Data-driven design of glasses with desirable optical properties using statistical regression

Yomei Tokuda,Misa Fujisawa,Daniel M Packwood,Yoshikatsu Ueda,Mei Kambayashi

doi:10.1063/5.0022451

Abstract

In this study, we used a data-driven approach to build models for assisting the design of new glasses with high refractive index and low dispersion. Our models, which are based on multiple linear regression and kernel ridge regression, achieved high accuracy in predicting optical properties of glasses based on their composition alone. Using the predictions of these models as a guide, we fabricated new glasses in our laboratory. In agreement with model predictions, these glasses had promising optical properties. This work therefore demonstrates a successful example of data-driven materials design and can be used as a template for designing glasses or other materials with other desirable properties.

Highlights

Most commercial glasses consist of a multi-component mixture of oxides
We focused on the 879 glasses with high refractive index between 1.9 and 2.0 because high refractive index is desirable for optical glass
The design of new glasses with targeted properties is an outstanding challenge in materials science

Summary

Introduction

Most commercial glasses consist of a multi-component mixture of oxides. The composition of this mixture can have dramatic effects on physical properties, and trial-and-error attempts at optimizing composition to obtain desirable properties are costly and have a large environmental load because of experimental waste. High refractive index and a large Abbe number are a desirable property for optical lenses with less chromatic aberration. According to the Abbe diagram, refractive index tends to decrease non-linearly as the Abbe number is increased. This trend is not favorable for making optical lenses, which require high refractive index and low dispersion. While several reports have proposed novel glasses with a large refractive index and large Abbe number, the discovery of these glasses via a purely experimental approach can be time-consuming

Methods

Results

Conclusion