Evaluation of the neutron background in CsI target for WIMP direct detection when using a reactor neutrino detector as a neutron veto system
aa r X i v : . [ a s t r o - ph . I M ] A p r Evaluation of the neutron background inCsI target for WIMP direct detectionwhen using a reactor neutrino detectoras a neutron veto system
Ye Xu a ∗ , Xiangpan Ji b , Haolin Li b , Yulong Feng b a Department of Mathematics and Physics, Fujian University of Technology,Fuzhou 350118, China b School of Physics, Nankai University, Tianjin 300071, China
Abstract
A direct WIMP (Weakly Interacting Massive Particle) detector with a neutron vetosystem is designed to better reject neutrons. An experimental configuration is studiedin the present paper: a WIMP detectors with CsI(Na) target is placed inside a reactorneutrino detector. The neutrino detector is used as a neutron veto device. The neutronbackground for the experimental design has been estimated using the Geant4 simulation.The results show that the neutron background can decrease to O(0.01) events per yearper tonne of CsI(Na). We calculate the sensitivity to spin-independent WIMP-nucleonelastic scattering. An exposure of one tonne × year could reach a cross-section of about3 × − pb. Keywords:
Dark matter, Neutron background, Neutrino detector, CsI(Na)
PACS numbers: 95.35.+d, 95.55.Vj, 29.40.Mc
It is indicated by seven year Wilkinson Microwave Anisotropy Probe data com-bined with measurements of baryon acoustic oscillations and Hubble constant that ∼ of the matter content in the Universe is non-baryonic dark matter [1, 2, 3].Weakly Interacting Massive Particles (WIMPs) [4], predicted by extensions of theStandard Model of particle physics, are a well-motivated class of candidates fordark matter. They are distributed in the halo surrounding the Milky Way. WIMPsmay be directly detected through measuring nuclear recoils in terrestrial detectorsproduced by their scattering off target nuclei [5, 6, 7]. The nuclear reoils is ex-pected to have a roughly exponential energy distribution with a mean energy in afew tens of keV [6, 8, 9].In direct searches for WIMPs, there are three different methods used to detectthe nuclear recoils, including collecting ionization, scintillation and heat signatures ∗ Corresponding author, e-mail address: [email protected] Detector description induced by them. The background of this detection is made up of electron recoilsproduced by γ and β scattering off electrons, and nuclear recoils produced byneutrons scattering elastically off target nucleus. It is very efficient discriminat-ing nuclear recoils from electron recoils with pulse shape discrimination, hybridmeasurements and so on. The rejection powers of these techniques can even reach [10, 11]. For example, the CDMS-II [10] and EDELWEISS-II [12] experimentsmeasure both ionization and heat signatures using cryogenic germanium detectorsin order to discriminate between nuclear and electron recoils, and the XENON100[13] and ZEPLIN-III [14] experiments measure both ionization and scintillationsignatures using two-phase xenon detectors. However, it is very difficult to dis-criminate between nuclear recoils induced by WIMPs and by neutrons. In somedark matter experiments and researches, tagging neutron is applied to reject neu-tron background. In the ZEPLIN-III experiment, the . Gadolinium (Gd) dopedpolypropylene is used as the neutron veto device, and its maximum tagging effi-ciency for neutrons reaches about [15]. In Ref. [16], the Gd-doped wateris used as the neutron veto, and its neutron background can be reduced to 2.2 (1)events per year per tonne of liquid xenon (liquid argon). In our past work [17], thereactor neutrino detector with Gd-doped liquid scintillator (Gd-LS) is used asthe neutron veto system, and its neutron background can be reduced to about 0.3per year per tonne of liquid xenon.Cesium iodide(CsI) crystals as a kind of dark matter target have been appliedto dark matter experiments, such as the KIMS experiment[18]. In a dark matterexperiment with CsI target, CsI crystals needn’t to be cooled with the liquidnitrogen or xenon, so this detector can be of the simpler structure. In the Sun, Luet al.’s work[19], the rejection power against electron recoil can reach O( ) withNa-doped CsI crystals. The feasibility of direct WIMPs detection with a neutronveto based on a neutrino detector had been validated in our past work[17]. So,in the present paper, a neutrino detector with Gd-LS ( Gd-doped) is still usedas a neutron-tagged device and WIMP detectors with CsI target are placed insidethe Gd-LS. Here we designed an experimental configuration: four WIMP detectorswith CsI(Na) target are individually placed inside four reactor neutrino detectormodules which are used as a neutron veto system. The experimental hall of theconfiguration is assumed to be located in an underground laboratory with a depthof 910 meter water equivalent (m.w.e.), which is similar to the far hall in the DayaBay reactor neutrino experiment[20]. Collecting scintillation signals is consideredas the only method of WIMPs detection in our work. The neutron background forthis design are estimated using the Geant4 [21] simulation.
Four identical WIMP detectors with CsI(Na) target are individually placed insidefour identical neutrino detector modules. The concentration of Na is 0.02% inCsI crystals. The experimental hall of this experimental configuration is assumedto be located in an underground laboratory with a depth of 910 m.w.e., which issimilar to the far hall in the Daya Bay reactor neutrino experiment. The detectoris located in a cavern of 20 × × m . The four identical cylindrical neutrinomodules (each 413.6 cm high and 393.6 cm in diameter) are immersed into a13 × × m water pool at a depth of 2.5 meters from the top of the pool andat a distance of 2.5 meters from each vertical surface of the pool. The detectorconfiguration is shown in Fig.1. Some features of simulation Each neutrino detector module is partitioned into three enclosed zones. Theinnermost zone is filled with Gd-LS, which is surrounded by a zone filled with un-load liquid scintillator (LS). The outermost zone is filled with transparent mineraloil. 366 8-inch PMTs are mounted in the mineral oil.Each WIMP detector (69.6 cm height, 59.9 cm in diameter) consists of threecomponents: a CsI(Na) crystal array, PMTs and Copper vessels. The CsI(Na)array is made up of 31 CsI(Na) crystals whose sections are regular hexagons (34cm height, 4.5 cm side length) and is placed inside a 1 cm thick copper vessel(69.2 cm height, 59.5 cm in diameter) filled with dry nitrogen gas. Fig.1 shows thearrangement of these crystals. 62 3-inch PMTs are individually mounted on twoends of these crystals. The copper vessel of each WIMP detector is surrounded byan 0.2 cm thick Aluminum reflector for photons produced in the Gd-LS.
The Geant4 (version 8.2) package[21] has been used in our simulations. The physicslist in the simulations includes transportation processes, decay processes, low en-ergy processes, electromagnetic interactions (multiple scattering processes, ion-ization processes, scintillation processes, optical processes, cherenkov processes,Bremsstrahlung processes, etc.) and hadronic interactions (lepton nuclear pro-cesses, fission processes, elastic scattering processes, inelastic scattering processes,capture processes, etc.). The hadronic processes include the low energy(<20 GeV),high energy(>20GeV) and neutron high-precision(<20 MeV) models. The cuts forthe productions of gammas, electrons and positrons are 1 mm, 100 µ m and 100 µ m,respectively. The quenching factor is defined as the ratio of the detector responseto nuclear and electron recoils. The Birks factor for protons in the Gd-LS is setto 0.01 g/cm /MeV, corresponding to the quenching factor 0.17 at 1 MeV, in oursimulation. Besides, We utilize 10 Intel Core i5 CPUs (four cores, each core offera base of speed of 2.8GHz) in the neutron background evaluation. The recoil energies for WIMP interactions with CsI(Na) nuclei were set to a rangefrom 20 keV to 100 keV in this work. Proton recoils induced by neutrons andneutron-captured signals are used to tag neutrons which reach the Gd-LS. Theenergy deposition produced by proton recoils is close to a uniform distribution.Neutrons captured on Gd and H lead to a release of about 8 MeV and 2.2 MeVof γ particles, respectively. Due to the instrumental limitations of the Gd-LS, weassume neutrons will be tagged if their energy deposition in the Gd-LS is more than1 MeV, corresponding to 0.17 MeVee (electron equivalent energy). In the Gd-LS, itis difficult to distinguish signals induced by neutrons from electron recoils, whichare caused by the radioactivities in the detector components and the surroundingrocks. But these radioactivities can be controlled to less than ∼
50 Hz accordingto the Daya Bay experiment[20]. If we assume a 100 µ s for neutron tagging timewindow, the indistinguishable signals due to the radioactivities will result in a totaldead time of less than 44 hours per year.Neutrons are produced from the detector components and their surroundingrock. For the neutrons from the surrounding rock there are two origins: first byspontaneous fission and ( α , n) reactions due to U and Th in the rock (these Neutron background estimation neutrons can be omitted because they are efficiently shielded, see Sec.4.2), andsecondly by cosmic muon interactions with the surrounding rock.We estimated the number of neutron background in the CsI(Na) target of onetonne. This number has been normalized to one year of data taking and aresummarized in Tab.1. Neutrons from the detector components are induced by ( α , n) reactions due to Uand Th. According to Mei et al.[22], the differential spectra of neutron yield canbe expressed as Y i ( E n ) = N i X j R α ( E j ) S mi ( E j ) E j Z dσ ( E α , E n ) dE α dE α where N i is the total number of atoms for the i th element in the host material, R α ( E j ) refers to the α -particle production rate for the decay with the energy E j from T h or U decay chain, E α refers to the α energy, E n refers to the neutronenergy, and S mi is the mass stopping power of the i th element. The U and Th contaminations in the
SiO material are considered as the onlyneutron source in the PMTs in our work. Neutrons from SiO are emitted withtheir average energy of 2.68 MeV[22]. The PMTs in the copper vessels of thefour WIMP detectors amounts to 248. The U and Th concentrations in the PMTcomponents can reach ten or even less ppb[23], so a rate of one neutron emittedper PMT per year is conservatively estimated[24]. Consequently, there are 248neutrons produced by all the PMTs in the copper vessels per year. A simulatedsample of 2.48 × events is used to study this neutron background. These eventsgenerated isotropically are uniformly distributed in the SiO material of the PMTs.The simulation result is summarized in Tab. 1. 2.9 neutron events/(ton · yr)reach the CsI(Na) target and their energy deposition falls in the same range asthat of the WIMP interactions, as seen in Tab. 1. Because 0.04 of them arenot tagged in the Gd-LS, these background events cannot be eliminated. Theuncertainties of the neutron background from the PMTs in Tab.1 are from thebinned neutron spectra in Ref.[22]. But the neutron background errors from thestatistical fluctuation (their relative errors are less than 1 % ) are too small to betaken into account. In the copper vessels, neutrons are produced by the U and Th contaminationsand emitted with their average energy of 0.81 MeV[22]. Their total volume isabout . × cm . The radioactive impurities Th can be reduced to . × − ppb in some copper samples[25]. If we conservatively assume a 0.001 ppb U/Thconcentrations in the copper material[26], a rate of one neutron emitted per × cm per year is estimated[16]. Consequently, there are 1.8 neutrons produced bythe all copper vessels per year. A simulated sample of 1.8 × events is used tostudy this neutron background. These events generated isotropically are uniformlydistributed in the copper material of the copper vessels. Neutron background estimation The simulation result is summarized in Tab.1. 0.03 neutron events/(ton · yr)reach the CsI(Na) target and their energy deposition falls in the same range asthat of the WIMP interactions (see Tab.1). As 0.001 of them are not tagged inthe Gd-LS, these background events cannot be eliminated. The uncertainty of theneutron background from the copper vessels in Tab.1 are from the binned neutronspectra in the Ref.[22]. But the neutron background errors from the statisticalfluctuation (their relative errors are less than 1 % ) are too small to be taken intoaccount. The U and Th contaminations in other detector components also contribute tothe neutron background in our experiment setup. Neutrons from the Aluminumreflectors are emitted with the average energy of 1.96 MeV[22]. The U and Thcontaminations in the carbon material are considered as the only neutron source inthe Gd-LS/LS. Neutrons from the Gd-LS/LS are emitted with the average energyof 5.23 MeV[22]. The U and Th contaminations in the
SiO material are consideredas the only neutron source in the PMTs in the oil. Neutrons from PMTs areemitted with the average energy of 2.68 MeV[22]. The U and Th contaminationsin the iron material are considered as the only neutron source in the stainlesssteel tanks. Neutrons from the stainless steel tanks are emitted with the averageenergy of 1.55 MeV[22]. We evaluated the neutron background from the abovecomponents using the Geant4 simulation. All the nuclear recoils in the CsI(Na)target, which fall in the same range as that of the WIMP interactions, are tagged.The neutron background from these components can be ignored. The simulatedsamples that amount to 1000 years of data taking are used to evaluate theseneutron backgrounds. In the surrounding rock, almost all the neutrons due to natural radioactivity arebelow 10 MeV [16, 27]. Water can be used for shielding neutrons effectively, espe-cially in the low energy range of less than 10 MeV [28]. The WIMP detectors aresurrounded by about 2.5 meters of water and more than 1 meter of Gd-LS/LS, sothese shields can reduce the neutron contamination from the radioactivities to anegligible level.
Neutrons produced by cosmic muon interactions constitute an important back-ground component for dark matter searches. These neutrons with a hard energyspectrum extending to several GeV energies, are able to travel far from producedvertices.The total cosmogenic neutron flux at a depth of 910 m.w.e. is evaluated by afunction of the depth for a site with a flat rock overburden [29], and it is 1.31 × − cm − s − . The energy spectrum (see Fig.2) and angular distribution of theseneutrons are evaluated at the depth of 910 m.w.e. by the method in [29, 30]. Theneutrons with the specified energy and angular distributions are sampled on thesurface of the cavern, and the neutron interactions with the detector are simulated Rough estimation of other background with the Geant4 package. A simulated sample of 1.25 × events is used to studyin this neutron background.Tab.1 shows that 2.5 neutron events/(ton · yr) reach the CsI(Na) target and theirenergy deposition falls in the same range as that of the WIMP interactions. 0.28of them are not tagged by the Gd-LS/LS. Muon veto systems can tag muons veryeffectively, thereby most cosmogenic neutrons can be rejected. In the Daya Bayexperiment, the contamination level can even be reduced by a factor of more than30[20]. We assume the neutron contamination level from cosmic muons decreasesby a factor of 30 using a muon veto system. This could lead to the decrease ofcosmogenic neutron contamination to 0.01 events/(ton · yr). The uncertainties ofthe cosmogenic neutron background in Tab.1 are from the statistical fluctuation. Besides neutron background, other background events are mainly from reactorneutrino events and electron recoils in the experimental design in the presentpaper. The contamination caused by electron recoils consists of bulk electronrecoil events and surface events.
Since neutrino detectors are fairly close to nuclear reactors (about 2 kilometersaway) in reactor neutrino experiments, a large number of reactor neutrinos willpass through the detectors, and nuclear recoils will be produced by neutrino elasticscattering off target nucleus in the WIMP detectors. Although neutrinos may be asource of background for dark matter searches, they can be reduced to a negligiblelevel by setting the recoil energy threshold of 10 keV[31]. Besides, nuclear recoilsmay also be produced by low energy neutrons produced by the inverse β -decayreaction ¯ ν e + p → e + + n . But their kinetic energies are almost below 100 keV[32],and their maximum energy deposition in the WIMP detectors is as large as a fewkeV. Thus the neutron contamination can be reduced to a negligible level by theenergy threshold of 10 keV. The intenal sources of electron recoils are mainly caused by the radioisotopes of Cs and Rb in the CsI crystals, and the external ones are mainly from theradioisotopes of U , Th and K in the 3" PMTs inside the copper vessels.Here we assume that the concentrations of the Cs and Rb in the CsI crystalsare about 2 mBq/kg and 1 ppb[33, 34], respectively. Then their background rateswill be about 1.7 counts/keV/kg/day (cpd) in the region 10keV[34]. While theconcentrations of U/Th/K are about 78, 25 and 504 mBq/PMT, respectively[34],and their background rates will be about 0.5 cpd in the region 10keV. Consideringthe fact that the rejection power against electron recoils can reach O( )[19], weroughly estimate that the bulk electron recoil contamination due to the internaland external sources of the CsI(Na) crystals is about 0.9 events/(ton · yr). Discussion and conclusion The surface events for CsI(Na) crystals are caused by the deliquescence whichreduces the Na + concentration on the crystal surface[19]. Signals produced by α particles from the progenies of Rn in the air on the "old" surface of thecrystal mimic nuclear recoils seriously. The surface events can be prevented fromby avoiding the deliquescence of the crystal surface. For example, the coppervessels are filled with dry nitrogen gas[34], and thus there is no air between thecopper vessels and CsI(Na) crystals. Hence the surface event contamination canbe reduced to a negligible level by the above methods. Neutron background can be effectively suppressed neutrino detectors used as aneutron veto system in direct dark matter searches. Tab.1 shows the total neutroncontamination are 0.05 events/(ton · yr). And compared to Ref.[17], it is reducedby a factor of about 6. If the electron recoils contaminations are considered, thetotal background amounts to about 1 events/(ton · yr).According to our work, the neutron background is mainly from the PMTs inthe copper vessels in this configuration with the CsI(Na) targets. After finishinga precision measurement of the neutrino mixing angle θ , we can utilize the ex-isting experiment hall and neutrino detectors. This will not only save substantialcost and time for direct dark matter searches, but the neutron background couldalso decrease to O(0.01) events per year per tonne of CsI(Na) in the case of theDaya Bay experiment. According to Ref.[29], The neutron fluxes in the RENO(in an underground laboratory with a depth of 450 m.w.e.), Double CHOOZ (inan underground laboratory with a depth of 300 m.w.e.) experiments[35, 36] arerespectively about 5 and 3 times more than that of the Daya Bay experiment.Their neutron backgrounds are roughly estimated to be about 0.1 events/(ton · yr),if their detector configurations are the same as the one described above. Theneutron contamination is one order of magnitude smaller than the electron recoilcontamination, so neutron contamination can be ignored in this detector configu-ration.To evaluate the detector capability of directly detecting dark matter, we assumea standard dark matter galactic halo[9], an energy resolution that amounts to 25 % for the energy range of interest and 40 % nuclear recoil acceptance[18].If no signals are significantly observed, a sensitivity to WIMP-nucleon spin-independent elastic scattering can be calculated via the same method as Ref.[37].Our calculation shows that an exposure of one tonne × year could reach a cross-section of about 3 × − pb at the 90 % confidence level (see Fig.3). This work is supported in part by the National Natural Science Foundation ofChina (NSFC) under contract No. 11235006 and the science fund of Fujian Uni-versity of Technology under contract No. GY-Z13114.
References [1] N. Jarosik, et.al., ApJS, 192, 14 (2011)
Acknowledgements [2] W. J. Percival, et al., MNRAS, 401, 2148 (2010)[3] A. G. Riess, et al., ApJ, 699, 539 (2009)[4] G. Steigman and M. S. Turner, Nucl. Phys. B253, 375 (1985)[5] M. W. Goodman and E. Witten, Phys. Rev. D31, 3059 (1985)[6] G. Jungman, M. Kamionkowski, and K. Griest, Phys. Rep. 267, 195 (1996)[7] R. J. Gaitskell, Ann. Rev. Nucl. Part. Sci. 54, 315 (2004)[8] G. Bertone, D. Hooper and J. Silk, Phys. Rept. 405, 279 (2005)[9] J. D. Lewin and P. F. Smith, Astropart. Phys. 6, 87 (1996)[10] Z. Ahmed, et al., Science 327, 1619 (2010)[11] W.H.Lippincott, et al., Phys. Rev. C78, 035801 (2008)[12] E. Armengaud, et al., Phys. Lett. B702, 329 (2011)[13] E. Aprile, et al., Astropart. Phys. 35, 573 (2012)[14] D. Akimov, et al., Phys. Lett. B709, 14 (2012)[15] D. Akimov, et al., Astropart, Phys. 34, 151 (2010)[16] A. Bueno, M. C. Carmona and A. J. Melgarejo, JCAP. 08, 019 (2008).[17] Ye Xu, et al., JCAP06, 009 (2011)[18] S.C. Kim, et al., Phys. Rev. Lett. 108, 181301 (2012)[19] Xilei Sun, Junguang Lu, et al., Nucl. Instrum. Meth. A642, 52-58, 2011[20] F.P.An, et al., Phys. Rev. Letts, 108, 171803 (2012); Daya Bay Collaboration,arXiv:hep-ex/0701029v1[21] S. Agostinelli, et al., Nucl. Instru. Meth. A506, 250 (2003)[22] D. M. Mei, C. Zhang and A. Hime, Nucl. Instrum. Meth. A606, 651 (2009)[23] M.J.Carson et al.,Astropart. Phys. 21 667 (2004)[24] A. Bueno, M.C. Carmona and A.J. Melgarejo, JCAP08, 019 (2008), arXiv:0805.0694[25] Marin E. Keillor, et al., J. Radioanal. Nucl. Chem. 282, 703-708 (2009)[26] D. S. Leonard, et al., Nucl. Instrum. and Meth. A591, 490 (2008)[27] M. J. Carson, et al., Astropart. Phys. 21, 667 (2004)[28] J. M. Carmona, et al., Astropart. Phys. 21, 523 (2004)[29] D. M. Mei and A. Hime, Phys. Rev. D73, 053004 (2006)[30] Y.F.Wang, et al., Phys. Rev.D64, 013012 (2001) Acknowledgements [31] Jocelyn Monroe and Peter Fisher, Phys. Rev. D76, 033007 (2007)[32] M. Apollonio, et al., CHOOZ Collaboration, Eur. Phys. J. C27, 331-374(2003), arXiv: hep-ex/0301017[33] T. Y. Kim, et al., Nucl. Instrum. and Meth. A500, 337-344 (2003)[34] H. S. Lee, et al., Nucl. Instrum. and Meth. A571, 644-655 (2007)[35] J.K. Ahn, et al., Phys. Rev. Letts. 108, 191802 (2012)[36] F. Ardellier, et al., Double Chooz: A Search for the Neutrino Mixing Angle θ , arXiv: hep-ex/0606025[37] G.J.Feldman and R.D.Cousins, Phys. Rev. D57, 3873 (1998)[38] http://pisrv0.pit.physik.uni-tuebingen.de/darkmatter/limits/index.php Acknowledgements E recoil <100keV Not TaggedPMTs in copper vessels 2.9 ± ± ± ± ± ± ± ± ± ± Tab. 1:
Estimation of neutron background from different sources for an un-derground laboratory at a depth of 910 m.w.e. The column labeled"20keV< E recoil <100keV" identifies the number of neutrons whose energydeposition in the CsI(Na) target is in the same range as WIMP interac-tions. The column labeled "Not Tagged" identifies the number of neutronswhich are misidentified as WIMP signatures (their energy deposition in theCsI(Na) target is in the same range as WIMP interactions while their recoilenergies in the Gd-LS/LS are less than the energy threshold of 1 MeV).The row labeled "copper vessel" identifies the number of neutrons fromthe copper vessels. The row labeled "PMTs in copper vessel" identifies thenumber of neutrons from the PMTs in the copper vessels. The row labeled"cosmic muons" identifies the number of cosmogenic neutrons in the caseof not using the muon veto system. The row labeled "muon veto" iden-tifies the number of cosmogenic neutrons in the case of using the muonveto system. We assume that neutron contamination level from cosmicmuons decreases by a factor of 30 using a muon veto system. Only thetotal background in the case of using the muon veto system is listed in thistable. The terms after ± are errors. (a) (b)(c) (d) PMTCsIWaterNeutrino Detector VetoWIMP DetectorMineral OilLSGd-LSWIMP Detector Copper Vessel
Fig. 1: (a) Cross-section for WIMP detector with CsI targets. (b) Longitudinalsection for WIMP detector with CsI targets. (c) Longitudinal section for aneutrino detector where a WIMP detector is placed inside. (d) Four WIMPdetectors are individually placed inside four neutrino detectors in a watershield.
Acknowledgements Neutron Energy (MeV)0 500 1000 1500 2000 2500 3000 3500 4000 A r b i t r a r y un i t s -1 Fig. 2:
The energy spectrum of cosmogenic neutrons at depth of 910 m.w.e. /GeV
WIMP M ( pb ) W I M P - nu c l eon s -11 -10 -9 -8 -7 -6 -5 Fig. 3:
We calculate the sensitivity to spin-independent WIMP-nucleon elasticscattering assuming an exposure of one tonne × year. The calculationshows this exposure could reach a cross-section of about 3 × − pb at the90 %%