Resistive Switching by Voltage-Driven Ion Migration in Bipolar RRAM—Part II: Modeling

TLDR

Resistive‑switching memory based on transition‑metal oxides is a promising alternative to Flash and DRAM, yet its underlying mechanism remains poorly understood. This study develops a model for resistive switching in bipolar metal‑oxide RRAM devices. The model represents reset and set as voltage‑driven ion migration within a conductive filament, employing drift–diffusion equations with Arrhenius‑activated diffusivity and mobility, and solving self‑consistently for local temperature and electric field in a 3‑D axis‑symmetric geometry. It reproduces the abrupt set and gradual reset transitions, accurately predicts set–reset characteristics, and supports scaling projections for metal‑oxide RRAM.

Abstract

Resistive-switching memory (RRAM) based on transition metal oxides is a potential candidate for replacing Flash and dynamic random access memory in future generation nodes. Although very promising from the standpoints of scalability and technology, RRAM still has severe drawbacks in terms of understanding and modeling of the resistive-switching mechanism. This paper addresses the modeling of resistive switching in bipolar metal-oxide RRAMs. Reset and set processes are described in terms of voltage-driven ion migration within a conductive filament generated by electroforming. Ion migration is modeled by drift–diffusion equations with Arrhenius-activated diffusivity and mobility. The local temperature and field are derived from the self-consistent solution of carrier and heat conduction equations in a 3-D axis-symmetric geometry. The model accounts for set–reset characteristics, correctly describing the abrupt set and gradual reset transitions and allowing scaling projections for metal-oxide RRAM.