Performance of a prototype active veto system using liquid scintillator for a dark matter search experiment

1 MB·

by  J. S. Park

Performance of a prototype active veto system using liquid scintillator for a dark
matter search experiment
J.S. Park,1 P. Adhikari,2 G. Adhikari,2 S.Y. Oh,2 N.Y. Kim,1
Y.D. Kim,1,2 C. Ha,1 K.S. Park,1 H.S. Lee,1 and E.J. Jeon1
1Center for Underground Physics, Institute for Basic Science (IBS), Daejon 34047, Korea
2Department of Physics, Sejong University, Seoul 05006, Korea
WereporttheperformanceofanactivevetosystemusingaliquidscintillatorwithNaI(Tl)crystals
for use in a dark matter search experiment. When a NaI(Tl) crystal is immersed in the prototype
detector, thedetector tags 48% of the internal 40K background in the 0–10 keV energy region. We
7 also determined the tagging efficiency for events at 6–20 keV as 26.5 ± 1.7% of the total events,
1 which corresponds to 0.76 ± 0.04 events/keV/kg/day. According to a simulation, approximately
0
60%ofthebackgroundeventsfromU,Th,andKradioisotopesinphotomultipliertubesaretagged
2
atenergiesof0–10keV.Fullshieldingwitha40-cm-thickliquidscintillatorcanincreasethetagging
n efficiency for both theinternal 40K and external background to approximately 80%.
a
J PACSnumbers: 29.40.Mc,85.60.Ha

7
1
I. INTRODUCTION as photomultiplier tubes (PMTs), encapsulating copper

] cases,andshielding materials,arenotperfectly rejected.
M
Weconsideranewshieldingstructureincludinganactive
Numerous astronomical observations have indicated
I veto system that surrounds the NaI(Tl) crystals with a
thatmostofthematterintheuniverseisinvisible,exotic,
h. and nonrelativistic dark matter [1–4]. However, despite liquid scintillator (LS) in order to tag the internal 40K
p background as well as reduce the external background
considerable experimental effort, the nature of the dark
- contributions. We build a prototype active veto detec-
o matter remains unknown. The weakly interacting mas-
tor to verify the performance of the full shielding before
r siveparticle(WIMP)isacandidatethatconformstothe
t finishing its final design. In this article, we describe the
s stringent standards for particle dark matter, and it is
a prototype active veto detector from construction to per-
supported by astronomicalobservations as well as parti-
[ formance. In addition, a shielding design for a future
clephysicstheoriesthatextendthestandardmodel[5,6].
NaI(Tl) experiment is presented.
1 Numerous experiments have searched for WIMPs by di-
v rectly detecting nuclei recoiling from WIMP–nucleus in-
4
teractions [7, 8].
1
5 It is essential to reduce the background level as much II. PROTOTYPE LIQUID SCINTILLATOR
4 as possible for dark matter searches because of the ex- VETO SYSTEM
0 tremely low cross-section of the WIMP–nucleon interac-
1. tion. The dominant background contributions produced Weproducealinearalkylbenzene(LAB)-basedLS[16,
0 by external sources, such as decaying radioisotopes in 17] consistingof 3 g of 2,5-diphenyloxazole(PPO)/ℓand
7 surroundingmaterialsandcosmogenicmuons,canbere- 30 mg of 1,4-bis[2-methylstyryl]benzene (bis-MSB)/ℓ for
1 duced by shielding the detector in a deep underground a high light yield and long attenuation length. We use
v: laboratoryusing heavy materials consisting of lead, cop- a water extraction method [18] and nitrogen purging
i per, and polyethylene layers. However, remnants of this to remove possible contamination from natural radioiso-
X
contribution,aswellasthebackgroundcontributionfrom topesintheLS.Themeasuredcontaminationlevelsfrom
r materials close to the detector, still exist. In addition to 238U and 232Th in the LS are less than 7 and 4 ppt, re-
a
passive shielding, it is beneficial to have an active veto spectively. These contaminants provide negligible back-
system that can reduce the background level by tagging ground to the NaI(Tl) crystal.
neutron and γ events [9–11] from external sources. An To take advantage of a low-background facility, the
activevetosystemcanalsoreducetheleveloftheinternal prototype detector is designed to fit in an existing shield
background by tagging the escaping γ-rays. forthe KIMS-CsI experimentatYangyangUnderground
The Korea Invisible Mass Search (KIMS) Collabora- Laboratory(Y2L)[19–21]. AsshowninFig.1,theproto-
tion is developing ultralow-background NaI(Tl) crystal type detector is relatively narrow because of the limited
detectors [12–14] to verify an observation of an annual space in the shielding. The maximum available space
modulation in the detection rate of an array of NaI(Tl) surroundedbythelayersof5cmpolyethyleneand10cm
crystals[15]thatcanbeinterpretedastheresultofinter- copperaccommodatesa110cm(L)×44cm(H)×25cm
action with WIMPs. Even though the internal radioiso- (W)rectangularacrylicbox,includinganadditional6cm
topesoftheNaI(Tl)crystalswereidentifiedandmightbe of copper on the outside. The acrylic structure can con-
further reduced by future development, the background tain 140 ℓ of the LAB-based LS. A separate acrylic sup-
contributionsoriginatingfromdetectorcomponents,such port is installed to hold a NaI(Tl) crystal in the middle
2
(a) (b) (c)
FIG. 1: (a) Design of prototypeLS vetodetector and installation structureconsisting of (b) acrylic box and (c) copper box.
of the box. The crystalis wrapped in a Teflon sheet and NaI andLS,are shownin Fig.2. Low-energycalibration
encapsulated in oxygen-free high-conductivity (OFHC) oftheNaI(Tl)crystalwasdoneusingthe59.54keVγ-line
copperwithtwosyntheticquartzwindowsthatallowcou- froma 241Am source. An extrapolationofthe 59.54keV
pling to the PMTs. The PMTs are encased in another linetolowenergywascheckedusingthe3keVX-rayline
OFHC copper housing to decouple light signals between from internal 40K, which agreed within 5%.
the crystal and the LS. Figure 3 (a) and (b) show the energy spectra of the
The acrylic box is wrapped in Tyvek reflective LS veto detector in various energy ranges. Figure 3 (c)
films [see Fig. 1 (b)] to increase the light collection effi- showsthe energyofthe LSvetodetectorasafunctionof
ciencyoftheLS.ScintillationsignalsproducedbytheLS the time difference between the LS veto signal and the
aremeasuredbyten3inchHamamatsulow-radioactivity NaI(Tl) signal (∆t). With the requirement of 10 keV of
and high-quantum-efficiency PMTs (R12669SEL). The energy from the LS, one can remove dominant random
PMTsignalsareamplifiedbyafactorof30;theamplified events so that the crystal-related background coinciding
signals are digitized by 64 MHz flash analog-to-digital with the LS veto detector can be efficiently tagged. In
converters(FADCs). Thetotalrecordedtimewindowfor addition, werequire ∆t to be −30ns<∆t<70ns. The
an event is 8 µs. The energy scale of the LS veto detec- event rate of the LS veto detector passing the criteria is
toriscalibratedusingtheComptonedgesofγ-rayevents only 0.12 Hz, which makes the veto-induced dead time
froma 60Co (1173keVand1333keVγ lines) sourceand negligible. We can calculate a fake event rate of 0.1%
extrapolated to low energy. The energy calibration is in the coincident event region from the events occurring
checkedwith57Co(122.1keVγ line)and137Cs(661keV outside the coincident region (−1000 ns < ∆t < −30 ns
γ line) sources, which show good agreement. or 70 ns<∆t<1000 ns).
After the structure is installed, an already tested We count the event rate of the NaI(Tl) detector with
NaI(Tl)crystal(NaI-002)[12]isplacedinthecenter,and andwithouttheLSvetorequirement,asshowninFig.4.
the LS is fitted into the acrylic box. NaI-002 was grown The detector realizes a significant reduction in the back-
from selected ultrapure NaI powder by the Alpha Spec- groundrate by requiringthat no signalappear inthe LS
traCompany. Thecrystal(9.2kgmass)hasacylindrical veto detector. The tagging efficiency is 26.5 ± 1.7% in
shapewithadiameterof4.2inchandalengthof11inch. the 6–20 keV energy region and 63 ± 1% in the 100–
Two R12669SEL PMTs are attached at each end of the 1500 keV energy region. The tagged event rate between
crystal. The PMT signals from the crystal are split and 6 and 20 keV is 0.76 ± 0.04 events/keV/kg/day(dru).
amplified by factors of 30 and 2. The amplified signals We simulate various background sources with the
are digitized by 400 and 64 MHz FADCs with readout Geant4 [22] simulation package to obtain the veto effi-
time windowsof10 and8µs. A triggerconditionfor the ciency of the prototype. We first consider the internal
NaI(Tl)crystalis the presenceofone ormorephotoelec- 40K decay,which is one of the strongestbackgroundsfor
trons in each PMT within a 200 ns time window. If the darkmattersearcheswithNaI(Tl)crystals[23–26]. This
NaI(Tl)crystalistriggered,allthesignalsfromboththe decay generates an X-ray at approximately 3 keV with
NaI(Tl) crystal and the LS veto detector are recorded. a 1460 keV γ-ray. If the accompanying 1460 keV γ-ray
The typical event shapes, which show signals from both escapes from the crystal, the event consists of a single
3
550 450
500
nit) nit)400
U U
y 450 y
r r
a a
rbitr400 rbitr350
A A
( (
unt 350 unt 300
o o
c c
C C
D300 D
A A250
250
200
0 2000 4000 6000 8000 0 2000 4000 6000 8000
Time (ns) Time (ns)
NaI crystal
1100 2500 1600 2000 2200
Unit)1090000 Unit)2000 Unit)1400 Unit)11680000 Unit)12800000
ADC count (Arbitrary 456780000000000 ADC count (Arbitrary 11055000000 ADC count (Arbitrary 11026800000000 ADC count (Arbitrary 111024680000000000 ADC count (Arbitrary 1111024668000000000000
300 400 400 400
200
1000 1500 2000 2500 3000 1000 1500 2000 2500 3000 1000 1500 2000 2500 3000 1000 1500 2000 2500 3000 1000 1500 2000 2500 3000
Time (ns) Time (ns) Time (ns) Time (ns) Time (ns)
LS veto(left side)
500 700
1400 450 600 700 800
ADC count (Arbitrary Unit)11026800000000 ADC count (Arbitrary Unit)122334505050000000 ADC count (Arbitrary Unit)234500000000 ADC count (Arbitrary Unit)234560000000000 ADC count (Arbitrary Unit)234567000000000000
100
400 100 100 100
50
1000 1500 2000 2500 3000 1000 1500 2000 2500 3000 1000 1500 2000 2500 3000 10000 1500 2000 2500 3000 10000 1500 2000 2500 3000
Time (ns) Time (ns) Time (ns) Time (ns) Time (ns)
LS veto(right side)
FIG.2: Pulse shapesfrom PMTs fortypicalcoincident eventsof theNaIcrystal and LSvetodetectorwith measured energies
of 37 and 850 keV,respectively.
3 keV hit, which mimics a dark matter signal. However, ting assuming a flat background. The measured rate is
if we tag the escaping 1460 keV γ-ray with the LS veto 2.10±0.32counts/kg/day,whichisfoundtocorrespond
detector, we can effectively reduce the low-energy con- to43.7±6.4ppbKcontaminationbyacomparisonwith
tribution of the 3 keV X-ray in the NaI(Tl) crystal [27]. the simulation. This value is consistent with a previous
The veto efficiency for energies between 0 and 10 keV is measurement (49.3 ± 2.4 ppb) of the same crystal in a
about 48% from the internal 40K simulation. We esti- different veto system in which neighboring CsI(Tl) crys-
mate the K contamination of the NaI(Tl) crystal using tals are used to tag the 40K events [12].
the events vetoed by the LS veto detector to verify our
Other than the internal background of the NaI(Tl)
simulation. The filled (blue) circles in Fig. 4 (b) show
crystal, one of the main backgroundcontributions arises
the low-energy spectrum of the NaI(Tl) crystal that is from the natural radioisotopes, 238U, 232Th, and 40K, in
taggedbytheLSvetodetector. Inthisfigure,thetagged
the PMTs. We simulate these isotopes from the photo-
3 keV X-ray events can be extracted by Gaussian fit-
cathode as well as the glass envelope and estimate the
4
105 105 100
V V 90
y/ke104 y/ke104 80
da da 70
of events/110023 of events/110023 E (keV)LS456000
er er 30
umb 10 umb 10 20
N N 10
1
0 500 1000 1500 2000 2500 3000 3500 4000 10 10 20 30 40 50 60 70 80 90 100 −00.3 −0.2 −0.1 0 0.1 0.2 0.3
Energy(keV) Energy(keV) ∆ t(µs)
(a) (b) (c)
FIG. 3: Energy spectra of LS veto detector. (a) Energy spectrum for γ background region of 0–4 MeV. (b) Zoomed energy
spectruminlow-energyregion(0–100 keV)with10keVenergythreshold(redline)forcoincidencecondition. (c)EnergyofLS
vetodetector versus time differencebetween LS veto signal and NaI(Tl) signal (∆t) for coincidence condition (red line).
18 14
Total Events Total Events
16
12
Coincident Events Coincident Events
14
V NaI only signal V 10 NaI only signal
e e
k 12 k
y/ y/
da 10 da 8
g/ g/
k k
s/ 8 s/ 6
nt nt
ou 6 ou
C C 4
4
2
2
0 0
100 200 300 400 500 600 700 800 900 1000 0 2 4 6 8 10 12 14 16 18 20
Energy(keV) Energy(keV)
(a) High Energy (b) Low Energy
FIG.4: BackgroundenergyspectraoftheNaI(Tl)crystalintheLSvetosystem. Vetoedevents(bluesymbols)havehitsonthe
LS vetodetectorwith energies greater than10 keV,whereas hitson NaI only (red) havenosuch hitson theLS vetodetector.
The total hits (black) represents all events(thesum of eventsin both categories).
veto efficiencies in the 0–10 keV measured energy range, efficiency. The optimization is based on the simulation
which are 66%, 56%, and 63% for 238U, 232Th, and 40K, of the internal 40K background. We consider a simple
respectively. Other background contributions from out- cubic structureofthe LScontainerinwhichone NaI(Tl)
side the LS veto system are tagged at a level that is at crystal (same geometry as NaI-002) is immersed in the
least similar to that of the PMTs. center. The thicknesses of the NaI(Tl) crystal ends and
the LS container wall are the same and vary from 10 to
150 cm. Figure 5 shows the veto efficiency as a function
ofthe LS thickness. We consideran LSwith a minimum
III. DESIGN OF A NEW LS VETO SYSTEM
thickness of 40 cm for the final design by considering
FOR A 200 KG NAI EXPERIMENT
the size of the shielding as well as the veto efficiency.
The veto efficiency measured by the prototype detector
Thevetoefficiencyofthe prototypedetectorislimited is consistent with the simulated efficiency for a 17-cm-
by the narrow thickness of the container necessitated by thick container, as shown in Fig. 5.
thesmallavailablespaceintheexistingshieldingatY2L.
We plan to construct a dedicated shielding structure for AnLSvetodetector40cmthickcantagapproximately
the 200 kg array experiment using NaI(Tl) crystals [14]. 75%oftheinternal40Keventsat0–10keV.Theuntagged
Inthisshieldingstructure,wecanconsideramuchbigger 25%of events are due to 1460 keVγ-raysescaping with-
container for the LS veto system to increase the veto out any heat in the LS veto detector. The background
5
the 25%ofuntaggedeventsrepresentsareductionofap-
100 proximatelytwotimes comparedtothe 52%ofuntagged
90 events with the prototype detector.
%) 80 We finalized the design of the new LS veto system, as
y ( 70 shown in Fig. 6, considering the discussion above. The
c
en 60 minimum thickness of the LS from each end of the crys-
ci
effi 50 tal is 40 cm. The average tagging efficiency of the LS
ng 40 Prototype vetosystemandtheothercrystalsfortheinternal40Kis
ggi 30
a estimated to be 81%. The veto efficiencies for the PMTs
T
20
arepredictedtobeapproximately80%forU,Th,andK.
10
00 20 40 60 80 100 120 140 160
Thickness of LS veto (cm)
IV. CONCLUSION
FIG. 5: Veto efficiency of internal 40K as a function of the
thickness of theNaI(Tl) crystal and the LS container for en-
We constructed a prototype LS veto detector to de-
ergies between 0and 10 keVobtained from a simple geomet-
termine the feasibility of tagging the background of a
ric simulation. Square symbol represents the veto efficiency
NaI(Tl) crystal detector for a dark matter search ex-
of the prototypedetector from the simulation.
periment. The tagged background rate of the NaI(Tl)
crystal is 0.76 ± 0.04 dru at 6–20 keV. The tagging ef-
ficiency of the 3 keV X-rays from internal 40K is about
48% from the simulation. The measured event rates of
3keVX-rayswiththerequiredLSvetosignalareingood
agreement with a previous measurement. A new shield
with an appropriate LS thickness will increase this rate
to approximately 80%.
Acknowledgments
FIG. 6: Drawing of an LS veto system for a 200 kg NaI(Tl)
We thank the Korea Hydro and Nuclear Power
crystal array. Nine 5 in. Hamamatsu PMTs will beattached
(KHNP) Company for providing the underground lab-
to each end of the Cu container.
oratory space at Yangyang. This research was funded
by Grant No. IBS-R016-A1 and was supported by the
Basic Science Research Program through the National
contributionfromtheinternal40Kwiththenewvetode- ResearchFoundationofKorea(NRF)fundedbytheMin-
signcanbeeffectivelyreducedbyafactoroftwobecause istry of Education (NRF-2011-35B-C00007).
[1] D.Clowe et al.,Astrophys.J. 648 (2006) L109. [12] K. W. Kim et al.,Astropart. Phys. 62 (2015) 249.
[2] E.Komatsuetal.(WMAPCollaboration),Astrophys.J. [13] H. S. Leeet al.,J. High Energy Phys.08 (2015) 093.
Suppl.192 (2011) 18. [14] P. Adhikariet al., Eur.Phys. J. C 76 (2016) 185.
[3] P. A. R. Ade et al. (Planck Collaboration), Astron. As- [15] R.Bernabeiet al.(DAMACollaboration), Eur.Phys.J.
trophys.571 (2014) A16. C 73 (2013) 2648.
[4] G. Bertone and D. Hooper, arXiv:1605.04909. [16] Y.Dinget al.,Nucl.Instrum.MethodsPhys.Res.,Sect.
[5] B. W. Lee and S. Weinberg, Phys. Rev. Lett. 39 (1977) A 522 (2007) 439.
165. [17] J. S. Park et al., Nucl. Instrum. Methods Phys. Res.,
[6] G. Jungman, A. Kamionkowski, and K. Griest, Phys. Sect. A 707 (2013) 45.
Rep.267 (1996) 195. [18] M.Yeh,Nucl.Instrum.MethodsPhys.Res.,Sect.A578
[7] P.Cushman et al., arXiv:1310.8327. (2007) 329.
[8] T. M. Undagoitia and L. Rauch, J. Phys. G 43 (2016) [19] H.S.Leeetal.,(KIMSCollaboration),Phys.Lett.B633
013001. (2006) 201.
[9] A. Wright et al., Nucl. Instrum. Methods Phys. Res., [20] H.S.Leeetal.,Nucl.Instrum.MethodsPhys.Res.,Sect.
Sect.A 644 (2011) 18. A 517 (2007) 644.
[10] R. Calkins and B. Loer, AIP Conf. Proc. 1672 (2015) [21] S.C.Kimetal.,(KIMSCollaboration),Phys.Rev.Lett.
140002. 108 (2012) 181301.
[11] E. Shields, J. Xu, and F. Calaprice, Phys. Procedia 61 [22] S. Agostinelli et al., Nucl. Instrumen. Methods Phys.
(2015) 169. Res., Sect. A 506 (2003) 250; J. Allison et al., IEEE
6
Trans. Nucl. Sci. 53 (2006) 270. E. Barbosa deSouza et al., arXiv:1602.05939.
[23] J. Amare et al., Nucl. Instrumen. Methods Phys. Res., [25] K. Fushimi et al.,arXiv:1407.3542.
Sect. A 742 (2014) 187; J. Amare et al., Phys. Procedia [26] J. Xu et al.,AIPConf. Proc. 1672 (2015) 040001.
61 (2015) 157. [27] J. Amare et al.,Eur. Phys. J. C 76 (2016) 429.
[24] J. Cherwinka et al., Phys. Rev. D 90 (2014) 092005;