Resistive Switching Devices for Neuromorphic Computing: From Foundations to Chip Level Innovations.

Area of Science:

• Materials Science
• Computer Engineering
• Neuroscience

Background:

• Neuromorphic computing addresses growing data demands with brain-inspired architectures.
• Resistive random-access memory (RRAM) devices emulate neuronal behaviors, offering advantages like compactness and low power consumption.

Purpose of the Study:

• This review comprehensively analyzes RRAM devices for brain-inspired computing.
• It covers biological concepts, RRAM switching behaviors, physical mechanisms, materials, and applications.

Main Methods:

• Literature review of RRAM devices and neuromorphic applications.
• Analysis of resistive switching behaviors and underlying physical mechanisms.
• Discussion of various material choices and benchmark results.

Main Results:

• RRAM devices exhibit diverse resistive switching behaviors and mechanisms.
• Various materials are suitable for RRAM, with recent research showing promising benchmark results.
• Emerging trends and state-of-the-art results in RRAM-based neuromorphic applications are highlighted.

Conclusions:

• RRAM devices are crucial for advancing neuromorphic computing.
• The review provides insights into RRAM materials, mechanisms, applications, challenges, and future outlook.
• It summarizes key findings and identifies salient research results for commercial chip-level applications.