Sustainable by All-Biobased Design: Recyclable Flax Fiber/Silk Matrix Composites with Machine Learning-Assisted Mechanical Performance Prediction

Abstract

Abstract Fiber-reinforced polymer composites are indispensable structural materials, yet their sustainability is limited by two key issues: the environmental impact of manufacturing and the difficulty of recycling at the end of life. Here, we develop fully biobased composites by combining flax fibers with a regenerated silk fibroin matrix to overcome both obstacles. Owing to the room-temperature self-assembly capability of the silk matrix, chemical additives such as crosslinkers, curing agents, and catalysts are completely avoided during fabrication. The resulting composite achieves a tensile strength and modulus of up to ∼70 MPa and 2.4 GPa, comparable to conventional petroleum-based counterparts, while the silk matrix can be dissolved to non-destructively recover flax fibers for repeated reuse. To complement this all-biobased material design, we further construct a machine learning model trained on experimental datasets for predicting the mechanical properties of fiber-reinforced silk matrix composites. By narrowing the experimental formulation search space and identifying a promising fiber-content range for subsequent experimental validation, this data-driven approach serves as a low-resource design aid that reduces trial-and-error experimentation. Collectively, the combination of fully biobased composite design with machine learning-guided formulation offers a promising pathway toward sustainable composite development from the present to the future.