Inhibitory neuristor based on metal-to-insulator transition

TL;DR

This paper introduces an inhibitory neuristor based on metal-to-insulator transition materials, demonstrating controllable self-oscillations that mimic neuronal inhibition for neuromorphic computing.

Contribution

It presents the first experimental demonstration of inhibitory-like self-oscillations in MIT materials, expanding neuromorphic device functionalities.

Findings

Achieved ~0.1 - 1 MHz oscillations with minimal variation

Controlled oscillations via voltage, temperature, and inductance

Demonstrated inhibitory-like behavior in MIT-based devices

Abstract

Mimicking the collective excitatory and inhibitory behaviors of biological neurons remains a critical challenge in the development of neuromorphic computing systems that rival the complexity and performance of the human brain. Volatile high-to-low resistance switching in insulator-to-metal transition (IMT) materials produces an abrupt increase in current flow, resembling neuronal excitation. This electrical excitation enables IMT materials to be driven into a neuron-like spiking self-oscillation regime using simple RC circuits. Here, we report a new type of self-oscillation dynamics that occurs in the opposite class of metal-to-insulator transition (MIT) materials. Electrical triggering of the MIT suppresses current flow, resembling neuronal inhibition. Using a prototypical MIT material, we experimentally demonstrate inhibitory-like self-oscillations in two-terminal switching devices incorporated into a simple RL circuit. Our results show robust ~0.1 - 1 MHz electric current oscillations with minimal cycle-to-cycle variation, which can be controlled by varying the applied DC voltage, temperature, and inductance. This work demonstrates a new type of inhibitory MIT-based artificial neuron that can complement the excitatory functionalities of IMT-based neuristors in biologically plausible neuromorphic systems.