神经形态工程学
材料科学
同质结
光电子学
计算机科学
适应(眼睛)
明视
数字光处理
微电子机械系统
人工智能
图像传感器
特征(语言学)
过程(计算)
光电导性
电阻随机存取存储器
暗视
人工神经网络
显示设备
电子工程
计算机视觉
作者
Xin Lin,Z L Li,Fang Wang,Junqing Wei,Jiale Li,Song Li,Yemei Han,S Q Wang,Tianling Ren,Wentao Xu,K L Zhang
摘要
ABSTRACT Optoelectronic synapses, constructed with two‐dimensional transition metal dichalcogenides (2D‐TMDCs) with atomic‐scale thickness, offer a promising solution to break through limitations of conventional artificial visual system with separating sensors and memory. In this work, an optoelectronic memtransistor with asymmetric‐contact featuring mono‐bi‐MoS 2 homojunctions is presented, successfully integrating light perception, information processing and memory functions. The device demonstrates visually plastic transition from short‐term to long‐term memory and further achieves “learning‐forgetting‐relearning” behavior under different light pulse stimuli. We reveal conductive mechanism driven by photoconductive effect dominated by separation of photo‐generated carriers and carrier trapping‐releasing with and without illumination. The device highly simulates dynamic visual adaptation process based on photopic and scotopic adaptation under electrical‐optical co‐stimulation, achieving target detection in low‐light complex environments. The neuromorphic visual system based on MoS 2 ‐memtransistor achieves recognition accuracy of 94.3% for normal‐light and accuracy of 92.4% for dim‐light with image feature enhancement. This work provides new insights for next‐generation multifunctional neuromorphic vision and intelligent sensing applications.
科研通智能强力驱动
Strongly Powered by AbleSci AI