Genetic Algorithms to Automate the Design of Metasurfaces for Absorption Bandwidth Broadening

• Published in: ACS applied materials & interfaces Vol. 13; no. 6; pp. 7792 - 7800
• Main Authors: Zhang, Junming, Wang, Guowu, Wang, Tao, Li, Fashen
• Format: Journal Article
• Language: English
• Published: United States American Chemical Society 17.02.2021
• Subjects: absorption; algorithms; finite element analysis; Surfaces, Interfaces, and Applications; broad-band absorption; genetic algorithms; design strategy; metasurfaces; neural networks
• ISSN: 1944-8244; 1944-8252; 1944-8252
• DOI: 10.1021/acsami.0c21984

In this paper, we present a method to automate the design of an efficient metasurface, which widens the bandwidth of the substrate. This strategy maximizes the potential of the substrate for the application of broad-band absorption. The design is achieved by utilizing the coding metasurface and a combination of two types of intelligent algorithms. First, inspired by the coding metasurface, a large number of structures are generated to act as potential metasurface unit patterns by randomly generating the associated binary codes. Then, the binary codes are directly substituted as optimization objects into a genetic algorithm to find the optimal metasurface. Finally, a neural network is introduced to replace the finite element analysis method to correlate the binary codes with the absorbing bandwidth. With the participation of neural networks, the genetic algorithm can find the optimal solution in a considerably short time. This method bypassed the prerequisite physical knowledge required in the process of metasurface design, which can be used for reference in other applications of the metasurface.