In the recent years the concept of metasurface has progressively emerged as a revolutionary means to manipulate the behavior of light at the nanoscale. These devices are nanostructured two-dimensional materials offering unprecedented control over the optical properties of light, leading to previously unattainable applications in flat lenses, holographic imaging, and emission control among others. The design of nanophotonics based optical components involves complex light matter interaction at nanoscale, in regimes for which the crude ray optics approximation does not hold, thus requiring advanced numerical modeling methods. Among those different methods, inverse design approaches have recently been reported to be effective solutions for achieving highly efficient and rather robust optical performances.
We study different approaches for the geometric parametrization of nanoantennas based on B-Splines or Gielis superformula. Numerical characterization of metasurfaces relies on a high order time-domain finite element solver from the DIOGENeS software suite.
We exploit statistical learning methods for the optimization of phase gradient metasurfaces. This study is conducted in collaboration with the group of Patrice Genevet at the Research Center for Heteroepitaxy and its Applications (CRHEA, CNRS).