Predicting Sliding Angles on Random Pit-Distributed Textures Using Probabilistic Neural Networks

Area of Science:

• Surface science
• Materials science
• Tribology

Background:

• Superhydrophobic surfaces are crucial for droplet manipulation.
• Existing models for sliding angle (SA) are limited to regular textures.
• Random textures present challenges in predicting droplet behavior.

Purpose of the Study:

• To investigate the sliding angle (SA) of superhydrophobic surfaces with random microtextures.
• To understand the influence of pit location on droplet sliding.
• To develop a predictive model for SA on complex surfaces.

Main Methods:

• Fabrication of randomly pitted microtextured surfaces with a 19% area ratio.
• Analysis of the three-phase contact line movement and contact angle (CA).
• Development of a Probabilistic Neural Network (PNN) model using pit coordinates and SA.

Main Results:

• Identical contact angles (CA) but varying sliding angles (SA) were observed for random textures.
• Pit location was identified as a key factor influencing SA.
• The continuity of the three-phase contact angle (T) showed a poor linear correlation (R² = 74%) with SA.
• The PNN model achieved 90.2% accuracy in predicting SA.

Conclusions:

• Droplet sliding on random textures is complex and influenced by pit arrangement.
• SA prediction for random textures is challenging with simple models.
• A PNN model offers a promising approach for accurate SA estimation on complex surfaces.