The neutron background of the XENON100 dark matter experiment
E.Aprile, M.Alfonsi, K.Arisaka, F.Arneodo, C.Balan, L.Baudis, B.Bauermeister, A.Behrens, P.Beltrame, K.Bokeloh, A.Brown, E.Brown, G.Bruno, R.Budnik, J.M.R.Cardoso, W.-T.Chen, B.Choi, A.P. Colijn, H.Contreras, J.P. Cussonneau, M.P. Decowski, E.Duchovni, S.Fattori, A.D. Ferella, W.Fulgione, F.Gao, M.Garbini, C.Ghag, K.-L. Giboni, L.W.Goetzke, C.Grignon, E.Gross, W.Hampel, F.Kaether, A.Kish, J.Lamblin, H.Landsman, R.F. Lang, M.Le Calloch, C.Levy, K.E. Lim, Q.Lin, S.Lindemann, M.Lindner, J.A.M.Lopes, K.Lung, T.Marrodan Undagoitia, F.V. Massoli, A.J. Melgarejo Fernandez, Y.Meng, M.Messina, A.Molinario, K.Ni, U.Oberlack, S.E.A. Orrigo, E.Pantic, R.Persiani, G.Plante, N.Priel, A.Rizzo, S.Rosendahl, J.M.F. dos Santos, G. Sartorelli, J.Schreiner, M.Schumann, L.ScottoLavina, P.R. Scovell, M.Selvi, P.Shagin, H.Simgen, A.Teymourian, D.Thers, E.Tziaferi, O.Vitells, H.Wang, M.Weber, C.Weinheimer
TThe neutron background of the XENON100 dark matter search experiment
E. Aprile, M. Alfonsi, K. Arisaka, F. Arneodo, C. Balan, L. Baudis, B. Bauermeister, A. Behrens, P. Beltrame,
8, 3
K. Bokeloh, A. Brown, E. Brown, G. Bruno, R. Budnik, J. M. R. Cardoso, W.-T. Chen, B. Choi, A. P. Colijn, H. Contreras, J. P. Cussonneau, M. P. Decowski, E. Duchovni, S. Fattori, A. D. Ferella,
4, 6
W. Fulgione, F. Gao, M. Garbini, C. Ghag, K.-L. Giboni, L. W. Goetzke, C. Grignon, E. Gross, W. Hampel, F. Kaether, A. Kish, ∗ J. Lamblin, H. Landsman, R. F. Lang, M. Le Calloch, C. Levy, K. E. Lim, Q. Lin, S. Lindemann, M. Lindner, J. A. M. Lopes, K. Lung, T. Marrod´an Undagoitia,
15, 6
F. V. Massoli, A. J. Melgarejo Fernandez, Y. Meng, M. Messina, A. Molinario, K. Ni, U. Oberlack, S. E. A. Orrigo, † E. Pantic, R. Persiani, G. Plante, N. Priel, A. Rizzo, S. Rosendahl, J. M. F. dos Santos, G. Sartorelli, J. Schreiner, M. Schumann,
16, 6
L. Scotto Lavina, P. R. Scovell, M. Selvi, P. Shagin, H. Simgen, A. Teymourian, D. Thers, E. Tziaferi, O. Vitells, H. Wang, M. Weber, and C. Weinheimer (The XENON100 Collaboration) Physics Department, Columbia University, New York, NY 10027, USA Nikhef and the University of Amsterdam, Science Park, Amsterdam, Netherlands Physics & Astronomy Department, University of California, Los Angeles, USA INFN, Laboratori Nazionali del Gran Sasso, Assergi, 67100, Italy Department of Physics, University of Coimbra, R. Larga, 3004-516, Coimbra, Portugal Physics Institute, University of Z¨urich, Winterthurerstr. 190, CH-8057, Z¨urich, Switzerland Institut f¨ur Physik, Johannes Gutenberg-Universit¨at Mainz, 55099 Mainz, Germany Department of Particle Physics and Astrophysics, Weizmann Institute of Science, 76100 Rehovot, Israel Institut f¨ur Kernphysik, Wilhelms-Universit¨at M¨unster, 48149 M¨unster, Germany Department of Physics, Purdue University, West Lafayette, IN 47907, USA SUBATECH, Ecole des Mines de Nantes, CNRS/In2p3, Universit´e de Nantes, 44307 Nantes, France INFN-Torino and Osservatorio Astrofisico di Torino, 10100 Torino, Italy Department of Physics, Shanghai Jiao Tong University, Shanghai, 200240, China University of Bologna and INFN-Bologna, Bologna, Italy Max-Planck-Institut f¨ur Kernphysik, Saupfercheckweg 1, 69117 Heidelberg, Germany Albert Einstein Center for Fundamental Physics, University of Bern, Sidlerstrasse 5, 3012 Bern, Switzerland Department of Physics and Astronomy, Rice University, Houston, TX 77005 - 1892, USA
The XENON100 experiment, installed underground at the Laboratori Nazionali del Gran Sasso(LNGS), aims to directly detect dark matter in the form of Weakly Interacting Massive Particles(WIMPs) via their elastic scattering off xenon nuclei. This paper presents a study on the nuclearrecoil background of the experiment, taking into account neutron backgrounds from ( α ,n) andspontaneous fission reactions due to natural radioactivity in the detector and shield materials, aswell as muon-induced neutrons. Based on Monte Carlo simulations and using measured radioactivecontaminations of all detector components, we predict the nuclear recoil backgrounds for the WIMPsearch results published by the XENON100 experiment in 2011 and 2012, 0.11 +0 . − . events and0.17 +0 . − . events, respectively, and conclude that they do not limit the sensitivity of the experiment. PACS numbers: 95.35.+d, 29.40.-n, 34.80.Dp
I. INTRODUCTION
The XENON100 detector aims at the direct detectionof dark matter in the form of Weakly Interacting MassiveParticles (WIMPs), and is taking data at the LaboratoriNazionali del Gran Sasso (LNGS) in Italy. It is a doublephase (liquid-gas) time projection chamber (TPC) with62 kg of liquid xenon (LXe) in the active volume viewedby two photomultiplier tube (PMT) arrays on the top ∗ Corresponding author: [email protected] † Present address: IFIC, CSIC-Universidad de Valencia, E-46071Valencia, Spain and bottom. The design of the detector and its work-ing principle are described in detail in Ref. [1], and thedata analysis procedure is explained in Ref. [2]. To-dateXENON100 is the most sensitive detector for direct darkmatter detection, and has set the most stringent limitson the spin-independent elastic WIMP-nucleon scatter-ing for WIMP masses above 8 GeV/c , with a minimumcross section of 2 × − cm at 55 GeV/c (at 90% confi-dence level) [3], and on the spin-dependent scattering forWIMP masses above 6 GeV/c , with a minimum crosssection of 3.5 × − cm at a WIMP mass of 45 GeV/c ,at 90% confidence level [4].A WIMP is expected to elastically scatter off a nu-cleus in the target, producing a low energy nuclear recoil(NR) [5]. There are two types of background for a dark a r X i v : . [ a s t r o - ph . I M ] S e p matter search with the xenon-based detectors: NRs fromhadronic interactions of neutrons, and electronic recoils(ERs) from electromagnetic interactions of γ -rays andelectrons. The different ionization density characteristicof a NR and an ER results in a different probability ofelectron-ion pair recombination, and thus a different ratioof scintillation to ionization [6]. This provides the possi-bility of rejecting ER background, which is performed inXENON100 with an efficiency higher than 99% at ∼ γ -rays in LXe. It is therefore crucial to min-imize and accurately characterize this potentially dan-gerous background. Due to intrinsic contamination with U, U, and
Th of materials in the detector andshield systems, radiogenic neutrons in the MeV rangeare produced in ( α ,n) reactions and spontaneous fission(SF). Additionally, cosmogenic neutrons with energiesextending to a few GeV are induced by muons penetrat-ing through the rock into the underground laboratory.This makes the neutron yield dependent on laboratory’sdepth.A study of the electronic recoil background in theXENON100 experiment was published in Ref. [9]. Inthis paper we summarize results from a comprehensiveMonte Carlo study, predicting the neutron induced NRbackground originating from natural radioactivity andcosmic muons. The study of the radiogenic neutron back-ground is based on calculations with the SOURCES-4Acode [10]. Simulations of the cosmogenic neutron back-ground employ the muon energy spectrum and angulardistribution generated with MUSIC and MUSUN [11].Neutron production and propagation is performed withthe GEANT4 toolkit [12]. The detector model developedfor the Monte Carlo simulations is described in detail inRefs. [9] and [13]. The results of this work were usedto predict the neutron background in the dark mattersearch data acquired with XENON100 and published inRefs. [3, 7, 8]. II. NEUTRON PRODUCTION DUE TONATURAL RADIOACTIVITY
The radiogenic neutron production rates and energyspectra were calculated with the SOURCES-4A software,modified by the group of the University of Sheffield in or-der to extend the cross sections for ( α ,n) reactions from6.5 MeV to 10 MeV, based on available experimentaldata [14]. The calculation was performed with the as-sumption that the α -emitters are uniformly distributedwithin a homogeneous material.The systematic uncertainty on the neutron produc-tion rate of the SOURCES-4A code is ±
17% [10]. Across-check of the calculations of the neutron production was performed with independent software described inRef. [15], showing agreement in neutron rates to within20%.The program takes into account the energy-dependent( α ,n) cross sections and Q -values for all target nuclides,particle stopping cross sections for all elements, the en-ergy of each α -particle, and the SF branching ratio foreach source nuclide. As an input, the code requires infor-mation about source and target nuclides, and the neutronenergy range to be considered. The fractions of atoms inthe target material were calculated using the chemicalcomposition of the detector and shield components pre-sented in Table I, and the natural isotopic abundancefrom Ref. [17]. The simulation takes into account all α -emitters from each decay chain, and their half-lives,assuming secular equilibrium.The cross section for the ( α ,n) reaction is suppressedby the Coulomb barrier for heavy nuclei, and increaseswith decreasing atomic number Z of the target. This ex-plains why the neutron production is dominated by ( α ,n)reactions for materials consisting of light elements, suchas polyethylene (only C and H) or polytetrafluoroethy-lene (PTFE, C and F). For high- Z materials, such ascopper and lead, the neutron production is almost en-tirely due to SF of U (see Fig. 1). Thus, the neutronproduction rate in such materials is dependent only onthe contamination level and not on the chemical compo-sition of the material.The neutron production rate was calculated as num-
Energy [MeV]0 1 2 3 4 5 6 7 8 9 10 ] M e V s N e u t r on p r odu c t i on r a t e [ k g SF,n) in PTFE α ( ,n) in copper α ( FIG. 1: Differential neutron production rates in ( α ,n) andspontaneous fission reactions in materials of the XENON100detector and its shield due to contamination of U, U,and
Th (assuming 1 Bq of
U and 1 Bq of
Th). PTFEis the material with the highest production rate among theXENON100 materials. The neutron production is dominatedby ( α ,n) reactions due to the low atomic mass Z of elementalconstituents. Copper is an example of a high Z material,where the neutron production rate is almost entirely due toSF reactions. Both, PTFE and copper, have been used tobuild the TPC field cage in the XENON100 detector. TABLE I: Neutron production rates for the materials used in the XENON100 experiment calculated with SOURCES-4Aas number of generated neutrons over number of disintegrating U/Th nuclei. The natural abundance of
U is taken intoaccount. The systematic uncertainty of the calculation is ±
17% [10]. Details on the detector and its components can be foundin Refs. [1, 9].Material Density Chemical composition Neutron production[g/cm ] U (incl. U) Th Cryostat and TPC (6.3 ± × − (1.0 ± × − Copper 8.92 Cu 100% (1.1 ± × − (3.6 ± × − Ceramics 1.00 NaAlSiO (1.1 ± × − (2.0 ± × − PMT parts
Kovar metal 8.33 Fe 55%, Ni 29%, Co 16%; (13 g/PMT) (1.2 ± × − (1.0 ± × − Stainless steel 7.64 Fe 71.8%, C 0.1%, Si 0.5%, Mn 0.7%, (1.5 ± × − (2.3 ± × − Ni 8.6%, Cr 18.3%; (7 g/PMT)Synthetic silica 2.20 SiO ; (2 g/PMT) (2.2 ± × − (2.1 ± × − Borosilicate glass 2.21 SiO O O ± × − (1.5 ± × − Li O 1.0%, Na O 6.0%, BaO 2.0%; (1 g/PMT)Aluminum 2.70 Al 100%; (0.1 g/PMT) (1.4 ± × − (2.8 ± × − Cirlex H N O ; (1.4 g/PMT base) (4.8 ± × − (2.4 ± × − Shield components
Polyethylene 0.92 CH (1.9 ± × − (1.4 ± × − Lead 11.34 Pb 100% (1.1 ± × − (3.0 ± × − Environment laboratory concrete 2.4 H 0.89%, C 7.99%, O 48.43%, Na 0.6%, Mg 0.85%, (1.9 ± × − (1.5 ± × − Al 0.9%, Si 3.86%, P 0.04%, S 0.16%, K 0.54%,Ca 34.06%, Ti 0.04%, Fe 0.43%TABLE II: Neutron production due to natural radioactivity in the stainless steel (type 316Ti), mainly used for the XENON100cryostat and its support bars. The radioactive contamination was measured with gamma- and mass-spectrometry (the mostsensitive result of the two methods is given) and shows that secular equilibrium in the chains is broken. U − Th Ra − Pb U − Pb Th − Ac Th − PbContamination [mBq/kg] 4.9 ± < ± < ± ± × − (3.1 ± × − (4.1 ± × − (1.8 ± × − (2.0 ± × − TABLE III: Neutron production rates in the XENON100 components due to ( α ,n) and SF reactions. Some detector components,such as the copper parts inside the cryostat, the TPC resistor chain, the bottom and top electrodes made of 316Ti SS, andPMT cables, are combined into ‘Additional detector parts’ due to their small contribution to the total neutron production.Radioactive contamination of the laboratory concrete is taken from Ref. [16], and the neutron production is calculated as aflux. Component Amount Contamination [mBq/kg] Neutron production U Th [neutrons/year]Cryostat and pumping ports (316Ti SS) 73.61 kg see Table II 16 ± ± ± ± ± ± × kg 0.083 0.012 3.2 ± × kg 0.23 0.094 37 ± × kg 0.66 0.55 162 ± × kg 4.20 0.52 (4.3 ± × LNGS concrete 2.6 × × (8.7 ± × − cm − s − ber of generated neutrons over number of disintegratingU/Th nuclei in a given material, with the contamina-tion of U computed from the measured contaminationof
U, assuming a natural abundance of 0.72%. Thesimulated neutron spectra for some of the materials areshown in Fig. 1, and the neutron production rates for allmaterials are presented in Table I.In the 316Ti stainless steel (SS) used in XENON100,the secular equilibrium is broken in the
U and
Thdecay chains. This was established by measuring theintrinsic radioactive contamination by inductively cou-pled plasma mass spectrometry (ICP-MS), performed inaddition to γ -spectrometry with germanium detectors.Hence, the neutron production in this material was cal-culated for the different parts of the chains separately: U − Th and Ra − Pb, Th − Ac and
Th- − Pb, and U − Pb. The results are presented inTable II. Ignoring the disequilibrium, the neutron back-ground would be underestimated.The total neutron production rates were calculatedby scaling the results of SOURCES-4A to the mass ofthe components in the detector and shield and to theirmeasured radioactive contamination, using the massmodel and the radioactive screening results introducedin Refs. [9, 18]. The results are presented in Table III.The detector components with the highest total neu-tron production rates are the lead and polyethylene in theshield, and the detector cryostat and support bars madefrom 316Ti SS. Neutron production in the TPC resistorchain is negligible, due to the small mass involved. Eventhough the neutron production rate in the aluminum ofthe PMTs is relatively high, its contribution to the totalneutron production in the PMTs is negligible due to thevery low amount of material (0.1 g per PMT), since it isused only as strips deposited on the window in order toimprove the resistivity of the photocathode at cryogenictemperature.Neutron production due to natural radioactivity in theconcrete walls of LNGS was calculated using the mea-sured chemical composition of Ref. [19]. Radiogenic pro-duction in the rock has been ignored in the present study,since our simulation showed that almost all neutronswhich originate in the rock are absorbed by the concreteshell. The results of our simulations agree well with themeasurements of neutron flux summarized in the samepaper. In particular, the discrepancy with the valuesmeasured by [20, 21] is less than 15%.The neutron energy spectra calculated fromSOURCES-4A and the total production rates wereused as an input for Monte Carlo simulations to predictthe neutron-induced NR background in the XENON100experiment. The neutron propagation was performedwith GEANT4.9.3.p02, using the neutron data filesG4NDL 3.13 with thermal cross sections, which arebased on the ENDF/B-VI/B-VII databases [22]. Foreach material and neutron source, 1 million events weresimulated, resulting in a negligible statistical uncertaintyof ∼ Xewith T / = 8.9 days, and Xe with T / = 11.8 days.These events have a signature of a prompt NR, followedby an ER produced by a γ from de-excitation of themetastable state. However, by computing the ratio ofthe cross sections for elastic and inelastic neutron inter-actions with xenon nuclei, restricted to these particularnuclear levels, we estimated that the contribution of theseevents to the total NR rate is < ∼ L eff described in Ref. [8]. Due to a change in themeasured value of the LXe response to 122 keV ee gammarays, used to normalize the scale, the energy ranges usedfor the 2011 and 2012 results are slightly different, andcorrespond to (8.4 − nr and (6.6 − nr ,respectively.The spatial distribution of single scatter NRs in theenergy region of interest for a WIMP search is shown inFig. 4. The radiogenic NR background was predictedfor the entire 62 kg target volume and for two fidu-cial volumes, 48 kg and 34 kg, which were used in theanalyses published in 2011 [8] and 2012 [3], respectively.Fiducialization is less efficient for reducing the NR back-ground because of the longer mean free path of neutrons( ∼
14 cm for 1 MeV), compared to γ -rays of similar en-ergy ( ∼ γ -ray, for example produced by an inelastic neutron in-teraction. ERs and NRs in the veto volume cannot be ] nr Energy [keV0 10 20 30 40 50 ] n r ke V d ay R a t e [ eve n t s k g single scattersdouble scattersall multiple scatters FIG. 2: Energy spectra of NRs in 34 kg fiducial volume fromneutrons produced in ( α ,n) and spontaneous fission reactions.The total energy deposited in multiple scatter interactions ison average higher than that of single scatter events. ] ee Energy threshold in the veto volume [keV1 10 F r ac t i on o f r e m a i n i ng eve n t s [ % ] FIG. 3: Efficiency of the veto coincidence cut as a functionof the energy threshold in the veto volume. A veto cut withthe measured volume-averaged energy threshold of 100 keV ee (shown as a vertical dashed line) provides a ∼
25% reductionof the single scatter NR rate. distinguished through the ratio of scintillation and ion-ization signals, as it is done within the TPC. Hence, en-ergy depositions from all interactions in the veto volumeare summed up, taking into account the light quench-ing for NRs [8]. As shown in Fig. 3, by applying a cutwith the measured volume-averaged energy threshold of100 keV ee [1, 9], the NR background can be reduced by ∼
25% with respect to the ‘passive veto’ mode, when onlyself-shielding of the external LXe layer is taken into ac-count.The predicted NR background rates in the WIMPsearch region from radiogenic neutrons are presented inTable IV. The contribution of the different componentsto the total background is shown in Table V. The rela-tive contributions do not significantly change by applyingfiducial volume and veto coincidence cuts. The dominant ] Area [cm0 100 200 300 400 500 600 700 Z [ c m ]
30 25 20 15 10 50 ] ke V d ay R a t e [ eve n t s k g × FIG. 4: Spatial distribution of single scatter NRs producedby radiogenic neutrons in the energy range of interest for theWIMP search. The solid (dashed) line shows the 48 kg (34 kg)fiducial volume boundary. The lower density at the edge isdue to the specific shape of the TPC defined by interlockingPTFE panels. This leads to a smaller active LXe volumerepresented by the last radial bins. part of the background comes from the 316Ti SS compo-nents, PMTs, and the PTFE parts of the TPC. This wasexpected due to the rather high neutron production ratesin these components and their location close to the liq-uid xenon target. Despite the high neutron productionrates in the lead shield, the contribution of this sourceto the total NR background rate is negligible, since it islocated outside of the polyethylene neutron shield. Thecontribution from radiogenic neutrons originating fromthe concrete lining of the laboratory’s cavern results in(6.6 ± × − events/year in the entire target volume(62 kg of LXe) in the energy range of interest for theWIMP search, even without using a veto coincidence cut,and hence can be considered negligible. III. MUON-INDUCED NEUTRONPRODUCTION
The cosmic muon flux underground at LNGS is re-duced by six orders of magnitude with respect to thevalue measured at the surface, due to the 3.6 km-water-equivalent, obtained averaging over the muon arrival di-rection, of overburden rock [23]. High-energy muons pen-etrating into the underground laboratory produce neu-trons through photo-nuclear reactions in electromagneticshowers, in deep inelastic muon-nucleus interactions, andin several secondary processes ( π -n, π -absorption, p-n,etc.) [25, 26]. The deeper the experimental site, thehigher the mean muon energy and hence the average en-ergy of muon-induced neutrons. Moreover, neutron pro-duction due to negative muon capture, which is relevantfor low energy stopping muons, becomes negligible. Theenergy of muon-induced neutrons extends up to a few TABLE IV: Predicted NR background rate in the given energy range from neutrons produced in ( α ,n) and SF reactions dueto natural radioactivity in the detector and shield components. The statistical error of the GEANT4 simulation is ∼ ee in the LXe active veto. Only single scatter NRs are relevant asbackground for WIMP searches. The acceptance to NRs, while relevant for the WIMP search, is not taken into account here,but in Section IV. The multiplicity of neutron interactions is indicated with the parameter n .Predicted background rate [year − ]Target volume 62 kg 48 kg 34 kgEnergy range 8.4 − nr − nr − nr Veto passive active passive active passive activesingle scatter events ( n =1) 0.49 ± ± ± ± ± ± n =2) 0.46 ± ± ± ± ± ± n >
1) 1.19 ± ± ± ± ± ± n >
0) 1.69 ± ± ± ± ± ± GeV, hence the hydrocarbon and water neutron shield,as employed in XENON100, cannot fully moderate andcapture them.In order to simulate the muon-induced neutron back-ground, the GEANT4 model introduced in Ref. [9] hasbeen extended by including the rock and concrete liningof the underground site at LNGS, taking into accounta rock thickness of 5 m. The Gran Sasso rock is com-posed mainly of CaCO and MgCO , and has an averagedensity of (2.71 ± [27]. The location of theexperiment, situated in a cavity at the corner of the in-terferometer tunnel at LNGS, has been described with asimplified geometry. The model is shown in Fig. 5, to-gether with an example of a muon interaction in the rockof the laboratory, which generates two neutrons and anelectromagnetic shower: one of the neutrons is stoppedby the passive shield, while the other one penetrates intothe target volume.An average muon flux of 1.2 h − m − was assumedfor the simulations, as measured by MACRO [28] andLVD [29]. The seasonal variation of the muon inten-sity [31], being smaller than 2%, is neglected. The muonenergy spectrum and angular distribution were simulatedwith the MUSIC-MUSUN package [11], taking into ac-count the depth of the experimental hall and the de-tails of the mountain profile; the results are shown inFig. 6. The average muon energy is 273 GeV, which isin good agreement with the results in Ref. [24] and themeasurements reported in Ref. [28]. The azimuthal andzenith distributions agree well with the measurement ofthe LVD experiment [23, 29]; there is agreement also withthe predictions done with FLUKA and the measurementsby MACRO and Borexino [30]. Most muon trajectorieshave zenith angles < ◦ . The µ + / µ − ratio is assumed tobe 1.4, as shown by recent observations for high energymuons [32, 33].The propagation of the high energy muons was per-formed with GEANT4.9.3.p02 using the QGSP BIC HPphysics list [34], which is based on a quark gluon stringmodel for high energy hadronic interactions [35], andwith a data-driven high-precision neutron package totransport neutrons below 20 MeV down to thermal ener- Muon energy [GeV]0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 ] G e V s R a t e [ c m (a) energy spectrum ] ° Azimuth [0 50 100 150 200 250 300 350 ] ° Z e n i t h [ E ve n t s (b) angular spectrum FIG. 6: Energy and angular spectra of the muons under-ground at LNGS from simulations with MUSUN. The aver-age muon energy is 273 GeV, and most of the muons have azenith angle < ◦ . gies. For primary protons and neutrons with energiesbelow 10 GeV, the GEANT4 binary cascade is used,which describes production of secondary particles in in-teractions with nuclei [34] more accurately than otherGEANT4 models. The direct interaction between muonsand nuclei is modeled with the G4MuNuclearInteractionprocess [36], which describes it by producing virtual pho-tons and treating them as a combination of π + and π − interactions.About 300 million muons were simulated, correspond-ing to about 185 years of livetime. This results in a sta-tistical uncertainty of ∼
10% on the background predic-tion. Information from the literature was used in or-der to evaluate the systematic uncertainty of the sim-ulations. The muon-induced neutron production withGEANT4 was validated in Ref. [37] via comparison withmeasured data from the NA55 experiment, resulting ina factor of ∼ ∼ ± ee rejectsmore than half of the single scatter NRs in the target vol-ume. IV. TOTAL PREDICTED NUCLEAR RECOILBACKGROUND
The Monte Carlo simulations of the NR backgroundpresented above assume 100% detection efficiency. Themeasured trigger efficiency is >
99% at 300 (150) PE inthe 2011 (2012) WIMP search results and rolls off atlower values [1], which reduces the detection efficiency.In addition, the acceptance for single scatter events is
TABLE VI: Predicted NR background rate due to muon-induced neutrons. The active veto coincidence cut assumes a volume-averaged energy threshold of 100 keV ee . The statistical error of the GEANT4 simulation is ∼ n . No deficit of acceptance for NRs is considered yet (see Section IV).Predicted background rate [year − ]Target volume 62 kg 48 kg 34 kgEnergy range 8.4 − nr − nr − nr Veto passive active passive active passive activesingle scatter events ( n =1) 2.2 +2 . − . +0 . − . +1 . − . +0 . − . +0 . − . +0 . − . double scatter events ( n =2) 1.8 +1 . − . +0 . − . +1 . − . +0 . − . +1 . − . +0 . − . multiple scatter events ( n >
1) 5.6 +5 . − . +2 . − . +4 . − . +1 . − . +3 . − . +1 . − . all events ( n >
0) 7.8 +7 . − . +2 . − . +5 . − . +2 . − . +4 . − . +1 . − . TABLE VII: Relative contribution from the detector andshield components to the muon-induced neutron background.It is given for all neutrons that produce NRs in the targetvolume (left column), and also for those neutrons that haveonly a true single scatter NR signature in the energy regionof interest (ROI).Component/Material Contribution [%]all NRs single scatterNRs in ROIRock and concrete < < < < finite, and is determined by the size of the proportionalscintillation (S2) signal. If one interaction of a doublescatter event generates an S2 which is below the thresh-old, such an event is mis-identified as a single scatterinteraction. Since in the measured data the energy of anevent is computed from the scintillation signal, the av-erage energy of these events is higher than that of truesingle scatter interactions. Electronic recoils are less af-fected by this detection efficiency, as the typical S2 sig-nals are much larger than the threshold value. In order toselect single scatter NR interactions in the Monte Carlodata, we apply cuts which are similar to the ones used inthe analysis of the measured data.The energy-dependent acceptance of these cuts wascalculated with Monte Carlo simulations, taking into ac-count the L eff parametrization [8]. The simulation re-sults cannot be validated via direct comparison with themeasured background data due to the very low rate ofNRs. Hence, the results were verified by comparing theratio of single and double scatter events using measured AmBe neutron calibration data and a corresponding ] ee Energy threshold in the veto volume [keV1 10 F r ac t i on o f r e m a i n i ng eve n t s [ % ] FIG. 7: Efficiency of the veto coincidence cut as a function ofthe energy deposited in the veto volume for reduction of thecosmogenic neutron background. A veto cut with the mea-sured volume-averaged energy threshold of 100 keV ee providesa ∼
60% reduction of the single scatter NR rate.
Monte Carlo simulation, showing excellent agreementwithin the known uncertainties of L eff and the sourcestrength [42].The total NR background rates predicted for the2011 [7] and the 2012 [8] data analyses, taking into ac-count the detection efficiency, the energy range, and theexposure, are given in Table VIII. The contribution fromenvironmental radioactivity (contamination in the rockand concrete of the LNGS laboratory) is at the percentlevel and has no significant impact on the XENON100 sci-ence goals. About 70% comes from muon-induced neu-trons. In future experiments, such as XENON1T [44],this background will be significantly reduced by using amuon veto system. The energy spectra of the backgroundfrom radiogenic and cosmogenic neutrons predicted forthe results published in 2012 are shown in Fig. 8(a). Fig-ure 8(b) presents the total NR background expected inXENON100, converted to the S1 PE energy scale andcorrected for the acceptance due to the analysis cuts,taking into account fluctuations of the light signal forevents below the threshold. V. CONCLUSIONS
The NR background in the XENON100 experiment,originating from radiogenic and cosmogenic neutrons hasbeen predicted for the XENON100 experiment based onMonte Carlo studies, using a detailed model of the de-tector and its shield.The total NR background in the 100.9 days dataset(2011, [8]), which used a fiducial mass of 48 kg and (8.4- ] nr Energy [keV0 10 20 30 40 50 60 70 80 90 100 ] n r ke V d ay R a t e [ eve n t s k g radiogenic BGmuon induced neutrons (a) raw Monte Carlo data Energy [PE]0 10 20 30 40 50 ] PE R a t e [ eve n t s d ay total neutron backgroundwith acceptance (b) converted to PE, with energy resolution and acceptance FIG. 8: Energy spectra of the NR background due to radio-genic and cosmogenic neutrons predicted for the 2012 WIMPsearch analysis (34 kg fiducial volume). The prediction takesinto account the measured trigger threshold which is the causeof the roll-off at the lowest energies. The bottom plot showsthe total NR background on the energy scale converted toPE, with the energy resolution and acceptance of the analy-sis cuts taken into account. Fluctuations of the light signalfor events below the threshold have also been included in thisanalysis. The thin vertical dashed lines indicate the WIMPsearch energy region. TABLE VIII: Nuclear recoil backgrounds predicted for the2011 [8] and the 2012 [3] WIMP searches, taking into accountthe detection efficiency. The NR acceptance after applying anER discrimination cut based on S2/S1 ratio is not applied asit is not used in the standard Profile Likelihood analysis [43].
Data release 2011 [8] 2012 [3]Live time 100.9 days 224.6 daysFiducial volume 48 kg 34 kgEnergy range 8.4 − nr − nr (4 −
30 PE) (3 −
30 PE)Radiogenic neutrons 0.10 ± ± +0 . − . events 0.34 +0 . − . eventsTotal NR background 0.31 +0 . − . events 0.48 +0 . − . events − nr energy range, is (0.31 +0 . − . ) events. Thedetector’s energy resolution as well as the active vetoefficiency are taken into account here. The prediction forthe 224.6 days data used for the WIMP search results of2012 [3] is (0.48 +0 . − . ) events. A fiducial target mass of34 kg and an energy range of (6.6 − nr were usedfor this search.With the reduced NR acceptance after applying theS2/S1 electronic recoil discrimination cut to define abenchmark WIMP search region, these values translateinto (0.11 +0 . − . ) events, and (0.17 +0 . − . ) events, respec-tively. Compared to the total ER background esti-mates of (1.7 ± ± − nr energy range and 48 kg fiducial target, thecontributions from radiogenic and muon-induced neu-trons add up to (0.12 +0 . − . ) background events/year inthe full discrimination space. In a 34 kg target it is only(0.07 +0 . − . ) events/year, which is about 10% of the back-ground in the energy range below 30 PE. VI. ACKNOWLEDGEMENTS
We gratefully acknowledge support from NSF, DOE,SNF, UZH, FCT, R´egion des Pays de la Loire, STCSM,NSFC, DFG, Stichting voor Fundamenteel Onderzoekder Materie (FOM), the Max Planck Society and theWeizmann Institute of Science.We are grateful to LNGS for hosting and supportingXENON100.0 [1] E. Aprile et al. (XENON100 Collaboration), Astropart.Phys. :9, 573590 (2012).[2] E. Aprile et al. (XENON100 Collaboration),arXiv:1207.3458 (2012).[3] E. Aprile et al. (XENON100 Collaboration), Phys. Rev.Lett. et al. (XENON100 Collaboration), Phys. Rev.Lett. , 021301 (2013).[5] J.D. Lewin and P.F. Smith, Astropart. Phys.
87 (1996);G. Jungman, M. Kamionkowski, and K. Griest, Phys.Rept.
195 (2006).[6] E. Aprile, T. Doke, Rev. Mod. Phys. , 20532097(2010).[7] E. Aprile et al. (XENON100 Collaboration), Phys. Rev.Lett. et al. (XENON100 Collaboration), Phys. Rev.Lett. et al. (XENON100 Collaboration), Phys. Rev.D , 082001 (2011).[10] W.B. Wilson et al. , SOURCES-4A, Technical Report LA-13539-MS, Los Alamos (1999).[11] V. Kudryavtsev, Comp. Phys. Comm.,
339 (2009).[12] J. Sulkimo et al.
GEANT4 Collaboration, Nucl. Instrum.Methods Phys. Res. Sect. A
250 (2003).[13] A. Kish, PhD thesis, University of Zurich, Switzerland(2011), http://opac.nebis.ch/ediss/20121322.pdf.[14] M. Carson et al. , Astropart. Phys.
667 (2004).[15] D. Mei, C. Zhang, and A. Hime, Nucl. Instr. Meth. Phys.Res. Sec. A
651 (2009).[16] M. Haffke et al. , Nucl. Instr. Meth. Phys. Res. Sec. A , 36 (2011).[17] S.Y.F. Chu, L.P. Ekstroem, and R.B. Fire-stone. The Lund/LBNL Nuclear Data Search.http://nucleardata.nuclear.lu.se/nucleardata/toi/.[18] E. Aprile et al. (XENON100 Collaboration), Astropart.Phys. 35, 34-49 (2011) .[19] H. Wulandari et al. , Astropart. Phys. et al. , Nucl. Instrum. Methods Phys. Res.Sect. A
203 (1989).[21] P. Belli et al. , Il Nuovo Cim.
959 (1989).[22] http://geant4.web.cern.ch/geant4/support/datafiles origin.shtml. [23] M. Aglietta et al. (LVD Collaboration), Phys. Rev. D , 688 (2003).[26] H.M. Araujo et al. , Nucl. Instrum. Meth. Phys. Res. Sect.A
398 (2005).[27] P.G. Catalano et al. , Mem. Soc. Geol. It.
647 (1986).[28] M. Ambrosio et al. (MACRO Collaboration), Phys. Rev.D et al. (LVD Collaboration), Phys. Rev. D et al. , arXiv:1210.2708 (2012).[31] M. Selvi et al. (LVD Collaboration), Proceedings of the31 st ICRC, 1–4 (2009).[32] P. Adamson et al. (MINOS Collaboration), Phys. Rev. D et al. (OPERA Collaboration), Eur. Phys.J. C et al. , IEEE Trans. Nucl. Sc. et al. , Nucl. Instr. Meth. Phys. Res. Sect.A et al. (ZEPLIN-II Collaboration), Astropart.Phys. et al. , Astropart. Phys. et al. (KamLAND Collaboration), Phys. Rev. C et al. (ZEPLIN-III Collaboration),arXiv:1302.4275 (2013).[42] E. Aprile et al. (XENON100 Collaboration), Phys. Rev.D , 012006 (2013).[43] E. Aprile et al. (XENON100 Collaboration), Phys. Rev.D84