Design, Simulation, and Optimization of 3D Si/SiO2/Si Asymmetric Grating for Mid-InfraRed Filters

This paper investigates and optimizes mid-wave infrared (MWIR) optical filters based on three-dimensional asymmetric grating structures of Si/SiO₂/Si. MWIR filters hold significant importance due to their extensive applications in thermal imaging, spectroscopy, and sensing. The primary aim of this study is to achieve the maximum figure of merit (FOM) through precise tuning of the grating's structural parameters. Initially, the transmission spectra of the filters in the 9–11 µm wavelength range were simulated for various structures using the finite element method (FEM). Analysis of these spectra revealed two resonance modes exhibiting different interference behaviors based on changes in asymmetry parameters. Key parameters of each resonance mode, including transmission, peak wavelength, linewidth, and dip depth, were evaluated to calculate quality factors (Q-factor) and FOM. To optimize the FOM, two approaches were employed: artificial neural network (ANN) fitting on FEM data and direct optimization using the genetic algorithm (GA). The ANN model predicted a maximum FOM of 11.51 1/µm with optimal parameters: w_sx = 80%, w_sy = 10%, t_sy = 10%, and t_sx = 80%. Subsequently, the GA achieved an optimal FOM of 13.55 1/µm through direct parameter space search, with the best parameters being t_sx = 86.98%, w_sx = 65.39%, t_sy = 45.76%, and w_sy = 31.45%. These findings demonstrate that precise tuning of asymmetry parameters and applying appropriate optimization methods can significantly enhance the performance of MWIR filters and improve their spectral resolution.