Neural network modelling on contact angles of liquid metals and oxide ceramics

Abstract

A neural network model was developed in this paper to predict the contact angles of 21 metals and 14 solid oxides. 15 factors were used in the neural network model to distinguish different metal and oxide categories and experimental conditions. With 1 120 contact angle values as the learning data, the neural network model was successfully developed. It can properly reproduce the experimental data on contact angles of molten metals and solid oxides under various conditions. Specifically, only three predictions among the total 1 155 predictions were over 20% deviation from the experimental data. All the predictions on the 35 test data are within 20% deviation from the experimental values. Factors such as oxygen partial pressure and surface tension of molten metal were found to be important for a good model prediction. With the developed model, contact angle values of Fe and CeO2 were predicted.