[Ifeffit] Self absorption questions

Bruce Ravel bravel at bnl.gov
Wed Nov 7 10:07:09 CST 2007

Hi Richard,

I'll take a quick stab at this.

> I'm trying to explain self absorption for our group's guide to EXAFS and I
> need some help understanding some of the finer details of self absorption.
>  In the realm of fluorescence EXAFS, am I correct in that at a given
> energy there is a certain penetration depth of the X-rays that is expected
> (in a thick, dilute sample), but in a thick, concentrated sample, the
> penetration depth is actually less than the expected depth in thick,
> dilute samples (causing the decrease in the amplitude of the oscillation
> peak in Bruce's "wiggle up" of the XAFS oscillation)?  Then, in the
> "wiggle down" of the XAFS oscillation, the penetration depth is actually
> more than what would be expected so the depth of the oscillation valley
> decreases.  As a result, for thick concentrated samples like a metal foil,
> in the "wiggle up," the fluorescence accounts for fewer than expected
> absorbing atoms and in the "wiggle down" the fluorescence accounts for
> more than expected absorbing atoms.

So, why is "wiggling up" attributed to me?  I don't think I wiggle up
any more than anyone else in the XAS community! ;-)

Dani Haskel and I wrote up a decent explanation of this some years
ago.  Look at chapter 2 of this:


When you work out the math, you end up with a term that depends on
mu(E) in the numerator *and* the denominator of the expression.
Bepending on the reltive sizes of the absorption of the absorber and
the matrix, you get more or less attenuation due to the presence of
the mu(E) term in the denominator.

(If someone wanted to replicate the discussion in that link in the
FAQ, that would be a valuable addition.)

> Next, in our amorphous silicates (doped with Ti, V, Cr, W, or Sn), is the
> self absorption effect greater in the lower Z elements (like titanium)
> than in the higher Z elements (like tin)?  Does the high background due to
> the silicon at the lower energies (4.9 keV) in relation to the higher
> energies (29.2 keV) have an effect or is it just an underlying,
> independent issue that is more pronounced with the lower Z elements when
> self absorption occurs?

I think the best way to look at this is to consider what is meant when
you call an XAS sample "thin".  That is, what we consider thin at the
Ti edge is very differnet from thin at the Sn edge.  So, yes,
self-absortion tends to be more of an issue at the Ti edge.  But I am
uncomfortable making a blanket statement because self-absorption
depends on the details.  One can certainly make a Sn edge sample that
will have self-absorption.  Again, it has to do with the relative
amount of absorption of the absorber compared to the matrix.  Follow
the math in that link above, and it should be quite clear.

You didn't ask about this, but it always seems worth repeating.
Self-absorption is an amplitude problem.  You can still measure
distances accurately even in a severely attenuated data set.
Sometimes transmission is not possible and your choices are
fluorescence with attenuation or going home.  Obviously, fluo with
attenuation is the better choice. ;-)


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

 National Institute of Standards and Technology
 Building 535A, Room M7
 c/o Brookhaven National Laboratory
 Upton NY, 11973, USA

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

More information about the Ifeffit mailing list