Machine Learning Phase Classification of Thermoelectric Materials.

Area of Science:

• Materials Science
• Computational Materials Science
• Machine Learning Applications

Background:

• Thermoelectric (TE) materials are crucial for energy conversion, but exploring their phases is time-consuming and expensive.
• A vast number of TE alloys exist, necessitating faster methods for phase identification and material discovery.
• Machine learning (ML) models show promise in predicting material phases, including complex multi-principal element alloys.

Purpose of the Study:

• To develop and apply a Support Vector Machine (SVM) model for efficient classification of thermoelectric alloy phases.
• To address the need for time-efficient computational methods in accelerating the discovery of new TE materials.
• To demonstrate the robustness and accuracy of the ML model in distinguishing between various TE phases.

Main Methods:

• Implementation of a Support Vector Machine (SVM) classification model.
• Training and validation of the SVM model on datasets of thermoelectric alloys.
• Utilizing cross-validation techniques to assess the model's performance across different TE phases.

Main Results:

• The SVM model achieved high prediction accuracies for TE alloy phase classification, ranging from 77% to 92%.
• Cross-validation confirmed the model's robustness and reliability in differentiating various thermoelectric phases.
• The computational approach proved to be time-efficient compared to traditional experimental and ab initio methods.

Conclusions:

• The developed SVM model offers a computationally efficient method for classifying TE alloy phases.
• This approach can significantly aid in the evaluation and design of novel, high-performance thermoelectric materials.
• The study highlights the potential of machine learning to accelerate materials discovery in the field of thermoelectrics.