Light-Regulated Mott Transition for On-Demand Multilevel Memory Storage, Processing, and Energy Efficient Machine Vision

Abstract

Photosensing, data processing, and sequential memory storage are indispensable components for advanced optoelectronic devices. Nevertheless, despite enormous independent research efforts, achieving well-controlled cofunctionality such as sensing, processing, and manifold memory storage within a single unit remain a critical issue. Here, a Mott-insulator-based monolithic photosensor that, depending on illuminating intensity and applied bias, can simultaneously store tunable multilevel data without latency is demonstrated. In particular, the threshold voltage to initiate the insulator-to-metal transition has been regulated from 4.5 to 1.5 V by changing the photon illumination intensity from dark to 6 mW cm−2. Microscopic evidence of Mott transition at the nanoscale is revealed through current maps obtained using conductive atomic force microscopy. Further, as a front-end image sensor, the authors’ device offers on-demand photo-intensity sensing and edge detection with direct optical input, which is confirmed by simulation. This study intends an essential breakthrough toward on-demand photosensing, data storage, and processing even at nanoscale, and offers the opportunity to utilize it for various applications including in volatile memory storage, neuromorphic cameras, and emergency alert systems.