神经形态工程学
光子学
光电子学
物理
突触后电流
电容
能源消耗
量子
量子点
材料科学
电压
记忆电阻器
突触重量
突触
量子计算机
电容感应
逻辑门
费米能量
拓扑(电路)
能量(信号处理)
计算机科学
量子电容
人工神经网络
量子阱
电子工程
兴奋性突触后电位
和大门
晶体管
低温冷却器
延迟(音频)
范德瓦尔斯力
调制(音乐)
量子位元
光学计算
作者
Xiangyu Zeng,Yang Zhang,Xu Wang,Cizhe Fang,Zhuoqing Gao,Fazhi Yang,Xiaoxi Li,Liang Zhang,Yan Liu,Yue Hao,Genquan Han
出处
期刊:ACS Nano
[American Chemical Society]
日期:2026-02-25
标识
DOI:10.1021/acsnano.5c19004
摘要
Emerging neuromorphic systems demand devices that seamlessly integrate sensing, memory, and computing in a single element to overcome the energy and latency constraints inherent to conventional architectures. Here, we introduce a two-dimensional van der Waals photonic synapse (MoS2/h-BN/WTe2/h-BN) with the Weyl semimetal (WTe2) employed as the floating gate. As the Fermi level is set near the Weyl nodes by a minute charge tunneling, the enhanced electron-electron correlation triggers the negative carrier compressibility and finally results in the negative quantum capacitance effect. This quantum effect amplifies the gate voltage and creates a strong built-in electric field at the h-BN/MoS2 interface, which benefits both the long-term retention (approaching one year at room temperature) and short-term response (electric energy consumption as low as 0.26 fJ per event) of the photonic synapse. Additionally, the device supports key synaptic functions and applications, including excitatory postsynaptic current potentiation, paired pulse facilitation, logic gate operation, and artificial neural network handwritten-digit classification. These findings establish quantum effects as a robust approach to ultralow-energy, long-retention neuromorphic devices with compelling prospects for integrated optoelectronic computing.
科研通智能强力驱动
Strongly Powered by AbleSci AI