Currently, a 3 x 3 BGO detector is incorporated into the design of the PELAN.  However, the high γ-ray efficiency of this detector has a drawback in that it detects γ-rays from the environment surrounding the object under interrogation. These γ-rays from the environment are essentially noise that must be filtered from the signal from the object under interrogation. Since there is no practical way to focus the neutrons the signal-to-noise ratio (SNR) of the detector must be modified. We have tried several approaches to solve this problem.

The traditional way of removing the noise from the ground or from the neutron generator is to place a very heavy, dense material such as lead between the objects creating the noise and the detector. Unfortunately, this has two drawbacks: 1) for a portable system, this increases the weight, and 2) these materials contribute to the noise.

Another way of removing the noise from the spectrum is to use an active filtering device.  For example, Compton-suppression systems are often used in nuclear physics to eliminate the continuum caused by Compton scattering within the detector.

Carbon and oxygen measurements are most likely to be affected by the noise from the environment and the neutron generator.Thus, an active filter must be optimized for γ-rays with energies between 4 and 6 MeV. In this energy range, the two dominant photon interactions with matter are pair production and scattering (primarily Compton scattering).

Figure 1 is a schematic drawing showing γ-rays entering the detector. On the left-hand side (A), three γ-rays are entering PELAN’s detector. The top-most γ-ray is scattered and there is an appreciable probability that the Compton scattered γ-ray will leave the detector. The bottom-most γ-ray undergoes pair production with a 511 keV γ-ray escaping the detector. Only the γ-ray that is perpendicular to the front surface of the detector deposits its entire energy within the detector.

• Figure 1. Schematic representation of high energy γ-rays entering the PELAN detector (A). The PELAN detector surrounded by a veto detector (B). See text for details.

On the right-hand side of Figure 1, the same detector is now surrounded by another detector. The two γ-rays that deposited only a fraction of their energy have a high probability of being absorbed in this second detector. Data are collected in the PELAN detector only when there is no corresponding signal from this second detector (anti-coincidence mode). Since this second detector vetoes data in the PELAN detector, we are describing it as a “veto detector”.

Figure 2 shows a veto detector designed for the PELAN project. The PELAN detector is a 7.6 cm x 7.6 cm (3”x3”) BGO detector. The veto detector is also composed of BGO with a thickness of 2.5 cm (1”) with an inner diameter just over 7.6 cm

• Figure 2. The PELAN detector surrounded by the veto detector.

The veto detector is comprised of four optically isolated pieces of BGO. Each piece has two photomultiplier tubes in order to efficiently gather the light from the scintillations. In practice, the signals from the pair of the photomultipliers from each piece are tied together. There is no attempt to gather position information from the veto detector.

An anti-coincidence circuit for the veto detector has also been designed. The timing resolution between PELAN’s detector and the veto detector is approximately 10 ns when using a ²²Na γ-ray source.

The veto detector is not fully annular and only completes three-quarters of the circumference. Tests completed during the design phase of the veto detector construction indicated that a complete annulus was not required. This is due to the shielding intrinsic to PELAN which is sufficient to limit the γ-ray noise emanating from the neutron generator.

Figure 3 shows two γ-ray spectra: the dashed line spectrum is taken without the veto detector, and the solid line spectrum is taken with the veto detector surrounding the PELAN detector. The shielding effect of the veto detector is apparent, since the solid line spectrum is lower than the dashed line one. In the region around 5.1 MeV, the second escape peak of the 6.1 MeV γ-ray is totally absent from the spectrum taken with the veto detector.

• Figure 3. The spectrum from the PELAN detector in anti-coincidence mode with veto detector (solid line) and a spectrum taken with the PELAN detector alone. Both spectra were taken with the detector near the ground.

The veto detector does indeed reduce the “noise” generated by the background. However, the change in the SNR was equivalent to the SNR produced by surrounding the BGO with a 1” thick piece of Pb. We could not justify the expense and the complexity in utilizing the veto detector.

A third way is to change the geometry of the detector. The gamma rays that PELAN analyzes range in energy from 1.0 MeV to 8 MeV. These are highly penetrating and require a long free mean path within the BGO for full energy deposition in the crystal. If the geometry is changed so that the cross-sectional area of the detector is smaller than the length of the detector, then the higher energy gamma rays will deposit preferentially their full energy along the long axis of the detector (see Figure 4). A large detector can then be created having a close-packed array of these detectors. To suppress scattering, any event in which two or more detectors are in coincidence is vetoed out. In this manner, the solid angle is maintained and gamma rays in the direction of the long axis are preferentially selected. We refer to this design as a “segmented” detector.

• Figure 4. Segmented detector.

However, there are several technical problems associated with a segmented detector. First, the resolution of the detector decreases due to the fact that smaller diameter photo-multiplier tubes have a poorer light collection efficiency. Because of this limitation, the segments could not have a diameter much smaller than 2.5 cm (1”) with a 7.6 cm (3”) length. Second, the detectors must be in close proximity to increase the efficiency of the anti-coincidence circuit. Finally, compact PC-based electronics must be designed.

A proof of principle test was performed utilizing 4 NaI detectors which matched the desired shape. They were connected to an anti-coincidence circuit. The analysis of the effectiveness of this detector is still ongoing.

For PELAN, we are still seeking an improved detection system. We desire improvements in directionality which we believe will change the SNR ratio and eliminate the problems caused by clutter.

Acknowledgements: This work was supported in part by the Department of Defense under contracts DAAD07-98-C-0116, DAAD05-98-C-022 and DACA72-01-C-0017

For more detailed information, please see the Optimizing the Signal to Noise Ratio for the PELAN System Publications portion of this web site.