An Integrated Single Photodetector-Memristor System for Light-Modulated Multi-State Resistance Memory

Abstract

Neuromorphic devices have attracted outstanding attention as next-generation devices due to their high efficiency in complex data, using the parallel operation inspired by the natural signal processing system of the neurons and synapses. Recently, research on visual information processing using neuromorphic devices has been intensely conducted with computing features, such as low-power and quick response to a dynamic environment. Especially, resistive random-access memory, one of the crucial components in neuromorphic systems, stands out due to its simplicity, non-volatility, multilevel operation, and fast resistance modulation. In this study, we introduce an in-pixel neuromorphic system, integrating an optically efficient silicon-based PIN photodetector (PD) with an oxide-based memristor. The proposed system enables multi-state resistance memory through the generated photocurrent of PD depending on incident light intensity. This system aims to offer advanced progress on neuromorphic architecture for sophisticated applications, such as self-driving cars and drones.