Helium-3 proportional counters are generally used to detect thermal neutrons and therefore produce thermal neutron spectrum, which looks the same regardless of the energy of the neutron source.
Obviously, in order to detect thermal neutrons fast neutrons first must be thermalized (i.e. slowed down) using moderator such as high-density polyethylene (HDPE). So, fast neutrons must pass through several inches of moderator to loose their energy (i.e. thermalize) so they can be absorbed by a proportional counter. Without the moderator 99.99% of the fast neutrons will pass through the detector without producing any signal because neutron absorption cross-section of Helium-3 (and most other elements used for neutron detection) decays exponentially with neutron energy – Fig. 1.
As a result Helium-3 is ~10,000 times is less efficient at capturing fast (E ~ MeV) neutrons as it is at capturing thermal (E < eV) neutrons. However, if the neutron flux is high enough one can use proportional counters without moderator to measure true fast neutron spectrum.
Conversely, recording neutron spectrum from a weak source for a long enough time will yield some counts outside of the thermal neutron spectrum. The trick is to situate the neutron source in such way as to maximize neutron path through the detector. So, I took an 12″ long SNM-18 proportional counter tube and situated a 5 mCi Po-Be neutron source at the end of the tube such that a good deal of neutrons could travel along the tube and thus maximize the absorption probability – Fig. 2.
Although the moderator was not necessary for this particular measurement I still used it to obtain a nice thermal neutron spectrum. Even without the moderator I would have gotten the same thermal neutron spectrum but f lower magnitude as some neutrons will thermalize naturally as the interact with the environment in the lab.
The neutron spectrum that I have obtained after ~6 days of counting is shown on Fig. 3.
As you can see from fig. 3, I was able to get some higher-energy neutrons to register: they correspond to the ‘tail’ extending past the 764 keV thermal peak. There were 982,144 counts in the 0 to 800 keV range and only 824 counts in the 800 keV to 2.5 MeV range and only 77 counts with energies over 2.5 MeV. This count difference is inline with about 1,000-fold reduction in sensitivity for detecting non-thermal neutrons using Helium-3 counter.
However, I was not able to recover the true fast neutron spectrum of Po-Be source, which is shown on FIg. 4.
The key reasons why despite counting for 5.5 days I was not able to recover the true Po-Be neutron energy spectrum is that the 5 mCi Po-Be source is too weak; a source on the order of a few Ci range is required for fast neutron spectrum acquisition.
Also, the ideal environment should be free of moderator, moisture and hydrogen-rich materials to prevent neutron thermalization and thermal neutron spectrum build-up. Partial moderation could have also skewed the high-energy spectrum by increasing the counts of partially scattered neutrons.
Last but not least, for energies greater than 1 MeV Helium-4 (rather than Helium-3) based neutron detectors are more suitable due to their grater fast neutron sensitivity and superb gamma / neutron discrimination ability.