Surface Wettability Prediction Using Image Analysis and an Artificial Neural Network.

Area of Science:

• Materials Science
• Surface Chemistry
• Computational Modeling

Background:

• Controlling surface wettability is crucial for various applications.
• Predicting wettability based on surface nanostructures is challenging.
• Existing methods often lack precision in correlating structure with function.

Purpose of the Study:

• To develop a novel wettability-predicting method using artificial neural networks (ANN).
• To quantify surface nanostructures from digital images for predictive modeling.
• To identify optimal nanostructures for achieving superhydrophobic surfaces.

Main Methods:

• Polyester films were treated with oxygen plasma to create diverse nanostructured surfaces.
• Surface structural characteristics were quantified using box-counting, gray-level co-occurrence matrix, and binary image analysis algorithms.
• An ANN model was trained with quantified surface data and measured wettability to predict surface behavior.

Main Results:

• The developed ANN model accurately predicted surface wettability based on quantified nanostructure parameters.
• Key surface structural parameters significantly influencing wettability were identified.
• The study successfully suggested an optimal nanostructure for achieving superhydrophobicity.

Conclusions:

• ANNs provide a powerful tool for predicting surface wettability from nanostructure images.
• Quantitative analysis of surface morphology is essential for accurate wettability prediction.
• This approach enables the rational design of surfaces with desired hydrophobic properties.