1. Dark matter Search with 100kg of CsI(Tl) crystals Seung Cheon Kim 1

2. Conclusion of previous presentatation Total exposure, 32793 kg days of data analyzed. More simple cut was developed for PMT noise rejection and Muon veto is applied. Surface alpha background are identified and subtracted. New improved limit on WIMP nucleon interaction was obtained. We set most stringent limit on spin dependent WIMP proton interaction among direct dark matter searches. 2

3. 3 Mean time distribution of SA, NR & gamma

4. 4 Fitting WIMP search data with 3 components, SA, NR, gamma det0 SA NR(WIMP) gamma

5. How to set the limit on NR (WIMP event rate) => Bayesian analysis with PSD Modeling the distribution of logrmt10 with 3 components  Pdf = f0 x F NR + f1 x F SA + (1-f0-f1) x F gamma F NR :pdf of Nuclear recoil events F SA :pdf of surface alpha events F gamma :pdf of gamma background 5

6. Bayesian analysis with PSD P: posterior pdf L: likelihood P0:prior pdf => f0, f1 are flat in (0,1), practically (0,0.3) for simulation time  : data  : parameter 6

7. 7 SA NR 2D posterior pdf for SA & NR

8. 8 Posterior pdf for NR NR

9. Nuclear recoil Event rate 9 90% limit 68% interval

10. w/o det8,11 90% limit of nuclear recoil event rate 10

11. 11 Compare this with NR rate  Minimize (NR – p*sim)  From p, cross section obtained Detector simulation for WIMP Considering the Maxwellian WIMP velocity distribution with a spherical halo, quenching factor, form factor, and the detector performance.

12. Limit from 12 detectors SI interaction 12

13. SD proton interaction 13

14. Systematic study Main systematic factors 1) Uncertainty of SA, NR, Gamma p.d.f modeling 2) The effect of Prior p.d.f in bayesian analysis 3) Uncertainty of expected spectrum of WIMP

15. Systematic error from PDF uncertainty Uncertainty of pdf(surface alpha, neutron, gamma) and crystal dependences are considered. From the Am calibration data (59.54keV), the variation of logrmt10 of gamma events for 12 detectors: 0.62  0.035 => uncertainty of gamma pdf logrmt10 of 12 detectors

16. 345678910 SA&N0.00550.00240.00040.0030.00310.00380.00270.0039 gamma0.00050.00000.00080.00070.00150.0008 0.0006 total0.00550.00240.00120.00320.00340.00390.00290.0039 lim900.0077 5 0.0055 5 0.0029 5 0.0023 5 0.0073 5 0.0082 5 0.0131 5 0.0140 5 Systematic error of f0(fraction of Nsig = Nsig/Ntot) including all the detectors The deviation of the most probable value, Nsig, from 1 sigma variation effect of from certain factor. The systematic error from pdf uncertainty is around 30%.

17. Flat prior is a very popular prior representing complete ignorance about the parameters. But, it has a problem that if it’s flat on , then it’s not flat for non linear function of . So, different parameterization would in general lead to non-equivalent posterior pdf. Consideration about prior

18. Jeffrey’s prior  Non informative,objective prior  Parameterization invariant

20. Jeffrey’s Prior for our model

21. f1=0 f1=0.2 f1=0.5 f0+f1<1 f0=0 f0=0.2 f0=0.5

22. Jeffrey prior flat prior For det0, 3keV bin There’s no difference. The results are determined by the data dominantly.

23. 23 Systematic error from simulation The cross section limit error from simulation uncertainty is 3%.

24. Trigger effect (8ms dead time) muon tail rejection cut > 0.05s * Muon coincidence event rate ~6evts/hr Muon tail event rejection

25. time span from the previous events Events below 20keV (WIMP search region) are drawn. After muon tail rejection cut After fit quality cut After all the event selection cuts 8ms Trigger dead time effect High energy event tail rejection

26. Discussions on further study with this result =>Inelastic dark matter (idm) iDM is proposed to reconcile the DAMA observation and null results from other experiments. (Tucker-Smith, Weiner 2001) There exists an excited state  * of the dark matter particle with a mass m  * - m  =  ~ 100keV. Elastic scatterings off of the nucleus,  N ->  N are suppressed, compared with the inelastic scatterings  N ->  * N.

27. Minimum velocity for recoil E, E R m N : mass of target Nucleus µ:reduced mass of WIMP/target Nucleus  :mass splitting between  and  * ERER  min

28. Three key features *Spectrum of events is dramatically changed, suppressing or eliminating low energy events. ->Spectrum of Conventional WIMP rises exponentially at low energy. But iDM peaks at E R ~20keV or above. ERER rate Conventional WIMP spectrum iDM spectrum 20keV

29. Three key features *Heavier targets favored over lighter targets. For a given E R and , Heavier target mass -> lower  min -> greater available range of velocity I(DAMA) is better than Ge(CDMS).

30. Three key features *modulation of the signal enhanced significantly. Mainly high velocity WIMPs are detectible. In these high velocity region, WIMP numbers is changing rapidly. visible to DAMA visible to DAMA and CDMS f(v) Neal Weiner, NYU; IDM 2008 Modulation ratio of DAMA/LIBRA according to 

31. Benchmark Points (S.Chang et al, PRD 79 043513 (2009)) Vesc =500km/s, V0=220km/s

32. NR event rate of KIMS 2007 w/o det8,11 NR event rate of KIMS 2010 This level is below 0.03 dru, touching Bench mark region of DAMA.

33. Conclusion New improved limit on WIMP nucleon interaction was obtained. We set most stringent limit on spin dependent WIMP proton interaction among direct dark matter searches. Systematic study was done.  Most dominant uncertainty is from the modeling of logrmt10 distribution. It’s around 30% conservatively. The discussion on further study about idm model was shown.  It is expected to give the strong limit on the bench mark region of DAMA. 33

34. Back up 34

35. Dark matter ? Most matter of the universe is unknown. => It doesn’t emit the light. It rarely interacts. But, its gravitational effect is evident. Rotation curve for Galaxy NGC6503 35

36. Dark matter ? Ordinary matter like atoms, or other known particles can’t explain these unknown matter effect. The existence of exotic dark matter is supported with various astronomical observation. X-ray from Hot clusterBullet cluster 36

37. WIMP(Weakly Interacting Massive Particle)? The dark matter is very likely to be (or preferred) Stable (i.e, long-lived with relic origin) Weakly Interacting (i.e, explaining present relic density) Massive (i.e, non-relativistic) Particles. We have plenty of candidates for WIMP from various motivations other than dark matter problem.  Lightest Supersymmetric Particle(LSP), Lightest Kluza-klein Particle(LKP), Massive Sterile Neutrino, Axino... 37

38. How to sense WIMP? WIMP will recoil nucleus. It is expected to deposit around a few tens keV. It will interact very rarely. Thermal signal Scintillation, e-h pair 38

39. KIMS’ main detector: CsI(Tl) scintillator CsI crystal? Easy to get large mass with an affordable cost High light yield ~60,000/MeV Easy fabrication and handling Enabling pulse shape discrimination -> Statistically, nuclear recoil event rates can be estimated 39 So, it’s very natural idea to test WIMP interaction with CsI(Tl) scintillator. And, it’s KIMS who did it, for the first time.

40. KIMS’ main detector: CsI(Tl) scintillator CsI crystal? Sensitive to both SD and SI WIMP interactions A(Cs) =133, A(I) =127 IsotopeJAbun 133 Cs7/2100%-0.3700.003 127 I5/2100%0.3090.075 73 Ge9/27.8%0.030.38 129 Xe1/226%0.0280.359 131 Xe3/221%-0.009-0.227 19 F1/2100%0.441-0.109 40

41. But, there’re various inherent radioisotopes background. Cs134, Cs137, Rb87 We had tried to reduce the most dominant background, Cs137. Now, We reduced the background level to 2-3count/day/kg/day below 10keV. Geant Simulation 41 CsI(Tl) scintillator for WIMP search?

42. WIMP search of KIMS WIMP search using CsI(Tl) scintillator =>12 detectors (103.4kg) YangYang underground Lab(Y2L) =>Located in Yangyang pumped storage power plant =>700m minimum depth, 2000m water-equivalent =>accessible by car (tunnel~2km) 42

43. Neutron shield(30cm mineral oil) Lead shield (15cm) Polyethylene(5cm) Copper (10cm) CsI(Tl) Scintillator Neutron detector Muon detector (Neutron sheild) KIMS Detector system N 2 gas flow inside the Cu shield 43

44. CsI(Tl) crystal detector One detector module : one CsI Crystal + 2 PMTs PMT : 3” PMT (9269QA), Quartz window, RbCs photo cathode (green extended) Crystal size: 8x8x30 cm 3 (8.7 kg) (Beijing Hamamatsu Photon Techniques Inc.) Trigger condition: In 2us, 2 more photons in each PMT + high energy event 8ms dead time after high energy event 50ms dead time after muon coincidence Event window is 40µs. Digitized with 400MHz FADC Muon event sec 44 Event window is 40µs. Digitized with 400MHz FADC

45. Trigger effect muon tail rejection cut Muon coincidence events rejection * Muon coincidence event rate ~6evts/hr 45

46. 46 Relative intensity..... Am calibration -> ~5 p.e /keV Am241 calibration E(keV) 13.9keV Np L  X-ray 17.8keV Np L  X-ray 20.8keV Np L  X-ray 26.35keV gamma Cs, I X –ray escape 59.54 gamma

47. 47 PE yield of 12 detectors

48. Full range Energy spectrum Crystal label B0510A B0510BB0511B0601 B0605A B0605BB0606AB0606BB0609A B0609BS0501AF0711 ▬ Total events ▬ Multiple hit events det0det1 det2 det3 det4 det9 det10det11 det8 det5 det6det7 48 Event rate of 12 detector ~ 6Hz

49. Full range Energy spectrum ▬ Total events ▬ Multiple hit events E of det10 Total E except det10 Energy spectrum of one detector Spectrum of total energy sum of whole detectors 605 KeV 134 Cs, 660 KeV 134 Cs 796 KeV 134 Cs 1401 KeV 134 Cs 1970 KeV 134 Cs 49 1168+802 = 1970 KeV 1365+605 = 1970 KeV β decay

50. Backgrounds in WIMP search with CsI(Tl) scintillator Gamma  Almost from internal radioisotope can be subtracted by PSD Surface alpha(SA)  Alpha events escaping from the surface of the detector faster than nuclear events can be subtracted by PSD PMT noise background  PMT produces dark current and typically dark current event has large cluster in it. PMT, itself, produces the scintillation. 50

51. 51 Surface alpha? in 10  s

52. Mechanism of surface alpha(SA) Radon(Rn222) is a noble gas and well distributed in the air. It decays to Po(Po218), which is a reactive metal, readily deposited on all surfaces. The intermediate progeny, Pb210 has long life time,22year. It provides alpha emitter, Po210 constantly. Alpha escaping from the surface doesn’t leave the full energy. 52

53. 53 PMTs CsI(Tl) crystal A: 0609A, Rn progeny contaminated -exposed to Rn gas for one week B: 0512, clean Aluminum foil t=2  m x 3 layers AB Sides are wrapped by teflon Study of SA events with Rn progeny contaminated crystal Rn contaminated Double crystal Detector(RDD)

54. 54 SA event selection Part A Tagged as Alpha at part B

55. 55 Energy spectrum of SA event

56. 56

57. PMT noise limits the sensitivity at the low energy level experiment seriously. Source of PMT noise =>Thermionic Emission =>Afterpulse: the drift of ionized residual gas in the PMT toward photocathode excitation of metastable state of photocathode or glass => Cerenkov radiation from Cosmic ray or background radio-isotope => Fluorescence of glass PMT noise event is asymmetric along 2PMTs in size and time. PMT noise event study 57

58. Characters of PMT noise event Clean acryl box PMT  Data from PMTs + acryl box system We have 345days data for this. Event rate: a few tens to a few hundreds events per day 58 pmt noise eventtypical scintillation event called as clusters Usually one cluster equals one s.p.e. But, pmt noise event includes abnormly big cluster larger than typical cluster. pmt0 pmt1

59. 59 Biggest cluster cut pmt0.qc / pmt0 bclust >3 & pmt1.qc / pmt1 bclust >3 qc asymmetry cut Cut range is determined respectively for each detector since the difference of gain, yield and noise of PMTs. PMT noise rejection cut

60. 60 time difference cut -0.3  s < start time difference < 0.3  s

61. PMT noise event rejection And with other cuts ( multiple hit veto …) For 345 days pmt dummy detector data, After applying the cut developed above, 98 events survived at 3-4keV =>~ 0.03 cpd, What are they? 61 Examples of events of pmt only detector, passing cut pmt0 pmt1

62. 62 Short tail rejection cut 0.002

63. 63 The effectiveness of PMT noise rejection cut

64. The description about the data Period: September 2009 – August 2010 => 317.15 days Detector mass: 103.4kg Total exposure: 32793 kg days 64

65. Event selection 65 PMT noise rejection cut +... High energy even tail rejection The rejection of the event containing dark current

66. 66 The effect of cut Multiple rejection Biggest cluster cut Short tail event rejection

67. 67 The efficiency of cut obtained from the compton scattering events

68. 68 Total background level

69. Combining pdf by multiplication at 3keV 69

70. SA Event rate 70

71. Detector simulation for WIMP Comparison between SIM and data for Am calibaration Considering the gain, quantum efficiency of PMTs, and energy calibration. 71

72. Limit from 12 detectors SI interaction 72

73. SD proton interaction 73

74. 345678910 SA&N0.00120.00320.00080.00150.00260.00660.00250.0033 gamma0.00050.00000.00080.00070.00150.0008 0.0006 total0.00120.00320.00120.00170.00290.00660.00260.0033 lim900.0077 5 0.0055 5 0.0029 5 0.0023 5 0.0073 5 0.0082 5 0.0131 5 0.0140 5 Systematic error of f0(fraction of Nsig = Nsig/Ntot) Excluding det11 det11 is contaminated by SA more seriously than other detectors.

75. 345678910 SA&N0.00000.00370.00110.00230.00380.00200.00180.0024 gamma0.00050.00000.00080.00070.00150.0008 0.0006 lim900.0077 5 0.0055 5 0.0029 5 0.0023 5 0.0073 5 0.0082 5 0.0131 5 0.0140 5 Systematic error of f0(fraction of Nsig = Nsig/Ntot) Excluding det2,3,4,11

76. Systematic error from PDF uncertainty Uncertainty of pdf(surface alpha, neutron, gamma) are considered. The deviation of the most probable value, Nsig, from 1 sigma variation effect of from certain factor. 345678910 SA&N0.00080.0040.00070.00150.00130.0040.0010.003 gamma0.00250.0020.00870.00180.00490.00570.00290.0023 lim900.0076 5 0.0066 5 0.0037 5 0.0025 5 0.0098 5 0.0072 5 0.0091 5 0.0096 5 Systematic error of Fsig(fraction o Nsig = Nsig/Ntot)

77. Benchmark Points (S.Chang et al 2008) Expected energy spectrum simulation for KIMS. The peak is above our threshold energy, 3keV! E measured Eee(keV)

78. SD proton interaction 78 Indirect searches

79. cpd combined including 12 detectors 79

80. cpd combined excluding det8 and det11 80