Realize low-power artificial photonic synapse based on (Al,Ga)N nanowire/graphene heterojunction for neuromorphic computing

Abstract

The fast development of brain-inspired neuromorphic computing systems has stimulated urgent requirements for artificial synapses with low-power consumption. In this work, a photonic synaptic device based on (Al,Ga)N nanowire/graphene heterojunction has been proposed and demonstrated successfully. In the artificial synaptic device, the incident light, the nanowire/graphene heterojunction, and the light-generated carriers play the roles of action potential, pre-synaptic/post-synaptic membrane, and neurotransmitter in a biological synapse, respectively. As a key synaptic function, the paired pulse facilitation index of the photonic synapse can reach 202%, which can be modulated by the interval time between two adjacent light pulses. It is found that the graphene defects, the surface band bending, and the Al vacancies on the surface of (Al,Ga)N nanowires can be the key reasons contributing to the synaptic characteristics of artificial photonic devices. Hence, the dynamic “learning–forgetting” performance of the artificial synaptic device can resemble the “learning–forgetting” behavior of the human brain. Furthermore, the hand-written digits are set up to mimic a typical characteristic of human perceptual learning. After only three training epochs, the simulated network can achieve a high recognition rate of over 90% based on the experimental conductance for long-term potentiation and long-term depression. In supervised learning processes, such few training times are beneficial to reduce energy consumption significantly. Therefore, in the area of neuromorphic computing technology and artificial intelligence systems requiring low-power consumption, this work paves a potential way to develop the optoelectronic synapse based on semiconductor nanowires.