Bidirectionally Modulated Synaptic Plasticity with Optically Tunable Ionic Transistors

Recently, hardware implementation of neuromorphic device in the optical domain is considered as one of the most promising routes to realize energy-efficient neuromorphic computing systems. Especially, a complete plasticity modulation by all-photonic stimulation has been one of the most important challenges for implementation of an optoelectronic neuromorphic device. Here, we demonstrate a fully optically driven bidirectional synaptic device using ionic electrolyte transistors. The photovoltaic divider enables wavelength-selective light-to-voltage conversion and subsequently induces ionic migration in the electrolyte, resulting in the synaptic potentiation or depression. Based on the synaptic characteristics, pattern recognition with an accuracy up to 90.1% is obtained in the Modified National Institute of Standards and Technology simulation.

Automated summary

The recent development of hardware implementation of neuromorphic devices in the optical domain has shown promising results in energy-efficient computing systems. By utilizing ionic electrolyte transistors, a bidirectional synaptic device has been successfully demonstrated, enabling synaptic potentiation or depression with high pattern recognition accuracy of up to 90.1% in simulations.