Automated and High-Throughput Phase Separation Control for Supramolecular Polymer Blends Enabled by Machine Learning

Area of Science:

• Materials Science
• Polymer Science
• Computational Chemistry

Background:

• Supramolecular polymer blends (SPBs) possess tunable morphologies crucial for macroscopic properties.
• Rational design of SPBs is hindered by a lack of predictive structure-morphology models.

Purpose of the Study:

• To establish a data-driven, high-throughput workflow for accelerated discovery of supramolecular polymer blends.
• To develop predictive models for correlating SPB structure with morphology.

Main Methods:

• Modular synthesis of 33 hydrogen-bonding end-functional homopolymers.
• Robotic formulation to create 260 SPBs.
• Automated atomic force microscopy (AFM) for morphology characterization.
• Machine learning (ML), specifically Support Vector Regression (SVR), for model training.

Main Results:

• Generated 260 SPBs and 2340 AFM morphology datasets rapidly.
• Developed an SVR model that accurately predicted phase-separation sizes.
• Experimental validation confirmed the model's predictive accuracy for target sizes (50, 100, 150 nm).

Conclusions:

• Coupling high-throughput experimentation with ML significantly accelerates morphology discovery in SPBs.
• This work provides a large-scale dataset for supramolecular polymer systems.
• The workflow enables more rational design of SPBs with targeted properties.