Understanding and predicting the electrochemical behavior of high-nickel cathode materials remains a central challenge in developing advanced lithium-ion energy storage systems. Although recent machine learning methods have achieved remarkable predictive performance, their generic architectures seldom embody the underlying physical and chemical mechanisms governing electrochemical processes, which limits both interpretability and generalization. We present MEMFNet, a deep learning framework specifically designed to reflect materials science knowledge through a dual-pathway architecture that mirrors the distinction between static material properties and dynamic electrochemical processes. Trained on 158,200 voltage-capacity data points from 791 discharge profiles of high-nickel cathode materials, MEMFNet reduces prediction error by 48.64 % compared to state-of-the-art methods. More importantly, the knowledge-guided architecture transforms the model from a black box into an interpretable system whose learned representations align with established electrochemical principles. By integrating domain knowledge, MEMFNet enables interpretable and scientifically meaningful learning in materials informatics.

The detection of trace nitrogen dioxide (NO2) in port atmospheres is crucial for protecting occupational health and maintaining air quality in coastal cities. Nonetheless, the reliable monitoring of NO2 remains a challenge in complex environments featuring high humidity, diverse gas compositions, and dynamically fluctuating concentrations, which impose substantial interference. Here, the NO2 sensors based on the CeO2/ZnO/WO3 (CZWT) heterostructure films were fabricated via the template-assisted magnetron sputtering. The optimal device exhibited outstanding NO2 sensing performance at 280 °C with high response (81.69 to 50 ppm NO2), rapid response/recovery rate (25/10 s), ultralow detection limit (10 ppb), ideal selectivity, and excellent long-term stability (90 days). Further investigations suggested that the exceptional hydrophobicity of the CZWT heterostructure film and the dynamic Ce3+/Ce4+ redox cycle endowed the sensors with humidity-tolerant response properties. In parallel, a wireless gas-detection device was developed to achieve the monitoring of NO2 in humid environments. This work demonstrates an effective approach for designing humidity-independent MOS gas sensors.

Herein, spent ternary lithium-ion batteries (NCM) are upcycled into a NiCo/MnO@CC interfacial catalyst for Li-CO2 batteries. Integrated theoretical calculations and experimental analyses reveal that the interfacial affinity facilitates CO2 adsorption and induces small-sized and uniform Li2CO3 deposition, endowing the catalyst with reduced polarization and 1200 h cycling stability.

ABSTRACT For sodium‐ion batteries (SIBs), the central obstacle to artificial intelligence (AI)‐guided discovery is not simply data volume or algorithm choice, but the conditional nature of electrochemical labels. Capacity, voltage, initial Coulombic efficiency (ICE), rate capability, and retention depend on synthesis, disorder, electrolyte, interphase, and cell format rather than composition alone. Existing SIB and AI‐battery reviews are commonly organized by material class, value chain, or chemistry, but they rarely explain how a prediction becomes a defensible cell‐level decision. This review develops a data‐to‐interface framework linking literature, computation, processing, electrolyte, and full‐cell data to descriptors, models, mechanistic hypotheses, and validation under realistic constraints. Ordered cathodes are matched to voltage, phase stability, and Na‐ion migration; hard carbon is treated as a disordered‐anode case governed by precursor history, pore structure, storage mechanism, and electrolyte‐sensitive interphases. Electrolytes and electrode/electrolyte interfaces define whether electrode‐level predictions remain valid in practical cells. A structured assessment of representative studies shows that no study in the selected evidence set covers all seven metadata layers, with full‐cell matching and interphase reporting as the most common gaps. We propose a sodium‐specific minimum reporting profile for uncertainty‐aware, descriptor‐driven, and interface‐validated decision‐making.

ABSTRACT Rational electrolyte design for high‐energy‐density lithium‐ion batteries (LIBs) urgently demands precise and quantitative molecular descriptors of solvation power to enable deep learning (DL)‐accelerated screening, yet such descriptors remain lacking. Here, we introduce the electrostatic potential ratio |ESP min |/ESP max (ESP ratio ) as a quantitative descriptor capturing the balance between electron‐donating and electron‐accepting capacities, and identify a solvation modulation zone (0.9 < ESP ratio < 2.4) through unsupervised clustering of 344 molecules encompassing 196 experimentally reported LIB electrolyte molecules. By combining this descriptor with self‐supervised pre‐trained DL models fine‐tuned on small experimental datasets, we enable hierarchical screening of ∼10 6 PubChem molecules and prioritize electrolyte candidates from previously unexplored chemical space. Experimental evaluation of representative candidates, including TBDN and PIV as co‐solvents and additional nitrile‐containing molecules as electrolyte additives, confirms that the ESP ratio ‐guided workflow can enrich chemically meaningful electrolyte candidates for high‐voltage Li||LiCoO 2 .

ABSTRACT The practical deployment of aqueous zinc‐ion batteries (AZIBs) is impeded by irreversible dendritic growth and parasitic interfacial reactions. Here, we report a dynamic‐static shielding strategy that employs trace amounts of 2‐methylimidazole (2H‐MZ) and Na + as dual‐functional additives. 2H‐MZ spontaneously adsorbs onto the Zn (002) plane, forming a hydrophobic barrier that excludes water and suppresses hydrogen evolution, while the low reduction potential of Na + offers electrostatic shielding, which helps to homogenize the Zn 2+ flux and reduce concentration polarization. Notably, 2H‐MZ modulates the electrolyte's H‐bond network and weakens Zn 2+ ‐H 2 O binding without entering the primary solvation shell, thereby accelerating Zn 2+ transport. The resulting interfacial synergy facilitates highly oriented (002) plating and stabilizes anions. Consequently, Zn||Zn symmetric cells exhibit stable cycling for over 650 h at 5 mA cm −2 /5 mAh cm −2 , and Zn||Cu cells achieve an average Coulombic efficiency of 99.29% over 750 cycles. Full Zn||I 2 cells retain > 90% capacity after 2400 cycles at 5 A g −1 , while a flexible pouch cell maintains 81.1% capacity after 250 cycles. This work demonstrates a low‐concentration, non‐consumptive dual‐additive design that combines physical adsorption and cation shielding, offering a new avenue for durable zinc anodes.

Dye wastewater containing toxic substances and high resistance to degradation poses a significant threat to environmental integrity and human well-being. Notably, the low cost and highly efficient treatment of high concentration azo dyes, such as methyl orange (MO), remains a significant challenge. This study developed a novel composite adsorbent by in-situ growth of cobalt-based zeolitic imidazolate framework (ZIF-L(Co)) on waste eggshell membrane (ESM), designated as ZIF-L(Co)/ESM. The physical and chemical characteristics revealed that ZIF-L(Co) was uniformly grown and vertically dispersed on the three-dimensional fiber network of ESM, increasing the contact area between ZIF-L(Co) and MO. Meanwhile, cobalt ions coordinated with the functional groups on the surface of ESM to form chemical bonds, further strengthening the stability of growth sites and effectively preventing cobalt ion leaching. The adsorption experiments showed that the removal rate of 0.01 g ZIF-L(Co)/ESM for 30 mL MO with a concentration of 100 mg/L was 90.6 %, which was significantly better than that of pure ZIF-L(Co) (81.6 %) and unmodified ESM (33.1 %). The maximum adsorption capacity of ZIF-L(Co)/ ESM for MO was 1216.75 mg/g. The kinetic analysis revealed that the adsorption process followed the pseudo-second-order kinetic model and was predominantly chemical. The adsorption mechanism was primarily attributed to the synergistic effects of hydrogen bonding, it-it interaction, and electrostatic attraction between ZIF-L (Co)/ESM and MO. ZIF-L(Co)/ESM exhibits excellent adsorption performance and economic and environmental friendliness. It presents a promising green and sustainable strategy for treating high concentration dye wastewater.

Biomass-derived hard carbon has become the most promising anode material for sodium-ion batteries (SIBs) due to its high capacity and excellent cycling stability. However, the effects of synthesis parameters and structural features on hard carbon's (HC) electrochemical performance are still unclear, requiring time-consuming and resource-intensive experimental investigations. Machine learning (ML) offers a promising solution by training on large datasets to predict hard carbon performance more efficiently, saving time and resources. In this study, four ML models were used to predict the capacity and initial Coulombic efficiency (ICE) of HC. Data augmentation based on the TabPFN technique was employed to improve model robustness under limited data conditions, and the relationships between features and electrochemical performance w ere examined. Notably, the XGBoost model achieved an R2 of 0.854 and an RMSE of 23.290 mA h g-1 for capacity prediction, and an R2 of 0.868 and an RMSE of 3.813% for ICE prediction. Shapley Additive Explanations (SHAP) and Partial Dependence Plot (PDP) analyses identified carbonization temperature (Temperature_2) as the most important factor influencing both capacity and ICE. Furthermore, we used bamboo as the precursor to synthesize four hard carbons based on the predictive approach. The electrochemical performance of these samples closely matched our predictions. By leveraging machine-learning approach, this study provides an efficient framework for accelerating the screening process of biomass-derived hard carbon candidates. (c) 2026 Science Press and Dalian Institute of Chemical Physics, Chinese Academy of Sciences. Published by Elsevier B.V. and Science Press. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

Lithium metal batteries (LMBs) are widely regarded as promising next-generation energy storage solutions because of their high theoretical energy density. Nonetheless, their practical application is hindered by complex interfacial phenomena, including unstable solid electrolyte interphases (SEI), dendritic lithium growth, and ongoing electrolyte consumption, which cause rapid performance decline and safety concerns. Traditional experimental and computational methods often struggle to capture the multiscale and dynamic nature of these processes. Recent progress in machine learning (ML) offers new opportunities to speed up LMB development through data-driven materials discovery, mechanistic insights, and performance prediction. This review summarizes recent advances in applying ML to LMBs from three key perspectives: materials design, interfacial processes, and lifetime prediction. It highlights how ML models facilitate electrolyte discovery, elucidate the mechanisms of SEI formation and dendrite growth, and predict battery degradation through early-cycle signals and morphological features. Finally, this review explores emerging opportunities for combining ML with highthroughput experimentation and autonomous laboratories to facilitate the closed-loop discovery of advanced battery technologies, as well as discusses the key challenges and future directions of ML-driven LMB research. This review is expected to provide not only a critical assessment of the state of the art but also a practical roadmap for integrating predictive analytics into next-generation battery research.

Abstract Na4Fe3(PO4)2P2O7 (NFPP) is regarded as a highly promising cathode material for sodium-ion batteries (SIBs) in energy storage systems. In particular, the low cost and environmental friendliness of NFPP render it especially suitable for large-scale commercialization. Nevertheless, the inherently low ionic/electronic conductivity of NFPP severely impedes its practical deployment in SIBs. Herein, employing a bulk-phase regulation strategy through elemental doping, we prepared Ti-doped Na4Fe3–xTix(PO4)2P2O7/C cathode materials using iron phosphate as both the iron and phosphorus source via a low-cost solid-state method, aiming to further reduce cost and enhance overall material performance. The results demonstrate that moderate Ti4+ doping shortens the average Fe–O bond length, reinforces the structural stability of the crystalline framework, and effectively improves Na+ diffusion kinetics. Among the compositions investigated, the optimized Na4Fe2.94Ti0.06(PO4)2P2O7/C (denoted NFPP-Ti0.06) exhibits the best comprehensive electrochemical properties, delivering outstanding rate capability from 0.2 to 20 C (with a specific capacity of 96.11 mAh g–1 at 20 C, retaining 83.51% of the 0.2 C capacity) and exceptional cycling stability (capacity retention of 99.18% after 400 cycles at 2 C and 82.59% after 3000 cycles at 20 C). Furthermore, a full cell assembled with NFPP-Ti0.06 as the cathode and hard carbon as the anode also exhibited favorable electrochemical performance.

Faradaic capacitive deionization (Faradaic CDI) is a promising technology for tackling the global water crisis through efficient desalination. However, its practical implementation has been hindered by inherent challenges, such as sluggish reaction kinetics and limited mass transfer, especially under low-salinity conditions. To address these issues, this study proposes an induced-charge Faradaic CDI (IC-Faradaic CDI) system. The key innovation lies in the strategic integration of Faradaic CDI with an electric double-layer mechanism through a wireless induced-charge IC unit. This design modulates the electric field and ion transport, optimizes the concentration distribution, and establishes a universal pathway to reconcile the trade-off between kinetics and capacity. As a result, the IC-Faradaic CDI system delivers a desalination capacity of 0.269 mg cm-2 and an outstanding average desalination rate of 0.027 mg cm-2 min-1, surpassing most existing Faradaic CDI systems. Furthermore, to validate its real-world applicability, a larger-scale system (324 cm2) was constructed, achieving 77% salt removal from simulated brackish water and the effective desalination of real brackish water from the Yangtze River estuary. This work provides a novel and universal strategy to alleviate the kinetic limitations of Faradaic CDI and offers a low-cost, membrane-free, and energy-efficient solution for high-performance water treatment.

Solar interfacial evaporation technology has attracted widespread attention as an environmentally friendly and low-cost freshwater production method. However, hydrogel evaporators are prone to issues like salt crystallization and limited water transport rate, which restrict their evaporation performance and long-term stability. In this study, a sponge-like hydrogel evaporator with a three-dimensional porous structure was successfully fabricated via a simple foaming method. The synergistic effect between its porous structure and the Donnan effect not only provides high-speed water transport channels but also inhibits surface salt crystallization, thereby significantly enhancing the hydrogel's evaporation performance, brine transport rate, and long-term stability. The PAC hydrogel evaporator showed no salt deposition after 10 h of continuous evaporation in a 10 wt% NaCl solution. Under one sun, the evaporation rate of PAC hydrogel reaches 2.85 kg m- 2 h- 1. Additionally, the negatively charged -SO3- groups on the PAC hydrogel's polymer chains can specifically adsorb dye molecules in organic wastewater, and its porous structure shortens mass transfer paths for rapid dye capture. This study provides a highly promising solution in the fields of seawater desalination and clean water production, and is expected to play a pivotal role in practical applications.

Porous fibers are recognized as an ideal functional carrier in areas such as toxic gas adsorption, industrial catalyst recovery, membrane fouling mitigation, and antibacterial agent immobilization. To extend the service life of polymeric porous fibers and maximize the exposure and immobilization of functional components on the fiber surface, this study fabricated activated carbon (AC)-loaded porous polyphenylene sulfide (PPS) fibers via melt-spinning of PPS/AC/polyethersulfone (PES) ternary blends, followed by annealing and PES phase extraction. To clarify the pore formation mechanism and AC migration behavior, the effects of blending sequence (AC/PPS-PES, AC/PES-PPS, and AC/PES/PPS), AC content, and melt extrusion temperature were systematically investigated in terms of rheological behavior, AC distribution, pore morphology, and mechanical properties of the resulting fibers. Based on rheological tests, scanning electron microscopy, N₂ adsorption-desorption analysis, and mechanical measurements, the rules governing the exposure-immobilization state of AC on the PPS fiber surface were summarized. Although the AC/PPS-PES blend exhibited the highest continuity of the PES phase, it produced the smallest pores, leading to significant embedding of AC within the PPS matrix. In contrast, AC/PES-PPS and AC/PES/PPS blends showed stronger affinity between PPS and AC, promoting AC migration toward the PES-PPS interface. After PES extraction, these blends not only developed richer pore structures but also immobilized a substantial amount of AC on the pore walls of the PPS matrix. Furthermore, porous fibers from the AC/PPS-PES sequence demonstrated higher tensile strength, which increased progressively with AC content.

A novel metal-carbon dual-gradient strategy was developed using multivalent ions from laterite nickel leachate to construct spinel-type multi-metallic oxide microspheres for lithium-ion battery anodes. A controlled coprecipitation process enabled the self-assembled formation of a metal-gradient structure from Fe/Mn-rich cores to Ni/Mg/Al-enriched shells, while an appropriate glucose addition during the subsequent liquid-phase coating produced a uniformly distributed amorphous carbon shell and internal conductive network. This moderate carbon content effectively stabilized the microspherical architecture, mitigated thermal stress during calcination, and promoted the integration of a dual-gradient structure combining metallic and carbon gradients. The optimized G-FNMMA@0.3C electrode delivered a high initial discharge capacity of 1268.1 mAh g- 1 at 100 mA g- 1 and retained 1156 mAh g- 1 after 500 cycles at 1000 mA g- 1, outperforming most reported Fe-based spinel anodes. Electrochemical analyses (CV, EIS, GITT) confirmed that this dual-gradient configuration enhances ion diffusion, suppresses interfacial degradation, and stabilizes the multi-metal framework. This work provides a scalable pathway for converting industrial nickel laterite leachates into high-efficiency lithium storage materials.

It is crucial for advancing supercapacitor technology to synthesize low-cost, high-performance electrode materials. Although research on biomass-derived porous carbon is advancing rapidly, it remains challenging to construct hierarchical porous structures under sustainable and facile conditions. This paper proposes a low-alkali activation strategy that utilizes the natural potassium-rich ash component in peanut shells to facilitate the preparation of hierarchical porous carbon. By utilizing the uniformly distributed potassium-containing components and a KOH/precursor mass ratio of 2:1, a porous carbon material (HUPS-18-2) with a specific surface area as high as 1795.1 m(2) g(-1) and a balanced microporous/mesoporous structure was synthesized. In 6 M KOH aqueous electrolyte, HUPS-18-2 exhibited a specific capacitance of 304.2 F g(-1) at a current density of 1 A g(-1) in a three-electrode test system, with a capacity retention rate of 74% at a high current density of 50 A g(-1). In addition, a symmetric supercapacitor assembled using this material exhibited an energy density of 11.1 Wh kg(-1) at a power density of 100 W kg(-1) and maintained a capacity retention rate of 97.5% after 10,000 cycles. Furthermore, in 1 M Na2SO4 neutral electrolyte, the device achieved an energy density of 37.5 Wh kg(-1) at a power density of 200 W kg(-1) and retained a capacity retention rate of 93.4% after 10,000 cycles. This study demonstrates that self-etching using uniformly distributed potassium-containing components can reduce activator consumption and promote the formation of hierarchical pores, thereby providing a practical and environmentally friendly preparation method for industrial supercapacitor electrodes.

Aqueous zinc-iodine batteries are promising for grid-scale energy storage but suffer from irreversible capacity loss when pursuing the high-energy four-electron redox chemistry, primarily due to the hydrolysis of high-valent iodine species (I+) and severe corrosion of the zinc anode. Herein, we propose a polyhalide ionic-liquid phase-separation strategy enabled by the dual-functional additive 1-ethyl-3-methylimidazolium ([EMIm]+). We find that [EMIm]+ preferentially coordinates with the electrogenerated polyhalide [IBr2]- to form a hydrophobic ionic liquid (EMImIBr2), which spontaneously separates from the aqueous electrolyte. This phase separation physically isolates I+ from water, effectively suppressing hydrolysis and enabling highly reversible I0/I+ conversion. Meanwhile, [EMIm]+ mitigates Br--induced corrosion, guides Zn deposition along the dendrite-suppressing (002) plane, and improves plating/stripping reversibility. As a result, Zn||I2 cells achieve a high specific capacity of 391.0 mAh g-1 at 0.1 A g-1 (approaching the theoretical limit of 422 mAh g-1), with an excellent rate performance (302.4 mAh g-1 at 3 A g-1), and long-term cycling stability (70% capacity retention over 2000 cycles). Practical viability is demonstrated by high-loading pouch cells delivering 190 mAh and powering electronic devices.

Conventional treatment processes for aquaculture wastewater address phosphorus and antibiotics separately. However, achieving effective simultaneous removal of them remains a significant challenge. This study proposed a method of in-situ compositing using ions leached from MgAlFe-LDH under hydrothermal conditions, and an insitu growth of metal-organic framework compounds (M-BTC, M = Mg, Al, Fe) composed of organic ligands of trimesic acid on its surface, MgAlFe-LDO/Ox catalyst with Z-scheme heterojunction was ingeniously constructed after calcination. MgAlFe-LDO/Ox showed the maximum phosphorus adsorption capacity was 280.21 mg/g, and the tetracycline removal rate reached 88.17% under the 50 mg/L tetracycline with visible light. After phosphorus adsorption, the tetracycline removal rate of MgAlFe-LDO/Ox increased to 95.47%. Phosphorus adsorption enhanced the response of MgAlFe-LDO/Ox to visible light, and the negative charge on the surface of the catalyst accelerated the migration of photogenerated holes to the surface to participate in the oxidation reaction, improving the photogenerated carriers separation rate. Phosphorus adsorption also promoted the formation of & sdot;O2- in the photocatalytic process and accelerated the degradation of tetracycline. It provided a new composite material-based method for the simultaneous, effective removal of phosphorus and tetracycline from water.