Nitrogen-doped rock-salt Li₃V₂O₅ nanosheet arrays with improved rate capability as an anode for thin film lithium-ion microbatteries†

Wei Liu‡ , Chenyang Xu‡ , Fan Kong , Qiuying Xia *, Feng Zan , Jing Xu and Hui Xia *
School of Materials Science and Engineering, Nanjing University of Science and Technology, Nanjing 210094, China. E-mail: xiahui@njust.edu.cn; qiuyingxia@njust.edu.cn

First published on 4th July 2025

---

---

Qiuying Xia

Qiuying Xia is an associate professor at the School of Materials Science and Engineering, Nanjing University of Science and Technology. He received his B.E. (2014) and PhD (2020) degrees from the Nanjing University of Science and Technology. His research interests focus on lithium-ion batteries and all-solid-state thin film lithium batteries.

1 Introduction

The rapid development of the Internet of Things (IoT) has catalysed swift progress in miniaturized and highly integrated electronic and information technologies, giving rise to a surge in small autonomous wireless devices (e.g., microsensors, microelectromechanical systems, and microrobots) that constitute IoT devices.¹ This has also led to an increasing demand for appropriate integrated micro-sized energy storage devices. Among current battery technologies, the all-solid-state thin film lithium/lithium-ion battery (TFB) stands out as the optimal energy solution for microelectronic devices due to its inherent safety, flexible size control, on-chip power design, and low self-discharge rate.² However, the anode materials currently used in TFBs are predominantly lithium metal anodes, which suffer from low chemical stability, high air sensitivity, and a low melting point of approximately 180 °C. Given that the solder-reflow process during microelectronics integration typically involves a harsh gas environment and temperatures of over 180 °C, lithium anodes are ill-suited to meet the on-board integration requirements of TFBs.³ Thus, it is urgent to develop new thin-film anodes with high chemical stability, a high melting point, and good electrochemical performance for future TFBs.

Among various attractive candidates to replace lithium anodes, disordered rock-salt Li₃V₂O₅ (LVO) has recently attracted extensive attention due to its high chemical stability, high melting point, and good structural stability.⁴ Besides, the LVO anode can yield a much higher capacity (∼300 mA h g⁻¹) and lower operating voltage (∼0.6 V vs. Li/Li⁺) than lithium titanate, TiO₂, and Nb₂O₅-based anodes that possess good structural stability and low volume expansion.^5,6 Additionally, the LVO anode experiences less volume expansion during cycling than Si and Sn-based anodes with high theoretical capacities.^7,8 Despite these advantages, the inherently low electronic conductivity and Li⁺ diffusion coefficient of LVO still restrict its enhancement in rate capability. Unfortunately, the current research on LVO electrodes mainly focuses on LVO materials produced through the electrochemical prelithiation process, morphology engineering of LVO, and intercalation chemistry of LVO.^4,9 Addressing the restricted Li-ion diffusion and suboptimal electrical conductivity in LVO remains a significant challenge. Moreover, most studies on LVO anodes have focused on testing LVO powders, while research on LVO anodes in thin films remains underexplored.

Previous reports have shown that nitrogen doping is an effective method for improving the electrochemical properties of metal oxide electrodes (e.g., MnO₂ and TiO₂) by creating defect states and increasing their electronic conductivity.^10,11 Inspired by these findings, we hypothesize that nitrogen doping could optimize LVO's electronic structure and Li⁺ transport dynamics. In this work, a nitrogen-doped rock-salt LVO (N–LVO) thin-film anode is prepared via an in situ electrochemical transition of nitrogen-doped layered V₂O₅ (N–V₂O₅) nanosheet arrays, which are deposited by reactive magnetron sputtering at room temperature. Unlike conventional nitrogen doping strategies for metal oxides that involve nitriding treatment at high temperatures, we can achieve nitrogen doping in N–LVO at room temperature and optimize the electrode's performance facilely by adjusting the flow ratios of nitrogen in the reactive working gases. A comprehensive comparison study between the N–LVO and LVO thin films has been carried out to investigate the benefits of nitrogen doping. The results show that nitrogen doping can efficiently improve the electrode kinetics of the N–LVO electrode by improving the electronic conductivity and reducing the Li⁺ ion migration barriers of N–LVO. Consequently, N–LVO exhibits a high reversible capacity of 350.1 mA h g⁻¹ at 0.1C, exceptional rate capability (160.7 mA h g⁻¹ at 10C), and remarkable cycling stability (80% capacity retention after 2000 cycles). These performances significantly outshine those of the LVO electrode, rendering N–LVO a highly promising candidate as an anode for advanced TFBs.

2 Experimental section

2.1 Preparation of V₂O₅ and N–V₂O₅ thin films

The V₂O₅ and N–V₂O₅ thin films were deposited on stainless steel foil and glass substrates by reactive radio frequency (RF) magnetron sputtering (Kurt J. Lesker) at room temperature. A pure V metal (76 mm in diameter) was used as the target. The distance between the target and the substrate was adjusted to 10 cm. Before deposition, the chamber was vacuumed to a pressure lower than 1 × 10⁻⁵ Pa. A pre-sputtering period of at least 5 minutes was conducted before each deposition. The deposition of the V₂O₅ film was performed at an RF power of 120 W at a flow of 40 sccm Ar and 10 sccm O₂. And the depositions of N–V₂O₅ films were performed at an RF power of 120 W under the flow of N₂ and O₂. The samples deposited under the flow of 15 sccm N₂ and 55 sccm O₂, 20 sccm N₂ and 50 sccm O₂, 35 sccm N₂ and 35 sccm O₂, 60 sccm N₂ and 10 sccm O₂, and 65 sccm N₂ and 5 sccm O₂ are marked as sample 15–55NVO, 20–50NVO, N–V₂O₅, 60–10NVO, and 65–5NVO, respectively.

2.2 Preparation of N–V₂O₅/LiPON/Li TFB

A 2 μm thick LiPON electrolyte film was deposited on the N–V₂O₅ film by RF magnetron sputtering (Kurt J. Lesker) using a Li₃PO₄ target (76 mm in diameter) at a sputtering power of 100 W and a gas flow of 90 sccm N₂. Then, a 2 μm thick Li anode film was deposited on the N–V₂O₅/LiPON film by thermal evaporation (Kurt J. Lesker) to construct the N–V₂O₅/LiPON/Li TFB. The depositions of LiPON and Li films were carried out in a glove box.

2.3 Material characterization

The crystallinity and structure of the samples were characterized by X-ray diffraction (XRD, Bruker AXS D8). The chemical composition and bonding information of the samples were investigated by X-ray photoelectron spectroscopy (XPS, Escalab250Xi, Thermo Scientific). The Raman spectra were acquired using a 532 nm wavelength incident laser on a Renishaw inVia Reflex Raman microprobe. The morphologies and microstructures of the samples were characterized using a field emission scanning electron microscope (FESEM, FEI Quanta 250F), transmission electron microscope (TEM), and high-resolution transmission electron microscope (HRTEM, FEI Tecnai G2) equipped with energy dispersive X-ray spectroscopy (EDS) and electron energy loss spectroscopy (EELS). Ultraviolet-visible (UV-vis) absorption spectra of the samples were collected using a Shimadzu UV-3600 UV/vis/NIR spectrophotometer.

2.4 Electrochemical measurements

To investigate the electrochemical performance, R2025-type coin cells were assembled in an Ar-filled glove box. The V₂O₅ and N–V₂O₅ thin films deposited on stainless steel foil were directly used as the working electrode, and pieces of Li foil were used as both the counter and reference electrodes. Celgard 2400 was used as the separator, and 1 M LiPF₆ in ethylene carbonate and diethyl carbonate (EC/DEC, v/v = 1:1) solution was used as the electrolyte. The electrodes based on the V₂O₅, 15–55NVO, 20–50NVO, N–V₂O₅, 60–10NVO, and 65–5NVO films are denoted as LVO, 15–55NLVO, 20–50NLVO, N–LVO, 60–10NLVO, and 65–5NLVO, respectively. The mass loading of the films was measured using a Sartorius analytical balance (model CPA225D, with a resolution of 10 μg). The mass loading was found to be 0.12–0.15 mg cm⁻². The galvanostatic charge/discharge measurements and galvanostatic intermittent titration technique (GITT) measurements were carried out in the voltage range between 0.01 and 2 V at different current densities using a Neware BTS4000 battery test system. The GITT test was conducted at a current density of 10 mA g⁻¹ with a duration of 10 min and an open-circuit relaxation for 30 min. The Li⁺ diffusion coefficient (D_Li⁺) was calculated based on Fick's law:

	(1)

where i is the applied current, V_m represents the molar volume, F is Faraday's constant, S represents the contact area, is the slope of the coulometric titration curve, and is the slope of the linearized plot of the potential E during the current pulse. Cyclic voltammetry (CV) measurements were carried out in the voltage range between 0.01 and 2 V on an electrochemical working station (Biologic VSP). Electrochemical impedance spectroscopy (EIS) measurements were performed by applying an AC amplitude of 10 mV over the frequency range of 100 kHz to 0.01 Hz at open circuit potential on an electrochemical workstation (Biologic VSP).

2.5 Density function theory (DFT) calculations

All the calculations were performed using the Vienna Ab initio Simulation Package (VASP) software with the projector augmented wave (PAW) method based on the density functional theory.¹² The Perdew–Burke–Ernzerhof (PBE) functional within the generalized gradient approximation (GGA) was used to describe the electronic interaction relation terms. The Li₃V₂O₅ structure was generated using a 2 × 2 × 5 supercell of the primitive rock-salt unit cell. The occupancy of the octahedral sites (Wyckoff symbol 4b) was set at 0.6Li: 0.4 V.¹³ In order to obtain a more precise structure, a plane wave energy cutoff of 520 eV and the Brillouin zone k-point sampling of the 4 × 4 × 9 Monkhorst–Pack (MP) mesh were employed for geometry optimization. All systems were initialized with a high-spin ferromagnetic configuration for the least error in relative energies.¹⁴ In addition, the convergence of total energy between two adjacent iteration steps is less than 10⁻⁵ eV and the forces on each atom were less than 0.05 eV Å⁻¹. Meanwhile, a Hubbard U extension (U value) of 3.25 eV for vanadium was considered for the optimization of the structure and energy calculations.⁴ The N-doped model was similarly obtained based on the optimization of the above conditions.

3 Results and discussion

The N–LVO and LVO thin films were prepared via the in situ electrochemical transition of N–V₂O₅ and V₂O₅ nanosheet arrays, respectively. The N–V₂O₅ and V₂O₅ nanosheet arrays were prepared using reactive RF magnetron sputtering, as schematically illustrated in Fig. 1a. During the sputtering process, reactive N₂/O₂ and Ar/O₂ gases were introduced into the chamber to ionize into the corresponding ions to bombard the V metal target, resulting in the depositions of N–V₂O₅ and V₂O₅ films on the substrates, respectively. At a low substrate temperature of room temperature, the growth of thin films favors 3D island growth via the Volmer–Weber growth mode.¹⁵ This occurs because once the sputtered ions arrive at the substrate, they possess low energy for diffusion and thus tend to aggregate into clusters. Fig. 1b and S1a (ESI†) show the top-view FESEM images of the N–V₂O₅ and V₂O₅ thin films, respectively. Both samples display 3D architecture characterized by nearly vertically aligned nanosheet skeletons with a lateral size of around 0.5–1 μm, accompanied by smaller nanosheet branches that are 100–200 nm in lateral size. Fig. 1c and S1b† show the cross-sectional FESEM images of the N–V₂O₅ and V₂O₅ films, respectively, revealing their highly porous morphology with a thickness of about 1.7 μm. These results indicate that nitrogen doping into V₂O₅ will not alter the film's morphology and thickness. According to previous reports, such a 3D architecture may help expand the electrode/electrolyte interface and enable fast ion transport kinetics for both films.¹Fig. 1d shows the EDS elemental mapping images of the N–V₂O₅ film. Uniform distributions of V, O, and N elements throughout the film were observed, verifying the nitrogen doping in the film. And the N:V ratio is determined to be about 14.02% for the N–V₂O₅ film.

Based on the experimentally measured N/V ratio and the valence balance of V, O, and N, the chemical formula of N–V₂O₅ after N doping is determined to be about V_2.06O_4.71N_0.29. Fig. 1e presents the XRD patterns of the N–V₂O₅ and V₂O₅ films. Apart from the diffraction peaks from the substrate, all other diffraction peaks of the two samples match those of layered V₂O₅ (JCPDS card no. 89-0611),^4,9 suggesting that both samples exhibit high crystallinity without any impurities. The absence of some V₂O₅ peaks based on the JCPDS card can be attributed to the preferential growth of the layered V₂O₅ film along specific crystallographic orientations during magnetron sputtering deposition, thereby reducing the exposure of certain crystal planes. Although both samples are pure layered V₂O₅ phases, their diffraction peaks exhibit significant differences in intensity, indicating that nitrogen doping in N–V₂O₅ affects the exposed facets and the crystal structure. This phenomenon can be confirmed by the Raman results of the N–V₂O₅ and V₂O₅ films in Fig. S2 (ESI†). Both samples exhibit Raman peak positions located at around 148, 199, 286, 304, 407, 483, 526, 701, and 996 cm⁻¹, which are well aligned with the characteristic Raman peaks of orthorhombic V₂O₅.¹⁶ The peak at 148 cm⁻¹ is assigned to the B_1g/B_3g mode, representing the skeleton bend vibration of the V₂O₅ layers. The 199 cm⁻¹ peak corresponds to the Ag mode and is associated with the bending vibrations of O–V–O bonds. The Raman peak at 286 cm⁻¹ is attributed to the B_1g/B_3g mode, indicative of the bending vibrations of O–V–O bonds. The peak at 304 cm⁻¹ is linked to the Ag mode and corresponds to the bending vibrations of V–O bonds. The 407 cm⁻¹ peak is related to the Ag mode and is associated with the stretching vibrations of V–O bonds. The Raman peak at 483 cm⁻¹ is attributed to the Ag mode, representing the bending vibrations of V–O–V bonds. The peak at 526 cm⁻¹ is linked to the Ag mode and corresponds to the stretching vibrations of V–O bonds. The Raman peak at 701 cm⁻¹ is associated with the B_3g mode and represents the stretching vibrations of O–V–O bonds. The peak at 996 cm⁻¹ corresponds to the Ag mode and is attributed to the stretching vibrations of terminal oxygen (VO) in the V₂O₅ structure. A red shift from 148 to 143 cm⁻¹ can be observed for N–V₂O₅, suggesting that the substitution of oxygen by nitrogen results in the relaxation of the V–O–V bonds within the V₂O₅ framework.¹⁷ Besides, the N–V₂O₅ film shows blue shifts at around 414, 488, 536, and 709 cm⁻¹, indicating an increase in the vibrational frequency of the V–O bonds due to the incorporation of nitrogen, which likely results from the strengthening of these bonds or a change in the local bonding environment. The Raman results, along with the EDS elemental mapping and XRD results, provide definitive evidence of nitrogen doping in the N– V₂O₅ film synthesized through the reactive magnetron sputtering method. Based on this finding, we prepared different N–V₂O₅ films with varying doped nitrogen content by simply adjusting the reactive N₂/O₂ gas ratio during sputtering (Fig. S3–S5, ESI†), aiming to explore how varying nitrogen doping levels influence the electrochemical properties of the resultant rock-salt N–LVO electrodes. The elemental composition and chemical state of the N–V₂O₅ and V₂O₅ films were investigated by XPS. The survey-scan XPS spectra in Fig. S6 (ESI†) reveal the presence of V, N, and O elements in the N–V₂O₅ film and V and O elements in the V₂O₅ film without impurities. The N:V ratio was measured to be approximately 14.51%, consistent with the EDS elemental mapping result. The V 2p core-level XPS spectra of the N–V₂O₅ and V₂O₅ films are presented in Fig. 1f. Both samples exhibit V 2p_1/2 and V 2p_3/2 peaks at 524.78 and 517.31 eV, respectively, corresponding to the V⁵⁺ oxidation state. In addition to the peaks ascribed to V⁵⁺, small peaks corresponding to V³⁺ can be observed for the N– V₂O₅ film, suggesting that nitrogen doping reduces the average valence state of V.^18,19Fig. 1g shows the N 1s core-level XPS spectra of the N–V₂O₅ and V₂O₅ films. Two peaks centered at 401.42 and 399.83 eV can be observed from the N–V₂O₅ film, which can be assigned to V–O–N and V–N bonds, respectively, further confirming the presence of nitrogen in N–V₂O₅.¹⁸ Fig. S7a (ESI†) presents the UV-vis spectra of the N–V₂O₅ and V₂O₅ films, revealing a blue shift in the absorption sharp edge of the N–LVO film from 579 to 543 nm. The optical band gap of the N– V₂O₅ and V₂O₅ films are measured to be 2.16 and 2.30 eV (Fig. S7b†), respectively, suggesting that nitrogen doping improves the electrical conductivity of the N–V₂O₅ film.

Based on the N–V₂O₅ film, the N–LVO electrode can be synthesized via an in situ electrochemical lithiation process. Fig. 2a shows the initial two discharge and charge curves of the N–LVO electrode at a current density of 0.1C (1C = 250 mA g⁻¹). As the cell discharges from the open circuit voltage to 3.3, 2.5, 2.2, and 2 V (vs. Li/Li⁺), N–V₂O₅ may undergo sequential transformations into the ε, δ, γ, and ω phases. This transformation is evidenced by four distinct plateaus in the initial discharge curve, which matches well with previously reported discharge profiles for V₂O₅ anodes.¹⁹ When the cell is further discharged to a lower potential below 2.0 V with 3-Li⁺ insertion, layered V₂O₅ can be irreversibly transformed into a rock-salt ω-LVO phase. The initial coulombic efficiency of the N–LVO electrode is about 23.6%, which is close to those of the LVO electrode (24.74%) and previously reported rock-salt LVO electrodes.^4,9 The very similar coulombic efficiencies of the two electrodes indicate that the introduction of N in N–LVO does not alter its phase transition process. Ex situ XRD and Raman characterization studies were performed to reveal the phase transformation from the layered N–V₂O₅ phase to the disordered rock-salt N–LVO phase. Fig. 2b presents the XRD patterns of pristine N–V₂O₅ and the transformed N–LVO at different charge/discharge states. After the initial discharging process, the characteristic diffraction peaks of the layered V₂O₅ phase disappear, while a new diffraction peak (∼43°) that is assigned to the (200) plane of the rock-salt LVO phase appears.⁴ And the XRD patterns of the transformed N–LVO remain in a disordered rock-salt phase in subsequent cycles, suggesting the exceptional structural stability of the rock-salt phase throughout the cycling process. Fig. 2c compares the Raman spectra of pristine N–V₂O₅ and the transformed N–LVO at different charge/discharge states. Notably, the Raman bands of N– V₂O₅ that correspond to the layered V₂O₅ phase vanish after the initial discharging process, and a new Raman band emerges at approximately 818 cm⁻¹ for the N–LVO samples at 0.01 V and 2 V, verifying the formation of the rock-salt N–LVO phase.¹⁹ To directly visualize this phase transition, comprehensive TEM investigations were performed. Fig. 2d shows the TEM image of the N–V₂O₅ film, confirming the nanosheet morphology of N–V₂O₅. The HRTEM image of N–V₂O₅ in Fig. 2e exhibits a well-ordered layered structure with a lattice fringe of 0.589 nm, corresponding to the (200) plane of V₂O₅. Fig. 2f displays the STEM image and the corresponding EDS elemental mapping images of N–V₂O₅. Uniform distribution of V, O, and N elements can be observed within the film, confirming the nitrogen doping in the layered V₂O₅ phase. The TEM image of the transformed N–LVO film at the 2 V state is shown in Fig. 2g, revealing that its nanosheet morphology was well retained after the phase transition. Fig. 2h presents the HRTEM image of the N–LVO at the 2 V state. Clear lattice fringes with an interplanar spacing of about 0.210 nm can be observed, corresponding to the (200) crystal plane of rock-salt LVO. The selected area electron diffraction (SAED) pattern of the N–LVO inserted in Fig. 2h can be well indexed to the rock-salt LVO phase, confirming the cubic rock salt structure of N–LVO. The STEM image and EDS elemental mapping images of N–LVO in Fig. 2i reveal the homogeneous nitrogen distribution throughout the film, confirming the nitrogen doping in the transformed N–LVO film. Combined with the proportion of the N element measured in N–V₂O₅, the chemical formula of the transformed N–LVO material could be Li₃V_2.06O_4.71N_0.29.

Ex situ TEM measurements of the N–LVO electrode at 0.01 V and 2 V states were further performed to investigate the Li storage mechanism. Fig. 3a–f show the HRTEM images and corresponding FFT patterns of N–LVO at 0.01 V and 2 V states, indicating the well-retained rock-salt structure with ordered atomic arrangement of N–LVO throughout the lithiation and delithiation processes. For N–LVO at the 0.01 V state, the interplanar spacing of lattice fringes assigned to the (200) plane was determined to be 0.213 nm. As Li⁺ ions were removed from N–LVO, the interplanar spacing of these lattice fringes decreased to 0.210 nm for N–LVO at 2 V, suggesting a small volume change of about 4.3% for N–LVO between the fully lithiated and delithiated states. Electron energy-loss spectroscopy (EELS) was performed to detect the valence state changes in the N–LVO electrode during charging and discharging processes. Fig. 3g shows the V-M edge and Li-K edge EELS spectra of N–LVO at 2.0 and 0.01 V states. The Li-K edge is evident in both samples, suggesting that Li⁺ ions are not completely removed from N–LVO at the fully charged state. The presence of Li⁺ ions enables the rock-salt structure of N–LVO to be maintained at the fully charged state. Fig. 3h compares the V L-edge EELS spectra of N–LVO at 2.0 and 0.01 V states. The energy difference between V L₃-edge and O K-edge is found to be 10.6 eV for N–LVO at the 2 V state and 11.6 eV for the N–LVO at the 0.01 V state, corresponding to a V oxidation state of about 3.4+ and 2.0+, respectively.^20,21 The observed changes in the V oxidation states align with the specific capacity of the N–LVO electrode. The EELS results, along with the XRD, Raman, and HRTEM results, demonstrate that vanadium redox reactions occur in N–LVO with a specific amount of Li⁺ ions being reversibly inserted into and extracted from the rock-salt structure. And the lithium storage mechanism in N–LVO is schematically depicted in Fig. 3i.

To understand the influence of nitrogen doping on the rock-salt LVO electrode, the electrochemical performances of the LVO and N–LVO electrodes were systematically studied and compared. The electrochemical properties of various nitrogen-doped LVO electrodes, including 60–10NLVO, 20–50NLVO, 15–55NLVO, and 65–5NLVO electrodes, were also investigated to determine the influence of different nitrogen doping contents on the electrochemical performances of rock-salt LVO. Fig. 4a presents the CV curves of the N–LVO and LVO electrodes between 0.01 and 2.0 V at a scan rate of 0.2 mV s⁻¹. Both CV curves exhibit a pair of redox peaks at around 0.6 V, corresponding to Li⁺ ion insertion and extraction into/from the rock-salt LVO phase.²² The N–LVO electrode shows a smaller gap between its redox peaks compared to the LVO anode, indicating that the N–LVO electrode possesses smaller polarization. The typical charge and discharge curves of the N–LVO and LVO electrodes at 0.1C during the third cycle are presented in Fig. 4b. Notably, the LVO and N–LVO electrodes exhibit a high specific capacity of about 309.3 mA h g⁻¹ (37.1 μAh cm⁻²) and 350.1 mA h g⁻¹ (42 μAh cm⁻²) at a current rate of 0.1C, respectively, which are higher than those of previously reported LVO electrodes.²³ These capacities of the LVO and N–LVO electrodes measured at low C-rates exceed the theoretical capacity of LVO (∼264.1 mA g⁻¹), which may be attributed to the extra Li⁺ ion adsorption on the surface of the nanosheet arrays with high surface area.²⁴Fig. 4c compares the rate performances of the LVO and N–LVO electrodes. At current rates of 0.2C, 0.5C, 1C, 2C, 5C, and 10C, the N–LVO electrode delivers discharge capacities of 307.4, 292.3, 265.0, 235.9, 200.6, and 160.7 mA h g⁻¹ respectively, which are obviously higher than 222.2, 179.5, 103.7, 79.8, 44.4, and 30.2 mA h g⁻¹ of the LVO electrode. These results reveal that the N–LVO electrode possesses enhanced rate capability compared to the LVO electrode. When the current density reduces back to 2C directly, the specific capacity recovers to about 246.4 mA h g⁻¹, suggesting that the Li⁺ ion insertion and extraction processes in N–LVO are highly reversible. Moreover, when compared to the 60–10NLVO, 20–50NLVO, 15–55NLVO, and 65–5NLVO electrodes, the N–LVO electrode delivers the highest specific capacities at all current rates (Fig. S8, ESI†), indicating that an appropriate amount of N doping significantly enhances the reversible capacity and rate capability of rock-salt LVO. However, the excessive nitrogen doping in LVO significantly reduces the specific capacity of the LVO electrode. This may be attributed to that the nitrogen doping in LVO induces the formation of V³⁺, thereby reducing the overall valence state of V in LVO that can participate in redox reactions. Fig. 4d shows the cycle performances of the N–LVO and LVO electrodes at 2C. The fluctuations in capacity for the electrodes can be attributed to variations in environmental temperature, as the cycle performances of the cells in our work were tested at room temperature in a laboratory environment. The room temperature was susceptible to fluctuations due to weather changes. Nevertheless, these capacity fluctuations occur within limited cycle intervals, which exert minimal influence on the overall cycling stability or subsequent cycle performance analysis of the cells. As observed, both the N–LVO and LVO electrodes exhibited fast capacity fading during the initial 50 cycles. Nevertheless, the capacity of the N–LVO electrode began to decline more slowly after the 50th cycle and maintained a specific capacity of 235.2 mA h g⁻¹ after 2000 cycles (80% capacity retention), which is obviously higher than 75.1 mA h g⁻¹ of the LVO electrode after 2000 cycles (38.9% capacity retention). This result reveals that the nitrogen doping in N–LVO contributes to improved cycling stability. Besides, the coulombic efficiency of the N–LVO electrode remained close to 100% throughout the entire cycling, demonstrating its exceptional suitability as an anode material for TFBs. Fig. S9a and b (ESI†) show the top-view FESEM images of the LVO and N–LVO electrodes, respectively. As observed, the LVO surface is fully covered by a SEI film, which completely obscures the initial nanosheet array morphology. A thick SEI layer on the LVO electrode implies severe electrolyte decomposition, which is pinpointed as the main reason for its rapid capacity decay. Unlike the LVO electrode, the N–LVO electrode retains its original nanosheet array structure without cracking after 50 cycles, demonstrating superior stability with a stable SEI layer formed on the surface. In addition, the rate performance and cycle performance of the present binder-free N–LVO electrode are better or comparable to those of previously reported LVO electrodes (Table S1, ESI†), indicating that nitrogen doping is a highly effective approach to improve the lithium storage properties of rock-salt LVO. To get insight into the mechanism of the enhanced electrochemical performances of the N–LVO electrode, EIS measurements were carried out (Fig. 4e). The Nyquist plots of the N–LVO and LVO electrodes at the pristine state show an intercept of the semicircle with the Z′ axis, one semicircle in the high-frequency region, and a straight line in the low frequency region, which represent the system resistance, charge transfer resistance, and Warburg impedance of Li ion diffusion into the active material, respectively.^9,22 As observed, the N–LVO electrode exhibits smaller charge transfer resistance (∼402 Ω) as compared to the LVO electrode (∼467 Ω), suggesting that the N–LVO electrode facilitates faster charge transport.²² Fig. S9c† presents the EIS spectra of the LVO and N–LVO electrodes after 50 cycles. Both electrodes after 50 cycles exhibited elevated charge transfer resistance compared to their pristine states, which can be attributed to the formation of the SEI layer. Notably, the LVO electrode demonstrates substantially higher impedance than the N–LVO electrode, confirming thicker SEI formation due to more extensive electrolyte decomposition. This consequently deteriorated charge transfer kinetics and accelerated capacity fading in the LVO electrode. CV measurements at different scan rates were further carried out to understand the improved kinetics of the N–LVO electrode. Fig. 4f and g show the CV curves of the N–LVO and LVO electrodes at various scan rates from 0.1 to 1 mV s⁻¹. The peak currents (i) in the CV curves follow a power-law relationship with the scan rate (v) according to the the following relationship (eqn (2)):^25,26

i = avb	(2)

where a and b are constants. The value of b can provide information about the nature of the electrode process. In particular, a b-value close to 0.5 corresponds to a diffusion-controlled process, while a b-value close to 1.0 suggests capacitive behavior. Following this concept, the capacitive and diffusive contributions in the CV curve for a particular scan rate can be determined using Dunn's equation:²⁷

i = k1v + k2v0.5	(3)

where k₁v and k₂v^0.5 represent the diffusive part and capacitive part, respectively. By simply plotting the relationship between i and v^0.5, we can calculate the k₂ value. And the testing current i at a specific potential can be quantitatively analysed using the values of k₁ and k₂. By calculating the k₁ and k₂ at different voltages at various scan rates, we can plot the capacitive capacity contribution in the CV curves (Fig. S10, ESI†). The ratios between the capacitive and diffusion-controlled capacity contributions for the N–LVO and LVO electrodes at various scan rates were calculated and are shown in Fig. 4h. It is observed that approximately 68% of the total capacity of the N–LVO electrode is contributed by capacitive capacity at 0.1 mV s⁻¹, which is higher than the 52% for the LVO electrode. The dominance of capacitive-controlled capacity contributions of both electrodes could be attributed to their nanosheet array morphology with high surface area and short ion diffusion pathways. Notably, the capacitive-controlled capacity contributions of the N–LVO electrode are higher than those of the LVO electrode at all scan rates, suggesting that the N–LVO electrode has fast redox reactions and rate-independent behavior during the electrochemical process.

To further explore the mechanism behind the superior rate capability of the N–LVO electrode, GITT measurements were performed. The typical GITT potential profiles of the N–LVO and LVO electrodes are shown in Fig. 5a, and the calculated Li⁺ diffusion coefficients are compared in Fig. 5b. The Li⁺ diffusion coefficients of the N–LVO electrode are in the range of 9.56 × 10⁻¹⁴ to 1.49 × 10⁻¹² cm² s⁻¹, surpassing those of the LVO electrode. Fig. 5c presents the UV-vis absorption spectra of the N–LVO and LVO films. The optical band gap of the N–LVO film was found to be 1.43 eV, lower than that of the LVO film (1.52 eV), indicating enhanced electron transfer in the N–LVO film due to nitrogen doping. These results indicate that the N–LVO film can simultaneously achieve fast Li⁺ diffusion and efficient electron transfer, thereby resulting in improved electrode kinetics. DFT calculations were carried out to elucidate the improved Li⁺ diffusion and electron transfer mechanisms of N–LVO arising from nitrogen doping. To optimize the crystal structure of N–LVO, five oxygen coordination environments in N–LVO for potential nitrogen substitution were explored, classified as 3Li–3V, 4Li–2V, 2Li–4V, 6Li, and 5Li–1V based on the surrounding Li and V atomic configurations (Table S2, ESI†). Total energy calculations revealed that the 3Li–3V configuration exhibited the highest stability. Consequently, oxygen sites with 3Li–3V coordination were chosen for subsequent nitrogen substitution simulations. And the crystal structure information of the simulated N–LVO and LVO is presented in Table S3 (ESI†). Fig. 5d shows the optimized structural models, the bond lengths, and Li⁺ ion migration paths in the structural models of LVO and N–LVO. To elucidate the effects of nitrogen doping on the electronic structure, the Densities of States (DOS) of N–LVO and LVO were determined and are depicted in Fig. 5e. No band gap near the Fermi level (E_F) can be observed for LVO and N–LVO, indicative of a metallic nature. This is due to the random distribution of transition metal cations in the disordered rock-salt phase, which introduces more defects or electronic states, leading to changes in the band structure and altering the electronic energy levels. Nevertheless, a significant increase in peak intensity near the Fermi level is observed for N–LVO, indicating that nitrogen doping enhances the density of electronic states in N–LVO. This leads to higher electron occupancy in the energy structure, thereby resulting in improved electronic conductivity of N–LVO.²⁸ Besides, the bond length between nitrogen and lithium is found to be longer than that between oxygen and lithium. This indicates that the bond energy between nitrogen and lithium is likely lower, making the chemical bond relatively weaker. The weaker bond may allow Li⁺ ions to escape the constraints of nitrogen atoms more easily, thus promoting the diffusion of Li⁺ ions. To demonstrate this, the diffusion energy barriers of Li⁺ ions in N–LVO and LVO were investigated, as shown in Fig. 5f. The Li⁺ ion diffusion energy barriers in N–LVO were calculated to be 1.36 and 1.15 eV along paths 1 and 2, respectively, which are obviously lower than 2.17 and 1.34 eV in LVO. This suggests that the N doping in N–LVO facilitates the uptake and release of Li⁺ ions more effectively, thereby accounting for the rapid capacitive charge storage. The DFT calculation results, which are in good agreement with the experimental findings, demonstrate that nitrogen doping in rock-salt LVO significantly impacts the electronic structure and optimizes the Li⁺ diffusion environment, thereby leading to the enhanced electrochemical performance of the N–LVO electrode.

With the advantages of the room-temperature preparation process, high specific capacity, good rate capability, and superior cycling stability, N–LVO is well-suited as an anode material for TFBs. To demonstrate this, an N–LVO/LiPON/Li TFB (N–LVO TFB) was constructed by sequentially depositing the LiPON film and Li film on the N–V₂O₅ film (Fig. S11a, ESI†), followed by an in situ electrochemical transition process. Fig. S11b† displays the initial two charge and discharge curves of the N–LVO TFB at 0.1C. The N–LVO TFB exhibited a low open circuit voltage of below 2 V, which is attributable to the electrolyte Li⁺ infusion into the N–V₂O₅ film during LiPON deposition. Benefiting from this, N–V₂O₅ can consume fewer Li⁺ ions to transform into N–LVO during the initial discharging process, thereby resulting in an obviously higher initial coulombic efficiency (65.9%) than that of the N–LVO tested in a liquid half-cell (23.6%). This relatively higher initial coulombic efficiency also makes the application of N–LVO in TFBs more feasible than in liquid half-cells. In the subsequent cycles, the N–LVO TFB exhibited charge and discharge curves that are similar to those in the N–LVO half cells and a reversible specific capacity of 250.1 mA h g⁻¹ (30.0 μAh cm⁻²). Fig. S11c† shows the rate performance of the N–LVO TFB. The specific capacity and rate performance of the N–LVO TFB are relatively lower than those of the N–LVO tested in the liquid half-cell, which can be attributed to the limited charge transfer kinetics of the solid-state battery architecture using LiPON as the electrolyte. As can be seen from Fig. S12 (ESI†), the N–LVO TFB exhibits a larger charge transfer resistance (∼480 Ω) than the N–LVO electrode (∼402 Ω). Besides, the N–LVO TFB shows a flatter Warburg region slope compared to the N–LVO electrode, indicating slower ionic diffusion kinetics. Nevertheless, the N–LVO TFB exhibited relatively higher specific capacity (about 315.2 mA h g⁻¹/37.8 μAh cm⁻² at 0.05C) and lower operating voltage than most currently reported anodes for TFBs tested in solid-state battery architecture (Table S4, ESI†). Fig. S11d† shows the cycle performance of the N–LVO TFB at 0.2C. The capacity fluctuations of the N–LVO TFB can be ascribed to environmental temperature variations. Notably, the N–LVO TFB achieved good cycling stability, maintaining a capacity of 192.2 mA h g⁻¹ after 200 cycles, positioning it as a viable anode material for future TFBs. To evaluate structural stability of the N–LVO TFB upon cycling, XRD and FESEM characterization studies on the N–LVO TFB after 200 cycles were performed. The N–LVO TFB before cycling showed an XRD pattern (Fig. S13a, ESI†) similar to that of the N–LVO electrode after the initial discharging process (Fig. 2b), which was attributable to the electrolyte Li⁺ infusion into the N–V₂O₅ film during LiPON deposition. The XRD patterns of the N–LVO TFB at the 2 V and 0.01 V states after 200 cycles remain consistent with those before cycling, with no additional peaks detected, indicating the high structural stability of N–LVO during cycling. Fig. S13b† shows the cross-sectional FESEM image of the N–LVO TFB after 200 cycles. The nanosheet array morphology of the N–LVO film is well-preserved after 200 cycles. Additionally, the LiPON electrolyte maintains intimate contact interfaces with both the N–LVO and Li films, demonstrating the exceptional stability of the N–LVO TFB.

4 Conclusions

In summary, a disordered rock-salt N–LVO thin film with nitrogen doping was prepared through the in situ electrochemical transition of nitrogen-doped V₂O₅ nanosheet arrays that are deposited by reactive magnetron sputtering at room temperature. By adjusting the reactive N₂/O₂ gas ratios, the doping contents of nitrogen in the films can be manipulated. With the synergistic benefits of improved Li⁺ diffusion coefficient and enhanced electric conductivity resulting from optimal nitrogen doping, the optimized N–LVO electrode achieved substantially enhanced rate capability over the bare LVO electrode. Specifically, the N–LVO electrode exhibited reversible specific capacities of 350.1 mA h g⁻¹ at 0.1C and 160.7 mA h g⁻¹ at 10C, which are superior to 309.3 mA h g⁻¹ at 0.1C and 30.8 mA h g⁻¹ at 10C of the LVO electrode. More importantly, the N–LVO anode possessed excellent cycling stability with 80% capacity retention after 2000 cycles at 2C, making it a promising anode material for TFBs. This research offers a valuable structural regulation approach that is advantageous for advancing thin-film electrodes in TFBs.

Data availability

All relevant data are within the manuscript and its additional files.

Author contributions

Wei Liu: investigation, methodology, and writing – original draft. Chenyang Xu: investigation, methodology, and visualization. Fan Kong: investigation and methodology. Qiuying Xia: methodology, visualization, and writing – review & editing. Jing Xu: supervision. Feng Zan: supervision. Hui Xia: supervision, funding acquisition, and writing – review & editing.

Conflicts of interest

The authors declare that they have no conflict of interest.

Acknowledgements

This work was funded by the National Natural Science Foundation of China (No. 52372202 and 52272218) and the Fundamental Research Funds for the Central Universities (No. 30924010204). The authors would like to thank Meihong Tan from Shiyanjia Lab (https://www.shiyanjia.com) for support of XPS characterization.

References

1. Q. Xia, F. Zan, Q. Zhang, W. Liu, Q. Li, Y. He, J. Hua, J. Liu, J. Xu, J. Wang, C. Wu and H. Xia, Adv. Mater., 2023, 35, 2200538 CrossRef CAS .
2. Y. Yu, S. Cao, C. Dong, H. Wang, C. Zhou, Z. Miao, K. Yan, M. Gong, L. Mai and X. Xu, Adv. Funct. Mater., 2025, 2423990 CrossRef .
3. X. Wang, G. Pawar, Y. Li, X. Ren, M. Zhang, B. Lu, A. Banerjee, P. Liu, E. J. Dufek, J.-G. Zhang, J. Xiao, J. Liu, Y. S. Meng and B. Liaw, Nat. Mater., 2020, 19, 1339–1345 CrossRef CAS .
4. H. Liu, Z. Zhu, Q. Yan, S. Yu, X. He, Y. Chen, R. Zhang, L. Ma, T. Liu, M. Li, R. Lin, Y. Chen, Y. Li, X. Xing, Y. Choi, L. Gao, H. S.-y. Cho, K. An, J. Feng, R. Kostecki, K. Amine, T. Wu, J. Lu, H. L. Xin, S. P. Ong and P. Liu, Nature, 2020, 585, 63–67 CrossRef CAS .
5. J. Yang, M. Huang, L. Xu, X. Xia and C. Peng, Chem. Eng. J., 2022, 445, 136638 CrossRef CAS .
6. J. Y. Cheong, J.-W. Jung, D.-Y. Youn, C. Kim, S. Yu, S.-H. Cho, K. R. Yoon and I.-D. Kim, J. Power Sources, 2017, 360, 434–442 CrossRef CAS .
7. M. Kim, Z. Z. Yang, S. E. Trask and I. Bloom, ACS Appl. Mater. Interfaces, 2022, 14, 15103–15111 CrossRef CAS .
8. Z. Zhao, Q. Liu, D. Huang, F. Zhou, Y. Liu and D. Zhang, Mater. Mech. Eng., 2012, 36, 1–7 CAS .
9. X. Lan, L. Wang, L. Yu, Y. Li and X. Hu, ACS Mater. Lett., 2021, 3, 1394–1401 CrossRef CAS .
10. R. Feng, M. Li, Y. Wang, J. Lin, K. Zhu, J. Wang, C. Wang and P. K. Chu, Electrochim. Acta, 2021, 370, 137716 CrossRef CAS .
11. W.-L. Bai, S.-M. Xu, C.-Y. Xu, Q. Zhang, H.-H. Wang, Z. Zhang, X. Chen, S.-Y. Dong, Y.-S. Liu, Z.-X. Xu, X.-G. Zhang, Z. Wang, K.-X. Wang and J.-S. Chen, J. Mater. Chem. A, 2019, 7, 23046–23054 RSC .
12. G. Kresse and J. Hafner, Phys. Rev., 1993, 48, 13115–13118 CAS .
13. G. W. L. Hart and R. W. Forcade, Phys. Rev. B:Condens. Matter Mater. Phys., 2008, 77, 22411577 CrossRef .
14. M. F. V. Hidalgo, Y. Lin, A. Grenier, D. Xiao, J. Rana, R. Tran, H. Xin, M. Zuba, J. Donohue, F. O. Omenya, I. Chu, Z. Wang, X. Li, N. A. Chernova, K. W. Chapman, G. Zhou, L. Piper, S. P. Ong and M. S. Whittingham, J. Mater. Chem. A, 2019, 7, 8423–8432 RSC .
15. H. Q. Ta, D. J. Perello, D. L. Duong, G. H. Han, S. Gorantla, V. L. Nguyen, A. Bachmatiuk, S. V. Rotkin, Y. H. Lee and M. H. Rümmeli, Nano Lett., 2016, 16, 6403–6410 CrossRef CAS .
16. Prashant, A. Kumar, V. Dhiman, A. Somvanshi, R. Parmar, J. M. Rosolen, D. B. de Feritas Neto and N. Kondal, Ceram. Int., 2023, 49, 31815–31828 CrossRef .
17. A. Akande, A. Machatine, B. Masina and G. Chimowa, J. Phys. D: Appl. Phys., 2018, 51, 015106 CrossRef .
18. M. C. Biesinger, L. W. M. Lau, A. R. Gerson and R. S. C. Smart, Appl. Surf. Sci., 2010, 257, 887–898 CrossRef CAS .
19. H. E. M. Hussein, R. Beanland, A. M. Sànchez, D. Walker, M. Walker, Y. Han and J. V. Macpherson, J. Mater. Chem. A, 2022, 10, 8515–8527 RSC .
20. A. Mukherjee, N. Sa, P. J. Phillips, A. Burrell, J. Vaughey and R. F. Klie, Chem. Mater., 2017, 29, 2218–2226 CrossRef CAS .
21. H. Tan, J. Verbeeck, A. Abakumov and G. Van Tendeloo, Ultramicroscopy, 2012, 116, 24–33 CrossRef CAS .
22. X. Lan, X. Liu, T. Meng, S. Yang, Y. Shen and X. Hu, Small Methods, 2023, 7, 2201290 CrossRef CAS .
23. X. Lan, L. Wang, L. Yu, Y. Li and X. Hu, ACS Mater. Lett., 2021, 3, 1394–1401 CrossRef CAS .
24. B. Q. Hu, X. D. Wang, J. Xu, J. N. Ding and S. H. Ge, J. Energy Storage, 2023, 66, 107472 CrossRef .
25. W.-K. Han, J.-X. Wei, K. Xiao, T. Ouyang, X. Peng, S. Zhao and Z.-Q. Liu, Angew. Chem. Int. Ed., 2022, 61, e202206050 CrossRef CAS .
26. N. Liu, X. Wu, L. Fan, S. Gong, Z. Guo, A. Chen, C. Zhao, Y. Mao, N. Zhang and K. Sun, Adv. Mater., 2020, 32, 1908420 CrossRef CAS .
27. M. Liu, Z. Cai and K. Ramani, J. Comput. Appl. Math., 2025, 467, 116596 CrossRef .
28. Y. Qin, H. Zhao, M. Hua, J. Gao, J. Lu, S. Zhang, X. Lin, J. Sun, L. Yin and R. Wang, J. Power Sources, 2023, 588, 233722 CrossRef CAS .

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ta03758a

‡ Both authors contributed equally to this work.

This journal is © The Royal Society of Chemistry 2025