Accelerating Heusler Alloy Discovery Using Machine Learning

Heusler alloys, discovered at the end of the nineteenth century, have become exciting materials in the twenty-first century due to their fascinating properties, such as half-metallicity, martensitic transformation, and ferromagnetism. These properties make Heusler alloys potential candidates for various applications, including spintronics and thermoelectric devices. In this study, we aim to discover new Heusler alloys using state-of-the-art machine learning approaches to facilitate the discovery and applications of these materials for potential application.

Over a thousand Heusler alloys and their variants have been identified in the past century, and the search for more is ongoing. To support this effort, a comprehensive dataset of X2YZ Heusler alloys has been compiled. This dataset is one of the few comprehensive databases accessible for researchers in this field, offering a complementary source of data alongside the Open Quantum Materials Datasets (OQMD). In this work, we use machine learning-assisted tools to facilitate the discovery of new Heusler alloys. We perform supervised learning on the dataset using various machine learning models, including Kernel Ridge Regression (KRR), Support Vector Machines (SVMs), Random Forest, and Gaussian Process Regression. Among these models, the KRR is found to be the most accurate for our dataset. After carefully validating the model, we apply it to predict new Heusler alloys.

Our study will provide fundamental insights into the computational and experimental synthesis of new Heusler alloys. By leveraging advanced machine learning techniques, we can accelerate the discovery process, leading to faster development of materials with tailored properties. This has significant implications for the fields of spintronics, thermoelectric devices, and beyond, potentially revolutionizing how we design and utilize new materials for various technological applications.