aa r X i v : . [ a s t r o - ph . I M ] N ov Tests on NaI(Tl) crystals for WIMP search at the Yangyang UndergroundLaboratory
K.W. Kim b , W.G. Kang a , S.Y. Oh e , P. Adhikari e , J.H. So a , N.Y. Kim a , H.S. Lee c, ∗ , S. Choi b , I.S. Hahn d ,E.J. Jeon a , H.W. Joo b , B.H. Kim b , H.J. Kim f , Y.D. Kim a,e, ∗ , Y.H. Kim a,g , J.K. Lee b , D.S. Leonard h ,J. Li a , S.L. Olsen b , H.S. Park g a Center for Underground Physics, Institute for Basic Science (IBS), Daejon 305-811, Korea b Department of Physics and Astronomy, Seoul National University, Seoul 151-747, Korea c Department of Physics, Ewha Womans University, Seoul 120-750, Korea d Department of Science Education, Ewha Womans University, Seoul 120-750, Korea e Department of Physics, Sejong University, Seoul 143-747, Korea f Department of Physics, Kyungpook National University, Daegu 702-701, Korea g Korea Research Institute of Standards and Science, Daejon 205-340, Korea h Department of Physics, University of Seoul, Seoul 130-743, Korea
Abstract
Among the direct searches for WIMP-type dark matter, the DAMA experiment is unique in that it hasconsistently reported a positive signal for an annual-modulation signal with a large (9.3 σ ) statistical sig-nificance. This result is controversial because if it is interpreted as a signature for WIMP interactions, itconflicts with other direct search experiments that report null signals in the regions of parameter space thatare allowed by the DAMA observation. This necessitates an independent verification of the origin of theobserved modulation signal using the same technique as that employed by the DAMA experiment, namelylow-background NaI(Tl) crystal detectors. Here, we report first results of a program of NaI(Tl) crystalmeasurements at the Yangyang Underground Laboratory aimed at producing NaI(Tl) crystal detectors withlower background levels and higher light yields than those used for the DAMA measurements. Keywords:
Dark Matter, WIMP, KIMS, NaI(Tl) crystal
PACS:
1. Introduction
Numerous astronomical observations have led tothe conclusion that most of the matter in the Uni-verse is invisible, exotic, and nonrelativistic darkmatter [1, 2]. However, in spite of a considerableamount of experimental effort, the nature of thedark matter remains unknown. Weakly interactingmassive particles (WIMPs) are one of the most at-tractive candidates for dark matter particles [3, 4].In supersymmetric models for beyond the standardmodel physics, the lightest supersymmetric particle(LSP) is a natural candidate for WIMP-type dark ∗ Corresponding authors, Tel:+82-2-3277-3413; fax:+82-2-3277-2372
Email addresses: [email protected] (H.S. Lee), [email protected] (Y.D. Kim ) matter. A number of experiments have made directsearches for a WIMP component of our Galaxy bylooking for evidence for WIMP-nucleus scatteringby detecting the recoiling nucleus in ultra-sensitivelow-background detectors [5, 6].The DAMA experiment searches for an annualmodulation in the detection rate of nuclear recoilsin an array of ultra-low-background NaI(Tl) crys-tals caused by the Earth’s orbital motion throughour Galaxy’s dark-matter halo [7, 8, 9]. This ex-periment, which has been operating for over 15years, has consistently reported a positive signalfor an annual modulation with a phase that is con-sistent with expectations for motion of the Earthrelative to the Galactic rest frame. The statisti-cal significance of the DAMA annual modulationsignal has now reacehed 9 . σ . Other experiments,including CoGeNT [10, 11, 12], CRESST [13], and Preprint submitted to Elsevier August 23, 2018
DMS [14], have also reported signals that couldbe interpreted as being possibly due to WIMP in-teractions. However, these signals have marginalsignificance at the < σ confidence level and someinconsistencies with null results using similar detec-tors that were already reported [15, 16, 17]. Never-theless, these marginal signals have motivated thesame experimental groups and, in some cases, in-dependent groups to devise experiments to checkthese signals using similar techniques but withhigher sensitivity. In contrast, to date, no inde-pendent verification of the DAMA signal in an ex-periment using the same technique has been done.The DAMA signal, in particular its interpre-tation as being due to WIMP-nucleus scattering,has been the subject of a continuing debate thatstarted with the first DAMA report 15 years ago.This is because the WIMP-nucleon cross sectionsinferred from the DAMA modulation are in con-flict with limits from other experiments that di-rectly measure the total rate of nuclear recoils,such as XENON100 [18], LUX [19], and Super-CDMS [20]. However, room remains for explainingall of these experimental results without conflict interms of nontrivial systematic differences in detec-tor responses [21, 22] and the commonly used astro-nomical model for the WIMP distribution [23]. Anunambiguous verification of the DAMA signal byan independent experiment using similar NaI(Tl)crystals is mandatory.Reproducing the DAMA measurement re-quires the independent development of ultra-low-background NaI(Tl) crystals. This is becausethe crystal-growing company that supplied theDAMA NaI(Tl) crystals will not produce the samecrystals for other experimental groups. Recently,the ANAIS group has been developing ultra-low-background NaI(Tl) crystals with the goal of repro-ducing the DAMA result [24, 25] and the DM-Icegroup reports background measurements of NaI(Tl)crystals [26]. However, to date, no experimentalgroup has produced NaI(Tl) crystals with back-ground levels at or below than those used in theDAMA experiment.The Korea Invisible Mass Search (KIMS) Collab-oration has been performing direct WIMP searchesusing ultra-low-background CsI(Tl) crystal detec-tors, which are similar experimental devices to theNaI(Tl) crystals used in the DAMA experiment. Crismatec, France
This implementation required extensive researchand development aimed at identifying and reducinginternal backgrounds in the CsI crystals [27, 28].Null results from KIMS reject WIMP-Iodine nu-clei interactions as the source of the DAMA signalwith very little model dependence [29, 30]. How-ever, because the DAMA results can be interpretedas being primarily due to WIMP-sodium nuclei in-teractions, which, for example, would be the casefor low-mass WIMPs, it remains necessary to con-firm the DAMA observations with NaI(Tl) crys-tal detectors. This motivated a program to de-velop ultra-low-background NaI(Tl) crystal detec-tors with lower background levels and and higherlight yields (and, thus, a lower energy threshold)than those of the DAMA experiment in order toidentify unambiguously the origin of the DAMAmodulation signature.
2. Experimental Setup
To evaluate the NaI(Tl) crystals, we use theexperimental setup that was used for the KIMSCsI(Tl) detector measurements at the YangyangUnderground Laboratory (Y2L), which has a 700 mearth overburden (2400 m water equivalent). Thisincludes a 12-module array of CsI(Tl) detectors(total mass of 103.4 kg) inside a shield that con-sists, from inside out, of 10 cm of copper, 5 cmof polyethylene, 15 cm of lead, and 30 cm ofliquid-scintillator-loaded mineral oil to stop exter-nal neutrons, gamma rays, and veto cosmic raymuons. Each detection module consists of a low-background CsI(Tl) crystal (8 × ×
30 cm ) with aphotomultiplier tube (PMT) mounted at each end.Two NaI(Tl) crystals were mounted inside theCsI(Tl) detector array as shown in Fig. 1. InNovember 2013, we installed the first NaI(Tl) crys-tal, NaI-001, in the CsI array; the second NaI(Tl)crystal, NaI-002, was added in February 2014, soonafter that crystal was delivered and assembled. The two NaI(Tl) crystals were produced by theAlpha Spectra Company. The crystals were grownfrom NaI powder that was not of the highest attain-able purity; the initial purification was carried outby Alpha Spectra. The crystals have a cylindricalshape and were cut from a 32 inch ingot that was igure 1: Schematic test setup for two NaI(Tl) crystals (cir-cles) with 12 CsI(Tl) crystals (squares). grown by the Kyropoulos method. The detailedsizes of the crystals are listed in Table 1. After thecrystal surfaces were polished they were wrappedwith a Teflon reflector and inserted into an oxigen-free electronic (OFE) copper cylinder and encap-sulated in a nitrogen gas environment. There is a12.7-mm-thick quartz-plate window at each end ofthe cylinder, with optical grease between the crys-tal and the quartz windows. A 3 ′′ PMT is mountedat each end of the cylinder.
Each NaI crystal PMT signal was split and am-plified by factors of 30 and 2; the amplified signalswere digitized by 400- and 64-MHz flash analog-to-digital converters (FADC), respectively. The cor-responding amplification factors for the CsI crystalPMTs were ×
100 and ×
10. The total recorded timewindow for an event was 40 µ s, of which 5 µ s is ana-lyzed for the NaI(Tl) crystals and 25 µ s is analyzedfor the CsI(Tl) crystals, reflecting the different de-cay times of the two materials.Figure 2 shows a schematic diagram of the de-tector setup. The trigger condition for the CsI(Tl)crystals is two or more photoelectrons (PEs) in eachPMT within a 2- µ s time window. The NaI(Tl) crys-tals have a reduced PE requirement of one PE ineach PMTs within a 200 ns window in order to havea minimal trigger bias and a low energy threshold.The muon rate at Y2L is (7 . ± . × − / s/cm /sr [31]. Since the KIMS muon vetosystem was under maintenance, no muon veto is ap-plied to the data reported here. Approximately 3 ∼ < Figure 2: Schematic electronics diagram of the KIMS NaIexperiment. keV during a few second time interval after a muonpasses through the crystal because of a long decay-time component of the NaI(Tl) scintillation process.However, most of these events are rejected by theselection requirements discussed in Section 6 and,so, cosmic ray muon-related events are negligible inthe data reported here.
3. Signal Calibrations
The energy calibration of the NaI(Tl) crystalswas done with a
Am source. The detector hadone hole of 10 mm in diameter covered by 127- µ m-thick aluminum foil in the center of the encapsulat-ing Cu container. The source was located in front ofthe hole. The CsI(Tl) crystal calibration procedureis described in Refs. [29, 30]. Two different types of low-background PMTswere tested: a metal-packed R11065 and a glass-packed R12669, both manufactured by HamamatsuPhotonics. Table 2 shows the specifications foreach PMT. The R12669 PMT is a modified ver-sion of the R6233 Super Bialkali (SBA) PMT usedby DAMA in its recent upgrade [32]. The PMTswe used were selected for their high quantum effi-ciency. The radioactivity levels of the PMTs weremeasured at underground with a HPGe detector.As expected, the metal PMT has a lower radioac-tivity level than the glass PMT as listed in Table 2.However, the 3 ′′ R12669 PMTs suffer from seriousnonlinearity when the signal height is > able 1: Specifications of the NaI(Tl) crystals used in this study. The last two columns are the dates the crystals were grownand transported to Y2L. “Transp” indicates the means by which the crystals were transported from the United State to Korea. Crystal Size Mass Transp T (Growth) T (underground)NaI-001 5 ′′ (D) × ′′ (L) 8.26 Air 2011.9 2013.9NaI-002 4.2 ′′ (D) × ′′ (L) 9.2 Sea 2013.4 2014.1 Table 2: Specifications for PMTs tested in this study. Radioactivity levels measured with a HPGe detector at Y2L. SEL means“selected for high quantum efficiency.”
PMT R12669SEL b R11065SEL b Photocathode SBA BialkaliWindow Borosilicate QuartzBody Borosilicate KovarStem Glass GlassGain (HV) 1 × × Radioactivity a U( Bi) 25 ± ± Ac) 12 ± ± K) 58 ± ± Ra 60 ±
10 5 ± Tl 4 ± For low-energy events, the 400-MHz FADC wave-forms were analyzed to identify clusters of singlePEs (SPEs) [33]. The charge distribution of a SPE,obtained by identifying isolated clusters at the de-cay tail of the signal (2–5 µ s after the signal start)in order to suppress multiple PE clusters, is shownin Fig. 3. Cluster Charge (pC) N u m be r o f c l u s t e r s Figure 3: The SPE charge distribution measured with theNaI-001 crystal readout by an R12669 PMT. This distribu-tion is produced with a
Am calibration source in a timewindow that is 2–5 µ s after the signal start to reduce PEpileup. Figure 4 shows the distribution of the number ofPEs obtained using an
Am source with the NaI-001 and NaI-002 detectors read out with differenttypes of PMTs. The mean number of PEs detectedwith the R12669 PMTs is 22% greater than thatdetected by the R11065 PMTs. This is consistentwith the ANAIS test [24] and reflects the higherquantum efficiency of the SBA photocathode.
Number of PEs N u m be r o f e v en t s NaI-001/R11065NaI-002/R12669
Figure 4: The measured number of PEs with the
Amsource calibration for the NaI-001 and NaI-002 crystals readout by two different PMT types.
We study the scintillation decay time of theNaI(Tl) crystals from the time response to signals4 able 3: Backgrounds from the internal radioactive contami-nants in the NaI(Tl) crystals. The units for all the values aremBq/kg. For “Total alphas,” each alpha particle is countedas one decay.
Radionuclei NaI-001 NaI-002
U (by
Bi) < < Th (by
Po) < ± K 1.25 ± ± Pb 3.28 ± ± ± ± γ -ray source. Figure 5shows the integrated signal shapes for 59.54 keV γ -rays from a Am source. The data are fitted withtwo exponentials with time constants of 0.22 and1.17 µ s. The fast component accounts for ∼
83% ofthe total light yield. s) µ Time ( A cc u m u l a t e d c h a r g e ( a r b i t r a r y un i t s ) Figure 5: Scintillation decay time spectrum of the NaI-001crystal obtained with
Am source data and fitted with twoexponential functions.
4. Natural background
To produce ultra-low-background crystals, con-tamination from internal natural radioisotopes hasto be reduced. Table 3 shows the measured resultsof the internal backgrounds for the two crystals. Inthis section we describe some details of these mea-surements. K background The most serious internal background contami-nation is K because of the low-energy x-ray that is produced during its electron capture decay pro-cess, which proceeds via a transition to an excitedstate of Ar with a branching ratio of 10%. Thisdecay generates a ∼ γ -ray. If the accompanying 1460 keV γ -ray escapes from the crystal, the event consistsof a single 3 keV hit. The K level in the DAMAcrystals is in the 10–20 ppb range [34].We studied coincidence signals between NaI ( ∼ K decays. Figure 6(a) shows a scat-terplot of NaI versus
CsI detector reponse with-out the application of any of the requirements usedto remove PMT noise-induced signals that are dis-cussed in Section 6 below; Fig. 6(b) shows the sameplot after the application of these requirements. InFig. 6(b), the event cluster near 3 keV in NaI and1460 keV in a surrounding CsI crystal is from Kdecays. Another, smaller cluster near ∼ Na decays. By comparing the rate for Kinduced events with a Geant4-based detector sim-ulation [35] for the CsI and NaI crystal setup, wedetermine the K contamination in the NaI crys-tals to be 41 . ± . . ± .
09 mBq/kg) and49 . ± . . ± .
07 mBq/kg), respectively,for NaI-001 and NaI-002. This is close to the Klevels measured in the ANAIS-25 crystals [24, 36]. U background Although the PMTs are nonlinear at high lightoutput, alpha-induced events inside the crystal canbe identified by the mean time of the signal, definedas h t i ≡ P i A i t i P i A i . (1)Here A i and t i are the charge and time of eachcluster (for low energies) or digitized bin (for highenergies). Figure 7 shows a scatter plot of the pulseheight versus mean time for event signals from NaI-001. Alpha-induced events are clearly separatedfrom gamma-induced events in the high-energy re-gion because of the faster decay times of alpha-induced signals.Because of the nonlinearity of high-energy sig-nals, alpha particles from different nuclides can-not be distinguished event-by-event by their mea-sured energies. Instead we determine the level of U chain contaminants by exploiting the
Po237 µ s mean lifetime that occurs between Bi β -decay and Po α -decay, a technique that was used5 (keV) CsI E ( k e V ) N a I E (keV) CsI E ( k e V ) N a I E (a) No cut (b) After PMT noise cuts Figure 6: Scatter plots of energy in a NaI(Tl) crystal ( E NaI ) versus that in a surrounding CsI(Tl) crystal ( E CsI ) before (a)and after (b) the application of event selection requirements that remove PMT noise events. The cluster at 3 keV is due to K and that at 1 keV is due to Na. s) µ Mean time ( -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 P u l s e he i gh t ( a r b i t r a r y un i t s ) Figure 7: A scatter plot of maximum height versus meantime of event signals from NaI-001. Alpha-induced eventsare well separated from γ -induced signals because of theirshorter decay time. successfully for contamination measurements in theKIMS CsI crystals [28]. Figure 8 shows the distribu-tion of measured time intervals between an alpha-induced event and its immediately preceding event.Since there is no significant exponential componentobserved with the 237- µ s decay time of Po, anupper limit on the activity level is determined. Thisanalysis shows that the contamination levels fromthe
U chain are already sufficiently low for aWIMP dark matter search. Th background Contamination from the
Th chain is studiedby using α – α time interval measurements in thecrystals. In this case we look for a Po α -decay t (ms) ∆ N u m be r o f e v en t s Figure 8: The β – α coincidence time spectrum from the NaI-001 crystal. There is no significant exponential decay froma Bi β and Po α -decay component and an upper limitis determined from the fit. component with a mean time of 209 ms followingits production via Rn → Po α -decay. Figure9 shows the distribution of the time difference be-tween two alpha events. As one can see in Fig. 9,there is a small exponential component with the Po decay time in NaI-001; there is no such sig-nal in NaI-002 and we set an upper limit for thecontamination in that crystal. These α – α eventrates can translate into contamination levels fromthe Th series in the
Th chain. The
Thcontamination levels of the two crystals, listed inTable 3, are also sufficiently low. Pb background The levels of
U and
Th contamination mea-sured in both crystals are too low to account for6 t (s) ∆ N u m be r o f e v en t s Figure 9: The time difference distribution between two alphaevents in the NaI-001 crystal. The exponential componentbelow 1 s is due to the sequential decays of
Rn and
Po. the total observed alpha particle rate, which sug-gests that they are due to decays of
Po nucleithat originate from
Rn contamination that oc-cured sometime during the powder and/or crystalprocessing stages. This is confirmed by the obser-vation of a 46-keV γ peak that is characteristic of Pb.The time change in the total alpha rate providesinformation about when the
Rn contaminationoccurred. After
Rn contamination, the num-ber of
Pb nuclei increases as does the Po α -decay rate. After about three years, equilibrium isreached and the Po activity becomes constant.Figure 10 shows the total alpha rates in the twocrystals as a function of data-taking time (days).The NaI-001 and NaI-002 crystals emit about 2344and 1334 alpha particles per day, respectively. Af-ter considering the crystal masses, we find that theNaI-002 alpha activity is less than that for NaI-001by almost a factor of two. Moreover, we also seethat the NaI-002 crystal’s alpha activity is increas-ing with time.For
Po, the alpha activity will increase as R α ( t ) ≈ A (1 − e − ( t − t ) /τ Po ) , (2)where τ Po is the decay time of Po (200 days) and t is the time the initial Pb contamination oc-curred, assuming that the contamination happenedsuddenly. The measured alpha rate was fitted tothis equation, and the results indicate that the con-tamination occurred at the end of April, 2013. Thiscoincides with the time that the crystal was grown,and we conclude that the contamination occurredthen.
5. Background from cosmic excitation
The two crystals were transported from theU.S. to Korea by different means, NaI-001 by airand NaI-002 by sea, in order to understand thecosmogenic-activation-dependence on the deliverymethod. Figure 11(a) shows the energy spectra forthe NaI-001 crystal during the first week and fora week-long period after a 64-day delay after thearrival of the crystal underground. Figure 11(b)shows the difference between the first and secondmeasurements. The peak at 68.7-keV is the sumof γ -rays and x-rays from I electron capture de-cay. The lower energy peak also can be identi-fied as originating from iodine and tellurium de-cays. We found significantly lower cosmogenic ac-tivation in the NaI-002 crystal, and we concludethat surface transportation is mandatory for low-background crystals. Na can be produced through the ( n, n ) re-action on Na by energetic cosmic neutrons atsea level. It decays via positron emission (90%)and electron capture (10%) followed by 1270-keVgamma emission with a mean lifetime of 3.8 years.The electron capture decay produces ∼ Na decay willproduce 0.8-keV x-rays and 1270-keV γ -rays at thesame time. The γ - γ coincidences show up in Fig. 6as a cluster of events below the K decay events.The more frequent β + decay channel does not gen-erate low-energy x-rays. The 0.8-keV events areuseful for studying selection efficiencies in the 1-keV energy region.
6. PMT noise background
Photomultiplier tubes are known to generate low-energy noise signals primarily via four differentmechanisms. First, radioactive decays of U, Th,and K inside the PMT materials generate ultra-violet and/or visible photons directly inside thePMTs. Second, the high voltage applied to thePMT can cause charge to accumulate somewhere inthe PMT and subsequently discharge, producing aflash. Third, PMT dark currents will produce acci-dental coincidences between two PMTs that satisfythe trigger condition. Fourth, large pulses can re-sult from afterpulsing produced by ionized residualgas inside the PMT. In fact, PMT noise involvescomplex phenomena that are far from being com-pletely understood.7 ime (day) e v en t s α N u m be r o f NaI-001NaI-002
Date(Mon/Day/Year) e v en t s α N u m be r o f (a) (b) Figure 10: (a) Total number of alpha particles per day for the two NaI detectors. (b) The alpha activity increase in NaI-002is fitted with a model in which a nearly instantaneous
Pb contamination is assumed.
Energy (keV) c oun t s / da y / k g / k e V A first weekA week after 64 days Energy (keV) C oun t s / da y / k g / k e V (a) (b) Figure 11: (a) Energy spectra of the NaI-001 crystal during the first week after the arrival of the crystal underground and for aweek following a 64 day interval. The peak around 50-keV is due to
Pb decay inside the crystal. (b) The difference betweenthe initial and the delayed energy spectra, which shows short life-time cosmogenic activations.
All PMTs normally have some level of dark cur-rent; this is essentially due to SPEs that are spon-taneously emitted from the photocathode. In thePMTs used in these measurements, the SPE ratesvary from PMT to PMT, and are typically of order ∼ ∼ > The DAMA group reported a signal selection cri-teria for efficiently removing the PMT noise eventsfrom their NaI(Tl) detectors that exploits the factthat noise pulses are generally fast. The DAMA re-quirement places restrictions on the ratio of “fast”charge (0–50 ns), X1, and “slow” charge (100–600 ns), X2 [32, 34]. We examined the DAMA pa-rameters for our NaI(Tl) crystals. Figure 12 shows8 ( s l o w c h a r g e ) X ( f a s t c ha r ge ) N u m be r o f e v en t s X ( s l o w c h a r g e ) X ( f a s t c ha r ge ) N u m be r o f e v en t s (a) Background data (b) Fe calibration data
Figure 12: Two-dimensional plots of “fast” and “slow” charges for background (a) and low-energy x-ray source calibration (b)data. PMT noise events have larger “fast” charges and signal events have larger “slow” charges. a two-dimensional X2 versus
X1 scatter-plot forevents in the 2–4 keV energy range both for back-ground data (a) and for data taken with an Fesource (b). The figure shows that our discrimina-tion between noise and signal is very efficient, sim-ilar to the DAMA results [34]. A rejection rate ofthe 2–4 keV WIMP search data is approximately84%, however approximately 86% of the ∼ K coincidence events is remained.
Although a large fraction of PMT noise events be-low 5 keV are removed by the DAMA requirement,we find that some PMT noiselike events remain.We, therefore, developed further requirements toremove these events. We define the asymmetry be-tween the two PMT signals asasymmetry ≡ Q − Q Q + Q , (3)where Q and Q are the charges measured by thetwo PMTs. This asymmetry allows us to locatewhere the event occurred inside the crystal. A sim-ilar study of asymmetry cuts with NaI crystals wasreported previously [37].To characterize PMT noiselike events, we usedmultiple hit events in which two or more detector modules satisfy the trigger condition. Figure 13shows a two-dimensional asymmetry versus energyscatter plot for single-hit (a) and multiple-hit (b)events. These data are obtained from the NaI-001crystal coupled with R12669 PMTs. In the datashown in these figures, the DAMA requirement hasalready been applied. The multiple-hit events have | asymmetry | < .
6, while many single-hit eventswith energy below 3 keV have asymmetries that areeven larger than those for real events that occurnear the edge of the crystal. This suggests thatthese events are caused by visible light producednear one of the PMTs. In contrast, the NaI-002detector with the R11065 metal PMTs do not showmany events with such large asymmetries. Eventswith | asymmetry | > . We found a peculiar class of low energy eventsin the single hit sample that are made up of SPEclusters that are spread nearly uniformly over afew hundred nanosecond interval. These are ev-ident in the scatter-plot of the total energy ver-sus the average charge of the energy clusters (totalcharge/number of clusters) shown in Fig. 14, wherethe black dot entries are for single-hit events andthe red circles are for multiple-hit events. These9 symmetry -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 E ne r g y ( k e V ) Asymmetry -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 E ne r g y ( k e V ) (a) Single-hit events (b) Multiple-hit events Figure 13: Asymmetry versus energy plots of the NaI-001 coupled with an R12669 PMT for single-hit events (a) and multiple-hitevents (b). Many single-hit events with energy below 3 keV have large asymmetries that are attributed to PMT noise. data were obtained with the NaI-001 crystal cou-pled to the glass R12669 PMTs, and the DAMA andasymmetry selection requirements were applied. Adistinct cluster of low energy signals with an aver-age cluster charge consistent with that for a SPEshows up for single-hit events. While the source ofthese events is still not understood, since they donot show up in multiple hit events, they are consid-ered likely to be induced by PMT-noise. Althoughthis phenomenon requires additional study, for nowwe veto events that lie to the left of the solid curveshown in the figure.
Average cluster charge (pC) E ne r g y ( k e V ) Single hit eventsMultiple hit events
Figure 14: Energy versus the average charge of clusters forsingle-hit events (dots) and multiple-hit events (open cir-cles). There are additional noise events for the single-hitevents in lower cluster charge regions. The solid line showsthe cut condition to remove such noise events.
7. Background model
Figure 15 shows the background levels of the twocrystals coupled to R12669 PMTs after the appli-cation of all of the event selection criteria discussedbefore. NaI-002 has a much lower background levelthan that of NaI-001, because of its lower cosmo-genic activation as a result of its surface deliveryand its lower
Pb contamination. Its backgroundlevel at 6 keV is ∼ Energy (keV) C oun t s / da y / k g / k e V NaI-001NaI-002
Figure 15: Background levels in the two crystals after theapplication of the PMT noise rejection requirements. Herewe show data obtained with the R12669 glass PMTs. .1. Background simulations We simulated the background spectra withGeant4-based detector simulations of
Pb, U, Th, and K with contamination levels set atthe measured values for each crystal. Figure 16shows the data and the simulated spectra for thetwo crystals. For the current NaI(Tl) crystals, thesignificant remaining backgrounds are from
Pb, K, and PMT noise. The U, Th, and Ra internalcontamination levels in the crystal produce back-grounds at low energies that are already sufficientlysmall: i.e. < ∼ γ -rays from sourcesthat are exterior to the crystal, i.e. , the PMTs, thesurrounding CsI crystals, and the materials of thesurrounding shield. In addition to these constantbackgrounds, there are also I and
Te cosmo-genic backgrounds that are continuously decreasingas a function of time. Further simulations will clar-ify and quantify the contributions from each exter-nal source as well as the contributions from cosmo-genic activations to the total background level.
8. Perspectives
The DAMA experiment has been consistentlyshowing a significant annual modulation with twodifferent experimental arrangements that has per-sisted over the past 15 years. To check theseresults, it is necessary to use the same targetmaterial and preferable to have a lower thresh-old and reduced background levels. Achieving asoftware energy threshold of 1 keV seems feasi-ble because of the high crystal light output andthe high quantum efficiency of the new PMTs.Background levels can be significantly improvedfrom the current measured background level of ∼ Pb and K, we are attempting toreduce these contaminations below the backgroundlevel of 0.2 counts/keV/kg/day for each source, andthe prospects of starting with purer powder fromSigma Aldrich are promising. In addition to the internal backgrounds, the ex-ternal backgrounds need to be controlled well below0.5 counts/keV/kg/day. Low-background, metal-housed PMTs with lower radioactivity specifica-tions are commercially available and it is possible touse high efficiency SBA photocathodes with thesetubes. We are working closely with the HamamatsuCompany to develop a PMT that is better suited fora low-background NaI(Tl) crystal detector module.Further, we expect a significant reduction in theinternal or external backgrounds by the immersionof the NaI(Tl) crystal array inside a liquid scintil-lator box that provides an active veto capability.A naive simulation shows that ∼
70% of the PMT-initiated backgrounds below 10 keV can be vetoed.A performance test with a single NaI crystal is inprogress.
9. Conclusion
We tested the performance of two large NaI(Tl)crystals as part of a program to develop ultra-low-background NaI crystals for WIMP searches. Wedeveloped selection requirements that are effectivefor reducing PMT-noise induced background sig-nals. Based on this effort, we achieved a back-ground level of ∼ < Acknowledgments
We thank the Korea Hydro and Nuclear Power(KHNP) company for providing the undergroundlaboratory space at Yangyang. This research wasfunded by Grant No. IBS-R016-D1 and was sup-ported by the Basic Science Research Programthrough the National Research Foundation of Korea(NRF) funded by the Ministry of Education (NRF-2011-35B-C00007).
References [1] E. Komatsu et al. (WMAP Collaboration), Astrophys.J. Suppl. 192 (2011) 18.[2] P. A. R. Ade et al. (Planck Collaboration),arXiv:1303.5076. nergy (keV) C oun t s / k g / da y / k e V -1 DataTotal Simulation K Pb U Th Energy (keV) C oun t s / k g / da y / k e V -1 DataTotal Simulation K Pb U Th (a) NaI-001 (b) NaI-002 Figure 16: The measured background levels in the NaI crystals compared with simulations of backgrounds from internalcontamination in the crystals.[3] B. W. Lee and S. Weinberg, Phys. Rev. Lett. 39 (1977)165.[4] G. Jungman, A. Kamionkowski, and K. Griest, Phys.Rep. 267 (1996) 195.[5] R. Gaitskell, Annu. Rev. Nucl. Part. Sci. 54 (2004) 315.[6] L. Baudis, Phys. Dark Univ. 1 (2012) 94.[7] R. Bernabei et al. (DAMA Collaboration), Eur. Phys.J. C 56 (2008) 333.[8] R. Bernabei et al. (DAMA Collaboration), Eur. Phys.J. C 67 (2010) 39.[9] R. Bernabei et al. (DAMA Collaboration), Eur. Phys.J. C 73 (2013) 2648.[10] C. E. Aalseth et al. (CoGeNT Collaboration), Phys.Rev. Lett. 107 (2011) 141301.[11] C. E. Aalseth et al. (CoGeNT Collaboration), Phys.Rev. D 88 (2013) 012002.[12] C. E. Aalseth et al. (CoGeNT Collaboration),arXiv:1401.3295.[13] G. Angloher et al. (CRESST Collaboration), Eur. Phys.J. C 72 (2012) 1971.[14] R. Agnese et al. (CDMS Collaboration), Phys. Rev.Lett. 111 (2013) 251301.[15] Q. Yue et al. (CDEX Collaboration), arXiv:1404.4946.[16] G.K. Giovanetti et al. , arXiv:1407.2238.[17] G. Angloher et al. (CRESST Collaboration),arXiv:1407.3146.[18] E. Aprile et al. (XENON100 Collaboration), Phys. Rev.Lett. 109 (2012) 181301.[19] R. Agnese et al. (SuperCDMS Collaboration), Phys.Rev. Lett. 112 (2014) 041302.[20] D. S. Akerib et al. (LUX Collaboration), Phys Rev.Lett. 112 (2014) 091303.[21] G. Plante et al. , Phys Rev. C 84 (2011) 045805.[22] D. Barker and D. M. Mei, Astropart. Phys. 38 (2012)1.[23] Y. Y. Mao, L. E. Strigari, and R. H. Wechsler, Phys.Rev. D 89 (2014) 063513.[24] J. Amare et al. , Nucl. Instrum. Methods Phys. Res.,Sect. A 742 (2014) 187.[25] J. Amare et al. , arXiv:1404.3564.[26] J. Cherwinka et al. (DM-Ice Collaboration),arXiv:1401.4804. [27] Y. D. Kim et al. , J. Korean. Phys. Soc. 40 (2002) 520;Y. D. Kim et al. , Nucl. Instrum. Methods Phys. Res.,Sect. A 552 (2005) 456.[28] T. Y. Kim et al. , Nucl. Instrum. Methods Phys. Res.,Sect. A 500 (2003) 337; H. S. Lee et al. , Nucl. Instrum.Methods Phys. Res., Sect. A 571 (2007) 644.[29] H. S. Lee et al. , (KIMS Collaboration), Phys. Rev. Lett.99 (2007) 091301.[30] S. C. Kim et al. , (KIMS Collaboration), Phys. Rev.Lett. 108 (2012) 181301.[31] J. J. Zhu et al. , High Energ. Phys. Nucl. 29 (2005) 8.[32] R. Bernabei et al. , JINST 7 (2012) P03009.[33] H. S. Lee et al. (KIMS Collaboration), Phys. Lett. B633 (2006) 201.[34] R. Bernabei et al. , Nucl. Instrum. Methods Phys. Res.,Sect. A, 592 (2008) 297.[35] S. Agostinelli et al. , Nucl. Instrumen. Methods Phys.Res., Sect. A 506 (2003) 250; J. Allison et al. , IEEETran. Nucl. Sci. 53 (2006) 270.[36] C. Cuesta et al. , arXiv:1403.3580.[37] M. Robinson et al. , Nucl. Instrum. Methods Phys. Res.,Sect. A, 545 (2005) 225., Nucl. Instrum. Methods Phys. Res.,Sect. A, 545 (2005) 225.