A group of scientists headed by Rensselaer Polytechnic Institute’s Trevor David Rhone, Assistant Professor in the Department of Physics, Applied Physics, and Astronomy, has discovered novel van der Waals (vdW) magnets with the use of advanced tools in artificial intelligence (AI).
Trevor David Rhone. Image Credit: Rensselaer Polytechnic Institute
Specifically, the researchers found transition metal halide vdW materials with large magnetic moments that are anticipated to be chemically stable using semi-supervised learning. Such 2D (two-dimensional) vdW magnets have possible applications in spintronics, data storage, and quantum computing.
Rhone specializes in using materials informatics to find novel materials with unforeseen characteristics that progress science and technology. Materials informatics, an emerging field of study, is the combination of materials science and AI. Recently, his team’s recent study was featured on the cover of Advanced Theory and Simulations.
2D materials can be as thin as a single atom. They were found only in 2004 and have been the topic of great scientific interest due to their unforeseen properties. 2D magnets are important as their long-range magnetic ordering continues when they are thinned down to one or more layers. This is due to magnetic anisotropy.
The interplay with this low dimensionality and magnetic anisotropy could produce exotic spin degrees of freedom, like spin textures that could be employed in the development of quantum computing architectures. The complete range of electronic properties can be spanned by 2D magnets. These magnets can be made use of energy-efficient and high-performance devices.
Rhone and his group integrated high-throughput density functional theory (DFT) calculations, to find the vdW materials’ characteristics, with AI to apply a type of machine learning known as semi-supervised learning. Semi-supervised learning utilizes a combination of unlabeled and labeled information to find patterns in data and make predictions. Semi-supervised learning alleviates a key difficulty in machine learning—the shortage of labeled data.
Using AI saves time and money. The typical materials discovery process requires expensive simulations on a supercomputer that can take months. Lab experiments can take even longer and can be more expensive. An AI approach has the potential to speed up the materials discovery process.
Trevor David Rhone, Assistant Professor, Department of Physics, Applied Physics, and Astronomy, Rensselaer Polytechnic Institute
With the use of an initial subset of 700 DFT calculations on a supercomputer, an AI model was trained that could make predictions about the characteristics of several thousands of material candidates on a laptop in milliseconds. Then, the team found potential candidate vdW materials with low formation energy and huge magnetic moment. Low formation energy is a sign of chemical stability, a significant need for synthesizing the material in a lab and consequent industrial applications.
Our framework can easily be applied to explore materials with different crystal structures, as well. Mixed crystal structure prototypes, such as a data set of both transition metal halides and transition metal trichalcogenides, can also be explored with this framework.
Trevor David Rhone, Assistant Professor, Department of Physics, Applied Physics, and Astronomy, Rensselaer Polytechnic Institute
Dr. Rhone’s application of AI to the field of materials science continues to produce exciting results. He has not only accelerated our understanding of 2D materials that have novel properties, but his findings and methods are likely to contribute to new quantum computing technologies.
Curt Breneman, Dean, Rensselaer’s School of Science.
Rhone was joined in the study by Romakanta Bhattarai and Haralambos Gavras of Renselaer; Bethany Lusch and Misha Salim of Argonne National Laboratory; Marios Mattheakis, Daniel T. Larson, and Efthimios Kaxiras of Harvard University; and Yoshiharu Krockenberger of NTT Basic Research Laboratories.
Journal Reference:
Rhone, Trevor David, et al. (2023) “Artificial Intelligence Guided Studies of van Der Waals Magnets.” Advanced Theory and Simulations. doi.org/10.1002/adts.202300019.