[Ifeffit] Re: a question about EXAFS fitting result

SCalvin100 at aol.com SCalvin100 at aol.com
Wed Nov 12 16:14:41 CST 2003


My two cents (understanding that we have no specifics on Steven's particular 
sample):

This particular comparison to a standard would make me exceedingly nervous as 
a general rule. Any sample which has a coordination 5/9 of the empirical 
standard is hardly a minor variation on the standard. And since the DW factor is 
likely to correlate highly to coordination number for a single-shell fit, part 
of an apparent change in coordination number might actually be due to a change 
in DW factor.

Again, without knowing the details of the project, or even Bruce's reply as 
to what happened with the cobalt foil, I highly recommend analyzing more than 
just the nearest neighbor when trying to determine coordination number. For 
example, average coordination number for the first shell is theoretically reduced 
in nanoparticles, but due to other effects this is a highly unreliable 
measure of particle size. When the effect of scattering off of neighbors beyond the 
nearest is included, however, Anatoly Frenkel (among others) has demonstrated 
that particle size can often be determined with reasonable accuracy.

--Scott Calvin
Sarah Lawrence College


In a message dated 11/12/03 8:45:45 AM Eastern Standard Time, 
ravel at phys.washington.edu writes:
> I am analyzing cobalt. As I spoke with you before, I got the
> coordination number of cobalt foil as around 9 instead of 12. You
> explained me why it can happen. What I want to know is If we got
> coordination number as 5 for a sample we are interested in and got
> the cobalt foil coordination number as 9, can I proportionally
> correct the value? (Consider 9 as 12 and multiply 5 by the ratio) Or
> just disregard the foil's coordination number? Could you please let
> me know what would be the better way to get the correct value?

Steven,

This is the sort of question for which we created the mailing list, so
I am forwarding it and my brief answer there.  If you are not already
subscribed, you might find it useful to do so.

The simple answer is "probably".  The point of measuring a standard
under identical conditions to the sample is to recognize any
systematic effects in the measurement, the smaple, the detectors, or
the theory.  If all else besides the sample is equal, then you can
probably make the argument you suggest above.  It will be, of course,
incumbant upon you to convince your reader that what you did was
reasonable when you publish.  You most certainly CANNOT cite "Bruce
Ravel, private communication" as your justification!!!

That said, if you want more specific help from the crowd on the
mailing list, you may wish to provide some more specifics about the
details of your problem.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://millenia.cars.aps.anl.gov/pipermail/ifeffit/attachments/20031112/adbb8c4e/attachment.html>


More information about the Ifeffit mailing list