IGZO Photonic-Synaptic Transistors with Outstanding Linearity by Controlling Oxygen Vacancy for Neuromorphic Computing

Abstract

A significantly high linearity was achieved by controlling oxygen vacancy in IGZO photonic-synaptic device. A substrate bias during sputtering process and subsequent nitrogen plasma treatment were used to control the oxygen vacancies in IGZO thin film. As the oxygen vacancies were reduced, the synaptic transistor mimics biological synapses. The nonlinear factor of weight update was considerably improved from 1.55 to 0.41, which is a sufficient level for highly accurate artificial neural network.