Yang Jianji, Fan Jonathan A
Opt Lett. 2017 Aug 15;42(16):3161-3164. doi: 10.1364/OL.42.003161.
Topology optimization is a powerful iterative inverse design technique in metasurface engineering and can transform an initial layout into a high-performance device. With this method, devices are optimized within a local design phase space, making the identification of suitable initial geometries essential. In this Letter, we examine the impact of initial geometric layout on the performance of large-angle (75 deg) topology-optimized metagrating deflectors. We find that when conventional metasurface designs based on dielectric nanoposts are used as initial layouts for topology optimization, the final devices have efficiencies around 65%. In contrast, when random initial layouts are used, the final devices have ultra-high efficiencies that can reach 94%. Our numerical experiments suggest that device topologies based on conventional metasurface designs may not be suitable to produce ultra-high-efficiency, large-angle metasurfaces. Rather, initial geometric layouts with non-trivial topologies and shapes are required.
拓扑优化是超表面工程中一种强大的迭代逆向设计技术，可将初始布局转变为高性能器件。使用这种方法时，器件在局部设计相空间内进行优化，因此确定合适的初始几何形状至关重要。在本信函中，我们研究了初始几何布局对大角度（75°）拓扑优化超光栅偏转器性能的影响。我们发现，当基于介电纳米柱的传统超表面设计用作拓扑优化的初始布局时，最终器件的效率约为65%。相比之下，当使用随机初始布局时，最终器件具有可达到94%的超高效率。我们的数值实验表明，基于传统超表面设计的器件拓扑可能不适用于制造超高效率、大角度的超表面。相反，需要具有非平凡拓扑和形状的初始几何布局。
