Understanding the statistical behaviors from films to devices is crucial for performance prediction and materials innovation. Here, we present the first fully automated high-throughput experimental platform for metal-halide perovskite research in China, integrating solution preparation, film fabrication, electrode evaporation, and comprehensive optical/optoelectronic characterization. This platform enables human-interference-free data collection with high repeatability, facilitating reliable statistical analysis. Through systematic investigation of over 1000 perovskite samples, we first identify the key factor of solvent atmosphere affecting experimental repeatability, and then introduce a super-absorbent resin to effectively mitigate solvent-related variability. By quantitative tracking of statistical distributions across the film-to-device transformation, we reveal that the deposition of charge transport layers also alters the bulk properties of perovskite films, as manifested by statistical changes in bandgap and Urbach energy. Finally, we develop a machine learning-based predictive model that links thin-film optical features to device performance, demonstrating the feasibility of AI-driven approaches to accelerate the evolution of perovskite materials.

Addressing the dual challenges of global energy sustainability and dynamic optical management, we present an innovative flexible photochromic transparent fluorescent wood composite film (PT-FWF) with molecular-scale engineered design, fabricated through in situ Eu³⁺ coordination on TEMPO-oxidized cellulose scaffolds. This hierarchically structured material combines fluorescent wood film with hot-pressing, impregnation, and coating (PMMA/WO₃) to achieve multimodal optical control. PT-FWF demonstrates exceptional multifunctionality: 88% optical transparency, 107.5° ± 1.0° hydrophobicity surface, and thermal insulation (ΔT ≈ 5.5°C). A unique dual-mode photoresponsive mechanism enables through synergistic photochromic-fluorescent effects: instantaneous fluorescence under UV light and coloring/bleaching with or without light-assisted (UV or simulated sunlight). The smart window model exhibits over 90% UV-blocking efficiency, and the transmittance of the smart window can be reversibly switched between 88% and 5% under prolonged light conditions, showing a high modulation of visible light (∆T_lum = 83%) at 1030 nm, enabling simultaneous daylight optimization and energy conservation. This molecular-scale engineered wood composite defines a transformative platform for adaptive optical materials, merging energy-efficient architectural solutions with information encryption through sunlight-regulated smart windows that simultaneously enable environmental protection and anti-counterfeiting.

Flexible pressure sensors are a key component of electronic skins (e-skins), converting mechanical stimuli into easily analyzed electrical signals. These sensors need to be highly sensitive to respond to small changes in external stimuli. However, balancing the trade-off between sensitivity and pressure monitoring range remains a significant challenge. Here, we fabricated a capacitive tunable pressure sensor (TPS), based on the synergistic effect within a composite material, composed of a sponge-like porous structure and thermoplastic expandable microspheres (TEMs). By adjusting the temperature to drive the expansion of the TEMs, mode switching between low and high compression modulus was achieved. This enables high sensitivity (2.39 kPa⁻¹) in low compression modulus mode and a wide pressure monitoring range (up to 953.96 kPa) in high compression modulus mode. TPSs are applicable in diverse fields, from detecting subtle pressures like human pulse and respiration to measuring larger pressures based on touch, and even vehicle loads. These sensors can also be integrated with machine learning algorithms for object recognition. The success of TPS is expected to provide new ideas in solving the trade-off between sensitivity and pressure monitoring range of flexible pressure sensors.

The rapid accumulation of retired lithium-ion batteries demands sustainable recycling technologies, particularly for lithium iron phosphate (LFP) cathodes, to alleviate resource constraints and curb environmental hazards posed by conventional disposal. Here, we propose a tunable pre-oxidization and microencapsulation strategy for the direct regeneration of unhomogenized spent LFP. Through controlled pre-oxidation, heterogeneous spent LFP is converted into a stoichiometric intermediate of Li₃Fe₂(PO₄)₃ and Fe₂O₃, resetting structural heterogeneity and removing binder/carbon residues. Polarity-modified encapsulation spatially confines Li₂CO₃/PVA (polyvinyl alcohol) around intermediates by non-solvent induced phase separation (NIPS), enabling uniform Li replenishment. Subsequently, annealing reconstructs the olivine lattice and concurrently generates an in situ carbon coating. The regenerated LFP exhibits restored crystallinity with Fe-Li antisite defects reduced from 6.1% to 1.41%, and a 5 nm in situ carbon coating, delivering a specific discharge capacity of 161 mAh g⁻¹ at 0.1 C with a ~30% reduction in polarization voltage, exhibiting 82% capacity retention over 1000 cycles at 2 C. This work establishes a facile pathway for LFP recycling by integrating defect correction with carbon coating in a scalable process, offering a viable solution to industrial battery reclamation and the circular economy.

Sodium dual-ion batteries (SDIBs) based on organic active materials have attracted extensive attention due to their low cost, environmental friendliness, high safety, and superior stability. However, limitations such as poor conductivity, high solubility in electrolytes, kinetics constraints, and low active site utilization caused by dense layer stacking impede further advancement. Herein, a sawtooth polyimide anode with wide layer spacing, abundant active sites, and an extended π-conjugated system on nonplanar surfaces was designed through interlayer molecular engineering. The material exhibits a stable structure, fast transport and reaction kinetics, and high active-site utilization. Proof-of-concept SDIBs delivered a high discharge capacity of 162.4 mAh g⁻¹ with 200 stable cycles without degradation, robust fast-charging capability, and a low self-discharge rate of 0.11% h⁻¹. Excellent electrochemical performance with a reversible capacity of 107.6 mAh g⁻¹, outstanding rate capability, low polarization, and 2000 stable cycles without attenuation were achieved even at high active mass loading. Mechanistic studies reveal a dual-storage mechanism involving diffusion and pseudocapacitance, expanding the diversity of redox-active polymers. These findings provide new insights and theoretical guidance to designing high-performance organic materials for Na⁺ storage.

Developing metal/carbon materials as durable electrocatalysts for electrochemical CO₂ reduction is of great importance for maintaining long-term activity of metal sites. However, the uncertainty associated with the interaction of metal–carbon restricts the exposure of active sites and the inhibition of the hydrogen evolution reaction. Herein, we have successfully synthesized a hierarchical bimetal/carbon catalyst with unconventional rectifying interfaces (Bi-Sn@C), which works as a charge emitter for efficiently bending CO₂ to enhance the adsorption and hydrogenation of activated *CO₂ and the generation of *OCHO intermediate by the nucleophilic reaction process due to the electronic perturbation at rectifying interfaces and electron delocalization of the bimetallic cores. The Bi-Sn@C demonstrates up to HCOOH faradic efficiency of 93.06% with energy efficiency of 70.6% at −0.52 V (vs. RHE) and low overpotential of 320 mV in a flow electrolyzer, and operates continuously for more than 160 h due to the protective mechanisms of the carbon shell. Experimental results and theoretical calculations reveal that the hierarchical rectifying interfaces of Bi-Sn@C show an apparent non-uniform distribution of charge and low energy barrier of *OCHO-to-*HCOOH for facilitating the reaction kinetics of formate production.

Nonlinear physical systems hold great promise for energy-efficient and low-hardware-cost information processing. However, their computational capabilities remain constrained by the complexity and tunability of system nonlinearity. Here we report a dual-ferroelectric gate-tunable memristor with a dipole coupling effect, achieving enlarged hysteresis, rich temporal dynamics, and nonvolatile heterosynaptic plasticity. By harnessing the dynamic nonlinearity of the dual-ferroelectric memristor, multimodal reservoir computing with an in-material fusion strategy has been achieved, which is demonstrated with a multimodal object recognition task. By exploring the static nonlinearity of the dual-ferroelectric memristor, nonlinear in-memory computing is realized with gate-tunable nonlinear functions, which successfully accelerates the Euclidean distance computation in the K-means clustering task. This work achieves strong coupling between the intrinsic physical dynamics and computational functionalities, offering new opportunities for more efficient hardware-accelerated systems.

The rapid growth of the aging population and the rising prevalence of motor disorders demand intelligent, user-centric rehabilitation technologies. Integrating artificial intelligence and the Internet of Things (AIoT) into sensor devices offers a powerful means of capturing limb motion data and assisting rehabilitation, thereby helping patients regain confidence and functional independence. This work presents a self-powered sensor based on a Kresling-structured thermoplastic polyurethane (TPU) substrate that integrates triboelectric nanogenerators (TENGs) and electromagnetic generators (EMGs). Optimizing the Kresling geometry and stiffness of the Kresling structure achieves high adaptability to human motion and high-sensitivity monitoring. The bistable design enables synergistic TENG–EMG signal outputs under axial compression and circumferential torsion, leveraging TENG sensitivity and EMG stability for reliable low-frequency motion detection. Using machine learning framework extracts multi-scale motion features, enabling identity verification, limb activity monitoring, and precise wrist tracking with classification accuracy all above 98%. Based on composite sensor signals and human-machine interaction (HMI), immersive and assistive wrist rehabilitation training is achieved through real-time feedback and applications such as claw machine. Additionally, interactive platforms including a “Dancing Machine” and a “Driving Simulator” integrate the sensor to explore brain–body collaborative rehabilitation. This work provides a low-cost, energy-efficient, and scalable solution for next-generation intelligent rehabilitation, paving the way for personalized, immersive, and user-centric therapy systems.

In artificial visual systems, optimizing the dynamic range (DR) of optoelectronic synapses is essential for achieving robust and environment-adaptive perception. However, the inherent trade-off between photoresponse and dark current noise presents significant challenges in realizing a high DR. This study introduces a flat-band heterojunction strategy to achieve high DR optoelectronic synapses through a zinc oxide (ZnO) nanowires and graphene oxide (GO) sheets heterostructure, which enables efficient minority carrier trapping under minimal external bias. Through multi-defect-engineering in the heterojunction structure, the device demonstrates enhanced persistent photoconductivity (PPC), improved photocurrent gain, and significantly suppressed dark current, achieving an ultra-high DR of 74.9 dB in two-terminal optoelectronic synaptic devices while reducing energy consumption to 23 fJ/spike at a bias voltage of 1 mV. Additionally, the devices can emulate typical synaptic functionalities and attain 92.84% pattern recognition accuracy in artificial neural network simulations, offering an energy-efficient platform for advanced neuromorphic systems. This work offers a generalizable strategy for low-power, high-fidelity visual perception systems, advancing intelligent sensing and neuromorphic computing.

Cu(I) halides have emerged as promising scintillator candidates for underwater x-ray imaging applications, owing to their exceptional stability in water environment and outstanding optical properties. However, the reliance on toxic organic solvents and low production yield in conventional synthesis methods pose obstacles to practical application. In this work, an aqueous grinding method is presented for the efficient synthesis of Cu(I)-iodine cluster halides powder (C₁₂H₂₄O₆)₂X₂Cu_mI_n (XCuI, X represents different alkali metals or alkaline earth metals), which exhibits bright luminescence with a high photoluminescence quantum yield (PLQY). The proposed approach is characterized by its simplicity, cost-effectiveness, environmental friendliness and safety. Given its high PLQY of 95.86% and rich photophysical properties, the scintillation performance of NaCuI is systematically investigated. The NaCuI powder achieves a low detection limit of 54.8 nGy_air s⁻¹ and a high relative light yield of 61 986 photons MeV⁻¹. Building upon these foundations, we fabricated a large-area and highly flexible NaCuI scintillation screen, which achieves an outstanding spatial resolution of 10.84 lp mm⁻¹. Furthermore, integration of the scintillator screen with a thin-film transistor backplane array enabled real-time digital imaging of various test objects. The resulting x-ray flat-panel detector demonstrated exceptional imaging performance, capturing well-defined contours of imaged subjects and remarkably detecting underwater objects with notch defects. This research provides a new strategy for the synthesis and application of highly efficient luminescent scintillators with low-cost, high yield, and environmental friendliness.

Tandem solar cells offer a pathway beyond the Shockley–Queisser limit of single-junction devices. Among these, all-perovskite tandems are especially appealing for their low cost and facile fabrication. However, non-radiative recombination at the interfaces between perovskite absorbers and charge-transport layers continues to impede their translation from theoretical potential to experimental realization. Here, we develop a molecular-design strategy for dual interface engineering of the perovskite photoactive layer, addressing the vertical inhomogeneity inherent to solution-processed films. We demonstrate that the efficacy of surface modification hinges on matching the alkyl-chain length of diammonium cations to the local lattice dimensions of each sub-cell. By applying tailored alkyl diammonium salts to both the top and bottom interfaces, we achieve dramatic reductions in non-radiative loss, lowered interfacial energy barriers, and suppressed vacancy formation. As a result, the power conversion efficiencies (PCEs) of single-junction cells improved from 16.7% to 20.5% for the high-bandgap sub-cell and from 18.9% to 22.4% for the low-bandgap sub-cell. Integration into a monolithic tandem architecture yields a PCE of 27.5%, and the device retains 90% of its initial performance under maximum-power-point operation (AM 1.5G, 100 mW cm⁻²) at room temperature in ambient air for over 500 h. This work establishes a clear, structure-guided paradigm for interface passivation in perovskite tandems, unlocking both high efficiency and operational durability.

Aqueous rechargeable Zn-MnO₂ batteries are considered one of the most promising energy storage systems and have been extensively studied in recent years, owing to their high energy density, low cost, and intrinsic safety. However, the practical application of conventional Zn-MnO₂ batteries is hindered by poor cycling stability, corrosion, and unwanted side reactions. Recently, dual electrode-free Zn-MnO₂ batteries have emerged as a promising alternative. Their simplified battery configurations and lightweight design, achieved by eliminating the need for pre-fabricated bulk electrodes, offer higher energy density. Nevertheless, such designs can, in principle, suffer from limited cycle life due to the poor reversibility of the Zn-MnO₂ deposition/stripping process. This review critically examines recent advances aimed at overcoming these challenges, highlighting the transition from conventional to anode-free, cathode-free, and ultimately dual electrode-free configurations. We also present key strategies including electrolyte engineering, current collector modification via 3D printing, and interfacial engineering to enable stable long-term cycling, along with insights from advanced in situ characterization techniques such as electrochemical quartz crystal microbalance (EQCM) and optical microscopy. Finally, we outline future opportunities required to advance this promising field toward practical applications.