(De)lithiation mechanism of Li/SeSx (x = 0-7) batteries determined by in situ synchrotron x-ray diffraction and X-ray absorption spectroscopy

Article abstract of DOI:10.1021/ja402597g

Electrical energy storage for transportation has gone beyond the limit of converntional lithium ion batteries currently. New material or new battery system development is an alternative approach to achieve the goal of new high-energy storage system with energy densities 5 times or more greater. A series of SeS_x-carbon (x = 0-7) composite materials has been prepared and evaluated as the positive electrodes in secondary lithium cells with ether-based electrolyte. In situ synchrotron high-energy X-ray diffraction was utilized to investigate the crystalline phase transition during cell cycling. Complementary, in situ Se K-edge X-ray absorption near edge structure analysis was used to track the evolution of the Se valence state for both crystalline and noncrystalline phases, including amorphous and electrolyte-dissolved phases in the (de)lithiation process. On the basis of these results, a mechanism for the (de)lithiation process is proposed, where Se is reduced to the polyselenides, Li₂Se_n (n ≥ 4), Li₂Se₂, and Li₂Se sequentially during the lithiation and Li₂Se is oxidized to Se through Li₂Se_n (n ≥ 4) during the delithiation. In addition, X-ray photoelectron spectroscopy and electrochemical impedance spectroscopy demonstrated the reversibility of the Li/Se system in ether-based electrolyte and the presence of side products in the carbonate-based electrolytes. For Li/SeS₂ and Li/SeS₇ cells, Li ₂Se and Li₂S are the discharged products with the presence of Se only as the crystalline phase in the end of charge.

Full text of DOI:10.1021/ja402597g

Article
pubs.acs.org/JACS
(De)Lithiation Mechanism of Li/SeSx (x = 0−7) Batteries Determined
by in Situ Synchrotron X‑ray Diffraction and X‑ray Absorption
Spectroscopy
Yanjie Cui,† Ali Abouimrane,*,† Jun Lu,† Trudy Bolin,‡ Yang Ren,‡ Wei Weng,† Chengjun Sun,‡
Victor A. Maroni,† Steve M. Heald,‡ and Khalil Amine†,§
‡
†Chemical Science and Engineering Division and X-ray Science Division, Argonne National Laboratory, Argonne, Illinois 60439,
United States
§King Abdulaziz University, Jeddah, Saudi Arabia
S
* Supporting Information
ABSTRACT: Electrical energy storage for transportation has gone beyond
the limit of converntional lithium ion batteries currently. New material or
new battery system development is an alternative approach to achieve the
goal of new high-energy storage system with energy densities 5 times or
more greater. A series of SeSx−carbon (x = 0−7) composite materials has
been prepared and evaluated as the positive electrodes in secondary lithium
cells with ether-based electrolyte. In situ synchrotron high-energy X-ray
diffraction was utilized to investigate the crystalline phase transition during
cell cycling. Complementary, in situ Se K-edge X-ray absorption near edge
structure analysis was used to track the evolution of the Se valence state for
both crystalline and noncrystalline phases, including amorphous and
electrolyte-dissolved phases in the (de)lithiation process. On the basis of
these results, a mechanism for the (de)lithiation process is proposed, where
Se is reduced to the polyselenides, Li2Sen (n ≥ 4), Li2Se2, and Li2Se sequentially during the lithiation and Li2Se is oxidized to Se
through Li2Sen (n ≥ 4) during the delithiation. In addition, X-ray photoelectron spectroscopy and electrochemical impedance
spectroscopy demonstrated the reversibility of the Li/Se system in ether-based electrolyte and the presence of side products in
the carbonate-based electrolytes. For Li/SeS2 and Li/SeS7 cells, Li2Se and Li2S are the discharged products with the presence of
Se only as the crystalline phase in the end of charge.
the solubility of intermediate lithium polysulfide species during
cycling, which causes the so-called redox shuttle effect and,
thus, poor cyclability.12−14
1. INTRODUCTION
The desire for effective utilization of renewable resources,
including solar, tidal, and wind, for transportation applications
as a replacement for fossil fuels and the widespread use of
portable electronics has created great opportunities for
development of new energy storage systems. State-of-the-art
rechargeable batteries are mainly based on conventional lithium
intercalation chemistry, using lithium transition metal oxides as
cathode material with typical capacities of 120−160 mA·h/g.1,2
The low energy density and/or high cost of these cathode
materials have limited their large-scale production and
application in Li ion batteries. Discovery of new cathode
materials with higher energy density is, thus, a key to realizing
more efficient energy storage systems. Recently, lithium−sulfur
(Li/S)3−5 and lithium−oxygen (Li/O2) cells6−9 have been
demonstrated to possess the potential to provide 2−5 times the
energy density of conventional Li ion cells. However, both Li/S
and Li/O2 cells suffer from poor cycling performance, which
had impeded their commercial utilization. For instance, the
cyclability of Li/O2 cells is limited by severe electrolyte
decomposition and large cell polarization under deep
discharge/charge conditions,10,11 while Li/S cells suffer from
In a previous paper, we demonstrated the potential of a new
class of rechargeable lithium and sodium batteries based on
selenium as the active cathode material.15 The choice of
selenium, which is a d-electron-containing member of group 16
in the periodic table, is due to its higher electrical conductivity
compared to sulfur and high theoretical gravimetric capacity
(678 mA·h/g) and volumetric capacity (3268 mA·h/cm3). In
that work, we reported, for the first time, that Se represented an
attractive cathode material for not only rechargeable lithium ion
batteries but also sodium batteries. Notably, unlike the existing
Na/S batteries that only operate at high temperatures,16,17
these new Se-based electrodes are capable of cycling against Na
at room temperature. This encouraging result indicated that the
selenium system possesses favorable electrochemical properties
compared to its sulfur analogues. However, there are still some
fundamental issues that need to be clarified. For example, the
Received: March 13, 2013
Published: April 30, 2013
© 2013 American Chemical Society
8047
dx.doi.org/10.1021/ja402597g | J. Am. Chem. Soc. 2013, 135, 8047−8056

Journal of the American Chemical Society
Article
Grinding Mill with ball to powder ratio of 7:1. The Se and SeS2 were
purchased from Aldrich and used directly. The SeS7 was synthesized
by mixing SeS2 and S (Aldrich) in a 1:5 ratio and then heating at 160
°C for 16 h. Also, Li2Se was synthesized by lithium triethylborohydride
reduction of selenium pellets in tetrahydrofuran (THF) to serve as a
standard material.22 X-ray diffraction (XRD) patterns indicated that
the desired phases were well crystallized, as shown in Figure 1S
(Supporting Information).
charge and discharge voltages are evolving in the Li/Se system
during the initial cycles in the carbonate-based electrolyte.
Significant polarization occurs once the charge and dicharge
voltages are stabilized after five cycles, which leads to a low
roundtrip efficiency. In addition, the Coulombic efficiency is
quite low during the first 20 cycles. The mechanisms or
underlying reasons for this unsatisfactory performance are still
not well understood due to inadequate characterization of the
battery materials during electrochemical cycling. In particular,
the effects of organic electrolytes on the (de)lithiation process
(i.e., both lithiation and delithiation) of the Li/SeSx cells, if any,
are still not clear, nor are the effects of the sulfur content in
SeSx.
Organic electrolytes are known to be particularly important
in Li/O2 and Li/S batteries. For instance, the electrolyte plays a
key role in Li/O2 cell performance because, at high voltage,
electrolyte decomposition may occur, especially in carbonate-
based electrolyte.10,18 It has also been reported that cycle life
could be extended for Li/S batteries by tuning the composition
of the electrolyte.19,20 Selenium, since it is located in the same
group as oxygen and sulfur in the periodic table, should exhibit
a similar reactivity preference when used as a cathode material.
Moreover, the selenium anion is more nucleophilic than the
sulfur anion, showing even higher reactivity with carbonyl
groups in the traditional carbonate-based electrolyte.21 We,
therefore, expected that performance could be enhanced for the
Li/Se system by utilizing a noncarbonate electrolyte, as is the
case for Li/S and Li/O2 batteries.
In this study, we first adopted an ether-based electrolyte in
the Li/SeSx cell to investigate its effect on the electrochemical
performance. The results demonstrate that cell performance is
significantly improved in terms of voltage profile and
Coulombic efficiency. The voltage profiles of these cells
indicate that complete lithiation of selenium to Li2Se is
occurring through the formation of intermediate phases, i.e.,
Li2Sen. This behavior differs from the single-phase transition in
carbonate-based electrolyte, as reported earlier.15 This result
clearly suggests that cell performance highly depends on the
nature of the electrolyte. In order to better understand the
relationship between the formation of the intermediate phases
and voltage profile of the cells with the ether-based electrolyte,
we performed an in-depth study by different characterization
techniques, including high-energy X-ray diffraction (HEXRD)
and X-ray absorption spectroscopy (XAS). The powerful
capability of in situ HEXRD and XAS, available at the
Advanced Photon Source (APS) at Argonne National
Laboratory, allowed us to monitor instantly the phase
transition, the intermediate phases, and their oxidation states
during cycling. The mechanism behind the (de)lithiation
processes for Li/SeSx cells is proposed here on the basis of
the results obtained from the in situ X-ray probes. This work
has helped us to better understand and correlate the formation
of intermediate phases with the electrochemical performance of
Li/SeSx cells and shines new light on how to improve the cell
performance by tuning the electrolyte composition.
2.2. Electrochemical Characterization. The electrochemical
performance of the selenium−carbon composites was assessed by
fabricating CR2032 coin cells. Metallic Li foil was used as the negative
electrode. The positive electrode was prepared by first mixing a slurry
containing 70 wt % composite material, 10 wt % carbon black, and 20
wt % polyvinylidene fluoride (PVDF) or polytetrafluoroethylene
(PTFE) binder, previously dissolved in N-methyl-2-pyrrolidinone
(NMP), and the proper amount of NMP as dispersant. The slurry was
then coated on aluminum foil using a doctor blade. The NMP was
removed by first drying the electrode at 75 °C for 2 h and then further
drying the electrode in a vacuum oven at 100 °C overnight. A half-cell
was constructed in an argon-filled glovebox. The electrolyte
(designated D2) was 1 M lithium bis(trifluoromethanesulphonyl)-
imide (LiTFSI) in 1,3-dioxolane (DOL):1,2-dimethoxyethane (DME)
(1:1 v/v ratio), and the separator was a Celgard 2325 film. For
comparison, several cells were assembled with a carbonate electrolyte
(GenII): LiPF6 (1.2 M) dissolved in ethylene carbonate (EC)/ethyl
methyl carbonate (EMC) solvents (3:7 v/v ratio). The cells were
tested with a MACCOR electrochemical analyzer at a current density
of 50 mA/g between 0.8 and 4.0 V. Electrochemical tests were
conducted at room temperature and duplicated to check reproduci-
bility. The capacity was calculated on the basis of the active material
mass, which corresponds to 49 wt % of the electrode weight.
Cyclic voltammetry (CV) was conducted at a scan rate of 0.2 mV/s
in the potential range of 0.8−4.0 V on the cycled cell using a frequency
response analyzer (Solartron, model 1400). Electrochemical impe-
dance spectroscopy (EIS) data were collected over a frequency range
from 5 × 105 to 2 × 10−2 Hz, with a potential amplitude of 5 mV on
the Solartron analyzer.
Pouch cells for in situ XAS measurements were fabricated using the
Cell Fabrication Facility at the Argonne National Laboratory with Li
metal as the negative electrode.
2.3. X-ray Diffraction. The HEXRD measurements were carried
out at Beamline 11-ID-C of the APS. The X-ray wavelength was 0.108
Å. In situ HEXRD patterns were collected with custom-designed coin
cells cycled with a MACCOR cycler at a constant rate (about C/6)
between 0.8 and 4.0 V. During the charging and discharging, the XRD
patterns were collected for 1 min, with 15−26 min intervals, using a
Perkin-Elmer 2D X-ray detector. 2D images were converted into a 1D
plot of 2θ versus intensity by using the FIT2D program calibrated
against a CeO2 standard.
2.4. X-ray Absorption Spectroscopy. The XAS measurements
for the Se K-edge (12 658 eV) were made in transmission mode on the
bending-magnet beamline of the APS (20-BM-B and 9-BM-C). The
incident beam was monochromatized by using a Si(111) fixed-exit,
double-crystal monochromator. In situ XAS experiments were made
on the pouch cells cycled with a MACCOR cycler. Standard materials
were mixed with boron nitride and then pressed into a self-supported
pellet sealed in 50-μm-thick Kapton tape. X-ray absorption near edge
structure (XANES) and extended X-ray absorption fine structure
(EXAFS) data reduction followed standard methods using the
ATHENA software package.23 The normalized EXAFS data were
converted from energy to k-space by k2 weighting.
2.5. X-ray Photoelectron Spectroscopy (XPS). Selected
electrodes were analyzed by XPS using a Kratos Axis Ultra DLD
surface analysis instrument. The base pressure of the analysis chamber
during these experiments was 3 × 10−10 Torr, with operating pressures
around 1 × 10−9 Torr. Spectra were collected with a monochromatic
Al Kα source (1486.7 eV) and a 300 × 700 μm spot size. The Al
source was operated at 12 mA of emission current with the target
anode set to 15 kV; the resulting power was 180 W. For survey spectra
the data were collected at a pass energy of 160 eV (fixed analyzer
2. EXPERIMENTAL SECTION
2.1. Preparation and Characterization of the Carbon−
Selenium Composite. Composite materials were prepared by
mixing selenium (or SeSx) with multiwalled carbon nanotubes
(Aldrich) in a 7:3 mass ratio, grinding for 65 h, then heating at 260
°C (or 160 °C for SeS2- and SeS7-containing composites) under argon
for 16 h. The dry grinding was carried out on a SWECO Vibro-Energy
8048
dx.doi.org/10.1021/ja402597g | J. Am. Chem. Soc. 2013, 135, 8047−8056

Journal of the American Chemical Society
Article
Figure 1. Cycle performance of Li cells with (a, b) Se−, (c, d) SeS2−, and (e, f) SeS7−carbon composite as cathodes in ether-based electrolyte. The
second discharge capacities for Li/Se, Li/SeS2, and Li/SeS7 cells are ∼0.6, ∼0.6 and ∼0.9 mA·h/cm2, respectively.
transmission mode), a step size of 1 eV, and a dwell time of 200 ms.
High-resolution regional spectra were collected with pass energy of 20
eV (fixed analyzer transmission mode), a step size of 0.1 eV, and a
dwell time of 300 ms. For low signal-to-noise regions, multiple passes
were made, and the results averaged. Peak position correction was
further corrected by referencing the C 1s peak position of adventitious
carbon for a sample (284.8 eV24) and shifting all other peaks in the
spectrum accordingly.
Fitting was done by using the program CasaXPS. Each relevant
spectrum was fit to a Shirley-type background to correct the rising
edge of the backscattered electrons, which shift the baseline higher at
high binding energies. Peaks were fit as asymmetric Gaussian/
Lorentzian line shapes, with 0−30% Lorentzian character. The full
width at half-maximum of all subpeaks was constrained to 0.7−2.0 eV,
as dictated by instrumental parameters, lifetime broadening factors,
and broadening due to sample charging. With this native resolution
set, peaks were added, and the best fit was obtained from a least-
squares fitting routine.
efficiency during the initial 20 cycles, although much higher
capacity. This behavior is likely due to the greater amount of S
in the SeS7 compound. The Li/(Se + 2S) and Li/S cell were
also tested, as shown in Figures 2S and 3S (Supporting
Information), for comparison. Among the three composite
materials tested in this study, SeS2 showed the most favorable
trade-off among capacity, efficiency, and capacity retention.
Parts b, d, and f of Figure 1 present the first, second, and
30th voltage profiles of these same three cells. These profiles
suggest overall reversible electrochemical reactions for all the
cells. In the case of the Li/Se cell (Figure 1b), two voltage
plateaus are visible above 1.5 V in the discharge profiles with an
inflection point at about 2.03 V vs Li+/Li. This finding implies
that multistep phase transitions take place during the discharge
when the ether-based electrolyte is applied, since a single-step
reaction was detected for the same composite using the
carbonate-based electrolyte, as we reported earlier.15 This
observation indicates that the intermediate phases and their
solubility are varied in different electrolytes. Upon charging, a
single plateau at about 2.2 V is present for the Li/Se cell.
Comparable discharge and charge plateau potentials are also
present for the Li/SeS2 and Li/SeS7 cells. The voltage profiles
shown in Figure 1b,d,f indicate that the polarization of the cells
is significantly reduced in comparison to that for the carbonate-
based electrolyte, which leads to the higher efficiency of the
cells in the ether-based electrolyte. Note that the voltage
variation below 1.5 V in the first discharge profile is ascribed to
the decomposition of the PTFE binder, since it is not observed
with the use of PVDF binder. This conclusion is confirmed by
the XPS analysis on the Se cathode after the first discharge, as
shown in Figure 4S (Supporting Information). A clear Li−F
signal in the F 1s XPS spectrum is present due to the
decomposition of the PTFE binder, which is less stable than
PVDF in the low-voltage range when used as a binder for the
3. RESULTS AND DISCUSSION
3.1. Electrochemical Performance in Ether-Based
Electrolyte. Active cathode materials containing Se, SeS2,
and SeS7 combined with carbon were tested against Li metal
anodes between 0.8 and 4.0 V, and the capacity retention over
50 cycles is shown in parts a, c, and e of Figure 1, respectively.
The ether-based electrolyte D2, i.e., 1.0 M LiTFSI in DOL/
DME, was selected for this study due to its greater stability
compared to the carbonate-based electrolyte in Li/O2 and Li/S
systems. A discharge capacity of 350, 571, and 833 mA·h/g for
the Li/Se, Li/SeS2, and Li/SeS7 cells, respectively, is maintained
for over 50 cycles. The cell capacity increases with increasing S
content in the composites due to its contribution to the overall
capacity. The capacity faded a little for all three cells tested. The
Coulombic efficiency was nearly 100% for the Li/Se and Li/
SeS2 cells. The Li/SeS7 cell showed a relatively low Coulombic
8049
dx.doi.org/10.1021/ja402597g | J. Am. Chem. Soc. 2013, 135, 8047−8056

Journal of the American Chemical Society
Article
Figure 2. Cyclic voltammetry curves of (a) Li/Se and (b) Li/SeS2 batteries in ether-based electrolyte with PVDF binder.
Figure 3. HEXRD patterns of Li/Se cell during the first cycle (a, c) with the voltage profile (b). HEXRD patterns of Li/SeS2 cell in the first cycle (d,
f) with the voltage profile (e).
anode applications.25 We only observed the decomposition of
PTFE in the first discharge, and the cell reaction stabilized
afterward. Because the PTFE-containing cathode had showed
better capacity retention and cycle life than those with PVDF
binder,26 we adopted this binder for cycle testing of the cells.
To avoid the complexity introduced by the PTFE binder during
the further characterizations of the cells, including CV and in
situ XAS, as presented below, PVDF binder was applied
instead.
The CV curves for the Li/Se and Li/Se2S cells are shown in
Figure 2. In the Li/Se system, the CV starts from an open
circuit voltage of 3.2 V and sweeps to 0.8 V during the initial
discharge, converting elemental selenium to selenides with two
reduction peaks at 1.83 and 2.02 V. These features are
analogous to those of Li/S cells, in which two separate
reduction peaks at 2.4 and 2.1 V were observed, corresponding
to the conversion of high-order lithium polysulfides (e.g., Li2S8)
to low-order lithium polysulfides (Li2Sx, 8 > x ≥ 4) and then to
Li2S2/Li2S, respectively. In the subsequent anodic sweep to 4.0
V, only one oxidation peak at 2.32 V is observed for the Li/Se
system. In the Li/S system, some studies suggest that since only
one oxidation peak is observed at 2.5 V, the “recovered”
2‑
product after the charge should be intermediate S8 instead of
elemental S.27−29 On the basis of the CV results and the
similarity between Se and S, we can propose a mechanism for
the electrochemical reactions in the Li/Se cell that has been
further confirmed by the in situ X-ray probes, as will be
presented in the later sections of this paper.
According to this mechanism, during the discharge,
elemental Se is first reduced to lithium polyselenide Li2Sen (n
≥ 4), which is further reduced to Li2Se2 and Li2Se, while the
discharge product Li2Se is oxidized into intermediate
polyselenides Li2Sen (n ≥ 4) and Se during the subsequent
charge. In addition, the two reduction peaks of the following
two cycles are relatively stable, while the oxidation activity
decreases significantly, indicating improved Coulombic effi-
ciency with cycling. As for the Li/SeS2 cell, the CV curve shows
four reduction peaks: peaks at 2.32 and 2.03 V can be assigned
to the conversion of high-order polysulfides (due to the
introduction of sulfur) to low-order polysulfides and then to
8050
dx.doi.org/10.1021/ja402597g | J. Am. Chem. Soc. 2013, 135, 8047−8056

Journal of the American Chemical Society
Article
Figure 4. (a) Normalized Se XANES spectra of standard Se and Li2Se, (b) Se EXAFS spectra of standard Se and Li2Se, and also selected normalized
Se XANES spectra of Li/Se cell during discharge (c) and charge (d).
Li2S2 and Li2S, and peaks at 2.10 and 1.95 V can be attributed
to the conversion of Se to polyselenides and then to Li2Se.
Correspondingly, two oxidation peaks at 2.30 and 2.39 V
during the charge sweep are attributed to the oxidation of Li2Se
and Li2S, respectively. To better understand the lithium
(de)insertion mechanism of the Se-containing material in the
ether-based electrolyte, in situ HEXRD and XAS investigations
were carried out on Se and SeS2 composites.
3.2. In Situ HEXRD Analysis. In situ HEXRD is a powerful
technique for monitoring reacting systems, since it can produce
high-quality data in a very short time period (e.g., 1 min or even
less depending on the material properties). Specially designed
coin cells were used for the in situ HEXRD studies. High-
energy X-rays, which can penetrate the Al current collector and
the Li foil anode, make the measurements possible. The
resulting HEXRD patterns revealed a few very strong peaks at
2.65°, 3.05°, and 4.32° for the Al current collector and 2.50°,
3.53°, and 4.32° for the Li anode. Figure 3 shows the HEXRD
patterns for the Se and SeS2 composites in lithium coin cells
during cell cycling. As shown in Figure 3a,b, the peak intensity
of the starting material, Se, decreased, while the Li2Se peaks
were growing during the discharging, and the process was
reversed during charging, even though the peak intensities for
Se and Li2Se are relatively low compared to the peaks for the Al
current collector.
grow steadily with accompanying disappearance of Li2Se.
During the discharge, the amount of crystalline Se decreases
with Li insertion; however, Li2Se does not form until a capacity
of ∼500 mA·h/g is reached without any other phases detected.
At this point, several questions arose concerning where the Li+
goes during discharging, why Li2Se forms only after a critical
potential of 1.25 V, and/or whether Li2Se forms earlier as an
amorphous phase. We could not resolve these questions
conclusively on the basis of the HEXRD analysis, since it can
only detect crystalline phases.
For SeS2, as shown in Figure 3d−f, no crystalline phase is
present at the beginning of discharge. When the cell is
discharged to 2.08 V, the Li2S phase first appears followed by
Li2Se crystallization at ∼2.04 V. During the delithiation, Li2Se
and Li2S remain until the cell is charged to 2.19 and 2.30 V,
respectively. When the cell is charged to 2.25 V, Se peaks
appear and keep increasing to the end of charge. These
phenomena suggest that the charge product is a mixture of Se
and S, but the S is not detected due to its weak diffraction
ability. Other evidence for the charge products come from the
high capacity of the Li/SeS2 cell obtained in the second cycle
and the HEXRD of the second discharge, which show the
presence of Li2Se and Li2S (Figure 5S, Supporting
Information).
3.3. In Situ XANES Analysis. Synchrotron radiation based
XAS is a well-established method to probe the environment
around atoms of all states of matter.30 In situ XAS can be very
informative when applied to a complex battery system that
involves liquid and solid, as well as crystalline and/or
amorphous phases. Also XANES is especially useful for
determining the average oxidation state of the elements in all
phases. With these considerations in mind, we carried out XAS
measurements during cycling of a pouch cell. As shown in
Figure 4a, the two curves are XANES from commercial
An enlarged portion of the HEXRD patterns (Figure 3c)
exhibits additional details about the phase change. The intensity
of the Se peaks (1.64° and 2.06°) decreases slowly with the
lithiation process, while no new phase appears until the cell is
discharged to ∼1.69 V. The voltage quickly drops to ∼1.25 V
after a 26-min measurement interval. The Li2Se phase starts to
appear when the Se peaks have completely vanished. During
the charge, the Li2Se phase remains until the potential reaches
∼2.22 V. When charging to 2.24 V, the Se structure starts to
8051
dx.doi.org/10.1021/ja402597g | J. Am. Chem. Soc. 2013, 135, 8047−8056

Journal of the American Chemical Society
Article
Figure 5. (a) Normalized XANES spectra of Li/Se cell during cycling, (b) battery voltage profile, and (c) derivative of normalized XANES spectra of
Li/Se cell during cycling.
Figure 6. Se XANES spectra linear combination fitting, (a) XANES 19th spectrum fitted with two components (Se and Li2Se) showing huge residue,
(b) reduced χ2 × 1000 of the linear combination fitting of XANES spectra using different component combinations, (c) XANES 19th spectrum fitted
with three components (Li2Se4, Li2Se2, and Li2Se), and (d) relative composition evolution of possible phases during cycling.
selenium powder and synthesized Li2Se mixed with boron
nitride. The selenium K-edge absorption, at 12 658 eV, arises
from the transition of Se 1s core electrons to the unoccupied 4p
states.31 The Se K-edge of Li2Se shifts to a higher energy at 12
660 eV, which is inconsistent with the commonly observed
phenomenon that the edge shifts to a higher energy with the
increasing oxidation state.31,32 However, that the Se K-edge of
Li2Se shifts to a high energy might also be explained by a
reduced screening effect, as in the S case.20 The S K-edge of
Li2S shifted to a higher energy, as reported by Abruna et al. in
2011, due to the strong Coulombic interaction between the S2‑
ion and the eight neighboring Li+ ions, which reduces the
screening effect.20,33 In addition, the change in composition
from Se to Li2Se can be seen in k-space as the shape of the
EXAFS spectra changes significantly (Figure 4b). Figure 4c,d
compare selected XANES spectra from the first discharge and
charge of the Li/Se cell. To better illustrate the XANES spectra
evolution, 2D contour plots of Se XANES spectra recorded
during the cell cycling are presented in Figure 5a. In addition,
the Li/Se battery voltage profile is displayed on the side of the
XANES spectra (Figure 5b). During lithiation, the critical
points (at 2.04 and 1.67 V) of the electrochemical process are
well-correlated with the evolution of the XANES spectra. For
example, the inflection point at 2.04 V in the voltage profile
coincides with the shift of the XANES spectrum to lower
energy. The precipitous voltage drop at around 6 h of
discharging corresponds to the sudden appearance of the Se
K-edge of Li2Se.
Generally, the energy position of the edge is determined as
the lowest energy peak of the first derivative of the XANES
spectra. To determine the edge positions, we show the first
derivatives of the XANES spectra in Figure 5c. This plot shows
that the Se edges are shifted significantly with (de)lithiation.
When the cell is first discharged through the inflection point at
2.04 V, the edge shifts from 12 657.9 to 12 656.7 eV. In the
following 3 h (more than 300 mA·h/g capacity), the edge
positions do not change much and are accompanied by the
voltage plateau around 2 V. When the cell is discharged to 1.67
8052
dx.doi.org/10.1021/ja402597g | J. Am. Chem. Soc. 2013, 135, 8047−8056

Journal of the American Chemical Society
Article
Figure 7. (a) Normalized XANES spectra, (b) voltage profile, and (c) derivative of normalized XANES spectra of Li/SeS2 cell during cycling.
Figure 8. XPS spectra of Se cathodes with different electrolytes: (a) Se 3d/Li 1s and (b) Se 3p for 1 M LiTFSI in DOL/DME (D2) as electrolyte,
and (c) Se 3d/Li 1s and (d) Se 3p for 1.2 M LiPF6 in EC/EMC (GenII) as electrolyte.
V, the Se K-edge jumps to 12 660 eV, suggesting the formation
of Li2Se. During charge, the Se edges slowly shift back to lower
energy. These observations indicate that polyselenides might
have formed during the cycling, but in a noncrystalline state;
hence, they are not detected by XRD. At the beginning of the
discharge, Se reacts with Li+ ions, forming polyselenides and
leading to the decrease of Se concentration seen in the in situ
HEXRD studies. As a consequence, the reduced Se in the
lithium polyselenides shifts the Se K-edge to a lower energy, as
expected. With increasing lithiation, the elemental Se is
depleted, and polyselenides are reduced to crystallized Li2Se,
which contains Se2‑ ions in an environment surrounded by
eight Li+. The strong Coulombic interaction significantly
reduces the screening effect and shifts the Se edge to a higher
energy.
combination fitting is conveniently available in the Athena
software package.23 By selecting XANES spectra of the starting
materials and discharge products as standards, we can
determine the composition of the electrodes at different states
of charge by linear combination fitting. With the hypothesis
that we have a two-phase, one-step transition from Se to Li2Se
directly, we first fit all the spectra using the spectrum of the
pristine electrode and the 25th spectrum measured at the fully
discharged point as standards. Unfortunately, most of the
spectra could not be well fitted; Figure 6a is an example of a
poorly fitted spectrum, as illustrated by the large reduced χ2
from 0.020‰ to 5.676‰ shown in Figure 6b (purple squares).
However, when carrying out the fitting on any combination of
Se, Li2Se, Li2Sen (n ≥ 4), and Li2Se2 as standards, we achieved
greatly improved fits, with reduced χ2 values from 0.017‰ to
0.07‰, as also shown in Figure 6b. Details of the standards
selection procedure are discussed in the Supporting Informa-
tion. An example of a good fit is plotted in Figure 6c. The
Another common use of XANES is for fingerprinting by
linear combination fitting of XANES spectra of known species
to determine the components in a mixture.34 Linear
8053
dx.doi.org/10.1021/ja402597g | J. Am. Chem. Soc. 2013, 135, 8047−8056

Journal of the American Chemical Society
Article
relative composition of each component in the process of
cycling is shown in Figure 6d. The exact value of the
composition may not be completely accurate, considering the
standards selection procedure. However, the trend for each
component’s evolution is of significance. For example, with Se
composition decreasing, Li2Sen (n ≥ 4) forms, increases in
content, and then decreases with Li2Se2 formation (Figure 6d).
During discharge, Li2Sen (n ≥ 4), Li2Se2, and Li2Se coexist. The
fully charged electrode has a composition of 74% Se, 2% Li2Sen
(n ≥ 4), and 24% Li2Se, as indicated by the fully charged Se
edge being about 0.2 eV higher than the pristine Se edge. This
finding suggests that some Li2Se might have formed on the
anode side due to the dissolution of polyselenides. This
hypothesis is supported by Raman spectra, as Li2Se was
observed on the Li anode of the charged cells (Figure 6S,
Supporting Information).
In situ Se XANES measurement on a Li/SeS2 cell was also
carried out at APS Sector 9. A 2D contour plot of Se XANES
spectra of the cycled cell is shown in Figure 7a, with the
corresponding voltage profile plotted in Figure 7b. These
results further confirm that SeS2 is reduced to Li2Se and Li2S
with the Se K-edge shifting to higher energy. The 2D contour
plot of the first derivative of the XANES spectra also exhibits a
slight shift of the Se K-edge to lower energy, with the formation
of lithium polyselenides, followed by a jump to higher energy,
indicating the formation of Li2Se (Figure 7c), as observed in
the Li/Se case.
reaction between Se and Li2Se is also observed during the first
cycle, but at the end of the discharge, the presence of a peak at
lower binding energy can be related to a reduced selenium
species other than Li2Se. After the first charge, the peaks at
higher binding energy could be attributed to Se−O species.
The Se−O species might form as a result of the reaction of Se
anions with the carbonyl groups of the carbonate solvent. Such
reactions, because of the formation of an insulating layer on the
cathode surface that impedes the transportation of lithium ions
and electrons, will lead to poor cycle performance and large
overpotential, as reported earlier.15
Electrochemical impedance spectroscopy (EIS) provides
further evidence for the formation of an insulating layer on
the electrode surface of Li/Se cell with GenII electrolyte. The
EIS data were obtained right after the cell assembly and after
five cycles with D2 and GenII electrolyte (selected voltage
profiles with different electrolyte are compared in Figure 7S,
Supporting Information). The Nyquist plots for the Li/Se cells
(Figure 9) are composed of a semicircle at high frequencies and
3.4. XPS and EIS Analysis. We also performed XPS
analysis, a surface sensitive technique that can provide valuable
information about the solid reaction products on the cathode
surface. We derived high-resolution C 1s, F 1s, Li 1s, Se 3d, and
Se 3p XPS spectra of the cathodes from the Li/Se system with
the two electrolytes. The C 1s peak position of adventitious
carbon at 284.8 eV was used as an internal reference. The fitted
Li 1s, Se 3d, and Se 3p XPS spectra are shown in Figure 8. Note
that, because the XPS spectra of Li 1s and Se 3d overlap, we
provided XPS spectra of Se 3p as additional evidence for Se
evolution between discharge and charge.
Figure 8a,b shows the Se 3d (Li 1s) and Se 3p XPS spectra
for cathodes recovered from the first discharge and first charge
using the ether-based (D2) electrolyte. For comparison, the
XPS spectra of the pristine Se cathode and Li2Se powder are
also shown at the bottom of these figures. In the pristine Se
cathode, Se 3d5/2 and 3d3/2 peaks (Figure 8a) located at ∼55.2
and ∼56.1 eV with a spin−orbit splitting of 0.86 eV are
attributed to metallic selenium.24 The Se 3d5/2 peak moves
toward higher binding energy (about 55.6 eV) when Se binds
with Li, forming Li2Se. This trend is consistent with the Se 3p
(Figure 8b) XPS spectra, as well as with the XANES results
presented earlier. On the basis of both Se 3d/Li 1s and Se 3p
XPS spectra, as shown in Figure 8a,b, Li2Se appears to form as
the dominant discharge product of the electrochemical
reactions. After the first charge, the Li 1s signals from Li2Se
completely disappear (Figure 8a), and the Se 3p peak moves
back to the binding energy for metallic Se (Figure 8a),
suggesting that Li2Se reversibly oxidizes to Se. This encouraging
result suggests that the desired discharge products (Li2Se) can
be produced without any side reactions on the electrode surface
when the D2 electrolyte is used. It is noteworthy that this is not
the case with carbonate-based electrolyte.
Figure 9. Nyquist plots after fabrication of Li/Se cell and after five
cycles with LiTFSI in DOL/DME (D2) or LiPF6 in EC/EMC (GenII)
as electrolyte.
an inclined line at low frequency. The intercepts of the
semicircles and the real axis can be used to determine the
resistance of the electrolyte (Re, in the high-frequency region)
and the resistance to charge transfer (Rct, at relatively lower
frequency). The data show that the resistances (both Re and
Rct) of the Li/Se cell with D2 electrolyte barely changed after
five cycles. For the cell with GenII electrolyte, the Re of the
fresh cell is comparable to that of the cell with D2 electrolyte,
but it increased after five cycles, indicating the instability of the
carbonate-based electrolyte. The Rct of the fresh cell in the G2
electrolyte is more than twice that of the cell with D2
electrolyte and increases to more than 10 times larger with the
D2 electrolyte after five cycles. These findings can be ascribed
to an insulting layer formed on the cathode, as observed in the
XPS analysis. These results further illustrate the advantage of
D2 electrolyte over GenII electrolyte.
4. (DE)LITHIATION MECHANISM
We can now propose a mechanism for the Li/Se cell lithiation
process by correlating the voltage profiles, the linear
combination fitting of the XANES spectra, and the HEXRD
analysis. Figure 10 shows the proposed mechanism in the form
of a voltage profile with the associated phases that form during
charge and discharge. Selenium is first reduced to Li2Sen (n ≥
4) at 2.04 V, which is then further reduced to Li2Se2 and Li2Se
after discharging to 2.06 and 0.8 V, respectively. During
After the first discharge and the first charge, the Se 3d (Li 1s)
and Se 3p XPS spectra of cathodes using the GenII electrolyte
showed extra peaks, as evident in Figure 8c,d. The reversible
8054
dx.doi.org/10.1021/ja402597g | J. Am. Chem. Soc. 2013, 135, 8047−8056

Journal of the American Chemical Society
Article
HEXRD. Complementary XANES spectra analysis provides
detailed evidence on the evolution of the selenium valence state
in both crystalline and noncrystalline phases during cycling. It is
proved that Se is first reduced into polyselenides, Li2Sen (n ≥
4), then Li2Se2, and finally Li2Se with discharge. During the
charge, Li2Se is directly oxidized to Li2Sen (n ≥ 4) and Se
without the formation of Li2Se2. The existence of the
polyselenides is further manifested by a solubility test of
oversaturated Se and Li2Se solutions in the electrolyte solvent
and a mixture of the solutions. The solubility of polyselenides
could facilitate the electrochemical reaction in the Li/Se cell but
may cause the deposition of Li2Se on the lithium anode. In
addition, we demonstrated that the performance of the Li/SeSx
system is greatly improved by using the ether-based electrolyte.
In D2 electrolyte, the Li/SeSx cell shows reversible cycle
performance with a longer cycle life and improved efficiency
compared to those assembled with carbonate-based electrolyte.
For Li/SeS2 and Li/SeS7 cells, Li2Se and Li2S are the discharged
products, with the presence of Se only as the crystalline phase
in the end of charge. More than 350, 571, and 833 mA·h/g
discharge capacities were retained after 50 cycles for Li/Se, Li/
SeS2, and Li/SeS7 cells, respectively. Carbonate-based electro-
lyte is unstable for Li/SeSx systems, as shown by XPS and
confirmed further by EIS measurements, which showed
significant increase of the cell impedance. This phenomenon
is quite comparable to the Li/O2 and Li/S battery systems,
where the instability of the carbonate-based electrolyte has
been reported.
Figure 10. Representation of cathode phase evolution during charge
and discharge of a Li/Se cell in an ether-based electrolyte.
charging, Li2Se is directly oxidized to Li2Sen (n ≥ 4) and then
Se. We believe that the only oxidation peak in the CV curve
(Figure 2a) actually corresponds to two processes: Li2Se to
Li2Sen (n ≥ 4) and Li2Sen (n ≥ 4) to Se. Due to the different
kinetics of the processes, the two oxidation peaks completely
overlap. This mechanism is quite similar to that of the Li/S
battery, which shows two reduction peaks, ascribed to the
reduction of high-order polysulfide (e.g., Li2S8) to low-order
polysulfides (Li2Sx, 8 > x ≥ 4) and polysufides to Li2S2 and
Li2S, where the one oxidation peak is attributed to the oxidation
of Li2S to polysulfides without the formation of S.27−29 In our
HEXRD and XANES results, we did see the Se recovery after
charge, as well as the coexistence of Li2Sen (n ≥ 4), Li2Se, and
Se. This difference between S and Se is related to the high
atomic number of Se, which is easily detected by various
techniques, such as XRD and XANES, in contrast to S.
In this study, we carried out a comprehensive investigation of
selenium-based lithium battery systems. In situ XANES
provides valuable information on the oxidation states of
crystalline and amorphous phases, including intermediate
species dissolved in the electrolyte. A mechanism is proposed
by a combination of in situ synchrotron X-ray techniques and
electrochemical methods. These results also highlight the
importance of appropriately combining advance measurement
techniques and the wide range of issues in other lithium ion
battery systems or other complicated processes that could be
investigated by these combined techniques.
To further demonstrate the formation of lithium poly-
selenides, we completed a solubility test. We added 3.95 mg of
Se to 5 mL of DOL and DME solvent (equivalent to 0.01 mol/
L solution) in one container and 4.64 mg of Li2Se to 5 mL of
the same solvent in a second container (again equivalent to
0.01 mol/L solution). As shown in the inset of Figure 10,
neither material is completely dissolved when present alone, the
solubility of Se being extremely low. However, the precipitated
Se particles did dissolve when we added a few drops of
saturated Li2Se solution to the Se suspension, and the solution
took on a dark brown color. The intense color of the solution
suggests the formation of highly soluble polyselenides.35
Polyselenide speciation studies had been carried out as early
as 1977 with various techniques, such as ultraviolet−visible
spectroscopy, nuclear magnetic resonance, electrospray mass
spectrometry, and electrochemical techniques.35−40 Accord-
ASSOCIATED CONTENT
■
S
* Supporting Information
XRD patterns for synthesized SeS7 and Li2Se, cycle perform-
ance of Li/S and (Se + 2S)−carbon composite materials, F 1s
XPS spectra of pristine Se cathode with PTFE binder and after
first discharge, in situ HEXRD patterns of Li/SeS2 cell in the
second cycle, XANES spectra linear combination fitting
polyselenide standards selection procedures, Raman spectra of
Li2Se powder and a charged anode recovered from a Li/Se cell
with D2 electrolyte, and voltage profile comparation of Li/Se
cell with LiTFSI in DOL/DME (D2) or LiPF6 in EC/EMC
(GenII) as electrolyte. This material is available free of charge
via the Internet at http://pubs.acs.org.
35,37,38,40
2‑
ingly, Sen species with n = 1−8 were identified.
This
test confirmed that the lithium polyselenides do exist in D2
electrolyte and are more soluble than either Se or Li2Se alone.
We believe that the formation of soluble polyselenides could
facilitate the Li/Se battery redox reaction, but it may also cause
some dissolutive transport of Se to the anode, which would
result in the loss of active cathode material, as observed in the
Li/S system.14,41
AUTHOR INFORMATION
■
5. CONCLUSIONS
Corresponding Author
abouimrane@anl.gov
A combination of HEXRD and XANES analysis has been used
to investigate the (de)lithiation mechanism of the Li/SeSx cell
system. The total conversion of Se into Li2Se with discharge
and recovery of Se with charge was monitored by in situ
Notes
The authors declare no competing financial interest.
8055
dx.doi.org/10.1021/ja402597g | J. Am. Chem. Soc. 2013, 135, 8047−8056

Journal of the American Chemical Society
Article
(28) Fu, Y.; Manthiram, A. RSC Adv. 2012, 2, 5927.
(29) Yamin, H.; Penciner, J.; Gorenshtain, A.; Elam, M.; Peled, E. J.
Power Sources 1985, 14, 129.
(30) Koningsberger, D. C.; Prins, R. X-ray Absorption: Principles,
Applications, Techniques of EXAFS, SEXAFS and XANES; John Wiley
and Sons: New York, 1988.
(31) Myneni, S. C. B.; Tokunaga, T. K.; Brown, G. E. Science 1997,
278, 1106.
(32) Pickering, I. J.; Brown, G. E.; Tokunaga, T. K. Environ. Sci.
Technol. 1995, 29, 2456.
(33) Mori, R. A.; Paris, E.; Giuli, G.; Eeckhout, S. G.; Kavcic, M.;
Zitnik, M.; Bucar, K.; Pettersson, L. G. M.; Glatzel, P. Anal. Chem.
2009, 81, 6516.
(34) Lengke, M. F.; Ravel, B.; Fleet, M. E.; Wanger, G.; Gordon, R.
A.; Southam, G. Environ. Sci. Technol. 2006, 40, 6304.
(35) Ahrika, A.; Paris, J. New J. Chem. 1999, 23, 1177.
(36) Sharp, K. W.; Koehler, W. H. Inorg. Chem. 1977, 16, 2258.
(37) Lyons, L. E.; Young, T. L. Aust. J. Chem. 1986, 39, 511.
(38) Raymond, C. C.; Dick, D. L.; Dorhout, P. K. Inorg. Chem. 1997,
36, 2678.
ACKNOWLEDGMENTS
■
Work done at Argonne and use of the Advanced Photon Source
and the Center for Nanoscale Materials at Argonne National
Laboratory was supported by the U.S. Department of Energy,
Office of Science, Office of Basic Energy Sciences, under
Contract No. DE-AC02-06CH11357. The pouch cells were
produced at the U.S. Department of Energy’s (DOE) Cell
Fabrication Facility, Argonne National Laboratory. The Cell
Fabrication Facility is fully supported by the DOE Vehicle
Technologies Program (VTP) within the core funding of the
Applied Battery Research (ABR) for Transportation Program.
We thank Xiangyi Luo for part of in situ HEXRD data
collection. J. Lu was supported by the Department of Energy
(DOE) Office of Energy Efficiency and Renewable Energy
(EERE) Postdoctoral Research Award under the EERE
Vehicles Technology Program administered by the Oak Ridge
Institute for Science and Education (ORISE) for the DOE.
ORISE is managed by Oak Ridge Associated Universities
(ORAU) under DOE contract number DE-AC05-06OR23100.
(39) Goldbach, A.; Johnson, J.; Meisel, D.; Curtiss, L. A.; Saboungi,
M. L. J. Am. Chem. Soc. 1999, 121, 4461.
(40) Cusick, J.; Dance, I. Polyhedron 1991, 10, 2629.
(41) Hassoun, J.; Scrosati, B. Adv. Mater. 2010, 22, 5198.
REFERENCES
■
(1) Tarascon, J. M.; Armand, M. Nature 2001, 414, 359.
(2) Whittingham, M. S. Chem. Rev. 2004, 104, 4271.
(3) Rao, M. L. B. U.S. Patent 3413154, 1968.
(4) Ji, X.; Lee, K. T.; Nazar, L. F. Nat. Mater. 2009, 8, 500.
(5) Zhang, B.; Qin, X.; Li, G. R.; Gao, X. P. Energy Environ. Sci. 2010,
3, 1531.
(6) Abraham, K. M.; Jiang, Z. J. Electrochem. Soc. 1996, 143, 1.
(7) Ogasawara, T.; Debart, A.; Holzapfel, M.; Novak, P.; Bruce, P. G.
J. Am. Chem. Soc. 2006, 128, 1390.
(8) Read, J. J. Electrochem. Soc. 2002, 149, A1190.
(9) Peng, Z. Q.; Freunberger, S. A.; Chen, Y. H.; Bruce, P. G. Science
2012, 337, 563.
(10) Freunberger, S. A.; Chen, Y.; Peng, Z.; Griffin, J. M.; Hardwick,
L. J.; Barde,
8040.
́ ́
F.; Novak, P.; Bruce, P. G. J. Am. Chem. Soc. 2011, 133,
(11) Padbury, R.; Zhang, X. J. Power Sources 2011, 196, 4436.
(12) Yeon, J.-T.; Jang, J.-Y.; Han, J.-G.; Cho, J.; Lee, K. T.; Choi, N.-
S. J. Electrochem. Soc. 2012, 159, A1308.
(13) Mikhaylik, Y. V.; Akridge, J. R. J. Electrochem. Soc. 2004, 151,
A1969.
(14) Yamin, H.; Gorenshtein, A.; Penciner, J.; Sternberg, Y.; Peled, E.
J. Electrochem. Soc. 1988, 135, 1045.
(15) Abouimrane, A.; Dambournet, D.; Chapman, K. W.; Chupas, P.
J.; Weng, W.; Amine, K. J. Am. Chem. Soc. 2012, 134, 4505.
(16) Rydh, C. J.; Sanden, B. A. Energy Convers. Manage. 2005, 46,
1957.
(17) Lu, X.; Kirby, B. W.; Xu, W.; Li, G.; Kim, J. Y.; Lemmon, J. P.;
Sprenkle, V. L.; Yang, Z. Energy Environ. Sci. 2013, 6, 299.
(18) Mizuno, F.; Nakanishi, S.; Kotani, Y.; Yokoishi, S.; Iba, H.
Electrochemistry 2010, 78, 403.
(19) Chang, D.-R.; Lee, S.-H.; Kim, S.-W.; Kim, H.-T. J. Power
Sources 2002, 112, 452.
(20) Gao, J.; Lowe, M. A.; Kiya, Y.; Abruna, H. D. J. Phys. Chem. C
2011, 115, 25132.
(21) Steinmann, D.; Nauser, T.; Koppenol, W. H. J. Org. Chem. 2010,
75, 6696.
(22) Detty, M. R.; Seidler, M. D. J. Org. Chem. 1982, 47, 1354.
(23) Ravel, B.; Newville, M. J. Synchrotron Radiat. 2005, 12, 537.
(24) Moulder, J. F.; Stickle, N. F.; Sobol, P. E.; Bomben, K. D. (Eds.)
Handbook of X-ray Photoelectron Spectroscopy; Physical Electronics Inc.
(PHI): Eden Prairie, MN, 1995.
(25) Liu, W.; Huang, X.; Li, G.; Wang, Z.; Huang, H.; Lu, Z.; Xue, R.;
Chen, L. J. Power Sources 1997, 68, 344.
(26) Manickam, M.; Takata, M. Electrochim. Acta 2003, 48, 957.
(27) Ji, X.; Nazar, L. F. J. Mater. Chem. 2010, 20, 9821.
8056
dx.doi.org/10.1021/ja402597g | J. Am. Chem. Soc. 2013, 135, 8047−8056