Posted on by maxfomitchev

The shape of a thermal neutron spectrum produced by a Helium-3 filled proportional counter tube generally depends on bias voltage. Three things happen with the increase in bias voltage:

  1. Pulse amplitude increases (which helps with noise rejection);
  2. Pulse rise time increases;
  3. The thermal peak at 764 keV gradually disappears (i.e. the tube looses ability to resolve neutron energy).

Further increase in bias voltage will transition the tube’s operation from proportional mode (where pulse-height depends on neutron energy) to corona mode, where pulse-height is independent of energy. Detector pulse magnitude in corona mode is generally much higher than in proportional mode, but there are many noise pulses, which need to be rejected by increasing the magnitude of the trigger level.

The neutron count rate should remain constant and remain independent of the applied bias, however. The increase in the count rate with the increase in bias is indicative of EM noise or detection of gammas. Although if the trigger level was set too high initially (thus rejecting low-amplitude neutron pulses) the neutron counts will increase with the increased in bias since low-energy neutron events will grow in amplitude and thus will eventually surpass the trigger level and be counted.

Fig. 1 compares thermal neutron spectrum of an SNM-18 proportional counter acquired at 1350V bias (red) vs the spectrum acquired at 1400V. Thermal neutron peak is clearly defined in both cases although the height of the peak is greater (FWHM is lower hence energy resolution is better) at lower bias voltage. The signal amplitude increased from ~20 mV to ~50 mV, count rate remains the same at ~9 CPS.

Fig. 1. Comparison of thermal neutron spectrum of an SNM-18 proportional counter acquired at 1350V bias (red) vs the spectrum acquired at 1400V (blue).

At 1450V bias the thermal peak disappears – Fig. 2, the pulse amplitude increases to ~100 mV, count rate remains at 9 CPS,

Fig. 2. Comparison of thermal neutron spectrum of an SNM-18 proportional counter acquired at 1350V bias (red) vs the spectrum acquired at 1450V (blue).

At 1500V bias the gamma peak appears in the low channels – Fig. 3, the pulse amplitude increases to ~250 mV, count rate increases to ~ 10 CPS (due to the contribution from the gamma peak).

Fig. 3. Comparison of thermal neutron spectrum of an SNM-18 proportional counter acquired at 1350V bias (red) vs the spectrum acquired at 1500V (blue).

At ~1500V the pulse voltage maxes out and starts declining with further increase in bias: e.g. the amplitude drops to ~150 mV at 1550V and goes down to ~100 mV at 1650V, at which point the tube transitions to corona mode.

The actual voltages will depend on detector type and will vary even among detectors of the same type. Also, corona mode tubes generally require higher resistor in series with the power supply (e.g. 50 MOhm) to quench the discharge whereas 1 MOhm is sufficient for proportional mode. What is surprising though that this particular SNM-18-1 counter did not produce higher magnitude pulses in corona mode.

Conclusion

There is a trade off between neutron detector signal magnitude and energy resolution. When energy resolution does not matter one can raise the bias voltage to increase the pulse magnitude, although one has to be careful to adjust the trigger to reject gammas that may appear as a low-energy peak on spectrum.