[Ifeffit] How to assign measure unit to XAFS data

[Matt Newville]

Hi Matteo,


On Thu, May 12, 2016 at 7:11 PM, Matteo Busi <basebush at gmail.com> wrote:

> Hi Bruce, Matt, and  mailing list readers,
>
> I have one question regarding the XAFS data. I have collected absorption
> coefficient in fluorescence regime of a CuSO4 penta hydrate sample(however
> my question can be generalized to any compound).
> Through Athena I then obtained the x\mu(eV) spectra as well as \chi(k) and
> \chi(R). I am wondering how can I assign a unit to the x\mu values. I would
> like to have it in the cm^-1 or cm^2 kg^-3 unit so than I can match it with
> tabulated data for the Copper absorption coefficient and proceed with my
> studies.
>


The analysis tools treat "xmu" as a dimensionless value.   To get values
for the mu in cm^-1 or grm/cm^2, you'll have to convert the measured values
used for XAFS analysis (say, readings from ion chambers) into actual X-ray
intensities, and try to account for anything else in the beam path that
will attenuate the X-rays.  For measurements made in transmission, this can
be done.   In fact, for most transmission measurements, the *change* in
-ln(I0 / I1) is usually a very good measure of the change in absorption of
the sample, so that if the observed jump is a value of 2, the sample really
did change its absorption by Delta_xmu = 2.    But the absolute values for
xmu are definitely not in meaningful units.  Typically, the photo-current
from the ion chamber or diode is converted to a voltage and then summed
over some time period in a data acquisition system.   You'd have to know
the details of that system to convert the "I0" and "I1" numbers into actual
beam intensities.

For fluorescence measurements, the situation is worse. You'd need a very
good measure of the solid angle of your detector, fluorescence efficiency,
and the attenuation of the fluorescence getting from the sample to the
detector.  Depending on the detector type, you'd also need a good
assessment of the backgrounds.   It's probably not impossible, but I think
it would not be easy unless that was the actual point of the experiment and
you made many careful measurements of the detection system.    If you're
asking this question here, you'd didn't make those measurements.



>
> In case a treatment is needed for the normalized \mu , that would also
> suit my situation.
>
> My idea was to divide by the thickness of the sample but I don't find a
> reasonable justificiation for this.
>


If you're measuring in fluorescence, dividing by the thickness of the
sample is not what you want.


>
> I really hope I made the question clear to any reader this time, and if
> that is not the case I apologyze in advance.
>
>

Cheers,

--Matt
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://millenia.cars.aps.anl.gov/pipermail/ifeffit/attachments/20160513/ddd0427f/attachment.html>