[Ifeffit] Frustrating monochromator glitch

[Matt Newville]

Hi Mark,

> Dear mailing list - my last round of EXAFS experiments performed
> on the Fe-K edge in fluorescence mode gave data extending nicely
> to 15 in k. One problem was a nasty monochromator glitch at
> approx 13.5 in k which is so great in some cases it causes
> sufficient distortion to the data that even upon deglitching I
> must cut back the data to 13 in order to gain a good fit using
> my models. Firstly I was wondering if someone here could
> describe to me how a monochromator glitch arises (some
> systematic defect in the monochromator crystal or more? - the
> only problem with this explanation is that some data sets seem
> more affected than others and it has not manifested itself as
> badly in previous trips). Also is there any other way to handle
> the removal of a monochromator glitch besides deglitching? My
> supervisor seems to think that there must be a scale factor or
> offset in the I0 monitor counts which could be corrected in
> order to overcome a glitch in the existing data - has anyone
> ever !
>  thought of other ways to salvage data in this way?

The convential mono glitch comes about when another Bragg
reflection from the monochromator becomes allowed.  For example,
with a Si(111) mono, you can get to an angle where a higher order
non-harmonic reflection is in the diffraction condition -- say, a
(311) or (220) or (11 7 7) reflection.  The presence of these
reflections robs intensity from the main reflection, giving less
intensity at that energy.

Glitches seem to come in groups of 2 or 3 (sometimes 4), which
probably corresponds to just grazing throught the Ewald sphere of
a family of reflections with same nominal q. 

Glitches can come from either the first or second crystal of a
double crystal monochromator.  Ideally, you'd like to be able to
rotate the crystals azimuthally -- I don't know of any beamlines
where this is under the user's control!!  But there are a few
thing you can try to reduce the impact of glitches.
   First, try changing the detuning of the 2nd crystal to one 
      side and then the other of the rocking curve.  It might 
      move the glitches, or it might do nothing.
   Next, try moving both the first and second crystal with 
      whatever other motions you have. Again, this might do
      nothing, but you might get lucky.
   Third, find out about how 'feedback' is done, and play with 
      this.  Sometimes, you can stabilize the intensity this way, 
      minimizing the impact of the glitch.

But before you start messing around with the mono, a loss in I0
intensity is supposed to normalize out of EXAFS, so question why
it does not rather than blame the glitch for screwing up your
data.  Here are some questions to ask when glitches don't
normalize away: Are the harmonics really being rejected ?  Are the
ion chambers really working in a linear regime with flat
voltage-intensity curves?  Are the dark currents measured and
being subtracted correctly?  Is the beam being clipped somewhere
or is incompletely defined by slits and any mirrors?  How uniform
is the sample?

Hope that helps,

--Matt