Base Elements for Artificial Neural Network: Structure Modeling, Production, Properties

Abstract

A radical reduction in power consumption is becoming an important task in the development of supercomputers. Artificial neural networks (ANNs) based on superconducting elements of spintronics seem to be the most promising solution. A superconducting ANN needs to develop two basic elements - a nonlinear (neuron) and a linear connecting element (synapse). The theoretical and experimental results of this complex and interdisciplinary problem are presented in this paper. The results of our theoretical and experimental study of the proximity effect in a stacked superconductor/ferromagnet (S/F) superlattice with Co-ferromagnetic layers of various thicknesses and coercive fields and Nb-superconducting layers of constant thickness equal to the coherence length of niobium and some studies using computer simulation of the formation of such multilayer nanostructures and their magnetic properties are presented in this article.