Abstract
It has been known experimentally for some time that Al3 Ti is a powerful nucleant for the solidification of aluminum from the melt; however, a full microscopic understanding is still lacking. To develop this understanding, we have performed molecular dynamics simulations of the nucleation and early stages of growth using published embedded atom method potentials for Al-Ti, but modified by us to stabilize the D 022 structure. We discover that Al3 Ti can indeed be very effective in promoting the growth of solid Al but the manner in which growth takes place depends sensitively on the surface on which the Al nucleates. In particular, complete growth of solid Al from the liquid on the (001) and (110) surfaces of Al3 Ti occurs at a lower temperature than on the (112) surface. This anisotropy agrees with observations in previous experiments. We explain this observation in terms of interfacial energies. On the preferential (111) surface of Al the solid-liquid interfacial energy is highest while the solid-vacuum energy is lowest. Our simulations also show that the extent of ordering taking place in liquid Al close to the Al 3 Ti substrate above the melting point correlates well with the effectiveness of the substrate as a nucleant below the melting temperature: this could provide a computationally efficient scheme to identify good nucleants. © 2010 The American Physical Society.
Original language | English (US) |
---|---|
Journal | Physical Review B |
Volume | 82 |
Issue number | 14 |
DOIs | |
State | Published - Oct 27 2010 |
Externally published | Yes |
Bibliographical note
KAUST Repository Item: Exported on 2020-10-01Acknowledgements: The authors would like to acknowledge funding from King Abdullah University of Science and Technology (KAUST), and support from Thomas Young Centre (TYC) at Imperial College London. The authors also acknowledge many useful discussions both with colleagues at Imperial College London, Oak Ridge National Laboratory, and Ford Research and Advanced Engineering Laboratory and, especially Stefano Angioletti-Uberti, Mike Finnis, James R. Morris, and Mei Li and John Allison.
This publication acknowledges KAUST support, but has no KAUST affiliated authors.