摘要
大数据和物联网时代的到来使得传统冯·诺依曼架构的计算机在数据处理过程中面临极大的挑战,存算分离的架构从根本上限制着计算机的计算速度和能效,迫切地需要开发一种新的计算范式来应对当前面临的问题和挑战。近年来,神经形态计算以高度的并行处理、极低功耗和存算一体的特征受到广泛关注。其中,具有独特物理机制的新型神经形态器件是构建神经形态芯片的基本底层单元。在构建神经形态器件的众多候选电子材料中,低维材料相比传统三维材料具有优异的物理特性和电学特性,并且弱的层间范德华力使其易于堆叠,有利于异质整合集成。本文详述了基于低维材料的人工突触器件和人工神经元器件的研究进展,总结了不同类型神经形态器件的工作机制、性能指标和技术优势。在此基础上,介绍了低维材料的神经形态器件在视觉、听觉、运动控制和规模集成芯片等领域的应用,并对神经形态器件未来发展趋势进行了展望。
The arrival of the era of big data and the Internet of Things makes the traditional Von Neumann architecture computer face great challenges in the process of data processing.The architecture of storage and computing separation fundamentally limits the computing speed and energy efficiency of the computer.It is urgent to develop a new computing paradigm to overcome the current challenges.Neuromorphic computing has attracted wide attention because of its high parallelism,low power consumption and integrated storage,and the novel neuromorphic devices with unique physical mechanisms are the basic units of neuromorphic computing systems.Among many candidate materials,low-dimensional materials have unique physical and electrical properties.Weak interlayer Van der Waals forces enable them to be arbitrarily stacked,which is conducive to heterogeneous integration.In this paper,the research progress of artificial synaptic devices and artificial neural devices based on low-dimensional materials is reviewed.The working mechanisms,performance indicators and technical advantages of different types of neuromorphic devices are summarized.On this basis,the applications of neuromorphic devices based on low-dimensional materials in the fields of vision,hearing,motion control and large-scale integration are introduced.Finally,the future development of artificial neuromorphic devices is analyzed and prospected.
作者
刘依婷
万军
邱晨光
赵建文
王华
LIU Yiting;WAN Jun;QIU Chenguang;ZHAO Jianwen;WANG Hua(Institute of Metrology and Testing Engineering,China Jiliang University,Hangzhou 310020,China;School of Electronics,Peking University,Beijing 100871,China;Device Department,Suzhou Institute of Nanotechnology and Nanobionics,Chinese Academy of Sciences,Suzhou 215125,China;Key Laboratory of Interface Science and Engineering in Advanced Materials,Ministry of Education,Taiyuan University of Technology,Taiyuan 030024,China)
出处
《发光学报》
EI
CAS
CSCD
北大核心
2023年第6期1085-1111,共27页
Chinese Journal of Luminescence
基金
国家自然科学基金(61971009,62122006)。
关键词
低维材料
人工突触器件
人工神经元器件
神经形态芯片
low dimensional materials
artificial synaptic devices
artificial neural devices
neuromorphic chips