- Email: [email protected]
An investigation of HPGe gamma efficiency calibration software (ANGLE V.3) for applications in nuclear decommissioning
Applied Radiation and Isotopes 70 (2012) 2737–2741
Contents lists available at SciVerse ScienceDirect
Applied Radiation and Isotopes journal homepage: www.elsevier.com/locate/apradiso
An investigation of HPGe gamma efficiency calibration software (ANGLE V.3) for applications in nuclear decommissioning S.J. Bell a,n, S.M. Judge b, P.H. Regan a a b
Department of Physics, University of Surrey, Guildford, Surrey, GU2 7XH, UK National Physical Laboratory, Hampton Road, Teddington, Middlesex, TW11 0LW, UK
H I G H L I G H T S c c c c c
The efficiency transfer software, ANGLE V.3, has been assessed. Computed efficiency calibration curves were compared to experimental results. A variety of geometries and matrices was investigated. A new approach is suggested for correcting results for inaccurate detector spec. The software was found fit-for-purpose for use in a nuclear decommissioning lab.
a r t i c l e i n f o
abstract
Article history: Received 14 June 2012 Accepted 11 August 2012 Available online 23 August 2012
High resolution gamma spectrometry offers a rapid method to characterise waste materials on a decommissioning nuclear site. To meet regulatory requirements, measurements must be traceable to national standards, meaning that the spectrometers must be calibrated for a wide range of materials. Semi-empirical modelling software (such as ANGLETM) offers a convenient method to carry out such calibrations. This paper describes an assessment of the modelling software for use by a small laboratory based on a nuclear site. The results confirmed the need for accurate information on the detection construction if the calibration were to be accurate to within 10%. & 2012 Elsevier Ltd. All rights reserved.
Keywords: ANGLE Efficiency calibration Efficiency transfer Nuclear decommissioning Gamma spectrometry
1. Introduction Gamma spectrometry has an important role to play in supporting the decommissioning of civil nuclear power plants. Waste materials from buildings or obsolete plant have to be characterised for radionuclide content so that a decision can be taken on the safe disposal route (Nuclear industry code of practice (NICoP), 2005). High-resolution gamma spectrometry offers a rapid, non-destructive method to measure the principal radionuclides (such as 137Cs, 60Co and 241Am) found on nuclear sites, to inform such decisions. The gamma spectrometers must be calibrated using certificated reference materials so that results are traceable to primary radioactivity standards (Dean et al., 2007). Due to self-absorption of gamma radiation, reference materials should match the composition, density and format of the samples to be measured. However, a wide variety of materials has to be measured, such as building materials, pipework, exhaust stacks and low density
insulation. Few suitable reference materials are available and calibrating the spectrometers for a wide range of materials is both time-consuming and expensive. Modelling software offers an alternative method to direct calibration of the spectrometers; the technique is time efficient and can be adapted for any sample delivered to the laboratory for measurement. The method has been adopted by commercial instrument companies, which has the advantage that the calculations are an integral part of the analysis software, simplifying routine operation. This paper summarise the results from an experimental verification of one commercial software package (ANGLETM, distributed by ORTEC). A brief summary is given of the algorithm and the results from calculations are compared to experimental data. A new approach is suggested for correcting results for incomplete or inaccurate information on the detector construction which is a key input to the software.
2. Detector efficiency calculation algorithms n
Corresponding author. Tel.: þ44 79 04 338 996. E-mail addresses: [email protected], [email protected] (S.J. Bell), [email protected] (S.M. Judge), [email protected] (P.H. Regan). 0969-8043/$ - see front matter & 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.apradiso.2012.08.007
The detection efficiency of a typical high resolution germanium spectrometer is a function of gamma energy. The efficiency
2738
S.J. Bell et al. / Applied Radiation and Isotopes 70 (2012) 2737–2741
calibration depends on the sample (composition, density, geometry and position) and the construction of the spectrometer itself, such as the shape of the germanium crystal, detector housing materials and thickness. Three different approaches have been used to model gamma spectrometers. The most sophisticated method is to use a MonteCarlo simulation of the measurement; this requires accurate information on the spectrometer construction, which can involve X-ray radiography of the instrument. The second method is semiempirical, combining experimental data for the detection efficiency for one type of sample with an analytical model of the detector (often referred to as efficiency transfer). The third technique uses a simple point detector model, but requires that the instrument is calibrated using a reference source in the same container and geometry as the sample to be measured (GAMATOOL, www.hightechsource.co.uk/Legacy/Resources/Geometry.pdf). ANGLETM is an example of the second method, which has the advantage that detailed information on the spectrometer construction is not needed. An initial calibration is required which can be achieved using readily available point source or solution standards. The algorithm was developed by (Jovanovic´ et al., 1997, 2010; ANGLE, 2012) based on earlier work by Moens et al. (1981) and assumes that the photopeak detection efficiency is proportional to the attenuation of gamma radiation in the detector crystal (excluding coherent scattering), taking into account attenuation in any intervening materials (sample matrix, container, detector window etc.). This assumption was supported first by experimental data (Moens et al., 1981) and later by Monte Carlo modelling (Vidmar et al., 2001). Previous work to verify efficiency calibration codes have included a comprehensive study by Lepy et al. (2001). This study compared software-produced efficiency curves to experimental results for a number of geometries, for example, varying the source-detector distance for point sources and using cylindrical vials with different matrices, varying in density from 0.25 g cm 3 to 1.54 g cm 3. The full Monte Carlo simulations gave the best results, with accuracies better than 5%, however the semiempirical model was also within 7% of the experimental result. The results highlighted the importance of accurate information on the detector dimensions. In a similar way, Abbas et al. (2002) compared ANGLETM and a full Monte Carlo programme (LabSOCS) to experimental results. Again, results from the Monte Carlo simulation were closer than the semi-empirical model to the experimental data, but most results were within 10% and the authors concluded that both codes would be reliable for applications requiring this level of accuracy. A recent study by Vidmar et al. (2010) compared dedicated software packages and general purpose Monte Carlo codes but did not include experimental data. The conclusion was that the results from the different codes were broadly equivalent except at low energies, where differences in the cross section data used may be responsible for larger discrepancies. For nuclear decommissioning, waste material is usually assessed by measuring the radionuclide content (Bq/g) in a batch of samples taken from random locations. The results are used to estimate the true mean activity and to compare the result to a regulatory limit. The spread in the results is dominated by the variation in the samples themselves, so an accuracy of 5–10% in the detector calibration is acceptable for most applications. The cost of disposal of radioactive waste normally depends on the total activity, so higher uncertainties could result in significantly higher costs if site operators allow a margin of error on stating the activity. A semi-empirical code, such as ANGLETM, therefore could offer a practical ‘fit-for purpose’ approach, but for a regulatory application the software requires verification before it can be used.
3. Experimental verification To test the ANGLETM software package, detection efficiency calibration curves produced using the model were compared to experimental results from reference materials using common sample dimensions and a typical HPGe detector, similar to instruments available on most nuclear sites. Samples were measured in typical containers used commonly in the UK nuclear industry: 250 ml bottles and 20 ml vials. These containers were not specifically designed for gamma spectrometry and did not have well specified dimensions. The dimensions of the containers were therefore determined by measurement using calibrated callipers. The containers were also not simple cylinders, and some judgement had to be used to model the dimensions in the software. Assumptions also had to be made on the composition of the container material; for example, ‘plastic’ was taken to be the default setting of 57% carbon, 9% hydrogen, 34% oxygen with a density of 0.94 g cm 3. The instrument used in the study was an Ortec GMX series coaxial HPGe, mounted vertically within a 10 cm thick lead shield. The initial detector dimensions were obtained from the manufacturer and are shown in Fig. 1. The starting point for the calculations by ANGLETM is a calibration using a reference source of known gamma emission rate. Three sources were used: a mixed radionuclide point reference source (product code QCD1, supplied by Eckert & Zeigler Isotope Products Inc. (EZ)), and aqueous standards in a 250 ml bottle and a 20 ml vial, prepared from a mixed radionuclide standardised solution (QCY48, also supplied by EZ). The QCY48 solution was used to prepare the other test materials used in these measurements: a low density 250 ml source was prepared by depositing drops of the solution onto cloth placed in layers inside the bottle, and a high density source was prepared in a 20 ml vial by mixing the solution with sodium sulphate. Initial results using the manufacturer’s data for the detector dimensions showed large discrepancies between the measured and calculated detection efficiencies. An example is shown in Fig. 2, which compares the measured detection efficiency for a 250 ml bottle to the computed detection efficiency based on a point source calibration. Discrepancies of up to 31% were observed, which is unlikely to be acceptable for use in the industry. As pointed out by Lepy et al. (2001), quoted detector dimensions can be unreliable; this risks restricting the usefulness of the software. A sensitivity analysis was carried out, varying aspects of the detector construction to determine the important parameters. It was found that the calculations depended mainly on the distance between the detector window and the Ge crystal, so it was decided to conduct a simple measurement to check the accuracy of this parameter using the inverse square law. A series of measurements was therefore carried out with a point reference source at different distances from the detector along the axis of symmetry of the detector. The effective centre of the detector was calculated using the inverse square law, and the offset was plotted as a function of gamma energy and extrapolated to zero energy (at zero energy, the effective centre of the detector is the top face of the crystal). The results indicated a distance of 3.4 cm between the top face of the detector and the surface of the germanium crystal compared with the manufacturer’s specification of 0.5 cm. This result is subject to high uncertainties, but casts doubt on the accuracy of the manufacturer’s value. The distance between the top face of the detector and the crystal was therefore fine-tuned by ‘trial and error’. A setting of 1.2 cm was found to give the best results; a summary is given in Table 1 of a comparison between calculated and measured detection efficiencies between 88 keV and 1836 keV (further details are given in Bell, 2010).
S.J. Bell et al. / Applied Radiation and Isotopes 70 (2012) 2737–2741
2739
Fig. 1. Detector specifications, provided by ORTEC, for the GMX series n-type HPGe detector used during the experiment.
The reference calibrations used to calculate the results presented in Figs. 3–5 were created by positioning a point source at three different heights above the detector (close, mid and far). A notable degree of variation was observed in the results calculated from these reference geometries, particularly at energies below 662 keV. It is likely that an optimum reference calibration exists for a particular sample geometry and that potentially extensive experimental verification would be required to determine which reference calibration this is. An alternative where time-consuming experimental verification is not possible would be to average the efficiency curves calculated from a range of reference calibrations. At the cost of moderately degrading the accuracy across the whole energy range, this approach would remove the most extreme inaccuracies (for example, at the low energy range).
The ANGLETM software also allows the user to select a volumetric source as the reference calibration in place of a point source at a known distance. This is likely to be a convenient approach for small laboratories that may not have access to a point reference source or accurate source positioning jigs. The tests were repeated using the 250 ml and 20 ml aqueous standards as the calibration sources, but with inconsistent results. Good agreement (generally better than 5%) was found if the 250 ml calibration was used to estimate the calibration for the 20 ml vial (and vice versa) but discrepancies of up to 40% were found applying the same calibrations to the high and low density sources. In practice, a typical germanium crystal is not a perfect cylinder, the top circular edge of the crystal is rounded off so
S.J. Bell et al. / Applied Radiation and Isotopes 70 (2012) 2737–2741
that the crystal resembles a bullet. It has been suggested that this ‘bulletisation’ of the detector crystal, if not accurately modelled, may lead to discrepancies between the computed and measured efficiency values (Gasparro, 2008; Vidmar and Gasparro, 2009). The detector used in the current study was indeed bulletised, however this was reflected in the detector input parameters (a new feature in ANGLETM V.3, 2012). Other authors have found the detector crystal to be tilted within the cryostat (for example, Lepy et al., 2001). It is not known whether the detector used in the current study was titled as it was not radiographed, however it has been previously shown that tilting up to 51 has no significant influence on computed efficiency values (Gasparro, 2008).
8 6 Relative Deviation (%)
2740
4 2 0
0
200
400
600
Close Point Source
-4
Mid Point Source
-6
Far Point Source
-8
Semi-empirical detector efficiency modelling codes (such as ANGLETM) have the potential to produce results that are fit-forpurpose for measurements to support the nuclear decommissioning industry. The code was easy to use and implement in an industrial setting with limited resources. The efficiency calibration was more accurate if a point source was used as the reference calibration rather than a volumetric calibration (even with the variation observed from one point source position to another). Uncertainties related to the dilution and matrix of the volumetric sources were likely to be responsible for this, however further work is needed to investigate the reasons for the larger (40%) discrepancies. This study has confirmed the work of previous authors that accurate information on the detector construction is essential, and users of the software would be advised to carry out some basic verification of the calibration. This verification should include an assessment of which reference calibration is most suited for a particular sample geometry. Where this is not possible, an alternative would be to take the average efficiency curve calculated from a range of reference calibrations.
Energy (keV)
Fig. 3. The relative deviation of the computed values from the experimental values for a 250 ml bottle containing an aqueous standard homogeneously distributed throughout the volume. Close, mid, and far point source geometries have been used as a reference to calculate the results.
8 6 Relative Deviation (%)
4. Conclusions and recommendations for future work
800 1000 1200 1400 1600 1800 2000
-2
4 2 0 -2
0
200
400
600
800 1000 1200 1400 1600 1800 2000
-4 -6
Close Point Source Mid Point Source Far Point Source
-8 -10 -12
Energy (keV)
Fig. 4. The relative deviation of the computed values from the experimental values for a 20 ml vial containing an aqueous standard homogeneously distributed throughout the volume. Close, mid, and far point source geometries have been used as a reference to calculate the results.
6
0.04
Experimental (250ml Bottle)
4
Computed (Ref: Far Point Source)
2
0.03 0.02 0.01 0.00
Relative Deviation (%)
Efficiency (no units)
0.05
0 -2
0
200
400
600
800 1000 1200 1400 1600 1800 2000
-4 -6 Close Point Source
-8
Mid Point Source
-10
Far Point Source
-12 10
100
1000
Energy (keV)
10000
Energy (keV) Fig. 2. The computed efficiency curve produced by the far point source reference using the manufacturer’s detector specification verbatim.
Fig. 5. The relative deviation of the computed values from the experimental values for a 250 ml bottle containing layered cloth, with an aqueous standard homogeneously distributed throughout the volume. Close, mid, and far point source geometries have been used as a reference to calculate the results.
Table 1 A summary of the agreement found between calculated and measured detection efficiencies for the range of geometries and matrices investigated. Sample geometry
Matrix
Reference calibration used
Agreement between calculated and measured detection efficiency (%)
250 ml bottle 20 ml vial 250 ml bottle 20 ml vial
Aqueous Aqueous Low density (0.15 g cm 3) High density (2.3 g cm 3)
Point Point Point Point
Generally Generally Generally Generally
source source source source
better better better better
than than than than
6 (see Fig. 3) 7 (see Fig. 4) 6 (see Fig. 5) 10 (see Fig. 6)
Relative Deviation (%)
S.J. Bell et al. / Applied Radiation and Isotopes 70 (2012) 2737–2741
2741
15
References
10
Abbas, K., et al., 2002. Reliability of two calculation codes for efficiency calibration of HPGe detectors. Appl. Radiat. Isotopes 56, 703–709. ANGLE 3 – Semiconductor detector efficiency calculations software, 2012. /http:// www.dlabac.com/angle/home.htmlS. Jovanovic´, S., & Dlabac, A. Bell, S.J., 2010. An investigation and verification of ANGLE V.3: HPGe gamma efficiency calibration software for application within the nuclear decommissioning industry MSc Dissertation. University of Surrey, Guildford, UK. Dean, J.C.J., Adsley, I., Burgess, P.H., 2007. Traceability for measurements of radioactivity in waste materials arising from nuclear site decommissioning. Metrologia 44, S140–145. GAMATOOL. /www.hightechsource.co.uk/Legacy/Resources/Geometry.pdfS. High Technology Sources Ltd. Gasparro, J., 2008. Monte Carlo modelling of germanium crystals that are tilted and have rounded front edges. Nucl. Instrum. Methods Phys. Res. A 594, 196–201. Jovanovic´, S., et al., 1997. ANGLE: a PC-code for semiconductor detector efficiency calculations. J. Radioanal. Nucl. Chem. 218 (1), 13–20. Jovanovic´, S., et al., 2010. Angle V2.1: New version of the computer code for semiconductor detector gamma-efficiency calculations. Nucl. Instrum. Methods Phys. Res. A 622 (2), 385–391. Lepy, M.-C., et al., 2001. Intercomparison of efficiency transfer software for gamma-ray spectrometry. Appl. Radiat. Isotopes 55, 493–503. Moens, L., et al., 1981. Calculation of the absolute peak efficiency of gamma-ray detectors for different counting geometries. Nucl. Instrum. Methods 187, 451–472. Nuclear industry code of practice (NICoP), 2005. /http://www.cewg.co.uk/images/ uncontrolled-issue-1.01-nuclear-industry-code-of-practice-aug-2006.pdfS. Stoyell, L. Vidmar, T., et al., 2001. A physically founded model of the efficiency curve in gamma-ray spectrometry. J. Phys. D: Appl. Phys. 34, 2555–2560. Vidmar, T., Gasparro, J., 2009. Crystal rounding and the efficiency transfer method in gamma-ray spectrometry. Appl. Radiat. Isotopes 67 (11), 2057–2061. Vidmar, T., et al., 2010. Testing efficiency transfer codes for equivalence. Appl. Radiat. Isotopes 68, 355–359.
5 0
0
200
400
600
800 1000 1200 1400 1600 1800 2000
-5 -10
Close Point Source Mid Point Source
-15 -20
Far Point Source Energy (keV)
Fig. 6. The relative deviation of the computed values from the experimental values for a high density 20 ml vial containing sodium sulphate/aqueous standard mixture. Close, mid, and far point source geometries have been used as a reference to calculate the results.
Acknowledgements The authors would like to thank Adam Wright (Magnox Ltd.) for his help in carrying out the study, Trevor Hatt (Ortec), Slobodan Jovanovic´ (University of Montenegro), and Aleksandar Dlabac (University of Montenegro) for many useful discussions. Support from the UK’s Nuclear Decommissioning Authority (through the EmPower scheme) is gratefully acknowledged. Thanks are also due to Magnox Ltd, Dungeness A Site, Romney Marsh, Kent TN29 9PP for hosting the work.
Contents lists available at SciVerse ScienceDirect
Applied Radiation and Isotopes journal homepage: www.elsevier.com/locate/apradiso
An investigation of HPGe gamma efficiency calibration software (ANGLE V.3) for applications in nuclear decommissioning S.J. Bell a,n, S.M. Judge b, P.H. Regan a a b
Department of Physics, University of Surrey, Guildford, Surrey, GU2 7XH, UK National Physical Laboratory, Hampton Road, Teddington, Middlesex, TW11 0LW, UK
H I G H L I G H T S c c c c c
The efficiency transfer software, ANGLE V.3, has been assessed. Computed efficiency calibration curves were compared to experimental results. A variety of geometries and matrices was investigated. A new approach is suggested for correcting results for inaccurate detector spec. The software was found fit-for-purpose for use in a nuclear decommissioning lab.
a r t i c l e i n f o
abstract
Article history: Received 14 June 2012 Accepted 11 August 2012 Available online 23 August 2012
High resolution gamma spectrometry offers a rapid method to characterise waste materials on a decommissioning nuclear site. To meet regulatory requirements, measurements must be traceable to national standards, meaning that the spectrometers must be calibrated for a wide range of materials. Semi-empirical modelling software (such as ANGLETM) offers a convenient method to carry out such calibrations. This paper describes an assessment of the modelling software for use by a small laboratory based on a nuclear site. The results confirmed the need for accurate information on the detection construction if the calibration were to be accurate to within 10%. & 2012 Elsevier Ltd. All rights reserved.
Keywords: ANGLE Efficiency calibration Efficiency transfer Nuclear decommissioning Gamma spectrometry
1. Introduction Gamma spectrometry has an important role to play in supporting the decommissioning of civil nuclear power plants. Waste materials from buildings or obsolete plant have to be characterised for radionuclide content so that a decision can be taken on the safe disposal route (Nuclear industry code of practice (NICoP), 2005). High-resolution gamma spectrometry offers a rapid, non-destructive method to measure the principal radionuclides (such as 137Cs, 60Co and 241Am) found on nuclear sites, to inform such decisions. The gamma spectrometers must be calibrated using certificated reference materials so that results are traceable to primary radioactivity standards (Dean et al., 2007). Due to self-absorption of gamma radiation, reference materials should match the composition, density and format of the samples to be measured. However, a wide variety of materials has to be measured, such as building materials, pipework, exhaust stacks and low density
insulation. Few suitable reference materials are available and calibrating the spectrometers for a wide range of materials is both time-consuming and expensive. Modelling software offers an alternative method to direct calibration of the spectrometers; the technique is time efficient and can be adapted for any sample delivered to the laboratory for measurement. The method has been adopted by commercial instrument companies, which has the advantage that the calculations are an integral part of the analysis software, simplifying routine operation. This paper summarise the results from an experimental verification of one commercial software package (ANGLETM, distributed by ORTEC). A brief summary is given of the algorithm and the results from calculations are compared to experimental data. A new approach is suggested for correcting results for incomplete or inaccurate information on the detector construction which is a key input to the software.
2. Detector efficiency calculation algorithms n
Corresponding author. Tel.: þ44 79 04 338 996. E-mail addresses: [email protected], [email protected] (S.J. Bell), [email protected] (S.M. Judge), [email protected] (P.H. Regan). 0969-8043/$ - see front matter & 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.apradiso.2012.08.007
The detection efficiency of a typical high resolution germanium spectrometer is a function of gamma energy. The efficiency
2738
S.J. Bell et al. / Applied Radiation and Isotopes 70 (2012) 2737–2741
calibration depends on the sample (composition, density, geometry and position) and the construction of the spectrometer itself, such as the shape of the germanium crystal, detector housing materials and thickness. Three different approaches have been used to model gamma spectrometers. The most sophisticated method is to use a MonteCarlo simulation of the measurement; this requires accurate information on the spectrometer construction, which can involve X-ray radiography of the instrument. The second method is semiempirical, combining experimental data for the detection efficiency for one type of sample with an analytical model of the detector (often referred to as efficiency transfer). The third technique uses a simple point detector model, but requires that the instrument is calibrated using a reference source in the same container and geometry as the sample to be measured (GAMATOOL, www.hightechsource.co.uk/Legacy/Resources/Geometry.pdf). ANGLETM is an example of the second method, which has the advantage that detailed information on the spectrometer construction is not needed. An initial calibration is required which can be achieved using readily available point source or solution standards. The algorithm was developed by (Jovanovic´ et al., 1997, 2010; ANGLE, 2012) based on earlier work by Moens et al. (1981) and assumes that the photopeak detection efficiency is proportional to the attenuation of gamma radiation in the detector crystal (excluding coherent scattering), taking into account attenuation in any intervening materials (sample matrix, container, detector window etc.). This assumption was supported first by experimental data (Moens et al., 1981) and later by Monte Carlo modelling (Vidmar et al., 2001). Previous work to verify efficiency calibration codes have included a comprehensive study by Lepy et al. (2001). This study compared software-produced efficiency curves to experimental results for a number of geometries, for example, varying the source-detector distance for point sources and using cylindrical vials with different matrices, varying in density from 0.25 g cm 3 to 1.54 g cm 3. The full Monte Carlo simulations gave the best results, with accuracies better than 5%, however the semiempirical model was also within 7% of the experimental result. The results highlighted the importance of accurate information on the detector dimensions. In a similar way, Abbas et al. (2002) compared ANGLETM and a full Monte Carlo programme (LabSOCS) to experimental results. Again, results from the Monte Carlo simulation were closer than the semi-empirical model to the experimental data, but most results were within 10% and the authors concluded that both codes would be reliable for applications requiring this level of accuracy. A recent study by Vidmar et al. (2010) compared dedicated software packages and general purpose Monte Carlo codes but did not include experimental data. The conclusion was that the results from the different codes were broadly equivalent except at low energies, where differences in the cross section data used may be responsible for larger discrepancies. For nuclear decommissioning, waste material is usually assessed by measuring the radionuclide content (Bq/g) in a batch of samples taken from random locations. The results are used to estimate the true mean activity and to compare the result to a regulatory limit. The spread in the results is dominated by the variation in the samples themselves, so an accuracy of 5–10% in the detector calibration is acceptable for most applications. The cost of disposal of radioactive waste normally depends on the total activity, so higher uncertainties could result in significantly higher costs if site operators allow a margin of error on stating the activity. A semi-empirical code, such as ANGLETM, therefore could offer a practical ‘fit-for purpose’ approach, but for a regulatory application the software requires verification before it can be used.
3. Experimental verification To test the ANGLETM software package, detection efficiency calibration curves produced using the model were compared to experimental results from reference materials using common sample dimensions and a typical HPGe detector, similar to instruments available on most nuclear sites. Samples were measured in typical containers used commonly in the UK nuclear industry: 250 ml bottles and 20 ml vials. These containers were not specifically designed for gamma spectrometry and did not have well specified dimensions. The dimensions of the containers were therefore determined by measurement using calibrated callipers. The containers were also not simple cylinders, and some judgement had to be used to model the dimensions in the software. Assumptions also had to be made on the composition of the container material; for example, ‘plastic’ was taken to be the default setting of 57% carbon, 9% hydrogen, 34% oxygen with a density of 0.94 g cm 3. The instrument used in the study was an Ortec GMX series coaxial HPGe, mounted vertically within a 10 cm thick lead shield. The initial detector dimensions were obtained from the manufacturer and are shown in Fig. 1. The starting point for the calculations by ANGLETM is a calibration using a reference source of known gamma emission rate. Three sources were used: a mixed radionuclide point reference source (product code QCD1, supplied by Eckert & Zeigler Isotope Products Inc. (EZ)), and aqueous standards in a 250 ml bottle and a 20 ml vial, prepared from a mixed radionuclide standardised solution (QCY48, also supplied by EZ). The QCY48 solution was used to prepare the other test materials used in these measurements: a low density 250 ml source was prepared by depositing drops of the solution onto cloth placed in layers inside the bottle, and a high density source was prepared in a 20 ml vial by mixing the solution with sodium sulphate. Initial results using the manufacturer’s data for the detector dimensions showed large discrepancies between the measured and calculated detection efficiencies. An example is shown in Fig. 2, which compares the measured detection efficiency for a 250 ml bottle to the computed detection efficiency based on a point source calibration. Discrepancies of up to 31% were observed, which is unlikely to be acceptable for use in the industry. As pointed out by Lepy et al. (2001), quoted detector dimensions can be unreliable; this risks restricting the usefulness of the software. A sensitivity analysis was carried out, varying aspects of the detector construction to determine the important parameters. It was found that the calculations depended mainly on the distance between the detector window and the Ge crystal, so it was decided to conduct a simple measurement to check the accuracy of this parameter using the inverse square law. A series of measurements was therefore carried out with a point reference source at different distances from the detector along the axis of symmetry of the detector. The effective centre of the detector was calculated using the inverse square law, and the offset was plotted as a function of gamma energy and extrapolated to zero energy (at zero energy, the effective centre of the detector is the top face of the crystal). The results indicated a distance of 3.4 cm between the top face of the detector and the surface of the germanium crystal compared with the manufacturer’s specification of 0.5 cm. This result is subject to high uncertainties, but casts doubt on the accuracy of the manufacturer’s value. The distance between the top face of the detector and the crystal was therefore fine-tuned by ‘trial and error’. A setting of 1.2 cm was found to give the best results; a summary is given in Table 1 of a comparison between calculated and measured detection efficiencies between 88 keV and 1836 keV (further details are given in Bell, 2010).
S.J. Bell et al. / Applied Radiation and Isotopes 70 (2012) 2737–2741
2739
Fig. 1. Detector specifications, provided by ORTEC, for the GMX series n-type HPGe detector used during the experiment.
The reference calibrations used to calculate the results presented in Figs. 3–5 were created by positioning a point source at three different heights above the detector (close, mid and far). A notable degree of variation was observed in the results calculated from these reference geometries, particularly at energies below 662 keV. It is likely that an optimum reference calibration exists for a particular sample geometry and that potentially extensive experimental verification would be required to determine which reference calibration this is. An alternative where time-consuming experimental verification is not possible would be to average the efficiency curves calculated from a range of reference calibrations. At the cost of moderately degrading the accuracy across the whole energy range, this approach would remove the most extreme inaccuracies (for example, at the low energy range).
The ANGLETM software also allows the user to select a volumetric source as the reference calibration in place of a point source at a known distance. This is likely to be a convenient approach for small laboratories that may not have access to a point reference source or accurate source positioning jigs. The tests were repeated using the 250 ml and 20 ml aqueous standards as the calibration sources, but with inconsistent results. Good agreement (generally better than 5%) was found if the 250 ml calibration was used to estimate the calibration for the 20 ml vial (and vice versa) but discrepancies of up to 40% were found applying the same calibrations to the high and low density sources. In practice, a typical germanium crystal is not a perfect cylinder, the top circular edge of the crystal is rounded off so
S.J. Bell et al. / Applied Radiation and Isotopes 70 (2012) 2737–2741
that the crystal resembles a bullet. It has been suggested that this ‘bulletisation’ of the detector crystal, if not accurately modelled, may lead to discrepancies between the computed and measured efficiency values (Gasparro, 2008; Vidmar and Gasparro, 2009). The detector used in the current study was indeed bulletised, however this was reflected in the detector input parameters (a new feature in ANGLETM V.3, 2012). Other authors have found the detector crystal to be tilted within the cryostat (for example, Lepy et al., 2001). It is not known whether the detector used in the current study was titled as it was not radiographed, however it has been previously shown that tilting up to 51 has no significant influence on computed efficiency values (Gasparro, 2008).
8 6 Relative Deviation (%)
2740
4 2 0
0
200
400
600
Close Point Source
-4
Mid Point Source
-6
Far Point Source
-8
Semi-empirical detector efficiency modelling codes (such as ANGLETM) have the potential to produce results that are fit-forpurpose for measurements to support the nuclear decommissioning industry. The code was easy to use and implement in an industrial setting with limited resources. The efficiency calibration was more accurate if a point source was used as the reference calibration rather than a volumetric calibration (even with the variation observed from one point source position to another). Uncertainties related to the dilution and matrix of the volumetric sources were likely to be responsible for this, however further work is needed to investigate the reasons for the larger (40%) discrepancies. This study has confirmed the work of previous authors that accurate information on the detector construction is essential, and users of the software would be advised to carry out some basic verification of the calibration. This verification should include an assessment of which reference calibration is most suited for a particular sample geometry. Where this is not possible, an alternative would be to take the average efficiency curve calculated from a range of reference calibrations.
Energy (keV)
Fig. 3. The relative deviation of the computed values from the experimental values for a 250 ml bottle containing an aqueous standard homogeneously distributed throughout the volume. Close, mid, and far point source geometries have been used as a reference to calculate the results.
8 6 Relative Deviation (%)
4. Conclusions and recommendations for future work
800 1000 1200 1400 1600 1800 2000
-2
4 2 0 -2
0
200
400
600
800 1000 1200 1400 1600 1800 2000
-4 -6
Close Point Source Mid Point Source Far Point Source
-8 -10 -12
Energy (keV)
Fig. 4. The relative deviation of the computed values from the experimental values for a 20 ml vial containing an aqueous standard homogeneously distributed throughout the volume. Close, mid, and far point source geometries have been used as a reference to calculate the results.
6
0.04
Experimental (250ml Bottle)
4
Computed (Ref: Far Point Source)
2
0.03 0.02 0.01 0.00
Relative Deviation (%)
Efficiency (no units)
0.05
0 -2
0
200
400
600
800 1000 1200 1400 1600 1800 2000
-4 -6 Close Point Source
-8
Mid Point Source
-10
Far Point Source
-12 10
100
1000
Energy (keV)
10000
Energy (keV) Fig. 2. The computed efficiency curve produced by the far point source reference using the manufacturer’s detector specification verbatim.
Fig. 5. The relative deviation of the computed values from the experimental values for a 250 ml bottle containing layered cloth, with an aqueous standard homogeneously distributed throughout the volume. Close, mid, and far point source geometries have been used as a reference to calculate the results.
Table 1 A summary of the agreement found between calculated and measured detection efficiencies for the range of geometries and matrices investigated. Sample geometry
Matrix
Reference calibration used
Agreement between calculated and measured detection efficiency (%)
250 ml bottle 20 ml vial 250 ml bottle 20 ml vial
Aqueous Aqueous Low density (0.15 g cm 3) High density (2.3 g cm 3)
Point Point Point Point
Generally Generally Generally Generally
source source source source
better better better better
than than than than
6 (see Fig. 3) 7 (see Fig. 4) 6 (see Fig. 5) 10 (see Fig. 6)
Relative Deviation (%)
S.J. Bell et al. / Applied Radiation and Isotopes 70 (2012) 2737–2741
2741
15
References
10
Abbas, K., et al., 2002. Reliability of two calculation codes for efficiency calibration of HPGe detectors. Appl. Radiat. Isotopes 56, 703–709. ANGLE 3 – Semiconductor detector efficiency calculations software, 2012. /http:// www.dlabac.com/angle/home.htmlS. Jovanovic´, S., & Dlabac, A. Bell, S.J., 2010. An investigation and verification of ANGLE V.3: HPGe gamma efficiency calibration software for application within the nuclear decommissioning industry MSc Dissertation. University of Surrey, Guildford, UK. Dean, J.C.J., Adsley, I., Burgess, P.H., 2007. Traceability for measurements of radioactivity in waste materials arising from nuclear site decommissioning. Metrologia 44, S140–145. GAMATOOL. /www.hightechsource.co.uk/Legacy/Resources/Geometry.pdfS. High Technology Sources Ltd. Gasparro, J., 2008. Monte Carlo modelling of germanium crystals that are tilted and have rounded front edges. Nucl. Instrum. Methods Phys. Res. A 594, 196–201. Jovanovic´, S., et al., 1997. ANGLE: a PC-code for semiconductor detector efficiency calculations. J. Radioanal. Nucl. Chem. 218 (1), 13–20. Jovanovic´, S., et al., 2010. Angle V2.1: New version of the computer code for semiconductor detector gamma-efficiency calculations. Nucl. Instrum. Methods Phys. Res. A 622 (2), 385–391. Lepy, M.-C., et al., 2001. Intercomparison of efficiency transfer software for gamma-ray spectrometry. Appl. Radiat. Isotopes 55, 493–503. Moens, L., et al., 1981. Calculation of the absolute peak efficiency of gamma-ray detectors for different counting geometries. Nucl. Instrum. Methods 187, 451–472. Nuclear industry code of practice (NICoP), 2005. /http://www.cewg.co.uk/images/ uncontrolled-issue-1.01-nuclear-industry-code-of-practice-aug-2006.pdfS. Stoyell, L. Vidmar, T., et al., 2001. A physically founded model of the efficiency curve in gamma-ray spectrometry. J. Phys. D: Appl. Phys. 34, 2555–2560. Vidmar, T., Gasparro, J., 2009. Crystal rounding and the efficiency transfer method in gamma-ray spectrometry. Appl. Radiat. Isotopes 67 (11), 2057–2061. Vidmar, T., et al., 2010. Testing efficiency transfer codes for equivalence. Appl. Radiat. Isotopes 68, 355–359.
5 0
0
200
400
600
800 1000 1200 1400 1600 1800 2000
-5 -10
Close Point Source Mid Point Source
-15 -20
Far Point Source Energy (keV)
Fig. 6. The relative deviation of the computed values from the experimental values for a high density 20 ml vial containing sodium sulphate/aqueous standard mixture. Close, mid, and far point source geometries have been used as a reference to calculate the results.
Acknowledgements The authors would like to thank Adam Wright (Magnox Ltd.) for his help in carrying out the study, Trevor Hatt (Ortec), Slobodan Jovanovic´ (University of Montenegro), and Aleksandar Dlabac (University of Montenegro) for many useful discussions. Support from the UK’s Nuclear Decommissioning Authority (through the EmPower scheme) is gratefully acknowledged. Thanks are also due to Magnox Ltd, Dungeness A Site, Romney Marsh, Kent TN29 9PP for hosting the work.