Transient Prediction of Nanoparticle-Laden Droplet Drying Patterns through Dynamic Mode Decomposition

Area of Science:

• Materials Science
• Fluid Dynamics
• Machine Learning

Background:

• Nanoparticle-laden sessile droplet drying is crucial for various applications.
• Predicting droplet evaporation dynamics and particle self-assembly remains challenging.
• Accurate prediction of drying patterns is essential for controlling material deposition.

Purpose of the Study:

• To develop a data-driven machine learning algorithm for predicting transient drying patterns of aluminum oxide (Al2O3) nanoparticle-laden sessile droplets.
• To utilize a single top-view camera data point for prediction.
• To assess the algorithm's accuracy in predicting drying patterns, droplet diameter, and coffee-ring thickness.

Main Methods:

• A data-driven machine learning algorithm was developed.
• Dynamic Mode Decomposition (DMD) was employed as the learning model.
• Transfer learning was applied to individual nanoparticle-laden droplet systems.

Main Results:

• The algorithm predicted transient drying patterns with less than 10% error.
• Predictions for droplet diameter showed less than 0.13% error.
• The model accurately predicted coffee-ring thickness (2.0-6.7 μm) and recognized liquid volume and drying regime transitions.

Conclusions:

• The developed data-driven approach accurately predicts nanoparticle-laden sessile droplet drying patterns.
• Machine learning, specifically DMD with transfer learning, offers a powerful tool for analyzing complex droplet dynamics.
• This method enables precise control over material deposition in applications involving nanoparticle drying.