Light-Controlled Reconfigurable Optical Synapse Based on Carbon Nanotubes/2D Perovskite Heterostructure for Image Recognition.

Two-dimensional (2D) halide perovskite material is characterized by a mixed conducting behavior that possesses both electronic and ionic conductivity. The study on the influence of the light on ion migration in the 2D perovskite is helpful to improve the performance of perovskite-based optoelectronic devices. Here, we constructed an exfoliated 2D perovskite/carbon nanotubes (CNTs) heterostructure optical synapse, in which CNTs can be used as nanoprobes to qualitatively observe the ion aggregation or dissipation process in 2D perovskite, and found that light significantly changes the memory curve of the reconfigurable optical synapses. Through the molecular dynamic simulation, the dynamic process of ion migration in the heterostructure was simulated and the electrostatic interaction effect of nonequilibrium charge distribution of CNTs on iodide ion was demonstrated. Finally, an effective light-controlled process was realized through the synapses, which in situ regulated the performance of the weight-value discretized BP (WD-BP) neural network. This work lays a foundation for the future development of intelligent nano-optoelectronic devices.