Machine learning assisted inverse design on mechanically tunable lateral hybrid metasurface

The proliferation of artificial intelligence (AI) has resulted in a surge of metasurface design for various purposes. Our latest work [1] has introduced the lateral hybrid system for the first time, where resonators composed of two different materials are placed horizontally next to each other to form a lattice. Such design demonstrates a sensitivity that is five times greater than conventional tunable metasurfaces. With this work, we have successfully implemented an AI-assisted design process to develop an optimized lateral hybrid metasurface for structural colour applications. Our results have demonstrated that the lateral hybrid concept can enhance the performance and optical tuning of various semiconductor and high dielectric metasurfaces, such as TiO2 and Si3N4. This approach has the potential to unlock new opportunities for using metamaterials in high precision sensors with increased sensitivity, and represents a major step forward for the field.