Curating Benzothiophene Experimental Absorption and Emission Spectra to Design Fluorescent Organic Polymer Chemical...

Area of Science:

• Organic Chemistry
• Materials Science
• Computational Chemistry

Background:

• Benzodithiophene (BDT) chromophores are key components in fluorescent organic polymers.
• Predicting photophysical properties of these materials is crucial for designing new applications.
• Machine learning offers a powerful approach to accelerate materials discovery.

Purpose of the Study:

• To develop and apply machine learning models for predicting photophysical properties (λmax and λe) of BDT-based fluorescent organic polymers.
• To design novel BDT-based polymers with targeted emission characteristics.
• To assess the synthetic accessibility of the designed polymers.

Main Methods:

• Utilized Rapid Discovery Kit (RDKit) to generate molecular descriptors for known BDT chromophores.
• Employed machine learning models, including Linear Regression, Random Forest, and Decision Tree, to predict λmax and λe.
• Analyzed SHapley Additive exPlanations (SHAP) values to identify key molecular features influencing properties.
• Designed 5,000 new polymers and evaluated their synthetic accessibility using the Synthetic Accessibility Likelihood Index (SALI).

Main Results:

• Machine learning models achieved high prediction accuracy, with R² values between 0.96 and 0.98.
• Labute Accessible Surface Area and the number of Rotatable Bonds were identified as influential features.
• Designed polymers exhibited predicted emission wavelengths extending up to 987 nm.
• Top-ranked polymers showed high SALI scores (up to 3.21), indicating good synthetic accessibility.

Conclusions:

• Machine learning provides an effective framework for designing novel fluorescent organic polymers based on BDT chromophores.
• The study advances the understanding of structure-property relationships in BDT-based materials.
• The designed polymers represent promising candidates for various optoelectronic applications.