Guided-mode resonance filters generated with genetic algorithms

Abstract

This paper focuses on the application of genetic algorithms to the study and design of reflection and transmission filters based on the guided-mode resonance (GMR) effect in waveguide gratings.1–3 As genetic algorithms are well suited for problems with multidimensional, large search spaces4, they may be used effectively for optical filter design involving multiple periodic and homogeneous layers. In this work, the genetic algorithm library PGAPACK5 is combined with a forward code based on rigorous coupled-wave analysis6 in a new computer program that optimizes the merit function of a multilayer diffractive optical structure. Thus, a GMR-filter response with a given central wavelength, linewidth and sideband levels can be specified with a corresponding diffractive structure yielding approximately the specified response provided by the program. The net effect of this approach is that the inverse problem of finding a structure (i. e., layer thicknesses, refractive indices, fill factors, grating period) that yields a given filter response can be solved. In addition to providing useful filter designs, this approach may aid in the discovery of diffractive structures with profiles that may differ significantly from those ordinarily treated.