A Review of Machine Learning Potentials in the Study of Materials Properties

With the rapid advancement of artificial intelligence technologies and hardware capabilities, AI has gradually become a revolutionary tool driving transformative changes across multiple scientific research domains. In the field of materials science, machine learning methods have played a significant role in high-throughput materials design and property prediction. Over the past decade, machine learning-based approaches for constructing interatomic potentials have been widely applied in the study of material properties, providing crucial support for the theoretical design of novel materials and the in-depth understanding of their underlying microscopic mechanisms. This article reviews the development of machine learning potentials, introduces their fundamental workflows, outlines the principles of mainstream methods and their applications in materials property research, briefly discusses recent progress in emerging universal potential models, and concludes with an analysis of current challenges and future research directions.