1. WIMPS, 0-  decay & xenon High-pressure 136 Xe Gas TPC: An emerging opportunity David Nygren Physics Division - LBNL

2. Stockholm October 20072 Outline 0-  decay & WIMPs Conclusion Background rejection Energy resolution TPC options - “due diligence” Scaling to the 1000 kg era Issues Perspective

3. Stockholm October 20073 WIMPs Dark matter: ~23% of universe mass –visible ordinary matter: ≤1% –atoms, ordinary matter ~4% –WIMPs - lightest supersymmetric particle? Mass range: ~100 - 1000 GeV/c 2 Interaction strength not pinned down Recoil energy: ~ 0 - 100 keV –For a positive claim: Very robust evidence shall be required!

4. Stockholm October 20074 Two Types of Double Beta Decay If this process is observed: Neutrino mass ≠ 0 Neutrino = Anti-neutrino! Lepton number is not conserved! Neutrinoless double beta decay lifetime Neutrino effective mass A known standard model process and an important calibration tool Z 0  2 

5. Stockholm October 20075 Conclusions: An attempt to optimize a detector for a 0-  decay search in 136 Xe has led me to the following three conclusions: –High-pressure xenon gas (HPXe ~20 bar) will offer much better resolution than liquid xenon –Optimal energy resolution is achieved with a proportional scintillation readout plane –Result: An optimum WIMP detector too!  Two identical TPCs, filled with different isotopes…

6. Stockholm October 20076 Two Identical HPXe TPCs “  Detector” –Fill with enriched Xe mainly 136 Xe –Isotopic mix is mainly even-A –Events include all  events + background –WIMP events include more scalar interactions “WIMP Detector” –Fill with normal Xe or fill with “depleted” Xe –Isotopic mix is ~50% odd-A: 129 Xe 131 Xe –Events include only backgrounds to  –WIMP events include more axial vector interactions

7. Stockholm October 20077 Double beta decay Only 2- v decays! Rate EnergyQ-value Only 0- v decays! No backgrounds above Q-value! The robust result we seek is a spectrum of all  events, with negligible or very small backgrounds. 0

8. Stockholm October 20078 Perils of backgrounds Sensitivity to active mass deteriorates if backgrounds begin to appear: –m ~ {1/MT} 1/2 –m ~ {(b  E)/MT} 1/4 b = number of background events/unit energy  E = energy resolution of detector system M = active mass T = run time of experiment Ouch!

9. Stockholm October 20079 Current Status Present status (partial list) : –Heidelberg-Moscow ( 76 Ge) result:  m v  = 440 +14 -20 meV - disputed ! –Cuoricino ( 130 Te): taking data, but - background limited... –EXO ( 136 Xe): installation stage, but -  E/E shape/resolution issue... –GERDA ( 76 Ge): under construction at LNGS –Majorana ( 76 Ge): proposal & R&D stage –NEMO  Super-NEMO (foils): proposal & R&D stage Global synthesis:  m i < 170 meV (95% CL) *  100’s to 1000 kg active mass likely to be necessary! Rejection of internal/external backgrounds in ~10 27 atoms! Excellent energy resolution:  E/E <10 x 10 -3 FWHM - at least! Target (from oscillations):  m   ~50 meV * Mohapatra & Smirnov 2006 Ann. Rev. Nucl. Sci. 56 - (other analyses give higher values)

10. Stockholm October 200710  - Background rejection 1. Charged particles With a TPC, a fiducial volume surface can be defined with all the needed characteristics: –fully closed –Precise spatial definition –100% active - no partially sensitive surfaces –deadtime-less operation –surface is variable ex post facto Charged particles cannot penetrate fiducial surface unseen; –100.000% rejection

11. Stockholm October 200711  - Background rejection 2. Neutral Particles  -rays (most serious source, and needs real work) –Radon daughters ( 208 Tl, 214 Bi) plated out on surfaces –Photo-conversion fluorescence: 85% ( ~1 cm @ 20 bar) –Multiple Compton events more likely at large scale – Cannot get adequate self-shielding for MeV  -rays Neutrons can activate detector components, –Excitation of xenon may be useful calibration signal Neutral particles need serious study - high priority

12. Stockholm October 200712 HPXe, LXe TPC Fiducial volume -HV plane Readout plane. ionselectrons Fiducial volume surface * Real eventBackgrounds

13. Stockholm October 200713 EXO-200: LXe TPC Installation now, at WIPP: 200kg of 80% enriched 136 Xe Liquid xenon TPC Localization of the event in x,y,z (using scintillation for T 0 ) Avalanche Photo-Diodes (APD) used to detect scintillation Strong anti-correlation of scintillation & Ionization components Expect  = 3.3% FWHM for  For a subsequent phase of experiment: Barium daughter tagging by optical spectroscopy Electrostatic capture & mechanical transfer of ion to trap - Can this ambitious idea really work? Liquid or gas under consideration

14. Stockholm October 200714 EXO Experimental Design

15. Stockholm October 200715 EXO: Anti-correlation observed between scintillation and ionization signals Anti-correlation also observed in other LXe data  = 3.3% FWHM for  expected resolution (2480 KeV) What about the tails?

16. Stockholm October 200716 Energy resolution in xenon shows complexity: ionization  scintillation For  >0.55 g/cm 3, energy resolution deteriorates rapidly! Ionization signal only

17. Stockholm October 200717 Molecular physics of xenon Complex multi-step processes exist: *... –Ionization process creates regions of high ionization density in very non-uniform way –As density of xenon increases, aggregates form, with a localized quasi-conduction band –Recombination is ~ complete in these regions –Excimer formation Xe * + Xe  Xe 2 *  h + Xe –Also: Xe * + Xe *  Xe **  e - + Xe + + heat –Density dependence of signals is strong! * not completely understood - most likely, not by anyone

18. Stockholm October 200718 Impact for WIMP search? But...wait a second! WIMP searches in LXe use the ratio  = Primary ionization Primary scintillation to discriminate nuclear from electron recoils  nuclear <  electron

19. Stockholm October 200719

20. Stockholm October 200720 Impact for WIMP Search... The anomalous fluctuations in LXe must degrade the nuclear recoil discriminant S 2 /S 1 = ... So: Maybe HPXe would be better!...but Do the fluctuations depend on 1.atomic density? 2.ionization density? 3.particle type? 4.all of the above? Many  -induced events reach down to nuclear band in plots of S 2 /S 1 versus energy

21. Stockholm October 200721 Impact for WIMP Search... The anomalous fluctuations in LXe must degrade the nuclear recoil discriminant S 2 /S 1 = ... So: Maybe HPXe would be better!...but Do the fluctuations depend on 1.atomic density? 2.ionization density? 3.particle type? 4.all of the above? - YES, in LXe Many  -induced events reach down to nuclear band in plots of S 2 /S 1 versus energy

22. Stockholm October 200722 Gamma events (  ) Neutron events (NR) Latest Xenon-10 results look better, but nuclear recoil acceptance still needs restriction Log10 S2/S1

23. Stockholm October 200723 Fluctuations Let R = [log 10 {S 2 /S 1 (  )}] [log 10 {S 2 /S 1 (N)}] –R for LXe: ~10 1/2 (from previous slide) –R for HPXe:unmeasured Let D  = discrimination power vs density D = R/  T  T = [  2 {log 10 S 2 /S 1 (  )} +  2 {log 10 S 2 /S 1 (N)}] 1/2

24. Stockholm October 200724 Fluctuations… –S 2 /S 1 ( , HPXe) (  < 0.5 g/cm 3 ) is roughly known anomalous fluctuations are surely absent;  (Q/S (  )) is normal in HPXe for sure! But: –S 2 /S 1 ([N, , , energy,E-field], HPXe) unmeasured At least, haven’t found any relevant data – R could be smaller/larger in HPXe, but: D could/should be much larger!

25. Stockholm October 200725

26. Stockholm October 200726 What about “nuclear” recoil? v/c  0.001 for 50 keV recoiling xenon v/c  0.003 for a thermal electron  atomic shells stay with nucleus…  recoiling atom, not just nucleus  Xe-Xe collisions should be different from ,   “resonant charge exchange” isolates electrons  fast ion moves away quickly  isolated electrons are easily swept up by field

27. Stockholm October 200727 “Due Diligence” Evaluation of TPC operating modes Emphasis placed on: energy resolution for  decays Stability of operation One unlikely mode stands out

28. Stockholm October 200728 Pioneer HPXe TPC detector for 0-  decay search “Gotthard tunnel TPC” –5 bars, enriched 136 Xe (3.3 kg) –No scintillation detection!  no TPC start signal –  E/E ~ 80 x 10 -3 FWHM (1592 keV)  66 x 10 -3 FWHM (2480 keV) Reasons for this poor resolution are not clear... What is the intrinsic resolution? Can resolution near the intrinsic value be reached?

29. Stockholm October 200729 “Intrinsic” energy resolution for HPXe (  < 0.55 g/cm 3 ) Q-value of 136 Xe = 2480 KeV W =  E per ion/electron pair = 22 eV (depends on E-field) N = number of ion pairs = Q/W N  2.48 x 10 6 eV/22 eV = 113,000  N 2 = FN (F = Fano factor) F = 0.13 - 0.17 for xenon gas   N = (FN) 1/2 ~ 130 electrons rms  E/E = 2.7 x 10 -3 FWHM (intrinsic fluctuations only) Compare: Ge diodes (energy for pair) √3/22 = 0.37 Compare: LXe/HPXe Fano factors: √20/.15 = 11.5 !!

30. Stockholm October 200730 Energy resolution issues in traditional gas detectors Main factors affecting ionization: –Intrinsic fluctuations in ionization yield Fano factor (partition of energy) –Loss of signal Recombination, impurities, grids, quenching, –Avalanche gain fluctuations Bad, but wires not as bad as one might imagine... –Electronic noise, signal processing, calibration Extended tracks  extended signals

31. Stockholm October 200731 Loss of signal Fluctuations in collection efficiency  introduce another factor: L L = 1 -  similar to Fano factor (assuming no correlations)  N 2 = (F + L)N –Loss on grids is small: L grid < F seems reasonable If L grid = 5%, then  E/E = ~3 x 10 -3 FWHM –Other sources of L include: Recombination - ionization density, track topology Electronegative impurities that capture electrons (bad correlations) Escape to edges Quenching - of both ionization and scintillation! Xe* + M  Xe + M*  Xe + M + heat (similarly for Xe 2 *, Xe**, Xe 2 * + … ) Xe + + e – (hot) + M  Xe + + e – (cold) + M *  Xe + + e – (cold) + M + heat  e – (cold) + Xe +  Xe*

32. Stockholm October 200732 A surprising result: adding a tiny amount of simple molecules - (CH 4, N 2, H 2 ) quenches both ionization and scintillation (  particle)  dE/dx is very high; does this effect depend on  too? (yes...) Impact for atomic recoils?… Gotthard TPC: 4% CH 4 how much ionization for  particles was lost? K. N. Pushkin et al, IEEE Nuclear Science Symposium proceedings 2004 ( ~25 bars )  particles

33. Stockholm October 200733 Avalanche gain –Gain fluctuations: another factor, “G”  N = ((F + G)N) 1/2 0.6 < G < 0.8 *  N = ((0.15 + 0.8)N) 1/2 = 328  E/E = 7.0 x 10 -3 FWHM - not bad, but: No more benefit from a small Fano factor Very sensitive to density (temperature) Monitoring and calibration a big effort Space charge effects change gain dynamically Micromegas, GEM, may offer smaller G * Alkhazov G D 1970 Nucl. Inst. & Meth. 89 (for cylindrical proportional counters)

34. Stockholm October 200734 “Conventional” TPC Pure xenon (maybe a small molecular admixture is OK...) –Modest avalanche gain is possible –Diffusion is large, up to ~1 cm @ 1 meter –drift velocity very slow: ~1 mm/  s –Add losses, other effects,…  N = ((F + G + L)N) 1/2  E/E ~ 7 x 10 -3 FWHM Avalanche gain may be OK, but can it be avoided?

35. Stockholm October 200735 “Ionization Imaging” TPC 1. No avalanche gain analog readout (F + L) dn/dx ~ 1.5 fC/cm:  ~9,000 (electron/ion)/cm gridless “naked” pixel plane (~5 mm pads) very high operational stability   E/E = 3 x 10 -3 FWHM (F + L) only –but, electronic noise must be added! 50 pixels/event @ 40 e – rms  N ~280 e – rms  E/E ~ 7 x 10 -3 FWHM Complex signal formation Need waveform capture too Ionization imaging: Very good, but need experience

36. Stockholm October 200736 “Negative Ion” TPC “Counting mode” = digital readout, (F + L) –Electron capture on electronegative molecule –Very slow drift to readout plane; –Strip electron in high field, generate avalanche –Count each “ion” as a separate pulse: Ion diffusion much smaller than electron diffusion Avalanche fluctuations don’t matter, and Electronic noise does not enter directly, either Pileup and other losses: L~ 0.04 ?  E/E = ~3 x 10 -3 FWHM – Appealing, but no experience in HPXe...

37. Stockholm October 200737 Proportional Scintillation (PS): “HPXe PS TPC” Electrons drift (m) to high field region (5 mm) –Electrons gain energy, excite xenon, make UV –Photon generation up to ~1000/e, but no ionization –Multi-anode PMT readout for tracking, energy –PMTs see primary and secondary light “New” HPXe territory, but real incentives exist –Sensitivity to density smaller than avalanche –Losses should be very low: L < F –Maybe: G ~ F ?

38. Stockholm October 200738 H. E. Palmer & L. A. Braby Nucl. Inst. & Meth. 116 (1974) 587-589 

39. Stockholm October 200739 From this spectrum: G ~0.19

40. Stockholm October 200740 Fluctuations in Proportional Scintillation G for PS contains three terms: Fluctuations in n uv (UV photons per e):  uv = 1/√n uv –n uv ~ HV/E  = 6600/10 eV ~ 660 Fluctuations in n pe (detected photons/e):  pe = 1/√n pe –n pe ~ solid angle x QE x n uv x 0.5 = 0.1 x 0.25 x 660 x 0.5 ~ 8 Fluctuations in PMT single PE response:  pmt ~ 0.6 G =  2 = 1/(n uv ) + (1 +  2 pmt )/n pe ) ~ 0.17 Assume G + L = F, then Ideal energy resolution (  2 = (F + G + L) x E/W) :  E/E ~4 x 10 -3 FWHM

41. Stockholm October 200741 HPXe PS TPC is the “Optimum”  detector Readout plane is high-field layer with PMT array –Transparent grids admit both electrons and UV scintillation –Multi-anode PMT array “pixels” match tracking needs –Two-stage gain: optical + PMT: Noise < 1 electron! –No secondary positive ion production –Much better stability than avalanche gain Major calibration effort needed, but should be stable –Beppo-SAX: 7-PMT HPXe (5 bar) GPSC in space!

43. Stockholm October 200743

44. Stockholm October 200744

45. Stockholm October 200745 Some Gas Issues: –Operation of PS at 20 bar? should work... –Maintain E/P, but surface fields are larger… –Large diffusion in pure xenon –   tracking good enough? (I think so… needs study) –Integration of signal over area? (… needs study) –are pixels on the track edge adding signal or noise? –Role of additives such as: H 2 N 2 CH 4 CF 4 ? –molecular additives reduce diffusion, increase mobility, but –Do molecular additives quench  signals? (atomic recoils?)

46. Stockholm October 200746 More gas issues: Neon –Multiple roles of neon (must add a lot) –to diminish Bremmstrahlung? –to increase drift velocity? –to diminish multiple scattering? –to add Penning effect, thereby reducing F? –to facilitate topology recognition by B field? –to add higher energy WIMP-nuclear recoil signals? –Operation of PS with neon? should work... –neon excitation energy transfers to xenon –May be “invisible”

47. Stockholm October 200747 More issues Detection of xenon primary UV scintillation (175 nm - ~7.3 eV) n electrons  n photons (depends on E-field, particle, etc) 0-  decays (Q = 2480 keV): No problem! WIMP signals (5 - 100 keV): Need very high efficiency!  Design augmentation: Must maximize primary UV detection! Add reflective surfaces along sides - for sure... Add optimized dielectric mirror to cathode plane (corner reflectors?)

48. Stockholm October 200748 Even more issues For given mass M, HPXe detector has ~9 x more surface area than LXe…  backgrounds larger !? –Background rejection may be so much better that overall performance is better. –Scintillator shell along sides may help…

49. Stockholm October 200749 Scaling to 1000 kg Minimum S/ V for cylindrical TPC: L = 2r Fixed mass M & track size l: –l /L   -2/3 –higher density  less leakage High voltage  L  M 1/3 (fixed  ) Number of pixels involved in event: –Diffusion  L 1/2 so: very little change Scaling to large M is fairly graceful

50. Stockholm October 200750 1000 kg Xe:  = 225 cm, L =225 cm  ~ 0.1 g/cm 3 (~20 bars) A.Sensitive volume B. HV cathode plane C.GPSC readout planes, optical gain gap is ~1-2 mm D.Flange for gas & electrical services to readout plane E.Filler and neutron absorber, polyethylene, or liquid scintillator, or … F.Field cages and HV insulator, (rings are exaggerated here) possible site for scintillators

51. Stockholm October 200751 Is HPXe PS TPC the “Optimum” WIMP detector? –Do the fluctuations persist in “nuclear” recoils at HPXe densities of interest? threshold density  ~ 0.2 g/cm 3 for protons, tritons* Recoil atoms appear to behave differently –Can we bridge the dynamic range? Yes,  instantaneous ionization signal is ~50 keV But, scintillation signal is instantaneous Augmentations needed: #1: scintillation signal needs high dynamic range #2: photon detection maximum efficiency * private communication, Aleksey Bolotnikov (n + 3 He  3 H + p)

52. Stockholm October 200752 An HPXe System at LLNL

53. Stockholm October 200753 An HPXe System at LLNL Built for Homeland Security  spectroscopy goals Dual 50 bar chambers! Bake-out, with pump Gas purification Gas storage Excellent R&D platform

54. Stockholm October 200754 Two identical HPXe TPCs Two distinct physics goals “  Detector” –Fill with enriched Xe mainly 136 Xe –Events include all  events + backgrounds –Isotopic mix is mainly even-A –WIMP events include more scalar interactions “WIMP Detector” –Fill with normal Xe or fill with “depleted” Xe –Events include only backgrounds to  –Isotopic mix is ~50% odd-A: 129 Xe 131 Xe –WIMP events include more axial vector interactions

55. Stockholm October 200755 What to do? 1.Measure response of HPXe to n &  ’s ratio of ionization (S 2 )/scintillation (S 1 ) resolution of S 2 & S 1 signals (fluctuations!) must use a good proportional scintillation system measure as f (E- field, density, Neon, N 2,… ?) 2.Most of the apparatus and expertise needed has been developed already by the WIMP community (TAMU) 3.Can two distinct communities collaborate? 1.LXe  effort EXO just getting going at WIPP, NM 2.LXe and WIMP community: enthusiasm under control for new approach by outsider, at least so far...

56. Stockholm October 200756 Perspective It is extremely rare that two distinct physics goals are not only met, but enhanced, by the realization of two essentially identical detector systems (differing only in isotopic content). Because the impact of success would be so significant, and future costs so large, this approach needs to be explored seriously.

57. Stockholm October 200757 Thanks to: Azriel Goldschmidt - LBNL Katsushi Arisaka -UCLA Adam Bernstein - LLNL Alexey Bolotnikov - BNL Doug Bryman - TRIUMF Gianluigi De Geronimo - BNL Madhu Dixit - Carleton J J Gomez-Cadenas - Valencia Cliff Hargrove - Carleton Mike Heffner - LLNL John Kadyk - LBNL Norm Madden - LLNL Jeff Martoff - Temple Jacques Millaud - LLNL Leslie Rosenberg - U Wash Hanguo Wang - UCLA James White - Texas A&M

58. Stockholm October 200758

59. Stockholm October 200759

60. Stockholm October 200760 Barium daughter tagging and ion mobilities… Ba + and Xe + mobilities are quite different! –The cause is resonant charge exchange –RCE is macroscopic quantum mechanics occurs only for ions in their parent gases no energy barrier exists for Xe + in xenon energy barrier exists for Ba ions in xenon RCE is a long-range process: R >> r atom glancing collisions = back-scatter RCE increases viscosity of majority ions

61. Stockholm October 200761 Barium daughter tagging and ion mobilities… –Ba ++ ion survives drift: IP = 10.05 eV –Ba ++ ion arrives at HV plane, well ahead of all other Xe + ions in local segment of track –Ba ++ ion liberates at least one electron at cathode surface, which drifts back to anode –arriving electron signal serves as “echo” of the Ba ++ ion, providing strong constraint –Clustering effects are likely to alter this picture!

62. Stockholm October 200762 A small test chamber can show whether ion mobility differences persist at higher gas density (no data now). This could offer an auto- matic method to tag the “birth” of barium in the decay, by sensing an echo pulse if the barium ion causes a secondary emission of one or more electrons at the cathode. 

63. Stockholm October 200763

64. Stockholm October 200764

65. Stockholm October 200765