[Ifeffit] experimental resolution

[Matt Newville]

Dear Irina,

On Wed, 25 Sep 2002, Irina Pirog wrote:
>
> Dear Dr. M. Newville. I'm a researcher of X-Ray department at
> Physical Research institute of Rostov State university. In our
> laboratory we use UWXAFS code treating experimental data. I will be
> very gratitude to you, if you would answer me: how is experimental
> resolution is taken into account in UWXAFS.
>
> Sincerely yours
>
> Irina Pirog.

This is an excellent question that I'm sure other people may have, so
I hope it's OK with you that I am responding to the Ifeffit mailing
list, which we've set up to share such questions about EXAFS and
Feff, Feffit, Ifeffit, etc.

By itself, the UWXAFS programs (atoms, autobk, and feffit) don't know
what the experimental resolution is.  In principle, the experimental
resolution is important only to the extent that it is different
between two measurements or between a measurement and a theoretical
calculation.  For the UWXAFS programs, the only place this becomes
important is when comparing Feff calculations to data in feffit.

Generally speaking, Feff includes terms that effect the resolution
and all other loss terms in the mean-free-path, lambda.  Feff
includes contributions from the core-hole lifetime and some many-body
electron loss terms in lambda, but it does not explicitly put in an
experimental resolution.  But Feff usually has a little too much loss
(ie, lambda is a little smaller than it should be), which turns out
to be fine for experiments , with a resolution of ~1eV or so.

But sometimes you know what the experimental resolution is and
want to include it in the analysis.  To account for that,
Feffit has an energy-broadening Path Parameter, ei, that gives
an additional broadening term in the EXAFS equation through
  p^2 = (k_real + i/lambda)^2 +   ei * 2m / hbar^2

where k_real and lambda are taken from the feff0001.dat file for
each path (though they are the same for any one Feff calculation),
and then, the k and lambda terms used in the EXAFS equation are:
  k      = Re[sqrt(p)]
  lambda = 1/Im[sqrt(p)]

To use this, you would have a feffit.inp with something like this:

 Path   1  feff0001.dat
 id     1  path 1
 s02    1  amp_factor
 ei     1  1.0            # add 1 eV broadening to Feff calc.
 ...

I mention the s02 parameter here because ei will be very highly
correlated with s02 (and sigma2).  In fact, I'd guess that much of
the variation in the observed s02 parameter for different
measurements is actually due to differences in energy resolution.


Finally, you may be interested in using the greatly revised versions
of the UWXAFS programs, including graphical user interfaces for data
processing and fitting:
   http://cars.uchicago.edu/ifeffit

Ifeffit also uses a Path Parameter called ei with the same effect,
though the syntax would be slightly different:

  path(1, file=feff0001.dat, s02=amp_factor,  ei= 1.)

Anyway, I hope that answers the question,

--Matt

|= Matthew Newville        mailto:newville at cars.uchicago.edu
|= GSECARS, Bldg 434A      voice: (630) 252-0431 / 1713
|= Argonne Natl Lab        fax:   (630) 252-0436
|= 9700 South Cass Ave     http://cars9.uchicago.edu/~newville/
|= Argonne, IL 60439 USA