Logic Circuits Based on Neuron-Like Antiferromagnetic Spin Hall Oscillators

Neuromorphic computing is field of electronics which use the similar principals as human brain does. Neurons in nervous system can be presented as non-linear oscillators with different regimes of operating. In contradiction to traditional computers neuromorphic systems demonstrate better results in tasks of imaging recognition, tasks of synthesis and predictions. In this paper we show that antiferromagnetic spin Hall oscillator in certain regime can demonstrate behavior similar to the one of a biological neuron. Form of generation pulses, existence of the threshold of driving signal, which divides different operation regimes, and other common features of a spin Hall oscillator and biological neuron gives an opportunity to propose the use of such a generator as a neuron in neuromorphic circuits. Also such artificial "neuron" based on an antiferromagnetic spin Hall oscillator can perform different logical operations. We demonstrate that AND, OR, Majority and Q gates, and full adder operations can be realized on that such "neurons". As single peak duration is several picoseconds, the operation frequency of such neuromorphic systems can be tens of GHz.