Shield design, geometry, mass, etc.

[Cosmic]

Hi,

I want to build a shield for gamma spectrometry of low activity samples with a GS device and a 2" NaI probe.
A ton of lead would probably do the job, but that would be way too expensive and the floor in my appartment might not be strong enough.
I would like to find some kind of optimum between cost/weight and minimum reasonable detection level.

Assuming the same amount of lead is used, what would give better results:
- thick shielding, space for a small sample vs. thin shielding, but space for a larger sample/Marinelli beaker. Is a Marinelli geometry desirable?
- probe and sample fully surrounded by shielding vs. "open end", no shielding at the bottom of the probe

For the latter I made an experiment with a lead "tube" 200mm long with 20mm wall thickness and 20mm lead cover on one side, the results are:

Background: 171 cps
0mm, probe just outside the shield: 144 cps
50mm of probe in shield: 76 cps
100mm: 29 cps
150mm: 20 cps
200mm (total length of shield available): 17 cps

It seems to me, that the "open end" design would save a lot of lead or at least the shielding at the bottom of the probe could be much thinner.

I'm thinking about a design like this
grey - probe, green - lead, red - sample/Marinelli beaker
space for the probe is 120mm diameter, 100mm height - the standard Marinelli beakers I found are larger but I would 3D print my own ones.
The drawing shows 30mm thick lead (20mm at the bottom) totalling at 70kg, maybe it could be reduced to 20mm and 50kg. The empty space at the bottom could be filled with concrete.

For the Cu/Sn lininig I would go with the calculations by wraith_oz in this post viewtopic.php?f=10&t=38&p=139#p100

Btw. I had this idea before, but then found this commercially available shielding:
http://www.interphysix.com/index.php?id ... er=product
Details can be seen in the download/pdf

What do you think?

[cicastol]

Hi Stefan,
if you wanna go for low activity samples route a proper shield is a must, 4Pi Marinelli geometry is essential and at least 5cm of lead all around is needed, i've found also that a "cap" over the detector improve the background/shielded ratio counts as well.
I've made a modular shield of pure lead rings, maximum weight ring is the base (around 28kg) intermediate rings are near 15kg, outside diameter of 255mm inside 150mm for Marinelli placement, during times i added an internal block of round lead for an hefty total base thickness of 10,5cm of lead.
With such configuration i can attain a 50:1 background reduction with my Luuk 3"x3" and even more reduction if i apply a top cap over the divider.

[Cosmic]

Hi Ciro,

wow that looks like lol (lots of lead) - 300-400kg?
Thank you for your opinion - I didn't expect the shield to be more expensive than sensor + electronics.
I like the idea of a modular design with lead rings.
Maybe I'll start with small rings and an "open end", then add more rings at the bottom later and eventuelly enclose this shield with larger rings.

[Sesselmann]

Stefan,

The good news is that most of the background noise is low energy, just look at a typical background spectrum and see how it decays exponentially towards the higher energy channels.

This means that there are diminishing returns for adding more lead, even 1 mm of lead all the way around is going to reduce the count rate significantly, so my advise is make it as thick and as heavy as you can afford, and consider adding a layer of tin and copper.

[cicastol]

Hi Ciro,

wow that looks like lol (lots of lead) - 300-400kg?

After doing some math my shield is not so much LOL (i thout was more heavy), maybe in the 180 kg range or less with the heavier part of 28kg, despite this relatively low amount of lead is highly effective .
My first shield was a simple 2/2,5cm wrapped lead with end cap, it was good for a 2"x2" detector giving a 20:1 background ratio but moving on with a 3"x3" it had turned to be insufficient.
In my test tin and copper remove Pb XRF lines at the expense of increased scatter all around the spectrum, so imho adding lining should be evaluated based on the task.