The trend of synthesized 2D materials toward artificial intelligence: Memory technology and neuromorphic computing

Abstract

2D materials, specifically transition metal dichalcogenides (TMDs), have gained massive attention for their potential use in high-integration memory technologies due to their exceptional carrier transport, atomically thin structure, and superior physical and electronic properties. High-density memory processors and complex hardware neural architectures based on TMDs have been developed and shown to have exceptional memory properties, making them a potential competitor to conventional Si technology. However, TMDs are still facing challenges with achieving high yields at high-density levels when compared to Si-based semiconductor technology. This review article covers the synthesis methods, memory device structures, high-volume circuits, and neuromorphic computing of TMD materials. We briefly discuss a plethora of synthesis methods that are utilized to achieve large-area uniform distribution in the fabrication of memory arrays. Various memory device architectures based on two-terminal and three-terminal designs are introduced, offering comprehensive prospects for utilizing TMDs in neuromorphic computing and developing energy-efficient and low-power neural networks for complex computational tasks beyond conventional Si-based architecture. Finally, the potential and challenges of utilizing TMDs in neuromorphic circuits are briefly discussed, including perspectives on system architecture and performance, synaptic functionalities, implementing ANN algorithms, and applications to artificial intelligence at high-density levels.