A Molecular Descriptor-Based Analysis of Organic Compounds and Designing New Promising Candidates by Machine...

Area of Science:

• Computational Chemistry
• Materials Science
• Machine Learning

Background:

• Energetic materials are crucial for various applications.
• Developing new energetic materials with enhanced properties like high heats of sublimation (ΔH) is an ongoing challenge.
• Data-driven approaches offer a promising avenue for accelerating materials discovery.

Purpose of the Study:

• To employ a data-driven approach using Python and machine learning (ML) to discover novel energetic materials with high heats of sublimation (ΔH).
• To identify key molecular descriptors influencing ΔH and synthetic accessibility (SA).
• To design and predict new organic compounds with high ΔH and favorable SA.

Main Methods:

• Trained multiple ML models (linear, random forest, gradient boosting, extra trees) on a dataset of 307 energetic compounds with experimental ΔH values.
• Utilized descriptors like ATS1s, mZagreb1, and Wpath, and employed SHapley Additive exPlanations (SHAP) for feature importance analysis.
• Designed 705 new organic compounds using optimized ML models and evaluated their predicted ΔH and synthetic accessibility (SA) based on Simplified Molecular Input Line Entry System (SMILES) length.

Main Results:

• ML models achieved high R-squared values (0.95-0.98), indicating excellent predictive performance.
• ATS1s and mZagreb1 descriptors showed high correlation with ΔH; ATS1s and Wpath significantly impacted model performance.
• Predicted ΔH values reached up to 170 kJ/mol for new compounds.
• Molecules with SMILES length between 500-700 exhibited the highest synthetic accessibility (SA).

Conclusions:

• ML-assisted design is a powerful strategy for discovering new energetic materials.
• Considering synthetic accessibility (SA) alongside performance metrics like ΔH is crucial for practical material development.
• The designed compounds with high ΔH and optimal SA show potential for applications in propellants, explosives, and other high-energy systems.