J Solid State Electrochem (2017) 21:3269–3279

DOI 10.1007/s10008-017-3671-7

ORIGINAL PAPER

Effect of overdischarge (overlithiation) on electrochemical

properties of LiMn2O4 samples of different origin
V. V. Kosilov 1 & A. V. Potapenko 1 & S. A. Kirillov 1

Received: 20 March 2017 / Revised: 29 May 2017 / Accepted: 31 May 2017 / Published online: 18 June 2017
# Springer-Verlag GmbH Germany 2017

Abstract Overdischarge poses a significant threat to the extra lithium ions can be electrochemically inserted into
safety and reliability of lithium ion batteries. In this paper, the 16с voids in the spinel structure resulting in the
overdischarge (overlithiation) effect on electrochemical Li1.5Mn2O4 compound, which is able to reversibly work
parameters of lithium manganese spinels of different origin within the 2.4–4.6 V potential range.
is studied. Charge/discharge tests for 240 cycles are per-
formed within recommended and lowered potential ranges.
Keywords Lithium ion batteries . Cathode materials . Lithium
Insertion of lithium ions into spinels by decreasing poten-
manganese spinels . Overdischarge . Overlithiation
tial to either +2.4 or +1.5 V against Li/Li+ is found to lead
to the formation of the overlithiated Li2-xMn2O4 spinel
phase. Repeated cycling involving the low-potential re-
gions worsens electrochemical parameters: cell resistance Introduction
increases and specific capacity decreases. It appears that
even small overdischarge is totally unacceptable for the Overdischarge, i.e., overreaching certain minimum allow-
nanosized home-made samples and microsized EQ–Lib– able voltage region upon discharge of a lithium battery is
LMO (MTI) samples. The microsized BE–30 (NEI) sample considered to result in the irreversible degradation of
better tolerates cycling at lowered potentials. The greatest cathode materials up to their complete destruction. In or-
resistance to overdischarge demonstrates the aluminum- der to avoid this phenomenon, lithium batteries are usual-
doped microsized HPM–9051C (Toda) sample. Being ly cathode limited after assembly, viz., the anode capacity
treated at lowered voltages, this material acquires a steady is available in excess, being in the range of ca. 10–20% in
trend towards an increase in specific capacity by half. This capacity [1, 2], and cell balancing control schemes are
may signify that at least half of the gram equivalent of employed in many commercial batteries [3]. If a failure
in a battery equalizer occurs or an occasional battery
asymmetry happens, severe problems arise for batteries
with cells subjected to overdischarge or overcharge [4].
Highlights
Possible consequences of these issues are settled by user
• Insertion of Li+ into LiMn2O4 by decreasing potential to 2.4 Vor 1.5 V is
studied. normative acts [5] and International Electrotechnical
• Even small overdischarge is totally unacceptable for nanosized samples. Commission standards [6].
• Microsized samples better tolerate cycling at lowered potentials. Overdischarge phenomena have been studied for batte-
• Al-doped microsized samples demonstrate the greatest resistance to
ries with LiCoO2 [7], LiFePO4, LiNiO2, and LiMn2O4
overdischarge.
cathodes [8]. It should be mentioned that some authors
* S. A. Kirillov
question the destructive effect of overdischarge on batte-
[email protected] ries with lithiated oxide cathodes. Say, a conclusion
drawn for cells with LiCoO2 cathodes [9] states that de-
1
Joint Department of Electrochemical Energy Systems, 38A creasing voltage to below than 1.5 V causes no significant
Vernadsky Ave., Kiev 03680, Ukraine capacity losses.
3270 J Solid State Electrochem (2017) 21:3269–3279

If the lithium manganese spinel LiMn2O4 is used as a to a fast capacity loss (from 814 to 452 mAh g−1 for 100 cycles
cathode, the electrochemical intercalation/deintercalation [19]). Such extremely deep overdischarge leads to the
of lithium ions into/from the lattice occurs at about 4 V, amorphization of LiMn2O4 and to the fast descent of its elec-
trochemical parameters [8, 20].
LiMn2 O4 ‐xLiþ ‐xe‐ ⇄Li1‐x Mn2 O4 ð1Þ The formula of Li2Mn2O4 formed in the process of electro-
chemical lithiation of LiMn2O4 can be formally written as
In the composition ranges of 0 ≤ х ≤ 0.5 and 0.5 ≤ х ≤ 1.0, the
LiMnO2. Numerous attempts have been made so as to synthe-
reaction takes place in two stages, but the spinel keeps its cubic
size LiMnO2, which is similar to Li2Mn2O4 by composition
structure (space group Fd3m, Z = 8) [5, 10, 11] with the lattice
and differs in structure, by a non-electrochemical route [18,
parameter а = 8.248 Å and the unit call volume V = 539.75 Å3.
20–24], and to use it as a cathode material instead of expensive
These stages are well observed on galvanostatic curves, which
and toxic LiCoO2. It appears however that Li1+хMn2O4 mate-
consist of two separated flat regions, and on cyclic voltammetry
rials cannot be obtained in air since manganese tends to be
curves having two typical cathodic and anodic peaks corre-
sponding to the intercalation/deintercalation of lithium ions. oxidized forming Li 2О∙yMnO2 systems, e.g., Li2MnO 3
(y = 1) of rock-salt structure and Li2O∙yMnO2 (2.5 ≤ у ≤ 4.0)
The theoretical capacity of LiMn2O4 according to Eq. (1)
of spinel structure [18, 25]. Furthermore, orthorombic LiMnO2
equals to 148 mAh g−1.
In 1983, Thackeray et al. [12] have shown an ability of (similar to layered α–NaFeO2, the Pmnm space group) dem-
onstrates better cycling ability than LiMn 2O 4 charged/
LiMn 2 O 4 to insert lithium ions into its lattice. In the
discharged in accord with reactions (1) and (2) [23, 24]. On
LiMn2O4 structure, oxygen ions occupy the tetrahedral posi-
tions (32e) and form the cubic dense packing. The octahedral the other hand, X-ray studies demonstrate the irreversible
transformation of the layered LiMnO2 to a spinel-like structure
positions (16d) are inhabited by manganese ions (Mn3+/
just upon the first charging [18, 26]. This is consistent with the
Mn4+). Lithium ions hold the tetrahedral (8a) positions. Each
of them is separated from four neighboring ions by voids emergence of two plateaus on galvanostatic curves at 4.0 and
2.9 V upon cycling.
(16c). This means that three-dimensional channels (8a–16c–
A common problem of lithium-manganese spinel cathodes
8a–16c) exist for possible insertion and transport of lithium
is their interaction with an electrolyte causing manganese
ions [12, 13], and extra lithium ions can be electrochemically
losses through disproportionation reaction [27],
inserted into the 16с voids [13, 14] randomly filling half of
them and resulting in the Li2Mn2O4 (LiMnO2) compound
according to Eq. (2), 2Mn3þ ðsolidÞ →Mn4þ ðsolidÞ þ Mn2þ ðsolutionÞ ð3Þ

Jang and Oh have determined [28] that disproportionation

LiMn2 O4 þ xLiþ þ xe− ⇄Li1þx Mn2 O4 ; where 0 < x < 1:ð2Þ
significantly depends on the composition and acidity of the
Its unit cell expands by 6.2% [12], and the unit cell edge electrolyte. Mn2+ dissolution results in the loss of cathode
ratio a/c (а = 8.007 Å, с = 9.274 Å) increases by 16% [15] material, influences anode material, and accelerates on rising
compared with the parent cubic structure (а = 8.248 Å). The temperature.
newly formed tetragonal structure belongs to the I41/amd Thackeray and co-workers stress that upon long-term cy-
space group, Z = 2 [13]. This transformation is determined cling, especially at small overdischarge, the surface of
by the Jahn–Teller effect [12, 16, 17]. LiMn2O4 particles enriches with lithium. This entails the
The theoretical capacity of LiMnO2 is 285.4 mAh g−1, i.e., Jahn–Teller transformation, tetragonal Li2Mn2O4 emerges,
is almost twice as great as that of LiMn2O4. The process de- and finally Li2MnO3 and soluble MnO are forming through
scribed by Eq. (2) is showing up as one cathodic and one the disproportionation reaction (3) [15, 29] as
anodic peak at ~2.9 V on cyclic voltammetry curves [14]
and as a flat region on galvanostatic curves [11]. The cycling Li2 Mn2 O4 →Li2 MnO3 þ MnO ð4Þ
ability of LiMn2O4-based electrodes in the low-voltage re-
gion, as corresponds to Eq. (2), is poor [17]. This is explained This means that upon long-term cycling, a combined
by the electrode degradation due to asymmetric expansion/ overdischarge and dissolution effects transform the better part
contraction of Li1+хMn2O4 upon charging/discharging and of electrode material from LiMn2O4 to Li2MnO3 well discern-
by the Jahn–Teller effect [14, 18]. Some other causes of deg- ible in XRD studies.
radation of spinel cathodes to be described in what follows Substituting a part of manganese in LiMn2O4 with polyva-
may also add to this phenomenon. lent cations (Li, Zn, Mg, etc.) causes the rising of the oxidation
Further lithium insertion at voltages lower than 1.5 V leads degree of manganese. This results in the decrease in the
to the reduction of Mn3+ to Mn0. This process manifests itself amount of unstable Mn3+ and material acquires resistance to
by two plateaus on galvanostatic curves at 0.1–3.0 Vand leads the disproportionation reaction (3) [23, 24, 30]. A striking
J Solid State Electrochem (2017) 21:3269–3279 3271

example of cationic substitution is doping LiMn2O4 with alu- Overdischarge studies of spinels may cast some light on phe-
minum. As shown in Refs. [31, 32], changing manganese by nomena occurring upon such processes.
aluminum decreases the unit cell parameters and improves the
working ability of the cathode material (decreases capacity
losses upon cycling) compared with the neat LiMn2O4. Experimental
Replacing Mn3+ with Al3+ strengthens Mn–O bonds in spi-
nels, and a two-phase behavior upon cycling (peak separation Materials
according to Eq. (2)) becomes less vivid [33, 34]. All authors
unanimously agree with enhancing the cycling ability of spi- A sample synthesized in our laboratory by means of a citric
nels by means of doping, in spite of some losses in capacity acid route [41, 42] (designated hereinafter as JDEES) and
arising from electrochemical inactivity of dopants. commercial materials EQ–Lib–LMO by MTI, BE–30 by
Summarizing, one can conclude that the formation of Li1+ NEI, and HPM–9051C by Toda have been studied in this
хMn 2O4 in modern lithium ion batteries is a negative phenom- work. Physico-chemical and electrochemical characteristics
enon, and overdischarge significantly increases the likelihood of these electrode materials are summarized in Table 1. In
of its occurrence. It may be caused by mishandling, a failure of particular, it is worth noticing that the JDEES sample is
cell balancing control schemes and/or entire battery manage- nanosized, whereas the commercial MTI, NEI, and Toda sam-
ment system, self discharge, etc. Overdischarge involves such ples are microsized materials. It should be also mentioned that
adverse events as capacity decrease, increase in internal resis- the JDEES, MTI, and NEI materials are stoichiometric spi-
tance up to internal short circuit, case swelling, rupture, or nels, and the Toda sample contains 1.52 wt.% of aluminum.
damage [7, 9, 35, 36]. It should be mentioned, however, that
there is no data in literature describing more or less extended
cycling of spinel electrodes within a wide voltage range (2.5– Procedures and measurements
4.5 V) where reactions (1) and (2) flow, thereby simulating
overdischarge phenomenon. Sample 2016 coin cells with a lithium metal anode serving as
Therefore, the primary aim of this work is looking for gen- a counter and reference electrode were studied. A working
eral regularities of the overstoichiometric electrochemical electrode was made by means of coating an aluminum foil
lithiation of ordinary and aluminum-substituted lithium-manga- with a slurry prepared in an RW 20 (IKA, Germany) high-
nese spinels and studying the impact of overlithiation speed mixer and containing the material in question, Super P
(overdischarge) on their further behavior in the high-voltage and SFG6 (Timcal) conducting additives, and a binder
region. In order to reach this goal, four samples of different (Kureha #1100 poly(vinylidenedifluoride)) with the dry com-
origin, including commercial ones, have been tested. In practi- ponent ratio of 80:7:5:8 suspended in N–methylpyrrolidone
cal terms, long-term cycling these materials in a wide voltage (Sigma–Aldrich). The quantity of LiMn2O4 in a remainder
range will enable one to conclude regarding their overdischarge obtained after drying under an IR radiator was of 4–
resistance and possible response to emergency situations. 5 mg∙cm−2. Cells containing electrodes divided by a Celgard
Clearly, overdischarge studies would look more complete 2500 separator membrane impregnated with a 1 mol L−1 so-
if tests in the full-cell configuration using commercially avail- lution of LiPF6 in a mixture of ethylene carbonate/dimethyl
able graphite negative electrode were performed. Regretfully, carbonate (1:1 by mass) were assembled in a dry glove box.
in the potential region we intended to probe, this idea is diffi- Electrochemical measurements were performed on a
cult to implement. Being overdischarged to around 1.5 V, the home-made automated electrochemical workstation operated
full cell containing a carbonaceous anode rolled onto a copper in cyclic voltammetry (CVA, 0.1 mV S−1 scanning rate), gal-
foil would inevitably experience severe erosion due to the vanostatic (constant current, CC), and combined (constant
anodic dissolution of the substrate with further crumbling of current–constant voltage, CCCV) modes at 25 °C. More de-
the anode material up to short circuit. Therefore, instead of full tails regarding working regimes are presented in respective
cells, we confined ourselves with sample cells containing lith-
Table 1 Characteristics of samples studied
ium anode, which enable us to operate within a very broad
potential range. Material JDEES MTI NEI Toda
Another point to be considered here is the overstoichiometric
Recommended working 3.4–4.5 3.4–4.5 3.0–5.0 3.0–4.3
lithiation of common cathode materials for giving them an an- range (V)
ode function [37, 38]. The first study in this field has been Nominal capacity (mАh/g) 100 115 100 96
presented by Tarascon and co-workers [39]. Later on, it has been Crystallite size (nm) 30 300 300 300
found that obtaining overstoichiometric lithiated oxides by Particle size (μm) 0.1 12 0.34 10
means of electrochemical lithiation of cathode materials [40] is Specific surface (m2/g) 6.3 0.4–1.0 4.1 0.51
advantageous over electrochemical lithiation of oxides.
3272 J Solid State Electrochem (2017) 21:3269–3279

sections describing preliminary low-rate deep overdischarge LiMn2O4 [10, 11, 13, 14, 17–19, 21–25, 29, 34–36, 41, 42],
CVA tests and high-rate overdischarge galvanostatic tests. namely two waves are present corresponding to a stepwise
Parameters of discharge expressed in C [mA g−1] values were intercalation/deintercalation of 0.5 g ions of lithium, Fig. 1.
calculated based on the theoretical discharge capacity of Another wave at potentials above 4.4 V is observed for the
LiMn2O4 equal to 148 mAh g−1. Internal resistance RDС NEI sample. It adds of about 20% to the overall specific ca-
[Оhm] of cells was calculated from measurements of the value pacity and can be explained by the presence of local structure
of potential drop Udrop [V] upon charge/discharge switching defects [24]. For the aluminum-doped Toda sample, cathodic
[43, 44] within the time of 10 mS as and anodic CVA peaks are less sharp and more overlap than for
other lithium-manganese spinels [33, 34]. Average values of
U drop
RDC ¼ ð5Þ specific capacities are 84.6, 113.4, 96.7, and 90.4 mAh g−1 or
I JDEES, MTI, NEI ,and Toda, respectively, Fig. 2.
where І [A] is the value of discharge current. Further electrochemical tests have been executed to/from
lower potentials (2.4–3.5 V, 6–15 cycles) necessitating reac-
tion (2) to occur. CVA curves show up a reversible peak in the
cathode and anode region at the potential of around 2.9 V,
Results and discussion which corresponds to the Mn4+↔Mn3+ oxidation/reduction
process and the transition of the cubic phase to the tetragonal
Preliminary low-rate deep overdischarge CVA tests one [10–12, 16, 17] (Fig. 1). In this case, in accord with spe-
cific capacity values, the JDEES sample reveals the ability to
In these experiments, the scan rate upon cyclic voltammetry absorb of about 1 g equivalent of lithium ions, Fig. 2. Unlike
experiments for all four electrodes has been settled as 0.1 mV this sample, other ones (MTI, NEI, Toda) insert/deinsert
S−1 in all ranges of potentials studied. Such regime is the most ~0.5 g-equivalent of lithium ions. In addition, a gradual de-
close to stationary conditions. Electrochemical tests have cov- crease of capacity is fixed for the JDEES and MIT samples,
ered recommended (Table 1) and lowered charge/discharge similar to that noticed in Ref. [17].
potential ranges. Specifically, 1–5, 15–20, and 26–30 cycles After tests in a reduced potential range, electrode materials
have been obtained in the recommended potential range, have been investigated at the recommended potentials (> 3 V,
whereas 6–15 and 21–25 cycles have been registered at 2.4– 16–20 cycles), Figs. 1 and 2. Unlike the MTI, NEI, and Toda
3.5 V and 1.5–3.5 V, respectively. Representative CVA curves samples, which recover their initial capacities almost
are shown in Fig. 1; respective dependences of the specific completely, a loss of capacity is observed for the JDEES sam-
capacity on the cycling potential range and the number of ple. It should be mentioned that the shape of CVA curves for
cycle are given in Fig. 2. commercial samples is changed insignificantly compared with
Cycling electrodes in the recommended ranges of potentials fifth cycle (Fig. 1), their profile remains constant and the
(> 3 V, 1–5 cycles) causes reaction (1) with a capacity value height of peaks lower insignificantly.
close to the manufacturers data listed in Table 1. Herewith, for Then electrochemical tests in a range of potentials of 1.5–
JDEES and MTI samples, CVA characteristics are typical of 3.5 V have been carried out simulating a deep overdischarge

Fig. 1 CVA data for LiMn2O4

samples of different origin
J Solid State Electrochem (2017) 21:3269–3279 3273

Fig. 2 Dependence of the

specific charge/discharge capacity
of LiMn2O4 samples of different
origin on the number of cycle
within the recommended voltage
range and upon overdischarge

(21–25 cycles). In this case, insertion/deinsertion of ~0.75 g doped lithium manganese spinels and requires additional
ions of lithium is observed for the JDEES sample. This is less physical, chemical, and electrochemical studies.
than in first overdischarge cycles within 2.4–3.5 V and can be Furthermore, a cathodic wave at ~3.25 V with no anodic
considered as direct evidence of degradation of the nanosized counterpart emerges for the NEI sample (see 21st cycle,
material. The МТІ and NEI samples intercalate/deintercalate Fig. 1). Its height depends on deinsertion depth and becomes
~0.5 and ~0.7 g ions of lithium, respectively, as the same negligible at charging to 4.1 V. Probably, the reverse process
amount as in case of the first overdischarge. Unlike other for this wave lies far in high-voltage region, i.e., above 4.1 V,
samples, Toda absorbs/releases up to 1.25 g ions of lithium, but its origin calls for further investigations.
i.e., much more than at the first overdischarge cycling. Finally, the last cycling in the recommended ranges of poten-
These peculiarities are well seen in CVA curves (Fig. 1). tials has been performed (26–30 cycles). In this case, the specific
Deep lithiation/delithiation is reflected in the shift of cathodic capacity of the nanosized JDEES sample falls down significant-
peak to lower potentials and anodic peak to higher potentials ly, i.e., overdischarge appears detrimental to it and leads to a
for all samples. Compared with the processes occurring at profound degradation. As for the microcrystalline MTI and
lowering potentials to 2.5 V, deep overdischarge to 1.5 V NEI samples, their capacity decreases not more than by 10–
causes decreasing current peaks for the JDEES and MTI sam- 15%, and for the Toda sample, the capacity loss is less than 8–
ples and increasing for the NEI and Toda samples (cf. 15 and 9% (Fig. 2). Even at the last cycle (Fig. 1, cycle 30), CVA curves
25 cycles in Fig. 1). An increase in peak currents has been retain their typical shape inherent to lithium-manganese spinels.
observed in Refs. [24, 45] for cathodes made of orthorhombic A significant reduction in peak currents for the JDEES sample
LiMnO2 and interpreted as showing signs of the transforma- reflects its deep degradation. Smaller changes in current peak
tion to the spinel structure upon cycling in the range of 4.2– heights are observed for the MIT, NEI, and Toda samples. This
1.0 V. According to X-ray diffraction data, the plateau at 2.0 V decrease in capacity is presumably caused by increasing the
has been attributed to the formation of lithium-rich metastable resistance of electrodes due to the formation of passive films
Li2MnO2 phase [45]. Interestingly, this effect has not been created on the surface of spinels at low potentials.
found for stoichiometric LiMn2O4 samples. It should be mentioned that in all recommended ranges of
CVA curves signify an increase in capacity for the NEI and potentials, the charge current for each first scan (Fig. 2, 1st,
Toda samples arising from the presence of a washed-up wave 16th, 26th cycles) is significantly higher than the discharge
at 2.0 V, which emerges at the first deep overdischarge cycles current. It may be caused by the formation of a solid electro-
(21st cycle in Fig. 1). On the next cycles, this wave flattens. lyte interface (SEI) destroyed in the course of overdischarge
Such flattening appears more noticeable in the cathodic region cycles and created again upon charging spinels to 4.3–5.0 V.
and entails an increase in intensity of the peak at 3.0 V (25th
cycle, Fig. 1) to the value observed upon cycling within the High-rate overdischarge galvanostatic tests
2.4–3.5 V range (15th cycle, Fig. 1). An effect of a significant
increase in the specific capacity upon widening the working Preliminary low-rate deep overdischarge CVA tests lead to the
voltage region may be associated with structural features of following conclusions. At stationary conditions, the common
3274 J Solid State Electrochem (2017) 21:3269–3279

feature of all samples studied is a transformation showing up Fig. 3. 10th charge/discharge cycle has been chosen so as to
by a characteristic Li+ insertion/deinsertion peak in the 2.9 V show the shape of curves before cycling. The last cycle of
range. All spinels except the JDEEE sample successfully tol- each range (110, 147, 189, and 240) demonstrates the effect
erate potential lowering to 2.4 V and retain their working of multiple overdischarge.
ability. This fact is in line with recommendations regarding All materials studied show up two plateaus at ~4 and
the limiting discharge voltage of 2.0–2.5 V for lithium-ion ~2.9 V. These correspond to the peaks at CVA curves and
batteries [46]. Therefore in further high-rate galvanostatic may be described by Eqs. (1) and (2), respectively.
tests, it would be safe to confine ourselves with the cutoff Qualitatively, all samples except Toda demonstrate fast capac-
potential of 2.4 V. From the point of view of practice, in ity fade upon cycling.
discharged cathode limited cells, at overall voltages around For JDEES and MTI materials, fast rising electrical resis-
0 V, both cathode and anode potentials tend to the same value tance of sample cells occurs upon repeated cycling (Fig. 4).
of ca. 4 V, and an emergency situation should arise so as to Probably this is due to the growth of a SEI layer on LiMn2O4
shift the cathode potential to 2.0–2.5 V [2]. and the products of decomposition of the electrode material.
Galvanostatic tests were organized as follows. Unlike inves- NEI is less sensitive to multiple overdischarge than JDEES
tigations described above, overdischarge was performed in a and MTI, and its resistance increases slower. One may con-
single step (to 2.4 V) instead of a stepwise procedure (first to clude that single discharges are of small destructive effect on
3.0–3.4 V and then to 2.4.V). Before these tests, four condition- cathodes made of the NEI BE–30 material. The stability of
ing galvanostatic cycles by 0.1C current were executed followed specific capacity and resistance of the Toda material upon
by three CVA cycles at 0.1 mV S−1 scanning speed. In further treatment for 240 cycles both in recommended and extended
cycles, charging cells in the CCCV mode was carried out. First, potential ranges evidences that the structure of the aluminum-
1.0 C constant current was applied until the charge voltage doped spinel remains unchanged.
reached the higher recommended value, Table 1. Then cells More detailed conclusions regarding the long-term cycling
were charged by a tapered current at this voltage kept constant. can be made from Fig. 5. It appears that in the first cycling
These charge/discharge conditions are to some extent similar to series (8–75 cycles, recommended voltage range), JDEES and
a real operating mode of lithium-ion batteries. Three series of especially Toda samples demonstrate the best capacity reten-
cycling (8–75, 116–147, and 193–240 cycles) were carried out tion. In absolute values, specific capacity decreases from 109
in voltage ranges recommended by producers, Table 1. Between to 100 mAh g−1 for JDEES, from 118 to 105 mAh g−1 for
them, two series of cycling (80–110 and 153–189 cycles) were MTI, from 80 to 76 mAh g−1 for NEI, and remains stable
done with the discharge potential of 2.4 V. These were preceded (90 mAh g−1) for Toda.
and followed by CVA cycling at 0.1 mV S−1 scanning speed In the second cycling series (80–110 cycles, overdischarge
(76–79, 111–115, 190–192, and 240–243 cycles), so as to con- to 2.4 V), Fig. 5, all samples show the growth in specific
clude regarding possible causes of capacity losses (material deg- capacity corresponding to uptake/release of ~1.5 g equivalent
radation, increase in resistance) on cycling. of lithium ions: 1 Li+ in the recommended voltage range and
Representative charge/discharge characteristics of lithium- ~0.5 Li+ upon overdischarge. For МТІ, NEI, and Toda such a
manganese spinel electrodes of different origin are given in behavior well agrees with that found in preliminary deep

Fig. 3 Charge/discharge curves

upon cycling LiMn2O4 samples
of different origin. For the
description of cycle numbers see
text
J Solid State Electrochem (2017) 21:3269–3279 3275

voltage range, Fig. 5) confirm the degradation trend for the

stoichiometric samples. The capacity fade is fastest for JDEES.
For MTI and NEI it is slower and obeys an approximately linear
dependence on cycle number, which is common to all three
series executed in the recommended voltage range, see Fig. 5.
Unlike these samples, the aluminum-doped Toda sample dem-
onstrates exceptional durability. After 240 cycles, which include
68 overdischarge cycles, it retains 92% of its initial capacity.
Valuable conclusions follow from the analysis of CVA
curves registered before and after galvanostatic tests (Fig. 6).
Slow scan rate (0.1 mV S−1) enables one to perform control
cycles at more equilibrium conditions than those used at
Fig. 4 Resistance of sample cells upon cycling
CCCV mode with 1.0C discharge current. This gives more
realistic charge/discharge capacity estimates and allows for
overdischarge CVA tests. For JDEES this differs from prelim- some speculations regarding the role of resistance.
inary experiments where the sample is able to insert/deinsert As follows from Fig. 6, a gradual decrease in peak values
1 g equivalent of lithium ions in the voltage range of 4.5–3.4 V and current slopes takes place for the JDEES sample upon
and 1 g equivalent of lithium ions in the voltage range of 3.4– cycling, although the very shape of CVA curves remains sim-
2.4 V. We guess such differences may be caused by rising the ilar. Upon overdischarge cycling, specific capacity is not re-
charge/discharge rate. In absolute specific capacity values, cy- covered even at small current loads characteristic of the CVA
cling JDEES demonstrates a rapid drop from 140 to 114 mAh mode. This evidences of the degradation of the cathode mate-
g−1. MTI manifests a growth from 152 to 155 mAh g−1 follow- rial and the growth of the electrode resistance. The MTI sam-
ed by a descent to 135 mAh g−1. NEI shows a slower descent ple tells almost the same story.
from 106 to 97 mAh g−1. As far as the aluminum-modified The behavior of the NEI sample is similar to that of the
Toda sample is concerned, its specific capacity grows by half, JDEES and MTI samples. At the same time, it is clear from
from 90 to 128 mAh g−1, and remains stable upon cycling. Fig. 6 that changes in specific capacity upon cycling are grad-
As expected [18, 24, 26–34], returning to the recommend- ual and demonstrate no jumps after overdischarge.
ed voltage range (third cycling series, 116–147 cycles, Fig. 5), Interestingly, long-term overdischarge cycling does not influ-
reveals the loss in the specific capacity and the degradation of ence the form of extra waves at 3.2 and 4.6 V.
the JDEES and MTI samples. The NEI sample is less affected The analysis of CVA tests of the Toda sample confirms
by overdischarge. The aluminum-modified Toda sample conclusions drawn from galvanostatic cycling. For 180 cycles,
completely reproduces its initial specific capacity. the form of CVA curves does not vary significantly. After this,
Fourth and fifth cycling series (153–189 cycles, some decrease in peak values and current slopes is showing up
overdischarge to 2.4 V, and 193–240 cycles, recommended accompanied by insignificant capacity fade (Fig. 6).

Fig. 5 Galvanostatic life tests for

LiMn2O4 samples of different
origin in the recommended
voltage range and upon
overdischarge
3276 J Solid State Electrochem (2017) 21:3269–3279

Fig. 6 CVA tests for LiMn2O4

samples of different origin in the
recommended voltage range and
upon overdischarge

All these data are summarized in Fig. 7, where the depen- viewpoint, some efforts should be undertaken so as to justify
dence of capacity retention on the cycle number is built for all the use of a lowered discharge voltage for this material in a
four materials studied. As follows from it, non-modified spinels regular operating mode of a lithium-ion battery. However, it
by JDEES and MTI quickly degrade upon overstoichiometric should be asserted with confidence that overdischarges are not
lithiation. These materials can be steadily cycled just in the critical to the normal operation of cathodes made of this
recommended voltage range, without entering the voltage re- material.
gion of < 3.0 V. In practical terms, this means that being used in
lithium-ion batteries the JDEES and MTI materials would not Possible causes of overdischarge stability of LixMn2O4
tolerate even occasional overdischarge cycles with potential
drops below 3.0 V. It is of some interest to discuss the ability of the lithium man-
The NEI material is less sensitive to overdischarge than ganese spinel structure to uptake extra lithium ions from the
non-doped spinels by JDEES and MTI, and degrades more point of view of existing structural, spectroscopic, and quan-
slowly. Its behavior upon cycling at lowered voltages signifies tum chemistry information. As mentioned in Introduction,
that single overdischarges should not significantly worsen Li2Mn2O4 belongs to the I41/amd space group and represents
electrochemical properties of this sample. a distorted spinel. In accord with relations between the parent
The best result is obtained for the aluminum-doped mate- cubic Fd3m space group and the tetragonal I41/amd space
rial by Toda. It is characterized by high electrochemical sta- group, the lattice parameters can be determined as follows:
bility both in the recommended voltage range and upon pffiffiffi
aT ≈aC = 2, cT ≈ aC, where subscripts T and C stand for cubic
overdischarge to the voltage of 2.4 V. From a practical and tetragonal lattices [13], and the 4a, 8c, 8d, and 16 h posi-
tions in the tetragonal structure correspond to the 8a, 16c, 16d,
and 32e positions in the cubic structure, respectively [12, 13].
Hereby, in the Li2Mn2O4 structure, oxygen ions occupy the
tetrahedral positions (16 h), the octahedral positions (8d) are
inhabited by manganese ions, half of lithium ions hold the
tetrahedral (4a) positions, and another half of lithium ions
dwells in half of the (8c) voids, half of voids being free.
Interestingly, Thackeray et al. managed to synthesize com-
pounds of the composition of LixMn2O4 where x ≤ 2.25
[12]. This suggests that the free 8c voids may be further filled
with lithium ions, possibly up to the Li3Mn2O4 composition
with the formal charge of +2.5 on manganese ions, hence the
overlithiated lithium manganese spinel may be of the
Fig. 7 Capacity retention for materials studied berthollide nature. In view of this fact, it is not surprising that
J Solid State Electrochem (2017) 21:3269–3279 3277

the samples studied are able to uptake up to 1.25 mol of extra Conclusions
lithium ions just reaching the Li2.25Mn2O4 composition (Toda
in the 1.5–3.5 V range). For other samples, smaller x values In this paper, four lithium-manganese spinels of different ori-
are obtained (~2 for JDEES, ~1.5 for MTI, 1.5 for NEI and gin, viz., nanosized stoichiometric home-made material
Toda in the 2.4–3.5 V range; ~1.7 for JDEES, ~1.5 for MTI, (JDEES), two microsized commercial stoichiometric spinels
~1.7 for NEI in the 1.5–3.5 V range). This may be caused by (MTI and NEI) and microsized commercial aluminum-doped
fast degradation of the nanosized JDEES material and to dif- spinel (Toda) have been subjected to cyclic voltammetric and
ferences in synthetic procedures used and further treatment galvanostatic overdischarge studies for 240 cycles. The results
applied to commercial samples. obtained can be summarized as follows.
Existing papers dealing with the electronic structure of Intercalation of lithium ions into lithium-manganese spi-
LixMn2O4 lead to contradictory results and do not clarify the nels by decreasing potential to either +2.4 or +1.5 V against
picture of the lithiation/delithiation process. Some studies con- Li/Li+ leads to the formation of the overlithiated Li2−xMn2O4
clude charge transfer from lithium to oxygen ions upon lithi- spinel phase discernible by a characteristic Li+ insertion/
um insertion [47, 48], while others claim that charge transfer deinsertion peak in the 2.9 V region. No noticeable changes
takes place from lithium to manganese 3d states [13, 49]. The of this peak with scan number signifies no transformations of
most detailed picture of the lithiation/delithiation process ex- the spinel structure, and materials studied retain their electro-
tended on the overlithiated compositions has been presented chemical activity in the working region of spinel electrodes,
by Suzuki and co-workers [50]. Based on high-resolution X- i.e., > 3.5 V.
ray Compton scattering and an ab initio Mulliken population An ability to intercalate extra lithium ions at low potentials
analysis, these authors conclude that in LixMn2O4 with significantly depends on material. Decreasing potential to
0 ≥ x ≥ 1.5, the orbital involved in the lithium insertion and +1.5 V enables one to reversibly insert/deinsert 1 g equivalent
extraction process is found to be mainly the oxygen 2p orbital. of extra Li+ forming Li2Mn2O4 in the case of the nanosized
It should be mentioned that in Ref. [50], no characteristic JDEES sample. For other (microsized) samples, uptake/
points on electron density-composition dependences at x = 1 release of extra Li+ is limited by no more than half gram
exist evidencing the possible berthollide nature of the equivalent. Being electrochemically overlithiated by decreas-
overlithiated compound. One may speculate that the uptake ing potential to +2.4 V, all samples uptake/release half of the
of 1.25 mol of extra lithium ions leading to the Li2.25Mn2O4 gram–equivalent of extra Li+. The easiness of overlithiation of
composition, as in Ref. [12] and in the case of the Toda sample nanosized lithium-manganese spinels compared with
in the 1.5–3.5 V range in this study, corresponds to a critical microsized ones may be advantageous for obtaining
filling value, which cannot be exceeded without the collapse overstoichiometric lithiated oxides by means of electrochem-
of the crystal lattice. Differences in the maximal x values for ical lithiation.
the samples studied could be attributed to differences in their Repeated cycling involving the low-potential regions
morphology (particle and crystallite size, see Table 1). worsens electrochemical parameters: cell resistance in-
Unlike LixMn2O4, much less considerations exist regard- creases and specific capacity decreases. The degradation
ing the role of Al substitution in the spinel lattice. Replacing degree is determined by the material. Comparisons reveal
greater Mn3+ (0.72 Å) for smaller Al3+ (0.68 Å) leads to the that even small overdischarge is totally unacceptable for the
appreciable contraction of the lattice [31, 51, 52], so that one JDEES and MTI samples. The NEI sample better tolerates
might expect difficulties in intercalation/deintercalation cycling at lowered potentials. The greatest resistance to
processes. Amatucci and co-workers [31] notice that the overdischarge demonstrates the aluminum-doped Toda
decrease in lattice parameter should lead to the increase in sample. Interestingly, after treatment at lowered voltages,
Mn-O covalency. However, they stress that the effect the this material acquires a steady trend towards an increase in
aluminum substitution has on the electronic characteristics specific capacity by half. This may signify that at least half
of the bonding between the transition metal and oxygen of the gram equivalent of extra lithium ions can be electro-
sublattice cannot be ignored. Aluminum forms a highly co- chemically inserted into the 16с voids in the spinel struc-
valent bond with neighboring oxygens. Manganese ions ture resulting in the Li1.5Mn2O4 compound, which is able
share these same oxygens, so that a subsequent increase in to reversibly work within the 2.4–4.6 V potential range.
Mn-O ionicity should be observed. In this case, one may Most likely, such feature is related to the changes in
suggest that the distribution of charge in the lattice and es- Coulombic potential on the 16c voids originating from re-
pecially on vacancies significantly changes and may facili- placing greater Mn3+ (0.72 Å) for smaller Al3+ (0.68 Å).
tate filling of the 16с voids with lithium ions. Indeed, such This finding calls for additional studies, first of all, by
speculation needs further proofs by means of charge distri- means of structure-sensitive techniques, which is outside
bution calculations, quantum mechanical calculations, and the scope of this work and will be described in forthcoming
modern experimental techniques. papers.
3278 J Solid State Electrochem (2017) 21:3269–3279

Acknowledgements The authors thank Mr. S Chernukhin for prelimi- 20. Vitinc G, West K (1997) Lithium intercalation into layered
nary tests and discussions. LiMnO2. J Electrochem Soc 144:2587–2592
21. Reimers JN, Fuller EW, Rossen E, Dahn JR (1993) Synthesis and
electrochemical studies of LiMnO2 prepared at low temperatures. J
Electrochem Soc 140:3396–3401
References 22. Armstrong AR, Bruse PG (1996) Synthesis of layered LiMnO2 as
an electrode for rechargeable lithium batteries. Nature 381:499–500
1. Brandt K (1994) Historical development of secondary lithium bat- 23. Gummow RJ, de Kock A, Thackeray MM (1994) Improved capac-
teries. Solid State Ionics 69:173–183 ity retention in rechargeable 4 V lithium/lithium-manganese oxide
2. Krueger S, Kloepsch R, Li J, Nowak S, Passerini S, Winter M (spinel) cells. Solid State Ionics 69:59–67
(2013) How do reactions at the anode/electrolyte interface deter- 24. Tarascon JM, McKinnon WR, Coowar F, Bowmer TN, Amatucci
mine the cathode performance in lithium–ion batteries? J G, Guyomard D (1994) Synthesis conditions and oxygen stoichi-
Electrochem Soc 160:A542–A548 ometry effects on li insertion into the spinel LiMn 2 O 4 . J
3. Lee YS, Cheng MW (2005) Intelligent control battery equalization Electrochem Soc 141:1421–1431
for series connected lithium-ion battery strings. IEEE Trans Ind 25. Tarascon JM, Guyomard D (1991) Li metal-free rechargeable bat-
Electr 52:1297–1307 teries based on Li1+XMn2O4 cathodes (0≤x≤1) and carbon anodes. J
4. Doughty D, Roth EP (2012) A general discussion of Li ion battery Electrochem Soc 138:2864–2868
safety. Electrochemical Society Interface 21:37–44 26. Gummow RJ, Liles DC, Thackeray MM (1993) Lithium extraction
5. Jeevarajan JA, Winchester CS (2012) Battery safety qualifications from orthorhombic lithium manganese oxide and the phase trans-
for human rating. Electrochem Soc Interface 21:51–55 formation to spinel. Mater Res Bull 28:1249–1256
27. Hunter JC (1981) Preparation of a new crystal form of manganese
6. International Electrotechnical Commission (2012) Secondary cells
dioxide: λ-MnO2. J Solid State Chem 39:142–147
and batteries containing alkaline or other non-acid electrolytes—
28. Jang DH, Oh SM (1997) Electrolyte effects on spinel dissolution
safety requirements for portable sealed secondary cells, and for bat-
and cathodic capacity losses in 4 V li/LixMn2O4 rechargeable cells.
teries made from them, for use in portable applications. IEC 62133
J Electrochem Soc 144:3342–3334
7. Li HF, Gao JK, Zhang SL (2008) Effect of overdischarge on swell-
29. Cho J, Thackeray MM (1999) Structural changes of LiMn2O4 spi-
ing and recharge performance of lithium ion cells. Chin J Chem 26:
nel electrodes during electrochemical cycling. J Electrochem Soc
1585–1588
146:3577–3581
8. Shu J, Shui M, Xu D, Wang D, Ren Y, Gao S (2012) A comparative
30. Robertsson AD, Lu SH, Howard WF (1997) M3+ -modified
study of overdischarge behaviors of cathode materials for lithium-
LiMn2O4 spinel intercalation cathodes. J Electrochem Soc 144:
ion batteries. J Solid State Electrochem 16:819–824
3505–3512
9. Maleki H, Howard JN (2006) Effects of overdischarge on perfor-
31. Amatucci GG, Pereira N, Zheng T, Tarascon JM (2001) Failure
mance and thermal stability of a li-ion cell. J Power Sources 160:
mechanism and improvement of the elevated temperature cycling
1395–1402
of LiMn2O4 compounds through the use of the LiAlxMn2−xO4−zFz
10. Rossouw MH, de Kock A, de Piciotto LA, Thackeray MM, David solid solution. J Electrochem Soc 148:A171–A182
WIF, Ibberson RM (1990) Structural aspects of lithium-manganese- 32. Liu Q, Wang S, Tan H, Yang Z, Zeng J (2013) Preparation and
oxide electrodes for rechargeable lithium. Mater Res Bull 25:173– doping mode of doped LiMn2O4 for Li-ion batteries. Energies 6:
182 1718–1730
11. Ohzuku T, Kitagawa M, Hirai T (1990) Electrochemistry of man- 33. Kim KW, Lee SW, Han KS, Chung HJ, Woo SI (2003)
ganese dioxide in lithium nonaqueous cell. J Electrochem Soc 137: Characterization of Al-doped spinel LiMn2O4 thin film cathode
769–775 electrodes prepared by liquid source misted chemical deposition
12. Thackeray MM, David WIF, Bruce PG, Goodenough JB (1983) (LSMCD) technique. Electrochem Acta 48:4223–4231
Lithium insertion into manganese spinels. Mater Res Bull 18: 34. Sun Y, Li C (2012) Synthesis and properties of the modification of
461–472 micro–crystal LiMn2−xAlxO4 cathode material for Li–ion batteries.
13. Berg H, Goransson K, Nolang B, Thomas J (1999) Electronic struc- J Wuhan Univ Technol Mater Sci Educ 27:79–81
ture and stability of the LixMn2O4 (0<x<2) system. Mater Chem 9: 35. Wu C, Sun J, Zhu C, Ge Y, Zhao Y (2015) Research on overcharge
2813–2820 and overdischarge effect on lithium-ion batteries, Vehicle Power
14. Winter M, Besenhard JO, Spahr ME, Novak P (1998) Insertion and Propulsion Conference (VPPC) 2015 I.E. 1–6
electrode materials for rechargeable lithium batteries. Adv Mater 36. Guo R, Lu L, Ouyang M, Feng X (2016) Mechanism of the entire
10:725–763 overdischarge process and overdischarge-induced internal short cir-
15. Thackeray MM, Yang S, Kahaian AJ, Kepler KD, Skinner E, cuit in lithium-ion batteries. Sci Rep 6:30248
Vaughey JT, Hackney SA (1998) Structural fatigue in spinel elec- 37. Chen K, Donahoe AC, Noh YD, Komarneni S, Xue D (2014)
trodes in high voltage (4 V) Li/LixMn2O4 cells. Electrochem Solid- LiMn2O4-based materials as anodes for lithium-ion battery. Funct
State Lett 1:7–9 Mater Lett 7:1350070
16. Thackeray MM, de Picciotto LA, de Kock A, Johnson PJ, Nicholas 38. Chen K, Xue D (2014) Anode performances of mixed LiMn2O4
V, Adendorff K (1987) Spinel electrodes for lithium batteries—a and carbon black toward lithium-ion battery. Funct Mater Lett 7:
review. J Power Sources 21:1–8 1450017
17. Thackeray MM, de Kock A, Rossow MH, Liles D (1992) Spinel 39. Poizot P, Laruelle S, Grugeon S, Dupont L, Tarascon JM (2000)
electrodes from the Li-Mn-O system for rechargeable lithium bat- Nano-sized transition-metaloxides as negative-electrode materials
tery applications. J Electrochem Soc 139:363–366 for lithium-ion batteries. Nature 407:496–499
18. Gummow RG, Thackeray MM (1994) An investigation of spinel- 40. Doh CH, Jin BS, Lim JH, Moon SI (2002) Electrochemical char-
related and orthorhombic LiMnO2 cathodes for rechargeable lithi- acteristics of lithium transition-metal oxide as an anode material in a
um batteries. J Electrochem Soc 141:1178–1182 lithium secondary battery. Korean J Chem Eng 19:749–755
19. Feng J, Song B, Lai MO, Lu L, Zeng X, Huang Z (2012) 41. Potapenko AV, Kirillov SA (2014) Lithium manganese spinel ma-
Electrochemical property of LiMn2O4 in over–discharged condi- terials for high-rate electrochemical applications. J Energy Chem
tions. Funct Mater Lett 5:1250028 23:543–558
J Solid State Electrochem (2017) 21:3269–3279 3279

42. Potapenko AV, Chernukhin SI, Kirillov SA (2015) A new method 48. Liu Y, Fujiwara T, Yukawa H, Morinaga M (1999) Electronic struc-
of pretreatment of lithium manganese spinels and high-rate electro- tures of lithium manganese oxides for rechargeable lithium battery
chemical performance of Li[Li0.033Mn1.967]O4. Mater renew sus- electrodes. Solid State Ionics 126:209–218
tain. Energy 4:40 49. Grechnev GE, Ahuja R, Johansson B, Eriksson O (2002) Electronic
43. IEC 62576 (2009) Electric double-layer capacitors for use in hybrid structure, magnetic, and cohesive properties of LixMn2O4: theory.
electric vehicles—test methods for electrical characteristics. Phys Rev B 65:174408
International Electrotechnical Commission, Geneva, p 51 50. Suzuki K, Barbiellini B, Orikasa Y, Go N, Sakurai H, Kaprzyk S,
44. Diab Y, Venet P, Gualous H, Rojat G (2008) Electrical, frequency Itou M, Yamamoto K, Uchimoto Y, Wang YJ, Hafiz H, Bansil A,
and thermal measurement and modelling of supercapacitor perfor- Sakurai Y (2015) Extracting the redox orbitals in Li battery mate-
mance. In ESSCAP rials with high-resolution X-ray Compton scattering spectroscopy.
45. Shu ZX, Davidson I, McMillan R, Murray JJ (1997) Electrochemistry Phys Rev Lett 114:087401
of LiMnO2 over an extended potential range. J Power Sources 68: 51. Song D, Ikuta H, Uchida T, Wakihara M (1999) The spinel
618–622 LiAlyMn2−yO4 (y=0, 1/12, 1/9, 1/6, 1/3) and Li(Al,M)1/6Mn11/6O4
46. Moore SW, Schneider PJ (2001) A review of cell equalization (M=Cr, Co) as the cathode for rechargeable lithium batteries. Solid
methods for lithium ion and lithium polymer battery systems (№. State Ionics 117:151–156
2001–01–0959). SAE Tech. Paper Series 1–5 52. Lee YS, Kumada N, Yoshio M (2001) Synthesis and characteriza-
47. Aydinol MK, Ceder G (1997) First-principles prediction of inser- tion of lithium aluminum-doped spinel (LiAlxMn2−xO4) for lithium
tion potentials in Li-Mn oxides for secondary Li batteries. J secondary battery. J Power Sources 96:376–384
Electrochem Soc 144:3832–3835