Design of thin-film quasi-optical terahertz components using evolutionary algorithms

In this paper, the design of THz thin-film components using single and multi-objective evolutionary algorithms is introduced. Each structure is optimized through a HFSS-MATLAB interface with infinitely periodic boundary conditions and plane wave incidence. First, terahertz band-stop filters are investigated based on a dual-cross structure. The filters are optimized for rejection, bandwidth, and multi-resonant properties. A bandwidth of approximately 45 GHz with -25 dB of rejection at 300 GHz is observed, as well as multi-resonant filters at 250 and 300 GHz. The miniaturization of a THz bolometer is also investigated using a single-objective genetic algorithm in order to improve its noise equivalent power. From the optimization, a unit cell approximately λ₀=25 in size at 100 GHz is found, which is significantly smaller than the typical size constraints of traditional periodic structures.