- Email: [email protected]
γ discrimination by an emission-based phoswich approach
Radiation Measurements 129 (2019) 106203
Contents lists available at ScienceDirect
Radiation Measurements journal homepage: http://www.elsevier.com/locate/radmeas
Neutron/γ discrimination by an emission-based phoswich approach G. Zorloni a, F. Cova b, M. Caresana a, *, M. Di Benedetto a, J. Hosta�sa e, M. Fasoli b, I. Villa b, I. Veronese c, d, A. Fazzi a, A. Vedda b a
Department of Energy, Politecnico di Milano, Via Lambruschini 4, 20156, Milan, Italy Department of Materials Science, University of Milano - Bicocca, Via Cozzi 55, 20125, Milan, Italy c Dipartimento di Fisica, Universit� a degli Studi di Milano, Via Celoria 16 I, 20133, Milan, Italy d INFN Sezione di Milano, Via Celoria 16 I, 20133, Milan, Italy e CNR ISTEC, Institute of Science and Technology for Ceramics, Via Granarolo 64, 48018, Faenza, Italy b
A R T I C L E I N F O
A B S T R A C T
Keywords: Phoswich detector Particle discrimination Neutron detector
Particle discrimination in a mixed radiation field using scintillators is a challenging topic in radiation detection research. We propose a novel approach relying on the possibility of identifying particles interacting in a phos wich detector from the emission spectrum of the produced scintillation signal. As a proof of concept, we focus on the discrimination between neutrons and gamma rays. Lu3 Al5 O12 :Pr and Gd3 Al2 Ga3 O12 :Ce thin single crystal scintillators, coupled to two different silicon photo multipliers equipped with optical filters, are simultaneously used in the same phoswich detector. Their optical emissions peak at approximately 310 nm and 545 nm respectively, and thus they can easily be distinguished by optical filtering. While both crystals are sensitive to gamma rays, neutrons interact only with the Gd3 Al2 Ga3 O12 : Ce thanks to the presence of Gd acting as neutron converter. Optical filtration and an anti-coincidence algorithm are therefore used to perform particle discrimination, rejecting coincidence signals arising from gamma rays, which simultaneously deposit energy in both crystals, and counting anti-coincidence signals due to neutrons, which deposit energy only in the Gd3 Al2 Ga3 O12 :Ce. The simple neutron counter developed here is intended to be a proof of the principle of the feasibility of the color-based particle discrimination technique.
1. Introduction In recent years, new advances in inorganic solid state scintillator materials science, mainly triggered by the needs of high energy physics experiments and medical imaging (Lecoq, 2016), have led to the dis covery and the development of a wide variety of new scintillators (Nikl and Yoshikawa, 2015; Dujardin et al., 2018). Recent progress in mate rials science permits tuning and optimization of scintillator properties and characteristics, with studies of band gap engineering (Fasoli et al., 2011) and energy level positioning (Dorenbos, 2013). In this study we take advantage of some of these novel materials to propose a new par ticle discrimination technique. The standard way to perform particle discrimination with scintilla tors is so-called Pulse Shape Discrimination (PSD) (Knoll, 1999), a method based on the electronic analysis of the voltage pulse time behavior. As a potential alternative, in this work we investigate the
feasibility of a passive optical-based technique using phoswich detectors.1 As a proof of concept, we developed a simple system for neutrongamma discrimination consisting of a composite Gd3 Al2 Ga3 O12 :Ce/ Lu3 Al5 O12 :Pr (GGAG:Ce/LuAG:Pr) sensor. While both scintillators are sensitive to gamma rays, only GGAG:Ce interacts with thermal neutrons because of its high Gd content. After neutron capture, the Internal Conversion (IC) electrons deposit their energy only in the GGAG:Ce crystal, thus producing an anti-coincidence signal, and they are counted. On the other hand, because the scintillators are thin, the background gamma rays can deposit their energy in both scintillators, thus pro ducing a coincidence signal, and they are rejected. Since the two scin tillator species emit in well distinguished spectral regions, passive discrimination could be obtained by coupling the two scintillators to two distinct photodetectors through optical filters. In principle, this approach could lead to an improvement with
* Corresponding author. E-mail address: [email protected] (M. Caresana). 1 A phoswich, or phosphor sandwich, is the combination of two different scintillators usually coupled to a single photodetector (Knoll, 1999). https://doi.org/10.1016/j.radmeas.2019.106203 Received 18 March 2019; Received in revised form 11 October 2019; Accepted 16 October 2019 Available online 26 October 2019 1350-4487/© 2019 Elsevier Ltd. All rights reserved.
G. Zorloni et al.
Radiation Measurements 129 (2019) 106203
respect to an equivalent system relying on electronic signal analysis, which is affected by random noise and jitter effects: the PSD algorithms are substituted by the two optical filters, which perform discrimination prior to signal formation, with a passive physical-based quasi-digital filtering. The simple neutron counter developed in this work is not aimed at replacing existing detectors for neutron measurements, but for demon strating the feasibility of the proposed technique. Therefore, our inves tigation shows how a compact detector composed of different materials, coupled together in engineered geometries, can display attractive properties, useful for particle discrimination, leading to novel specific functionalities.
0.12 and 0.48 for 155 Gd and 157 Gd respectively (Kandlakunta et al., 2013). The energies of the main IC lines are around 30 keV, 70 keV and 130 keV. The continuous slowing down approximation range for 180 keV electrons, i.e. the highest energy IC line with a significant branching ratio, in the GGAG:Ce is about 90 μm. For the 70 keV elec trons the range in the GGAG:Ce is about 20 μm, while for the 30 keV is only a few microns. The calculations were performed using the ESTAR database (https://physics.nist.gov/, 2019). Since these electrons are extracted from the atomic shell of the nuclei, the atoms are, in turn, in excited states. They de-excite by emitting X rays, mainly around 40–50 keV for the K shell and 6–7 keV for the L shell, and low-energy Auger electrons (Cerullo et al., 2009; Kandlakunta and Cao, 2012).
2. Materials and methods 2.1. Characteristics of the sources
2.4. Experimental phoswich set-up
We tested the proposed methodology by developing a small detector system for thermal neutron counting in a mixed neutron/γ field. The tests were performed with two sources, one of thermal neutrons and one of gamma rays, thus observing the effects of the two types of radiation field independently on the system. Tests with neutrons were performed with an Am–Be neutron source placed at the bottom of a polyethylene cylinder, providing the neutron thermalization. Measurements were performed in the hollow upper part of the cylinder. Between the source and the cavity, a thin layer of lead attenuates the 60 keV gammas from 241 Am. This configuration mitigates the low energy photon contamination and provides a quasi-pure thermal neutron field (about 87%) with a fluence rate of about 400 neutrons per square centimeter per second within the measurement cavity. Some additional high-energy gamma contamination arises from neutron cap ture on carbon and hydrogen. Tests with gamma rays were performed with certified metrological grade gamma sources of 137 Cs (0.1, 1, 10 Ci). 137 Cs was chosen to be representative of the background gamma rays in neutron measurements.
A simple set-up has been developed to prove the proposed concept. Two twin circuits to read the photodetectors signals are realized. Two Silicon PhotoMultipliers (SiPMs) SensL (Cork, Ireland) ArrayJ-600354P-EVB are directly supported by the circuits, one in front of the other, acting also as supports to filters and crystals, without adding other structural components. The photon detection efficiency of the SiPMs is between 0.3 and 0.4 for both scintillator species at 6 V overvoltage. The phoswich is mounted between the photodetectors. The whole system is protected from light exposure by a sealed aluminum case. The neutron sensitive component consists of a commercial 5 � 5 � 0.1 mm3 slice of GGAG:Ce, produced by Kinheng Crystal Mate rial (Shangai, China). The small thickness was chosen from physical considerations about the neutron/γ interactions: i) it guarantees a low gamma sensitivity (i.e. 5‰ attenuation for a 137 Cs expanded aligned photon source); ii) the range of IC electrons from the (n,γ) reactions on Gd is short enough for them to be fully stopped in the crystal slice, since they are the reaction products the system should reveal. The second scintillator, insensitive to neutrons, is a commercial 5 � 5 � 0.5 mm3 LuAG:Pr crystal, produced by Kinheng Crystal Material (Shangai, China). In Table 1 the main characteristics of the crystals are briefly summarized. The set-up is completed by the two optical filters. The long-pass filter is an Andover Corporation (Salem, US) model 450FH90–12.5, the shortpass is a Knight Optical (Harrietsham, UK) model 465FCS5050. The phoswich setup, as described in Fig. 1, consists of, from the top to the bottom: first SiPM, long-pass filter, GGAG:Ce, LuAG:Pr, short-pass filter, second SiPM. The optical coupling between the phoswich components is provided by optical grease. The GGAG:Ce sensitive SiPM facing the longpass filter (hereafter denoted as n-channel) is connected to a CAEN (Viareggio, Italy) DT5770 multichannel analyzer, with the capability of anti-coincidence veto counting. The logic signal, which opens the veto window in the multichannel anti-coincidence module, is obtained by connecting the LuAG:Pr sensitive SiPM, i.e. the one facing the short-pass filter (denoted as γ-channel), to a programmable single channel analyzer unit. The single channel threshold is selectable by the user, thus it is possible to observe the effects of the rejection logic of the system as a function of the γ-channel veto threshold.
2.2. Description of the optical characterization apparatus The crystals were characterized by means of radioluminescence (RL) spectroscopy. RL measurements were performed using a homemade apparatus featuring a liquid nitrogen-cooled, back-illuminated and UVenhanced, charge-coupled device (CCD) (Jobin-Yvon Spectrum One 3000, Bensheim, Germany) coupled to a monochromator (Jobin-Yvon Triax 180) with a 100 grooves/mm grating as the detection system. RL excitation was obtained by X-ray irradiation through a beryllium win dow, using a Philips PW2274 (Andover, US) X-ray tube with tungsten anode operated at 20 kV. RL emission spectra were corrected for the spectral response of the detection system. The transmittance of the filters was measured with a Agilent (Santa Clara, US) Varian Cary 50 spectrophotometer from 190 to 1100 nm. 2.3. Neutron interactions with gadolinium Natural gadolinium comprises 7 natural isotopes, out of which 155 Gd and 157 Gd show the highest thermal cross sections (60,900 b and 254,000 b respectively, for a total cross section for natural Gd of about 48,800 b) (Dumazert et al., 2018). The products of the reaction are two stable isotopes, 156 Gd* and 158 Gd* in excited states, 8.536 MeV and 7.937 MeV respectively. The prompt γ cascade is characterized by the emission of about 3 photons per reaction on average, with a quasi-continuous energy distribution, with a maximum emission prob ability around 2 MeV (Dumazert et al., 2018). A competitive mechanism to the γ emission is the IC process, mainly at low energies (the maximum energy line of the IC electrons is 246 keV) (Harms and McCormack, 1974). IC yield per neutron capture is around
Table 1 Scintillation properties of LuAG:Pr and GGAG:Ce. ρ: density; Zeff: Z effective; τ: main decay time; LY: light yield; λpeak : emission wavelength peak. Parameter 3
ρ (g/cm ) Zeff ( ) τ main (ns) LY (phot/MeV) λpeak (nm) a b
2
LuAG:Pra
GGAG:Ceb
6.67 63 20 16,000–19,000 310
6.3 54 90 50,000–60,000 545
Data taken from (Nikl et al., 2013). Data taken from (Kamada et al., 2012).
G. Zorloni et al.
Radiation Measurements 129 (2019) 106203
energy of the photons (the attenuation of 2 MeV photons in the LuAG:Pr is about 1%). On the other hand, the low energy relaxation X rays will probably interact, with about 80% attenuation in the LuAG:Pr, so that the veto threshold of the γ-channel must be properly chosen not to trigger the veto on these interactions. False positive signals can arise from background gamma interactions which deposit energy only in the GGAG:Ce crystal, without interacting in the LuAG:Pr. 2.5. Description of the Monte Carlo simulations Monte Carlo (MC) simulations were performed to compare the experimental data and the theoretical predictions in the case of thermal neutron capture. MC results can also be used for the energy calibration of the system. In fact, the low thickness of the GGAG:Ce crystal prevents the 137 Cs photopeak formation. On the other hand, the resulting energy spectrum after neutron capture, which can provide, in principle, two main peaks around 30 and 70 keV, is unknown a priori. With the MC simulation it is possible to predict the system response to neutron cap ture, and this information can be used to assess the energy calibration of the detector. Calculations were performed using the MCNPX version 2.6 code (X-5 Monte Carlo Team, 2003) and a Matlab script was written for the post-processing elaboration of the MC results, mainly to account for summation effects of the atomic de-excitation processes. Simulations were performed in five steps:
Fig. 1. Schematic of the phoswich arrangement: the GGAG:Ce crystal is dis played in yellow, the LuAG:Pr crystal is represented in gray. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
The discrimination is obtained as follows: 1. When a neutron interacts in the GGAG:Ce, it produces a signal in the n-channel. The short range IC electrons are fully stopped in the GGAG:Ce slice, while the prompt gammas escape from the system. If the signal falls within a user-defined acceptance window (see 3.3 for the details), the event is recorded as a true positive neutron count; 2. When a background γ-ray interacts in the GGAG:Ce, or in the sur rounding materials, the secondary electrons could pass through both scintillators, because of the small thickness of the GGAG crystal. This event produces a simultaneous signal in both channels. If the γ-channel signal is above threshold, a logic signal is triggered, thus opening a veto time window in the anti-coincidence module of the nchannel. Any signal occurring within the veto window in the nchannel is flagged as a gamma-induced count, and is consequently rejected as a true negative.
1. Evaluation of the reaction rate along the GGAG:Ce thickness due to the thermal neutron capture; 2. Pulse height tally (energy deposition spectrum) of the prompt gamma rays after capture on Gd in the GGAG:Ce crystal; 3. Pulse height tally (energy deposition spectrum) of the prompt IC electrons after capture on Gd in the GGAG:Ce crystal; 4. Pulse height tally (energy deposition spectrum) of the X-rays and Auger electrons after the atomic relaxation process following the IC emission in the GGAG:Ce crystal; 5. Weighted combination (summation and superposition) of point 2, 3 and 4 contributions, accounting also for the Gaussian energy broadening of the crystal response, i.e. scintillator resolution.
Fig. 2 shows a schematic diagram of the readout acquisition logic. False negative signals can arise from interaction of a prompt γ, emitted simultaneously to IC electrons, in the Pr-doped scintillator. However, the interaction probability is rather low, because of the high mean
The first step was designed to evaluate the reaction rate profile along the crystal, since the high Gd cross section could in principle perturb the thermal neutron fluence along the crystal thickness. An isotropic Max wellian neutron source was simulated for modeling the experimental setup. The result was scored independently for the 155 Gd and 157 Gd isotopes. Other Gd isotopes supposedly do not significantly contribute to the reaction rate, since their cross section is several orders of magnitude lower than that of 155 Gd and 157 Gd (Dumazert et al., 2018). The same simulation was used to calculate the pulse height tally due to the prompt gamma contribution, using the ENDF/B-VII.0 library. The reaction rate profile obtained in the first step was used as an input parameter for the spatial distribution of the IC electrons, X-rays and Auger electrons treated as isotropic volumetric sources distributed within the GGAG:Ce crystal. Reaction rate was estimated independently for the two isotopes since the IC electrons and X-rays energy distributions are isotope dependent. The energy distribution spectra were finally combined considering superposition of the effects in the case of independent events and sum mation effects in the case of correlated events, e.g. X-rays cascade of the atomic relaxation after IC emission. The multiple gamma ray cascade was treated as a series of independent events, because of their high mean energy, i.e. negligible probability of multiple simultaneous correlated events.
Fig. 2. Logic block diagram of the readout circuit. The acceptance window of the n-channel is the region of the differential energy spectrum of the n-channel in which the signals are eligible to be counted as neutrons (description in 3.3). 3
G. Zorloni et al.
Radiation Measurements 129 (2019) 106203
3. Results and discussion
anti-coincidence logic is turned on, and if only waveshifting occurs, some positive signals will be collected. Otherwise, if no waveshifting occurs, or, for instance, if a limited fraction of the photons emitted by Pr generates waveshifting, the anticoincidence logic rejects the n-channel signal as a true negative. Measurements without external sources proved that the second condition was verified, thus demonstrating that, for the present prototype, waveshifting effects can be neglected.
3.1. Characterization of the crystals Both scintillators and filters were investigated by RL and optical absorption spectroscopy to verify if the chosen materials addressed the desired criteria. GGAG:Ce and LuAG:Pr normalized RL emission spectra are shown in Fig. 3. For both crystals, the main emission bands are ascribed to the allowed 5d-4f transition of Ce and Pr ions, peaking at around 545 nm and 310–375 nm, respectively. Emission lines from 4f states of Pr ions in the VIS - NIR region are also weakly visible in the RL spectrum of LuAG: Pr; they are related to forbidden transitions which are much slower (in the ms time scale) than the main 5d–4f bands (Blasse and Grabmaier, 1994) and can therefore be considered negligible for the scope of this work (the shaping time of the scintillation measurements was kept in the μs time scale). Fig. 3 displays also the transmittance spectra of the two selected optical filters, aimed to be sensitive only to the light of one out of the two materials. The transmittance of the filters in the region of interest is around 80%. With reference to Fig. 3, it can be noted that the short-pass filter is actually a band-stop filter. In any case, it addresses the desired criteria, that are i) transparency to LuAG:Pr emission, and ii) opacity to GGAG:Ce emission. It is worth noting that the optical emission of the LuAG:Pr crystal overlaps with the optical absorption of the GGAG:Ce crystal (Wu et al., 2015), which may give rise to false positive neutron signals by wave shifting effects.2 However, this kind of behavior was found to be negli gible from experimental evidence. In fact, the LuAG:Pr possesses an internal radioactivity, estimated to be roughly 20 Bq, due to the pres ence of the natural occurring 176 Lu isotope, which produces some signals in the LuAG:Pr or in both scintillators (176 Lu is a β emitter). If the
3.2. Monte Carlo simulations and comparison with experiments MC calculations were performed considering the complete set of reactions, as described in 2.3. This procedure is aimed at verifying that the system is able to reveal the IC electrons and relaxation cascade products with the 100 μm thick crystal. Moreover, MC simulations provided two points for the energy calibration of the detector. In Fig. 4 the numerical spectrum and the experimental n-channel spectrum are compared. MC calculations demonstrate that the main contribution to the growth of the spectrum is due to the convolution of IC electrons, Auger electrons and X rays signals. Thus, the two main peaks are centered around 32 keV and 74 keV. To better visualize the IC electron contribution, Fig. 4 shows also the main IC lines of natural Gd. The multichannel analysis and the comparison with MC simulations demonstrate that the system is actually capable of counting the IC electrons. The presence of the two peaks centered in correspondence to the main IC lines proves that the thickness of the crystal is sufficient for completely stop the IC electrons. The flat background is due to Compton scattered electrons after the interaction of prompt γ-rays. 3.3. Test of the phoswich setup Measurements were performed independently with the 137 Cs source and with the thermal neutron source. The counts recorded in the nchannel correspond to those seen within a 30–100 keV acceptance window of the multichannel analyzer (i.e. the area subtended by the two main peaks in Fig. 4). The lower acceptance window bound was chosen to eliminate the SiPM noise. The upper bound was chosen to subtend the main IC electron emission peak, excluding the high energy Compton
Fig. 3. RL emission spectra (solid lines) of LuAG:Pr crystal (purple, triangles) and GGAG:Ce crystal (green, circles) and transmittance spectra (short-pass blue dashed line, long-pass orange dotted line) of the two selected optical filters to be coupled to the two photodetectors. Transmittance values are reported on the right Y axis. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Fig. 4. Comparison between the simulated spectra (dashed line) and the measured one (solid line) in the n-channel after neutron capture. Counts are normalized. The yellow vertical lines correspond to the normalized IC emission probabilities of natural Gd after neutron capture. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
2 The term waveshift denotes the effect of a high energy radiation, such as the LuAG emitted photons, which in principle can be absorbed by a material, such as GGAG, and then re-emitted in a low energy spectral region.
4
G. Zorloni et al.
Radiation Measurements 129 (2019) 106203
electrons from the γ background. Fig. 5 (a) shows th Counts Per Second (CPS) recorded in the nchannel for events generating signals within the n-channel acceptance window, normalized per 1 Sv/h ambient dose equivalent (Report 39, 2016; Report 57, 2016), under different experimental conditions. The upper dashed line, in blue, represents the CPS recorded by irradiating with thermal neutrons without implementing the rejection logic. The circle dots represent the CPS recorded in the same condition as above, as a function of the γ-channel veto threshold but switching on the anti coincidence gamma rejection. The neutron efficiency is weakly depen dent on the γ-channel voltage threshold. The lower dashed line, in red, represents the CPS recorded by irra diating the detector with gamma rays only, without implementing the rejection logic. In this case the contribution represents a false positive, i. e. gamma induced neutron signal. The triangle dots represent the CPS recorded in the same condition as above, as a function of the γ-channel veto threshold but switching on the anticoincidence gamma rejection. The variation of the false positive rate is greatly reduced by triggering the veto signal with low voltage. In both neutron and gamma irradiation the veto signal time exten sion was set to 1 μs, which was found to be sufficiently long to account for jitter effects, and sufficiently short not to include independent/un correlated events. Normalization per unit ambient dose equivalent was used because it is a traditional way to compare radiation protection instrumentation, in particular neutron dosimeters and counters (Care sana et al., 2014; Aza et al., 2014). In Fig. 5 (b) the same information is expressed in terms of rejection percentage calculated as the ratio be tween the CPS recorded with the anticoincidence logic switched on divided by the CPS recorded without implementing the anticoincidence logic. The analysis was done for both neutrons (circle dots) and photons (triangle dots). With respect to the counts recorded without the rejection logic, lower dashed line in Fig. 5 (a), the false positives decrease as the veto threshold decreases, thus demonstrating the feasibility of the emissionbased discrimination method. In fact, the lower the γ-channel veto threshold, the higher the probability that a background gamma ray deposits enough energy for triggering the veto, thus increasing the discrimination capability of the system. On the other hand, a lower veto threshold leads also to a slight decrease of neutron detection efficiency. In this case a larger number of true positives are recorded as false neg atives because of the cross-talk between the two crystals. As an example, after neutron capture the prompt γ and X rays may deposit a certain amount of energy in LuAG:Pr, thus triggering a false coincidence window.
The 17% neutron rejection, Fig. 5 (b), even at high rejection threshold has to be analyzed in details for future improvements of the system. The exact cause of this percentage of rejection is currently un known, but two possible contributions can be mentioned. Firstly, the background gamma contamination of the source, which does not pro vide a purely neutron field, may cause an overestimation of the neutron sensitivity without the implementation of the anti-coincidence logic. Secondly, the dark counts due to the natural radioactivity of Lu, which are completely counteracted by the anti-coincidence logic at low threshold, may cause again an overestimation of the neutron sensitivity in the case of no anti-coincidence logic implemented. In particular, the 20 Bq activity of the internal source is not negligible with respect to the counts due to the 400 neutrons per square centimeter per second of the Am–Be source. A proper quantification of the two contributions is planned for future activities. However, these two issues may cause an overestimation of the neutron sensitivity without the anti-coincidence,
Fig. 6. Comparison between the n-channel multichannel spectra obtained by irradiating with neutrons (blue solid line) and 137 Cs (red dashed line) without implementing the anticoincidence logic. Results are normalized per unit ambient dose equivalent. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Fig. 5. (a): n-channel counts per unit ambient dose equivalent due to neutrons (blue, circles) and 662 keV photons (red, triangles) independently, as a function of the veto trigger threshold on the γ-channel. The dashed lines represent the counts recorded in the n-channel without implementing the anti-coincidence logic. (b): percentage of rejection with respect to the counts recorded without implementing the rejection logic. Lines connecting points are guides for the eye. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
5
G. Zorloni et al.
Radiation Measurements 129 (2019) 106203
but they are actually rejected by the discrimination logic. In Fig. 6, the n-channel multichannel spectra obtained by irradiating with neutrons and 137 Cs are compared in the case where no anti coincidence logic is implemented. The partial Compton edge produced by 137 Cs mostly overlaps with the acceptance window. The low crystal thickness prevents the formation of both the complete Compton edge and the photopeak. Even if from the multichannel point of view this is the worst case scenario (i.e. energy response overlap), the technique proved to be efficient up to 45% of rejection, as seen in Fig. 5 (b). It should be noted that the actual configuration is directional, i.e. a twoslice phoswich is far from the isotropic response, leaving wide room for geometrical improvements. We have demonstrated by a simple case that particle discrimination can be pursued by an emission-based approach. Among the most valu able aspects of this method, is the possibility to use a quasi-digital logic by performing the rejection prior to the electronic signal formation. This uniquely identifies in which out of the two phoswich components an event occurred. The simplicity of the electronics results because no al gorithms and no additional elements/electronic devices for pulse shape analysis are required. The drawbacks of the proposed technique are the traditional ones of a phoswich arrangement, for instance optical requirements and spillover/cross-talk due to the stochastic behavior of the particles in teractions in matter. Two additional drawbacks are typical of the pro posed approach itself, i) an expected worsening in the energy resolution, because more components are present in the phoswich (i.e. optical fil ters), and ii) the requirement of two independent readout circuits/ photodetectors. Regarding the first point, a preliminary test with the GGAG:Ce directly coupled without a filter to the SiPM showed that most of the gamma signal overlaps the acceptance window encompassing the neutron signal even with a better resolution. This feature makes impossible the separation of the two components. Despite the low percentage of gamma rejection, which can be regarded as still far from the value required for discrimination appli cations of this system, the results presented in this work leave wide room for improvements of the proposed approach. These can include an optimization of the crystals thickness through MC simulations, and the introduction of a three-slice phoswich i.e. LuAG:Pr-GGAG:Ce-LuAG:Pr, to enhance the isotropy of the response and the rejection capability.
Further perspectives could look towards a more optimized and compact geometry, also exploiting the use of a single specifically engi neered composite material, or replacing the optical filters with a reflecting coating layer between the two scintillators. Then, new possi bilities could be explored: besides neutron detection, other phoswich applications, like surface contamination detection (e.g. α=β; β=γ), and energy dispersive X-ray radiography could be foreseen by a specific engineering of composition and geometry of the phoswich configuration. References Aza, E., Caresana, M., Cassell, C., Charitonidis, N., Harrouch, E., Manessi, G., Pangallo, M., Perrin, D., Samara, E., Silari, M., 2014. Instrument intercomparison in the pulsed neutron fields at the cern hiradmat facility. Radiat. Meas. 61, 25. https:// doi.org/10.1016/j.radmeas.2013.12.009. Blasse, G., Grabmaier, B.C., 1994. Luminescent Materials. Springer-Verlag. Caresana, M., Denker, A., Esposito, A., Ferrarini, M., Golnik, N., Hohmann, E., Leuschner, A., Luszik-Bhadra, M., Manessi, G., Mayer, S., Ott, K., Rohrich, J., Silari, M., Trompier, F., Volnhals, M., Wielunski, M., 2014. Intercomparison of radiation protection instrumentation in a pulsed neutron field. Nucl. Instrum. Methods Phys. Res., Sect. A 737, 203. https://doi.org/10.1016/j.nima.2013.11.073. Cerullo, N., Bufalino, D., Daquino, G., 2009. Progress in the use of gadolinium for NCT. Appl. Radiat. Isot. 67, 157. Dorenbos, P., 2013. Ce3þ 5d-centroid shift and vacuum referred 4f-electron binding energies of all lanthanide impurities in 150 different compounds, 135, 93. Dujardin, C., Auffray, E., Bourret, E., Dorembos, P., Lecoq, P., Nikl, M., Vasil’ev, A.N., Yoshikawa, A., Zhu, R., 2018. Needs, Trends and Advances in Inorganic Scintillators. IEE Transactions on Nuclear Science, p. 1. Dumazert, J., Coulon, R., Lecomte, Q., Bertrand, G.H.V., Hamel, M., 2018. Gadolinium for neutron detection in current nuclear instrumentation. Nucl. Instrum. Methods Phys. Res. 882, 53. Fasoli, M., Vedda, A., Nikl, M., Jiang, C., Uberuaga, B.P., Andersson, D., McClellan, K.J., Stanek, C.R., 2011. Band-gap engineering for removing shallow traps in rare-earth Lu3Al5O12 garnet scintillators using Ga3þ doping. Phys. Rev. B 84, 081102(R). Harms, A.A., McCormack, G., 1974. Isotopic conversion in gadolinium-exposure neutron imaging. Nucl. Instrum. Methods 118, 583. https://physics.nist.gov/PhysRefData/Star/Text/ESTAR.html (accessed on July 19, 2019). Kamada, K., Yanagida, T., Endo, T., Tsutumi, K., Usuki, Y., Nikl, M., Fujimoto, Y., Fukabor, A., Yoshikawa, A., 2012. 2 inch diameter single crystal growth and scintillation properties of Ce: Gd3Al2Ga3O12. J. Cryst. Growth 352, 88. Kandlakunta, P., Cao, L., 2012. Gamma-ray rejection, or detection, with gadolinium as a converter. Radiat. Prot. Dosim. 151 (3), 586. Kandlakunta, P., Cao, L.R., Mulligan, P., 2013. Measurement of internal conversion electrons from Gd neutron capture. Nucl. Instrum. Methods Phys. Res. 705, 36. Knoll, G.F., 1999. Radiation Detection and Measurement, third ed. Wiley and Sons, Inc. Lecoq, P., 2016. Development of new scintillators for medical applications. Nucl. Instrum. Methods Phys. Res. 809, 130. Nikl, M., Yoshikawa, A., 2015. Recent R&D trends in inorganic single-crystal scintillator materials for radiation detection. Adv. Opt. Mater. 3, 463. Nikl, M., Yoshikawa, A., Kamada, K., Nejezchleb, K., Stanek, C.R., Mares, J.A., Blazek, K., 2013. Development of LuAG-based scintillator crystals – a review. Prog. Cryst. Growth Charact. Mater. 59, 47. Report 39, 2016. Int. Comm. Radiat. Units Meas. os20 (2) https://doi.org/10.1093/ jicru/os20.2.Report39. NP–NP. Report 57, 2016. Int. Comm. Radiat. Units Meas. os29 (2) https://doi.org/10.1093/ jicru/os29.2.Report57. NP–NP. Wu, Y., Luo, Z., Jiang, H., Meng, F., Koschan, M., Melcher, C., 2015. Single crystal and optical ceramic multicomponent garnet scintillators: a comparative study. Nucl. Instrum. Methods Phys. Res., Sect. A 780, 45–50. X-5 Monte Carlo Team, Apr. 2003. MCNP – A General Monte Carlo N-Particle Transport Code, Version 5. Volume I: Overview and Theory. Los Alamos National Laboratory, Oak Ridge.
4. Conclusions We investigated the feasibility of performing particle discrimination with an emission-based technique using a phoswich. The main requirement of this approach is the choice of two scintillating materials with no overlap in their emission spectra The scintillator in which the interaction occurs is thus uniquely identified based on the spectrum of the emitted light. We developed a phoswich set-up for neutron-gamma discrimination, using gadolinium as a neutron converter, for proving the proposed method. Even if the results are still preliminary, we demonstrated with a simple example case that it is possible to perform particle discrimination with optical filtering, instead of using active methods such as PSD.
6
Contents lists available at ScienceDirect
Radiation Measurements journal homepage: http://www.elsevier.com/locate/radmeas
Neutron/γ discrimination by an emission-based phoswich approach G. Zorloni a, F. Cova b, M. Caresana a, *, M. Di Benedetto a, J. Hosta�sa e, M. Fasoli b, I. Villa b, I. Veronese c, d, A. Fazzi a, A. Vedda b a
Department of Energy, Politecnico di Milano, Via Lambruschini 4, 20156, Milan, Italy Department of Materials Science, University of Milano - Bicocca, Via Cozzi 55, 20125, Milan, Italy c Dipartimento di Fisica, Universit� a degli Studi di Milano, Via Celoria 16 I, 20133, Milan, Italy d INFN Sezione di Milano, Via Celoria 16 I, 20133, Milan, Italy e CNR ISTEC, Institute of Science and Technology for Ceramics, Via Granarolo 64, 48018, Faenza, Italy b
A R T I C L E I N F O
A B S T R A C T
Keywords: Phoswich detector Particle discrimination Neutron detector
Particle discrimination in a mixed radiation field using scintillators is a challenging topic in radiation detection research. We propose a novel approach relying on the possibility of identifying particles interacting in a phos wich detector from the emission spectrum of the produced scintillation signal. As a proof of concept, we focus on the discrimination between neutrons and gamma rays. Lu3 Al5 O12 :Pr and Gd3 Al2 Ga3 O12 :Ce thin single crystal scintillators, coupled to two different silicon photo multipliers equipped with optical filters, are simultaneously used in the same phoswich detector. Their optical emissions peak at approximately 310 nm and 545 nm respectively, and thus they can easily be distinguished by optical filtering. While both crystals are sensitive to gamma rays, neutrons interact only with the Gd3 Al2 Ga3 O12 : Ce thanks to the presence of Gd acting as neutron converter. Optical filtration and an anti-coincidence algorithm are therefore used to perform particle discrimination, rejecting coincidence signals arising from gamma rays, which simultaneously deposit energy in both crystals, and counting anti-coincidence signals due to neutrons, which deposit energy only in the Gd3 Al2 Ga3 O12 :Ce. The simple neutron counter developed here is intended to be a proof of the principle of the feasibility of the color-based particle discrimination technique.
1. Introduction In recent years, new advances in inorganic solid state scintillator materials science, mainly triggered by the needs of high energy physics experiments and medical imaging (Lecoq, 2016), have led to the dis covery and the development of a wide variety of new scintillators (Nikl and Yoshikawa, 2015; Dujardin et al., 2018). Recent progress in mate rials science permits tuning and optimization of scintillator properties and characteristics, with studies of band gap engineering (Fasoli et al., 2011) and energy level positioning (Dorenbos, 2013). In this study we take advantage of some of these novel materials to propose a new par ticle discrimination technique. The standard way to perform particle discrimination with scintilla tors is so-called Pulse Shape Discrimination (PSD) (Knoll, 1999), a method based on the electronic analysis of the voltage pulse time behavior. As a potential alternative, in this work we investigate the
feasibility of a passive optical-based technique using phoswich detectors.1 As a proof of concept, we developed a simple system for neutrongamma discrimination consisting of a composite Gd3 Al2 Ga3 O12 :Ce/ Lu3 Al5 O12 :Pr (GGAG:Ce/LuAG:Pr) sensor. While both scintillators are sensitive to gamma rays, only GGAG:Ce interacts with thermal neutrons because of its high Gd content. After neutron capture, the Internal Conversion (IC) electrons deposit their energy only in the GGAG:Ce crystal, thus producing an anti-coincidence signal, and they are counted. On the other hand, because the scintillators are thin, the background gamma rays can deposit their energy in both scintillators, thus pro ducing a coincidence signal, and they are rejected. Since the two scin tillator species emit in well distinguished spectral regions, passive discrimination could be obtained by coupling the two scintillators to two distinct photodetectors through optical filters. In principle, this approach could lead to an improvement with
* Corresponding author. E-mail address: [email protected] (M. Caresana). 1 A phoswich, or phosphor sandwich, is the combination of two different scintillators usually coupled to a single photodetector (Knoll, 1999). https://doi.org/10.1016/j.radmeas.2019.106203 Received 18 March 2019; Received in revised form 11 October 2019; Accepted 16 October 2019 Available online 26 October 2019 1350-4487/© 2019 Elsevier Ltd. All rights reserved.
G. Zorloni et al.
Radiation Measurements 129 (2019) 106203
respect to an equivalent system relying on electronic signal analysis, which is affected by random noise and jitter effects: the PSD algorithms are substituted by the two optical filters, which perform discrimination prior to signal formation, with a passive physical-based quasi-digital filtering. The simple neutron counter developed in this work is not aimed at replacing existing detectors for neutron measurements, but for demon strating the feasibility of the proposed technique. Therefore, our inves tigation shows how a compact detector composed of different materials, coupled together in engineered geometries, can display attractive properties, useful for particle discrimination, leading to novel specific functionalities.
0.12 and 0.48 for 155 Gd and 157 Gd respectively (Kandlakunta et al., 2013). The energies of the main IC lines are around 30 keV, 70 keV and 130 keV. The continuous slowing down approximation range for 180 keV electrons, i.e. the highest energy IC line with a significant branching ratio, in the GGAG:Ce is about 90 μm. For the 70 keV elec trons the range in the GGAG:Ce is about 20 μm, while for the 30 keV is only a few microns. The calculations were performed using the ESTAR database (https://physics.nist.gov/, 2019). Since these electrons are extracted from the atomic shell of the nuclei, the atoms are, in turn, in excited states. They de-excite by emitting X rays, mainly around 40–50 keV for the K shell and 6–7 keV for the L shell, and low-energy Auger electrons (Cerullo et al., 2009; Kandlakunta and Cao, 2012).
2. Materials and methods 2.1. Characteristics of the sources
2.4. Experimental phoswich set-up
We tested the proposed methodology by developing a small detector system for thermal neutron counting in a mixed neutron/γ field. The tests were performed with two sources, one of thermal neutrons and one of gamma rays, thus observing the effects of the two types of radiation field independently on the system. Tests with neutrons were performed with an Am–Be neutron source placed at the bottom of a polyethylene cylinder, providing the neutron thermalization. Measurements were performed in the hollow upper part of the cylinder. Between the source and the cavity, a thin layer of lead attenuates the 60 keV gammas from 241 Am. This configuration mitigates the low energy photon contamination and provides a quasi-pure thermal neutron field (about 87%) with a fluence rate of about 400 neutrons per square centimeter per second within the measurement cavity. Some additional high-energy gamma contamination arises from neutron cap ture on carbon and hydrogen. Tests with gamma rays were performed with certified metrological grade gamma sources of 137 Cs (0.1, 1, 10 Ci). 137 Cs was chosen to be representative of the background gamma rays in neutron measurements.
A simple set-up has been developed to prove the proposed concept. Two twin circuits to read the photodetectors signals are realized. Two Silicon PhotoMultipliers (SiPMs) SensL (Cork, Ireland) ArrayJ-600354P-EVB are directly supported by the circuits, one in front of the other, acting also as supports to filters and crystals, without adding other structural components. The photon detection efficiency of the SiPMs is between 0.3 and 0.4 for both scintillator species at 6 V overvoltage. The phoswich is mounted between the photodetectors. The whole system is protected from light exposure by a sealed aluminum case. The neutron sensitive component consists of a commercial 5 � 5 � 0.1 mm3 slice of GGAG:Ce, produced by Kinheng Crystal Mate rial (Shangai, China). The small thickness was chosen from physical considerations about the neutron/γ interactions: i) it guarantees a low gamma sensitivity (i.e. 5‰ attenuation for a 137 Cs expanded aligned photon source); ii) the range of IC electrons from the (n,γ) reactions on Gd is short enough for them to be fully stopped in the crystal slice, since they are the reaction products the system should reveal. The second scintillator, insensitive to neutrons, is a commercial 5 � 5 � 0.5 mm3 LuAG:Pr crystal, produced by Kinheng Crystal Material (Shangai, China). In Table 1 the main characteristics of the crystals are briefly summarized. The set-up is completed by the two optical filters. The long-pass filter is an Andover Corporation (Salem, US) model 450FH90–12.5, the shortpass is a Knight Optical (Harrietsham, UK) model 465FCS5050. The phoswich setup, as described in Fig. 1, consists of, from the top to the bottom: first SiPM, long-pass filter, GGAG:Ce, LuAG:Pr, short-pass filter, second SiPM. The optical coupling between the phoswich components is provided by optical grease. The GGAG:Ce sensitive SiPM facing the longpass filter (hereafter denoted as n-channel) is connected to a CAEN (Viareggio, Italy) DT5770 multichannel analyzer, with the capability of anti-coincidence veto counting. The logic signal, which opens the veto window in the multichannel anti-coincidence module, is obtained by connecting the LuAG:Pr sensitive SiPM, i.e. the one facing the short-pass filter (denoted as γ-channel), to a programmable single channel analyzer unit. The single channel threshold is selectable by the user, thus it is possible to observe the effects of the rejection logic of the system as a function of the γ-channel veto threshold.
2.2. Description of the optical characterization apparatus The crystals were characterized by means of radioluminescence (RL) spectroscopy. RL measurements were performed using a homemade apparatus featuring a liquid nitrogen-cooled, back-illuminated and UVenhanced, charge-coupled device (CCD) (Jobin-Yvon Spectrum One 3000, Bensheim, Germany) coupled to a monochromator (Jobin-Yvon Triax 180) with a 100 grooves/mm grating as the detection system. RL excitation was obtained by X-ray irradiation through a beryllium win dow, using a Philips PW2274 (Andover, US) X-ray tube with tungsten anode operated at 20 kV. RL emission spectra were corrected for the spectral response of the detection system. The transmittance of the filters was measured with a Agilent (Santa Clara, US) Varian Cary 50 spectrophotometer from 190 to 1100 nm. 2.3. Neutron interactions with gadolinium Natural gadolinium comprises 7 natural isotopes, out of which 155 Gd and 157 Gd show the highest thermal cross sections (60,900 b and 254,000 b respectively, for a total cross section for natural Gd of about 48,800 b) (Dumazert et al., 2018). The products of the reaction are two stable isotopes, 156 Gd* and 158 Gd* in excited states, 8.536 MeV and 7.937 MeV respectively. The prompt γ cascade is characterized by the emission of about 3 photons per reaction on average, with a quasi-continuous energy distribution, with a maximum emission prob ability around 2 MeV (Dumazert et al., 2018). A competitive mechanism to the γ emission is the IC process, mainly at low energies (the maximum energy line of the IC electrons is 246 keV) (Harms and McCormack, 1974). IC yield per neutron capture is around
Table 1 Scintillation properties of LuAG:Pr and GGAG:Ce. ρ: density; Zeff: Z effective; τ: main decay time; LY: light yield; λpeak : emission wavelength peak. Parameter 3
ρ (g/cm ) Zeff ( ) τ main (ns) LY (phot/MeV) λpeak (nm) a b
2
LuAG:Pra
GGAG:Ceb
6.67 63 20 16,000–19,000 310
6.3 54 90 50,000–60,000 545
Data taken from (Nikl et al., 2013). Data taken from (Kamada et al., 2012).
G. Zorloni et al.
Radiation Measurements 129 (2019) 106203
energy of the photons (the attenuation of 2 MeV photons in the LuAG:Pr is about 1%). On the other hand, the low energy relaxation X rays will probably interact, with about 80% attenuation in the LuAG:Pr, so that the veto threshold of the γ-channel must be properly chosen not to trigger the veto on these interactions. False positive signals can arise from background gamma interactions which deposit energy only in the GGAG:Ce crystal, without interacting in the LuAG:Pr. 2.5. Description of the Monte Carlo simulations Monte Carlo (MC) simulations were performed to compare the experimental data and the theoretical predictions in the case of thermal neutron capture. MC results can also be used for the energy calibration of the system. In fact, the low thickness of the GGAG:Ce crystal prevents the 137 Cs photopeak formation. On the other hand, the resulting energy spectrum after neutron capture, which can provide, in principle, two main peaks around 30 and 70 keV, is unknown a priori. With the MC simulation it is possible to predict the system response to neutron cap ture, and this information can be used to assess the energy calibration of the detector. Calculations were performed using the MCNPX version 2.6 code (X-5 Monte Carlo Team, 2003) and a Matlab script was written for the post-processing elaboration of the MC results, mainly to account for summation effects of the atomic de-excitation processes. Simulations were performed in five steps:
Fig. 1. Schematic of the phoswich arrangement: the GGAG:Ce crystal is dis played in yellow, the LuAG:Pr crystal is represented in gray. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
The discrimination is obtained as follows: 1. When a neutron interacts in the GGAG:Ce, it produces a signal in the n-channel. The short range IC electrons are fully stopped in the GGAG:Ce slice, while the prompt gammas escape from the system. If the signal falls within a user-defined acceptance window (see 3.3 for the details), the event is recorded as a true positive neutron count; 2. When a background γ-ray interacts in the GGAG:Ce, or in the sur rounding materials, the secondary electrons could pass through both scintillators, because of the small thickness of the GGAG crystal. This event produces a simultaneous signal in both channels. If the γ-channel signal is above threshold, a logic signal is triggered, thus opening a veto time window in the anti-coincidence module of the nchannel. Any signal occurring within the veto window in the nchannel is flagged as a gamma-induced count, and is consequently rejected as a true negative.
1. Evaluation of the reaction rate along the GGAG:Ce thickness due to the thermal neutron capture; 2. Pulse height tally (energy deposition spectrum) of the prompt gamma rays after capture on Gd in the GGAG:Ce crystal; 3. Pulse height tally (energy deposition spectrum) of the prompt IC electrons after capture on Gd in the GGAG:Ce crystal; 4. Pulse height tally (energy deposition spectrum) of the X-rays and Auger electrons after the atomic relaxation process following the IC emission in the GGAG:Ce crystal; 5. Weighted combination (summation and superposition) of point 2, 3 and 4 contributions, accounting also for the Gaussian energy broadening of the crystal response, i.e. scintillator resolution.
Fig. 2 shows a schematic diagram of the readout acquisition logic. False negative signals can arise from interaction of a prompt γ, emitted simultaneously to IC electrons, in the Pr-doped scintillator. However, the interaction probability is rather low, because of the high mean
The first step was designed to evaluate the reaction rate profile along the crystal, since the high Gd cross section could in principle perturb the thermal neutron fluence along the crystal thickness. An isotropic Max wellian neutron source was simulated for modeling the experimental setup. The result was scored independently for the 155 Gd and 157 Gd isotopes. Other Gd isotopes supposedly do not significantly contribute to the reaction rate, since their cross section is several orders of magnitude lower than that of 155 Gd and 157 Gd (Dumazert et al., 2018). The same simulation was used to calculate the pulse height tally due to the prompt gamma contribution, using the ENDF/B-VII.0 library. The reaction rate profile obtained in the first step was used as an input parameter for the spatial distribution of the IC electrons, X-rays and Auger electrons treated as isotropic volumetric sources distributed within the GGAG:Ce crystal. Reaction rate was estimated independently for the two isotopes since the IC electrons and X-rays energy distributions are isotope dependent. The energy distribution spectra were finally combined considering superposition of the effects in the case of independent events and sum mation effects in the case of correlated events, e.g. X-rays cascade of the atomic relaxation after IC emission. The multiple gamma ray cascade was treated as a series of independent events, because of their high mean energy, i.e. negligible probability of multiple simultaneous correlated events.
Fig. 2. Logic block diagram of the readout circuit. The acceptance window of the n-channel is the region of the differential energy spectrum of the n-channel in which the signals are eligible to be counted as neutrons (description in 3.3). 3
G. Zorloni et al.
Radiation Measurements 129 (2019) 106203
3. Results and discussion
anti-coincidence logic is turned on, and if only waveshifting occurs, some positive signals will be collected. Otherwise, if no waveshifting occurs, or, for instance, if a limited fraction of the photons emitted by Pr generates waveshifting, the anticoincidence logic rejects the n-channel signal as a true negative. Measurements without external sources proved that the second condition was verified, thus demonstrating that, for the present prototype, waveshifting effects can be neglected.
3.1. Characterization of the crystals Both scintillators and filters were investigated by RL and optical absorption spectroscopy to verify if the chosen materials addressed the desired criteria. GGAG:Ce and LuAG:Pr normalized RL emission spectra are shown in Fig. 3. For both crystals, the main emission bands are ascribed to the allowed 5d-4f transition of Ce and Pr ions, peaking at around 545 nm and 310–375 nm, respectively. Emission lines from 4f states of Pr ions in the VIS - NIR region are also weakly visible in the RL spectrum of LuAG: Pr; they are related to forbidden transitions which are much slower (in the ms time scale) than the main 5d–4f bands (Blasse and Grabmaier, 1994) and can therefore be considered negligible for the scope of this work (the shaping time of the scintillation measurements was kept in the μs time scale). Fig. 3 displays also the transmittance spectra of the two selected optical filters, aimed to be sensitive only to the light of one out of the two materials. The transmittance of the filters in the region of interest is around 80%. With reference to Fig. 3, it can be noted that the short-pass filter is actually a band-stop filter. In any case, it addresses the desired criteria, that are i) transparency to LuAG:Pr emission, and ii) opacity to GGAG:Ce emission. It is worth noting that the optical emission of the LuAG:Pr crystal overlaps with the optical absorption of the GGAG:Ce crystal (Wu et al., 2015), which may give rise to false positive neutron signals by wave shifting effects.2 However, this kind of behavior was found to be negli gible from experimental evidence. In fact, the LuAG:Pr possesses an internal radioactivity, estimated to be roughly 20 Bq, due to the pres ence of the natural occurring 176 Lu isotope, which produces some signals in the LuAG:Pr or in both scintillators (176 Lu is a β emitter). If the
3.2. Monte Carlo simulations and comparison with experiments MC calculations were performed considering the complete set of reactions, as described in 2.3. This procedure is aimed at verifying that the system is able to reveal the IC electrons and relaxation cascade products with the 100 μm thick crystal. Moreover, MC simulations provided two points for the energy calibration of the detector. In Fig. 4 the numerical spectrum and the experimental n-channel spectrum are compared. MC calculations demonstrate that the main contribution to the growth of the spectrum is due to the convolution of IC electrons, Auger electrons and X rays signals. Thus, the two main peaks are centered around 32 keV and 74 keV. To better visualize the IC electron contribution, Fig. 4 shows also the main IC lines of natural Gd. The multichannel analysis and the comparison with MC simulations demonstrate that the system is actually capable of counting the IC electrons. The presence of the two peaks centered in correspondence to the main IC lines proves that the thickness of the crystal is sufficient for completely stop the IC electrons. The flat background is due to Compton scattered electrons after the interaction of prompt γ-rays. 3.3. Test of the phoswich setup Measurements were performed independently with the 137 Cs source and with the thermal neutron source. The counts recorded in the nchannel correspond to those seen within a 30–100 keV acceptance window of the multichannel analyzer (i.e. the area subtended by the two main peaks in Fig. 4). The lower acceptance window bound was chosen to eliminate the SiPM noise. The upper bound was chosen to subtend the main IC electron emission peak, excluding the high energy Compton
Fig. 3. RL emission spectra (solid lines) of LuAG:Pr crystal (purple, triangles) and GGAG:Ce crystal (green, circles) and transmittance spectra (short-pass blue dashed line, long-pass orange dotted line) of the two selected optical filters to be coupled to the two photodetectors. Transmittance values are reported on the right Y axis. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Fig. 4. Comparison between the simulated spectra (dashed line) and the measured one (solid line) in the n-channel after neutron capture. Counts are normalized. The yellow vertical lines correspond to the normalized IC emission probabilities of natural Gd after neutron capture. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
2 The term waveshift denotes the effect of a high energy radiation, such as the LuAG emitted photons, which in principle can be absorbed by a material, such as GGAG, and then re-emitted in a low energy spectral region.
4
G. Zorloni et al.
Radiation Measurements 129 (2019) 106203
electrons from the γ background. Fig. 5 (a) shows th Counts Per Second (CPS) recorded in the nchannel for events generating signals within the n-channel acceptance window, normalized per 1 Sv/h ambient dose equivalent (Report 39, 2016; Report 57, 2016), under different experimental conditions. The upper dashed line, in blue, represents the CPS recorded by irradiating with thermal neutrons without implementing the rejection logic. The circle dots represent the CPS recorded in the same condition as above, as a function of the γ-channel veto threshold but switching on the anti coincidence gamma rejection. The neutron efficiency is weakly depen dent on the γ-channel voltage threshold. The lower dashed line, in red, represents the CPS recorded by irra diating the detector with gamma rays only, without implementing the rejection logic. In this case the contribution represents a false positive, i. e. gamma induced neutron signal. The triangle dots represent the CPS recorded in the same condition as above, as a function of the γ-channel veto threshold but switching on the anticoincidence gamma rejection. The variation of the false positive rate is greatly reduced by triggering the veto signal with low voltage. In both neutron and gamma irradiation the veto signal time exten sion was set to 1 μs, which was found to be sufficiently long to account for jitter effects, and sufficiently short not to include independent/un correlated events. Normalization per unit ambient dose equivalent was used because it is a traditional way to compare radiation protection instrumentation, in particular neutron dosimeters and counters (Care sana et al., 2014; Aza et al., 2014). In Fig. 5 (b) the same information is expressed in terms of rejection percentage calculated as the ratio be tween the CPS recorded with the anticoincidence logic switched on divided by the CPS recorded without implementing the anticoincidence logic. The analysis was done for both neutrons (circle dots) and photons (triangle dots). With respect to the counts recorded without the rejection logic, lower dashed line in Fig. 5 (a), the false positives decrease as the veto threshold decreases, thus demonstrating the feasibility of the emissionbased discrimination method. In fact, the lower the γ-channel veto threshold, the higher the probability that a background gamma ray deposits enough energy for triggering the veto, thus increasing the discrimination capability of the system. On the other hand, a lower veto threshold leads also to a slight decrease of neutron detection efficiency. In this case a larger number of true positives are recorded as false neg atives because of the cross-talk between the two crystals. As an example, after neutron capture the prompt γ and X rays may deposit a certain amount of energy in LuAG:Pr, thus triggering a false coincidence window.
The 17% neutron rejection, Fig. 5 (b), even at high rejection threshold has to be analyzed in details for future improvements of the system. The exact cause of this percentage of rejection is currently un known, but two possible contributions can be mentioned. Firstly, the background gamma contamination of the source, which does not pro vide a purely neutron field, may cause an overestimation of the neutron sensitivity without the implementation of the anti-coincidence logic. Secondly, the dark counts due to the natural radioactivity of Lu, which are completely counteracted by the anti-coincidence logic at low threshold, may cause again an overestimation of the neutron sensitivity in the case of no anti-coincidence logic implemented. In particular, the 20 Bq activity of the internal source is not negligible with respect to the counts due to the 400 neutrons per square centimeter per second of the Am–Be source. A proper quantification of the two contributions is planned for future activities. However, these two issues may cause an overestimation of the neutron sensitivity without the anti-coincidence,
Fig. 6. Comparison between the n-channel multichannel spectra obtained by irradiating with neutrons (blue solid line) and 137 Cs (red dashed line) without implementing the anticoincidence logic. Results are normalized per unit ambient dose equivalent. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Fig. 5. (a): n-channel counts per unit ambient dose equivalent due to neutrons (blue, circles) and 662 keV photons (red, triangles) independently, as a function of the veto trigger threshold on the γ-channel. The dashed lines represent the counts recorded in the n-channel without implementing the anti-coincidence logic. (b): percentage of rejection with respect to the counts recorded without implementing the rejection logic. Lines connecting points are guides for the eye. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
5
G. Zorloni et al.
Radiation Measurements 129 (2019) 106203
but they are actually rejected by the discrimination logic. In Fig. 6, the n-channel multichannel spectra obtained by irradiating with neutrons and 137 Cs are compared in the case where no anti coincidence logic is implemented. The partial Compton edge produced by 137 Cs mostly overlaps with the acceptance window. The low crystal thickness prevents the formation of both the complete Compton edge and the photopeak. Even if from the multichannel point of view this is the worst case scenario (i.e. energy response overlap), the technique proved to be efficient up to 45% of rejection, as seen in Fig. 5 (b). It should be noted that the actual configuration is directional, i.e. a twoslice phoswich is far from the isotropic response, leaving wide room for geometrical improvements. We have demonstrated by a simple case that particle discrimination can be pursued by an emission-based approach. Among the most valu able aspects of this method, is the possibility to use a quasi-digital logic by performing the rejection prior to the electronic signal formation. This uniquely identifies in which out of the two phoswich components an event occurred. The simplicity of the electronics results because no al gorithms and no additional elements/electronic devices for pulse shape analysis are required. The drawbacks of the proposed technique are the traditional ones of a phoswich arrangement, for instance optical requirements and spillover/cross-talk due to the stochastic behavior of the particles in teractions in matter. Two additional drawbacks are typical of the pro posed approach itself, i) an expected worsening in the energy resolution, because more components are present in the phoswich (i.e. optical fil ters), and ii) the requirement of two independent readout circuits/ photodetectors. Regarding the first point, a preliminary test with the GGAG:Ce directly coupled without a filter to the SiPM showed that most of the gamma signal overlaps the acceptance window encompassing the neutron signal even with a better resolution. This feature makes impossible the separation of the two components. Despite the low percentage of gamma rejection, which can be regarded as still far from the value required for discrimination appli cations of this system, the results presented in this work leave wide room for improvements of the proposed approach. These can include an optimization of the crystals thickness through MC simulations, and the introduction of a three-slice phoswich i.e. LuAG:Pr-GGAG:Ce-LuAG:Pr, to enhance the isotropy of the response and the rejection capability.
Further perspectives could look towards a more optimized and compact geometry, also exploiting the use of a single specifically engi neered composite material, or replacing the optical filters with a reflecting coating layer between the two scintillators. Then, new possi bilities could be explored: besides neutron detection, other phoswich applications, like surface contamination detection (e.g. α=β; β=γ), and energy dispersive X-ray radiography could be foreseen by a specific engineering of composition and geometry of the phoswich configuration. References Aza, E., Caresana, M., Cassell, C., Charitonidis, N., Harrouch, E., Manessi, G., Pangallo, M., Perrin, D., Samara, E., Silari, M., 2014. Instrument intercomparison in the pulsed neutron fields at the cern hiradmat facility. Radiat. Meas. 61, 25. https:// doi.org/10.1016/j.radmeas.2013.12.009. Blasse, G., Grabmaier, B.C., 1994. Luminescent Materials. Springer-Verlag. Caresana, M., Denker, A., Esposito, A., Ferrarini, M., Golnik, N., Hohmann, E., Leuschner, A., Luszik-Bhadra, M., Manessi, G., Mayer, S., Ott, K., Rohrich, J., Silari, M., Trompier, F., Volnhals, M., Wielunski, M., 2014. Intercomparison of radiation protection instrumentation in a pulsed neutron field. Nucl. Instrum. Methods Phys. Res., Sect. A 737, 203. https://doi.org/10.1016/j.nima.2013.11.073. Cerullo, N., Bufalino, D., Daquino, G., 2009. Progress in the use of gadolinium for NCT. Appl. Radiat. Isot. 67, 157. Dorenbos, P., 2013. Ce3þ 5d-centroid shift and vacuum referred 4f-electron binding energies of all lanthanide impurities in 150 different compounds, 135, 93. Dujardin, C., Auffray, E., Bourret, E., Dorembos, P., Lecoq, P., Nikl, M., Vasil’ev, A.N., Yoshikawa, A., Zhu, R., 2018. Needs, Trends and Advances in Inorganic Scintillators. IEE Transactions on Nuclear Science, p. 1. Dumazert, J., Coulon, R., Lecomte, Q., Bertrand, G.H.V., Hamel, M., 2018. Gadolinium for neutron detection in current nuclear instrumentation. Nucl. Instrum. Methods Phys. Res. 882, 53. Fasoli, M., Vedda, A., Nikl, M., Jiang, C., Uberuaga, B.P., Andersson, D., McClellan, K.J., Stanek, C.R., 2011. Band-gap engineering for removing shallow traps in rare-earth Lu3Al5O12 garnet scintillators using Ga3þ doping. Phys. Rev. B 84, 081102(R). Harms, A.A., McCormack, G., 1974. Isotopic conversion in gadolinium-exposure neutron imaging. Nucl. Instrum. Methods 118, 583. https://physics.nist.gov/PhysRefData/Star/Text/ESTAR.html (accessed on July 19, 2019). Kamada, K., Yanagida, T., Endo, T., Tsutumi, K., Usuki, Y., Nikl, M., Fujimoto, Y., Fukabor, A., Yoshikawa, A., 2012. 2 inch diameter single crystal growth and scintillation properties of Ce: Gd3Al2Ga3O12. J. Cryst. Growth 352, 88. Kandlakunta, P., Cao, L., 2012. Gamma-ray rejection, or detection, with gadolinium as a converter. Radiat. Prot. Dosim. 151 (3), 586. Kandlakunta, P., Cao, L.R., Mulligan, P., 2013. Measurement of internal conversion electrons from Gd neutron capture. Nucl. Instrum. Methods Phys. Res. 705, 36. Knoll, G.F., 1999. Radiation Detection and Measurement, third ed. Wiley and Sons, Inc. Lecoq, P., 2016. Development of new scintillators for medical applications. Nucl. Instrum. Methods Phys. Res. 809, 130. Nikl, M., Yoshikawa, A., 2015. Recent R&D trends in inorganic single-crystal scintillator materials for radiation detection. Adv. Opt. Mater. 3, 463. Nikl, M., Yoshikawa, A., Kamada, K., Nejezchleb, K., Stanek, C.R., Mares, J.A., Blazek, K., 2013. Development of LuAG-based scintillator crystals – a review. Prog. Cryst. Growth Charact. Mater. 59, 47. Report 39, 2016. Int. Comm. Radiat. Units Meas. os20 (2) https://doi.org/10.1093/ jicru/os20.2.Report39. NP–NP. Report 57, 2016. Int. Comm. Radiat. Units Meas. os29 (2) https://doi.org/10.1093/ jicru/os29.2.Report57. NP–NP. Wu, Y., Luo, Z., Jiang, H., Meng, F., Koschan, M., Melcher, C., 2015. Single crystal and optical ceramic multicomponent garnet scintillators: a comparative study. Nucl. Instrum. Methods Phys. Res., Sect. A 780, 45–50. X-5 Monte Carlo Team, Apr. 2003. MCNP – A General Monte Carlo N-Particle Transport Code, Version 5. Volume I: Overview and Theory. Los Alamos National Laboratory, Oak Ridge.
4. Conclusions We investigated the feasibility of performing particle discrimination with an emission-based technique using a phoswich. The main requirement of this approach is the choice of two scintillating materials with no overlap in their emission spectra The scintillator in which the interaction occurs is thus uniquely identified based on the spectrum of the emitted light. We developed a phoswich set-up for neutron-gamma discrimination, using gadolinium as a neutron converter, for proving the proposed method. Even if the results are still preliminary, we demonstrated with a simple example case that it is possible to perform particle discrimination with optical filtering, instead of using active methods such as PSD.
6