- Email: [email protected]
Study of the response of an ORTEC GMX45 HPGe detector with a multi-radionuclide volume source using Monte Carlo simulations
Applied Radiation and Isotopes 113 (2016) 47–52
Contents lists available at ScienceDirect
Applied Radiation and Isotopes journal homepage: www.elsevier.com/locate/apradiso
Study of the response of an ORTEC GMX45 HPGe detector with a multiradionuclide volume source using Monte Carlo simulations A. Saraiva a, C. Oliveira a, M. Reis a,n, L. Portugal a,b, I. Paiva a, C. Cruz a a b
Instituto Superior Técnico, Universidade de Lisboa, Estrada Nacional 10 (km 139.7), 2695-066 Bobadela, LRS, Portugal Agência Portuguesa do Ambiente, Rua da Murgueira, 9/9ª-Zambujal, 2611-865 Amadora, Portugal
H I G H L I G H T S
High resolution gamma-ray spectrometry. Computation of efficiencies by using Monte Carlo methods. Use of traceable multi-gamma volume sources to validate the MC model. Enabling the calculations of true coincidence summing effects and efficiency transfer factors for geometry deviations.
art ic l e i nf o
a b s t r a c t
Article history: Received 27 October 2015 Received in revised form 17 April 2016 Accepted 20 April 2016 Available online 22 April 2016
A model of an n-type ORTEC GMX45 HPGe detector was created using the MCNPX and the MCNP-CP codes. In order to validate the model, experimental efficiency was compared with the Monte Carlo simulations results. The reference source is a NIST traceable multi-gamma volume source in a waterequivalent epoxy resin matrix (1.15 g cm 3 density) containing several radionuclides: 210Pb, 241Am, 137Cs and 60Co in a cylinder shape container. Two distances of source bottom to end cap front surface of the detector have been considered. The efficiency for the nearest distance is higher than for longer distance. The relative difference between the measured and the simulated full-energy peak efficiency is less than 4.0% except for the 46.5 keV energy peak of 210Pb for the longer distance (6.5%) allowing to consider the model validated. In the absence of adequate standard calibration sources, efficiency and efficiency transfer factors for geometry deviations and matrix effects can be accurately computed by using Monte Carlo methods even if true coincidence could occur as is the case when the 60Co radioisotope is present in the source. & 2016 Elsevier Ltd. All rights reserved.
Keywords: ORTEC GMX detector Monte Carlo simulation Efficiency transfer factors
1. Introduction An ORTEC GMX45 HPGe detector was recently acquired by the radionuclide laboratory of CTN (Campus Tecnológico e Nuclear) to measure radioactivity content in NORM samples, radioactive waste samples, radioactivity (natural and anthropogenic) content in industrial samples, as for example in metallurgic industry and any radioactive contaminated material. A good performance of a gamma spectrometry system depends among other performances on the quality of its calibration, and in particular the efficiency calibration. This can and is normally done experimentally using activity standards (commonly called calibration sources). But having calibration sources for all the geometries and matrices needed may not be economically and physically possible thus the use of Monte Carlo methods for efficiency determinations comes n
Corresponding author.
http://dx.doi.org/10.1016/j.apradiso.2016.04.016 0969-8043/& 2016 Elsevier Ltd. All rights reserved.
up as simple, little expensive and reliable solution. The purpose of this work is to create the model of the HPGe detector and validate it comparing experimental efficiencies with efficiencies determined by Monte Carlo simulations. This work is part of the European Metrology Research Programme (EMRP) project IND04 “Ionising Radiation Metrology for the Metallurgical Industry” (MetroMetal). The general aim of the project is to improve radioactivity measurements in metallurgical industry at each stage of the smelting process (Solc et al., 2015). A large set of works have been published using the Monte Carlo simulations of the detector for several purposes, namely for optimizing processes or for studying the influence of some parameters on the results or even for characterization of the detector system (Luis et al., 2010; Li et al., 2015; Stefanakis, 2014; Dryak and Kovar, 2006; Solc et al., 2015). The goal of this work is different from those works and relies on having a model of the detector as accurate as possible in order to be able to determine, using Monte Carlo simulations of the system sample-detector, the efficiency for
48
A. Saraiva et al. / Applied Radiation and Isotopes 113 (2016) 47–52
a given energy when there are no adequate standard radioactive calibration sources or in terms of energy and/or in terms of activity, or to calculate the efficiency transfer factor due to differences in geometry (Solc et al., 2015) and/or density and chemical composition between standards and samples to measure (matrix effect) and/or to determine the coincidence-summing corrections factors.
2. Monte Carlo simulation codes There are a set of Monte Carlo codes that have been used by several authors to perform simulations of the detectors. MCNP family codes, EGSnrc, GEANT3, GEANT4, GESPECOR, and PENELOPE are the most used codes for this purpose. The MCNPX (Pelowitz, 2011) and the MCNP-CP (Berlizov, 2007) codes were used in this work to simulate the detector based on the established model. The model and the simulation calculus was validated by comparing calculated and experimental full energy efficiencies in the 46.5–1332.5 keV energy range for a multiradionuclide volume source and two source-to-detector distances. The MCNP-CP code uses the full decay scheme including the gamma emission probability and provides a radionuclide cascade source of correlated photons as well as versatile capabilities for modeling the response of a detector with true coincidence summing. So, for this reason it becomes possible to compare directly the experimental results with the simulations results even in situations where there are true coincidence summing.
3. Detector geometrical parameters For detector simulation purposes it is necessary to have the full
characterization of the detector in order to be able to design the model of the detector and use it in the Monte Carlo code. The detector to be characterized was an n-type ORTEC GMX45 Hyper Pure Germanium (HPGe) detector with a nominal 52.5% relative efficiency and resolution of 1.9 keV (for 1332.5 keV Co-60 photons). The generic characterization of the detector provided by the manufacturer is usually insufficient to create the model, since some of the parameters are unknown or not related to the specific detector but instead averaged over a set of detectors of the same type. The study of this detector has confirmed that the measured values differ from nearly all of the values established by the manufacturer. Thus, a full characterization of an HPGe detector is not a straightforward procedure. To reach this goal it was necessary to get some information about this specific detector namely to take some radiographies and to determine the value of some relevant parameters. Additionally, the knowledge of the dead layers thickness raises some problems because they were not directly measurable. The value adopted is taken from the technical report (ORTEC, 2003). Two radiographies, obtained using an X-ray beam of 120 kV, are shown in Fig. 1. The complete setup of the detector includes a rectangular lead shielding with 5 cm thickness, copper lining of 0.2 cm thickness and an inner volume of 49.6 49.6 69.6 cm3. The list of the measured parameters taken from the radiographies is shown in Table 1, along with the values that were provided by the manufacturer and the values adopted in the Monte Carlo model. Twenty one parameters were considered. Fig. 2 shows the physical detector parameters. Of the five parameters that was possible to measure directly on the radiograph there are two values lower and three values higher than the values provided by the manufacturer. The higher difference between them was found in the “End cap to crystal gap”. The value from the radiography is higher more than 50% than the value
Fig. 1. Detector radiographies obtained with a 120 keV X-ray. Different details are shown in the radiographies: (a) Besides the Ge crystal, it is visible the aluminum end-cap wall; (b) On the top it is visible the small bars used as a scale to calibrate the measuring system integrated in the X-ray image processing.
A. Saraiva et al. / Applied Radiation and Isotopes 113 (2016) 47–52
Table 1 Detector geometrical parameters. Parameter (mm)a
Manufacturer Radiography
MC input
Detector-crystal diameter Detector-crystal length Detector-crystal end radius Hole diameter Hole depth Hole bottom radius Mount cup length End cap to crystal gap Mount cup base End cap window thickness Insulator/Shield Outside contact layer (mm) Hole contact layer Mount cup wall End cap wall End cap wall to mount cup wall gap Superior dead layer (mm) Lateral dead layer Inferior dead layer Hole dead layer End cap window diameter
63.9
63.077 0.50
63.07
70.2
69.077 0.50
a b c
69.07 c
8.9 61.9 4.45 105 4
h ¼9.56 70.50 ; w¼ 4.0770.50 – – – – 6.56 7 0.50
h ¼9.56; w¼ 4.07 8.9 60.77 4.45 105 6.56
3.2 0.5
– –
3.2 0.5
0.05 0.3
– –
0.05 0.3
0.9 0.76 1.0 –
– 1.077 0.50 1.337 0.50 7.107 0.50
0.9 1.35 1.0 7.37
0.3
–
0.3
0.5 0.7 0.9 71.44b 70.05
– – –
0.5 0.7 0.9 71.44
8
The outside contact layer and the superior dead layer are presented in mm. Value measured with a caliper. See Fig. 2.
49
values. This was done by changing the values of those parameters in order to bring the simulated efficiencies to as close as possible to the measured efficiencies in a way that the consistency of the model was not compromised. The uncertainty associated to each parameter takes into account the uncertainty associated to the calibration of the measurement device. The used voltage was too low to allow the visualization of the hole structure in the crystal.
4. The model The model describes a high-purity germanium coaxial detector with a thin beryllium window based on a model of ORTEC GMX series inside a shielding (Fig. 3). The detector is in vacuum, and it is constituted by three parts: the sensitive part of the germanium crystal, the dead layer (a not active germanium layer) and the detector assembly. The dead layer is lining the whole crystal: in the top, in the lateral sides, in the bottom and inside the hole. The detector assembly consists of an end-cap wall with a window and a mount cup. The end-cap wall is made of aluminum, with a beryllium window. The mount-cup has a lateral wall, a base and a window. The window covers all the length of the mount cup. The shielding is composed by three layers. The inside layer of copper (0.1 cm), a layer of lead in the middle (7 cm thickness on the bottom and lateral sides and 5 cm thickness on the top) and an iron layer at the bottom (0.2 cm thickness). The shielding has a circular hole at the bottom. The Monte Carlo model was created to work with the MCNPX code, version 2.7.0., and also with the MCNP-CP code. The used source is a NIST traceable multi-gamma volume source in a water-equivalent epoxy resin matrix with 1.15 g cm 3 density. It is a cylinder shape source with 3.39 cm radius and 4.76 cm height. Two distances of source bottom to end cap front surface of the detector have been considered, being the shortest 4.54 mm and the longer 26.12 mm. In Fig. 4 the MC models mimic the experimental configuration with the source positioned at the closer distance and at the longer distance.
5. Monte Carlo simulations
Fig. 2. Physical detector parameters.
provided by the manufacturer which clearly shows the importance of measuring these parameters. The quasi symmetry of the values (number of values higher (3) and number of values lower (2) than the values provided by the manufacturer) makes useless a model designed only with the manufacturer values because these wrong measures may originate contributions with opposite effects and the final result can reveal a good agreement with experimental values. However these results are based on a wrong model. So, a rigorous selection process of the geometric parameters values of the detector is needed. It should be noted that there are two parameter values (mount cup wall and gap between end cap wall and mount cup wall), that are different from both the measured values and the manufacturer
The Monte Carlo simulation code used was the MCNP-CP, version 3.2. The MCNP-CP code is a MCNP-based code created by (Berlizov, 2007). The MCNP-CP code uses the full decay scheme including the gamma emission probability and provides a radionuclide cascade source of correlated particles as well as versatile capabilities for modeling the response of a detector with true coincidence summing. The comprehensive ENSDF – Evaluated Nuclear Structure Data File is used by MCNP-CP as the source of information about decay properties of radionuclides to perform a full-scale simulation of the radioactive decay process, which yields random composition (quantity and type) and probabilistic characteristics (energies, times, directional cosines) of a set of emitted correlated nuclear particles. The code samples photons according to the decay scheme table and sums up the total absorbed energy in the detector from all photons emitted in one decay. The crosssection, form-factor, and fluorescence data are the same as the data used by the standard MCNP code and were all derived from the ENDF/B-VI.8, which is in turn based upon the EPDL97 library (Cullen et al., 1997). The calculated efficiency is based on the counts which are calculated using the F8 tally. Only the transport of photons was considered. The cut-off energy for both photons and electrons is 3 keV. In order to understand which parts of the HPGe crystal were important in the detection process, a plot of the energy deposition
50
A. Saraiva et al. / Applied Radiation and Isotopes 113 (2016) 47–52
Fig. 3. (a) The Monte Carlo model; (b) Detail of the torus: h ¼ 9.56 mm, w¼ 4.07 mm.
Fig. 4. Monte Carlo model with the volume source. (a) Source positioned at 4.54 mm (nearest position); (b) Source positioned at 26.12 mm (farthest position).
A. Saraiva et al. / Applied Radiation and Isotopes 113 (2016) 47–52
210
Fig. 5. Energy deposition on the detector for: (a) 46.5 keV energy of
in the detector was generated using the MCNPX, version 2.7.0 and the MCNPX Visual Plotter Version X_24E-2. For this study the model with a single radioisotope source and the nearest distance was used. The results for 46.5 keV photons of 210Pb and 1332.5 keV photons of 60Co are shown in Fig. 5. Each point represent the local where the photon as a given interaction. It is possible to observe that different parts of the detector are involved in the local deposition of the energy. For the lower energy practically only the superficial layer of the crystal is involved in the process whereas for the higher energy there is a distribution of the energy deposition in the whole volume of the detector, which means that the shaping details of the top of the crystal could be important for the efficiency, in particular for the lower energies.
6. Efficiency measurements The used multi-gamma volume source contains several radionuclides: 210Pb, 241Am, 137Cs and 60Co, emitting on the gamma energies of 46.5 keV, 59.5 keV, 661.7 keV and 1173.2 and 1332.5 keV, respectively. The activity at measurement time was 4.6 kBq, 0.41 kBq, 0.14 kBq and 0.14 kBq, respectively. Two sourceto-detector distances, 4.54 mm and 26.12 mm, were used. The acquisition time was 72 h, for both source-to-detector distances, and 60 h for the background. In what regards the MC calculated efficiency, the number of the photons simulated (NPS) was 2.8 108 and 3.4 108 respectively for the shorter and the longer distances.
51
Pb; (b) 1332.5 keV energy of
60
Co.
7. Results The experimental efficiency and the MC simulation efficiency are shown in Tables 2 and 3. For the determination of the experimental efficiency, the uncertainty components considered were activity and peak area counting statistics. For the determination of the simulated efficiency, the statistical uncertainty, the activity uncertainty and the emission probability uncertainty have been considered. The statistical uncertainty (type A uncertainty) is calculated by MC code and it is one of the outputs of the code. In order to estimate the uncertainty associated to the efficiency calculated by MCNP-CP due the uncertainty on the activity and on the emission probability, a sensitivity study has been performed for the nearest source-to-detector distance. These values were considered also valid for the longest distances. When the statistical uncertainty is also included, the uncertainty increases lightly. All uncertainties are determined for k ¼1. Its values are included in the Tables 2 and 3. As it can be observed, the efficiency for the nearest distance is higher than for longer distance by a factor between 1.67 and 1.83, for experimental results or by a factor between 1.76 and 1.91, for simulation results. The relative difference between the measured and the simulated full-energy peak efficiency is shown in Tables 2 and 3 with the associated uncertainties calculated with the square root of the sum of the squares of the uncertainties associated with experimental and simulated efficiencies. The relative difference is equal or less than 4.0% except for the 46.5 keV energy peak of 210Pb for the longer distance, allowing to consider
Table 2 Experimental and simulated efficiencies for the shorter distance. 210
Energy (keV) Experimental efficiency, εexp (counts Bq 1 s 1) Activity uncertainty (%) Peak area uncertainty (%) Combined std uncertainty (k¼ 1) (%) Monte Carlo efficiency, εMC (counts Bq 1 s 1) (NPS¼ 2.8 108) Statistical uncertainty (%) Combined uncertainty (%) (a) Combined std uncertainty (k¼ 1) (%) Deviation, (εMC-εexp )/εexp (%) Uncertainty (%)
Pb
46.5 2.106E-03 1.40 0.15 1.41 2.102E-03 0.13 0.28 0.31 0.2 1.4
241
137
60
60
59.5 1.535E-03 0.90 0.16 0.91 1.578E-03 0.15 0.95 0.96 2.9 1.3
661.7 3.662E-04 1.40 0.29 1.43 3.810E-04 0.30 1.50 1.53 4.0 2.1
1173.2 2.744E-04 1.30 0.33 1.34 2.854E-04 0.35 1.50 1.54 4.0 2.0
1332.5 2.536E-04 1.30 0.34 1.34 2.599E-04 0.37 1.50 1.54 2.5 2.0
Am
Cs
Co
Co
NPS-Number of photons emitted in one run of the MC code. a Combined standard relative uncertainties of uncertainty associated to the efficiency calculated by MCNP-CP due to the uncertainty on the activity and on the emission probability.
52
A. Saraiva et al. / Applied Radiation and Isotopes 113 (2016) 47–52
Table 3 Experimental and simulated efficiencies for the longer distance. 210
241
137
46.5 1.179E-03 1.40 0.18 1.41 1.102E-03 0.16 0.28 0.32 6.5 1.4
59.5 8.399E-04 0.90 0.22 0.93 8.323E-04 0.19 0.95 0.97 0.9 1.3
661.7 2.064E-04 1.40 0.39 1.45 2.096E-04 0.38 1.50 1.55 1.5 2.1
Pb
Energy (keV) Experimental efficiency,εexp (counts Bq 1 s 1) Activity uncertainty (%) Peak area uncertainty (%) Combined std uncertainty (%) Monte Carlo efficiency,εMC (counts Bq 1 s 1) (NPS ¼ 3.4 108) Statistical uncertainty (%) Combined uncertainty (%)a Combined std uncertainty (%) Deviation, (εMC-εexp )/εexp (%) Uncertainty (%)
Am
Cs
60
60
1173.2 1.640E-04 1.30 0.45 1.38 1.625E-04 0.43 1.50 1.56 0.9 2.1
1332.5 1.499E-04 1.30 0.45 1.38 1.478E-04 0.45 1.50 1.57 1.4 2.1
Co
Co
a Combined standard relative uncertainties of uncertainty associated to the efficiency calculated by MCNP-CP due to the uncertainty on the activity and on the emission probability.
that the model of the detector is validated and, consequently, the detector is well characterized. This model of the detector was used to calculate by MCNP-CP the true coincidence summing effects for standards of cast steel, slag and fume dust. The results were presented in Solc et al., 2015 as MCNP-CP IST.
Union on the basis of Decision No. 912/2009/EC. The EMRP is jointly funded by the EMRP participating countries within EURAMET and the European Union.
References 8. Conclusions A model describing a high-purity germanium coaxial detector with a thin beryllium window based on a model of an n-type ORTEC GMX45P4 HPGe detector has been created. A good agreement between experimental and Monte Carlo simulations results was achieved for the two considered distances, confirming that the model corresponds to the real detector. This means that true coincidence summing effects can be determined and in a more general way, in the absence of adequate standard calibration sources, efficiency and efficiency transfer factors for geometry deviations and matrix effects can be accurately computed by using Monte Carlo methods even if true coincidence could occurs as is the case when the 60Co radioisotope is present in the source.
Acknowledgment The authors wish to thanks to IPOFG Lisboa the possibility to perform the radiographies on the detector and to José Afonso all the help to obtain the excellent set of radiographies. This work was supported by the EMRP joint research project “Ionising radiation metrology for the metallurgical industry” (MetroMetal) which has received funding from the European
Berlizov, A.N., 2007. MCNP-CP: a correlated particle radiation source extension of a general purpose Monte Carlo N-particle transport code. Institute for nuclear research, National Academy of Science of Ukraine 47, Prospekt Nauki, MSP 0380. Kyiv, Ukraine. In: Semkow, T. (Ed.), Applied Modeling and Computations in Nuclear Science 13. ACS Symposium Series; American Chemical Society, Washington, DC, pp. 183–194. Cullen, D.E., Hubbell, J.H., Kissel, L.D., 1997. EPDL97: The Evaluated Photon Data Library, ’97 Version Lawrence Livemore National Laboratory, UCRL-50400, (6), Rev. 5. Pelowitz, D.B., MCNPXTM, 2011. A General Monte Carlo N-Particle Transport Code. User's Manual, Version 2.7.0. Luis, R., Bento, J., Carvalhal, G., Nogueira, P., Silva, L., Teles, P., Vaz, P., 2010. Parameter optimization of a planar BEGe detector using Monte Carlo simulations. Nucl. Instrum. Methods Phys. Res. A 623, 1014–1019. Li, G., Bai, L., Xu, J., Liang, Y., 2015. Optimization of cylindrical sample geometry in determining the radioactive activity of environmental soil samples by using a Broad Energy Germanium detector. J. Instrum. 10. Stefanakis, N., 2014. Detection Efficiency of a BEGe Detector using the Monte Carlo Method and a Comparison to other Calibration Methods, November. 〈http:// xxx.tau.ac.il/pdf/1411.4225.pdf〉. Dryak, P., Kovar, P., 2006. Experimental and MC determination of HPGe detector efficiency in the 40–2754 keV energy range for measuring point source geometry with the source-to-detector distance of 25 cm. Appl. Radiat. Isot. 64, 1346–1349. Solc, J., Dryák, P., Moser, H., Branger, T., Garcia-Toraño, E., Peyrés, V., Tzika, F., Lutter, G., Capogni, M., Fazio, A., Luca, A., Vodnik, B., Oliveira, C., Saraiva, A., Szucs, L., Dziel, T., Burda, O., Arnold, D., Martinkovic, J., Siiskonen, T., Mattila, A., 2015. Characterisation of a radionuclide specific laboratory detector system for the metallurgical industry by Monte Carlo simulations. Radiat. Phys. Chem. 116, 189–193. ORTEC, 2003. The Best Choice of High Purity Germanium (HPGe) Detector.
Contents lists available at ScienceDirect
Applied Radiation and Isotopes journal homepage: www.elsevier.com/locate/apradiso
Study of the response of an ORTEC GMX45 HPGe detector with a multiradionuclide volume source using Monte Carlo simulations A. Saraiva a, C. Oliveira a, M. Reis a,n, L. Portugal a,b, I. Paiva a, C. Cruz a a b
Instituto Superior Técnico, Universidade de Lisboa, Estrada Nacional 10 (km 139.7), 2695-066 Bobadela, LRS, Portugal Agência Portuguesa do Ambiente, Rua da Murgueira, 9/9ª-Zambujal, 2611-865 Amadora, Portugal
H I G H L I G H T S
High resolution gamma-ray spectrometry. Computation of efficiencies by using Monte Carlo methods. Use of traceable multi-gamma volume sources to validate the MC model. Enabling the calculations of true coincidence summing effects and efficiency transfer factors for geometry deviations.
art ic l e i nf o
a b s t r a c t
Article history: Received 27 October 2015 Received in revised form 17 April 2016 Accepted 20 April 2016 Available online 22 April 2016
A model of an n-type ORTEC GMX45 HPGe detector was created using the MCNPX and the MCNP-CP codes. In order to validate the model, experimental efficiency was compared with the Monte Carlo simulations results. The reference source is a NIST traceable multi-gamma volume source in a waterequivalent epoxy resin matrix (1.15 g cm 3 density) containing several radionuclides: 210Pb, 241Am, 137Cs and 60Co in a cylinder shape container. Two distances of source bottom to end cap front surface of the detector have been considered. The efficiency for the nearest distance is higher than for longer distance. The relative difference between the measured and the simulated full-energy peak efficiency is less than 4.0% except for the 46.5 keV energy peak of 210Pb for the longer distance (6.5%) allowing to consider the model validated. In the absence of adequate standard calibration sources, efficiency and efficiency transfer factors for geometry deviations and matrix effects can be accurately computed by using Monte Carlo methods even if true coincidence could occur as is the case when the 60Co radioisotope is present in the source. & 2016 Elsevier Ltd. All rights reserved.
Keywords: ORTEC GMX detector Monte Carlo simulation Efficiency transfer factors
1. Introduction An ORTEC GMX45 HPGe detector was recently acquired by the radionuclide laboratory of CTN (Campus Tecnológico e Nuclear) to measure radioactivity content in NORM samples, radioactive waste samples, radioactivity (natural and anthropogenic) content in industrial samples, as for example in metallurgic industry and any radioactive contaminated material. A good performance of a gamma spectrometry system depends among other performances on the quality of its calibration, and in particular the efficiency calibration. This can and is normally done experimentally using activity standards (commonly called calibration sources). But having calibration sources for all the geometries and matrices needed may not be economically and physically possible thus the use of Monte Carlo methods for efficiency determinations comes n
Corresponding author.
http://dx.doi.org/10.1016/j.apradiso.2016.04.016 0969-8043/& 2016 Elsevier Ltd. All rights reserved.
up as simple, little expensive and reliable solution. The purpose of this work is to create the model of the HPGe detector and validate it comparing experimental efficiencies with efficiencies determined by Monte Carlo simulations. This work is part of the European Metrology Research Programme (EMRP) project IND04 “Ionising Radiation Metrology for the Metallurgical Industry” (MetroMetal). The general aim of the project is to improve radioactivity measurements in metallurgical industry at each stage of the smelting process (Solc et al., 2015). A large set of works have been published using the Monte Carlo simulations of the detector for several purposes, namely for optimizing processes or for studying the influence of some parameters on the results or even for characterization of the detector system (Luis et al., 2010; Li et al., 2015; Stefanakis, 2014; Dryak and Kovar, 2006; Solc et al., 2015). The goal of this work is different from those works and relies on having a model of the detector as accurate as possible in order to be able to determine, using Monte Carlo simulations of the system sample-detector, the efficiency for
48
A. Saraiva et al. / Applied Radiation and Isotopes 113 (2016) 47–52
a given energy when there are no adequate standard radioactive calibration sources or in terms of energy and/or in terms of activity, or to calculate the efficiency transfer factor due to differences in geometry (Solc et al., 2015) and/or density and chemical composition between standards and samples to measure (matrix effect) and/or to determine the coincidence-summing corrections factors.
2. Monte Carlo simulation codes There are a set of Monte Carlo codes that have been used by several authors to perform simulations of the detectors. MCNP family codes, EGSnrc, GEANT3, GEANT4, GESPECOR, and PENELOPE are the most used codes for this purpose. The MCNPX (Pelowitz, 2011) and the MCNP-CP (Berlizov, 2007) codes were used in this work to simulate the detector based on the established model. The model and the simulation calculus was validated by comparing calculated and experimental full energy efficiencies in the 46.5–1332.5 keV energy range for a multiradionuclide volume source and two source-to-detector distances. The MCNP-CP code uses the full decay scheme including the gamma emission probability and provides a radionuclide cascade source of correlated photons as well as versatile capabilities for modeling the response of a detector with true coincidence summing. So, for this reason it becomes possible to compare directly the experimental results with the simulations results even in situations where there are true coincidence summing.
3. Detector geometrical parameters For detector simulation purposes it is necessary to have the full
characterization of the detector in order to be able to design the model of the detector and use it in the Monte Carlo code. The detector to be characterized was an n-type ORTEC GMX45 Hyper Pure Germanium (HPGe) detector with a nominal 52.5% relative efficiency and resolution of 1.9 keV (for 1332.5 keV Co-60 photons). The generic characterization of the detector provided by the manufacturer is usually insufficient to create the model, since some of the parameters are unknown or not related to the specific detector but instead averaged over a set of detectors of the same type. The study of this detector has confirmed that the measured values differ from nearly all of the values established by the manufacturer. Thus, a full characterization of an HPGe detector is not a straightforward procedure. To reach this goal it was necessary to get some information about this specific detector namely to take some radiographies and to determine the value of some relevant parameters. Additionally, the knowledge of the dead layers thickness raises some problems because they were not directly measurable. The value adopted is taken from the technical report (ORTEC, 2003). Two radiographies, obtained using an X-ray beam of 120 kV, are shown in Fig. 1. The complete setup of the detector includes a rectangular lead shielding with 5 cm thickness, copper lining of 0.2 cm thickness and an inner volume of 49.6 49.6 69.6 cm3. The list of the measured parameters taken from the radiographies is shown in Table 1, along with the values that were provided by the manufacturer and the values adopted in the Monte Carlo model. Twenty one parameters were considered. Fig. 2 shows the physical detector parameters. Of the five parameters that was possible to measure directly on the radiograph there are two values lower and three values higher than the values provided by the manufacturer. The higher difference between them was found in the “End cap to crystal gap”. The value from the radiography is higher more than 50% than the value
Fig. 1. Detector radiographies obtained with a 120 keV X-ray. Different details are shown in the radiographies: (a) Besides the Ge crystal, it is visible the aluminum end-cap wall; (b) On the top it is visible the small bars used as a scale to calibrate the measuring system integrated in the X-ray image processing.
A. Saraiva et al. / Applied Radiation and Isotopes 113 (2016) 47–52
Table 1 Detector geometrical parameters. Parameter (mm)a
Manufacturer Radiography
MC input
Detector-crystal diameter Detector-crystal length Detector-crystal end radius Hole diameter Hole depth Hole bottom radius Mount cup length End cap to crystal gap Mount cup base End cap window thickness Insulator/Shield Outside contact layer (mm) Hole contact layer Mount cup wall End cap wall End cap wall to mount cup wall gap Superior dead layer (mm) Lateral dead layer Inferior dead layer Hole dead layer End cap window diameter
63.9
63.077 0.50
63.07
70.2
69.077 0.50
a b c
69.07 c
8.9 61.9 4.45 105 4
h ¼9.56 70.50 ; w¼ 4.0770.50 – – – – 6.56 7 0.50
h ¼9.56; w¼ 4.07 8.9 60.77 4.45 105 6.56
3.2 0.5
– –
3.2 0.5
0.05 0.3
– –
0.05 0.3
0.9 0.76 1.0 –
– 1.077 0.50 1.337 0.50 7.107 0.50
0.9 1.35 1.0 7.37
0.3
–
0.3
0.5 0.7 0.9 71.44b 70.05
– – –
0.5 0.7 0.9 71.44
8
The outside contact layer and the superior dead layer are presented in mm. Value measured with a caliper. See Fig. 2.
49
values. This was done by changing the values of those parameters in order to bring the simulated efficiencies to as close as possible to the measured efficiencies in a way that the consistency of the model was not compromised. The uncertainty associated to each parameter takes into account the uncertainty associated to the calibration of the measurement device. The used voltage was too low to allow the visualization of the hole structure in the crystal.
4. The model The model describes a high-purity germanium coaxial detector with a thin beryllium window based on a model of ORTEC GMX series inside a shielding (Fig. 3). The detector is in vacuum, and it is constituted by three parts: the sensitive part of the germanium crystal, the dead layer (a not active germanium layer) and the detector assembly. The dead layer is lining the whole crystal: in the top, in the lateral sides, in the bottom and inside the hole. The detector assembly consists of an end-cap wall with a window and a mount cup. The end-cap wall is made of aluminum, with a beryllium window. The mount-cup has a lateral wall, a base and a window. The window covers all the length of the mount cup. The shielding is composed by three layers. The inside layer of copper (0.1 cm), a layer of lead in the middle (7 cm thickness on the bottom and lateral sides and 5 cm thickness on the top) and an iron layer at the bottom (0.2 cm thickness). The shielding has a circular hole at the bottom. The Monte Carlo model was created to work with the MCNPX code, version 2.7.0., and also with the MCNP-CP code. The used source is a NIST traceable multi-gamma volume source in a water-equivalent epoxy resin matrix with 1.15 g cm 3 density. It is a cylinder shape source with 3.39 cm radius and 4.76 cm height. Two distances of source bottom to end cap front surface of the detector have been considered, being the shortest 4.54 mm and the longer 26.12 mm. In Fig. 4 the MC models mimic the experimental configuration with the source positioned at the closer distance and at the longer distance.
5. Monte Carlo simulations
Fig. 2. Physical detector parameters.
provided by the manufacturer which clearly shows the importance of measuring these parameters. The quasi symmetry of the values (number of values higher (3) and number of values lower (2) than the values provided by the manufacturer) makes useless a model designed only with the manufacturer values because these wrong measures may originate contributions with opposite effects and the final result can reveal a good agreement with experimental values. However these results are based on a wrong model. So, a rigorous selection process of the geometric parameters values of the detector is needed. It should be noted that there are two parameter values (mount cup wall and gap between end cap wall and mount cup wall), that are different from both the measured values and the manufacturer
The Monte Carlo simulation code used was the MCNP-CP, version 3.2. The MCNP-CP code is a MCNP-based code created by (Berlizov, 2007). The MCNP-CP code uses the full decay scheme including the gamma emission probability and provides a radionuclide cascade source of correlated particles as well as versatile capabilities for modeling the response of a detector with true coincidence summing. The comprehensive ENSDF – Evaluated Nuclear Structure Data File is used by MCNP-CP as the source of information about decay properties of radionuclides to perform a full-scale simulation of the radioactive decay process, which yields random composition (quantity and type) and probabilistic characteristics (energies, times, directional cosines) of a set of emitted correlated nuclear particles. The code samples photons according to the decay scheme table and sums up the total absorbed energy in the detector from all photons emitted in one decay. The crosssection, form-factor, and fluorescence data are the same as the data used by the standard MCNP code and were all derived from the ENDF/B-VI.8, which is in turn based upon the EPDL97 library (Cullen et al., 1997). The calculated efficiency is based on the counts which are calculated using the F8 tally. Only the transport of photons was considered. The cut-off energy for both photons and electrons is 3 keV. In order to understand which parts of the HPGe crystal were important in the detection process, a plot of the energy deposition
50
A. Saraiva et al. / Applied Radiation and Isotopes 113 (2016) 47–52
Fig. 3. (a) The Monte Carlo model; (b) Detail of the torus: h ¼ 9.56 mm, w¼ 4.07 mm.
Fig. 4. Monte Carlo model with the volume source. (a) Source positioned at 4.54 mm (nearest position); (b) Source positioned at 26.12 mm (farthest position).
A. Saraiva et al. / Applied Radiation and Isotopes 113 (2016) 47–52
210
Fig. 5. Energy deposition on the detector for: (a) 46.5 keV energy of
in the detector was generated using the MCNPX, version 2.7.0 and the MCNPX Visual Plotter Version X_24E-2. For this study the model with a single radioisotope source and the nearest distance was used. The results for 46.5 keV photons of 210Pb and 1332.5 keV photons of 60Co are shown in Fig. 5. Each point represent the local where the photon as a given interaction. It is possible to observe that different parts of the detector are involved in the local deposition of the energy. For the lower energy practically only the superficial layer of the crystal is involved in the process whereas for the higher energy there is a distribution of the energy deposition in the whole volume of the detector, which means that the shaping details of the top of the crystal could be important for the efficiency, in particular for the lower energies.
6. Efficiency measurements The used multi-gamma volume source contains several radionuclides: 210Pb, 241Am, 137Cs and 60Co, emitting on the gamma energies of 46.5 keV, 59.5 keV, 661.7 keV and 1173.2 and 1332.5 keV, respectively. The activity at measurement time was 4.6 kBq, 0.41 kBq, 0.14 kBq and 0.14 kBq, respectively. Two sourceto-detector distances, 4.54 mm and 26.12 mm, were used. The acquisition time was 72 h, for both source-to-detector distances, and 60 h for the background. In what regards the MC calculated efficiency, the number of the photons simulated (NPS) was 2.8 108 and 3.4 108 respectively for the shorter and the longer distances.
51
Pb; (b) 1332.5 keV energy of
60
Co.
7. Results The experimental efficiency and the MC simulation efficiency are shown in Tables 2 and 3. For the determination of the experimental efficiency, the uncertainty components considered were activity and peak area counting statistics. For the determination of the simulated efficiency, the statistical uncertainty, the activity uncertainty and the emission probability uncertainty have been considered. The statistical uncertainty (type A uncertainty) is calculated by MC code and it is one of the outputs of the code. In order to estimate the uncertainty associated to the efficiency calculated by MCNP-CP due the uncertainty on the activity and on the emission probability, a sensitivity study has been performed for the nearest source-to-detector distance. These values were considered also valid for the longest distances. When the statistical uncertainty is also included, the uncertainty increases lightly. All uncertainties are determined for k ¼1. Its values are included in the Tables 2 and 3. As it can be observed, the efficiency for the nearest distance is higher than for longer distance by a factor between 1.67 and 1.83, for experimental results or by a factor between 1.76 and 1.91, for simulation results. The relative difference between the measured and the simulated full-energy peak efficiency is shown in Tables 2 and 3 with the associated uncertainties calculated with the square root of the sum of the squares of the uncertainties associated with experimental and simulated efficiencies. The relative difference is equal or less than 4.0% except for the 46.5 keV energy peak of 210Pb for the longer distance, allowing to consider
Table 2 Experimental and simulated efficiencies for the shorter distance. 210
Energy (keV) Experimental efficiency, εexp (counts Bq 1 s 1) Activity uncertainty (%) Peak area uncertainty (%) Combined std uncertainty (k¼ 1) (%) Monte Carlo efficiency, εMC (counts Bq 1 s 1) (NPS¼ 2.8 108) Statistical uncertainty (%) Combined uncertainty (%) (a) Combined std uncertainty (k¼ 1) (%) Deviation, (εMC-εexp )/εexp (%) Uncertainty (%)
Pb
46.5 2.106E-03 1.40 0.15 1.41 2.102E-03 0.13 0.28 0.31 0.2 1.4
241
137
60
60
59.5 1.535E-03 0.90 0.16 0.91 1.578E-03 0.15 0.95 0.96 2.9 1.3
661.7 3.662E-04 1.40 0.29 1.43 3.810E-04 0.30 1.50 1.53 4.0 2.1
1173.2 2.744E-04 1.30 0.33 1.34 2.854E-04 0.35 1.50 1.54 4.0 2.0
1332.5 2.536E-04 1.30 0.34 1.34 2.599E-04 0.37 1.50 1.54 2.5 2.0
Am
Cs
Co
Co
NPS-Number of photons emitted in one run of the MC code. a Combined standard relative uncertainties of uncertainty associated to the efficiency calculated by MCNP-CP due to the uncertainty on the activity and on the emission probability.
52
A. Saraiva et al. / Applied Radiation and Isotopes 113 (2016) 47–52
Table 3 Experimental and simulated efficiencies for the longer distance. 210
241
137
46.5 1.179E-03 1.40 0.18 1.41 1.102E-03 0.16 0.28 0.32 6.5 1.4
59.5 8.399E-04 0.90 0.22 0.93 8.323E-04 0.19 0.95 0.97 0.9 1.3
661.7 2.064E-04 1.40 0.39 1.45 2.096E-04 0.38 1.50 1.55 1.5 2.1
Pb
Energy (keV) Experimental efficiency,εexp (counts Bq 1 s 1) Activity uncertainty (%) Peak area uncertainty (%) Combined std uncertainty (%) Monte Carlo efficiency,εMC (counts Bq 1 s 1) (NPS ¼ 3.4 108) Statistical uncertainty (%) Combined uncertainty (%)a Combined std uncertainty (%) Deviation, (εMC-εexp )/εexp (%) Uncertainty (%)
Am
Cs
60
60
1173.2 1.640E-04 1.30 0.45 1.38 1.625E-04 0.43 1.50 1.56 0.9 2.1
1332.5 1.499E-04 1.30 0.45 1.38 1.478E-04 0.45 1.50 1.57 1.4 2.1
Co
Co
a Combined standard relative uncertainties of uncertainty associated to the efficiency calculated by MCNP-CP due to the uncertainty on the activity and on the emission probability.
that the model of the detector is validated and, consequently, the detector is well characterized. This model of the detector was used to calculate by MCNP-CP the true coincidence summing effects for standards of cast steel, slag and fume dust. The results were presented in Solc et al., 2015 as MCNP-CP IST.
Union on the basis of Decision No. 912/2009/EC. The EMRP is jointly funded by the EMRP participating countries within EURAMET and the European Union.
References 8. Conclusions A model describing a high-purity germanium coaxial detector with a thin beryllium window based on a model of an n-type ORTEC GMX45P4 HPGe detector has been created. A good agreement between experimental and Monte Carlo simulations results was achieved for the two considered distances, confirming that the model corresponds to the real detector. This means that true coincidence summing effects can be determined and in a more general way, in the absence of adequate standard calibration sources, efficiency and efficiency transfer factors for geometry deviations and matrix effects can be accurately computed by using Monte Carlo methods even if true coincidence could occurs as is the case when the 60Co radioisotope is present in the source.
Acknowledgment The authors wish to thanks to IPOFG Lisboa the possibility to perform the radiographies on the detector and to José Afonso all the help to obtain the excellent set of radiographies. This work was supported by the EMRP joint research project “Ionising radiation metrology for the metallurgical industry” (MetroMetal) which has received funding from the European
Berlizov, A.N., 2007. MCNP-CP: a correlated particle radiation source extension of a general purpose Monte Carlo N-particle transport code. Institute for nuclear research, National Academy of Science of Ukraine 47, Prospekt Nauki, MSP 0380. Kyiv, Ukraine. In: Semkow, T. (Ed.), Applied Modeling and Computations in Nuclear Science 13. ACS Symposium Series; American Chemical Society, Washington, DC, pp. 183–194. Cullen, D.E., Hubbell, J.H., Kissel, L.D., 1997. EPDL97: The Evaluated Photon Data Library, ’97 Version Lawrence Livemore National Laboratory, UCRL-50400, (6), Rev. 5. Pelowitz, D.B., MCNPXTM, 2011. A General Monte Carlo N-Particle Transport Code. User's Manual, Version 2.7.0. Luis, R., Bento, J., Carvalhal, G., Nogueira, P., Silva, L., Teles, P., Vaz, P., 2010. Parameter optimization of a planar BEGe detector using Monte Carlo simulations. Nucl. Instrum. Methods Phys. Res. A 623, 1014–1019. Li, G., Bai, L., Xu, J., Liang, Y., 2015. Optimization of cylindrical sample geometry in determining the radioactive activity of environmental soil samples by using a Broad Energy Germanium detector. J. Instrum. 10. Stefanakis, N., 2014. Detection Efficiency of a BEGe Detector using the Monte Carlo Method and a Comparison to other Calibration Methods, November. 〈http:// xxx.tau.ac.il/pdf/1411.4225.pdf〉. Dryak, P., Kovar, P., 2006. Experimental and MC determination of HPGe detector efficiency in the 40–2754 keV energy range for measuring point source geometry with the source-to-detector distance of 25 cm. Appl. Radiat. Isot. 64, 1346–1349. Solc, J., Dryák, P., Moser, H., Branger, T., Garcia-Toraño, E., Peyrés, V., Tzika, F., Lutter, G., Capogni, M., Fazio, A., Luca, A., Vodnik, B., Oliveira, C., Saraiva, A., Szucs, L., Dziel, T., Burda, O., Arnold, D., Martinkovic, J., Siiskonen, T., Mattila, A., 2015. Characterisation of a radionuclide specific laboratory detector system for the metallurgical industry by Monte Carlo simulations. Radiat. Phys. Chem. 116, 189–193. ORTEC, 2003. The Best Choice of High Purity Germanium (HPGe) Detector.