Why is the O3 to O1 phase transition hindered in LiNiO 2 on full delithiation? † – |

Ni-enriched layered materials are utilized as positive electrode materials of high-energy Li-ion batteries. Because electrode reversibility is gradually lostfor stoichiometric LiNiO 2 after continuous cycles,Niions are partially substituted by other metal ions (Co, Mn, Al etc .). However, the origin of deterioration in stoichiometric LiNiO 2 is still not fully understand yet. Moreover, the loss of capacities is observed only in the high voltage region (>4.1 V), which is obviously di ﬀ erent from the failure mode observed in other electrode materials. Here, we report for the ﬁ rst time the origin of deterioration, which is revealed by an in situ X-ray di ﬀ raction study. For fully charged NiO 2 , Ni ions migrate from original octahedral sites in NiO 2 slabs to face-sharing tetrahedral sites in Li layers, by which the O3 to O1 phase transition is suppresed. Note that Ni migration is a reversible process, and the Ni ions migrate back to the original octahedral sites on discharge. However, after continuouscycles, thereversibility ofNimigration is gradually lost, and Ni ions are partially left at the tetrahedral sites in Li layers. Electrode kinetics are also deteriorated because of the Ni occupation in Li layers, and the accumulation of Ni ions at tetrahedral sites results in the loss of reversible capacities in the high voltage region. This ﬁ nding opens a new way to design high-capacity Ni-enriched electrode materials, leading to the development of high-energy Li-ion batteries.

Li-containing layered oxides are used for energy storage applications since the historic inception of LiCoO 2 in 1980. 1 LiCoO 2 is isostructural with a-NaFeO 2 whose structure consists of a cubic close packed (ccp) array of oxide ions, and Co and Li ions are lled at vacant octahedral sites in the ccp array of anionic species. 2 Neighboring CoO 6 or LiO 6 octahedra share three edges, forming CoO 2 and LiO 2 slabs. These CoO 2 and LiO 2 slabs are alternately stacked, forming the a-NaFeO 2 -type layered structure. The a-NaFeO 2 -type layered structure is also classied as an O3-type layered structure according to Delmas' notation. 3 Herein, "O" indicates that Li ions are located at octahedral Li sites in LiCoO 2 , and "3" denotes that three CoO 2 slabs are found in the unit cell. When LiCoO 2 is charged in a Li cell, trivalent Co ions are oxidized to tetravalent Co ions. Simultaneously, Li ions are extracted from the host structure for charge compensation. Li x CoO 2 shows an operating voltage of approximately 4 V vs. metallic Li and delivers a reversible capacity of $180 mA h g À1 with cationic redox of Co ions. LiCoO 2 , combined with a graphite negative electrode, is still widely used as a power source for mobile electronics applications.
Charging to high voltage (4.8 V) and removal of all Li ions from LiCoO 2 result in gliding of CoO 2 slabs, forming CoO 2 with an O1-type layered structure, in which octahedral sites of Li layers share faces with the CoO 6 octahedra. Large repulsive electrostatic interaction between Co and Li ions is anticipated for the O1 phase. The O1 phase transition is, therefore, allowed only for a fully delithiated phase, or a small number of Li ions is probably found at neighboring tetrahedral sites, which share edges with the CoO 6 octahedra. This observation was rst evidenced in 1996 by Amatucci et al. using an in situ X-ray diffraction (XRD) study. 4 O1-type CoO 2 is isostructural with CdI 2 with space group P 3m1. The formation of O1 CoO 2 has also been reported in other publications. 5,6 In contrast to LiCoO 2 , O1-type NiO 2 cannot be easily obtained even though LiNiO 2 is isostructural with LiCoO 2 . The formation of O1 NiO 2 was reported aer long time exposure at 4.45 V. 7 The partial formation of the O1 NiO 2 phase was also reported in LiNiO 2 aer acid treatment. 8 The synthesis of stoichiometric LiNiO 2 is difficult, and oen results in the formation of non-stoichiometric Li 1Àx Ni x NiO 2 . Ni 2+ ions are located at Li sites in Li 1Àx Ni x NiO 2 , and these and extra Ni ions disturb the phase transition from the O3 to O1 phase. 8 Thermal instability of NiO 2 is another practical issue as an electrode material. 9,10 Recently, Sun et al. developed the synthesis method of nearly stoichiometric LiNiO 2 at 650 C in an oxygen atmosphere, and nearly stoichiometric LiNiO 2 delivers a reversible capacity of 240 mA h g À1 , which corresponds to 90% of the theoretical capacity of LiNiO 2 , at a rate of 18 mA g À1 . 11 Nevertheless, the formation of O1 NiO 2 was not evidenced aer the extraction of nearly all Li ions, and instead, an increase in the peak intensity at the 101 diffraction line was noted. 11 In this article, the phase transition process of nearly stoichiometric LiNiO 2 is carefully examined by in situ XRD and high-resolution ex situ synchrotron XRD studies. Our analysis reveals that Ni ions migrate to adjacent tetrahedral sites in Li layers on charging, which is a reversible process for the initial few cycles. Nevertheless, the reversible capacity in the high voltage region is gradually lost because Ni ions are in part le at tetrahedral sites in the Li layers aer continuous cycles. This fact results in the loss of the reversible capacity only in the high voltage region (>4.1 V), but not in the low voltage region (<4.1 V). The nding, including the suppression of O1 phase formation in NiO 2 , is clearly a new criterion to explain the deterioration of Li y NiO 2 , which potentially provides alternative approaches to design high capacity Ni-based electrode materials.
Nearly stoichiometric LiNiO 2 was synthesized by the methodology developed by Sun et al. LiOH$H 2 O (Wako Chemical Co.) was crushed into ne powder by using a mortar and pestle. Thus obtained LiOH$H 2 O powder was mixed with Ni(OH) 2 (Wako Chemical Co.) by using a mortar and pestle, and then a mixture of LiOH$H 2 O and Ni(OH) 2 was pressed into a pellet. 3 mol% excess LiOH was utilized to compensate for the vaporization of LiOH on heating. The pellet was heated at 650 C in an oxygen atmosphere for 12 h. The sample aer calcination was stored in an Ar-lled glove box because LiNiO 2 reacts with moisture in air. 12 The XRD pattern and morphological character of LiNiO 2 discussed in this study are shown in ESI Fig. S1a, † and the sample delivers an initial reversible capacity of 220 mA h g À1 at 10 mA g À1 as shown in ESI Fig. S1b. † Selected galvanostatic charge/discharge curves are shown in Fig. 1a. Coulombic efficiency is also plotted in ESI Fig. S1c. † Although the reversible capacity deteriorated from 220 to 160 mA h g À1 aer 30 continuous cycles, the reversible capacity is lost only in the high voltage region (>4.1 V). No capacity reduction is evidenced in the voltage region below 4.1 V, which is clearly displayed in the differential dQ/dE curves shown in Fig. 1b. The peak at 4.2 V is gradually lost aer 30 continuous cycles while symmetric oxidation/reduction proles are retained in other peaks below 4.1 V. Observed discharge capacities in different voltage regions are also plotted in Fig. 1c. Similar deterioration of the electrode reversibility in the high-voltage region is also reported in a recent publication. 13 The loss of reversible capacity only in the high voltage region suggests that the loss of reversible capacities originates from intrinsic phase transition behavior of the high voltage region as discussed in a later section. This observation also indicates that the below three failure mechanisms are eliminated and/or negligible; (1) the loss of electrical contact and isolation of particles in the composite electrode, associated with pulverization of LiNiO 2 , (2) dissolution of active materials in electrolyte, (3) the damage to the crystal structure, for instance oxygen loss from the crystal lattice. These failure processes inevitably result in the loss of reversible capacity in the entire voltage region, which cannot explain the degradation process of LiNiO 2 as observed in Fig. 1. Note that the impedance is gradually increased upon electrochemical cycling (ESI Fig. S1d †). Similar degradation of reversible capacities and the growth of impedance were also reported with 4.3 V cut-off for Li y NiO 2 (ref. 11) (4.5 V cut-off was used in this study). Because electrolyte decomposition would be ignored with 4.3 V cut-off, it is anticipated that electrolyte decomposition and surface lm formation are also less responsible for the degradation process.
To assess the loss of electrode reversibility in Li y NiO 2 in the high voltage region, in situ XRD data were collected, and the data were compared with those of Li y CoO 2 . The XRD pattern and electrochemical data of LiCoO 2 used in this study are shown in ESI Fig. S2. † In situ XRD patterns were obtained using an electrochemical cell equipped with a Be window and XRD diffractometer (Bruker, D8 Advance). The slurry was pasted on thin Al foil (5 mm) and used as a composite electrode for in situ measurements. The cell was charged at a rate of 5 mA g À1 and the XRD data were collected in an initial charge process. Fig. 2 compares the in situ XRD patterns of Li y NiO 2 and Li y CoO 2 with charge curves at a slow rate of 5 mA g À1 . Phase transition processes observed in Li y CoO 2 are consistent with the literature, 5 and the original rhombohedral phase with a cubic close- Fig. 1 (a) Galvanostatic charge/discharge curves of LiNiO 2 at a rate of 10 mA g À1 , (b) differential dQ/dE plots obtained from (a), and (c) variations in discharge capacities with different voltage ranges. packed lattice changes into the O1 phase with a hexagonal close-packed lattice via the intermediate phases, monoclinic Li 0.5 CoO 2 and rhombohedral Li 0.25 CoO 2 with the O3/O1 intergrown layered structure (oen denoted as the H1-3 phase 5 ). The O3 to O1 phase transition is accompanied by a larger change in interlayer distances (a gap of 0.54Å for Li 0.5 CoO 2 and CoO 2 ). Gliding of CoO 2 layers is also required for the O3 to O1 phase transition. Moreover, to form the O1 phase, charging to high voltage (>4.5 V) is needed, which destabilizes the surface of oxides. 14,15 Therefore, deterioration of capacity retention is observed for LiCoO 2 when nearly all Li ions are extracted from the host structure (ESI Fig. 2b). In contrast, a clearly different trend is noted for LiNiO 2 . The formation of the O1 phase is not evidenced and rhombohedral symmetry is retained aer charging to 4.5 V. The 003 diffraction line for the fully charged sample is found at 19.6 , which is located at a lower diffraction angle when compared with the 001 diffraction line for O1 CoO 2 at 20.6 . In addition, a relatively small gap for interlayer distances (0.33Å) is observed for Li 0.3 NiO 2 and NiO 2 which have the longest and shortest interlayer distances, respectively. Note that the peak intensity of 101 and 110 diffraction lines is clearly increased for NiO 2 aer charging to 4.5 V. The highest intensity for the 101 diffraction line is observed aer charging to 4.5 V. Ex situ synchrotron XRD patterns of before and aer charging to 4.5 V are shown in Fig. 3a. Synchrotron XRD data were collected at the beamline BL16B2 in SPring-8 synchrotron facility in Japan. 16 The measurement was conducted using an automatic powder diffraction system for Debye-Scherrer geometry using a capillary sample. The wavelength of X-rays was calibrated to be 0.500Å with the CeO 2 reference. A similar XRD pattern to our observation has also been reported in a publication by Sun et al. 11 The increase in the peak intensity of the 101 diffraction line has also been reported for Li x Ni 0.5 Mn 0.5 O 2 , 17,18 and such a change originates from Ni(Li) ion occupation at tetrahedral sites (6c sites). In Na insertion layered materials, Fe migration from metal oxide slabs to face-shared tetrahedral sites in Na layers has also been reported. 19 Therefore, the XRD patterns of NiO 2 are simulated with different cation distributions (Fig. 3b). In the simulation, Ni ion migration from the original octahedral sites to face-shared tetrahedral sites in the Li layers is assumed. The intensity of the 101 and 110 lines is intensied with an increase in the enrichment of Ni ions at tetrahedral sites and less occupation at original octahedral sites. The simulated patterns in Fig. 3b clearly show that a non-negligible amount of Ni ions is occupied at tetrahedral sites in Li layers of NiO 2 . When 15% of Ni ions migrate to the tetrahedral 6c sites, the intensity of the 101 diffraction line becomes higher than both 003 and 104 diffraction lines. However, peak broadening is also observed for 10l diffraction lines in the experimentally observed diffraction pattern of NiO 2 . This observation is indicative of the presence of stacking faults, and O1 domains are partially formed in O3 NiO 2 . 7 The peak broadening associated with stacking faults results in the difficulty to estimate the detailed cation distribution in NiO 2 , but Ni ion migration to the tetrahedral sites is clearly non-negligible as shown in Fig. 3. To further examine possibility of the O1 phase transition for NiO 2 , voltage holding is applied for 100 h at 4.5 V. However, a similar phase without voltage holding is observed as shown in ESI Fig. S4. † In addition, two new peaks at 4.07 and 8.13 appear as shown in the enlarged pattern of Fig. 3a. These two peaks are successfully assigned to 002 and 004 diffraction lines, which would originate from the presence of additional through-plane Ni ion ordering at the tetrahedral sites. The structural model of cation ordering at tetrahedral sites has not been solved yet. From these results, it is concluded that the O1 phase transition is hindered by the Ni ions at tetrahedral sites in Li layers, and thus the O3 phase is retained in NiO 2 .
Note that such Ni ion migration is a reversible process. Aer reduction to 2.5 V in a Li cell, Ni ions migrate back to the original octahedral sites. However, changes in the peak intensity for 101 and 104 diffraction lines upon electrochemical cycling are noted in LiNiO 2 aer 20 cycles as shown in Fig. 3a. This nding suggests that the reversibility of Ni ion migration is gradually lost upon cycling, and Ni ions are in part le at tetrahedral sites in Li layers. The results of Rietveld analysis of synchrotron XRD patterns of the as-prepared LiNiO 2 and the samples aer the initial cycle and 20 cycles are shown in ESI Fig. S5 † and 4. Rietveld renement was carried out using RIETAN-FP soware. 21 A good tting result is obtained for asprepared LiNiO 2 without the consideration of Ni ion occupation in the Li layer. Aer the initial cycle, a similar result is obtained without Ni ions at 6c tetrahedral sites, indicating that Ni ion migration is a reversible process. However, the intensity of the 101 diffraction line is obviously increased aer 20 cycles, and an insufficient tting result is obtained when Ni ion occupation at the tetrahedral 6c sites is not considered (Fig. 4a). The tting result is signicantly improved with the assumption of Ni ion occupation at tetrahedral sites (7%) in the Li layers as shown in Fig. 4b. In the analysis, the chemical composition is hypothesized as Li 0.82 NiO 2 , which is the same as that of the sample aer the initial cycle because similar voltage proles are obtained for <4.1 V region for both the initial and 20 th cycles. Li ions are assumed to be kinetically trapped in the structure, leading to the gradual loss of reversible capacity in the high voltage region, associated with the enrichment of Ni ions at the tetrahedral sites. Indeed, the gradual increase in impedance upon electrochemical cycling in the fully charged state has been found as shown in ESI Fig. S1d. † The gradual loss of Ni migration reversibility is expected to originate from the kinetic limitation. Li migration is also impeded by Ni ions in the Li layer, leading to the sluggish electrode kinetics. Note that such unfavorable Ni ion migration to tetrahedral sites is effectively suppressed by substitution of Ni ions by other metal ions, leading to the improvement of electrode reversibility in the high-voltage region, which will be reported in detail elsewhere. Although stoichiometric layered materials, like LiNiO 2 derivatives, show excellent capacity retention as electrode materials, 22,23 now the experimentally available reversible capacity is approaching its theoretical limit. To further increase reversible capacities, Li-enrichment in the host structure is necessary. [24][25][26][27] A comparative study of degradation mechanisms for the stoichiometric and Li-enriched electrode materials leads to better understanding to design durable and high-energy positive electrode materials in the future.

Conflicts of interest
There are no conicts to declare.