Accelerating Materials Discovery of Novel Europium(II)-Activated Phosphors through Machine Learning Classification of Europium Valences

Abstract

An approach is presented to accelerate the discovery of host compounds for novel Eu²⁺-activated phosphor materials by integrating systematic data collection, machine learning, and experimental validation. A data set of Eu²⁺- and Eu³⁺-activated phosphors has been constructed using systematic data collection methodology from numerous academic articles. A machine-learning classification model has been developed using the collected data set to predict the oxidation states of Eu ions in potential hosts regarding luminescence. The model considers the nonexclusive nature of the divalent and trivalent oxidation states of Eu ions in phosphor applications. A comprehensive exploration of a materials database was conducted to identify host candidates for novel Eu²⁺-activated phosphor materials, leading to attempts to synthesize them. Photoluminescence analysis revealed the successful synthesis of 12 new Eu²⁺-activated phosphors, demonstrating the potential of the proposed approach for accelerating material discovery.