Machine Learning-Based Phase Prediction and Structural Stability Analysis of High Entropy Nitrides

Abstract

High Entropy Nitrides (HENs) represent a distinctive class of advanced materials characterized by the incorporation of multiple principal elements bonded with nitrogen. Their diverse compositional space and configurational complexity endow them with exceptional properties, including high hardness, oxidation resistance, and thermal stability. Despite their promising potential, the prediction of structural stability and phase classification of HENs poses a significant challenge due to the absence of robust computational models. This research addresses this critical issue by employing a machine learning-based approach to classify the phases and predict the structural stability of HENs. By utilizing a combination of structural and thermodynamic descriptors, several machine learning models were developed, including K-Nearest Neighbors (KNN), Random Forest (RF), Support Vector Machine (SVM), and Gaussian Naive Bayes (GNB). Techniques such as ADASYN were implemented to balance the dataset, thereby enhancing the performance of the models. Among the various models, the KNN demonstrated the highest prediction accuracy and robustness. This work significantly contributes to the data-driven discovery of stable HENs, providing a pathway to expedite experimental synthesis and characterization efforts.