Optimization of Reflection and Transmission Characteristics of a Waveguide Window

Waveguide windows are known to be major components of transmission lines used with highpressure or vacuum applicators, particle accelerators, and microwave plasma devices [1-5]. The function of the window is to provide vacuum/gas isolation of the source from the cavity while transmitting microwaves with minimum attenuation. It essentially consists of a dielectric plate surrounded by a metal frame and sandwiched between special flanges so that the cross-sectional dimensions of the plate are larger than the corresponding waveguide dimensions. In practice, to reduce reflections of a single window, the latter is matched by an additional inductive or capacitive iris [2]; when dealing with a double window, the mutual orientation of the two dielectric plates is chosen the way that the total reflection is minimized [3]. Known theoretical characterizations of the waveguide window include the approaches based on substantial idealizations of the problem and exploitation of transmission-line equations [1, 3, 4] as well as more adequate considerations supported by advanced numerical techniques – finite elements (code ANSYS) [5] and finite integration (code MAFIA) [2]. However, neither validation, nor estimation of accuracy of these models has been given. Moreover, the entire structure of a waveguide window has never been optimized in terms of the low power reflections. This paper, for the first time, presents an accurate numerical analysis of reflections and transmissions of a double window, demonstrates a possibility of detection, identification, and visualization of trapped (ghost) modes [6], and generates a series of optimized geometries guaranteeing fairly low reflection and high transmission. The structure has been analyzed with the use of 3D conformal FDTD method (code QuickWave-3D) [7] and optimized by the artificial neural network procedure outlined in our earlier paper [8].