Magnetic MXene: A Machine-Learning Model With Small Data

MXenes, comprising atomically thin layers of transition metal nitrides, carbides, and carbonitrides, exhibit properties that are not found in their corresponding bulk materials. Interestingly, because of the presence of transition metal, MXenes may also provide the candidate materials for observing low-dimensional magnetism. This can be of interest to various applications such as data storage, electromagnetic interference shielding, and spintronic devices. Here, we focus on the magnetic MXenes, which are only a few in number out of known MXenes. We propose machine-learning models to predict the magnetic moments of the MXenes and to classify the MXenes based on their chemical stability. Using these models, we propose four new chemically stable MXene materials having a potentially high magnetic moment.

I. Introduction

Two-dimensional (2-D) nitrides and carbides of transition metals called MXene are gaining popularity due to their tremendous applications in energy storage [1], [2], as catalysts [3], [4] Click or tap here to enter text.photonics [5], and many more fields. The MXenes are hexagonal layered nitrides, carbides, or carbonitrides of transition metals with the chemical formula where range from 1 to 3 and represents the early transition metal, is C or N, are the surface terminations like OH, F, O, or Cl. These 2-D materials are produced by etching out the layers from the MAX phase, which is a family of hexagonal layered ternary transition metal nitrides, carbides, and carbonitrides with a composition of where stands for a group element (Al, Si, Sn, and In). Therefore, a large number of chemical combinations are possible, and after the discovery of the first MXene (Ti₃C₂T_x) in 2011, more than 30 compositions were published [6]. Interestingly, the presence of 3d transition metals makes them a potential candidate for low-dimensional spintronics [7]. So far, among 2-D MXenes, Cr₂C, Cr₂N, Ta₃C₂, and Cr₃C₂ are reported to be ferromagnets [8], [9], Ti₃C₂, and Ti₃N₂ are antiferromagnets and Ti₂C and Ti₂N are nearly half-metallic ferromagnets [10]. An intrinsic ferromagnetism is reported in 2-D Cr₃C₂ MXene having a magnetic moment of 3.9 per formula unit using the Kohn–Sham density functional theory (DFT) [8]. Though many experimental and theoretical studies are reported to discover and understand the properties of MXenes, a machine learning (ML) approach may be the most inexpensive approach for this purpose and can speed up the process of discovering new MXenes.