Search for: All records

Supramolecular polymer blends (SPBs) represent a versatile class of polymers whose morphology directly determines their macroscopic properties. However, rational design of SPBs remains hindered by the lack of predictive models describing how molecular features and intermolecular interactions determine morphology. Here, we report a data-driven high-throughput workflow integrating modular synthesis, robotic sample formulation and processing, automated morphology characterization, and machine learning (ML) for SPBs discovery. Using a plug-and-play modular synthetic strategy, 33 hydrogen-bonding end-functional homopolymer precursors were prepared and orthogonally paired to fabricate 260 SPBs within one day. A custom automated atomic force microscopy (AFM) protocol enabled systematic morphological characterization, producing 2340 images with little human intervention. Average phase separation sizes (e.g. domain spacings) was extracted from processed AFM data using multiple complementary approaches and applied to ML model training. Leveraging the high-throughput sample formation and characterization, a high-quality database was curated for SPBs, allowing training of ML models. Guided by support vector regression (SVR) model, target morphologies of 50, 100, and 150 nm were successfully predicted and experimentally validated. This work demonstrates the potential of coupling high-throughput experimentation with ML to accelerate polymer blends phase discovery, providing one of the first large-scale, experimentally derived datasets specifically designed for supramolecular polymer research.