神经形态工程学
晶体管
材料科学
光电子学
调制(音乐)
电子迁移率
排水诱导屏障降低
阈值电压
电气工程
栅极电介质
电压
计算机科学
物理
人工神经网络
工程类
声学
机器学习
作者
C. Zhang,Baofeng Yang,Dong Wang,Zhihao Zhou,Chuanyu Han,L. X. Qian,P.T. Lai,Xiaodong Huang
标识
DOI:10.1109/ted.2023.3295789
摘要
Synaptic devices are essential for constructing neuromorphic computing. For the current synaptic transistors, their gate dielectrics are always required to possess special functions (e.g., ion migration and charge trapping) for realizing weight modulation, which degrades the carrier mobility and thus the device performance due to the Coulomb scattering. A new synaptic transistor is presented to address this issue. This device uses the metal-hydroxyl (M-OH) defect at the back channel rather than the gate dielectric for weight modulation. Driven by the gate voltage, electron trapping, and detrapping occur in the M-OH defect, which changes the channel carrier density to result in a delicate weight modulation. This transistor displays relatively high carrier mobility by suppressing the Coulomb scattering existing in the current synaptic transistors. Typical synaptic functions are also well demonstrated for this device. Moreover, simulation results prove that this synaptic transistor can provide high recognition accuracy for neuromorphic computing.
科研通智能强力驱动
Strongly Powered by AbleSci AI