2D materials-based photo-memristors with tunable non-volatile responsivities for neuromorphic vision processing

Abstract Conventional artificial-intelligence (AI) machine vision technology, based on the von Neumann architecture, uses separate computing and storage units to process the huge amounts of vision data generated in sensory terminals. The frequent movement of redundant data between sensors, processors and memory, however, results in high-power consumption and latency. A more efficient approach is to shift some tasks of the memory and computational to sensory elements which can perceive and process optical signal simultaneously. Here, we proposed a non-volatile photo-memristor, in which reconfigurable responsivity can be modulated by charge and/or photon flux through it and further stored in the device. The non-volatile photo-memristors consist of simple two-terminal architecture, in which photoexcited carriers and oxygen-related ions are coupled, leading to a displaced and pinched hysteresis of current-voltage characteristics. The non-volatile photo-memristors sets first implemented computationally complete logic for the photoresponse-stateful logic operations, for which the same photo-memristor serves simultaneously as logic gates and memory unit that uses photoresponse instead of light, voltage and memresistance as the physical state variable. Further changing the polarity of photo-memristors demonstrate great potential for in-memory sensing and computing with feature extraction and image recognition for neuromorphic vision processing.