Operando Observation of Lithiation and Delithiation Reactions of a LiCoO₂-Li₃BO₃ Composite Electrode Formed on a Li_6.6La₃Zr_1.6Ta_0.4O₁₂ Solid Electrolyte Sheet by Laboratory-based Hard X-ray Photoelectron Spectroscopy

Abstract

A positive electrode composed of LiCoO₂ (LCO) and Li₃BO₃ (LBO) was formed on one side of a Li_6.6La₃Zr_1.6Ta_0.4O₁₂ (LLZT) solid electrolyte sheet by applying LCO fine powder to LLZT sheet precoated with a Nb thin layer, placing a droplet of aqueous solution of LiOH and H₃BO₃, and being annealed at ∼700 °C in an oxygen atmosphere. After binding a negative electrode, a Li metal foil, on the other side of LLZT sheet precoated with a Li thin layer, electrochemical reaction at the LiCoO₂-Li₃BO₃ composite positive electrode was observed in an all-solid-state battery configuration, i.e., a LCO-LBO/Nb/LLZT/Li cell, by a newly-developed laboratory-based hard x-ray photoelectron spectroscopy (HAXPES) apparatus equipped with a Cr-Kα source (5414.9 eV) and bias application system. A sharp main peak and a broad satellite peak characteristic to LCO were observed in the Co 2p_3/2 region at the pristine state. During charging, i.e., delithiation from LCO, the main peak was asymmetrically broadened to a higher binding energy due to the partial oxidation of Co³⁺ ions at 780 eV to Co⁴⁺ ions at ∼781 eV. In addition, the full width half maximum (FWHM) of the Co⁴⁺ peak increased with increasing the amount of lithium insertion, while that of the Co³⁺ peak remained unchanged. The decrease of satellite peak further confirms the oxidation of Co³⁺ ions. During the subsequent discharging, i.e., lithiation of LCO, those recovered to the original states, confirming the reversible reduction of Co⁴⁺ ions to Co³⁺ ions. When all the peaks were calibrated with respect to B 1s peak corresponding to LBO as a bulk electrolyte, the Co³⁺ peaks shifted consistently with the change in cell voltage during charge/discharge cycles, due to the shift of Fermi level of LCO.

1. Introduction

All-solid-state lithium-ion batteries (ASS-LIBs) are expected to be one of the next-generation rechargeable batteries due to their superior performance, especially, very high safety and durability, compared with the conventional lithium-ion batteries by replacing liquid organic electrolyte solutions with solid-electrolytes.^1,2

In contrast to the sulfide-based ASS-LIBs of which electrode materials and solid electrolytes can be bound by mechanical press at relatively low temperature,^1,3 co-sintering at high temperature is required to join electrode active materials and oxide-based solid electrolytes due to their poor plasticity. Co-sintering of oxide-based solid electrolytes and electrode active materials often causes undesired side reactions, resulting in the decomposition of materials⁴ or the formation of highly resistive interfacial layers.⁵ For example, a highly lithium-ion-conducting garnet-type solid electrolyte, Li₇La₃Zr₂O₁₂ (LLZO) and a high-capacity cathode material, LiCoO₂ are bound together by co-sintering but react with each other to produce a lithium-ion insulating layer.^4,6,7

Over the past decade, various sintering additives have been reported to promote the formation of excellent physical contact between oxide-based solid electrolytes and electrode materials without interfacial reaction layers.^6,8–11 Ohta et al. used Li₃BO₃ (LBO) which is a Li-ion conductive (2 × 10⁻⁶ S cm⁻¹ at RT) and low melting point (∼700 °C) material, as a sintering additive to join LCO and Nb-doped LLZO (Nb-LLZ) pellet.⁸ At 700 °C, LBO melted and infiltrated into the electrode layer composed of LBO and LCO particles on Nb-LLZ pellet, and LCO particles were bound to Nb-LLZ via the lithium-ion conducting LBO with suppressing mutual elemental diffusion and side reactions between LCO and Nb-LLZ.^6,8,10 Thus, the sinterability and electrochemical performance were significantly improved by adding LBO.^6,8

X-ray photoelectron spectroscopy (XPS) is a powerful technique to probe the electronic states and chemical compositions of sample surfaces. In contrast to the conventional XPS using an Al-Kα source which has high surface sensitivity, typically shallower than 10 nm,¹² hard x-ray photoelectron spectroscopy (HAXPES) using synchrotron facilities or laboratory-based x-ray sources such as Cr-Kα (5414.9 eV) and Ga-Kα (9251.7 eV), enables us the non-destructive analyses of buried deeper regions at the depth of ∼50 nm because it generates photoelectrons with a kinetic energy higher than those obtained by the conventional Al-Kα source. Kiuchi et al. applied this technique to a thin layer type ASS cell composed of Al current collector, LiCoO₂ positive electrode, Li_1+x+yAl_x(Ti, Ge)_2−xSi_yP_3−yO₁₂ (LASGTP) solid electrolyte, Li₃PO₄ buffer layer and Li counter electrode and successfully demonstrated operando HAXPES measurements using synchrotron facility.¹³ Later, Hikima et al. unambiguously clarified the electronic state of Li₂MnO₃/LASGTP junction by taking full advantage of the relatively long information depth.¹⁴

In the present study, we constructed a laboratory-based operando HAXPES apparatus and observed electrochemical reaction of a composite positive electrode layer of LCO-LBO mixture formed on a Li_6.6La₃Zr_1.6Ta_0.4O₁₂ (LLZT) solid electrolyte sheet precoated with a Nb thin layer in a bulk type ASS cell in LCO-LBO/Nb/LLZT/Li configuration. We used an oxide-based solid electrolyte LLZT with a relatively high Li ion conductivity, to avoid any possible contamination of vacuum chamber, potentially produced by using sulfide-based solid electrolytes because of their relatively high vapor pressure. In addition, LCO was used as a target for the operando HAXPES observation because its spectral changes in response to charge/discharge cycles are well-documented.^13,15,16 Moreover, the composite electrode of LCO can be relatively easily prepared on the LLZT surface by using LBO as a sintering additive.^6,8

2. Experimental

2.1 Fabrication of an all-solid-state (ASS) cell

One side of a Li_6.6La₃Zr_1.6Ta_0.4O₁₂ solid electrolyte sheet (10 mm × 10 mm × 500 µm; LLZT, Toshima Manufacturing Co., Ltd., Japan) was sandblasted with 220-grit alumina powder to obtain a rough surface. After annealing at 700 °C for 2 hours, the LLZT was immersed in an aqueous solution saturated with LiOH for 1 hour to remove resistive species, such as LiOH and Li₂CO₃. A 20 nm-thick Nb thin layer was deposited on the sandblasted surface of LLZT by DC sputtering using Ar gas, and 1.892 mg of LiCoO₂ (LCO) fine powder with an average particle size of around 200 nm (Fig. S2 in the Supporting Information) was applied to the LLZT precoated with Nb layer. A 10 µL droplet of aqueous solution saturated with H₃BO₃ was placed on the LCO-applied LLZT and then dried in a vacuum. Subsequently, a 10 µL droplet of aqueous solution saturated with LiOH was placed on the LCO-applied LLZT and then dried in a vacuum. Finally, the LLZT was annealed at 700 °C for 2 hours under an oxygen flow to form a LCO-LBO composite positive electrode on a Nb interlayer-coated LLZT. The estimation of loading amount of LBO is described in the Supporting Information. The thickness of LCO-LBO composite electrode layer is around 10 µm (Figs. S1 and S2). Pt thin layer with a thickness of 200 nm was overcoated on a 10 mm × 4 mm rectangular area of LCO-LBO composite positive electrode, as a current collector, as shown in Fig. 1a. The remaining exposed area of LCO-LBO composite positive electrode was the target of HAXPES measurements.

Figure 1.

Schematic illustration of (a) the ASS cell, and (b) the sample holder for HAXPES measurements.

The other side of LLZT was mechanically polished by 600-grit sandpaper and then a 1 µm-thick Li thin layer was formed by thermal evaporation. After placing a piece of Li metal foil with a thickness of 100 µm as a negative electrode, the LLZT was heated at 200 °C for 30 min to yield an ASS cell in the LCO-LBO/Nb/LLZT/Li configuration. Although the Nb thin layer sandwiched by the LCO-LBO composite positive electrode and LLZT should be converted to a LiNbO₃ (LNO) after heating,¹⁷ the interlayer is denoted as Nb because no experimental evidence was obtained in the present study.

2.2 HAXPES measurements

The ASS cell was mounted onto the sample holder as shown in Fig. 1b. The 200 nm-thick Pt thin layer on LCO-LBO positive electrode was electrically connected to the hemispherical analyzer through the terminal A and base plate, while the Li negative electrode was insulated from the LCO-LBO positive electrode side by the insulating plate on the base plate. The cell mounted on sample holder was transferred from an Ar-filled glove box to HAXPES apparatus (Scienta omicron, ULVAC-PHI) without air exposure. After transferring the sample into the analysis chamber, terminals B and C encounter the Li negative electrode and LCO-LBO positive electrode sides, respectively. This allows us to apply a bias between the Li negative electrode and LCO-LBO positive electrode by a potentiostat outside vacuum.

HAXPES measurements were performed in the analysis chamber kept under a pressure of ∼2 × 10⁻⁸ mbar. X-rays from monochromatic Cr-Kα (5414.9 eV) source at a power of 50 W with a spot size of 200 µm were incident to the exposed area of the LCO-LBO composite positive electrode with an incident angle θ₁ of 75°. The pass energy of the photoelectrons was fixed at 200 eV. The obtained spectra were fitted using the Voigt function after the background subtraction by using a Shirley method.¹⁸

In addition to static HAXPES measurements of pristine and electrochemically treated cell 30 min after each charge/discharge process, dynamic HAXPES measurements in which B 1s and Co 2p_3/2 photoelectron spectra were alternately measured throughout the charge/discharge cycles.

2.3 Electrochemical measurements

A galvanostatic charge/discharge cycles were carried out at a constant current of ±12.96 µA corresponding to 0.05 C in the potential ranges of 3.0–4.2 V vs. Li⁺/Li using a potentio-galvanostat (Bio-Logic VSP-300) at room temperature. The potential given herein is expressed versus Li⁺/Li unless otherwise noted.

3. Result

Figure 2 shows typical galvanostatic charge/discharge potential profiles of the ASS cell. In the first a few charge/discharge cycles, capacities for charging are somewhat larger than those of successive discharging. The charge integrations of the first charging and successive discharging were 121 mAh g⁻¹ and 83.3 mAh g⁻¹, respectively, with a Coulombic efficiency of 69.1 %. Those in the second charging and discharging were 99.8 mAh g⁻¹ and 85.2 mAh g⁻¹, respectively, with a Coulombic efficiency of 85.4 %. We denote such initial irreversible cycles as conditioning. The irreversible capacity loss during the conditioning cycles should be due to the anodic instability of LLZT^19,20 and/or irreversible reaction characteristic to low temperature LCO formed by wet chemical process, followed by annealing.²¹ After the conditioning, galvanostatic potential profiles become reversible; with a capacity of ∼85 mAh g⁻¹ and a Coulombic efficiency of ∼100 %.

Figure 2.

(a) Galvanostatic charge/discharge potential profiles, (b) cycle performance and corresponding Coulombic efficiency of the ASS cell.

A characteristic plateau corresponding to the Co^3+/4+ accompanying with lithiation/delithiation of LCO at the potential of ∼3.9 V was observed, confirming the reversible redox reaction.^22,23 The capacity and Coulombic efficiency of the present ASS cell were higher than those of previous report on LBO-free LCO cathode layer directly formed on a garnet-type solid electrolyte⁶ and LCO-LBO composite electrode,⁶ confirming the role of Li-ion conducting Li₃BO₃ (LBO) interlayer at the LCO/LLZT interfaces.⁶ Part of LBO was segregated on the top and bottom of composite positive electrode layer (Figs. S1 and S2). The charge/discharge capacity after conditioning cycles is around 85 mAh g⁻¹ (Fig. 2), which is 62 % of theoretical capacity of LCO, 137 mAh g⁻¹. The ratio of active LCO is still higher than that reported previously by other group.⁶ However, such segregation is considered as the origin of inactive LCO.

Figure 3 shows the Co 2p_3/2 and B 1s photoelectron spectra of the LCO-LBO composite positive electrode before and after conditioning cycles. In the photoelectron spectra without calibration (Figs. 3a and 3b), the Co 2p_3/2 peak shifted by 0.5 eV to a lower binding energy from the pristine to first charged state although the Pt thin layer (current collector) deposited on LCO-LBO composite positive electrode was grounded to the hemispherical analyzer. This shift is attributed to the change in valence state of LCO due to the transition from a p-type semiconducting to a metallic state induced by delithiation.¹³ After the first charging, the position of Co 2p_3/2 peak remained unchanged because the Fermi level of LCO, Pt and hemispherical analyzer were aligned.

Figure 3.

Co 2p_3/2 and B 1s photoelectron spectra of the LCO-LBO composite electrode before and after charge/discharge cycles (a, b) without and (c, d) with calibration using the B 1s peak of LBO at 191.5 eV.

Thus, hereafter, all the spectra were calibrated using B 1s peak of LBO at 191.5 eV as a standard (Figs. 3c and 3d) because the electrochemical potential of electron in bulk electrolyte should be constant.²⁴ Although the potential distribution occurs in the electric double layer (EDL) at the electrode/electrolyte interfaces, thickness of EDL, in the range of ∼10 nm,²⁵ should be much smaller than the size of primary LBO particles, that is in the sub-micrometer range as imaged by SEM (Fig. S1 in the Supporting Information).

In the Co 2p_3/2 region (Fig. 3c), a sharp main peak and a broad satellite peak characteristic to LCO were observed at around 780 eV and 790 eV, respectively.^13,15,26 The main peak can be deconvoluted into two peaks at around 780 eV and 781 eV attributed to Co³⁺ ions of the LCO and final state effect,¹⁵ respectively. After the 1st charging, the main peak was asymmetrically broadened toward a higher binding energy due to the appearance of Co⁴⁺ peak overlapping with the component attributed to the final state effect, with maintaining the Co³⁺ peak unchanged (Fig. 3c). It is noted that the Co⁴⁺ peak cannot be deconvoluted from that of final state effect because their peak positions are very close to each other. Furthermore, the relative areas of satellite peak decreased (Fig. 3c), confirming the oxidation of Co³⁺ to Co⁴⁺ ions.^13,15

After the subsequent discharging, those observed changes were recovered; the FWHM of Co⁴⁺ peak decreased and the relative areas of satellite peak increased (Fig. 3c), confirming that the Co⁴⁺ ions were reversibly reduced to Co³⁺ ions. The magnified graphs of Fig. 3c were shown in Fig. S7 in the Supporting Information. The FWHMs of Co⁴⁺ peak and relative area of satellite peak were repeatedly changed in response to the successive charge/discharge cycles, suggesting the reversible oxidation-reductions of Co ions in LCO. The shape and intensity of B 1s peak corresponding to LBO were almost unchanged before and after charge/discharge cycles (Fig. 3d), showing that it serves as a bulk electrolyte in the composite electrode layer.

Figure 4 shows the time course of photoelectron spectra in the B 1s and Co 2p_3/2 regions of LCO-LBO composite positive electrode repeatedly measured throughout the charge/discharge cycles after conditioning. Upon charging, Co⁴⁺ peak and the satellite peak started to broaden and decrease, respectively, due to the oxidation of Co³⁺ ions as observed in static HAXPES measurements (Fig. 3). Those changes became more prominent with increase in the capacity. In addition to the broadening of Co⁴⁺ peak and decrease of satellite peak, the positions of the Co³⁺ and Co⁴⁺ peaks shifted to a higher binding energy consistently with the cell voltage, i.e., the electrode potential of a Pt thin layer deposited on the LCO-LBO composite electrode, throughout charging process as shown in Fig. 4g. Since the B 1s peak of LBO at 191.5 eV is used as a standard, this shift is attributed to the change in cell voltage, i.e., the electrochemical potential of LCO, in response to the decrease of Li content x in Li_xCoO₂. In the subsequent discharging process, the position of Co³⁺ and Co⁴⁺ peaks reversibly shifted to a lower binding energy consistently with the electrode potential. In addition, FWHM of Co⁴⁺ peak and intensity of satellite peak gradually recovered with the decrease of Li content x in Li_xCoO₂, confirming the reversibility.

Figure 4.

B 1s and Co 2p photoelectron spectra generated during (a, b) charging and (d, e) discharging processes in the potential range of 3.0–4.2 V vs. Li⁺/Li. (c, f) Magnified graphs of dashed box shown in (b, e). (e) Position of Co³⁺ peak and FWHM of Co³⁺ and Co⁴⁺ peak plotted as functions of capacity, together with galvanostatic charge/discharge potential profiles.

4. Conclusion

We developed the laboratory-based HAXPES apparatus equipped with a Cr-Kα source (5414.9 eV) and bias application system and demonstrated operando observation of electrochemical reactions of LCO-LBO composite electrode deposited on a LLZT in an ASS cell configuration during charge/discharge cycles. Reversible broadening/narrowing of Co⁴⁺ peak and decrease/increase of satellite peak were observed in response to the oxidation/reduction of Co³⁺ ions of LCO during charge/discharge cycles. In addition, LCO peak shifted consistently with the change of cell voltage because the B 1s peak due to LBO was used as an internal standard.

Application of the present operando HAXPES apparatus to the non-destructive electronic and chemical analysis of buried electrode/electrolyte interfaces is now on-going using a thin film type ASSBs.

Acknowledgment

The present work was supported by the Materials Processing Science project “Materealize” of the Ministry of Education, Culture, Sports, Science and Technology of Japan (MEXT; Grant Number JPMXP0219207397), and the Center of Innovation NEXT Program (COI-NEXT) of Japan Science and Technology Agency (JST; Grant Number JPMJPF2016). We also acknowledge the NIMS Battery Platform, especially Ms. Makiko Oshida, for the technical support on SEM measurements.

Data Availability Statement

The data that support the findings of this study are openly available under the terms of the designated Creative Commons License in J-STAGE Data at https://doi.org/10.50892/data.electrochemistry.24425332.

• 1) A positive electrode composed of LiCoO₂ (LCO) and Li₃BO₃ (LBO) was formed on one side of a Li_6.6La₃Zr_1.6Ta_0.4O₁₂ (LLZT) solid electrolyte sheet by applying LCO fine powder to LLZT sheet precoated with a Nb thin layer, placing a droplet of aqueous solution of LiOH and H₃BO₃, and being annealed at ∼700 °C in an oxygen atmosphere. After binding a negative electrode, a Li metal foil, on the other side of LLZT sheet precoated with a Li thin layer, electrochemical reaction at the LiCoO₂-Li₃BO₃ composite positive electrode was observed in an all-solid-state battery configuration, i.e., a LCO-LBO/Nb/LLZT/Li cell, by a newly-developed laboratory-based hard x-ray photoelectron spectroscopy (HAXPES) apparatus equipped with a Cr-Kα source (5414.9 eV) and bias application system. A sharp main peak and a broad satellite peak characteristic to LCO were observed in the Co 2p_3/2 region at the pristine state. During charging, i.e., delithiation from LCO, the main peak was asymmetrically broadened to a higher binding energy due to the partial oxidation of Co³⁺ ions at 780 eV to Co⁴⁺ ions at ∼781 eV. In addition, the full width half maximum (FWHM) of the Co⁴⁺ peak increased with increasing the amount of lithium insertion, while that of the Co³⁺ peak remained unchanged. The decrease of satellite peak further confirms the oxidation of Co³⁺ ions. During the subsequent discharging, i.e., lithiation of LCO, those recovered to the original states, confirming the reversible reduction of Co⁴⁺ ions to Co³⁺ ions. When all the peaks were calibrated with respect to B 1s peak corresponding to LBO as a bulk electrolyte, the Co³⁺ peaks shifted consistently with the change in cell voltage during charge/discharge cycles, due to the shift of Fermi level of LCO.

CRediT Authorship Contribution Statement

Tsukasa Iwama: Investigation (Lead), Writing – original draft (Lead)

Tsuyoshi Ohnishi: Investigation (Equal), Methodology (Equal), Writing – review & editing (Supporting)

Takuya Masuda: Conceptualization (Lead), Funding acquisition (Lead), Investigation (Supporting), Methodology (Supporting), Project administration (Lead), Resources (Lead), Supervision (Lead), Writing – review & editing (Lead)

Conflict of Interest

The authors declare no conflict of interest in the manuscript.

Funding

MEXT: JPMXP0219207397

Japan Science and Technology Agency: JPMJPF2016

Footnotes

T. Iwama: ECSJ Student Member

T. Masuda: ECSJ Active Member

References

• 1)  K. Takada, Acta Mater., 61, 759 (2013).
• 2)  M. G. Boebinger, J. A. Lewis, S. E. Sandoval, and M. T. McDowell, ACS Energy Lett., 5, 335 (2020).
• 3)  A. Kato, M. Nose, M. Yamamoto, A. Sakuda, A. Hayashi, and M. Tatsumisago, J. Ceram. Soc. Jpn., 126, 719 (2018).
• 4)  F. Ichihara, S. Miyoshi, and T. Masuda, Phys. Chem. Chem. Phys., 24, 25878 (2022).
• 5)  K. Nagao, A. Hayashi, and M. Tatsumisago, J. Ceram. Soc. Jpn., 124, 915 (2016).
• 6)  K. Park, B.-C. Yu, J.-W. Jung, Y. Li, W. Zhou, H. Gao, S. Son, and J. B. Goodenough, Chem. Mater., 28, 8051 (2016).
• 7)  M. Kotobuki, H. Munakata, K. Kanamura, Y. Sato, and T. Yoshida, J. Electrochem. Soc., 157, A1076 (2010).
• 8)  S. Ohta, S. Komagata, J. Seki, T. Saeki, S. Morishita, and T. Asaoka, J. Power Sources, 238, 53 (2013).
• 9)  S. Ohta, J. Seki, Y. Yagi, Y. Kihira, T. Tani, and T. Asaoka, J. Power Sources, 265, 40 (2014).
• 10)  F. Han, J. Yue, C. Chen, N. Zhao, X. Fan, Z. Ma, T. Gao, F. Wang, X. Guo, and C. Wang, Joule, 2, 497 (2018).
• 11)  T. Okumura, T. Takeuchi, and H. Kobayashi, Solid State Ionics, 288, 248 (2016).
• 12)  K. Hikima, K. Shimizu, H. Kiuchi, Y. Hinuma, K. Suzuki, M. Hirayama, E. Matsubara, and R. Kanno, J. Am. Chem. Soc., 144, 236 (2022).
• 13)  H. Kiuchi, K. Hikima, K. Shimizu, R. Kanno, F. Toshiharu, and E. Matsubara, Electrochem. Commun., 118, 106790 (2020).
• 14)  K. Hikima, K. Shimizu, H. Kiuchi, Y. Hinuma, K. Suzuki, M. Hirayama, E. Matsubara, and R. Kanno, Commun. Chem., 5, 52 (2022).
• 15)  L. Dahéron, R. Dedryvère, H. Martinez, M. Ménétrier, C. Denage, C. Delmas, and D. Gonbeau, Chem. Mater., 20, 583 (2008).
• 16)  J. C. Dupin, D. Gonbeau, H. Benqlilou-Moudden, P. Vinatier, and A. Levasseur, Thin Solid Films, 384, 23 (2001).
• 17)  N. Ohta, K. Takada, I. Sakaguchi, L. Q. Zhang, R. Z. Ma, K. Fukuda, M. Osada, and T. Sasaki, Electrochem. Commun., 9, 1486 (2007).
• 18)  M. Schmid, H.-P. Steinrück, and J. M. Gottfried, Surf. Interface Anal., 46, 505 (2014).
• 19)  K. Kanamura, T. Kimura, M. Shoji, and H. Munakata, ECS Meet. Abstr., MA2016-03, 45 (2016).
• 20)  M. Sakakura, K. Mitsuishi, T. Okumura, N. Ishigaki, and Y. Iriyama, ACS Appl. Mater. Interfaces, 14, 48547 (2022).
• 21)  Y. Ren, T. Liu, Y. Shen, Y. Lin, and C.-W. Nan, Ionics, 23, 2521 (2017).
• 22)  J. N. Reimers and J. R. Dahn, J. Electrochem. Soc., 139, 2091 (1992).
• 23)  S. Li, S. Tsutsumi, S. Shironita, and M. Umeda, Electrochemistry, 90, 037004 (2022).
• 24)  X. Wu, C. Villevieille, P. Novak, and M. El Kazzi, Phys. Chem. Chem. Phys., 20, 11123 (2018).
• 25)  Y. Nomura, K. Yamamoto, T. Hirayama, S. Ouchi, E. Igaki, and K. Saitoh, Angew. Chem., Int. Ed. Engl., 58, 5292 (2019).
• 26)  A. T. Appapillai, A. N. Mansour, J. Cho, and Y. Shao-Horn, Chem. Mater., 19, 5748 (2007).