Machine learning refractive index model and nitrogen implantation studies of zinc arsenic tellurite glasses

Abstract

The first time machine learning-based refractive index model proposed based on the density parameter using a glass dataset of 2000 oxide glass samples to predict refractive index of the xZnF₂-(20-x)ZnO-40As₂O340TeO₂. The study uses various machine learning techniques such as gradient decent, artificial neural network, and random forest regression to predict the refractive index and density of glasses. The random forest regression (RFR) model is found to be the most effective with a maximum R² value of 0.950 in the case of refractive index prediction and 0.926 for density prediction. The study also investigates the effects of nitrogen ion implantation on the glasses, finding that increased nitrogen dose causes a reduction in density and an increase in refractive index. The glass transition temperature decreases with increased nitrogen dose, possibly due to implantation defects. However, the glass stability increases with increasing implantation dose for low and high fluorine content glasses, likely due to the development of band gap defect levels and an increase in carrier concentration.