Solid State Ionics 148 (2002) 405 – 416
www.elsevier.com/locate/ssi
A short review of failure mechanisms of lithium metal and lithiated
graphite anodes in liquid electrolyte solutions
Doron Aurbach *, Ella Zinigrad, Yaron Cohen, Hanan Teller
Department of Chemistry, Bar-Ilan University, Ramat Gan 52900, Israel
Abstract
Li electrodes in any relevant electrolyte solution (i.e., polar aprotic) are covered by surface films of a very complicated
structure. It was found that even in cases where the surface films formed on lithium contain elastomers, or where the lithium
metal reactivity is reduced by doping with elements such as N, As, Al, Mg, Ca, etc., it is impossible to achieve sufficient
passivation with lithium electrodes and liquid solutions. Passivation is considerably worsened when Li electrodes are operated
at high rates (especially at high charging, Li deposition rates). Thus, there is no way that rechargeable Li batteries can compete
with Li-ion batteries in any application that requires high charging rates (e.g., in powering portable electronic devices). The
electrochemical behavior of lithiated graphite electrodes also depends on passivation phenomena. The surface films formed on
lithiated graphite are similar to those formed on Li metal in the same solutions. The volume changes of graphite electrodes
during Li insertion – deinsertion are small enough to enable their reasonable passivation in a variety of electrolyte solutions. A
critical factor that determines the stability of graphite electrodes is their morphology. It was found that the shape of graphite
particles plays a key role in their application as active mass in anodes for Li-ion batteries. D 2002 Elsevier Science B.V. All
rights reserved.
Keywords: Lithium; Liquid electrolyte solutions; Li-ion batteries
1. Introduction
The beginning of the third millennium marks over
30 years of intensive R&D of rechargeable Li (metal)
batteries, and about 15 years of very successful R&D
of Li-ion batteries. In spite of the intensive efforts to
develop and commercialize rechargeable batteries
with lithium metal anodes and liquid electrolyte
solutions (i.e., for improved low temperature performance, high rates, and cheap production), except for
*
Corresponding author. Tel.: +972-3-5318317; fax: +972-35351250.
E-mail address: [email protected] (D. Aurbach).
two or three short-term ventures (e.g., Li – MoS2
batteries from MOLI Canada [1] and Li– MnO2 from
Tadiran Israel [2]), these batteries have never become
a commercial reality. In contrast, shorter-term efforts
to develop Li-ion batteries (i.e., based on the ‘rocking
chair’ concept [3]) proved very successful.
In recent years, rechargeable Li-ion batteries have
conquered the market and powered many types of portable electronic and optoelectronic devices throughout
the world. In their current common version, Li-ion
batteries are comprised of graphitic anodes, LiCoO2
cathodes, and liquid electrolyte solutions based on
LiPF6 as the electrolyte in a mixture of alkyl carbonate
solvents. These include ethylene carbonate (EC) as a
necessary component, and dialkyl carbonates from the
0167-2738/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 7 - 2 7 3 8 ( 0 2 ) 0 0 0 8 0 - 2
406
D. Aurbach et al. / Solid State Ionics 148 (2002) 405–416
following list: dimethyl, diethyl, ethyl – methyl carbonates (DMC, DEC, and EMC), etc. [4]. Both graphite
and LiCoO2 are layered compounds that intercalate
reversibly with lithium in processes which are mainly
phase transitions between stages of different degrees of
lithiation. While LiCoO2 electrodes insert lithium
reversibly in any polar aprotic Li salt solution of
sufficient anodic stability, the process of Li insertion
of graphitic carbon, as well as the stability of graphite
electrodes in Li salt solutions, is highly dependent on
the composition of the electrolyte solution [5].
Hence, in both cases of Li metal-based or Li-ion
rechargeable batteries, their successful operation
depends primarily on the performance/stability/reversibility of the anode side (i.e., Li metal or graphite
electrodes, respectively). Both Li metal and graphite
electrodes in a large variety of non-aqueous Li salt
solutions have an important common feature—in the
same solutions, they are covered by very similar
surface films, which in fact control their electrochemical behavior.
Lithium reacts spontaneously with all atmospheric
gases except noble gases, all polar aprotic solvents,
and most of the commonly used salt anions (e.g.,
ClO
4 , AsF 6 , PF 6 , BF 4 , CH 3 CO 3 , and (CH 3 SO2)2N ) to form Li salts, most of which are usually
insoluble in the mother precursor solution [6]. These
film formation processes passivate the active metal
and protect it from corrosion when the surface films
become sufficiently thick to block electron transfer.
Due to the special properties of the lithium ion (e.g.,
its small size) and the nature of its ionic bonds with
other atoms, most of the Li salts precipitated as thin
films conduct electricity under an electric field. This is
due to the fact that Li-ions are mobile in thin films of
Li salts and can thus easily migrate. This is known as
the solid electrolyte interphase (SEI) model for Li
electrodes [7].
It is generally known that Li deposition in many
electrolyte solutions is highly dendritic, and this
makes these systems (electrode-solutions) unsuitable
for secondary batteries [8]. Another highly important
aspect regarding practical batteries is that of safety
issues. As already explained in detail, many lithiumsolution systems undergo a thermal runaway [9].
Hence, poorly designed rechargeable Li batteries can
explode easily under abuse conditions such as short
circuit and overcharge, when they are punctured or
crushed or when they are exposed to high temperatures. In recent years, we have developed rechargeable batteries with internal safety mechanisms that
prevent explosion of the batteries in the event of short
circuit, overcharge, and heating up to 150 jC [10].
Over the past 15 years, we have intensively explored
the behavior of Li electrodes in a large variety of
solutions [11 –13], using a variety of in situ methods,
including FTIR spectroscopy [14], AFM [15], EQCM
[16], and EIS [17]. Together with Tadiran [10,18 – 20],
we then explored completed, practical, and rechargeable AA Li –LixMnO2 batteries. These studies included
a rigorous postmortem analysis of cycled batteries [20].
These studies provide us with a broad view of the
intrinsic limitation of rechargeable Li battery systems
with liquid electrolyte solutions, and of possible useful
future directions for R&D of practical rechargeable Li
battery systems.
Similar to lithium, graphite electrodes polarized to
low potentials in non-aqueous solutions, containing Li
salts, reduce solution species to insoluble salts, and
hence, become covered by surface films of a chemical
structure, similar to that of surface films formed on
lithium in the same solutions [21]. For lithiated graphite, the major failure mechanism is cointercalation of
solvent molecules that are pushed together with Liions between the graphene planes. In the absence of
passivation, cointercalation of solvent molecules
together with Li-ions simply exfoliates the graphite
and decomposes it to a dust of graphene sheets [22].
Hence, a key factor that determines the stability of
graphite electrodes in Li insertion processes is to what
extent protective surface films are formed rapidly
enough before cointercalation can take place [23]
(i.e., these processes are definitely competitive). During the past 10 years, we have extensively studied the
behavior of graphite electrodes in a large variety of
electrolyte solutions [24 –26]. Recently, these studies
have also included postmortem analysis of practical
Li-ion batteries [27].
This paper aims at reviewing our very recent
conclusions on the factors that limit the use of Li
metal anodes in rechargeable battery systems, and
what are the possible practical directions in R&D of
new rechargeable Li (metal) batteries. In addition, this
paper reviews our conclusions regarding capacityfading mechanisms of graphite electrodes in Li-ion
batteries.
D. Aurbach et al. / Solid State Ionics 148 (2002) 405–416
2. Experimental
The experimental aspects of our studies of Li
electrodes in non-aqueous solutions using FTIR spectroscopy, EQCM, AFM, EIS, SEM, EDAX, and
standard electrochemical techniques are described in
Refs. [28 – 33], respectively. The study of completed
battery systems and the use of postmortem analysis is
described in Refs. [2,10,19,20]. The experimental
aspects of our studies of graphite electrodes are described in Refs. [23 – 26,35]. These include descriptions of electrode preparation and their study by
standard electrochemical techniques, EIS, FTIR spectroscopy, and XPS. In situ analysis of graphite electrodes by in situ AFM is described in Ref. [36], and
postmortem analysis of practical Li-ion batteries is
described in Ref. [27].
3. Results and discussion
3.1. Li (metal) battery systems
Scheme 1 presents, as an example, a list of surface
reactions of lithium in a single electrolyte solution: ECDMC/LiPF6 [35]. Both solvents react to form several
products, both ROCO2Li and ROLi species, as well as
species with LiUC bonds. Further reduction of
ROCO2Li, as well as secondary reactions to these
species with trace H2O, leads to the formation of surface
Li2CO3. The salt anion (PF
6 ) and its related contaminants, HF and PF5, also react with lithium, and hence,
contribute to the complicated surface chemistry of
lithium. In addition to this variety of surface reactions
that form a mosaic of insoluble surface species, the
sequence of the surface reactions further complicates
the structure of the surface films formed on lithium [37].
Consequently, the surface films formed on lithium have
a multilayer structure [17,37] (also confirmed by depth
profiling and XPS [38]), and are laterally non-uniform,
as confirmed by in situ AFM [39]. Imaging of Li
electrodes by AFM (see Scheme 1) clearly reflects the
non-uniform morphology of the surface films on Li
electrodes (in the nanometer scale), dictated by the
complicated Li surface chemistry and the chemical
non-uniformity of the surface films thus formed.
In fact, a similar complicated picture of the surface
chemistry of lithium in non-aqueous solutions was
407
found for almost every system studied, including
ethers, esters, other alkyl carbonate solutions, and different salts, including LiClO4, LiAsF6, LiN(SO2CF3)2,
LiBF4, and LiC(SO2CF3)3 (each of the anions of these
salts is reactive with lithium) [11 – 13]. This lateral and
vertical non-uniformity of the surface films induces a
highly non-uniform current distribution of Li deposition –dissolution, which was probed by in situ AFM
imaging [15,39]. The typical surface phenomena that
relate to the non-uniform Li deposition – dissolution
processes as observed by imaging with AFM [39] are
schematically presented in Fig. 1. Because the surface
films formed on lithium in most of the relevant and
commonly used electrolyte solutions are comprised of
Li salts, their cohesion and flexibility is very limited.
Hence, as seen in Fig. 1, these surface films cannot
properly accommodate the morphological changes of
the Li metal upon Li deposition and dissolution (due to
the non-uniformity of these processes). The surface
films formed on lithium can be easily cracked, and the
passivation is broken during both Li deposition and
dissolution. This leads to dendrite formation and to a
massive loss of both lithium and solution species due
to the surface reactions and the ‘repair’ of the surface
films (on an increasing Li surface area). This situation
is, of course, intolerable for rechargeable battery systems, and indeed, most of the commonly used nonaqueous solvents and Li salts are useless for secondary
Li batteries. However, it should be noted that when
pressure is applied to the Li electrodes in solutions
(e.g., by packing Li batteries with external physical
pressure, so that the separator between the Li anode
and the cathode pressurizes the Li surface), phenomena such as dendritic Li deposition are largely avoided
[40,41].
In many Li battery systems, such as cylindrical
cells of a jelly-rolled configuration, the application of
uniform pressure on the Li anode is not possible and
therefore, application of pressure is definitely not the
solution to the problems addressed above. In order to
obtain sufficiently good passivation of Li electrodes,
despite the unavoidable phenomena of non-uniform
electrochemical behavior, the surface films on lithium
must be flexible, thus containing elastomers in addition to ionic species. Indeed, we found one electrolyte
solution in which the passivation of lithium is excellent, and hence, Li deposition occurs in flat, flake-like
morphology. This solution includes 1– 3 dioxolane
408
D. Aurbach et al. / Solid State Ionics 148 (2002) 405–416
Scheme 1. A list of surface reactions on lithium surfaces in the electrolyte solution EC-DMC/LiPF6 [1 – 13,35] and a relevant AFM image of the
Li surface in this solution.
D. Aurbach et al. / Solid State Ionics 148 (2002) 405–416
Fig. 1. A description of the morphology and failure mechanisms of lithium electrodes during Li deposition and Li dissolution and relevant AFM images describing selected
phenomena: the beginning of dendrite formation and non-uniform Li dissolution accompanied by breakdown and repair of the surface films. (Li electrodes in an EC-DMC/LiPF6
solution.)
409
410
D. Aurbach et al. / Solid State Ionics 148 (2002) 405–416
(DN) as the solvent, LiAsF6 as the salt, and tributylamine (TBA) as a basic stabilizer (to prevent immediate polymerization of the DN, which is an acetal, in
the presence of trace Lewis acids that come with the
salt, e.g., AsF5 and AsF3). DN is reduced on lithium to
different alkoxy species. ROLi and AsF
6 is reduced
to LiF and LixAsFy species, which may also include
As0 and Li3As. The arsenic species on the Li surface
modifies its reactivity [10]. DN reduction, and/or the
presence of ROLi species on the surface, leads to
anionic partial polymerization of DN to form species
of the LiO(CH2CH2OCH2O)n Li type. These species
are elastomers, and probably provide the surface films
formed on Li in these electrolyte solutions, with the
necessary flexibility needed to maintain sufficient
passivation during Li deposition – dissolution processes.
Indeed, Li cycling efficiency in the DN/LiAsF6/
TBA electrolyte solutions is close to 100%, as confirmed by both standard electrochemical measurements
[34] and EQCM [29]. Based on these solutions, we
commercialized AA rechargeable Li (metal) batteries
with LixMnO2 (0.3 < x < 1), 3-V cathode [2]. The use of
DN/LiAsF6/TBA provided an additional important
bonus, namely, internal safety mechanisms that shut
down the battery in abuse cases such as short circuit,
overcharge, and overheating (see Ref. [10]). We then
discovered a serious drawback to these batteries: while
they could be discharged at high rates (C/1 – 3 h), and
could deliver hundreds of full DOD (cathode limited)
discharge– charge cycles, their charging rate had to be
very slow (C/10 – 12 h). Otherwise, the number of
cycles available was very small. Our extensive studies
of this battery system, which included postmortem
analysis of dead batteries [20], is summarized in Fig.
2. This figure shows the effect of charging rate,
translated to the current density of Li deposition in
the batteries, on the available charge–discharge cycles
and on the morphology of the lithium anodes (see the
attached SEM micrographs), and the dependence of the
average grain size of lithium deposits on the current
density of Li deposition. It should be noted that the
discharge rate in all of these experiments was C/3 h,
corresponding to 250 mA per AA battery, and 1.25
mA/cm2 of Li dissolution.
The life span of the batteries ended for a single
reason only: the solution reacted with lithium and
disappeared. In fact, both electrodes from dead cells
could be operated in fresh solutions. Of significance is
the fact that as the current density of Li deposition
increases, the grain size of the Li deposits decreases.
However, this decrease levels off at some high current
density, and the low number of cycles available for the
batteries at a high charging rate levels off accordingly.
Hence, from these results we concluded that although
Li deposition in this system is not dendritic, the
passivation of the lithium deposits can never be
hermetic. High current densities lead to Li deposition
in small average grain size. Moreover, the average
grain size of the Li deposit is smaller, and the number
of grains simultaneously growing on the electrode
surface is higher as the current density increases [42].
As a result, the overall available surface area for Lisolution reactions increases as well. Therefore, high
charge rates cause enhanced surface reactions of
solution species. Since the amount of solution in
practical batteries is relatively small (mostly contained
in the pores of the separator and in the composite
cathode), the solution simply disappears rapidly if the
surface reactions are too intense.
These findings clearly demonstrate a critical limitation of secondary Li batteries with liquid electrolyte
solutions: their charging rates have to be very slow,
overnight or so, and hence, these batteries will never
be able to compete with Li-ion batteries, whose
energy density is nearly the same and can be charged
at C rates (i.e., the total capacity is recovered during
less than 1 h of charging).
In an attempt to modify the morphology of Li
deposition at fast rates, and to reduce the reactivity of
the Li surface towards Li anode solution species, we
followed reports on Li– Li3N solid solutions as excellent anodes for Li batteries [43], and produced Li
anodes containing different amounts of Li3N (3 –
11%). We also tried using Li foils containing 3– 6%
aluminum, since it was found that Li – Al alloys are
much less reactive towards solution species than Li
metal [44]. Another modification that we tried was to
dope the Li surfaces by Ca and Mg using Ca(ClO4)2
or Mg(ClO4)2 additives in solution (Li surface reduces
both Ca2+ and Mg2+ ions, thus forming Li– Ca and
Li –Mg surface alloys). We also tried to modify the
solution reactivity by the use of derivatives of 1– 3
dioxolane, such as 1 –3 dioxolane (the 6-member ring
acetal), and methyl and dimethyl derivatives of 1– 3
dioxolane. It was logical to assume that both the 6-
D. Aurbach et al. / Solid State Ionics 148 (2002) 405–416
Fig. 2. Typical behavior of Li electrodes in DN/LiAsF6/TBA solutions in Li – LixMnO2 AA batteries at different charging rates. The number of available charge – discharge cycles of
Li – LixMnO2 AA batteries discharged at C/3 h and the average size of Li deposits are plotted as a function of charging rates. Typical SEM micrographs of Li deposits at high and low
charging rates are also presented. (Practical Li anodes from cycled Li – LixMnO2 AA batteries.)
411
412
D. Aurbach et al. / Solid State Ionics 148 (2002) 405–416
member acetal (1 – 4 DN) and methyl substituted 1 –3
dioxolanes should be less reactive towards lithium
than 1 –3 dioxolane, but would still have a similar
surface chemistry: the surface films on Li in these
solvents should also contain elastomers, because these
modified solvents are also acetals that can easily
undergo different routes of polymerization.
None of the above modifications, neither doping
the Li surfaces nor changing the solvent, improved the
performance of Li electrodes or of AA secondary
batteries containing these modifications, at fast charging rates. In fact, it was found that the best performance of AA secondary batteries in terms of cycle life
at any rate was obtained when the anode was pure
lithium and the solvent was 1– 3 dioxolane (which is,
as discussed above, insufficient for practical use since
the competing Li-ion system appeared on the market).
We therefore conclude that for most applications,
secondary Li metal batteries with liquid electrolyte
solutions are no longer practical. Further recent work
by Zaghib et al. [45] may provide a promising direction for the practical use of Li metal-based batteries. Li
metal is highly stable at elevated temperatures with
polymeric electrolytes based on polyethylene oxide
(PEO). At temperatures above 60 jC, quite reasonable
conductivities ( c 10 3 s/cm) can be obtained with
PEO-based electrodes, especially with their branched
derivatives [46]. As the temperature of operation
increases, the lithium metal softens, and this has a
very positive influence on dendrite-free lithium deposition. Consequently, Li batteries comprised of Li
metal, PEO-based electrolytes, and cathodes such as
Li x V 2 O 5 or Li x FePO 4 ( c 3 and 3.5-V systems,
respectively) can be cycled many hundreds of times
at elevated temperatures, showing excellent capacity
retention and stability.
This performance opens the way for applications
such as electrical vehicles (as part of a hybrid system),
use in space, load leveling, etc., where operation at
high temperatures is tolerable. The use of Li metalbased battery systems may then be advantageous in
terms of cost and high energy density.
3.2. Graphite electrodes for Li-ion batteries
We studied a matrix of several types of graphitic
carbons and a variety of electrolyte solutions, including ethers (THF and glymes), esters (butyrolactone
and methyl formate), and alkyl carbonates (EC, PC,
DMC, DEC, EMC with salts such as LiAsF6, LiPF6,
LiBF4, LiN(SO2CF3)2, and LiC(SO2CF3)3).
Fig. 3 shows schematically three types of morphologies of the graphite particles that we used:
1.
2.
3.
synthetic or natural graphite flakes;
spheric graphite particles, the so-called mesocarbon microbeads (MCMB);
graphite fibers.
The graphite flakes are characterized by relatively
wide x –y dimensions (2 –50 Am) in parallel to the basal
planes [22 – 27], and a much smaller z dimension,
Fig. 3. A schematic presentation of the morphology of three types of graphite particles: synthetic graphite flakes, mesocarbon microbeads
(MCMB), and graphitic carbon fibers.
D. Aurbach et al. / Solid State Ionics 148 (2002) 405–416
perpendicular to the basal planes (along the edge planes
of the flakes, the flakes may be even less than 1 Am
thick). Very important is the fact that the surface of the
edge planes (i.e., the facets perpendicular to the basal
planes) is much higher than the geometric surface area.
It contains many entries, as illustrated in Fig. 3. In
contrast, the surface area of both the MCMB and the
graphite fibers is smoother. In contrast to the parallel
order of the basal planes in the graphite flakes, there is
some significant, unavoidable degree of turbostatic
disorder in the spherical and fibrous graphite particles.
We found a pronounced difference in the behavior of
graphite electrodes comprised of graphite flakes compared with that of electrodes comprised of MCMB or
fibers. This difference could be clearly probed, using
solutions containing propylene carbonate as a major
component. Both MCMB or graphite fiber electrodes
can intercalate highly reversibly with lithium in PC
solutions, while electrodes comprised of graphite
flakes do not behave reversibly in PC solutions. Pronounced surface reactions are recorded in the latter
case, which lead to deactivation of the electrodes.
We explored intensively the behavior of graphite
electrodes in PC solutions and found that they are
deactivated not because the graphite particles are
destroyed by surface processes (e.g., exfoliation of
the graphene planes), but rather, because the graphite
particles become electrically isolated from the current
collector due to surface reactions and precipitation of
relatively thick surface films [26]. It is both important
and interesting to note that such a deactivation of
graphite electrodes does not occur with MCMB or
fibers, or when the major solvent is EC. EC is very
similar in its chemical structure to PC, except for the
extra methyl group of the latter compound. Moreover,
both solvents undergo very similar reduction paths on
both Li and Li carbon surfaces:
PC þ 2e þ 2Liþ
! CH3 CHðOCO2 LiÞCH2 ðOCO2 LiÞ #
þCH3 CH ¼ CH2 z
ð1Þ
EC þ 2e þ 2Liþ
! CH2 ðOCO2 LiÞCH2 OCO2 Li #
þCH2 ¼ CH2 z:
ð2Þ
413
We explain these results as follows: most of the
alkyl carbonate solvents, as well as salt anions such as
AsF
6 , PF6 , and N(SO2CF3)2 , are reduced on graphite
surfaces in the presence of Li-ions at potentials below
1.5 V (the major reduction processes usually occur
around 1 V), and continue gradually as the potential is
lowered. Reduction of the cyclic alkyl carbonates also
forms gases (see Eqs. (1) and (2)). In addition, in
parallel to the surface film formation, Li intercalation,
as well as cointercalation of solvent molecules, can
occur at potentials below 1 V (e.g., formation of dilute
stage 1, LixC6, x < 0.25) [5]. Cointercalation of solvent
molecules leads to exfoliation of graphite [5], while gas
formation between graphene planes and the build-up of
pressure within the particles can quite easily crack and
split them. The question is, what happens more
quickly: cointercalation or build-up of internal pressure
within the particles due to surface reactions, or fast
precipitation of surface films that protect the particles
from internal destructive processes.
EC is highly reactive towards lithium, and its
reduction products, which contain two Li –carbonate
groups, precipitate quickly and form cohesive, highly
passivating surface films. PC is similarly reduced on
the graphite surfaces, but the extra methyl group of its
reduction products interferes badly with the formation
of sufficiently cohesive surface films. Thus, the surface film formation in PC solutions cannot compete
successfully enough with the possible destructive
processes addressed above. Hence, the surface area
of the graphite particles increases during the PC
reduction processes by some exfoliation and cracking
of the particles due to a build-up of internal pressure
(CH3CH = CH2 gas formation as a coproduct of the
ROCO2Li species). The cracking of graphite particles
is the most critical destructive process in these systems, because it allows further surface reactions on the
cracked particles, which leads to their complete electrical isolation from the current collector, due to the
complete coverage by passivating surface films. However, the importance of the scenario described above,
depends primarily on the morphology of the graphite
particles, especially on the shape of the edge planes
(the facets perpendicular to the basal planes). In the
case of MCMB or fibers, the edge planes are relatively smooth, and thereby, a build-up of pressure due
to gas formation inside the particles is largely avoided.
The turbostatic disorder in these particles further helps
414
D. Aurbach et al. / Solid State Ionics 148 (2002) 405–416
to prevent easy cracking and exfoliation, because the
graphene planes are better bound together. In the case
of the flakes, the edge planes are very rough and have
many entries in which gas bubbles can be formed,
trapped between the graphite mass and solid surface
species which are continuously formed at the low
potentials.
In addition to the above-described deterioration
mechanisms of flaky graphite electrodes, we identified destructive mechanisms in which the graphite
particles are completely exfoliated, due to solvent
cointercalation with the Li-ions. This usually happens
with ethers [13]. Polarization of graphite electrodes in
Li salt solutions of ethers such as THF, 2Me –THF,
and polyethers of the glyme family lead to an exfoliation of graphite particles at potentials below 0.2 V (Li/
Li+ ) [13,26]. Exfoliation of graphite electrodes at Li
insertion potentials ( > 300 mV vs. Li/Li+ ) also occurs
in Li salt solutions of esters such as g-butyrolactone
and methyl formate [25]. As mentioned above, highly
reversible behavior of composite graphite electrodes is
obtained in LiClO4, LiAsF6, LiPF6, and LiN(SO2
CF3)2 solutions in mixtures of EC and linear alkyl
carbonate solvents from the DMC, EMC, DEC, etc.,
list, due to formation of highly passivating and stable
surface films when EC is a major component of the
electrolyte solution [13,35]. However, upon prolonged
cycling of practical Li battery systems, their capacity
gradually fades, due to an increase of the electrode’s
impedance, especially that of the carbon anodes.
Recent studies of composite graphite electrodes by in
situ AFM [36] clearly showed that Li insertion– deinsertion into composite graphite electrodes leads to
periodic volume changes in the graphite particles.
While the basic morphology of the graphite particles
and the surface films that cover them are very stable
upon repeated charge–discharge cycling in EC-based
solutions (DMC, DEC, or EMC as co-solvents, LiAs
F6, or LiPF6 electrolytes), the particles expand during
Li insertion and contract during Li deinsertion, as
could be imaged in situ by AFM [36].
Hence, we assume that the surface films usually
formed on graphite electrodes may not be able to
fully accommodate the volume changes of the graphite particles due to intercalation with lithium. Therefore, there are continuous, small scale reactions between solution species, which percolate through the
surface films with the highly reactive lithiated graph-
ite. These reactions form insoluble surface species that
gradually thicken the surface films and lead to the
observed gradual increase of the impedance of graphite electrodes upon prolonged Li insertion– deinsertion
cycling.
4. Conclusion
Lithium electrodes fail in rechargeable Li battery
systems upon fast charging because at high rates of Li
deposition, small Li grains are formed and the number
of simultaneously growing gains on the electrode
surface is higher. The passivation of lithium metal
by surface films in any polar aprotic non-aqueous
electrolyte solution can never be hermetic. Thus, there
are unavoidable reactions between the active metal
and solution species, which intensify as the deposited
Li grains become smaller (i.e., high average surface
area of the active metal). Hence, in practical secondary Li batteries charged at rates that are too fast, the
cycle life obtained is very limited, because the entire
electrolyte solution of the battery may react with the
lithium anode.
In the case of graphite anodes in Li-ion batteries,
we can define three cases.
(1) A capacity-fading mechanism due to exfoliation of the graphite particles because of cointercalation of solvent molecules together with Li-ions. This
failure mechanism is relevant to ethereal solutions of
Li salts.
(2) A capacity-fading mechanism in which the
active mass retains its basic structure. However,
graphite particles are electrically isolated from the
current collector due to coverage by surface films
(reaction products of the reduction of solution species
in the presence of Li-ions, by cathodically polarized
graphite). This mechanism, related to electronic isolation of the active mass, depends on the morphology
of the particles and possible gas formation in the
reactions between the reactive electrode’s surface
and solution species. This situation is relevant to
propylene carbonate solutions.
(3) A stable situation in which highly passivating
surface films with very good cohesion are formed
before detrimental processes related to gas formation
between graphene planes or solvent cointercalation
can take place. This situation is relevant to EC so-
D. Aurbach et al. / Solid State Ionics 148 (2002) 405–416
lutions and solutions containing reactive additives
such as CO2, SO2, ethylene sulfite, etc., which react
predominantly on the electrode’s surface, thus forming highly protective surface films. However, in all of
these solutions (even in the ‘good’ ones), there are
continuous small scale surface reactions between
lithiated graphite and solution species that lead to a
gradual increase in the electrode’s impedance upon
cycling. These small-scale reactions take place because of unavoidable volume changes in the graphite
particles during lithium insertion– deinsertion cycles.
[12]
[13]
[14]
[15]
[16]
[17]
[18]
Acknowledgements
[19]
[20]
Partial support for the work described in this paper
was obtained from the BMBF, the German Ministry of
Science, in the framework of the DIP project, from the
Israeli Ministry of Science and Technology, in the
framework of the Infrastructure Research Program,
and from the Israel National Foundation of the Israel
Academy of Science.
[21]
[22]
[23]
[24]
References
[25]
[1] L. Dominey, in: D. Aurbach (Ed.), Non Aqueous Electrochemistry, Marcel Dekker, New York, 1999, pp. 437 – 460, Chap. 8.
[2] P. Dan, E. Mengeritsky, Y. Geronov, D. Aurbach, I. Weissman,
J. Power Sources 54 (1995) 143.
[3] Y. Nishi, in: M. Wakihara, O. Yamamoto (Eds.), Lithium Ion
Batteries. Fundamentals and Performance, Wiley-VCH, Weinheim, 1998, pp. 181 – 198, Chap. 8.
[4] K. Nagajima, Y. Nishi, in: T. Osaka, M. Datta (Eds.), Energy
Storage Systems for Electronics, Gordon and Breach Science
Publisher, Singapore, 2000, pp. 109 – 129, Chap. 5.
[5] M. Winter, J.O. Besenhard, M.E. Spahr, P. Novak, Adv. Mater.
10 (1998) 725.
[6] D. Aurbach, in: D. Aurbach (Ed.), Non Aqueous Electrochemistry, Marcel Dekker, New York, 1999, pp. 289 – 409, Chap. 6.
[7] E. Peled, in: J.P. Gabano (Ed.), Li Batteries, Academic Press,
London, 1983, pp. 43 – 72, Chap. 3.
[8] K.M. Abraham, S.B. Brummer, in: J.P. Gabano (Ed.), Lithium
Batteries, Academic Press, London, 1963, pp. 371 – 406,
Chap. 14.
[9] J.R. Dahn, A.K. Sleigh, H. Shi, B.M. Way, W.J. Weydanz, J.N.
Reimers, Q. Zhong, U. von Sacken, in: G. Pistoia (Ed.), Lithium Batteries, New Materials, Developments and Perspectives,
Elsevier, Amsterdam, 1994, p. 1, and references therein.
[10] D. Aurbach, I. Weissman, A. Zaban, Y. Ein-Eli, E. Mengeritsky, P. Dan, J. Electrochem. Soc. 143 (1996) 2110.
[11] D. Aurbach, A. Zaban, Y. Ein-Eli, I. Weissman, O. Chusid, B.
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
415
Markovsky, M. Levi, E. Levi, A. Schechter, E. Granot, J.
Power Sources 86 (1997) 91.
D. Aurbach, B. Markovsky, M.D. Levi, E. Levi, A. Moshkovich, M. Moshkovich, Y. Cohen, J. Power Sources 81 (1999)
95.
D. Aurbach, B. Markovsky, K. Gamolsky, E. Levi, Y. Ein-Eli,
Electrochim. Acta 45 (1999) 67.
O. Chusid, D. Aurbach, J. Electrochem. Soc. 140 (1993) L1.
D. Aurbach, Y. Cohen, J. Electrochem. Soc. 143 (1996) 3525.
D. Aurbach, A. Zaban, J. Electrochem. Soc. 142 (1995)
L108.
D. Aurbach, E. Zinigrad, A. Zaban, J. Phys. Chem. 100 (1996)
3098.
P. Dan, E. Mengeritsky, I. Weissman, E. Zinigrad, J. Power
Sources 68 (1997) 443.
D. Aurbach, I. Weissman, A. Zaban, P. Dan, Electrochim. Acta
45 (1999) 1135.
D. Aurbach, E. Zinigrad, H. Teller, P. Dan, J. Electrochem.
Soc. 147 (2000) 2486.
E. Peled, D. Golodnitsky, J. Penciner, in: J.O. Besenhard (Ed.),
Handbook of Battery Materials, Wiley-VCH, Weinheim, 1999,
pp. 419 – 456, Part III, Chap. 6.
R. Yazami, in: G. Pistoia (Ed.), Lithium Batteries, New Materials, Developments and Perspectives, Elsevier, Amsterdam,
1994, p. 49, and references therein.
D. Aurbach, Y. Ein-Eli, O. Chusid, M. Babai, Y. Carmeli, H.
Yamin, J. Electrochem. Soc. 141 (1994) 603.
D. Aurbach, O. Youngman Chusid, Y. Carmeli, M. Babai, Y.
Ein-Eli, J. Power Sources 43 (1993) 47.
D. Aurbach, Y. Ein-Eli, B. Markovsky, Y. Carmeli, H. Yamin,
S. Luski, Electrochim. Acta 39 (1994) 2559.
D. Aurbach, M.D. Levi, E. Levi, A. Schechter, J. Phys. Chem.
B 101 (1997) 2195.
D. Aurbach, B. Markovski, A. Rodkin, M. Cojocaro, H.J.
Kim, An analysis of rechargeable Lithium ion batteries after
prolonged cycling, (2002), In press.
E. Goren, O. Chusid, D. Aurbach, J. Electrochem. Soc. 138
(1991) L6.
D. Aurbach, M. Moshkovich, J. Electrochem. Soc. 145 (1998)
2629.
A. Zaban, D. Aurbach, J. Electroanal. Chem. 348 (1993) 155.
Y. Cohen, D. Aurbach, Rev. Sci. Instrum. 70 (1999) 4668.
D. Aurbach, Y. Gofer, Y. Langzam, J. Electrochem. Soc. 136
(1989) 3198.
D. Aurbach, I. Weissman, H. Yamin, E. Elster, J. Electrochem.
Soc. 145 (1998) 1421.
O. Youngman, Y. Gofer, A. Meitav, D. Aurbach, Electrochim.
Acta 35 (1990) 625, 639.
D. Aurbach, B. Markovsky, A. Schechter, Y. Ein-Eli, H. Cohen, J. Electrochem. Soc. 143 (1996) 3809.
D. Aurbach, M. Koltypin, The study of Li insertion – deinsertion into composite graphite electrodes by in situ AFM, Electrochem. Com. 4 (2002) 17.
M. Vorotyntsev, M.D. Levi, A. Schechter, D. Aurbach, J. Phys.
Chem. B 105 (2001) 188.
K. Kanamura, T. Tamura, S. Shiraishi, Z. Takehara, J. Electroanal. Chem. 394 (1995) 49.
416
D. Aurbach et al. / Solid State Ionics 148 (2002) 405–416
[39] D. Aurbach, Y. Cohen, J. Phys. Chem. B 104 (51) (2000)
12282.
[40] D.P. Wilkinson, D. Wainwright, J. Electroanal. Chem. 355
(1993) 193.
[41] T. Hirai, I. Yoshimatsu, J.I. Yamaki, J. Electrochem. Soc. 141
(1994) 611.
[42] E. Zinigrad, D. Aurbach, Electrochim. Acta 46 (2001) 1836.
[43] a) C.D. Desjardins, G.K. Maclean, H. Sharibia, European Patent Appl. 68301752.7, publication number 0281352, 1987;
b) C.D. Desjardins, T.G. Green, G.K. Maclean, J. Power Sources 12 (1988) 489.
[44] N. Kumagai, Y. Kikuchi, K. Tanno, J. Appl. Electrochem. 22
(1992) 620.
[45] K. Zaghib, M. Armand, M. Gautier, J. Electrochem. Soc. 145
(1998) 3135.
[46] M. Kono, E. Hayashi, M. Vishiura, M. Watanabe, J. Electrochem. Soc. 147 (2000) 2517.