Neuromorphic computing based on an antiferromagnet-heavy metal hybrid structure under the action of laser pulses

Abstract

The paper proposes a model of a neuromorphic processor, consisting of excitatory and processing neurons that are oscillators and detectors. The concept of neuromorphic computing, implemented by generating a spin current due to optical excitation of magnetic oscillations in an antiferromagnet is considered. The inverse spin Hall effect causes the generation of an electric current in the heavy metal layer. A constant driving current flows through the common bus. Magnetic oscillations in the receiving neuron occur due to the spin Hall effect. A biaxial nickel oxide crystal was used as a material for the base cells of AFM insulators and platinum was utilized as a heavy metal. The use of optical excitation can significantly increase the processing speed of neuromorphic computing with low power consumption. The presented model implements the simplest operations of neuromorphic computations, such as logical “OR”, “AND”.