Mapping Boryl Radical Properties and Reactivity Using Machine Learning: The B-Rad and React-B-Rad Maps.

Area of Science:

• Organic Chemistry
• Computational Chemistry
• Machine Learning

Background:

• Boryl radicals are crucial in organic synthesis but their reactivity is difficult to predict.
• Understanding their steric and electronic properties is key to harnessing their synthetic potential.

Purpose of the Study:

• To develop a comprehensive classification system for boryl radicals.
• To create a predictive tool for boryl radical reactivity in organic reactions.
• To facilitate the rational design and virtual screening of boryl-radical reagents.

Main Methods:

• Generated a database of 141 boryl radicals with DFT-calculated features.
• Applied unsupervised machine learning (k-means, PCA/UMAP) to create a "B-rad map".
• Integrated global electrophilicity/nucleophilicity indices and DFT-computed activation energies.
• Trained supervised machine learning (random forest) models to predict reactivity.

Main Results:

• The "B-rad map" visualizes steric and electronic trends across five clusters of boryl radicals.
• Polarity-annotated and React-B-rad maps link intrinsic properties to reaction performance.
• Machine learning models successfully predict boryl radical reactivity across various reaction types.

Conclusions:

• The developed platform provides a practical guide for experimentalists and a foundation for data-driven discovery.
• This approach enables rational design and virtual screening of boryl-radical reagents for synthetic applications.