Ordering nuclear surplus from Russia and custom crystals from China to perform gamma-ray spectroscopy at home
In the fall of 1998 I played Fallout 2:
It permanently altered the course of my life - I have been obsessed with radioactivity ever since.
The ability to detect/see radioactive particles feels like a superpower. I had already built a geiger counter and a cloud chamber, what was left?
A lot.
Whenever I start a new hobby or interest, I take the same approach:
Radiation detection was no different. My homemade geiger counter was built around a Soviet SBM-20 Geiger–Müller tube - a fun starting point, but limited in its detection capabilities. My goal was to have the ability to measure (in useful ways) every major type of ionizing radiation (alpha α, beta β, gamma γ, and X-rays). (Neutron radiation will be out-of-scope until I get around to building a fusor)
The process was:
And with all of that completed, I unlocked a new ability:
Gamma rays are the highest-energy form of electromagnetic radiation. Each isotope produces gamma rays with energy focused at particular levels. By measuring this spectrum of energies it is possible to distinguish radioactive isotopes.
Radiation-measuring instruments come in various forms:
I wanted to have at least one of each of these instruments.
Survey meters are the classic handheld radiation measuring devices. Models differ in how the detector attaches (or the detector may be internal), how much control you have over the pulse detection (threshold, high voltage), the quantity displayed (raw counts vs rates), and how rugged they are.
I started by getting a consumer geiger counter, aiming for a portable device which offers data logging capabilities and easy modification. The GQ GMC-300E Plus is a popular unit with an active forum documenting successful modifications.
With modifications the GMC-300E Plus can work with a variety of Geiger-Müller tubes, but I planned on using many other detector types. For that I needed a more serious instrument.
Modular analog survey meters fit the bill - they are rugged, reliable, and highly customizable. Ludlum is the gold-standard for this market. Their Model 3 is the classic, I went with the Model 3-97 variant (which includes an internal 1"x1" NaI(Tl) detector):
For external probes, the high-voltage (HV) may be adjusted from 400V to 1500V - this gives some flexibility, but even more range would be useful. The Ludlum Model 12 has a wider HV range (400V to 2500V) and adds an adjustable discriminator (to set a threshold for which pulses should be counted):
I had my lineup of pulse-counting instruments, so I moved on to the next category.
The purpose of a dosimeter is to measure the cumulative absorbed dose of ionizing radiation. They are pocket-sized and are affixed to the clothing (usually the chest or torso area) in situations where occupational exposure may occur. By measuring the dose in a standardized location, the dosimeter can estimate doses for major organs and different depths in the human body.
In the classic pen form factor, I started with an Arrow-Tech Model 725:
It has a huge range of 0-5 Roentgen(R), which is really only useful in the case of an accident or if it were placed directly in the beam of an x-ray. For practical purposes, I wanted a dosimeter with a finer resolution. The Arrow-Tech Model 138 has a range of 0–200 mR, much more practical:
Speaking of being placed directly in the beam of an x-ray, here is what the Model 138 looks like when imaged with x-rays:
They are surprisingly simple: a filament is placed in a tiny air-filled ionization chamber and when X- or gamma-rays ionize the air in the chamber, the filament deflects electrostatically. This deflection is visible on a dose-calibrated scale visible through a built-in microscope.
This process depends upon the ionization chamber starting at a known voltage. To achieve this, a charger is used. Two types exist, the first uses batteries and a DC-DC step-up to smoothly adjust the voltage applied to the center electrode:
The other type presses a plunger against a piezo crystal that generates the necessary voltage:
With the classic pen dosimeter obtained, I moved on to the modern implementation: an electronic personal dosimeter (EPD). As these instruments are intended for monitoring the cumulative dosage of workers who may be exposed to ionizing radiation, they add additional functionality to the pen dosimeter: continuous real-time readings, long-term dose logging, and alarms.
I went with the Thermo EPD Mk2.3:
That covers general-purpose dose-measuring, but there’s a special-purpose instrument in this category which may come in handy if (when) I play with X-rays: an X-ray exposure-rate meter. The Nuclear Associates RAD-CHECK Plus Model 06-526 is a special-purpose instrument containing an ionization chamber (with optional external chamber), calibrated for the range of typical diagnostic X-ray beams (30kVp-150kVp).
It can measure X-ray dosages from 0.001-1.999R, and exposure rates of 0.01-19.99R/min.
By attaching the Model 6000-528 external ionization chamber I gain flexibility in where I can take measurements:
When working with X-rays, exposure time is a primary concern (for safety reasons, but also for ensuring proper exposure of images). The Electronic Control Concepts Model 8700 measures total exposure time of an X-ray beam (for calibration of an X-ray source), and also the number of X-ray pulses produced. It does this using a silicon PIN photodiode-based sensor.
The last piece of X-ray-specific instrumentation allows me to measure the peak energy level of an X-ray generator. The Nuclear Associates kVp Meter Model 07-479 contains two silicon PIN photodiodes with two different sets of filters. By comparing the ratio of the accumulated doses between the two diodes, the instrument can derive the peak kVp of an X-ray beam.
Owning these X-ray measuring instruments is overkill, but who can resist a good eBay auction?
I returned to my ultimate goal: gamma spectroscopy. Each gamma-ray interaction in a detector produces a voltage pulse. The peak amplitude of that pulse is proportional to the deposited energy from the ray. A multichannel analyzer (MCA) takes these voltage pulses and bins them by amplitude, producing a spectrum / histogram.
As an example, here is the gamma spectrum of background radiation on the earth’s surface:
Peaks for potassium-40 and the daughter products of thorium and uranium (lead-212, bismuth-214) are marked in the image.
Thanks to the open-source Theremino project, I was able to make my own MCA. Binning occurs in the Theremino MCA software, but acquiring the voltage peaks requires supporting electronics. First I needed an adapter which could provide high-voltage to the gamma spectroscopy probes (detailed later) and stretch out the length of the pulses, then I needed an analog-to-digital converter (ADC) to sample the pulses and digitize them.
The probe adapter was a relatively simple build. I started by having the PCBs made:
Then I ordered all the necessary components:
For a shielded enclosure, I used my go-to: an electrical junction box. After soldering everything together, I had the probe adapter:
For the ADC, China came to the rescue: I modified a dirt-cheap USB audio card.
I made the following modifications:
After making these modifications, I had an ADC:
Together with the Theremino software, I had an MCA. I moved on to the detectors.
All radiation detectors work in the same way: an interaction occurs in a medium, causing an effect that can be measured.
Charged particles (alpha, beta) directly cause ionization or excitation along their tracks.
Photons (X-rays, gamma-rays) interact via:
A Geiger-Müller tube consists of a chamber filled with a gas mixture. A large voltage difference (hundreds of volts) is established between a central electrode and the conductive walls of the chamber. When a radioactive particle strikes the tube, ionization occurs in some of the gas molecules (either directly, or as a secondary effect of interaction with the walls). These ions migrate in the gas due to the strong electric field caused by the voltage difference and cause further ionization, starting an electron avalanche. The avalanche quickly turns the gas near the electrode into a conductive plasma, completing a circuit - current briefly flows causing a sudden voltage dip which is detected as a pulse. The electric field collapses and the tube recharges, ready for the next event.
By designing a GM tube to limit (or allow) particular types of ionizing radiation to penetrate the chamber, you can produce a variety of different detectors.
I’ve already mentioned the vintage Soviet SBM-20, which can detect beta and gamma radiation (its thick metal walls prevent detection of alpha radiation), as well as X-rays above 40keV.
The M4011 is a modern GM tube with a similar construction (using glass walls instead of metal), and the same detection capabilities.
If I want to measure alpha radiation as well as low-energy X-rays (<=40keV, the range where bone and tissue absorption increases) I need a tube with a thin window to admit these forms of ionizing radiation.
The vintage Soviet SBT-9 GM tube (used in Russian spacecraft!) has a mica end-window, which enables it to detect low-energy X-rays and alpha radiation (as well as the usual beta and gamma radiation).
Another vintage Soviet tube, the SBT-11A has a larger mica window, making it even more sensitive to alpha radiation and low-energy X-rays.
And while I am at it, since I already have Ludlum meters I will get a Ludlum Model 44-9 GM probe which contains an even larger mica window:
With a collection of GM tubes acquired, I moved on to modifying my digital geiger counter to accept them.
Thanks to the experimentation of other owners, I have proof that the GMC-300E Plus can handle all of my GM tubes. By adding a DIP switch, I can swap between my choice of GM tube or use more than one GM tube at once.
The SBT-9 needed a hole cut out of the side to expose the mica window:
And the SBT-11A was mounted externally:
With GM tube detectors taken care of, I moved on to a more exotic detection technology: scintillation detectors.
A scintillation detector measures ionizing radiation by combining two components: a scintillator (usually a crystal) that emits photons after an interaction, and a photomultiplier tube (PMT) which uses a photocathode to detect photons via the photoelectric effect (multiplying the signal with a series of dynodes).
The intensity of the light produced by the scintillator is proportional to the energy deposited by the incident radiation - this is what allows the creation of the histograms used in gamma-ray spectroscopy. The energy-level peaks in the histogram unambiguously reveal the source(s) of the incident radiation.
With an understanding of how they work, I next built some scintillation detectors.
When it comes to scintillators, Thallium-doped Sodium Iodide (NaI(Tl)) is the most well-established and extensively used. It has high stopping power, very high luminescence, and the blue light they emit (~415nm) matches well with the optimal wavelength (420nm) for the bialkali photocathode found in most PMTs. The drawback is that NaI(Tl) is hygroscopic and must be hermetically sealed in a metal container (with an optical window) to prevent absorption of moisture from the environment (which degrades their performance).
I bought two NaI(Tl) crystals. The first a 1" (diameter) x 1" (height) cylinder:
The second a more impressive 2.5" x 2.5" cylinder:
Aside from the obvious (a bigger detector captures more radioactive particles), the thicker crystal increases sensitivity to high-energy gamma-rays (>= 1 MeV) by increasing the probability that Compton-scattered photons are detected before leaving the crystal.
There are some drawbacks to NaI(Tl), however. The metal encapsulation stops alpha-radiation and most beta-radiation, and low-energy X-rays are heavily attenuated. They are designed to prioritize gamma-radiation. If I wanted to be able to detect alpha, beta, and low-energy X-rays, I needed an alternative scintillator.
Thalium-doped Cesium Iodide (CsI(Tl)) was the solution. It is less hygroscopic which allows a thin metal film to be used in place of a thick metal enclosure - as a result it can detect alpha, beta, and low-energy X-rays. In addition, if the crystal is kept thin, the chances of a gamma-ray interaction are reduced. The drawback of CsI(Tl) is that it emits green light (~550nm) which is not a good match for typical bialkali PMTs (resulting in smaller pulses and lower resolution).
As a reference point, the commercial RAP47 scintillation probe uses a 1" diameter x 1mm thick CsI(Tl) crystal tuned for 20-40keV X-rays. I used the same size for my CsI crystal, and experimented with one even thinner (to be even more selective for 20-40keV X-rays).
Obtaining a polished cylindrical CsI(Tl) crystal in these dimensions required a custom order. I emailed numerous suppliers around the world and found one who could produce a small batch of 1" x 1mm and 1" x 0.5mm discs at a reasonable price: OST Photonics in China.
With my scintillation crystals obtained, I moved on to the other half of the scintillation probes: the photomultiplier tubes.
As described earlier, a photomultiplier tube (PMT) detects photons (as little as a single photon!) with a photocathode which emits electrons when hit. These electrons are multiplied via secondary emission in a high-voltage (500-2000V) dynode chain, multiplying at each stage until collected at the anode and producing a measurable pulse.
To achieve the multiplying effect, each successive dynode requires a higher voltage than the last. This is achieved through a voltage divider attached to the end of the PMT.
When choosing PMTs for my scintillation detectors, I had two major considerations:
I started with the 2.5" NaI(Tl) crystal. The Hamamatsu R6233 is a 3" diameter PMT, with a peak sensitivity near 420nm:
To align the 2.5" crystal with the center of the 3" PMT optical window, I modeled and printed a gasket:
For the PMT voltage divider, I modeled and printed a socket:
For connecting to the PMT pins, Molex 08-50-0106 was the perfect size. I soldered the necessary components (resistors and a capacitor) directly onto the crimps. This allows easy removal/replacement of the PMT without having to desolder anything.
Next I tackled the 1" NaI(Tl) crystal. The Hamamatsu R6095 is a 1" diameter PMT, with a peak response near 435nm:
For the 1" CsI(Tl) crystals, the Hamamatsu R268 was a suitable PMT with 1" diameter and quite good response at 550nm:
Both the R6095 and the R268 use the same 14-pin style socket, so there was an opportunity to create a generic voltage divider PCB. The R6095 has 11 stages of dynode, while the R268 is a 10-stage tube - I simply leave out one resistor in the divider when populating the PCB for the R268.
The finished PCB:
I also modeled and printed a socket which works for both PMTs:
By designing the socket in two pieces I was able to embed female DB25 crimps in the adapter for connecting to the PMT pins and soldering directly to the PCB.
The scintillation detectors were ready to be assembled.
I started by coupling the scintillation crystal to the PMT. Both glass surfaces were cleaned with acetone, and optical grease was applied to one surface. Optical interface compounds have a refractive index very close to glass, and their purpose is to eliminate glass-to-air and air-to-glass transitions. At each transition, a photon may be reflected back or refracted in another direction - improving this coupling increases the amount of photons which reach the photocathode.
My NaI(Tl) crystals are encapsulated in metal housings lined with a reflective coating (to maximize the amount of photons that leave the optical window), which made coupling with the PMTs easy. My CsI(Tl) crystals are bare, however - which meant after coupling with the optical grease I needed to take the additional step of wrapping the exposed crystal with a reflective material. White PTFE tape worked great for this purpose.
For my CsI(Tl) detectors, I wanted the end of the crystal to have as transparent a barrier as possible (since I was targeting low-energy X-rays). To do this, I designed a window of 0.8 mg/cm² aluminized polyester known as Alpha-Film (aka Rad-Film). This material blocks light without substantially reducing low-energy X-ray transmission.
With the crystals coupled to the PMTs, I wrapped everything in black electrical tape to prevent light leakage (external photons must not reach the photocathode).
The PMTs were plugged into their sockets:
PMTs are extremely sensitive to magnetic fields - even the earth’s magnetic field can cause noticeable shifts in their output. To combat this, I surrounded each detector by 0.8mm of solid mu-metal:
I modeled and 3d printed an opaque enclosure for each of the probes:
For the CsI(Tl) probes, the bottom was left open to maximize transmission:
For the connectors I went with BNC, and for the cables I went with RG58/U coaxial cable. A chassis-mount BNC adapter worked great:
I finally had my lineup of gamma spectroscopy probes:
And a complete gamma spectroscopy system:
If I wanted to use my gamma spectroscopy probes for their intended purpose (producing a spectrum of energy peaks to identify radioactive isotopes), I needed a way to shield them from background radiation. Beyond cosmic rays, scintillation probes will count gamma rays given off by concrete and other materials in the surrounding environment. I did my best to block as much of this noise so that the spectrums produced by the probes have distinct peaks without getting lost in the noise (especially problematic for low-activity samples and trace isotopes).
Lead is a well-known shield against radiation. It hits the sweet-spot: it is dense, soft/malleable, has a low melting point (so is castable), is radioactively quiet, and is reasonably priced (relative to the denser and harder tungsten, for example). It is also commonly used in roofing as a weatherproofing material (lead flashing) which makes it easy to obtain.
There is one problem, however. When lead is hit with high-energy gamma rays, it emits X-rays (~75-85keV). To ensure a clean spectrum, these X-rays need their own layer of shielding. A graded shielding method addresses this - adding a layer of copper attenuates the X-rays from the lead. Of course, copper itself exhibits a similar phenomenon, so another layer is added to address that, and so on. Unfortunately, each additional layer of shielding adds additional backscatter, which adds noise back to the spectrum. A balance has to be struck.
Thanks to shielding calculators created by users at the Gammaspectacular forums I calculated target dimensions and materials. To cut the background radiation in half, I would need 13.27mm of Lead, and to cut the lead X-rays by 75% I would need 2.04mm of copper.
I ended up with 10.5mm of lead surrounding a 4" diameter copper tube (2.0mm thick), 12" in length. The entire outer surface was covered with tape to reduce lead exposure when moving the shield.
This got close enough to the goal:
For a cap, I used a similar arrangement: a copper disk (2.82mm thick) on top of lead flashing (3.84mm thick).
When in use, the top is left open and the sample is placed at the very bottom of the tube for maximum shielding:
(The ideal graded shield would consist of lead, tin, then copper. I will consider adding a tin layer if the current design proves to be insufficient.)
With everything built, I needed to calibrate the spectrum generated by the MCA. To do this I needed a material which would produce a known spectrum - a radioactive source.
I purchased Cs-134 and Tl-204 check sources:
Cs-134 produces strong peaks at 605 keV and 796 keV. By maintaining source-to-detector distance and keeping our shield in place, I was able to use both of these lines to calibrate the MCA and probes. Once calibrated, I characterized the performance of my probes.
The measure commonly used to quantify the resolution of a gamma spectroscopy probe is the “full width at half maximum” (FWHM) of the spectral peak:
With the resolution defined as a ratio:
Resolution(%) = FWHM (keV) / Epeak (keV)
Expected resolution for my probes would be:
I measured the following:
The full collection of radiation detection gear:
Background radiation measured with the 2.5" x 2.5" NaI(Tl):
TODO
Background radiation measured with the 2.5" x 2.5" NaI(Tl) from inside the lead shield:
TODO
){kind=link}