A data-driven strategy to explore magnetism in 2D materials

0
February 21, 2022

(News from Nanowerk) The discovery of 2D magnetism not only opened a new avenue for potential breakthrough spintronics and valleytronics applications, but also raised fundamental questions about understanding and predicting ferromagnetic and antiferromagnetic ordering in 2D compounds.

The description of the magnetic order in 2D materials is complex. More than 50 years ago, based on Heisenberg’s isotropic model, Mermin and Wagner demonstrated that, unlike bulk compounds, in 2D materials the long-range magnetic order is suppressed by thermal fluctuations.

This suggests that 2D materials can exhibit magnetic ordering only in the presence of large magneto-crystalline anisotropy, and therefore magnetism in 2D and bulk materials is assumed to have different physical mechanisms.

In recent years, a lot of effort has been invested in the systematic prediction of two-dimensional magnetic systems, mainly coming from density functional theory (DFT) calculations. These predictions are based on a direct approach which involves computing all possible candidates. However, these trial and error processes can be time-consuming and costly.

In a recent article by Applied materials and ACS interfaces (“Machine Learning Study of the Magnetic Ordering in 2D Materials”), the researchers propose a data-driven strategy to explore magnetism in 2D materials: unlike the commonly used direct approach, they aim to explore the material- attribute to provide the simplest correlation between magnetic order and characteristics, for example, composition, crystal symmetry, and atomic properties.

Illustration of the proposed machine learning strategy to determine and predict the existence of magnetism (step I) in a given 2D compound and its specific magnetic order (step II). The starting point of this strategy is the grouping of species and atomic structures. Atomic species are inherently grouped according to the organization of the periodic table, while an ML model is made to group structures according to possible composition, crystal point group and local distortions. This cluster is then used to identify the tendency of a specific element and structure to form a magnetic (blue) or non-magnetic (red) 2D structure. If the specific combination of atoms and structures for the 2D compound tends to be magnetic, we identify its position in a map of materials separating the antiferromagnetic order from the ferromagnetic order. These positions are calculated using the atomic characteristics of the constituent atoms. Finally, this ML strategy allows us to predict and select compounds for specific applications in electronics and spintronics. (Reproduced with permission from the American Chemical Society) (click image to enlarge)

Specifically, the authors train machine learning algorithms using a database of 2D magnetic materials to obtain descriptors capable of classifying materials as nonmagnetic, ferromagnetic, or antiferromagnetic.

Their strategy is divided into two main steps, namely, 1) first develop a random forest model to separate magnetic compounds from non-magnetic compounds based on trends in crystal structure and atomic composition, and 2) on the basis of the sure independence screening and scattering operator (SISSO) method, to find a mathematical model (i.e. a function of atomic characteristics) that uses composition to provide a map of materials with regions defined for 2D ferromagnetic and antiferromagnetic materials.

The researchers note that the random forest algorithm used in their work was able to differentiate magnetic compounds from non-magnetic compounds with high accuracy. Non-magnetic compounds were predicted with an accuracy of around 96%, while for magnetic compounds this accuracy was around 85%.

In conclusion, the authors write that their findings may be useful in the search for new 2D magnetic materials, expanding the possibilities of this expanding and exciting field to compounds that may be more stable and easier to synthesize than the candidates observed so far. here.

Share.

About Author

Comments are closed.