On the origin of filamentary resistive switching in oxides-based memristive devices

TL;DR

This paper investigates the dynamics of oxygen vacancies in oxide memristors, revealing that a critical vacancy concentration is necessary for resistive switching and that defects influence filament formation and device performance.

Contribution

It introduces a detailed model of oxygen vacancy behavior, highlighting the threshold conditions for filament formation and the effects of static defects on resistive switching.

Findings

A percolation path alone does not induce low resistance.

A critical oxygen vacancy concentration is required for switching.

Defects can modulate the ON-OFF ratio and filament formation.

Abstract

The control and manipulation of filamentary resistive switching (FRS) is essential for practical applications in fields like non-volatile memories and neuromorphic computing. However, key aspects of the dynamics of conductive filament formation and their influence on device resistance remain incompletely understood. In this work we study FRS in binary oxides based memristors by investigating the dynamics of oxygen vacancies (OV) on a two dimensional lattice and their role in forming low-resistance paths that facilitate the transition between high and low global resistance states. We reveal that the mere formation of an OV percolation path is insufficient to induce a transition to a low-resistance state. Instead, an OV concentration exceeding a critical threshold across all sites in the filament is required to generate a low-resistivity conducting path. Furthermore, we simulate the impact of static defects -which block OV migration and would correspond to voids in real porous samples-, on filament formation. We show that there is a range of defect density values where OV percolate through the sample, leading to the formation of OV filaments, but conductive paths remain absent. Additionally, a small concentration of defects can reduce the final value of the low-resistance state, thereby increasing the ON-OFF ratio. These findings provide valuable insights into optimizing defective nanomaterials with memristive properties, which are crucial for advancing in-memory and neuromorphic computing technologies.