Artificial Intelligence for Science (AI4S) has emerged as a transformative paradigm that fundamentally reshapes modern scientific discovery across physical and material disciplines. This systematic review comprehensively examines recent AI4S advances based on peer-reviewed publications spanning 2024 to 2026, covering atomic, molecular, and optical physics; condensed matter and materials physics; quantum information science; and cross-domain intelligent applications. The surveyed literature consolidates six core research directions: machine learning-enabled spectroscopic analysis (primarily laser-induced breakdown spectroscopy, LIBS), data-driven functional materials discovery and property prediction (including hierarchical neural network-based superconducting critical temperature forecasting), memristor-based neuromorphic devices and hardware-algorithm co-design, intelligent optical communication and quantum information processing, and reinforcement learning-governed complex systems for autonomous experimental exploration. However, persistent challenges remain: data scarcity, the disconnect between physical principles and algorithms, limited generalization ability, and the critical gap between computational prediction and experimental synthesis. Furthermore, the long-term sustainability of AI4S innovation faces systemic barriers including the lack of standardised databases, acute talent shortages, and complex legal and regulatory issues surrounding cross-border data sharing. This review argues that the future of AI4S lies not in replacing physical insight with data-driven methods, but in achieving a symbiotic integration of both — where AI accelerates discovery while being grounded in deep chemical and physical understanding. As the field matures, the conventional boundary between “AI for Science” and “Science for AI” will gradually dissolve: intelligent models will assist in uncovering unknown physical laws, while domain scientific insights will guide the design of more robust, interpretable, and discipline-adaptive AI architectures. The ultimate measure of success will be the tangible scientific outputs, including novel functional materials, high-performance devices, and practical technological breakthroughs that emerge from interdisciplinary endeavor.
The orbital angular momentum (OAM) of photons provides a pivotal resource for carrying out high-dimensional quantum Pauli-X operations due to its unique discrete infinite orthogonality, which opens enticing perspectives for high-capacity quantum communication and multi-parallel quantum computation. However, the state-of-the-art Pauli-X gates that require cyclic transformation of OAM modes are mostly limited to four modes, face scalability challenges owing to the system redundancy and fidelity degradation induced by increased processing modes. Here, we propose a general strategy to design scalable quantum Pauli-X gates in high-dimensional OAM space by taking full advantage of the order-correlated spatial divergence property of OAM modes. Leveraging the mechanism of balance mode filtering, partitioned phase is embedded between two structurally symmetric Q-plate elements that enable bidirectional mode shifting to filter out the highest-order OAM mode for spatially selective phase modulation. Derived from radial matching between modes and partitioned phase profiles, this design of parallel mode transformation underpins the extension to higher dimensions by merely adjusting filter parameters. We demonstrate the five-mode cyclic transformation with mode purity and diffraction efficiency higher than 92% and 92.7%, respectively. By incorporating polarization-entangled photon sources, we further realize five-dimensional quantum Pauli-X gates encoded on OAM space and experimentally verify the entanglement of output mode states. Low modulation loss and efficient quantum state tomography (QST) techniques endow the reconstructed density matrices with fidelities exceeding 92.3% ± 0.1%, indicating the well-performed entanglement preservation of output states. These results confirm that our approach achieves scalable quantum Pauli-X gate operations through a simple balanced architecture with high coherence, consolidating a theoretical and experimental foundation for large-scale quantum computation encoded on high-dimensional OAM space.
In recent years, magic-angle twisted bilayer graphene has emerged as a focal point in condensed matter physics due to its novel quantum properties arising from ultra-flat electronic bands. Subsequent research has also revealed analogous flat-band physics in other twisted multilayer graphene systems, such as twisted monolayer-bilayer graphene, twisted trilayer graphene, and twisted double-bilayer graphene. These graphene-based twisted systems host a rich variety of strongly correlated and topological quantum states — including ferromagnetism, superconductivity, Chern insulators, and the quantum anomalous Hall effect — significantly advancing the investigation of strongly correlated and topological physics in real materials. Here, we conduct a systematic review of the recent advances in the exploration of flat-band and strong correlation physics using scanning tunneling microscopy/spectroscopy in magic-angle twisted multilayer graphene. First, we briefly introduce the theoretical background of the flat-band physics in magic-angle twisted multilayer graphene. Subsequently, we present in detail the strongly correlated electronic states and exotic quantum phenomena observed by scanning tunneling microscopy/spectroscopy in various twisted graphene platforms, including twisted bilayer graphene, twisted monolayer?bilayer graphene, twisted trilayer graphene, and twisted double-bilayer graphene. Lastly, we provide a summary and an outlook for future research in this field.
With the continuous scaling down of channel dimensions,the relentless minia-turization of CMOS-based integrated circuits is increasingly constrained by short-channel effects,leakage currents,and quantum tunneling phenomena,suggesting that Moore's Law is approaching its physical limits.Two-dimen-sional(2D)materials,distinguished by their ultrathin nature,exceptional carrier mobility,and intrinsic passivation characteristics,offer a promising solution to mitigate short-channel effects while maintaining ultra-low leakage currents.Consequently,they are considered a viable pathway for further device miniaturization.In this review,we focus on 2D transistors,elaborating on recent advancements in synthesizing wafer-scale single-crystal 2D semicon-ductor materials,with particular emphasis on low-temperature(＜450 ℃)synthesis techniques.Additionally,this review summarizes innovative strategies for synthesizing 2D dielectric layers that exhibit superior compatible with 2D semiconductor materials.Subsequently,this review discusses strategies to opti-mize the contact between 2D semiconductor materials and metal electrodes.Crucial advancements in 2D transistor technology,encompassing large-scale integration and unconventional device architectures,are also highlighted.Finally,this review delineates the major technical challenges and potential optimization directions for 2D transistors.This review is confident that resolving these bottlenecks will catalyze unprecedented advancements in post-Moore integrated circuit technology.
In view of the fast development of vector beams and their broad application prospects,the systematical review about the characterizations,generation methods,propagation mechanisms and diverse micro-nano optical appli-cations of various vector beams are provided.Four mathematical charac-terizations of vector beams are generalized and the relation of different characterization methods is uncovered.The direct and indirect generation methods including some innovation techniques are summarized.The propagation fields of vector beams in different circumstances are analyzed and the related phenomena are explained.The diverse applications in various fields are enumerated.The detailed description of vector beams provides a theoretical basis for deeply understanding of the polarization properties of vector beams.The comparison of several generation methods brings the researchers the purposeful choices in practice works.The anal-ysis about propagation mechanisms of vector beam lays the foundation for predicting the unique phenomena and utilizing reasonably vector beams.The presented applications confirm the significance of vector beams in many fields.We expect this review will bring more help for the deep studies about foundational and applications of vector beams.
