Crystal Lattice Structure Prediction of Fe-Based Compounds by a Molecular Dynamics Method

Fe-based hard magnetic compounds such as α″-Fe16N2Fe16N2 have been the potential candidates for rare-earth (RE)-free permanent magnets. Fe-based soft magneticSoft magnetic materialsmaterialsMaterials such as MinnealloyMinnealloy, α″-Fe16(NC)2, have been the potential candidates for critical-element-free magnetic materialsMagnetic materials. Their intrinsic magnetic properties such as giant saturation magnetization and high or low magnetic anisotropyAnisotropy energy (MAE) are the results of their unique crystalline and electronic structuresStructure. Their electronic and magnetic structuresStructure are largely dependent on their crystal latticeLatticesstructuresStructure. First-principles methods can be used to study the electronic structuresStructure of these materialsMaterials and to calculate the ground-state latticeLattices parameters. However, when examining a new composition, first-principles methods typically use existing known latticeLatticesstructuresStructure that may or may not represent the global minimum of the true ground state. In this paper, we present another approach to address this challengeChallenges, to use a molecular dynamics (MDMolecular dynamics (MD))-based simulated annealing method during energy minimization to predict the unit crystal latticeLatticesstructureStructure. In this method, a building-block latticeLatticesstructureStructure is formed first as the initial point that is closely related to the predicted structureStructure with a matched spatial atomic ratio. Then a series of local energy minima (EM) searches are evaluated within confined latticeLattices moving volumes that are continuously fractionally divided from the previous one. A global minimum state can be evaluated from these local minima as it is the closest ground state. Stress conditions can also be applied during this EM processProcess to simulate a textured latticeLattices growth at experimental condition. We show that this method can be computationally efficient and provide great insights in predicting the crystalline latticeLatticesstructuresStructure of Fe-based compounds such as α″-Fe16N2Fe16N2, α″-Fe16C2, and-Fe16CxN2-x α″-Fe16CxN2−x (x ∈ (0,2)).