[Ifeffit] Debye-Waller factors for MS paths

John J. Rehr jjr at leonardo.phys.washington.edu
Wed Jan 14 16:36:22 CST 2004

 I have to agree with Matt Newville, that "The overall answer to
'is there a general method for defining MS  Debye-Waller Factors'
is still no."

 One of the reasons for this is that there seems to be insufficient
information in the experimental spectra, even with temperature dependent
data, to determine all the relevant MS DW factors with reasonable accuracy.

 In my view it is more efficient to fit, not to the DW factors directly,
but instead to the much smaller number of spring constants that implicitly
define all the MS DW factors. Krappe and Rossner have implemented this
with their Bayesian analysis approach [Phys. Rev. B66, 184303 (2002)]
with reasonable results. They used the fast recursion approach of
Poiarkova and myself [J. Synchrotron Rad. 8, 313 (2001)] to caclulate
the MS DW factors from the springs.   The recursion method is implemented
in FEFF8 to 2nd order, but it is reasonably straightforward to extend it,
as Krappe and Rossner have done. An advantage of their Bayesian approach
is that one can take advantage of a priori data, such as ab initio
calculations of spring constants (e.g., as Grant has described or from
other ab initio codes like GAUSSIAN, WASP, etc.) as good starting points
in a fit and for calculations of the spring constants that cannot (due the
limited information in the data) be fit.   Work along these lines is
in progress here at UW, which hopefully, will help with the goal "It sure
would be nice to have something like this available in ifeffit"

  J. Rehr

More information about the Ifeffit mailing list