Li Zhu, Feng Zhang, Pengyu Chen, Xiang Wan, Chee Leong Tan, Huabin Sun, Yong Xu, Shancheng Yan, Gang He, Changjin Wan, Zhihao Yu
Neuromorphic visual systems, integrating broadband optical perception, memory, and computation, offer a promising pathway toward next-generation intelligent sensing and brain-inspired computing. Here, we report a TeSeO/InZnO heterojunction optoelectronic synaptic transistor that leverages a narrow-bandgap/wide-bandgap coupling strategy to achieve broadband photoresponse spanning from UV to NIR (360–1550 nm). Under UV–visible–NIR illumination, the device exhibits tunable synaptic functionalities, including excitatory postsynaptic current (EPSC), paired-pulse facilitation (PPF), and controllable memory transition processes. Moreover, the reservoir computing framework constructed on this device demonstrates a rich dynamic state space, achieving up to 93% accuracy in colored object recognition tasks. Furthermore, it effectively solves second-order nonlinear dynamical equations with a minimal normalized root-mean-square error of 4.19 × 10 –4 . This work establishes a material and device foundation for broadband neuromorphic visual perception and high-dimensional dynamic computing.