Tungsten Boride Stabilized Single‐Crystal LiNi0.83Co0.07Mn0.1O2 Cathode for High Energy Density Lithium‐Ion Batteries: Performance and Mechanisms

Transition metal doped LiNiO2 layered compounds have attracted significant interest as cathode materials for lithium‐ion batteries (LIBs) in recent years due to their high energy density. However, a critical issue of LiNiO2‐based cathodes is caused particularly at highly delithiated state by irreversible phase transition, initiation/propagation of cracks, and extensive reactions with electrolyte. Herein, a tungsten boride (WB)‐doped single‐crystalline LiNi0.83Co0.07Mn0.1O2 (SNCM) cathode is reported that affectively addresses these drawbacks. In situ/ex situ microscopic and spectroscopic evidence that B3+ enters the bulk of the SNCM, enlarging the interlayer spacing, thus facilitating Li+ diffusion, while W3+ forms an amorphous surface layer consisting of LixWyOz (LWO) and LixByOz (LBO), which aids the construction of a robust cathode‐electrolyte interphase (CEI) film, are shown. It is also shown that WB doping is effective in controlling the degree of the c‐axis contraction and release of oxygen‐containing gases at high voltages. The best doping concentration of WB is 0.6 wt.%, at which the capacity retention rate of the SNCM reaches 93.2% after 200 cycles at 2.7–4.3 V, while the morphology and structure of the material remain largely unchanged. The presented modification strategy offers a new way for the design of new stable SNCM cathodes for high‐energy‐density LIBs.


Introduction
The realization of a sustainable and decarbonized future relies critically upon the use of carbon-free energy such as renewables and electric vehicles (EVs) charged by renewable electricity. However, a widespread deployment of renewable energy and therefore EVs into the energy market requires efficient and low-cost energy storage (ES) technologies to compensate the renewables, power generation intermittency. Among all types of ES technologies available, the fastresponse, high-energy-density and modular electrochemical batteries stand out to be one of the best options to meet the grand ES challenge. LIBs, viewed as a benchmark battery technology, are currently being used in EVs and have been considered for largescale renewable ES applications. Unfortunately, the low energy density, limited Li and Co resources, and high cost along with safety concern have considerably constrained the applications of LIBs in the large-scale clean energy sectors. Increasing the energy density and using widely available materials have been deemed high-priority research for LIBs in recent decades. [1] LiMO 2 (MCo, Mn, and Ni) layered oxides have been the cathodes of choice for the modern LIBs technology for several decades. This is because the 2D layered structure is amenable to Li-ion intercalation/storage, while the transition metal cation can vary its oxidation states to maintain the charge neutrality. LiCoO 2 is the first cathode material used in commercial LIBs. [2] However, the concerns over the cost and limited resource of Co have triggered new interest in search for alternative layered LiMO 2 cathodes in recent years. One promising class of low-Co LiMO 2 is Ni-rich alternative, LiNi x Co y Mn 1-x-y O 2 (x ≥ 0.8, denoted as NCM), which features higher working voltage and energy density, and lower cost than LiCoO 2 . [3] Currently, most commercial NCM cathodes are made of polycrystalline powders composed of agglomerated nanoparticles. Although polycrystalline NCM powders allow for the penetration of the liquid electrolyte into its internal structure, which boosts the numbers of reactive sites, they are prone to significant capacity fading and safety issue during longterm cycling. This is because, upon deep delithiation, cracks are easily generated within grains due to the substantial anisotropic volume change, provoking parasitic interphase reactions between the electrolyte and NCMs, and destabilizing the CEI layer, [4] all of which will lead to battery's capacity decay. At higher Ni-contents, e.g., x > 0.6, the capacity fading becomes more pronounced and faster due to the extensive intergranular cracks. [5] Compared with polycrystalline NCMs, single-crystal NCMs (SNCMs) with monocrystalline primary particles have been shown as a potential solution to address the issues facing polycrystalline NCM. [4b,6] However, a drawback for SNCM is the prolonged Li + diffusion pathways, causing uneven distribution of Li + , [7] and ultimately resulting in cracks inside micron-sized primary particles after long-term cycling at high voltage or high current density. More importantly, when cycling at a high delithiated state, NCM experiences an irreversible phase transition from the second hexagonal structure (H2) to the third hexagonal structure (H3), causing a sudden contraction of the lattice and creation of nanocracks. [7b] As in the polycrystalline NCMs, once the cracks are initiated, capacity decay will happen.
To tackle the above problems, elemental doping and surface modification are often employed in SNCM. [ 3 coating, which maintains the structure integrity during cycling. [11] Overall, it remains a critical challenge to further improve the structural stability of SNCM to the level of commercial applications by simple surface modifications without affecting the transport of Li + .
In this work, we report on WB as a performance stabilizer of single-crystal LiNi 0.83 Co 0.07 Mn 0.1 O 2 . In one study in the literature, B was reported to occupy the tetrahedral interstitial site of packed oxygen lattice as a dopant, which forms stronger BO covalent bond, blocking the migration channels of transition metal ions and thus stabilizing the structure of the crystal. [12] In another independent study, W was added to NCM, in which W was found to react with NCM to form LWO layer on the surface of polycrystalline NCM particles during heat treatment; the former was also found to help enhance mechanical strength. [13] However, there is no related work in the literature on simultaneously using W and B as an additive to improve the stability of SNCM and understanding the dual stabilization mechanisms. The objective of this work is to fill this knowledge gap by using WB as an additive to stabilize SNCM and concentrating on unravelling the fundamental dual stabilization mechanisms.

Phase Compositions
The overall synthesis process of the SNCM is schematically illustrated in Figure 1a. The actual chemical compositions of the as-prepared all SNCM samples as determined by ICP-MS analysis (see Table S1, Supporting Information) agrees well with the expected values. The phase compositions of pristine and WB-modified SNCM samples are shown in Figure 1b of XRD patterns. All samples exhibit a rhombohedral structure (S.G:R3m) without any detectable impurity within the resolution of our XRD, indicating that a small amount of WB either dissolves into SNCM or forms amorphous phase. As the amount of WB increases, the (003) peak gradually shifts to smaller angle, implying the enlargement of the interlayer spacing. However, the (003) peak shift seems to reach a maximum at 0.6 wt.% WB (sample WB0.6-SNCM). The Rietveld refinement shown in Figure 1c and Figure S1 (Supporting Information), yields the lattice parameters of these materials, see Table S2-S5 (Supporting Information). It is worth noting that the c value increases with WB and reaches the peak of 14.2256 Å at 0.6 wt.% WB; the latter is consistent with the maximum shift of (003) peak shown in Figure 1b. In addition, the degree of Li + /Ni 2+ cation mixing is found to be 1.13% for WB0.6-SNCM, which is much <2.50% of the Bare-SNCM (Table S6, Supporting Information). This reduction in Li + /Ni 2+ cation mixing implies that there are more Li + active sites available for delithiation/lithiation, thus improving the electrochemical performance.
To further investigate the states of B and W in the WB-doped NCM, we also synthesized samples with B and W individually doped SNCM for comparison; the XRD patterns of these samples are shown in Figure S2 (Supporting Information). For the 5 wt. %B doped sample, the (003) peak shifts to lower angles, indicating that B can be dissolved into the lattice. However, for the 5 wt. %W doped sample, the (003) peak position remains unchanged compared to the undoped sample, except that there appears an amorphous hump at ≈20°. To further confirm the composition of the amorphous hump, 10 wt.% WB doped SNCM with Li-excess was analyzed by XRD. The 10 wt.% WB doped SNCM also has a layered structure, and the (003) peak not only shifts toward lower angle, but also the amorphous hump appears to be Li 6 WO 6 /Li 4 WO 5 accompanied with a weak Li 3 BO 3 / LiBO 2 peak. These observations further confirm B-doping into O-site and the existence of OLWO and LBO compounds. [13b,14] The possibility of WB reacting with LiOH is independently verified by Figure S3 (Supporting Information). The XRD results indicate that WB reacts with LiOH, possibly via the following reaction (not mass balanced for illustration purpose): Our XPS analysis further confirms the XRD findings. Compared to the W 4f and B 1s spectra in Bare-SNCM, see Figure 1d,e, the characteristic peaks belonging to W and B appear in the spectra of WB0.6-SNCM. The peaks of Li 2 CO 3 / LiOH and OM (lattice oxygen) located at 531.5 and 528.7 eV, respectively, were also detected in Figure 1f for Bare-SNCM and WB0.6-SNCM samples. The peak of OM in WB0.6-SNCM is significantly weaker than Bare-SNCM, which is likely caused by the blocking effect of LWO-LBO layer on the surface.
To support the above experimental result and gain fundamental insights, we also performed density functional theory (DFT) calculations. We assume that W 6+ (0.60 Å) substitutes into M-sites (MNi, Co or Mn), while B 3+ (0.27 Å) occupying the tetrahedral interstice of the oxygen lattice. The reactions are assumed to take place on the surface and in the bulk; the doping configurations and the results are shown in Figure S4 Figure 1g, respectively. Note that B 3+ has the lowest formation energy (−1.76 eV) on the M-site in the bulk, while W 6+ has the lowest formation energy (−1.16 eV) on the surface. These calculation results are in agreement with our early experimental observations, i.e., B 3+ entering into the bulk and enlarging the interlayer spacing, while W 6+ prone to gather on the surface. To investigate the effect of B bulk doping and W surface reaction on the structure, 3D and 2D charge density difference distribution of WB0.6-SNCM is calculated; the results are shown in Figure 1h and Figure S5 in the Supporting Information. For comparison, the charge differential densities without B in the bulk ( Figure S6a, Supporting Information) and without W at the surface ( Figure S6b, Supporting Information) are also calculated, respectively. On one hand, after B is doped into the bulk, more electrons are gathered around B, indicating that stronger BO bond is formed, which is beneficial to the improvement of the internal structure stability. On the other hand, compared with no W-doping, the charge accumulation on the whole surface after W-doping is significantly improved, indicating that the stability of the surface is improved. The charge accumulation around W is more than that of Ni, forming a stronger WO bond than NiO, which further strengthens the interface. To further examine the distribution of B and W inside the WB0.6-SNCM, depth-profiling XPS was performed; the results are shown in Figure S7 (Supporting Information). With increasing depth, the peak intensity of B is well maintained, while the peak intensity of W shows a drastic decrease. Hence, we speculate that B has been successfully doped into the bulk, while W is more likely enriched on the surface of WB0.6-SNCM; this assessment is consistent with the DFT calculations.

Microstructures
The particle morphologies of Bare-SNCM and WB0.6-SNCM are shown in Figure 2a,b as SEM images. For reference, the particle morphologies and elemental distribution of the precursor mixed with WB after ball milling are shown in Figure S8 in the Supporting Information. Both SNCMs are composed of smooth and clear-edge particles with mono-modal distribution of 2 µm in size; no grain boundaries are observed within each particle. There is virtually no difference in morphology between the Bare-SNCM and WB0.6-SNCM. The particle morphologies and structures are further analyzed by TEM and HRTEM; the results are shown in Figure 2c-e. The inset of Figure 2c is the pattern derived from the selected area electron diffraction (SAED), revealing a characteristic of single crystal diffraction pattern, and proving that the synthesized material contains fully and well crystallized particles. From the diffraction pattern, it is determined that the particle of the Bare-SNCM has an exposing (003) surface with a lattice spacing of 0.472 nm in S.G. = R3m, which is further confirmed by Figure 2d of the fast Fourier transformation (FFT). For the WB0.6-SNCM sample, Figure 2e reveals an ≈3 nm thick, amorphous layer on the surface of the particle, whereas the bulk of the particle exhibits the (101) surface of R3m structure with a lattice spacing of 0.245 nm. This observation confirms that the structure of the SNCM crystal was not changed by WB doping, which is consistent with the XRD result. From the high-angle annular dark-field imaging (HAADF) results shown in Figure S9   all uniformly distributed inside the WB0.6-SNCM, More importantly, from Figure 2g and Figure S10 in the Supporting Information of cross-sectional EPMA results, it provides additional evidence that most of W and some of B are concentrated on the surface of the sample, suggesting the outer amorphous layer shown in Figure 2e might be B-and W-rich, likely LWO and LBO as mentioned above, while a small amount of B also goes into the bulk of the WB0.6-SNCM.

Phase compositions
The effect of WB on the electrochemical properties of SNCM is evaluated in coin cells. Figure S11, the Supporting Information, shows the initial charge/discharge curves of SNCM with different contents of WB. These curves are very close, implying that the LWO-LBO layer formed on the surface of SNCM does not significantly impact on the electrochemical reactions. The initial discharge specific capacities decrease slightly with WB content, i.e., 204.7, 203.1, 202.5, and 194.1 mAh g −1 for Bare-SNCM, WB0.3-SNCM, WB0.6-SNCM, and WB1.0-SNCM, respectively, with the corresponding initial coulombic efficiencies of 88.6, 89.3, 89.9, and 87.0%, respectively. What distinguishes among the samples is, however, the cycle stability. Figure S12 in the Supporting Information shows that WB0.6-SNCM has the most outstanding cycling stability and the highest capacity. Therefore, we selected WB0.6-SNCM for further study. Figure 3a,b compares the CV curves of Bare-SNCM and WB0.6-SNCM samples, both of which exhibit three pairs of redox peaks. It is obvious that the voltage difference (ΔV) of WB0.6-SNCM, 0.110 V, is <0.154 V of Bare-SNCM, illustrating that WB0.6-SNCM has a lower overpotential and higher reversibility of Li + insertion/extraction than Bare-SNCM. In addition, the discharge specific capacities at different current densities shown in Figure 3c indicate a similar capacity at 0.1 C but increased difference at higher C-rates for both samples. The  better rate performance of WB0.6-SNCM than Bare-SNCM suggests that the former has a lower cell resistance, which may benefit from the larger interlayer spacing and Li + -conduction friendly LWO-LBO layer. Notably, the rate performance of the sample with only B-doping is better than that of only W-doping, implying that the excellent rate performance of WB0.6-SNCM comes more from the B-doping, see Figure S13, Supporting Information.
To further understand the influence of WB doping on Li + diffusion kinetics, GITT measurements were conducted to measure the diffusion coefficient of lithium ions (D Li + ) in charging process; the results are shown in Figure 3d,e and Figure S14 (Supporting Information). The average D Li + of WB0.6-SNCM is 1.33 × 10 −10 cm 2 S −1 , >4.37 × 10 −11 cm 2 S −1 of Bare-SNCM. The improved Li + diffusion kinetics is again originated from the enlarged interlayer spacing by B-doping and Li + conducting LWO-LBO surface layer formed by W-doping. Figure 3e,f compares voltage profiles of Bare-SNCM and WB0.6-SNCM at different cycles. Note that the first three cycles were conducted at 0.1 C in this measurement for the activation purpose. It is evident that WB0.6-SNCM outperforms Bare-SNCM in capacity and retention rate over 100 cycles. The corresponding cycle stability comparison for Bare-SNCM and WB0.6-SNCM is shown in Figure 3g,h at two different C-rates. It is evident that Bare-SNCM has a faster capacity fading than WB0.6-SNCM., i.e., 168 mAh g −1 (over 200 cycles) versus 156 mAh g −1 (over 100 cycles) and 93.2% (over 200 cycles) versus 86.2% (over 100 cycles) at 1 C rate. For reference, the corresponding differential capacity (dQ/dV) curves in discharge are calculated and shown in Figure  S15 in the Supporting Information.
Given the fact that both samples are subject to phase transformation (to be shown in the following) from hexagonal H3 to H2 (H3→H2), to monoclinic (H2→M), and to hexagonal 1 (M→H1) during lithation, and vice versa during delithiation, the resultant structural instability will negatively impact the stability of the battery performance. [3a] However, from Figure 3g,h, it is evident that WB0.6-SNCM exhibits more stable performance than Bare-SNCM. The WB doping clearly play a key role in this improvement by synergistically facilitating Li + transport kinetics and maintaining structural stability.

X-ray Absorption Spectroscopy
X-ray absorption spectroscopy (XAS) was also employed to study the oxidation state of Ni in the two SNCM cathodes. Figure 4 and Figure S16 (Supporting Information) compare their XANES spectra of Ni K-edge under full-charged state at the 3 rd and 100 th cycle. As shown in Figure 4a, the edge position of K-edge in the WB0.6-SNCM sample shifts to higher energy compared to Bare-SNCM after the 3 rd cycle (also see inset of Figure 4a), suggesting that the oxidation state of Ni is elevated and more Ni 3+ /Ni 4+ is responsible for compensating more Li + (thus higher capacity). The raise of Ni oxidation state has been previously found to arise from the construction of robust and low-resistant CEI layer. [15] After 100 cycles, Figure 4b shows almost identical position and intensity of the K-edge for both samples. But the white line intensity of WB0.6-SNCM is still higher than that of the Bare-SNCM. Note that the spectra of the K-edge of WB0.6-SNCM at the 3 rd and 100 th cycle almost completely overlap, see inset of Figure 3b, which implies that WB0.6-SNCM has a better reversibility of Ni 3+ /Ni 4+ redox reaction during cycling process.
The Fourier transformed extended X-ray absorption fine structure (EXAFS) of the Ni-XAS has a better sensitivity to identify local chemical and structural change. Figure 4c,d show that the interatomic distances of NiO and NiM for Bare-SNCM after the 100 th cycle are shorter than after the 3 rd cycle, while there is no change in WB0.6-SNCM. The change in interatomic distances is due to the presence of unfilled Li vacancies during lithiation, thus leading to changes in the chemical state and atomic environment of Ni. In addition, Wavelet Transform (WT) analysis, a powerful technique to further denoise and parse the EXAFS data, is employed to identify the contributions of backscattered atoms, and study the local environment of Ni. From Figure S17 (Supporting Information), the positions and intensities of NiO and NiM in the Bare-SNCM are different after the 3rd and 100th cycles, where there are no discernable changes in the WB0.6-SNCM, between the 3 rd and 100 cycles, see Figure 4e,f, demonstrating stability and reusability of Ni in WB0.6-SNCM during cycling process. Considering the above characterization and analysis, we conclude that the Ni 3+ /Ni 4+ redox couple in WB0.6-SNCM has a better reversibility than WB0.6-SNCM during long-term cycling.

Post-Test Analysis
To investigate the reason for the apparent difference in cycling stability between Bare-SNCM and WB0.6-SNCM, we disassembled the coin cells and extracted cathode section after 100 cycles at 1 C for SEM examination.  Figure 5b, is shown with not only nanocracks, but also large cross-grain cracks. These cracks will lead to pulverization of the cathode and deterioration of the electrochemical performance. [16] In contrast, there is no visible intragranular cracks in the WB0.6-SNCM, see Figure 5d, which ensures the interface stability and stable cycle performance.
The surface chemical compositions of the cycled cathodes were also examined by XPS. The C1s spectra of Figure 5e can be divided into five peaks assignable to CC, CH, CO, CO and OCO 2 . The characteristic peaks of ROCO 2 Li, Li 2 CO 3 /LiOH, and OM shown in Figure 5f are detected in the O1s spectra. The intensity of the CO and CO peaks in C1s and the ROCO 2 Li peak located near 533 eV in O1s spectra are significantly stronger in Bare-SNCM than in WB0.6-SNCM, suggesting that more organic components from electrolyte decomposition is present on the surface of Bare-SNCM. In addition, it is clear from Figure 5g of F1s spectra that the concentration of LiF in Bare-SNCM is much higher than that in WB0. 6 than that of WB0.6-SNCM, which could imply a thicker CEI film formed on the surface of Bare-SNCM that weakens the signal of Ni. The overgrown CEI layer with large amount of LiF leads to increase of cell impedance, thereby hindering the Li + transport and reducing the electrochemical activity of the material. [17] The increased impedance is supported by the EIS results shown in Figure S20 and Table S7 (Supporting  Information).
The two peaks located in 854.7 and 856 eV in Figure 5h are related to Ni 2+ and Ni 3+ , respectively. It is important to note that the ratio of Ni 3+ /Ni 2+ in WB0.6-SNCM, 106.8%, is much >51.5% of Bare-SNCM. This result suggests that there are more Ni 2+ in the cycled Bare-SNCM than the cycled WB0.6-SNCM. The production of Ni 2+ may originate from the side reactions at the electrolyte/electrode interface, leading to the reduction of Ni 4+ to Ni 2+ . [18] The increased Ni 2+ can exacerbate the Li + /Ni 2+ cation mixing due to the close ionic radius between Ni 2+ (0.69 Å) and Li + (0.69 Å). [19] The severe Li + /Ni 2+ cation mixing in the Bare-SNCM induces a faster structural transformation on the particle surface (to be demonstrated by the following STEM analysis), which is a key reason for the performance deterioration of the Bare-SNCM.

In Situ Studies
In the early cycle stage of LIBs, especially in the first cycle, the formation of CEI film is often accompanied by the evolution of SNCM's structure and the underlying electrochemical reactions. Therefore, it will be valuable to understand what these changes are during the initial cycles by performing real-time in-situ measurements. We first performed in-situ XRD measurement to reveal the structural evolutions of Bare-SNCM and WB0.6-SNCM at the first lithiation/delithiation cycle. During the first cycle, Figure S18a,b (Supporting Information) show that both Bare-SNCM and WB0.6-SNCM exhibit the characteristic diffraction peaks of layered structures, e.g., (003), (101), (006), (102), and (104), and shifts consistent with the stateof-charge and return to original positions after discharging. Specifically, the diffraction peaks of WB0.6-SNCM almost completely return to the original positions after discharging, while Bare-SNCM has a visible shift. This difference indicates that WB0.6-SNCM has a better reversibility of the phase transition than Bare-SNCM. Focusing on (003) peak, Figure 6a-c,e-g show its shift under different state-of-charge. From the opencircuit voltage to near 4.1 V, it gradually shifts to a lower angle accompanied by a continuous phase transition from first hexagonal phase (H1) to monolithic (M), and then to H2 phase, accompanied by a continuous expansion of the c-axis due to the electrostatic repulsion of the adjacent oxygen layers. As the delithiation continues beyond 4.1 V, a sudden and quick shift of (003) peak to a higher angle is observed; this significant change is due to the H2→H3 transition accompanied by a significant lattice contraction in the c-axis direction. [9,20] The (003) peak shifts to the highest angle at the end of the transition, i.e., 4.3 V. During the discharge, the phase undergoes a reverse transition to the charge process. A comparison of the maximum angle shift of Bare-SNCM and WB0.6-SNCM during the irreversible H2-H3 phase transition, see Figure 6b,f, finds less shift for WB0.6-SNCM than Bare-SNCM, i.e., 0.47° versus 0.62°. We also observe in Figure 6d that the variation (Δc) of the c-axis lattice parameter during delithiation is strongly dependent on the shift of the (003) peak. Compared to Δc = 3.2% of Bare-SNCM, WB0.6-SNCM exhibits smaller c-value change, Δc = 2.6%, meaning less volume contraction and thus better structural stability.
We then conducted in situ DEMS to monitor the gas evolution of O 2 and CO 2 during the first charge process; the results are shown in Figure 6i. The gas production can be divided into four stages associated with the phase transitions from H1 to M, M to H2, and H2 to H3 phase. For both samples, only minor amounts of O 2 and CO 2 are observed in the stage 1 (3.5 -3.7 V). During stage 2 (3.7 -3.9 V), WB0.6-SNCM shows a change at the beginning in O 2 and CO 2 evolution, followed by a plateau, whereas Bare-SNCM shows a continuous increase. Upon entering stage 3 (3.9 -4.1 V), WB0.6-SNCM first shows a sudden decrease in O 2 , then increase, while there is no change in CO 2 evolution. In contrast, Bare-SNCM shows no significant change in O 2 and CO 2 during this stge. At > 4.1 V (stage 4), there is a decrease in O 2 and no change in CO 2 for WB0.6-SNCM, whereas there is no change in O 2 , but significant increase in CO 2 for Bare-SNCM. Since the structure of Ni-rich cathodes becomes unstable at a highly delithiated state, Ni 4+ tends to form electrochemically inert Ni 2+ and reconstructed layers such as spinel and rock-salt phases to stabilize the structure, a process that cause decomposition and O 2 evolution. [21] The formation of CO 2 is attributed to the reaction of the active oxygencontaining species with the organic electrolyte. [22] The suppression of gas evolution observed in the WB0.6-SNCM is mainly resulted from a more stable internal structure and robust interface protected by an LWO-LBO hybrid layer.
All the above in-situ results support that the WB-doped SNCM can effectively suppress the deleterious H2-H3 phase transition with smaller volume change. The oxygen vacancies created by O 2 evolution on the surface and in the bulk after cycling were also assessed by EPR. Figure 6i shows a g-factor of 2.0 that proves the existence of oxygen vacancies. [23] There are significantly less oxygen vacancies in WB0.6-SNCM than in Bare-SNCM, which is consistent with less O 2 and CO 2 evolution as revealed by in-situ DEMS and reversible structural evolution by in situ XRD.

Microstructural Evolution with State-of-the-Charge
To further study the mechanism of WB0.6-SNCM with enhanced cycling performance, we also studied morphological and structural evolutions of Bare-SNCM and WB0.6-SNCM at surface and in bulk after 100 cycles using scanning transmission electron microscopy (STEM). To observe the cross section, the particles were milled-out by a focused ion beam (FIB), see Figure S19 (Supporting Information). For the Bare-SNCM, Figure 7a clearly shows intra-grain nanocracks. In contrast, Figure 7f shows that the interior of WB0.6-SNCM still maintains a crack-free intact bulk structure as that before the cycle. Furthermore, Figure 7b-d show a HAADF STEM analysis on the areas around the cracks from the near surface to interior bulk of the Bare-SNCM sample. Based on the results of in-situ XRD analysis, Bare-SNCM suffers from a more severe H2-H3 phase transition accompanied by a drastic anisotropic volume change, which causes it to be more vulnerable to cracking and irreversible rock-salt phase. Only the Fm3m rock-salt phase is detected in region 1, see Figure 7b, indicating that serious phase transition has occurred around the nanocracks. In region 2, Figure 7c, near region 1, layered structures with slight disorder are observed. In region 3, Figure 7d, a region far away from region 1, the initial ordered layered hexagonal (R3m) structure is detected, which gradually transforms into disordered layered structure. Additionally, a continuous irreversible phase transformation from ordered layered phase to disordered rock-salt phase is revealed in region 4 in Figure 7e, which is located at the subsurface area of Bare-SNCM. The Li layer and transition metal layer can be clearly identified inside the bulk, see the left part in Figure 7e. However, defective disordered layered structures gradually appear with the structure degradation, see the middle part in Figure 7e. It finally transforms into disordered Fm3m rock-salt structure on the surface of Bare-SNCM, see the right part in Figure 7e.
The irreversible deterioration of the structure of Bare-SNCM around the nanocracks and near the surface is likely caused by the parasitic interface reactions. In the long-term cycling, the decomposition products of the electrolyte will constantly erode the cathode interface from the outside in, resulting in the transformation of the crystal structure on the surface, which will destroy the mechanical stability and structural integrity and eventually lead to the formation of nanocracks. On the contrary, the structural damage in WB0.6-SNCM is effectively inhibited by WB-modification. The layered structure of WB0.6-SNCM is well maintained in the inner bulk as well as outer surface, see Figure 7g-i, accompanied by a slightly disordered layer on the outermost surface, which demonstrates WB0.6-SNCM as a structurally more stable cathode for long-term cycle.
A summary schematic of the structural evolutions of Bare-SNCM and WB0.6-SNCM after long cycling is shown in Figure 7j. During repeated delithiation/lithiation in Bare-SNCM, the accumulation of anisotropic lattice strain due to the volume change associated with the H2-H3 phase transition initiates cracks, which is accompanied by the generation of oxygen vacancies and release of oxygen-containing species. The volume changing phase transitions lead to structural instability and formation of disordered rock-salt phases. During prolonged cycles, the cracks will propagate throughout the cathode, allowing the electrolyte to continuously erode the exposed electrode surface along the cracks. Consequently, the parasitic cathodic/electrolyte reactions will lead to surface reconstruction with an overgrown CEI film at the interface and formation of detrimental by-products, which will in turn generate more electrochemically inert rock-salt phases and cracks, resulting in deterioration of the electrochemical performance. In contrast, the morphological and structural stability of WB0.6-SNCM enabled by the LWO-LBO surface layer ensures a robust cathode/electrolyte interface and stable CEI film, thus stable performance.

Conclusion
In summary, WB-doped single-crystalline LiNi 0.83 Co 0.07 Mn 0.1 O 2 (SNCM) cathode has been synthesized by a simple ball milling and solid-state reaction method. Both experimental and theoretical data support that B enters into the bulk, enlarging the interlayer spacing, while W gathers at the surface forming an LWO/LBO layer to protect the cathode. Among all WB contents studied, the 0.6 wt.% WB doped SNCM exhibits the best rate capacity, cycle retention and structural stability. It delivers a discharge capacity of 164 mAh g −1 at 5C, while maintaining a superior capacity retention rate of 92.3% under 1C for 200 cycles in the voltage range of 2.7-4.3 V. The excellent performance is attributed to the following reasons: i) B-doping enlarges the interlayer spacing and creates a stronger BO bond to achieve a more stable structure, all of which will suppress volume changing H2 to H3 phase transition, reduce oxygen vacancies, lowers Li + /Ni 2+ cation mixing, and improves the Li + diffusion coefficient. ii) The amorphous hybrid LWO-LBO surface layer favors the formation of a thinner CEI layer, resulting in a robust electrode and electrolyte interface with suppressed parasitic reactions and the initiation of cracks. Materials Characterization: The crystal structure of pristine and WB-added SNCMs were examined by X-ray diffraction (XRD) using Bruker D8 Advance with Cu Kα radiation (λ = 1.5418 Å) in a 2θ range of 10 to 80°. The collected XRD data was refined by General Structure Analysis System (GSAS, USA). In situ XRD was performed in a specialized in situ battery cell holder (Beijing Scistar Technology Co. Ltd, China); it is consisted of a φ12 mm inner diameter electrochemical cell with a Be window for X-ray transmission. Morphology and microstructure of the samples were captured by a field emission scanning electron microscopy (FESEM, Hitachi SU8010) and high-resolution transmission electron microscopy (HRTEM, JEM-2100F). The elemental distribution in SNCM was analyzed by electron-probe (EPMA, JXA-8530F PLUS). Elemental oxidation state and composition were determined by an X-ray photoelectron spectrometer (XPS, Thermo Fisher Scientific K-Alpha) with an Al-Kα radiation (1486.6 eV). Inductively coupled plasma mass spectrometry (ICP-MS, Agilent 7700) was applied to analyze the concentration of each element in SNCM. To analyze the Ni oxidation state in pristine and WB-modified SNCM samples under fully charged state after different cycles, X-ray absorption near-edge structure (XANES) measurement was performed at 21A X-ray nanodiffraction beamline of Taiwan Photon Source (TPS) at National Synchrotron Radiation Research Center (NSRRC). The concentrations of oxygen vacancies in pristine and modified SNCM samples were determined by electron spin resonance spectra (ESR, Bruker EMXPLUS).

Experimental Section
Electrochemical Measurements: The coin-type CR2032 cells were used to evaluate the electrochemical performance of the prepared SNCMs. To prepare the cathode, active SNCM material (90 wt.%), conductive agent (carbon black, 5 wt.%) and binder (polyvinylidene fluoride, PVDF, 5 wt.%) are dispersed in the solvent (N-methyl-2-pyrrolidone, NMP) to form a homogeneous slurry. The slurry was then coated onto an aluminum foil and dried in a vacuum oven at 90 °C for 6 h. Then, the obtained cathode assembly was cut into disk with a diameter of 12 mm. The average mass loading of active material is ≈2.0 mg cm −2 . The CR2032 cells were assembled in an Ar-filled glove box (H 2 O<0.1 ppm, O 2 <0.1 ppm) with Li metal foil as counter electrode, a polypropylene film (Celgard 2400) as separator saturated with 80 µL liquid electrolyte consisted of 1 M LiPF 6 in ethylene carbonate (EC)/dimethyl carbonate (DMC) with EC:DMC volume ratio = 1:1.
Galvanostatic charge/discharge tests were conducted using the Neware battery test system (CT-4008, Shenzhen, China) within a voltage of 2.7-4.3 V. For this study, 1.0 C is equivalent to 200 mA g −1 . Cyclic voltammetry (CV) with a sweep rate of 0.1 mV s −1 from 2.7 to 4.3 V and electrochemical impedance spectroscopy (EIS) with a 5-mV perturbation and 100 kHz to 0.01 Hz frequency range were carried out by an electrochemical workstation (CHI 660E, Chenhua, China). Galvanostatic intermittent titration techniques (GITT) were carried as follows. First, the cell was charged at 0.1C for 10 min followed by relaxed for 50 min. Then, the above process was repeated until the voltage reaches at the cut-off voltage.
The in situ differential electrochemical mass spectrometry (DEMS, HPR-40 Hiden, Germany) was studied using a three-electrode cell configuration with SNCM as the working electrode and Li metal as the counter and reference electrodes.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.