[Ifeffit] amplitude values too high (>>100%)

[Bruce Ravel]

On Wednesday 11 August 2010 05:41:41 am Edgar Macias wrote:
> I am looking for some help on trying to understand what is wrong with our
> data. We are very impressed with the high values of amplitude (up to
> 390.24)  that we got working with Artemis and we know are wrong. Hope
> somebody could help us and take a look into these files, we will appreciate
> it.

Hi Edgar,

There is much to be concerned about in your fitting model, but the
concern that prompted yor post to the list is not one of them.

It would seem that the version of Feff that you are using did not
rescale its importance factor correctly when it ran into the situation
that the second path has a higher importance than the first.  The
importance factor is not used for any real purpose either in Feff or
in Ifeffit.  Artemis displays as an aid to understanding the paths,
but does not use it in any manner in the fit.  So I think you are
worried about something that isn't a problem.

So ... what should you be worried about?  Certainly a couple of
things:

1. I don't understand the purpose of importing the exact same data
   twice and then assigning different Feff calculations to each copy.
   Is you intent to model a mixed phase system?  That seems likely
   given that you defined an "x" parameter then did things like
   "x*amp" and "(1-x)*amp" in the paths.

   If you are trying to model a mixed phase system, you need to import
   your data *once* and assign multiple Feff calculations to it.  The
   data and path list should look something like this (my little
   sketch will probably make sense only by using fixed-width font):

        - Data
          - Feff #1
            Path 1
            Path 2
          - Feff #2
            Path 1
            Path 2

2. Perhaps you did have a reason to import the data twice.  If so, it
   is not clear why you set the k-range to a sensible [2:13.3] in the
   second copy, but to [1:6] in the first copy.

3. I'd be sceptical that you could meaningfully measure independent E0
   parameters for the two phases.  Maybe, but you should look at that
   with a critical eye.

If you have not spent some time with the information about using
Artemis that can be found at http://xafs.org , then you probably
should do so.

HTH,
B




-- 

 Bruce Ravel  ------------------------------------ bravel at bnl.gov

 National Institute of Standards and Technology
 Synchrotron Methods Group at NSLS --- Beamlines U7A, X24A, X23A2
 Building 535A
 Upton NY, 11973

 My homepage:    http://xafs.org/BruceRavel
 EXAFS software: http://cars9.uchicago.edu/~ravel/software/exafs/