As of July 2022, a large quantity of nifty PMT modules is available on eBay at around US $20 each. Consisting of a subminiature photomultiplier tube bonded to a thallium-doped cesium iodide scintillator crystal, these modules appear to be designed for use in 'radiation pagers' for counterterrorism and border security applications.

While insensitive to alpha and beta radiation, the scintillator/sensor combo is very good at detecting higher-energy sources such as gamma and neutron emitters, offering noticeably better sensitivity than a conventional Geiger-Mueller counter. Even more interesting, it seems likely that these modules can be used for pulse-height spectroscopy applications given appropriate hardware and software support.

As a quick experiment, I tried powering up the PMT with a voltage divider and -800 volt power supply comparable to that recommended by Hamamatsu for the R7400U, depicted below. (Also see catalog here.) It wasn't necessary to rig up an amplifier or pulse-shaping circuit to see some results. Simply connecting a scope probe to the PMT anode was (somewhat disturbingly) effective at picking up energetic background radiation.

In the silent 5-minute video below, each vertical division corresponds to 10 volts at the PMT anode. Sweep rate is 200 us/division.

Pulses exceeding 10 volts in height arrived several times more frequently than background clicks from a nearby Geiger counter.

With gain exceeding 700,000x and quantum efficiency near 20%, the R7400U is extremely sensitive to low levels of light. Oversimplifying a bit, the PMT anode acts as a current source that delivers 50 amperes per lumen (typical) into the 10-megohm load presented by the probe tip, yielding an overall sensitivity near 2E-9 lumens per volt in this setup.

The PMT's data sheet specifies its maximum linear output current at 13 uA, or 130 volts into 10 megohms. Pulses due to background radiation seem to reach at most 60 volts in amplitude, corresponding to about 1.2E-7 lumens from the CsI(Tl) crystal. The crystal's emission peak in the graph at right doesn't quite match the PMT's response curve, so the actual light output is presumably higher.

Using similar sensors with a ²⁴¹Am source (γ=59.5409 keV) for X-ray spectroscopy.

Solid Scintillators (1/6/2010): Informative slide deck that discusses both scintillator materials and PMT characteristics in detail (108 pages, 10 MB)

Open Gamma Detector and Gamma-MCA: Gamma spectroscopy with Raspberry Pi Pico 2 and PWA UI

Some further tests with a Faxitron MX-20 indicate that the scintillator's response below about 12 keV is strongly attenuated, probably by the shielding around the crystal along with the copper tape used to secure it to the test board. Observed output amplitude at 33 kVp was about half that seen at 37 kVp.