Data-Driven Studies of van der Waals Magnetic Heterostructures

Area of Science:

• Materials Science
• Condensed Matter Physics
• Computational Materials Science

Background:

• Magnetic van der Waals (vdW) materials possess unique physical properties with potential to transform the semiconductor industry.
• Existing research on magnetic vdW heterostructures, like MnBi2Te4/Sb2Te3, highlights their promise but requires efficient design strategies.

Purpose of the Study:

• To investigate magnetic vdW heterostructures of the form AiAiiBi4Xi8/Bii4Xii6 using a data-driven framework.
• To explore how combining magnetic and nonmagnetic vdW monolayers influences magnetic properties and band gaps.
• To accelerate the discovery of novel vdW heterostructures for spintronics, optoelectronics, and topological quantum computing.

Main Methods:

• Employed a data-driven framework utilizing density functional theory (DFT) generated data.
• Trained various machine learning (ML) models to predict properties of a vast number of heterostructures.
• Screened 16,431,660 AiAiiBi4Xi8/Bii4Xii6 heterostructures for promising candidates.

Main Results:

• Demonstrated that combining magnetic AiAiiBi4Xi8 and nonmagnetic Bii4Xii6 monolayers can effectively tune magnetic properties and band gaps.
• Successfully predicted properties for a large-scale dataset of heterostructures using ML models.
• Identified promising candidate heterostructures based on ML predictions.

Conclusions:

• The data-driven ML approach significantly accelerates the design and discovery of magnetic vdW heterostructures.
• The investigated heterostructure design offers a pathway to tailor materials for advanced electronic and spintronic applications.
• This work lays the foundation for developing next-generation materials in spintronics, optoelectronics, and topological quantum computing.