Machine-Learning-Driven G-Quartet-Based Circularly Polarized Luminescence Materials

Circularly polarized luminescence (CPL) materials have garnered significant interest due to their potential applications in chiral functional devices. Synthesizing CPL materials with a high dissymmetry factor (g_lum ) remains a significant challenge. Inspired by efficient machine learning (ML) applications in scientific research, this work demonstrates ML-based techniques for the first time to guide the synthesis of G-quartet-based CPL gels with high g_lum values and multiple chiral regulation strategies. Employing an "experiment-prediction-verification" approach, this work devises a ML classification and regression model for the solvothermal synthesis of G-quartet gels in deep eutectic solvents. This process illustrates the relationship between various synthesis parameters and the g_lum value. The decision tree algorithm demonstrates superior performance across six ML models, with model accuracy and determination coefficients amounting to 0.97 and 0.96, respectively. The screened CPL gels exhibiting a g_lum value up to 0.15 are obtained through combined ML guidance and experimental verification, among the highest ones reported till now for biomolecule-based CPL systems. These findings indicate that ML can streamline the rational design of chiral nanomaterials, thereby expediting their further development.