Uncovering Fundamental Composition–Property Relationships in Halide Glasses Through Machine Learning
Journal of the American Ceramic Society · 2026-07 · 109 (7)
DOI: 10.1111/jace.71025 ↗At a glance
Journal article in Journal of the American Ceramic Society, published online on 2026-07-13. Written by Zhongming Wen and 9 co-authors. 41 references. Funded by: National Natural Science Foundation of China.
Abstract (original language)
Halide glasses, endowed with a broad infrared transmission window and low optical loss, stand as highly promising candidates for advanced specialty optical applications. However, reliable composition–property correlations for these materials remain insufficiently explored, which hinders their rational design and practical deployment. To address this gap, this study systematically investigates four key properties of halide glasses—density, glass transition temperature, coefficient of thermal expansion, and refractive index—and constructs a comprehensive, high‐quality corresponding dataset. Four state‐of‐the‐art machine learning (ML) algorithms—support vector machine (SVM), artificial neural network (ANN), multilayer perceptron (MLP), and random forest (RF)—were employed to develop predictive models, among which the RF model exhibited the optimal performance in terms of prediction accuracy and generalization ability. Leveraging Shapley Additive exPlanations (SHAPs) analysis, the regulatory mechanisms of individual components on each target property were elucidated, offering intuitive insights into the composition‐property interplay. Furthermore, SHAP interaction analysis revealed the significant synergistic and antagonistic effects between key component pairs, which are crucial for fine‐tuning material properties. These findings deepen the fundamental understanding of halide glass chemistry and provide robust, data‐driven guidance for the targeted design of novel glass systems for advanced optical applications.
Funding
- National Natural Science Foundation of China
Cite
APA
Wen, Z., Yang, H., Lu, J., Xu, R., Tang, G., Wu, M., Li, J., Song, X., Liu, X., & Qian, Q. (2026). Uncovering Fundamental Composition–Property Relationships in Halide Glasses Through Machine Learning. Journal of the American Ceramic Society, 109(7). https://doi.org/10.1111/jace.71025
Vancouver
Wen Z, Yang H, Lu J, Xu R, Tang G, Wu M, et al. Uncovering Fundamental Composition–Property Relationships in Halide Glasses Through Machine Learning. Journal of the American Ceramic Society. 2026;109(7). doi:10.1111/jace.71025
BibTeX
@article{Wen2026,
author = {Wen, Zhongming and Yang, Haolin and Lu, Junyan and Xu, Runzi and Tang, Guowu and Wu, Minbo and Li, Jiahong and Song, Xiangyang and Liu, Xueyun and Qian, Qi},
title = {Uncovering Fundamental Composition–Property Relationships in Halide Glasses Through Machine Learning},
journal = {Journal of the American Ceramic Society},
year = {2026},
volume = {109},
number = {7},
doi = {10.1111/jace.71025}
}
EUVIMED