光探测
神经形态工程学
响应度
光电子学
晶体管
异质结
材料科学
计算机科学
场效应晶体管
电场
逻辑门
范德瓦尔斯力
电压
和大门
电子工程
光电探测器
NOR门
门阵列
光学计算
集成电路
纳米技术
纳米机电系统
光通信
物理
柔性电子器件
微电子
领域(数学)
电气工程
光开关
数码产品
作者
An Gui,Yutao Song,T F Li,Kaiwen Gong,Haozhe Li,Haoran Mu,Rong Yang,Lei Liao,G L Zhang,Shenghuang Lin
出处
期刊:SmartMat
[Wiley]
日期:2026-07-16
卷期号:7 (4)
摘要
ABSTRACT Machine vision systems in real‐world applications require devices capable of adapting to multiple operational modes, including high‐sensitivity photodetection, neuromorphic synaptic behavior, and optical memory. However, integrating these distinct optoelectronic functionalities into a single device presents a challenge due to their conflicting requirements for photoresponse time and retention. Here, a gate‐programmable multifunctional transistor based on a MoS 2 /MAPbBr 3 van der Waals (vdW) heterostructure is fabricated. By using the gate field to tune the interfacial electric field and trapping/release kinetics associated with shallow‐ and deep‐level states, the device can be programmed to operate as a photodetector, synaptic transistor, and gate‐bias‐assisted multilevel optical memory. Under a positive gate voltage ( V G ), a pronounced photogating effect enables photodetection with responsivity ( R ) of ∼2397 A/W. At V G = –20 V, the device switches to synaptic mode. At a more negative gate bias, the device enters a gate‐bias‐assisted optical‐memory mode under sustained gate bias. To illustrate the utility of this programmable multifunctionality, frequency‐encoded image recognition is demonstrated with 96.4% accuracy. In addition, a device‐enabled reservoir‐computing (RC) scheme with software‐assisted readout improves classification accuracy by ~20% through deep delay‐based processing. This work demonstrates a multimodal transistor architecture that integrates multiple optoelectronic functions in a single device and suggests a compact hardware route toward adaptive in‐sensor optical information processing.
科研通智能强力驱动
Strongly Powered by AbleSci AI