Empirical model for morphological evolution of crystallisation process using artificial neural networks

Abstract

Understanding and modelling crystals evolution is a major concern in crystallography science and engineering fields. A new approach for modelling nucleation patterns of ZnO-TeO2 crystallisation process is introduced. This approach utilises artificial neural network (ANN) models to estimate time-dependent nucleation and to predict the crystallisation directions that are based on prior formations, which were extracted from processed images of the crystal. Quantitatively, crystals evolution is predicted by a systematic combination of image analysis associated with ANN modelling systems. Different crystallisation stages were characterised by image analysis to distinguish each stage, and to extract created crystal information. It was found that the model is able to successfully predict the crystal evolution with respect to used nucleation seeds.