Finding heterogeneous nucleating agents for ice using a data-driven approach

Area of Science:

• Materials Science
• Crystallography
• Physical Chemistry

Background:

• Heterogeneous nucleation is crucial for phase transitions like ice formation.
• Identifying effective nucleating agents is challenging due to complex structural requirements.
• Current methods for discovering nucleating agents are often slow and empirical.

Purpose of the Study:

• To develop a high-throughput computational workflow for identifying potential heterogeneous ice nucleating agents.
• To screen a large database of inorganic compounds for ice nucleation capabilities.
• To experimentally validate the predictions of the computational workflow.

Main Methods:

• A data-driven workflow was developed to evaluate the structural fit between ice Ih and potential nucleator surfaces.
• The model considered crystal morphology features and Miller index planes up to (333).
• Bulk water immersion experiments were used to establish temperature-based criteria for nucleation behavior, leading to numerical tolerance limits.

Main Results:

• The workflow screened 3500 metal oxides and halides, predicting a small percentage as potential ice nucleators based on geometric matching.
• Experimental validation of 22 compounds showed a 64% prediction accuracy.
• Four new ice nucleators (CeO2, WO3, Bi2O3, Ti2O3) were identified, and copper tubing demonstrated ice nucleation properties.

Conclusions:

• The developed high-throughput workflow provides an efficient initial screening method for discovering heterogeneous nucleating agents.
• Geometric interface matching, despite its simplicity, is a valuable predictor of ice nucleation activity.
• This approach accelerates the discovery of new materials for phase change applications.