Numerical demonstration of a novel multi-strategy algorithm for multilayer terahertz metasurface wave plate

Abstract Metasurface is an emerging artificial two-dimensional periodic structure with strong electromagnetic modulation capability, which is of great significance for realizing polarization conversion in terahertz(THz) band. Metal is highly electromagnetically active, strong electromagnetic resonance effects can also be generated between metal layers, so multilayer metal-structured metasurface wave plates have gradually become a research hotspot. However, the design of multilayer metasurface structural relies on traditional trial-and-error methods which is undoubtedly a huge waste of computational resources. In order to solve these problems, this paper develops and designs a novel multi-strategy parallel genetic-particle swarm (MSP-GA-PSO) algorithm by combining genetic algorithm (GA) with particle swarm algorithm (PSO), which is able to achieve over 100% efficiency improvement, meaning the computation time is approximately half times that of the traditional algorithm. Six metasurface quarter wave plates (QWPs) with bilayer metallic C-ring are derived by using the developed algorithm for simulation optimization in 0.1-0.5 THz and 0.5-1.8 THz. By analyzing the phase difference tolerance and the ellipticity tolerance of these structures, we can conclude that the phase difference tolerance has to play a greater impact for the achromatic bandwidth during the optimization process. This conclusion is conducive to saving computational resources and is significant for the design of multilayer ultra-bandwidth achromatic metasurface wave plates.