Data driven studies of magnetic ground state and transition temperature in two-dimensional magnets

Abstract

The magnetic characteristics of two dimensional (2D) van der Waals (vdW) magnets are governed by a delicate balance among various factors, posing a significant challenge in the design of novel 2D magnets. In this work, we employ a data-driven approach to investigate the magnetic properties of monolayers composed A₂B₂X₆, building upon the well-established ferromagnetic Cr₂Ge₂Te₆. Here, using random forest and gradient lift regression algorithms, we perform a high-throughput scan of 696 materials from a database to classify ferromagnetic and antiferromagnetic compounds based on their magnetic ground state. First principles-based computations and Monte Carlo simulations, followed by Heisenberg model-based, are employed to estimate the transition temperature ($T_c$) of these magnets. The classification accuracy reaches approximately 84%, while the regression accuracy is around 81%. Our results not only enrich the family of 2D magnets and present high-temperature ferromagnetic materials but also offer insights into the realization of high temperature magnets. This work paves the way for accelerating the discovery of novel magnetic compounds with high transition temperatures for spintronic applications.