Unlocking High-Valent Chalcogen Redox with Halide-Rich Electrolytes for High-Energy Lithium Batteries Authors: Yinlin Shen, Hao Zhang, Yuxuan Wu, Mingzi Sun, Doudou Feng, Jiaqian Qin, Zhengyu Shi, Bolong Huang, Jijian Xu Published: 2026 ABSTRACT AbstractElemental chalcogens (Ch) are promising positive materials for sustainable, high-energy lithium batteries, yet their chemistry is generally limited by the two-electron Ch0/Ch2− conversion below 2.5 V. This leaves the high-valent redox regime largely unexplored due to the instability of oxidized chalcogen species. Here, we report a halide-rich electrolyte design that enables reversible high-valent chalcogen redox in lithium batteries. By using soluble organic halide salts with asymmetric cations, the electrolyte provides active chloride or bromide anions to promote high-valent redox and stabilize oxidized intermediates, thereby enabling a redox-amphoteric selenium (Se) conversion pathway. This transition from reduction-only chemistry (Se2−/Se0) to three-electron conversion (Se2−/Se0/Se+) is evidenced by a distinct plateau at ~2.6 V, corresponding to the Se0/Se+ process. Consequently, the Li | |Se cell achieves a reversible discharge capacity of 980 mAh g−1 and a specific energy of 2003 Wh kgSe−1 with stable cycling performance over 200 cycles at 400 mA g−1. This strategy is further extended to sulfur and selenium sulfide materials, activating high-valent conversion. These results establish a potentially general route to access high-valent, multi-electron chalcogen chemistry, broadening the energy density limits for next-generation batteries.Subject terms: Batteries, Batteries Introduction Rechargeable lithium-ion batteries remain the cornerstone of electrified transportation and large-scale energy storage, yet the energy density of state-of-the-art intercalation-based positive electrodes is increasingly constrained by the single-electron redox limits of transition metals and the rigid structural requirements of host lattices1–3. Lithium-sulfur (Li | |S) and lithium-selenium (Li | |Se) conversion chemistries offer a compelling alternative, boasting high theoretical capacities (1675 mAh g−1 for S and 679 mAh g−1 for Se) by utilizing earth-abundant active materials3–5. Compared with S, Se possesses orders-of-magnitude higher electronic conductivity (Se: 1 × 10−5 S cm−1; S: 5 × 10−30 S cm−1), which intrinsically facilitates high-rate operation and active-material utilization. Despite these advantages, practical Li | |Se chemistry has remained largely confined to the classical two-electron Se0/Se2− conversion. Challenges such as polyselenide dissolution and shuttle, slow heterogeneous phase evolution, and unstable electrode/electrolyte interphases collectively prevent even this two-electron limit from being fully realized6,7. A transformative strategy to break this ceiling is to activate additional Faradaic processes of Se at higher potentials, thereby accessing previously untapped high-valent redox states and enabling genuinely multi-electron storage within the same elemental positive electrode. Realizing such high-valent redox chemistry is fundamentally a challenge of simultaneous activation and stabilization8. The oxidation of Se0 to Sen+ (n = 1, 2, or 4) typically yields electrophilic selenium halides (e.g., Se2Cl2, SeCl2, SeCl4) or complexed cationic species that are prone to hydrolysis and irreversible side reactions in conventional electrolytes. Fundamentally, accessing these high-valent Se states requires a charge-compensating anion/ligand capable of activating Se0/Sen+ oxidation and stabilizing the resulting high-valent intermediates through coordination and speciation control. While high-valent chalcogen chemistry has been recently explored in aqueous zinc (Zn), aluminum (Al), and magnesium (Mg) batteries, the electrolyte plays a critical role in providing strongly coordinating anions to activate high-valent states and to suppress deleterious side reactions to ensure reversibility. In Zn-based systems, strategies such as replacing water with organic media9, employing hydrogels10 or ionic-liquid-based electrolytes11,12, operating at subzero temperatures (down to −45 °C)13, or adopting superconcentrated salts10,12,14 have been used to mitigate hydrolysis while maintaining sufficient ionic transport. In these electrolytes, Cl− (or ClO4−) acts as the key charge-compensating anion that activates chalcogen oxidation and stabilizes the resulting oxidized products. Analogously, in Al15–17 and Mg batteries18, ionic-liquid and organic electrolytes are used to avoid hydrolysis and, critically, the strongly Lewis acidic chloroaluminate species (e.g., Al2Cl7−, AlCl4−) promote oxidation of chalcogens to higher valent states (S0/S+, Se0/Se+, Te0/Te4+). Recent advances, such as the 3.6 V anode-free sodium-sulfur (Na | |S) battery utilizing a thionyl chloride (SOCl2)-based electrolyte, have demonstrated the potential of high-valent S0/S4+ (S/SCl4) chemistry19. However, the reliance on aluminum chloride (AlCl3)/SOCl2 raises practical concerns regarding corrosivity and handling. Another approach involves the direct introduction of lithium bromide (LiBr) powder into the sulfur positive electrode13. Yet, embedding electronically insulating LiBr significantly reduces overall electronic conductivity and limits the high-rate performance of the Li | |S/LiBr cell, as the S/LiBr reaction proceeds via a solid-solid pathway with a large kinetic barrier. These limitations highlight an urgent need for a rational electrolyte design that provides high and mobile halide activity to activate and stabilize the high-valent Se0/Se+ redox couple. Here, we address this challenge by designing a nonaqueous lithium electrolyte incorporating organic halide salts with asymmetric cation architectures (Tributylmethylammonium chloride, TBMACl; Tributylmethylammonium bromide, TBMABr). This electrolyte design delivers high halide solubility and effective halide activity, unlocking a reversible high-valent Se0/Se+ conversion in a nonaqueous lithium system, featuring a distinct ~2.6 V plateau. A redox-amphoteric Se conversion is thus achieved, transitioning from a conventional reduction-only mechanism to a three-electron Se2−/Se0/Se+ pathway in Li | |Se cells. Comprehensive analysis, combined with ex situ and in situ spectroscopic characterizations, reveals the formation of Se2Cl2 or Se2Br2 at high voltages and their reversible conversion. As a result, the optimized Li | |Se cells deliver a reversible discharge capacity of 980 mAh g−1, retaining a reversible capacity of 633 mAh g−1 after 200 cycles. Importantly, this electrolyte concept generalizes to activate high-valent reactions in Li | |S and Li | |SeS2, delivering pronounced gains in discharge capacity. Results Reversible redox-amphoteric Se conversion chemistry Se is uniquely positioned to support ambipolar redox chemistry (Fig. 1a). On the one hand, its strong electron affinity (first electron affinity, −195 kJ mol−1) enables efficient uptake of electrons and storage of metal cations (Li, Na, K, Mg) through conversion to metal selenides20–22. On the other hand, Se also exhibits a comparatively low ionization energy (first ionization energy, 940.9 kJ mol−1), implying that oxidation to high-valent cationic selenium species (Se+, Se2+, Se4+) is accessible if the electrolyte can effectively activate and stabilize the oxidized states. When an organic electrolyte simultaneously supplies electrochemically active Li+ and coordinating halide anions (Cl−/Br−), Se can therefore be driven into a redox-amphoteric conversion manifold: during discharge, Se is reduced by two electrons and couples with Li+ to form Li2Se; during charge, Se is oxidized to high-valent states that are stabilized by halides, resulting in the formation of cationic selenium halides (e.g., Se2Cl2/Se2Br2) within a defined potential window. The optimized molecular structures and electrostatic potential (ESP) mapping of Se8, Se2Cl2, and Se2Br2 (Supplementary Fig. 1) indicate that halogenation disrupts the Se8 ring framework. In the resulting Se2Cl2 and Se2Br2, Se serves as the electron-deficient center, while the more electronegative halogen atoms exhibit localized electron density; notably, the greater polarization observed in Se2Cl2 suggests that Cl− is a more effective charge-compensating/coordination anion for activating the Se0/Se+ process. Importantly, combining the conventional reductive Se0/Se2− conversion with an oxidative Se0/Se+ reaction is expected to increase the theoretical specific capacity by ~50% relative to the two-electron Se0/Se2− chemistry (Fig. 1b). Furthermore, the elevated potential (~2.6 V) of the Se0/Se+ reaction significantly enhances the theoretical energy density by >50%. Reversible redox-amphoteric Se conversion chemistry. a Schematic showing the Se redox-amphoteric conversion process. b Theoretical projection of capacity and energy density enhancements enabled by activating the additional Se0/Se+ redox chemistry compared to conventional Se0/Se2− conversion. c Discharge-charge profiles of a Li | |Se cell at 400 mA g−1 and d corresponding differential capacity (dQ/dV) curve during the charging process, identifying the distinct Se0/Se2− and Se0/Se+ redox peaks. Blue and red regions represent the low-valent and high-valent reaction regions, respectively. To evaluate this three-electron Se2−/Se0/Se+ conversion chemistry, we assembled Li | |Se cells using a lithium metal negative electrode and a Se/Ketjen black composite positive electrode (denoted as Se/KJ). The cells were operated in a conventional electrolyte (1 M lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) in a Dimethoxyethane/Dioxolane (DME/DOL) mixture (1:1, v/v), with 5 wt.% lithium nitrate (LiNO3) added as an additive, denoted as Baseline electrolyte) and in a designed chloride-containing electrolyte (1 M LiTFSI + 1 M lithium bis(fluorosulfonyl)imide (LiFSI) + 0.5 M LiNO3 + 1 M TBMACl in 1,3-Dimethyl-2-imidazolidinone (DMI), denoted as Designed-Cl electrolyte). The Designed-Cl electrolyte exhibits good compatibility with Se2Cl2, the charged product formed during high-valent Se redox (Supplementary Fig. 2). The highly porous, conductive KJ framework is expected to enhance electronic percolation within the Se/KJ electrode and physically confine both charged products (Se2Cl2) and discharge intermediates (lithium polyselenides). This confinement strategy effectively mitigates active-material migration to the Li electrode, thereby suppressing the shuttle effect and minimizing capacity loss. Structural characterization confirms effective Se confinement: after mechanical ball milling, the characteristic Se diffraction peaks remain visible but undergo obvious broadening (Supplementary Fig. 3a), indicative of a reduction in Se domain size. After thermal treatment at 260 °C, Se reflections are barely detectable in the Se/KJ composite, indicating that Se successfully infiltrated into the KJ pore network. Scanning electron microscopy (SEM) further corroborates this inference: pristine Se consists of aggregated ~1 μm particles (Supplementary Fig. 3b), whereas no exposed Se particles are observed in Se/KJ (Supplementary Fig. 3c, d). These observations are consistent with efficient encapsulation of Se within the porous carbon matrix. Additional Se2Cl2 adsorption tests (Supplementary Fig. 4), post-cycling SEM/energy-dispersive X-ray spectroscopy (EDS) analysis of the Se/KJ electrode (Supplementary Fig. 5), and SEM/EDS/X-ray photoelectron spectroscopy (XPS) characterization of the Li electrode (Supplementary Figs. 6 and 7) further demonstrate that the synergistic combination of the KJ host and tailored electrolyte formulation effectively suppresses active-species diffusion and the shuttle effect, thereby mitigating potential corrosion or passivation of the Li negative electrode. The electrochemical activation of high-valent states was investigated via a charge-first protocol, which provides direct evidence for electrolyte-dependent accessibility of the oxidative Se0/Se+ reaction. When charged from the initial state (Supplementary Fig. 8), the Li | |Se cell in Baseline electrolyte exhibits no capacity. In stark contrast, the cell in Designed-Cl electrolyte delivers a clear charge capacity of 180 mAh g−1, demonstrating that the Se0/Se+ oxidation pathway is uniquely activated by the high chloride activity of the designed electrolyte. The corresponding differential capacity (dQ/dV) analysis (Supplementary Fig. 8d) assigns this oxidation to a distinct process at 2.61 V. Upon the subsequent first discharge, the Li | |Se cells in the Baseline and Designed-Cl electrolytes delivered specific capacities of 591 and 1050 mAh g−1, respectively, corresponding to gravimetric specific energies of 1218 and 2003 Wh kgSe−1 based on Se mass, or 584.6 and 961.4 Wh kgelectrode−1 based on the total positive electrode mass, respectively; thus, the Li | |Se cell in Designed-Cl electrolyte increases discharge capacity and specific energy by 77% and 64%. The carbon cloth (CC) current collector contributes negligible additional capacity (Supplementary Fig. 9). The advantage persists upon cycling: in the 2nd/3rd cycles, the Li | |Se cell in Baseline electrolyte delivers discharge capacities of 503/482 mAh g−1, whereas the Li | |Se cell in Designed-Cl electrolyte maintains capacities of 980/958 mAh g−1, indicating markedly improved reversibility and stability in the halide-containing electrolyte. The 2nd-cycle voltage profile in Designed-Cl electrolyte (Fig. 1c) exhibits high Coulombic efficiency, consistent with a highly reversible three-electron conversion process. Moreover, the charging dQ/dV curve (Fig. 1d) displays well-resolved oxidation features at 2.10 V, 2.24 V, and 2.63 V: the lower-potential peaks (2.10 and 2.24 V) are attributed to the Se2−/Se0 conversion, while the high-potential peak at 2.63 V corresponds to the Se0/Se+ process. These distinct electrochemical signatures corroborate the successful activation of a three-electron Se2−/Se0/Se+ reaction manifold, fundamentally enabled by the halide-anion chemistry of the designed electrolyte. The three-electrode Li | |Se measurements (Supplementary Fig. 10) further confirm the reversible three-electron Se2−/Se0/Se+ conversion. Mechanistic origin of multivalent Se conversion The conversion mechanism of the Se electrode was elucidated using various characterization techniques. To investigate the structural evolution during electrochemical cycling, in situ Raman spectroscopy was performed utilizing a customized cell configuration as shown in Supplementary Fig. 11. Figure 2a–c present the galvanostatic charge-discharge profile of the Li | |Se cell at 400 mA g−1 and the corresponding Raman spectra, respectively. Upon discharge, the Raman band at 233 cm−1, characteristic of Se−Se vibrations in elemental Se, progressively attenuates and becomes undetectable by 2.0 V, indicating depletion of the Se electrode. Further discharge to 1.7 V gives rise to an additional feature at 250 cm−1, which we assign to Li−Se interactions on the basis of reference spectra of Se and Li2Se powders (Supplementary Fig. 12). Notably, the apparent “silent” region between 2.0 and 1.7 V is consistent with the formation of soluble polyselenide intermediates during discharge. Dissolution of these transient species into the electrolyte reduces their local concentration at the interrogated electrode surface and consequently suppresses their Raman detectability. At full discharge to 1.5 V, the associated band of Li−Se reaches maximal intensity, consistent with accumulation of Li2Se as the dominant discharged phase. Reaction mechanism elucidated by spectroscopic characterizations. a Voltage curve of the Li | |Se cell at 400 mA g−1 and 25 °C used for in situ Raman test. b Contour map of Raman patterns ranging from 200 to 360 cm−1. c Representative Raman spectra of the Se electrode at selected potentials. d Ex situ XPS spectra of Se electrode collected at selected potentials during the first cycle (initial, discharge to 1.5 V, and charge to 3.3 V) at a specific current of 400 mA g−1, confirming the evolution of chemical states across Se2−/Se0/Se+. During the subsequent charge, the Raman evolution reverses, reflecting re-oxidation of reduced selenium species. The Li−Se feature gradually diminishes, while the Se−Se vibration recovers. At 2.4 V (Fig. 2c), the spectrum is dominated by the Se−Se signature attributable to the Se8 structure. Upon further charging, the Se−Se band undergoes a discernible red shift from 233 to 231 cm−1, and an additional band emerges at 326 cm−1, which is characteristic of Se−Cl bonding and consistent with the reference Raman spectrum of Se2Cl2 dissolved in the Designed-Cl electrolyte (Supplementary Fig. 13). The red shift of the Se−Se vibration is consistent with a shortened Se−Se bond in Se2Cl2 relative to Se8 (Se−Se in Se8: 2.36 Å; Se−Se in Se2Cl2: 2.23 Å; Supplementary Fig. 1), supporting assignment of the charged-phase species to selenium chlorides. The concomitant appearance of the Se−Cl mode and the Se−Se red shift thus provides direct spectroscopic evidence for electrolyte-enabled formation of Se2Cl2 at the positive electrode upon charging. Importantly, the presence of these bands indicates that Se2Cl2 remains stabilized within the electrode. The ultraviolet−visible (UV−vis) spectrum (Supplementary Fig. 14) of the solution obtained by immersing the fully charged electrode in acetonitrile further shows a characteristic absorption peak of Se2Cl2 at 301 nm23, confirming the formation of Se2Cl2 in the charged state. On the subsequent discharge, the Raman signatures evolve in a stepwise manner: the Se−Cl band disappears concomitantly with a blue shift of the Se−Se vibration (consistent with reduction of Se2Cl2), followed by re-emergence of the Se−Se band associated with Se8, and ultimately replacement by the Li−Se feature as discharge proceeds. This stepwise spectral progression is consistent with an electrochemical reaction manifold that traverses Se+/Se0/Se2− during cycling. To further examine the chemical states of the selenium species, XPS was performed on Se electrodes collected at representative states of charge (Fig. 2d). To minimize artefacts arising from air/moisture sensitivity of the electrochemically generated selenium species, all electrodes were sealed in the glovebox and transferred to the XPS chamber using an inert-atmosphere transfer holder. The pristine Se electrode exhibits a characteristic Se 3d spin-orbit doublet corresponding to elemental Se (Se0), with the Se 3d5/2 and Se 3d3/2 components located at 55.7 eV and 56.6 eV, respectively. After full discharge to 1.5 V, the Se0 doublet vanishes and is replaced by an additional Se 3d doublet at 55.3 eV (3d5/2) and 56.1 eV (3d3/2), consistent with the −2 valence position of Se and therefore with the formation of Li2Se as the discharged product. Upon charging to 3.3 V, the Se 3d features shift to 55.1 eV and 55.9 eV (aligned with the Se+ state)24. To support this assignment, a reference spectrum was obtained from a synthesized Se2Cl2 drop-cast onto carbon cloth. The Se 3d doublet of the fully charged electrode coincides quantitatively with that of the Se2Cl2 reference, providing strong evidence for the formation of Se2Cl2 during the charging process. Notably, no residual Se0 contribution is detectable in the charged electrode, indicating that selenium undergoes essentially complete oxidation under the applied charging conditions. Furthermore, XPS analysis (Supplementary Fig. 15) of the Se electrodes collected at the charged states after the 5th and 10th cycles consistently shows clear Se+ signals, further confirming the highly reversible and sustained formation of high-valent Se species over extended cycling. Synchrotron-based X-ray absorption spectroscopy (XAS) was employed to track the evolution of the local chemical environment of Se in the Se/KJ electrode during electrochemical cycling. For comparison and spectral assignment, reference spectra of elemental Se, Na2Se, and Se2Cl2 were collected under identical conditions. In the Se K-edge X-ray absorption near-edge structure (XANES) profiles recorded during the first discharge (Fig. 3a), the absorption edge exhibits a progressive blue-shift as the cell voltage decreases, reaching its maximum shift at 1.5 V. At full discharge, the edge position nearly coincides with that of the Na2Se reference, indicating the formation of Se2− species and the generation of Li2Se as the discharge product. During the subsequent charge (Fig. 3b), the Se K-edge exhibits a non-monotonic evolution: it first red-shifts from 1.5 to 2.4 V and then blue-shifts from 2.4 to 3.3 V. At 2.4 V, the edge position overlaps with that of the Se reference, consistent with regeneration of Se0. Upon further charging to 3.3 V, the edge position becomes indistinguishable from that of the Se2Cl2 reference, identifying Se2Cl2 as the charged product and thereby supporting a stepwise Se2−/Se0/Se+ reaction sequence during charge. The apparent blue-shift accompanying Se0/Se+ oxidation is associated with the distinct bonding and electronic structure of Se2Cl2, and is consistent with the XPS observation that the Se0/Se+ process is accompanied by a shift of Se 3d binding energies to lower values (Fig. 2d). Reaction pathway established by XAS and theoretical calculations. a–c Se K-edge XANES spectra of Se at different states of charge during the first cycle at a specific current of 400 mA g−1. Wavelet-transformed Se K-edge EXAFS of d initial Se, e fully discharged Se (1.5 V), and f fully charged Se (3.3 V) during the first cycle at 400 mA g−1. Theoretical calculations of reaction pathways for the Se oxidation/reduction process in (g) chlorine- and (h) bromine-containing electrolytes. To gain deeper insights into the local structural evolution, extended X-ray absorption fine structure (EXAFS) was analyzed for the pristine state (Fig. 3d), fully discharged (1.5 V; Fig. 3e) and fully charged (3.3 V; Fig. 3f) states. To disentangle backscattering contributions, we performed wavelet transform (WT) analysis on the k3-weighted EXAFS data, visualized as two-dimensional intensity maps in R and k space. The pristine Se electrode shows a dominant WT maximum at R = 2.10 Å, corresponding to Se−Se scattering. At 1.5 V, an additional WT feature emerges at R = 2.16 Å, consistent with Li−Se scattering and supporting the formation of Li2Se. At 3.3 V, the WT maximum shifts to R = 1.97 Å, which we assign to Se−Cl scattering in Se2Cl2. This is consistent with the refined structural parameters (Supplementary Fig. 1) showing that the Se−Cl bonds in Se2Cl2 are shorter than the Se−Se bonds in Se8. Based on the cumulative spectroscopic and electrochemical evidence, the overall reaction sequence of the Li | |Se cell—comprising the high-valent Se+/Se0 and conventional Se0/Se2− redox couples—can be summarized as follows. In particular, the Se electrode undergoes two steps as shown below. For the Li electrode: The overall reaction is shown as: In the chloride-containing electrolyte (Fig. 3g), the calculated redox potentials are 2.64 V and 1.84 V versus Li+/Li, in close accord with the experimentally observed discharge plateaus (Fig. 1c). Upon substituting Cl− with Br− (Fig. 3h), the calculated potential for the Se0/Se+ reaction shifts upward to 2.78 V. This slightly higher potential suggests a larger activation barrier for Br−-assisted Se0/Se+ conversion relative to Cl−. Collectively, these results indicate that introducing Cl− into the electrolyte not only modulates the solvation structure but also improves the overall reaction kinetics of the Li | |Se system. Design of non-aqueous electrolyte The oxidation of elemental Se is kinetically demanding. Consequently, accessing the Se0/Se+ reaction requires an electrolyte that provides high-activity, coordinatively competent charge-compensating anions within a sufficiently wide stability window. Anion screening is primarily governed by nucleophilicity and ionic size (Fig. 4a): more nucleophilic, smaller anions are capable of more effectively stabilizing the highly electrophilic Se⁺-based intermediates through stronger ion pairing/coordination25,26. Consequently, conventional anions (bis(trifluoromethanesulfonyl)imide (TFSI−), trifluoromethanesulfonate (OTf−), hexafluorophosphate (PF6−)) are generally unsuitable because they are bulky and weakly nucleophilic. In contrast, halides (Cl−, Br−) possess the requisite combination of compact size and higher nucleophilicity for effective coordination. Activating the high-valent I0/I+ redox chemistry of I2 by introducing Cl⁻ or Br− into the electrolyte has been widely reported in aqueous Zn batteries27–31, translating this strategy to non-aqueous media remains a challenge. This is primarily due to the very limited solubility of conventional inorganic halides (LiCl, LiBr) in organic carbonates or ethers, which prevents maintaining high concentrations and high activity of free Cl−/Br− in the electrolyte. We address this by using organic halide salts, whose cation structures tune ion pairing and dissociation (Supplementary Fig. 16). In particular, quaternary ammonium halides with asymmetric cations (Tributylmethylammonium, TBMA+; Trimethylbutylammonium, TMBA+) exhibit lower ion-pair dissociation energies than symmetric tetraalkylammonium analogues (Tetramethylammonium, TMA+; Tetraethylammonium, TEA+; Tetrapropylammonium, TPA+; Tetrabutylammonium, TBA+) (Fig. 4b and Supplementary Fig. 17). This lower dissociation energy is expected to promote ion separation and increase the effective activity of halide anions in solution, thereby providing a more favorable chemical environment for Se0/Se+ activation. Non-aqueous electrolyte design enabling high-valent redox. a Comparison of anion size and nucleophilicity for various anions. b Calculated dissociation energies of symmetric and asymmetric chloride salts. c Relative dielectric constants and viscosities of various solvents. d ESP mapping of DMI molecular. Cyan balls = C, Blue balls = N, Silver-white balls = H, and Red ball = O. e, f Raman spectra and g 1H NMR spectra of pure solvent and electrolytes. h 35Cl NMR spectra of designed chloride-containing electrolyte. i Snapshot of the designed chloride-containing electrolyte after MD simulations and three typical local solvation structures of Li+. Purple balls = Li+ and green balls = Cl−. Grey sticks = C, blue sticks = N, red sticks = O, yellow sticks = S, cyan sticks = F, and white balls = H. j RDF results (g(r), solid line) and coordination number (dotted line) of Li−Cl and Li−O. k The MSD and simulated diffusion coefficient of Li+ and Cl−. Solvent selection is equally critical for achieving both high salt solubility and controlled intermediate transport32–34. Compared with ether solvents (DME, Diglyme (DEGDME), DOL), DMI has a high dielectric constant (ε = 37.6), which promotes dissolution of organic halide salts and ion separation, while its moderate viscosity can suppress dissolution and long-range diffusion of oxidized products (Se2Cl2) and lithium polyselenide intermediates, mitigating the shuttle effect and capacity loss. The molecular structure and electrostatic potential distribution of DMI are shown in Fig. 4d. Solubility measurements of a series of organic chloride and bromide salts in DME and DMI (Supplementary Fig. 18) confirm this design rationale: owing to their low dissociation energies combined with the strong solvating capability of DMI, TBMACl and TBMABr reach solubilities exceeding 3 M in DMI, providing a practical route to halide-rich organic electrolytes suitable for unlocking high-valent Se+/Se0 chemistry. The solvation structures of the designed halide-containing electrolytes were elucidated through an integrated approach combining Raman spectroscopy, nuclear magnetic resonance (NMR) spectroscopy and molecular dynamics (MD) simulations. Raman spectroscopy (Fig. 4e, f and Supplementary Figs. 19–21) provides a direct probe into the local coordination environment of the DMI solvent. In neat DMI, the symmetric C=O stretching band is centered at 1690 cm−1. Upon addition of 1 M LiTFSI or 1 M LiFSI, this vibration upshifts to 1696 cm−1, indicative of a strong Li+−O(C=O) coordination35. By contrast, adding 1 M TBMACl to DMI produces only a modest shift of the C=O band to 1692 cm−1, indicative of comparatively weak TBMA+−solvent interactions. Notably, the Raman spectrum of TBMACl/DMI exhibits no additional discernible bands beyond those of DMI in the probed region, suggesting that the organic halide salt does not introduce prominent additional vibrational features and that its primary role is to modulate ionic speciation rather than solvent structure. In comparison, LiTFSI/DMI and LiFSI/DMI solutions display characteristic anion bands associated with Li+ coordination: a feature at 740 cm−1 attributable to Li+-coordinated TFSI− and a feature at 720 cm−1 attributable to Li⁺-coordinated FSI−, respectively35,36. In both solutions, a signal near 778 cm−1 is observed, which can be assigned to the S−N−S vibration common to TFSI⁻ and FSI⁻. In the fully formulated Designed-Cl electrolyte, the DMI carbonyl band shifts further to a higher wavenumber, indicating a stronger perturbation of the C = O group. 1H NMR (Fig. 4g and Supplementary Figs. 22, 23) provides complementary evidence for Li+−solvent coordination. Upon adding LiTFSI or LiFSI to DMI, the resonances of the −CH2− and −CH3 protons in DMI shift upfield, indicating increased local electronic shielding. This trend is consistent with the coordination of the carbonyl (C=O) group in DMI to Li+, which redistributes electron density within the DMI framework. The upfield shifts become more pronounced in the higher-salt-concentration Designed-Cl and Designed-Br electrolytes, consistent with strengthened Li+−DMI interactions. In addition, clear halogen signals are detected in the 35Cl NMR spectrum (Fig. 4h) of the Designed-Cl electrolyte and in the 81Br NMR spectrum (Supplementary Fig. 23) of the Designed-Br electrolyte, supporting the presence of highly available halide species and thus high halide activity in these electrolytes. Classical molecular dynamics (MD) simulations on the Designed-Cl electrolyte were conducted with the resulting data shown in Fig. 4i–k and Supplementary Data 1. Most Li+ ions exhibit a stable coordination number (CN) of 4 (Fig. 4i), but with heterogeneous ligand identities that reflect a competitive solvation landscape. In solvation structure I, Li+ is coordinated exclusively through oxygen donors from DMI and TFSI−. In solvation structure II, one Cl− is incorporated into the first coordination shell, while the remaining ligands are contributed by DMI and anions such as TFSI−, FSI−, and NO3−. We also observe a less abundant but chemically notable three-coordinate motif comprising one DMI molecule and two Cl− ligands (CN = 3). The extensive participation of anions in the Li+ solvation sheath suggests an anion-rich solvation sheath, which is expected to facilitate interfacial Li+ desolvation by weakening purely solvent-dominated coordination. Radial distribution function (RDF) analysis (Fig. 4j) defines the first-shell peak positions for Li−Cl, Li−O(DMI), Li−O(TFSI−), Li−O(FSI−), and Li−O(NO3−) at 1.90 Å, 1.89 Å, 1.95 Å, 1.96 Å, and 1.69 Å, respectively. The corresponding coordination numbers are 0.62, 2.24, 0.53, 0.23, and 0.25, indicating that DMI dominates the first solvation shell while a large number of anionic contributions—especially from Cl− and sulfonylimide species—remain integral to the local coordination environment. Finally, ion-transport properties were evaluated from MD-derived mean-squared displacement (MSD) analyses (Fig. 4k). The extracted tracer diffusion coefficients (also termed self-diffusion coefficients) from the MSD of individual ionic species are 5.36 × 10−6 cm2 s−1 for Li+ and 6.15 × 10−6 cm2 s−1 for Cl−, revealing relatively fast dynamics in the designed electrolyte. The tracer diffusion coefficients were calculated to quantitatively reveal the intrinsic dynamic behavior of Li+ and Cl− in the designed electrolyte under equilibrium conditions, free from interference from external fields or concentration gradients. Li plating/stripping measurements in Li | |Li symmetric cells (Supplementary Fig. 24) demonstrate stable cycling in the Designed-Cl electrolyte for over 320 h. Comparison with an otherwise identical electrolyte formulation without LiNO3 highlights the critical role of LiNO3 in stabilizing the lithium negative electrode. In addition, linear sweep voltammetry (LSV, Supplementary Fig. 25) and chronoamperometry (CA, Supplementary Fig. 26) measurements using Li | |Al and Li | |Ti cells verify that both the Designed-Cl and Designed-Br electrolytes maintain an oxidative stability window exceeding 3.7 V. Electrochemical performance of the Li | |Se cells The electrochemical performance of Li | |Se cells was evaluated using the Designed-Cl electrolyte. Cyclic voltammetry (CV) curves recorded at 0.1 mV s−1 (Fig. 5a) exhibit two well-defined oxidation peaks centered at 2.20 V and 2.63 V, which are assigned to the Se2⁻/Se0 and Se0/Se+ processes, respectively. Notably, the peak positions remain essentially unchanged over the first three cycles, indicating good electrochemical reversibility in this electrolyte environment. As the scan rate is increased from 0.2 to 1.0 mV s−1 (Fig. 5b), the current associated with the Se0/Se+ couple increases markedly, suggesting that the high-potential Se0/Se+ conversion becomes increasingly dominant under rapid polarization. Moreover, the log(i)-log(v) analysis (Supplementary Fig. 27) yields a slope of 0.57 for the Se0/Se+ peak, implying that the process is primarily diffusion-controlled rather than capacitive. Galvanostatic intermittent titration (GITT) measurement further resolves the potential response of the Se electrode (Fig. 5c and Supplementary Fig. 28). The linear relationship between potential and concentration during the GITT measurements supports the validity of the calculated diffusion coefficients. Compared with the low ion diffusivity ( ~ 10−12 cm2 s−1) in the Se0/Se2− plateau region, a much higher diffusivity ( ~ 10−9 cm2 s−1) is achieved in the Se+/Se0 plateau region. Such a high diffusion coefficient suggests that selenium oxidation in our Li | |Se system proceeds with markedly faster kinetic characteristics than the selenium reduction process in a conventional Li | |Se system. Electrochemical performance of the three-electron-conversion Li | |Se cells. a CV profiles of Li | |Se cells in designed electrolyte at 0.1 mV s−1 and b different scan rates of 0.2, 0.4, 0.6, 0.8, and 1.0 mV s−1. c GITT profiles and the corresponding diffusivity coefficient as a function of the state of charge. d Charge-discharge profiles and e cycling performance of Li | |Se cells at 400 mA g−1. f Comparison of the gravimetric specific energy (based on active mass) of this work against representative conversion-type, intercalation-type, and hybrid conversion-intercalation-type positive electrodes37–40. NCM represents lithium nickel cobalt manganese oxide. The source of the literature data shown in this figure can be found in Supplementary Information, Table 1. Galvanostatic charge-discharge profiles (Fig. 5d) acquired at 400 mA g−1 overlap well from the 2nd to the 5th cycle, confirming stable cycling behavior under practical specific currents. The corresponding dQ/dV curves (Supplementary Fig. 29) display peak pairs at 2.66/2.43 V and 2.16/1.82 V, assigned to the Se0/Se+ and Se2⁻/Se0 redox couples, respectively. The smaller hysteresis for the Se0/Se+ couple relative to the Se2⁻/Se0 couple further supports faster kinetics in the halide-mediated high-voltage conversion, consistent with the high diffusivity of Cl⁻ in the electrolyte. Cycling performance (Fig. 5e) at 400 mA g−1 shows that the cell retains a reversible capacity of 633 mAh g−1 after 200 cycles, corresponding to a capacity retention of 62.2%, which is comparable to most reported conversion-type positive electrodes (Supplementary Fig. 30). The initial capacity decrease followed by recovery is likely associated with the gradual electrochemical activation of the Se cathode during cycling, which improves the utilization of active Se material. In self-discharge tests (Supplementary Fig. 31), the Li | |Se cell using the Designed-Cl electrolyte retained 92.66% of its initial discharge capacity after 48 h of storage, indicating that the combination of the KJ carbon host and tailored electrolyte can effectively anchor the charged products and suppress their diffusion. On the basis of the Se active mass, the positive electrode delivers a gravimetric specific energy of 2003 Wh kgSe−1, which is competitive with those of recent conversion-type and intercalation-type positive electrodes (Fig. 5f and Supplementary Table 1)37–40. Constraining the operating window to 2.3−3.3 V (Supplementary Fig. 32) isolates the Se0/Se+ chemistry and yields a redox couple at 2.66/2.57 V with a polarization of only ~90 mV, further evidencing the high effective Cl− activity in the electrolyte. Notably, the Li | |Se cell involving only the high-valent Se0/Se+ reaction exhibits better cycling stability than that limited to the low-valent Se2−/Se0 reaction within 1.5−2.3 V (Supplementary Fig. 33). Extending the upper cutoff voltage to 3.9 V (1.5−3.9 V window; Supplementary Fig. 34) activates an additional high-valent Se+/Se4+ process, delivering a discharge capacity of 1863 mAh g−1 at 400 mA g−1, corresponding to a ~5.5-electron redox reaction. After 50 cycles, the cell still maintains 1186 mAh g−1 (63.7% retention). XPS spectra collected at the fully charged state (3.9 V, Supplementary Fig. 35) reveals Se4+ signatures, corroborating activation of the Se+/Se4+ pathway. The multi-electron conversion behavior is robust with respect to electrolyte composition: varying the TBMACl concentration between 0.5 M and 2.0 M (Supplementary Fig. 36) or replacing TBMACl with TBMABr (Supplementary Figs. 37, 38) still preserves three-electron Se conversion. Finally, replacing KJ with ordered mesoporous carbon CMK3 (Supplementary Fig. 39) yields comparable voltage profiles and capacities over the first 60 cycles, but rapid fading thereafter, which we attribute to the more open pore architecture of CMK3 that is less effective than KJ at confining soluble species generated during cycling. To demonstrate the versatility of the electrolyte-enabled high-valent chalcogen redox strategy, the Designed-Cl electrolyte was extended to Li | |S (Supplementary Figs. 40, 41) and Li | |SeS2 (Supplementary Figs. 42, 43) cells. Discussion In summary, we report an electrolyte-mediated route to unlock high-valent chalcogen redox chemistry in nonaqueous lithium batteries. By rationally designing halide-rich (Cl− or Br−) non-aqueous electrolytes, we enable a three-electron redox-amphoteric Se chemistry in Li | |Se cells, accessing a reversible Se2−/Se0/Se+ conversion that transcends the conventional two-electron Se0/Se2− limit. As a result, the optimized Li | |Se cells deliver a reversible discharge capacity of 980 mAh g−1 and a gravimetric specific energy of 2003 Wh kgSe−1. These metrics represent a 77% increase in capacity and a 64% increase in specific energy relative to the baseline electrolyte. The Se2−/Se0/Se+ reaction mechanism is supported by in situ Raman spectroscopy in conjunction with ex situ XPS and XAS, which collectively confirm the formation and reversible interconversion of high-valent selenium species (Se2Cl2 or Se2Br2). The selection of halide salts with an asymmetric organic cation simultaneously ensures high solubility and high activity of halide anions. These halide anions are essential not only for activating the Se0/Se+ oxidation process but also for stabilizing oxidized Se+ products, enabling reversible cycling. Furthermore, this electrolyte-enabled strategy can be extended to other chalcogen positive electrodes. When applied to Li | |S and Li | |SeS2 cells, the designed electrolyte activates analogous high-valent reactions of S and SeS2, leading to increased discharge capacity. Overall, this electrolyte strategy provides a potential approach for accessing high-valent chalcogen redox chemistry and improving the energy density of lithium batteries. Methods Materials All chemicals were used directly without any further purification: Selenium (99.99%, Macklin), Sublimed sulfur (99.99%, Sigma-Aldrich), 1,3-Dimethyl-2-imidazolidinone (DMI, 99.8%, Aladdin), Dimethoxymethane (DME, 99.5%, Macklin), 1,3-Dioxolane (DOL, 99.5%, Aladdin), Acetonitrile (ACN, 99.5%, Aladdin), Lithium bis(trifluoromethanesulfonyl)imide (LiTFSI, 99.9%, Macklin), Lithium bis(fluorosulfonyl)imide (LiFSI, 99.9%, Macklin), Lithium nitrate (LiNO3, 99.99%, Aladdin), Tetramethylammonium chloride (TMACl, 98%, Macklin), Tetramethylammonium bromide (TMABr, 98%, Macklin), Tetraethylammonium chloride (TEACl, 98%, Aladdin), Tetrylammonium bromide (TEABr, 99%, Aladdin), Tetrapropylammonium chloride (TPACl, 98%, Macklin), Tetrapropylammonium bromide (TPABr, 98%, Macklin), Tetrabutylammonium chloride (TBACl, 98%, Macklin), Tetrabutylammonium bromide (TBABr, 98%, Macklin), Tributylmethylammonium chloride (TBMACl, 98%, Aladdin), Tributylmethylammonium bromide (TBMABr, 98%, Aladdin), Ketjenblack Carbon (KJ, 99.9%, Dodochem), Super P (SP, 99.9%, Dodochem), polyvinylidene fluoride (PVDF, 99.5%, weight-average molecule weight, Mw ≈ 1,100,000, Dodochem), and N-methylpyrrolidone (NMP, 99%, Aladdin). Preparation of Se/KJ, Se/CMK3, S/KJ, and SeS2/KJ materials Se/KJ and Se/CMK3 materials: Elemental Se and carbon materials (KJ or CMK3) were mixed at a mass ratio of 3:2 by mechanical ball milling at 500 rpm for 5 hours using a planetary ball mill (MITR, YXQM-2L). The milling was conducted in a 50 mL stainless-steel jar with stainless-steel balls of 5 mm diameter. The total mass of the starting materials was 2 g, and the ball-to-powder mass ratio was 40:1. The resulting Se+KJ or Se+CMK3 composite was then transferred to an autoclave and heat-treated at 260 °C for 15 h. Unless otherwise specified, all Li | |Se cells in this work employed Se/KJ as the positive material. S/KJ material: Elemental S and KJ were mixed at a mass ratio of 3:2 by mechanical ball milling at 500 rpm for 5 h under the same milling conditions described above. The resulting S+KJ composite was then transferred to an autoclave and heat-treated at 155 °C for 15 h. SeS2/KJ material: Elemental SeS2 and KJ were mixed at a mass ratio of 3:2 by mechanical ball milling at 500 rpm for 5 h under the same milling conditions described above. The resulting SeS2+KJ composite was then transferred to an autoclave and heat-treated at 250 °C for 12 h. Preparation of Se2Cl2 Typically, 0.1 mol of SeO2 (99%, Aladdin) was placed in a flask and dissolved in 50 mL of concentrated HCl (37%, Anaqua). Subsequently, 0.3 mol of Se was added, followed by the dropwise addition of concentrated H2SO4 (98%, Anaqua) until the evolution of HCl gas ceased. After cooling, the red product Se2Cl2 was separated using a separatory funnel and washed several times with concentrated H2SO4. Residual H2SO4 in the product was removed by treatment with anhydrous BaCl2 (99.9%, Aladdin), after which the solid was filtered. Preparation of electrolytes All electrolytes were prepared in an Ar-filled glovebox ( < 1 ppm of water and O2). The baseline electrolyte consisted of 1 M LiTFSI dissolved in a DME/DOL mixture (1:1, v/v), with 5 wt.% LiNO3 added as an additive. The designed halogen-rich electrolyte comprised 1 M LiTFSI, 1 M LiFSI, and 0.5 M LiNO3 dissolved in DMI, with the addition of either 1 M TBMACl or 1 M TBMABr. Characterizations Unless otherwise specified, all sample preparation for testing was conducted under ambient laboratory conditions (25 ± 2 °C) in an argon-filled glovebox. The morphologies and compositions of materials were studied by scanning electron microscope (SEM, Quattro S, 15 kV) and X-ray diffraction (XRD, Rigaku SmartLab) applying Cu-Kα radiation (λ = 1.5418 Å, 9 kW, 5 ° min−1, and step size of 0.01°). X-ray photoelectron spectroscopy (XPS) was studied in a Thermo K alpha X-ray Photoelectron Spectrometer with an Al Ka X-ray source. All samples were transferred to characterization instruments using inert atmosphere sample holders to prevent degradation. The spectra were calibrated using the C 1s peak (284.8 eV) and fitted using CasaXPS. Raman spectra were recorded with a Renishaw inViaTM confocal Raman microscope with a laser of 532 nm wavelength. All liquid samples were tested in sealed cuvettes under an argon atmosphere. The in situ Raman measurements were performed using the ECC-Opto-10 electrochemical cell purchased from EL-CELL, and its structure is shown in Supplementary Fig. 5. The instrument used for 1H, 19F, 35Cl, and 81Br NMR tests was a Bruker 400 MHz. To avoid the influence of deuterated reagents on the solvation structure, the NMR test used coaxial nuclear magnetic tubes, and the deuterated reagent was deuteroxide. The chemical environment of the Se edge was analyzed by X-ray absorption spectroscopy (XAS). The Se K-edge X-ray absorption near-edge structure (XANES) spectra were processed and normalized using Athena software. A linear pre-edge background was subtracted, and the post-edge absorption was scaled to unity to allow direct comparison of the edge positions and spectral features. UV-vis absorption spectra were recorded on a Lambda 1050 spectrophotometer. Electrochemical measurements A slurry composed of active material, Super P, and PVDF binder in a mass ratio of 8:1:1, mixed in NMP, was used to prepare the positive electrodes. The solvent-to-solid mass ratio in the slurry was 3:1. The slurry was homogenized using a centrifugal stirring defoaming machine (TMAX-XXH-1500S) at 600 rpm for 4 h and then coated on carbon cloth (W0S1011, CeTech, thickness: 0.35 mm) using a doctor blade (MTI Corporation), followed by drying in a vacuum oven at 80 °C for 12 hours. Prior to electrode fabrication, the carbon cloth was rendered hydrophilic by treatment with concentrated nitric acid (Oriental Chemicals & Lab Supplies Ltd.) at 80 °C for 12 h. The obtained single-side-coated positive electrodes (thickness: ~0.35 mm) were punched into round pellets with a diameter of 9 mm using a manual hole punch (MTI Corporation). The mass loading of Se, S, and SeS2 materials on carbon cloth was around 2.0 mg cm–2. The chalcogenide constituted 48 wt.% of the electrode material. Coin-2032 cells were assembled for electrochemical measurements. The coin-cell cases, spacers, and springs were made of 316 L stainless steel. A single-sided, coated positive electrode with a diameter of 9 mm was used. A Li foil (100 μm in thickness, 14 mm in diameter) was used as the negative electrode, and a glass fiber film (Whatman GF/A, 1.6 μm pore size, 260 μm in thickness, 16 mm in diameter) was used as the separator. The Li foil was used directly without any surface cleaning, polishing, or other pretreatment before cell assembly. The amount of electrolyte for each cell is 80 μL. For the three-electrode measurements, two Li foils were used as the reference electrode and counter electrode, respectively. CV curves (0.2 mV s−1, 1.5–3.3 V) and LSV curves (1.0 mV s−1, from open circuit potential to 5.0 V) were collected from CHI660E electrochemical workstation. Galvanostatic tests were conducted using the Neware thermostatic battery test system (CT-4008Tn-5V10mA-HWX) and the LAND battery test system (CT2001A) at 25 ± 0.5 °C. Unless otherwise specified in the figure captions, Li | |Se cells were cycled within 1.5−3.3 V at 400 mA g−1, Li | |S cells were cycled within 1.5−3.5 V at 400 mA g−1, and Li | |SeS2 cells were cycled within 1.5−3.3 V at 200 mA g−1. The cycling data were obtained from at least three independently assembled cells tested in parallel. The Coulombic efficiency (CE) was calculated as the ratio of the discharge capacity to the charge capacity in the preceding charge cycle. Density functional theory (DFT) calculations Density functional theory (DFT) calculations were performed with the CASTEP code41. The exchange–correlation interactions were treated using the Perdew–Burke–Ernzerhof (PBE) functional under the generalized gradient approximation (GGA) framework42–44. Ultrasoft pseudopotentials were adopted, and the kinetic energy cutoff for the plane-wave basis set was set to 380 eV. Structural relaxation was conducted with the Broyden–Fletcher–Goldfarb–Shanno (BFGS) optimization scheme using a coarse k-point mesh45. The geometry optimization was considered converged when the maximum atomic force was below 0.001 eV Å−1 and the total energy change was less than 5 × 10−5 eV per atom. The solvation effect has not been considered for DFT calculations since the adsorption behaviors of different ions in the solution have not been investigated in this work. The geometry optimizations of DFT calculations were performed to further demonstrate the ESP and the dissociation energy. Molecular dynamics (MD) simulation For the MD simulations, we have constructed the electrolyte models in a 43 × 43 × 43 Å box, which consists of 32 TBMACl, 16 LiNO3, 32 LiTFSI, and 32 LiFSI in the ionic condition with 296 DMI atoms. All ions and solvent molecules are randomly distributed in the simulated box. The MD simulations are performed by the Forcite package with the COMPASSII force field applied for all elements. The cut-off distance of nonbonded interactions was 12.5 Å. The long-range electrostatic and van der Waals interactions are calculated by the Ewald method and the atom-based method, respectively. Before MD simulations, the initial electrolyte structures were pre-equilibrated via geometry optimizations and annealing to release excessive strain. For the annealing process, the system was equilibrated under the NVT ensemble by heating from 300 K to 500 K and maintaining 500 K for 500 ps, followed by gradual cooling back to 300 K. There will be a total of 10 heating ramps per cycle with 200 dynamic steps for each cycle. After each cycle, we will perform a geometry optimization. For the MD simulations, we apply the NPT ensemble under 298 K and 1 atm for 2 ns with each step of 1 fs. The Nosé-Hoover thermostat has been applied for the temperature control during MD simulations. We adopted the NPT ensemble instead of NVT because the density of this highly concentrated multi-salt electrolyte cannot be reliably predefined. The NPT ensemble allows the system volume to self-adjust to an equilibrium density under realistic temperature and pressure conditions, ensuring more reliable radial distribution functions, coordination numbers, and diffusion coefficients. For MD simulations, the radial distribution function (RDF), g(r), describes the probability of finding a neighboring species at a distance r from a reference particle, relative to that of an ideal gas at the same density. The coordination number (CN) is directly derived from the RDF and represents the cumulative number of neighboring particles within a given cutoff distance. Mathematically, the coordination number is obtained by integrating the RDF according to: where ρ is the number density of the coordinating species. In practice, the coordination number is typically evaluated by integrating the RDF up to the first minimum after the primary peak, corresponding to the first solvation shell In this work, the reported diffusion coefficients are tracer diffusion coefficients (also termed self-diffusion coefficients) obtained from the mean-squared displacement (MSD) of individual ionic species in MD simulations. Specifically, the diffusion coefficient D was calculated using the Einstein relation: where ∣rit−ri0∣2 is the ensemble-averaged MSD of ion i. The slope was obtained from the linear fitting of the MSD-time curve after the system reached equilibrium and exhibited clear diffusive behavior. Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Transparent Peer Review file Source data Source data Author contributions Y.S. and J.X. conceived the idea for the project. Y.S., H.Z. and D.F. prepared the materials and performed the electrochemical experiments. Y.W. and Z.S. performed the in situ measurements. Y. S., H. Z. and J. Q. carried out the XAS experiments. M.S. and B.H. conducted the DFT and MD calculations. All the authors discussed the results, analyzed the data, and drafted the manuscript. Peer review Peer review information Nature Communications thanks Hao Sun and the other, anonymous, reviewers for their contribution to the peer review of this work. [A peer review file is available.] Funding J.X. discloses support for the research of this work from the Natural Science Foundation of China (grant numbers 22522904, 52402312), the Hong Kong Research Grants Council (CityU 21306326), NSFC/RGC Joint Research Scheme sponsored by the Research Grants Council of Hong Kong and the National Natural Science Foundation of China (Project No. N_CityU116/25), Croucher Foundation (9509012), the Environment and Conservation Fund (ECF Project 20/2023), Guangdong Natural Science Fund (grant numbers 2025A1515012976, 2026A1515010320), and the Science Technology and Innovation Committee of Shenzhen Municipality (JCYJ20240813153120027). Other authors declare no relevant funding. Data availability The data supporting the findings of this study are available within the article, and its Supplementary Information files. Additional data or analyses are available from the corresponding authors upon request. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Bolong Huang, Email: b.h@cityu.edu.hk. Jijian Xu, Email: jijianxu@cityu.edu.hk. Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77258-w. References Associated Data Supplementary Materials Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Transparent Peer Review file Source data Data Availability Statement The data supporting the findings of this study are available within the article, and its Supplementary Information files. Additional data or analyses are available from the corresponding authors upon request. Source data are provided with this paper.