Li2TiO3 Dopant and Phosphate Coating Improve the Electrochemical Performance of LiCoO2 at 3.0–4.6 V

A sol–gel tandem with a solid-phase modification procedure was developed to synthesize Li2TiO3-doped LiCoO2 together with phosphate coatings (denoted as LCO-Ti/P), which possesses excellent high-voltage performance in the range of 3.0–4.6 V. The characterizations of X-ray diffraction, high-resolution transmission electron microscopy, and X-ray photoelectron spectroscopy illustrated that the modified sample LCO-Ti/P had the dopant of monoclinic Li2TiO3 and amorphous Li3PO4 coating layers. LCO-Ti/P has an initial discharge capacity of 211.6 mAh/g at 0.1 C and a retention of 85.7% after 100 cycles at 1 C and 25 ± 1 °C between 3.0 and 4.6 V. Nyquist plots reflect that the charge transfer resistance of LCO-Ti/P after 100 cycles at 1 C is much lower than that of the spent LCO, which benefits Li-ion diffusion. Density functional theory calculations disclose the superior lattice-matching property of major crystal planes for Li2TiO3 and LiCoO2, the lower energy barriers for Li-ion diffusion in Li2TiO3, and the suppressed oxygen release performance resulting from phosphate adsorption. This work provides useful guidance on the rational design of the high-voltage performance of modified LiCoO2 materials in terms of lattice-matching properties aside from the phosphate coating to reduce the energy barriers of Li-ion diffusion and enhance cycling stability.


Introduction
LiCoO 2 has been the dominant cathode material for lithiumion batteries in portable electronic products because of its high volumetric energy density [1,2]. Increasing the operating voltage of LiCoO 2 can effectively enhance its specific capacity; however, achieving stable cycling performance of LiCoO 2 at high voltage faces many challenges [3], including the adverse phase transition at an operating voltage above 4.5 V (vs. Li/Li + ) [4,5], the lattice distortion caused by internal mechanical strain, the loss of oxygen, dissolution of cobalt and side reactions with the electrolyte [6,7].
Moreover, chemically inert phosphate can inhibit the undesired reactions between active materials and electrolytes and suppress the growth of the resistance to Li-ion diffusion [29,30]. For example, Wang et al. [31] constructed a composite surface on LiCoO 2 consisting of amorphous coating layers of AlPO 4 and Li 3 PO 4 as well as Al and P doping, and the modified sample showed an initial discharge capacity of 214.1 mAh/g (27.4 mA/g) and a retention of 88.6% after 200 cycles under a current density of 137 mA/g and RT in the range of 3.0-4.6 V. Gu et al. [32] used a spraydrying method of NH 4 ·H 2 PO 4 to synthesize a phosphaterich coating layer on LiCoO 2 , reporting an initial capacity of 179.3 mAh/g (0.2 C) and a retention of 80.9% after 500 cycles at 1 C (140 mA/g) and RT in the range of 3.0-4.5 V. Hence, we are enlightened to explore an effective method for improving the high-voltage performance of LiCoO 2 via the Li 2 TiO 3 dopant as well as a phosphate coating layer.
In this article, using the precursors of Ti(OC 4 H 9 ) 4 and NaH 2 PO 2 , a sol-gel tandem with a solid-phase modification procedure was developed to synthesize the modified LiCoO 2 (denoted as LCO-Ti/P), which possesses excellent highvoltage performance in the range of 3.0-4.6 V. Based on the characterizations of X-ray diffraction (XRD), highresolution transmission electron microscopy (HRTEM), X-ray photoelectron spectroscopy (XPS), in situ XRD, etc., illustrated that the modified sample LCO-Ti/P has a multiphase interface consisting of monoclinic Li 2 TiO 3 , spinel Co 3 O 4 , and amorphous Li 3 PO 4 coating layers, leading to an initial discharge capacity of 211.6mAh/g (0.1 C) and a retention of 85.7% after 100 cycles at 1 C and RT, which is far superior to the individual-element modified samples LCO-Ti and LCO-P. This work provides useful guidance on the rational design of the high-voltage performance of LiCoO 2 with crystalline phases that match properly with the bulky, layered structure at the interface.

Material Synthesis
Pristine LiCoO 2 (LCO) was provided by Kejing Materials Technology. Three types of modified samples were prepared, including the individual-element modified samples of LCO-Ti and LCO-P and the dual-element modified LCO-Ti/P. To prepare the sample with the modification of Ti (LCO-Ti), 4.0 g LiCoO 2 was added to 100 mL ethanol solution containing 0.4 mL tetrabutyl titanate (Ti(OC 4 H 9 ) 4 ) (98.0%, Aladdin, Shanghai) and 0.4 mL acetic acid (36.0 wt%) under vigorous stirring at RT. After stirring for 12 h, the mixture was evaporated at 80 ℃, and then, the obtained powder was ground thoroughly, followed by calcination at 800 ℃ for 1 h at a heating rate of 5 ℃/min under the atmosphere. The final product was named LCO-Ti.
To prepare the sample with the modification of phosphorus (LCO-P), 4.0 g LiCoO 2 was mixed with 0.024 g NaH 2 PO 2 (99.0%, Aladdin, Shanghai), i.e., the NaH 2 PO 2 additive accounts for 0.6 wt%, ground in a mortar in the presence of a small amount of ethanol for 10 min, and then dried and heated at 200 ℃ for 1 h. The obtained modified sample was denoted as LCO-P.
The dual-element modified sample LCO-Ti/P was prepared based on the Ti-modified sample LCO-Ti. A 1.0 g LCO-Ti sample was mixed with 0.006 g NaH 2 PO 2 (the mass fraction of NaH 2 PO 2 is 0.6 wt%) and ground in the presence of a small amount of ethanol for 10 min. Then, the mixture was dried and heated at 200 ℃ for 1 h. The obtained sample was named as LCO-Ti/P.

Material Characterization
X-ray diffraction (XRD) (Rigaku Smartlab) was used to determine the crystal structures of the samples using Cu Kα radiation in a range of 15-70° with a scanning rate of 1°/min. General structure analysis system (GSAS) software was used for Rietveld refinement. X-ray photoelectron spectroscopy (XPS) (ThermoFisher Scientific K-Alpha + ) measurements were performed to detect the valence state of surface elements with the peak calibration by C 1 s (284.80 eV). Advantage software was used to fit the peaks of the XPS spectra. The surface morphology and microstructures of the samples were observed with scanning electron microscopy (SEM) (FEI Apreo S LoVac), high-resolution transmission electron microscopy (HRTEM) (JEOL JEM-F200), and energy-dispersive spectrometry (EDS). Fourier-transform infrared spectroscopy (FT-IR) measurements of prepared samples were performed using a Thermo Scientific Nicolet iZ10 spectrometer. The sample materials were well mixed with KBr for FT-IR experiments. Each sample was scanned in the wavenumber range of 400-4000 cm −1 . The in situ XRD patterns were collected with a scanning rate of 10°/ min in the range of 15-70° in the charging and discharging process at 0.2 C (1 C = 200 mA/g).

Electrochemical Characterization
The electrochemical performance of the cathode material was tested using half-cells (CR-2032) with a Li metal anode. The cathode electrodes were prepared by mixing the active material, polyvinylidene fluoride, and carbon black (8:1:1 by weight) with N-methyl-1,2-pyrrolidone (NMP) solution. The solvent was evaporated at 120 °C overnight after casting the slurry on Al foil. The active mass loading of the cathode material was approximately 2.5-2.8 mg/cm 2 . The electrolyte consisted of 1 mol/L LiPF 6 dissolved in a mixture of EC, DEC, and EMC (1:1:1 by volume). Cycling and rate performance and galvanostatic intermittent titration technique (GITT) tests were conducted on a Land system (CT2001A) at 3.0-4.6 V and 25 ± 1 °C (RT). The half-cells were first activated for two cycles at 0.1 C, and then, the long-term cycles were performed at 1 C (1 C = 200 mA/g). Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) tests were implemented on an electrochemical workstation (PARSTAT 4000A). The EIS test was conducted at a fully charged state (4.6 V) over a frequency range of 100 kHz-0.01 Hz, and CV tests were cycled from 3.0 to 4.6 V at a scan rate of 0.1 mV/s.

First-Principles Calculations
Based on the PAW method, DFT (density functional theory) calculations were performed using the VASP package [33]. The exchange-correlation between electrons was described by the PBE functional under the GGA method [34]. Hubbard U parameters of 4.91 and 5.0 eV were employed to treat the strongly correlated d-orbital electrons of Co and Ti, respectively, referring to previous studies [35]. The electronic wave functions were expanded in a plane-wave basis set with an energy cutoff of 520 eV. To obtain a stable structure, the structural geometry was optimized until the average force on each atom was < 0.05 eV/Å. A resolution of 2π × 0.04 Å −1 for the Monkhorst-Pack k-point mesh was used to integrate the Brillouin zone [36]. The nudge elastic band method was used to evaluate Li-ion diffusion pathways and the diffusion energy barrier [37].
The binding energies between the crystal planes of LiCoO 2 and Li 2 TiO 3 were calculated using the following equation: where E(LCO-LTO) represents the total energy of the system of interfacial structural models, and E(LCO) and E(LTO) represent the energies of the different crystal planes of LiCoO 2 and Li 2 TiO 3 , respectively.
The PO 4 3− adsorption energy was calculated using the following formula: where E(LCO-PO 4 3− ) indicates the total energy of the LiCoO2 system with the adsorbed PO 4 3− , E(LCO) is the total energy of LiCoO2, and E(PO 4 3− ) represents the energy of PO 4 3− . The oxygen vacancy formation energy was defined as follows.
where E(LCO-O) represents the total energy of LiCoO 2 with one oxygen vacancy, E(LCO) is the total energy of LiCoO2, and E(O 2 ) indicates the energy of an oxygen molecule.

Results
XRD patterns were recorded to determine the crystal structures of pristine LCO and the modified sample (LCO-Ti/P) together with the control samples with single-element modification (LCO-P, LCO-Ti). As shown in Fig. 1a, the main structures of all the samples are classified as the α-NaFeO 2 -type layered structure (JCPDS: 50-0653) with an R-3 m space group. Notably, for LCO-Ti and LCO-Ti/P, small peaks appear at 31.27° and 36.85°, which are indexed to spinel Co 3 O 4 (JCPDS: 42-1467) with a space group of Fd-3 m, as well as the peak at 43.72°, which is indexed to monoclinic Li 2 TiO 3 (JCPDS: 33-0831) with a space group of C 2/c. The Rietveld refinement patterns of these samples are given in Fig. 1b, c, and Fig. S1, respectively, and the corresponding lattice parameters are listed in Table 1. The c/a values of all samples are > 4.9, indicating a well-ordered layered structure.
The morphologies of all samples were observed through SEM images. As shown in Fig. S2, the pristine LCO sample has a smooth and clean surface, whereas the surfaces of the modified samples (LCO-P, LCO-Ti, and LCO-Ti/P) are rough to some extent, suggesting that coating layers were formed on the surfaces.
HRTEM images were observed to disclose the phase compositions of LCO and LCO-Ti/P. As shown in Fig. 2a, pristine LCO exhibits a uniform crystal phase with a layered structure. The fast Fourier transform (FFT) pattern in Region 1 shows the (104), (110), and (1-4) planes of the layered phase (R-3 m), and the lattice spacing of 0.21 nm corresponds to the (104) plane. In Region 2, the FFT pattern shows the (012) planes of the layered phase (R-3 m), and the lattice spacing of 0.23 nm also corresponds to the (012) plane. However, LCO-Ti/P has a multiphase surface structure (Fig. 2b). The FFT pattern of the inner area of LCO-Ti/P (cyan dashed rectangle in Region 2) shows the layered phase (R-3 m) in the bulk area, and the lattice spacing of 0.21 nm corresponds to the (104) plane of the layered phase. In contrast, the yellow dashed rectangle in Region 2 corresponds to the monoclinic phase (C2/c) of Li 2 TiO 3 , the FFT pattern shows the (33-1), (− 206), and (135) planes, and the lattice spacing of 0.15 nm is near the interplanar distance of the (− 206) plane of monoclinic Li 2 TiO 3 . In Region 1, the diffraction pattern of the surface structure (green dashed rectangle) shows an identifiable spinel structural feature with an FFT pattern due to the (400), (040), and (440) planes, and the lattice spacing of 0.22 nm approximates the interplanar distance of the (400) plane of spinel Co 3 O 4 . These results agree with the above XRD results, confirming the presence of spinel Co 3 O 4 and monoclinic Li 2 TiO 3 in the modified LCO-Ti/P sample. In the area of the outmost surface of LCO-Ti/P (marked by the red dashed line in Region 2), an amorphous phase layer is observed. The amorphous phase is probably due to Li 3 PO 4 , according to the synthetic process, which is confirmed by XPS spectra in this context. EDS spectra of LCO-Ti/P were obtained to detect the uniformity of the Ti and P elements on the surface, as shown in Fig. S3, suggesting the uniform distribution of Ti and P elements in LCO-Ti/P. XPS spectra were recorded to distinguish the surface compositions and identify the valence states of elements. For the individual-element modified sample LCO-Ti, in the Ti 2p XPS spectra (Fig. 3a), the peaks at the binding energies of 458.28 eV and 464.08 eV correspond to Ti 4+ [38], which agrees with the above results of Li 2 TiO 3 reflected by XRD  and HRTEM analyses. For the dual-element modified sample LCO-Ti/P, Ti 2pXPS spectra have similar peak positions with those of LCO-Ti, in addition to the slight shifts of binding energies, which are probably caused by the influence of PO 4 3− . In the case of the P 2p XPS spectra, LCO-Ti/P and LCO-P have a peak at 132.78 eV (Fig. 3b), corresponding to the P−O bonding of PO 4 3− [39,40]. Figure 3c displays the O 1 s XPS spectra of the modified samples, exhibiting peaks at 533.17 and 532.00 eV assigned to C−O and C=O bonds, peaks near 529.10 and 529.50 eV assigned to the lattice oxygen, and a peak at 531.04 eV due to overlapping Ti−O (531.40 eV for LCO-Ti) and P−O (530.80 eV for LCO-P) peaks [41,42]. Table S2 lists the relative percentages of different oxygen species and the binding energies of these four samples. The O 1 s peaks of LCO-Ti have high intensities, probably due to the greater amount of hydrolyzed product from the Ti-precursor on the surface. As for Co 2p XPS spectra ( Fig. 3d and Table S3), these four samples have similar binding energies for Co 2+ 2p3/2 and Co 3+ 2p3/2 [43]. In combination with XRD and HRTEM results, this result suggests that the modified sample LCO-Ti/P has a Li 2 TiO 3 dopant as well as a Li 3 PO 4 coating layer.
The FT-IR spectra of LCO and LCO-Ti/P were recorded, as shown in Fig. S2. The bands at 550 and 600 cm −1 correspond to CoO 6 stretching and bending vibrations [44], the broad bands near 1630 cm −1 and 3400 cm −1 are assigned to adsorbed water molecules, and the bands near 1100 cm −1 correspond to the C=O vibration due to the adsorption of CO 2 [45]. Notably, additional characteristic peaks appear from 900 to 1100 cm −1 in the LCO-Ti/P spectrum, and the small absorption peaks at 1050, 1085, and 1180 cm −1 are due to asymmetric P−O stretching vibrations and asymmetrical stretching of P=O in PO 4 3− , confirming the formation of Li 3 PO 4 [46,47].
The electrochemical performances of the individualelement modified samples of LCO-Ti and LCO-P and the dual-element modified LCO-Ti/P were evaluated by galvanostatic tests in the range of 3.0-4.6 V. As shown in Fig. 4a, the initial discharge capacities of LCO, LCO-Ti, LCO-P, and LCO-Ti/P at RT and 0.1 C are 214.5, 212.3, 213.6, and 211.6 mAh/g, respectively. After 100 cycles at 1 C, LCO, LCO-Ti, LCO-P, and LCO-Ti/P show capacity retention of 27.4%, 65.5%, 65.9%, and 85.7%, respectively. After 300 cycles at 1 C, LCO-Ti/P maintains a capacity of 130.4 mAh/g, i.e., having a retention of 67.3%, far superior  Table S1 compares the highvoltage electrochemical performance of LCO-Ti/P with that of the modified LiCoO 2 material published in the recent literature. The dual-element modified sample LCO-Ti/P shows a comparable high-voltage performance. Figure 4b displays the rate capabilities of LCO-Ti/P and LCO. LCO-Ti/P delivers a higher capacity under high current density, and a reversible capacity of 199.7 mAh/g with retention of 97.5% is achieved while returning to 0.1 C, indicating that the dual-element modification can provide a fast Li-ion diffusion pathway. The charge/discharge profiles at the selected cycle numbers are presented in Fig. 4c, d. Pristine LCO shows severely degraded profiles compared with LCO-Ti/P, and the corresponding discharge voltage fading during cycling is shown in Fig. S4. Rapid voltage fading is caused by severe structural degradation. The ameliorative voltage decay of LCO-Ti/P indicates that the dual-element modification can efficaciously stabilize the crystal structure during Li-ion deintercalation and intercalation.
CV curves within the first five cycles for these samples are compared in Fig. 5. All samples have redox peaks near 3.8 V and 4.4 V, which correspond to the H2/H1 and M2/ H3 phase transitions, respectively [48]. For pristine LCO, these peaks gradually weaken with the scanning cycles (Fig. 5a), and similar trends appear for LCO-Ti and LCO-P (Fig. 5b, c). The broadened redox peaks in LCO-P (Fig. 5b) indicate suppressed charge transfer kinetics due to the low electronic conductivity of Li 3 PO 4 . In the case of LCO-Ti/P, slight decreases in the peak intensities are apparent (Fig. 5d), suggesting a reversible phase transition. Additionally, the potential difference between the main redox peaks in the first cycle for LCO-Ti/P equals 0.202, which is smaller than that of LCO (0.222), reflecting the lower polarization of LCO-Ti/P compared with pristine LCO. The lower polarization leads to a better electrochemical property; therefore, LCO-Ti/P possesses excellent cycling stability.
In situ XRD patterns were tested for LCO and LCO-Ti/P during the first cycle in the range of 3.0-4.6 V at 0.2 C. The evolution of the (003) and (107) diffraction peaks of LCO and LCO-Ti/P is shown in Fig. 6a, b, respectively. The insulator-metal two-phase transition at the beginning of the charging stage leads to the splitting of (003) and (107) peaks [49]. The order-disorder transition process at 4.2 V can be observed in LCO-Ti/P, featuring (107) peak splitting [50]; however, this transition process disappears in the case of LCO, and this phenomenon agrees with the disappearance of anodic and cathodic peaks at 4.2 V in the CV curves of LCO (Fig. 5a), suggesting that LCO-Ti/P possesses high structural reversibility. When the voltage is charged to 4.6 V, the phase transition from O3 to H1-3 is observed in the (003) peak evolution pattern, and the variation in the c-value during the phase transition is reflected in the amplitude of the (003) peak variation. LCO has a larger amplitude (1.277°) than LCO-Ti/P (1.222°), demonstrating the mitigation of structural change for LCO-Ti/P under high voltage. To obtain more detailed structure variations during charging and discharging, Rietveld refinements of in situ XRD were obtained, as shown in Fig. 6c. With the deintercalation and intercalation of Li ions, the lattice parameter, c-value, and volume value change significantly, and LCO-Ti/P exhibits less structural variation than that of LCO, indicating that dual-element modification can alleviate the structure deterioration [51,52]. EIS was performed to analyze the kinetic behaviors of these samples. Nyquist plots are shown in Fig. 7a, b, and Fig. S6, respectively, and the fitting results by the Zview2 software are shown in Table 2 and Table S4. The intersection of the starting point and the x-axis in the Nyquist plots represents the ohmic resistance (R s ), and the semicircles at high frequency and medium frequency represent the electrode surface film impedance (R f ) and charge transfer impedance (R ct ), respectively; the sloping line at low frequency reflects the Li-ion solid-state diffusion (W o ) [53]. The transfer kinetics at the interface can be reflected by the change in R ct , and the reduced charge transfer resistance benefits Li-ion diffusion [54]. The growth of R ct is significant for LCO, and R ct drastically increases to 1412.00 Ω after 50 cycles at a high voltage of 4.6 V. The suppressed growth of R ct for LCO-Ti and LCO-P is visually displayed in Fig. 7c, and concrete numerical results are listed in Table S4. Although R ct gently increases to 675.20 for LCO-Ti/P (Table 2), the smaller variation in R ct in cycling is due to the special surface structure resulting from the dualelement modification.
GITT characterizations were conducted for all samples during the 1 st and 100 th cycles, as shown in Fig. 7d and Fig. S7, respectively. The analysis details are shown in Supporting Information. The GITT result during the 1 st cycle shows that the Li-ion diffusion coefficients of LCO-Ti/P and pristine LCO are almost identical, slightly higher than that of LCO-P, and lower than that of LCO-Ti. After 100 cycles, LCO-Ti/P shows a slightly reduced Li-ion diffusion coefficient compared with the first cycle, indicating that the dual-element modification generates a superior interfacial structure that can provide a fast Li-ion diffusion pathway. Moreover, the measurably reduced charge transfer resistance and unimproved Li-ion diffusion kinetics of LCO-P probably originate from the amorphous Li 3 PO 4 coating layer, which is chemically inert to the electrolyte [55,56].
Characterizations of the spent samples were performed for LCO and LCO-Ti/P after 100 cycles in the range of 3.0-4.6 V at 1 C. SEM images of spent LCO after 100 cycles show an obvious disintegration of particles (Fig. 8a), whereas spent LCO-Ti/P has intact particles with dense cathode electrolyte interphase (CEI) layers formed on the surface (Fig. 8b). The accumulated stress originating from anisotropic expansion and contraction of the lattice during the deep lithiation/delithiation process leads to the fragmentation of the particles, whereas after dual-element modification, the structural deterioration of LCO-Ti/P is alleviated. Figure 8c, d shows the HRTEM images of spent LCO and LCO-Ti/P, respectively. The LCO surface (Fig. 8c) reconstructs extensively and forms a large area of Co 3 O 4 -like spinel structure (dashed cyan rectangle in Region 1 and dashed green rectangle in Region 2) under long-term cycling due to oxygen loss and cobalt dissolution. The irreversible phase transition leads to rapid capacity decay [57,58]. In contrast, the layered structure with a space group of R-3 m (Region 2) still dominates the crystal structure of LCO-Ti/P (Fig. 8d). The lattice fringe of 0.47 nm in the bulk corresponds to atypical (003) plane of the R-3 m layered structure. The FFT pattern of the outermost surface in Region 1 shows the (0-24), (113), and (13-1) planes of monoclinic Li 2 TiO 3 , and the amorphous phase (marked by a dashed red line) in the coating layer on the surface of LCO-Ti/P can be identified as Li 3 PO 4 , indicating a coating layer tightly attached to the bulk during cycling. In addition, the microcrack formation can be observed in the spent LCO, whereas no microcracks appear in the spent LCO-Ti/P. The robust coating layer hinders the direct contact of LCO with electrolytes, thus suppressing the side reaction originating from the electrolyte decomposition, and the oxygen loss and cobalt dissolution are effectively inhibited at high voltage. The optimized interfacial structure leads to the suppressed growth of charge transfer resistance, thus improving the cycling stability.

Discussion
The characterizations of XRD (Fig. 1), HRTEM (Fig. 2), and XPS (Fig. 3) illustrate that at the interface of the modified sample LCO-Ti/P, monoclinic Li 2 TiO 3 and phosphate coatings are present. To further analyze the mechanism of dual-element modification, DFT calculations were performed to explain the improved cycling stability of LCO-Ti/P. First, the lattice-matching property between monoclinic Li 2 TiO 3 and layered LiCoO 2 was analyzed by calculating the binding energies of the individual crystal planes of these two materials. Figure 9a shows the interfacial structure models of the Li 2 TiO 3 (002) and LiCoO 2 (003) crystal planes. The binding energy between these two crystal planes is calculated as −17.49 eV, which suggests a strong affinity between them. Shim et al. [27] analyzed the microstructure of LiCoO 2 modified by tetrabutyl titanate precursor through HAADF-STEM and found that the coating of monoclinic Li 2 TiO 3 with space group C2/c was epitaxially grown on the LiCoO 2 surface, and the (003) planes of LiCoO 2 matched the (002) planes of Li 2 TiO 3 , which agreed with the simulation results. In addition, the lattice matching for the Li 2 TiO 3 (202) and LiCoO 2 (104) crystal planes is spontaneous (Fig. 9b), with a calculated binding energy of − 10.9 eV. Similarly, Fig. S8 shows the interfacial structure models of matched crystal planes, including Li 2 TiO 3 (002) and LiCoO 2 (104) as well as Li 2 TiO 3 (202) and LiCoO 2 (003). The corresponding binding energies of these matched crystal planes are negative, as shown in Fig. 9c, illustrating that the major crystal planes of monoclinic Li 2 TiO 3 and layered LiCoO 2 match perfectly. This excellent lattice matching between LiCoO 2 and Li 2 TiO 3 is beneficial for achieving the superior structural stability of the cathode material by inhibiting particle cracking and breakage.
Moreover, the GITT results (Fig. 7d) indicate that LCO-Ti has the highest Li-ion diffusion coefficients. Herein, the effects of Li 2 TiO 3 on the Li-ion diffusion energy barriers were investigated using molecular simulation. For pristine LiCoO 2 , as shown in Fig. 10a, there are two Li-ion diffusion paths in LiCoO 2 [59,60]: (1) Li ions diffuse through the adjacent octahedra in the Li-O layer (oxygen dumbbell hopping, ODH path), and (2) Li ions migrate among the octahedra through the intermediate tetrahedra (tetrahedral site hopping path, TSH path). Figure 10b indicates that in   pristine LiCoO 2 , the Li-ion diffusion energy barrier is lower through the ODH path (0.68 eV) than through the TSH path.
For monoclinic Li 2 TiO 3 , there exists a 3D Li-ion diffusion path, i.e., Li ions can migrate through the Li-O and Ti-O layers, as displayed in Fig. 10c, e. Li 2 TiO 3 has a lower Li-ion diffusion energy barrier through the Li-O layer (0.48 eV) (Fig. 10d) and the Ti-O layer (0.46 eV) (Fig. 10f), suggesting faster Li-ion diffusion in Li 2 TiO 3 compared with LiCoO 2 . Therefore, the 3D Li-ion diffusion in monoclinic Li 2 TiO 3 results in a higher Li-ion diffusion coefficient in the modified sample of LCO-Ti/P. Moreover, the above experimental results indicate the presence of an amorphous Li 3 PO 4 coating on the modified sample LCO-Ti/P. Herein, the potential location sites of PO 4 3− on the major crystal plane (104) of LiCoO 2 were studied, considering three typical sites: (1) the adsorption site by the Li atom, (2) the adsorption site by the Co atom, and (3) the adsorption site midway between Li and Co atoms (Fig. 11a). The calculated adsorption energies are shown in Fig. 11b. The negative adsorption energies suggest that PO 4 3− can be easily adsorbed on the (104) plane of LiCoO 2 .
The oxygen release phenomenon is one of the main problems of the capacity fading of LiCoO 2 at high voltage. The effects of PO 4 3− adsorption on the formation of an oxygen vacancy were studied, considering four positions of potential oxygen vacancies (O1-O4 in Fig. 11c). In the absence of PO 4 3− adsorption, the formation energy for an oxygen vacancy is calculated as − 2.23 eV. The negative formation energy suggests that active oxygen atoms can easily escape from the lattice framework of LiCoO 2 (104) under a deep delithiation state, which would further lead to structural collapse and more interfacial side reactions. In contrast, in the presence of PO 4 3− adsorption, as shown in Fig. 11d, the formation energies of the four types of oxygen vacancies in LiCoO 2 (104) are positive, indicating that PO 4 3− adsorption can inhibit oxygen release from the deep delithiation state of LiCoO 2 , resulting in superior cycling stability for LCO-Ti/P.

Conclusion
A dual-element Ti and P modification method involving a sol-gel treatment and a solid-phase reaction with the precursors of Ti(OC 4 H 9 ) 4 and NaH 2 PO 2 was developed to improve the electrochemical performance of LiCoO 2 at high voltage. The dual-element modified sample LCO-Ti/P shows an initial discharge capacity of 211.6 mAh/g (0.1 C) and excellent capacity retention of 85.7% after 100 cycles at 1 C and RT (3.0-4.6 V). The morphology, phase compositions, and surface microstructure of modified samples were analyzed using SEM, XRD, HRTEM, and XPS to disclose the mechanism for improving the performance at high voltage. It is illustrated that LCO-Ti/P has the monoclinic Li 2 TiO 3 and the amorphous Li 3 PO 4 coating layers. Analyses of in situ XRD, CV, EIS, and GITT tests show that LCO-Ti/P undergoes a mitigated structural change and phase transition. The alleviated growth of charge transfer resistance and an elevated Li-ion diffusion coefficient for LCO-Ti/P stem from the reconstructed interfacial structure. Furthermore, DFT calculations disclose the superior lattice-matching property of the major crystal planes of Li 2 TiO 3 and LiCoO 2 , the lower energy barriers for Li-ion diffusion in Li 2 TiO 3 , and the suppressed oxygen release performance resulting from the phosphate coating.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.