Characterization and performance of the DTAS detector

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY-NC-ND 4.0 https://creativecommons.org/licenses/by-nc-nd/4.0/ Characterization and performance of the DTAS detector © 2018 Elsevier B.V. Accepted version (Final draft) Guadilla, V.; Tain, J. L.; Algora, A.; Agramunt, J.; Äystö, Juha; Briz, J. A.; Cucoanes, A.; Eronen, Tommi; Estienne, M.; Fallot, M.; Fraile, L. M.; Ganioğlu, E.; Gelletly, W.; Gorelov, Dmitry; Hakala, Jani; Jokinen, Ari; Jordan, D.; Kankainen, Anu; Kolhinen, Veli; Koponen, Jukka; Lebois, M.; Meur, L. Le; Martinez, T.; Monserrate, M.; Montaner-Pizá, A.; Moore, Iain; Nácher, E.; Orrigo, S. E. A.; Penttilä, Heikki; Pohjalainen, Ilkka; Porta, A.; Reinikainen, Juuso; Reponen, Mikael; Rice, S.; RintaAntila, Sami; Rubio, B.; Rytkönen, Kari; Shiba, T.; Sonnenschein, Volker; Sonzogni, A. A.; Valencia, E.; Vedia, V.; Voss, Annika; Wilson, J. N.; Zakari-Issoufou, A.-A. Guadilla, V., Tain, J. L., Algora, A., Agramunt, J., Äystö, J., Briz, J. A., Cucoanes, A., Eronen, T., Estienne, M., Fallot, M., Fraile, L. M., Ganioğlu, E., Gelletly, W., Gorelov, D., Hakala, J., Jokinen, A., Jordan, D., Kankainen, A., Kolhinen, V., . . . Zakari-Issoufou, A.-A. (2018). Characterization and performance of the DTAS detector. Nuclear Instruments and Methods in Physics Research Section A : Accelerators, Spectrometers, Detectors and Associated Equipment, 910, 79-89. https://doi.org/10.1016/j.nima.2018.09.001 2018 Accepted Manuscript Characterization and performance of the DTAS detector V. Guadilla, J.L. Tain, A. Algora, J. Agramunt, J. Äystö, J.A. Briz, A. Cucoanes, T. Eronen, M. Estienne, M. Fallot, L.M. Fraile, E. Ganio˘ glu, W. Gelletly, D. Gorelov, J. Hakala, A. Jokinen, D. Jordan, A. Kankainen, V. Kolhinen, J. Koponen, M. Lebois, L. Le Meur, T. Martinez, M. Monserrate, A. Montaner-Pizá, I. Moore, E. Nácher, S.E.A. Orrigo, H. Penttilä, I. Pohjalainen, A. Porta, J. Reinikainen, M. Reponen, S. Rice, S. Rinta-Antila, B. Rubio, K. Rytkönen, T. Shiba, V. Sonnenschein, A.A. Sonzogni, E. Valencia, V. Vedia, A. Voss, J.N. Wilson, A.-A. Zakari-Issoufou PII: S0168-9002(18)31079-9 DOI: https://doi.org/10.1016/j.nima.2018.09.001 Reference: NIMA 61162 To appear in: Nuclear Inst. and Methods in Physics Research, A Received date : 31 May 2018; Revised date : 31 July 2018; Accepted date : 1 September 2018 Please cite this article as: V. Guadilla, et al., Characterization and performance of the DTAS detector, Nuclear Inst. and Methods in Physics Research, A (2018), https://doi.org/10.1016/j.nima.2018.09.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Characterization and performance of the DTAS detector V. Guadillaa,1,∗, J.L. Taina, A. Algoraa,b, J. Agramunta, J. ¨ Ayst¨oc, J.A. Brizd, A. Cucoanesd, T. Eronenc, M. Estienned, M. Fallotd, L.M. Frailef, E. Ganio˘glug, W. Gelletlya,h, D. Gorelovc, J. Hakalac, A. Jokinenc, D. Jordana, A. Kankainenc, V. Kolhinenc, J. Koponenc, M. Leboisi, L. Le Meurd, T. Martineze, M. Monserratea, A. Montaner-Piz´aa, I. Moorec, E. N´acherj, S.E.A. Orrigoa, H. Penttil¨ac, I. Pohjalainenc, A. Portad, J. Reinikainenc, M. Reponenc, S. Riceh, S. Rinta-Antilac, B. Rubioa, K. Rytk¨onenc, T. Shibad, V. Sonnenscheinc, A.A. Sonzognik, E. Valenciaa, V. Vediaf, A. Vossc, J.N. Wilsoni, A.-A. Zakari-Issoufoud aInstituto de F´ısica Corpuscular, CSIC-Universidad de Valencia, E-46071, Valencia, Spain bInstitute of Nuclear Research of the Hungarian Academy of Sciences, Debrecen H-4026, Hungary cUniversity of Jyv¨askyl¨a, FIN-40014, Jyv¨askyl¨a, Finland dSubatech, IMT-Atlantique, Universit´e de Nantes, CNRS-IN2P3, F-44307, Nantes, France eCentro de Investigaciones Energ´eticas Medioambientales y Tecnol´ogicas, E-28040, Madrid, Spain fUniversidad Complutense, Grupo de F´ısica Nuclear, CEI Moncloa, E-28040, Madrid, Spain gDepartment of Physics, Istanbul University, 34134, Istanbul, Turkey hDepartment of Physics, University of Surrey, GU2 7XH, Guildford, UK iInstitut de Physique Nucl`eaire d’Orsay, 91406, Orsay, France jInstituto de Estructura de la Materia, CSIC, E-28006, Madrid, Spain kNNDC, Brookhaven National Laboratory, Upton, NY 11973-5000, USA Abstract DTAS is a segmented total absorption γ-ray spectrometer developed for the DESPEC experiment at FAIR. It is composed of up to eighteen NaI(Tl) crystals. In this work we study the performance of this detector with laboratory sources and also under real experimental conditions. We present a procedure to reconstruct offline the sum of the energy deposited in all the crystals of the spectrometer, which is complicated by the effect of NaI(Tl) light-yield non-proportionality. The use of a system to correct for time variations of the gain in individual detector modules, based on a light pulse generator, is demonstrated. We describe also an event-based method to evaluate the summing-pileup electronic distortion in segmented spectrometers. All of this allows a careful characterization of the detector with Monte Carlo simulations that is needed to calculate the response function for the analysis of total absorption γ-ray spectroscopy data. Special attention was paid to the interaction of neutrons with the spectrometer, since they are a source of contamination in studies of β-delayed neutron emitting nuclei. Keywords: βdecay, total absorption γ-ray spectrometer, exotic nuclei, NaI(Tl) detector, non-proportional scintillation light yield, Monte Carlo simulations 1. Introduction Decay studies of exotic nuclear species at the focal plane of the FAIR-NUSTAR Super Fragment Separator in the DESPEC experiment [1] will provide information on the nuclear structure and the5 astrophysics impact of exotic nuclei. Far from stability, the Qβvalues are very large, and the corre- ∗Corresponding author Email address: [email protected] (V. Guadilla) 1Current address: Subatech, IMT-Atlantique, Universit´e de Nantes, CNRS-IN2P3, F-44307, Nantes, France sponding increase in level density implies, on the one hand, the fragmentation of the βfeeding into many levels populated in the decay and, on the10 other hand, the fragmentation of the γintensity between many possible cascades. Total Absorption γ-Ray Spectroscopy (TAGS) has been shown to be an accurate tool to determine β-decay intensity distributions for such nuclei far from the valley of β15 stability. This technique avoids the so-called Pandemonium effect [2], related to the relatively poor efficiency of HPGe detectors. Instead of detecting individual γrays as in high-resolution experPreprint submitted to Nuclear Instruments and Methods A July 31, 2018 *Manuscript Click here to view linked References iments with HPGe detectors, TAGS aims to detect20 the full β-delayed electromagnetic cascade. This is achieved with spectrometers made of large scintillator crystals covering a solid angle of ∼4π. To extract the β-intensity distributions a de-convolution procedure is applied to the measured energy spec-25 trum using the detector response, as will be explained later. For this reason, a new spectrometer has been designed and constructed for the DESPEC experiment [3]. The Decay Total Absorption γ-ray Spec-30 trometer (DTAS) is a segmented detector that consists of a maximum of eighteen NaI(Tl) crystals with dimensions 150 mm ×150 mm ×250 mm [3]. The choice of the material, as well as the geometry, comes from a careful study where several goals and35 constraints were taken into account: the efficiency, the amount of dead material, the energy and time resolution, the neutron sensitivity, the sensitivity limit for high-lying β-intensity, the coupling to ancillary detectors, and the cost [3]. The advantage40 of the segmentation in this case is threefold: the possibility to extract information from the modulemultiplicity spectra, as will be explained later, the possibility of using the individual modules as single γdetectors, and the mechanical flexibility of the45 set-up. In fact, we consider two main configurations for DTAS: a sixteen-module configuration designed for experiments at fragmentation facilities, as proposed in [3], and an eighteen-module configuration for experiments at ISOL-type facilities. Both con-50 figurations without shielding can be seen in Fig. 1. In the eighteen-module configuration side holes can be made by moving away the modules of the horizontal central plane, thus allowing access from both sides of the detector, as shown in Fig. 1 bottom.55 In this way DTAS can be combined with ancilliary detectors and it is possible to position a beam pipe in the centre of the spectrometer. This configuration has recently been commissioned at IGISOL [4], with holes of 10 cm used to place a HPGe detector60 from one side and the beam pipe with a βdetector from the other side. The two central modules were separated by 16 cm instead of 10 cm in order to lower their counting rate, so that it was comparable to the external modules. The configuration65 foreseen for FAIR [3], with sixteen modules, will be coupled to the Advanced Implantation Detector Array (AIDA) [5]. In order to place AIDA in the center of DTAS, the two central modules in the eighteen-module configuration are removed and the70 two modules above the central hole are supported by a specially designed aluminium frame with external dimensions identical to a module, as shown in Fig. 1 upper panel. The shielding surrounding DTAS is composed of75 stainless steel sheets, lead bricks and aluminium, and it served to reduce the background counting rate by one order-of-magnitude in the measurements of this work. The allocation of individual modules to positions in the arrangement was done80 according to their resolutions, ranging from 7% to 9% at 661.7 keV, so that the positions associated with the lowest counting rates (the eight corners of the assembly shown in Fig. 1) were occupied by the modules with the poorest resolution.85 Figure 1: DTAS detector in the sixteen-module configuration (top) and in the eighteen-module configuration (bottom) without radiation shielding. The outline of the article is the following: in section 2 we will describe the procedure to reconstruct the full energy deposited in the detector from the signals of the individual modules. In section 3 a method to evaluate the summing-pileup contami-90 nation will be explained, and its validation with calibration sources will be discussed. Finally, the Monte Carlo (MC) response function of the detector will be described in section 4, and the repro2 duction of several calibration sources and the neu-95 tron contamination coming from β-delayed neutron emitters will be discussed. 2. Total energy reconstruction: hardware sum and software sum In this section we will describe the electronic100 chain employed to process the signals from the individual modules of DTAS, and the procedure to reconstruct the total energy deposited in the detector. In particular, two methods to calculate the total energy sum will be discussed: the hardware105 sum and the software sum. 2.1. Signal processing In order to analyse data from DTAS we have to reconstruct accurately, for each event, the energy deposited in the full spectrometer and its110 module-multiplicity, Mm(number of modules that fire above the threshold, often known as fold). The full energy released in the spectrometer is obtained by summing the energy deposited in the individual modules, either electronically or via software.115 The electronic chain to process the signals from the modules was designed with this idea in mind, and it is represented in Fig. 2. We use Mesytec MSI-8p preamplifiers [6] for both anode and dynode signals from the photomultiplier120 tubes (PMTs). MSI-8p are custom adapted units with fixed gain, preamplifier constants optimized for these PMTs and without timing filter amplifier. After the preamplifier, dynode signals are split into two branches; one branch is sent to a CAEN N625125 Quad Linear FAN-in FAN-out [7], and the other to Mesytec MSCF-16 shapers. The N625 module acts as an analog signal adder and one of the outgoing signals is processed in an ORTEC 671 amplifier [8] to produce the sum energy signal (hardware sum)130 sent to the analog to digital converter (ADC), a CAEN V785 module, of the data acquisition system (DACQ). Another output from the N625 module is used to construct a common stop signal sent to a time to digital converter (TDC), CAEN V775,135 using an ORTEC 474 Timing Filter Amplifier and an ORTEC 584 Constant Fraction Discriminator. The MSCF-16 shapers provide individual energy and timing output signals that are sent to the individual channels of the ADC and TDC modules140 respectively. The anode signals after the preamplifier are sent to sampling digitizers of a second digital DACQ, running in self-triggered mode, which is not discussed in this publication. In order to carry out the hardware sum prop-145 erly we need to match the gains of the different PMTs by adjusting the high voltage (HV) applied to them, so that the signals of individual modules are aligned. Note that aligned here means having the same amplitude for the same energy deposited.150 The software sum is reconstructed offline from the individual signals processed with the MSCF-16 shapers. In the following subsections we will show a method of correcting possible changes in the gain of the modules, as well as the way to perform properly155 the alignment and determine the software sum of these signals. 2.2. Gain correction system A system to correct changes in the gain of individual modules has been developed, as planned160 in [3]. These changes may be due to temperature variations [9], drift of the PMT current and fluctuations in the HV supply. In this system the gain of each module is monitored checking the position of the peak produced by a pulsed light source. An165 additional external reference detector, with a weak 137Cs radioactive source, is used to monitor the stability of the light pulse generator. The following elements are employed in this system: •An external reference well-type NaI(Tl) detec-170 tor of 3” diameter ×3” length manufactured by Saint Gobain [10]. The well has 15 mm diameter and 40 mm depth. The crystal is mounted on a 3” diameter ETI 9305 PMT as shown in Fig. 3.175 •A 490 nm light pulse generator model 6010 from BNC [11]. The generator is triggered with an external 100 Hz clock signal. •A 2 m long bundle of borosilicate glass fibres split into 20 bundles of 2 mm diameter, manu-180 factured by FiberTech Optica [12]. The fibres are terminated with SMA type connectors. •A weak 137Cs source of ∼300 Bq. The fibre bundle splitter is used to distribute the light pulse from the generator to the reference de-185 tector and to each of the eighteen modules. The 137Cs source is placed inside the well of the reference detector. The reference detector is surrounded by lead shielding and is placed close to DTAS. Since 3 Figure 2: Schematic diagram of the electronic chain. The labels correspond to: Preamplifier (Preamp), Spectroscopic Amplifier (Amp), Timing Filter Amplifier (TFA), Constant Fraction Discriminator (CFD), Gate/Delay Generator (GDG), Time to Digital Converter (TDC), Analog to Digital Converter (ADC). Figure 3: NaI(Tl) reference well type detector. Inset: a view of the front face with the hole where the weak 137Cs source is placed. both the reference detector and DTAS have shield-190 ing, this weak source does not affect the DTAS measurements. The position of the 661.7 keV peak in the well detector provides a reference for possible changes in the gain of this detector. Comparing the position of the light pulser peak with this peak195 we can determine if there are variations of the intensity of the light source. With this information we can separate in each module variations in the gain from variations in the light source intensity. The gain correction is calculated for short time inter-200 vals, and the procedure will be detailed in the next subsection. An example of the spectra of the reference detector and one individual module of DTAS showing the light pulser peaks can be seen in Fig. 4.205 In order not to disturb the measured individual spectra, the peak due to the light pulser has to be located beyond the energy region of interest, see Fig. 6 as an example. When choosing an optical fibre bundle for each module, we took into account210 that each of the 20 bundles does not transport the same amount of light, and the individual modules do not convert the same amount of incident light into the same signal amplitude in the PMT. For both reasons, in order to minimize the difference in215 position of light pulser peaks between modules we assigned the bundles that transport more light with the worst modules in terms of light conversion. Apart from applying the gain correction offline, the gain correction system could also be used for220 maintaining the alignment of the signals of the modules during the measurement by applying periodic HV corrections to the PMTs. This requires information about the dependence of the gain with the HV for each module. Although we have tested225 this online correction method, it was not used in the actual measurements. 4 Channels 500 1000 1500 2000 2500 Counts 1 10 2 10 3 10 4 10 Source Light peak Channels 500 1000 1500 2000 2500 3000 Counts 1 10 2 10 3 10 Cs peak 137 Light peak Figure 4: Individual DTAS detector spectrum with the light pulser peak in a 60Co measurement (top). Reference detector spectrum with the 661.7 keV peak from the weak 137Cs source, and the light pulser peak (bottom). 2.3. Software sum Just as in the case of the hardware sum, before performing the software sum the amplitude230 of the signals stored for each event must be properly aligned. Although signal amplitudes were gainmatched before the FAN-in FAN-out for the hardware sum, and even though the gains of the shapers are set to a common value, the stored amplitude in-235 formation needs to be corrected due to slight variations in gain and offset of the individual electronic channels. The first idea for making this alignment was to convert signal amplitude (proportional to light col-240 lected) into energy for each of the individual channels. This conversion between light collected and deposited energy is what we will call energy calibration. A solution like this has been successfully adopted for a 12-fold segmented BaF2spectrometer245 in previous works [13, 14, 15, 16]. Nevertheless, we soon realized that it can not be applied in the case of a segmented detector made of NaI(Tl) because of the non-proportionality of the light yield in this material [17, 18]. The reason is related to what was250 pointed out in [19], explaining the shift of the position of full energy peaks due to γ-ray cascades with respect to single γ-ray peaks of the same energy. For every primary electron created by γ-ray interactions there is a shift of about 10 keV in the appar-255 ent energy. Since γ-rays of several hundreds of keV to a few MeV typically require of the order of three interactions (two Compton, one photoelectric) to deposit the full energy this explains why for a γcascade of two γ-rays (γ-multiplicity, Mγ=2) the260 shift is approximately 30 keV, while for Mγ=3 it is 60 keV and so on. In the case of a segmented detector the situation is more complicated, and the shift depends not only on the γ-multiplicity, Mγ, but also on the number of modules where the energy is265 deposited, Mm, which determines the distribution of the number of primary electrons in each module. Taking into account the different ways that 3 ×Mγ electrons can be distributed in Mmmodules one can determine that the apparent energy shifts follow ap-270 proximately the numbers in Table 1. The first row in the table corresponds to the behaviour of a single NaI(Tl) crystal spectrometer like LUCRECIA at ISOLDE [20] or the LBNL spectrometer used at GSI [21].275 Mγ1234 Mm 1 0 +30 +60 +90 2−30 0 +30 +60 3−60 −30 0 +30 4−90 −60 −30 0 Table 1: Shift in keV of the sum peak position due to the non-proportionality of the light yield in a segmented NaI(Tl) spectrometer when the individual modules are calibrated in energy before the software sum. For a single crystal spectrometer the nonproportionality is not a problem as far as this effect is included in the MC simulations in the way detailed in [19]. Likewise, it does not present any problem for the hardware sum of a segmented280 NaI(Tl) spectrometer, as long as the PMTs are gain-matched. However, the consequence of applying an energy calibration to individual modules before summing, is that the resolution of the sum peaks is worsened due to the displacement of285 the different Mmmodule-multiplicities contributing to the sum. The non-proportionality of the light yield in NaI(Tl) is known to have an important contribution to the resolution of single crystal 5 detectors [22, 23], and by extension of the hard-290 ware sum of multi-crystal detectors. Here we are referring to an additional effect that must be taken into account when reconstructing the software sum for multi-crystal detectors. In Fig. 5 these shifts are shown for a measurement of 22Na (Mγ=3) and295 for the corresponding MC simulation of this source that includes the non-proportionality of the light yield as in [19]. In both cases an energy calibration has been applied to all the individual modules before summing. The vertical black line cor-300 responds to 2296.5 keV, the sum of the energies of the three γ-rays involved: 511 keV, 511 keV and 1274.5 keV. The sum peaks of the different Mmmodule-multiplicities are not aligned, showing a displacement in agreement with Table 1. Only305 Mm=3 is aligned with the nominal sum, since it corresponds to a 0 keV shift in Table 1, with three γ-rays detected in three crystals. Note that the experimental spectra are not background subtracted, whereas the MC is only widened by the light func-310 tion from [19], without taking into account additional contributions to the resolution. In summary, in order to maintain the resolution, we have to align the stored amplitudes of the individual modules, thus reproducing with our software315 sum the same behaviour of the hardware sum, and, equivalently, of a single crystal detector. In addition there are other effects that may worsen the resolution, like changes in the gain of the PMTs and the electronic chain. The correction we have to320 apply to counteract these effects is applied sequentially on a reduced number of events. The number of events should be sufficiently large to determine the peak positions accurately and sufficiently small to limit the effect of gain variations during325 the acquisition time. We have verified that 1 million events (that corresponds to approximately 4 minutes for a typical counting rate of 4-5 kHz in DTAS), fulfill this condition. The stored amplitude is represented by the bin330 number in the histogram accumulated by each ADC channel (detector module) and the first step is to determine the offset and the gain. To determine the ADC offset for each channel we use the position of the peak due to the electronic noise. The335 gain is obtained from the position of the two peaks from a calibration run with a 22Na source (511 keV and 1274.5 keV). With the offset and gain so obtained the alignment of the first million events is performed choosing one arbitrary module as a ref-340 erence. Energy [keV] 2000 2200 2400 2600 2800 Counts 1 10 2 10 3 10 Total =1 m M =2 m M =3 m M =4 m M =5 m M =6 m M MC Energy [keV] 2000 2200 2400 2600 2800 Counts 1 10 2 10 3 10 4 10 Total =1 m M =2 m M =3 m M =4 m M =5 m M =6 m M EXP Figure 5: 22Na software sum of the light produced in the individual detectors calibrated in energy. Both the MC (top) and the experimental measurement (bottom) show the shifts from Table 1. The vertical black line corresponds to the sum of the energies of the three γ-rays involved. (For interpretation of the references to color in this figure caption, the reader is referred to the web version of this paper.) After the alignment, the reference values of the parameters involved in the gain correction procedure are determined. The ADC offset is represented by aj, with j= 0...18, with j= 0 being the well345 detector and j= 1...18 the DTAS modules. The reference position of the light pulser peak for each module, Lj, is obtained by peak fitting. Analogously, the 137Cs peak and the light pulser peak reference positions for the well detector, P0and L0 350 respectively, are determined. The next group of one million events is then processed. We define L0 jas the new light pulser peak position of module j,bjas the gain change factor of module j, and Cas the change factor in the light355 source intensity. The procedure described below is followed in order to calculate the gain corrections and sum the amplitudes of all modules stored in each event: •The new position of the 137Cs peak for the well360 detector, P0 0, is determined, as well as the po6 sition of the light pulser peak L0 0. •The change in the gain of the PMT of the well detector is calculated: b0=P0−a0 P0 0−a0 (1)365 •The change in the light produced by the light pulse generator, C, is calculated: C=L0−a0 b0(L00−a0)(2) •L0 jis determined for each of the DTAS modules, and with this value the gain change factor370 is calculated taking into account the change of intensity of the light source from Equation 2: bj=Lj−aj C(L0 j−aj)(3) Once the parameters bjare determined we reprocess the same group of events applying the gain375 correction factor in order to align the amplitudes of all modules to the first group of events used as a reference. As a result of applying this procedure, the software sum can be performed properly, as can be seen in Figure 6 for a 60Co source. At this point380 an energy calibration can be applied (a conversion between light collected and energy) by using single peaks (Mγ=1), as in the case of the hardware sum. Energy [keV] 0 5000 10000 Counts 1 10 2 10 3 10 4 10 Total Sum Individual modules Light pulser peaks Figure 6: Software sum spectrum of a 60Co source measurement and individual module spectra showing the alignment. The peaks in individual modules above 8 MeV are due to the light pulser. The software sum reconstructed in this way exhibits the same behaviour as the hardware sum in385 terms of the non-proportionality of the light yield, as seen in Fig. 7 for two calibration sources. In both cases the segmented detector behaves as a single crystal detector in terms of the position of the sum peak. The main differences are related to a390 slightly better resolution in the software sum with respect to the hardware sum due to the gain corrections, and a different shape in the pileup region that will be commented on next section. The final resolution achieved for the software sum of the eighteen395 modules is 8.6% at 661.7 keV, worse than the 7.7% value obtained for the single prototype module in [3]. Energy [keV] 0 1000 2000 3000 4000 5000 Counts 1 10 2 10 3 10 4 10 5 10 Software sum Hardware sum Energy [keV] 0 1000 2000 3000 4000 5000 Counts 1 10 2 10 3 10 4 10 5 10 Software sum Hardware sum Figure 7: Comparison of the hardware sum (grey) and the software sum (black) in DTAS for a 60Co source (top) and a22Na source (bottom). In order to ensure that the treatment of the nonproportionality is correct, we can check the spectra400 of the Mmmodule-multiplicities after this process. In Fig. 8, we show the good alignment of the different Mmmodule-multiplicity spectra achieved with this method for a 22Na source (Mγ=3) and a 60Co source (Mγ=2), in contrast with the results shown405 in Fig. 5. The vertical black lines correspond to the sum peak positions calculated using the shift associated with single crystals according to [19] (first row of Table 1): 2296.5 keV+60 keV for the 22Na 7 ture is quite sensitive to the de-excitation pattern after neutron emission. For 95Rb, the use of the evaluated decay scheme available in ENSDF pro-755 duced a wrong shape for the spectrum. However when we use the de-excitation scheme in 94Sr measured by Kratz et al. [36] a good reproduction was obtained as can be seen in Fig. 16. Thus except when the neutron emission proceeds entirely to the760 ground state, the effect of the neutron energy distribution on the shape of the capture peak is obscured by the final nucleus γspectra. Since the latter is often unknown or poorly known, it seems difficult to obtain reliable information about the shape of the765 β-delayed neutron spectrum from TAGS spectra in the general case. Ideally the normalization of the β-delayed neutron contribution to the total spectrum should be determined by the Pnvalue, as we did in previous770 works with a BaF2spectrometer [13, 14, 15, 16]. However, in BaF2there is no clear peak to check the accuracy of this normalization. Recently it was suggested that precise Pnvalues could be obtained from TAGS measurements with a NaI(Tl) spec-775 trometer [29]. In the present case, from the normalization of the simulated and measured counts in the capture bump at 6.8 MeV we have obtained Pnvalues of 6.8% and 10.9% for 137I and 95Rb respectively, after taking into account all the contam-780 inants (summing-pileup and activity of the descendants). Note that in comparison with the numbers given in Table 3, we found a 5% smaller value for 137I, while for 95Rb a 25% larger value is obtained. Compared to our recently measured values [34] the785 differences are -12% and +20% respectively. We studied the dependence of the extracted Pnon the length of the time window applied to the experiment (coincidence gate) and to the MC simulation. We found that in the range 300-500 ns the results790 were stable within 3%. It should be noted that the value we obtain is 14% lower than the value of 7.9(4)% obtained by a similar procedure with MTAS [29]. We suspect that the differences with respect to the reference numbers given in Table 3795 are due to uncertainties in the MC simulations of the interaction of neutrons. In view of this discrepancy and the fact that for 95Rb we also obtain a large difference but of opposite sign, we conclude that further investigations are needed before decid-800 ing on the reliability of Pnextraction from TAGS spectra [29]. In any case, the key point for us is the reproducibility of the shape of the spectra of the βdelayed neutron contamination that affects the ex-805 traction of Iβ(Ex) from the analysis of TAGS spectra. A proper determination of this background component is particularly relevant when extracting an accurate value for the βintensity above the neutron separation energy that proceeds by γemission,810 Pγ[15]. The investigation of γ/neutron competition from neutron unbound states is a topic of current active research. The importance of the correction of the background due to β-delayed neutrons for the determination of Pγusing TAGS spectrom-815 eters made of NaI(Tl) can not be overlooked. This material has a large capture cross-section resulting in large β-delayed neutron detection efficiencies, of the order of 40%. We note, in particular, that this correction has been ignored in a recent measure-820 ment of Pγfor 70Co decay with the SuN spectrometer [37] and might change their result significantly. The sensitivity of the MC simulation of the β-n decay contamination to the knowledge of the decay (neutron spectrum and γ-cascades after neutron825 emission) represents a challenge for very neutronrich nuclei in the general case where this information is poorly known or not known at all. Given that γ-rays produced by neutron interactions are delayed with respect to β-particle emission one830 can use timing information to discriminate between these signals [3]. We have tested this idea for 137I and 95Rb with reasonable results, as will be shown in a forthcoming publication. However, this type of time discrimination cannot be applied for the γ-835 ray de-excitation in the final nucleus after neutron emission since they are prompt with respect to the β-particles. The best option here seems to use the spectrometer itself to obtain information about this type of contamination as was suggested in [28]. The840 modularity of DTAS helps here, since there will be a certain degree of spatial separation between γ-rays coming from the final nucleus and those coming from neutron interactions. This can be exploited to tag β-delayed neutron events by setting a co-845 incidence gate on the neutron capture ”peak” observed, for example, in one half of the spectrometer and looking at the spectra in the other half of the spectrometer. Work to demonstrate the feasibility of this approach is in progress.850 5. Conclusions The characterization of the DTAS detector has been carried out. A gain stabilization system based 14 on a light pulse generator has been tested successfully. The non-proportionality of the light yield ef-855 fects in a NaI(Tl) multi-crystal spectrometer were taken into account to reconstruct properly the sum of the total energy deposited in the spectrometer. The summing-pileup distortion of the spectrum was successfully computed using a revision of a method860 previously developed, and for high-rate measurements an improvement in this method has been introduced with the help of MC simulated data. A careful Geant4 MC simulation of the DTAS detector response to β-decays has been performed. The865 quality of the response function, needed for any TAGS analysis, has been validated after obtaining excellent agreement when comparisons were made with measurements of calibration sources. This includes in particular a good agreement of Mm 870 module-multiplicity gated spectra. A nice agreement between the measured and simulated shape of the β-delayed neutron background was also obtained for two well known neutron emitters. 6. Acknowledgements875 This work has been supported by the Spanish Ministerio de Econom´ıa y Competitividad under grants FPA2011-24553, AIC-A-2011-0696, FPA2014-52823-C2-1-P and the program Severo Ochoa (SEV-2014-0398), by the European Commis-880 sion under the FP7/EURATOM contract 605203, and by the Spanish Ministerio de Educaci´on Cultura y Deporte under the FPU12/01527 grant. The work was also supported by the UK Science and Technology Facilities Council (STFC) grant885 ST/P005314/1. E. Ganio˘glu was supported by the Istanbul University Scientific Research Project Unit under FYO-2017-24144 project. References [1] B. Rubio, Int. J. Modern Phys. E 15 (2006) 1979.890 [2] J. Hardy, et al., Phys. Lett. B 71 (2) (1977) 307. [3] J. L. Tain, et al., Nucl. Instrum. and Methods A 803 (2015) 36. [4] V. Guadilla, et al., Nucl. Instrum. and Methods B 376 (2016) 334.895 [5] T. Davinson, et al., Technical Report for the Design, Construction and Commissioning of the Advanced Implantation Detector Array (AIDA)[link]. URL http://www.fair-center.eu/fileadmin/fair/ publications_exp/aida_TDR.pdf900 [6] [link]. URL http://www.mesytec.com/ [7] [link]. URL http://www.caen.it/ [8] [link].905 URL http://www.ortec-online.com/ [9] M. Moszyn´ski, et al., Nucl. Instrum. and Methods A 568 (2006) 739. [10] [link]. URL http://www.crystals.saint-gobain.com/910 [11] [link]. URL http://www.berkeleynucleonics.com/ [12] [link]. URL https://fibertech-optica.com/ [13] J. L. Tain, et al., Phys. Rev. Lett. 115 (2015) 062502.915 [14] A.-A. Zakari-Issoufou, et al., Phys. Rev. Lett. 115 (2015) 102503. [15] E. Valencia, et al., Phys. Rev. C 95 (2017) 024320. [16] S. Rice, et al., Phys. Rev. C 96 (2017) 014320. [17] D. Engelkemeir, Rev. Sci. Instr. 27 (1956) 589.920 [18] S. Payne, et al., IEEE Transactions on Nuclear Science 56 (2009) 4. [19] D. Cano-Ott, et al., Nucl. Instrum. and Methods A 430 (1999) 333. [20] B. Rubio, et al., J. Phys. G: Nucl. Part. Phys. 44 (2017)925 084004. [21] M. Karny, et al., Nucl. Instrum. and Methods B 126 (1997) 411. [22] P. Dorenbos, et al., IEEE Transactions on Nuclear Science 42 (1995) 6.930 [23] J. Valentine, et al., IEEE Transactions on Nuclear Science 45 (1998) 512. [24] B. Rasco, et al., Nucl. Instrum. and Methods A 788 (2015) 137. [25] D. Cano-Ott, et al., Nucl. Instrum. and Methods A 430935 (1999) 488. [26] S. Agostinelli, et al., Nucl. Instrum. and Methods A 506 (2003) 250. [27] J. L. Tain, D. Cano-Ott, Nucl. Instrum. and Methods A 571 (2007) 728.940 [28] M. Karny, et al., Nucl. Instrum. and Methods A 836 (2016) 83. [29] B. C. Rasco, et al., Phys. Rev. C 95 (2017) 054328. [30] J. L. Tain, D. Cano-Ott, Nucl. Instrum. and Methods A 571 (2007) 719.945 [31] J. L. Tain, et al., Nucl. Instrum. and Methods A 774 (2015) 17. [32] E. Mendoza, et al., IEEE Transactions on Nuclear Science 61 (2014) 4. [33] Evaluated Nuclear Structure Data Fle, http://www.950 nndc.bnl.gov/ensdf/. [34] J. Agramunt, et al., Nucl. Instrum. and Methods A 807 (2016) 69. [35] M. Brady, Evaluation and Application of Delayed Neutron Precursor Data, Los Alamos National Laboratory,955 1989. [36] K.-L. Kratz, et al., Z. Phys. A 306 (1982) 239. [37] A. Spyrou, et al., Phys. Rev. Lett. 117 (2016) 142701. 15 Energy [keV] 0 1000 2000 3000 4000 5000 Counts 1 10 2 10 3 10 4 10 Na 22 Experiment MC Summing-Pileup Energy [keV] 0 2000 4000 6000 Counts 1 10 2 10 3 10 4 10 =1 m Na M 22 Experiment MC Summing-Pileup Energy [keV] 0 2000 4000 6000 Counts 1 10 2 10 3 10 4 10 =2 m Na M 22 Experiment MC Summing-Pileup Energy [keV] 0 2000 4000 6000 Counts 1 10 2 10 3 10 4 10 =3 m Na M 22 Experiment MC Summing-Pileup Energy [keV] 0 2000 4000 6000 Counts 1 10 2 10 3 10 4 10 =4 m Na M 22 Experiment MC Summing-Pileup Energy [keV] 0 2000 4000 6000 Counts 1 10 2 10 3 10 =5 m Na M 22 Experiment MC Summing-Pileup Figure 15: 22Na experimental spectra after subtracting the environmental background (solid grey) compared with the MC simulations (solid black) taking into account the summing-pileup contamination (dashed blue). The sum energy spectrum without conditions and with a condition on module-multiplicity Mmfrom 1 to 5 is shown. 16 Energy [keV] 0 5000 10000 Counts 1 10 2 10 3 10 4 10 I 137 -gatedβDTAS MC Energy [keV] 0 5000 10000 Counts 1 10 2 10 3 10 4 10 Rb 95 -gatedβDTAS MC Figure 16: Simulation of the β-delayed neutron decay branch for neutron emitters measured in the commissioning of DTAS at IGISOL: 137I (top) and 95Rb (bottom). Experimental β-gated spectra (in grey) are compared to simulations (in black). 17