[Ifeffit] Case of "nonequivalent multiple atomic sites of absorbing atoms"

Matt Newville newville at cars.uchicago.edu
Wed Jul 28 06:46:24 CDT 2010


Dear Rana,

Delta E0 is a shift in the Energy origin of k, that is where k=0.  For
the experimental data, this was chosen empirically (and so somewhat
arbitrarily) from the measured data.  For the FEFF calculation, it was
chosen at the calculated absorption threshold (lowest available
electronic level), but is known to not be perfect at this, especially
in Feff6.

Though E0 is meant to be the absorption threshold, it is more
important that the FEFF calculation and measured data agree.  Because
of this, Delta E0 is best left as an adjustable parameter in the
analysis.

It is generally found that a single Delta E0 parameter is needed *per
measured data set and per FEFF calculation*, because measured data can
be shifted in energy and because each calculation can have a different
definition of E0.  It is possible to read the FEFF outputs and try to
figure out how different these energies are and so set (not allow to
float) the different E0 values.  I believe this can work well, but it
is not automated, and so not often used.  In addition, usually one
uses a different Delta E0 for each data set, but well-aligned data
that re-use a FEFF calculation (say, data taken on the same system at
different external conditions) can often get by with sharing a single
Delta E0 parameter.

S02 is meant to express the incomplete overlap of the "passive
electrons" in the absorbing atom.  Removing the deep core electron can
allow the other (L, M, N, O) electrons to relax slightly from their
ground state.  That relaxation reduces the overlap of the "initial"
and "final" states of the transition that gives rise to XAFS, and so
reduces the full effect.  FEFF6 does not do a very good job of
estimating this effect.

In practice, S02 can also mask other "amplitude issues", some
theoretical (the inelastic losses that fold into the mean-free-path
are not perfect) and some experimental (it is easier to mis-measure
the amplitude than the phase of an oscillation).  For XAFS, these
experimental problems all tend to spoil the amplitude of the
meaurement more than the phase: pin-holes in the sample, slight
misalignment of the sample, over-absorption for fluorescence
measurements, energy bandwidth of the incident beam and harmonic
content of the incident beam.

The rule of thumb is that it is OK to use the same S02 for all samples
of a particular absorbing atom.  Many would add "measured at the same
beamline" to that rule, so that effects of energy bandwidth and
harmonic content are compensated.

Cheers,

--Matt



More information about the Ifeffit mailing list