Multi-objective optimization of meta-atoms

Metasurfaces are thin structures that can control electromagnetic propagation in a variety of ways. This is done by designing discrete elements (meta-atoms) with desired properties that form the building blocks of the metasurface. In the past, these meta-atoms have often been comprised of simple shapes (e.g., rectangles, ellipses, split-rings). However, greater throughput and multiple functionalities can be achieved by more complex shapes. In the present work, a single method to represent most shapes is demonstrated. By combining this shape generation method with a multi-objective optimizer, the best meta-atom designs can be found that achieve good performance for multiple objectives. This method is demonstrated by designing a polarization-selective, beam steering application where transmitted electromagnetic radiation deflects into the +1 or-1 diffraction order depending on the incident wave polarization.