`to 300°C have been reported for this conversion. The
`conversion of the spinel could limit the usefulness of
`the LiMn,O,phase in a practical battery if it occurs
`at low temperature, so we investigated this phase
`decomposition by means of both differential scan-
`ning calorimetry (DSC) and X-ray diffraction mea-
`surements. We found, as shownin Fig. 4, that this
`phase transformation occurs at 180°C, but can also
`be obtained at lower temperatures if lower heating
`rates are used. In contrast to previous reports, we
`did not observe a direct phase conversion to f-
`MnO,, but rather an evolution of the /-phase to the
`é and then finally to the B-MnO, phase. Finally,
`powdersleft at 55°C for three months werestill pure
`4-MnO,, as determined by X-ray analysis, ensuring
`that rocking-chair cells based on this material could
`operate to 55°C without risk of failure due to an
`electrode phase conversion.
`In a rocking-chair cell the amountof positive and
`negative electrode materials has to be adjusted in a
`
`Specific Capacity (mAh/g)
`200
`100
`
`0
`
`Voltage
`
`0.4
`
`Discharge
`
`(voltsvsLi)
`
`
`
`Voaitage(voltsvsLi)
`
`Specific Capacity (mAh/g)
`100
`50
`
`
`
`0
`
`0.2
`
`0.4
`
`0.6
`
`0.8
`
`1
`
`x in Li, Mn2 O4
`
`
`xin Li, Cg
`
`0 poI0 0.2 0.4 0.6 0.8
`
`
`
`
`
`
`
`Fig. 3. The typical cycling behavior at 25 and 55°C for a LiMn,0,/EC + DEE(S0:50) + 1M LiClO,/Li
`(a) and a C (petroleum coke)/EC + DEE(50:50) + 1M LiCIO,/Li(b) is shown. The spinel-based cell was
`cycled at a C/10 rate in contrast
`to C/20 for
`the carbon-based cell. EC = ethylene carbonate,
`DEE= diethoxyethane.
`
`3
`
`

`

`1224
`
`mCal/s
`
`a<3
`2oO2
`2&
`&>2acog
`
`£
`44 28
`36
`28=36
`44
`44
`28
`36
`
`160°C
`s Spinel Mn,O,
`311
`400
`
`222
`
`Diffraction angle (2 6) J. M. TARASCON and D. GuYOMARD
`
`50
`
`100
`
`150 200 250 300 350 400
`
`Temperature (C)
`Fig. 4. Differential scanning calorimetric measurements (DSC)for a delithiated LiMn,O, (ie A-MnO,)
`sample(see text). The phase transformation occurs at 195°C. The inset shows the X-rays taken at position
`marked by asterisks on the DSC curve. The ambient wasair and the temperature rate of 2°C per min.
`
`waythat the reversible capacity of each electrode is
`equalized for both safety and optimum performance.
`In short, with this technology it
`is necessary to
`balance the weight of the two electrodes so that
`when LiMn,O,is fully charged (ie delithiated) the
`negative intercalation electrode has
`to be fully
`lithiated with its voltage close to 0.02 V vs. Li. To
`determine accurately such a ratio, we have used a
`three electrode Swagelock cell (Fig. 5a) where we
`simultaneously monitor the voltage of each electrode
`vs. a Li reference electrode as well as the output
`voltage of the cell. From these measurements we
`deduced that the optimum ratio is 2.1 (Fig. 5b). For
`such a ratio, the LiMn,O,spinel phase is fully deli-
`thiated when the voltage of the negative electrode vs.
`Li reaches 0.02 V. The effect of changing the ratio r
`on both the safety and performance behavior of a
`rocking-chair cell
`is illustrated in Fig. 6. Ratios
`greater than 2.1 may result in Li plating that could
`jeopardize the cell safety. Ratios smaller than 2.1
`result in poor use of the rocking-chair system and
`decreases its safety, since the potential can reach
`values above that of electrolyte decomposition of the
`electrolyte. This situation is illustrated in Fig. 7 where
`the cycling behavior of a rocking-chair with a ratio
`of 1.7 is shown. The overcharge cut-off voltage is
`limited to 4.5 V (4.6 V being the electrolyte decompo-
`sition potential at room temperature). However, note
`that on charge the voltage of the cell stabilizes at
`4.5V because of the electrolyte oxidation, by the
`positive electrode. (The potential at the positive elec-
`trode is 4.8 V vs. Li.) On subsequent cycles, even if
`the cell is kept lower than 4.4V vs. Li, the voltage
`composition curve shows lower capacity and larger
`polarization than a cell
`that has not been over-
`charged. Most likely, the electrolyte decomposition
`results in the formation offilm at the positive elec-
`trode that increases the internal cell resistance. This
`trend is enhanced by further cycling until the cell has
`
`no more capacity. These results indicate that for
`rocking-chair cells using LiClO, based electrolytes,
`once the oxidation of the electrolyte has been initi-
`ated, it cannot be stopped even by lowering the over-
`charge cut-off voltage on subsequentcycles.
`An optimized cell with a ratio of positive to nega-
`tive electrode of 2.1 was assembled andits first cycles
`are shownin Fig. 8. The cell is cycled at a rate of C/5
`between 2 and 4.35 V. After the first cycle, the differ-
`ence in voltage between charge and discharge is
`small
`indicating a good rate capability. The cell
`capacity decreases rapidly from the first
`to the
`second cycle and then levels off after the fifth and
`sixth cycle. This cell can reversibly intercalate 0.32 Li
`atoms per Mn atom at an average output voltage of
`3.7V, yielding an effective energy of 250Whkg =!
`(both positive and negative masses included). This is
`2.5 times greater than that obtained with the well
`known Ni-Cdbatteries.
`rocking-chair
`The above results indicated that
`cells can indeed be made with LiMn,O,as the posi-
`tive electrode. However, before such a system can be
`used in a commercial product several problems need
`to be solved in order to enhance safety and improve
`performance.
`The use of a LiClO, based electrolyte is not a
`viable option with respect to safety because of the
`explosive nature of the LiClO, salt. In addition, the
`narrow voltage window between the end of the
`lithium deintercalation in the spinel and the begin-
`ning ofthe electrolyte oxidation, 0.2 V at room tem-
`perature and OV at 55°C,limits the capacity of our
`system at 55°C, and does not provide a safety
`margin large enough in case of overcharge.
`When a LiMn,O,/LiClO, based electrolyte/C cell
`is discharged to 0 V, the voltage of the negative elec-
`trode is equal to that of the positive electrode nearly
`4V vs. Li (a LiMn,O,/Electrolyte/Li has an open
`circuit voltage of 3.5V). A reaction may occur
`
`4
`
`

`

`The Li, ,,.Mn,0,/C rocking-chair system
`
`1225
`
`
`
`Spring
`Current Collector
`
`Counter Electrode
`
`
`Separators
`
`Working Electrode
`Ref. Electrode (Li)
`
`Plungers
`
`Dats
`
`een
`
`LiMn2O,/COKE CELL AT 25°C
`IN EC + DEE + LiCIOg
`
`29 charge
`
`1% discharge ~—
`
`
`
`Voltage(volts)
`
`Time (hours)
`Fig. 5. The schematic of a three-electrode cell is shown in (a) and thefirst cycles at room temperature of
`a LiMn,0O,/EC + DEE(50:50) + 1M LiClO,/C (petroleum coke) three-clectrode cell, under constant
`current, are shown in (b). The voltage measured at the positions denoted 1, 2 and 3 on thecell schematic
`is plotted as a function of time in the bottom plot, curves 1, 2 and 3, respectively.
`
`60
`
`80
`
`100
`
`F—I—-—+—S
`
`between the copper current collector and the C elec-
`trode at voltages greater than 3.4V vs. Li. Thus, a
`requirementfor this rocking-chair system is the pro-
`vision of an overdischarge protection allowing the
`cell to operate safely.
`
`25°C
`
`Voltage
`
`(volts)
`
`"0.0
`
`0.2
`
`0.6
`0.4
`xin LiyMngO4
`
`0.8
`
`1.0
`
`Fig. 7. The cycling behavior of a cell using a weight ratio
`(positive to negative electrode) of 1.7 (51/30mg) is shown.
`The cell was cycled at 0.5SmAcm~? between 0 and 4.5V
`during the first charge and then between 2.5 and 4.35 on
`subsequentcycles.
`
`A
`LiMn,O,
`4.5r
`a a“ AL c
`
`4
`
`a
`
`s‘
`
`50
`
`660
`
`40
`
`50
`
`«40
`
`Time (hours)
`
`Fig. 6. Voltage profiles vs. Li (as measured from a three
`electrode cell) of LiMn,O, and carbon electrodes at the
`end of the first charge and the beginning of the first dis-
`charge with different ratios r= 3.1 (a), r= 2.1 (b) and
`r = 1.7 (c). For these experiments the mass of the carbon
`electrode was maintained constant (30mg), the electrolyte
`used was EC+ DEE (50:50)+1M LiClO, and the
`current rate was 0.4mA cm ~?,
`
`oo> 30
`
`é
`
`> £°2oo
`
`5
`
`

`

`1226
`
`J. M. Tarascon and D. GUYOMARD
`
`5.0
`
`80
`
`Specific Capacity (mAh/g)
`60
`40
`65 mAhj/g
` --------- >
`
`20
`
`0
`
`
`
`Voltage(volts)
`
`Charge
`
`Discharge
`
`Li,MngO4/GOKE CELL
`
`IN EC + DEE+ LiClO,
`
`system. The
`the Li-Mn-O based rocking-chair
`spinel LiMn,O, possesses
`a
`second reversible
`lithium intercalation plateau[23, 24] located at 3V
`and we propose to use this second lithium inter-
`cation plateau as a lithium reservoir[18]. Indeed, by
`increasing the amount of Li in Li,,,Mn,0,, the
`value of x may be adjusted to compensate for the
`loss of Li associated with the negative electrode,
`barely affecting the weight of the positive electrode.
`To take advantage of this unique property of the
`spinel,
`it
`is mecessary
`to
`prepare
`air-stable
`Li,,,Mn,0, powders so that
`the manufacturing
`cost will
`remain unchanged. The synthesis of
`Li, Mn,O, has been reported[30] using n-BuLi as a
`reducing reagent and it was claimed that the reacting
`material was moisture sensitive. We identified the
`room temperature of a
`Fig. 8. Cycling behavior at
`use of a strong reducing reagent such as n-BuLi as
`LiMn,0,/EC + DEE(50:50)+ 1M_LiClO,/C rocking-
`the source of the problem. We have tried a mild
`chair cell at a rate of C/10 with the mass of the positive
`reducing reagent, such as Lil, and demonstrated the
`electrode being 2.1 times that of the negative electrode.
`possibility
`of
`preparing
`air-stable Li,.Mn,O,
`powders. Our chemical
`reaction simply involves
`refluxing LiMn,O, powders in an acetonitrile solu-
`tion of Lil for several hours.
`The key advantage of this chemical reaction is
`that it allows preparation of Li,,,Mn,O, powders
`of well defined composition x as described in detail
`elsewhere. By fixing the amount of LiMn,O, powder
`and the volume of acetonitrile and by using the
`Nernst equation, the redox potential of the solution
`at any state of the reaction x, and the amountof Lil
`needed to form a Li, ,,Mn,0, powderof defined x
`maybecalculated. Figure 9 illustrates this. Note that
`the reduction reaction can only occur when the
`redox potential of the solution is lower than the one
`of
`the LiMn,O,/Li,Mn,O, plateau. Finally,
`it
`should be possible to scale this process up.
`Using the three electrode cell, we determined that
`the right amount of excess lithium in Li, ,,Mn,0,
`to compensate for the loss associated with the nega-
`tive electrode during the first cycle is x = 0.1. The
`behavior of a rocking-chair cell using Li, ;Mn,O,
`
`2.05
`
`0.2
`
`0.6
`0.4
`x in Li,MngO4
`
`0.8
`
`1
`
`Finally, although the petroleum coke negative
`electrode gives the best compromise between capac-
`ity and cycling performance over other graphitic
`carbons
`it would be of tremendous benefit
`to
`improve the capacity of this negative electrode (ie
`using graphite instead of petroleum coke). The Li
`intercalation into graphite has been intensively
`studied for more than twenty years. Even the best
`results obtained so far
`show irreversible losses
`during the first cycle of 50% and poorrate capabil-
`ities, never exceeding C/30[15, 16]. Such a large irre-
`versibility during the first cycle is not critical for cells
`having a Li negative electrode, however,
`in a
`rocking-chair type configuration where the amount
`of lithium is well defined by the amount of positive
`electrode, such a large loss is detrimental to per-
`formance. There is also a need to enhance the
`kinetics of the lithium intercalation process with
`graphite. In the next section we present our solutions
`to the above problems.
`
`Li reservoir within the spinel LiMn,O,
`We have previously shown that our petroleum
`coke electrode exhibits a 25% loss in reversibility
`during the first cycle similar to all the rocking-chair
`systems using a carbon negative electrode. This
`implies that a large amount of Li from the positive
`electrode is lost during the first recharge of the
`rocking-chair cell. Our LiMn,O,/petroleum coke
`rocking-chair system has been optimized with a mass
`ratio of 2.1 which takes into account the 25% capac-
`ity loss at the carbon electrode during thefirst cycle.
`However,in these cells, an excess amountof positive
`electrode [(1 + x)LiMn,0,] material
`is used to
`provide the extra amount of Li that is lost on the
`carbon negative electrode during the first cycle. This
`excess reduces the gravimetric and volumetric capac-
`ities of these cells. A way to overcome this problem
`would be to use the spinel structure with extra Li
`(Li, ,,,.Mn,O,) in such a way that the excess Li could
`be totally used to compensate for the 25% loss at
`the
`negative
`electrode. The
`construction
`of
`Li, ,,Mn,O,/Cinstead of (1 + x)LiMn,0,/C would
`result
`in a lower mass ratio (<2.1),
`leading to
`enhanced gravimetric and volumetric capacities for
`
`NERNST : E=E°% RT/2F Log [Iq |/11'|°
`
`Voltage(VvsLi/Li’) 2.91
`
`1.2
`
`1.4
`
`8.0
`
`0.2
`
`#04
`
`1.0
`O08
`#06
`x in Li,,,.Mn,Q,
`Fig, 9. A calculation of the Nernst equation at 82°C for the
`reaction: LiMn,O, + 3x/2 Lil > Li,,,Mn,0, + x/2 Lil,
`has been done in order to determine the amount of Lil
`needed to prepare Li,,.Mn,O0, powders of any wanted
`lithium excess (x). The amount of LiMn,O, has been fixed
`at 4g, while nx4g of Lil is used in 100 cm? ofacetonitrile.
`The
`dashed
`curve
`represents
`the
`voltage
`of
`the
`LiMn,0,/Li,Mn,O, phase transformation measured at
`82°C.
`
`6
`
`

`

`1227
`
`3.8
`
`42 44 46
`Voitage (V)
`
`48
`
`5
`
`4 3
`
`.8
`
`4°
`
`#42 44 46
`Voltage (V)
`
`48
`
`5
`
`(A) 42
`Current
`
`44
`
`#5
`
`5.2
`
`48 °
`#46
`(V)
`Vaitage
`Fig. 11. Comparative plots of cell current against charging
`voltage at both room temperature (a) and 55°C (b) for two
`Li, , .Mn,0,/Licells using our prior LiClO, based electro-
`lyte (dashed curve) and our new LiCO, free electrolyte
`(solid curve). The voltammogram in the high voltage range
`4,3 and 5.1 V for a Li, ,,Mn,O,/Licell using the new elec-
`trolyte is shownin (c).
`
`perature (Fig. 11a) and at 4.3 V at 55°C (Fig. 11b) for
`our LiClO,-based electrolyte corresponds to its oxi-
`dative decomposition. In contrast, this rapid increase
`of
`the current upon increasing voltage is not
`observed for our new electrolyte above 4.3 V. Strik-
`ingly, the oxidizing current at 5 V remains very small
`even at 55°C,indicating an increasedstability of this
`electrolyte against oxidation. However, several weak
`features can be observed on the I vs. V curve at volt-
`ages greater than 4.3V. These anomalies become
`more pronounced in Fig. 11c where this part of the
`curve has been enlarged. This new set of reversible
`oxidation-reduction peaks, observed for
`the first
`time, corresponds to the removal of 0.06 lithium ion
`from the spinel.
`to remove
`is difficult
`This result shows why it
`more than 0.9 Li from the spinel. The two capacity
`peaks are related to local structural defects (ie differ-
`ent environments for the Li ions) and their ampli-
`tude is extremely sensitive to the conditions of
`synthesis of LiMn,O,. The electrochemical per-
`formance of LiMn,O, was found to be closely tied
`to the relative amplitude of these two peaks. In
`short, they can simply be used as fingerprints for pre-
`paring LiMn,O, powders which will give the best
`electrochemical performance when used as positive
`electrode in a rocking-chair cell. A complete study
`
`The Li, ,,Mn,0,/C rocking-chair system
`0.14
`0.12
`
`
`
`Current(mA) ° o@
`
`|.
`
`The Li reservoir
`
`
`
`
`
`
`Voltage(volts)
`
`‘Time (hours)
`cycles
`at
`room temperature of a
`first
`Fig. 10. The
`Li, ,Mn,0,/EC + DEE(50:50) + 1M LiClO,/C (petrol-
`eum coke)cell cycled between 4.35 and 2.3 V, as measured
`with the three-electrode configuration. The solid curve rep-
`resents the output voltage of the cell, while the dashed and
`the dot-dashed curves represent, respectively, the voltage
`vs. Li of the LiMn,O, and the carbonelectrode.
`
`powderas the positive electrode is shown in Fig. 10.
`Weobserved the removal of 0.1 Li at 3 V during the
`first cycle, and subsequent cycles without any further
`loss in capacity. The use of the Li reservoir results in
`an increase in the cell capacity by 10%. In addition,
`this extra lithium will limit the voltage of the nega-
`tive electrode to 3V (vs. Li) when discharging the
`rocking-chaircell to 0 V, and thereby acts as a safety
`switch for these cells in case of overdischarge.
`
`New electolyte composition
`We showed that 1 M LiClO, in a 50:50 mixture of
`ethylene carbonate (EC) and diethoxyethane (DEE)
`electrolytes begins to undergo oxidation at 4.5 V,
`and 4.3V at 55°C. The electrolyte oxidation was
`found to result in solid and gaseous products, with
`the solid products forming a film that increases the
`polarization of the cell and decreases its output
`voltage, while the gaseous products increase the risk
`of explosion. Thus, to increase the cycle life of these
`batteries and to improve their
`temperature per-
`formance it is necessary to minimize the electrolyte
`oxidation. This search is purely empirical since the
`catalytic activity of the positive electrode material
`against electrolyte oxidation cannot be predicted.
`Numerous
`electrolyte
`compositions
`containing
`various ratios of salts, solvents and cosolvents were
`investigated for their ionic conductivity and resist-
`ance to oxidation. From this study a new LiClO,-
`free electrolyte composition, resistant against oxida-
`tion to 5 V, compatible with carbon (both petroleum
`coke and graphite) and with high ionic conductivity
`over a wide temperature range was discovered. The
`oxidation current for this proprietary new electrolyte
`under highly oxidizing conditions (up to voltages of
`5 V) is compared(in Fig. 11) to the oxidation current
`of our previous electrolyte under similar test condi-
`tions. In both plots, the peaks located at 4.0 and
`4.15V correspond to the removal of Li while the
`large increase of the current at 4.5V at room tem-
`
`7
`
`

`

`1228
`
`J. M. Tarascon and D. GuYOMARD
`
`dwelling on the importance of these two peaks
`with
`respect
`to
`the
`optimization
`of
`the
`Li, +,Mn,0,/Electrolyte/C system will be described
`elsewhere[31].
`Besides allowing a better understanding of the
`importance of the synthesis conditions on the elec-
`trochemical behavior of Li, .,Mn,O, with respect to
`lithium,
`this newly discovered electrolyte has also
`increased the cycle-life of these rocking-chair bat-
`teries. Figure 12 shows the capacity plotted as a
`function of the cycle-number both at room tem-
`perature and at 55°C. It is remarkable to note that
`the cell capacity of a Li, ,,.Mn,O,/Electrolyte/C cell,
`charged and discharged at a rate of C/2 between 2
`and 4.7V,
`is still higher than 60% of the initial
`capacity after 2300 cycles. Even at 55°C, when cycled
`under similar conditions, the loss in cell capacity is
`less than 20% ofthe initial capacity measured after
`the fifth cycle.
`
`Carbonvs. graphite
`The effective use of graphite as an intercalation
`material in a Li battery has been prevented by the
`inability of finding an electrolyte that: (1) minimizes
`the large irreversibility of the Li intercation process
`into graphite during the first cycle; and (2) allows for
`a fast intercalation rate. The best results so far were
`reported by J. Dahn et al[32] (Fig. 13a) which
`showssuccessful intercalation of twolithiumsper six
`carbons (“Li,C,”) during the first cycle and one
`reversibly on subsequentcycles at a rate of C/40. He
`suggested that the formation of a superficial passiva-
`tion layer on carbon (resulting from electrolyte
`reduction) controls the kinetics of the Li
`inter-
`
`1.6
`
`a
`
`Prior Art
`
`
`
`Voltage(Volts) o— ai
`
`°aS
`
`o
`
`0.5
`
`1.0
`
`1.5
`
`2.0
`
`x in LIC,
`
`
`
`Voltage(Volts)
`
`0.5
`
`300
`
`°
`
`° N
`
`0.6
`0.4
`x in Licx
`6
`
`0.8
`
`200
`
`
`
`Capacity(mAh/g) 100
`
`120
`
`FOO bern
`
`wR oO oO
`
`Capacity(mAh/g)
`100 f Capacity
`(mAh/g)
`
`b&oaowoOQOQo
`
`120
`
`oa Q
`
`0
`
`500
`
`1000
`
`1500 2000 2500
`
`Discharge number
`
` L
`300
`100
`200
`400
`500
`
`0
`
`Discharge number
`Fig. 12. Variation of the capacity as a function of the cycle
`number for two Li,,,Mn,O,/new electrolyte/petroleum
`coke laboratory test cells operating at 20°C (a) and at
`55°C (b).
`
`°
`
`0.5
`
`1.5
`1.0
`Cycle Rate (hr.!)
`
`2.0
`
`25
`
`Fig. 13. The previous state of the art for the electrochemi-
`cal intercalation of Li into graphite is shown in (a). In (b)
`the typical cyling behavior at 25°C of a graphite/new
`electrolyte/Li cell charged and discharged at a rated of
`C/15 is shown with (c) the capacity as a function of various
`discharging rates for the samecell.
`
`calation process into carbon, limiting the rate capa-
`bility of the C/Electrolyte/Licell. In spite of the low
`rate capability, the 50% loss duringthe first cycle is
`not
`tolerable with the present rocking-chair tech-
`nology, because the amount of available lithium is
`defined by the amountof Liin the positive electrode.
`Thus, electrolytes that are compatible with graphite
`are needed.
`Electrochemical cells using a graphite electrode,
`1M LiC10,-based electrolyte and Li as the negative
`electrode have shown a high degree of irreversibility
`duringthefirst cycle (> 60%) and poorrate capabil-
`ity. For comparison, the cycling behavior of a cell
`(denoted cell 1) using our new electrolyte and dis-
`charged and charged at C/15 rate is shown in Fig.
`13b. Note that the reversibility lost during the first
`cycle