晶体管
阈下传导
计算机科学
电气工程
物理
电压
工程类
作者
Jun Yu,Jiawei Fu,Candong Zhao,Fuwei Zhuge,Qi Chen,Yuhui He,Xiangshui Miao
标识
DOI:10.1109/imw59701.2024.10536953
摘要
We demonstrate a novel edge-contact floating gate transistor (ECFGT) based on phase engineered MoS 2 for in-memory computing. The ECFGT performs ultra-high operation speed (10ns for programming and 100ns for erasing), low operation voltage (15V) and a significantly reduced power consumption (~11.5 fJ/per program, ~41.3 fJ/per erase). Besides, good endurance (>8×10 4 ) for multi-level conductance is demonstrated, and 5-bit distinguishable states in our device are utilized as synapses for neuromorphic computing. The ultra-small read currents (10 -12 ~10 -6 A) in the subthreshold mode of the individual devices illuminate the promise of constructing very large-scale neural network through device integration.
科研通智能强力驱动
Strongly Powered by AbleSci AI