Review Articles MORE

• 

  From device non-idealities to computation accuracy: A cross-layer review of error mechanisms and mitigation in analog in-memory computing

  Abstract Full Text PDF Analog in-memory computing (AIMC) has attracted significant attention as an energy-efficient computing paradigm for data-intensive workloads. However, its practical deployment is fundamentally limited by various non-idealities across multiple abstraction levels, which degrade computational accuracy and reliability. This paper presents a comprehensive review of non-idealities in AIMC systems from a cross-layer perspective. A common computational representation is introduced to illustrate how diverse physical non-idealities may manifest as weight perturbations, multiplicative distortions, nonlinear transfer effects, additive errors, and quantization or accumulation errors. From this perspective, we review design techniques across device, array, circuit, data-conversion, and system levels, and discuss how errors are generated, transformed, and accumulated throughout the computation flow. Furthermore, algorithm-hardware co-design strategies are discussed as an effective approach to mitigate cross-layer errors and enhance system robustness. By providing a structured overview of error generation, transformation, and propagation, this work highlights the importance of cross-layer co-design and offers insights into the design trade-offs among accuracy, efficiency, and scalability in AIMC systems.

  Analog in-memory computing (AIMC) has attracted significant attention as an energy-efficient computing paradigm for data-intensive workloads. However, its practical deployment is fundamentally limited by various non-idealities across multiple abstraction levels, which degrade computational accuracy and reliability. This paper presents a comprehensive review of non-idealities in AIMC systems from a cross-layer perspective. A common computational representation is introduced to illustrate how diverse physical non-idealities may manifest as weight perturbations, multiplicative distortions, nonlinear transfer effects, additive errors, and quantization or accumulation errors. From this perspective, we review design techniques across device, array, circuit, data-conversion, and system levels, and discuss how errors are generated, transformed, and accumulated throughout the computation flow. Furthermore, algorithm-hardware co-design strategies are discussed as an effective approach to mitigate cross-layer errors and enhance system robustness. By providing a structured overview of error generation, transformation, and propagation, this work highlights the importance of cross-layer co-design and offers insights into the design trade-offs among accuracy, efficiency, and scalability in AIMC systems.
• 

  A review of metastable ε/κ-Ga₂O₃: Phase identification, epitaxy, and emerging device physics

  Abstract Full Text PDF Ga₂O₃ is a leading candidate for next-generation high-power electronics and deep-ultraviolet optoelectronics due to its ultrawide bandgap and high theoretical breakdown field. While the stable β-phase has been extensively studied, interest is shifting toward metastable polymorphs to broaden functionality. Among these, the ε/κ phase is distinguished by its non-centrosymmetric orthorhombic structure, offering unique ferroelectric and piezoelectric properties. Despite progress in epitaxial stabilization, a comprehensive understanding of its crystal structure, physical behavior, and device potential remains limited. This review critically examines the ε−κ phase relationship and heteroepitaxial growth mechanisms on Al₂O₃, GaN, and other substrates, highlighting the roles of supersaturation, doping, and buffer layer engineering in stabilizing metastable phases while suppressing secondary phases. Furthermore, the work explores the unique physical properties of ε/κ-Ga₂O₃, including the substantial spontaneous polarization, a large piezoelectric coefficient, and the direct-bandgap nature favorable for DUV detection. The practical efficacy of these attributes is showcased across a diverse technological portfolio, including ferroelectric memristors for neuromorphic computing, high-frequency acoustic wave resonators, and high-sensitivity solar-blind photodetectors. Finally, the review identifies remaining challenges-such as rotational domain mitigation and polarization saturation-providing a strategic roadmap for realizing multifunctional ε/κ-Ga₂O₃-based devices.

  Ga₂O₃ is a leading candidate for next-generation high-power electronics and deep-ultraviolet optoelectronics due to its ultrawide bandgap and high theoretical breakdown field. While the stable β-phase has been extensively studied, interest is shifting toward metastable polymorphs to broaden functionality. Among these, the ε/κ phase is distinguished by its non-centrosymmetric orthorhombic structure, offering unique ferroelectric and piezoelectric properties. Despite progress in epitaxial stabilization, a comprehensive understanding of its crystal structure, physical behavior, and device potential remains limited. This review critically examines the ε−κ phase relationship and heteroepitaxial growth mechanisms on Al₂O₃, GaN, and other substrates, highlighting the roles of supersaturation, doping, and buffer layer engineering in stabilizing metastable phases while suppressing secondary phases. Furthermore, the work explores the unique physical properties of ε/κ-Ga₂O₃, including the substantial spontaneous polarization, a large piezoelectric coefficient, and the direct-bandgap nature favorable for DUV detection. The practical efficacy of these attributes is showcased across a diverse technological portfolio, including ferroelectric memristors for neuromorphic computing, high-frequency acoustic wave resonators, and high-sensitivity solar-blind photodetectors. Finally, the review identifies remaining challenges-such as rotational domain mitigation and polarization saturation-providing a strategic roadmap for realizing multifunctional ε/κ-Ga₂O₃-based devices.
• 

  Artificial intelligence enhances a new generation of intelligent flexible wearable systems

  Abstract Full Text PDF AI-integrated flexible wearable sensors combine smart materials, conformal device design, and data-driven algorithms to continuously capture and interpret physiological, motion, and environmental signals. Here, we summarize how representative machine learning and deep learning models are used for signal processing, feature extraction, multimodal fusion, and system optimization in next-generation wearable sensing platforms. We further discuss their applications in personalized healthcare, human–machine interaction, and sensory augmentation. Finally, we identify key challenges, including data quality, model generalization, interpretability, power-efficient deployment, and privacy and security, together with perspectives on the development of more reliable, scalable, and trustworthy intelligent sensing systems.

  AI-integrated flexible wearable sensors combine smart materials, conformal device design, and data-driven algorithms to continuously capture and interpret physiological, motion, and environmental signals. Here, we summarize how representative machine learning and deep learning models are used for signal processing, feature extraction, multimodal fusion, and system optimization in next-generation wearable sensing platforms. We further discuss their applications in personalized healthcare, human–machine interaction, and sensory augmentation. Finally, we identify key challenges, including data quality, model generalization, interpretability, power-efficient deployment, and privacy and security, together with perspectives on the development of more reliable, scalable, and trustworthy intelligent sensing systems.
• 

  Functional photoelectronic devices: leveraging two-dimensional van der Waals device configuration

  Abstract Full Text PDF Two-dimensional (2D) van der Waals (vdW) materials offer atomically thin, dangling-bond-free interfaces and strong electrostatic control characteristics, enabling a new paradigm for optoelectronic devices. This review focuses on vdW contact engineering and heterostructure construction. The vdW contact can suppress Fermi-level pinning, reduce contact resistance, and preserve the intrinsic properties of 2D semiconductors. Leveraging such clean interfaces, 2D vdW materials show great potential in implementing functional devices, including photodetectors, miniaturized computational spectrometers, reconfigurable devices, and neuromorphic systems. In these cases, the tunable vdW interfaces eliminate lattice-matching constraints and allow feasible heterogeneous integration. This review discusses the recent progress and future prospects in this field, providing guidance for realizing functional applications based on vdW device configuration.

  Two-dimensional (2D) van der Waals (vdW) materials offer atomically thin, dangling-bond-free interfaces and strong electrostatic control characteristics, enabling a new paradigm for optoelectronic devices. This review focuses on vdW contact engineering and heterostructure construction. The vdW contact can suppress Fermi-level pinning, reduce contact resistance, and preserve the intrinsic properties of 2D semiconductors. Leveraging such clean interfaces, 2D vdW materials show great potential in implementing functional devices, including photodetectors, miniaturized computational spectrometers, reconfigurable devices, and neuromorphic systems. In these cases, the tunable vdW interfaces eliminate lattice-matching constraints and allow feasible heterogeneous integration. This review discusses the recent progress and future prospects in this field, providing guidance for realizing functional applications based on vdW device configuration.