ICAMS / Interdisciplinary Centre for Advanced Materials Simulation

Atomistic modelling of the kinetics of solid-solid phase transformations

In solidification theory the properties of solid-liquid interfaces such as the interface energy, interface mobility, anisotropy etc., are of great interest since they determine to a large extend the nucleation and growth processes. Because solid-liquid interface energies are very difficult to measure experimentally accurate values are widely unknown even for rather simple systems. Large uncertainties in the measured values arise from the possible occurrence of heterogeneous nucleation from impurities. Therefore simulation methods can aid in providing reliable theoretical values for solid-liquid interface energy. But usually solid-liquid phase transformations require the system to overcome a rather high energy barrier for nucleation making these transitions between liquid and solid states extremely rare and therefore very inefficient to be studied using regular molecular dynamics.

Schematic representation of the free energy surface at the melting temperature vs. a one-dimensional collective variable. The atomic structures for the two stable states together with the one containing the liquid-solid interface are shown along with the free energy surface.

Here we employ transition path sampling (TPS) to study solid-liquid phase transformations in different metallic systems. TPS samples the path space, explores transition state region and generate an ensemble of reactive trajectories that contain the full dynamical information of the sampled process, in our case the phase transformation. Using a reweighting scheme for the path ensemble we can extracted interfacial properties such as interface free energies, nucleation and formation of interface. Also dynamical properties like anisotropic interface mobilities as a function of temperature and interface orientation can be calculated. The results from our simulations concerning the interfacial properties can then be used as input parameters in mesoscale models.

┬ź back