Si-doped vanadium pentoxide/graphene xerogel nanocomposite cathodes with excellent cycle life for Li-ion batteries

Abstract

Vanadium pentoxide (V₂O₅) is a promising cathode material for lithium-ion batteries due to its high theoretical specific capacity (443 mA h g⁻¹) and specific energy (1218 Wh kg⁻¹ at 2.75 V). However, its large-scale application is limited by its poor electrical conductivity and structural instability, leading to irreversible capacity loss, consequently, short cycle life, and low-rate capability. In this work, we developed a 3D nanostructured xerogel composite by uniformly doping silicon (Si) into the V₂O₅ framework, which anchors on the surface of graphene sheets via a sol–gel process. The optimized Si-doped V₂O₅ on the graphene (Si-V₂O₅@G) composite, containing 10 wt% Si and 2 wt% graphene, delivers a high specific capacity of 392 mA h g⁻¹ at 0.1C rate and excellent cycling stability: 589 cycles with 80% capacity retention at a high rate of 1.0C, and a very low capacity fading rate of 0.03% capacity loss per cycle—significantly surpassing the best performance of up-to-date V₂O₅ work (e.g. the best performance of 160 mA h g⁻¹ and 300 cycles at 0.75C rate and 0.13% capacity loss per cycle). Scanning Transmission Electron Microscopy (STEM) reveals uniform Si distribution, which leads to much larger V₂O₅ nanoribbons over the graphene sheets (observed using TEM). Such a nanocomposite structure is capable of tolerating more structural change during the (de)lithiation process, resulting in a much improved cycle life. Electrochemical impedance spectroscopy (EIS) and cyclic voltammetry (CV) further demonstrate the improved conductivity and structural integrity. X-ray photoelectron Spectroscopy (XPS) manifests the interaction of doped Si with the V–O system. The in operando XANES and EXAFS analyses suggest the increased multi-valence change among V⁵⁺, V⁴⁺, and V³⁺ states in the (de)lithiation process after Si-doping. Additionally, EXAFS analysis indicates that Si doping effectively stabilizes the local V–O coordination environment, facilitating better Li⁺ insertion/extraction reversibility and reducing the degradation of Si-V₂O₅@G systems. This work demonstrates a cost-effective, non-metal doping strategy for enhancing the electrochemical performance of high-energy-density metal oxide cathodes for next-generation LIBs.

---

1. Introduction

Lithium-ion batteries (LIBs) have become indispensable energy storage devices in many applications in today's society, ranging from portable electronics to electric vehicles,^1–7 driving the demand for higher energy density to meet the increasing needs of the energy storage market. However, conventional intercalation-type cathode materials such as LiFePO₄ and LiCoO₂ are limited by a one-electron transfer mechanism, limiting their theoretical capacities and impeding further improvements in energy density.^8–12 Among alternative high-capacity cathode candidates, vanadium pentoxide (V₂O₅) has drawn considerable attention due to its low cost, natural abundance, and a high theoretical capacity of 443 mA h g⁻¹, stemming from its ability to intercalate up to three lithium ions per formula unit over a broad voltage window (1.5–4.0 V vs. Li⁺/Li).^13–21 Despite these advantages, practical implementation of V₂O₅ as commercial cathodes is hampered by severe structural evolution and rapid capacity fading during deep (de)lithiation, especially when engaging the full three-electron redox process.²² These challenges arise from phase transitions involving crystalline and amorphous states (e.g., ω-Li_xV₂O₅ to β-Li_0.3V₂O₅)^14,23–25 and significant short-range order (SRO) evolution that destabilizes the electrode materials. Recent advances in cation-disordered rock-salt-type cathode materials suggest that the heterogeneous element doping effect can suppress detrimental SRO evolution and enhance structural resilience.^{11,19,20,22,25,26} Multiple doping of heteroatoms into a material can result in the increased thermodynamic intrinsic stability, which has been demonstrated in recent high entropy catalysts²⁷ and battery electrode materials.^18,28 In our previous work, we took the approach of using the graphene sheet to (1) improve the local electronic conduction on the nanoscale and (2) stabilize the V₂O₅ structure by sandwiching V₂O₅ nanoribbons between two graphene sheets to maintain the structural integrity during (de)lithiation.^29,30 Such an approach achieved theoretical specific capacity and 120 cycles with 80% initial capacity.³¹

A cost-effective, nonmetal entropy-engineering strategy by introducing silicon (Si), the second most abundant element on Earth's crust, into the V₂O₅ lattice and incorporating graphene to form a robust 3D nanostructure via a sol–gel process has been developed in the past work.^{9,17,26,32,33} Based on this concept, we developed a Si-doped V₂O₅/graphene composite (Si-V₂O₅@G) cathode herein to achieve uniform Si distribution and enhanced structural stability. Electrochemical testing demonstrates the outstanding performance, with the optimized Si-V₂O₅@G (10 wt% Si, 2 wt% graphene) delivering a high specific capacity of 392 mA h g⁻¹ at 0.1C and achieving 589 cycles at a fairly high 1C rate with 80% of its initial capacity and a very low capacity fading rate of 0.03% capacity loss per cycle—significantly surpassing the performance of most of the V₂O₅ work (e.g. the best performance of 160 mA h g⁻¹ and 300 cycles at 0.75C rate and 0.13% capacity loss per cycle²⁶) reported up to now (Table S1). STEM manifested that Si doping could lead to much larger V₂O₅ nanoribbons over the graphene sheets, which can tolerate more structural change during the (de)lithiation process, resulting in a much improved cycle life. Cyclic voltammetry and impedance spectroscopy confirm that Si doping enhances electronic conductivity and reduces charge transfer resistance. The XPS analysis displays the doping effect of Si on the V₂O₅ structures, which indicates strong binding effects of Si–O–V. Moreover, in operando XANES and EXAFS analyses³⁴ reveal that Si doping stabilizes the V–O framework and promotes reversible valence-state transitions of vanadium, enabling a complete and symmetric (de)lithiation process. These results underscore the efficacy of Si doping simultaneously enhancing both structural and electrochemical stability, offering a promising pathway for the development of next-generation high-energy cathode materials.^35–41

2. Experimental methods

2.1 Materials

LiPF₆ (battery grade), ethylene carbonate (EC) and ethyl carbonate (EMC) were obtained from Novolyte Technologies (Independence, OH, USA) for electrolyte preparation. NaVO₃ (99.5%) and Na₂SiO₃ (99%) were purchased from Sigma-Aldrich. Proton-exchange resin (Dowex-50-WX2) with 50–100 mesh was obtained from Thermo Scientific Chemicals. Super P conductive carbon black was purchased from MTI Corporation (Richmond, CA, USA) and used without further treatment.

2.2 Synthesis of a Si-doped V₂O₅ xerogel

The V₂O₅ xerogel was synthesized using our previous modified ion exchange method.^42,43 Around 100 mL of 0.1 M NaVO₃ aqueous solution was eluted through a column loaded with a proton exchange resin. The obtained yellow solution of HVO₃ was aged in a glass container for 3 weeks to obtain a mature homogeneous V₂O₅ hydrogel. The dried V₂O₅ xerogel was obtained by freeze-drying the above hydrogel under vacuum for about a week. The Si-doped V₂O₅ xerogel was prepared by additionally mixing Na₂SiO₃ into the 0.1 M NaVO₃ solution and following the same procedures as for the preparation of the V₂O₅ xerogel.^15,16,44

2.3 Synthesis of GO

Graphene oxide was prepared based on our previous method^42,43 and more details can be found in the SI. The synthesized GO was dispersed in purified deionized (DI) water with a concentration of about 10 mg mL⁻¹.^17,31

2.4 Synthesis of V₂O₅@G and Si-V₂O₅@G

The V₂O₅@G hybrid was prepared by mixing the prepared GO suspension and the yellow solution of HVO₃ generated by proton exchange at the desired ratio. The obtained dark yellow solution was aged in a glass container for 3 weeks to obtain a completely cured, homogeneous V₂O₅/GO hydrogel. The dried V₂O₅@GO xerogel was obtained by freeze-drying the V₂O₅@GO hydrogel under vacuum in a freeze-dryer. The formed V₂O₅@GO xerogel was heated and annealed in an inert environment (N₂) at a rate of 5 °C min⁻¹ up to 400 °C and kept at 400 °C for 6 h, during which the GO was reduced to graphene and the graphene-modified V₂O₅ hybrid (V₂O₅@G) was formed. The graphene-modified Si-doped V₂O₅ hybrid (Si-V₂O₅@G) was prepared by mixing the GO suspension with the solution of HVO₃ and H₂SiO₃ generated by proton exchange and following the same procedures as for the V₂O₅@G preparation. For comparison, additional two samples, V₂O₅ and Si doped V₂O₅ (Si-V₂O₅), were prepared without mixing with the GO suspension and then underwent the same treatment as V₂O₅@G and Si-V₂O₅@G. The mass percentage of vanadium in V₂O₅, V₂O₅@G, Si-V₂O₅ and Si-V₂O₅@G is 56.0 wt%, 54.9 wt%, 49.4 wt% and 49.2 wt%, respectively.

2.5 Electrochemical characterization

The electrodes were prepared by casting a slurry of 80 wt% active materials (V₂O₅@G and Si-V₂O₅@G, respectively), 10 wt% polyvinylidene difluoride (PVDF), and 10 wt% Super P onto a 10 µm-thick sheet of Al foil. For comparison, the electrodes of V₂O₅ and Si-V₂O₅, which were synthesized using the same methods except for the addition of GO, were prepared. The prepared electrodes were placed in a vacuum oven and allowed to dry at 90 °C for 24 h. The electrolyte consisted of 1.2 M LiPF₆ in a mixture of ethylene carbonate (EC) and ethyl-methylene carbonate (EMC) solvent (3:7, by weight). The prepared electrodes were assembled into 2016-type coin cells using Li metal anodes (Aldrich, USA) and Celgard 2400 separators (Celgard, Ohio, USA) to characterize their electrochemical performance. These cells were tested using an Arbin battery cycler (BT-2000, Arbin, TX, USA) with different C-rates between 1.5 and 4.0 V. A Solartron 1287A/1260A Potentiostat/Impedance System (Solartron Analytical, England, UK) was used to measure the AC impedance of these cells in the frequency range of 0.01 Hz–1 MHz with a voltage amplitude of 5 mV. The results were fitted using the equivalent circuit, where R₀ is the contact resistance, R_e and C_e refer to the resistance and capacitance of the V₂O₅ electrode, R_ct and C_dl stand for the charge-transfer resistance of the redox reaction of vanadium in V₂O₅ and the double-layer capacitance in the electrode, respectively, and W_d refers to the Warburg diffusion impedance, which could reflect the diffusion of Li ions in V₂O₅.

2.6 Materials characterization

Scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDXS) were conducted using a JEOL 7800 SEM (Japan) operating at 5 kV. High-resolution transmission electron microscopy (HRTEM) was performed on a FEI Talos microscope (USA) operated at 80 kV. Synchrotron radiation-based in operando X-ray absorption spectroscopy (XAS) was employed to investigate the dynamic electrochemical behavior of the V₂O₅ cathodes. X-ray photoelectron spectroscopy (XPS) was performed using a PHI 5000 VersaProbe II system attached to an Ar-filled glovebox with monochromatic Al Kα radiation. X-ray powder diffraction (XRD) was conducted using Bruker D8 Discover, Bruker USA. In particular, in operando X-ray absorption near-edge structure (XANES) spectroscopy was used to probe the oxidation state and local structural changes of vanadium across all phases during cycling. In operando XANES measurements were carried out on the coin cells incorporating V₂O₅@G and Si-V₂O₅@G electrodes at beamline 20-BM-B of the Advanced Photon Source (APS), Argonne National Laboratory. Coin cells with 2 mm diameter holes at the center were prepared for the in operando study, sealed with Kapton tape to allow X-ray penetration while maintaining an inert environment. The vanadium K-edge (5463.76 eV) was monitored in transmission mode using a Si (111) monochromator. Energy calibration was performed using the first derivative of the vanadium metal reference spectrum, which was simultaneously collected in a dedicated reference channel for each measurement. Identical electrodes were used in all cells to ensure consistency and improve signal-to-noise ratios. Electrochemical cycling was conducted at a constant current of ∼0.5C between 1.7 V and 3.6 V. XANES data processing and analysis were performed using the ATHENA software package following standard protocols.

3. Results and discussion

3.1 Synthesis and morphology of Si-doped V₂O₅ on graphene

The graphene supported Si-doped V₂O₅ (Si-V₂O₅@G) was synthesized via a modified ionic exchange method.³¹ As shown in Fig. 1a, a well-mixed aqueous solution of NaVO₃ and Na₂SiO₃ was eluted through an ion exchange/separation column filled with proton exchange resin. During the process, the sodium ions were exchanged with protons (H⁺) and immobilized in the resin, while VO₃⁻ and SiO₃²⁻ further acquired a proton and became HVO₃ and H₂SiO₃ in the resulting solution, followed by uniform mixing with GO suspension. An olation reaction along the H₂O–V–OH direction to expand the coordination of vanadium, incorporated with Si, occurred during the subsequent aging procedure, during which the resulting solution turned dark red with increasing viscosity. Finally, V₂O₅@G was obtained after freeze drying and annealing processes. The SEM image of Si-V₂O₅@G is shown in Fig. 1b, in which a sheet stacked morphology of Si-V₂O₅ sheets supported on the planar graphene surface. The corresponding EDXS mapping (Fig. 1c–e) shows that V and Si are uniformly distributed over the surface of sheets, which indicates that Si has been uniformly doped into the V₂O₅ structure. The EDXS mapping also demonstrates the uniform distribution of carbon, indicating the distribution of V₂O₅ on graphene sheets.¹⁷ The TEM analysis in Fig. 1f shows that Si-V₂O₅@G displays the nanoribbon-like structure supported on the graphene sheet. The average diameter of the Si-V₂O₅ nanoribbons is 100 nm with an outstanding morphology, as shown in Fig. 1g. The corresponding selected area electron diffraction (SAED) (Fig. 1h) on the single nanoribbon in Fig. 1f shows the lattice structure corresponding to the (010), (100) and (110) crystal planes of V₂O₅, while no SiO₂ lattice was detected, indicating the successful doping of Si into the V₂O₅ structure instead of the formation of a SiO₂ phase. The above phenomenon is also consistent with the XRD results (Fig. S2), where no SiO₂ phase was detected in both Si-V₂O₅ and Si-V₂O₅@G systems. Compared with Si-V₂O₅@G, the V₂O₅@G without Si doping reveals much smaller sized nanoribbons, with an average diameter of 50 nm, as shown in Fig. 1i. The comparatively larger size of nanoribbons in Si-V₂O₅@G suggests that the doping of Si into the V₂O₅ lattice can increase the crystal size of the structure, making it more robust during the (de)lithiation process, resulting in a significantly improved cycle life. The STEM EDXS mapping of V₂O₅@G (Fig. S1) manifests the uniform distribution of V, O and C, suggesting the outstanding formation of the V₂O₅ structure sitting on a graphene sheet. The STEM EDXS mapping of Si-V₂O₅@G (Fig. 1j–n) further demonstrates the uniform distribution of Si, V and O within the Si-V₂O₅ nanostructure, indicating the excellent doping of Si into the V₂O₅ structure at the nanocrystalline scale. The XRD analysis exhibits that both V₂O₅@G and Si-V₂O₅@G are non-crystalline structures when integrated with the graphene matrix, while pure V₂O₅ and Si-V₂O₅ display a crystalline structure after thermal treatment at 400 °C. Besides, the XRD analysis of Si-V₂O₅@G shows that Si doping introduces more peaks around the smaller 2-theta region compared with Si-V₂O₅@G, which may result in the larger nanoribbons.

Fig. 1 Synthesis schematic and morphology analysis. (a) Schematically illustrated synthesis of the self-assembled Si-V₂O₅@G sample. (b) SEM images and (c–e) corresponding EDXS analysis. (f) TEM images of Si-V₂O₅@G, (g) high-resolution TEM image and (h) corresponding SAED pattern. (i) TEM images of pure V₂O₅/G. (j–n) TEM EDXS analysis of the Si-V₂O₅@G sample.

3.2 Electrochemical performance and stability

To evaluate the electrochemical performance and cycling stability of the Si-V₂O₅@G materials, coin-type cells were assembled with Li metal as the counter and reference electrode. The cycling stability was evaluated based on galvanostatic charge–discharge at 1.0C. As shown in Fig. 2a, Si-V₂O₅@G demonstrates significantly improved stability and achieved 589 cycles at a fairly high 1C rate with 80% of its initial capacity with a very low capacity fading rate of 0.03% capacity loss per cycle—significantly surpassing the performance of most V₂O₅ cathodes (e.g. the best performance of 160 mA h g⁻¹ and 300 cycles at 0.75C rate with 0.13% capacity loss per cycle²⁶) reported up to now (Table S1). This improvement is much better than that of V₂O₅@G (102 cycles with 80% capacity retention). In addition, it was found that the cycling stability of both V₂O₅@G and Si-V₂O₅@G is higher than that of pure V₂O₅ and Si-V₂O₅, which indicates that the incorporation of graphene sheets can also improve the stability of the V₂O₅ nanoribbons. Such an improvement from the graphene sheets can be attributed to the greatly enhanced electric conduction between V₂O₅ units within the V₂O₅ nanoribbons while the graphene sheet provides an electron transport path and enhances the structural integrity of V₂O₅ nanoribbons by sandwiching them between two neighboring graphene sheets. Moreover, for both V₂O₅ cathodes with or without incorporation of graphene, the addition of Si doping can significantly improve stability. The enhanced stability should originate from the doping of Si into the V₂O₅ nanostructure, possibly the larger sized V₂O₅ nanoribbons, and thus improving the structural stability and reversibility. Furthermore, the rate performance of these cathode materials was recorded at various current densities and is illustrated in Fig. 2b. Both Si-V₂O₅@G and V₂O₅@G have the same rate capability until 1C rate while V₂O₅@G has increasingly improved rate capability as the rate is beyond 1C. This also supports the hypothesis that doping Si leads to larger sized V₂O₅ nanoribbons, which are responsible for improved stability. The larger V₂O₅ nanoribbons help tolerate more lattice strain change during (de)lithiation, maintain the structural integrity and enable higher cycle life. On the other hand, the graphene sheet can improve the electronic conductivity between the V₂O₅ units (i.e. bipyramids) of the V₂O₅ nanoribbon, while the graphene sheet cannot enhance the electronic conductivity within the V₂O₅ nanoribbon. Larger-sized ribbons may result in lower electronic conduction within the V₂O₅ nanoribbon due to the elongated electron transport length. Besides, the non-crystalline structure (i.e. amorphous) of Si-V₂O₅@G and V₂O₅@G, which is displayed in XRD (Fig. S2), provides much more configurational stability during cycling compared with pure V₂O₅ and Si-V₂O₅ with much more crystalline structures. The corresponding specific capacity of Si-V₂O₅@G is 397, 360, 315, 274 and 172 mA h g⁻¹, similar to that of V₂O₅@G (419, 354, 312, 288, and 247 mA h g⁻¹, respectively) at the current densities of 0.1, 0.5, 1.0, 2.0 and 5.0C. In addition, compared with Si-V₂O₅, Si-V₂O₅@G exhibits significantly higher specific capacity, especially at larger current densities, which suggests that the incorporation of graphene can effectively increase the large-current capacity due to the improved conductivity. From our previous work,⁴² the initial cycle of charge and discharge curves of pristine V₂O₅ displays three plateaus, which corresponds to the crystalline-to-crystalline transitions with a sequence of α-Li_xV₂O₅, ε-Li_xV₂O₅, δ-Li_xV₂O₅, γ-Li_xV₂O₅, and ω-Li_xV₂O₅. While for V₂O₅@G, Si-V₂O₅, and Si-V₂O₅@G in Fig. 2c, no plateaus were observed in the initial discharge process, which indicates the non-crystalline properties compared with pristine V₂O₅.^23,44–47 The non-crystalline (i.e. amorphous) V₂O₅ has higher cycling stability compared with crystalline V₂O₅ due to its reversible structural change. Furthermore, there are much smaller differences for the charge and discharge curves between the initial and final (590th) cycles for Si-V₂O₅@G compared with V₂O₅ (Fig. 2d), which demonstrates the outstanding cycling stability of Si-V₂O₅@G and shows much better performance compared with the other samples. Besides, Si-V₂O₅@G also displays outstanding and improved stability compared with other recent published studies (Table S1). Overall, the introduction of Si leads to large V₂O₅ nanoribbons, which change the V₂O₅ long-term structure from crystalline to non-crystalline and significantly improves the cycling stability.

Fig. 2 Electrochemical performance of different V₂O₅ cathodes; electrolyte: 1.2 M LiPF₆ in EC : EMC (3 : 7 by weight), 2016-type coin cell, anode: Li metal, separator: Celgard 2400, and room temperature. (a) Capacity retention cycled at 1.0C. (b) Rate capability performance. (c) Initial discharge and charge curve at 1.0C rate. (d) The end-of-life discharge and charge curve at 1.0C rate.

3.3 EIS, CV curves and XPS analysis on Si doping

To investigate the cause of the significantly improved cycling stability, the electrochemical processes upon lithium insertion and extraction were studied using electrochemical impedance spectroscopy (EIS) and cyclic voltammetry (CV), as shown in Fig. 3. The CV curves show the positions of different redox peaks for the reversible reactions between different valence states of vanadium. As shown in Fig. 3a, broader peaks were found in the CV curves of Si-V₂O₅@G, which correspond to the initial charge and discharge processes. A large oxidation peak was seen around 3.25 V after a shallow oxidation peak around 2.25 V, while reduction peaks at 2.6 V and 1.7 V were observed, respectively, for the 1st CV cycle of Si-V₂O₅. In the following cycles, only two major peaks appear: an oxidation peak at 3.80 V and a reduction peak at 1.75 V, corresponding to the oxidation state change of vanadium. Compared with Si-V₂O₅, Si-V₂O₅@G shows an oxidation peak at 3.3 V and a reduction peak at 2.4 V for the 1st cycle with reduction peak shifting toward 2.3 V for the 3rd cycle. Si-V₂O₅@G demonstrates much better cycling stability with each cycle as the CV curves keep the same shape and peaks, similar to those in the first cycles. To better understand the effects of Si doping in V₂O₅ on the cycling stability and electrochemical behaviors, the EIS analysis was conducted. The Nyquist plots of pure V₂O₅, V₂O₅@G, Si-V₂O₅@G and Si-V₂O₅@G after initial 3 cycles at a current density of 0.1C are shown in Fig. 3b, along with the corresponding equivalent electrical circuit for fitting. The total impedance was divided into four components, the contact resistance R₀, the electrode resistance R_e, the charge-transfer resistance R_ct and the Warburg diffusion impedance W_d. The diameter of the semicircle in the high frequency range represents the charge transfer resistance, R_ct, including the resistance of desolvation of the solvated Li ions in the electrolyte, Li-ion transport in the solid electrolyte interphase or cathode–electrolyte interphase layer, and the final acceptance of electrons in the host structures. Therefore, charge transfer resistance is significantly relevant to lithium insertion and extraction kinetics. As shown in Fig. 3a, the charge transfer resistance of Si-V₂O₅ decreases obviously from 46.9 Ω of pristine V₂O₅ to 28.2 Ω, and that of Si-V₂O₅@G decreases from 10.9 Ω of V₂O₅@G to 9.8 Ω, which indicates that the introduction of Si promotes the lithium reaction kinetics. In addition, both V₂O₅@G and Si-V₂O₅@G show smaller charge transfer resistance, which corresponds to the improved electron transport between V₂O₅ bi-pyramid units.

Fig. 3 (a) Voltammetry cycling analysis of Si-V₂O₅-based materials. (b) Electrochemical impedance spectroscopy (Nyquist plot) of Si-V₂O₅-based materials. (c–h) XPS analysis of V_2p and Si_2p of V₂O₅-based materials.

The XPS analysis (Fig. 3c–f) was performed to investigate the electronic structures and valence-state evolution of vanadium in the V₂O₅ structures as a function of Si doping and graphene incorporation. Fig. 3c and d present the high-resolution V_2p spectra of Si-V₂O₅@G and V₂O₅@G, respectively. For pristine V₂O₅@G, the V_2p3/2 peak is centered at 517.18 eV, which represents the state of V⁵⁺ in V₂O₅. Upon Si doping, this peak shifts remarkably to a lower binding energy of 516.35 eV in Si-V₂O₅@G. This significant negative shift indicates an increased electron density around the vanadium centers, suggesting a partial reduction of V⁵⁺ to V⁴⁺ species. The observed binding-energy shift can be attributed to electronic interactions induced by Si incorporation into the V₂O₅ lattice, where Si acts as an electron donor and modifies the local coordination environment of vanadium. Such an electronic redistribution facilitates charge transfer toward the V–O framework, leading to a lowered average oxidation state of vanadium. This effect is further stabilized by the graphene support, which enhances interfacial charge delocalization and suppresses electron localization.

Additionally, more evidence for successful Si incorporation is provided by the Si_2p XPS spectrum (Fig. 3g and h), which reveals a dominant Si–O bonding feature, confirming that Si is chemically integrated into the oxide framework rather than existing as a separate phase, such as SiO₂. Notably, the Si–O peak exhibits a slight binding-energy shift compared to crystalline SiO₂, which can be attributed to the distorted and partially amorphous local environment arising from graphene incorporation. This shift strongly suggests the formation of interfacial Si–O–V bonds within the Si-V₂O₅@G structure, further corroborating lattice-level doping rather than surface adsorption of Si. The XPS results demonstrate that Si doping effectively tailors the electronic structure of V₂O₅ by inducing partial vanadium reduction and strengthening interfacial bonding, which plays a significant role in enhancing the electrochemical stability and activity of the Si-V₂O₅@G nanostructure.

3.4 In operando XANES and EXAFS analysis

In operando X-ray absorption near-edge structure (XANES) spectroscopy was employed to investigate the (de)lithiation mechanism of V₂O₅-based cathodes by monitoring the changes in the electronic structure during electrochemical cycling.^13,35,48 Specifically, the evolution of the vanadium oxidation state was tracked during the charge and discharge processes, offering a critical insight into the structural reversibility of the V₂O₅-based cathodes. The binding energy shift observed in XANES spectra serves as a direct reflection of changes in the vanadium valence states and thus can be effectively used to evaluate the reversibility and stability of the cathode material structure. The comparative in operando XANES spectra of blank Li/V₂O₅@G and Li/Si-V₂O₅@G coin cells during the first and second discharge/charge cycles are presented in Fig. 4a. For both V₂O₅@G and Si-V₂O₅@G, the energy shifts (contour plots Fig. 4a and b) observed during lithiation and delithiation were largely symmetric, suggesting the good structural reversibility in both systems. However, the degree of the energy shift and the shape of the spectra are significantly different. Pristine V₂O₅@G exhibits more pronounced changes in its XANES spectra across both cycles compared to the Si-doped variant. These more substantial shifts suggest that the undoped V₂O₅ structure undergoes more significant electronic and structural fluctuations during cycling, which indicates greater instability and irreversibility. In contrast, the smaller energy shifts and smoother spectral evolution observed in Si-V₂O₅@G indicate a more stable electronic structure and enhanced structural robustness during lithium insertion and extraction. Fig. 4c and d display typical K-edge XANES spectra of the two cathodes during the first discharge process. V₂O₅@G exhibits substantial shifts in the absorption edge, indicating the notable changes in the oxidation state and coordination environment of vanadium during lithiation. This large shift implies a significant alteration in the electronic structure, which can be associated with more severe lattice distortion and lower cycling stability. In contrast, Si-V₂O₅@G showed moderate shifts in the absorption edge, implying less drastic changes in the local electronic structure and suggesting a more stable lithiation process, which is in consistent with the XPS analysis. Additionally, the Si-doped V₂O₅ demonstrates more symmetric spectral changes during charging and discharging compared to pristine V₂O₅, reinforcing the notion that Si doping contributes to improved structural reversibility and cycling stability.^49–52

Fig. 4 In operando X-ray Absorption Spectroscopy (XAS) characterization. Normalized X-ray Absorption Near Edge Spectroscopy (XANES) 2D contour plots of (a) Li/V₂O₅@G and (b) Li/Si- V₂O₅@G coin cells during 1st and 2nd discharge/charge cycling. Typical K-edge XANES spectra of (c) Li/V₂O₅@G and (d) Li/Si-V₂O₅@G coin cells during the first discharge/charge process. EXAFS 2D contour plots of (e) Li/V₂O₅@G and (g) Li/Si-V₂O₅@G coin cells during 1st discharge/charge cycling. Valence fitting results of the first discharge/charge process of (f) pristine V₂O₅@G and (h) Si-V₂O₅@G.

To further explore the local structural environment of vanadium in these two systems, extended X-ray absorption fine structure (EXAFS) analysis was performed during the initial lithiation and delithiation cycles. The EXAFS spectra provide detailed information about the short-range ordering and local atomic interactions, particularly the V–O–Li coordination that forms during lithium insertion. Fig. 4e presents the EXAFS 2D contour plots for V₂O₅@G during the initial charge/discharge process. A prominent peak at approximately 1.3 Å was observed, corresponding to the V–O–Li interaction. As the state of charge (SOC) increases, the intensity of this peak gradually rises, reflecting the enhanced V–O–Li interaction due to the progressive intercalation of Li⁺ ions into the V₂O₅ lattice. However, during the delithiation process, the peak intensity did not fully return to its original state, indicating incomplete recovery of the V–O coordination environment. This asymmetry in the EXAFS intensity between discharge and charge cycles strongly suggests that the V₂O₅ structure is not fully reversible after a complete lithiation–delithiation cycle. The inability to fully extract Li⁺ ions from the V₂O₅ structure and restore the original lattice contributes to long-term capacity fading and performance degradation. In contrast, Si-V₂O₅@G shows a highly reversible pattern in the EXAFS 2D contour plot in Fig. 4g under the same cycling conditions. Similar to the pristine sample, a V–O–Li peak at ∼1.3 Å was observed, and its intensity increased with SOC. However, unlike V₂O₅@G, the peak intensity in the Si-doped system returned to nearly its original state after delithiation. This symmetric behavior in EXAFS intensity suggests that the V–O–Li coordination in Si-V₂O₅@G is more reversible, implying improved structural integrity during cycling. The data provide strong evidence that Si doping effectively stabilizes the local V–O coordination environment, facilitating better Li⁺ insertion/extraction reversibility and reducing the degradation typically seen in pristine V₂O₅ systems.

To further quantify the valence state changes of vanadium during the first discharge and charge process, linear combination modeling (LCM) analysis³¹ was conducted on the in operando XANES spectra, and the results are displayed in Fig. 4f and h. For pristine V₂O₅@G, the initial vanadium valence was found to be approximately 71% V⁵⁺, 14% V⁴⁺, and 15% V³⁺, with an average valence state of 4.5. During the discharge process, the portion of V⁵⁺ gradually decreases while the V³⁺ and V⁴⁺ fraction increases simultaneously. Notably, the V⁴⁺ content remained relatively unchanged after discharging to 50% SOC, suggesting that the reduction of vanadium during lithiation primarily involves transitions between V⁵⁺ and V³⁺ states, i.e. V⁵⁺ + 2e⁻ → V³⁺. Upon charging, a partial recovery of V⁵⁺ was observed, but approximately 40% of vanadium remained constantly in the V⁴⁺ state. This incomplete return to the initial oxidation state suggests that the structure of V₂O₅@G does not fully recover, contributing to the loss of capacity over extended cycling. In contrast, Si-V₂O₅@G exhibits a remarkably different valence evolution profile (Fig. 4h). Initially, the vanadium was composed of 76% V⁵⁺ and 24% V³⁺, with no detectable V⁴⁺ and an average valence of 4.5. During the early stages of discharge (down to 80% SOC), the V⁵⁺ fraction decreases primarily through conversion to V³⁺, with no significant V⁴⁺ formation. From 80% to 20% SOC, V⁵⁺ (green line) transitioned into both V⁴⁺ (red line) and V³⁺ (blue line), with V⁴⁺ increasing more rapidly than V³⁺. As the SOC approached 10%, V⁵⁺ nearly disappears, and the portion of V⁴⁺ begins to decline while V³⁺ continues to increase. This evolution indicates a multistep redox mechanism in which vanadium transitions from V⁵⁺ → V³⁺ → V⁴⁺ and then to V³⁺ again as lithiation progresses. Notably, during the subsequent charging process, the valence state transitions are nearly symmetric to those observed during discharge, with V³⁺ converting back to V⁴⁺ and V⁵⁺ in reversed order. This symmetry suggests a highly reversible redox mechanism in Si-V₂O₅@G, highlighting its structural and electrochemical robustness.

In summary, the in operando XANES and EXAFS results, supported by LCM analysis, clearly demonstrate that Si doping significantly improves the structural stability and electrochemical reversibility of V₂O₅ during the (de)lithiation process. The undoped V₂O₅@G electrode undergoes more severe electronic structure changes and exhibits irreversible valence shifts, indicative of structural degradation. In contrast, Si-doped V₂O₅@G maintains a more stable and reversible electronic structure, as reflected in symmetric spectral features, moderate valence fluctuations, and recoverable local environments. These findings strongly support the conclusion that Si doping enhances both the structural stability and performance of V₂O₅-based cathodes, offering a promising pathway for developing high-energy density and long-life lithium-ion batteries.

4 Conclusions

We developed a method to dope silicon (Si) uniformly into the V₂O₅ framework, integrated with graphene sheets through a sol–gel process to form a 3D nanostructured composite. The optimized Si-doped V₂O₅/graphene nanocomposite (Si-V₂O₅@G) with 10 wt% Si and 2 wt% graphene exhibits exceptionally stable electrochemical performance, delivering a specific capacity of 392 mA h g⁻¹ and extraordinary cycling stability with capacity retention of 80% after 589 cycles, which shows significantly better stability than pristine V₂O₅@G. Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) analysis demonstrate that Si doping improves conductivity and stabilizes the V₂O₅ structure. The XRD analysis exhibits that both V₂O₅@G and Si-V₂O₅@G are in a non-crystalline (amorphous) structure when integrated with the graphene matrix. XPS results demonstrate that Si doping effectively tailors the electronic structure of V₂O₅ by inducing partial vanadium reduction and strengthening interfacial bonding. Furthermore, in operando XANES and EXAFS analyses reveal that Si doping significantly enhances the structural stability and electrochemical reversibility of V₂O₅ during the (de)lithiation process. While pristine V₂O₅@G exhibits asymmetric energy shifts and incomplete valence recovery—indicating poor reversibility, Si-V₂O₅@G shows symmetric spectral evolution and consistent V valence transitions. These results demonstrate that Si incorporation not only stabilizes the V–O framework but also promotes more complete Li-ion extraction and insertion, ultimately leading to improved cycling performance. The findings highlight the effectiveness of Si doping in optimizing the electrochemical behavior of V₂O₅-based cathodes.

Author contributions

Guangqi Zhu: conceptualization, data curation, in operando XAS data analysis, investigation, mechanism validation, and writing – original draft. Yadong Liu: conceptualization, material synthesis, electrochemical testing and in operando XAS experiment. Yikang Yu, Mohammad Behzadnia, Qi Zhang and Yi-Kai Lien: material synthesis, SEM analysis and writing. Chengjun Sun: in operando XAS experiment. Yuzi Liu: TEM analysis. Wenquan Lu and Mangilal Agarwal: materials analysis. Jian Xie: designed the project, manuscript writing, conceived the mechanism, funding acquisition, and supervision.

Conflicts of interest

There are no conflicts to declare.

Data availability

All the data supporting this article have been included in the supplementary information (SI). Supplementary information: Fig. S1 and S2, Table S1 and further experimental details. See DOI: https://doi.org/10.1039/d5ta06057e.

Acknowledgements

This research used resources of Center for Nanoscale Materials and Advanced Photon Source, both Office of Science User Facilities operated for the U.S. Department of Energy (DOE) Office of Science by Argonne National Laboratory, and was supported by the U.S. DOE under Contract No. DE-AC02-06CH11357, and the Canadian Light Source and its funding partners.

References

1. M. S. Whittingham, Chem. Rev., 2004, 104, 4271–4301.
2. Y. Wang, K. Takahashi, K. H. Lee and G. Z. Cao, Adv. Funct. Mater.,, 2006, 16, 1133–1144.
3. Y. J. Mai, X. L. Wang, J. Y. Xiang, Y. Q. Qiao, D. Zhang, C. D. Gu and J. P. Tu, Electrochim. Acta, 2011, 56, 2306–2311.
4. Y. Tang, X. Rui, Y. Zhang, T. M. Lim, Z. Dong, H. H. Hng, X. Chen, Q. Yan and Z. Chen, J. Mater. Chem. A, 2013, 1, 82–88.
5. J. Wang, N. Yang, H. Tang, Z. Dong, Q. Jin, M. Yang, D. Kisailus, H. Zhao, Z. Tang and D. Wang, Angew. Chem., Int. Ed. Engl., 2013, 52, 6417–6420.
6. Y. Liu, Q. Liu, Z. Li, Y. Ren, J. Xie, H. He and F. Xu, J. Electrochem. Soc., 2014, 161, A620–A632.
7. Y. Lu, H. Andersen, R. Wu, A. M. Ganose, B. Wen, A. Pujari, T. Wang, J. Borowiec, I. P. Parkin, M. De Volder and B. D. Boruah, Small, 2024, 20, e2308869.
8. B. D. Boruah, B. Wen and M. De Volder, Nano Lett., 2021, 21, 3527–3532.
9. A. Chen, C. Li, C. Zhang, W. Li and Q. Yang, J. Solid State Electrochem., 2022, 26, 729–738.
10. S. A. Hevia, J. Orive, F. Guzmán, E. Cisternas, F. Dietrich, R. Villarroel and J. Lisoni, Appl. Surf. Sci., 2022, 576, 151710.
11. X. Li, Y. Su, X. Lang, L. Li, C. Yao and K. Cai, Ionics, 2022, 28, 1511–1521.
12. X. Niu, N. Li, Y. Chen, J. Zhang, Y. Yang, L. Tan, J. Wu, L. Guo and Y. Zhu, Batter. Supercaps, 2023, 6, e202300238.
13. F. Coustier, Solid State Ionics, 1997, 100, 247–258.
14. F. Leroux, G. Goward, W. P. Power and L. F. Nazar, J. Electrochem. Soc., 2019, 144, 3886–3895.
15. F. Coustier, G. Jarero, S. Passerini and W. H. Smyrl, J. Power Sources, 1999, 83, 9–14.
16. M. Lira-Cantú and P. Gómez-Romero, J. Electrochem. Soc., 2019, 146, 2029–2033.
17. H. Liu and W. Yang, Energy Environ. Sci., 2011, 4, 4000–4008.
18. X. Ping, X. Pei, D. Zou, C. Wang, X. Li, C. Cao, H. Lei, C. Yang, Q. Cheng, W. Liu, X. Cao, M. Liu and Y. Wang, Electrochim. Acta, 2025, 530, 146297.
19. Y. Zhang, Z. Zou, J. Liu, S. Zhang and H. Zhang, Mater. Technol., 2020, 35, 887–895.
20. P. Chen, G. Zheng, G. Guo, Z. Wang, J. Tang, S. Li, Z. Wen, S. Ji and J. Sun, J. Alloys Compd., 2019, 784, 574–583.
21. K. Feng, Y. Li, C. Xu, M. Zhang, X. Yang, Y. Cheng, Y. Wang, L. Yang and S. Yin, Electrochim. Acta, 2023, 444, 142022.
22. T. Partheeban, T. Kesavan, M. Vivekanantha and M. Sasidharan, Appl. Surf. Sci., 2019, 493, 1106–1114.
23. D. B. Le, S. Passerini, A. L. Tipton, B. B. Owens and W. H. Smyrl, J. Electrochem. Soc., 2019, 142, L102–L103.
24. A. L. Tipton, S. Passerini, B. B. Owens and W. H. Smyrl, J. Electrochem. Soc., 2019, 143, 3473–3477.
25. P. Juntree, S. Siriroj, J. Padchasri, P. Songsiriritthigul and P. Kidkhunthod, J. Alloys Compd., 2025, 1016, 178961.
26. M. Wilhelm, R. Adam, A. Bhardwaj, I. Neumann, S. H. Cho, Y. Yamada, T. Sekino, J. Tao, Z. Hong, T. Fischer and S. Mathur, Adv. Eng. Mater., 2022, 25, 2200765.
27. Z. Shi, L. Wang, Y. Huang, X. Y. Kong and L. Ye, Mater. Chem. Front., 2024, 8, 179–191.
28. Y. Yu, G. Zhu, Q. Zhang, M. Behzadnia, Z. Yang, Y. Liu and J. Xie, ACS Appl. Energy Mater., 2024, 7, 11031–11037.
29. S. Liu, Z. Ding, T. Jiang, T. Chi, H. San, J. Cui, S. Liu, X. Li, X. Li and L. Zhang, J. Energy Storage, 2024, 84, 110822.
30. N. Aliahmad, P. K. Biswas, H. Dalir and M. Agarwal, Energies, 2022, 15, 552.
31. Q. Liu, Z. F. Li, Y. Liu, H. Zhang, Y. Ren, C. J. Sun, W. Lu, Y. Zhou, L. Stanciu, E. A. Stach and J. Xie, Nat. Commun., 2015, 6, 6127.
32. Q. Liu, Y. Liu, C.-J. Sun, Z.-f. Li, Y. Ren, W. Lu, E. A. Stach and J. Xie, Electrochim. Acta, 2014, 136, 318–322.
33. B. Wang, Y. Wang, B. Sun, P. Munroe and G. Wang, RSC Adv., 2013, 3, 5069–5075.
34. K. Omosako, T. Watanabe, S. Ogawa and I. Kanno, Sci. Rep., 2025, 15, 14477.
35. J. Wong, F. W. Lytle, R. P. Messmer and D. H. Maylotte, Phys. Rev. B, 1984, 30, 5596–5610.
36. C. Y. Lee, A. C. Marschilok, A. Subramanian, K. J. Takeuchi and E. S. Takeuchi, Phys. Chem. Chem. Phys., 2011, 13, 18047–18054.
37. X. Lai, J. E. Halpert and D. Wang, Energy Environ. Sci., 2012, 5, 5604–5618.
38. F. Xu, H. Zhang, J. Ilavsky, L. Stanciu, D. Ho, M. J. Justice, H. I. Petrache and J. Xie, Langmuir, 2010, 26, 19199–19208.
39. Y. Qian, A. Vu, W. Smyrl and A. Stein, J. Electrochem. Soc., 2012, 159, A1135–A1140.
40. M. X. Wang, Q. Liu, Z. F. Li, H. F. Sun, E. A. Stach and J. Xie, J. Phys. Chem. Lett., 2013, 4, 1484–1488.
41. Z. F. Li, H. Zhang, Q. Liu, L. Sun, L. Stanciu and J. Xie, ACS Appl. Mater. Interfaces, 2013, 5, 2685–2691.
42. Y. Liu, Y. Yu, F. Yang, G. Zhu, K. Yu, R. Kou, C. Sun, Y. Liu, J. Xu, C. Liu, C. Li, T. Liu, Y. Ren, W. Lu, R. Ferreira, P. Ferreira, Z. Zhang and J. Xie, Small, 2023, 19, e2206947.
43. J. Livage, Coord. Chem. Rev., 1998, 178–180, 999–1018.
44. S. Wang, Z. Lu, D. Wang, C. Li, C. Chen and Y. Yin, J. Mater. Chem., 2011, 21, 6365–6369.
45. M. G. Kanatzidis and S. P. Huang, J. Am. Chem. Soc., 2002, 111, 760–761.
46. J. Livage, Chem. Mater., 2002, 3, 578–593.
47. J. Cocciantelli, Solid State Ionics, 1995, 78, 143–150.
48. S. Passerini, Solid State Ionics, 1996, 90, 5–14.
49. M. Giorgetti, S. Passerini, W. H. Smyrl, S. Mukerjee, X. Q. Yang and J. McBreen, J. Electrochem. Soc., 2019, 146, 2387–2392.
50. A. N. Mansour, P. H. Smith, W. M. Baker, M. Balasubramanian and J. McBreen, Electrochim. Acta, 2002, 47, 3151–3161.
51. E. Almeida, Solid State Ionics, 2003, 160, 61–67.
52. A. M. Cao, J. S. Hu, H. P. Liang and L. J. Wan, Angew. Chem., Int. Ed. Engl., 2005, 44, 4391–4395.

---

Footnote

† Contribute equally to this work.

---