材料科学
神经形态工程学
异质结
光电子学
可重构性
二硫化钼
突触
范德瓦尔斯力
突触可塑性
记忆电阻器
非易失性存储器
纳米技术
计算机科学
人工神经网络
电子工程
神经科学
人工智能
分子
物理
化学
电信
生物化学
受体
量子力学
工程类
冶金
生物
作者
Yuan Li,Min Zhao,Xikui Ma,Liang Zhang,Shangzhou Zhao,Włodek Strupiński,Xiangyu Zeng,Mingjia Zhang,Yufeng Hao
标识
DOI:10.1002/adfm.202423333
摘要
Abstract Memtransistors with nonvolatile storage, reconfigurability, and simulated synaptic functions are critical to overcoming the traditional von Neumann computer architecture bottleneck. Emerging two‐dimensional van der Waals heterostructures (vdW) are promising candidates for constructing advanced three‐terminal memtransistors by integrating the intriguing features of different materials and offering additional controllability over their existing optoelectronic properties. Herein, molybdenum disulfide (MoS 2 )/Tellurene (Te) vdW p‐n junction memtransistors are fabricated to mimic the plasticity, multi‐bit memory, and paired‐pulse facilitation behavior of biological synapses. The high surface potential difference and charge trapping of the MoS 2 /Te heterostructure can endow the device with reconfigurable functionality through the transformation from short‐term plasticity to long‐term plasticity under illumination. Meanwhile, optoelectronic synaptic memtransistors also demonstrate nonvolatile behavior with a long retention time up to several hours, which can realize optical potentiation and electrical depression in one synaptic activity. On this basis, a logical operation of “OR” is realized by controlling the optical and electrical inputs. Moreover, artificial neural network training is performed to achieve a high recognition accuracy of 87.8% for handwritten digit recognition, demonstrating the potential of the artificial optoelectronic synapses in neuromorphic calculation.
科研通智能强力驱动
Strongly Powered by AbleSci AI