All-oxide-based and metallic electrode-free artificial synapses for transparent neuromorphic computing

Abstract

The extraordinary abilities of the human brain rely on the functioning of biological synapses in the nervous system. Emulating human brain activity for neuromorphic computing has been an alluring prospect in artificial intelligence, and this would become possible with the development of an efficient artificial synapse. The human brain assimilates most information through visual perception. Therefore, optoelectronic synapses can be considered an important keystone of neuromorphic computing due to their efficient ability to process optoelectronic input signals. In this work, a metallic electrode-free, all-oxide-based optoelectronic synapse was developed that demonstrated efficient and repeatable electronic and photonic synaptic plasticity behaviour. A unique artificial synapse with an n–n heterostructure was developed using ZnO and V2O5. This artificial synapse structure has high optical transparency and does not require a metallic electrode for data acquisition. The device exhibits resistive switching memory behaviour, along with electronic and photonic synaptic behaviour. The storing and erasing of information, as well as learning-experience behaviour, was also successfully demonstrated in the V2O5/ZnO artificial synapse. This work could therefore be beneficial to the development of artificial intelligence when transparent and metallic electrode-free synaptic devices are considered necessary.