Magnetic Nanowire Networks as 3D Platforms for Neuromorphic Computing

Three-dimensional (3D) nanomagnetic systems can accommodate complex spin textures and magnetization dynamics [1, 2], while mimicking brain-like structure. These attributes make them an excellent platform for implementing neuromorphic computing. In this work, we present novel 3D interconnected nanowire (NW) networks to explore this.



Previously, we have studied quasi-ordered, free-standing 3D Co NW networks fabricated using multiple-angle ion tracking of polycarbonate membranes and electrodeposition [3, 4]. Magnetoresistance (MR) measurements exhibited multiple discrete jumps during field cycling, suggesting step-by-step magnetization switching. Furthermore, first-order reversal curve (FORC) measurements of MR revealed that partial switching within the network enables access to new magnetization reversal pathways, ultimately determining the final magnetic state at a given field. This allows the network to be conditioned to host diverse magnetic states, promising for implementing multistate memristors.



More recently, we have explored random networks fabricated using electrodeposited, self-assembled Ni NWs and sintered to form conductive pathways [5]. Multiple electrode pairs were used for MR measurement which showed the networks contain multiple unique transport pathways each hosting many discrete magnetization states. These pathways can be selectively addressed by activating different electrode pairs. Moreover, the magnetic states within them can be electrically controlled by applying current pulses and switching specific sections of the networks. Consequently, the pathways can act as synaptic weights that can be programmed by adjusting the magnitude and duration of the applied current pulses. These properties were utilized to demonstrate applicability of these networks in neural network architectures, showcasing their potential as an efficient and versatile neuromorphic computing platform.



1. Fischer et al, APL Mater. 8, 010701 (2020).

2. Fernández-Pacheco et al, Nat. Commun. 8, 15756 (2017).

2. Burks et al, Nano Lett., 21, 716 (2021).

3. Bhattacharya et al, Nano Lett. 22, 10017 (2022).

4. Bhattacharya et al, Submitted.