Amorphous Silicon Oxide Nanoflakes Coupled With Single‐Layer Graphene as Reliable Bio‐Photonic Synapse Integrating With Dual‐Functionalities of Photoreceptor and Memory Effects

Abstract Neuromorphic computing, characterized by self‐learning and highly parallel computing, addresses the von Neumann bottleneck inherent in traditional computing architectures. The photosynaptic device holds great potential in the past few years to mimic the pivotal behavior of biological synapses, facilitating information storage and processing for neuromorphic computing. Despite many materials and synaptic architecture has been well documented in previous literature, there is much debate concerning the solution toward effectively prolonging the retention time of photosynaptic devices without necessitating an increase in program voltages. In this study, defect engineering through modification of amorphous silicon oxide nanoflakes (SiONFs) with polymethyl methacrylate (PMMA) polymeric matrix, ruling dual‐functional optical receptor and floating gate, is demonstrated. It is revealed that the dangling configurations of SiONFs are passivated via PMMA modification, mitigating the non‐radiative recombination process; whereas deep trap states of oxygen vacancies associated with long‐level trapping of photoexcited electrons states challenge to be passivated, thereby inducing the hole currents in the graphene channel and achieving longer retention time through capacitive coupling. The integrated devices demonstrate the discernible memory effect to emulate the intricate functionalities observed in biological synapses, including photonic‐driven excitatory postsynaptic current, paired pulsed facilitation, short‐ and long‐term memory, and learning‐forgetting‐relearning behaviors.