www.advenergymat.de
FULL PAPER
www.MaterialsViews.com
Solid State Enabled Reversible Four Electron Storage
Thomas A. Yersak, H. Alex Macpherson, Seul Cham Kim, Viet-Duc Le, Chan Soon Kang,
Seoung-Bum Son, Yong-Hyun Kim, James E. Trevey, Kyu Hwan Oh, Conrad Stoldt,
and Se-Hee Lee*
advantage of a sulfide based glass electrolyte to address the problems commonly
associated with FeS2’s rapid capacity fade
at lower temperatures. Along the way, we
identify orthorhombic-FeS2 (marcasite) as
a charge product and use this discovery to
come to a better understanding of the FeS2
conversion chemistry.
The effort to design lower temperature FeS2 batteries focuses on the management of electro-active species formed
upon full charge (3.0 V versus Li+/Li) and
full discharge (1.0 V versus Li+/Li). Two
particularly troublesome species are polysulfides (Sn2−) and
elemental iron (Fe0). To prevent diffusion and agglomeration
of Fe0 nanoparticles in conventional cells, a variety of polymer
electrolytes have been employed with limited success.[5,7] A
similar approach is applied to the confinement of intermediate
polysulfides in conventional S/Li batteries. Notable methods
for addressing polysulfide dissolution and Li2S irreversibility
include polysulfide adsorption on high surface area CMK-3
nano-porous carbon electrodes,[8] polymer electrolytes,[9] and
polyacrylonitrile-sulfur composites.[10] Another approach–often
used concurrently with the previously mentioned methods–
is to limit the upper and/or lower voltage limits of the FeS2
cells.[4–7,11] In this manner, the formation of Fe0 and Sn2− is limited by avoiding full discharge and/or charge. However, limiting
the cell voltage range diminishes achievable energy density.
The basic nature of a solid-state battery architecture makes
it ideal for the confinement of electro-active species. Highly
reversible solid-state sulfur cells have been demonstrated, albeit
with a large volume of carbon additive[12] and, FeS2 and Li2FeS2
were both utilized reversibly as a solid-state anode.[13,14] Despite
the advantage of good electro-active species confinement, solid
electrolytes still have conductivities that can be several orders of
magnitude lower than that of organic liquid electrolytes at room
temperature. Yet, promising new solid electrolytes are increasingly demonstrating higher conductivities up to 10−2 S cm−1 at
room temperature.[15] Sulfide based glass-ceramic solid electrolytes are also stable at elevated temperatures unlike organic
liquid electrolytes which are troublesome at high temperatures.
Due to the inherent safety advantages of a solid-state construction, all-solid-state lithium batteries (ASSLBs) can utilize
a metallic lithium anode (3861 mAh g−1) and operate over
a broader temperature range than conventional Li-ion batteries.[16] With the higher safety, energy density, and voltage
enabled by using a lithium anode, ASSLBs have the potential
to usurp other leading technologies for integration into a wide
We report that a solid-state battery architecture enables the reversible, four
electron storage of fully utilized solvothermally synthesized cubic-FeS2
(pyrite). With a sulfide based glass electrolyte we successfully confine electroactive species and permit the safe use of a lithium metal anode. These FeS2/
Li solid-state cells deliver a theoretical specific capacity of 894 mAh g−1 at
60 °C. We find that nanoparticles of orthorhombic-FeS2 (marcasite) are generated upon recharge at 30–60 °C which explains a coincident change in rate
kinetics.
1. Introduction
The late 1980s and early 1990s saw research efforts focused on
the development of molten salt FeS2/Li-Al batteries for transportation applications. Given that these cells required an operating
temperature around 400 °C, thermal FeS2 battery research was
soon abandoned in favor of the emerging room temperature
lithium-ion and lithium-polymer technologies.[1] Yet interest
in FeS2 remained as it is inexpensive, environmentally benign
and energy-dense. The FeS2 four electron conversion reaction
exhibits a theoretical capacity of 894 mAh g−1 with two discharge plateaus at roughly 2.1 and 1.5 V versus Li+/Li. Research
continued to examine the utilization of FeS2 as an active material at ambient to moderate temperatures. To this end, FeS2
has been successfully commercialized in high energy density
primary cells.[2] Unfortunately, the FeS2 conversion chemistry
is irreversible in cells with a variety of polymer or liquid electrolytes at temperatures <135 °C.[3–7] In this paper, we take
T. A. Yersak, H. A. Macpherson, C. Stoldt, S.-H. Lee
Department of Mechanical Engineering
University of Colorado at Boulder
Boulder, CO 80309-0427 USA
E-mail: [email protected]
S. C. Kim, C. S. Kang, S.-B. Son, K. H. Oh
Department of Materials Science and Engineering
Seoul National University
Seoul, 151-742, Korea
V.-D. Le, Y.-H. Kim
Graduate School of Nanoscience and Technology (WCU)
KAIST, Daejeon 305-701, Korea
J. E. Trevey
HRL Laboratories Limited Liability Company
Malibu, CA 90265-4797 USA
DOI: 10.1002/aenm.201200267
120
wileyonlinelibrary.com
© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Energy Mater. 2013, 3, 120–127
www.advenergymat.de
www.MaterialsViews.com
FULL PAPER
Figure 1. (a) Indexed x-ray diffraction of synthetic cubic-FeS2. (b) FESEM image of synthetic cubic-FeS2 that confirms cubic structure with 2–3 μm
cubes.
variety of portable electronics or large scale applications such as
electric vehicles (EV’s) and hybrid electric vehicles (HEV’s).[15]
The solid-state architecture not only permits the safe use of a
lithium metal anode, but also the reversible full utilization of
FeS2 as a cathode material.
2. Results and Discussion
The morphology of synthetically prepared FeS2 was characterized with field emission scanning electron microscopy
(FESEM) and the crystalline configuration by x-ray diffraction
(XRD). Cu-Kα x-ray diffraction analysis of synthetically prepared FeS2 exhibits diffraction peaks that match well with the
cubic-FeS2 phase (Figure 1a). FESEM images reveal cubic FeS2
particles with ∼2.5 μm wide faces (Figure 1b). Raman spectroscopy also confirms the cubic-FeS2 phase (Supporting Information Figure 1).
Synthetic FeS2 was tested in both a solid-state and liquid
cell configuration. To achieve full utilization of FeS2, the cells
are cycled between 1.0 V and 3.0 V vs. Li+/Li. The results of
cycling at ambient temperature (30 °C) and moderate temperature (60 °C) are presented in Figure 2. Both solid-state cells are
observed to have a stable capacity and a high degree of FeS2
utilization. The gradual increase in capacity with cycling is
observed and attributed to better FeS2 utilization and not utilization of the Li2S solid electrolyte component. This conclusion is
supported by differential capacity (dQ/dV) analysis (Supporting
Information Figure 2). By the 20th cycle, the cell tested at 30 °C
exhibits a discharge capacity of nearly 750 mAh g−1 while the
cell tested at 60 °C exhibits a theoretical discharge capacity of
894 mAh g−1. It is likely that the temperature dependence of the
solid electrolyte’s ionic conductivity contributes to the full FeS2
utilization at 60 °C. At 60 °C, the conductivity of the 77.5Li2S22.5P2S5 solid electrolyte increases to 4.4 × 10−3 Ω−1 cm−1 from
9.17 × 10−4 Ω−1 cm−1 at 30 °C (Supporting Information Figure
3). Improved reaction kinetics of the Fe0 + Li2S/FeS2 conversion
reaction may contribute to better FeS2 utilization as well.
The liquid cells’ discharge capacities rapidly fade upon
cycling. By the 20th cycle, the liquid cell tested at 30 °C exhibits
Adv. Energy Mater. 2013, 3, 120–127
a discharge capacity of only 190 mAh g−1 while the cell tested at
60 °C exhibits no discharge capacity. Decomposition processes
are accelerated at 60 °C leading to such a fast rate of capacity
fade that negligible capacity is observed after the second cycle.
On the other hand, we have just shown that cycling a solidstate FeS2 cell at 60 °C improves its performance. This result
is important when it is considered that most traction battery
packs are designed to operate at temperatures near 60 °C. The
superior performance of solid-state batteries at higher temperatures may reduce the need for extensive thermal management
systems.
It is well documented by Mössbauer spectroscopy, near-edge
X-ray absorption spectroscopy (XANES) and density functional
theory (DFT) simulation that the products of FeS2 reduction
are elemental iron (Fe0) and Li2S.[4,17–20] The initial discharge
of FeS2 has been documented as proceeding in two steps. Each
reaction can occur at one voltage or two depending on the
kinetics of the system.[4]
FeS2 +2Li+ +2e− ↔ Li2 FeS2
+
−
(1)
0
Li2 FeS2 +2Li +2e ↔ 2Li2 S + Fe
(2)
In agreement with the literature, we observe that a cell’s initial discharge profile has one plateau when the cell is cycled at
30 °C (Figure 2a,c) and two plateaus when the cell is cycled at
60 °C (Figure 2b,d). At 30 °C the reduction of cubic-FeS2 particles is limited by the low diffusivity of Li+ into cubic-FeS2 particles such that Equations (1) and (2) proceed simultaneously at
1.5 V versus Li+/Li. At 60 °C, Equations (1) and (2) can proceed
at 1.7 and 1.5 V respectively due to the higher diffusivity of Li+.
The shoulder at 1.3 V in the ambient temperature liquid cell’s
initial discharge profile (Figure 2c) is attributed to a new phase
related to the reaction of Fe0 with organic liquid electrolyte.[21]
The initial discharge profile for each FeS2 cell is different from
subsequent discharge profiles. We will propose that the change
in discharge profiles is due to the formation of nano-crystalline
orthorhombic-FeS2 particles at full charge.
As stated earlier, solid-state FeS2 batteries are reversible
because of their superior confinement of electro-active species.
The confinement of Fe0 by the solid electrolyte partially explains
© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
wileyonlinelibrary.com
121
www.advenergymat.de
FULL PAPER
www.MaterialsViews.com
Figure 2. Comparison of synthetic cubic-FeS2 cycled at 30 °C and 60 °C in conventional liquid coin cells and in solid-state cells: a) solid-state cell
at 30 °C, b) solid-state cell at 60 °C, c) liquid coin cell at 30 °C, d) liquid coin cell at 60 °C, e) capacity retention comparison of cells cycled at 30 °C,
and f) capacity retention comparison of cells cycled at 60 °C. All cells except for the 30 °C solid-state cell were cycled at a current of 144 μA which
corresponds to a rate of C/10 for charge and discharge. The 30 °C solid-state cell was cycled at rate of C/10 for the first cycle and C/20 (72 μA) for all
subsequent cycles.
the better capacity retention. Fe0 takes the form of superparamagnetic atoms or small aggregates of atoms of about 3.6 nm
in diameter.[18] In agreement with the literature, our DFT simulation of fully reduced Li4FeS2 confirms the aggregation of Fe0.
The fully-discharged, amorphous-like Li4FeS2 model (Figure 3a)
shows nanoscale separation of a Fe0 nanocluster from Li2S.
The average Fe-Fe interatomic distance (dFe-Fe) at full discharge,
x = 4, is much shorter than that of Fe in the bulk (Figure 3b).
122
wileyonlinelibrary.com
A shorter dFe-Fe indicates that Fe0 should be very catalytically
active. Indeed, nanoparticles of Fe0 have a high reactivity which
is related to the nanoparticle’s large surface area. Should Fe0
particles agglomerate into larger particles with an overall
smaller surface area, then these particles will also have a
lower reactivity.[22] It is the high reactivity of the Fe0 nanoparticles that maintains the electro-activity of Li2S. Unfortunately,
Fe0 is susceptible to continuous agglomeration upon cycling.
© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Energy Mater. 2013, 3, 120–127
www.advenergymat.de
www.MaterialsViews.com
Agglomeration of Fe0 results in the isolation of Li2S species
and the observed capacity fade when cells are discharged to low
voltages. A solid-state architecture prevents the agglomeration
of Fe0 nanoparticles.[14] It is the atomic proximity of Fe0 nanoparticles with Li2S that maintains the electro-activity of Li2S
without the excessive amount of conductive additive needed in
S/Li batteries.
A solid-state architecture is also successful at confining
polysulfides (Sn2−) formed from the electro-active species
present at full charge. At lower temperatures, it is generally
accepted that cubic-FeS2 is not regenerated by the four electron
oxidation of Fe0 and Li2S. The same is not true for molten salt
FeS2 cells which operate reversibly at temperatures in excess of
400 °C. A previous DFT study suggests it is thermodynamically
favorable to regenerate cubic-FeS2 upon full charge at room
temperature.[20] However, it may not be kinetically favorable for
such a reaction to take place at standard pressure and temperatures below 200 °C. Fong et al. suggest that non-stoichiometric
pyrrhotite, FeSy, and elemental sulfur are the products of full
charge at room temperature.[4] Another study also supports this
view with findings concluding that cubic-FeS2 is not regenerated upon charging at 55 °C.[19] While the exact nature of the
intermediates formed during the oxidation of Fe0 and Li2S are
still the subject of controversy, reaction steps have been proposed to proceed according to the following reactions:[4,7,17,23]
Fe0 +Li2 S ↔ Li2 FeS2 +2Li+ +2e−
(3)
Li2 FeS2 ↔ Li2−x FeS2 +xLi+ +xe− (0.5 < x < 0.8)
(4)
Li2−x FeS2 ↔ FeSy +(2 − y)S + (2 − x)Li+ +(2 − x)e−
(5)
Adv. Energy Mater. 2013, 3, 120–127
FULL PAPER
Figure 3. DFT simulation of lithiated LixFeS2 indicating material amorphization and Fe agglomeration for x = 4; a) Ball-and-stick representation
of LixFeS2 along a charging cycle from x = 4 to x = 0 (see Computational
Methods in the Supporting Information) and b) average Fe-Fe distance
(dFe-Fe) at each state in comparison with the Fe bulk value. Green, yellow,
and brown balls in (a) represent Li, S, and Fe atoms, respectively.
A product of oxidation in Equation (5) is elemental sulfur.
The direct reduction of sulfur by Li+ upon subsequent discharges will introduce intermediate polysulfides (Sn2−) into the
system. In a liquid cell, polysulfides dissolve into the electrolyte and participate in a parasitic “shuttle” mechanism which
causes rapid capacity fade and self-discharge. The “shuttle”
mechanism has been well documented as the primary degradation process occurring in S/Li cells[24] and is known to occur in
FeS2/Li cells as well. However, polysulfides cannot dissolve into
the solid electrolyte and the confinement provided by the solid
electrolyte inhibits the “shuttle” mechanism.
Contrary to Equation (5), we have found that orthorhombicFeS2 is produced electrochemically from discharge products
along with elemental sulfur at 30–60 °C. This conclusion is
based upon the results of DFT simulation (Figure 3), coulometric titration (Figure 4a), differential capacity (dQ/dV) analyses (Figure 4b,c) and TEM observation (Figure 5). Literature
generally agrees that the formation of FeSy and S instead of
cubic-FeS2 upon full charge explains why subsequent discharge profiles differ from the initial discharge profile. The
initial discharge of cubic-FeS2 would follow Equations (1) and
(2) while subsequent charges and discharges would follow
Equations (3), (4) and (5). Only one study that we are aware
of has used coulometric titration to support the claim that
cubic-FeS2 is not produced electrochemically. However, the
time needed for the cubic-FeS2 electrode to reach equilibrium
is much longer than the 24 hours provided in that study.[25]
When a cubic-FeS2 solid-state cell is allowed up to 144 hours to
establish equilibrium during its initial discharge at 60 °C, the
open circuit voltage (OCV) of the cell approaches the voltage
of a subsequent discharge at the appropriate reaction coordinate, x. The results of coulometric titration are compared to 1st,
2nd, and 10th discharges of the solid-state cell cycled at 60 °C
(Figure 4a). This result indicates that the difference between
the initial discharge profile and subsequent discharge profiles
can be explained by kinetics and not by an entirely different
reaction pathway.
Particle morphology and a more open regenerated crystal
structure would result in faster reaction kinetics. As Fong et al.
have already indicated, the initial reduction of pyrite is limited
both by the rate and the temperature of the reaction.[4] The slow
diffusion of Li+ into 3 μm pyrite cubes severely limits the reduction reaction kinetics. If electrochemically produced FeS2 particles are nano-crystalline, the greatly increased interfacial surface
area will facilitate a fast reaction rate despite poor Li+ diffusivity.
The diffusivity of Li+ may also be improved by regenerating a
phase other than cubic-FeS2. For example, orthorhombic-FeS2
has a more open structure than cubic-FeS2. The formation of
orthorhombic-FeS2 instead of cubic-FeS2 may result in faster
Li+ diffusion, thus further increasing the reduction reaction
kinetics.
High resolution transmission electron microscopy (HRTEM) supports these claims with direct observation of
orthorhombic-FeS2 nanoparticles upon charge. We recovered
electrode material from the solid-state cell cycled at 60 °C upon
completion of its 20th charge (Figure 2b). This cell exhibits full
utilization of FeS2 so it is unlikely that a significant mass of
electrochemically inactive synthetic cubic-FeS2 remains in
the cell by the 20th charge. Figure 5a shows a bright field (BF)
© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
wileyonlinelibrary.com
123
www.advenergymat.de
FULL PAPER
www.MaterialsViews.com
Figure 4. a) Coulometric titration results for the solid-state cell titrated
at 60 °C compared with the 1st, 2nd, and 10th discharge profiles for the
solid-state cell cycled at 60 °C (Figure 2b). b) dQ/dV of solid-state cell
cycled at 30 °C. c) Deconvolution of the dQ/dV peaks at 2.1 and 2.2V with
fitted peaks and residual.
TEM image of the 20th cycled charged FeS2 solid-state electrode. This image depicts nano-crystalline domains (darker) of
100–200 nm in diameter encased by an amorphous material
(lighter). Fast Fourier transform (FFT) analyses of HR-TEM
images matches well with orthorhombic-FeS2 along the [−110]
zone axis (Figure 5b).
124
wileyonlinelibrary.com
DFT simulation supports the HR-TEM observation of
orthorhombic-FeS2. The average Fe-Fe interatomic distance
(dFe-Fe) increases for decreasing x over the range 4 ≤ x < 2
(Figure 3b). Delithiation of the first two lithium ions corresponds to the oxidation of Fe0 to Fe2+. At x = 2, we observe the
presence of a disordered layer of lithium and also observe that
the local Fe-S4 tetrahedral motif is restored. This result is consistent with the hexagonal structure of Li2FeS2 and what has
been characterized as an intercalation chemistry for 2 ≤ x <
1.2.[4] For decreasing x over the range 2 ≤ x ≤ 0, dFe-Fe varies very
slightly (2.7 to 2.6 Å) and the Fe-S4 tetrahedral network remains
practically unchanged. We observe a stable Fe-S4 network
because the oxidation state of Fe remains unchanged. The delithiation of the second two lithium ions instead corresponds to
the oxidation of (S2)4− to (S2)2−. This two-step oxidation process
reflects the two plateaus observed in the experimental charge/
discharge voltage profiles. At x = 0, our atomic model depicts
some degree of FeS2 crystallization with a rather open structure. Perhaps more remarkably, the x = 0 model also depicts the
presence of a S2 dimer. It is the presence of elemental sulfur
that inhibits the full crystallization of FeS2 in our simulation. It
is possible that the observed FeS2 nanoclusters could crystallize
into orthorhombic-FeS2 rather than cubic-FeS2 because of the
former’s lower density.
To gain a better understanding of the amorphous regions
that we observed with HR-TEM, we examined the differential
capacity of the solid-state cell cycled at 30 °C (Figure 4b). The
green peaks in Figure 4b correspond to the oxidation of Li2S
and the reduction of S in a solid-state S/Li cell.[12] The purple
peaks correspond to reaction plateaus observed during the
1st, 2nd and 9th discharges of our solid-state FeS2 cell cycled at
30 °C. When the solid-state FeS2 cell is charged we observe no
peaks corresponding to the oxidation of Li2S. An absent Li2S
oxidation peak indicates that the Li2S component in the solid
electrolyte is not electrochemically utilized in the cell. Instead,
elemental sulfur is produced only by the disproportionation of
Li2–xFeS2 at approximately 2.4 V. However, upon discharge we
observe a peak at 2.2 V which corresponds to the direct reduction of sulfur to Li2S. The same is true for the solid-state cell
cycled at 60 °C (Supporting InformationFigure 2a,b). This
result indicates that the cycling of a FeS2 solid-state cell still
follows Equation (5) to some degree. To roughly quantify the
mass of elemental sulfur produced upon charging, we assume
all elemental sulfur is directly reduced to Li2S at 2.2V. The
solid-state cell cycled at 30 °C exhibited a discharge capacity
of 737 mAh g−1 upon its 9th discharge. If the peaks at 2.1 and
2.2 V correspond to the reaction of charge products with the
equivalent of 2 electrons, then it is expected that integrating
the dQ/dV curve between 1.6 and 2.5 V will yield a capacity of
368 mAh g−1. When these two peaks are deconvoluted by fitting
each with a Voigt profile, then the calculated total area gives a
capacity of 342.2 mAh g−1 (Figure 4c). This value matches well
with the expected capacity of 368 mAh g−1. The peak at 2.2 V
has an area of 57.14 mAh g−1 while the peak at 2.1 V has an area
of 285.79 mAh g−1. If we assume that (2-y)S is directly reduced
to Li2S, then the remaining capacity may be attributed to FeSy
using Equation (5). We calculate (2–y) to equal approximately
0.175. If subsequent discharges follow Equation (5), then the
chemical formula of FeSy is approximately FeS1.82. Yet, we have
© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Energy Mater. 2013, 3, 120–127
www.advenergymat.de
www.MaterialsViews.com
FULL PAPER
Figure 5. Electrode material from the solid-state cell cycled at 60 °C (Figure 2b) was recovered after the 20th charge for TEM analysis. a) Bright field TEM
image of the 20th cycle sample. Darker areas correspond to nano-crystalline orthorhombic-FeS2 while the lighter areas correspond to a amorphous region
composed of FeSy and elemental sulfur. b) HR-TEM of the 20th cycle sample. FFT analysis matches with orthorhombic-FeS2 along the [−110] zone axis.
confirmed the presence of nano-crystalline orthorhombic-FeS2
particles. For this reason, our charge products are likely a multiphase mixture of nano-crystalline orthorhombic-FeS2, non-stoichiometric FeSy and elemental sulfur. If we assume that FeSy
predominantly takes the form of pyrrhotite (Fe7S8), then we can
propose Equation (6) for the final oxidation step.
Li2−x FeS2 → 0.8ortho − FeS2 + 0.2FeS8/ 7 + 0.175S
+ (2 − x)Li+ + (2 − x)e−
(6)
3. Conclusions
Our assertion that charge products are nano-crystalline
orthorhombic-FeS2 particles encased in non-stoichiometric
FeSy and sulfur is consistent with prior research and our
observations. Coulometric titration indicates that the
initial discharge is kinetically limited and that subsequent discharges largely follow the same reaction path.
Nano-crystalline orthorhombic-FeS2 particles enable faster
reduction kinetics such that subsequent discharges are not
kinetically limited and can occur at a higher potential. From
dQ/dV analysis, we observe evidence of sulfur reduction at
2.2 V but not Li2S oxidation. Evidence of direct sulfur reduction supports our conclusion that the observed amorphous
region contains elemental sulfur. The amorphous region
also contains FeSy because the absence of the Li2S oxidation peak indicates that elemental sulfur is produced by
the disproportionation of Li2–xFeS2 during charging.[4] And
finally, orthorhombic-FeS2 exhibits very weak temperature
independent paramagnetism.[26] For this reason, it is likely
that 57Fe Mössbauer spectroscopy used in previous studies
was not capable of distinguishing orthorhombic-FeS2 from
other magnetic phases like FeS8/7 and unreacted cubic-FeS2.
To this end, we have demonstrated the reversible, ambient to
moderate temperature cycling of FeS2 and proposed a new
Adv. Energy Mater. 2013, 3, 120–127
charge product for a better understanding of the FeS2 discharge process.
4. Experimental Section
FeS2 was synthesized solvothermally utilizing a reaction scheme adapted
from Wang et al.[27] Polyvinylpyrrolidone (PVP, Mw,avg = 10,000, Sigma),
FeCl2∗4H2O (>99%, Sigma), ethylene glycol (99%, Mallinckrodt Baker
Inc.) and sulfur (Fischer Scientific) were used as starting materials.
HPLC grade water, analytical grade NaOH, and absolute ethanol were
used without further purification. Using a magnetic stir bar, ethylene
glycol (17 mL), PVP (600 mg), FeCl2∗4H2O (0.64 mmol, 127 mg), NaOH
(1 M, 8 mL) and sulfur (180 mg) are mixed sequentially. This solution
is stirred for 20 minutes and then dielectrically heated in a microwave
reactor (Discover SP, CEM Inc.).[28] The sample is irradiated with 75 W
of power until it reaches 190 °C and is then held at this temperature
for 12 hours. After the reaction is finished it is cooled by compressed
air. The resulting silver colored precipitate is separated by centrifugation
and washed 3 times by sonication in ethanol. It is then stored in ethanol
and vacuum dried overnight at 50 °C for battery utilization. Synthetic
FeS2 was characterized by Cu-Kα x-ray diffraction (XRD) measurement,
field emission scanning electron microscopy (FESEM, JEOL JSM-7401F),
and Raman spectroscopy (Jasco NRS-3100).
Cell fabrication and cell testing for this study was carried out under
an inert Argon gas environment. Solid electrolytes were prepared by
planetary ball milling (Across International). The solid electrolyte is
an amorphous 77.5Li2S:22.5P2S5 binary glass.[29] The glass electrolyte
is prepared by milling an appropriate ratio of Li2S (Aldrich, 99.999%,
reagent grade) and P2S5 (Aldrich, 99%) with a planetary ball mill (Across
International PQ-N2). 2g net weight of material is milled in a 500mL
stainless steel vial (Across International) with ×2 16mm diameter and
×20 10 mm diameter stainless steel balls at 500 rpm for 20 hours.
The composite positive electrode is a 10:20:2 weight ratio mixture of
synthetic FeS2, 77.5Li2S:22.5P2S5, and carbon black (Timcal Super C65)
respectively. The composite positive electrode is mixed using an agate
mortar and pestle. Stabilized lithium metal powder (SLMP) is used as the
negative electrode (FMC Lithium Corp.). The construction and testing of
solid state batteries utilizes a titanium-polyaryletheretherketone (PEEK)
test cell die.[30] 200 mg of solid electrolyte powder is pressed at 1 metric
ton in the PEEK cell die. 5 mg of composite positive electrode and the
© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
wileyonlinelibrary.com
125
www.advenergymat.de
FULL PAPER
www.MaterialsViews.com
stabilized lithium metal powder are then attached to opposite sides
of the solid electrolyte pellet by pressing at 5 metric tons. The solidstate sulfur cell used in dQ/dV analysis was fabricated using a process
adapted from Nagao et al.[12]
Liquid cells were fabricated by spreading an electrode slurry with a
6:2:2 weight ratio of synthetic FeS2, polyvinylfluorine (PVDF) binder (Alfa
Aesar) and acetylene black (Alfa-Aesar, 50% compressed) respectively.
The following electrode preparation description was adapted from Riley
et al.[31] PVDF binder was first dissolved into N-methyl-2-Pyrrolidone
(NMP, Alfa-Aesar) solvent. FeS2 and acetylene black are then stirred into
the PVDF binder. A 50 μm thick layer of slurry was spread onto aluminum
foil (ESPI Metals, 0.001” thick) and dried at 60 °C in a single wall gravity
convection oven (Blue M) for 12 hours. The electrode sheet was then
calendared with a Durston rolling mill to 75% of the total thickness.
9/16” diameter electrodes were punched and heat treated at 200 °C in
an Argon environment overnight. FeS2 electrodes were then assembled
into coin cells with a lithium foil negative electrode (Alfa-Aesar, 0.25 mm
thick) and 1 M LiPF4 electrolyte.
Cells were cycled galvanostatically using an Arbin BT2000 battery
tester at room temperature (30 °C) and elevated temperature (60 °C).
Stated C-rates are based upon FeS2’s theoretical capacity of 894 mAh g−1.
Reaction equilibrium was studied by use of coulometric titration. For
this experiment, the cell was cycled in increments of x = ½ (x in LixFeS2)
at a temperature of 60 °C and allowed not less than 6 days to reach
equilibrium for x < 2 and not less than 3 days for 2 < x < 4.
For the detailed observation of lithiated FeS2 composite electrodes,
high resolution transmission electron microscopy (TEM, JEOL 3000F)
samples are prepared using our focused ion beam (FIB, FEI NOVA200
dual beam system) machine equipped with air-lock system. The air-lock
system enables our composite electrode to remain in a vacuum state
while samples are loaded from the glove box to the FIB chamber. The
Gatan Digital Micrograph fast Fourier transform (FFT) software was
used in the analysis to determine the crystalline phases present in the
fully charged FeS2 electrode.
Supporting Information Figure 4 shows a typical sample preparation
sequence using the airlock equipped FIB. The SEM image in Supporting
Information Figure 4a is a surface of composite electrode that contains
FeS2, acetylene black and SSE. A 2 μm thick Pt protective layer was
deposited on a FeS2 particle. The FeS2 particle was sectioned on both
the front and back side by using a 30 keV Ga+ ion beam (Supporting
Information Figure 4b). A cross sectional sample of 5 μm × 5 μm ×
0.1 μm was recovered by the lift-out technique (Supporting Information
Figure 4c) and then attached to Cu TEM grid using a manipulating
probe (100.7TM, Omniprobe). Further thinning of the lift-out sample was
performed by milling parallel to the cross-sectional plane with low ion
beam currents (30–50 pA) until the final sample thickness of 50–80 nm
is achieved (Supporting Information Figure 4d).
First-principles density functional theory (DFT) calculations were
performed with the Perdew–Burke–Ernzerhof (PBE) exchange–correlation
functional[32] and all-electron-like projector augmented-wave potentials
(PAW) as implemented in VASP code.[33] A plane-wave basis set with a
cutoff energy of 450 eV and a (4 × 4 × 5) mesh for k-point integration
were used for four formula-unit supercells. To generate an atomic
model for the fully-discharged sample, we started from the 4(Li2FeS2)
hexagonal crystal structure and added two lithium atoms successively
until we obtained a low-energy, amorphous 4(Li4FeS2) model. At each
insertion step, we fully relaxed atomic forces and structural stresses. To
simulate the charging (delithiation) process, we successively removed
two randomly-selected lithium atoms. We performed a full structural
optimization at each delithiation step. The structural analysis was
performed using the atomic models constructed during the charging
(delithiation) process.
Supporting Information
Supporting Information is available from the Wiley Online Library or
from the author.
126
wileyonlinelibrary.com
Acknowledgements
Funding for this study was provided by the National Science Foundation
Graduate Research Fellowship Program (NSF-GRFP), a grant from the
Defense Advanced Research Project Agency (DARPA, FA8650-08-1-7839),
a grant from the Fundamental R&D program for Technology of World
Premier Materials through the Ministry of Knowledge Economy of Korea
(10037919), grants from the World Class University (WCU) program
through the National Research Foundation (NRF) of Korea (R31-2008000-10071-0, R31-2008-000-10075-0), and a grant from the Center for Iron
and Steel Research Institute of Advanced Materials (RIAM, D-BB04-11,
0417-20110105).
Received: April 9, 2012
Published online: August 27, 2012
[1] G. L. Henriksen, A. N. Jansen, in Handbook of Batteries, (Eds:
D. Linden, T. B. Reddy), McGraw-Hill, New York 2002.
[2] Y. Shao-Horn, S. Osmialowski, Q. C. Horn, J. Electrochem. Soc.
2002, 149, A1499.
[3] J. Cabana, L. Monconduit, D. Larcher, M. R. Palacín, Adv. Mater.
2010, 22, E170; L. A. Montoro, J. M. Rosolen, Solid State Ionics 2003,
159, 233; J. W. Choi, G. Cheruvally, H. J. Ahn, K. W. Kim, J. H. Ahn, J.
Power Sources 2006, 163, 158.
[4] R. Fong, J. R. Dahn, C. H. W. Jones, J. Electrochem. Soc. 1989, 136,
3206.
[5] E. Strauss, D. Golodnitsky, K. Freedman, A. Milner, E. Peled, J. Power
Sources 2003, 115, 323; E. Peled, D. Golodnitsky, E. Strauss, J. Lang,
Y. Lavi, Electrochim. Acta 1998, 43, 1593; G. Ardel, D. Golodnitsky,
K. Freedman, E. Peled, G. B. Appetecchi, P. Romagnoli, B. Scrosati,
J. Power Sources 2002, 110, 152.
[6] J. Barker, E. Kendrick, J. Power Sources 2011, 196, 6960.
[7] D. Golodnitsky, E. Peled, Electrochim. Acta 1999, 45, 335.
[8] X. Ji, K. T. Lee, L. F. Nazar, Nat. Mater. 2009, 8, 500.
[9] J. Hassoun, B. Scrosati, Adv. Mater. 22, 5198.
[10] J. L. Wang, J. Yang, J. Y. Xie, N. X. Xu, Adv. Mater. 2002, 14, 963.
[11] E. Peled, D. Golodnitsky, G. Ardel, J. Lang, Y. Lavi, J. Power Sources
1995, 54, 496.
[12] M. Nagao, A. Hayashi, M. Tatsumisago, Electrochim. Acta 2011, 56,
6055.
[13] K. Takada, K. Iwamoto, S. Kondo, Solid State Ionics 1999, 117, 273.
[14] K. Takada, Y. Kitami, T. Inada, A. Kajiyama, M. Kouguchi, S. Kondo,
M. Watanabe, M. Tabuchi, J. Electrochem. Soc. 2001, 148,
A1085.
[15] N. Kamaya, K. Homma, Y. Yamakawa, M. Hirayama, R. Kanno,
M. Yonemura, T. Kamiyama, Y. Kato, S. Hama, K. Kawamoto,
A. Mitsui, Nat. Mater. 2011, 10, 682.
[16] F. Mizuno, A. Hayashi, K. Tadanaga, M. Tatsumisago, Adv. Mater.
2005, 17, 918.
[17] C. H. W. Jones, P. E. Kovacs, R. D. Sharma, R. S. McMillan, J. Phys.
Chem. 1991, 95, 774.
[18] C. H. W. Jones, P. E. Kovacs, R. D. Sharma, R. S. McMillan, J. Phys.
Chem. 1990, 94, 832.
[19] D. A. Totir, I. T. Bae, Y. Hu, M. R. Antonio, M. A. Stan, D. A. Scherson,
J. Phys. Chem. B 1997, 101, 9751.
[20] Y. Yamaguchi, T. Takeuchi, H. Sakaebe, H. Kageyama, H. Senoh,
T. Sakai, K. Tatsumi, J. Electrochem. Soc. 2010, 157, A630.
[21] E. Strauss, S. Calvin, H. Mehta, D. Golodnitsky, S. G. Greenbaum,
M. L. den Boer, V. Dusheiko, E. Peled, Solid State Ionics 2003, 164,
51.
[22] P. Poizot, S. Laruelle, S. Grugeon, L. Dupont, J. M. Tarascon, Nature
2000, 407, 496.
[23] R. Brec, A. Dugast, A. Le Mehaute, Mater. Res. Bull. 1980, 15, 619.
© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Energy Mater. 2013, 3, 120–127
www.advenergymat.de
www.MaterialsViews.com
Adv. Energy Mater. 2013, 3, 120–127
FULL PAPER
[24] X. Ji, L. F. Nazar, J. Mater. Chem. 2010, 20, 9821.
[25] N. G. Bannik, L. I. Neduzhko, I. M. Maksiuta, A. V. Markevich,
E. M. Shembel, Russ. J. Electrochem. 2009, 45, 1245.
[26] R. Garg, V. K. Garg, Y. Nakamura, Hyperfine Interact. 1991, 67, 447.
[27] D.-W. Wang, Q.-H. Wang, T.-M. Wang, Cryst. Eng. Comm. 2010, 12,
755.
[28] I. Bilecka, M. Niederberger, Nanoscale 2010, 2, 1358.
[29] J. Trevey, J. S. Jang, Y. S. Jung, C. R. Stoldt, S.-H. Lee, Electrochem.
Commun. 2009, 11, 1830.
[30] T. A. Yersak, J. Trevey, E. S.-H. Lee, J. Power Sources 2011, 196, 9830.
[31] L. A. Riley, S. H. Lee, L. Gedvilias, A. C. Dillon, J. Power Sources 2010,
195, 588.
[32] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77, 3865.
[33] G. Kresse, D. Joubert, Phys. Rev. B 1999, 59, 1758.
© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
wileyonlinelibrary.com
127