Single Particle All‐Optical Synapses

ABSTRACT

Artificial synapses for emulating the synaptic behaviors and neuromorphic computing have been extensively exploited based on the temporal response of diverse physical and chemical processes. In these synaptic devices, operating in an all‐optical fashion offers significant benefits in parallelism, bandwidth, and energy efficiency; however, most of these optical synapses are based on complex optical architectures or ensembles of particles. Here, we show that the optical response of single long persistent phosphor (LPP) particles can be manipulated to simulate typical synaptic behaviors in biological systems with enhanced energy efficiency. Both short‐term and long‐term plasticity can be optically simulated by the response of single LPP particles to pulse light excitation, and the synaptic plasticity is dependent on the particle size, shape, as well as the inter‐particle coupling. Mechanistically, the synapse‐like response in single LPP particles is kinetically controlled by the temporal optical response involving trap and the slow detrapping processes. We demonstrate further that the synaptic response in single LPP can be used to build an artificial neuron network for image classification, offering accuracies close to electronic synaptic devices. Our results suggest the temporal response in single LPP particles could provide an all‐optical platform for optical neuromorphic computing.