Are all atoms on the regular hcp lattice sites? In that case, most likely, bulk Ru FEFF calculation will properly describe theoretical EXAFS signal (provided that you vary coordination number from your geometry and define it to be equal to that value when you construct your EXAFS signal). You will need to find S02 independently from experiment.

If the atoms are distorted from the lattice sites, but the bond lengths historgam is reasonably symmetric and narrow, you can run FEFF calculation for a mean distance (by constructing it from bulk Ru, for example, with the model FEFF distance being the mean distance of your histogram). Then you can assign Debye Waller factor term that has two components: the static (from your histogram standard deviation) and dynamic (that you can theoretically estimate from the Einstein temperature and the cluster temperature.

If the atoms are distorted asymmetrically (e.g., your histogram is strongly non-Gaussian), there are two possibilities:

1) Keep single FEFF calculation but correct for the geometry by using a method described in Yevick and Frenkel, Phys. Rev. B 2010. This approach assumes that the interatomic distances do not change more than 10% (you can check that) and thus FEFF-calculated amplitudes and phases are transferrable. In that case, you will need to add thermal factor (See above) but you have to be careful since the Einstein model would not so obviously work for the strongly distorted surface bonds.

2) If the histogram is more broad than that, you will need to calculate FEFF on each atom and add 80 FEFF calculations. Thermal factor will still be a problem. All in all, a MD calculation is the best approach, and that is what, I think, Bruce Ravel is working on.

Anatoly