Measuring kinetic coefficients by molecular dynamics simulation of zone melting

Abstract

Molecular dynamics simulations are performed to measure the kinetic coefficient at the solid-liquid interface in pure gold. Results are obtained for the (111), (100) and (110) orientations. Both Au(100) and Au(110) are in reasonable agreement with the law proposed for collision-limited growth. For Au(111), stacking fault domains form, as first reported by Burke, Broughton and Gilmer [J. Chem. Phys. 89, 1030(1988]. The consequence on the kinetics of this interface is dramatic: the measured kinetic coefficient is three times smaller than that predicted by collision-limited growth. Finally, crystallization and melting are found to be always asymmetrical but here again the effect is much more pronounced for the (111) orientation.

Atomistic and continuum modeling of dendritic solidification

“…Anisotropies in the BGJ method can be found by repeating the procedure for different solid-liquid interface normals. A related ''forced-velocity'' method has also been developed recently by Celestini and Debierre [129]. Since the original BGJ study a number of other methods have been proposed to extract m from MD simulation as summarized in a recent review [128].…”

“…The effect also leads to a sizeable dependence of the [1 1 1] growth rate on the periodic dimensions parallel to the solid-liquid interface in MD simulations, with smaller systems being found to solidify at higher rates. Although there is evidence that stacking fault drag plays a role in the case of the low stacking fault energy EAM Au system [54,129], no direct evidence for this effect has been found in the most recent MD studies by Sun, Asta and Hoyt, to be published, for the higher stacking fault energy EAM Ni system for which the computed ratio m 100 =m 111 ¼ 1:49 AE 0:26 is much larger than the prediction based on the ratio of interplanar spacings. As will be discussed in the next section, the DFT-based model of Mikheev and Chernov [126] yields predictions for the crystalline anisotropy in m that appear to be in rather close agreement with recent simulation results for m in each of the lowindex interface orientations.…”

Kinetic coefficient of steps at the Si(111) crystal-melt interface from molecular dynamics simulations

“…Since the pioneering work of Broughton et al 1 much of the most detailed information concerning the intrinsic properties of crystal-melt interfaces has been derived from atomic-scale molecular dynamics ͑MD͒ and Monte Carlo simulations. Such simulations have determined the magnitudes and anisotropies of crystal-melt interfacial free energies [2][3][4][5][6][7][8][9][10] and kinetic coefficients 7,[11][12][13][14][15][16][17] for numerous elemental systems, modeled with Lennard-Jones, hard-sphere, and repulsive power-law potentials, as well as embedded atom method models for metals. For these systems, which crystallize in simple fcc, bcc, and hcp crystal structures, crystal-melt interfaces are atomically rough, with properties that are relatively weakly anisotropic.…”

Towards Multiscale Modeling of Metals via Embedded Particle Computer Simulation

“…The generalized method is not significantly more complicated to use than pair potentials. It had been used in a number of recent works; see, e.g., [12,13,14,15,16,17,18]. Solution of the Schrödinger equation yields the electron density established by a given potential, and the energy is a functional of that potential.…”