材料科学
视网膜
神经形态工程学
遗忘
光电子学
神经科学
人工神经网络
人工智能
计算机科学
生物
哲学
语言学
作者
Xuemei Zheng,Mengji Dong,Qi Li,Lu Wang,Yanli Liu,Xuan Di,Qilin Hua,Longfei Wang,Jianping Meng,Zhou Li
标识
DOI:10.1002/adfm.202420612
摘要
Abstract As the most promising candidates for implementing in‐sensor computing and lower power consumption, retina‐inspired morphometric vision sensors must be able to perceive and discriminate images under various illumination conditions. However, integrating invisible ultraviolet (UV) light sensing and memory in a device is valuable for exploring advanced neuromorphic visual perception systems. Here, a retina‐inspired artificial neuromorphic vision sensor based on GaN/BiFeO 3 heterojunction, which integrates sensing, memory, and processing functions, is presented. The device exhibits immediate switchable synapse plasticity‐modulated ultra‐weak UV illumination, including excitatory postsynaptic current (EPSC), paired‐pulse facilitation (PPF), short‐term and long‐term plasticity (STP/LTP), and the learning‐forgetting‐relearning process. The maximum PPF index of the device is 200%, and the minimum excitation light intensity is 0.68 µW cm −2 . The phenomenon is attributed to the effective UV light modulation for negative polarization, ionization and de‐ionization of the oxygen vacancies, further enhancing the persistent photoconductive (PPC) effect. In addition, the device successfully mimics human‐like visual memory behavior by identifying the “BINN” letter pattern, which is extracted and visualized by invisible UV information from the learning‐forgetting‐relearning process. This work provides valuable insights into next‐generation low‐power and multi‐functional photovoltaic artificial synapses.
科研通智能强力驱动
Strongly Powered by AbleSci AI