Optoelectronic-Driven Multi-Functional MoS2:PVA Flexible Memristor for In-Sensor Neuromorphic Computing.

Abstract

This article presents a multifunctional Ag/MoS2:PVA/SnO2/Pt memristive device that demonstrates resistive memory, synaptic learning, and optical information processing. The device exhibited stable bipolar resistive switching with low operating voltages with SET and RESET at ∼  0.57/-0.86 V, respectively, an ON/OFF ratio of ∼  400 at 0.2 V, endurance around 10 4 cycles, and retention stability over 10 4 s with resistance separation greater than 10 3 . The device also maintained reliable operation under repeated bending and different bending radii, confirming excellent mechanical flexibility. Gradual conductance modulation enabled repeatable potentiation and depression behavior for neuromorphic functionality. Under optical stimulation, wavelength- and intensity-dependent conduction evolution was observed and followed an exponential learning model, enabling optical memory characteristics. Furthermore, image memorization and traffic signal recognition were demonstrated using experimentally extracted conductance reconstruction. A comprehensive conduction switching model based on ion migration, conductive pathway modulation, and photo-induced carrier modulation is proposed to explain the multifunctional response. These results demonstrate the potential of the proposed platform for artificial vision and in-sensor neuromorphic computing.