Date: Thu, 29 SEP 94 13:48:24 EST
Subject: Re: EDS icing
From: WALCKSD@ml.wpafb.af.mil

Just a note on icing of EDS detectors.

Several years ago, when I was at UAB, we had a Kevex detector on our SEM
with a turret window system ( Be, thin, and no window). It developed an
icing problem with the Be window. Over a period of about 6 months, you
could see a significant oxygen peak develop with the Be window in place.
We could see this using an aluminum sample. Just prior to the point
where this peak was noticeable, fitting the X-ray background became just
about impossible. When we warmed the detector and cooled it back down,
the peak was gone and the fitting of the background could once again be
easily accomplished. We did not see a noticeable change in detector
performance from the warming. You could monitor the ice formation on
the detector by the changes in the parameters in the program to match
the background curve, especially the low energy regions. When it became
too difficult to match, we just warmed up the detector. I believe that
it was eventually fixed by Kevex.

Incidently, this gave me the basis of a bonus question on an exam that I
asked in an SEM Xray microanalysis course. I don't know how close
some of my assumptions are, but it brought a lot of principles of
X-ray/specimen interaction into play. For your enjoyment, here is the
question:

"We believe that a layer of ice formed on the face of the detector
crystal. When a spectrum from a pure aluminum sample was acquired with
the Be window in place, a small oxygen peak was seen. Asuume that 25%
of the ionization events were due to Al Ka X-rays and that three
ionization events occur for each absorbed Al X-ray. The number of
oxygen counts were 550 an the number of Al counts were 140,000. The
density of ice is .92 g/cm3. The mass absorption coefficient for Al Ka
X-rays in oxygen is 1500 cm2/g. the fluorescence yield for oxygen is
0.002. Estimate the thickness of the ice on the crystal. State
assumptions."

-Scott