摘要
现阶段计算与存储分离的"冯·诺依曼"体系在功耗和速率方面已经不能满足人工智能、物联网等新技术的发展需求,存算一体化的类脑计算方案有望解决这一问题,迅速成为研究热点。忆阻器是一种新型微电子基础器件,其电阻可通过外场连续调节且具有非易失性、小尺寸、低能耗、高速和CMOS兼容等优良特性,被认为是快速实现存算一体化计算最具潜力的类突触器件。与此同时,光电子器件和神经元遵从动力学数学同构性,借助这种同构性可用光电子器件模拟神经元行为并实现类脑计算,基于光子器件的类脑芯片正在往更高集成度、更低功耗、更高性能方向发展,其将会在类脑计算领域发挥越来越重要的作用。介绍忆阻器材料和器件方面的研究进展,具体包括石墨烯材料低温生长,小尺寸钙钛矿忆阻器件、Parylene忆阻器件、WTiO_x忆阻器件以及光子集成类突触器件及芯片等方面研究,并讨论忆阻器在类脑芯片和人工智能领域的应用前景。
Nowadays,the Von Neumann system,which separates computing from storage,can not satisfy the requirement of new technologies such as artificial intelligence and Internet of Things in terms of the power consumption and speed.Brain-like computing based on integration of storage and computation is the potential solution of this problem and becomes a hot research area.As a new type of basic microelectronic device,the memristor,its resistance can be regulated continuously through external field.Due to the non-volatility,small size,low energy consumption,high speed and CMOS compatibility,the memristor is considered as the most potential brain-like device for fast integration of storage and computation.Meanwhile,the optoelectronic devices and neurons obey the dynamic mathematical isomorphism.Based on this isomorphism,the optoelectronic devices can simulate neuron behaviors and realize brain-like computing.The photonic-devices-based brain-like chips with higher integration,lower power consumption and higher performance are developed and play the important role in the field of brain-like computing.This report will introduce the research progress of memristor materials and devices,including low temperature growth of graphene materials,small size perovskite memristor devices,parylene memristor devices,WTiOxmemristor devices and photonic integrated brain devices and chips,and discuss the application prospects of memristors in the field of brain-like chips and artificial intelligence.
作者
陈子龙
程传同
董毅博
张欢
张恒杰
毛旭瑞
黄北举
CHEN Zilong;CHENG Chuantong;DONG Yibo;ZHANG Huan;ZHANG Hengjie;MAO Xurui;HUANG Beiju(Syracuse University,New York,13244,USA;Institute of brain machine fusion intelligence,Jiangsu industrial technology research institute,Suzhou,215133,China;Institute of Semiconductors,Chinese Academy of Sciences,Beijing,100083,China;Beijing University of Technology,Beijing,100124,China)
出处
《微纳电子与智能制造》
2019年第4期58-70,共13页
Micro/nano Electronics and Intelligent Manufacturing
基金
国家重点研发计划“宽光谱高响应度纳米结构探测器”(2018YFA0209004)项目资助.
关键词
忆阻器
光电子器件
类脑芯片
人工智能
memristor
photoelectronic devices
brain-like chips
artificial intelligence
作者简介
通信作者:黄北举,副研究员,主要研究方向为石墨烯、忆阻器材料与器件。E-mail:bjhuang@semi.ac.cn
