Two‐Dimensional Neuromorphic Electronic Devices and Their Applications in Artificial Neural Network Computing
TL;DR
This review systematically reviews representative 2D material systems and typical neuromorphic device architectures, establishes the mapping relationship between device characteristics and artificial neural network computation, and clarifies the advantages of 2D neuromorphic electronic devices in terms of synaptic plasticity, integration density, and power efficiency.
Abstract
Inspired by the human brain, neuromorphic computing offers an effective way to overcome the efficiency bottleneck of the von Neumann architecture. Artificial neural networks (ANNs) are gradually becoming the mainstream paradigm for intelligent computing, but their hardware implementation requires efficient, low‐power devices. Two‐dimensional (2D) materials, with their atomic‐level thickness and unique superior electronic/optical properties, provide an ideal platform for constructing high‐performance neuromorphic devices. This review systematically reviews representative 2D material systems and typical neuromorphic device architectures (memristors and transistors), establishes the mapping relationship between device characteristics and artificial neural network computation, and clarifies the advantages of 2D neuromorphic electronic devices in terms of synaptic plasticity, integration density, and power efficiency. Finally, key challenges such as scalability, stability, and array integration are discussed, and forward‐looking solutions for practical artificial neural network applications are proposed.