Grain boundaries (GB) represent planar defects in a crystal structure where two or more grains with different orientations are connected. Such an imperfection in the lattice often comes with an extra space that usually serves as a location where impurities tend to segregate. Since impurity decorated as well as clean GBs have substantial impact on mechanical properties, they have been extensively studied for decades. Ab initio calculations and molecular dynamics are the most used theoretical methods to study mechanical properties of GBs at atomistic scale. However, molecular dynamics is limited by reliability of available interatomic potentials and the ab initio methods can be performed only for relatively small computational cells and at absolute zero temperature. To overcome these limitations, one can use the ab initio molecular dynamics with the machine learning (ML) as it is implemented in the current version of the Vienna Ab initio Simulation Package (VASP) which makes it possible to perform reliable molecular dynamics simulations with thousands of atoms in the simulation cell at finite temperatures.

In this work, we employed the aforementioned tools to study segregation of Sn atom at the Σ5 (310) GB in bcc iron. Segregation energies were obtained using small cells with approx. 200 atoms and also for larger cells containing thousands of atoms and compared with results of static ab initio calculations. This comparison serves as a benchmark of the force field dataset received from the VASP machine learning. The results of the segregation energies clearly show that Sn atom tends to segregate at the GB as a substitutional impurity.