Abstract Wearable magnetic human‐machine interfaces offer unique advantages for unobtrusive and robust interaction, yet their practical deployment is limited by two fundamental challenges: rapid signal attenuation with distance at the hardware level (low signal‐to‐noise ratios) and ambiguous signal interpretation at the algorithmic level. Here, we present a multi‐command magnetic HMI that synergistically integrates low‐noise sensor engineering with time‐series deep learning to overcome these limitations. At the device level, we realized a planar Hall magnetoresistive sensor with an optimized concentric‐ring configuration and a low‐noise readout architecture, achieving a noise floor approaching the thermal limit. The system resolves magnetic‐field variations down to 0.4 μT corresponding to 2 mm displacement at a 12 cm sensor‐magnet separation while extending the interaction range to 20 cm. At the algorithmic level, we developed a temporal learning framework and systematically compared thresholding, a rule‐based classifier, multilayer perceptron, 1D convolutional neural network (1D‐CNN), long short‐term memory (LSTM), and several modern sequence‐learning architectures. The LSTM model achieved an average accuracy of 99.4% for four gesture primitives and maintained 99.3% accuracy after extending the gesture set to six commands by introducing two temporally more complex gestures. Augmentation‐enhanced training further improved performance across acquisition days, previously unseen users, wearing‐position variations, environmental disturbances, and motion artifacts, with accuracies exceeding 90% under all evaluated conditions, demonstrating the system's generalization capability and practical robustness. By combining near‐thermal‐noise magnetic sensing with temporal learning, this work establishes a scalable framework for long‐range, high‐resolution wearable interaction.
Abstract GeTe, a prototypical phase‐change material, has attracted pronounced attention for next‐generation storage and neuromorphic computing, yet its nucleation mechanism remains an ongoing pursuit. The challenge stems from the limited scales of conventional simulations and the intrinsic structural disorder of nucleation. To bridge the scale gap, we developed a machine learning potential (ML potential) trained on an extensive density functional theory dataset, enabling large‐scale molecular dynamics simulations that capture the complete crystallization pathway with quantum‐level fidelity. By employing a unified short‐ and medium‐range structural framework, we simplified the complexity and disorder inherent to nucleation, allowing us to unravel the intricate atomic environment, pinpoint essential structural motifs, and track their dynamic evolution. Through this approach, our simulations reveal an unconventional nucleation process: the initial formation of Ge‐rich clusters with defective octahedral coordination, followed by their Te‐mediated assembly into the final rock‐salt structure. This sequential mechanism presents a different scenario from classical nucleation theory's expectation of alternating Ge/Te incorporation, wherein the Te sublattice preferentially forms a face‐centered‐cubic structure ahead of Ge ordering. The observed two‐stage nucleation process adds a valuable perspective on the structural ordering kinetics in PCMs, helping to explain their fast crystallization characteristics at the atomic level.
Abstract Lead‐based perovskites exhibit exceptional optoelectronic properties, making them highly suitable for light‐emitting diodes (LEDs). However, their intrinsic toxicity, primarily arising from soluble Pb 2+ ions, poses significant environmental and regulatory barriers to large‐scale commercialization. Tin‐based perovskites have emerged as the most promising lead‐free alternatives, retaining favorable charge transport properties and tunable bandgaps while substantially reducing ecotoxicological and human health risks. Nevertheless, the external quantum efficiency (EQE) of tin‐based perovskite LEDs (PeLEDs) remains capped at ∼20%, reflecting a pronounced performance gap relative to state‐of‐the‐art lead‐based counterparts. In this review, we systematically summarize the fundamental structural motifs and photophysical behavior of tin‐based perovskites to identify the root causes limiting film quality, particularly Sn 2+ oxidation under ambient conditions and uncontrolled crystallization kinetics during film formation. We then outline recent device‐level advances, including rational ligand engineering, interfacial dipole modulation, and compositional grading that have enabled measurable EQE improvements. Moreover, we evaluate the practical viability of tin‐based emitters in next‐generation optoelectronic applications, with an emphasis on narrowband pure‐red displays and near‐infrared optical communication. Finally, we highlight three persistent mechanistic challenges: (i) severe nonradiative recombination and imbalanced charge injection stemming from facile Sn 2+ oxidation and inherent p ‐type self‐doping; (ii) pronounced efficiency roll‐off at elevated current densities driven by Auger recombination and localized Joule heating; and (iii) intrinsic operational instability under continuous optoelectronic and thermal stress. We conclude by proposing a unified research framework that integrates rational materials design, machine learning‐guided high‐throughput screening, and scalable fabrication paradigms to guide the development of tin‐based PeLEDs. This roadmap aims to achieve efficiencies and operational robustness comparable to lead‐based devices, thereby enabling truly sustainable high‐performance optoelectronics.
Abstract Two‐dimensional heterostructure‐based computational spectrometers with broad operational bandwidths demonstrate significant potential for laboratory and industrial applications, including medical diagnostics, chemical analysis, and environmental monitoring. However, the fabrication difficulty of two‐dimensional heterostructures hinders their practical and scalable applications. Multilayer indium selenide (InSe), with its high carrier mobility (>10 3 cm 2 V −1 s −1 ) and direct bandgap of ∼1.3 eV, is a promising candidate for high‐performance broadband spectrometers. Here, we develop a homostructure InSe spectrometer, combining advanced computational spectral‐reconstruction algorithms with a gate‐tunable spectral response matrix to demonstrate a room‐temperature InSe‐based spectrometer with a broad operational bandwidth ranging from 450 to 973 nm. The device features a miniaturized footprint of approximately 20 × 50 μm 2 , achieves nanometer‐level resolution (<0.33 nm), and operates at low voltages ranging from −4 to+4 V. Our work paves the way for high‐performance computational spectrometers with diverse applications, including sensing, surveillance, proof‐of‐concept spectral imaging, and spectral measurements.
Abstract Due to the rapid development of electric vehicles and energy storage systems, a more severe demand is proposed for the energy density and lifespan of lithium‐ion batteries. The performances depend on the structure of the materials, such as the cathode, anode, and electrolyte. However, the traditional material research pipeline requires massive preliminary surveys, high experimental investment, and lengthy time span, which makes the battery iteration in years. The application of artificial intelligence (AI) and machine learning (ML) subverts the traditional research pattern. AI‐driven battery research has broken through the bottleneck of traditional battery investigation. ML can decouple high‐dimensional data to excavate the clues for the structure–activity relationship of batteries. Herein, the essential concepts, models, and techniques for engaging ML, as well as the mechanisms and active components of lithium‐ion batteries, are summarized. The basic theoretical research for deciphering the structure–activity relationship of lithium‐ion batteries is further collected to mentor electrode and electrolyte design. The ML‐driven mining pipeline for handling complex data and exploring relationships for different materials is revealed. The challenges confronted in AI‐driven battery designation are discussed, and the ML‐driven research paradigm progress in battery research is prospected.
This paper comprehensively reviews the development of millimeter‐wave (MMW) and terahertz (THz) near‐field imaging technologies, with an emphasis on the state of synthetic aperture radar (SAR)‐based imaging technologies. Near‐field imaging technologies are categorized into passive and active imaging modes, among which active imaging is favored because of its strong signal‐to‐noise ratio and three‐dimensional (3D) reconstruction capabilities. This paper discusses SAR‐based active imaging systems with various antenna array structures, including planar SISO (Single‐Input Single‐Output)‐SAR, cylindrical SISO‐SAR, planar MIMO (Multiple‐Input Multiple‐Output)‐SAR and cylindrical MIMO‐SAR. Specifically, the paper emphasizes the advancements in SISO‐SAR and MIMO‐SAR technologies, highlighting the advantages of MIMO‐SAR in improving imaging speed and reducing costs. Finally, the paper provides a summary and outlook on SAR‐based MMW and THz near‐field imaging technologies.
Memristive materials (MMs) with adjustable conductance states are becoming the fundamental building blocks of artificial neural networks. However, challenges such as nonlinear weight conductance updates and the trade-off between increasing melting uniformity and reducing solid-to-liquid transition time have impeded the demonstration of in situ learning on a large-scale multiple-layer memristive network. Here, we modulate the dynamic interactions between MMs and multiple-pulse excitations in the melting-based voltage-pulse length diagram and temperature dependent mean-square displacement. We utilize all four melting states (FMSs) to create an integrated framework for attaining rapid in-memory computing and in situ deep neural network applications. We achieved near-ideal R 2 value uniformity and a partial melting time below random-access memory (RAM) device's switching times. Reversible switching using below-RAM switching time pulses was also demonstrated. The FMS network discloses in situ learning capability and competitive classification accuracy on a conventional machine learning dataset. Simulations suggest that increasing the number of hidden neurons would further improve classification accuracy. Ab initio molecular-dynamics simulations provide a clear insight into the melting kinetics in MMs and the structural origins of the prior complete melting-facilitated decrease in output conductance. The memristive neural network represents a potential hardware platform for artificial intelligence, offering high speed and energy efficiency.
Magnetic van der Waals (vdW) semiconductors offer unique opportunities to integrate spin and charge degrees of freedom in atomically thin devices. Here, we fabricate dual-gated heterostructures consisting of exfoliated CrSBr, stacked beneath monolayer graphene and encapsulated by hexagonal boron nitride ( h -BN). Magnetotransport measurements performed at temperatures down to 2 K and magnetic fields up to 13.5 T demonstrate that CrSBr acts as an exceptionally strong p -type doping source to graphene in this vdW heterostructure, driving the graphene Fermi level deep into the valence band with hole density reaching ∼10 13 cm −2 . High-frequency Shubnikov-de Haas oscillations and well-quantized quantum Hall plateaus were observed, confirming the high quality of the doped graphene channel. Furthermore, nearly field-independent resistance ridges emerged at a top gate bias of ≈ −3 V, which we attribute to graphene/ h -BN moiré minibands with a twist angle of >3°. These miniband features become experimentally accessible solely due to the extreme p -doping by CrSBr, a regime that is inaccessible via conventional gating techniques.
Achieving high selectivity, high sensitivity, and superior conductivity simultaneously for chemiresistive gas sensors (CGS) remains a significant challenge due to inherent trade-offs among the material's properties. Inspired by biological olfaction, we design and fabricate a Ni-MOF/SnO 2 heterojunction as a perception-transduction-transmission separation membrane for bionic CGS, where Ni-MOF acts as a selective perception layer for H 2 S, SnO 2 serves as the electron transmission and proton blocking layer, while the heterojunction enables efficient signal transduction from perception to transmission. The sensor demonstrates exceptional performance, including fingerprint-level selectivity, high moisture stability, and an ultra-low detection limit of 10 ppb. Notably, the sensor operates based on the heterojunction rectification enhancement model. The sensitivity is solely determined by the Ni-MOF perception layer, while the baseline current is governed by the SnO 2 transmission layer. By reducing the resistance of SnO 2 , the net response is geometrically amplified. Additionally, applying a bias voltage enables geometric acceleration of the response/recovery rates. This study proposes a novel strategy for designing high-performance CGS.
Surface passivation and reconstruction in quantum dot (QD) materials are crucial for enhancing the performance of optoelectronic devices, particularly in high-sensitivity, low-noise short-wave infrared (SWIR) photodetectors (PDs). This study presents an optimized approach for PbS/CdS core-shell QDs through optimized surface engineering through controlled CdS shell modulation and solution-phase ligand exchange with concentrated lead halides. The refined surface reconstruction significantly reduces QD aggregation and reduces trap states, resulting in ordered QD stackings with narrower energy distributions. Consequently, the QD PDs achieve a significantly reduced dark current density of 192 nA cm − 2 and an enhanced detectivity of 5.06 × 10 12 Jones, resulting in a 29.6% reduction in dark current and a 7.4-fold improvement in detectivity compared to pristine QD PDs. Electrochemical impedance spectroscopy confirms the reduction in trap-assisted recombination, supported by extended photoluminescence lifetimes and higher quantum efficiencies. These findings underscore the potential of surface reconstructed QDs for advanced SWIR PD applications, particularly in achieving high sensitivity in imaging systems.
Artificial neurons are pivotal for neuromorphic hardware, but the development of compact and uniform devices remains challenging. Conventional volatile memristors suffer from abrupt switching, which hinders spatiotemporal consistency. In this study, we developed a two‐terminal artificial neuron with intrinsic leaky integrate‐and‐fire (LIF) dynamics, eliminating the need for bulky capacitors or additional reset circuits and enabling exceptional compactness. Crucially, the device exhibited superior spatiotemporal uniformity across arrays compared to typical volatile memristors—which show abrupt transitions—achieved through gradual volatile switching. Combined theoretical and experimental analyses revealed that this behavior resulted from the controlled formation and self‐rupture of pure oxygen vacancy–based conductive filaments, which were modulated by electric field and Joule heating. Neuronal dynamics, including the firing threshold and relaxation, were tuned by adjusting the input amplitude and frequency. To validate functionality, a two‐layer spiking neural network leveraging these neurons was developed, which achieved 97.4% accuracy on MNIST classification, rivaling ideal LIF models even under noisy conditions. This highlights the remarkable noise tolerance of the device, which is crucial for real‐world applications. This study elucidates filament‐driven volatility mechanisms and establishes a scalable approach to energy‐efficient neuromorphic systems, advancing the development of bio‐inspired computing hardware.
With the rapid advancement of information technology and artificial intelligence, the energy efficiency bottleneck of the von Neumann architecture and the scaling limits of silicon-based semiconductors are increasingly constraining improvements in computing performance. Neuromorphic computing, by emulating the cooperative behavior of neurons and synapses in the brain, enables the integration of computation and memory, offering a promising route toward next-generation low-power and high-performance computing. As a key strategy for artificial synapse implementation, optoelectronic synapses exhibit ultrafast response, broad bandwidth, and contactless optical signal transmission, conferring significant advantages in information sensing, transmission, and storage. In particular, two-dimensional (2D) transition metal dichalcogenides (TMDs) with their atomic-scale thickness, tunable bandgaps, and outstanding optoelectronic properties provide an ideal material platform for highly integrated optoelectronic synaptic devices. This review presents a comprehensive overview of the fundamental concepts of artificial synapses and synaptic plasticity, examines the device architectures and operating mechanisms of optoelectronic synapses based on 2D TMDs along with their applications in neuromorphic visual systems and neuromorphic computing, and further discusses the key challenges and future research directions in this area.
Hydrogen peroxide (H 2 O 2 ), an environmentally benign oxidizer, finds extensive applications in pulp bleaching, wastewater treatment, and medical sterilization. Photocatalytic H 2 O 2 synthesis via water and oxygen activation on semiconductor surfaces presents a sustainable production strategy. Notably, structurally tunable organic photocatalysts have emerged as promising candidates, in which targeted molecular engineering can boost the photocatalytic performance by enlarging specific surface areas, extending light absorption ranges, and facilitating charge carrier transport-separation dynamics. Given the growing significance of organic photocatalysts in H 2 O 2 synthesis, a comprehensive review of this field has become imperative. This paper offers a systematic examination of visible-light-driven H 2 O 2 synthesis using various organic photocatalysts, including graphitic carbon nitride (g-C 3 N 4 ), resorcinol-formaldehyde (RF) resin, covalent organic frameworks (COFs), and linear conjugated polymers (LCPs). The focus lies on fundamental mechanistic elucidation, design of reaction pathways and active sites, modification strategies, and establishment of efficient photocatalytic systems. Extensive studies have correlated photocatalytic efficiency with interfacial electron transfer kinetics and spatial charge separation. Therefore, we methodically analyze key determinants governing photogenerated carrier dynamics and present engineering strategies for performance enhancement. Furthermore, we discuss emerging application scenarios enabled by photocatalytic H 2 O 2 generation. Importantly, this review critically evaluates persistent challenges and cutting-edge solutions in visible-light-mediated H 2 O 2 synthesis, ultimately providing design principles for developing high-efficiency organic photocatalysts.
Photocatalytic overall water splitting (OWS) can convert solar energy into hydrogen (H 2 ) and oxygen (O 2 ), which is significant in reducing the reliance on fossil fuels. Constructing S‐scheme heterojunctions is an effective method for facilitating charge transfer, but the huge interfacial charge transfer barrier poses a challenge to advance the efficiency of photocatalytic OWS. Here, a low‐interfacial barrier Ce‐S bond‐enhanced Mo‐doped ZnIn 2 S 4 /oxygen‐deficient CeO 2 (Mo‐ZIS/O V ‐CeO 2 ) S‐scheme heterojunction photocatalyst was designed via a doping‐defect coupling strategy. The abundant unsaturated S atoms generated by doping Mo atoms in ZnIn 2 S 4 combine with the unpaired electrons on the Ce atom in O V ‐CeO 2 , forming the interfacial Ce‐S bonds, which induce a 43% decrease in carrier transport activation energy and a 2.1‐fold increase in build‐in electric field intensity compared to ZIS/O V ‐CeO 2 . Reduced carrier transport activation energy and increased built‐in electric field intensity provide a strong driving force for charge separation following the S‐scheme pathway. Benefiting from the interfacial Ce‐S bonds and the S‐scheme transfer path, Mo‐ZIS/O V ‐CeO 2 exhibits H 2 and O 2 evolution rates of 512.7 and 256.3 μmol g −1 h −1 , respectively, along with a solar‐to‐hydrogen efficiency of 0.14%. This study proposes an innovative insight into developing and constructing S‐scheme heterojunction photocatalysts with efficient charge migration interfaces.
Owing to its ability to reduce charge recombination and enhance redox capability, the step‐scheme (S‐scheme) heterojunction has manifested appealing prospect for photocatalysis. In this work, an organic‐inorganic S‐scheme heterojunction based on CdS nanorods and conjugated polymer 2‐hexyl‐carbazole‐benzothiadiazole (CBT) is constructed. The obtained catalyst exhibited impressive photocatalytic hydrogen production performance (14.02 mmol g −1 h −1 ) with a high apparent quantum efficiency of 5.4% at 420 nm. The charge transfer mechanism and the enhancement of photocatalytic hydrogen production in S‐scheme heterojunctions were investigated by density functional theory calculations, in situ X–ray photoelectron spectroscopy, and in situ Kelvin probe force microscopy. The successful construction of organic‐inorganic S‐scheme heterojunctions and the formation of Cd–S bonds at the interface effectively promoted the separation and transfer of charge carriers.
Electrochemical deposition technique, a method widely recognized for its precision and versatility in the electronics industry, is gaining attraction in the energy field, particularly in developing solid oxide fuel cells (SOFCs). Its ability to deposit metal compounds with nanostructures under simple ambient conditions makes it invaluable for modifying conventional electrodes with refined morphologies and compositions. In this mini-review, we explore the principles of electrochemical deposition and highlight its recent applications in SOFC technology. Our focus lies on its pivotal role in fabricating coating layers or catalysts on electrodes with improved functionalities to build more efficient and durable fuel cells. Furthermore, we discuss emerging strategies for electrode surface modification and the potential of electrochemical deposition in advancing SOFC design and functionality. Our review also outlines future research directions aimed at harnessing and expanding the capabilities of electrochemical deposition in energy conversion applications.
Metal halide perovskites (MHP)-based electrically pumped vertical-cavity surface-emitting lasers (EPVCSEL) are promising candidates in optoelectronics due to low-carbon footprint solution processing method. However, significant challenges impede MHP-EPVCSEL manufacturing: (1) Distributed Bragg Reflectors (DBRs) composed of typical electron transport layers (ETLs) and hole transport layers (HTLs) are not conductive enough. (2) Due to large mobility difference of typical ETLs and HTLs, carriers-unbalanced injection leads to severe performance degradation. Herein, we propose a potential strategy to address such challenges using MAPbCl 3 and CsSnCl 3 as carrier transport layers with mobility 3 orders larger than typical ETLs and HTLs. Via transfer matrix method calculations, we find that the reflectance of DBRs composed of MAPbCl 3 (130.5 nm)/CsSnCl 3 (108 nm) is larger than 91% with 10 pairs of DBRs. Furthermore, the proposed EPVCSEL device simulation shows that MHP-EPVCSEL has the potential to achieve room temperature continuous wave lasing with a threshold current density of ∼69 A cm −2 and output optical power ∼10 −4 W. This work can provide a deep insight into the practical realization of MHP-EPVCSEL.
Emerging-wide bandgap semiconductor Ga 2 O 3 shows distinct characteristics for optoelectronic applications and a stable crystal phase of Ga 2 O 3 is highly desired. Herein, we have first reported a metal-semiconductor-metal structure photonic synaptic device based on the ε -Ga 2 O 3 thin film. The ε -Ga 2 O 3 epilayer is grown on the c -sapphire with a low temperature nucleation layer, which presents a crystal orientation relationship with the c -sapphire ( ε -Ga 2 O 3 <010> // c -sapphire <1–100> and ε -Ga 2 O 3 <001> // c -sapphire <0001>). The ε -Ga 2 O 3 photonic device was stimulated by UV pulses at different pulse widths, pulse intervals, and reading voltages. Under the UV pulse excitation, the photonic device exhibits primary synaptic functions including excitatory postsynaptic current, short term memory, pair pulse facilitation, long term memory, and STM-to-LTM conversion. In addition, stronger and repeated stimuli can naturally contribute to the higher learning capability, thus prolonging the memory time.
Semiconductor photocatalysis is a promising tactic to simultaneously overcome global warming and the energy crisis as it can directly convert inexhaustible solar energy into clean fuels and valuable chemicals, hence being employed in various energy applications. However, the current performance of photocatalysis is largely impeded by the fast recombination of photogenerated charge carriers and insufficient light absorption. Among various materials, bismuth-based photocatalysts have stood out as excellent candidates for efficient photocatalysis due to their unique controllable crystal structures and relatively narrow band gap. These features endow the selective exposure of active facets (facet engineering) and wide light absorption range, resulting in tunable photocatalytic activity, selectivity, and stability. Therefore, it is of great potential to use facet-engineered bismuth-based photocatalysts for efficient energy applications (e.g., water splitting, CO 2 reduction, N 2 fixation, and H 2 O 2 production) to achieve sustainable development. Herein, the introduction provides the overview of this research, while the synthesis, modification strategy, and the latest progress of facet-engineered bismuth-based photocatalysts in energy application were summarized and highlighted in this review paper. Lastly, the conclusion and outlooks of this topic were concluded to give some insights into the direction and focus of future research.
Electrochemical ammonia oxidation reaction (AOR) presents a promising avenue for realizing sustainable nitrogen cycling in various energy and environmental applications. However, sluggish catalytic activity, catalyst poisoning effects, and low stability pose significant challenges. Developing efficient electrocatalysts with high activity and stability necessitates a thorough understanding of the complex mechanisms and various reaction intermediates. In this review, we first discuss the AOR mechanism and the operando/in-situ characterization techniques employed for elucidating the reaction mechanisms. Subsequently, we summarize the development of AOR electrocatalysts, including noble-metal-based catalysts, non-noble-metal-based catalysts, and homogeneous catalysts. We also highlight the primary practical applications of AOR in energy, environment and chemical production fields, including direct ammonia fuel cells, chemical production of nitrates, nitrites, hydrogen, and wastewater treatment. Finally, based on the progress in electrochemical AOR, we discuss the challenges and propose future directions for advancing this field.
