The evolution of artificial intelligence, spurred by advancements in neuromorphic computing and large language models, demands increasingly energy-efficient computing systems. Photonic neural networks (PNNs), leveraging the high-speed, parallel processing capabilities of light, have emerged as a promising avenue towards achieving this goal. This project explores the integration of an archetypal volatile phase-change material, vanadium dioxide (VO2), into PNNs to address the limitations of their current forms. By capitalizing on the unique properties of VO2, including its lower energy thresholds and potential for ultrafast phase changes, we aim to develop more compact and efficient PNN components, study their functions in contemporary and advanced neuromorphic topologies, and eventually experimentally demonstrate neuromorphic computing tasks with these innovative devices. Through a multidisciplinary approach combining expertise in nanostructure fabrication, optical simulations, and circuit design, we thus seek to unlock the untapped potential of VO2 in PNNs

Keywords
Vanadium dioxide;Tunable photonic circuits;Integrated photonic circuits;Metasurfaces