I'm trying to build a basic detector for x-ray or low energy gamma spectroscopy. I have a big photodiode with a CsI(Tl) scintillation crystal glued on top (Hamamatsu S8559 for x-ray detection [1]) and want to detect for example gamma radiation from a ~1uCi Am-241 source.

To do spectroscopy and being able to discern the pulses by energy I would first have to be able to measure single pulses coming from radiation hitting the detector. So I tried to calculate the charge of these pulses:
The CsI(Tl) scintillation crystal [2] has a light yield of 54 photons/keV and Am-241 decays partly via gamma decay with an energy of 59.541keV, therefore I calculate the number of 550nm photons produced by the crystal after being hit with a gamma photon to be 54 photons/keV * 59.541keV = 3215.2 photons.
Quantum efficiency has been calculated to be about 0.586 for my photodiode at 550nm, so I would expect 3215.2 photons * 0.586 = 1884.1 electrons being generated by the photodiode, which equals a charge of about 0.3 femtocoulombs.

This seems really low, so my questions are:

  • I realize this is not the physics stackexchange, but can someone check if my calculations are realistic?

  • If this is really the charge I have to detect, how would I do this? Should I use a charge amplifier circuit with a tiny feedback capacitance, for example with an LMC662? [3]

  • Is this photodiode + crystal combination even usable for this application or would I have to use a silicon photomultiplier?

  • Sadly my photodiode has a huge junction capacitance of over 500pF even with the highest allowed reverse bias. Can I use photodiode bootstrapping to reduce the capacitance? [4]

I found a circuit that seems close to what I need in [5], but it uses a photodiode without scintillation crystal that is also much lower capacitance than mine, so I'm not sure it would work.

[1] https://www.hamamatsu.com/resources/pdf/ssd/s8559_kspd1051e.pdf
[2] https://www.crystals.saint-gobain.com/sites/imdf.crystals.com/files/documents/csitl-and-na-material-data-sheet.pdf
[3] http://www.ti.com/lit/ds/symlink/lmc662.pdf
[4] http://www.analog.com/en/design-center/reference-designs/circuit-collections/large-area-photodiode-bootstrapping.html
[5] http://einstlab.web.fc2.com/Gamma/spectroscopy.html

  • 1
    \$\begingroup\$ I use NaI + PMT. I think you have made a good choice, though, as CsI(TI) does have a nice emission near 550 nm that fits well with silicon photodiode quantum efficiencies. The light yield of CsI(TI) is good, as you cite. You may want to include a multilayer teflon reflector to maximize your detector's light collection (you can use an optical grease to get needed optical coupling.) A PIN diode might be a good choice to match up with this crystal, but you will definitely need to use a guard-ring, a carefully selected pre-amp design, and pulse shaper. Crystal reset time is slow; 1 us or longer. \$\endgroup\$
    – jonk
    Jul 7, 2018 at 4:29

1 Answer 1


You may consider using an avalanche photodiode so very weak radiation could be detectable.

Avalanche photodiodes generate current, so basically a single photon generates thousands of electrons and makes it easier to measure the number of particles hitting the detector. There are ready devices that use single-photon detection avanalche diodes, but I suspect they are quite pricey.

The other thing is that:

"Americium-241 decays mainly via alpha decay, with a weak gamma ray byproduct."[1]

Alpha rays are basically 4|2 He and penetrate very tiny distances. Am is a bad gamma ray source, so you may want to use other sources of radiation to detect gamma rays.


  • 1
    \$\begingroup\$ Avalanche diodes produce the same signal amplitude for each hit, so they are unusable for spectroscopy. Silicon photomultipliers are arrays of avalange diodes, which give signals linear to the number of fired diodes. They are suitable, but show big temperature and voltage dependence. And if Am is good or not, depends on the targeted energy range. \$\endgroup\$
    – sweber
    Jul 7, 2018 at 16:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.