Vanadium pentoxide (purity > 99%) and 30% hydrogen peroxide were purchased from ChuanDong Chemicals Co. Magnesium nitrate hexahydrate (purity > 99%) and acetonitrile (AR > 98%) were purchased from Aladdin Co. CTAB (purity > 99%), while N-methyl pyrrolidone and polyvinylidene fluoride were purchased from Shanghai Maclin Biochemical Technology Co. The APC and Mg(TFSI)₂ electrolyte were purchased from Suzhou Duoduo Chemical Technology Co. All reagents and solvents are used directly without any treatment.

In detail, 2-mmol V₂O₅ powder was dissolved in 50 mL of deionized water and continuously stirred for 30 min at 25 °C. Then, 8-mL 30% hydrogen peroxide was added to the V₂O₅ aqueous solution stirring continuously for 60 min to remove the bubbles. Afterward, 0.1-mmol magnesium nitrate was dissolved in the above homogeneous solution, followed by adding 35-mL 2.59 mg mL⁻¹ GO aqueous solution and placing it in an ice bath ultrasound for 60 min. After being thoroughly mixed, it was then placed in a blast oven at 190 °C for 12 h. Subsequently, the as-prepared Mg_0.07V₂O₅·1.4H₂O/rGO was freeze-dried for 60 h. The GO aqueous solution was synthesized by the previous report work [35].

The surface morphology and microstructure of the as-prepared material were characterized by environmental scanning electron microscopy (ESEM, Thermo Fisher) and field emission transmission electron microscopy (FETEM, Talos). Elemental valence state and chemical bond compositions were characterized by X-ray photoelectron spectrometry (XPS, Thermo Fisher ESCALAB) with Al-Kα radiation as the excitation source. The internal elemental valence and chemical information were characterized by XPS after Ar ion etching, where the etched spot size and voltage were 1.5 mm and 3000 eV, respectively. X-ray diffractometer (XRD, Panalytical) with Cu-Kα radiation (λ = 0.15406 nm) was used to test the phase composition of the samples. Thermogravimetric analyzer (TGA, Mettler Toledo) was conducted to calculate the mass of structural water. The functional groups of graphene oxide and vanadium oxide were confirmed by Raman spectroscopy (HORIBA).

The active material, acetylene black conductor, and polyvinylidene fluoride binder were ground with the mass ratio of 7:2:1 as a homogeneous slurry, which was then coated on the carbon paper collector and vacuum dried at 80 °C for 12 h. Subsequently, the dried slurry was cut into disks with a diameter of 12 mm to obtain the electrodes, with a loading mass of 1.5-2.0 mg. Mg foils were cut into disks with a diameter of 14 mm and polished by metallographic sandpaper (800 mesh) to remove the magnesium surface passivation layer. 0.4 M APC-CTAB electrolyte was synthesized with the reported literature [23]. Half-cells were assembled by active carbon (AC) anode, Mg_0.07V₂O₅·1.4H₂O/rGO cathode, Whatman GF/D separator, and 0.5 M Mg(TFSI)₂ in AN electrolyte in Ar glove box (H₂O < 0.1 ppm, O₂ < 0.1 ppm). Likewise, the full cell was assembled by the same cathode and separator in Ar glove box (H₂O < 0.1 ppm, O₂ < 0.1 ppm), while the 0.1-mm Mg foil and 0.4 M APC-CTAB were used as the anode and electrolyte. The electrochemical performance of both half-cells and full cells was evaluated at room temperature. Galvanostatic charge-discharge tests were carried out on a Neware-CT8000 with a voltage window of -1.2-1.4 V vs. AC for half-cell and 0.2-2.0 V vs. Mg²⁺/Mg for full cell. Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) were evaluated on a CHI660E electrochemical workstation.

Vienna ab initio simulation package (VASP) [36] was utilized to carry out density functional theoretical (DFT) calculations in the Perdew-Burke-Ernzerhof (PBE) formulation with projected augmented wave (PAW) potentials [37]. The 2 × 7 × 2 K-points and 500 eV of cutoff energy were used to fully optimize the structures. For vanadium, we combined the DFT + U methods to resolve the electron correlation effect, the U value was set to 3.25 eV. Regarding the structure, convergence was achieved when each atomic force was reduced to less than 0.01 eV Å^-1. The Visualization for Electronic and Structural Analysis (VESTA) software was employed for the visualization and analysis of the crystal structure.

We employed a facile one-step hydrothermal method to synthesize rGO-substrated Mg_0.07V₂O₅·1.4H₂O nanobelts (MVOH/rGO) with dual defects, namely, interlayer P-Mg_d and surface O_d in V-O layers (Table S1). The XRD pattern shown in Fig. 1a displays a predominant diffraction peak at 6.1°, corresponding to the (001) plane, indicating a typical layered structure with an impressively large interlayer spacing of 14.5 Å [17,38]. The Rietveld refinement method was employed to analyze the further detailed crystal structure of MVOH/rGO (Table S2), which reveals a monoclinic structure with space group C2/m, and the refined parameters are calculated to be a = 14.19 Å, b = 3.77 Å, and c = 14.34 Å; α =  γ = 90°, and β = 83.9°. Notably, MVOH/rGO comprises anisotropic nanobelts several micrometers in length, and the finite width (50-70 nm) of the nanobelts effectively shortens the migration pathway of Mg²⁺ (Fig. 1b and c). The corresponding elemental mapping results (Figs. 1d and S1) confirm the presence of Mg, V, O, and C, aligning with the results of ICP-OES (Table S1). The P-Mg_d generates MgO₅ pyramids between the V-O layers (Figs. 1e and S2). These MgO₅ pyramids function as pins, fixing the lamellar structure of V₂O₅ in MVOH/rGO and thereby steadily restraining volume changes caused by Mg²⁺ insertion/extraction during discharging/charging. In addition, the P-Mg_d induces the polarization of V-O bonds [39]. According to ab initio calculations, this polarization leads to 3d orbital spitting, altering orbital hybridization in V-O bonds (Figs. 1f and S3a, b). Consequently, the valence band of V₂O₅ with P-Mg_d overlaps with its conduction band, resulting in substantially higher electronic conductivity of V₂O₅ in MVOH/rGO than that of V₂O₅·nH₂O (Figs. 1g and S3c, d). Furthermore, the P-Mg_d causes the reduction of trace V⁵⁺ to V⁴⁺, as verified by the average vanadium valence of + 4.93 in MVOH/rGO according to the chemical formula of Mg_0.07V₂O₅·1.4H₂O (Fig. S5). The 3d electrons of V⁴⁺ can easily hop to the vacant 3d orbitals of V⁵⁺, further increasing the electronic conductivity of V₂O₅ nanobelts in MVOH/rGO. The O_d in the V-O layers is generated by rGO and reduces the negative charge density of the layers. This leads to looser bonding between the V-O layers and inserted Mg²⁺, resulting in a considerable interlayer spacing of 14.5 Å between the V-O layers of V₂O₅ in MVOH/rGO. In comparison, V₂O₅·1.5H₂O (VOH) synthesized under similar conditions exhibits an interlayer spacing of 14.5 Å, while Mg_0.05V₂O₅·1.0H₂O (MVOH) synthesized under the same conditions with only P-Mg_d has an interlayer spacing as small as 13.5 Å (Fig. S6); this result shows the interlayer spacing enlargement effect of the O_d endowed by rGO [40]. The O_d also causes the reduction of superficial V⁵⁺ to V⁴⁺ at the interface between rGO and V₂O₅, further increasing the electronic conductivity of V₂O₅ in MVOH/rGO. According to the Raman analysis (Fig. S7), the signal intensity ratio of sp² graphite carbon (D band) and sp³ disordered carbon (G band) in MVOH/rGO (I_D/I_G = 1.14) is significantly higher than that in GO (I_D/I_G = 0.8), implying that the deoxidization of GO is realized by the redox reaction between carboxyl groups in GO and V⁵⁺ [41,44]. This is verified by the gas bubbles being observed after the hydrothermal synthesis of MVOH/rGO, which are attributed to the CO₂ gas produced by the oxidation of carboxyl groups in GO. Due to the excellent electronic conductivity of rGO, the V₂O₅ nanobelts are connected by the conductive rGO network, ensuring rapid electrochemical insertion/extraction of Mg²⁺.

To confirm the presence of dual defects (P-Mg_d and O_d) in MVOH/rGO, high-resolution transmission electron microscopy (HRTEM) was first employed (Fig. 1h and i). MVOH/rGO exhibited curved and discontinuous lattice fringes, indicating lattice disorder and suggesting the presence of lattice defects. Spherical aberration-corrected transmission electron microscopy (SAC-TEM) was employed to further investigate the lattice defects. As shown in Fig. 2a and b, V⁵⁺/V⁴⁺ vacancies are largely observed, implying the existence of O_d. The results of the pair distribution function analysis shown in Fig. S8 reveal faint and indiscernible signals at > 24 Å, further confirming the existence of V⁵⁺/V⁴⁺ vacancies in MVOH/rGO. The O_d in MVOH/rGO was clearly detected at 3468 and 3576 G by electron paramagnetic resonance (EPR) characterization (Fig. 2c). In contrast, no O_d was detected in MVOH without rGO, confirming that rGO produces O_d in MVOH/rGO. Furthermore, argon ion etching X-ray photoelectron spectroscopy (A-XPS) was employed to investigate the dual defects in MVOH/rGO (Fig. S9). High-resolution XPS spectra of O 1s at an etching depth of 0 nm showed three characteristic peaks at 533.5, 531.6, and 530.4 eV, representing adsorbed oxygen, O_d, and lattice oxygen, respectively [45,47]. Results showed that the ratio of O_d decreases from 29.5% to 24.9% as the etching depth increases from 0 to 20 nm (Fig. 2d-f), demonstrating the presence of more O_d on the surface of V₂O₅ nanobelts than in the bulk, as O_d is produced by composited rGO. Similarly, moving from the MVOH/rGO surface to 20 nm inside, the V⁴⁺ content decreased from 38.9% to 35.3% (Fig. 2g-i), aligning with the decreasing trend of O_d content. The V⁴⁺ content in the superficial area of MVOH/rGO is substantially higher than the stoichiometric content of V⁴⁺ according to the Mg_0.07V₂O₅ formula for MVOH/rGO and that in MVOH (Fig. S10), clearly confirming the coexistence of P-Mg_d and O_d in MVOH/rGO.

The Mg²⁺ storage performance of MVOH/rGO as the cathode material with dual defects was investigated. A coin cell was assembled using activated carbon as the anode and 0.5 M Mg(TFSI)₂ in acetonitrile as the electrolyte, with the activated carbon anode having a stable potential and simultaneously functioning as a reference electrode and counter electrode [48]. The cyclic voltammetry (CV) curve of the MVOH/rGO cathode exhibited the lowest and highest redox peaks at 1.5/2.7 V and maximum area within the potential window of 1.2-3.8 V (vs. Mg²⁺/Mg) at 1.0 mV s⁻¹ (Fig. 3a). Moreover, the peak potentials of these redox peaks are in high accordance with the galvanostatic charge/discharge (GCD) curves shown in Fig. 3b, indicating superior charge-transfer efficiency and Mg²⁺ storage capacity of the MVOH/rGO cathode compared to VOH and MVOH. Consequently, the MVOH/rGO cathode exhibits a superior discharge capacity of 197 mAh g⁻¹ at 0.02 A g⁻¹ (Fig. 3c), corresponding to 0.15 mol of Mg²⁺ inserted per mole of V₂O₅ [5]. The chemical reaction for magnesiation is as follows:

As the current density increases to 3.0 A g⁻¹, the MVOH/rGO cathode still exhibits a considerable rate performance of 64 mAh g⁻¹. In contrast, VOH without defects cannot discharge and charge with acceptable Coulombic efficiency (CE), which is due to the decomposition of electrolyte; according to the charging curve in Fig. 3d, the VOH cathode cannot be charged to higher than 3.5 V, and the electrolyte decomposed when the voltage is higher than 3.5 V. However, according to the initial GCD curves for MVOH and MVOH/rGO (Fig. S11), the decomposition of electrolyte was not observed for MVOH and MVOH/rGO cathodes. MVOH with only P-Mg_d shows increased electronic conductivity due to the overlap of its conduction and valence bands. However, its interlayer spacing is smaller than that of MVOH/rGO, resulting in an inferior rate capability (148/51 mAh g⁻¹ at 0.02/3.0 A g⁻¹; Fig. 3e) compared to that of MVOH/rGO. The GCD curves of MVOH display standard surface adsorption behavior without a redox plateau, indicating a capacity lower than that of MVOH/rGO (Fig. 3f). Owing to the pinning effect of P-Mg_d, the MVOH/rGO cathode presents remarkably similar GCD curves in the initial three cycles at 0.02 A g⁻¹, consistent with the CV results shown in Fig. S12, indicating the high reversibility of Mg²⁺ insertion/extraction (Fig. 3g). In addition, the MVOH/rGO cathode exhibits a notable capacity retention of 103 mAh g⁻¹ after 900 cycles at 0.5 A g⁻¹ with a capacity decay rate of only 0.014% per cycle, demonstrating its high cycle stability (Fig. S14a). In comparison, the VOH-O_d cathode (Fig. S13) exhibits a significantly lower cycle capacity of 57 mAh g⁻¹ at 0.5 A g⁻¹ and experiences much higher capacity decay rate of 0.04% per cycle (Fig. S14b); moreover, the MVOH cathode exhibits a substantial capacity decline (0.05% per cycle), remaining at a low capacity of 66 mAh g⁻¹ after 900 cycles at 0.5 A g⁻¹ (Fig. S14c). This comparison highlights the disability of P-Mg_d or O_d alone to improve the cathode cycle lifespan. Furthermore, even at a higher current density of 1.0 A g⁻¹, the MVOH/rGO cathode can still maintain an impressive capacity of 91 mAh g⁻¹ with an average CE of 99.6% after cycling for over 500 h (Figs. 3h and S15). The durability of the dual-defect MVOH/rGO cathode is higher than those of previously reported cathodes (Table S3), including PEDOT-intercalated V₂O₅ [10], O_d-V₂O_5-x [49], and NaV₂O₂(PO₄)₂F composited with rGO [50]. These results demonstrate the success of dual-defect engineering in prolonging the cathode cycle lifespan.

To investigate the Mg²⁺ storage mechanism of the dual-defect MVOH/rGO cathode, CV tests were conducted at scan rates of 0.1–0.8 mV s⁻¹ (Fig. 4a). The redox peaks exhibited a constant shape as the response currents expanding. The voltammetric response of the MVOH/rGO cathode follows the equation i =  av^b, where “i” represents the peak current, and a and b are adjustable constants [51,52]. Further, b = 1 indicates a capacitive-controlled process, while b = 0.5 indicates a diffusion-controlled process. For MVOH/rGO, the b values of peaks A and B at various scan rates are 0.96 and 0.81, respectively, indicating that the Mg²⁺ storage process is dominated by capacitive behavior (Fig. S16). Compared with VOH without defects and MVOH with only P-Mg_d, the contribution ratio of pseudocapacitive reactions in MVOH/rGO increases, which demonstrates the positive effect of dual defects in enhancing the capacitive contribution (Fig. S17). At various scan rates, the capacitive contribution in MVOH/rGO is > 72% (Fig. 4b), consistently surpassing those in VOH and MVOH. This underscores the superior ionic/electronic conductivity elicited by the dual defects. Owing to the dual defects, ion and electron diffusion in the MVOH/rGO cathode are rapid, and the charge-storage process is solely controlled by surface pseudocapacitive reactions, indicating fast Mg²⁺ insertion/extraction and the compatibility of the cathode with magnesium-metal anodes. The charge transport dynamics in MVOH/rGO were analyzed by electrochemical impedance spectrometry (EIS) measurements. As anticipated, the MVOH/rGO cathode exhibited the smallest charge-transfer resistance (R_ct), confirming its highest ionic/electronic conductivity (Figs. 4c and S18). The Mg²⁺ diffusion coefficient (${\text{D}}_{{\text{Mg}}^{2+}}$) of MVOH/rGO was as high as 5.8 × 10⁻¹⁴ cm² s⁻¹, substantially surpassing those of VOH (9.8 × 10⁻¹⁵ cm² s⁻¹) and MVOH (2.0 × 10⁻¹⁴ cm² s⁻¹), indicating the highest ionic conductivity of MVOH/rGO.

To further investigate the charge-storage mechanism, we conducted ex situ XRD, XPS, and TEM characterizations. Figure 4d and e shows the XRD patterns of the MVOH/rGO cathode at different discharge/charge states (A-F). No new phase was detected; however, the (001) diffraction peak slightly shifted during discharging/charging. In particular, the (001) diffraction peak shifted toward a lower angle for States A-C, indicating increased spacing between the V-O layers. During charging in States C-F, the (001) diffraction peak shifted back to its original location, indicating full recovery of the interlayer spacing after Mg²⁺ extraction. This demonstrates the high reversibility and cycle stability of the MVOH/rGO cathode with dual defects. In addition, the variation in the (001) peak location is only within 1°, verifying the high structural stability provided by the pinning effect of P-Mg_d. Figure 4f shows a slight increase in lattice spacing by 0.03 Å (from 0.210 to 0.213 nm) at the discharged State B, and the spacing further expands to 0.216 nm at fully discharged state with total variation ratio less than 3%. Subsequently, upon charging to charged State D, the interlayer spacing contracts to 0.214 nm and further decreases to 0.211 nm at full dicharged state. Such a negligible interlayer spacing variation further confirms that the combination of the pinning effect of P-Mg_d and the interlayer force relaxation effect of O_d can effectively restrain irreversible structural changes (Figs. S19 and S20). Figure 4g and h shows the high-resolution XPS spectra of V 2p and Mg 1s. The valence percentages of V⁴⁺ and V⁵⁺ in the pristine MVOH/rGO cathode are 27% and 73%, respectively. In the fully discharged state, the V⁴⁺ content increased to 56%, aligning with the reduction process accompanied by Mg²⁺ insertion. After full charge, the V⁴⁺ content decreased to 32%, which corresponds to the oxidation reaction during Mg²⁺ extraction. Compared to the pristine MVOH/rGO cathode, the slight increase in the V⁴⁺ content in fully charged MVOH/rGO indicates the presence of trace inserted Mg²⁺. The V⁴⁺ ratio in Fig. 4g is lower than that in Fig. 2g is due to the oxidation of partial surface V⁴⁺ during the electrode preparation process, which does not influence the relative variation trend of V⁴⁺ ration during charing/discharging. The Mg 1s signal at 1304.3 eV corresponded to the P-Mg_d species in MVOH/rGO, indicating that all pre-intercalated Mg²⁺ was pinned to the lamellar structure of MVOH/rGO (Fig. 4h). In the fully discharged state, a new signal appeared at 1304.9 eV, which was attributed to electrochemically inserted Mg²⁺. The presence of these two different chemical environments of Mg²⁺ is due to the distinct interaction modes of Mg²⁺ with the lattice O atoms [5]. After full charge, the signal intensity of inserted Mg²⁺ substantially decreased, with trace inserted Mg²⁺ remaining in the interlayer, which is in agreement with previously reported results [53,54].

To comprehensively evaluate the performance of the MVOH/rGO cathode, we assembled a Mg foil//MVOH/rGO full cell using 0.4 M APC-CTAB electrolyte because the magnesium-metal anode tends to passivate in the Mg(TFSI)₂ electrolyte (Fig. 5a) [55,56]. Leveraging the favorable ionic and electronic diffusion performance of MVOH/rGO, the Mg foil//MVOH/rGO full cell demonstrated a considerable initial discharge capacity of 123 mAh g⁻¹ at 0.02 A g⁻¹, operating at 1.1 V and exhibiting an energy density of 86 Wh kg⁻¹ (Fig. 5b). Even with an increased current density to 1.0 A g⁻¹, it retained a capacity of 29 mAh g⁻¹, showcasing superior rate performance (Fig. 5c). The dual-defect engineering of MVOH/rGO significantly restricted volume changes during magnesiation/demagnesiation, leading to an outstanding capacity retention of 75 mAh g⁻¹ after 400 cycles at 0.05 A g⁻¹, with a slight capacity decay of 0.027% per cycle and an average CE as high as 99.5% (Fig. S21). At a current density of 0.1 A g⁻¹, the cell maintained an impressive lifespan of 850 cycles, with a capacity retention ratio of 84% (Fig. 5d). Furthermore, it could continuously power an orange LED, showcasing its promising practical applications. In comparison with those of previously reported vanadium-based or carbon-substrated cathode materials, such as RFC/V₂O₅ [57], V₂O₅-PEDOT [23], PA-VOPO₄ [9], PVO [22], NaV₃O₈·1.69H₂O [58], MoS₂/graphene [59], Ag₂Se@C [60], and Ti₃C₂T_x@C [61], the cycle lifespan of the dual-defect MVOH/rGO cathode is significantly higher (Fig. 5e and Table S4).