Anti-swelling dual-network organohydrogel fiber sensors for deep-learning-assisted underwater communication

Conductive hydrogel fibers are ideal for wearable underwater sensing due to their weavability, conformal deformation adaptability, and ionic conductivity. However, conventional hydrogel fibers swell severely in water, causing mechanical deterioration, disrupted conductive pathways and signal instability, which limits their underwater applications. Herein, dual-network polyvinyl alcohol (PVA)/sodium alginate (SA)/glutaraldehyde (GA) (PSG) organohydrogel fibers were fabricated by combining wet-spinning, freeze-thawing and solvent replacement. The covalently crosslinked PVA-GA network acts as a stable structural skeleton, the ionically coordinated SA-Ca²⁺ network serves as reversible sacrificial bonds for energy dissipation, and freeze-induced PVA microcrystals enhance crosslinking density to restrain swelling. The glycerol/water binary solvent system endows the fibers with excellent low-temperature tolerance. The resultant fibers exhibit a tensile strength of 2.32 MPa, elongation at break of 610%, an ultra-low 7 day swelling ratio of 4.95%, wide-temperature conductivity from -25 °C to 60 °C, and outstanding strain-sensing performance with a maximum gauge factor of 3.53, 150 ms response time and stable operation over 1500 loading-unloading cycles. They can monitor full-scale human motions, and combined with bidirectional long short-term memory deep learning, achieve an underwater Morse code communication system with 97.3% recognition accuracy for 26 English letters, supporting wearable sensing in underwater scenarios.