Erlangen 2022 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

Q: Fachverband Quantenoptik und Photonik

Q 31: Photonics I

Q 31.4: Vortrag

Mittwoch, 16. März 2022, 11:15–11:30, Q-H15

Deep learning assisted design of high reflectivity metamirrors — •Liam Shelling Neto^{1, 2}, Johannes Dickmann^{1, 2}, and Stefanie Kroker^{1, 2, 3} — ¹Institut für Halbleitertechnik, Braunschweig, Deutschland — ²Laboratory for Emerging Nanometrology, Braunschweig, Deutschland — ³Physikalisch-Technische Bundesanstalt, Braunschweig, Deutschland

Manipulating light in an ever so complex manner can be a complicated task. Metasurfaces, i.e. two-dimensional periodic nanostructures of sub-wavelength size, allow exotic applications in wavefront manipulation for the price of nonintuitive design of the surfaces building blocks. Since the mapping of a given design to the underlying electromagnetic response is highly non-linear, common approaches involve numerous simulations to optimize the device performance to given requirements. With increasing functionality of the metasurface, the parameter space that necessary to provide enough flexibility can be rather large and thus, difficult to control. When it comes to the application of metasurfaces as focusing mirrors in ultra-stable cavities or future gravitational wave detectors, those devices face unprecedented requirements such as high reflectivity, optimal phase agreement etc. Here, we utilize powerful deep learning algorithms to implement an inverse design framework that handles large parameters spaces with ease in order to design high-reflectivity metamirrors.