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

Area of Science:

• Materials Science
• Supramolecular Chemistry
• Machine Learning Applications

Background:

• Circularly polarized luminescence (CPL) materials are crucial for chiral functional devices.
• Achieving high dissymmetry factors (g_lum) in CPL material synthesis is challenging.

Purpose of the Study:

• To apply machine learning (ML) for guiding the synthesis of G-quartet-based CPL gels.
• To identify optimal synthesis parameters for high g_lum values.
• To explore ML-driven rational design of chiral nanomaterials.

Main Methods:

• Utilized an "experiment-prediction-verification" approach.
• Developed ML classification and regression models for solvothermal synthesis in deep eutectic solvents.
• Employed a decision tree algorithm, evaluating six ML models.

Main Results:

• The decision tree ML model achieved high accuracy (0.97) and determination coefficient (0.96).
• Identified G-quartet CPL gels with g_lum values up to 0.15.
• Demonstrated ML's effectiveness in correlating synthesis parameters with g_lum.

Conclusions:

• ML significantly streamlines the rational design and synthesis of chiral nanomaterials.
• This work establishes a precedent for using ML in developing advanced CPL materials.
• The developed ML models provide a pathway for rapid discovery of high-performance CPL gels.