Two-Dimensional Memristors: Materials, Switching Physics, Architectures, and Computing Applications

Abstract For the upcoming generation of non-volatile memory, neuromorphic computing, and in-memory artificial intelligence (AI) hardware, two-dimensional (2D) material-based memristors have emerged as a promising device type. Their van der Waals (vdW) interfaces, atomic-scale thickness, and tunable electronic characteristics enable remarkable control over resistive switching mechanisms. They have an advantage over conventional bulk oxide memristors in terms of device uniformity, scalability, and power efficiency. An overview of recent developments in 2D memristive technologies is provided in this review, which also discusses material systems like black phosphorus, hexagonal boron nitride, graphene derivatives, transition metal dichalcogenides, and new 2D ferroelectrics. Additionally, it covers important resistive switching mechanisms such as interface-modulated switching, electrochemical metallization, defect and vacancy migration, and phase-transition-driven behavior. The effectiveness of various device architectures, such as planar and vertical structures, vdW heterostructures, and large-scale crossbar arrays, for in-memory computing is examined. Furthermore, neuromorphic characteristics such as very low-energy matrix-vector multiplication, spiking neural networks, analog weight modulation, and synaptic plasticity are emphasized. Important problems pertaining to endurance, environmental stability, variability, and large-area growth are explored, along with fresh approaches to these problems. Finally, future directions that highlight the revolutionary potential of 2D memristors for AI, edge computing, and next-generation memory technologies are described, including 3D integration, atomically engineered defects, and ultra-flexible neuromorphic systems.