Bio-Based Thermosetting Polymers from Microbial Oils for Machine Learning-Assisted Design.

Designing bio-based polymers with pre-defined thermo-mechanical properties remains a central challenge in sustainable materials science. We present a novel closed-loop materials design framework that integrates synthetic biology, polymer chemistry, and machine learning (ML) to engineer triglyceride-based thermoset polymers with tunable performance. Broader range of microbial oils are polymerized with divinylbenzene (DVB) and n-butyl methacrylate (BMA) to generate thermosets that are characterized to establish structure-property relationships, correlating fatty acid chain length, number and position of carbon-carbon double bonds, and fatty acid composition to thermo-mechanical properties, such as glass transition temperature (Tg), storage modulus (E′), and crosslink density (ε). Using a data-driven ML approach trained on composition–property datasets, we establish correlations between fatty acid structural features and bulk material properties. The resulting predictive models enable both forward (property prediction from composition) and inverse (composition design for target properties) pathways. This approach offers a paradigm shift toward the rational design of sustainable thermosetting materials and paves the way for developing high-performance, bio-derived polymers for structural and thermal applications.