Metal Oxide based resistive switching memristor for neuromorphic computing

Abstract

Brain-inspired neuromorphic computing has attracted great interest considering its advantages of massive parallelism and high energy efficiency. Metal oxide based resistive switching memristors are emerging as the promising nanodevices toward hardware implementation of neuromorphic computing, thanks to their structural and functional resemblance to the biological counterpart and excellent compatibility to the advanced CMOS technique. In this article, we review recent progress in metal oxide memristor and its application on neuromorphic computing. First, the metal oxide based memristor devices with different structural features are summarized, including the two-terminal device and three-terminal device. Furthermore, various memristive mechanisms are systematically discussed, such as the formation of conductive filaments, Mott transition, ferroelectric polarization. We examine the applications of different forms of memristor devices in artificial intelligence, such as electrical and optoelectronic synapse and neuron, and neuromorphic perception system. Finally, we discuss the challenges and prospects of materials, devices, and integrations in this rapidly progressing field of research.