Predicting Adhesive Free Energies of Polymer-Surface Interactions with Machine Learning

Area of Science:

• Polymer Science
• Materials Science
• Computational Chemistry

Background:

• Polymer-surface interactions are fundamental in biological and industrial contexts.
• Predicting these interactions accurately is challenging but crucial for material design.

Purpose of the Study:

• To develop a machine learning (ML) method for predicting polymer adhesion to decorated surfaces.
• To establish a connection between a polymer's sequence and its adhesive properties.

Main Methods:

• Simulated adhesive free energies for 20,000 coarse-grained polymer sequences.
• Employed support vector regression (SVR) models for prediction.
• Utilized coarse-grained simulations integrated with data-driven ML.

Main Results:

• Achieved inexpensive and reliable prediction of adhesive free energy.
• Demonstrated the predictive power of ML models based on polymer sequence.
• Established a quantitative relationship between sequence and adhesion.

Conclusions:

• The integration of coarse-grained simulation and ML offers a powerful tool for designing functional polymers.
• This work advances the understanding and prediction of polymer-surface interactions.
• Represents a significant step toward linking polymer composition to interfacial behavior.