神经形态工程学
对偶(语法数字)
激发
计算机科学
化学
神经科学
光电子学
材料科学
物理
计算机体系结构
纳米技术
心理学
人工智能
人工神经网络
艺术
文学类
量子力学
作者
Xiang He,Jie Wang,Jingwei Fu,Jianyu Ming,L. Wang,Wen Li,Chaoyue Zheng,Haifeng Ling
摘要
Implementing reconfigurable excitatory–inhibitory bidirectional weight updates in a unipolar transistor is highly desirable for developing compact neuromorphic hardware but remains a significant challenge. In this study, we employ a dual-mechanism approach by designing an organic electrolyte gate-selected memtransistor (OESmT) to reliably and rapidly update synaptic weights in both directions. The switching between synaptic inhibition and excitation depends on the connection state of the gate (modulation terminal), which can be either grounded or floating. This switching mechanism is driven by the dynamics between the electric-double-layer (EDL) and charge-trapping effects, resulting from the antagonistic effect of the two mechanisms. Long-term synaptic plasticity in both directions is demonstrated. The reconfigurable characteristic of the OESmT remains stable for 5000 pulses applied. Besides, the device has been applied to the autonomously guided vehicle. Our research highlights the importance of coupling of EDL and charge-trapping effects in reconfigurable neuromorphic systems.
科研通智能强力驱动
Strongly Powered by AbleSci AI