FeNi nanowires with permalloy composition as possible processing units in neural network applications

Abstract

Progress in artificial intelligence requires the development of new types of devices, such as artificial neural networks. Magnetic nanowires made of permalloy can be ideal materials for their production. The presented experimental and theoretical investigations of the FeNi nanowire matrix are the first stage of studies aimed at designing artificial neural networks. FeNi nanowires with a diameter of 40 nm were synthesized in alumina membranes by electrodeposition at potentials ranging from -1.0 V to -1.3 V vs. Ag/AgCl. Increasing voltage caused changes in the chemical composition of nanowires with a gradual increase in Ni content. Nanowires obtained at lower potentials showed an anomalous character of co-deposition. The elemental content influenced the phase composition. At higher potentials, Ni-rich alloys with fcc structure dominated (with Ni hcp phase), at a potential of -1.05 V vs. Ag/AgCl the FeNi${}_3$ phase was identified with a lattice parameter a = 0.355 nm, while at the lowest potential FeNi phase appeared. The samples demonstrated a preferred growth direction that varied with the chemical composition. All samples showed magnetic anisotropy with an easy axis along the nanowires. Low coercivity and squareness with high saturation magnetization were observed for the samples exhibiting permalloy composition. Micromagnetic simulations of nanowires with permalloy composition revealed the magnetization reversal mechanism, which proceeded through the nucleation and propagation of vortex domain walls. Analysis of the demagnetizing field distribution confirmed strong magnetostatic interactions between nanowires in the array and the important role of the vortex domain walls in reducing the demagnetization field value.