Accepted
, , Received Date: 2025-08-25
Abstract +
,
Abstract +
Density functional theory (DFT) stands as the predominant workhorse for electronic structure calculation across physics, chemistry, and materials science. However, its practical application is fundamentally constrained by a computational cost that scales cubically with system size, rendering high-precision studies of complex or large-scale materials prohibitively expensive. This review addresses the pivotal challenge by surveying the rapidly evolving paradigm of integrating machine learning (ML) with first-principles calculations to dramatically accelerate and scale electronic structure prediction. Our primary objective is to provide a comprehensive and critical overview of the methodological advances, physical outcomes, and transformative potential of this interdisciplinary field.
The core methodological progression involves a shift from black-box property predictors to symmetry-preserving, transferable models that learn the fundamental Hamiltonian—the central quantity from which diverse electronic properties derive. We detail this evolution, beginning with pioneering applications in molecular systems using graph neural networks (e.g., SchNOrb, DimeNet) to predict energies, wavefunctions, and Hamiltonian matrices with meV-level accuracy. The review then focuses on the critical extension to periodic solids, where preserving symmetries like E(3)-equivariance and handling vast configurational spaces are paramount. We systematically analyze three leading model families that define the state-of-the-art: the DeepH series, which employs local coordinate message passing and E(3)-equivariant networks to achieve sub-meV accuracy and linear scaling; the HamGNN framework, built on rigorous equivariant tensor decomposition, excelling in modeling systems with spin-orbit coupling and charged defects; and the DeePTB approach, which leverages deep learning for tight-binding Hamiltonian parameterization, enabling quantum-accurate simulations of millions of atoms.
These methods yield significant physical results and computational breakthroughs. Key outcomes include: 1) Unprecedented accuracy and speed. Models consistently achieve Hamiltonian prediction mean absolute errors (MAE) below 1 meV (e.g., DeepH-E3: ~0.4 meV in graphene; HamGNN: ~1.5 meV in QM9 molecules), coupled with computational speedups of 3 to 5 orders of magnitude compared to conventional DFT. 2) Scale bridging. Successful applications now span from small molecules to defect-containing supercells with over 10,000 atoms (e.g., HamGNN-Q on a 13,824-atom GaAs defect) and even to millions of atoms for optoelectronic property simulations (DeePTB). 3) Expanded application scope. The review highlights how these ML-accelerated tools are revolutionizing research in previously intractable areas: predicting spectroscopic properties of molecules (e.g., DetaNet for NMR/UV-Vis spectra), elucidating electronic structures of topological materials and magnetic moiré systems, computing electron-phonon coupling and carrier mobility with DFT-level accuracy but far greater efficiency (HamEPC framework), and enabling high-throughput screening for materials design.
In conclusion, ML-accelerated electronic structure calculation has matured into a powerful paradigm, transitioning from a proof-of-concept to a tool capable of delivering DFT-fidelity results at dramatically reduced cost for systems of realistic scale and complexity. However, challenges remain, including model interpretability ("black-box" nature), transferability to unseen elements, and seamless integration with existing plane-wave DFT databases. Future directions point towards physics-constrained unsupervised learning (e.g., DeepH-zero), development of more universal and element-agnostic architectures, and the creation of closed-loop, artificial intelligence (AI)-driven discovery pipelines. By overcoming current limitations, these methods hold the potential to fundamentally reshape the materials research landscape, accelerating the journey from atomistic simulation to rational material design and discovery.
The core methodological progression involves a shift from black-box property predictors to symmetry-preserving, transferable models that learn the fundamental Hamiltonian—the central quantity from which diverse electronic properties derive. We detail this evolution, beginning with pioneering applications in molecular systems using graph neural networks (e.g., SchNOrb, DimeNet) to predict energies, wavefunctions, and Hamiltonian matrices with meV-level accuracy. The review then focuses on the critical extension to periodic solids, where preserving symmetries like E(3)-equivariance and handling vast configurational spaces are paramount. We systematically analyze three leading model families that define the state-of-the-art: the DeepH series, which employs local coordinate message passing and E(3)-equivariant networks to achieve sub-meV accuracy and linear scaling; the HamGNN framework, built on rigorous equivariant tensor decomposition, excelling in modeling systems with spin-orbit coupling and charged defects; and the DeePTB approach, which leverages deep learning for tight-binding Hamiltonian parameterization, enabling quantum-accurate simulations of millions of atoms.
These methods yield significant physical results and computational breakthroughs. Key outcomes include: 1) Unprecedented accuracy and speed. Models consistently achieve Hamiltonian prediction mean absolute errors (MAE) below 1 meV (e.g., DeepH-E3: ~0.4 meV in graphene; HamGNN: ~1.5 meV in QM9 molecules), coupled with computational speedups of 3 to 5 orders of magnitude compared to conventional DFT. 2) Scale bridging. Successful applications now span from small molecules to defect-containing supercells with over 10,000 atoms (e.g., HamGNN-Q on a 13,824-atom GaAs defect) and even to millions of atoms for optoelectronic property simulations (DeePTB). 3) Expanded application scope. The review highlights how these ML-accelerated tools are revolutionizing research in previously intractable areas: predicting spectroscopic properties of molecules (e.g., DetaNet for NMR/UV-Vis spectra), elucidating electronic structures of topological materials and magnetic moiré systems, computing electron-phonon coupling and carrier mobility with DFT-level accuracy but far greater efficiency (HamEPC framework), and enabling high-throughput screening for materials design.
In conclusion, ML-accelerated electronic structure calculation has matured into a powerful paradigm, transitioning from a proof-of-concept to a tool capable of delivering DFT-fidelity results at dramatically reduced cost for systems of realistic scale and complexity. However, challenges remain, including model interpretability ("black-box" nature), transferability to unseen elements, and seamless integration with existing plane-wave DFT databases. Future directions point towards physics-constrained unsupervised learning (e.g., DeepH-zero), development of more universal and element-agnostic architectures, and the creation of closed-loop, artificial intelligence (AI)-driven discovery pipelines. By overcoming current limitations, these methods hold the potential to fundamentally reshape the materials research landscape, accelerating the journey from atomistic simulation to rational material design and discovery.
,
Abstract +
This paper develops a regularized lattice Boltzmann method for efficiently simulating the flow of N-phase immiscible incompressible fluids based on the phase field model that satisfies conservation and compatibility. By designing auxiliary moments, this method can accurately recover the second-order Allen-Cahn equation and the modified momentum equation. The correctness and effectiveness of the developed N-phase regularized lattice Boltzmann method are validated through numerical simulations of three-phase liquid lens spreading and Kelvin-Helmholtz instability phenomena. Finally, numerical simulations and analyses of three-phase Rayleigh-Taylor instabilities (RTI) were conducted, focusing on the evolution of the phase interface within the Reynolds number range of $ 500 \leqslant Re \leqslant 20000 $ (particularly under high Reynolds number condition of $ Re = 20000 $). Quantitative analyses were performed on the amplitude variations of bubbles and spikes at the two interfaces, as well as the changes in dimensionless velocity. We found that as the Reynolds number increases, the phase interface curls up at multiple locations due to Kelvin-Helmholtz instability, making the fluid more prone to dispersion and fragmentation. This study also simulated the evolution of RTI under different interface perturbations. The results indicate that RTI first develops at the perturbed interface, and its evolution gradually triggers instability at another interface.
, , Received Date: 2025-08-10
Abstract +
The optical vortex (OV) and spatiotemporal optical vortex (STOV) are special beams carrying different forms of orbital angular momentum (OAM). The OV has longitudinal OAM, whereas the STOV has transverse OAM and is coordinated with time to achieve control. Due to their reliance on different physical mechanisms, traditional optical platforms are difficult to independently control these two vortex beams on the same platform. This limitation, to some extent, hinders the understanding of the unified physical mechanism underlying spatial and spatiotemporal orbital angular momentum and also slows the development of multi-dimensional light field manipulation technology. This paper proposes a terahertz (THz) metasurface device based on vanadium dioxide (VO2) phase change material. The device integrates in-plane asymmetry, provided by triangular pores and required to excite STOV, with anisotropic phase units, realized by VO2 broken rings and needed to generate OV, into one metasurface platform, This integration enables dynamic switching of OV and STOV on the same metasurface platform. The uniqueness of its design and the key to achieving functional integration lie in the selection of Si and VO2 materials for the upper layer of the metasurface. When VO2 is in the insulating state, its dielectric constant in the THz band is similar to that of Si and its conductivity is very low. Different rotation angles of the units can still be considered as a periodic structure with the same symmetry on a macroscopic scale. The structure uses circularly polarized waves for reflection, generating a topological dark point at approximately 1.376 THz and a topological dark line between 1.3765 THz and 1.378 THz, which excites STOV. When VO2 transforms into a metallic state, its high conductivity makes the broken ring a dominant scatterer. By reasonably arranging the encoded units of the metasurface and combining the Pancharatnam-Berry (PB) phase, not only can OV with different topological charges be generated, but also multi-channel and multi-functional OV can be created through convolution theorem and shared aperture theorem. Subsequently, the influence of structural parameters is analyzed in detail. By changing the shape of the triangular pores and the thickness of the broken ring, the two vortex beams are adjusted, and it is found that they have strong topological stability under different conditions and can be reversibly switched through temperature control. This research provides a new idea for realizing multifunctional vortex light generation in the terahertz frequency band and opens up new avenues for the application of vortex light in terahertz communication and optical information processing.
, , Received Date: 2025-08-13
Abstract +
Understanding the intrinsic correlation between helium concentration and the evolution of defects as well as mechanical properties in low-activation steel on an atomic scale is crucial for designing fusion materials with excellent resistance to swelling and embrittlement. This study investigates the effect of helium concentration on single-crystal iron through molecular dynamics simulations, thereby clarifying the mechanisms by which helium concentration affects helium defect evolution, mechanical properties, and plastic deformation behavior of low-activation steel on an atomic scale. Models of body-centered cubic (BCC) iron with different helium concentrations (0.5%—4.5%) are established. Wigner-Seitz cell analysis and cluster clustering methods are employed to track the evolution of Frenkel Pairs (FPs) and cluster defects, revealing the mechanism of helium concentration-induced FPs and cluster formation at 500 ℃. Furthermore, combined with tensile mechanical simulations, the effects of helium behavior on the mechanical properties of single-crystal iron, such as elastic modulus, yield strength, and toughness, are analyzed, and the correlation mechanisms between helium concentration-induced defect evolution, mechanical properties, and plastic deformation behavior are revealed. The results show that when NHe < 3.0%, the number of FPs linearly reaches to a peak and then stabilizes. This is because helium behavior causes a rapid increase in the number of FPs and a large number of interstitial atoms are generated, some of which recombine. The annihilation rate of FPs increases with their number increasing and eventually equals the generation rate, resulting in a stable number of FPs. When NHe ≥ 3.0%, the initial increase and stabilization are the same as those for NHe < 3.0%. However, after the formation of large interstitial clusters, they absorb interstitial atoms and grow, hindering recombination and reducing the annihilation rate of FPs, thus leading to a secondary increase. The large clusters are surrounded by vacancies and no longer hinder FP recombination, and a new balance is achieved, resulting in a secondary stabilization of the FP number. When NHe increases to 3.0%, the elastic modulus, yield strength, and toughness of single-crystal iron decrease by 21%, 88%, and 57%, respectively; beyond this concentration, the mechanical properties no longer decrease. This is because when NHe < 3.0%, as helium concentration increases, helium-induced defects increase, leading to a decrease in toughness and promoting dislocation nucleation, thus reducing the elastic modulus and yield strength. When NHe ≥ 3.0%, dislocations exist in the initial defects, and the number of clusters changes slightly; toughness no longer decreases, and dislocation nucleation is not affected, leading to the stabilization of elastic modulus and yield strength. At NHe = 3.0%, the formation of large clusters hinders the movement of slip systems, changes the orientation of slip planes, weakens the effectiveness of the main slip system, which leads to an increase in small slip bands and causes the plastic deformation mechanism to transform from cross-slip to decomposition into discrete dislocations and point defects once the slip bands intersect with each other. This study reveals the influence patterns and key mechanisms of helium concentration on defect evolution and mechanical properties of single-crystal iron, providing a theoretical basis for designing fusion iron-based materials.
, , Received Date: 2025-07-09
Abstract +
The spatial attitude and dynamic performance of the cold mass support system for superconducting magnets are critical for engineering applications. This study aims to develop a design method for the spatial attitude of tie rods through a series of theoretical derivations and simulations, enabling superconducting magnets to possess a certain degree of dynamic environmental adaptability. This paper first constructs a mathematical model of the three-dimensional cold mass support system under impact loads. Stress formulas for the tie rod under vertical 5g, axial 3g, and lateral 3g impact loads are derived. Based on this, a penalty term for stress differences is introduced to construct the objective function, and the spatial inclination angle of the tie rod is optimised. After determining the acute angle between the tie rod and the coordinate axis, the cold mass support structure exhibits four different attitudes. In order to keep the natural frequency of the magnet away from the main excitation frequency band of vehicle transportation, this study uses the finite element method to perform modal analysis and proposes a method for posture design based on the principle of maximising the first-order natural frequency. Finally, random vibration simulations are conducted for the vibration environment of highway transportation. Reference points are established at both ends of the axis of the magnet body components and the room-temperature tube axis. The displacement response power spectral density (PSD) curves and root mean square values of the reference points during vibration are analysed. The conclusions of this study are as follows. 1) When the acute angles α, β, and γ included between the tie rod and the vertical, axial, and lateral directions are 31.22°, 68.50°, and 68.50°, respectively, the mechanical performance of the three-dimensional cold mass support system reaches its optimal state. 2) When the tie rod is installed in the spatial attitude configuration, the first-order natural frequency of the cold mass system is the highest, with a value of 125.99 Hz. 3) During long-distance integrated vehicle transportation, the maximum values of the vertical and lateral displacements of the magnet assembly axis relative to the room-temperature tube axis are both less than 0.1 mm. The maximum stress locations are both at the root of the carbon fibre tie rod, far below the strength limit of carbon fibre composite materials, indicating that the superconducting magnet possesses a certain degree of dynamic environmental adaptability. These analysis results provide theoretical guidance and data support for the structural safety and stability of this type of superconducting magnet during long-distance integrated vehicle transportation.
Abstract +
,
Abstract +
Covalent organic frameworks (COFs) have emerged as promising substrates for surface-enhanced Raman scattering (SERS) due to their highly ordered crystalline porous architecture, superior molecular adsorption and enrichment capabilities, and excellent thermal and chemical stability. However, pure COFs inherently lack plasmonic resonance and free electron density, resulting in limited electromagnetic enhancement and overall weak SERS signal, which hinders their practicality in ultrasensitive molecular detection applications. To overcome these limitations, this study aims to design and synthesize a novel ruthenium-based covalent organic framework composite (Ru-COF) by integrating ruthenium complexes directly into the COF skeleton, thereby creating a metal-organic, synergy-enhanced SERS substrate suited for trace analysis in real water.A Ru-COFis synthesized by solvothermal condensation of 1, 2, 4, 5-benzenetetramine (BTA·4HCl) with tris (4, 4’-dicarboxy-2, 2’-bipyridyl) ruthenium, forming Ru-N/O coordinated nodes within the framework. The material is characterizedusing X-ray diffraction (XRD) to confirm enhanced π-π stacking and new crystalline peaks at 10.2° and 16° in Ru-COF, Fourier-transform infrared spectroscopy (FT-IR) to verify amide and benzimidazole bond formations with shifts indicating Ru integration, Brunauer-Emmett-Teller (BET) analysis to reveal the increased specific surface areas (22.5 m2/g for Ru-COF vs. 17.2 m2/g for COF), and scanning electron microscopy (SEM) with energy-dispersive X-ray spectroscopy (EDS) mapping to show uniform distribution of C, N, O, and Ru elements in a dense layered morphology. SERS performance is evaluated using methylene blue (MB) as a probe molecule on a Renishaw InVia Raman spectrometer (514.5 nm excitation, 40 mW power, 10 s exposure), with additional tests on 4-mercaptobenzoic acid (4-MBA) for universality assessment. Enhancement mechanisms are analyzed through energy level alignments, with Ru-COF’s HOMO/LUMO at –0.95 eV/–1.12 eV (vs. vacuum) facilitating hole-injection charge transfer to MB’s levels (–2.34 eV/–4.15 eV), enhancing polarizability derivatives and Raman cross-sections via Herzberg-Teller coupling. The results demonstrate that Ru-COF exhibits superior SERS activity compared with pure COF and Ag-COF. For MB detection, the characteristic peak at 1624 cm–1 shows an analytical enhancement factor (EF) of 1.83 × 1010, calculated from normalized intensities and molecular densities, which far exceeds COF’s performance. Concentration-dependent spectra reveal a linear response from 10–3 to 10–13 M (R2 = 0.997), with a limit of detection (LOD, S/N = 3) of 4.16 × 10–12 M. Signal reproducibility is excellent, with a relative standard deviation (RSD) of 3.41% across 10 random spots. Cycling tests (5 repetitions) retain 90.2% of initial intensity, and long-term stability assessment shows 85.7% signal retention after four-months of air exposure. For 4-MBA, non-resonant enhancement yields an LOD of 10–12 mol/L, dominated by CM via interfacial coordination and π-π interactions. In complex matrices such as tap and river water, Ru-COF maintains LODs of 5.2 × 10–12 mol/L and 6.8 × 10–12 mol/L, respectively, with 91% signal retention after five cycles, demonstrating robust anti-interference against ions (e.g., Cl–, SO42–) and organic impurities, attributed to the hydrophobic porous structure and stable Ru coordination. In conclusion, the Ru-COF composite represents a breakthrough in SERS substrate design by achieving ultrasensitive detection through EM-CM synergy, with key physical outcomes including high EF, sub-picomolar LODs, and exceptional spatiotemporal stability. This work provides a novel paradigm for metal-embedded COFs in plasmonic sensing and lays the groundwork for practical applications in environmental monitoring, food safety, and biomedical diagnostics.
,
Abstract +
Motivation : With the rapid development of the low-altitude economy, increasing attention has been paid to the radiation environment safety of low-altitude aircraft such as drones and electric vertical takeoff and landing (eVTOL) aircraft. Traditional views hold that the dense lower atmosphere is an effective barrier against cosmic radiation, but the shrinking feature sizes of modern integrated circuits (ICs) have significantly increased their susceptibility to single-event effects (SEEs). Most conventional studies have focused on the effects of particles such as neutrons and protons, while systematic evaluations of the risks induced by muons —the most abundant charged particles at sea level—remain scarce, particularly during extreme solar events. Therefore, this study quantitatively evaluates the muon-induced SEE risks of lowaltitude aircraft in different regions of China under both static cosmic ray backgrounds and Ground Level Enhancements (GLEs), aiming to provide critical insights for the operational safety of next-generation low-altitude aviation platforms.
Methods : This study employs city-specific atmospheric models and simulates atmospheric shower processes over different cities within the CORSIKA framework, yielding reliable energy spectra of lowenergy muons (10–100 MeV) across diverse regions. Drawing on electrical simulation data from other research groups, this study estimates muon-induced SEE cross sections in transistors with different process nodes, covering Bulk, FD-SOI, and FinFET processes. Subsequently, by integrating solar energetic particle (SEP) energy spectra associated with Ground Level Enhancement (GLE) events, we evaluate muoninduced SEE risks for systems of varying sizes under both static conditions (only cosmic-ray injection) and GLE event scenarios.
Results : Our results indicate that under static conditions, flight control systems (with 1 MB of memory) incorporating advanced process-node (≤ 45nm) Bulk transistors are exposed to non-negligible muon-induced SEE risks across all cities in China. In contrast, systems utilizing FD-SOI transistors can effectively alleviate such risks. For systems with large memory capacities (1 GB), irrespective of the process technology employed, redundancy and other radiation-hardening measures must be adopted. Regarding GLE events, this study innovatively introduces the concept of muon hazard levels to evaluate regional variations in risk. Specifically, during GLEs, the aggravation of muon-induced SEE risks in mid-to-low latitude regions is negligible, whereas high-latitude regions experience a significant rise in such risk.
Methods : This study employs city-specific atmospheric models and simulates atmospheric shower processes over different cities within the CORSIKA framework, yielding reliable energy spectra of lowenergy muons (10–100 MeV) across diverse regions. Drawing on electrical simulation data from other research groups, this study estimates muon-induced SEE cross sections in transistors with different process nodes, covering Bulk, FD-SOI, and FinFET processes. Subsequently, by integrating solar energetic particle (SEP) energy spectra associated with Ground Level Enhancement (GLE) events, we evaluate muoninduced SEE risks for systems of varying sizes under both static conditions (only cosmic-ray injection) and GLE event scenarios.
Results : Our results indicate that under static conditions, flight control systems (with 1 MB of memory) incorporating advanced process-node (≤ 45nm) Bulk transistors are exposed to non-negligible muon-induced SEE risks across all cities in China. In contrast, systems utilizing FD-SOI transistors can effectively alleviate such risks. For systems with large memory capacities (1 GB), irrespective of the process technology employed, redundancy and other radiation-hardening measures must be adopted. Regarding GLE events, this study innovatively introduces the concept of muon hazard levels to evaluate regional variations in risk. Specifically, during GLEs, the aggravation of muon-induced SEE risks in mid-to-low latitude regions is negligible, whereas high-latitude regions experience a significant rise in such risk.
Abstract +
Non-Hermitian physics has emerged as a rapidly advancing field of research, revealing a range of novel phenomena and potential applications. Traditional non-Hermitian Hamiltonians are typically simulated by constructing asymmetric couplings or by introducing dissipation and gain to realize non-Hermitian systems. The quadratic bosonic system (QBS) with squeezing interaction is intrinsically Hermitian; however, its dynamical evolution matrix in both real and momentum spaces is non-Hermitian. Based on this, applying a field-operator transformation $\{\hat{x},\hat{p}\}$ to the dynamical evolution matrix yields quadrature nonreciprocal transmission between the $\hat{x}$ and $\hat{p}$ operators. This nonreciprocal characteristic can be utilized in signal amplifiers. On the other hand, within the Bogoliubov–de Gennes framework in momentum space, one can observe non-Hermitian topological phenomena such as point-gap topology and the non-Hermitian skin effect, both induced by spectra with nonzero winding numbers. Additionally, QBS can be employed to realize non-Hermitian Aharonov–Bohm cages and to extend non-Bloch band theory. Previous studies in non-Hermitian physics have largely concentrated on classical systems. The influence of non-Hermitian properties on quantum effects remains a key issue awaiting exploration and has evolved into a research direction at the interface of non-Hermitian and quantum physics. In QBS, squeezing interactions without dissipation cause the dynamical evolution of the system to display effective non-Hermitian characteristics and induce quantum correlation effects, such as quantum entanglement. Recent studies have shown that the non-Hermitian exceptional points in QBS can alter squeezing dynamics and entanglement dynamics. Therefore, such systems not only offer a natural platform for realizing quantum non-Hermitian dynamics but also constitute an important basis for investigating the relationship between non-Hermitian dynamics and quantum effects, as well as for achieving quantum control based on non-Hermitian properties. Future research may further focus on elucidating the connections between non-Hermitian dynamics and quantum effects in QBS, which is expected to serve as a bridge linking non-Hermitian dynamics and quantum effects.
Abstract +
Bistable switching materials that enable reversible transitions between distinct stable states have emerged as a transformative platform for next-generation information technologies, optoelectronics, and quantum control. The application of high pressure serves as a powerful and precisely tunable stimulus for manipulating crystal structures, electronic configurations, and crystal fields, thereby enabling deterministic switching of diverse physical properties. This review systematically examines recent advances in pressure-induced bistable transitions, encompassing nonlinear optical switching via symmetry breaking, luminescence and color transitions mediated by bandgap engineering, insulator-metal transitions driven by electronic correlation effects, semiconductor carrier-type inversion, and spin crossover phenomena. Through comprehensive analysis integrating in situ high-pressure characterization techniques including synchrotron X-ray diffraction, vibrational spectroscopy, spatially resolved photoluminescence mapping, nonlinear optical microscopy, and transport measurements, we establish quantitative correlations between structural evolution, local coordination changes, and macroscopic switching responses. These multimodal investigations reveal fundamental mechanisms governing bistable transitions, particularly highlighting the critical roles of pressure-controlled symmetry breaking, coordination reconstruction, lone-pair stereochemical activity, and electronic correlation tuning. Notably, certain material systems exhibit extended multistate switching characteristics on complex energy landscapes, offering promising avenues for advanced applications in high-density data storage beyond conventional bistability. However, practical implementation faces significant challenges including the relatively high switching pressures required, limited reversibility in some systems, and difficulties in device integration. To solve current challenges, we proposed potential solutions including the development of diamond anvil cell-integrated micro/nanoelectrodes, fiber-optic coupled on-chip high-pressure cells, and strategies to reduce switching pressures to practical ranges. This work provides fundamental insights into the mechanisms of pressure-driven state switching while simultaneously outlining practical pathways toward realizing devices and reconfigurable optoelectronic systems. The integration of advanced in situ characterization techniques with theoretical understanding offers a robust framework for both fundamental research and technological applications of bistable switching materials under pressure.
,
Abstract +
Abstract Nuclear masses are fundamental observables that reflect nuclear structure and stability, playing a key role in nuclear physics and astrophysical processes. Most existing neural network studies focus on predicting either binding energies or neutron/proton separation energies individually, with limited attention to the physical correlations between these observables. Based on the relativistic point-coupling model PCF-PK1, a physics-informed artificial neural network (ANN) was developed to systematically predict nuclear binding energies along with single- and double-neutron/proton separation energies, while preserving the physical self-consistency of the predictions. To assess the impact of incorporating separation-energy constraints, networks were trained with varying loss function weight combinations, enabling a comparison between networks without separation-energy constraints (e.g., ANN1) and those including such constraints (e.g., ANN3).
The neural network significantly improves the overall prediction accuracy of binding energies compared with the PCF-PK1 model. Without separation-energy constraints, ANN1 already achieves high precision for binding energies (RMSE ≈ 0.147 MeV) and separation energies (RMSE ≈ 0.158– 0.185 MeV). Incorporating separation-energy constraints in ANN3 results in a slight improvement in overall prediction accuracy. The binding energy predictions improve by approximately 4.6%, while the separation energy predictions increase by 8.9–12.0%. The improvement is particularly noticeable for nuclei where the deviations of ANN1 predictions from experimental values exceed 0.2 MeV. Supporting datasets are publicly accessible at the Science Data Bank (https://doi.org/10.57760/sciencedb.j00213.00239). To facilitate the review process, a private access link is provided for reviewers during the review period (https://www.scidb.cn/s/bqyemq).
The neural network significantly improves the overall prediction accuracy of binding energies compared with the PCF-PK1 model. Without separation-energy constraints, ANN1 already achieves high precision for binding energies (RMSE ≈ 0.147 MeV) and separation energies (RMSE ≈ 0.158– 0.185 MeV). Incorporating separation-energy constraints in ANN3 results in a slight improvement in overall prediction accuracy. The binding energy predictions improve by approximately 4.6%, while the separation energy predictions increase by 8.9–12.0%. The improvement is particularly noticeable for nuclei where the deviations of ANN1 predictions from experimental values exceed 0.2 MeV. Supporting datasets are publicly accessible at the Science Data Bank (https://doi.org/10.57760/sciencedb.j00213.00239). To facilitate the review process, a private access link is provided for reviewers during the review period (https://www.scidb.cn/s/bqyemq).
Abstract +
Mechanoluminescence (ML) is a phenomenon in which photon emission is produced directly under mechanical stimulation. Owing to its high spatial selectivity, rapid response, and multimodal emission capabilities, ML exhibits great potential for applications in structural health monitoring, intelligent sensing, and optical anti-counterfeiting. However, due to the complexity of ML modes, categories, and underlying kinetic processes, the field still faces several challenges, including the lack of a well-established mechanism, the limited availability of high-performance ML materials, and the absence of standardized testing standards. Existing studies have demonstrated that crystal field strength, band structure, and lattice configuration play crucial roles in governing the ML properties. High-pressure, with its unique ability to tune these physical quantities, undoubtedly provides new pathways for advancing ML research. Recent breakthroughs in rapid loading techniques have further enabled the exploration of ML behaviors under high-pressure conditions. In the GPa pressure range, modulation of interatomic distances, electronic orbitals, and crystal structures has not only allowed effective control over emission intensity and color, but has also enabled the capture of ML kinetic processes over microsecond–second timescales, thereby supplying essential experimental data for revealing the microscopic mechanisms of ML. In this review, we first provide a brief overview of the historical development, classification, and mechanistic understanding of ML, together with commonly employed ML characterization methods under ambient and high-pressure conditions. We then summarize recent progress in the application of high-pressure techniques for optimizing ML performance and elucidating ML mechanisms, highlighting advances in enhancing emission intensity, modulating spectral characteristics, and uncovering dynamic processes. Finally, the future directions and challenges for high-pressure ML research are discussed.
,
Abstract +
As the field of artificial intelligence continues to evolve, it generates an escalating need for intensive computational resources and novel computing architectures. As a new generation of non-volatile memory, memristors can simulate biological synapses. This makes them ideal for neuromorphic computing, enabling brain-like learning and reasoning to significantly enhance computational capabilities. Current research on memristor dielectric materials primarily focuses on transition of metal oxides, perovskites, and organic polymers. Among these, the transition metal oxide TiO2 is widely used for the switching layer due to its high dielectric constant and excellent thermal stability. However, TiO2-based memristors face challenges including poor stability and inadequate analog performance, which limit their application in neuromorphic computing. This study developed a high-performance analog memristor using an aMoS2/a-TiO2 (amorphous MoS2/ amorphous TiO2) heterostructure, achieving over 200 stable cycles and a long data retention time exceeding 104 seconds. This device demonstrates a lower threshold voltage, higher endurance, and superior data retention, as compared to previously reported TiO2-based heterostructure memristors. Furthermore, various voltage sweep schemes were designed to successfully implement multi-level conductance modulation in the W/a-MoS2/a-TiO2/Pt device. The resistive switching mechanism of the W/a-MoS2/a-TiO2/Pt device was elucidated by combining conductive mechanism fitting with a physical model that attributes the switching to the localized formation and rupture of conductive filaments. Finally, synaptic functions like LTP and LTD were implemented in the device using square-wave pulses. A convolutional neural network leveraging these functions achieved a 95.8% accuracy in handwritten digit recognition. This study developed a W/a-MoS2/a-TiO2/Pt heterostructure that significantly enhances analog memristive performance, providing an effective strategy for improving transition metal oxide-based memristors.
- 1
- 2
- 3
- 4
- 5
- ...
- 16
- 17
