A hybrid memristor with oxide-polymer heterojunction

Abstract

A hybrid memristor based on the bilayer structure of indium gallium zinc oxide (IGZO)/polyvinyl alcohol (PVA) is developed, which demonstrates device state updates in an analog manner with high reliability. The IGZO/PVA heterojunction is crucial for the realization of the memristive characteristics, presumably associated with oxygen ion redistribution across the IGZO/PVA interface. The hybrid memristor may act as an electronic synapse, being capable of emulating synaptic potentiation with good linearity, synaptic depression, and paired-pulse facilitation. It highlights potential applications of the oxide-polymer heterojunction in the exploration of neuromorphic devices.

Programmable Retention Characteristics in MoS2-Based Atomristors for Neuromorphic and Reservoir Computing Systems

“…Among them, RRAM is one of the most promising candidates for application in neuromorphic computing owing to its high scalability, low-power consumption, increased switching speed, and high-switching versatility. Various materials such as metal-oxide, organic, and two-dimensional (2D) are being investigated for applications in neuromorphic systems to emulate synaptic functions. − In particular, the conductive bridge random access memory (CBRAM), a type of RRAM with a Ag or Cu electrode, has attracted attention owing to the capability of the CBRAM to implement synaptic functions via short- and long-term memory mechanisms based on active metal-ion transitions. − However, active metals have high reactivity and atomic mobility, thus causing diffusion through an insulator or oxidization. These problems result in device performance degradation; consequently, the active metal is not a fab-friendly material in the semiconductor industry.…”

Room‐Temperature‐Processable Highly Reliable Resistive Switching Memory with Reconfigurability for Neuromorphic Computing and Ultrasonic Tissue Classification

“…[6] Processing-in-memory (PIM) architecture is a promising solution to resolve those difficulties by collocating its processor and storage. [7][8][9][10][11] Especially, a crossbar array of analog two-terminal memory devices, composed of resistive switching (RS) materials exhibiting tunable resistances, has emerged as a leading candidate to construct this architecture using those devices as synaptic elements of an ANN. [3] Many research works have been conducted to reveal their operating mechanisms and various types of RS materials are developed.…”