Light Pulse Shapes in Liquid Scintillators Originating ... - IEEE Xplore

3846

IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 57, NO. 6, DECEMBER 2010

Light Pulse Shapes in Liquid Scintillators Originating From Gamma-Rays and Neutrons T. Szcz˛esńiak, Member, IEEE, M. Moszyn´ski, Fellow, IEEE, A. Syntfeld-Ka˙zuch, Member, IEEE, Ł. S´widerski, Member, IEEE, D. Wolski, M. Grodzicka, Member, IEEE, G. Pausch, Member, IEEE, J. R. Stein, F. Kniest, M. R. Kusner, P. Schotanus, and C. Hurlbut

Abstract—Liquid scintillators loaded with boron-10 or lithium-6 are capable to detect gamma rays, fast neutrons and also thermal neutrons. One of the popular methods applied in order to distinguish events originating from different particles is the pulse shape discrimination (PSD). The previously presented study of boron-10 loaded liquid scintillators using the PSD method showed different discrimination performance in scintillators such as BC523A, BC523A2, EJ339A2 and EJ309B5. It triggered a further study of the light pulse shapes in these scintillators originating from events related to gamma rays, fast and thermal neutrons. The light pulse shapes, measured using the single photon method, were recorded together with the 2-dimensional n/gamma discrimination data. Next, the recorded light pulses were gated using energy and the PSD information to extract pulses characteristic of the only one kind of particles. Finally, the analysis of the light pulse shapes with multi-exponential fits and calculation of decay time constants and intensities of components were performed. The results were compared with the data obtained for liquid scintillators not sensitive to thermal neutrons BC501A, EJ301 and EJ309. Index Terms—Light pulse shapes, liquid scintillators, neutrons, pulse shape discrimination.

I. INTRODUCTION

L

IQUID scintillators are known to be capable to detect both -rays and fast neutrons. Moreover, loaded with or may also detect thermal neutrons. In spectra, containing disevents related to all the three kinds of particles, the crimination methods are required to detect neutron events not contaminated by -rays background. In such scintillators a popdiscrimination between detected particles is ular method of the pulse shape discrimination (PSD) based on the zero-crossing

Manuscript received December 31, 2009; revised June 01, 2010; accepted July 25, 2010. Date of publication October 04, 2010; date of current version December 15, 2010. This work was supported in part by EU Structural Funds Project POIG.01.01.02-14-012/08-00. T. Szcz˛esńiak, M. Moszyn´ski, A. Syntfeld-Ka˙zuch, Ł. S´widerski, D. Wolski, and M. Grodzicka are with the Soltan Institute for Nuclear Studies, PL-05-400 Otwock-Swierk, Poland (e-mail: [email protected]). G. Pausch and J. R. Stein are with ICx Radiation GmbH, D-42653 Solingen, Germany. F. Kniest is with Saint-Gobain Crystals, Holland Office, 3760 DB Soest, The Netherlands. M. R. Kusner is with Saint-Gobain Crystals, Newbury, OH 77005 USA. P. Schotanus is with SCIONIX Holland B.V., 3980 CC Bunnik, The Netherlands. C. Hurlbut is with Eljen Technology, Sweetwater, TX 79556 USA. Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TNS.2010.2068311

(ZC) principle. The signals produced by different types of events can be separated because recoiled protons (related to interaction of fast neutrons in scintillator), alpha particles (related to slow neutrons) and electrons (related to -rays) produce scintillation light pulses with different intensities of slow components. Most of the producers of liquid scintillators provide information only about the fastest component of scintillation light or the mean decay times of the first 3 components [1], [2]. The accurate knowledge of decay time constants and intensities of each component of the light pulses originating from different particles allows improving neutron-gamma discrimination and better understanding of its limitations in both, analog techniques and digital PSD algorithms [3]. loaded liquid In our previous works [4], [5], a study of scintillators has been performed. The measurements were focused on thermal neutron discrimination over gamma rays component and showed different performance of scintillators such as BC523A, BC523A2, EJ339A2 and EJ309B5. Especially, a partial quenching of slow component of the light pulse due to was pointed out. It was confirmed that various loading of content down to 2% in BC523A2 leads to the reduction of increase of the light yield and improvement of the discrimination of thermal neutrons from fast neutrons and gamma-rays. The light pulse shapes following excitation by gamma rays and neutrons were measured in 1968 by Kuchnir and Lynch for Stilbene, NE-213, NE-213M and NE-218 liquids [6]. The BC501A from Saint-Gobain is a present equivalent of NE-213. The distribution of light emitted from scintillators with capability to detect thermal neutrons is less known. The aim of this work is to present measurements of the light pulse shapes and an analysis of decay time constants and intensities of components in liquid scintillators loaded with and sensitive to thermal neutrons. Scintillators not sensitive to thermal neutrons such as BC501A, EJ301 and EJ309 were also tested and comparison of both types of detectors was made. influence on light pulse shape should allow Knowledge of optimization of boron loading in a liquid scintillators leading to the best discrimination of slow/fast neutrons from gammas.

II. EXPERIMENTAL DETAILS A. Scintillators The studies were carried out on seven 2 in diameter and 2 high liquid scintillators from Saint-Gobain Crystals and Eljen

0018-9499/$26.00 © 2010 IEEE

SZCZES ˛ ´ NIAK et al.: LIGHT PULSE SHAPES IN LIQUID SCINTILLATORS ORIGINATING FROM GAMMA-RAYS AND NEUTRONS

3847

TABLE I MAIN PROPERTIES OF THE TESTED LIQUID SCINTILLATORS

a) Measured with XP5500 PMT (blue = 13:7 A=ImF) [5]

Fig. 2. An example of data acquisition system graphical interface together with the three measured parameters: channel 1—energy spectrum (blue), channel 2 and 3—single photons time distribution reflecting light pulse shapes at 500 ns and 2000 ns TAC range (red), channel 4—zero-crossing time (green). A 2D plot of ZC time versus energy is also presented. The data were recorded with BC523A2 scintillator and PuBe neutron source.

Fig. 1. The experimental setup for the light pulse shape measurements using Bollinger-Thomas single photon method and pulse shape discrimination (PSD) module.

Technology. Four of them were loaded and capable to detect thermal neutrons through a capture reaction:

(1) with and 94% branching. The reaction results in forming a distinct peak at energy corresponding to about 60 keV gamma-rays, which is often defined as 60 keV electron equivalent (60 keVee). The main properties of the tested scintillators are presented in Table I. The test scintillators were coupled via a perspex plate (see Fig. 1) to a 3 timing photomultiplier Photonis XP4312 [7] by means of silicone grease. B. Experimental Setup The light pulse shapes were recorded using the BollingerThomas single photon method [8]. Two photomultipliers, Photonis XP4312 (coupled with the tested scintillator) and Hamamatsu R5320 were placed in a light tight black box. Photons

that escaped through the light guide were detected by the fast Hamamatsu R5320 timing PMT characterized by a time jitter of 140 ps [9]. The Hamamatsu photomultiplier was not directly coupled to the light guide in order to minimize the number of detected photons. The space between the light guide and the photodetector was set in the way that assures detection of single photons only. Scintillators were irradiated by PuBe neu, placed in a paraffin ball 18 tron source, emitting cm in diameter. Examples of neutron spectra can be found in [4] and [5]. The recorded time distribution of the single photons detected in R5320 reflects the scintillation light pulse shape. The detailed diagram of the slow-fast experimental setup is presented in Fig. 1. In the fast signal electronic chain (red), the time spectra of single photons distribution were recorded simultaneously in 500 ns and 2000 ns ranges set in Time-to-Amplitude Converters (TAC). In the slow signal electronic chain (blue) the gate was generated, to select the energy range of single photoelectrons. Also pulse height spectrum was collected in this part. In the pulse shape discrimination (PSD) electronic chain (green), the time spectrum related to different particles due to different time of zero-crossing (ZC) stop signal was generated. The four parameters (pulse height, ZC time, single photons time distribution for 500 ns and 2000 ns intervals) were recorded in a list-mode by the Kmax multiparameter data acqusition system from Sparrow Corp. [10]. Additionally a 2D spectrum of the discrimination) versus energy was collected. Such ZC time ( plot allows distinguishing gamma events from fast and thermal neutrons and further extraction, from the recorded list-mode data, the light pulse shapes characteristic for one kind of particle. An example of acquisition system graphical interface together with measured parameters and 2D plot is presented in Fig. 2.

3848


Fig. 4. Normalized light pulse shapes obtained for different gates set at the data recorded with BC523A2 scintillator and at 500 ns TAC range.

Fig. 3. Examples of 2D plots of ZC time versus energy recorded with a BC523A2 scintillator. The spectra of gamma sources are presented in part a) and b). The spectrum recorded with a PuBe neutron source is presented in c) and the same spectrum with gates (light areas) chosen for different particles is showed in d).

C. Electronics Neutron-gamma discrimination was done by means of a modified NDE202 module, developed for the EUROBALL neutron wall [11], which provides two signals, fast timing signal from a constant fraction discriminator (CFD) and zero-crossing signal from discriminator working with a bipolar shaped input signal. These signals were used as a start and stop in Ortec 566 TAC. Stop signal, corresponding to single photons detected in the Hamamatsu PMT, was obtained using Ortec 935 CFD. The light pulse shape time spectra were measured with an Ortec 567 TAC/SCA modules and recorded by multiparameter data acqusition system Kmax. The Kmax electronics was based on three Camac modules: Ortec AD413A 4-channel ADC, Hytec List Processor LP 1341 and SCM-301-2 SCSI Crate Controller from Sparrow Corp. The time calibration of the Time-to-Amplitude Converter was done using a precise Time Calibrator (Ortec 462), based on a quartz clock. III. RESULTS AND DISCUSSION A. Discrimination of Events Originating From Gammas, Fast and Thermal Neutrons An example of 2-dimensional plots recorded using the Kmax data acquisition system is presented in Fig. 3. The X-axis shows the gammas or neutrons energies as detected in the tested liquid scintillator—1st recorded parameter. Y-axis represents the time difference between zero-crossing (ZC) signals corresponding to different events detected in the scintillator (gammas, thermal neutrons or fast neutrons)—4th recorded parameter.

The most pronounced spot on the left side in plots c) and . Events of fast d) represents thermal neutrons captured by neutrons scattered on protons are visible as a horizontal line around channel 120 on ZC time axis. Detected gammas are well separated from neutrons and situated below channel 100. Spot visible around channel 120 at energy axis is a result of Compton scattering of gammas with energy of 480 keV which appears . after thermal neutrons are captured by TACs’ data corresponding to time distributions of single photons emitted after gamma or neutron detection occurred in the scintillator were the 2nd and 3rd parameter recorded during the measurements. However, the final light pulse shape spectrum is a mix of all neutron and gamma events. Incorporating the data presented in Fig. 3 allowed us to set the gates (Fig. 3(d)) at listmode data and extract only the plots typical for a given kind of events. In the case of fast neutrons and gamma radiation the energy gates were set between 100 keV (above 59.5 keV of gamma peak) and 500 keV (Figs. 3(a) and 3(b)). Following the plots for NE-213 presented in [12] the light output in liquid scintillators due to response to electrons is 10 times higher than the one resulting from interaction with protons. Since the lower energy gate was set at about 100 keV, the minimum fast neutron energy can be estimated as 1 MeV. In the case of thermal neutrons the energy gate was set around 60 keV and ZC range was chosen in the way that assured minimal influence of fast neutrons and gamma radiation. An example of light pulse shapes after setting the gates is presented in Fig. 4. The bottom curve is a result of the gate set at gamma events. The two middle curves represent raw spectrum, without gating (lower one) and spectrum with the gates set at thermal neutrons (upper one). The top curve shows pulse shape originating from fast neutrons. B. Fitting Procedure The multi-exponential decay of the slow components in liquid scintillators is generally approximated by two components, medium one with the decay time constant of about 30 ns and the slow one with the time constant above 200 ns [6]. The measurements were performed in the two different time ranges, to determine separately the slow component from the time


Fig. 5. Light pulse shape recorded for the tested BC501A liquid scintillator and gated for gamma events at 2000 ns. The fit was made in time range between 400 ns and 1850 ns and was not sensitive to the fast component.

3849

Fig. 7. Light pulse shape recorded for the tested BC501A liquid scintillator and gated for fast neutrons events at 2000 ns. The fit was made in time range between 400 ns and 1850 ns and was not sensitive to the fast component.

at longer TAC ranges. Therefore the longest component was estimated using data from 2000 ns measurement and double exponential decay fit:

(2)

Fig. 6. Light pulse shape recorded for the tested BC501A liquid scintillator and gated for gamma events at 500 ns.

spectrum measured at 2000 ns range and then the medium and fast components of the light pulses using data of 500 ns range. As a result of gating the experimental data, six light pulse shape spectra (3 for 500 ns and 3 for 2000 ns TAC range) were loaded scintillator and four obtained for each of the tested spectra (only for gamma-rays and fast neutrons) in case of classic scintillators not sensitive to thermal neutrons. For each pair of 500 ns and 2000 ns spectra for a given particle, the spectrum obtained for 2000 ns range was normalized to the maximal number of counts in 500 ns range spectrum. Background level was calculated on the basis of 2000 ns spectra and linear fit to the data points, below main peak, in 20 ns up to 120 ns range. These data points reflect random events. The background level was subtracted from both analyzed spectra (for 500 ns and 2000 ns TAC ranges). First attempt of single exponential decay fit to the spectra recorded at 2000 ns showed that most of the obtained pulses contain additional, longer component (see examples in Figs. 5 and 7). Unfortunately the experimental set-up did not allow measuring the light pulse shapes with required low background level

The fit was made in time range between 400 ns and 1850 ns. was set to 0 and was set to 188 (ns), the value Parameter corresponding to the maximum amplitude of distribution in the time spectrum recorded in 2000 ns range. Additionally, the and was fixed longest component, containing parameters to go through the point with coordinate X equal to 1800 (ns) and Y equal to the mean value of counts in data points between and . In this way, in case of the longest component the decay time becames dependent on the amplitude . The final function contained three parameters to , , . fit: The two calculated long components were subtracted from the 500 ns spectrum. Then again the double exponential decay and fit in 500 ns spectrum was made with parameters fixed to 54 (ns), the position on time axis of the maximum amplitude in this time spectrum. This fit was made in the range 55 ns up to 500 ns. The sum of all the components determined the final fit and representation of the light pulse shape for a given type of particle detected in a scintillator. In summary, the decay of the light pulses of liquid scintillators was approximated by four exponential functions, the fast, medium and two more, describing the slow component. C. Classic Liquid Scintillators The measured light pulse shapes for gamma-rays and fast neutrons, together with multi-exponential fits calculated for the tested BC501A scintillator are presented in Figs. 5–8. Each plot contains experimental data points (squares) extracted from the measured spectra using gates related to a given particle, individual exponential components (dashed lines) and final multi-

3850


TABLE II FITTING PARAMETERS OF ALL THE COMPONENTS CALCULATED FOR THE TESTED LIQUID SCINTILLATORS

Fig. 8. Light pulse shape recorded for the tested BC501A liquid scintillator and gated for fast neutrons events at 500 ns.

exponential fit (solid line). Two long components are plotted as a sum in 500 ns spectra using one dashed line. All the results for the three tested scintillators, together with the decay time constants and intensities of components are collected in Table II. The two decay time constants obtained for events recorded in BC501A scintillator are in good agreement with 3.16 ns and 32.3 ns values reported in [6] for NE-213 liquid. The third component equal to 170 ns is much shorter than 270 ns shown in [6] but this is probably a consequence of fourth component assumed in case of 2000 ns spectra. The background component fitted at 2000 ns time range has the biggest influence on this third, long component and “steals” a part of its intensity. In consequence the long component became faster. For all the three tested liquid scintillators the light pulse shape induced by gamma rays is

Fig. 9. Light pulse shape recorded for the tested BC523A2 liquid scintillator and gated for gamma events at 500 ns.

dominated by a fast component and contribution of two slow decay time constants do not exceed 20%. Presence of fast neutrons enhances the two slow components up to 60%. D. Boron-10 Loaded Liquid Scintillators The light pulse shapes and exponential fits calculated for the loaded scintillators BC523A2 and EJ309B5 are presented in Figs. 9–14. loaded scintillators, together All the results for the tested with the decay time constants and intensities of components are collected in Table II. As it can be observed, fast component is dominating in almost all the scintillators for gamma events, covering about 90% of the emitted light. One exception is EJ309B5 with only 73%


Fig. 10. Light pulse shape recorded for the tested BC523A2 liquid scintillator and gated for fast neutrons events at 500 ns.

Fig. 11. Light pulse shape recorded for the tested BC523A2 liquid scintillator and gated for thermal neutrons events at 500 ns.

Fig. 12. Light pulse shape recorded for the tested EJ309B5 liquid scintillator and gated for gamma events at 500 ns.

intensity of fast component. In case of neutron events, fast component is reduced to about 70% for thermal energies and to 60% for fast neutrons. Again, results obtained for EJ309B5 scintillator differ slightly showing lower intensities of fast component,

3851

Fig. 13. Light pulse shape recorded for the tested EJ309B5 liquid scintillator and gated for fast neutrons events at 500 ns.

Fig. 14. Light pulse shape recorded for the tested EJ309B5 liquid scintillator and gated for thermal neutrons events at 500 ns.

with values equal to about 50% and 40% for thermal and fast neutrons, respectively. It is interesting that intensities of medium component are almost identical in a given scintillator comparing results for thermal and fast neutrons. The difference is observed only for EJ309B5 and here the pulse associated with thermal neutrons posses a little bit higher intensity than the one originating from fast neutrons. The results presented in Table II confirm that intensity of long components (a sum of long and background component) have dominating influence on the quality of pulse shape discrimination. For all the tested scintillators the intensity of these components is strongly related to different particles. The change in intensities reflects the discrimination capabilities reported in [5]. BC523A and EJ339A2 are characterized by low intensities of long components, starting from around 3% for gamma-rays and reaching the highest values of about 15%–19% for fast neutrons. Comparison of BC523A and BC523A2 liquids shows inloading. The higher quenching of fluence of 4.4% and 2% long decay scintillation light can be observed in the case of . In the case of BC523A2 BC523A loaded with 4.4% of all intensities of long components reloaded with 2% of lated to all three particles considerably increased. The EJ309B5

3852


scintillator which showed the best discrimination properties in [5] has the highest intensities of long components. effect on unComparison of EJ309 and EJ309B5 shows loaded liquid. As it can be seen the decay time constants are slightly changed for all the components with a little bit higher values in case of loaded detector. Similar effect can be observed comparing intensities of fast and medium component. More pronounced changes can be observed for long components which . A gamma-ray intensities were increased after loading with component increased by 50% whereas the neutron component reached 16% higher value. Very high intensities of long components obtained for a BC501A scintillator suggest that loading this liquid with can lead to similar discrimination capabilities as in the case of EJ309B5. In all the tested scintillators the decay time constants, for different particles do not differ significantly.

Loading scintillators with results in the increase of intensity of slow components and has minimal influence on fast and medium component.

IV. CONCLUSIONS Experimental setup, with perspex light guide plate placed between PMT and scintillator output window made possible measurement of the light pulse shapes in liquids and other scintillators encapsulated in traditional way (with only one output window). Application of multiparameter DAC system and offline analysis of list-mode data allowed easy discrimination of light pulses induced by three kinds of particles: gamma rays, fast neutrons or thermal neutrons. The recorded light pulse shapes can be described by multi-exponential decay with three or four components. The different intensities of these components are related to the kind of particle detected in the scintillator. The difference in intensity of long components is the most crucial for particle discrimination especially for separation of thermal neutrons from gamma-rays.

REFERENCES [1] Saint-Gobain Crystals BC501A and BC523A Liquid Scintillator Product DataSheet. [2] Eljen Technology, EJ301, EJ309 and EJ339A Data Sheets. [3] E. Gatti and F. de Martini, “A new linear method of discrimination between elementary particles in scintillation counters,” in Proc. Nuclear Electronics II Conf. (Belgrade, May 1961), Vienna, Austria, 1962, IAEA, 265. [4] L. Swiderski, M. Moszynski, D. Wolski, T. Batsch, A. Nassalski, A. Syntfeld-Kazuch, T. Szczesniak, F. Kniest, M. R. Kusner, G. Pausch, J. Stein, and W. Klamra, “Boron-10 loaded BC523A liquid scintillator for neutron detection in the border monitoring,” IEEE Trans. Nucl. Sci., vol. 55, no. 6, pp. 3710–3716, Dec. 2008. [5] L. Swiderski, M. Moszynski, D. Wolski, T. Batsch, J. Iwanowska, A. Nassalski, A. Syntfeld-Kazuch, T. Szczesniak, F. Kniest, M. R. Kusner, G. Pausch, J. Stein, W. Klamra, P. Schotanus, and C. Hurlbut, “Further study of boron-10 loaded liquid scintillators for detection of fast and thermal neutrons,” in Proc. IEEE Nuclear Science Symp. Conf. Rec., Oct. 19–25, 2008, pp. 2378–2382. [6] F. T. Kuchnir and F. J. Lynch, “Time dependence of scintillations and the effect on pulse-shape discrimination,” IEEE Trans. Nucl. Sci., vol. 15, no. 3, pp. 107–113, Jun. 1968. [7] T. Szczesniak, M. Gierlik, M. Kapusta, M. Moszynski, D. Wolski, P. Lavoute, and E. Rossignol, “The 75 mm diameter Photonis XP43D2 photomultiplier with the screening grid at the anode for timing experiments,” IEEE Trans. Nucl. Sci., vol. 53, no. 3, pp. 1540–1546, Jun. 2006. [8] L. M. Bollinger and G. E. Thomas, “Measurement of the time dependence of scintillation intensity by a delayed coincidence method,” Rev. Sci. Instrum., vol. 32, pp. 1044–1050, 1961. [9] M. Moszyn´ski, M. Kapusta, A. Nassalski, T. Szcz˛esńiak, D. Wolski, L. Eriksson, and C. L. Melcher, “New prospects for time-of-flight PET with LSO scintillators,” IEEE Trans. Nucl. Sci., vol. 53, no. 5, pp. 2484–2488, Oct. 2006. [10] [Online]. Available: http://www.sparrowcorp.com/products/software [11] O. Skeppstedt et al., “The EUROBALL neutron wall—Design and performance tests of neutron detectors,” Nucl. Instrum. Methods Phys. Res. A, vol. A421, pp. 531–541, 1999. [12] R. L. Craun and D. L. Smith, “Analysis of response data for several organic scintillators,” Nucl. Instrum. Methods, vol. 80, pp. 239–244, 1970.