Regensburg 2025 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

DS: Fachverband Dünne Schichten

DS 6: Thin Film Application

DS 6.8: Talk

Wednesday, March 19, 2025, 11:30–11:45, H3

High Accuracy Reflection Prediction Model for Multi-Layer Anti-Reflection Coatings Using Deep Learning and Machine Learning — •Iremnur Duru, Semih Oktay, and Timuçin Emre Tabaru — Department of Electrical Electronics Engineering, Sivas University of Science and Technology, 58000 Sivas, Turkey

In order to optimize the thickness parameters, this work employs Machine Learning (ML) and Deep Learning (DL) approaches to develop an accurate reflection prediction model that will direct the design of filters with multilayer Anti-Reflection Coating (ARC). A dataset of information derived from 3000 (1500 Ge- Al2O3, 1500 Ge- SiO2) computer simulations based on the thicknesses of multilayer structural materials has been used to create this model. Al2O3 and SiO2 served as the second layers in both coatings, with Ge serving as the substrate. Reflectance values for wavelengths ranging between the 3-5 *m and 8-12 *m bands characteristic of the mid-wave infrared (MWIR) and long-wave infrared (LWIR) bands are included in the data set. The average reflectance in the given 2-layer data set was at least 0.36 at thicknesses of 515 nm Ge and 910 nm SiO2. In terms of predicting reflectance values, the results demonstrate that machine learning (ML) models*specifically, decision tree, random forest and bagging methods perform better than the DL model and offer a useful guide for conceptualizing and manufacturing optical thin-film filters.

Keywords: Anti-reflection coating; Machine learning; Deep learning; Prediction model