Lithium intercalation in the surface region of an LiNi1/3Mn1/3Co1/3O2 cathode through different crystal planes

Epitaxial LiNi1/3Co1/3Mn1/3O2 film electrodes with orientations of (104), (1 18) and (003) were fabricated on SrRuO3/SrTiO3 by pulsed laser deposition. The films have a thickness of 23.8 to 25.0 nm and a flat surface with a roughness of approximately 2 nm, which offered a model system for clarifying the reaction plane dependencies of lithium intercalation at the LiNi1/3Co1/3Mn1/3O2 surface. All reaction planes delivered reversible lithium intercalation for electrochemical charging–discharging between 3.0 V and 4.3 V (vs. Li/Li). The (104) surface exhibited reversible behavior at a higher operation voltage between 3.0 V and 4.5 V, but the (1 18) and (003) planes showed fading of the discharge capacity and average discharge voltage. The anisotropic stability of the surface region indicates the importance of crystallographic facet control for the development of an LiNi1/3Co1/3Mn1/3O2 cathode with high cycle stability.


Introduction
LiNi 1/3 Co 1/3 Mn 1/3 O 2 with a layered rocksalt structure is one of the most promising cathode materials in lithium ion batteries due to its high discharge capacity, cost competitiveness, and high safety over conventional LiCoO 2 . 1-3 The LiNi 1/3 Co 1/3 Mn 1/3 O 2 delivers a reversible lattice change during lithium (de)intercalation under high-voltage operation to 4.5 V, 4 which is expected for use in high-energy-density batteries for large-scale applications such as electric vehicles or hybrid electric vehicles. However, the LiNi 1/3 Co 1/3 Mn 1/3 O 2 suffers from severe capacity fading and poor high-rate capability, which is associated with poor electronic conductivity and side reactions with electrolyte species at the electrode and electrolyte interface. 5 A cubic rock-salt-type phase is typically formed near the surface of LiNi 1/3 Co 1/3 Mn 1/3 O 2 and has been considered to increase the reaction resistance at the electrochemical interface, resulting in the fading of the chargedischarge capacity. 6,7 As with other cathode materials, it has been proposed that surface coating with oxide materials is an effective way to stabilize the LiNi 1/3 Co 1/3 Mn 1/3 O 2 surface. [6][7][8][9][10][11] A recent study using scanning transmission electron microscopy demonstrated that the formation of the rocksalt-type phase in the surface region was suppressed by an aluminum oxide coating on LiNi 1/3 Co 1/3 Mn 1/3 O 2 . 7 However, the detailed mechanisms of fading and stabilization in the surface region are still unclear due to insufficient experimental information for the elucidation of the complicated reaction eld and its associated reaction parameters.
Epitaxial thin-lm electrodes have the following features for clarifying the surface reaction mechanism in lithium ion batteries. [12][13][14][15][16][17] First, they have very at surfaces with roughness of approximately 2 nm and include no additives such as conductive carbon and binders, which simplify surface reactions. Second, the lm thickness can be controlled in the range of 10 to 100 nm; a very thin lm can function as a surface-enhanced electrode. Third, the reaction eld is restricted by the lattice orientation, which can provide information on anisotropic reaction mechanisms. Although epitaxial growths of some layered rocksalt oxides have been reported, 12,[18][19][20][21][22][23][24] there have been no reports of the successful synthesis of an epitaxial LiNi 1/3 Co 1/3 Mn 1/3 O 2 lm electrode. Previously, LiNi 1/3 Co 1/3 Mn 1/3 O 2 thin-lms were synthesized by magnetron sputtering and PLD methods on Pt/Si, Au, and Li 1+x+y Al x Ti 2Àx Si y P 3Ày O 12 substrates. 25,26 These lms were in a polycrystalline state with an oriented character in the (003) plane 26 or with random orientations. 25 Furthermore, they included amorphous LiNi 1/3 Co 1/3 Mn 1/3 O 2 owing to low synthetic temperatures, below 500 C. Recently, it has been reported that layered rock-salt-type Li(Mn,Co,Ni)O 2 (1À18) lms were epitaxially grown on an Nb:SrTiO 3 (110) single crystal by pulsed laser deposition. 27 However, the composition of the lms could not be determined owing to contamination with a large amount of disordered spinel and/or rock-salt-type phases, and thus the charge-discharge properties were not consistent with those observed for polycrystalline LiNi 1/3 Co 1/3 Mn 1/3 O 2 electrodes. Furthermore, no other orientations of epitaxial lms were fabricated.
In this work, we investigated synthetic conditions using pulsed laser deposition and successfully fabricated epitaxial LiNi 1/3 Co 1/3 Mn 1/3 O 2 lms with orientations of (1À18), (104), and (003) on Nb:SrTiO 3 (110), (100), and (111) substrates, respectively. The structure and chemical composition of the epitaxial lms were conrmed by X-ray diffraction (XRD), X-ray reectivity (XRR), inductively-coupled plasma mass spectrometry (ICP-MS), and X-ray absorption near-edge structure (XANES) measurements. The lithium intercalation properties of the LiNi 1/3 Co 1/3 Mn 1/3 O 2 surface were examined by chargedischarge measurements using epitaxial LiNi 1/3 Co 1/3 Mn 1/3 O 2 lm model electrodes with a thickness of approximately 25 nm. The anisotropy of lithium intercalation in the surface region of LiNi 1/3 Co 1/3 Mn 1/3 O 2 is discussed with respect to the reaction plane dependence of the charge-discharge capacity and the cycle retention of the model electrodes.
XRD and XRR measurements were performed in air using a thin-lm X-ray diffractometer (Rigaku ATX-G) with Cu Ka 1 radiation. The lm orientation was determined from XRD patterns collected along out-of-plane and in-plane directions. The lm thickness, X-ray scattering length density (SLD), and roughness were determined by XRR analysis using Parratt32 soware. 28,29 XRR spectra were plotted as a function of the scattering vector, Q z ¼ 4p sin q/l, where l is the X-ray wavelength (1.541Å) and q is the incident angle. Elemental ratios of Li, Ni, Co, and Mn in the thin lms were determined using ICP-MS (Agilent 7500cs, Agilent Technologies) from samples dissolved in aqua regia diluted with ultrapure water at 100 C for 30 minutes. XANES measurements were performed in the uorescence mode, using a germanium single-element solid-state detector installed at SPring-8 BL14B2, and data were collected at an oblique incidence angle of 4 . Pre-edge background and post-edge normalizations of spectra to unity were performed using the ATHENA soware package. 30 Charge/discharge characteristics were examined using a 2032-type coin cell. The cells were assembled inside an argon-lled glove box using lithium metal as counter electrode and the LiNi 1/3 Co 1/3 Mn 1/3 O 2 lm as the working electrode. The electrolyte comprised ethylene carbonate-diethyl carbonate with a volume ratio of 3 : 7 (EC/DEC; Kishida Chemical Co., Ltd., >99.5%) as the solvent and 1.0 mol dm À3 LiPF 6 as the supporting electrolyte. The charge-discharge characteristics were examined at room temperature using a multi-channel potentio/ galvanostat (TOSCAT 3100). The cut-off voltages were 3.0 and 4.3 V, 3.0 and 4.5 V, and 2.0 and 4.5 V for discharge and charge, respectively. The charge-discharge capacities of the cells were calculated by taking into account the area over which the lm was deposited (10 Â 7 mm), the lm thickness and the theoretical density of LiNi 1/3 Co 1/3 Mn 1/3 O 2 (4.77 g cm À3 ). The constant current applied was 1 mA. value of the LiNi 1/3 Co 1/3 Mn 1/3 O 2 (1À18) lm were rened to 25.0 nm, 2.2 nm, and 3.78 Â 10 À3 nm À2 , respectively (see Table S1 in ESI †). The SLD value was consistent with that expected for stoichiometric LiNi 1/3 Co 1/3 Mn 1/3 O 2 (3.81 Â 10 À3 nm À2 ). 31 The at SLD prole of the LiNi 1/3 Co 1/3 Mn 1/3 O 2 layer indicated a uniform lm composition in the depth direction. The lm orientation of (1À18) on the SrRuO 3 /Nb:SrTiO 3 (110) was consistent with that for Li-(Mn,Co,Ni)-O lms reported previously. 27 The LiNi 1/3 Co 1/3 Mn 1/3 O 2 lm fabricated in this study exhibited no diffraction peaks of other phases (except for those of SrRuO 3 /Nb:SrTiO 3 ) in the out-of-plane XRD pattern, although a 220 diffraction peak for a disordered spinel phase was observed at around 32 in the XRD patterns of the Li-(Mn,Co,Ni)-O(1À18) lm. This indicates that the LiNi 1/3 Co 1/3 Mn 1/3 O 2 (1À18) lm contained a much lower amount of the spinel-phase impurity. Table 1 lists the d values, intensities, and full width at half maximum (FWHM) values of the out-of-plane 1À18 and in-plane 110 diffraction peaks observed for the epitaxial lms fabricated in this study and in the previous study. 27 The positions and intensities of the 1À18 and 110 peaks were normalized using the values of the 110 and 2À20 peaks of the Nb:SrTiO 3 , respectively. The 1À18 and 110 diffraction peaks of the LiNi 1/3 Co 1/3 Mn 1/3 O 2 lm had higher relative intensities than those of the Li-(Mn,Co,Ni)-O lm, whereas no considerable difference in the FWHM value was observed. These XRD results demonstrated the formation of a well-crystallized LiNi 1/3 Co 1/3 Mn 1/3 O 2 (1À18) lm with high purity, compared to the Li-(Mn,Co,Ni)-O lm. A previous report proposed that the structural transformation from the layered rock salt to the cubic spinel is caused by lithium loss during the high temperature PLD process. 12,18,20,27,32 In this study, an Li 1.5 Ni 1/3 Co 1/3 Mn 1/3 O 2 target with a high lithium content (50% excess) was used, compared to that used in the previous study (Li 1.3 Ni 1/3 Co 1/3 Mn 1/3 O 2 ). The excess lithium in the target could compensate for the lithium loss, resulting in a LiNi 1/3 Co 1/3 Mn 1/3 O 2 (1À18) lm with high purity. Furthermore, the higher lithium content in the target enabled us to apply a high substrate temperature of 650 C, which was higher than that in the previous study (600 C). This could contribute to the high crystallinity of the lm. The purity and crystallinity of the LiNi 1/3 Co 1/3 Mn 1/3 O 2 lms were considerably improved by optimizing the synthetic conditions. LiNi 1/3 Co 1/3 Mn 1/3 O 2 lms were synthesized on the SrRuO 3 / Nb:SrTiO 3 (100) and SrRuO 3 /Nb:SrTiO 3 (111) substrates under the same PLD conditions used for the LiNi 1/3 Co 1/3 Mn 1/3 O 2 (1À18) lm. The XRD patterns and XRR analysis results of the LiNi 1/3 Co 1/3 Mn 1/3 O 2 lms are shown in Fig. 2. The LiNi 1/3 Co 1/3 Mn 1/3 O 2 lm deposited on the (100) substrate had h04h and h0À8h orientations along the out-of-plane [100] and in-plane [011] directions of the Nb:SrTiO 3 substrate, respectively ( Fig. 2a and b). No other diffraction peaks of the lm were observed in the XRD patterns, which indicates the epitaxial growth of the LiNi 1/3 Co 1/3 Mn 1/3 O 2 (104) lm. The same structural model provided a good tting result for the XRR spectrum (Fig. 2c). The thickness, surface roughness, and X-ray SLD value of the LiNi 1/3 Co 1/3 Mn 1/3 O 2 (104) lm were rened to 23.4 nm, 1.8 nm, and 3.76 Â 10 À3 nm À2 , respectively. The LiNi 1/3 Co 1/3 Mn 1/3 O 2 lm synthesized on the SrRuO 3 /Nb:SrTiO 3 (111) substrate also showed epitaxial growth with orientations of 00l and hh0 along the out-of-plane [111] and in-plane [1À10] directions of the Nb:SrTiO 3 substrate, respectively ( Fig. 2d and e). From the XRR analysis result (Fig. 2f), the thickness, surface roughness, and X-ray SLD value of the LiNi 1/3 Co 1/3 Mn 1/3 O 2 (003) lm were rened to 29.1 nm, 2.3 nm, and 3.78 Â 10 À3 nm À2 , respectively. Table 2  Furthermore, the oxidation states of Ni, Co, and Mn ions in the LiNi 1/3 Co 1/3 Mn 1/3 O 2 lm were investigated by XANES. Fig. 3 shows the Mn-K, Co-K, and Ni-K edges XANES spectra obtained from a LiNi 1/3 Co 1/3 Mn 1/3 O 2 (104) lm and polycrystalline standard materials. The XANES spectra indicated that the oxidation states of Ni, Co, and Mn were identical to those of polycrystalline NiO, LiCoO 2 , and Li 2 MnO 3 , respectively. The ICP and XANES results conrmed the growth of the epitaxial LiNi II 1/3 Co III 1/3 Mn IV 1/3 O 2 lms with no lithium deciency. Fig. 4 depicts the charge-discharge curves and dQ/dV curves of epitaxial LiNi 1/3 Co 1/3 Mn 1/3 O 2 lms with (1À18), (104), and (003) reaction planes. The electrochemical measurements were conducted between 3.0 V and 4.3 V during the initial ve cycles and between 3.0 V and 4.5 V from the 6th to the 10th cycles. For cut-off voltages of 3.0 and 4.3 V, the potential proles are almost independent of the reaction plane. During the charging process, the voltage rapidly increased to approximately 3.6 V and rose gradually to 4.3 V. The voltage gradually decreased to approximately 3.6 V. Then rapidly decreased to 3.0 V. Reversible oxidation and reduction peaks were observed at around 3.7 V in the dQ/dV curves, which corresponded to lithium    33,34 The rst charge/discharge capacities were 345/124 mA h g À1 , 291/123 mA h g À1 , and 275/106 mA h g À1 , for the (1À18), (104), and (003) lms, respectively. Large irreversible capacities of over 150 mA h g À1 were observed for all lms, which is a typical feature of nanosized lm electrodes because of side reactions that are quite prominent due to the small amount of active material. 16,23,35 The irreversible capacity gradually decreased during subsequent cycles. When the higher cutoff voltage was changed to 4.5 V at the 6th cycle, the LiNi 1/3 Co 1/3 Mn 1/3 O 2 (1À18), (104), and (003) lms delivered higher discharge capacities of 153, 136, and 122 mA h g À1 , respectively. A larger amount of lithium was deintercalated from the LiNi 1/3 Co 1/3 Mn 1/3 O 2 lattice at above 4.3 V during the charging process and then reversibly intercalated into it during the discharging process. 2,3 Fig. 5 depicts the cycle retention of the discharge capacity and the average discharge voltage of the (1À18), (104), and (003) lms, which were analysed using the charge-discharge curves shown in Fig. 4. The (104) lm exhibited no signicant fading of the discharge capacity or the average discharge voltage under battery operation between 3.0 and 4.3 V, whereas the (1À18) and (003) lms showed a small decrease in capacity and discharge voltage during the initial ve cycles. This behaviour was clearly observed when the batteries were operated between 3.0 and 4.5 V. In contrast, the discharge capacity of the (104) lm showed   (Table 2). This speculation is consistent in that the (1À18) and (104) lms delivered similar charge-discharge capacities between 3.0 and 4.3 V. However, the (104) lm exhibited a superior charge-discharge capacity and cycle retention compared with the (1À18) lm under battery operation between 3.0 and 4.5 V. This indicated the anisotropy of the electrochemical stability of the delithiated Li 1Àx Ni 1/3 Co 1/3 Mn 1/3 O 2 formed at above 4.3 V. The delithiated (1À18) surface might deteriorate by dissolution of transition metal ions, induced by side reactions with electrolyte species in high voltage regions, which has been generally observed for various electrode materials. 6,14,36,37 In contrast to the (1À18) and (104) lms, the (003) lm with the [LiO 6 ] layers parallel to the electrolyte is expected to exhibit poor lithium intercalative activity. However, the (003) lm with the [LiO 6 ] layers parallel to the electrolyte also exhibited reversible intercalative activity, although the charge-discharge capacities were smaller than those observed for the (1À18) and (104) lms. From atomic force microscopy, islands with the maximum height of #5 nm were observed at the surface of the 29.1 nm-thick (003) lm (see Fig. S1 in ESI †). Although the morphology indicates that the lithium (de)intercalation could proceed through the island edges in the top surface regions, it could not explain the discharge capacity of over 100 mA h g À1 observed for the (003) lm. These results suggested lithium diffusion along the h001i direction in the LiNi 1/3 Co 1/3 Mn 1/3 O 2 (003) lattice. It has been reported that a layered rock-salt-type LiNi 0.5 Mn 0.5 O 2 with a large degree of cation mixing between lithium and nickel sites exhibits reversible lithium intercalation. 32 In the LiNi 0.5 Mn 0.5 O 2 structure, lithium ions can diffuse along the c-axis between neighboring two-dimensional lithium layers, through the lithium sites in the transition metal layer. Thus, we can speculate that the LiNi 1/3 Co 1/3 Mn 1/3 O 2 (003) lm has a certain amount of lithium is introduced in the transition metal layer by the cation mixing, which contributes to the lithium diffusion from the LiNi 1/3 Co 1/3 Mn 1/3 O 2 lattice to the electrolyte through the (003) crystal plane. However, the (003) lm showed severe fading of the discharge capacity and the discharge voltage under the high cut-off voltage operation, compared to the (104) and (1À18) lms. This could be associated with the poor stability of the delithiated (003) surface. We can speculate, based on the lithium diffusion mechanism in the (003) lm, that the lithium vacancies are introduced into the twodimensional transitional layers when the state of charge increases above 4.3 V, which could destabilize the delithiated Li 1Àx Ni 1/3 Co 1/3 Mn 1/3 O 2 structure. The epitaxial-lm model electrodes indicated the importance of crystallographic facet control for stable lithium intercalation in the surface region of the cathode, although further investigations based on crystal structure and oxidation state analyses are needed for elucidating the detailed surface reaction mechanism.

Conclusion
Epitaxial LiMn 1/3 Co 1/3 Ni 1/3 O 2 lms with different reaction planes of (104), (1À18) and (003) were successfully synthesized on SrRuO 3 /SrTiO 3 by pulsed laser deposition using a target with a large excess of lithium. The lms have a thickness between 23.8 and 25.0 nm and a at surface with a roughness of below 2.0 nm. The similarity of the structures and compositions of the (104), (1À18), and (003) lms provided a model system for a mechanistic study of the anisotropy of lithium intercalation. All the reaction planes delivered reversible lithium intercalation for electrochemical charging/discharging between 3.0 V and 4.3 V. The (104) plane exhibited reversible behavior between 3.0 V and 4.5 V, but the (1À18) and (003) planes showed a decrease in discharge capacity and average discharge voltage. The (104) plane was highly stable during the deintercalation process above 4.3 V, compared to the (1À18) and (003) planes. This indicates the importance of crystallographic facet control for practical use: polycrystalline LiMn 1/3 Co 1/3 Ni 1/3 O 2 with a predominant (104) facet may deliver superior cycle stability.
(RISING Project) of the New Energy and Industrial Technology Development Organization (NEDO; Japan) and the Grant-in-Aid for Scientic Research (A, C) of Japan Society for the Promotion of Science (No. 25248051 and 16K05929). The synchrotron X-ray experiments were performed as projects approved by the Japan Synchrotron Radiation Research Institute (2014B1920).