Optimization of Pulse Processing Parameters for HPGe Gamma-Ray Spectroscopy

The use of HPGe detectors is chosen in situations where isotope identification is required due to their excellent resolution. In various circumstances, it is necessary to carry out these gamma-ray spectroscopy measurements in areas with high gamma-ray flux which means high counting rates, for example, nuclear medicine, radiochemistry, safeguards, and neutron activation analysis.

In other applications, a broad dynamic range in count rate can be encountered, for instance, samples obtained after a nuclear accident are calculated on a system that is normally used for monitoring environment. The count rate could be reduced by increasing the source-detector distance, shielding or collimation, or reducing the sample size in a laboratory counting situation. However, it may not be possible to lower count-rates by any of these means in a real situation. The challenge is to attain the “best” data possible in every measurement situation. “Best” is a combination of spectral quality (peak, width and position) and statistical (number of counts) considerations over a broad range of count rates.

The development of multichannel analyzers (MCA) through digital signal processing (DSP), where the detector signal is converted to a digital signal directly from an analog signal at the preamplifier output, allows many new ways of processing the signal without most of the compromises and approximations needed in analog signal processing ¹. It has produced a much broader range of values for shaping times (termed flattop, rise time and fall time). Furthermore, the processing of the detector signal can be performed in different ways to enhance the resolution or full-width at half-maximum (FWHM) performance, by measuring the preamplifier pulse and after that using the measurement to choose a range of digital filter to perform the pulse-by-pulse adjustments on the preamplifier pulse.

The FWHM of the spectrum peak or net full energy peak depends on the shaping time. A long shaping time includes too much signal noise and a short shaping time does not include all the preamplifier pulse ². The best and the smallest FWHM can be achieved with the shaping time that is suited to the detector preamplifier output signal. The total pulse processing time increases directly with the shaping time. The throughput is defined at the ratio of the pulses in the spectrum to the total number of gamma rays that enter the detector. The pulse processing time is basically the dead time or the time that the MCA is incapable of collecting the next pulse. The throughput is linked to the dead time and so the pulse processing time. Although longer shaping times normally produce better peak resolution, they mean larger dead times and lower throughput.

This article describes how two HPGe detectors (relative efficiency of 20% and 95%) were measured in order to determine the FWHM resolution of a low and high energy peak for the whole range of rise times and flattops. This range is much wider compared to the previously available analog systems and completely covers the range of HPGe detector output signals. The newest ORTEC DSP MCA, the DSPEC 50, was used to collect the data.

Nuclides and Gamma Rays

Detectors and Electronics

The low-efficiency HPGe detector was a GEM (p-type) detector of 56 mm diameter and 34 mm length for a relative efficiency of ~20%. The high-efficiency HPGe detector was a GEM detector of 79 mm diameter and 77 mm length for a relative efficiency of ~95%. Both the low-efficiency and high-efficiency HPGe detectors were mounted in horizontal cryostats and cooled with liquid nitrogen.

Using the same DSPEC 50, the detectors were measured one at a time. The DSPEC 50 supplies the low voltage power, high voltage bias, spectrum memory, and DSP. This detector is connected to the controlling computer via Ethernet.

The rise time ranges from 0.8 to 23.0 µs and the flattop ranges from 0.3 to 2.4 µs. Figure 1 shows the filter with rise time and flattop defined.

Figure 1. Rise time and Flattop definition

Although preamplifier pulse amplitude is proportional to the gamma ray energy, the shape, i.e., amplitude vs time, depends on the properties of the detector. Larger detectors likely to have longer total pulse widths than smaller detectors. Figure 2 shows the comparison of the preamplifier output pulses from the two detectors.

Figure 2. Preamplifier pulses from 20% and 95% HPGe

Sources

The sources used on this small detector were ⁵⁷Co and ⁶⁰Co and for the large detector, ¹⁰⁹Cd and ⁶⁰Co were used. All are point sources, and they were placed on axis before the detector at a distance to achieve reasonable count rate, i.e., in the range of 15 to 25 cm from the front of the endcap. The dead time was about 15% for the longest shaping time for each detector. The source-detector geometry continued to be stable for the entire data collection for each detector.

Methods

The data was collected for both flat top times and rise time covering the minimum to the maximum. Overall, 11 flat tops and 28 rise times were used. GammaVision was used to automate the data collection process. The peak area uncertainty was about 0.4% for the small detector and 0.5% for the large detector. The peak area, FW1/25M (full width at one twenty-fifth maximum) and FWHM for each peak was measured using the IEEE 325³ method as applied in GammaVision⁴. The width of the region for determining the background was selected to be relatively wide to reduce variations of the FWHM with calculation width.

Figure 3 shows the comparison of the 122 keV peak from the small detector for different flat tops and rise times. The low amplitude peak is for a flat top of 0.3 µs and rise time of 0.8 µs, the shortest times possible. The higher peak sitting on the base line is for a flat top of 1.0 µs and rise time of 5.0 µs, the low end of the uniform resolution region. The middle peak is for a flat top of 2.4 µs and rise time of 23.0 µs, the maximum possible values for each.

Figure 3. Comparison of spectrum peak shapes for different shaping times

Results

Resolution

FW1/25M and FWHM of the 122 keV peak for the 20% detector are shown in Figure 4. The shape of both curves follow the predictions in Ref 2, where the longer shaping times are displayed to include a greater fraction of noise. The 122 keV gamma rays will have only a single interaction in the crystal, because the charge pulse is short. The FWHM displays only a slight (~20%) increase, while the FW1/25M displays a larger increase (~35%), demonstrating that the added parallel noise is small in amplitude.

Figure 4. Resolution at 122 keV vs rise time for many flat top times (FWHM and FW1/25M) for 20% HPGe

The same data for the 95% detector at 88 keV is shown in Figure 5. The general dependence of shape on time is similar to the smaller detector but with the minimum occurring at higher times, which is fully consistent with the longer charge collection time in larger crystals. The short flattop of the 95% detector is very short for the low amplitude (low-energy gamma ray) detector pulse, which is covered by the longer rise times.

Figure 5. Resolution at 88 keV vs rise time for many flat top times (FWHM and FW1/25M) for 95% HPGe

The resolution data for the 1.33 MeV peak for the 20% detector is shown in Figure 6. The 1.33 MeV gamma rays will have a number of interactions in the crystal to deposit the entire energy and give rise to longer charge collection time. The FWHM displays some increase at short times, particularly the shortest flat top, while the FW1/25M displays more of an increase at the shorter times. Above a flat top of 1.0 µs, a little improvement is seen in either the FWHM or the FW1/25M. The rise time has minimum impact on the FWHM above ~ 4 µs. The FW1/25M has the best values at ~4 to 5 µs, however, it shows a small and steady increase starting at a rise time of ~ 13 µs.

Figure 6. Resolution at 1.33 MeV vs rise time for many flat top times (FWHM and FW1/25M) for 95% HPGe

Figure 7 shows the same data for the 95% detector. FWHM data for the combination of shorter rise time and flat top is not shown, but it displays the similar trends as the FW1/25M data in this region. For this detector with longer charge collection times, the FW1/25M and FWHM are large for short flat top times (below ~1 µs). Above a flat top time of 1 µs, the resolution depends on the rise time. The FW1/25M changes less than 7.5% from 5 µs to 23 µs, and the FWHM changes less than 6% from 4 µs to 23 µs. Using the rise time and flat top time at the lower end of the acceptable resolution area will reduce the processing time per pulse. The reduced dead time will improve the throughput and reduce the signal loss from random summing or pileup. This shows that the shorter shaping times can be used in almost all counting situations as the resultant peak shapes are nearly the same.

Figure 7. Resolution at 1.33 MeV vs rise time for many flat top times (FWHM and FW1/25M) for 20% HPGe

Throughput

The DSPEC 50’s throughput was measured for the “traditional” timing and for a number of values of protection time in the throughput enhancer mode for the shaping time at the low end of the acceptable range (rise time of 6.0 µs and flat top of 1.0 µs) and for the shortest shaping time possible (rise time of 0.8 µs and flat top of 0.6 µs).

Figure 8 shows the throughput vs input count rate for the shortest times. The improvement with shorter protection times needs to be noted. The resolution for the same set of parameters is shown in Figure 9. Also the FWHM does not differ with count rate or protection time over a broad range of dead times.

Figure 8. Throughput for different protection times as a function of dead time with Rise Time = 0.8 and Flat top = 0.6 µs.

Figure 9. Resolution at 1.33 MeV for different dead times and protection times

Both the resolution and throughput are useful at the rise time and flat top values at the low end of the acceptable range (rise time of 6.0 µs and flattop of 1.0 µs). The throughput as a function of input count rate is shown in Figure 10. It should be noted that the maximum throughput peaks at input count rates of approximately 40 kcps for traditional timing and 80 kcps for the minimum protection time. This is an increase of ~1.6 in throughput.

Figure 10. Throughput vs ICR

The FWHM and FW1/25M resolution for several pulse processing times from minimum to maximum is shown in Figure 11. Both FWHM and FW1/25M are not affected by the protection time. The FWHM does not begin to increase until the dead time reaches 95%, while the FW1/25M starts to increase at about 80%.

Figure 11. Resolution vs dead time for various protection times

The dead time vs input count rate is shown in Figure 12. Comparing the achieved results shown in Figures 10 and 11 with Figure 12, the maximum throughput at the shortest protection time is within the limits for the better resolution at both FWHM and FW1/25M.

Figure 12. Dead time vs input count rate for various protection times

Conclusion

According to the IEEE 325 standard, the the HPGe resolution should be measured at the 1.33 MeV peak of ⁶⁰Co with an analog shaping time of 6 µs and low count rates, which equivalent to digital pulse processing results using a 12 µs digital filter.

First published in 1971, the IEEE 325 standard was dependent on the technology of that particular time. Although this specification is a valid measure for the performance of the detector under the test conditions, it is not adequate to predict the performance of the detector in other operating conditions, and the 6 µs analog shaping time specification could be sub-optimal for any application of the existing detectors and MCAs. In order to obtain the system for a particular application, it is essential to specify performance parameters at values relevant to the application, which could be different compared to those specified in standards.

The resolution data reveals that in DSP MCAs, there is a lower limit of pulse shaping times (rise time and flat top time) dependent on the energy of the gamma rays and the size of the detector through which it possible to acquire better quality peak shapes. This may be accurately determined in contrast to earlier analog systems. Longer shaping times on small detectors can increase peak width and on large detectors do little to enhance low rate performance, but lower system maximum throughput at high count rates. Reducing the shaping time, which provides acceptable low count rate resolution, increases the dynamic range of count rate that can be achievable, and maximum throughput can be enhanced further by using throughput enhancement techniques.

The data also reveals that DSP MCAs have better performance (good throughput and resolution) over a broad range of shaping times and pulse processing times. This indicates that DSP MCAs can be set up to operate at low and high count rates without changing any adjustments and without reducing resolution performance or data collection at any count rate.

References

[1] Ronald Keyser, Timothy Twomey, and Russell Bingham, “Improved Performance in Germanium Detector Gamma Spectrometers based on Digital Signal Processing,” Proceedings of the ANS Fall meeting, Washington, D.C., November, 2004.

[2] Ron Jenkins, R. W. Gould, and Dale Gedcke, “Quantitative X-ray Spectrometry”, p 147, 1995, Marcel Dekker, Inc., New York

[3] IEEE Standard Test Procedures for Germanium Gamma-Ray Detectors Used in Digital Signal Processing Systems, The Institute of Electrical and Electronics Engineers, Inc., New York, NY 10017-2394

[4] GammaVision Operators Manual, ORTEC, Oak Ridge, TN

This information has been sourced, reviewed and adapted from materials provided by ORTEC.

For more information on this source, please visit ORTEC.