April 28 2003

MSD for 2624 atoms at 700K, 900K and 1000K

The equilibrated configurations went through further continuous constant temperature MD for 17.2ns to collect 500 configurations for MSD calculation.

Since the steps still exist, it should not only because of the discontinuous simulations. In this MSD calculations we used only unovelaping independent data. When the number of MSD data changed (e.g. from 2 to 1 when the interval is half of the total number of configurations), the poor statistics may introduce a big step.

Calculations with overlaping intervals (minimum 10 values) got rid of the steps and gave smoother curves.

The results titled Shift are the independent calculations with random start points assigned to each atom (if possible).

Dependent	Independent	Shift

Surface layer and sub-surface layer MSD

Why the average MSD of surface layer is smaller than sub-surface layer? Does it show that the atoms on sub-surface layer move faster than surface layer?

Let N₀ be the number of atoms on the surface layer (=860), and N₁ be that on the sub-surface layer (=600), and we redraw the average MSD of surface layer by a factor of N₀ / N₁ for T=700K and 900K. The following plots show that this problem was caused by averaging over all atoms, part of which do not diffuse.