Investigation of tungsten oxide based neuromorphic transistor and emulation of biological synapses

Abstract

An all-solid-state neuromorphic transistor composed of a WO3 thin film and a proton-conducting electrolyte was fabricated for application to next- generation information and communication technology including artificial neural networks. The inherent learning of brain has been emulated electronically with single neuromorphic transistor by using ionic/electronic mixed conductor (WO3) based on nanoionic principle. The electronic synaptic behaviours have been controlled by varying pulse voltage application conditions to the device. The device has been fabricated with epitaxially grown WO3 thin film on LaAlO3 (LAO) substrate by using the RF magnetron sputtering method. The drain current exhibited a 4-order-of-magnitude increment by redox reaction of the WO3 thin film deposited upon LAO substrate owing to proton migration. The learning, forgetting and memorizing functions are achieved by controlling the electrochemical redox reactions in WO3 channel. The drain current modulation has been recorded during the gate voltage sweep between 0 V to 2.5 V, a strong hysteretic behaviour is observed during the backward sweep. Learning and forgetting characteristics were well tuned by the gate control of WO3 redox reactions owing to the separation of the current reading path and pulse application path in the transistor structure. The Short Term Memory (STM) and Long Term Memory (LTM) functions have been obtained by using gate voltage pulses. This technique should lead to the development of versatile and low-power-consumption neuromorphic devices. newline