Lots of things change with wavelength and absorption is one of the more
prominent ones, but in the end its all just path lengths and Beer's law,
so the exact answer depends on the geometry. For example, if you are
talking about a flat detector 200 mm from the crystal, then 2 A spots
with 1 A radiation will be attenuated 1% more than spots near the beam
stop, but with 1.54 A radiation, these same 2 A spots will be attenuated
11% more than low-angle spots. So, yes, you will probably see this in
scaling.
A good place to look up things like this is here at LBL:
http://henke.lbl.gov/optical_constants/
This will even tell you the transmittance if you put in the path length.
The equation for air transmittance vs 2theta is:
exp(-mu_air*detector_distance/cos(2theta)). You can get mu_air from the
above website, but you may also notice this equation is NOT the equation
for a B factor: exp(-2*B*(sin(theta)/lambda)^2). As a result, scaling
programs can suck up much of this systematic error in a B factor, but
not all of it. Some processing programs can account for air
attenuation, but things get more complicated when you consider that the
phosphor in the detector will follow the opposite trend and steeper
incidence angles will actually make spots get brighter (because the
phosphor is effectively deeper and stops more photons). The latter is
seldom corrected for by processing programs, largely because it is hard
to know exactly how thick the phosphor is. This is why scala, etc. has
a fancy across-the-detector-face scale factor. I recommend scaling your
two data sets together in one scala run with this option on. It might
also be a good idea to include your working set of Fcalcs as a
"reference" to help guide the resolution dependence of the scaling.
-James Holton
MAD Scientist
James Stroud wrote:
Hello All,
I did diffraction with the exact same sample at 1 Å and at 1.54 Å and
noticed that the higher-resolution data from the 1 Å source is more
intense relative to the 1.54 Å source after normalizing cumulative
intensity between the two data sets. Is this effect from air
absorption or something else? If this is a known phenomenon, is there
some way to calculate what the wavelength dependent correction might
be (as a function of resolution) from the experimental conditions or
is the only hope to empirically determine the scaling parameters? The
sample is protein and it is shot in air at about 20 °C.
Thanks in advance for any input.
James
