Morphology prediction for polymer blend thin films using machine learning.

Area of Science:

• Materials Science
• Polymer Science
• Machine Learning

Background:

• Polymer blend thin films exhibit mesoscale morphologies dependent on processing parameters.
• Morphology is critical for determining thin film applications.
• Understanding phase separation in immiscible polymer blends is essential for materials design.

Purpose of the Study:

• To develop a machine learning (ML) framework for predicting the morphology of polystyrene/polymethyl methacrylate (PS/PMMA) blend thin films.
• To guide experimentalists in designing thin films with desired morphologies.
• To enhance the understanding of phase separation phenomena in polymer blends.

Main Methods:

• Utilized experimental parameters (PS weight fraction, PMMA molecular weight, concentration, substrate surface energy) as inputs for ML models.
• Employed multi-class classification to predict morphology types (column, hole, island).
• Implemented Support Vector Machine (SVM) and other ML algorithms, alongside explainable AI techniques.

Main Results:

• Achieved a highest prediction accuracy of 93.75% using the SVM algorithm.
• Explainable ML provided insights consistent with experimental observations, validating the model.
• Identified key parameters influencing phase separation and morphology formation.

Conclusions:

• The ML framework reliably predicts PS/PMMA blend thin film morphology.
• Insights from explainable AI deepen the understanding of polymer phase separation.
• Developed guidelines and a web tool to facilitate experimental design for specific thin film morphologies.