IGZO Nanofiber Photoelectric Synapse for Artificial Neural Networks

Abstract

Photoelectric synapses have attracted intensive attention due to their ultra-fast signal transmission, high bandwidth, low crosstalk and energy consumption. We proposed an indium gallium zinc oxide (IGZO) nanofiber based photoelectric synapse. The device has been demonstrated with versatile capabilities in mimicking biological synapse and the potential for constructing artificial neural networks (ANNs) with 5 bits precision and 15 fJ weight updating energy.