Kinetics of the formation of complex phases in steel
Within this project the growth of complex phases is investigated using molecular dynamics and adaptive kinetic Monte Carlo simulations. One of the focus points is the development of a polyhedron analysis to characterize the different structures during the simulations.
Currently available interatomic potentials are too simple to capture the complex interactions and ab initio methods as density-functional theory are computationally to expensive to treat the required system size. We will therefore employ suitable tight-binding models and bond-order potentials that are developed within ICAMS and in collaboration with the MML, Oxford. Applying atomistic simulations techniques to study the formation and growth of complex phases can provide detailed insight into the underlying mechanisms and help to find ways to prevent or control the formation of TCP phases. Since the kinetics of TCP phases involves so-called rare events, standard Molecular Dynamics simulations are not suitable to reach the necessary time scales. To circumvent this problem we use adaptive kinetic Monte Carlo. This coarse-grained model will be employed in large-scale simulations to investigate the mechanism of the transformation (displacive or diffusive) and to characterize suitable order parameters and transition paths. Another important aspect is the influence of the (local) composition of the material on the transformation path and the role of defects and environmental conditions such as temperature and pressure. The main setup to examine in this project is an interface between σ-FeCr and bcc-FeCr.
An important step in analysing the simulations is to identify the different phases in a structure. The rather complex structures of TCP phases are not easily characterised, but the topology of these phases determines that each atomic site in a TCP phase is enclosed by one of four so called Frank-Kasper polyhedra (containing 12, 14, 15 or 16 atoms) as coordination polyhedron.
Another property is the distinct formation of skeletons of these polyhedra in different phases, which makes it possible to determine the structures directly from the coordination polyhedra. For this we have developed algorithms to automatically find coordination polyhedra around each atom for a given structure and characterise them by comparing to a polyhedron database. We create fingerprints of the arrangement of coordination polyhedra around each atom and which enables us to determine the crystal structure by comparing this fingerprint to fingerprints from previously analysed structure prototypes.
Using adaptive kinetic Monte Carlo simulations and with the help of the structure analysis, we also simulate and track vacancy diffusion in copper bulk, interfaces and boundaries in cooperation with the group of Graeme Henkelman at the University of Texas, Austin.