- Email: [email protected]
Cross-correlation measurements with the EJ-299-33 plastic scintillator
Nuclear Instruments and Methods in Physics Research A 784 (2015) 460–464
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research A journal homepage: www.elsevier.com/locate/nima
Cross-correlation measurements with the EJ-299-33 plastic scintillator Mark M. Bourne a,n, Jeff Whaley a, Jennifer L. Dolan a,1, John K. Polack a, Marek Flaska a, Shaun D. Clarke a, Alice Tomanin b, Paolo Peerani b, Sara A. Pozzi a a b
Department of Nuclear Engineering and Radiological Sciences, University of Michigan, Ann Arbor, MI 48109, USA European Commission, Joint Research Centre, Institute for Transuranium Elements, via Enrico Fermi 2749, 21027 Ispra (VA), Italy
art ic l e i nf o
a b s t r a c t
Available online 31 October 2014
New organic–plastic scintillation compositions have demonstrated pulse-shape discrimination (PSD) of neutrons and gamma rays. We present cross-correlation measurements of 252Cf and mixed uranium– plutonium oxide (MOX) with the EJ-299-33 plastic scintillator. For comparison, equivalent measurements were performed with an EJ-309 liquid scintillator. Offline, digital PSD was applied to each detector. These measurements show that EJ-299-33 sacrifices a factor of 5 in neutron–neutron efficiency relative to EJ-309, but could still utilize the difference in neutron–neutron efficiency and neutron singleto-double ratio to distinguish 252Cf from MOX. These measurements were modeled with MCNPX-PoliMi, and MPPost was used to convert the detailed collision history into simulated cross-correlation distributions. MCNPX-PoliMi predicted the measured 252Cf cross-correlation distribution for EJ-309 to within 10%. Greater photon uncertainty in the MOX sample led to larger discrepancy in the simulated MOX cross-correlation distribution. The modeled EJ-299-33 plastic also gives reasonable agreement with measured cross-correlation distributions, although the MCNPX-PoliMi model appears to under-predict the neutron detection efficiency. Published by Elsevier B.V.
Keywords: EJ-299-33 plastic scintillator MCNP MCNPX-PoliMi Cross-correlation distribution Neutron measurements
1. Introduction The recent 3He shortage has driven efforts towards developing alternative neutron detectors for nuclear nonproliferation applications. Organic liquid scintillators offer numerous advantages for fast neutron detection due to fast pulse timing, good efficiency, and excellent pulse-shape discrimination (PSD) [1,2]. For use in field applications, however, liquid scintillators are limited due to the potential risk from leaks or fire [4]. The EJ-299-33 plastic scintillator developed in late 2012 is of particular interest in nuclear nonproliferation applications due to its proven PSD capabilities as a plastic while eliminating environmental risks specific to liquid scintillators in field applications [3,5–8]. Cross-correlation measurements have been previously used for characterization of fission sources through detection and analysis n Correspondence to: Department of Nuclear Engineering & Radiological Sciences, University of Michigan, 2355 Bonisteel Boulevard, Ann Arbor, MI 481092104, USA. E-mail addresses: [email protected] (M.M. Bourne), [email protected] (J. Whaley), [email protected] (J.L. Dolan), [email protected] (J.K. Polack), mfl[email protected] (M. Flaska), [email protected] (S.D. Clarke), [email protected] (A. Tomanin), [email protected] (P. Peerani), [email protected] (S.A. Pozzi). 1 Jennifer Dolan was attending the University of Michigan at the time of the Ispra measurement campaign. She has since graduated.
http://dx.doi.org/10.1016/j.nima.2014.10.052 0168-9002/Published by Elsevier B.V.
of correlated neutron and gamma-ray photon emissions. Neutron– neutron correlations consist primarily of spontaneous fission neutrons from 252Cf and even-numbered plutonium isotopes, allowing for easy discrimination from single neutron emitters such as (alpha,n) sources. Varying multiplicity of spontaneous fission sources means that the intensity of neutron–neutron correlations may allow for separate identification of 252Cf from plutonium-based samples. Photon–neutron or photon–photon correlations can be detected from correlated sources such as fission or annihilation or accidentally from decay product buildup; these have been previously used to identify varying levels of burnup in plutonium-based fuel samples [9]. In this study, we present the EJ-299-33 plastic used in cross-correlation measurements of separate 252Cf and mixed uranium–plutonium oxide (MOX) samples. This scintillation detector is directly benchmarked to EJ-309 liquid scintillators, with proven PSD capabilities [1,2], with the same scintillator dimensions, photomultiplier tube model with equivalent base, and digital readout system. Accurate modeling of scintillation detector response allows for fast, inexpensive testing of detector performance in various environments and geometries. To further this goal for cross-correlation applications, these measurements were modeled using MCNPXPoliMi. MPPost, a specialized FORTRAN post-processing script, was used to convert the detection physics calculated by MCNPX-PoliMi into the cross-correlation response generated by each [10].
M.M. Bourne et al. / Nuclear Instruments and Methods in Physics Research A 784 (2015) 460–464
461
2. Cross-correlation measurements The EJ-299-33 plastic scintillation detectors were manufactured and assembled by Eljen Technologies. The crystals are right circular cylinders with 7.62-cm diameter and 7.62-cm depth. They were mounted to 7.62-cm diameter photomultiplier tubes (PMTs) ETL 9821B [11]. The EJ-309 detector was manufactured by SCIONIX using the same scintillator dimensions and mounted to the same PMT model as that used for the plastic scintillators. Anode signals of each PMT were digitized using a CAEN V1720 digitizer, with a 250-MHz sampling rate, 12-bit resolution, and 2-V dynamic range. Pulses were then analyzed offline to apply PSD and compute correlation timing using custom algorithms.
Fig. 1. The setup of the
252
Cf cross-correlation measurement.
2.1. Experimental setup The 252Cf measurement was performed at the University of Michigan's Department of Nuclear Engineering and Radiological Sciences in Ann Arbor, Michigan. The 252Cf measurement geometry is shown in Fig. 1. Two EJ-309 detectors and two EJ-299-33 detectors were placed in 1801 pairs at a distance of 20 cm from a 252Cf spontaneous fission source emitting 175,000 neutrons/s. All four detectors were gain-matched such that the 137Cs Compton edge aligned with approximately 400 mV; this allowed for detection of 95% of the 252Cf neutron energy spectrum with EJ-309 and 98% of the 252 Cf neutron energy spectrum with EJ-299-33. Correlated gamma-ray photon and neutron emissions from the 252Cf source were measured for 18 h. The MOX measurement was performed at the Joint Research Centre in Ispra, Italy, with the geometry shown in Fig. 2a. Twelve EJ-309 detectors and two EJ-299-33 detectors were placed in similar 1801 pairs at a 16.8 cm source–detector distance. Detectors were arranged octagonally around a MOX sample, stacking the EJ-309 liquids for improved efficiency. All 14 detectors were gain-matched to a 137Cs Compton edge of 300 mV, increasing the detectable 252Cf neutron energy range to 98% at the cost of resolution. Due to the high gamma emission rate from the source, a 1-cm lead sleeve was placed around the MOX sample. The MOX has a mass of 1.01 kg and consists of 66.8 wt% 238U, 0.4 wt% 235U, 11.1 wt% 239Pu, 4.6 wt% 240Pu, 0.2 wt% 241 Pu, 0.3 wt% 242Pu, 0.02 wt% 238Pu, 0.5 wt% 241Am, and 16.4 wt% O2. The neutron source emissions from the MOX sample are given in Fig. 2b; the source emitted approximately 80,000 neutrons/s, of which 53,000 neutrons/s came from spontaneous fission and the remainder from (alpha,n) events in the oxide. Correlated photon and neutron emissions were measured for 68 min. 2.2. PSD by charge integration technique A charge-integration technique was used for classifying photons and neutrons. Pulses created from neutrons contain more light in their tail region than pulses created from gamma-ray photons; this behavior is quantified by integrating the total pulse and comparing it to the tail integral. Because the detectors used in the MOX measurement were operated at a different gain than in the 252Cf measurement, PSD was optimized separately for each measurement. Fig. 3 illustrates the PSD for each scintillator type in each measurement expressed as the tail and total integrals for each pulse, where the top region corresponds to neutron pulses and the bottom region corresponds to photon pulses. To obtain adequate PSD for the EJ-299-33 plastic, it was necessary to raise the light output threshold to 150 keVee, corresponding to a neutron energy threshold of 1.7 MeV [15]. The EJ-309 liquid was capable of PSD at a 70-keVee light output threshold, corresponding to 650 keV in neutron energy [16]. For each measurement the EJ-309 liquid gave better separation of the neutron and photon regions than the EJ-299-33 plastic, particularly at lower total integrals (energies).
Fig. 2. The setup of the mixed oxide cross-correlation measurement (a) and the mixed oxide neutron emissions (b) [17].
This holds true even when operating at a lower neutron energy threshold than the plastic. Each PSD plot was sliced using a specialized MATLAB routine designed to apply PSD to measured pulses that is dependent on the tail and total integral of each pulse [14]. The routine starts by slicing each integral plot in Fig. 3 perpendicular to the average of the slopes of the neutron and photon regions. Ideally, each slice would be perpendicular to both regions simultaneously, but this method was chosen as the closest approximation. Within each slice the pulses are histogrammed by their tail-to-total integral ratio, creating photon and neutron peaks such as those shown in Fig. 4. The resulting histogram is fit as a sum of two Gaussians, where each Gaussian represents particles classified as either photons or neutrons. The individual Gaussian fits are used to find the tail-to-total integral ratio that minimizes the misclassification rate of each particle, which is defined as the fraction of its Gaussian integral that is on the incorrect side of the discrimination point. The discrimination points from every slice were then fit to a 2nd-order polynomial to obtain a particle discrimination curve, which was used to classify every pulse as a neutron or photon.
462
M.M. Bourne et al. / Nuclear Instruments and Methods in Physics Research A 784 (2015) 460–464
Fig. 3. PSD tail-vs-total integral plots for EJ-309 (a, c) and EJ-299-33 (b, d) measuring 252Cf and mixed oxide, respectively. EJ-309 pulses were measured at a 650-keV neutron energy threshold, while EJ-299-33 pulses were measured at a 1.7-MeV neutron energy threshold.
Fig. 4. Tail-to-total integral ratio histogram for one slice from EJ-309, for illustrating the algorithm used by the MATLAB routine for finding PSD in each measurement.
detection physics obtained from the MCNPX-PoliMi simulation are then read by MPPost to calculate detector response, accounting for the neutron light output, detector resolution, and the time difference between correlated particles [15]. This approach allows for accurate modeling of a measured cross-correlation distribution for spontaneous fission sources. Fig. 5 shows the simulated geometry used for the 252Cf and MOX measurements. For the 252Cf measurement, the detectors were modeled 10 cm above a steel table with a thickness of 3 mm. The 252Cf source was simulated as a point source located at the center of the geometry 20 cm from each detector, and the built-in MCNPX-PoliMi source was used. Due to incomplete knowledge of the EJ-299-33 plastic detector assembly, only the PMTs for the EJ-309 liquids were modeled. In the MOX-sample simulation, the aged isotopics of the sample were used to estimate the source intensity from spontaneous fission or (alpha,n) of each isotope; these were simulated using the mixedoxide source specification in MCNPX-PoliMi. The aluminum detector holder, the measurement table, and dry air were simulated.
3. Monte Carlo modeling 4. Cross-correlation results When modeling time-correlated events, the Monte Carlo code MCNPX is limited when modeling single interactions [12]. For this reason, MCNP-PoliMi was chosen for modeling these measurements. The latest version, MCNPX-PoliMi, implements the new features available in MCNPX into MCNP-PoliMi [13]. Detailed
4.1. Cross-correlation distributions Constant fraction timing, taken at half of the pulse amplitude along the leading edge, was applied to all correlated pulses
M.M. Bourne et al. / Nuclear Instruments and Methods in Physics Research A 784 (2015) 460–464
4.2. Discriminating MOX from
463
252
Cf
The cross-correlation distributions of 252Cf and MOX allow for separate identification of each. 252Cf has higher neutron multiplicity than even-numbered isotopes of plutonium; this property is used by comparing total neutron–neutron correlations from each measurement. The NN efficiency EffNN is defined as follows: R NN correlations Ef f NN ¼ 2 sr 2 s 1 ; Ω ðmeasurement timeÞ
Fig. 5. Modeled geometry for the
252
Cf (a) and mixed oxide (b) measurements.
measured in 1801 detector pairs to determine the time at which a pulse occurred. Correlated pulse times were subtracted, and these time differences were histogrammed by correlation type to obtain the cross-correlation distribution for each measurement. Fig. 6 shows the measured and simulated cross-correlation distributions for 252Cf and MOX using EJ-309 liquid and EJ-299-33 plastic scintillators. Particles classified as neutrons (N) and photons (P) are labeled in each correlation type. It follows that the shape of each curve is dependent on the time-of-flight of each particle; NN and PP correlations appear symmetric about 0 ns, while NP and PN correlations appear shifted by 710 ns. The EJ-309 detector was more efficient at recording NN correlations than the EJ-299-33 detector, giving a factor-of-5 increase in the 252Cf measurement. Analysis of three EJ-309 detector pairs gave a factor-of-10 increase in NN efficiency per detector pair in the MOX measurement. The EJ-299-33 contains significant peaks in the PP correlation at 710 ns; since correlated photon emissions should be detected simultaneously, these were identified as NP and PN correlations that were misclassified as PP correlations. The EJ-299-33 also contains a higher peak in the NP and PN correlations at 0 ns relative to the EJ-309, identified primarily as PP correlations misclassified as NP and PN correlations. The MCNPX-PoliMi model shows that secondary photon production will also contribute to this peak. The MCNPX-PoliMi model is the most accurate when predicting the EJ-309 detector response to 252Cf, where NN, NP, and PN correlations agree to within 10%. Meanwhile, there is larger discrepancy in neutron components for the EJ-299-33 plastic, suggesting that the model underestimates the neutron efficiency in these detectors. The simulated PP contribution appears low because the aged 252Cf source includes numerous gamma-emitting isotopes that are not included in the model. This effect is magnified in the MOX sample, since the MOX sample has many more reaction channels through which to decay than 252Cf.
where Ω is the solid angle of each detector and is squared to account for the efficiency of two detectors simultaneously. Table 1 shows that both the EJ-309 and EJ-299-33 see a significant difference in EffNN when measuring 252Cf or MOX. This suggests that NN correlations can be used to determine if correlated neutron emissions are due primarily to 252Cf or to evennumbered isotopes of plutonium, although the NN efficiency is also dependent on source shielding and source intensity. If three or more scintillation detectors are used simultaneously, triples neutron correlations (NNN) would give a stronger metric for distinguishing 252Cf from plutonium since 252Cf emits an average of 3.8 neutrons per fission while plutonium emits only 2.4. Longer measurement times would be necessary due to the lowered efficiency for triples correlations. Significant (alpha,n) events in the MOX source add additional uncorrelated neutron counts to this measurement. This fact, when combined with lower multiplicity in even-numbered plutonium isotopes, suggests that the uncorrelated-to-correlated neutron ratio, or neutron single-to-double ratio, should be higher in MOX than 252Cf. Table 1 shows that the neutron single-to-double ratio is about a factor-of-8 greater in the MOX sample than 252Cf, confirming that this ratio can be used for discriminating MOX from 252 Cf. More importantly, single-to-double ratio accounts for differences in source intensity and is less dependent on source shielding since the singles and doubles rates are each dependent on source shielding.
5. Conclusions We have presented 252Cf and MOX cross-correlation distributions measured with 7.62-cm EJ-299-33 plastic and EJ-309 organic-liquid scintillators. A standard charge integration PSD technique was applied to each detector to distinguish neutrons and photons. EJ-299-33 sacrifices detection efficiency and PSD capability relative to EJ-309; despite this limitation, EJ-299-33 was still capable of utilizing neutron–neutron correlations and neutron single-to-double ratio to distinguish 252Cf from MOX. This suggests that EJ-299-33 can be utilized to characterize correlated samples in applications where accurate particle classification is not necessary at neutron energies lower than 1.7 MeV. MCNPX-PoliMi was used to model these measurements, and MPPost was used to predict the cross-correlation distribution for each detector. The model predicted the measured EJ-309 crosscorrelation function for 252Cf to within 10% for neutron contributions, while the EJ-299-33 cross-correlation function appears to under-predict the neutron efficiency in EJ-299-33. The aged MOX sample contains numerous photon emitters that were not included in the MCNPX-PoliMi model. Future work will seek to apply this method to various samples of plutonium metal and oxide to determine if EJ-299-33 can use cross-correlation to distinguish plutonium samples of varying composition and burn-up. Additionally, the MCNPX-PoliMi model for EJ-299-33 will be updated to better model light production from neutron scatters.
464
M.M. Bourne et al. / Nuclear Instruments and Methods in Physics Research A 784 (2015) 460–464
Fig. 6. Cross-correlation distributions measured with EJ-309 (a, c) and EJ-299-33 (b, d) for threshold, while EJ-299-33 pulses were recorded at a 1.7-MeV threshold.
Table 1 Numerical parameters used to discriminate computed for each scintillator. EffNN (sr 2 s 1)
252
Cf and mixed oxide samples,
Single-to-double ratio
Detector
252
EJ-309 EJ-299-33
1.4 0.31
Cf
MOX
252
0.32 0.025
160 370
Cf
MOX 1200 3100
Acknowledgment This research was funded by the National Science Foundation, USA, and the Domestic Nuclear Detection Office of the U.S. Department of Homeland Security through Academic Research Initiative Award # CMMI 0938909. References [1] F.D. Brooks, Nuclear Instruments and Methods 162 (1979) 477. [2] S.A. Pozzi, S.D. Clarke, M. Flaska, P. Peerani, Nuclear Instruments and Methods in Physics Research Section A 608 (2) (2009) 310. [3] Eljen Technology (2013), EJ-299-33 PSD Plastic Scintillator Provisional Data Sheet, Retrieved February 28, 2014 from 〈http://www.eljentechnology.com/ images/stories/Data_Sheets/Plastic_Scintillators/ej299-33%20psd%20data% 20sheet.pdf〉. [4] Eljen Technology (2014), EJ-309 Liquid Scintillator Pulse-Shape Discrimination Properties, Retrieved July 25, 2014 from 〈http://www.eljentechnology.com/ images/stories/Data_Sheets/Liquid_Scintillators/EJ309 data sheet.pdf〉.
252
Cf and mixed oxide, respectively. EJ-309 pulses were recorded at a 650-keV
[5] S.A. Pozzi, M.M. Bourne, S.D. Clarke, Nuclear Instruments and Methods A 21 (2013) 19. [6] Natalia Zaitseva, Benjamin L. Rupert, Iwona PaweŁczak, Andrew Glenn, H. Paul Martinez, Leslie Carman, Michelle Faust, Nerine Cherepy, Stephen Payne, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 668 (2012) 88. [7] D. Cester, G. Nebbia, L. Stevanato, F. Pino, G. Viesti, Nuclear Instruments and Methods in Physics Research A 735 (2014) 202. [8] S. Nyibule, E. Henry, W.U. Schröder, J. Tõke, L. Acosta, L. Auditore, G. Cardella, E. De Filippo, L. Francalanza, S. Gìani, T. Minniti, E. Morgana, E.V. Pagano, S. Pirrone, G. Politi, L. Quattrocchi, F. Rizzo, P. Russotto, A. Trifirò, M. Trimarchi, Nuclear Instruments and Methods in Physics Research A 728 (2013) 36. [9] S.D. Clarke, M. Flaska, S.A. Pozzi, P. Peerani, Nuclear Instruments and Methods in Physics Research A 604 (3) (2009) 618. [10] E.C. Miller, S.D. Clarke, M. Flaska, S. Prasad, S.A. Pozzi, E. Padovani, Journal of Nuclear Materials Management XL (2) (2012) 34. [11] ET Enterprises, (2011), Photomultipliers from ET Enterprises Ltd, Retrieved December 20, 2012 from 〈http://www.et-enterprises.com/〉. [12] D. Pelowitz (Ed.), MCNPX User's Manual, version 2.5.0, Los Alamos National Laboratory, 2005 (LA-CP-05-0369). [13] S.A. Pozzi, S.D. Clarke, W.J. Walsh, E.C. Miller, J.L. Dolan, M. Flaska, B.M. Wieger, A. Enqvist, E. Padovani, J.K. Mattingly, D.L. Chichester, P. Peerani, Nuclear Instruments and Methods A 694 (2012) 119. [14] J.K. Polack, M. Flaska, A. Enqvist, S.A. Pozzi, A computer-aided, visual, chargeintegration, pulse-shape-discrimination method for organic scintillators, in: Proceedings of the INMM 54th Annual Meeting, Palm Desert, CA, July 2013. [15] C.C. Lawrence, M. Febbraro, T.N. Massey, M. Flaska, F.D. Becchetti, S.A. Pozzi, Nuclear Instruments and Methods in Physics Research Section A 759 (2014) 16. [16] A. Enqvist, C.C. Lawrence, B.M. Wieger, S.A. Pozzi, T.N. Massey, Nuclear Instruments and Methods in Physics Research Section A 715 (2013) 79. [17] J.L. Dolan, M. Flaska, A. Poitrasson-Riviere, A. Enqvist, P. Peerani, D.L. Chichester, S. A. Pozzi, Nuclear Instruments and Methods in Physics Research Section A 763 (2014) 565.
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research A journal homepage: www.elsevier.com/locate/nima
Cross-correlation measurements with the EJ-299-33 plastic scintillator Mark M. Bourne a,n, Jeff Whaley a, Jennifer L. Dolan a,1, John K. Polack a, Marek Flaska a, Shaun D. Clarke a, Alice Tomanin b, Paolo Peerani b, Sara A. Pozzi a a b
Department of Nuclear Engineering and Radiological Sciences, University of Michigan, Ann Arbor, MI 48109, USA European Commission, Joint Research Centre, Institute for Transuranium Elements, via Enrico Fermi 2749, 21027 Ispra (VA), Italy
art ic l e i nf o
a b s t r a c t
Available online 31 October 2014
New organic–plastic scintillation compositions have demonstrated pulse-shape discrimination (PSD) of neutrons and gamma rays. We present cross-correlation measurements of 252Cf and mixed uranium– plutonium oxide (MOX) with the EJ-299-33 plastic scintillator. For comparison, equivalent measurements were performed with an EJ-309 liquid scintillator. Offline, digital PSD was applied to each detector. These measurements show that EJ-299-33 sacrifices a factor of 5 in neutron–neutron efficiency relative to EJ-309, but could still utilize the difference in neutron–neutron efficiency and neutron singleto-double ratio to distinguish 252Cf from MOX. These measurements were modeled with MCNPX-PoliMi, and MPPost was used to convert the detailed collision history into simulated cross-correlation distributions. MCNPX-PoliMi predicted the measured 252Cf cross-correlation distribution for EJ-309 to within 10%. Greater photon uncertainty in the MOX sample led to larger discrepancy in the simulated MOX cross-correlation distribution. The modeled EJ-299-33 plastic also gives reasonable agreement with measured cross-correlation distributions, although the MCNPX-PoliMi model appears to under-predict the neutron detection efficiency. Published by Elsevier B.V.
Keywords: EJ-299-33 plastic scintillator MCNP MCNPX-PoliMi Cross-correlation distribution Neutron measurements
1. Introduction The recent 3He shortage has driven efforts towards developing alternative neutron detectors for nuclear nonproliferation applications. Organic liquid scintillators offer numerous advantages for fast neutron detection due to fast pulse timing, good efficiency, and excellent pulse-shape discrimination (PSD) [1,2]. For use in field applications, however, liquid scintillators are limited due to the potential risk from leaks or fire [4]. The EJ-299-33 plastic scintillator developed in late 2012 is of particular interest in nuclear nonproliferation applications due to its proven PSD capabilities as a plastic while eliminating environmental risks specific to liquid scintillators in field applications [3,5–8]. Cross-correlation measurements have been previously used for characterization of fission sources through detection and analysis n Correspondence to: Department of Nuclear Engineering & Radiological Sciences, University of Michigan, 2355 Bonisteel Boulevard, Ann Arbor, MI 481092104, USA. E-mail addresses: [email protected] (M.M. Bourne), [email protected] (J. Whaley), [email protected] (J.L. Dolan), [email protected] (J.K. Polack), mfl[email protected] (M. Flaska), [email protected] (S.D. Clarke), [email protected] (A. Tomanin), [email protected] (P. Peerani), [email protected] (S.A. Pozzi). 1 Jennifer Dolan was attending the University of Michigan at the time of the Ispra measurement campaign. She has since graduated.
http://dx.doi.org/10.1016/j.nima.2014.10.052 0168-9002/Published by Elsevier B.V.
of correlated neutron and gamma-ray photon emissions. Neutron– neutron correlations consist primarily of spontaneous fission neutrons from 252Cf and even-numbered plutonium isotopes, allowing for easy discrimination from single neutron emitters such as (alpha,n) sources. Varying multiplicity of spontaneous fission sources means that the intensity of neutron–neutron correlations may allow for separate identification of 252Cf from plutonium-based samples. Photon–neutron or photon–photon correlations can be detected from correlated sources such as fission or annihilation or accidentally from decay product buildup; these have been previously used to identify varying levels of burnup in plutonium-based fuel samples [9]. In this study, we present the EJ-299-33 plastic used in cross-correlation measurements of separate 252Cf and mixed uranium–plutonium oxide (MOX) samples. This scintillation detector is directly benchmarked to EJ-309 liquid scintillators, with proven PSD capabilities [1,2], with the same scintillator dimensions, photomultiplier tube model with equivalent base, and digital readout system. Accurate modeling of scintillation detector response allows for fast, inexpensive testing of detector performance in various environments and geometries. To further this goal for cross-correlation applications, these measurements were modeled using MCNPXPoliMi. MPPost, a specialized FORTRAN post-processing script, was used to convert the detection physics calculated by MCNPX-PoliMi into the cross-correlation response generated by each [10].
M.M. Bourne et al. / Nuclear Instruments and Methods in Physics Research A 784 (2015) 460–464
461
2. Cross-correlation measurements The EJ-299-33 plastic scintillation detectors were manufactured and assembled by Eljen Technologies. The crystals are right circular cylinders with 7.62-cm diameter and 7.62-cm depth. They were mounted to 7.62-cm diameter photomultiplier tubes (PMTs) ETL 9821B [11]. The EJ-309 detector was manufactured by SCIONIX using the same scintillator dimensions and mounted to the same PMT model as that used for the plastic scintillators. Anode signals of each PMT were digitized using a CAEN V1720 digitizer, with a 250-MHz sampling rate, 12-bit resolution, and 2-V dynamic range. Pulses were then analyzed offline to apply PSD and compute correlation timing using custom algorithms.
Fig. 1. The setup of the
252
Cf cross-correlation measurement.
2.1. Experimental setup The 252Cf measurement was performed at the University of Michigan's Department of Nuclear Engineering and Radiological Sciences in Ann Arbor, Michigan. The 252Cf measurement geometry is shown in Fig. 1. Two EJ-309 detectors and two EJ-299-33 detectors were placed in 1801 pairs at a distance of 20 cm from a 252Cf spontaneous fission source emitting 175,000 neutrons/s. All four detectors were gain-matched such that the 137Cs Compton edge aligned with approximately 400 mV; this allowed for detection of 95% of the 252Cf neutron energy spectrum with EJ-309 and 98% of the 252 Cf neutron energy spectrum with EJ-299-33. Correlated gamma-ray photon and neutron emissions from the 252Cf source were measured for 18 h. The MOX measurement was performed at the Joint Research Centre in Ispra, Italy, with the geometry shown in Fig. 2a. Twelve EJ-309 detectors and two EJ-299-33 detectors were placed in similar 1801 pairs at a 16.8 cm source–detector distance. Detectors were arranged octagonally around a MOX sample, stacking the EJ-309 liquids for improved efficiency. All 14 detectors were gain-matched to a 137Cs Compton edge of 300 mV, increasing the detectable 252Cf neutron energy range to 98% at the cost of resolution. Due to the high gamma emission rate from the source, a 1-cm lead sleeve was placed around the MOX sample. The MOX has a mass of 1.01 kg and consists of 66.8 wt% 238U, 0.4 wt% 235U, 11.1 wt% 239Pu, 4.6 wt% 240Pu, 0.2 wt% 241 Pu, 0.3 wt% 242Pu, 0.02 wt% 238Pu, 0.5 wt% 241Am, and 16.4 wt% O2. The neutron source emissions from the MOX sample are given in Fig. 2b; the source emitted approximately 80,000 neutrons/s, of which 53,000 neutrons/s came from spontaneous fission and the remainder from (alpha,n) events in the oxide. Correlated photon and neutron emissions were measured for 68 min. 2.2. PSD by charge integration technique A charge-integration technique was used for classifying photons and neutrons. Pulses created from neutrons contain more light in their tail region than pulses created from gamma-ray photons; this behavior is quantified by integrating the total pulse and comparing it to the tail integral. Because the detectors used in the MOX measurement were operated at a different gain than in the 252Cf measurement, PSD was optimized separately for each measurement. Fig. 3 illustrates the PSD for each scintillator type in each measurement expressed as the tail and total integrals for each pulse, where the top region corresponds to neutron pulses and the bottom region corresponds to photon pulses. To obtain adequate PSD for the EJ-299-33 plastic, it was necessary to raise the light output threshold to 150 keVee, corresponding to a neutron energy threshold of 1.7 MeV [15]. The EJ-309 liquid was capable of PSD at a 70-keVee light output threshold, corresponding to 650 keV in neutron energy [16]. For each measurement the EJ-309 liquid gave better separation of the neutron and photon regions than the EJ-299-33 plastic, particularly at lower total integrals (energies).
Fig. 2. The setup of the mixed oxide cross-correlation measurement (a) and the mixed oxide neutron emissions (b) [17].
This holds true even when operating at a lower neutron energy threshold than the plastic. Each PSD plot was sliced using a specialized MATLAB routine designed to apply PSD to measured pulses that is dependent on the tail and total integral of each pulse [14]. The routine starts by slicing each integral plot in Fig. 3 perpendicular to the average of the slopes of the neutron and photon regions. Ideally, each slice would be perpendicular to both regions simultaneously, but this method was chosen as the closest approximation. Within each slice the pulses are histogrammed by their tail-to-total integral ratio, creating photon and neutron peaks such as those shown in Fig. 4. The resulting histogram is fit as a sum of two Gaussians, where each Gaussian represents particles classified as either photons or neutrons. The individual Gaussian fits are used to find the tail-to-total integral ratio that minimizes the misclassification rate of each particle, which is defined as the fraction of its Gaussian integral that is on the incorrect side of the discrimination point. The discrimination points from every slice were then fit to a 2nd-order polynomial to obtain a particle discrimination curve, which was used to classify every pulse as a neutron or photon.
462
M.M. Bourne et al. / Nuclear Instruments and Methods in Physics Research A 784 (2015) 460–464
Fig. 3. PSD tail-vs-total integral plots for EJ-309 (a, c) and EJ-299-33 (b, d) measuring 252Cf and mixed oxide, respectively. EJ-309 pulses were measured at a 650-keV neutron energy threshold, while EJ-299-33 pulses were measured at a 1.7-MeV neutron energy threshold.
Fig. 4. Tail-to-total integral ratio histogram for one slice from EJ-309, for illustrating the algorithm used by the MATLAB routine for finding PSD in each measurement.
detection physics obtained from the MCNPX-PoliMi simulation are then read by MPPost to calculate detector response, accounting for the neutron light output, detector resolution, and the time difference between correlated particles [15]. This approach allows for accurate modeling of a measured cross-correlation distribution for spontaneous fission sources. Fig. 5 shows the simulated geometry used for the 252Cf and MOX measurements. For the 252Cf measurement, the detectors were modeled 10 cm above a steel table with a thickness of 3 mm. The 252Cf source was simulated as a point source located at the center of the geometry 20 cm from each detector, and the built-in MCNPX-PoliMi source was used. Due to incomplete knowledge of the EJ-299-33 plastic detector assembly, only the PMTs for the EJ-309 liquids were modeled. In the MOX-sample simulation, the aged isotopics of the sample were used to estimate the source intensity from spontaneous fission or (alpha,n) of each isotope; these were simulated using the mixedoxide source specification in MCNPX-PoliMi. The aluminum detector holder, the measurement table, and dry air were simulated.
3. Monte Carlo modeling 4. Cross-correlation results When modeling time-correlated events, the Monte Carlo code MCNPX is limited when modeling single interactions [12]. For this reason, MCNP-PoliMi was chosen for modeling these measurements. The latest version, MCNPX-PoliMi, implements the new features available in MCNPX into MCNP-PoliMi [13]. Detailed
4.1. Cross-correlation distributions Constant fraction timing, taken at half of the pulse amplitude along the leading edge, was applied to all correlated pulses
M.M. Bourne et al. / Nuclear Instruments and Methods in Physics Research A 784 (2015) 460–464
4.2. Discriminating MOX from
463
252
Cf
The cross-correlation distributions of 252Cf and MOX allow for separate identification of each. 252Cf has higher neutron multiplicity than even-numbered isotopes of plutonium; this property is used by comparing total neutron–neutron correlations from each measurement. The NN efficiency EffNN is defined as follows: R NN correlations Ef f NN ¼ 2 sr 2 s 1 ; Ω ðmeasurement timeÞ
Fig. 5. Modeled geometry for the
252
Cf (a) and mixed oxide (b) measurements.
measured in 1801 detector pairs to determine the time at which a pulse occurred. Correlated pulse times were subtracted, and these time differences were histogrammed by correlation type to obtain the cross-correlation distribution for each measurement. Fig. 6 shows the measured and simulated cross-correlation distributions for 252Cf and MOX using EJ-309 liquid and EJ-299-33 plastic scintillators. Particles classified as neutrons (N) and photons (P) are labeled in each correlation type. It follows that the shape of each curve is dependent on the time-of-flight of each particle; NN and PP correlations appear symmetric about 0 ns, while NP and PN correlations appear shifted by 710 ns. The EJ-309 detector was more efficient at recording NN correlations than the EJ-299-33 detector, giving a factor-of-5 increase in the 252Cf measurement. Analysis of three EJ-309 detector pairs gave a factor-of-10 increase in NN efficiency per detector pair in the MOX measurement. The EJ-299-33 contains significant peaks in the PP correlation at 710 ns; since correlated photon emissions should be detected simultaneously, these were identified as NP and PN correlations that were misclassified as PP correlations. The EJ-299-33 also contains a higher peak in the NP and PN correlations at 0 ns relative to the EJ-309, identified primarily as PP correlations misclassified as NP and PN correlations. The MCNPX-PoliMi model shows that secondary photon production will also contribute to this peak. The MCNPX-PoliMi model is the most accurate when predicting the EJ-309 detector response to 252Cf, where NN, NP, and PN correlations agree to within 10%. Meanwhile, there is larger discrepancy in neutron components for the EJ-299-33 plastic, suggesting that the model underestimates the neutron efficiency in these detectors. The simulated PP contribution appears low because the aged 252Cf source includes numerous gamma-emitting isotopes that are not included in the model. This effect is magnified in the MOX sample, since the MOX sample has many more reaction channels through which to decay than 252Cf.
where Ω is the solid angle of each detector and is squared to account for the efficiency of two detectors simultaneously. Table 1 shows that both the EJ-309 and EJ-299-33 see a significant difference in EffNN when measuring 252Cf or MOX. This suggests that NN correlations can be used to determine if correlated neutron emissions are due primarily to 252Cf or to evennumbered isotopes of plutonium, although the NN efficiency is also dependent on source shielding and source intensity. If three or more scintillation detectors are used simultaneously, triples neutron correlations (NNN) would give a stronger metric for distinguishing 252Cf from plutonium since 252Cf emits an average of 3.8 neutrons per fission while plutonium emits only 2.4. Longer measurement times would be necessary due to the lowered efficiency for triples correlations. Significant (alpha,n) events in the MOX source add additional uncorrelated neutron counts to this measurement. This fact, when combined with lower multiplicity in even-numbered plutonium isotopes, suggests that the uncorrelated-to-correlated neutron ratio, or neutron single-to-double ratio, should be higher in MOX than 252Cf. Table 1 shows that the neutron single-to-double ratio is about a factor-of-8 greater in the MOX sample than 252Cf, confirming that this ratio can be used for discriminating MOX from 252 Cf. More importantly, single-to-double ratio accounts for differences in source intensity and is less dependent on source shielding since the singles and doubles rates are each dependent on source shielding.
5. Conclusions We have presented 252Cf and MOX cross-correlation distributions measured with 7.62-cm EJ-299-33 plastic and EJ-309 organic-liquid scintillators. A standard charge integration PSD technique was applied to each detector to distinguish neutrons and photons. EJ-299-33 sacrifices detection efficiency and PSD capability relative to EJ-309; despite this limitation, EJ-299-33 was still capable of utilizing neutron–neutron correlations and neutron single-to-double ratio to distinguish 252Cf from MOX. This suggests that EJ-299-33 can be utilized to characterize correlated samples in applications where accurate particle classification is not necessary at neutron energies lower than 1.7 MeV. MCNPX-PoliMi was used to model these measurements, and MPPost was used to predict the cross-correlation distribution for each detector. The model predicted the measured EJ-309 crosscorrelation function for 252Cf to within 10% for neutron contributions, while the EJ-299-33 cross-correlation function appears to under-predict the neutron efficiency in EJ-299-33. The aged MOX sample contains numerous photon emitters that were not included in the MCNPX-PoliMi model. Future work will seek to apply this method to various samples of plutonium metal and oxide to determine if EJ-299-33 can use cross-correlation to distinguish plutonium samples of varying composition and burn-up. Additionally, the MCNPX-PoliMi model for EJ-299-33 will be updated to better model light production from neutron scatters.
464
M.M. Bourne et al. / Nuclear Instruments and Methods in Physics Research A 784 (2015) 460–464
Fig. 6. Cross-correlation distributions measured with EJ-309 (a, c) and EJ-299-33 (b, d) for threshold, while EJ-299-33 pulses were recorded at a 1.7-MeV threshold.
Table 1 Numerical parameters used to discriminate computed for each scintillator. EffNN (sr 2 s 1)
252
Cf and mixed oxide samples,
Single-to-double ratio
Detector
252
EJ-309 EJ-299-33
1.4 0.31
Cf
MOX
252
0.32 0.025
160 370
Cf
MOX 1200 3100
Acknowledgment This research was funded by the National Science Foundation, USA, and the Domestic Nuclear Detection Office of the U.S. Department of Homeland Security through Academic Research Initiative Award # CMMI 0938909. References [1] F.D. Brooks, Nuclear Instruments and Methods 162 (1979) 477. [2] S.A. Pozzi, S.D. Clarke, M. Flaska, P. Peerani, Nuclear Instruments and Methods in Physics Research Section A 608 (2) (2009) 310. [3] Eljen Technology (2013), EJ-299-33 PSD Plastic Scintillator Provisional Data Sheet, Retrieved February 28, 2014 from 〈http://www.eljentechnology.com/ images/stories/Data_Sheets/Plastic_Scintillators/ej299-33%20psd%20data% 20sheet.pdf〉. [4] Eljen Technology (2014), EJ-309 Liquid Scintillator Pulse-Shape Discrimination Properties, Retrieved July 25, 2014 from 〈http://www.eljentechnology.com/ images/stories/Data_Sheets/Liquid_Scintillators/EJ309 data sheet.pdf〉.
252
Cf and mixed oxide, respectively. EJ-309 pulses were recorded at a 650-keV
[5] S.A. Pozzi, M.M. Bourne, S.D. Clarke, Nuclear Instruments and Methods A 21 (2013) 19. [6] Natalia Zaitseva, Benjamin L. Rupert, Iwona PaweŁczak, Andrew Glenn, H. Paul Martinez, Leslie Carman, Michelle Faust, Nerine Cherepy, Stephen Payne, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 668 (2012) 88. [7] D. Cester, G. Nebbia, L. Stevanato, F. Pino, G. Viesti, Nuclear Instruments and Methods in Physics Research A 735 (2014) 202. [8] S. Nyibule, E. Henry, W.U. Schröder, J. Tõke, L. Acosta, L. Auditore, G. Cardella, E. De Filippo, L. Francalanza, S. Gìani, T. Minniti, E. Morgana, E.V. Pagano, S. Pirrone, G. Politi, L. Quattrocchi, F. Rizzo, P. Russotto, A. Trifirò, M. Trimarchi, Nuclear Instruments and Methods in Physics Research A 728 (2013) 36. [9] S.D. Clarke, M. Flaska, S.A. Pozzi, P. Peerani, Nuclear Instruments and Methods in Physics Research A 604 (3) (2009) 618. [10] E.C. Miller, S.D. Clarke, M. Flaska, S. Prasad, S.A. Pozzi, E. Padovani, Journal of Nuclear Materials Management XL (2) (2012) 34. [11] ET Enterprises, (2011), Photomultipliers from ET Enterprises Ltd, Retrieved December 20, 2012 from 〈http://www.et-enterprises.com/〉. [12] D. Pelowitz (Ed.), MCNPX User's Manual, version 2.5.0, Los Alamos National Laboratory, 2005 (LA-CP-05-0369). [13] S.A. Pozzi, S.D. Clarke, W.J. Walsh, E.C. Miller, J.L. Dolan, M. Flaska, B.M. Wieger, A. Enqvist, E. Padovani, J.K. Mattingly, D.L. Chichester, P. Peerani, Nuclear Instruments and Methods A 694 (2012) 119. [14] J.K. Polack, M. Flaska, A. Enqvist, S.A. Pozzi, A computer-aided, visual, chargeintegration, pulse-shape-discrimination method for organic scintillators, in: Proceedings of the INMM 54th Annual Meeting, Palm Desert, CA, July 2013. [15] C.C. Lawrence, M. Febbraro, T.N. Massey, M. Flaska, F.D. Becchetti, S.A. Pozzi, Nuclear Instruments and Methods in Physics Research Section A 759 (2014) 16. [16] A. Enqvist, C.C. Lawrence, B.M. Wieger, S.A. Pozzi, T.N. Massey, Nuclear Instruments and Methods in Physics Research Section A 715 (2013) 79. [17] J.L. Dolan, M. Flaska, A. Poitrasson-Riviere, A. Enqvist, P. Peerani, D.L. Chichester, S. A. Pozzi, Nuclear Instruments and Methods in Physics Research Section A 763 (2014) 565.