Running FEFF

Where can I get Feff?

A version of Feff -- Feff6l -- is distributed with Ifeffit on all platforms.

More information about Feff, including later versions, is at


Where do I get crystallographic data for use with Atoms?

Here is a short, incomplete list of on-line resources for crystallographic data:

  1. Matt's atoms.inp database:

  2. MinCryst, a useful database hosted at Russia's Institute of Experimental Mineralogy:

  3. The American Mineralogist Crystal Structure Database:

  4. The non-free inorganic crystal structure data (ICSD):

  5. The Crystallography Open Database:

  6. The non-free Inorganic Crystal Structure Database (ICSD) at ILL: (unlicensed users can access 3325 structure subset of 89342 - release 2005-2)

  7. Another ICSD interface:

And, of course, the classic set of volumes edited by R.G.W. Wyckoff is an invaluable resource. Many libraries own a set of these.

Searching Google or Bing can also be useful. For example: here's what Google has to say about "NaCl crystal structure".

Note that recent versions of Atoms and Artemis can import CIF files directly.

Please add to this list!


Why does Feff give warnings about atoms being very close together?

Feff will issue an error message like

 atoms       40      44
   40  2.24156E+00  3.88224E+00  8.43300E-01
   44  2.24156E+00  3.88276E+00  8.43300E-01
 Run continues in case you really meant it.

in the situation where two positions in the atom list in feff.inp are less than 1.75 bohr (about 0.93 Angstrom) apart.

A very common cause of this is an error in the input information to Atoms used to generate the feff.inp file. The general rule in Atoms is that it considers 5 digits of precision after the decimal place.

This means that if a crystallographic position is at, say,

  1/3 2/3 0

then it is a mistake to type

  0.333 0.667 0

into Atoms or into the atoms.inp file. Instead you should enter

  0.33333 0.66667 0

The old version of Atoms, which was written in Fortran, could not accept fractions as crystallographic coordinates, but the modern version can. Thus if you use Artemis, the Web interface to Atoms, or the recent command line version of Atoms, you can actually use

  1/3 2/3 0

and not worry about precision issues.


How can I analyze polarized data with Feff?

See this this discussion from the mailing list and Bruce's answer

Here is the description of the POLARIZATION and ELLIPTICITY keywords in the Feff6 document:

And they are described on pages 18 and 19 of The Feff8 document (pdf)

Chapter 3 of this document (which is somewhat dated but still useful) written by Bruce also includes a discussion of polarization in Feff-based analysis.

All of the concepts described in those references fit easily into Artemis, although you will certainly have to edit the feff input data after running Atoms and before running Feff within Artemis. By default, the feff input data generated in Artemis does not include polarization effects. But once you understand all the stuff above, that will be trivial to do in Artemis.


What happened to TkAtoms?

The Atoms GUI, TkAtoms, was old and written rather sloppily. It proved too hard for Bruce to maintain, so he retired it. The most important features of TkAtoms have been incorporated into Artemis, thus tightly binding its functionality into the process of analyzing EXAFS data.

You can use Artemis as a feff.inp file generator. Just fire up Artemis and read in an atoms.inp or CIF file. Alternately, choose "New Atoms page" from the Theory menu, then "Blank page". That will give you an interface which is similar to the old TkAtoms for writing an atoms.inp from scratch. To save your feff.inp file, choose 'Write special output" from the Theory menu, then choose "feff6" or "feff8" from the submenu.

Alternately, you can use the command line version of Atoms. That will read an atoms.inp or CIF file and write out a feff.inp file, all without a GUI.

Another option is to use WebAtoms


What is the amplitude or "curved wave importance factor" of a path in Feff?

From the Feff6 document:

The curved wave importance ratios are the importance of a particular
path relative to the most important path encountered so far in the
calculation.  If the first path is the most important in the problem,
all the importance factors will be expressed as a fraction of that path.
However, if the third path considered is the most important, and the first
path is the next most important, path 1 will have a factor of 100%, path 2
will be a fraction of the first path's importance, path 3 will have an
importance of 100% (since it is now the most important path), and subsequent
paths will be expressed as a percent of path 3's importance.

The definition of this parameter is unchange in later versions of Feff, including Feff8.

The amplitude reported in the files.dat or list.dat files or in the Feff interpretation page in Artemis is, therefore, a rough estimate of the spectral weight of each path in relation to the largest path in the calculation. If you were to dig through the source code, you would find the calculation is quite crude, indeed. It is a very sparse trapezoidal integration of the magnitude of the calculated chi(k). A few caveats:

  1. Since the calculation is so crude, it may be unreliable in certain cases.

  2. For a material with very low symmetry, thus very small path degeneracies, the largest path won't be very big. Some paths with relatively large curved wave importance factors won't, in fact, be very large paths. The amplitude is a ratio. Ratios can be big when the numerator is big or when the denominator is small.

  3. The amplitudes are calculated without considering the effects of S02 or sigma2. Some paths with large amplitudes might become quite small when physically realistic fitting parameters are applied.


How do I add a new FAQ entry?

Edit this section, use copy and paste! Use a format like this, using the same level of header (i.e. the same number of = signs) and keeping the link back to the contents at the top of the page:

== How do I add a new FAQ entry? ==
     Edit this section, use copy and paste!
[#contents Contents]


FAQ/RunningFeff (last edited 2010-05-10 14:46:53 by BruceRavel)