Reducing
Voltage
Hysteresis
in
Li-Rich
Sulfide
Cathodes
by
Incorporation
of
Mn
Xiaotong
Li, Seong
Shik Kim, Michelle
D. Qian,
Eshaan
S. Patheria,
Jessica
L. Andrews,
Colin
T. Morrell,
Brent
C. Melot,
and Kimberly
A. See
*
Cite
This:
Chem.
Mater.
2024,
36, 5687−5697
Read
Online
ACCESS
Metrics
& More
Article
Recommendations
*
sı
Supporting
Information
ABSTRACT:
Conventional
intercalation-based
cathode
materials
in Li-ion
batteries
are based
on charge
compensation
of the redox-active
cation
and can only intercalate
one mole of electron
per formula
unit. Anion
redox,
which
employs
the anion
sublattice
to compensate
charge,
is a promising
way to achieve
multielectron
cathode
materials.
Most
anion
redox
materials
still face the problems
of slow kinetics
and large voltage
hysteresis.
One potential
solution
to reduce
voltage
hysteresis
is to increase
the
covalency
of the metal
−
ligand
bonds.
By substituting
Mn into the electrochemically
inert
Li
1.33
Ti
0.67
S
2
(Li
2
TiS
3
), anion
redox
can
be activated
in the
Li
1.33
−
2
y
/3
Ti
0.67
−
y
/3
Mn
y
S
2
(
y
= 0
−
0.5)
series.
Not only do we observe
substantial
anion
redox,
but the voltage
hysteresis
is significantly
reduced,
and the rate capability
is
dramatically
enhanced.
The
y
= 0.3 phase
exhibits
excellent
rate and cycling
performance,
maintaining
90% of the C/10
capacity
at
1C, which
indicates
fast kinetics
for anion
redox.
X-ray
absorption
spectroscopy
(XAS)
shows
that both the cation
and anion
redox
processes
contribute
to the charge
compensation.
We attribute
the drop in hysteresis
and increase
in rate performance
to the
increased
covalency
between
the metal
and the anion.
Electrochemical
signatures
suggest
the anion
redox
mechanism
resembles
holes
on the anion,
but the S K-edge
XAS data confirm
persulfide
formation.
The mechanism
of anion
redox
shows
that forming
persulfides
can be a low hysteresis,
high rate capability
mechanism
enabled
by the appropriate
metal
−
ligand
covalency.
This work
provides
insights
into how to design
cathode
materials
with anion
redox
to achieve
fast kinetics
and low voltage
hysteresis.
■
INTRODUCTION
Development
of rechargeable
batteries
is crucial
to achieve
grid-scale
energy
storage
to support
the transition
from fossil
fuels to renewable
energy
sources
such as wind and solar.
1
To
achieve
grid-scale
energy
storage
and other
next-generation
battery
uses,
the energy
density
of current
rechargeable
batteries
needs
to be dramatically
increased.
2
So far, Li-ion
batteries
(LIBs)
dominate
the energy
storage
market,
but their
capacities
are limited
by current
cathode
materials
that rely on
transition
metal
redox
and only reversibly
cycle up to one mole
of electron
per redox-active
transition
metal.
3,4
Employing
the electrons
that nominally
reside
on anions
to
compensate
charge,
termed
“anion
redox”,
is a promising
way
to achieve
multielectron
redox
reactions
and increase
cathode
energy
density.
5
Anion
redox
has been
explored
in Li-rich
oxides,
but there are still many
challenges
such as irreversible
oxygen
loss, large
voltage
hysteresis,
sluggish
kinetics,
and
significant
voltage
fade due to the large structural
changes.
6
−
8
Undesirable
side reactions
with the electrolytes
at high voltages
also complicate
the mechanistic
study
and practical
application
of anion
redox
in Li-rich
oxide
materials.
9,10
Moreover,
the
charge
compensation
mechanism
is still under
vigorous
debate.
11
−
15
Contrary
to oxides,
Li-rich
sulfide
materials
are promising
systems
to achieve
multielectron
redox.
S p states
are higher
in
energy
than O p states,
which
results
in lower-voltage
anion
redox
in sulfides
compared
to oxides.
However,
the extremely
high capacities
of sulfide
materials
yield energy
densities
that
are comparable
to those
of lower-capacity,
higher-voltage
oxides.
16,17
Additionally,
the low voltage
can be beneficial
from
a safety
perspective
as operation
can be constrained
to the
stability
window
of electrolytes.
16
Thus,
they can provide
more
reversible
cycling
performance
compared
to their
oxide
counterparts.
17
In Li-rich
oxides,
such as Li
2
RuO
3
, the first
cycle is usually
the “activation”
cycle,
and there is a completely
different
charge
curve
on the second
cycle,
suggesting
an
irreversible
structural
change.
15
In contrast,
Li-rich
sulfides
such as Li
2
FeS
2
can show nearly
identical
charge
and discharge
curve
shapes
upon cycling,
which
suggests
that the material
is
regenerated
upon
discharge,
allowing
for reproducible
features.
17,18
The (dis)charge
shapes
are similar
despite
the
significant
voltage
hysteresis,
defined
as the voltage
gap
between
the charge
and discharge
curves.
Interestingly,
the
Received:
March
12, 2024
Revised:
May 1, 2024
Accepted:
May 8, 2024
Published:
May 16,
2024
Article
pubs.acs.org/cm
© 2024
The Authors.
Published
by
American
Chemical
Society
5687
https://doi.org/10.1021/acs.chemmater.4c00736
Chem.
Mater.
2024,
36, 5687
−
5697
This article is licensed under CC-BY-NC-ND 4.0
hysteresis
is larger
in the anion
redox
region
than the transition
metal
redox
region.
The hysteresis
in the anion
redox
region
is
partially
due to kinetic
overpotentials;
however,
even
in
galvanostatic
intermittent
titration
(GITT)
experiments,
the
thermodynamic
hysteresis
remains
around
270 mV suggesting
path hysteresis.
17
Materials
can also exhibit
path hysteresis
when
the charge
and discharge
pathways
are dissimilar.
19
Contrarily,
Li-rich
oxides
commonly
show
very dissimilar
charge
curves
from cycle 1 to cycle 2 combined
with voltage
hysteresis.
The different
shape
originates
from the irreversible
structural
changes
and/or
the different
pathways
between
the
first charge
and discharge.
7,15
Though
Li-rich
sulfides
also
undergo
structural
changes
(e.g.,
formation
of persulfides
(S
2
)
2
−
), the changes
are reversible.
However,
it remains
a
challenge
to capitalize
on the benefits
of reversible
anion
redox
in sulfides
due to the large voltage
hysteresis.
Large
voltage
hysteresis
leads
to lower
discharge
energy
density
compared
to charge,
resulting
in energy
loss each cycle,
which
needs
to be minimized
for anion
redox.
Because
the
hysteresis
in materials
like Li
2
FeS
2
is both
kinetic
and
thermodynamic
and is localized
to the anion
redox
region,
the hysteresis
is likely
due to the making
and breaking
of S
−
S
bonds
and the nucleation
of phases
that support
them.
Thus,
we hypothesize
that preventing
the formation
of the S
−
S bond
during
sulfide
oxidation
is the key to reducing
hysteresis.
Such
a mechanism
could
be achieved
if holes can be stabilized
in the
S p bands
to yield
so-called
“ligand
holes”.
20
Persulfide
formation
is the consequence
of unstable,
depopulated
p
bands,
likely
nonbonding
p states,
that rehybridize
to form
more stable
S
−
S bonds.
Holes
may be stabilized
by increasing
p hybridization
with metal
d states
to stabilize
the depopulated
band,
i.e., increasing
metal-anion
covalency.
Such a phenom-
enon has been suggested
in the Li-rich
oxides.
21
−
26
To increase
covalency,
the energy
of the metal
d states
with
respect
to the S p states
must
be shifted
to yield
greater
overlap.
An ideal Li-rich
sulfide
system
to systematically
tune
the metal
−
ligand
bond
covalency
is Li
1.33
Ti
0.67
S
2
(Li
2
TiS
3
)
(Figure
1(a,b)),
since different
transition
metals
(Fe
2+
, Co
2+
,
and Ti
3+
) can be substituted
into the octahedrally
coordinated
metal
site.
27
−
29
Li
1.33
Ti
0.67
S
2
itself is electrochemically
inert
27
due to the d
0
Ti
4+
and the apparent
inability
to access
S p
states
during
initial
charge.
Interestingly,
substituting
redox-
active
metals
into Li
1.33
Ti
0.67
S
2
activates
anion
redox
and
significantly
improves
capacity,
27
−
29
but their
influence
on
voltage
hysteresis
has not been systematically
studied.
When
Ti
3+
and Co
2+
are substituted
into the structure,
there
is still
sizable
voltage
hysteresis
between
the charge
and discharge
profiles.
27,28
The Fe
2+
-substituted
Li
1.33
Ti
0.67
S
2
exhibits
a
significantly
reduced
hysteresis,
but the reason
behind
is not
clear.
29
Shortly
before
the submission
of this paper,
Louis
et al.
reported
the Mn
2+
-substituted
Li
1.33
Ti
0.67
Ch
2
(Ch = S/Se)
with
a focus
on the O
3
to O
1
phase
transition.
30
In this paper,
we
Figure
1.
Crystal
structure
of Li
1.13
Ti
0.57
Mn
0.3
S
2
from (a) side view and (b) top view. (c) Schematic
band structure
showing
relative
positions
of
Mn and Ti d bands
vs the S p manifold.
Figure
2.
(a) XRD
patterns
of Li
1.33
−
2
y
/3
Ti
0.67
−
y
/3
Mn
y
S
2
(
y
= 0
−
0.5)
series.
(b) Unit cell volumes
extracted
from Rietveld
refinement.
(c)
Synchrotron
XRD pattern
of Li
1.13
Ti
0.57
Mn
0.3
S
2
, with refinement
and difference
traces.
The tick marks
indicate
the Bragg
reflections
of the phases.
The Li
2
TiO
3
impurity
is less than 1 wt %.
Chemistry
of
Materials
pubs.acs.org/cm
Article
https://doi.org/10.1021/acs.chemmater.4c00736
Chem.
Mater.
2024,
36, 5687
−
5697
5688
mainly
focus
on the influence
of Mn
2+
substitution
on
electrochemistry,
especially
voltage
hysteresis,
and present
our theory
on how metal
−
ligand
covalency
can tune voltage
hysteresis.
Here,
we aim to substitute
a metal
with lower-energy
d states
compared
to Ti
4+
into Li
1.33
Ti
0.67
S
2
. We hypothesize
that lower
d band of Mn
2+
can yield a more
covalent
metal-anion
bond
(Figure
1(c)).
We synthesize
a series
of Mn-substituted
Li
1.33
Ti
0.67
S
2
phases
through
solid-state
techniques.
We
demonstrate
that Mn substitution
not only activates
anion
redox
in the otherwise
inert Li
1.33
Ti
0.67
S
2
but also significantly
reduces
the voltage
hysteresis.
The substituted
material
Li
1.13
Ti
0.57
Mn
0.3
S
2
exhibits
excellent
rate performance,
main-
taining
90% of the C/10
capacity
at 1C, which
suggests
that
fast kinetics
can be achieved
in anion
redox.
Using
metal
and
sulfur
K-edge
X-ray
absorption
spectroscopy,
we demonstrate
that both cation
and anion
redox
are responsible
for the charge
compensation.
Our results
suggest
that tuning
the metal
−
ligand
bond covalency
is an effective
way to control
the voltage
hysteresis
in anion
redox.
■
RESULTS
AND
DISCUSSION
Structural
Characterization.
Li
1.33
−
2
y
/3
Ti
0.67
−
y
/3
Mn
y
S
2
is
prepared
through
solid-state
synthesis
techniques,
which
yield
polycrystalline
black
pellets
that are subsequently
ground
to
black
powders.
The structures
of the series
are characterized
first by lab powder
X-ray
diffraction
(XRD).
The XRD patterns
of Li
1.33
−
2
y
/3
Ti
0.67
−
y
/3
Mn
y
S
2
are shown
in Figure
2(a).
There
are no extra
reflections
after
Mn substitution,
which
qualitatively
suggests
the space
group
and symmetry
remain
the same throughout
the series.
The patterns
are fit by Rietveld
refinement,
which
can be
found
in Figure
S1, to extract
the unit cell volumes.
All
reflections
are described
well by the
R
3
̅
m
space
group,
and the
unit cell volumes
increase
linearly
with
increasing
Mn
substitution
(Figure
2(b)),
which
indicates
the formation
of
the solid solution
across
all stoichiometries.
Among
the whole
series,
the
y
= 0.3 material
shows
the most
promising
electrochemical
performance
(vide
infra)
and is selected
for
further
study.
The
synchrotron
XRD
pattern
of
Li
1.13
Ti
0.57
Mn
0.3
S
2
is shown
in Figure
2(c) with the quantitative
Rietveld
refinement.
The major
phase
can be fit with the
R
3
̅
m
space
group,
with a small
amount
(<1 wt %) of Li
2
TiO
3
impurity
potentially
introduced
by air exposure
during
shipping.
The structure
of Li
1.13
Ti
0.57
Mn
0.3
S
2
is similar
to that
of LiTiS
2
, with additional
Li and Mn occupying
the Ti sites in
the metal
layer (Figure
1(a,b)).
The edge-sharing
octahedral
Ti/Mn/Li
layers
are separated
by the edge-sharing
octahedral
Li layers.
Other
2+ transition
metals
(e.g., Fe
2+
and Co
2+
) have
been substituted
into the Li
1.33
Ti
0.67
S
2
structure,
28,29
and the
Li
1.33
−
2
y
/3
Ti
0.67
−
y
/3
Fe
y
S
2
series
form a solid solution
with the
same
R
3
̅
m
space
group.
27
Electrochemical
Characterization.
The electrochemical
performance
of the Li
1.33
−
2
y
/3
Ti
0.67
−
y
/3
Mn
y
S
2
series
is first
examined
with galvanostatic
cycling,
as shown
in Figure
3. All
cells are cycled
at C/10
and charged
first, which
corresponds
to oxidation
at the cathode.
The parent
phase
(Li
1.33
Ti
0.67
S
2
)
exhibits
a limited
capacity
of 0.1 e
−
per formula
unit (f.u.),
since
Ti is in the 4+ oxidation
state and cannot
be further
oxidized
27,29,31
and anion
oxidation
is apparently
inaccessible.
After Mn substitution,
the capacity
significantly
increases
to
∼
1
e
−
per f.u. until
y
= 0.3 and then decreases
again for
y
= 0.4 and
0.5. We suspect
the decrease
in capacity
is due to limited
Li
+
inventory
since the amount
of Li in the formula
decreases
with
Mn
2+
substitution.
All substituted
materials
exhibit
a single
plateau
in the charge
profile
around
2.72 V. A single
plateau
above
2.5 V in other
similar
sulfides
has been attributed
to
sulfur
oxidation.
17,29,32,33
Even
assuming
multielectron
oxida-
tion from Mn
2+
to Mn
4+
, which
is unlikely
in a sulfide
anion
lattice
where
sulfur
oxidation
occurs
at lower
voltages
than
Mn
3+/4+
redox,
the measured
capacity
cannot
be fully
accounted
for (Figure
S2). This suggests
that the anions
contribute
to charge
compensation
(vide
infra).
More
interestingly,
the voltage
hysteresis,
which
is observed
as the gap between
the charge
and discharge
curves,
is reduced
from 500 mV in the parent
phase
to 150 mV in the
y
= 0.3
phase.
The charge
profile
for the second
cycle is also shown
in
Figure
3 and exhibits
a change
in shape
with now a short
sloping
region
at early states
of charge,
followed
by a plateau
region
at slightly
reduced
voltage
(
∼
2.7
V). The voltage
of the
first and second
charge
curves
is more
similar
at higher
Mn
substitution
ratios,
suggesting
a more reversible
redox
process.
In addition,
while
voltage
fading
is a common
problem
among
cathode
materials
with
rock-salt
structures,
6
the plateau
voltages
of the Mn-substituted
materials
are very stable
after
the first cycle (Figure
S3).
Cyclic
voltammetry
(CV)
is measured
to further
understand
the electrochemical
behavior,
and the data is shown
in Figure
4. For the Mn-substituted
materials,
there is one anodic
wave
at
∼
2.8 V in the first cycle,
which
matches
with the voltage
of
anion
redox
plateau
in galvanostatic
cycling.
From
the second
cycle on, it evolves
into two anodic
waves
with a minor
peak at
∼
2.3 V that matches
with the sloping
region
in the charge
curve,
likely
related
to transition
metal
oxidation,
and a major
peak centered
at
∼
2.7 V that matches
the voltage
of the plateau
region
associated
with anion
oxidation.
On the reverse
scan,
two overlapping
cathodic
waves
are observed
in the
y
= 0.1 and
0.2 phases.
One can be generally
correlated
to the small
cathodic
wave in the
y
= 0 material,
and we thus associate
it
with persulfide
reduction.
The other wave is more pronounced
as the Mn content
increases,
suggesting
association
with Mn
Figure
3.
Galvanostatic
cycling
of Li
1.33
−
2
y
/3
Ti
0.67
−
y
/3
Mn
y
S
2
(
y
= 0
−
0.5) series
at C/10
based
on 1 electron
per formula
unit.
Chemistry
of
Materials
pubs.acs.org/cm
Article
https://doi.org/10.1021/acs.chemmater.4c00736
Chem.
Mater.
2024,
36, 5687
−
5697
5689
reduction.
For
y
= 0.3 and above,
the higher-voltage
cathodic
wave associated
with Mn reduction
becomes
more pronounced
due to the higher
Mn content.
However,
anion
reduction
is
also occurring
so it is difficult
to assign
each peak discretely
to
an associated
redox
process.
The voltage
hysteresis
can be estimated
from the separation
between
the anodic
and cathodic
waves.
Li
1.33
Ti
0.67
S
2
exhibits
one anodic
wave
and one cathodic
wave
at 2.9 and 2.3 V,
respectively,
in the first cycle,
and the peak positions
match
the
plateau
voltages
during
charge
and discharge.
The large
separation
between
the two peaks
suggests
a large
voltage
hysteresis
during
the first cycle.
In the second
cycle,
the anodic
wave
shifts
to a lower
voltage
(2.7 V), indicating
reduced
voltage
hysteresis.
After
Mn substitution,
the anodic
wave of
the second
cycle shifts to a higher
voltage
and starts to overlap
with the peak from the first cycle.
For
y
= 0.3 and above,
the
peaks
of all cycles
nearly
overlap,
suggesting
a more
reversible
redox
process.
This is in sharp
contrast
to the Li-rich
oxides,
where
the first and following
charge
curves
are completely
different,
15,23
since
the structural
changes
in the sulfides
are
reversible
(vide
infra)
while
those
in the oxides
are not. More
importantly,
the separation
between
the anodic
and cathodic
peaks
is much
smaller
with Mn addition,
corresponding
to a
significantly
reduced
voltage
hysteresis.
GITT
is used to probe
the shape
of the profiles
under
near-
equilibrium
conditions
and estimate
kinetic
overpotentials
in
the Li
1.33
−
2
y
/3
Ti
0.67
−
y
/3
Mn
y
S
2
series,
and the traces
can be found
in Figure
5. The overpotentials,
which
can be approximated
by
the difference
between
the potential
under
bias and that after a
4 h open-circuit
voltage
(OCV)
hold, are much
larger
in the
parent
phase
(
∼
250
mV)
compared
to those
in the Mn-
substituted
materials.
The overpotential
is reduced
to 50 mV in
the
y
= 0.3 phase
and does not change
throughout
the plateau
in the first cycle,
suggesting
that a single
process
is happening
across
the whole
plateau
region.
This is different
from
the
electrochemistry
of Li
2
FeS
2
, where
it shows
distinct
cation
and
anion
redox
regions.
17
In Li
2
FeS
2
, the sloping
region
that
corresponds
to the cation
redox
exhibits
small
overpotential,
while
the plateau
region
of anion
redox
shows
much
larger
overpotential
because
the anion
redox
process
involves
larger
structural
changes
and slower
kinetics.
17
But in this material,
transition
metal
and anion
oxidation
are both occurring,
these
two processes
are concerted
and happen
simultaneously
(vide
infra).
The significantly
reduced
overpotential
suggests
that
after Mn substitution,
the kinetics
improves
as extraction
of
electrons
from the S p band becomes
more
facile
presumably
due to the significant
hybridization
of the Mn d states
with the
S p states.
The thermodynamic
hysteresis,
which
is the voltage
gap measured
under
near-equilibrium
conditions
(i.e., after the
4 h OCV
hold),
or the hysteresis
minus
the overpotential,
is
reduced
from 250 mV in the parent
phase
to 50 mV in the
y
=
0.3 phase.
Since
the equilibrium
hysteresis
is extracted
after
relaxation
at OCV
for several
hours,
it indicates
the electronic
states
after relaxation,
and the reduced
overpotential
suggests
lower
energy
barrier
for the redox
process.
To further
probe
the kinetics
of the
y
= 0.3 cathode
materials,
rate-dependent
galvanostatic
cycling
is performed.
Figure
6(a) shows
the galvanostatic
cycling
curves
of four
separate
cells at different
rates. A discharge
capacity
of around
230 mAh
g
−
1
can be achieved
at C/10,
which
is slightly
reduced
to 215 mAh g
−
1
at C/5 and still maintains
205 mAh
g
−
1
at C/2 and even 1C. The overpotential
is slightly
increased
at higher
rates as the cycling
rate outpaces
the kinetics
of the
redox
process.
These
results
suggest
that Mn substitution
not
only increases
the capacity
and decreases
the voltage
hysteresis
but also improves
the rate performance.
The latter
two
phenomena
both benefit
from the fast kinetics.
To evaluate
rate capability
over long-term
cycles,
three
replicate
cells are cycled
at different
rates.
The charge
and
discharge
capacities
along
with the cell-to-cell
reproducibility,
displayed
as the standard
deviation
between
the replicate
cells
shown
as error bars, are displayed
in Figure
6(b).
The small
decrease
in capacity
at the high rates is maintained
in the
sequential
cycling
experiment
from C/10
to 1C. It is worth
noting
that all of the electrochemical
characterizations
are
Figure
4.
Cyclic
voltammetry
of Li
1.33
−
2
y
/3
Ti
0.67
−
y
/3
Mn
y
S
2
(
y
= 0
−
0.5)
series
at 0.05 mV s
−
1
.
Figure
5.
GITT
curves
of Li
1.33
−
2
y
/3
Ti
0.67
−
y
/3
Mn
y
S
2
(
y
= 0
−
0.5)
series.
GITT
was obtained
at C/10
based
on 1 electron
per formula
unit for
40 min separated
by 4 h rest periods
at OCV.
Chemistry
of
Materials
pubs.acs.org/cm
Article
https://doi.org/10.1021/acs.chemmater.4c00736
Chem.
Mater.
2024,
36, 5687
−
5697
5690
performed
using
free-standing
electrodes
with
no further
optimization,
but the cells are able to maintain
85% of the
original
capacity
after
50 cycles
(18 of the cycles
were
performed
at higher
rates than C/10),
with only 0.3% capacity
fade per cycle.
Therefore,
we believe
that even better
cycling
performance
can be achieved
with more optimization
or using
cast electrodes.
Most
of the capacity
can be recovered
after
changing
back to C/10
indicating
minimal
material
degrada-
tion over this cycling
protocol.
Characterization
of
the
Redox
Processes.
In
Li
1.13
Ti
0.57
Mn
0.3
S
2
, around
one electron
per f.u. can be
extracted,
which
is likely
more than what can be compensated
by solely
cation
redox.
The excess
capacity
could
originate
from
multielectron
oxidation
of Mn,
34
oxidation
of S, or a
combination
thereof.
To further
understand
the redox
process
and uncover
the charge
compensation
mechanism,
a variety
of
characterization
techniques
measured
on materials
cycled
to
different
states
of charge
(SOCs)
are pursued.
First,
ex situ
X-
ray absorption
spectroscopy
(XAS)
is used to characterize
both
metal
and S K-edges
of materials
at the following
SOCs:
(1)
pristine,
(2) partial
charge
to 0.5 e
−
, (3) full charge,
(4) partial
discharge
to 0.5 e
−
, and (5) full discharge.
To probe
the activity
of metals
during
oxidation,
Ti and Mn
XAS are measured
at the SOCs
mentioned
above.
First,
we
Figure
6.
Galvanostatic
cycling
of Li
1.13
Ti
0.57
Mn
0.3
S
2
at various
rates
(indicated)
for (a) four independent
cells and (b) three replicate
cells
along
with error bars indicating
the standard
deviation.
Figure
7.
(a) Ti K-edge
XANES
and (b) first derivative
of the rising
edge, as well as (c) Mn K-edge
XANES
and (d) first derivative
of the rising
edge of Li
1.13
Ti
0.57
Mn
0.3
S
2
at different
states
of charge.
S K-edge
XANES
of Li
1.13
Ti
0.57
Mn
0.3
S
2
during
(e) charge
and (f) discharge.
Chemistry
of
Materials
pubs.acs.org/cm
Article
https://doi.org/10.1021/acs.chemmater.4c00736
Chem.
Mater.
2024,
36, 5687
−
5697
5691