The field of optoelectronic neuromorphic devices has witnessed remarkable progress in recent years. Significant breakthroughs in materials, device structures, functionalities, and integration strategies have been made. Here, we are proud to present this Special Issue on "optoelectronic neuromorphic devices", which includes five review articles and seven research articles. Briefly, Bi et al. comprehensively review recent advancements in electrolyte-gated optoelectronic neuromorphic transistors^[1]. In this review, they provide an overview of electrolyte-gated transistors (EGTs) for neuromorphic applications such as synaptic plasticity, spatiotemporal integration, and artificial perceptual systems. The potential applications of EGT optoelectronic synapses in different areas such as artificial visual system, pain system, and tactile perception systems are elaborated alongside with their future directions for neuromorphic computing development. Meng et al. give an overview on recent progress in ion-modulation optoelectronic neuromorphic devices^[2]. They elucidate the key mechanisms underlying ionic modulation of light fields, including ion migration dynamics and capture/release of charges through ions. In addition, the synthesis of active materials and the properties of these devices are analyzed in detail. In addition, the authors highlight the application of ion-modulation optoelectronic devices in artificial vision systems, neuromorphic computing, and other bionic fields. Wang et al. review the adaptive optoelectronic transistor for intelligent vision systems^[3]. Based on a description of the biological adaptive functions that are favorable for dynamically perceiving, filtering, and processing information in the varying environment, the article summarizes the representative strategies for achieving these adaptabilities in optoelectronic transistors, including the adaptation for detecting information, adaptive synaptic weight change, and history-dependent plasticity. The key points of the corresponding strategies are comprehensively discussed, and the applications of these adaptive optoelectronic transistors including the adaptive color detection, signal filtering, extending the response range of light intensity, and improving learning efficiency are also illustrated. Ye et al. review recent progress in silicon carbide (SiC)-based synaptic devices^[4]. An in-depth discussion is conducted regarding the categories, working mechanisms, and structural designs of SiC-based synaptic devices. Subsequently, several application scenarios for SiC-based synaptic devices are presented. Finally, a few perspectives and directions for their future development are outlined. Guo et al. summarize the recent advances of organic optoelectronic synaptic transistor arrays (OOSTAs)^[5]. Various strategies for manufacturing OOSTAs are introduced, including coating and casting, physical vapor deposition, printing, and photolithography. Furthermore, innovative applications of the OOSTA system integration are discussed, including neuromorphic visual systems and neuromorphic computing systems. At last, challenges and future perspectives of utilizing OOSTAs in real-world applications are provided.

In this context, optical signals present unique opportunities. Compared to electrical signals, optical signals offer inherent advantages such as ultra-high speed, broad bandwidth, immunity to electromagnetic interference, and heightened environmental sensitivity. By leveraging principles from neurobiology such as optogenetics, researchers have begun integrating optical capabilities into neuromorphic devices. This integration not only enables efficient optoelectronic signal sensing and conversion, but also enhances the performance and functionality of neuromorphic devices and their associated neural networks. These advances are paving the way toward fully integrated systems that combine sensing, memory, and computation, laying the groundwork for the next-generation computing architectures.

In addition to the aforementioned review articles, exciting research progresses have also been made in the research area of optoelectronic neuromorphic devices. Yang et al. report their recent research of synaptic nano-devices based on Ga₂O₃ nanowires^[6]. They demonstrate a deep-UV-photo-excited synaptic nano-device based on Ga₂O₃ nanowires with low energy consumption. Under 255 nm light stimulation, the biomimetic synaptic nano-device can stimulate various functionalities of biological synapses such as pulse facilitation, peak time-dependent plasticity and memory learning ability with a low energy consumption of 2.39 × 10^‒11 J for a synaptic event. The device achieves an accuracy of digit recognition exceeding 90% after 12 training sessions for the application of neural morphological computation. Zhong et al. report electropolymerized dopamine-based memristors using threshold switching behaviors for artificial current-activated spiking neurons^[7]. Their polydopamine-based memristors demonstrate the improvements in key performance, including a low threshold voltage of 0.3 V, a thin thickness of 16 nm, and a high parasitic capacitance of about 1 μF∙mm⁻². Furthermore, they construct a capacitor-free and low-power artificial spiking neuron capable of outputting the oscillation voltage, whose spiking frequency increases with the increase of current stimulation analogous to a biological neuron. Pu et al. report the development of an artificial visual synapse that integrates optical sensing and synaptic functions using a reconfigurable organic photovoltaic cell based on PM6: Y6 system^[8]. The device has stable and gradual resistance change, successfully simulating voltage-controlled long-term potentiation/depression (LTP/LTD) and exhibiting various photo-electric synergistic regulation of synaptic plasticity. The non-volatile Au/PM6: Y6/ITO memristor is used as an artificial synapse and neuron modeling, building a hierarchically coordinated processing single-layer perceptron-convolutional neural network (SLP-CNN) cascade neural network for visual image recognition training. The linear tunable photoconductivity of the device serves as the weight update of the network, enabling a recognition accuracy of up to 93.4%. Zhan et al. report their research progress on a nanowatt-level optoelectronic artificial synaptic device based on a GaN heterostructure for associative learning and neuromorphic computing^[9]. They fabricate a TiO₂/AlGaN/GaN synaptic phototransistor and mimic the synaptic weight and the synaptic cleft. Multiple synaptic neuromorphic functions such as short-term/long-term plasticity (STP/LTP) and paired-pulse facilitation (PPF) are effectively mimicked. In addition, they demonstrate that the device can achieve "retraining" process to extend memory time through enhancing the intensity of synaptic weight, which is similar to the working mechanism of a human brain. Jiang et al. report a programmable, low-loss optical nonlinear activation function (ONAF) device based on a silicon micro-ring resonator capped with antimony selenide (Sb₂Se₃) thin films, and ITO used as the microheater^[10]. Leveraging their self-developed phase-transformation kinetic and optical models, they successfully simulate the phase-transition behavior of Sb₂Se₃ and three different ONAFs—exponential linear unit (ELU), rectified linear unit (ReLU), and radial basis function (RBF), according to discernible optical responses of proposed devices under different phase-change extents. Classification results from the Fashion MNIST dataset demonstrate that these ONAFs can be considered as appropriate substitutes for traditional NAF. Ma et al. report a research progress on reconfigurable organic ambipolar optoelectronic synaptic transistors for information security access^[11]. They fabricate a reconfigurable ambipolar optoelectronic synaptic transistor based on poly (3−hexylthiophene) (P3HT) and poly [[N,N−bis(2−octyldodecyl)−napthalene−1,4,5,8−bis(dicarboximide)−2,6−diyl]−alt−5,5′−(2,2′−bithiophene)] (N2200) blended films. The resulting transistor exhibits a relatively large ON/OFF ratio of 10³ in both n- and p-type regions, and tunable photoconductivity after light illumination, particularly with green light. It is important that the reconfigurable optoelectronic properties of the device enable the realization of information light assisted burn after reading. Tian et al. show an optoelectronic memristor based on the structure of Au/a-C:Te/Pt for muti-mode reservoir computing^[12]. Synaptic functions including excitatory post-synaptic current and PPF are successfully mimicked with the memristor under electrical and optical stimulations. More importantly, the device exhibits distinguishable response currents by adjusting 4-bit input electrical/optical signals. A multi-mode reservoir computing (RC) system is constructed with the optoelectronic memristors to emulate tactile-visual fusion recognition with an accuracy of 98.7%.

The current special issue has captured some momentum of the rapidly evolving field of optoelectronic neuromorphic devices, serving as a platform for disseminating recent findings. We sincerely hope that this special issue will be a valuable resource for researchers, inspiring further innovation and collaboration. We extend our heartfelt gratitude to all the authors for their exceptional contributions to this special issue, and to the editorial and production staff of the Journal of Semiconductors for their invaluable assistance.

The rapid advancement of artificial intelligence (AI), recognized with Nobel Prizes in both Physics and Chemistry in 2024, has been revolutionizing countless aspects of modern life, driving innovations across diverse fields and reshaping how people interact with technology. Despite its great advances, the computing of AI systems is now approaching the limit of traditional computing based on the von Neumann architecture, which is characterized by separated memory and process units—a challenge known as the "von Neumann bottleneck". Neuromorphic computing inspired by the architecture and functionality of biological neural systems has emerged as a promising alternative to the computing based on the von Neumann architecture. This paradigm holds transformative potential for applications in AI, neural networks, brain-machine interfaces, and beyond.