Gating-induced reversible HxVO2 phase transformations for neuromorphic computing

Neuromorphic networks that consist of electronic synapses are very important for energy-efficient artificial intelligent applications. Therefore, in recent years, many efforts have been made to design and improve artificial synaptic devices to effectively mimic brain-spiking activity in biological synapses. In this work, we demonstrate a novel synaptic transistor based on the VO2 film that uses the electrolyte gating at room temperature. Through the gating-induced protonation and deprotonation, we realize reversible phase transformations between various H-doped phases, which is confirmed by many characterization measurements. The VO2 synaptic transistor based on the exploiting nonvolatile multi-level conductance states with various hydrogen doping concentrations can successfully emulate essential synaptic functions such as synaptic plasticity and spiking-time-dependent plasticity. An artificial neural network containing the VO2 synaptic transistors simulated with supervised learning shows high recognition accuracy for the MNIST handwritten recognition dataset. This study provides a promising approach to develop high-performance electronic synaptic transistors by utilizing advanced Mott materials.