XAX: a multi-ton, multi-target detection system for dark matter, double beta decay and pp solar neutrinos K. Arisaka, H. Wang, P. F. Smith, D. Cline, A. Teymourian, E. Brown, W. Ooi, D. Aharoni, C. W. Lam, K. Lung, S. Davies, and M. Price Department of Physics and Astronomy, University of California, Los Angles, USA A multi-target detection system XAX, comprising concentric 10 ton targets of 136Xe and 129/131Xe, together with a geometrically similar or larger target of liquid Ar, is described. Each is configured as a two-phase scintillation/ionization TPC detector, enhanced by a full 4π array of ultra-low radioactivity Quartz Photon Intensifying Detectors (QUPIDs) replacing the conventional photomultipliers for detection of scintillation light. It is shown that background levels in XAX can be reduced to the level required for dark matter particle (WIMP) mass measurement at a 10−10 pb WIMP-nucleon cross section, with single-event sensitivity below 10−11 pb. The use of multiple target elements allows for confirmation of the A2 dependence of a coherent cross section, and the different Xe isotopes provide information on the spin-dependence of the dark matter interaction. 9 The event rates observed by Xe and Ar would modulate annually with opposite phases from each 0 other for WIMP mass >∼100 GeV/c2. The large target mass of 136Xe and high degree of back- 0 2 ground reduction allow neutrinoless double beta decay to be observed with lifetimes of 1027-1028 years, corresponding to the Majorana neutrino mass range 0.01-0.1 eV, the most likely range from n observed neutrino mass differences. The use of a 136Xe-depleted 129/131Xe target will also allow a measurement of the pp solar neutrino spectrum to a precision of 1-2%. J 7 1. OVERALL DESCRIPTION OF DETECTION ] SYSTEM h p - This paper investigates the properties of the proposed o multi-tondetectorXAX1 comprisedofthreenobleliquid r t targets. XAX is designed to detect nuclear recoils from s a dark matter, events from neutrinoless double beta de- [ cay, and pp solar neutrino scattering events. The target materials are: 3 v Target 1: Liquid Xenon depleted in 136Xe (enriched in 8 6 129/131Xe) for: 9 3 (i) dark matter interactions with nuclei of non-zero . spin. 8 0 (ii) pp solar neutrino interactions (via νe scattering). FIG. 1: Conceptual view of XAX detector layout. 8 0 Target 2: Liquid Xenon enriched in 136Xe for: : v (i) dark matter interactions with zero-spin nuclei. and2mheight: (a)10tonsliquid129/131Xe,surrounding i 10 tons of liquid Xe enriched in 136Xe, and (b) 10 tons X (ii) neutrinoless double beta decay. (or larger) liquid Ar in a separate vessel. Two cryostats r a Target 3: Liquid Argon depleted in 39Ar (or optionally arelocatedwithinasingleshieldingenclosurewhichpro- liquid Neon) for: videspurifiedwatershieldingthickerthan4metersinall directions. Events in the water are separately monitored (i) comparison of dark matter interactions in Xenon by photomultipliers to provide an active Cerenkov veto with lower-A dark matter target. against the residual underground muon flux. (ii) additional pp solar neutrino target (in the case of Ne). The expected energy spectra for the three scientific objectives (dark matter, double beta decay and pp so- ApossibleconfigurationissketchedinFig1,consisting lar neutrinos) are shown in Fig 2. For simplicity, nat- of two cylindrical, active target volumes of 2 m diameter ural Xe is assumed here as a target material. In addi- tion, 100 GeV/c2, 1 and 10 TeV/c2 WIMP masses with 10−8 pb (=10−44cm2) of nuclear cross section and two- neutrino/neutrinoless double beta decays with life times 1 Acronymfor“129/131Xenon−Argon−136Xenon”,pronounced of1022/1027 yearsareassumedandplottedtogetherwith “ZAX” pp/7Be/8B solar neutrino spectra. 2 FIG. 2: Expected energy spectrums of WIMPs, solar neutrinos, and double beta decays in case of natural Xe as a target material. 100GeV/c2,1and10TeV/c2 WIMPmassesofWIMPwith10−8pbnuclearcrosssection;2νββ/0νββ withlifetimes of 1022/1027 years are assumed. The observed energy is smeared out assuming a resolution of σ= √1.5% . E(MeV) Detection of dark matter events is based on the well- than that of the current best low-radioactive photomul- establishedtwo-phaseTPC(TimeProjectionChamber), tipliers. giving both primary (S1) and secondary (S2) scintilla- In addition to the ionization/scintillation signal dis- tion signals for each event, from which nuclear recoils crimination provided by a double phase TPC, the Ar- can be discriminated from gamma background with less gon portion of the detector will also employ pulse shape than 1% overlap [1, 2, 3]. This method currently reaches discrimination in order to improve the efficiency of de- a sensitivity of < 10−7 pb as shown in Fig 7. The pur- termining nuclear recoils versus 39Ar beta decay events. poseofthispaperistodemonstratethatafactorof1000 The difference between the two is based upon the singlet improvement on the existing WIMP sensitivity can be and triplet dimer states of liquid Argon, each of which achieved in XAX by the use of larger target masses with has a unique lifetime. Thus, the timing information of improved self-shielding and a new ultra-low background the pulse provides a method for identifying nuclear and photondetectiondeviceQUPIDsurroundingeachtarget electronic events. Further information can be found in volume in a 4π array for position sensitivity. The same [6]. scintillationphotondetectionsystemwillalsobecapable ofrecordingeventsat2479keVfromneutrinolessdouble beta decay in 136Xe, as well as the spectrum of 0-250 2. CONCEPTUAL DETECTOR DESIGN keV electron recoils from pp solar neutrino scattering in Xe. Aconceptualdetectordesignofasinglevesselisshown ThebasicdetectionprinciplesofthedoublephaseTPC inFig3witha4π photodetectorarray. Alargetranspar- have been fully demonstrated on a smaller scale in pre- entvesselfortheactivetargetmaterial(aswellasacylin- vious/ongoing projects with both Xe [4] and Ar [5]. In drical container for 136Xe) is made from low-radioactive thispaperwedescribe,foreachtypeofsignal,howback- Acrylic, similar to the heavy-water Acrylic vessel devel- grounds can be reduced in much larger target masses to oped by SNO [7]. The internal wall is painted with a thenewlowlevelsneeded. Themostimportantproblem, WaveLengthShifter(WLS,suchasTPB)toconvertthe commontothenobleliquiddetectors,isthereductionof UV scintillation photons to visible light, similar to the gammaandneutronbackgroundsfromthephotodetector technique developed by WARP [8] and DEAP/CLEAN array. Toachievethis,wearedevelopinganewdevice-a [9]. Furthermore, to provide uniform vertical electric Quartz Photon Intensifying Detector, or QUPID, which fields for drifting ionized electrons upwards, the inter- weanticipatewillhaveaU/Thimpuritylevel(andhence nal cylindrical walls are either painted by a transparent gamma and neutron emission) a factor of ∼ 100 lower resistive layer, such as ATO (Antimony-Tin-Oxide) as 3 FIG. 3: A cross-sectional view of one of the XAX detectors, FIG.4: Simulationofelectricfieldlinescreatedbyacurrent showingadouble-layercoppercryostat,whichcontainsnoble throughthetransparentresistivelayertodriftelectronsinliq- liquid (Xe, Ar or optionally Ne), surrounded by a ∼10 cm- uidXe/ArverticallytotheXe/Argas. Applicationofacon- thick acrylic vessel (with TPB coating) and closely packed stant current along the side wall produces an extremely uni- 3950 units of 3 inch diameter QUPIDs. formpotentialgradient,creatingperfectlyhorizontalequipo- tential planes. in Fig 4, or coated by field shaping electrodes created by 3. LOW BACKGROUND LIGHT DETECTION - finealuminiumpatterns. Theinnersurfaceofthebottom QUPID plateoftheAcrylicvesselsispaintedwithacombination ofWLSandathintransparentconductinglayer, suchas ExistingdarkmatterdetectorsusingliquidXeorliquid ITO (Indium-Tin-Oxide) or Platinum coating. Ar can remove most gamma and neutron backgrounds through shielding and veto techniques, but a residual The outside of the Acrylic vessel is completely sur- backgroundremains,inparticularfromphotomultipliers, rounded by a closely-spaced 4π array of low background which are located close to the target. This background photo detectors (QUPIDs). This 4π arrangement not arisesprincipallyfromUandThimpuritiesinthephoto- only provides high detection efficiency of primary scin- multipliercomponents,givingbothagammabackground tillation light (S1), but also measures its position with directly from decay of the U/Th chains, and a neutron an accuracy of 10-20% of the detector diameter, giving background from alpha-n production. Over the past 10 a 3-D position sensitivity better than 1% of the detector years, improvements in radiopurity have reduced these volume. From this point of view, XAX takes full advan- by two orders of magnitude to the level of 2×103 gam- tage of conventional single-phase detectors with 4π ge- mas/dayand2×10−3 neutrons/dayper5cmPMTsuch ometry (such as XMASS [10] and DEAP/CLEAN [11]) as Hamamatsu R8778 [3]. in terms of S1 signal detection. In the case of XAX, Tofurtherreducebackgroundbyanotherfactorof100 however, the benefits of double-phase detectors are also (paving the way for the ultra-low background 4π arrays realized. S2 signals, obtained from the TPC operation, of photon detectors required for XAX), a design for an can provide 3-D position reconstruction accurately, as is all-quartzHybridAvalanchePhotodiode(HAPD),called the case in conventional double-phase detectors (such as the QUPID (Quartz Photon Intensifying Detector) has ZEPLIN-II [1], XENON10 [2], and WARP [12]). What been proposed by K. Arisaka and H. Wang. Advantages makesXAXuniqueisthepossibilitytoimpose4-D(posi- of the HAPD over a conventional photomultiplier are tionandtime)coincidencebetweentheS1andS2signals summarized in [13] and technical details of the QUPID for additional reduction of background. can be found elsewhere [14]. Briefly, the key features of The whole structure is housed in a double-layer vac- the QUPID are shown in Fig 5. The whole structure uum cryostat made of a low radioactive metal such is made by ultra-pure synthetic fused silica except for as OFHC (Oxygen-Free High Conductivity) Copper as the very small avalanche photodiode (APD) at the cen- shown in Fig 3. It is then immersed in a large water ter and its readout leads. Photoelectrons are emitted tank as shown in Fig 1. by the hemispherical photocathode and accelerated by 4 FIG. 5: (a) 3D cut-view of a 3 inch diameter QUPID, (b) simulated electric field lines and electron trajectories. -10 kV is supplied on the photocathode and photoelectrons are focused onto the center of the APD which is set to a ground level. 10 kV to the APD, producing ∼ 2,000 secondary elec- M GeV and an average velocity v = 220km · s−1. D 0 trons per photoelectron by direct bombardment, which If dark matter particles are incident on a nucleus of are then avalanched within the APD by a further factor atomic number A, producing a recoil energy E keV, R ∼ 30, giving a total gain of ∼ 60,000. The electron tra- the differential energy spectrum of the event rate R jectory simulation in Fig 5(b) shows that photoelectrons (events/kg/day) is given by [15] are well-focused towards the central region of the APD. The unit, currently being developed by UCLA and (cid:18) (cid:19) (cid:20) (cid:21) HamamatsuPhotonics(Hamamatsu,Japan),hasanout- dR = c1R0 exp −c2ER F2(E ,A) (1) side diameter of 70 mm, with a photocathode diameter dE E r E r R R 0 0 of65mm. Thisall-quartzdesignshouldachievenomore tthroanns/1y0e-1a0r0freommitetaecdhginamdimviadsuaalnQdU0.P0I1D-.0N.1oteemtihtatetdnonerue-- where E0 = 0.5MDvc20 keV, r = (M4MD+DMMTT)2, MT = 0.932A, and F2 is a nuclear form factor correction dis- sistor/capacitor chain is required, so additional radioac- cussed in [15]. The numerical constants c and c de- tivity from these components is eliminated. The quan- 1 2 pendonthemotionoftheEarth-Sunsystemthroughthe tum efficiency is expected to be 30-35% (at 170-450 nm Galaxy;foradetectoratrestwithrespecttoanisotropic wavelength), with an anticipated yield of approximately dark matter flux, c = c = 1, whereas motion through 4 photoelectrons/keV for a 4π array surrounding a 2 m 1 2 the Galaxy gives fitted values c =0.75, c =0.56, with diameterdetectorvolume. Toensuregoodphotocathode 1 2 a small modulation tabulated in [15]. Fig 6 shows the dynamicrangewithsuperiorlinearityatthelowtemper- energy spectra derived from this formula for several tar- ature of liquid Ar and Ne, a special photocathode ma- get materials being considered for WIMP search. Here, terial is being developed by Hamamatsu Photonics and a cross section of 10−8 pb is assumed for five target ele- applied to the QUPID. Combined with this new photo- ments: liquid132Xe,40Ar,and20Ne(proposedinXAX), cathode, the intrinsically superior dynamic range of the and solid 73Ge and 28Si. HAPD permits an extremely wide range of signal detec- From Equation (1) the experimental measurement of, tionfrom∼5keVdarkmattersignalsto2.5MeVdouble or limits on, the differential rate dR gives a value of, or beta decay signals. dER limit on, the quantity R , which is defined as the total 0 rate for a stationary Earth. This is related to the total 4. GALACTIC DARK MATTER (WIMP) nuclear cross section σA[pb] by DETECTION R Dσ 0 = A (2) The experimental requirements for galactic particle r µA2 dark matter detection and identification follow from the expected differential energy spectrum of dark matter where µA = MMDD+MMTT is the reduced mass of the collid- (WIMP) interactions with nuclei. Direct search experi- ingparticlesandDisanumericalfactorequalto94.3for mentsarebasedontheassumptionofalocalmassdensity an assumed dark matter density of 0.3 GeV/cm3 [15]. 0.3 GeV/cm3 of hypothetical neutral particles of mass Final limits are expressed in terms of the WIMP- nu- 5 detector in XAX, resulting in a total mass of 100 tons. Such enlargement may be feasible since Ar is several or- ders of magnitude less expensive than Xe. B. Confirmation of A2 dependence of cross section Considerable effort will be made to eliminate back- groundeventswhichimitaterecoilsfromdarkmatter: for example neutron scattering, radon decay nucleus recoils, or spurious data events. However, unambiguous signal identification can be achieved only by candidate signals from targets of differing A, using the A2 dependence in Equation (3). By using liquid Ar as a second target ma- terial, the event rate (events/kg/day) will be reduced by an order of magnitude as shown in Fig 7. For identifi- cation purposes, it is sufficient that the target masses of FIG. 6: Differential energy spectra of various target masses Xe and Ar are similar (eg 10 tons each), though a larger assuming a WIMP mass of 100 GeV. At lower recoil ener- mass of Ar would provide an event rate large enough for gies, Xenon is the preferred detection target because of its anindependentcalculationoftheWIMPmass. Afurther increased event rate. option is to replace the Ar with Ne for additional confir- mation of the A2 effect (and to provide an additional pp solar neutrino target, see §7). One should note however cleon cross section σ [pb] given by W−N that in the case of liquid Ne, double phase operation is (cid:18)µ (cid:19)2 σ (cid:18)1(cid:19)2(cid:18)R (cid:19) not feasible because of the slow electron drift velocity. σ = 1 A ∼0.01 0 (3) W−N µ A2 A r A where µ (∼ 0.92 GeV) is the reduced mass for A=1 C. Determination of WIMP mass and nuclear 1 and M > 50 GeV. The coherence factor A2 applies in cross section D the(expected)caseofapredominantlyspin-independent interaction, but would be absent in the case of an inter- The cross section, predicted by the so-called focus action only with a nucleon of unpaired spin. pointinminimalsupersymmetry,isoforder10−8pb[16]. Todemonstratethepossibilitiesinamorequantitative Therefore, it is not unreasonable to assume this order of way, MC simulations have been performed and yielded cross section. If so, 100-1000 events can be detected by the results shown below. 10 ton-years of XAX operation, in particular, by the Xe target. Taking this high statistics observation, from the shape of the energy spectrum, the WIMP mass can be A. Discovery potential deduced directly. Fig 8 shows 1σ and 2σ uncertainties in determining With the improvements in background reduction pos- WIMPmassandcross-sectionwiththeXetargetinXAX siblewiththeXAXsystem,darkmattersensitivitydown in10ton-yearsofdatataking,assuming(a)10−8 pband to10−11 pb(=10−47cm2)isattainable(showninFig7). (b) 10−9 pb. Above 100 GeV/c2 WIMP mass, there is a This should be compared with the current cross section degeneracybetweenthemassandcrosssection,resulting limits(2008)atthelevel10−7 -10−8 pb[1, 2], andtheo- in the 45◦ slope in this figure. reticalexpectations(fromsupersymmetrytheory)extend to the level 10−10 pb and below [16]. Thus, XAX could cover the most favored phase space predicted by super- D. Investigating spin-dependence symmetry models, if the neutralino is indeed the dark matter particle. Based on the hypothesis that the dark matter particle The order of magnitude difference in sensitivity be- is the lightest particle of supersymmetry theory, its in- tween the Xe and Ar curves, shown in Fig 7, is princi- teraction may be a combination of spin-dependence and pally the A2 coherence factor in the WIMP mass range spin-independence,butthelatterisalwayslikelytodom- above100GeV/c2. However,below100GeV/c2,aneven inate the cross section because of the multiplying factor larger loss of sensitivity in Ar arises from its higher en- A2. Nevertheless there is interest in setting limits sepa- ergy threshold (∼30 keV) coming from large photoelec- rately for spin-dependent and spin-independent interac- tron statistics required by pulse discrimination. To com- tions. In the case of Xe this can be done by separating pensateforsuchaneffect, itwouldbebeneficialtomake thoseisotopeswithspinfromthosewithoutspin. Forex- theArdetectoranorderofmagnitudelargerthantheXe ample, natural Xe is 48% odd-A isotopes with spin 3/2 6 FIG.7: 90%ConfidenceLimitofWIMP-NucleoncrosssectionexpectedfromXeandArdetectorswithoneyeardatataking. As a reference, the current best limits set by XENON10 and CDMS-II are plotted together. FIG. 8: 1σ and 2σ uncertainties in determining WIMP mass and cross-section by Xe target in XAX in 10 ton-years of data taking. WIMP masses 20, 50, 100, 200, 500 GeV, and cross sections of (a) 10−8 pb, (b) 10−9 pb are assumed. Above 100 GeV/c2 WIMP mass, there is degeneracy between mass and cross section, resulting in the 45◦ slope in this figure. or 1/2 (129Xe: 26.4%, 131Xe: 21.2%), and 52% even- toenrichspecifically136Xe,whichisacandidateforneu- A isotopes with spin 0 (132Xe: 26.9%, 134Xe: 10.5%, trinoless double beta decay, one of the three objectives 136Xe: 8.9%). For a target enriched in the isotopes 132 of the XAX project. This target isotope thus has the and above, one could obtain a target consisting domi- potential to provide not only a 0νββ signal, but also a nantly of spin zero nuclei. An important special case is dark matter signal for a spin-zero nucleus. This signal 7 FIG.9: (a)JuneandDecemberdarkmatterenergyspectraplottedinadimensionlessform. Thecurvescrossatapproximately a recoil energy E ∼ 0.78E r so that for data to the left of this point the annual modulation is reversed. (b) Normalized R 0 summer/winterdifferenceasafunctionofparticlemass,forbothArandXetargets,with±1σ errorsfora10ton-yearrunning period assuming a cross section of 10−8 pb. will be compared with the signal rate from a Xe target ence as a function of particle mass for both Ar and Xe depleted in isotopes ≥132 in order to search for any rate targets, together with ±1σ errors for a 10 ton-year run- difference related to the nuclear spin. ning period assuming a cross section of 10−8 pb. The data energy range used is that which is currently typical forthetwotargetmaterials. Theplotsshowthefeasibil- ity of estimating particle mass from the summer-winter E. Observation of annual modulation event rates, and also the expected sign reversal above 100GeVmassforthetypicalenergyrangeofArdata. It Theannualmodulationeffect(arisingfromtheEarth’s remains the case that with either target, one would see orbital motion superimposed on the Sun’s galactic mo- both signs of summer-winter difference in an experiment tion) can in principle prove the signal to be correlated capableofcoveringthefullenergyrange,asisclearfrom with motion in the Galaxy. This requires stable opera- Fig 9(a). tionoveraperiodofyears,togetherwithfullshieldingof Toobserveastatisticallysignificanteffectover2years, varying neutron and radon backgrounds. The difference we would need observation of a total of 800 events, or in total rate between June and December would be no 400/year, which requires a total of 20 tons liquid Xe at more than 4%, but taking into account the variation in a 10−10 pb signal strength, or 2 tons at 10−9 pb. energy spectrum, a combined parameter would vary by up to 18% [17]. Fig9(a)showstheJuneandDecemberdarkmatteren- 5. BACKGROUND ESTIMATE ergyspectraplottedinadimensionlessform. Thecurves cross at approximately a recoil energy E ∼ 0.78E r so R 0 To achieve a reasonable event rate at WIMP interac- that for data to the left of this point the annual modula- tion cross sections down to < 1×10−10 pb, the target tionisreversed. Foraparticlemass∼100GeV,thecross mass has to be increased to the 10 ton scale proposed over point occurs at a recoil energy ∼ 8 keV for an Ar here, three orders of magnitude larger than the ∼ 10 kg target, and ∼ 32 keV for a Xe target. As a result, the scaleoftoday’sexperiments. Atthesametime,theback- majority of experimental data for Ar will be to the right groundrate/kg/dayhastobereducedbyafactorof1000 of the crossover, and the majority of experimental data to match the factor of 1000 lower signal rate. There are for Xe will be to the left of the crossover, as indicated four major classes of backgrounds to be considered: by the ranges shown on Fig 9(b). The sign of the annual modulation will then be opposite for the majority of Xe (a) radiation from photon detectors (primarily from data to that of the majority of Ar data. Thus extended U/Th decay chains). operation of both Xe and Ar targets, and observation of (b) pp solar neutrinos. both signs, would provide an interesting cross check on (c) impurities in the noble liquid such as Kr and Rn, or the annual modulation expectations, and an additional internal radioactivity such as 39Ar (in case of Ar) or estimate of the dark matter particle mass. 136Xe(incaseofXe,duetotwo-neutrinodoublebeta Fig 9(b) shows the normalized summer/winter differ- decays). 8 (d) radiation from outside the detector, such as low en- 2. Neutron backgrounds ergy gammas/neutrons from shielding materials and rocks,andhighenergygammas/neutronsinducedby To observe a positive signal at the 10−10 pb level, we cosmic ray muon interaction in rocks. requiretheneutronbackgroundtobereducedtoatleast an order of magnitude below the signal level (i.e. to In the following section, it is shown that the back- <10−5 events/kg/day). ground level can be reduced by more than a factor of The major neutron background arises from U/Th in 1,000, to the level of < 0.1 event per 10 ton-years, by photodetectorsanddetectorvesselsthroughalpha-npro- combining (a) an ultra-low radiation photon detector, duction. Simulations based on the activity in R8778 QUPID, (b) a 10 cm-thick active self-shielding cut by PMTs and assuming 0.1ppb U/Th in nearby materials Liquid Xe/Ar, (c) a so-called S2/S1 cut for electromag- giveneutronbackgroundratesintheregion10−4/kg/day netic backgrounds (i.e. gamma rays and electrons), and [18],butthisisreducedbyafactorof100afteramultiple (d) a multiple scattering cut in particular to reject neu- scatteringcutanda10cmouterlayerofself-shieldingin trons. the target. As discussed in §3, a further factor of 100 decrease in photodetector activity would result from the use of QUPIDs in place of PMTs. Thus the neutron background is reducible to well below the level required A. Radiation from photon detectors for the detection of events at 10−10 to 10−11 pb WIMP- nucleon cross section. 1. Gamma backgrounds Again, to investigate the effectiveness of QUPIDs, ex- tensive GEANT4 simulations for neutron interactions The pair of signals in a two-phase noble liquid TPC, have been performed, assuming radioactivity of 1 mBq referred to as S1 (direct scintillation in liquid) and S2 of U/Th per QUPID. The obtained energy spectra from (proportional scintillation from the ionization avalanche neutroninteractioninthefiducialregionofLiquidXeare in gas), currently produces a gamma rejection factor of plottedinFig11. Heretheself-shieldingcutsof0cm,10 greater than a hundred (at 50% nuclear recoil efficiency) cm,20cm,and30cmareappliedafteramulti-scattering [2]. cut. Once the 10 cm self-shielding cut is applied, the As stated in §3, the most difficult gamma background neutron backgound from QUPIDs becomes below 10−8 to remove is that from U/Th in the nearby photodetec- events/keV/kg/day, which is completely negligible. tors. For conventional PMTs such as the Hamamatsu Lastly, to illustrate the effectiveness of active self- R8778, estimated target rates in the dark matter energy shielding more graphically, interaction points of gamma range are ∼10/kg/day for a 10 ton detector array. This reduces to ∼ 10−3 /kg/day after a multiple scattering ray and neutron backgrounds from QUPIDs are plotted inFig12fora2mdiameterand2mhightargetvolume. cut(sincetheweaklyinteractingWIMPscanscatteronly All the plots are made by GEANT4 after imposing the once in this volume) and a 10 cm layer of self-shielding; energywindowoftheWIMPsignalrange,corresponding the background reduces by an additional factor of 100 from an S2/S1 cut to ∼ 10−5 /kg/day, just achieving to 100-year-long data taking. the background level for detection of a 10−10 pb WIMP- Fig 12(a) shows the gamma backgrounds from U/Th nucleon cross section, but still masking a signal at 10−11 decays after a multiple scattering cut and an S2/S1 cut inthecaseofXe. Figs12(b)and12(c)showtheneutron pb. backgroundfromU/Thdecaysafteramultiplescattering The development of QUPIDs would gain a factor of cut in the cases of Xe and Ar respectively. Remarkably, 100 in reduction of gamma emission, and thus guarantee oncetheselfshieldingcutof∼10cmthicknessisimposed, achievement of the backgrounds needed for measuring a there would be only a few background events even for signal at 10−10 - 10−11 pb WIMP-nucleon cross section. 100-year-long data taking. More quantitative analysis of To further investigate the effectiveness of QUPIDs, ex- the effectiveness of self shielding will be summarized at tensiveGEANT4simulationhasbeenperformed,assum- the end of the background section. ing conservative radioactivity of 1 mBq per QUPID and natural Xe. The obtained energy spectra from gamma rayinteractioninthefiducialregionofliquidXeareplot- ted in Fig 10. Here the self-shielding cuts of 0 cm, 5 cm B. Solar neutrino background and10cmareappliedafteramulti-scatteringcutandan S2/S1cut(whichisassumedhavearejectionefficiencyof The pp solar neutrino background (from ν-e scatter- 99%). Once the 10 cm self-shielding cut is applied, the ing) below 20 keV electron-equivalent energy is constant gamma ray background from QUPID is reduced down at 10−5 events/keV/kg/day. This is an irreducible back- to 10−8 events/keV/kg/day, which is far below the level ground. However,becausethisisanelectronrecoilback- of pp solar neutrinos (the pp solar neutrino spectrum, ground in the S1 channel, it can be reduced by a factor which would become a dominant background at lower of 100 from S2/S1 discrimination to the level of 10−7 cross section levels, will be described later). events/keV/kg/day(asshowninFig10),or5events/10- 9 FIG. 10: Expected energy spectrum of WIMP interactions, solar neutrinos, double beta decays, and gamma ray backgrounds from QUPID as a function of self-shielding cuts. Natural Xe is assumed and both the S2/S1 cut and the multi-scattering cut have been applied in advance. Note that the unit of energy in this figure is keVee (= electron equivalent energy). FIG. 11: Expected energy spectrum of WIMP, and neutron backgrounds from QUPID as a function of self-shielding cuts. NaturalXeisassumedandthemulti-scatteringcuthasbeenappliedinadvance. NotethatunitofenergyinthisfigureiskeVr (= nuclear recoil energy). 10 FIG. 12: Distribution of gamma ray and neutron backgrounds in 2 m diameter and 2 m high target volume. (a) 100 years worth of gamma single scatter background events after an S2/S1 cut in the case of Xe. (b) 100 years worth of neutron single scatter background events in the case of Xe. (c) The same neutron background as in (b) except for Ar. ton/year. Thisisafactoroftwobelowthesensitivitycor- reduce the 85Kr content to a negligible level. respondingto10−11 pbWIMP-nucleoncrosssection,but would preclude observation of cross sections any lower than that, except by subtraction and/or by improved S2/S1 discrimination (whose electric field-dependence is still under investigation). On the other hand, above 100 keV thisbackgroundbecomes observable intheS1 chan- 2. Rn background in Xe or Ar nel as a clear signal, which can be used by XAX to mea- sure accurately the solar pp neutrino flux, as discussed below in §7. Radon has a lifetime of ∼4 days, and hence does not survive in the stored xenon gas, but can cause back- ground problems in the Xe detectors [1, 2, 3] by con- C. Impurity or internal radioactivity of the noble taminating the target or gas systems during assembly liquid targets andtransferringthroughdepositionofitsdecayproducts (218Po, 214Po) on the detector walls where they further 1. Kr background in Xe decay by alpha emission. The alphas themselves are in the MeV range, and hence do not result in low energy Kr is present as an impurity in commercial Xe, with events if emitted into the liquid, but if emitted into the typically 50 ppb. This Kr contains ∼ 10−11 cosmogenic wall, the nucleus recoils into the liquid with (by momen- 85Kr, which emits a continuous beta decay spectrum tum conservation) <100 keV recoil energy and thus can with a maximum energy of 690 keV and a 10-year half- produce a signal similar to that of a recoil Xe nucleus. life. Thiswouldcontributealowenergybetabackground These events can be eliminated from the data set by us- of ∼1 event/keV/kg/day to the gamma population lim- ing the S2 position sensitivity to make a radial cut, but iting the dark matter sensitivity to 10−8 pb. The need thiscutremovesasignificantfractionofthetargetmass. to remove this, to reach lower cross sections, has led to The future solution must be to prevent radon from con- the development of reflux distillation columns which will taminating the target system, and work is in progress reduce the level of Kr in Xe by a factor of 1000 per pass to understand this problem and develop procedures to [19, 20]. Thus, using several passes, it will be possible to eliminate it.