5.8 Final Considerations

As has already been emphasized, the key element in molecular dynamics is the interaction potential. Even if a considerable amount of work is spent to establish a potential function that is able to reproduce all known experimental results within acceptable accuracy, there always remains an uncertainty, since small deviations in the potential may lead to substantial differences in the MD results. But now, of all things, the use of different cut-off radii for the potential function which is a necessary and common technique of MD calculations leads to such deviations.

For asymptotically vanishing functions like that for krypton there is no problem: to be on the sure side one just has to increase the cut-off distance to a value where the potential is negligible. But for metallic bonds the according pseudo potentials show rather far-reaching oscillations that are only negligible for long distances. On the one hand, to consider all these oscillations would lead to ineffective MD algorithms. On the other hand, it is questionable at all whether the real nuclear interactions are this far-reaching. In the case of the Schommers potential for aluminium, the use of a cut-off radius in the range of the six nearest atomic neighbor distance (7.35 at 50 K) has delivered all the important characteristics and effects in good agreement with experiments.

Another point to discuss is the practical relevance of the observed cluster state transitions. Since standard mass production techniques for nanostructured materials work at rather high temperatures [FZJ, 1998]...