Machine-learning molecular simulations show role of atomic-scale roughness in hydrophilicity

Hydrophilicity governs key processes in catalysis, corrosion, and energy materials. However, the molecular origin of hydrophilicity remains poorly understood. Even for common materials such as aluminum oxide, experiments show that different crystal faces can have very different behaviors when interacting with water.

Choutipalli et al. used machine-learning molecular simulations to characterize the structure and dynamics of the hydrogen-bond networks at three distinct alumina surfaces. Using a neural network potential trained on data using density-functional theory, the group assessed the nanosecond molecular dynamics of water on three alumina crystallographic planes at the surface, showing how atomic-scale roughness and hydroxyl group arrangements dictate hydrophilicity.

Traditional simulations are typically too short or not accurate enough to capture multiple structural and dynamical observables to decode hydrophilicity at the atomic level. The team sought to overcome these limitations with machine learning.

“Our results establish a mechanistic foundation for hydrophilicity that can be extended to other oxides and heterogeneous interfaces,” said author Venkata Surya Kumar Choutipalli. “The identified link between surface corrugation, hydrogen-bond strength, and water ordering provides a quantitative framework for designing materials with tunable wetting properties. This can influence applications ranging from catalyst supports and anti-fouling coatings to membranes and energy storage interfaces.”

Water diffusion, density fluctuations, hydrogen-bond lifetimes, and vibrational spectra demonstrated that hydrophilicity increased with more corrugated surfaces. The most corrugated surface, (0112), traps water most strongly and forms the most robust hydrogen bonds.

“These results show that surface topography and hydroxyl arrangement — not just chemistry — govern how water adheres and organizes, providing a microscopic picture that unites structural, dynamic, and vibrational observations,” Choutipalli said.

They next look to explore amorphous and defect-rich alumina surfaces, where local disorder and roughness might further enhance hydrophilicity.

Source: “On the origin of hydrophilic interactions at alumina surfaces,” by Venkata Surya Kumar Choutipalli, Michael L. Klein, and Mark DelloStritto, Journal of Chemical Physics (2025). The article can be accessed at https://doi.org/10.1063/5.0294178 .