Prediction of luminescence lifetimes of Mn4+/Eu3+ doped phosphors based on interpretable machine learning

Jiayang Zhou, Haozhe Chen, Xiangfu Wang,
Prediction of luminescence lifetimes of Mn4+/Eu3+ doped phosphors based on interpretable machine learning,
Sensors and Actuators A: Physical,
Volume 394,
2025,
116975,
ISSN 0924-4247,
https://doi.org/10.1016/j.sna.2025.116975.
(https://www.sciencedirect.com/science/article/pii/S0924424725007812)
Abstract: Luminescence lifetime is a key parameter for evaluating the light energy conversion efficiency of luminescent materials. However, the magnitude of luminescence lifetime is affected easily by the material structure, defects, temperature, and synthesis method. It is necessary to detect an effective prediction method to cooperate with experiments and achieve the controllability. Consequently, this paper introduces an innovative machine learning framework designed to predict the luminescence lifetime of active ion doped phosphors. This framework aims to rapidly determine the luminescence lifetime of Mn4+/Eu3+ doped phosphors and investigate the relationship between material structure and luminescent properties. By constructing a multi-feature dataset and systematically evaluating 16 machine learning models across four categories, five high-performing algorithms have been identified. Among them, the determination coefficient R2 of the Random Forest regression optimized by Bayesian on the test dataset is 0.81, showing an excellent prediction performance. Furthermore, through explainability analysis, the relationship between material features and luminescence lifetime are thoroughly examined. This machine learning framework eliminates the need for experimental measurements. It offers an efficient and low-cost approach for guiding the selection and design of new phosphors.
Keywords: Interpretable Machine Learning; Luminescence lifetimes