Page 1 of 1

Newbie Question r.e. CsI Detector

Posted: 20 Feb 2019, 18:37
by isoenzyme
I hope someone can answer a newbie question about energy accuracy using the new 2" x 2" Gammaspectacular CsI(Tl) Scintillation Detector that I had after reading the post by Sebastien_billard on 29 Jan 2019. Helpful answers to his question (by luuk) indicated that NaI(Tl) detectors exhibit nonlinear behavior below ca. 100 keV, and that the use of energies below that threshold for calibration might cause problems. Does anyone know if that same nonlinear behavior is expected in a CsI(Tl) detector?

To test it for myself the following experiment was conducted (2"x2" CsI(Tl) detector, GS-USB-PRO driver in a GS-STANDUP-20 using the PRA software):
(1) 120 sec sampling times were collected for three disc sources (1 uCi Cs-137, 1 uCi Co-60, 1 uCi Na-22) and recorded to assign energies to the bins;
(2) the calibration function in the PRA software was used and selected;
(3) A background was collected for 10E6 pulses (86,400 seconds?, the default collection time in the PRA software). This background data was used as the background in the PRA v20 release;
(4) A spectrum was collected of a 2.5 pCi source of Ra-226 I had recently acquired for the default 10E6 pulses. (This source looks rather homemade and not one of the nice Spectrum Techniques sources - I'm a bit leery of touching it with my bare hands and use tweezers.)

During the calibration I used the 32 keV peak from Cs-137 (along with five other peaks [662, 511, 1275, 1173.2 & 1332.5 keV from the sources) and a "Fit linear" calibration mode. The attached image is the spectra collected on the four sources (assuming I can figure out how to attach an image):
Ra-226 Results.png
Spectra of Ra-226, Cs-137, Na-22 and Co-60 collected with the GS-USB-PRO and 2" CsI(Tl) detector. The Ra-226 spectra is background-corrected.
Though there are more peaks than I would expect from a pure Ra-226 source there is a small peak at 36.8 keV (expected 34.8), 178.0 keV (expected 186.2) and (if one uses their imagination) a peak at 452.1 keV (expected 449.4). Peaks at 414.6 and 660.7 keV are missing to my eye. The "close" peaks are shown in the attached image with blue arrows with other, major peaks indicated with the red arrows. Is this as accurate of energy calibration as I can expect or has using the 32 keV peak resulted in a 'wonky' calibration?

Full disclosure(s): I really am a newbie at using both the PRA software and my new hardware so I might be doing something sub-optimal here. Also, there is no lead shielding yet in the GS-STANDUP-20 as I am still making some modifications to it and haven't quite decided where I am going to keep it yet. (Pb shot will be added after a permanent location is selected - I'm not moving that thing after it's full!)

Re: Newbie Question r.e. CsI Detector

Posted: 20 Feb 2019, 18:58
by Steven Sesselmann

Thanks for posting...

I suspect your problem will fix itself if you choose "interpolate" rather than linear, this will interpolate linearily between points, so each point will be spot on.

Linear calibration draws a line of best fit through all points which means none of them will be accurate.

Using the GS-Standup with lead shot you will also get an X-ray peak around 80 keV.


Re: Newbie Question r.e. CsI Detector

Posted: 20 Feb 2019, 19:39
by isoenzyme
Thanks so much for your quick reply. I will give your suggestion a try! Though I agree with your assertion that calibration using a best fit line means none of the points will be accurate I have to believe in Gaussian statistics in this case (as radioactivity is the only natural phenomenon that's known to fit Gaussian statistics perfectly!). Thinking further along those lines the collected data was fit using and omitting the point at 32 keV with the following results:
Bin Calibration.png
Bin Calibration.png (41.03 KiB) Viewed 873 times
Seeing the difference between the linear fits might possibly describe the differences between the published and observed Ra-226 peaks. I'd still welcome any information on known nonlinear behavior areas for the Cesium Iodide detectors if known. Anyone (I'm still curious)?