As I recall, your setup down there is similar to mine (ALS 8.3.1), so
perhaps I can be of some assistance.

There are only two things that can cause changes in energy: d and
theta.  This is because lambda=2*d*sin(theta).   If the change is
entirely due to thermal expansion of the mono crystal, then the
fractional change in wavelength or energy (3 eV is 0.03% of 12657 eV)
must equal the fractional change in cell edge.  0.03% may not sound like
much, but the thermal coefficient of linear expansion for silicon is
only 2.6e-6 / K, so this change represents a shift in crystal
temperature of 121 C.  I doubt your cooling water is getting that hot.

This leaves theta (or rather 2*theta) as the culprit, and for this
change in energy you are looking for a tilt of 0.005 degrees or 0.1
millirads.  This is a change of 100 microns over 1 meter, which
represents a temperature difference of 8 C if you have two steel support
legs (expansion coefficient: 12e-6 /K) 1 meter tall and 1 meter apart.
This is not too hard to imagine.  If your supports are made of aluminum
(TCE = 23.5e-6 /K), then a 3 C change in one leg of a square will make a
0.1 mrad tilt.  Good news is, this change in tilt must be happening
either inside (or underneath) the mono or somewhere upbeam from it.
Focus your attention there.

As Liz pointed out, temperature fluctuations can always be a headache,
and it is important to remember that your x-ray beam is not the only
source of heat you need to think about.  Motors generate heat (such as
the ones in your mono).  I too saw a lot of drift in the second crystal
of my Khozu monochromator when we first got it, but this turned out to
be due to heating of the "Z2" motor.  This motor only moves when you
change energy and at first I thought it was just bad luck that we were
seeing big drift during every MAD data set.  Nowadays we only move this
motor if the beam is in danger of walking off the edge of the crystal
(~2 keV moves), and the mono is remarkably stable now.

Another things you might not expect is computers.  I recently narrowed
down a large (1 mm) beam drift problem on ALS 8.3.1 to tomography
reconstruction jobs at 8.3.2.  They had bought a new computer and
happened to accidentally point its exhaust fans at the downbeam support
leg of my M2 mirror tank.  The extra CPU heat from the tomography jobs
was enough to heat the support by ~10 C, which moves the beam at the
sample position (10 meters away) by 1 mm.  Took me a while to figure
that one out.

   Since your fluctuations are not correlated to ring current, I am
willing to bet that heat from the x-ray beam is not your problem.  More
likely a different source of radiation (with lower photon energy, but
higher total power) is to blame.  Your energy drifts seem to start at
sunrise and then turn around at sunset?  Although it is tempting to just
tell your facility that they need to buy a better air conditioner, a
more cost-effective solution will probably be to figure out exactly what
is moving and do something to fix it.  It is still possible that you are
doing something every morning that starts the drift (such as moving a
lot of motors to optimize the beam).

Measuring the temperature of the optics support legs is always useful,
but you can also get a lot of mileage of of a laser pointer, a mirror, a
piece of paper and a pen.  Rig the laser pointer to a remote power
source (you don't want to have to touch it to turn it on), attach it to
something "stable" (like the shield wall) and bounce the laser off a
mirror mounted on your mono tank, and back to a piece of paper near the
laser pointer.  Use the pen to mark where the beam falls at different
points of the day.  You can catch very small angles this way, and you
might get lucky.

Good luck,

-James Holton
MAD Scientist


Lucas Sanfelici wrote:
> Hello all!
>
> Does someone have experience in minimize energy instabilities in
> beamlines?
>
> MX2, our new beamline devoted to MX experiments, are facing problems
> with energy drifts. As far as we could notice, theses drifts are results
> of the contribution from several sources - possibly electron beam
> movements, heating of optical elements, etc...
>
> LNLS is a 2nd generation machine with 4 straight sections available for
> insertion devices. MX2 is a 2T wiggler-based beamline and produces a
> peak flux of 10^11 photons/s.
>
> What I'd like to know, before start performing calculations, how far
> should I expect the heating of a non-cooled 2nd crystal affects energy?
> Does someone know cases of a few eVs drifts?
>
> Thanks in advance and regards,
>
> Lucas Sanfelici
> Physicist
>
> Brazilian Synchrotron Ligth Source- LNLS (www.lnls.br)
> Diagnostics Group
> PO Box 6192 Postal Code 13083-970
> Campinas-SP Brazil
> Phone: +55-19-3512-1153/1152  Fax: +55-19-3512-1006
> E-mail: [log in to unmask]
>