Domain wall dynamics in (Co/Ni)n nanowire with anisotropy energy gradient for neuromorphic computing applications

Abstract

Artificial Intelligence (AI) has been gaining traction recently. However, they are executed on devices with the von Neumann architecture, requiring high power input. Consequently, brain-inspired neuromorphic computing (NC) has been gaining attention because it is expected to be more power efficient and more suitable for AI. Designing of NC circuits involves development of artificial neurons and synapses. More studies have hitherto been focused on artificial synapses instead of neurons because the latter should demonstrate leaky integrate-and-fire (LIF) properties, which is a challenge to replicate artificially. In this work, we propose a domain wall (DW) based device made from perpendicularly magnetized (Co/Ni)n nanowire (NW) with graded magnetic anisotropy and saturation magnetization. The DW is current-driven via spin-transfer-torque. Micromagnetic simulations demonstrated that the DWs in NWs with anisotropy field gradients can automatically return towards the initial position when electrical current is absent, indicative of the leakage process. The underlying physics of DW motion in such structure was studied in detail. To replicate the crystallinity of (Co/Ni)n structures, granular NWs were also defined. Depending on the grain structure of the NW, it was found that LIF properties were achieved under the conditions of steep anisotropy field gradients. Therefore, the proposed design has potential applications in neuron devices.