Photon-controlled memristive synapses: recent progress toward brain-inspired neuromorphic computing

Abstract

As Moore's law approaches its limit, there is an urgent need for alternative computing architectures capable of meeting rapidly growing data-processing and memory demands. Photonic information processing, which utilizes photons instead of conventionally used electrons, has emerged as a promising technological avenue. In recent years, optically controlled synaptic devices have garnered significant attention as building blocks for next-generation high-density computing, owing to their high bandwidth, ultrafast response times, low latency, and reduced energy consumption compared to their electronic counterparts. This article provides a comprehensive overview of recent advancements in photo-memristive mechanisms, materials, device architectures, synaptic emulation strategies, application prospects, and key performance metrics. A broad range of materials has been explored in device configurations that mimic the structural and functional characteristics of biological synapses. Efforts toward leveraging these devices for complex brain-inspired tasks, including their integration into large-scale arrays for neuromorphic hardware, are also highlighted. Despite notable progress, several challenges remain that must be addressed to enhance device performance further and enable their practical implementation in neuromorphic hardware. Finally, we discuss the current limitations and outline future perspectives for advancing optically driven synaptic devices.