Predictive Analysis of Wettability of Al-Si Based Multiphase Alloys and Aluminum Matrix Composites by Machine Learning and Physical Modeling

Machine-Learning Approach in Prediction of the Wettability of a Surface Textured with Microscale Pillars

Tuning the wettability of a flat surface by introducing an array of microscale pillars finds wide applications, especially in engineering a superhydrophobic surface. The wettability of such a pillared surface is quantified by the contact angle (CA) of a water droplet. It is desired to know the CA prior to construction of pillars, in order to obviate the trial-and-errors in experimenting with many different topographies. Given an accurate theoretical prediction of CA has been elusive, we propose a convolutional neural network (CNN) model of CA for a surface patterned with rectangular or cylindrical pillars. By employing a three-dimensional descriptor of the surface topography, the present CNN model can predict experimental CAs within errors comparable to the uncertainties in measuring CAs.

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

The three-phase contact line best reflects the sliding ability of droplets on solid surfaces. Most studies on the sliding angle (SA) of superhydrophobic surfaces are limited to regularly arranged microtextured surfaces, lacking definite models and effective methods for a complex surface of a random texture. In this study, random pits with an area ratio of 19% were generated on 1 mm × 1 mm subregions, and the subregions formed arrays on a sample surface of 10 mm × 10 mm to obtain a randomly distributed microtexture surface with no pit overlaps. Although the contact angle (CA) of randomly pitted texture was the same, the SA was different. The SA of surfaces was affected by the pit location. The location of random pits increased the complexity of the three-phase contact line movement. The continuity of the three-phase contact angle (T) can reveal the rolling mechanism of the random pit texture and predict the SA, but the relationship between the T and SA is a relatively poor linear relation (R² = 74%), and the SA of the random pit texture can only be roughly estimated. The quantized pit coordinates and SA were used as the input and output labels for the PNN model, respectively, and the accuracy of the model convergence was 90.2%.

Surface Wettability Prediction Using Image Analysis and an Artificial Neural Network

In this study, a wettability-predicting method that uses an artificial neural network (ANN) by learning from digital images of the actual surface structures was developed. Polyester film surfaces were treated with oxygen plasma to realize various nanostructured surfaces. Surface structural characteristics from SEM images were quantified in a multifaceted way using a box-counting algorithm, a gray-level co-occurrence matrix algorithm, and binary image analysis. An ANN model that can predict wettability from surface structures was developed using the quantified surface structure and the resulting wettability as learning data. Furthermore, a surface with an optimal nanostructure to achieve superhydrophobicity was suggested by considering extracted surface structural parameters that significantly affect the surface wettability.

Surface Stoichiometry and Roughness of a Degraded A380 Alloy after Casting, Technical Cleaning and Packaging

The surface stoichiometry of the degraded surface of a commercial Al–Si casting was investigated. The die-cast component was previously stored in a sealed plastic envelope. After that, surface stains in the form of white layers were observed. X-ray photoelectron spectroscopy (XPS) was used to study these layers. For comparison, a seemingly unaffected area as well as a freshly cut surface of the casting were also analysed. In order to additionally assess the surface condition, surface roughnesses were measured. Based on the binding energies (BEs) of the Al and O in the XPS spectra, and the stoichiometric results, it was concluded that the surface layers of the degraded and undegraded samples consist mostly of aluminium oxide and aluminium hydroxide. Furthermore, sodium phosphate from the leftover detergent and silicon oxide were detected in both analysed areas. Analyses of the Al KLL Auger transition were used to corroborate this. The relative shares of Al oxide vs. hydroxide based on the elemental concentrations were determined. The chemical compositions and chemical states of the elements in the top layers were thus obtained. The combination of surface-sensitive analytical techniques was found to be a suitable tool for the ex-post identification of the source of defects.