IMPROVED CONSTRAINTS ON WIMPS FROM THE INTERNATIONAL GERMANIUM EXPERIMENT IGEX 2 0 0 2 n a A. Moralesa1, C.E. Aalsethb, F.T. Avignone IIIb, R.L. Brodzinskic, S. Cebri´ana, E. Garc´ıaa, J I.G. Irastorzaa2, I.V. Kirpichnikovd, A.A. Klimenkoe, H.S. Mileyc, 6 J. Moralesa, A. Ortiz de Solo´rzanoa, S.B. Osetrove, V.S. Pogosovf, J. Puimed´ona, J.H. Reevesc, 1 M.L. Sarsaa, A.A. Smolnikove, A.G. Tamanyanf, A.A. Vasenkoe, S.I. Vasilieve, J.A. Villara 2 v aLaboratory of Nuclear and High Energy Physics, University of Zaragoza, 50009 Zaragoza, Spain 1 bUniversity of South Carolina, Columbia, South Carolina 29208 USA 6 cPacific Northwest National Laboratory, Richland, Washington 99352 USA 0 dInstitute for Theoretical and Experimental Physics, 117 259 Moscow, Russia 0 eInstitute for Nuclear Research, Baksan Neutrino Observatory, 361 609 Neutrino, Russia 1 1 fYerevan Physical Institute, 375 036 Yerevan, Armenia 0 / Abstract x e One IGEX 76Ge double-beta decay detector is currently operating in the Canfranc Underground - LaboratoryinasearchfordarkmatterWIMPs,throughtheGenuclearrecoilproducedbytheWIMP p elasticscattering. Anewexclusionplot,σ(m),hasbeenderivedforWIMP-nucleonspin-independent e h interactions. Toobtainthisresult,40daysofdatafromtheIGEXdetector(energythresholdEthr 4 ∼ : keV), recently collected, have been analyzed. These data improve the exclusion limits derived from v all the other ionization germanium detectors in the mass region from 20 GeV to 200 GeV, where a i X WIMP supposedly responsible for the annual modulation effect reported by the DAMA experiment would be located. The new IGEX exclusion contour enters, by the first time, the DAMA region by r a using only raw data, with no background discrimination, and excludes its upper left part. It is also shown that with a moderate improvement of thedetector performances, theDAMA region could be fully explored. 1 Introduction Experimental observations and robust theoretical arguments have established that our universe is essentiallynon-visible,theluminousmatterscarcelyaccountingforonepercentofthecriticaldensity of aflatuniverse(Ω=1). Thecurrent prejudiceis thattheuniverseconsists ofunknownespecies of DarkEnergy(ΩΛ 70%)andDarkMatter(ΩM 25 30%)ofwhichlessthan 5%isofbaryonic ∼ ∼ − ∼ origin. MostofthatDarkMatterissupposedtobemadeofnon-baryonicparticlesfillingthegalactic halos, at least partially according to a variety of models. Weak Interacting Massive (and neutral) Particles (WIMPs) are favourite candidates to such non-baryonic components. The lightest stable particles of supersymmetric theories, like theneutralino, describe a particular class of WIMPs. WIMPs can be detected by measuring the nuclear recoil produced by their elastic scattering off targetnucleiinasuitabledetector[1]. Inparticular,non-relativistic( 300km/s)andheavy(10 103 ∼ − GeV)galactichaloWIMPscouldmakeaGenucleusrecoilwithafewkeV,ataratewhichdependson thetypeofWIMPandinteraction. Onlyabout1/4ofthisenergyisvisibleinthedetector. Becauseof thelowinteraction rate(whichrangesfrom 10to10−5 counts/kg/dayaccording totheSUSYmodel and thechoice of parameters) and thesmall energy deposition (from a few to 100 keV),thedirect ∼ 1Correspondingauthor: [email protected] 2Presentaddress: CERN,EPDivision,CH-1211Geneva23,Switzerland 1 search for particle dark matter through scattering off nuclear targets requires ultralow background detectors with very low energy thresholds. Germanium detectors have reached one of the lowest background levels of any type of detector andhaveareasonableionizationyield(nuclearrecoilionizationefficiencyrelativetothatofelectrons ofthesamekineticenergy)rangingfrom 20%to30% dependingonthenuclearrecoil energy. Thus, with sufficiently low energy thresholds, they are attractive devices for WIMP searches. That is the case for IGEX. This paper presents new WIMPs constraints in the cross-section WIMP-nucleon versus WIMP massplot,derivedfrom agermanium detector(enrichedupto86%in76Ge) oftheIGEXcollabora- tion,whichimprovepreviouslimitsobtainedwithGeionizationdetectors,andenterbythefirsttime theso-called DAMAregion (correspondingtoaWIMPsupposedlyresponsiblefortheannualmodu- lation effectfoundintheDAMAexperiment[2])withoutusingmechanismsofbackgroundrejection, but relying only in theultra-low background achieved. 2 Experiment The IGEX experiment [3, 4], optimized for detecting 76Ge double-beta decay, has been described in detail elsewhere. One of the IGEX detectors of 2.2 kg, enriched up to 86 % in 76Ge, is being used to look for WIMPs interacting coherently with the germanium nuclei. Its active mass is 2.0 kg, measured with a collimated source of 152Eu. The full-width at half-maximum (FWHM∼) energy resolution is2.37keVat the1333 keVlineof60Co. Energycalibration andresolution measurements were made periodically using the lines of 22Na and 60Co. Calibration for the low energy region was extrapolated using the X-ray lines of Pb. The uncertainty induced by this extrapolation in the determination oftheenergyvaluesinthethreshold region hasbeenestimated tobesmaller than0.1 keV,as deduced from the checkof linearity performed systematically –with this and other detectors of IGEX– along several years since the arrival of the detector to the underground facility, when the activation peaks (at about 10 keV) were still visible. The Ge detector and its cryostat were fabricated following state-of-the-art ultralow background techniques and using only selected radiopure material components (see Ref [3, 5]). The first-stage field-effecttransistor (FET) ofthedetectorismountedon aTeflon block afew centimetersfrom the central contact of the germanium crystal. The protective cover of the FET and the glass shell of thefeedbackresistorhavebeenremovedtoreduceradioactive background. Thisfirst-stageassembly is mounted behind a 2.5-cm-thick cylinder of archaeological lead to further reduce the background. Further stages of preamplification are located at the back of the cryostat cross arm, approximately 70cmfromthecrystal. TheIGEXdetectorshavepreamplifiersmodifiedforthepulse-shapeanalysis used in thedouble-beta decay searches. The detector shielding has been recently modified with respect to that of the previous set-up of Ref. [5],improvingtheexternalneutronshieldingandincreasingthethicknessofleadsurroundingthe detector. Theshieldingisnowasfollows: theinnermostshieldconsistsofabout2.5tonsof2000-year- old archaeological lead of ancient roman origin (having <9 mBq/kg of 210Pb(210Bi), <0.2 mBq/kg of238U,and<0.3mBq/kgof232Th)formingacubicblockof60cmside. Thedetectorisfittedintoa precision-machined chambermadein thiscentral core, which minimizes theemptyspacearound the detector available to radon. Nitrogen gas, at a rate of 140 l/hour, evaporating from liquid nitrogen, is forced into the small space left in the detector chamber to create a positive pressure and further minimize radon intrusion. The archaeological lead block is surrounded, at its turn, by 20 cms of lead bricks made from 70-year-old low-activity lead ( 10 tons) having 30 Bq/kg of 210Pb. The ∼ ∼ whole lead shielding forms a 1-m cube, the detector being surrounded by not less than 40-45 cm of lead (25 cm of which is archaeological). A 2-mm-thick cadmium sheet surrounds the main lead shield, and two layers of plastic seal this central assembly against radon intrusion. A cosmic muon veto covers the top and sides of the shield, except where the detector Dewar is located. The veto consistsofBICRONBC-408plasticscintillators5.08cm 50.8cm 101.6cmwithsurfacesfinished × × by diamond mill to optimize internal reflection. BC-800 (UVT) light guides on the ends taper to 5.08 cm in diameter over a length of 50.8 cm and are coupled to Hamamatsu R329 photomultiplier tubes. The anticoincidence veto signal is obtained from the logical OR of all photomultiplier tube discriminator outputs with a count rate lesser than 40 Hz (i. e. using a threshold which allows to include events with a poor light collection). An external neutron moderator 40 cm thick formed by 2 polyethylenebricksand borated water tanks completes theshield. The entireshield is supported by anironstructurerestingonnoise-isolationblocks. Theexperimenthasanoverburdenof2450m.w.e., which reducesthe muon fluxto a (measured) valueof 2 10−7cm−2s−1. × The data acquisition system for the low-energy region used in this WIMP search is based on standardNIMelectronics. Ithasbeenimplementedbysplittingthenormalpreamplifieroutputpulses ofthedetectorandroutingthemthroughtwoCanberra2020amplifiershavingdifferentshapingtimes enabling noise rejection as first applied in Ref [6]. The minimum settled ratio of thetwo amplitudes processed with different shaping time depends on the energy and in the present case it ranges from 0.8 (at 4keV)to0.99 (at 50keV).These amplifieroutputsare convertedusing200 MHzWilkinson- type Canberra analog-to-digital converters, controlled by a PC through parallel interfaces. For each event,thearrival time (with an accuracy of 100 µs), theelapsed timesince thelast vetoevent(with an accuracy of 20 µs), and the energy from each ADC are recorded. Figure 1 shows the time-after- last-veto distribution of the events. Notice that it reflects properly the 200 µs delay included in the main trigger of the acquisition system for the computer to decide whether it acquires or not the digitized pulse after knowing its energy. The muon veto anticoincidence was done off-line with a software window up to 240µs. The probability of rejecting non-coincident events is less than 0.01. Therejected veto-coincidenteventsamount uptoabout the5% of thetotalrateand aredistributed in thelow energy region as shown by theFigure 2. Inaddition,thepulseshapesofeacheventbeforeandafteramplificationarerecordedbytwo800 MHz LeCroy 9362 digital scopes. These are analyzed one by one by means of a method based on wavelet techniques which allows us to assess the probability of this pulse to have been produced by a random fluctuation of the baseline. The method requires the calculation of the wavelet transform of f(x),therecorded pulse shape after amplification: ∞ 1 x b [Wψf](a,b)= √aZ−∞f(x)ψ(cid:16) −a (cid:17)dx (1) wherethe”mexicanhat”waveletfunctionψ(x) (1 x2)exp 1x2 waschosenforourpurposes. ∝ − −2 Following expression (1), a two-parameter function [Wψf](a,b)(cid:0)was n(cid:1)umerically obtained for each event. The relative maxima of this function were calculated, the highest one corresponding to the eventpulseandtheotherstorandomfluctuationsofthebaseline. Itwasproventhatthedistribution of the values of the wavelet transform at these points follows an exponential. By comparing the maximum corresponding to the event pulse with this exponential one can calculate the probability P for the first maximum to belong to the distribution of the other maxima. This value is the final outputof theanalysis for each event and isinterpreted as theprobability of themain pulseof being randomly generated bythe fluctuationsof thebaseline. In order to fix the rejection criterion this method was applied to a calibration set of data. The resultisshowninFigure3wheretheprobabilityobtainedwiththismethodversusenergyispresented for each event. The same plot but for a background set of data is shown in Figure 4. From these plots a criterion of P < 0.01 can be defined to distinguish the two populations of noise and data. Although it is hard to quantify the loss of efficiency with the available statistics, the figures show that it is very small for eventsabove4 keV. It is worth mentioning that in spite of its good efficiency, this technique was not able, by itself alone,toimprovethepreviouslowenergybackgroundpresentedinRef. [5]. Therefore,weconcluded thatnoiseandmicrophonicsdoesnotcontributesubstantiallytosuchbackgroundand,consequently, thereductionofbackgroundpresentedinthenextsectionisattributedtothechangesintheshielding. 3 Results and prospects The results presented in this paper are from a recent run with the modified shielding and analysis system previously described. They correspond to an exposure of Mt=80 kg days. The spectrum obtainedisshowninFigure5comparedwiththepreviousIGEXpublishedspectrumofRef. [5]. The numerical data are also given in Table 1. The high energy region up to 3 MeV is shown in Figure 6. Theenergythresholdofthedetectoris4keVandtheFWHMenergyresolutionatthe75keVPb X-raylinewasof800eV.Thebackgroundraterecordedwas 0.21c/keV/kg/daybetween4–10keV, ∼ 0.10c/keV/kg/daybetween10–20keV,and 0.04c/keV/kg/daybetween25–40keV.Asitcanbe ∼ ∼ seen,thebackgroundbelow10keVhasbeensubstantiallyreduced(aboutafactor50%)withrespect 3 to that obtained in the previous set-up [5], essentially due to the improved shielding (both in lead andinpolyethylene-water). Aswasstressedbefore,thisreductionwasnotduetotheimplementation of the Pulse Shape Analysis, which suggests that the neutrons could be an important component of thelow energy background in IGEX. Theexclusionplotsarederivedfromtherecordedspectruminone-keVbinsfrom4keVto50keV, by requiring the predicted signal in an energy bin to be less than or equal to the (90% C.L.) upper limit of the (Poisson) recorded counts. The derivation of the interaction rate signal supposes that the WIMPs form an isotropic, isothermal, non-rotating halo of density ρ = 0.3 GeV/cm3, have a Maxwellianvelocitydistributionwithvrms=270km/s(withanuppercutcorrespondingtoanescape velocity of 650 km/s), and have a relative Earth-halo velocity of vr =230 km/s. The cross sections are normalized to the nucleon, assuming a dominant scalar interaction. The Helm parameterization [7]isusedforthescalarnucleonformfactor. TocomparetheIGEXexclusionplotswiththatderived from the Heidelberg-Moscow data [8], the recoil energy dependent ionization yield used is the same that in Ref. [8], Evis=0.14 (Erecoil)1.19. TheexclusionplotderivedinthiswayisshowninFig.7(thicksolidline). ItimprovestheIGEX- DM previous result (thick dashed line) as well as that of the other previous germanium ionization experiments(includingthelast resultofHeidelberg-Moscow experiment[8]–thickdottedline–)fora massrangefrom20GeVto200GeV,whichencompassthatoftheDAMAmassregion. Inparticular, thisnewIGEXresultexcludesWIMP-nucleoncross-sectionsabove7 10−9nbformassesof 50GeV × ∼ and enters the so-called DAMA region [2] where the DAMA experiment assigns a WIMP candidate to their found annual modulation signal. IGEX excludes the upper left part of this region. That is thefirst time that a direct search experiment without background discrimination, butwith verylow (raw) background, enters such region. Also shown for comparison are the contour lines of the other experiments,CDMS [9]andEDELWEISS [10](thindashedline), which haveenteredthat region, as wellastheDAMAregion (closed line)correspondingtothe3σ annualmodulation effectreportedby that experiment [2] and the exclusion plot obtained by DAMA NaI-0 (thin solid line) [11] by using statistical pulse shape discrimination. A remark is in order: for CDMS two contour lines have been depicted according to a recent recommendation [12], the exclusion plot published in Ref. [9] (thin dotted line) and theCDMS expected sensitivity contour [12] (thin dot-dashed line). Datacollectioniscurrentlyinprogressandsomestrategiesarebeingconsideredtofurtherreduce thelowenergybackground. Another50%reductionfrom4keVto10keV(whichcouldbereasonably expected)wouldallowtoexplorepracticallyalltheDAMAregionin1kgyofexposure. Inthecaseof reducingthebackgrounddowntotheflatlevelof0.04c/kg/keV/day(currentlyachievedbyIGEXfor energies beyond 20 keV), that region would be widely surpassed. In Figure 8 we plot the exclusions obtained with a flat background of 0.1 c/kg/keV/day (dot-dashed line) and of 0.04 c/kg/keV/day (solid line) down to the current 4 keV threshold, for an exposure of 1 kg year. As can be seen, the complete DAMA region (m=52+10 GeV, σp=(7.2+0.4)x10−9 nb) could be tested with a moderate −8 −0.9 improvement of theIGEX performances. A new experimental project on WIMP detection using larger masses of Germanium of natural isotopic abundance (GEDEON, GErmanium DEtectors in ONe cryostat) is planned. It will use the technology developed for the IGEX experiment and it would consist of a set of 1 kg germanium ∼ crystals,ofatotalmassofabout28kg,placedtogetherinacompactstructureinsideoneonlycryostat. Thisapproachcouldbenefitfromanticoincidencesbetweencrystalsandalowercomponents/detector massratiotofurtherreducethebackgroundwithrespecttoIGEX.Adetailedstudyisinprogressto assess the physics potential of this device, but it can be anticipated that a flat background of 0.002 c/kg/keV/day down to a threshold below 4 keV is a reasonable estimate. The exclusion plot which could be expected with such proviso for 24 kg y of exposure is shown in the Figure 8. Moreover, following the calculations presented in [13], GEDEON would be massive enough to search for the WIMP annual modulation effect and explore positively an important part of the WIMP parameter space including theDAMA region. Acknowledgements TheCanfrancAstroparticleUndergroundLaboratoryisoperatedbytheUniversityofZaragozaunder contract No. AEN99-1033. This research was funded by the Spanish Commission for Science and Technology (CICYT), the U.S. National Science Foundation, and the U.S. Department of Energy. 4 The isotopically enriched 76Ge was supplied by the Institute for Nuclear Research (INR), Moscow, and the Institutefor Theoretical and Experimental Physics (ITEP), Moscow. References [1] For a recent survey of WIMP detection, see for instance A. Morales, Nucl. Phys. B (Proc. Suppl.)87(2000)477[astro-ph/9912554] andReviewTalkattheTAUP2001Workshop,LNGS, (September2001), to appear in Nucl.Phys. B (Proc. Suppl.)2002 [astro-ph/0112550]. [2] R. Bernabei et al. [DAMA Collaboration], Phys. Lett. B 450 (1999) 448; R. Bernabei et al. [DAMA Collaboration], Phys. Lett.B 480 (2000) 23. [3] C. E. Aalseth et al. [IGEX Collaboration], Phys. Rev.C 59 (1999) 2108. [4] D. Gonzalez et al. [IGEX Collaboration], Nucl.Phys. Proc. Suppl. 87 (2000) 278. [5] A. Morales et al. [IGEX Collaboration], Phys. Lett.B 489 (2000) 268 [hep-ex/0002053]. [6] J. Morales et al.,Nucl. Instrum.Meth. A 321 (1992) 410. [7] J. Engel, Phys. Lett. B 264 (1991) 114. [8] L. Baudis et al.,Phys.Rev.D 59 (1999) 022001 [hep-ex/9811045]. [9] R.Abusaidi et al. [CDMS Collaboration], Phys. Rev.Lett. 84 (2000) 5699 [astro-ph/0002471]. [10] A. Benoit et al. [EDELWEISS Collaboration], Phys.Lett. B 513 (2001) 15 [astro-ph/0106094]. [11] R.Bernabei et al., Phys.Lett. B 389 (1996) 757. [12] B. Sadoulet, private communication to A. Morales and talk given at TAUP 2001 Workshop, LNGS (September2001). To be published in Nucl.Phys. B (Proc. Suppl.)2002. [13] S. Cebrian et al.,Astropart. Phys. 14 (2001) 339 [hep-ph/9912394]. 5 140 120 100 s t n e v 80 e f o r e 60 b m u n 40 20 0 0 0.1 0.2 0.3 0.4 0.5 time (ms) Figure 1: Distribution of the time after last veto event. The distribution is centered at ∼ 200µs as expected (see text). 6 40 35 30 25 s unt 20 o c 15 10 5 0 0 10 20 30 40 50 60 70 80 90 100 Energy (keV) Figure 2: Distribution at low energies of the events rejected by the veto system. 7 0 -1 ) -2 g o -3 l ( y t -4 i l i b -5 a b o -6 r p -7 m u -8 m xi -9 a m -10 t s 4 keV r-11 i F -12 -13 -14 -15 40 60 80 100 120 140 160 180 200 Energy (ADC channel) Figure 3: Scatter plot of the probability assigned to each event by the wavelet technique (described in the text) versus energy for a calibration set of data. 8 0 -1 ) -2 g o -3 l ( y t -4 i l i b -5 a b o -6 r p -7 m u -8 m xi -9 a m -10 4 keV t s r-11 i F -12 -13 -14 -15 40 60 80 100 120 140 160 180 200 Energy (ADC channel) Figure4: SameasFigure3butforbackgrounddata. Thepopulationsofnoiseanddataarewellseparated above 4 keV. 9 E (keV) counts E (keV) counts E (keV) counts 4.5 18 19.5 4 34.5 4 5.5 25 20.5 5 35.5 4 6.5 16 21.5 1 36.5 6 7.5 11 22.5 4 37.5 3 8.5 23 23.5 4 38.5 3 9.5 9 24.5 4 39.5 3 10.5 12 25.5 4 40.5 5 11.5 17 26.5 4 41.5 4 12.5 12 27.5 9 42.5 0 13.5 7 28.5 4 43.5 2 14.5 6 29.5 3 44.5 3 15.5 6 30.5 2 45.5 5 16.5 8 31.5 2 46.5 2 17.5 6 32.5 1 47.5 3 18.5 1 33.5 1 48.5 4 Table 1: Low-energy data from the IGEX RG-II detector (Mt = 80 kg d). 10