Manipulation of Magnetic Skyrmions Motion in Confined Geometries for Potential Neuromorphic Applications

Abstract

Neuromorphic computing, inspired by the biological nervous system, has attracted enormous attention since it holds the promise of energy-efficient, intelligent, and highly adaptable computing systems [1-3]. Recently, magnetic skyrmions have been considered as promising candidates for neuromorphic computing design owing to their unique dynamic and topological characteristics such as high stability, low driving current density, and compact size in the nanometer range [4-6]. In this work, we report on a micromagnetic simulation study of the dynamics of the skyrmions and their manipulations in confined geometries, attempting to emulate the behaviors of biological synapses and neurons. By exploiting the topological characteristics in current-driven skyrmion motion accompanying with tunable geometrical potentials, we propose the design of skyrmion-based artificial neuron, mimicking the short-term plasticity and long-term potentiation functions of a biological neuron. The artificial neuron proposed here is built on a simple track-based structure in synaptic crossbar array architecture. Consequently, these results may enable to provide a dense and energy-efficient neuromorphic computing system with a simple single device implementation.