On the time response of background obtained in γ -ray spectroscopy experiments using LaBr3(Ce) detectors with different shielding

Nuclear Instruments and Methods in Physics Research A 811 (2016) 42–48

Contents lists available at ScienceDirect

Nuclear Instruments and Methods in Physics Research A journal homepage: www.elsevier.com/locate/nima

On the time response of background obtained in γ-ray spectroscopy experiments using LaBr3(Ce) detectors with different shielding J.-M. Régis n, M. Dannhoff, J. Jolie, C. Müller-Gatermann, N. Saed-Samii Institut für Kernphysik der Universität zu Köln, Zülpicher Str. 77, 50937 Köln, Germany

art ic l e i nf o Article history: Received 17 October 2015 Received in revised form 30 November 2015 Accepted 9 December 2015 Available online 18 December 2015 Keywords: Lifetime measurements The centroid-shift method Time response LaBr3(Ce) scintillators Active/passive shielding

a b s t r a c t Employing the γ-γ fast-timing technique with LaBr3(Ce) scintillator detectors allows the direct determination of lifetimes of nuclear excited states with a lower limit of about 5 ps. This limit is increased as soon as background is present in the coincidence spectra underneath the full-energy peaks of the γ-γ cascade. Our aim was to identify the components of the γ-ray background by systematic γ-γ fast-timing measurements using different types of γ shielding within a large γ-ray spectrometer. The energy dependent physical zero-time response was measured using background-free full-energy peak events from the 152Eu γ-ray source. This is compared with the time response of the (Compton-) background distribution as obtained using the prompt 60Co γ-ray source. The time response of the typical Compton background is about 15 ps faster than the time response of background-free full-energy peak events. Below about 500 keV, a second type of background contributes by the detection of Compton-scattered γ rays generated in the materials of the spectrometer around the detector. Due to the additional time-offlight of the Compton-scattered γ rays, this low-energy background is largely delayed. Compared with a bare cylindrical 1:5 in:
1:5 in: LaBr3(Ce) detector, the BGO-shielded detector in the Comptonsuppression mode improves the peak-to-total ratio by a factor of 1.66(5), while the Pb-shielded detector only slightly reduces the low-energy background. & 2015 Elsevier B.V. All rights reserved.

1. Introduction The use of electronic fast–timing techniques to measure lifetimes via γ–ray detection have become popular since the LaBr3(Ce) scintillators are commercially available [1–4]. The high scintillation light yield, expressed as the number of scintillation photons N ph: created per energy deposited in the scintillator (E 63 000 photons/MeV [5]), provides an excellent energy resolution of about 3% at the 137Cs energy of 662 keV [5,6]. This high resolution also improves the ratio of counts in the full-energy peak (FEP) to the counts of background (e.g. Compton background) which lie underneath the FEP compared to other scintillators used in γ-ray spectroscopy experiments, such as NaI or BaF2 with energy resolutions of 7% and 10%, respectively. In addition, theffi timing perpffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi formance, expressed by the figure of merit τsc: =N ph: [7] with the scintillator decay constant τsc: ( E15 ns for 5% Ce doped LaBr3 [5]), is comparable to the fastest scintillators available. This all makes it possible to perform electronic γ-γ time-difference measurements to access lifetimes of nuclear excited states which provide essential information on the nuclear structure of the considered atomic nucleus. Lifetimes of nuclear excited states down to about 10 ps n

Corresponding author. Tel.: þ 49 221 4703622; fax: þ49 221 4705168. E-mail address: [email protected] (J.-M. Régis).

http://dx.doi.org/10.1016/j.nima.2015.12.017 0168-9002/& 2015 Elsevier B.V. All rights reserved.

have been measured precisely by means of relative centroid-shift measurements using the γ-γ fast-timing technique with LaBr3(Ce) scintillator detectors [4,8,9]. Basically, two (start and stop) detectors are used to measure the time difference between the signals generated by FEP events of a γfeeder -γdecay cascade that interconnects an excited state of interest by using a linear time-to-amplitude converter (TAC). Constant fraction discriminators (CFD) are used to transform the fast negative anode output pulses of the LaBr3(Ce) plus photomultiplier detector assemblies into logic timing signals which are connected (uniquely) to the start and stop inputs of the TAC. The energy information is delivered by the pulse height of the positive dynode output pulse of the two photomultiplier tubes and are acquired in coincidence to the TAC output pulse height for off-line analysis. In any case, a “delayed” time-difference distribution is obtained if a transition which feeds the excited state, γfeeder , provides the start signal to the TAC, while the decay transition of the state, γdecay , provides the stop signal (for the inverted case, one obtains the “anti-delayed” time distribution). Assuming no background underneath the FEPs of the two γ rays, the delayed time distribution D(t) is a convolution of the prompt response function (PRF) of the set–up P(t) with an exponential decay as [10]: Z t 0 0 DðtÞ ¼ nλ Pðt 0  t 0 Þe  λðt  t Þ dt with λ ¼ 1=τ; ð1Þ 1

J.-M. Régis et al. / Nuclear Instruments and Methods in Physics Research A 811 (2016) 42–48

where n is the number of counts in the time distribution, λ is the transition (decay) probability and τ the mean lifetime of the nuclear excited state and t0 is the centroid (position) of the PRF which is obtained for τ r 1 ps. Lifetimes which are larger than the full width at half maximum (FWHM) of the PRF are obtained by a fit of the slope of D(t) outside the region of P(t) using an exponential function, according to Eq. (1) for t t 0 4 FWHM [10]. Using the centroid-shift method [11], also lifetimes smaller than the FWHM of the PRF can be measured. The centroid or center of gravity is defined as the first moment of the time distribution: R1 tDðtÞ dt CD ¼ o t 4 ¼  1 : ð2Þ n The statistical uncertainty of the centroid determination is related to pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi the variance of the time distribution as: δC D ¼ o t 2 4  ot 4 2 . For lifetimes τ oFWHM, the time distribution is nearly a Gaussian and the centroid uncertainty is approximately equal to the absolute time resolving power of the set-up δt related to the standard deviation σ ¼ FWHM=2:355 of the PRF and the statistics:

σ δC Dτ o FWHM  δt ¼ pffiffiffi:

ð3Þ

n

According to the centroid-shift method, the centroid of a delayed time distribution is displaced by the mean lifetime from the prompt centroid CP of its convoluted PRF [11]:

τ ¼ C D  C Pstop ;

ð4Þ

or τ for the anti-delayed time distribution. represents the FEP (zero-) time response of the stop detector. Using LaBr3(Ce) detectors, the energy-dependent FEP time response given by C P ðEγ Þ can be measured with an accuracy of about 5 ps for 40 keV o Eγ o 1408 keV by using the FEP events of about a dozen coincident transitions provided by the 152Eu γ-ray source [12]. The major systematic error that can be induced in centroid measurements is related to the background underneath the FEPs observed in γ-ray spectroscopy experiments. For energies larger than 300 keV, the background mainly consists of Compton events, where the Compton-scattered γ-ray escapes the detector after primary Compton interaction in the LaBr3(Ce) scintillator. It has been shown that the time response of single-interaction Compton background differs from the FEP time response related to multiple Compton scattering and final photo absorption in the creation of the FEP [13]. Additional γ-ray background is expected for Eγ o 300 keV, such as Compton-scattered γ-rays (e.g. backscatter peaks) which are detected after a primary interaction in the surrounding material of the LaBr3(Ce) crystal, the spectrometer (e.g. lead shields) and also the other detectors; the latter one describes inter-detector Compton scattering which provoke false γ-γ coincidences. In any case, the uncertainty of the lifetime determination will suffer from the time correction related to background contributions to the time spectrum dependent on the peak-tobackground ratio and the time response of the background [13]. The (full-energy) peak-to-total value, P=T, represents a fundamental detector characteristic and expresses the ratio of counts in the FEP to the counts in the total spectrum [14]. Analogous to γ-γ coincidence experiments with HPGe detectors, the FEP efficiency should be increased relative to the total, as the fraction F of meaningful events in a multifold γ-ray coincidence experiment is given by [15]: ¼ C Pstart  C AD

F ¼ ðP=TÞm ;

C Pstop

ð5Þ

where m is the γ-ray multiplicity. An improvement factor of about 3 has been reported for the P=T value of a HPGe detector, that was placed inside a large bismuth–germanate (BGO) anti-Compton shield [15]. We present the results of P=T and γ-ray background time response measurements obtained using LaBr3(Ce) detectors

43

with different shielding conditions for comparison. These informations are important for the design of future applications such as the Advanced Time Delay method (HPGe-gated β-γ fast-timing technique [13]) and especially the γ-γ lifetime measurement using fast–timing arrays consisting of many LaBr3(Ce) detectors. One aim is to construct the fast-timing array “FATIMA” [16,17] for the decay spectroscopy (DESPEC) experiment [18] at the focal plane of the fragment separator Super FRS at the Facility for Anti-proton and Ion Research (FAIR) within the Nuclear Structure Astrophysics and Reactions (NUSTAR) collaboration.

2. Experimental set-up and results Our aim was to compare P=T values and time response measurements in a realistic experimental scenario. For this purpose, the γ-ray cube spectrometer HORUS of the Institut für Kernphysik (IKP) in Cologne, Germany, was equipped with 6 BGO–shielded LaBr3(Ce) detectors and 8 bare HPGe detectors. This configuration is used to perform the HPGe-gated γ-γ fast–timing technique [19]. Using the high-resolution HPGe detectors to select a γ ray in coincidence with the γfeeder  γdecay cascade to be measured with the LaBr3(Ce) detectors, this can considerably clean the LaBr3(Ce) γ-ray coincidence spectra and thus improve the peak-tobackground ratio. The construction of the BGO anti-Compton detectors in use is similar to the one described in Ref. [15]. As schematically illustrated in Fig. 1, the wedge shaped BGO shield consists of 10 optically separated BGO segments, each connected to a photomultiplier tube. Each segment is made of a 14 cm long and 3–5 cm thick trapezoidal main BGO crystal coupled to a “nose” BGO crystal (about 3-cm thick) for the detection of backscattered γ rays. To avoid direct irradiation from the target, 3-cm thick lead (Pb) collimators with circular opening diameter of 3 cm are mounted on the front of the BGO nose. The LaBr3(Ce) detectors, consisting of cylindrical 1:5 in:
1:5 in: LaBr3 ðCeÞ crystals coupled to Hamamatsu R9779 photomultiplier tubes (PMT), were placed inside the BGO Compton suppressors at a distance of 9 cm to the target (source) position. The position of the LaBr3(Ce) detector relative to the BGO segments requires a certain optimization related to the solid angle of the detected γ rays and the Compton suppression. To obtain the best suppression of “backscattered” γ rays from the contribution of Compton events in the LaBr3(Ce) near the Compton-edge, the face of the LaBr3(Ce) detector should be placed behind the back end of the nose BGO (the dashed lines in Fig. 1). As shown in Fig. 1, our set-up was optimized for high γ-

PMT BGO

Pb

*

LaBr3

PMT

VD

anode dynode HV

γ −ray source BGO nose Fig. 1. Schematic drawing of the BGO-LaBr3(Ce) detector set-up used for this work. This set-up is optimized for high γ-ray detection efficiency in the LaBr3(Ce) detector, as explained in the text. Two outputs are delivered by the PMT voltage divider (VD) of the LaBr3(Ce) detector, adjustable by a high-voltage (HV) supply. The negative anode output pulses are used for fast timing and the positive dynode pulses for energy determination. The BGO detection system provides one output pulse which is directly connected to a CFD to generate an anti-coincidence window.

44

J.-M. Régis et al. / Nuclear Instruments and Methods in Physics Research A 811 (2016) 42–48

ray detection efficiency by placing the face of the LaBr3(Ce) detector close to the front end of the nose BGO. The center (target position) of the HORUS spectrometer is located inside a globated 1-mm thick aluminum target chamber with diameter of 6 cm connected to a beam line of the 10-MV tandem Van-de-Graaff accelerator of the IKP. The BGO-shielded LaBr3(Ce) detectors were arranged in central symmetry with respect to the center (beam spot) of the target position with ðϑ; φÞ angles of ð01; 7 901Þ, ð 7 551; 01Þ and ð 7 551; 1801Þ relative to the beam axis (z-axis). The HPGe detectors were placed in between the large BGOs at angles of ð 7 551; 7 901Þ, ð01; 7 451Þ and ð01; 7 1351Þ with a distance of 15 cm to the target position. An additional LaBr3(Ce) detector, also containing a cylindrical 1:5 in:
1:5 in: crystal, was placed at angles of ð301; 601Þ and a distance of 9 cm to the target. This detector is surrounded by BGO and HPGe detectors in a compact arrangement by almost touching the other detectors and Pb collimators and thus provides an ideal test detector within an environment with high fluence of Compton-scattered γ rays. This test detector has been used once without a shield and once with a 5-mm thick and 6-cm broad Pb wrapping to passively shield the sides of the LaBr3(Ce) scintillator. For picosecond sensitive γ-γ fast timing, an analogue electronics fast-timing circuitry as described in Ref. [20] was installed with 6 TACs to measure time differences between the very fast LaBr3(Ce) detector timing signals of all possible detector–detector combinations. The energy signals of the Ge and LaBr3(Ce) detectors and the TAC output signals were connected to synchronized 80 MHz digitizers. The CFD signals of the BGOs were individually connected to the according LaBr3(Ce)-detector veto inputs of the digitizers. For optimum Compton suppression, the thresholds of the BGO CFDs were carefully set just above the noise of the baseline. The CFD pulse of the BGO generates a window that needs to be optimized relative to the LaBr3(Ce) signal by adjusting the “gate delay” and “gate width” parameters in the configuration of the digital LaBr3(Ce)-signal processing. Each detector and TAC signal was used for digital pulse shape analysis to provide the detector or TAC ID, its pulse height and the related time information (12.5 ns time stamp) in a list-mode data format. By this trigger-less data acquisition, coincidence requirements are set in the off-line analysis as desired. The data are analyzed using the “SOCOv2” sorting code of the IKP described in Ref. [19]. 2.1. Reduction of

γ background

Considering the fast-timing technique using γ rays, the reduction of background is desired, as the background mainly results from Compton scattering of a primary γ ray inside or outside of the detector. If the primary γ ray creating the background is in coincidence with a γ ray of the cascade of interest and which is detected by a reference detector, the background event creates a false γ-γ coincidence with incorrect time information. It is experimentally impossible to distinguish between a true and a false γ-γ coincidence of same energies. According to Eq. (5), the improvement of the FEP efficiency relative to the total efficiency is crucial for the γ-γ fast-timing technique. This can be achieved by employing a larger detection volume, simply in order to reduce the escape of γ rays after a Compton interaction in the detection volume. However, the time resolution (FWHM of PRF) degrades rapidly with the detection volume due to a larger time spread related to the additional time-of-flight of the scintillation photons, and thus the increase of the volume is limited. The Compton background can be suppressed using a large BGO shield in anticoincidence mode to the γ-ray detector. In addition, the BGO Compton suppressor also act as a passive shield to protect the detector from Compton-scattered γ rays which are generated in the experimental surroundings and would hit the side of the bare

γ−ray source detection volume

a

c b 3

1 γ−ray detector

2 passive Pb or active BGO shield

Fig. 2. Schematic illustration of the background obtained in γ-ray spectroscopy experiments using a compact detector array including shielding materials. The “type a” case indicates an inter-detector Compton scattering with final absorption of the Compton-scattered γ ray in the second detector (detector 2). For cases of type b, the origin of the absorbed Compton-scattered γ ray is from the materials (e.g. from a lead shield) of which is made the whole spectrometer. Type c events generate the Compton background which can be suppressed using active BGO shielding.

detector. Often, a few-mm thick Pb shield is used to suppress the detection of Compton-scattered γ rays. In any case, the mean free path or penetration depth δ of γ rays in material is the decisive factor to improve the set-up and is given by

δðEγ Þ ¼

1

;

μðEγ Þ

ð6Þ

where μ is the energy-dependent linear attenuation coefficient of the material. Fig. 2 illustrates 2 types of background events which are connected to the detection of Compton-scattered γ rays. The type a event represents the inter-detector Compton scattering which provoke the so-called “cross-talk event” as one γ ray produces signals in two detectors; in detector 1 due to the energy released by the Compton effect (Compton background) and in detector 2 due to the absorption of the Compton-scattered γ ray. It is evident that cross-talk events are most unwanted in γ-γ fast-timing experiments. The type b events generate “background” as soon as a FEP (or Compton) event is detected simultaneously by another detector, e.g. by detector 3 in Fig. 2. In principle, the time response of such γ-ray background of type b can be measured and used for possible correction of the lifetime determination, similar to the correction related to the Compton background described in Refs. [13] and [19]. In a preparatory experiment, the suppression of cross-talk events has been measured in dependence of the thickness of a Pb shield. Two bare LaBr3(Ce) detectors were place nearly side-byside with distance of 5–10 mm from each other, similar to detectors 1 and 2 in Fig. 2. 10 cm
10 cm and 5-mm thick Pb sheets were inserted between the two detectors in order to fully cover the sides of the LaBr3(Ce) scintillators. The 152Eu γ-ray source has been shown to be ideal for testing the set-up for unwanted crosstalk events [19]. Using a narrow energy window (gate) on the FEP of the 444-keV transition detected by the reference detector (e.g. detector 2 in Fig. 2), the coincidence spectrum of the other detector should show 2 FEPs at 964 keV and 1085 keV, corresponding to the energy of two transitions which decay from the same excited state in 152Sm at 1085 keV fed by the 444-keV transition. Ideally, the FEP intensity ratio R ¼ I 964 =I 1085 should be 1.42(1) [21]. This intensity ratio, corrected by the energy dependent FEP efficiency, has been determined in dependence of the Pb thickness, as presented in Fig. 3. Note that the intensity ratio for the unshielded case is much larger than 1.42. The additional counts in the 964-keV FEP of the coincidence spectrum result from

J.-M. Régis et al. / Nuclear Instruments and Methods in Physics Research A 811 (2016) 42–48

1112−444 & 1089−444 & 1085−444

100

2000

without shield

R=3.18(10)

964−444 1085

50

counts per keV

150

964 & 1408−444

no shield (gate on 444)

45

1500

with 5−mm Pb shield 1000

with active BGO shield

500

0 150

5−mm Pb shield 0

counts per 4 keV

R=1.84(6)

200

100

400

600

800

1000

1200

1400

E γ [keV] Fig. 4. The singles energy spectra for different shielding conditions of LaBr3(Ce) detectors installed within the HORUS spectrometer as obtained using the 60Co γ-ray source. The spectra are normalized to the FEP at 1333 keV.

50

0

10−mm Pb shield R=1.46(5)

100

50

0 400

500

600

700

800

900

1000

1100

E γ [keV] Fig. 3. γ-γ Coincidence spectra from a pair of adjacent LaBr3(Ce) detectors in dependence of the thickness of the Pb shield installed in between the detectors. One of the two detectors selects the 444 keV transition in 152Sm, the electroncapture decay product of 152Eu. The top panel illustrates the effect of cross-talk events which provoke “ghost peaks” in the coincidence spectra. The intensity ratio R ¼ I 964 =I 1085 provides a sensitive measure for the detection of cross-talk events. The results are discussed in the text.

Compton scattering of the primary 1408-keV γ ray of 152Sm with detected energy transfer of 964 keV. The escaped Comptonscattered γ ray with energy of 444 keV, corresponding to the reference energy gate, is absorbed by the reference detector and thus provides the undesired cross-talk event. Assuming the sensitive Compton scattering angles φ for a pair of adjacent detectors to be in the range of 601 o φ o 1201, the energy Eγ 0 of the Compton-scattered γ ray from a primary Eγ ¼ 1:4 MeV γ-ray is in the range of 300–660 keV. The linear attenuation coefficient of lead μPb is larger than 1 cm  1 for Eγ 0 o662 keV [22] and thus the penetration depth is δ o 1 cm, according to Eq. (6). As indicated in Fig. 3, using a 10-mm thick lead shield (5 mm per detector), the cross-talk events generated by primary γ rays with energies lower than 1.4 MeV are almost completely suppressed. Concerning the HORUS set-up for γ-γ fast-timing experiments described in Section 2, no significant cross-talk events could be detected. Obviously, the 3–5 cm thick BGO segments almost fully absorb Compton-scattered γ rays, regardless where they are coming from. Therefore, also for the bare LaBr3(Ce) detector which is located between the BGOs of the other LaBr3(Ce) detectors, no cross-talk events have been observed. Peak-to-total values have been determined by employing the standard procedure using a 60 Co source for three different shielding conditions. The singles energy spectra for the three conditions are compared in Fig. 4 and

Table 1 Peak-to-total (P/T) values of LaBr3(Ce) detectors used with different shielding conditions inside the HORUS spectrometer, as obtained from the singles energy and γ-γ coincidence spectra generated by using 60Co and presented in Figs. 4 and 5. Shield

P=T (singles) (%)

P=T (γ-γ) (%)

Background at 150 keV (%)

No 5-mm Pb Passive BGO Active BGO

15.8(2) 16.8(2) 17.5(2) 28.5(1)

17.8(5) 18.6(5) 19.1(5) 30.1(3)

100 76 73 36

the corresponding P=T values are listed in Table 1. The P=T values were determined as the ratio of counts in the two FEPs of the 60Co source to the total counts in the spectrum for Eγ Z100 keV [15]. A remarkable result of this experiment is that the improvement of the P=T value of the Pb-shielded detector is only marginal relative to the bare detector. Note that in both cases, the same detector was used, thus at the same position relative to the reference detector which is a BGO-shielded one. Only for energies smaller than about 400 keV, a small decrease of the background is observed when using a lead shield. This is a hint for the low-energy γ background to be partially of type b (see Fig. 2). At 150 keV, the decrease of the background is about 24%. In total, the Compton background (type c in Fig. 2) dominates and presumably becomes the only component of the background for Eγ 4500 keV. It is clear that the use of a Compton suppressor remarkably improves the P=T value. As can be seen in Fig. 4, the Compton background over the total dynamic range is reduced almost uniformly by a factor of nearly 2. The peak–like structures at the Compton edges observed in the BGO suppression mode arise by the fact that the backscattered γ ray from a central Compton scattering in the detector escapes via the entrance window of the detector and thus cannot be detected by the BGO for suppression. About the same argument holds for the backscatter peaks seen at about 210 keV. Here, the central Compton event occurs in the material behind the source in front of the detector and the backscattered γ ray enters the detector via its entrance window. It can not be excluded that also cross-talk events are present in the region of the backscatter peaks. Other artifacts are seen in region of 60–90 keV, where the FEPs of the Pb and Bi K–X rays are overlapped. These γ-ray induced X rays are generated within femtoseconds [23] and thus are in coincidence with γ rays in a multifold γ coincidence experiment (see also Fig. 5). This can be a limiting factor for γ-γ fast-timing experiments in this energy region. To overcome this problem, one can think of resigning the use of any shield by arranging the LaBr3(Ce) detectors nearly side-by-side in a compact and intelligent geometry by

46

J.-M. Régis et al. / Nuclear Instruments and Methods in Physics Research A 811 (2016) 42–48

300

counts per keV

250

without shield

200

with 5−mm Pb shield 150

with active BGO shield

100 50 0 0

200

400

600

800

1000

1200

E γ [keV] Fig. 5. The γ-γ coincidence spectra for different shielding conditions of the LaBr3(Ce) detectors presented in Fig. 4. The results were obtained by setting a 50keV wide gate centered on the FEP events of the 1333-keV transition detected by a reference detector. The spectra are normalized to the FEP at 1173 keV.

almost touching each other, e.g. in a ring structure such as the two ones of the EXILL&FATIMA spectrometer described in Ref. [19]. There, γ-γ coincidence events detected in pairs of adjacent LaBr3(Ce) detectors are rejected from the analysis. This procedure eliminates cross-talk events and furthermore provides an active Compton suppressor. A P=T value of 22(1)% has been measured for the EXILL&FATIMA spectrometer. Considering the actual HORUS spectrometer in the γ-γ coincidence mode, the improvement factor for the P=T value of the BGO-LaBr3(Ce) detector combination is equal to 1.66(5) relative to the bare LaBr3(Ce) detector. A larger value is expected when optimizing the position of the LaBr3(Ce) detector within the BGO shield. As shown in Fig. 5 and presented in Table 1, P=T values were also determined from the γ-γ coincidence spectra as obtained by using a reference LaBr3(Ce) detector to gate on the FEP events at 1333 keV which are in coincidence with the 1173-keV γ-ray transition. As in this case, no (time-correlated) background is underneath the FEP at 1333 keV (see Fig. 4), similar P=T values are expected for either singles or coincidence spectra. The P=T value in the γ-γ coincidence mode is improved by 5–12% dependent on shielding. A small reduction is expected for uncorrelated background generated by “background radiation” from radioactive or activated materials around the detector. In our case, uncorrelated background is present due to “internal activity”, as about 0.09% of the natural lanthanum is radioactive. In total, the background distributions of the spectra presented in Figs. 4 and 5 and the related P=T values are similar. 2.2. Time response of

γ background

For the determination of lifetimes which are smaller than the FWHM of the PRF of the set-up, the centroid-shift method is used (Eq. (4)) to access the limits of the electronic fast-timing technique. The reduction of background represents an important issue for this purpose, as the background is considered to be nearly prompt. One goal is to provide time distributions which are free of background and thus, where the time information is directly related to the mean lifetime by Eq. (4) and no correction is needed. As will be demonstrated in this section, background-free time spectra can be obtained by using intelligent gates. In the most general case, the FEPs are sitting on time-correlated background. The experimental time distribution then consists of a superposition of the FEP time distribution Dðt; Eγ Þ and the time distribution of the background Bðt; Eγ Þ. Due to an energy dependency of the position and the FWHM of time distributions, the time

distribution of the background needs to be determined experimentally for a correction by taking into account the number of background events underneath the FEP. This can be achieved by an interpolation from the measurement of background time spectra Bðt; E0γ Þ at different energies E0γ below and above the FEP [13,19]. It is important to note that in general, the time response of the socalled prompt background C B ðEγ Þ, e.g. by using 60Co, differs from the (prompt) FEP time response C P ðEγ Þ of same energy [13,24]. In a realistic scenario, the FEP is sitting on a superposition of Compton continuum of primary γ rays with different energy which may be affected by significant lifetimes. Therefore, the analysis to derive the correction for background contributions needs to be performed for each individual FEP. Our prior objective was to determine the residual difference (“RD” [13]) between the FEP time response and the time response of the background generated using the prompt 60Co source. To measure the FEP time response, a 152Eu γ-ray source has been used for three good reasons: firstly, the 152Eu source produces more than 20 coincident γ rays in the energy range of 40–1408 keV which often corresponds to the energy region of the interest; secondly, the lifetimes of the relevant excited states in the daughters 152Sm and 152Gd are smaller than 100 ps and known with a precision better than 2.5 ps; and most importantly, no or nearly no background contributions are obtained by using appropriate gates. As already shown in Fig. 3, a background-free FEP is obtained at 1085 keV by gating the reference detector on the 444-keV FEP, as no coincident transition with Eγ 4 1085 keV is given for Eref: ¼ 444 keV. The peak–to–background ratio of the 964-keV FEP is very large ð 425Þ and the background contribution in this case is negligible. Using double events (mLaBr3 ðCeÞ ¼ 2 and mHPGe ¼ 0) and the 344-keV FEP as reference energy, FEPs with (virtually) no background are obtained at 779, 1089 and 1299 keV and for Eref : ¼ 244 keV, at 867 and 1213 keV. As shown in Fig. 6a, also the 40-keV Sm K-X ray FEP is free of background. This was achieved by using triple events (mHPGe ¼ 1 and mLaBr3 ðCeÞ ¼ 2) and taking advantage of the high-resolution HPGe detectors, which were used to gate on the 122-keV ground-state transition in 152Sm. By using in addition the reference LaBr3(Ce) detector to gate on the FEP of the 1408-keV γ-ray transition which feeds the 122-keV state of 152Sm, all parallel transitions with respective backgrounds are eliminated and only the 40-keV FEP remains. The 40-keV Sm K-X ray is emitted after the electron-capture decay of 152Eu within femtoseconds [23], which is faster than the lifetimes of the excited states in 152Sm populated by the electron-capture decay. Therefore, the resulting time spectrum of the 40–1408 keV coincidences shown in Fig. 6b is prompt and free of background. The integration region for the determination of the centroid is chosen such that no random coincidences on the left and right of the time spectrum are taken into account. For different reference energies, the corresponding data points (lifetime-corrected or prompt centroids) in the centroid diagram are shifted in parallel according to the energy dependent FEP time response of the reference detector. Therefore, the FEP time responses of both detectors have been measured by using an iterative procedure. The resulting centroids were fitted using a function which describes the timing uncertainty of the CFD timing principle reported in Ref. [12]. In Fig. 7, the FEP time response of the bare “stop” detector is presented and compared with the time response as obtained by “scanning” the background distribution of the 1173-keV transition in 60Ni using 20-keV wide gates and a gate on the coincident 1333-keV FEP by the reference start detector (see also Fig. 5). The resulting background centroids were shifted in parallel to fit the centroid of the 1333–1173 keV FEP-FEP time spectrum with the measured FEP time response at 1173 keV. The FWHM of the prompt 1333–1173 keV time spectrum was measured to be 226(3) ps.

J.-M. Régis et al. / Nuclear Instruments and Methods in Physics Research A 811 (2016) 42–48

47

30 600

20 gates: HPGe on 122 keV & LaBr3(Ce) on 1408 keV

400

10 RD [ps]

counts per keV

500

300

−10 −20 −30

0 10

50

100

150

200

250

300

350

60

start gate: 1408 keV

intergration region

stop gate: 40 keV

CP

20 10

3

& HPGe gate: 122 keV

1 −2

−1

0

1

2

t [ns] Fig. 6. (a) The LaBr3(Ce) coincidence spectrum obtained using the 152Eu γ-ray source with FEP gates as indicated. (b) The HPGe-gated background-free LaBr3(Ce)– LaBr3(Ce) time-difference distribution of the prompt 40–1408 keV cascade in 152Sm (τ ¼ 0:5ð1Þ ps [25]). The FWHM of this low-energy combination is 790(5) ps.

200

time response of "prompt" γ −ray background (60 Co)

150 100 50 0 −50

FEP time response (152Eu)

−100 −150 −200

0

200

400

600

−40 300 400 500 600 700 800 900 1000 1100 1200 E γ [keV] Fig. 8. The residual difference (RD) between the FEP time response and the time response of Compton events (data points from the 60Co source with Eref: ¼ 1333 keV) obtained using different shielding conditions.

E γ [keV]

counts per 10 ps

0

200 100

centroid [ps]

no shield 5−mm Pb active BGO

800

1000

1200

E γ [keV] Fig. 7. The centroid-shift analysis of the bare LaBr3(Ce) detector (“stop detector”) located inside the HORUS spectrometer of the IKP. Comparison between the energy-dependent zero-time response of FEP events generated by the 152Eu γ-ray source and the background time response obtained using the prompt 60Co γ-ray source. In both cases, background-free FEP events were selected by the reference start detector. The results are discussed in the text.

As can be seen in Fig. 7, the low-energy background below about 400 keV as obtained from a prompt γ ray source is delayed relative to the FEP time response. This is a clear indication for a part of the low-energy background to be created by Comptonscattered γ rays which are generated and coming from the

materials around the bare detector. Such type b background is largely delayed due to the additional path way and related timeof-flight of the scattered γ ray relative to the direct target to detector path way. Note the sudden increase of the delay observed at about the energy of the backscatter peak corresponding to 210 keV. For Eγ 4 400 keV, the background events are faster than the FEP time response. This indicates for single-interaction Compton events creating the Compton background compared to multiple interactions in the creation of the FEP, in agreement with the observations made in Refs. [13,24]. Similar centroid-shift analyses have been performed for Pband BGO-shielded LaBr3(Ce) detectors. Fig. 8 presents the residual difference (RD) between the FEP time response represented by the horizontal zero line and the background time response for Eγ 4 400 keV. No significant differences of the RDs for different shielding conditions are observed in this case. Obviously, the background with Eγ 4400 keV nearly completely consists of Compton background. Although the BGO remarkably suppresses Compton events, still some Compton background is obtained in cases where the Compton-scattered γ ray escapes the detector via its entrance or exit window. Independent of shielding, the shift (RD) between the Compton time response and the FEP time response becomes nearly constant for Eγ 4600 keV and corresponds to  14(5) ps at 950 keV. This shift is slightly larger than the values reported in Ref. [13] for a cylindrical 1 in:
1 in: BaF2 crystal. The latter comparison may confirm the statement that the RD for Compton events increases with the crystal size (detection volume) [13]. As demonstrated in Refs. [13] and [26], the Compton RD also is dependent on the energy Eref: of the FEP that is selected by the reference detector. When considering the low-energy region ðEγ o400 keVÞ, significant differences of the RDs are observed dependent on the type of shield, as can be seen in Fig. 9. Although only a small portion of the low-energy background is suppressed when using passive Pb shields, this results to a remarkably smaller delay between the background time response and the FEP time response by about 50 ps relative to the unshielded case for Eγ o 250 keV. The type b background cannot be suppressed completely, as part of the Compton–scattered γ-ray fluence encounter the detector via its entrance window. The use of a 3-cm large Pb collimator in the BGO-shielded case probably furthermore reduces the type b background and may be the reason for the small decrease of the low-energy RD compared to the Pb-shielded case used without collimator. In summary, the fast-timing results confirm the interpretations of the results presented in Section 2.1. Further test measurements or simulations, also by using unshielded adjacent LaBr3(Ce) detectors for active Compton suppression, may be

48

J.-M. Régis et al. / Nuclear Instruments and Methods in Physics Research A 811 (2016) 42–48

energy background includes type b background and is largely delayed (100–150 ps at 150 keV) relative to the FEP time response dependent on shielding and the set-up geometry. By reducing the low-energy background of type b, the residual difference (delay) between the remaining background and the FEP time responses is reduced by 50–70 ps at 150 keV.

350

without shield

300

RD [ps]

250 200

with 5−mm Pb shield

150 100

Acknowledgment

50 0

with active BGO shield 50

100

150

200

250

300

350

400

450

E γ [keV] Fig. 9. The residual difference between the FEP time response and the time response of low-energy background events (data points from the 60Co source with Eref: ¼ 1333 keV) obtained using different shielding conditions.

performed in order to provide the most effective background suppression and the highest γ-γ coincidence efficiency for a given set-up geometry.

3. Conclusion Extended γ-γ coincidence and γ-γ fast-timing measurements have been performed using cylindrical 1:5 in:
1:5 in: LaBr3(Ce) scintillator detectors with different shielding materials installed within the large γ-ray spectrometer HORUS of the IKP in Cologne, Germany. The studies are related to the three components of timecorrelated background obtained in γ-ray spectroscopy experiments. Inter-detector Compton scattering which create the crosstalk events (type a background) can be nearly completely suppressed by using 5-mm thick lead shields per detector or either massive BGO shields to protect the sides of the detectors for being hit by Compton-scattered γ rays which are generated from neighboring (adjacent) detectors. Such shielding also reduce the detection of low-energy Compton-scattered γ rays which are generated from the materials around the detector (type b background), such as the shields of other detectors. However, a passive lead shield only slightly improves the peak-to-total value by a factor smaller than 1.1. The typical and dominant Compton background (type c) can only be reduced by using active scintillator detectors as BGO shields or unshielded adjacent LaBr3(Ce) detectors. The use of active BGO shielding as presented in this paper improves the peak-to-total value by a factor of 1.66(5). A larger improvement factor is expected by an optimization of the BGO– LaBr3(Ce) geometry. In any case, the time response of the background generated by prompt mono-energetic γ rays differs from the FEP time response which represents the physical zero time of the γ-γ fast-timing set-up. While the Compton background has a slightly faster time response independent of shielding, the low-

This work was financially supported by the German Bundesministerium für Bildung und Forschung (BMBF) under Contracts 05P15PKFNA and 05P12PKNUF, the latter within the Nuclear Physics Network (NuPNET).

References [1] E.R. White, et al., Physical Review C 76 (2007) 057303. [2] J.-M. Régis, et al., Nuclear Instruments and Methods in Physics Research Section A 606 (2009) 466. [3] L. Bettermann, et al., Physical Review C 82 (2010) 044310. [4] J.-M. Régis, et al., Physical Review C 90 (2014) 067301. [5] K.S. Shah, et al., IEEE Transactions on Nuclear Science NS–50 (2003) 2410. [6] M. Moszyński, et al., Nuclear Instruments and Methods in Physics Research Section A 567 (2007) 31. [7] L.G. Hyman, Review of Scientific Instruments 36 (1965) 56. [8] J. Jolie, et al., Nuclear Physics A 934 (2015) 1. [9] M. Rudigier, et al., Physical Review C 91 (2015) 044301. [10] K.E.G. Löbner, in: W.D. Hamilton (Ed.), The Electromagnetic Interaction in Nuclear Spectroscopy, North-Holland, New York, 1975, p. 173. [11] Z. Bay, Physics Review 77 (1950) 419. [12] J.-M. Régis, et al., Nuclear Instruments and Methods in Physics Research Section A 684 (2012) 36. [13] H. Mach, et al., Nuclear Physics A 523 (1991) 197. [14] C. Michel, et al., Nuclear Instruments and Methods in Physics Research Section A 251 (1986) 119. [15] M. Moszyński, et al., Nuclear Instruments and Methods in Physics Research Section A 280 (1989) 73. [16] O.J. Roberts, et al., Nuclear Instruments and Methods in Physics Research Section A 748 (2014) 91. [17] L.M. Fraile, Technical Design Report for the DESPEC Fast Timing Array, http:// www.fair-center.eu/fileadmin/fair/publications_exp/TDR_HISPEC_DESPEC_ FATIMA_public.pdf, 2015. [18] B. Rubio, et al., International Journal of Modern Physics E 15 (2006) 1979. [19] J.-M. Régis, G.S. Simpson, et al., Nuclear Instruments and Methods in Physics Research Section A 763 (2014) 210. [20] J.-M. Régis, et al., Nuclear Instruments and Methods in Physics Research Section A 726 (2013) 191. [21] Y. Yosshizawa, Y. Iwata, Y. Iinuma, Nuclear Instruments and Methods 174 (1980) 133. [22] J.H. Hubbell, S.M. Seltzer, Radiation Research 136 (1993) 147. [23] J. Rzadkiewicz, et al., Nuclear Instruments and Methods in Physics Research Section B 235 (2005) 110. [24] J.-M. Régis, et al., Nuclear Instruments and Methods in Physics Research Section A 622 (2010) 83. [25] M.J. Martin, Nuclear Data Sheets 114 (2013) 1497. [26] M. Moszyński, H. Mach, Nuclear Instruments and Methods in Physics Research Section A 277 (1989) 407.