Neuromorphic Visual Perception through Multifunctional Lead-Free Perovskite Sensory Neurons

Abstract

Multifunctional artificial neurons capable of simultaneously sensing and processing optical information are highly desirable for next-generation neuromorphic vision systems. Here, we report a lead-free optoelectronic sensory neuron based on volatile memristors employing the mixed organic–inorganic perovskite (PDA)BiI₅ and nanoparticle-engineered interfaces. Soft-landed Ag nanoparticles (NPs) were incorporated into the device structure to promote controlled conductive filament formation (CF) and improve switching reliability without damaging the active layer. Compared with reference devices, the optimized memristors exhibited forming-free volatile switching, a memory window exceeding 10⁵, low operating voltages (<0.5 V), and pulse endurance beyond 10⁹ cycles, representing one of the highest endurance values reported for lead-free perovskite memristors. A physics-based numerical model was further employed to elucidate the switching mechanism and the role of the nanoparticle interfaces. By integrating the volatile memristor into a simple RC circuit, low-power optoelectronic leaky integrate-and-fire neurons were realized, exhibiting an energy consumption of only ∼50 nJ per spike. Owing to the transparent device architecture, neuronal operation could be controlled through optical stimulation. The proposed neurons demonstrated multifunctional perception capabilities by encoding optical wavelength, illumination intensity, and incident angle into distinct temporal spike trains. Selective responses were obtained under RGB illumination (470, 530, and 630 nm), different optical power densities, and varying illumination directions. These results establish nanoparticle-engineered lead-free perovskite memristors as promising building blocks for energy-efficient neuromorphic vision systems and future bioinspired artificial intelligence hardware.