Solidus and Liquidus Determination in Multi-Component Systems Using Hybrid Monte-Carlo-Molecular-Dynamics
John Copley, Princeton University
Molecular dynamics can provide valuable insight into the behavior of systems that are difficult to assess experimentally but is limited by computational cost to short timescales that may not allow for slow processes like solid state diffusion. Diffusion is especially important in the simulation of melting behavior of multi-component solutions, where the equilibrium melting point is described not as a single combination of pressure and temperature, as it is for unary systems, but by the solidus and liquidus curves. These curves, which describe the compositions of the solid and liquid that are in equilibrium at a given pressure, temperature combination, are difficult to investigate with conventional molecular dynamics approaches as partitioning of solute into the solid can be slow, as it requires long-range, solid-state diffusion.
To accelerate the effective diffusion rate, we perform hybrid Monte-Carlo Molecular-Dynamics simulations in which a Monte-Carlo swap takes the place of diffusion by translocating pairs of atoms of different species. We demonstrate that this method successfully recovers the solidus and liquidus of both fully miscible (Cu-Ni) and eutectic (Au-Si) phase diagrams. The method is then applied to study the melting behavior of Fe-Ni under Earth’s core conditions, using a Generalized Embedded Atom Potential. Under pressures from 135-360 GPa (approximately equal to the pressures at the core-mantle boundary and Earth’s center, respectively), we find the Fe-Ni phase diagram to be peritectic, with added Ni resulting in a melting point elevation and, at Ni concentrations several times larger than expected for the Earth’s core, a change from the hexagonal close packed structure preferred by Fe at core pressures, to face centered cubic. Our results also indicate narrow solidus-liquidus gaps at these high pressures, due to a strong preference for solubility from configurational entropy.