AI-Directed 3D Printing of Hierarchical Polyurethane Foams

Abstract

Hierarchical porous materials enable next-generation protective, thermal, and biomedical devices by leveraging multiscale architectures with tunable mechanical and thermal properties. Current 3D printing techniques mainly yield periodic lattices and support limited polymer types, restricting patterning possibilities and scalability. Stochastic foam architectures, with heterogeneous and interconnected pores, mimic biological structures for improved energy dissipation and functional adaptability. However, scalable additive manufacturing of such foams remains scarcely explored. Here, a direct ink writing (DIW) strategy couples static mixer-enabled reactive extrusion and in situ polymerization to manufacture stochastic polyurethane (PU) foams at ambient conditions, eliminating post-processing. Precise control over pore size (0.2 µm to 1.2 mm), porosity (65–95%), and open-cell architecture delivers thermal conductivities down to 0.067 W m−1 K−1 and elastic recovery exceeding 90% after 5000 cycles. A multi-agent artificial intelligence framework enables the patterning of spatially organized, bioinspired motifs into print-ready CAD geometries, resulting in architecturally structured foams with tailored anisotropy and spatial thermal management. Flow-rate modulation further tunes morphology, optimizing the balance between stiffness and damping. This manufacturing platform integrates stochastic pore formation, AI-guided patterning, and mechanical–thermal optimization to realize scalable, customizable materials for impact protection, wearable thermotherapy, and adaptive healthcare, advancing digital manufacturing, smart materials, and personalized function.