A Fully Printed ZnO Memristor Synaptic Array for Neuromorphic Computing Application

Abstract

In this letter, we report a fully printed metal-oxide memristor crossbar array based on the Ag/ZnO/Ag structure, and extend its neuromorphic computing application as artificial synapses. The wurtzite-type ZnO is printed as an active layer, which is sandwiched between two printed Ag electrodes, forming a memristor unit. The printed memristor exhibits volatile resistive switching behaviors under a low compliance current of 1 μA, and can emulate the short-term synaptic plasticity and biological "learning-relearning" processes. More importantly, the image learning and forgetting function is successfully demonstrated in a 3 × 3 printable memristor array. These results show that the printing technique offers a promising path towards large-scale and low-cost neuromorphic computing electronics.