Lower Power, Better Uniformity, and Stability CBRAM Enabled by Graphene Nanohole Interface Engineering

Abstract

Conductive-bridge random access memory (CBRAM) shows great potential as next-generation nonvolatile memory. However, the relatively high-power consumption, cycle-to-cycle and device-to-device variations are still drawbacks for industrial applications. A method of Monte Carlo simulation of CBRAM combined with a model to calculate the ${I}$ – ${V}$ curves has been developed, which investigate the formation and dissipation of filaments inside CBRAM cells. Through simulations, we conclude that the multi-filament structure is responsible for high-power consumption and the changing of regrowth filament location is responsible for the variations. We have enabled lower power, better uniformity, and stability CBRAM using graphene nanohole interface layer. Compared to CBRAM without graphene, CBRAM with inserted graphene with 0.66 nm hole shows better uniformity (60 times improvement), stability (5 times improvement), and lower power dissipation (10 times reduction). Moreover, our simulation results are matched with the experimental results of 45-nm graphene nanohole CBRAM. The reasons for better characters have been explained by the confining of the filament location with a single filament. This article shows great potential of integrating CBRAM with graphene nanohole structure.