I'm comparing a Geant4 (Monte Carlo based) simulated gamma spectrum of $^{137}$Cs to my experimental results using a NaI(Tl) detector. I find good agreement between the two for the main features such as for the photopeak, Compton edge and the Ba X-rays.

The deviations at the Compton continuum I think is a result of the simulation setup of the detector, as it doesn't include a reflector material around the scintillation crystal. The experimental setup consists of a shielding lead box inside which the Cs source is placed. Within the shielding lead box there is another smaller copper and tin box $\sim$ 1 mm thick each, and it is inside here the Cs source is placed.

Now, the two features/deviations I can't explain are indicated by arrows in the spectrum below. The first peak is the characteristic lead X-rays from the lead shielding, but this only shows up in the simulation and not the experiment. The second feature is the rise before the backscatter peak in the experiment, the origin of which I can't explain, only showing up in the experiment and not the simulation. I thought maybe that this might be due to the escape of characteristic X-rays (28 keV) of I in the crystal, but this should in that case show up as a peak, which it does not.

Basically the deviations in the whole segment from the Ba X-rays to the backscatter peak.

Edit: the lead X-ray peak is resolved, however the rise below the backscatter peak I can still not explain nor recreate in the simulation.

Simulated (blue) and experimental (red) Cs-137 spectrum

  • $\begingroup$ Did you simulate all likely scattering paths from your source to your detector? $\endgroup$
    – John Doty
    Commented Nov 12, 2023 at 18:35
  • $\begingroup$ Re "Ba X-rays": Do you mean "Ba gamma rays" (from the (a?) barium-137 isotope (approx. 660 keV), corresponding to the peak on the right in the spectrum)? If not, what does it refer to? $\endgroup$ Commented Nov 13, 2023 at 16:45
  • 2
    $\begingroup$ @Peter-Mortensen When Cs-137 decays, it is usually via a meta-stable Ba-137m state that decays via the 662 KeV gamma to the Ba-137 nuclear ground state. The Ba-137 atom is often not initially in its atomic ground state, and about 6% of the decays produce a 32 KeV Ba-137 $K_\alpha$ x-ray. The question is mostly about the simulation-data differences between the 32 KeV x-ray peak and the backscatter peak around 200 KeV. $\endgroup$ Commented Nov 13, 2023 at 16:58
  • $\begingroup$ @David Bailey: Thanks, that is a better lead. From The Rich Physics of Cs-137 Gamma Spectrum: "...the 137Cs gamma spectrum: the 32 keV peak is actually a barium x-ray fluorescence (XRF) line! One may find it confusing that both the 661.7 keV and 32 keV lines that we commonly associate with 137Cs actually originate from 137Ba." $\endgroup$ Commented Nov 13, 2023 at 17:13
  • $\begingroup$ cont' - Though there seems to be a contradiction, electrons vs. radioactive decay. $\endgroup$ Commented Nov 13, 2023 at 17:22

1 Answer 1


Backscattered photons can Compton scatter again, so the rise below the backscatter peak is likely the Compton edge of the backscatter peak. This feature due to multiple Compton scattering is mentioned and shown in R.L. Heath's "Scintillation Spectrometry Gamma-Ray Spectrum Catalogue". See page 13 and figure 5 of the 1957 edition and page 14 and figure 13 of the 1974 edition. Heath's figure gives spectra for 3 simple configurations that you could use to test your simulation.

Figures 4 & 5 from 1957 report, showing 3 shielding configurations and spectra

Cu-Sn-Pb graded shielding reduces the number of Pb X-rays entering and then backscattering out the Cu side, but the Cu-Sn layers do not make much difference in the the number coming out the Pb side. That just depends on how thick the Pb is. The graded shielding seems to be working since the real data does not show a Pb X-ray peak. The presence of the peak in the simulation makes me wonder if your Geant4 geometry is correct. Are you sure you have the Cu, Sn, Pb thicknesses and order correct in your simulation?


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.