Calibration of Ludlum 44-3 with 241Am failing

[ColoRad-o]

LiuEtAl20190001.png (134.91 KiB) Viewed 5268 times

Bad241AmPks.png (90.38 KiB) Viewed 5268 times

Howdy all,

I’ve been checking out a used eBay Ludlum model 44-3 low-energy gamma detector using a bias voltage of 700V [range: 500-1200 typical]. Many thanks to Steven for pointing out that all I had to do was replace the Ludlum “series C” connector by a standard female SHV connector (51494-2) in order to use my existing SHV cable. I could use the nut and the solder washer that came with the Ludlum; this ‘jack’ is the same diameter and is threaded the same way—I just had to re-solder the two wires inside. The detector is 1 inch diameter and 1 mm thick.

I later did a ‘plateau curve’ run: 700V is a good spot. Quick spectra for 241Am in AU taken with the Ludlum and with the GS 2020, both NaI(Tl), are attached; the little blue inset is a linear scale uncalibrated curve. The two peaks shown in blue would be assigned to energies of 26.34 and 59.54 keV. Spectra on gammaspectacular all agree that the higher peak is much brighter. I saw nothing unusual in the audio pulse shapes or the interval histogram, or pulse widths. The dead time is about 223 [processing LaTex...]\mu sec and the count rate is kinda high at 1569 cps. [700V is not far from the the flattest region, around 650V, nor do the dead times change very much at all.]

As is usual at the end of the detector range (often around 100 AU), the count rate plummets, as can be seen for the Ludlum beyond roughly 30 AU. This happens for much higher energies for the GS 2020. The nominal energy range for the Ludlum 44-3 is 10-60 keV, so rapid falloff past the second peak (IF it’s about 60 keV) is about as expected. What I did NOT expect is that (IF I’ve identified the peaks correctly) the relative heights and widths appear reversed—the higher AU peak is too low and too wide. Could this be neptunium X-rays? If so, what happened to the 26.3 keV peak? The entry window has density*thickness = 18.4 mg/cm^2; I’m not sure how this compares with the GS 2020 and what effect this might have.

Can anyone suggest (i) a reason why the Ludlum doesn’t even get the relative peak heights and widths correct, (ii) how to improve the resolution of the Ludlum? Any obvious blunders I should be aware of—light leaks, bad crystal, etc?

FYI: I’ve also attached an NaI(Tl) (St Gobain) spectrum from a fairly recent survey of a variety of halides by Liu et al. They conclude “Compared with other four scintillation crystals, NaI(Tl) scintillation crystal has the best resolution for X-rays below 25 keV, and the energy resolutions for 13.9 keV, 17.8 keV and 20.8 keV are 17.68% and 14.16%, and 13.42%, which make NaI(Tl) be the preferred scintillation crystal for low-energy X-ray detection, applied in X-ray imaging, port security and space X-ray detection.” https://iopscience.iop.org/article/10.1 ... 012083/pdf

Thanks!!
Dave W

[Sesselmann]

Dave,

Thanks for posting these results..

What you are seeing there is more or less expected. I actually had an old Ludlum 44-3 but sold it just recently.

So I guess you realize that the 44-3 is an x-ray detector, so it has a thin 1mm crystal with either beryllium or aluminium window (cant remember).

The GS-2020-NAL has a 1mm aluminium housing plus the crystal encapsulation 0.5mm, which means that x-rays and low energy gamma will be attenuated (absorbed in the aluminium). This becomes very obvious below about 30 keV.

The 44-3 on the other hand only has a thin foil over the crystal, so it will see many more counts at the low end of your spectrum, this is consistent with your findings.

Apart from that it looks like the resolution of the 44-3 is pretty ordinary, so maybe you can do a spectrum with Cs137 just to benchmark it.

Steven

[ColoRad-o]

Thanks for the reply, Steven. Yes, I'd known about the (nominally 10-60 keV) energy range of the 44-3.

I had imagined that because it had a narrow energy range (nominally 10-60 keV,) , the Ludlum would do a better job on that range. Alas no. It appears that this probe is MOSTLY used for count rates (not gamma spectrometry). I had some samples which gave a smooth, large bump at low energies and hoped the Ludlum would resolve these. Oh well--it was a not very expensive experiment.

I attach the count plateau curves for the Ludlum and the GS2020 for those who might be curious. The optimal bias voltage (at least on the basis of insensitivity of the count rate) would be 650 V for the Ludlum. The GS 2020 has an astoundingly large flat region, from roughly 675 to 825 V. So the choice of bias voltage probably should be made to make the calibration curves (see below) as flat as possible, to permit simple interpolation, but well above the 'knee'.

[The png named CalibLinesByBias is for the GS 2020 and are for a set of features from a 232Th lantern mantle spectrum, ranging as you can see from low energies up to 2614 keV. Some appreciation for how much AU values can change over the scan of bias voltages can be seen from comparing two distinct runs at 700 V, but generally speaking the lower the bias voltage, the more linear are the 'calibration curves'.]

Plateaux.png (77.41 KiB) Viewed 5240 times

[Sesselmann]

Dave,

The count rate plateau is a test of the PMT and that looks normal. Typical bias voltages are usually in the 650 to 750 range and that agrees well with your experiment.

My guess is that your thin 1mm crystral is damaged, most likely the encapsulation has failed and the crystal has become wet.

I think a simple solution would be to buy a new encapsulated crystal, if you like I can get you a quote for a replacement.

Steven

[ColoRad-o]

Steven,

Wow--thanks for the diagnosis of the wet crystal. There is no substitute for experience!

I'm having a hard time finding low-energy gamma spectra taken with a NaI(Tl) crystal size 1 inch x 1 mm thick or thereabouts (25mmx 1 mm), (the size of the existing crystal). Thus I don't have a clear feeling for what sort of resolution it would have in the range of, say, 10-100 keV. I assume I'd need to clean off the PMT window and use the Mystical Goop (I'll look it up on GS) others have used to attach/seal the PMT to the crystal, then wrap it all up. Probably would be fun!

Sure, I'd be delighted to get a quote on a replacement crystal. I looked around casually but have had a hard time finding skinny thickness ones. Thanks, Steven, as always.

[James Luck]

Dave,

Thanks for posting these results..

What you are seeing there is more or less expected. I actually had an old Ludlum 44-3 but sold it just recently.

So I guess you realize that the 44-3 is an x-ray detector, so it has a thin 1mm crystal with either beryllium or aluminium window (cant remember).

The GS-2020-NAL has a 1mm aluminum housing plus the crystal encapsulation 0.5mm, which means that x-rays and low energy gamma will be attenuated (absorbed in the aluminum). This becomes very obvious below about 30 keV.

The 44-3 on the other hand only has a thin foil over the crystal, so it will see many more counts at the low end of your spectrum, this is consistent with your findings.

Apart from that it looks like the resolution of the 44-3 is pretty ordinary, so maybe you can do a spectrum with Cs137 just to benchmark it.

Steven

Steven i don't mean to be a prude but more intending to correct the information on the detector.

The Ludlum 44-3 is a beta/gamma probe, designed to be most efficient at detecting and counting beta particles along with low energy gammas emitted by thallium and iodine in hospital environments.
The crystals casing has a bottom made from thin aluminum ( cant recall the thickness).
If i recall correctly the crystal was made thicker on one of Ludlum's revisions of the 44-3. From 1mm to 1.5mm, it is NaI, on very rare occasions you can come across ones with CsI, usually those are closer to 0.5mm thick.

Overall they work rather well at counting beta but also at low energy gamma counting during contamination sweeping, along with testing X-ray shielding for gaps.
I could see why spectrum results are rather low.

i have four 44-3 detectors, although one i rebuilt with BC-412 plastic and somehow managed to get the detector to be pretty close to one with NaI. counting tests only though, and this could be a topic for another post.

-James