Inspired by biology, researchers have achieved the highest performing underwater adhesive hydrogel technology to date through a data mining and machine learning approach.
Hydrogels are a permeable soft material consisting of polymer networks and water with applications ranging from bio-medical engineering to contact lenses. Intrinsic to hydrogels is the ability to endow diverse characteristics by modifying their polymer networks. Professor Gong's research lab at WPI-ICReDD, Hokkaido University, specializes in hydrogel technology and has engineered hydrogels with self-strengthening, self-healing, underwater adhesion properties and more. For adhesive hydrogels, achieving instant, strong, and repeatable underwater adhesion is a prevailing challenge.
Through a combination of data mining and machine learning, Professor Gong, Professor Takigawa, Professor Fan, graduate student Liao, and colleagues have recently developed the strongest underwater-adhesive hydrogels to date with adhesive strengths (Fa) exceeding 1 MPa. The gels' strength was both instant and repeatable and they are functional across various surfaces under variable levels of salinity from pure water to seawater. This research was published in Nature and was selected for the cover.
For reference, if these hydrogels were cut to the size of a single postage stamp (2.5 x 2.5 cm), they could theoretically support ~63 kg (e.g. an adult human). The researchers demonstrated the hydrogel's adhesive strength by applying it to a rubber duck on a seaside rock where it withstood repeated ocean tides and wave impacts.
Taking inspiration from biology, these hydrogels were designed with polymer networks derived from adhesive proteins found in archaea, bacteria, eukaryotes, and viruses. Despite the diversity across these organisms, these proteins share common sequence patterns that endow adhesion in wet environments. For this, ~25,000 adhesive protein datasets, collected from the National Center for Biotechnology Information (NCBI) protein database, were data mined for relevant amino acid sequences important for underwater adhesion.
They replicated these sequences into polymer networks and synthesized 180 hydrogels-each containing unique polymer networks. The data compiled from studying these hydrogels were analyzed with machine learning which further extrapolated the most significant polymer sequences. The original 180 gels synthesized from data mining demonstrated adhesive qualities greater than gels previously reported in the literature. However, the gels inspired by machine learning were more incredible, exceeding the highly desired qualities mentioned above.
Repeatable and instant adhesion are highly desired qualities for applications ranging from biomedical engineering and deep-sea exploration. These qualities are confirmed in an experiment in which the water leak from a damaged pipe could be covered instantly and repeatedly.
The significance of the data driven approach in this research is clearly highlighted upon comparison of these hydrogels with previous conventional models. Such a distinct advancement in overall performance should lead to exciting new discoveries and applications for adhesive hydrogel applications.
Original article:
Hongguang Liao, et al. Data-Driven De Novo Design of Super-Adhesive Hydrogels. Nature. August 7, 2025.
DOI: 10.1038/s41586-025-09269-4
Funding:
This research was supported by JSPS KAKENHI Grant (JP21K17745, JP21K14676, JP22K21342, JP22H04968, JP24K17728).
