Self-Powered, Low-Poling-Field and High-Photoresponsivity Perovskite-Based Photodetectors for Neuromorphic Vision

Reconfigurable photodetectors inspired by neural ion channel functions have attracted significant interest for neuromorphic vision technologies. Despite this, current devices often face challenges such as requiring high poling voltages or suffering from low photoresponsivity. In this work, we introduce a hybrid perovskite heterojunction photosensor that overcomes these limitations. The device offers a low poling field requirement, enhanced photoresponsivity, self-powered capability, and nonvolatility. Its photovoltaic behavior can be easily modified through ion migration induced by the poling field. Impressively, it achieves a high photoresponsivity of 137.4 mA W^–1 at a minimal poling field of 3 V μm^–1. This places it among the top-performing reconfigurable photodetectors. The device’s ability to switch photovoltaic properties enables advanced object imaging, while its adjustable photoresponsivity supports real-time sensing and computation. Notably, the perovskite-based artificial neural network achieves flawless pattern recognition, offering functions such as edge enhancement and resolution reduction, demonstrating great potential for energy-efficient machine vision applications.