MolPackL: Quantification and Interpretation of Intermolecular Interactions Driven by Molecular Packing.

Area of Science:

• Materials Science
• Computational Chemistry
• Organic Electronics

Background:

• Organic material properties depend heavily on molecular packing and intermolecular interactions (IMIs).
• Understanding and quantifying these IMIs is challenging due to their complexity and disorder.
• Existing methods struggle to accurately assess the impact of IMIs on material properties.

Purpose of the Study:

• To develop an innovative algorithm framework, Molecular Packing Learning (MolPackL), for quantifying IMIs.
• To extract intermolecular features for predicting organic solid properties.
• To advance the design of high-performance organic optoelectronic materials.

Main Methods:

• Constructed an algorithm framework, MolPackL.
• Utilized contact density histograms (CDHs) to quantify IMIs.
• Employed class activation mapping (CAM) for visual identification of interaction sites.
• Performed elemental importance analysis to validate feature contributions.

Main Results:

• MolPackL accurately quantifies elusive IMIs and extracts relevant intermolecular features.
• Predictive models for IMI-related properties in molecular crystals showed satisfactory performance.
• The band gap predictive model achieved state-of-the-art performance (MAE of 0.20 eV, R² of 0.92).
• CAM visually confirmed MolPackL's accurate identification of effective interaction sites.

Conclusions:

• MolPackL provides a novel framework for assessing and understanding the influence of IMIs.
• This advancement deepens comprehension of IMIs' effect on material properties.
• MolPackL is a powerful tool for designing high-performance organic optoelectronic materials.