跳到主要导航 跳到搜索 跳到主要内容

Functionalized dendritic memristor of Pt/MoS2@LCO-PVA/Si for Mimicking synaptic behavior

  • Xueli Geng
  • , Qin Gao*
  • , Jiangshun Huang
  • , Juan Gao
  • , Pengzhan Li
  • , Yihang Liu
  • , Qiaofeng Yang
  • , Xinghe Li
  • , Hao Zhang
  • , Zhisong Xiao
  • , Mei Wang
  • , Paul K. Chu
  • , Anping Huang
  • *此作品的通讯作者
  • Beihang University
  • Beijing Information Science & Technology University
  • City University of Hong Kong

科研成果: 期刊稿件文章同行评审

摘要

Artificial synaptic memristors, with their adjustable and non-volatile resistance, present a promising path for neuromorphic computing (NC). Inspired by the salting-out producing dense fractal dendritic morphologies, a dendritic ion framework composed of LiClO4·3H2O (LCO) and polyvinyl alcohol (PVA) is prepared by spin coating. The functionalized dendritic Pt/MoS2@LCO-PVA/Si memristor boasts 1,670 endurance cycles and a retention time exceeding 104 s. MoS2 nanosheets act as the intermediate to improve ion transportation and storage, and the multi-local enhanced electric field is verified by multi-physics COMSOL simulation. Synaptic plasticity, such as spike-voltage-dependent plasticity, long-term potentiation/depression, and spike-timing-dependent plasticity are also investigated. The dendritic device achieves a digit recognition accuracy of 91.4 % according to simulations of artificial neural networks. The results reveal a new approach that expands the applications of NC.

源语言英语
文章编号161487
期刊Chemical Engineering Journal
509
DOI
出版状态已出版 - 1 4月 2025

学术指纹

探究 'Functionalized dendritic memristor of Pt/MoS2@LCO-PVA/Si for Mimicking synaptic behavior' 的科研主题。它们共同构成独一无二的学术指纹。

引用此