Probing the Intrinsic Electron Recoil Rejection Power in Liquid Xenon for Dark Matter Searches

Transcription

1 Probing the Intrinsic Electron Recoil Rejection Power in Liquid Xenon for Dark Matter Searches Qing Lin, Kaixuan Ni* Shanghai Jiao Tong University Astroparticle Physics 4, Amsterdam, June 3-8 * nikx@sjtu.edu.cn

2 b Until now, the dominant background for liquid xenon DM experiments are electron recoils (ERs) Energy [kevnr] Radius [cm] Baudis et al., arxiv:39.74 XE -5 z [cm] Radius [cm] 3 wall corner wall face FIG. (color online). (Top) Event distribution in thegate discrimigrid nation parameter space log ðsb =SÞ, flattened by subtracting 5 the distribution s mean, as observed after unblinding using all cuts and a 34 kg fiducial volume (black squares). A lower analysis analysis 5 threshold of 6:6 kevnr (NR equivalent energy scale) is employed. The PL analysis uses an upperlux energy threshold of 43:3 kevnr (3 3 PE), and the benchmark WIMP search region is 5 to 3:5 kev (3 PE). The negligible impact of the limited nr S > 5 PE threshold cut is indicated by the dashed-dotted blue 3 cathode grid line,35 and the signal region is restricted by a lower border running =S discrimialong the NR quantile. hard S5 97% An additional b radiusdefines (cm) the benchmark WIMP nation cut at 99.75% ER rejection search region from above (dotted green line) but is only used to FIG.. Spatial distribution all events inwith cross-check the PL inference. Theof histogram red positionand gray corrected S in the range -3 phe from the 85.3 live-days indicates the NR band from the neutron calibration. Two events of WIMP search data. The cyan dashed line indicates the fall into the benchmark region where (: & :) are expected fiducial volume. The physical locations of the cathode and Rate in ROI [evts t- y-] Total - drift time (µs) decays of Xe, followed by decays of Kr and further reduced by noble gas purification, the sol traction of the -decay component. To dim background subtraction, and to extend the energ be observed beyond the 3 kev upper bound, on depleted in the 36 Xe isotope. The overall backgro in the same figure. Rate [evts t- y- kev-] log (S -.6 νββ Materials Rn 85 Kr pp+7be Energy [kev] For.future Xe experiments, Figure (Left):multi-ton Overall predicted backgrounder spect Table ) and internalfrom ( Kr, Rn,vb Xe)and contamina backgrounds Xe36 85 The solar expected background Kr decays (green, neutrinos will from be dominant (before 36 (black dashed,. µbq/kg) and from Xe -deca reaching the neutrino coherent the total neutrino signal (red, pp and 7 Be). (Right): background) -3scattering kev energy region as a function of fiducial liquid

3 The two-phase xenon detectors use S/S ratio to reject electron recoil background in the fiducial volume. WIMP, Neutron S S Nuclear Recoil Gamma, Beta Electron Recoil

4 lation (S=S) ratio, which is different for nucle e tested and defined, using low energy electron recoils in LXe. Figure shows the energy clear recoils from calibration data, as well ZEPLIN-II, arxiv:7858 XENON, PRL 8 of unmasked WIMP-search data. In this ret the data in terms of a spin-independent scattering cross section. Another Letter the spin-dependent interpretation of the. More details on the analysis and the ckgrounds will be reported elsewhere [7]. 4 electron-recoil events, in the a priori set week ending PHYSICAL REVIEW LETTERS f interest to 6.9(8) kev nuclear-recoil 8 JANUARY 8 PRL(4.5, 33 gy) for the WIMP search, were collected TABLE I. The software cut acceptance of nuclear recoils "c, dence of the logarithm of this ratio for electron recoils from NR ource. The number of events in the final the nuclear-recoil acceptance Anr, and the electron-recoil rejecthe 37 Cs gamma ray and for nuclear recoils from AmBe tion efficiency Rer for each of the seven energy bins (Enr in is aboutfast 4,.3 calibrations. times the statistics of the neutron The separation of the mean nuclear-recoil equivalent energy). The expected number of leaklogdetector s!s=s" valuesresponse between electron and nuclear recoils data. The to nuclear age events, Nleak, is based on Rer and the number of detected increases at lower energy. In addition, the width of the ined from hours band irradiation in situ, usingenergy. The events, Nevt, in each energy bin, for the 58.6 live days WIMPelectron-recoil is also smaller at lower search data, with 5.4 kg fiducial. Errors are the statistical e neutron source.of these two effects gives a better electron- uncertainty from the Gaussian fits on the electron-recoil combination recoil rejection efficiency at lower energies, al associated with each triggered event is reaching!log!s=s" distribution. 99.9%. The different ionization density and track structure FIG. (color online). log %S=S& as a function ofn ene the off-line By requiring "c Anr $ Rer ($3 ) Nevt Enr (kev) leak of low analysis. energy electrons and Xe ionsa incoinlxe resultelectron in a recoils (top) and nuclear recoils (bottom) from c %:7 at leastdifferent two PMTs, the efficiency of the Sfluctuations, recombination rate and associated :8$:4 3 :%: 99.9% $: va tion data. The colored lines are the mean log %S=S&%:3 might than explain the observed :7%:6 95 :3$: gorithmwhich is larger 99%, above behavior. a thresh$:9 theen-electron-recoil [upper (red)] and nuclear-recoil [lower The log!s=s" values increase with decreasing :%:9 83 :%: oelectrons (pe), or 4.5 kev nuclear-recoil $: $:5 ergy for both electron and nuclear recoils, within the bands. The region between the two %3:6 vertical dashed line ER :$: 9 :8%:7 $:4 gy. Theenergy S hardware trigger( threshold is window of interest kev nuclear-recoil energy window ( kev nuclear-recoil equivalent e %:8 %: : 33 :4 equivalent energy). The same behavior has been observed $:3 onding to about 4 electrons extracted from chosen for the WIMP search. An S software threshold$:4 of %:7 previously in small prototype detectors [4,5]. In our :3$: 38 :4%:5 $:4 h is the analysis, expectedwecharge from event with is also imposed (black lines). subtract theanenergy-dependent mean %: :$:9 374 :7%:9 98.5% ER rejection for band 99.3% $:7 Fig. 8. Calibration using from neutrons from an AmBe source (upper) and Compton!S=S" the electron-recoil to obtain log 6 scattered Co γ-rays (lower).nr calibrations were performed at a high trig~5% acceptance for These all events. After this band flattening,!log!s=s" Total 85 7:%:4 $: S/S energy, kevee S/S energy, kevee

5 of he of ent d al- nr ee b ee ee log (S /S) x,y,z corrected (S /S) x,y,z corrected log(sb/s)log x,y,z corrected b log(sb/s) x,y,z corrected b log (S /S)-ER mean considerably. The acceptance of most of the cuts is given ckvs the size of the measured S signal, used to infer the NR ent energy scale (vertical blue line in Fig. ). The S signal is he subject to large Poisson fluctuations due to the low number.6.3 ke.5 orlux, PRL 4 XENON, PRL (a) Tritium ER Calibration of quanta involved. Only the acceptance ofweek the S threshending.8 ER.4 E V I Eold Wcondition L E T TS E R>S5 PE is given vs the NOVEMBER S signal before gy kev.5 since the S Poisson smearing, S[PE]signal fluctuates indepen(a) Tritium ER Calibration kg dently from5 the S after the initial energy deposition cung kev Given the systematic uncertainties in the LXe light and 8,.8 en.5..8σ(b)(8%).5 charge yields and the resulting XENON response at. AmBe and Cf 5 NR Calibration ue he very low nuclear recoil.4 energies, we choose not to model in. a, deposits.8.4 below WIMP interactions kev with energy 3 kev nr.5 ed e is essentially With the mean Leff shown in Ref. [6], this (b) AmBe and Cf 5 NR Calibration. he he equivalent to neglecting upward fluctuations in S above 3 kev Xe m 99.75% ER rejection the-.6threshold of 3.5PE from energy deposits with S expec.4 nd n S x,y,z corrected (phe) tation values below PE. This approach results in a conng s servative upper limit for low mass WIMPs. For the central kev kev to dfigure : Left: Calibrations of4detector response i -. part of the WIMP mass range, its impact is <%. The ust ce ER (tritium, panel a) corrected and NR Cf, S x,y,z (phe) (AmBe and S x,y,z corrected (phe)is shown -. resulting acceptance of the S > 5 PE cut [] wo with the means (solid line) and ±.8 contours (d Energy [kevnr] in Fig. (red line). 9% d Figure : Left: Calibrations of detector response in the 8% kg fiducial volume. width (indicating band tails) is forthe presenta Radius [cm] ed d S [PE] 99.9%~99% ER rej the5high statistics tritium data, fits to simu ER (tritium, panela) and NR3 (AmBe and Cf, panel b) calibrations arewith depicted, ta, d. w/ 5% NR acpt. representing the lines). parameterizations forward 99% ers d with the means (solid line) and ±.8 contours (dashed This choicetaken of band.8-5 d eir ER plot also shows the anr band fits mean width (indicating % band tails) is for The presentation only. Panel shows to and v.6 as his energies using an shown S S combined s the high statistics tritium data, with fitsconstant to simulated NR data in panel energy b, %~% NR acceptance 99.9% d ys. dot-dashed line delineates the approx representing the parameterizations taken The forward to themagenta profile likelihood analysis.. se ere -5 cut.and Right: WIMP region. Events in The ER plot also shows the NR band mean vicelux versa. Graysignal contours indicate oper live-day exposure are shown. the Lines99.99% are a energy - constant energies using an S S combined scale (same contours on each plot). utof Energy [kevnr] nr discrimination z [cm]acceptance leakage fraction nr dashed cyan lines phe range S showing x,y,z corrected (phe) The dot-dashed magenta line delineates the approximate location ofthe the 3 minimum S used -5 FIG. (color online). Combined cut acceptance (solid blue cut. Right: LUX WIMP signal Events in fraction the 8 kg fiducial volume during FIG.region.. Plot showing the leakage (discrimination level) between electron and nuclear recoil pop line). The S threshold cut S > 5 PE (dashed red line) is acceptance (as calculated from flat-in-energy NR simulations), measured with the high-stat nuclear recoil fluctuations live-day Lines are as shown on (black the circles) left,andwith vertical Weshown. show the leakage from counting events in the dataset from projections of Gaussian independent ofthe possible inexposure S and hasare to be applied Result: Dark Matter WIM recoil population (red squares). An upper limit is shown for S bins without events. The blue dashed line

6 The Question: Is 99.9% ER rejection (5% NR acceptance) the fundamental limit for liquid xenon?

7 ZEPLIN III: achieved % ER rejection power, the best reported for LXe DM detectors..8 arxiv: WIMP-nucleon (SI) cross-section, pb.4 log(s/s) : ER rejection - ) energy (S channel), kevee

8 Field! (kv/cm)! Light yield (pe/kevee, for kev at zero field)! Energy ROI! (kevnr)! NR acceptance! ER rejection power! ZEPLIN-II!.!.! 4-58! 5%! 98.5%! XENON!.73! 5.4! 4.5~6.9! 45%~49%! 99.9~99.3%! ZEPLIN-III! 3.4! 3.-4.! 7-35! ~5%! %! XENON!.53! 3.8! ! 6%~%! 99.75%! LUX % 99.9~99% What is the key factor that determines the ER rejection power? Stronger field? Higher light yield? Something else?

9 Unfortunately it s difficult to answer these questions with the large Xe dark matter detectors.

10 energy as low as.4 kev and Sr for 56.5 kev rec s stable to within!m over two months. The xenon of the to ratio S=S is plotted as a function of nuclear recoil cali [4]. Other experiments [5 8] have measured L s purified with a high temperature commercial getter ef energy. In both detectors, the elastic nuclear recoil band is other tors higher and lower energies. The Columbia and d nimize trapping of the drifting electrons by electroneg: clearly from the band of electron recoil events. In gas except [7], were parameterized by L ve impurities. The Columbia detector accomplished thisseparated eff " :984E r Because the Columbia-measured value of Sr is close h a continuous gas circulation system developed for the unity, as seen in Fig. 3, we assume for simplicity that it t XENON prototype [,], while the Case detector no energy dependence. Following convention, nuclear d a similar recirculation system only at the start of its coil energies this way are denoted kev o-month run. Aprile calculated et al., PRL in 6 The detectors were calibrated with external gamma ray and electron recoil energies based on the linear S scale rces, including 57 Co, 33 Ba, and 37 Cs. The Case detecdenoted kevee. In the Case and Columbia detectors, also had an internal Po source deposited on the center S calibration gives, respectively, :5 pe=kevee he cathode plate, providing 5.3 MeV alpha particles. For 5 pe=kevee at zero electric field, which correspond tron data, external 5 Ci AmBe (Columbia) and 5!Ci :8 pe=kevr and pe=kevr for 56 kevr nuclear reco Calibration of the S ionization signal is also based Cf (Case) sources were used. 57 Figures and show the detectors responses to neutron Co kev gammas, but, as there is no published dat low energy Compton scattering events. The logarithm charge yield at these low energies, separate direct cha he ratio S=S is plotted as a function of nuclear recoil calibrations in the liquid phase were made in both de rgy. In both detectors, the elastic nuclear recoil band is tors. The amount of S light per electron is a function arly separated from the band of electron recoil events. In gas pressure and electric field in the gas phase [4], with Some earlier studies performed at small LXe prototypes (with no 3D sensitivity) FIG. (color). Columbia detector response to AmBe neutron 83- FIG. (color). Columbia detector response to AmBe neutron and Eric Dahl s thesis (9) 37 Cs gamma sources, at : kv=cm drift field. 83- Both studies obtained ~99% ER rejection (5% NR acceptance), suffered from edge effect.

11 A 3D sensitive two-phase xenon TPC used for this study Active diameter: 58 mm Meshes: mm pitch with um crossing bars Anode-Gate grid: 5 mm Liquid level: ~.5 mm Gate grid V: -4 kv Gas field: > kv/cm Drift length: cm Cathode: -(4.~6) kv Drift field:.- kv/cm

12 External sources (Cs37, Cf5) are used for electron recoils (ERs) and nuclear recoils (NRs) discrimination studies. Cs37, Cf5

13 Typical signals from the TPC

14 5 Cs37, kv/cm Electron lifetime = 5.±55.4µs 4 Counts kev S[PE] Light yield: equivalent to. PE/keV for kev at zero field S Bottom[PE] Gate Drift time[µs] Cathode Figure 8. (Left) The position dependence of S signal maximum Electron attenuation lifetime: of S ~ signals us for (or events 4 cm at the b is attenuation.4%. Blue dots length) indicate the median S values of ea drift time ranging from.5 µs to 5 µs. Dashed lines in X-Y dependence of S signals from 64 kev gamma ra 8 6 4

15 This TPC obtained excellent energy resolution for gamma rays. Q Lin et al 4 JINST 9 P44 (arxiv:39.556) keV: σ=3.85±.8 kev 36keV:. kev σ=5.± S[PE] 4 3 Counts[/keV] S[PE] θ E c,64kev [kev] -Drift Time[µs] Figure 9. (Left) Correlation of the S and S signals from activated xenon at kv/cm drift field. A -D Gaussian fit was applied to the S-SGate 8 profile and the correlation angle q was obtained. (Right) Combined 7 energy spectrum showing the two activated xenon lines at 64 kev and 36 kev. The combined scale was 6 defined based on 64 kev best fit. A fit using two independent Gaussian functions plus a flat background 5 gives resolutions (s/e) of.35% and.7% for the two lines, respectively. The reconstructed energy of 4 36Center kev under 64 kevedge gamma s best fit scale is 3 kev. The best resolution for 36 kev under its best fit 3 scale is.%. 4 Figure shows the combined Cathode energy spectrum from a 37 Cs run at.5 kv/cm with remaining activity from activated xenon. An energy resolution of.6% was achieved. For comparison, the X cog [mm] resolution for the 66 kev peak in the entire volume is.4%. The resolutions for the two activated S [PE]

16 ER and NR bands obtained at V/cm 5.5 kevnr σ ER band σ NR band 8 PE

17 ER and NR bands obtained at 7 V/cm

18 ER and NR bands obtained at V/cm

19 ER/NR calibration bands overlapped, V/cm

20 ER/NR calibration bands overlapped, 7 V/cm

21 ER/NR calibration bands overlapped, V/cm

22 Calculating the ER Gaussian leakage fraction : 4.3e-6 ER leakge below NR mean ( % ER Rej.) 8- PE 4.45σ ER 7.5e-5 ER leakage below NR mean (99.99% ER Rej.) 4-8 PE 3.79σ ER - - Probability Probability ER mean flattened log (S/S) ER flattened log (S/S)

23 S[PE] ER rejection power at different drift fields (central events) ER )/σ -µ (µ NR ER Nuclear recoil equivalent energy[kevnr] Preliminary V/cm 7V/cm V/cm % % 99.99% 99.9% 99%

24 Main%Observa,ons ER rejection power above 5 is observed (5% NR acceptance, 5.5~ kevnr) two orders of magnitude better than previous results No strong drift field dependence observed > 5 ER rejection at V/cm drift field If confirmed, HV will not be a concern for future multi-ton two-phase Xe detectors (low field operation) ER rejection power gets better at lower energy Xe is ideal for WIMP detection!

25 But%why? ER/NR%band%separa,on% ER%band%variance 4=8%PE.5.43± ± kevee.377±..9±.8 (b) AmBe and Cf 5 NR Calibration.349±..94± ±. kevnr ±.9 3 S x,y,z corrected (phe) σ (8%) kevee kevnr (b) AmBe and Cf 5 NR Calibration.6..4 b.±.9.6 (a) Tritium ER Calibration =4%PE.5.49±..3 kevee log (S /S) x,y,z corrected 6=%PE.6.5 b =6%PE.5 ER (a) Tritium ER Calibration log(sb/s) x,y,z corrected 8=%PE LUX, PRL 4 log (S /S) x,y,z corrected log (Sb /S) x,y,z corrected Energy'Bin' ΔLog(S/S).5 ('V/cm) S Sx,y,z x,y,zcorrected corrected(phe) (phe) kevnr 5 5 Figure : inleft: Calibrations of detector Figure : Left: Calibrations of detector response the 8 kg fiducial volume. response The 5 5 ER (tritium, panel a) and NR (AmBe and ER (tritium, panel a) and NR (AmBe and Cf, panel b) calibrations are depicted, Cf with the means (solid line) and ±.8 contours

26 Implica,ons%for%the%future Key factors to achieve a high ER background rejection (with high NR acceptance) S light collection S threshold Field uniformity (NOT strength) This result is very encouraging for a background-free operation of multi-ton two-phase Xe detectors towards a WIMP detection.