Re: Efficiency Measurement

[Jeff Sawyer <74652.2361@CompuServe.COM>]

Hi, All

The question of proper methodology for efficiency measurements has raised its
head again and so I thought I'd further muddy the already turbulent waters by
suggesting a couple of things for further discussion...

There has already been a lot of response to this, Judd Sills covered the issue
of back scatter quite well (and I quite agree with using accurately calibrated
sources of the same isotope under investigation - which admittedly can get quite
expensive) but I just thought I'd add my two cents anyway.

We (Eberline Instruments and in my case particlularly for personnel
contamination monitors) recommend that you use the actual 2 pi emission rate of
the calibration source when calibrating. This side steps the back scatter issue
by not requiring that you know the back scatter factor for the source/geometry.
Most manufacturers of the large area calibration sources can provide this
information - ideally as measured in a windowless proportional counter. (I'm
expecting a quite a few comments back about this one...) In a case of simple
"mathemagics" when calculating the "effective activity" of the source, we simply
double the 2 pi emission rate. This means that we are using a "larger" activity
number (as compared to the source certificate) for the efficiency calculation
which automatically corrects for back scatter. 

The other comment is that by using the activity of the source IF you don't
perform a correction for back scatter, you will always underestimate the
activity present because the measured efficiency (when using just the source
strength with no back scatter correction) will always be higher than that
measured with no back scatter. The scenario here is that you use a very nice,
typically very expensive, very rugged electroplated metal source (with a high
back scatter) to measure the efficiency and then measure contamination on
clothing or a person's hands with a much lower associated back scatter. If you
have an alarm set point based on activity, again in the case of the PCMs, then
you will routinely release activity greater than the alarm set point (compared
to the "ideal situation" where the system was calibrated with a true scatterless
source). The other problem with this approach is that the count times on many
instruments are determined using the measured detector efficiency. If you tell
the PCM that it has a higher efficiency than it actually does, the counting
system uses a shorter count time than it really needs for that alarm activity
which again works against your being able to monitor effectively.

I think someone has already commented on the problem of self absorption in
filter media with "prepared sources". If you were calibrating an instrument that
measured swipes, it would be appropriate to use the same media in the
calibration. It's probably not a good idea to use a "swipe" based calibration
when you're not measuring swipes... The other issue with this type of prepared
source is that of depositing the activity in such a way as to maximize source
uniformity and minimize self-absorption - but that's a whole other topic.

It's almost impossible to get a truely scatterless source. While you could
deposit the activity on a very thin low z film and do the measurements, that
type of source may be considered a bit fragile and hard to use. While metal
sources typically have back scatter factors around 40% for things like Sr/Y-90,
even the reasonably thick plastic backed sources add around 10 to 15% to the
"expected" emission. (Where "expected" was the usual half of the emitted
particles go up, half go down approach). Sure, I realize that the accuracy of
the source is probably around 5% in most cases, but please remember that all of
these errors add up (proper error propogation says its the square root of the
sums of the squares of the relative errors) and it all works against you. It's
not conservative at all! (The numbers for back scatter were dredged from a faint
memory of a conversation at a Health Physics Meeting a few years ago so if
anyone out has measured data that contradicts me, please correct me). There are
several good reference papers on back scatter but I have to admit I didn't take
the time to find 'em before I sent this missive through the ether.

Wes van Pelt suggested a c/d unit (counts per disintegration). I can only say
that it wouldn't correct for back scatter and unless the measurements were being
done on a similar material to the source (for example measurement of stainless
steel work benches and a steel calibration source), there would be an error in
the "wrong" direction - it would not be conservative.

I know the original post was about a survey meter calibration, but the same
arguments hold true for any type of contamination measurement and for most
filter based air monitoring applications as well.

So, any other comments out there?

Jeff Sawyer
Eberline Instruments
505-471-3232 ext. 237
Fax 505-473-9221
E-mail: 74652.2361@compuserve.com
Web Access: www.eberlineinst.com