Stability of quantized conductance levels in memristors with copper filaments: toward understanding the mechanisms of resistive switching

TL;DR

This study investigates the stability of quantized conductance in copper filament memristors, combining experimental retention analysis with a theoretical model that highlights quantum pressure effects on filament stability at voltages around 1V.

Contribution

It introduces a simplified theoretical model of quantum current effects on filament shape and experimentally demonstrates enhanced filament stability at low voltages, advancing understanding of resistive switching mechanisms.

Findings

Quantum pressure can stabilize thin filaments supporting quantized conductance.

Filament stability is enhanced at voltages around 1V due to quantum effects.

Recoil effects are significant for resistive switching and device retention.

Abstract

Memristors are among the most promising elements for modern microelectronics, having unique properties such as quasi-continuous change of conductance and long-term storage of resistive states. However, identifying the physical mechanisms of resistive switching and evolution of conductive filaments in such structures still remains a major challenge. In this work, aiming at a better understanding of these phenomena, we experimentally investigate an unusual effect of enhanced conductive filament stability in memristors with copper filaments under the applied voltage and present a simplified theoretical model of the effect of a quantum current through a filament on its shape. Our semi-quantitative, continuous model predicts, indeed, that for a thin filament, the "quantum pressure" exerted on its walls by the recoil of charge carriers can well compete with the surface tension and crucially affect the evolution of the filament profile at the voltages around 1V. At lower voltages, the quantum pressure is expected to provide extra stability to the filaments supporting quantized conductance, which we also reveal experimentally using a novel methodology focusing on retention statistics. Our results indicate that the recoil effects could potentially be important for resistive switching in memristive devices with metallic filaments and that taking them into account in rational design of memristors could help achieve their better retention and plasticity characteristics.