- Email: [email protected]
Beta–gamma coincidence counting efficiency and energy resolution optimization by Geant4 Monte Carlo simulations for a phoswich well detector
Applied Radiation and Isotopes 68 (2010) 2377–2381
Contents lists available at ScienceDirect
Applied Radiation and Isotopes journal homepage: www.elsevier.com/locate/apradiso
Beta–gamma coincidence counting efficiency and energy resolution optimization by Geant4 Monte Carlo simulations for a phoswich well detector Weihua Zhang n, Pawel Mekarski, Kurt Ungar Radiation Protection Bureau of Health Canada, 775 Brookfield Road, AL 6302D1, Ottawa, Ontario, Canada K1A 1C1
a r t i c l e in fo
abstract
Article history: Received 30 March 2010 Received in revised form 14 May 2010 Accepted 3 June 2010
A single-channel phoswich well detector has been assessed and analysed in order to improve beta– gamma coincidence measurement sensitivity of 131mXe and 133mXe. This newly designed phoswich well detector consists of a plastic cell (BC-404) embedded in a CsI(Tl) crystal coupled to a photomultiplier tube (PMT). It can be used to distinguish 30.0-keV X-ray signals of 131mXe and 133mXe using their unique coincidence signatures between the conversion electrons (CEs) and the 30.0-keV X-rays. The optimum coincidence efficiency signal depends on the energy resolutions of the two CE peaks, which could be affected by relative positions of the plastic cell to the CsI(Tl) because the embedded plastic cell would interrupt scintillation light path from the CsI(Tl) crystal to the PMT. In this study, several relative positions between the embedded plastic cell and the CsI(Tl) crystal have been evaluated using Monte Carlo modeling for its effects on coincidence detection efficiency and X-ray and CE energy resolutions. The results indicate that the energy resolution and beta–gamma coincidence counting efficiency of X-ray and CE depend significantly on the plastic cell locations inside the CsI(Tl). The degraded X-ray and CE peak energy resolutions due to light collection efficiency deterioration by the embedded cell can be minimised. The optimum of CE and X-ray energy resolution, beta–gamma coincidence efficiency as well as the ease of manufacturing could be achieved by varying the embedded plastic cell positions inside the CsI(Tl) and consequently setting the most efficient geometry. Crown Copyright & 2010 Published by Elsevier Ltd. All rights reserved.
Keywords: Geant4 Monte Carlo Conversion electron Peak energy resolution Beta–gamma coincidence counting Phoswich detector
1. Introduction In the comprehensive nuclear test ban treaty (CTBT) verification regime, the two-dimensional beta–gamma coincidence phoswich detector is widely used for radioxenon activity measurement due to its ability to distinguish 131mXe and 133mXe and to suppress no-coincident background events as well as for its high sensitivity to the coincident events characteristic of the xenon radioisotopes of interest. The phoswich detector basically consists of two scintillators, i.e. an inorganic scintillator and a plastic scintillator counting cell. To achieve high coincidence detection efficiency, a cavity is drilled inside the inorganic scintillator, which accommodates the plastic scintillator counting cell. The embedded plastic cell is filled with xenon gas to be counted. The xenon radioisotope decays by emitting gamma-rays or X-rays in coincidence with beta-particles or CE. The plastic scintillator is used to absorb all beta-particles and CE, while the longer range gamma-rays and X-rays are mainly detected by the inorganic scintillator. When a xenon radioisotope emits both CE and X-rays, beta–gamma
n
Corresponding author. Fax: + 1 613 957 1089. E-mail address: [email protected] (W. Zhang).
coincident signal can be distinguished. The detector enables the same energy (30.0 keV) X-rays emitted from 131mXe (abundance 44.2%) and 133mXe (abundance 45.7%) to be separated by the well-specified discrete conversion electron peaks at 129.4 keV (abundance 60.7%) and 198.7 keV (abundance 63.1%). The peak width of CE and X-rays is mostly limited by resolution of the scintillators, but the spatial variations in the light collection efficiency throughout the detector can significantly degrade the overall energy resolution of the detector (Meng et al., 2002). Especially for this crystal embedded phoswich detector, the light paths between inorganic scintillator and PMT could be obstructed, which spoils the uniformity of light collection efficiency and overall energy resolution of the detector. This is because when the same incident energy X-rays are deposited at different locations within the inorganic scintillator, different amounts of light will be collected by the PMT. For the same reason, the scintillation light resulting from CE energy depositions within the plastic cell needs to be collected throughout the inorganic scintillator as light passes through it. The different relative positions between PMT and plastic scintillator can also lead to non-uniformity of light collection efficiency. Thus, the additional contributions to the overall energy resolution of detector due to the embedded plastic cell have to be studied. Monte Carlo simulations, as a powerful tool to study
0969-8043/$ - see front matter Crown Copyright & 2010 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.apradiso.2010.06.007
2378
W. Zhang et al. / Applied Radiation and Isotopes 68 (2010) 2377–2381
detector behavior, have been used for these purposes. The study was based on a single-channel radioxenon beta–gamma coincidence phoswich detector as a prototype (Hennig et al., 2006). The paper will focus on characterization of the CE and X-ray energy peak resolution as well as beta–gamma coincidence efficiency at different relative positions of two scintilltors by Monte Carlo simulations. The Geant4 simulation toolkit (Geant4 Collaboration, 2003) was chosen to conduct the phoswich detector geometry, physical response and scintillation processes modeling. It has been demonstrated that the developed Geant4 simulation tool can correctly model the beta–gamma coincidence detection of xenon isotopes in the phoswich detector and produce accurate coincidence detection efficiencies for each individual xenon isotope in spectrum analysis (Mekarski et al., 2009; Hennig et al., 2009).
2. Simulation The phoswich detector consists of a spherical plastic scintillator BC-404 cell (decay constant 25 ns) optically coupled to a cylindrical CsI(Tl) crystal (decay constant 250 ns) and was read out by a single PMT. The signal output from PMT is directly connected to a digital pulse processor. The inner radius of BC-404 cell was 15.00 mm; the wall thickness was 2 mm. The radius of CsI(Tl) cylinder was 38.10 mm; height was 76.20 mm. The BC-404 cell was surrounded by CsI(Tl) on all sides. All interfaces between CsI and BC-404 had an optical couplant. The entire system was housed in a 1.8 mm sheet of reflective copper. The inner surfaces of copper sheet were optically defined as such that a very fine finish was painted onto them. The reflectivity of the metals surfaces was set to match the reflectivity of a reflector painted onto the surfaces (95%). The BC-404, CsI(Tl), and PMT surfaces were all coupled using an optical glue. There was a gap between the photocathode and the CsI(Tl) crystal. This gap, which was composed of borosilicate glass, served as the PMT window. The index of refraction of PMT borosilicate glass window and optical couplant was set at 1.50. The PMT bialkali photocathode reflectivity was set at 17%. The quantum efficiency of the photocathode surface was set to 30%. The index of refraction of the BC-404 was set at 1.58 and that of CsI(Tl) was set at 1.78. The light yields of 54,000 and 11,000 photons/MeV were set for CsI(Tl) and BC-404, respectively. The attenuation lengths of 2000 and 2800 mm were set for CsI(Tl) and BC-404, respectively. Due to the very short distances the photons travel in the PMT to reach the photocathode, the number of photons lost to absorption was assumed to be negligible and therefore no attenuation length was defined for borosilicate glass. The values were those corresponding to the peak emission wavelength of CsI(Tl) at 550 nm. These parameters were then set to vary according to the wavelengths in the CsI(Tl) light spectrum. The Geant4 toolkit requires the user to select the physics processes pertaining to their simulation. The processes defined within this simulation include the standard electromagnetic (EM) package, as well as both decay signal and radioactive decay. The standard EM package contains the following processes: photo-electric effect, Compton scattering, pair production, bremsstrahlung, ionization, multiple scattering, and annihilation. Its effective energy range is nominally between 1 eV and 100 TeV. Processes for optical photons have also been included in the simulation. These processes include Cerenkov, scintillation, Rayleigh scattering, absorption, and boundary processes. However, optionally the optical photon processes can be disabled if desired. Disabling them greatly reduces computational time and was done whenever scintillation was not needed. It should also be noted that energy conservation has not been considered while carrying out these processes using Geant4. However, since only
the number of photons is counted within the simulation and not the photons’ energies, this detail has no impact on the results. A great detail has been put into the definition of the optical parameters. This was done to ensure realistic light collection and to produce high quality light pulses and spectra. In a previous simulation study (Mekarski et al., 2009), monoenergetic photon was defined and the number of photons was counted. Subsequently, all the optical parameters were set to be those corresponding to that of photon energy. However, for this simulation study, the light spectrum produced by CsI(Tl) was defined. All the optical parameters were defined such that they varied with the wavelength of light, if applicable. A cluster computer consisting of four-processor 3.0 GHz Inter Xeon servers, with 4GB SDRAM each, was used to run the simulations. The radioactive decay module (G4RadioactiveDecay.3.2) was used for radioxenon isotopes. The beta-particles, conversion electrons (CEs), Auger electrons, gamma-rays, and X-rays associated with isomeric transition decays (131mXe and 133mXe) were included. Ten thousand MC samplings were used for each isotope simulation. The MC sample size was determined by 1% confidence interval (CI) at the confidence level of 95% for one million decay events. The computation time was about an hour for each isotope.
3. Results and discussion An important aspect of the study is that it can turn simulation results into a spectral file in the format described in IDC-3.4.1Rev.6 (Formats and Protocols for Messages, 2004). This can be compatible with standard beta–gamma coincidence spectra analysis software, in this case XECON made by the Swedish Defense Research Agency (Ringbom). This strategy enabled spectral analysis of the synthetic spectra in the same fashion as those experimentally collected. A synthetic 131mXe and 133mXe beta–gamma coincidence spectrum with 10,000 MC samplings each is illustrated in Fig. 1. The two-dimensional histogram at the bottom left shows a single well-defined peak resulting from X-rays energy depositions in CsI(Tl) (gamma-energy axis) from these two metastable isotopes. At the bottom right is the figure of distributions resulting from 129.4 and 198.7 keV CE energy depositions in BC-404 (beta-energy axis). The top view of threedimensional histogram at the top right contains information regarding detected beta–gamma coincidence decay events by the phoswich detector. The histogram is divided into several rectangularly shaped regions of interest (ROIs) based on gamma-ray energy and beta-particle energy. The boundaries of each ROI rectangular area are in energy units and depend on xenon radioisotope decay modes. The gross counts in each ROI are calculated by summing the counts per channel within the region. An analytical procedure described in literature (Axelson and Ringbom, 2003) is adopted to define various rectangular regions of interest (ROIs) in the simulated beta–gamma coincidence spectra. According to the procedure, various rectangularly shaped regions of interest are used to determine the number of coincidence counts associated with each xenon radioisotope. In Fig. 1, the highest-density horizontal bins correspond to the strongest coincident decay modes of 131mXe and 133mXe (30.0 keV X-rays following the 129.4 and 198.7 keV conversion electrons). The rectangular ROIs for these areas are defined as y-axis (CsI) 17.5–42.6 keV and x-axis (BC-404) 170.1–239.0 keV for 133mXe and 97.0–162.0 keV for 131mXe. The activities of 131mXe and 133m Xe can be determined by the counts falling within specific ROI bounds. It should be noted that the histogram data are stored in channels, whereas the ROI boundary definitions are initially in
W. Zhang et al. / Applied Radiation and Isotopes 68 (2010) 2377–2381
2379
Fig. 1. Typical synthetic 131mXe and 133mXe spectra from XECON based on 10,000 MC samplings for each. Top view of the 3D histogram regarding the beta–gamma coincidence decay events detected by phoswich detector (a), the 2D histograms of energy deposition events at CsI(Tl) (b), and BC-404 (c).
Fig. 2. Sketches of five full detector system geometry modelings produced by Geant4 simulations.
2380
W. Zhang et al. / Applied Radiation and Isotopes 68 (2010) 2377–2381
if there are some locations of plastic cell in the CsI(Tl) that can lead to a less deteriorated light collection efficiency and thus get improved CE peak resolution to make the system satisfy the requirement. Fig. 2 shows the five Geant4 geometric layouts used in this study. The plastic cell in the CsI(Tl) geometric centre is taken as zero; other positions are along the centre axis of the CsI(Tl) cylinder and 10, 20, 10, and 20 mm from the centre. In this kind of detector assembly, the embedded detector may obstruct the light path from CsI(Tl) crystal to PMT, and affect crystal uniformity in light collection efficiency of the CsI(Tl). A study on geometric light collection efficiency distribution using the DETECT2000 Monte Carlo code for this plastic cell embedded detector geometry was carried out (Hennig et al., 2007). The study indicated that the front part of CsI(Tl) crystal (volume between the embedded plastic cell and PMT) has better light collection efficiency and uniformity, while the light collection efficiency and
energy units. The ROI boundaries must be converted to channels with energy calibration information of the system. The peak separation of the overlapped doublet of 131mXe and 133m X CE is 69.3 keV, as shown in Fig. 1(c). The energy resolutions, expressed as peak full width at half maximum (FWHM) divided by peak centre position, are 35.9% and 31.6% for the 129.4 and 198.7 keV peaks, respectively. The calculations indicate that the two peaks can be separated only by 1.24 times FWHM. Deconvolution of the doublet with a Gaussian function indicates that the two CE peaks are overlapped by 10% of the entire counts distributions. To meet the requirements of negligible nuclide false-positive rate (nuclide misidentification) and small nuclide activity interference correction, the rule of thumb is that the doublet separation should be greater than 1.5 times FWHM resolution, so that at 5% risk level the false-positive rate is less than 1%. One primary motivation for the present study was to find
43.0
82.0
41.0
81.0
131
mXe
80.0
39.0
79.0 78.0
35.0
77.0
33.0
76.0
Efficiency, %
FWHM, %
37.0
75.0
31.0
74.0 29.0 73.0
CE peak width by BC404 X-ray peak width by CsI Coincidence efficiency
27.0
72.0
25.0
71.0 -25
-20
-15
-10
-5
0
5
10
15
20
25
Distance from center, mm
45.0
75 73
133
mXe
41.0
71
39.0
69
37.0
67 CE peak width by BC404 X-ray peak width by CsI Coincidence efficiency
35.0
65
33.0
63
31.0
61
29.0
59
27.0
57
25.0
Efficiency, %
FWHM, %
43.0
55 -25
-20
-15
-10
0 5 -5 Distance from centre, mm
10
15
20
25
Fig. 3. Simulated beta–gamma coincidence counting efficiency and energy resolution of these two metastable radioxenons at different plastic cell locations in the CsI(Tl).
W. Zhang et al. / Applied Radiation and Isotopes 68 (2010) 2377–2381
uniformity are lower in the back-part CsI(Tl) crystal (the volume on the other side of plastic cell). This could limit the achievable X-ray or CE peak energy resolution. Histograms of X-ray energy depositions in the CsI(Tl) detector and CE energy depositions in the plastic cell were obtained at each of the relative detector counting geometry. A Gaussian function was used to fit each energy histogram spectrum; thus energy resolution was obtained. The results are plotted in Fig. 3. When the plastic cell is at the CsI(Tl) geometric centre, the peak energy resolution of 30.0 keV X-ray is 32.5% for 131mXe and 31.6% for 133m Xe, as shown in Fig. 3. Moving the plastic cell forward to PMT along the centre axis, the simulated X-ray peak energy resolution is increased and significantly worsened to 35.7% at 20 mm plastic cell position. The results could be explained by an increasing back-part volume of CsI(Tl) crystal. As the plastic cell moves closer to the PMT, a larger back part of the CsI(Tl) crystal will be radiated by radioxenon source. The lower light collection efficiency and uniformity will lead to worse peak energy resolution. Moving the plastic cell in the opposite direction (away from PMT), the front part of CsI(Tl) crystal, which has better uniformity in light collection efficiency, will be increased. In these cases even some X-rays may escape out of the thinner part of the CsI(Tl) crystal; the results show that under the best case scenario (plastic cell at 20 mm), the peak energy resolution of 30.0 keV X-ray is slightly improved by 5–6% compared with the values obtained by the plastic cell at the CsI(Tl) geometric centre. It is likely that for the plastic cell embedded detector the spatial variation in light collection efficiency is a more important factor than statistical fluctuations for overall energy resolution improvement. Moving the plastic cell will make no impact on CE incident energy interacting with the detector, because the radiation source is also moving along with the cell. However, light produced by the plastic cell has to be transported through the CsI(Tl) crystal before reaching the PMT. Variations in light collection efficiency throughout the CsI(Tl) crystal will also result in CE energy peak changing. This has been observed in the simulation results. When the plastic cell is at the CsI(Tl) geometric centre, energy resolutions of the 129.4 keV (131mXe) and 198.7 keV (133mXe) CE peaks are 35.9% and 29.2%, respectively. Moving the plastic cell forward to the PMT along the centre axis, energy resolutions of these two CE peaks are degraded to 40.4% and 36.3%, respectively, at the plastic cell 20 mm position, as shown in Fig. 3. When moving the plastic cell away from the PMT, under the best case scenario (plastic cell at 20 mm), energy resolution of these two CE peaks can be improved by 10% compared with the plastic cell at the CsI(Tl) geometric centre. Also when the plastic cell moves away from the PMT, a significant improvement in beta–gamma coincidence efficiency is obviously observed as shown in Fig. 3. These results demonstrated that the narrower the X-ray and CE peaks, the more the coincidence events associated with 131mXe
2381
and 133mXe captured in the rectangularly shaped ROI region, which results in the improved beta–gamma coincidence counting efficiency.
4. Conclusions The evidence from this simulation study suggests that the X-ray and CE peak broadening caused by embedded plastic crystal should be one of the design considerations for this crystal imbedded single-channel phoswich detector. This is because when the plastic crystal is embedded within CsI(Tl), it will interrupt the light paths between all parts of the CsI(Tl) crystal and the PMT and result in spatial variations in light collection efficiency. The peak broadening due to the light interruption could be minimised by optimizing the positions of the plastic cell inside the CsI(Tl) crystal to reduce volumes of non-uniformity in light collection efficiency. The simulation results demonstrate that the detector configuration obtained by placing plastic cell furthest away from the PMT can provide the best overall detector energy resolution. Compared to the results obtained by placing the plastic cell at the CsI(Tl) geometric centre, such type of configuration can improve energy resolution of 30.0 keV X-ray by 8%, energy resolution of 129.4 keV CE of 131mXe by 10%, and energy resolution of 198.7 keV CE of 133mXe significantly by 9%. This energy resolution improvement results in 1.5 times FWHM peak separation for these two CE peaks from 131mXe and 133mXe, which is necessary to resolve these two CE peaks more clearly. Another key feature of the configuration is that the beta–gamma coincidence efficiency can be improved by 3.6% for 131mXe and 8.2% 133mXe.
References Axelson A., Ringbom,A., Swedish Defense Research Agency User Report. FOI-0913SE, 2003, pp. 1650–1942. Formats and Protocols for Messages, Preparatory Commission for the CTBTO. IDC Documentation, IDC-3.4.1 Revision 6, Vienna, 2004. Geant4 Collaboration, 2003. Nuclear Instruments and Methods in Physics Research A 506, 250–303 (Agostinelli, S. et al.). Hennig, W., Tan, H., Fallu-Labruyere, A., Warburton, W.K., McIntyre, J.I., Gleyzer, A., 2007. Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 579 (1), 431–436. Hennig, W., Tan, H., Warburton, W.K., Mcintyre, J.I., 2006. IEEE Transactions on Nuclear Science 53 (2), 620–624. Hennig, W., Warburton, W.K., Fallu-Labruyere, A., Sabourov, K., Cooper, M.W., McIntyre, J.I., Gleyzer, A., Bean, M., Korpach, E., Ungar, K., Zhang, W., Mekarski, P., 2009. Journal of Radioanalytical and Nuclear Chemistry 282 (3), 681–685. Mekarski, P., Zhang, W., Ungar, K., Bean, M., Korpach, E., 2009. Applied Radiation and Isotopes 67 (10), 1957–1963. Meng, L.J., Ramsden, D., Chirkin, V.M., Potapov, V.N., Ivanov, O.P., Ignatov, S.M., 2002. Nuclear Instruments and Methods in Physics Research A 485, 468–476. Ringbom, A., FOI Software XECON, /http://sourceforge.net/projects/excon/S.
Contents lists available at ScienceDirect
Applied Radiation and Isotopes journal homepage: www.elsevier.com/locate/apradiso
Beta–gamma coincidence counting efficiency and energy resolution optimization by Geant4 Monte Carlo simulations for a phoswich well detector Weihua Zhang n, Pawel Mekarski, Kurt Ungar Radiation Protection Bureau of Health Canada, 775 Brookfield Road, AL 6302D1, Ottawa, Ontario, Canada K1A 1C1
a r t i c l e in fo
abstract
Article history: Received 30 March 2010 Received in revised form 14 May 2010 Accepted 3 June 2010
A single-channel phoswich well detector has been assessed and analysed in order to improve beta– gamma coincidence measurement sensitivity of 131mXe and 133mXe. This newly designed phoswich well detector consists of a plastic cell (BC-404) embedded in a CsI(Tl) crystal coupled to a photomultiplier tube (PMT). It can be used to distinguish 30.0-keV X-ray signals of 131mXe and 133mXe using their unique coincidence signatures between the conversion electrons (CEs) and the 30.0-keV X-rays. The optimum coincidence efficiency signal depends on the energy resolutions of the two CE peaks, which could be affected by relative positions of the plastic cell to the CsI(Tl) because the embedded plastic cell would interrupt scintillation light path from the CsI(Tl) crystal to the PMT. In this study, several relative positions between the embedded plastic cell and the CsI(Tl) crystal have been evaluated using Monte Carlo modeling for its effects on coincidence detection efficiency and X-ray and CE energy resolutions. The results indicate that the energy resolution and beta–gamma coincidence counting efficiency of X-ray and CE depend significantly on the plastic cell locations inside the CsI(Tl). The degraded X-ray and CE peak energy resolutions due to light collection efficiency deterioration by the embedded cell can be minimised. The optimum of CE and X-ray energy resolution, beta–gamma coincidence efficiency as well as the ease of manufacturing could be achieved by varying the embedded plastic cell positions inside the CsI(Tl) and consequently setting the most efficient geometry. Crown Copyright & 2010 Published by Elsevier Ltd. All rights reserved.
Keywords: Geant4 Monte Carlo Conversion electron Peak energy resolution Beta–gamma coincidence counting Phoswich detector
1. Introduction In the comprehensive nuclear test ban treaty (CTBT) verification regime, the two-dimensional beta–gamma coincidence phoswich detector is widely used for radioxenon activity measurement due to its ability to distinguish 131mXe and 133mXe and to suppress no-coincident background events as well as for its high sensitivity to the coincident events characteristic of the xenon radioisotopes of interest. The phoswich detector basically consists of two scintillators, i.e. an inorganic scintillator and a plastic scintillator counting cell. To achieve high coincidence detection efficiency, a cavity is drilled inside the inorganic scintillator, which accommodates the plastic scintillator counting cell. The embedded plastic cell is filled with xenon gas to be counted. The xenon radioisotope decays by emitting gamma-rays or X-rays in coincidence with beta-particles or CE. The plastic scintillator is used to absorb all beta-particles and CE, while the longer range gamma-rays and X-rays are mainly detected by the inorganic scintillator. When a xenon radioisotope emits both CE and X-rays, beta–gamma
n
Corresponding author. Fax: + 1 613 957 1089. E-mail address: [email protected] (W. Zhang).
coincident signal can be distinguished. The detector enables the same energy (30.0 keV) X-rays emitted from 131mXe (abundance 44.2%) and 133mXe (abundance 45.7%) to be separated by the well-specified discrete conversion electron peaks at 129.4 keV (abundance 60.7%) and 198.7 keV (abundance 63.1%). The peak width of CE and X-rays is mostly limited by resolution of the scintillators, but the spatial variations in the light collection efficiency throughout the detector can significantly degrade the overall energy resolution of the detector (Meng et al., 2002). Especially for this crystal embedded phoswich detector, the light paths between inorganic scintillator and PMT could be obstructed, which spoils the uniformity of light collection efficiency and overall energy resolution of the detector. This is because when the same incident energy X-rays are deposited at different locations within the inorganic scintillator, different amounts of light will be collected by the PMT. For the same reason, the scintillation light resulting from CE energy depositions within the plastic cell needs to be collected throughout the inorganic scintillator as light passes through it. The different relative positions between PMT and plastic scintillator can also lead to non-uniformity of light collection efficiency. Thus, the additional contributions to the overall energy resolution of detector due to the embedded plastic cell have to be studied. Monte Carlo simulations, as a powerful tool to study
0969-8043/$ - see front matter Crown Copyright & 2010 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.apradiso.2010.06.007
2378
W. Zhang et al. / Applied Radiation and Isotopes 68 (2010) 2377–2381
detector behavior, have been used for these purposes. The study was based on a single-channel radioxenon beta–gamma coincidence phoswich detector as a prototype (Hennig et al., 2006). The paper will focus on characterization of the CE and X-ray energy peak resolution as well as beta–gamma coincidence efficiency at different relative positions of two scintilltors by Monte Carlo simulations. The Geant4 simulation toolkit (Geant4 Collaboration, 2003) was chosen to conduct the phoswich detector geometry, physical response and scintillation processes modeling. It has been demonstrated that the developed Geant4 simulation tool can correctly model the beta–gamma coincidence detection of xenon isotopes in the phoswich detector and produce accurate coincidence detection efficiencies for each individual xenon isotope in spectrum analysis (Mekarski et al., 2009; Hennig et al., 2009).
2. Simulation The phoswich detector consists of a spherical plastic scintillator BC-404 cell (decay constant 25 ns) optically coupled to a cylindrical CsI(Tl) crystal (decay constant 250 ns) and was read out by a single PMT. The signal output from PMT is directly connected to a digital pulse processor. The inner radius of BC-404 cell was 15.00 mm; the wall thickness was 2 mm. The radius of CsI(Tl) cylinder was 38.10 mm; height was 76.20 mm. The BC-404 cell was surrounded by CsI(Tl) on all sides. All interfaces between CsI and BC-404 had an optical couplant. The entire system was housed in a 1.8 mm sheet of reflective copper. The inner surfaces of copper sheet were optically defined as such that a very fine finish was painted onto them. The reflectivity of the metals surfaces was set to match the reflectivity of a reflector painted onto the surfaces (95%). The BC-404, CsI(Tl), and PMT surfaces were all coupled using an optical glue. There was a gap between the photocathode and the CsI(Tl) crystal. This gap, which was composed of borosilicate glass, served as the PMT window. The index of refraction of PMT borosilicate glass window and optical couplant was set at 1.50. The PMT bialkali photocathode reflectivity was set at 17%. The quantum efficiency of the photocathode surface was set to 30%. The index of refraction of the BC-404 was set at 1.58 and that of CsI(Tl) was set at 1.78. The light yields of 54,000 and 11,000 photons/MeV were set for CsI(Tl) and BC-404, respectively. The attenuation lengths of 2000 and 2800 mm were set for CsI(Tl) and BC-404, respectively. Due to the very short distances the photons travel in the PMT to reach the photocathode, the number of photons lost to absorption was assumed to be negligible and therefore no attenuation length was defined for borosilicate glass. The values were those corresponding to the peak emission wavelength of CsI(Tl) at 550 nm. These parameters were then set to vary according to the wavelengths in the CsI(Tl) light spectrum. The Geant4 toolkit requires the user to select the physics processes pertaining to their simulation. The processes defined within this simulation include the standard electromagnetic (EM) package, as well as both decay signal and radioactive decay. The standard EM package contains the following processes: photo-electric effect, Compton scattering, pair production, bremsstrahlung, ionization, multiple scattering, and annihilation. Its effective energy range is nominally between 1 eV and 100 TeV. Processes for optical photons have also been included in the simulation. These processes include Cerenkov, scintillation, Rayleigh scattering, absorption, and boundary processes. However, optionally the optical photon processes can be disabled if desired. Disabling them greatly reduces computational time and was done whenever scintillation was not needed. It should also be noted that energy conservation has not been considered while carrying out these processes using Geant4. However, since only
the number of photons is counted within the simulation and not the photons’ energies, this detail has no impact on the results. A great detail has been put into the definition of the optical parameters. This was done to ensure realistic light collection and to produce high quality light pulses and spectra. In a previous simulation study (Mekarski et al., 2009), monoenergetic photon was defined and the number of photons was counted. Subsequently, all the optical parameters were set to be those corresponding to that of photon energy. However, for this simulation study, the light spectrum produced by CsI(Tl) was defined. All the optical parameters were defined such that they varied with the wavelength of light, if applicable. A cluster computer consisting of four-processor 3.0 GHz Inter Xeon servers, with 4GB SDRAM each, was used to run the simulations. The radioactive decay module (G4RadioactiveDecay.3.2) was used for radioxenon isotopes. The beta-particles, conversion electrons (CEs), Auger electrons, gamma-rays, and X-rays associated with isomeric transition decays (131mXe and 133mXe) were included. Ten thousand MC samplings were used for each isotope simulation. The MC sample size was determined by 1% confidence interval (CI) at the confidence level of 95% for one million decay events. The computation time was about an hour for each isotope.
3. Results and discussion An important aspect of the study is that it can turn simulation results into a spectral file in the format described in IDC-3.4.1Rev.6 (Formats and Protocols for Messages, 2004). This can be compatible with standard beta–gamma coincidence spectra analysis software, in this case XECON made by the Swedish Defense Research Agency (Ringbom). This strategy enabled spectral analysis of the synthetic spectra in the same fashion as those experimentally collected. A synthetic 131mXe and 133mXe beta–gamma coincidence spectrum with 10,000 MC samplings each is illustrated in Fig. 1. The two-dimensional histogram at the bottom left shows a single well-defined peak resulting from X-rays energy depositions in CsI(Tl) (gamma-energy axis) from these two metastable isotopes. At the bottom right is the figure of distributions resulting from 129.4 and 198.7 keV CE energy depositions in BC-404 (beta-energy axis). The top view of threedimensional histogram at the top right contains information regarding detected beta–gamma coincidence decay events by the phoswich detector. The histogram is divided into several rectangularly shaped regions of interest (ROIs) based on gamma-ray energy and beta-particle energy. The boundaries of each ROI rectangular area are in energy units and depend on xenon radioisotope decay modes. The gross counts in each ROI are calculated by summing the counts per channel within the region. An analytical procedure described in literature (Axelson and Ringbom, 2003) is adopted to define various rectangular regions of interest (ROIs) in the simulated beta–gamma coincidence spectra. According to the procedure, various rectangularly shaped regions of interest are used to determine the number of coincidence counts associated with each xenon radioisotope. In Fig. 1, the highest-density horizontal bins correspond to the strongest coincident decay modes of 131mXe and 133mXe (30.0 keV X-rays following the 129.4 and 198.7 keV conversion electrons). The rectangular ROIs for these areas are defined as y-axis (CsI) 17.5–42.6 keV and x-axis (BC-404) 170.1–239.0 keV for 133mXe and 97.0–162.0 keV for 131mXe. The activities of 131mXe and 133m Xe can be determined by the counts falling within specific ROI bounds. It should be noted that the histogram data are stored in channels, whereas the ROI boundary definitions are initially in
W. Zhang et al. / Applied Radiation and Isotopes 68 (2010) 2377–2381
2379
Fig. 1. Typical synthetic 131mXe and 133mXe spectra from XECON based on 10,000 MC samplings for each. Top view of the 3D histogram regarding the beta–gamma coincidence decay events detected by phoswich detector (a), the 2D histograms of energy deposition events at CsI(Tl) (b), and BC-404 (c).
Fig. 2. Sketches of five full detector system geometry modelings produced by Geant4 simulations.
2380
W. Zhang et al. / Applied Radiation and Isotopes 68 (2010) 2377–2381
if there are some locations of plastic cell in the CsI(Tl) that can lead to a less deteriorated light collection efficiency and thus get improved CE peak resolution to make the system satisfy the requirement. Fig. 2 shows the five Geant4 geometric layouts used in this study. The plastic cell in the CsI(Tl) geometric centre is taken as zero; other positions are along the centre axis of the CsI(Tl) cylinder and 10, 20, 10, and 20 mm from the centre. In this kind of detector assembly, the embedded detector may obstruct the light path from CsI(Tl) crystal to PMT, and affect crystal uniformity in light collection efficiency of the CsI(Tl). A study on geometric light collection efficiency distribution using the DETECT2000 Monte Carlo code for this plastic cell embedded detector geometry was carried out (Hennig et al., 2007). The study indicated that the front part of CsI(Tl) crystal (volume between the embedded plastic cell and PMT) has better light collection efficiency and uniformity, while the light collection efficiency and
energy units. The ROI boundaries must be converted to channels with energy calibration information of the system. The peak separation of the overlapped doublet of 131mXe and 133m X CE is 69.3 keV, as shown in Fig. 1(c). The energy resolutions, expressed as peak full width at half maximum (FWHM) divided by peak centre position, are 35.9% and 31.6% for the 129.4 and 198.7 keV peaks, respectively. The calculations indicate that the two peaks can be separated only by 1.24 times FWHM. Deconvolution of the doublet with a Gaussian function indicates that the two CE peaks are overlapped by 10% of the entire counts distributions. To meet the requirements of negligible nuclide false-positive rate (nuclide misidentification) and small nuclide activity interference correction, the rule of thumb is that the doublet separation should be greater than 1.5 times FWHM resolution, so that at 5% risk level the false-positive rate is less than 1%. One primary motivation for the present study was to find
43.0
82.0
41.0
81.0
131
mXe
80.0
39.0
79.0 78.0
35.0
77.0
33.0
76.0
Efficiency, %
FWHM, %
37.0
75.0
31.0
74.0 29.0 73.0
CE peak width by BC404 X-ray peak width by CsI Coincidence efficiency
27.0
72.0
25.0
71.0 -25
-20
-15
-10
-5
0
5
10
15
20
25
Distance from center, mm
45.0
75 73
133
mXe
41.0
71
39.0
69
37.0
67 CE peak width by BC404 X-ray peak width by CsI Coincidence efficiency
35.0
65
33.0
63
31.0
61
29.0
59
27.0
57
25.0
Efficiency, %
FWHM, %
43.0
55 -25
-20
-15
-10
0 5 -5 Distance from centre, mm
10
15
20
25
Fig. 3. Simulated beta–gamma coincidence counting efficiency and energy resolution of these two metastable radioxenons at different plastic cell locations in the CsI(Tl).
W. Zhang et al. / Applied Radiation and Isotopes 68 (2010) 2377–2381
uniformity are lower in the back-part CsI(Tl) crystal (the volume on the other side of plastic cell). This could limit the achievable X-ray or CE peak energy resolution. Histograms of X-ray energy depositions in the CsI(Tl) detector and CE energy depositions in the plastic cell were obtained at each of the relative detector counting geometry. A Gaussian function was used to fit each energy histogram spectrum; thus energy resolution was obtained. The results are plotted in Fig. 3. When the plastic cell is at the CsI(Tl) geometric centre, the peak energy resolution of 30.0 keV X-ray is 32.5% for 131mXe and 31.6% for 133m Xe, as shown in Fig. 3. Moving the plastic cell forward to PMT along the centre axis, the simulated X-ray peak energy resolution is increased and significantly worsened to 35.7% at 20 mm plastic cell position. The results could be explained by an increasing back-part volume of CsI(Tl) crystal. As the plastic cell moves closer to the PMT, a larger back part of the CsI(Tl) crystal will be radiated by radioxenon source. The lower light collection efficiency and uniformity will lead to worse peak energy resolution. Moving the plastic cell in the opposite direction (away from PMT), the front part of CsI(Tl) crystal, which has better uniformity in light collection efficiency, will be increased. In these cases even some X-rays may escape out of the thinner part of the CsI(Tl) crystal; the results show that under the best case scenario (plastic cell at 20 mm), the peak energy resolution of 30.0 keV X-ray is slightly improved by 5–6% compared with the values obtained by the plastic cell at the CsI(Tl) geometric centre. It is likely that for the plastic cell embedded detector the spatial variation in light collection efficiency is a more important factor than statistical fluctuations for overall energy resolution improvement. Moving the plastic cell will make no impact on CE incident energy interacting with the detector, because the radiation source is also moving along with the cell. However, light produced by the plastic cell has to be transported through the CsI(Tl) crystal before reaching the PMT. Variations in light collection efficiency throughout the CsI(Tl) crystal will also result in CE energy peak changing. This has been observed in the simulation results. When the plastic cell is at the CsI(Tl) geometric centre, energy resolutions of the 129.4 keV (131mXe) and 198.7 keV (133mXe) CE peaks are 35.9% and 29.2%, respectively. Moving the plastic cell forward to the PMT along the centre axis, energy resolutions of these two CE peaks are degraded to 40.4% and 36.3%, respectively, at the plastic cell 20 mm position, as shown in Fig. 3. When moving the plastic cell away from the PMT, under the best case scenario (plastic cell at 20 mm), energy resolution of these two CE peaks can be improved by 10% compared with the plastic cell at the CsI(Tl) geometric centre. Also when the plastic cell moves away from the PMT, a significant improvement in beta–gamma coincidence efficiency is obviously observed as shown in Fig. 3. These results demonstrated that the narrower the X-ray and CE peaks, the more the coincidence events associated with 131mXe
2381
and 133mXe captured in the rectangularly shaped ROI region, which results in the improved beta–gamma coincidence counting efficiency.
4. Conclusions The evidence from this simulation study suggests that the X-ray and CE peak broadening caused by embedded plastic crystal should be one of the design considerations for this crystal imbedded single-channel phoswich detector. This is because when the plastic crystal is embedded within CsI(Tl), it will interrupt the light paths between all parts of the CsI(Tl) crystal and the PMT and result in spatial variations in light collection efficiency. The peak broadening due to the light interruption could be minimised by optimizing the positions of the plastic cell inside the CsI(Tl) crystal to reduce volumes of non-uniformity in light collection efficiency. The simulation results demonstrate that the detector configuration obtained by placing plastic cell furthest away from the PMT can provide the best overall detector energy resolution. Compared to the results obtained by placing the plastic cell at the CsI(Tl) geometric centre, such type of configuration can improve energy resolution of 30.0 keV X-ray by 8%, energy resolution of 129.4 keV CE of 131mXe by 10%, and energy resolution of 198.7 keV CE of 133mXe significantly by 9%. This energy resolution improvement results in 1.5 times FWHM peak separation for these two CE peaks from 131mXe and 133mXe, which is necessary to resolve these two CE peaks more clearly. Another key feature of the configuration is that the beta–gamma coincidence efficiency can be improved by 3.6% for 131mXe and 8.2% 133mXe.
References Axelson A., Ringbom,A., Swedish Defense Research Agency User Report. FOI-0913SE, 2003, pp. 1650–1942. Formats and Protocols for Messages, Preparatory Commission for the CTBTO. IDC Documentation, IDC-3.4.1 Revision 6, Vienna, 2004. Geant4 Collaboration, 2003. Nuclear Instruments and Methods in Physics Research A 506, 250–303 (Agostinelli, S. et al.). Hennig, W., Tan, H., Fallu-Labruyere, A., Warburton, W.K., McIntyre, J.I., Gleyzer, A., 2007. Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 579 (1), 431–436. Hennig, W., Tan, H., Warburton, W.K., Mcintyre, J.I., 2006. IEEE Transactions on Nuclear Science 53 (2), 620–624. Hennig, W., Warburton, W.K., Fallu-Labruyere, A., Sabourov, K., Cooper, M.W., McIntyre, J.I., Gleyzer, A., Bean, M., Korpach, E., Ungar, K., Zhang, W., Mekarski, P., 2009. Journal of Radioanalytical and Nuclear Chemistry 282 (3), 681–685. Mekarski, P., Zhang, W., Ungar, K., Bean, M., Korpach, E., 2009. Applied Radiation and Isotopes 67 (10), 1957–1963. Meng, L.J., Ramsden, D., Chirkin, V.M., Potapov, V.N., Ivanov, O.P., Ignatov, S.M., 2002. Nuclear Instruments and Methods in Physics Research A 485, 468–476. Ringbom, A., FOI Software XECON, /http://sourceforge.net/projects/excon/S.