Adaptive direct ink writing of dual-curable, silica-filled rubber latex composites via a hybrid physics-machine learning framework

Additive manufacturing of high - performance filled rubber composites that industrially relevant mechanical and chemical performance remains a significant challenge, particularly at silica loadings required for stiffness, durability, and chemical resistance. This study presents a dual - curable waterborne latex - silica ink system that enables the direct ink writing (DIW) of carboxylated styrene - butadiene rubber composites with up to 50 wt% silica, without compromising printability or final performance. The ink incorporates a photocurable N - vinylpyrrolidone/triethylene glycol divinyl ether interpenetrating network, which provides immediate green strength upon UV exposure, followed by a secondary ZnO - activated sulfur vulcanization step that constructs a robust polysulfidic network. Rheological characterization revealed tunable yield stress and shear-thinning behavior, which are suitable for extrusion across a wide composition range. To ensure consistent print fidelity across 30–50 wt% silica, a machine-learning-based adaptive viscous filament printing (ML-AVFP) framework was implemented, which uses experimentally determined rheological descriptors and line-printing calibration data to automatically rewrite G-code with composition-specific Z -offset and print-speed profiles, thereby improving filament flattening/interlayer overlap, suppressing printing-induced porosity, and reducing final post-vulcanization shrinkage by ~17–30% versus fixed-parameter DIW. The resulting dual-cured composites exhibited tensile strengths of up to 21 MPa, elongations approaching 800%, high toughness, low compression set from − 40 to 100 °C, and excellent oil resistance. Functional components, such as gaskets, bellows, and brake diaphragms, were successfully printed, dense, pore-free microstructures and reliable performance under realistic mechanical and fluid - exposure conditions. This study establishes a scalable hybrid materials - process strategy for the digital manufacturing of industrially relevant elastomer parts via DIW.