Prediction of viscosity behavior in oxide glass materials using cation fingerprints with artificial neural networks

Abstract

We propose a novel descriptor of materials, named 'cation fingerprints', based on the chemical formula or concentrations of raw materials and their respective properties. To test its performance, this method was used to predict the viscosity of glass materials using the experimental database INTERGLAD. Using artificial neural network models, we succeeded in predicting the temperature required for glass to have a specific viscosity within a root-mean-square error of 33.0°C. We were also able to evaluate the effect of particular target raw materials using a model trained without including the specific target raw material. The results show that cation fingerprints with a neural network model can predict some unseen combinations of raw materials. In addition, we propose a method for estimating the prediction accuracy by calculating cosine similarity of the input features of the material which we want to predict.

Structure and properties of alkali aluminosilicate glasses and melts: Insights from deep learning

“…For silicate melts, there have been several attempts to predict melt and glass properties using machine learning (see the reviews of Tandia et al, 2019;Liu et al, 2019), stretching back to the work of Dreyfus and Dreyfus (2003) on the prediction of liquidus temperature with artificial neural networks. Regarding melt viscosity, Hwang et al (2020) recently proposed an approach based on machine learning and the use of "cationic fingerprinting" to predict the temperatures of three reference viscosity points (10 1.5 , 10 6.6 and 10 12 Pa•s) for Na 2 O-SiO 2 -Al 2 O 3 -CaO melts with an error of ± 33 °C. Cassar (2021) also proposed the ViscNet model, a PGNN model that combines either the VFT or the MYEGA viscosity equations (see section 2.5.2) with a neural network to perform viscosity predictions of silicate and aluminosilicate melts.…”