Exploring the magnetic landscape of easily-exfoliable two-dimensional materials

Magnetic materials often exhibit complex energy landscapes with multiple local minima, each corresponding to a self-consistent electronic structure solution. Finding the global minimum is challenging, and heuristic methods are not always guaranteed to succeed. We apply an automated workflow to systematically explore the energy landscape of 194 magnetic monolayers from the Materials Cloud 2D crystals database and determine their ground-state magnetic order. Our approach enables effective control and sampling of orbital occupation matrices, allowing rapid identification of local minima. We reveal a diverse set of self-consistent collinear metastable states, further enriched by Hubbard-corrected energy functionals with U parameters computed from first principles using linear response theory. We categorize the monolayers by their magnetic ordering and highlight promising candidates for applications.

磁性材料通常呈现出复杂的能量景观，存在多个局部极小值，每个极小值均对应一个自洽的电子结构解。寻获全局极小值极具挑战性，且启发式方法并非总能保证成功。我们采用自动化工作流，系统探究来自材料云（Materials Cloud）二维晶体数据库的194种磁性单层膜的能量景观，并确定其基态磁序。本方法可实现对轨道占据矩阵的有效调控与采样，能够快速识别局部极小值。我们揭示了丰富多样的自洽共线亚稳态；借助基于线性响应理论从第一性原理计算所得U参数构建的哈伯德修正能量泛函，进一步扩充了该类亚稳态的集合。我们依据磁序对这些单层膜进行分类，并遴选出具备应用前景的候选体系。