Tuesday, July 21, 2020

Scintillation Gamma Spectroscopy: Setup

My DIY Gamma Spectroscopy setup is finally completed! 

The system is comprised of 6 major components:

1. Spectroscopy-Grade Scintillation Detector - this is the single most important component and the "heart" of the system.  

The detector is an RF and magnetically (Mu-Metal) shielded and light-tight assembly of a scintillation crystal or scintillation plastic (for example NaI(Tl) - Thallium doped Sodium Iodide , CsI(Tl), CeBr3, LaBr3, Bicron plastic, etc) coupled via optical interface (optical grease) with a Photomultiplier Tube (PMT) and a Dynode Voltage Divider for the PMT bias is usually mounted on the PMT's socket. 

In my setup, I use a Scionix Holland 38B57/1.5M-E1 detector with NaI(Tl) crystal 1.5" x 2.25", Hamamatsu R980 10-stage PMT and a dynode voltage divider I custom built with 2M for R or total impedance of 24M.

My modified detector.
 I took apart a commercial Sodium Iodide (NaI(Tl) Scionix Holland) detector from an Exploranium GR-135 and reworked the whole Dynode Voltage divider circuit. I also installed a BNC connector on the cap of the housing - the original detector just had cables coming from a rubber grommet on the side of the cylinder.
These are older but quite nice, all-around, spectroscopy-grade detectors that can be found used as part of surplus sales and are not outrageously expensive as most such detectors - mine came from a "Border Patrol contract" GR-135 unit.

I was lucky to find a unit with ~7.4% resolution at 662 keV - while this is typical of these detectors, if the device was abused, crystal was cracked or moisture got inside, performance will greatly deteriorate or it will be rendered completely useless for Spectroscopy.

These detectors are very expensive and fragile devices and should be protected at all times (!!!). The  housing must be sealed air-tight as ambient moisture will destroy the crystal, ambient daylight will destroy the PMT when operational and tiny light leaks will spoil the spectrum,  mechanical shock can damage both, the crystal and the glass envelope PMT (a vacuum tube in essence), over-voltage will shorten the PMT's life, etc.
 I am storing my detector for example in a well-padded Pelican case with some Silicagel to absorb moisture.
One should watch for the cosmetic condition (as everything is sealed anyways) when buying a used, untested detector as this usually can show evidence of rough handling. The bottom, aluminum part where the crystal is encapsulated should not have any dings (!) or deep scratches. Considering how fragile these devices are, the risk of buying a damaged / dropped detector will be still present. 

The original detector as just stripped off a GR-135 unit can not be used directly with GS-USB-PRO - the Dynode Voltage Divider must be completely reworked. 
The role of the PMT voltage-divider is to provide HV for the tube's Cathode-to-Anode bias and all of the dynodes in-between, as each dynode receives higher voltage than the preceding one, thus progressively increasing the gain with each consecutive dynode stage (Hamamatsu R980 PMT has 10 dynode stages). This process, achieves multiplication of the electrons in an "avalanche effect" until all electrons are collected by the anode. The overall multiplication factor follows the level of the High Voltage.
The original divider is a transistorized circuit and has a number of active components - it is designed to work with the firmware in GR-135 where the linearity is compensated for in the software and tailored for a specific energy range. Such divider will provide very poor linearity if used directly, in a conventional way - a "classic" PMT divider is needed.

This is the original circuit on the back of the PMT. Cables are fed thru a sealed rubber grommet on the side. Easiest way to rework the divider is to completely remove the board. As there is no socket for the PMT and the PMT itself is equipped with wire pins not suitable for a socket, all 12 pins must be de-soldered from the PCB. The PMT is a glass envelope vacuum tube and caution must be exercised when de-soldering.  If the outer housing is not removed, installing the board back onto the PMT wires can be quite difficult but not impossible - it took me about 20 min to re-align the stiff wires with the PCB holes as I didn't want to remove the main detector housing - it is glued and sealed to the crystal housing and I just didn't like the idea of disturbing it.

I removed all of the old components and had to figure out way how to apply the new divider circuit using the old PCB, traces and pads. 

The board was cleaned from solder and old flux before soldering the new voltage divider components - all of them are SMDs. The pin numbering on this picture is wrong!  This was a temp numbering as if using 14pin socket - this PMT has 12 pins - 10 stage dynodes, Anode and Cathode.

The reworked Dynode Voltage divider, using metal-film, high-voltage SMD resistors and HV ceramic 3kV capacitors. R=2M, 2R=4M for a total of 24M. All resistors are within 1% tolerance and actually better - I purchased extra resistors and used a LCR bridge to select and match 10 resistors as close as possible to the designated value and to each-other.
I was able to use the old pads and traces with only one additional small jumper-wire. BNC connector was installed on the cap, sealed and connected via small diameter short coax to the PCB. The PMT's Mu-metal shielding and the detector housing were all grounded together.
The resistors values in the voltage divider are specific for use with GS - the overall impedance is too low if used as a simple scintillation probe with a battery operated Ludlum meter for example - in such case 120M divider is expected with R=10M and 2R=20M correspondingly.

 The PMT's dynode voltage divider - a pretty standard circuit with the exceptions of the values in my case R is 2M and R1 is 2R or 4M. 
If this detector is modified for use as a standard scintillation probe (with a survey meter like Ludlum for example), R should be 10M with 2R = 20M - just like on the schematics. Higher impedance is needed in such case for these battery operated meters as their voltage supply is not as "stiff".

2. Detector Driver - This device is the electrical interface between the Detector and the MCA - it usually contains a High-Voltage Bias power supply for the PMT, coupling/decoupling circuit, an adjustable gain pre-amplifier, ADC and a Sound Card USB chip (in my specific case).
I use Gamma Spectacular GS-USB-PRO (www.gammaspectacular.com) as a Detector Driver and this thing is absolutely fantastic - I love it and highly recommend it! The unit connects to the USB port and powers the detector as well as it provides the analog (external sound card needed) and digital (built-in sound card in GS) interface to the MCA software. It is a highly configurable, flexible and adjustable device. GS-USB-PRO also provides a very nice stable HV bias supply /w voltmeter and allows for super-easy adjustment of the high-voltage to the PMT - range is 0 to 2000V. 
I run my detector at 650V - it is a good compromise - lower voltage gives you better linearity but low gain, while high-voltage affects the linearization but gives more gain to pick up weak gamma-emissions (and noise of course). The exact optimal voltage should be determined experimentally because it is tied to the particular condition of the crystal, PMT and voltage divider as well as to the targeted energies range.

The GS-USB-PRO Detector Driver - compact and attractive design. SHV and BNC connectors on the front panel alongside the USB port, analog audio jack and all trimmer adjustments for the high-voltage, pulse shape and volume.  On the bottom of the housing there are 2 switches for selecting detector wiring mode and the analog jack mode. The analog jack can serve as an input to connect two units for coincidence measurements. My GS unit is the latest v3.2 hardware version with all modifications to accommodate the Schmitt trigger daughterboard for use with Neutron Detectors.

There is a bright digital voltmeter on the top side of the unit. The trimmer pot on the front is adjustable with a small screwdriver for a range of 0 to 2000V. The voltage readout is absolutely stable and adjustable to 1V resolution.

Theremino MCA displaying a typical Thorium spectrum.


PRA displaying the results from a Lu-176 (LYSO Crystals)  Gamma-Spectrum Analysis.

3. MCA or Multi-Channel Analyzer - this can be either a stand-alone external hardware unit or a software component running on a PC. I use the software version - actually a few of them, running together on my laptop - PRA, Theremino MCA and BecqMoni (all free software btw.). Each application has its own pros and cons but fortunately they can run simultaneously on the same machine and listen to the same detector.
If the Detector is the "heart" and the Driver is the "spine", then the MCA is the "brains" of the operation. The software is not extremely complex but requires a good bit of understanding and knowledge in Gamma Spectroscopy and pulse detection.  What the MCA does is, it detects the pulses by their shape and classifies the pulses coming from the PMT based on their energy and timing and creates histograms displaying spectrum energies, counts, etc.


4. Interconnects - these are mainly the cables connecting the Detector to the Detector Driver and cable connections for audio and USB to the computer. 
GS-USB-PRO supports "Single cable" configuration (which is what I use) where HV bias and signal pulses share the same coax cable) as well as "Dual Cable Mode"  - 2 separate lines - one for the PMT HV bias and one for Signal. There are 2 connectors on the front - in a "Single cable mode" only the SHV connector is used. In "Dual Cable mode", configured by a switch, the signal is received over the BNC connector while SHV is used for the PMT bias only. 
The critical part is the Signal cable as it should be of a high-quality coaxial type with a fairly low capacitance - best is less than 60pF total (with both connectors installed on). Higher cable capacitance widens the pulses (which normally are very short) and makes their shape "mushy", especially the trailing edge.
The high-voltage cable should be done with a properly shielded coax cable, able to withstand up to 2000V without arcing. For anything above 1000V I would use SHV-to-SHV connectors - regular BNC are not really meant for high voltages. I installed a female BNC connector on my Detector housing so my single coax is SHV to BNC as the detector runs on only 650V.
Cables are part of the calibration process - i.e. changing cables WILL require a re-calibration.

The inner copper shielding is clearly visible inside the main shielding shells. The additional, removeable copper cylindrical inserts are not pictured here. They are made by rolling thick copper foil into a tube with thick wall and the exact diameter of the opening.

Main detector shield (top) and the "large sample holder" (bottom) mated together - the lead thickness is 22mm around the crystal or 32mm once the outer sleeve shield is installed. In the wooden shield-carrier box, this assembly is placed over 1" thick base of lead bricks and the outer shield (sleeve) is placed over the assembly.

Outer lead shielding - 9.5 mm of lead.
 Nothing beats the look of  a OD Green colored "canister" looking device with a bright yellow "Radioactive" sign 😁.

5. Detector Shielding - for more details on Lead shielding for my detector, please see here!

By now, I should have approximately 0.736 uCi of activity left in my Cs-137 Calibration Source.
 The uncertainty of these sources is generally +/- 20% of the specified activity.

6. Calibration Sources - these are extremely important aids for the proper energy calibration of the instrument. Without accurate calibration the measurements are more or less meaningless.
In theory, any strong radioactive source with a known spectrum and activity can work. 

Decay of Cs-137

Cs-137 Gamma-spectrum (peak @661.6 keV) is a good starting point for the calibration process.

I normally start with K-40 @ 1.46 MeV (actually, I use about 100 gr of KCl as a K-40 source) - this usually takes the longest time to get a good, measurable peak as the activity is fairly low - while doing this I can also observe the positron annihilation peak at 511 keV as a reference. Next step is calibration with Lu-176 source (LYSO crystals) at 201.83 keV and 306.78 keV and finally, the "classic" Cs-137 (1.0 uCi) disk source at 661.66 keV and the X-ray peak at 32.19 keV. 
Energy calibration, while using low-to-high activity source order allows me to keep the lower activity peaks in the histogram for a little bit as a reference, just to make sure my calibration is not drifting. These peaks stay until the higher activity source "pushes them down" but still it is a good confidence check.
Thorium (ThO2) also works well to check the linearization of the detector as it has many peaks from daughter products which are spread throughput the spectrum to as high as 2614.5 keV (Tl-208)

I love the portability and compactness of my setup - everything fits in 2 Pelican cases and can be carried in the field. One case contains the detector, driver, interconnects, etc. The other case holds all of the lead shielding components (and weights over 50lbs!). Calibration sources are also stored inside the shielding in their own sealed "lead pig" in order to prevent any possible contamination of the shield which will skew the measurements.


It is amazing to think that once available only in Scientific Laboratories and Universities, Gamma-Spectroscopy is now affordable and enjoyable for hobbyists and Nuclear / High-Energy Physics  enthusiasts!  
Special Thanks to Steven Sesselmann for hosting the Gamma Spectacular Science Forum.

No comments: