Chirality-induced spin-regulated synaptic dynamics in 2D perovskite memristors for polarimetric neuromorphic computing

Abstract

Conventional von Neumann architectures face fundamental energy-efficiency bottlenecks, while current optoelectronic neuromorphic devices remain confined to simple optoelectronic responses, unable to leverage continuous physical degrees of freedom such as light polarization for continuous, analog synaptic modulation. We report a continuously polarization-tunable photonic memristor based on chiral 2D (R/S-MBA)2PbI4 perovskites, which integrates polarimetric sensing with neuromorphic computing. By incorporating molecular chirality into the inorganic framework, we leverage the chirality-induced spin selectivity effect to regulate synaptic relaxation dynamics via spin-dependent carrier transport. This mechanism enables continuous, polarization-tunable synaptic weight updates, significantly extending the functional dimensionality of neuromorphic hardware. We validate this architecture through noise-resilient modified national institute of standards and technology (MNIST) database classification—where chiroptical filtering improves accuracy from 76% to 88%—and high-precision semantic segmentation, achieving a polarization phase resolution of 5° and a Sørensen–Dice coefficient exceeding 0.9. These results establish a physical foundation for integrating spin-dependent degrees of freedom into optoelectronic neuromorphic systems, offering a robust pathway for next-generation intelligent processing.