Abstract
Interface response functions (IRFs) for amorphous and crystalline forms of Si have been determined for several empirical atomic-scale models using Molecular Dynamics and compared to available experimental results fitted to a Wilson-Frenkel equation form. Stillinger–Weber (SW), the environment-dependent intermolecular potential (EDIP), and a version of the modified embedded atom method (MEAM) models were found to produce unacceptable representations of the IRFs of both solid phases; they were either unable to predict the amorphous melting point and/or the maximum solidification velocity. The best of these models was judged to be the SW potential, known to produce a very accurate IRF for crystalline silicon. Increasing the strength of the three-body term by up to 25% above that of the original SW potential improves the prediction of the melting characteristics of the amorphous phase. Above this limit, liquid phase properties are impaired. The resultant IRFs provide an important backdrop to understand the kinetics of explosive crystallization (EC) processes, as we shall show in comparison to recent experimental data on the EC of amorphous Ge. [A. Chojnacka and M.O. Thompson, in Growth, Evolution and Properties of Surfaces, Thin Films and Self-Organized Structures, edited by S.C. Moss, D.B. Poker, D. Ila, (Mat. Res. Soc. Symp. Proc. 648, Warrendale, PA 2001) p. P11.12.1–8]. We also provide evidence that homogeneous melting within the bulk of the amorphous material competes with heterogeneous melting at the planar amorphous/liquid interface.
Acknowledgements
The authors would like to thank Professor Michael O. Thompson and Dr Aleksandra Chojnacka for thier help in interpreting the simulation results in the light of their EC experiments and Professor Michael O. Thompson for reading this manuscript. The authors would also like to thank the National Science Foundation for funding this research through a KDI award (9980100) and the Cornell Center for Materials Research for providing much of the computational resources necessary for this computationally intensive study.
Notes
The percentage of solid values in samples containing both liquid and amorphous bulk samples are somewhat temperature-dependent, resulting in the ranges of 60–90% for amorphous and 0–45% for liquid samples. The crystal phase shows some temperature dependence, but over a far smaller range (99–100%).