Analog‐Type Resistive Switching Devices for Neuromorphic Computing

Abstract

Brain‐inspired neuromorphic computing has attracted considerable attention due to its potential to circumvent the “von Neumann bottleneck” and mimic human brain activity in electronic systems. The key to developing high‐performance and energy‐efficient neuromorphic computing systems lies in the realization of electronic devices that can closely mimic biological synapses. Resistive random‐access memory (RRAM) has shown some important properties for implementing synaptic functions, including analog weight storage and analog switching. Herein, the recent progress in analog‐type RRAM is reviewed. The mechanisms underlying the analog switching behavior in RRAM and different types of synaptic plasticity based on the analog switching behavior are discussed. Methods to improve the analog switching behavior and synaptic plasticity are then illustrated. Finally, a summary and a perspective on future research are presented.

Pseudo-Interface Switching of a Two-Terminal TaOx/HfO2 Synaptic Device for Neuromorphic Applications

“…The current range is fixed at 1 mA on the equipment, and it is difficult to obtain a value for significant additional conductance values. Figure 5b shows potentiation and depression curves in In order to use the memristor device as a synaptic element of a neuromorphic hardware system, implementation of multi-level conductance is essential [30,31]. The three well-controlled switching models discussed above were evaluated as to their suitability for synaptic devices.…”

Bridging Biological and Artificial Neural Networks with Emerging Neuromorphic Devices: Fundamentals, Progress, and Challenges

“…[158] The amplitude, duration, and numbers of electrical pulses all contribute to the change of filament size and the conductance of the device. [158] The amplitude, duration, and numbers of electrical pulses all contribute to the change of filament size and the conductance of the device.…”

High‐Uniformity Threshold Switching HfO2‐Based Selectors with Patterned Ag Nanodots

“…Due to its high speed, small cell size, and low cost, resistive random-access memory (RRAM) has been extensively studied as a promising candidate for highdensity memory technology, large-scale neuromorphic computing system, wearable electronics, and Internet of Things (IoTs). [1][2][3][4][5] A crossbar array, which takes the advantage of the simple structure and small footprint of RRAM, is an optimal structure for high-density integration. [6][7][8][9] However, the undesired parasitic leakage paths through unselected cells limit the array size and increase the power consumption in operating such crossbar array.…”