Dynamic Behaviors and Training Effects in TiN/Ti/HfO$_x$/TiN Nanolayered Memristors with Controllable Quantized Conductance States: Implications for Quantum and Neuromorphic Computing Devices

TL;DR

This study demonstrates precise control of quantized conductance states in TiN/Ti/HfO$_x$/TiN memristors through pulse-mode reset procedures, revealing insights into their structural evolution and potential for quantum and neuromorphic computing.

Contribution

It introduces a method for achieving controllable quantized conductance states with high precision and analyzes their dynamic behaviors and structural evolution at the atomic level.

Findings

Achieved precise quantized conductance states via pulse-mode reset.

Observed a training effect that accelerates switching.

Provided detailed analysis of filament evolution under different conditions.

Abstract

Controllable quantized conductance states of TiN/Ti/HfO$_x$/TiN memristors are realized with great precision through a pulse-mode reset procedure, assisted with analytical differentiation of the condition of the set procedure, which involves critical monitoring of the measured bias voltage. An intriguing training effect that leads to faster switching of the states is also observed during the operation. Detailed analyses on the low- and high-resistance states under different compliance currents reveal a complete picture of the structural evolution and dynamic behaviors of the conductive filament in the HfO$_x$ layer. This study provides a closer inspection on the quantum-level manipulation of nanoscale atomic configurations in the memristors, which helps to develop essential knowledge about the design and fabrication of the future memristor-based quantum devices and neuromorphic computing devices.