Facile Discovery of Pristine MoS₂ Magnetic Ordering via Machine Learning Prediction and DFT Validation

Understanding the magnetic ordering in molybdenum disulfide (MoS2) is crucial for the advancement of novel magnetic devices, given the fascinating physical and chemical properties of MoS2. Our approach combines a preliminary prediction from machine learning (ML) models, trained on a limited data set of only 390 data points, with rigorous validation through density functional theory (DFT) calculations. This hybrid method provides valuable insights into the underlying mechanisms driving the predictions. The analysis facilely reveals a complex relationship between crystal structure and net magnetization in pristine MoS2, demonstrating that an increased cell volume, achieved through plane expansion in a trigonal crystal structure, is associated with the emergence of magnetization in MoS2. Specifically, expanding the cell volume weakens atomic interactions, thereby facilitating the formation of sufficient energy states while simultaneously reducing electron saturation near the Fermi level. This efficient ML-DFT hybrid methodology offers a streamlined path for tailored material design, as exemplified by our exploration of pristine MoS2 for spintronic applications.