Anisotropic hyperelastic materials exhibit complex nonlinear constitutive behaviors. Machine learning approaches have shown promise in modeling these behaviors, but they usually require stress data and prior knowledge of material symmetry. In this paper, we develop a symmetry-aware, equilibrium-based neural network (SENN) for learning three-dimensional anisotropic hyperelasticity. The SENN is trained in an unsupervised manner, i.e., trained with full-field deformation and total force data instead of stress-strain pairs. The material symmetry is learned by a symmetry-aware fine-tuning method, where an algorithm identifies the rotations in material symmetry groups and the rotations are then embedded into the neural network model to improve its predictive accuracy. Combined with the physical constraints of objectivity, polyconvexity and a stress-free initial state, the SENN enables physically consistent constitutive modeling of hyperelastic materials with general symmetries, free of a priori symmetry assumptions or definitions of strain invariants. Numerical experiments on a variety of hyperelastic solids with diverse symmetry groups are conducted, demonstrating the capability of the network. The SENN enables data-efficient, unsupervised discovery of anisotropic constitutive hyperelastic laws, offering a generalizable tool for characterizing advanced materials with complex symmetries.

ABSTRACT Dual color detection technology spanning from the ultraviolet (UV) to near‐infrared (NIR) regions enables the acquisition of characteristic spectral information across multiple wavebands, which is critical for imaging, autonomous driving, and object recognition. However, the widespread adoption of such photodetectors remains challenges constrained by the heteroepitaxy difficulty of different semiconductors, low detection performance especially in IR band, and severe signal crosstalk. Therefore, achieving high performance low crosstalk UV‐NIR dual color detection in a single device has become an urgent priority. Here, we demonstrate a novel UV‐NIR dual color photodetector based on a p‐GaN/MoS 2 /BP van der Waals heterojunction, featuring a vertically stacked back‐to‐back diode configuration. The back illumination structure design significantly reduces optical crosstalk. This dual color photodetector achieves a responsivity of 125 mA W −1 and a specific detectivity of 1.8 × 10 11 Jones under 365 nm illumination at 0 V bias, while delivering a responsivity of 160 mA W −1 and a detectivity of 5.1 × 10 13 Jones under 1550 nm illumination at 0 V bias. Leveraging its unique spectral response characteristics, selective UV‐IR dual color imaging has been successfully realized. Our work paves the way to develop high performance low crosstalk dual color photodetectors through the van der Waals integration of different dimensional materials.

From Renaissance drapery to tissue morphogenesis, pattern formation exemplifies how geometry and constraints generate complex structures. In soft and architected matter, motifs such as creases, kinks, and domain walls function as order-parameter textures mediating structural transitions. Yet deterministic and reprogrammable control of such patterns remains a central challenge: conventional geometry-based strategies hardwire functionality into structure, leaving deformation modes defect-sensitive and difficult to reconfigure. Here we introduce a pseudo-dynamic mapping that interprets static deformation fields as trajectories of fictitious particles evolving in engineered energy landscapes. This paradigm provides a forward design strategy in which reshaping potential symmetry and bifurcation structure prescribes diverse reprogrammable solitonic domain-wall-lattices in a single, defect-free metamaterial solely under uniform loading. We demonstrate initiation, modulation, inversion, melting, and annihilation of these patterns, governed by a tunable bifurcation landscape. Predictions are validated through simulation and experiment, culminating in a mechanical display that encodes digital information via domain-wall-bits. This approach bridges nonlinear field theory with practical pattern reprogramming, offering a versatile route for programmable design in architected and adaptive materials.

Integrating vertical-cavity surface-emitting lasers (VCSELs) on flexible substrates offers significant opportunities for developing smart light sources and multifunctional photonic platforms. In this study, AlGaN-based deep ultraviolet VCSELs on a flexible substrate were demonstrated. The AlGaN quantum well heterojunction was separated from the sapphire substrate by selectively removing the thin n-GaN sacrificial layer using electrochemical etching and subsequently transferred onto a flexible substrate. Meanwhile, two dielectric distributed Bragg reflectors were deposited to construct the vertical resonant cavity. Single-mode lasing at 294.2 nm with a threshold power density of 7.4 MW/cm 2 and a linewidth of 0.39 nm was achieved at room temperature. Furthermore, multimode lasing attributed to non-uniformities within the distributed Bragg reflectors cavity was observed. This work opens up possibilities for advancing flexible VCSELs, as well as for the flexible photonic integration in the deep ultraviolet spectrum.

The integration of computational logic into mechanical metamaterials enables the development of matter with intelligence that can sense and respond to environmental stimuli. While recent advances have demonstrated diverse mechanical logic systems, a challenge is bridging the gap between continuous physical inputs and discrete digital outputs. In this study, we propose a mechanical metamaterial capable of nearly arbitrary digital encoding and computing of continuous stimuli through reprogrammable sequential deformation. The metamaterial is built upon an engineered multistable architecture that supports a series of snap-through events under uniaxial compression. Using an inverse design strategy, the deformation sequence can be tuned across a broad design space by tailoring the stiffness distribution of constituent units. Selective activation or deactivation of specific units in the metamaterial enables in situ reprogramming of the sequence without the need for remanufacturing. Experiments demonstrate that the inverse design and reprogramming allow a single material system to perform mechanical analog-to-digital conversion, signal processing, and field-programmable gate array (FPGA)-like logic operations. These capabilities open a new way for material-based computing, showing a promising route toward intelligent mechanical metamaterials.

The complete replacement of toxic mercury lamps requires III-nitride deep-ultraviolet (DUV) light emitters with high wall-plug efficiency (WPE at least 20%) under high-current operation, but their WPE is seriously limited by difficult forward light emission and low light extraction. Herein, a high-power DUV light emitter with a record 21.2% WPE is proposed by a cooperative photon-redirection strategy. The DUV chip architecture synergistically integrates an in situ nano-porous AlN scattering layer with an optimized reflective mesa and a double-sided patterned sapphire substrate. This strategy efficiently redirects laterally propagating photons into the escape cone through coupled reflection and multi-stage scattering, thereby increasing the TM-mode light extraction efficiency by 252.1%. Consequently, the fabricated DUV light emitter achieves a record-high WPE of 21.2% at an injected current of 70 mA, maintaining a WPE exceeding 16% over a wide current range from 10 to 350 mA. Furthermore, a DUV light emitter array module with a high irradiance of 1.33 W/cm(2) is integrated in a large-flow water-treatment system, making >99.999% inactivation of bacteria under an ultra-high water flow rate of 20 m(3)/h. This work will definitely accelerate the large-scale commercialization of the new-generation solid-state DUV source.

Fatigue crack growth behavior and deformation-induced nanocrystallization in (Cu0.5Zr0.5)100-xAlx (x = 5, 6, 8) metallic glasses (MGs) were investigated using three-point bending fatigue tests. Fatigue crack propagation follows Paris' law, with the Paris exponent m increasing with aluminium (Al) content, indicating reduced fatigue resistance; Cu46Zr46Al8 exhibits premature unstable fracture at lower Delta K. Three crack growth regions are identified: slow growth (Region I), fast growth (Region II), and unstable fracture (Region III). In Region I, fatigue striations form over multiple cycles via shear transformation zones (STZs), accompanied by dense nano-crystallization and nanotwinning that enhance energy dissipation. In Region II, multiple shear bands dominate crack propagation, producing fine and coarse striations, while nanocrystallization is reduced. Enhanced shear band activity in low-Al MGs leads to staircase-like crack paths and larger plastic zones, improving resistance. In Region III, fracture transitions from vein-like melted features to dimpled morphology, indicating a ductile-to-brittle shift with increasing Al content.

Far-ultraviolet light-emitting diodes hold great promise for applications in ultraviolet lithography, secure optical communication, and sterilization, owing to their minimal adverse effects on human tissues. However, producing milliwatt-level optical output at wavelengths below 240 nm remains a considerable technical challenge. In this study, we present the design, fabrication, and characterization of AlGaN-based far-ultraviolet micro-light-emitting diodes emitting at 230 nm. The finite-difference time-domain simulations indicated that reducing the mesa size improves light extraction efficiency. Based on these findings, the 30- mu m mesa arrays were fabricated, achieving a peak optical output power of 2.8 mW at 250 mA and a peak emission wavelength of 230 nm. To the best of our knowledge, this result represents the shortest emission wavelength reported for milliwatt-level far-ultraviolet micro-light-emitting diodes. These results demonstrate a major step forward in realizing compact, efficient, and high-performance far-ultraviolet light sources, with strong potential for next-generation applications in public health, environmental safety, and biomedical technologies.

From echoes and shadows to rainbows and mirages, the reflection and refraction of waves at spatial interfaces are ubiquitous in our daily lives. There is growing interest in exploring wave scattering at a time boundary, where an abrupt change in material properties throughout the entire space occurs and incident waves are nonadiabatically time refracted and reflected. Although it is long believed that reflection and refraction are wave phenomena, recent advances in space-time duality show the feasibility of generalizing these concepts to static systems. Here, we unveil a synthetic time boundary (STB) in static systems. Unlike its dynamic counterpart, the STB has a vanishing width and is free from energy input, while allowing for arbitrary parameter switch. On this basis, we develop a framework of static refraction/reflection for load-induced deformation, with its spatial trajectory undergoing an abrupt deflection upon crossing the STB. Similar to coherent wave control in temporal metamaterials, static refraction/reflection at an STB facilitates the manipulation of localized deformation. Our study discloses an STB prevalent in lattice materials, laying the foundation for exploring refraction and reflection in static systems.

Heterogeneous integration of various semiconductors is critical for broadband photodetection, yet current van der Waals integration techniques of 2D and 3D semiconductors face challenges in low process temperature, high‐quality heterointerface, and large‐scale fabrication. Here, a cryogenic‐temperature depositing strategy is presented for fabricating large scale and high‐quality tellurium (Te)/germanium (Ge) heterojunctions, leveraging the narrow bandgap (≈0.33 eV) of Te for broadband photodetection. By employing cryogenic thermal evaporation, wafer‐scale uniformity is achieved in 4‐inch Te films on Ge substrates, with a low roughness of ≈0.71 nm and large crystalline domains on the order of micrometers square. The fabricated Te/Ge heterojunction photodetector exhibits a broadband photoresponse from visible to mid‐infrared and a high performance including a linear dynamic range of 102.5 dB, a responsivity of 0.73 A W −1 , a specific detectivity of 8.47 × 10 10 cm Hz 1/2 W −1 , and rapid rise/fall times of 22 µs/14 µs at zero bias. The superior performance is attributed to the formation of a type‐II heterojunction with a native GeO x interlayer (≈3 nm) for quantum tunneling, which suppresses dark current. Furthermore, an 8 × 8 photodetector array is demonstrated and shows an exceptional uniformity. This work establishes Te/Ge heterojunctions as a versatile platform for next‐generation broadband photodetectors.

In this work, the electrical and optical performance of AlGaN-based ultraviolet-C light-emitting diodes (UVC-LEDs) with a tapered Al-content hole injection layer was investigated both theoretically and experimentally. A total of 1000 h of real-time electrical stress was conducted to study the degradation process of such devices. UVC-LED incorporating a hole injection layer with a larger gradient was found to significantly suppress the degradation process compared to a sample with a smaller tapering gradient. Marginal efficiency droop of only 4.55% as well as 66% improved light output power, were identified for the proposed design under a current density of approximately 100 A/cm2. It was unambiguously demonstrated that UVC-LED with a greatly tapered hole injection layer facilitates both electron blocking and hole injection, providing a promising pathway towards the development of high-efficiency UV emitters.

Crystalline/amorphous (C/A) nanolaminates offer a promising route to overcome intrinsic brittleness of bulk metallic glasses by combining high strength with enhanced plasticity. The mechanical performance of these materials is strongly governed by the crystalline-amorphous interfaces (CAIs), yet the underlying strengthening and toughening mechanisms remain poorly understood. Here, we employ large-scale molecular dynamics simulations to investigate the compressive deformation of C/A nanopillars composed of alternating equal-thickness crystalline Cu and amorphous Cu50Zr50 layers. The simulations reveal a nonmonotonic size effect, with the yield strength peaking at a critical layer thickness. As the layer thickness decreases, the dominant deformation mechanism shifts from shear localization in the amorphous layers to cooperative plasticity across both phases. At ultrathin layers (∼1-2 nm), shear transformation zone (STZ) activation and dislocation nucleation become dominant, enabling plastic strain to traverse interfaces and form sample-spanning shear bands. A theoretical model is proposed to explain the size-dependent strength by incorporating both amorphous and crystalline contributions. These findings provide atomic-scale insights into interface-mediated plasticity and offer guidance for designing C/A nanolaminates with superior mechanical properties.

Determining internal stress and strain fields in solid structures under external loads has been a central focus of continuum mechanics, playing a critical role in characterizing the mechanical behaviors and properties of both engineering and biological systems. With advancements in modern optical and electron microscopy techniques, strain fields can now be directly measured using sophisticated methods such as digital image correlation and digital volume correlation. However, direct measurement of stress fields remains limited to simple cases, such as photoelastic tests and standard uniaxial or shear tests. For elastoplastic solids, which exhibit complex irreversible and history-dependent deformations, stress fields are typically inferred through numerical calculations based on empirical constitutive models that are not always reliable or even available. Here, we introduce an unsupervised equilibrium-based neural network (ENN) that is trained using readily measurable strain fields and forces from a single specimen to directly predict the internal stress field. The ENN's structure aligns with the general framework of the incremental theory of elastoplasticity, without requiring prior knowledge of its detailed mathematical form. Once trained, the ENN, referred to as ENNStressNet, serves as an end-to-end stress mapper, enabling the direct determination of stress fields from measured strain fields in elastoplastic solids with arbitrary geometries and under various external loads. This approach thus bypasses the need for constitutive modeling and numerical simulations in conventional engineering analysis.

In this work, we propose four types of pre-compressed beam-based multistable mechanical metamaterials and use a combination of theoretical, simulation and experimental methods to systematically explore the effects of pre-compression and initial configurations on their mechanical properties. We found that curved beams with identical initial configurations but different pre-compressions have the same negative stiffness value, but their peak forces differ. Furthermore, the results demonstrate that applying pre-compression is a more effective programming strategy than geometric modulation for altering the stability of the beams. We also demonstrate that pre-compressed multistable mechanical metamaterials can robustly program unloading deformation sequences while maintaining consistent loading orders, thereby enabling the concealment of certain stable configurations. The proposed pre-compressed beam-based mechanical metamaterials offer potential benefits in mechanical computing and information encryption, paving the way for expanding the design concepts and application prospects of multistable mechanical metamaterials.

Structures capable of multiple stable configurations are increasingly attractive for applications in shape-morphing and adaptive systems. Among these, corrugated sheets are promising due to their ability to achieve different loading-position-dependent stable morphologies. In this work, the bistability of corrugated sheets is systematically investigated, where point loads at different positions can lead to distinct stability responses. To quantify the mechanical behavior, a theoretical model of the sheet is developed, combined with finite element analysis (FEA) and experimental validation. The analysis begins with a single-cell model, from which a phase diagram is derived for the transition between monostable and bistable regimes as a function of nondimensional geometric parameters. The model is then extended to multi-cell corrugated sheets to reveal the effects of intercellular interactions on the overall stability landscape of the structure. Finally, the theoretical model enables customization of bistable regions in the corrugated sheets—such as butterfly-like and diamond-like bistability regions—achieving programmable bistability through the geometric design of unit cells and their spatial arrangement. This work provides insights into how loading position influences the mechanical stability of corrugated sheets, presenting significant potential for advanced applications in shape-morphing structures, soft robotics, and sensor technologies, where tailored mechanical responses are crucial.

The complex spectrum of non-Hermitian topological systems manifests extreme sensitivity to boundary perturbations when the system size is large. Hence, despite precise manipulation of non-Hermitian boundaries and sizes remains a challenge, it is of fundamental importance in developing ultra-sensitive sensing devices. Here, we address this issue using a non-Hermitian static mechanical lattice platform, with the lower bound of the accessible boundary perturbation being 10-22, about tens of orders of magnitude better than current systems, for a maximal size exceeding 102. This performance facilitates the exploration of various extreme non-Hermitian phenomena. As a proof of concept, we demonstrate theoretically the braid topology of non-Hermitian non-Bloch bands, whose sensitivity increases exponentially with the size. Based on the static platform, we measure experimentally ultra-sensitive braid phase transitions. Our study unveils the nontrivial interplay among non-Hermiticity, braid topology, and spectral sensitivity, and reaches a much improved level of controllable non-Hermitian boundaries and sizes.

The efficiency of AlGaN based deep ultraviolet light-emitting diode (DUV LEDs) are mainly hindered by the light extraction issue. In this work, an innovative cooperative scattering structure is introduced that combines a nanopore configuration with an aluminum (Al) nanoparticle array on the n-AlGaN layer of the DUV LEDs. The integration of these two scattering arrays can enhance light extraction by mitigating total internal reflection at the device interface. The nanopores are formed on the n-AlGaN surface by electrochemical etching and optimized by varying the etching voltage, while the Al particles are formed by thermal annealing. With the help of the cooperative scattering structure, the light output power (LOP) of the optimized DUV LEDs is significantly increased by 77.6% and a notable 2.2 times is achieved in its light extraction efficiency (LEE) enhancement factor. Moreover, Finite-Difference Time-Domain (FDTD) simulations have validated that the cooperative scattering structure considerably enhances the LEE for both Transverse Electric (TE) and Transverse Magnetic (TM) modes, respectively. This work paves the way to fabricate high efficiency DUV LEDs via novel scattering structure designs.

Mechanical computing metamaterials, which utilize transitions among discrete configurations to process information autonomously when perceiving external stimuli, are important in the development of intelligent mechanical systems. However, most current designs involve tessellations of planar mechanisms or bistable structures, which typically offer only two distinct configurations, requiring many units to provide multiple input-outputs for complex logic operations with little consideration of system integration. Here, We propose a family of 2n-side kinematic polygonal modules introducing n decoupled inputs and 2n transitable extreme configurations, which are coupled with electrical circuits to construct conductive logic metamaterials for mechanical computing. To simplify the construction of logic computing systems, a minimized combinatorial logic canonical function, the parallel computing sum of the product (PCSoP) function, is developed. We first design and integrate seven basic logic gates on a quadrilateral module, followed by implementing all four types of 2-bit arithmetic operations with a single polygonal module, where the 2-bit divider on an octagonal module is designed for the first time. Moreover, information display and simultaneous recognition of three mathematical properties of the decoded decimal numbers 2-15 are implemented on one polygonal module. This kinematics-based design strategy for mechanical computing metamaterials will greatly advance the development of mechanical intelligence.

Haptic displays are crucial for facilitating an immersive experience within virtual reality. However, when displaying continuous movements of contact, such as stroking and exploration, pixel-based haptic devices suffer from losing haptic information between pixels, leading to discontinuity. The trade-off between the travel distance of haptic elements and their pixel size in thin wearable devices hinders solutions that solely rely on increasing pixel density. Here we introduce a continuity reinforcement skeleton, which employs physically driven interpolation to enhance haptic information. This design enables the off-plane displacement to move conformally and display haptic information between pixel gaps. Efforts are made to quantify haptic display quality using geometric, mechanical, and psychological criteria. The development and integration of one-dimensional, two-dimensional, and curved haptic devices with virtual reality systems highlight the impact of the continuity reinforcement skeleton on haptic display, showcasing its potential for improving haptic experience.

This paper presents a low-velocity impact study on two types of structures, namely (1) planar multi-cellular auxetic structures (AUS) composed of multiple re-entrant cell structures made of unidirectional carbon fiber reinforced composite (CFRP) laminate, and (2) sandwich CFRP-AUS structures (Al/CFRP-AUS), whereby the CFRP-AUS was sandwiched by aluminium plates. The experimental results reveal the mechanical behaviors of CFRP-AUS under quasi-static compression and drop hammer impact loading, and the mechanical behavior of Al/CFRP-AUS under drop impact loading. The energy absorption of the CFRP-AUS associated with quasi-static compression is greater than that associated with drop hammer impact, which is consistent with the observed differences in failure modes. The impact energy absorption capacity of the Al/CFRP-AUS is slightly higher than that of the CFRP-AUS due to the interaction between the plates and the AUS. The corresponding finite element analysis was performed and the drop hammer impact of the multi-layered CFRP-AUS was predicted. In conclusion, the CFRP-AUS structures have good energy absorption capacity during impact loading, and the known complex mechanical behaviors of deformation, failure and contact can provide guidance for the design of energy absorption box and bumper in engineering application.