Publication date: 23rd February 2022
Developing brain inspired devices emulating neurons and synapses is of great interest to the fields of materials and device physics, as it may lead to efficient hardware-based solutions in neuromorphic computing. Integrate-and-fire circuits have been used to simulate the firing of action potentials in neurons and can be implemented in self-oscillating circuits for oscillatory neural networks (ONN). Here we present self-oscillators based on epitaxially grown rare-earth nickelate thin films, which display a metal-insulator transition (MIT), as resistive switches to produce self-oscillating circuits. The MIT in rare-earth nickelates can be adjusted by choice of rare-earth element, substrate strain and film thickness, allowing for precise control of the device characteristics. Furthermore, the perovskite structure of rare-earth nickelate films allows for integration in ferroelectric, piezoelectric or ferromagnetic perovskite heterostructures. Large devices on the scale of 100 microns of 30 nm thick SmNiO3 were shown to give oscillations with a periodicity on the order of 10 ms at a potential of 17V over the device. Reducing the input potential and increasing the oscillation frequency are explored by scaling down the device size and optimizing device geometry for better cooling. Synchronization between oscillators is required for ONN implementation and might be achieved by thermal coupling when placing oscillators in close proximity of each other. The tunability of the MIT, together with the possibility of adding functionality in heterostructures, make of rare-earth nickelates interesting candidates for developing oscillating circuits.
We acknowledge the financial support of the CogniGron research center and the Ubbo Emmius Funds (Univ. of Groningen).