**About this Abstract** |

**Meeting** |
**2025 TMS Annual Meeting & Exhibition
** |

**Symposium
** |
**Computational Thermodynamics and Kinetics
** |

**Presentation Title** |
Calculation of thermodynamic properties of mixed oxides using modified polyhedron model |

**Author(s)** |
Jesus Alejandro Arias Hernandez, Sun Yong Kwon, Elmira Moosavi-Khoonsari |

**On-Site Speaker (Planned)** |
Jesus Alejandro Arias Hernandez |

**Abstract Scope** |
CALPHAD methodology relies on knowledge of the thermodynamic properties of compounds such as enthalpy, entropy, and heat capacity. However, access to these properties is limited for many oxides due to the complex nature of their experimental determination. The Polyhedron model is based on the assumption that every compound can be segmented into invariant polyhedra constituted by a cation surrounded by anions whose properties are estimated by linear regression. However, despite modifications done in the past to include magnetic and site disorders, it still overlooks non-vibrational contributions such as second nearest neighbors’ contributions, polyhedra distortion, and the linkages between them. Therefore, improvements are necessary to increase its predictability. This work focuses on further modifications of the Polyhedron model by including an artificial neural network model to account for the excluded contributions, which has been shown to improve the accuracy of the Polyhedron model for enthalpy and entropy. |

**Proceedings Inclusion?** |
Planned: |

**Keywords** |
Machine Learning, Modeling and Simulation, Other |