Journal of The Electrochemical Society, 161 (5) A663-A667 (2014) A6630013-4651/2014/161(5)/A663/5/$31.00 © The Electrochemical Society

Derivation of an Iron Pyrite All-Solid-State Composite Electrodewith Ferrophosphorus, Sulfur, and Lithium Sulfide as PrecursorsThomas A. Yersak,a,∗ Tyler Evans,b Justin M. Whiteley,b Seoung-Bum Son,cBrian Francisco,b Kyu Hwan Oh,c and Se-Hee Leeb,∗,z

aDepartment of Nano Engineering, University of California San Diego, La Jolla, California 92093-0448, USAbDepartment of Mechanical Engineering, University of Colorado at Boulder, Boulder, Colorado 80309-0427, USAcDepartment of Materials Science and Engineering, Seoul National University, Seoul, 151-742, Korea

In order to address the poor energy density of bulk all-solid-state lithium-ion batteries a new composite electrode preparation methodis proposed. This study demonstrates the concurrent synthesis of the FeS2 active material and a glass electrolyte which is similarto the (100-x)P2S5:xLi2S system of glass electrolytes. This is accomplished through the successive mechanochemical and thermaltreatments of Fe2P, S, and Li2S. The composite electrode’s ionic conductivity of 5.46 × 10−5 S cm−1 at 60◦C and specific energyof 1.2 Wh g−1 versus a lithium metal anode suggest that this method may provide a more intimate solid-solid interfacial contactbetween active material and glass electrolyte. The results of this study provide a completely new perspective on the preparation ofall-solid-state composite electrodes.© 2014 The Electrochemical Society. [DOI: 10.1149/2.003405jes] All rights reserved.

Manuscript submitted December 10, 2013; revised manuscript received February 6, 2014. Published March 7, 2014.

Bulk all-solid-state composite electrode engineering convention-ally approaches the problem of interfacial structuring from the per-spective that the active material and the solid-state electrolyte musteach be prepared separately and then combined via mechanical mix-ing. Wetting the porous electrode with a liquid electrolyte createsintimate interfacial contact. However, the power and energy densityof bulk all-solid-state batteries suffer because a solid-state electrolytecannot interface as intimately with the active material as a liquid elec-trolyte. For example, an all-solid-state electrode utilizing FeS2 andthe P2S5-Li2S glass electrolyte had an active mass loading of only32% in order to achieve FeS2’s theoretical capacity of 894 mAh g−1.1

Similarly structured cells utilizing S or Li2S as the active material suf-fer from the same limitations.2,3 While the bulk all-solid-state batteryarchitecture has demonstrated its value by enabling the reversible cy-cling of high capacity conversion active materials, further interfacialengineering should address how to achieve practical electrode levelenergy density.

Along these lines, hot isostatic pressing (HIP) was proposed asa method to densify composite electrodes with the P2S5-Li2S glass-ceramic for improved interfacial contact.4 HIP’ing is not appropriatefor active materials which may be reactive or decompose at hightemperatures. Pulsed laser deposition of P2S5-Li2S glass electrolytecoatings onto active material particles has also been suggested as away to improve interfacial contact and increase active material massloading.5,6 Another method involves mechanically preparing sulfurcomposite electrodes at a temperature above sulfur’s melting pointso that sulfur can flow about P2S5-Li2S glass-ceramic electrolyteparticles.7 An electrode prepared in such a manner was reported todeliver a specific energy in excess of 1000 Wh kg−1. Spark plasmasintering8,9 and impregnation of active material sols into macrop-orous 3D electrolyte membranes10–14 have also been suggested forall-solid-state batteries utilizing ceramic electrolytes other than theP2S5-Li2S system of glass-ceramics. Further, the development of poly-sulfidophosphates has blurred the distinction between active materialand solid-state electrolyte.15

In order to structure an intimate solid-solid interface this studyproposes that the active material and glass electrolyte former can besynthesized concurrently as products of a single solid-state chemi-cal reaction. Interfacial contact may then be dictated not only by themechanical properties of the composite electrode components, butalso by the process of phase nucleation and growth. In the study out-lined here the concurrently synthesized active material and electrolyteformer are suggested to be FeS2 and P2S5, respectively.

∗Electrochemical Society Active Member.zE-mail: sehee.lee@colorado.edu

Historically, the reaction of iron pyrite (FeS2) with ferrophos-phorus (Fe2P) at temperatures between 700 – 800◦C was thought toproduce P2S5 and FeS.16 In this two step process, FeS2 first thermallydecomposes into FeS and S.17 Sulfur then reacts with Fe2P to formP2S5 and more FeS. P2S5 can also be synthesized by the direct re-action of S with Fe2P.18 Equation 1 proposes FeS2 and P2S5 as thebyproducts for the reaction of S and Fe2P at a temperature of 350◦C.The enthalpy of this reaction at standard state is −762.6 kJ mol−119 or−674 kJ mol−120 depending upon the source of the P2S5 data so it isexpected that this reaction is thermodynamically favorable. The P2S5

component of this proposed composite can then be mechanochemi-cally modified with Li2S to synthesize the 22.5P2S5:77.5Li2S glasselectrolyte which will be intimately in contact with the FeS2 phase.Equation 2 outlines this reaction assuming that Equation 1’s reactionis complete with no other byproducts. This method should be con-trasted with our previous FeS2 composite electrode preparation tech-nique where FeS2 and carbon black are mixed with a pre-prepared22.5P2S5:77.5Li2S glass electrolyte.1

2Fe2P + 13S → 4FeS2 + P2S5(350◦C) [1]

22.5(4FeS2 + P2S5) + 77.5Li2S → 90FeS2 + 22.5P2S5 : 77.5Li2S[2]

Methods

All procedures to be outlined were conducted in a dry argon envi-ronment. The electrode composite was prepared via a three step pro-cess. First, 2 g total (2:13 molar ratio) of Fe2P (Aldrich, 99.5%) andSulfur (Aldrich, 99.98%) were combined by planetary ball milling.Material was milled in an airtight 500 mL stainless steel jar (AcrossInternational) with two 16 mm diameter and twenty 10 mm diameterstainless steel balls (Across International) for 20 hours at 500 rpm.Second, the composite was heat treated in an evacuated flame sealedborosilicate glass ampoule at 350◦C for 3 hours with controlled tem-perature ramping and cooling both taking 1 hour. Third, planetary ballmilling was used to combine 0.5 g total (22.5:77.5 molar ratio) of theheat treated composite and Li2S (Aldrich, 99.999%, reagent grade).Material was milled in an airtight 100 mL agate jar (Across Interna-tional) with fifty 6 mm diameter and five 10 mm diameter agate ballsfor 20 hours at 500 rpm. The materials resulting from the comple-tion of the first, second, and third step are referred to as (BM), (BM+ HT), and (BM + HT + BM) in the text and figures. BM is shorthandfor ball mill while HT is shorthand for heat-treatment.

To assemble test cells, a 200 mg glass electrolyte pellet separator(ϕ = 13 mm) was first pressed at 1 metric ton inside a polyetherether-ketone (PEEK) lined Ti test cell die.21 The electrolyte used for theseparator was the previously described mechanochemically prepared

) unless CC License in place (see abstract).  ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 147.46.237.212Downloaded on 2015-01-13 to IP

A664 Journal of The Electrochemical Society, 161 (5) A663-A667 (2014)

Figure 1. Powder XRD patterns for the (BM),(BM + HT) and (BM + HT + BM) samples.Peaks are indexed to Fe2P (open square), S (tri-angle), FeP (diamond), and cubic-FeS2 (star).

22.5P2S5:77.5Li2S glass.22,23 For increased electronic conductivity,10 wt% of carbon black (TimCal, C65) was added to the compositeelectrode using an agate mortar and pestle. 5 mg of the compositeelectrode was then pressed to one side of the glass electrolyte pelletat 5 metric tons and an LiIn alloy (FMC Lithium Corp., Lectro MaxPowder 100 and Indium powder, Alfa Aesar, Puratronic 99.999%)was attached to the opposite side of the pellet at 1 metric ton. The LiInalloy has a potential of 0.62 V vs. Li+/Li for a limited compositionalrange24 so voltage profile figures are given with respect to both LiInand Li metal. Calculations of specific energy assume Li metal as theanode and all specific capacities and current densities are given withrespect to the total composite electrode mass including carbon black.The test cells were cycled at 60◦C under constant current constantvoltage (CCCV) cycling parameters.

The composite electrode was characterized by Cu-Kα powderX-ray diffraction, FIB-SEM EDS analysis, cyclic voltammetry, andRaman spectroscopy. The FIB sample was prepared as previouslydescribed.1 Cyclic voltammetry was used to measure the electronicconductivity of the composite electrode by sweeping symmetric cellsas presented in Figure 4a. To measure ionic conductivity, a compositeelectrode pellet without a carbon black additive (150 mg) was sand-wiched between two layers of the 22.5P2S5:77.5Li2S glass electrolyte(150 mg each) in order to block electron flow as presented in Figure 4b.A similar cell configuration was used previously, but here junction po-tentials and resistance to charge transfer are considered negligible atlayer interfaces.25 The contributions from the electron-blocking glasselectrolyte layers can be subtracted out by considering that 1/σelectrode

= Rtotal – 2/σelectrolyte. The ionic conductivity of the 22.5P2S5:77.5Li2Sglass electrolyte used in the experiment was measured to have an ionicconductivity of 2.91 × 10−4 S cm−1 at 25◦C and 1.297 × 10−3 S cm−1

at 60◦C. DC polarization measurements at 10 mV were used to calcu-late the ionic conductivity of the electrodes. The calculations use thestabilized current values because, after a period of transient decay, thestabilized current is thought to be totally ionic. Raman spectra wererecorded using a JASCO NRS-3100 system equipped with a 532 nmlaser at a power level of 0.1 mW focused through a 20× objectivelens. Power levels higher than this were observed to cause sampledegradation. Raman shift was calibrated using a silicon standard andaccuracy was estimated to be ±7 cm−1. As the dark color of the ma-terial led to weak Raman scattering, 15 accumulations of 45 secondseach were typically averaged to increase the signal-to-noise ratio.

Results

Powder XRD was used to identify the phases present over thecourse of the FeS2 composite electrode preparation. Figure 1 pro-vides the XRD patterns for samples (BM), (BM + HT), and (BM +HT + BM). From the pattern for sample (BM), it was determinedthat Fe2P does not react with S during mechanical milling. Afterheat treating sample (BM) at 350◦C a solid-state reaction was ob-

served to occur. The pattern for sample (BM + HT) has new peakswhich can be assigned to iron pyrite (FeS2) and orthorhombic FeP26,27

while peaks for Fe2P disappear. Three weak peaks corresponding tosulfur’s three strongest diffractions are also observed. Residual sul-fur is consistent with the observation of sulfur crystallization on theinside surface of the evacuated glass ampoules after heat-treatment(Supporting Figure 1). After the addition of Li2S, no new phases areobserved in the pattern for sample (BM + HT + BM). Depending onthe composition residual crystalline Li2S is frequently observed whenthe (100-x)P2S5:xLi2S glass is mechanochemically prepared,21,22 butno crystalline Li2S peaks are observed here. This may be due to theamorphization of Li2S or a reaction between Li2S and a glass former.Diffraction peaks for FeS2 and FeP are found to broaden and re-duce in intensity after the second ball milling step which is consistentwith the reduction of average crystallite size by mechanical grinding.Diffraction peaks for crystalline P2S5 are not observed in any of thesamples.

FIB-SEM and EDS mapping of sample (BM + HT) provided inFigure 2 do not support the case for the formation of P2S5 or anotherglass former. A SEM image of a cross-sectioned sample is providedin Figure 3a. Close inspection reveals particles embedded in a porousmaterial. EDS mapping for P, S, and Fe are provided in Figure 2b, 2c,and 2d. It is noted that EDS cannot distinguish P in the sample fromthe Pt supporting structure. Dispersion of Fe throughout the entiresample is consistent with the formation of FeS2 and FeP. If P2S5 were

Figure 2. a) FIB-SEM image of a crossed sectioned (BM + HT) pressedpellet. Close inspection reveals particles embedded in a matrix. b-d) EDSmapping for P, S, and Fe. P and S are segregated while Fe is well dispersed.

) unless CC License in place (see abstract).  ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 147.46.237.212Downloaded on 2015-01-13 to IP

Journal of The Electrochemical Society, 161 (5) A663-A667 (2014) A665

Figure 3. Raman spectroscopy of the (BM + HT + BM) sample compared tocubic-FeS2, Li2S, and the 22.5P2S5:77.5Li2S glass electrolyte. Peaks at 422cm−1, 378 cm−1, and 340 cm−1 are attributed to cubic-FeS2 while the peakat 480 cm−1 is attributed to residual S. The PS4

3− ion, indicative of a glassyionic conductor, has been reported at 420 cm−1. The 422 cm−1 peak attributedto FeS2 overlaps is superimposed with the PS4

3− peak.

synthesized it would be expected that signals for P and S would be wellmixed, however, the opposite is true. Area mapped to P is segregatedfrom the area mapped to S. The particles observed in SEM have ahigh concentration of P signals so these particles are attributed to theFeP phase.

Raman spectroscopy provided in Figure 3 was used to character-ize the ionically conductive component of the (BM + HT + BM)electrode. The Raman spectrum for the (BM + HT + BM) electrodeis compared to that for FeS2, Li2S, and the 22.5P2S5:77.5Li2S glasselectrolyte. Four main peaks are observed at 480 cm−1, 422 cm−1,378 cm−1, and 340 cm−1. The peak at 480 cm−1 is attributed to unre-acted S in the composite28,29 while the peaks at 422 cm−1, 378 cm−1,and 340 cm−1 coincide with the peaks for FeS2. A peak at 420 cm−1

is typically assigned to the thiophosphate ion, PS43−, of the (100-

x)P2S5:xLi2S system of glass-ceramic electrolytes.30 Observation of

the PS43− ion would confirm that the composite electrode contains

the local structure of the glass electrolyte and indirectly suggests theformation of P2S5 or another glass former. Qualitatively, the ratio ofthe 422 cm−1 to 378 cm−1 peak intensities is much smaller for FeS2

than it is for the electrode. The increase in the ratio of these peaks isattributed to the presence of PS4

3−. Unfortunately, the (BM + HT+ BM) electrode powder exhibited poor Raman scattering so a mean-ingful peak deconvolution could not be performed. To substantiate ourclaim for the formation of a glassy electrolyte, the ionic transport of the(BM + HT + BM) sample was measured at 25◦C and 60◦C. A previ-ous study used cyclic voltammetry to measure the ionic conductivityof mixed conductors using the same symmetric cell configuration pre-sented in Figure 4b.31 However, for the (BM + HT + BM) samplewe observed a significant amount of non-Ohmic current using thecyclic voltammetry method. For this reason, the ionic conductivitywas measured by performing 10 mV DC polarization measurementson symmetric multi-layer test cells (Figure 4d). In the DC polarizationmethod, the ionic conductivity is calculated by measuring the stabi-lized current after transient decay. It is noted that the transient currentof the (BM + HT + BM) sample tested at 60◦C took far longer to de-cay than it did in other tests. The electronic conductivity was measuredby the cyclic voltammetry method. As a control, a conventional com-posite electrode was prepared by mixing nano-sized natural pyrite and22.5P2S5:77.5Li2S in a weight ratio of 56:44. This weight ratio cor-responds to the theoretical weight ratio of the products in Equation 2.The natural FeS2 is the same as that used in a recent study by Sonet al.32

The electronic conductivity of a conventional composite electrodewithout carbon black additive is only 1.81 × 10−5 S cm−1 while theelectronic conductivity of the (BM + HT + BM) composite electrodewithout carbon black is 4.94 × 10−4 S cm−1 (Figure 4c). The goodelectronic conductivity of the (BM + HT + BM) electrode is attributedto the presence of the unexpected FeP phase.33 On the other hand, theionic conductivity of the conventional electrode is superior to that ofthe (BM + HT + BM) electrode. At 60◦C, the conventional electrodehas an ionic conductivity of 2.82 × 10−4 S cm−1 while the (BM +HT + BM) electrode has an ionic conductivity of approximately 5.46× 10−5 S cm−1 (Figure 4d) at 60◦C.

The results of constant current constant voltage (CCCV) cycling at60◦C are provided in Figure 5. All specific capacities are normalizedto the total mass of the composite electrode including carbon black.In anticipation of excess Li2S utilization, the (BM + HT + BM)electrode was initially overcharged to provide an extra 114 mAh g−1

Figure 4. a) Schematic of the symmetric test cell used tomeasure the electronic conductivity of the (BM + HT +BM) sample. b) Schematic of the symmetric test cell usedto measure the ionic conductivity of the (BM + HT + BM)sample. c) Cyclic voltammetry i-V curves at 25◦C and cal-culated electronic conductivity for the (BM + HT + BM)sample (solid blue) and a conventional control electrode witha composition dictated by Equation 2 (dashed blue). d) 10mV DC polarization i curves for the same two electrodes at25◦C (blue) and 60◦C (red). The stabilized current was usedto calculate Ohmic ionic resistance in order to estimate ionicconductivity.

) unless CC License in place (see abstract).  ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 147.46.237.212Downloaded on 2015-01-13 to IP

A666 Journal of The Electrochemical Society, 161 (5) A663-A667 (2014)

of capacity. An overcharge capacity of 114 mAh g−1 indicates that59% of the Li2S added to the (BM + HT + BM) sample duringthe final ball milling step was electrochemically oxidized. We havepreviously reported on the electrochemical utilization of excess Li2S incomposite electrodes with the 22.5P2S5:77.5Li2S glass electrolyte.21

In the prior study activation of excess capacity was correlated with theconsumption of crystalline Li2S, however, possible contributions toexcess capacity from non-crystalline Li2S was not ascertained. Eventhough the XRD spectra for (BM + HT + BM) does not indicatethe presence of crystalline Li2S it is reasonable that Li2S may still beavailable for electrochemical utilization.

Assuming complete conversion of Equation 1, the (BM + HT+ BM) electrode composite should theoretically only deliver 500mAh g−1 through the complete reduction of FeS2. On the other hand,(BM + HT + BM) could also theoretically deliver 1024 mAh g−1 as-suming that no solid-state reactions occurred during electrode prepa-ration and that the S and Li2S precursors were fully utilized. On itsinitial discharge (cycle 2), the electrode delivers a specific discharge of721 mAh g−1. This discharge capacity, which is in between the twoextreme values of 500 mAh g−1 and 1024 mAh g−1, provides furtherevidence for the complex composition of the (BM + HT + BM) com-posite electrode. The sweeping voltage profile between 2.4 – 1.9 V isattributed to the electrochemical reduction of residual S and of S cre-ated by the oxidation of excess of Li2S during the initial overcharge.A small fraction of capacity may also be attributed to utilization ofLi2S in the glass electrolyte separator pellet.21 As expected, subse-quent discharge profiles are much different than the initial dischargewhich indicates either a change in chemistry or kinetics.1,21,34 Quiteimpressively, the (BM + HT + BM) electrode delivers a specificenergy density of 1,200 Wh kg−1 (vs. Li+/Li). On its 40th cycle the(BM + HT + BM) electrode delivers a specific energy of 1,130Wh kg−1 (vs. Li+/Li), which attests to a good stability (SupportingFigure 2). The lower discharge profiles, which are associated withthe redox of Fe2+ → Fe0, are observed to fade faster than the stableplateau at approximately 2.3 V which is attributed to the redox of S2

2−

→ 2S2−. This result is consistent with prior results obtained when cy-

Figure 5. 1st 10 cycles for a (BM + HT + BM) electrode with 10 wt% carbonblack. The cell is initially overcharged in order to activate 114 mAh g−1 ofadditional residual sulfur capacity.

cling a conventionally prepared composite electrode with FeS and Sas the active material and a 22.5P2S5:77.5Li2S glass as the solid-stateelectrolyte.21

A rate test was initiated after ten cycles at a low discharge currentdensity of 30 mA g−1. Three recovery cycles were applied for everytwo cycles of high discharge rate. The charge current density was keptconstant at 30 mA g−1. The average capacities of several different testcells are provided in Figure 6a with an estimate of standard deviation.The test cell provided in Figure 5 and Supplementary Figure 2 isincluded in this average. It is observed that the capacity of the (BM

Figure 6. a) Rate test of the (BM + HT + BM) elec-trode. Given capacities are an average with an estimatedstandard deviation. b) Voltage profiles for the cell pre-sented in Figure 5 at 30, 40, 80, 200, 400, and 800 mAg−1, or equivalently, 0.11, 0.15, 0.30, 0.75, 1.51, and3.01 mA cm−2. The initial discharge profile is given asa reference. c) Ragone plot summarizing Figure 6b).

) unless CC License in place (see abstract).  ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 147.46.237.212Downloaded on 2015-01-13 to IP

Journal of The Electrochemical Society, 161 (5) A663-A667 (2014) A667

+ HT + BM) electrode is not stable at higher rates. However, thecapacity recovers when the current density is reduced. Figures 6band 6c summarize the rate test for the cell presented in Figure 5.As the current density is increased, the lower voltage plateau, whichwe attribute to the reduction of Fe2+, is observed to disappear. At amoderate current density of 400 mA g−1, the Ragone plot (Figure 6c)indicates that the electrode delivers a specific power of 706 W kg−1.We suspect that the lower voltage plateau is associated with eitherthe reduction of FeS or an intermediary Li2-xFeS2 phase to Li2S andFe0.1,34–36 The lower plateau disappears because the kinetics of FeSor Li2-xFeS2 reduction may be limited by the solid state diffusion ofFe0 atoms needed to nucleate separate phases of Li2S and Fe0.

Discussion

The characterization of the (BM + HT) material yields data thatdoes not support the formation of the glass former, P2S5. As men-tioned, powder XRD peaks for crystalline P2S5 are absent and ele-mental mapping indicates that signals for P and S are segregated. Theobservation of FeP attests to the inadequacy of the proposed reactionmechanism. Nevertheless, the ionic conductivity and Raman spec-troscopy of the (BM + HT + BM) electrode indicate the presence ofan ionically conductive glassy electrolyte. Indeed, the ionic conduc-tivity of the (BM + HT + BM) electrode was nearly double that ofthe convention control electrode with prepared glass electrolyte.

Mechanochemical milling of (BM + HT) with Li2S will yield amaterial with a heavily altered structure that may include a glassyelectrolyte. Powder XRD results indicated that the initial ball millingstep did not induce a solid-state reaction, but the same may not betrue for the second ball milling step. The reactivities of FeP, S andother possible unidentified reaction products under mechanochemicalaction are not known. Preparation of sulfide glass electrolytes withoutP2S5 as a precursor is not without precedent. The mechanochemicalreaction of P, S and Li2S has produced functional ionically conductingglasses.28,29 P2S3 has also been used as a glass former.37 To confirmthis suspicion EDS elemental mapping of the (BM + HT + BM)sample should be performed to confirm the mixing of P and S signals.Unfortunately, this characterization was not completed due to a lackof resources.

Conclusions

We have demonstrated an electrode which delivers a specific en-ergy of 1,200 Wh kg−1 which ranks it among the most energy denseall-solid-state electrodes to be reported. Even though our proposedreaction mechanism is incomplete it does provide a good startingframework for understanding the coincident preparation of FeS2 andglass electrolyte precursors. While the composition of the compositeelectrode is not yet well understood, its high performance suggeststhat much promise lies ahead for structuring interfaces via solid-statereactions.

Acknowledgments

Funding for this study was provided by the National Science Foun-dation (NSF, CHE-1231048) and by the University of Colorado atBoulder Dean’s Graduate Student Research grant. Special thanks aregiven to Daniela Molina Piper and Dr. Chunmei Ban for discussionand assistance with preliminary XRD data collection.

References

1. T. A. Yersak, H. A. Macpherson, S. C. Kim, V. D. Le, C. S. Kang, S. B. Son,Y. H. Kim, J. E. Trevey, K. H. Oh, C. Stoldt, and S. H. Lee, Advanced Energy Mate-rials, 3(1), 120 (2013).

2. M. Nagao, A. Hayashi, and M. Tatsumisago, Electrochimica Acta, 56(17), 6055(2011).

3. M. Nagao, A. Hayashi, and M. Tatsumisago, Journal of Materials Chemistry, 22(19),10015 (2012).

4. H. Kitaura, A. Hayashi, T. Ohtomo, S. Hama, and M. Tatsumisago, J. Mater. Chem.,21(1), 118 (2010).

5. Y. Sakurai, A. Sakuda, A. Hayashi, and M. Tatsumisago, Solid State Ionics, 182(1),59 (2011).

6. K. Aso, A. Sakuda, A. Hayashi, and M. Tatsumisago, ACS Applied Materials andInterfaces, 5(3), 686 (2013).

7. M. Nagao, A. Hayashi, and M. Tatsumisago, Energy Technology, 1(2-3), 186 (2013).8. G. Delaizir, V. Viallet, A. Aboulaich, R. Bouchet, L. Tortet, V. Seznec, M. Morcrette,

J. M. Tarascon, P. Rozier, and M. Dolle, Advanced Functional Materials, 22(10),2140 (2012).

9. A. Aboulaich, R. Bouchet, G. Delaizir, V. Seznec, L. Tortet, M. Morcrette, P. Rozier,J. M. Tarascon, V. Viallet, and M. Dolle, Advanced Energy Materials, 1(2), 179(2011).

10. M. Hara, H. Nakano, K. Dokko, S. Okuda, A. Kaeriyama, and K. Kanamura, Journalof Power Sources, 189(1), 485 (2009).

11. K. Kanamura, N. Akutagawa, and K. Dokko, Journal of Power Sources, 146(1–2),86 (2005).

12. M. Kotobuki, Y. Isshiki, H. Munakata, and K. Kanamura, Electrochimica acta, 55(22),6892 (2010).

13. M. Kotobuki, Y. Suzuki, H. Munakata, K. Kanamura, Y. Sato, K. Yamamoto, andT. Yoshida, Journal of The Electrochemical Society, 157(4), A493 (2010).

14. M. Kotobuki, Y. Suzuki, K. Kanamura, Y. Sato, K. Yamamoto, and T. Yoshida,Journal of Power Sources, 196(22), 9815 (2011).

15. Z. Lin, Z. Liu, W. Fu, N. J. Dudney, and C. Liang, Angewandte Chemie InternationalEdition (2013).

16. P. Dutoit, U.S. Patent # 1,852,959, (1932).17. I. C. Hoare, H. J. Hurst, W. I. Stuart, and T. J. White, Journal of the Chemical Soci-

ety-Faraday Transactions I, 84, 3071 (1988).18. T. Ozturk, E. Ertas, and O. Mert, Chemical reviews, 110(6), 3419 (2010).19. O. Knacke, O. Kubaschewski, O. Hesselmann, and O. Editors, Thermochemical

Properties of Inorganic Substances, Springer-Verlag (1991).20. I. Barin, Thermochemical data of pure substances, VCH, New York (1993).21. T. A. Yersak, C. Stoldt, and S.-H. Lee, Journal of The Electrochemical Society,

160(8), A1009 (2013).22. A. Hayashi, S. Hama, H. Morimoto, M. Tatsumisago, and T. Minami, Journal of the

American Ceramic Society, 84(2), 477 (2001).23. J. Trevey, J. S. Jang, Y. S. Jung, C. R. Stoldt, and S.-H. Lee, Electrochemistry Com-

munications, 11(9), 1830 (2009).24. K. Takada, N. Aotani, K. Iwamoto, and S. Kondo, Solid State Ionics, 86-88(Part 2),

877 (1996).25. D. J. Vischjager, A. A. Van Zomeren, J. Schoonman, I. Kontoulis, and B. C. H. Steele,

Solid state ionics, 40, 810 (1990).26. G. Yunle, G. Fan, Q. Yitai, Z. Huagui, and Y. Ziping, Materials research bulletin,

37(6), 1101 (2002).27. S. Boyanov, J. Bernardi, F. Gillot, L. Dupont, M. Womes, J. M. Tarascon,

L. Monconduit, and M. L. Doublet, Chemistry of materials, 18(15), 3531 (2006).28. T. Ohtomo, F. Mizuno, A. Hayashi, K. Tadanaga, and M. Tatsumisago, Solid state

ionics, 176(31), 2349 (2005).29. F. Mizuno, T. Ohtomo, A. Hayashi, K. Tadanaga, and M. Tatsumisago, Solid State

Ionics, 177(26–32), 2753 (2006).30. M. Tachez, J.-P. Malugani, R. Mercier, and G. Robert, Solid State Ionics, 14(3), 181

(1984).31. S.-B. Son, S. C. Kim, C. S. Kang, T. A. Yersak, Y.-C. Kim, C.-G. Lee, S.-H. Moon,

J. S. Cho, J.-T. Moon, K. H. Oh, and S.-H. Lee, Advanced Energy Materials, 2(10),1226 (2012).

32. S.-B. Son, T. A. Yersak, D. M. Piper, S. C. Kim, C. S. Kang, J. S. Cho, S.-S. Suh,Y.-U. Kim, K. H. Oh, and S.-H. Lee, Advanced Energy Materials (in press).

33. G. L. Evarts and G. T. Miller, U.S. Patent # 4,828,614 A, (1989).34. R. Fong, J. R. Dahn, and C. H. W. Jones, Journal of The Electrochemical Society,

136(11), 3206 (1989).35. B.-C. Kim, K. Takada, N. Ohta, Y. Seino, L. Zhang, H. Wada, and T. Sasaki, Solid

State Ionics, 176(31–34), 2383 (2005).36. K. Takada, Y. Kitami, T. Inada, A. Kajiyama, M. Kouguchi, S. Kondo, M. Watanabe,

and M. Tabuchi, Journal of The Electrochemical Society, 148(10), A1085 (2001).37. N. Machida, H. Yamamoto, and T. Shigematsu, Chemistry Letters, 33(1), 30 (2004).

) unless CC License in place (see abstract).  ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 147.46.237.212Downloaded on 2015-01-13 to IP