Review and perspective on polyhedron model for estimating thermodynamic properties of oxides

Knowledge of thermodynamics and phase equilibria of oxidic materials is crucial for advancement in the field of ceramics and glass. With the development of computational thermodynamics, predicting phase diagrams and chemical reactions of multicomponent systems has become possible. However, there are still plenty of oxides, the thermodynamic properties of which have not been identified due to the challenges in conducting experiments. Therefore, a key to the advancement in thermodynamic modeling would be to develop a universal model that can be used to estimate the thermodynamic properties of oxides with reliable extrapolation capacity. Atomistic (or molecular) scale models are still insufficient in predicting the thermodynamic properties of oxides at any scale. Alternatively, among group contribution-based methods, the polyhedron model has presented its potential in the estimation of the thermodynamic properties of ionic crystals. However, this model still demands improvements that increase the model's accuracy and extrapolation capacity. In this paper, the background and the state-of-the-art of polyhedron model will be presented together with its strengths and shortcomings. Subsequently, it will be briefly discussed how the field of artificial intelligence could be exploited to devise the next generation of the polyhedron model, the modified polyhedron model.

Automated summary

Knowledge of thermodynamics and phase equilibria of oxidic materials is critical for advancements in ceramics and glass. Computational thermodynamics has made it possible to predict phase diagrams and chemical reactions, but challenges remain due to unidentified thermodynamic properties of many oxides. Developing a universal model for estimating oxide properties with reliable extrapolation capacity is key. Atomistic models are insufficient, while group contribution-based methods, like the polyhedron model, show potential but need improvements. This paper presents the background and current state of the polyhedron model, discussing its strengths, weaknesses, and the potential for developing an improved version using artificial intelligence.