SN Relic Neutrinos in Large Water Cherenkov Detectors

1. SN Physics Workshop September 17th 2009 SN Relic Neutrinos in Large Water Cherenkov Detectors Michael Smy UC Irvine Chandra/Hubble View of E0102-72

2. Outline • Super-Kamiokande Search • Published Analysis using SK-I Data • Analysis Improvements & Data Update: increase sensitivity by ~factor two • Search with Neutron Tagging • SN Relic n Prospects with a DUSEL Water Cherenkov Detector • Without Neutron Tagging • With Neutron Tagging Chandra/Optical/Radio View of SN1006

3. Super-Kamiokande

4. νe+ 16O  16N + e+ The main interaction mode for SRN’s in SK is charged current quasi-elastic interaction (inverse b decay) 10 νe+ p  e+ + n 0.1 10-3 SK Event Rate [/year /MeV] νe+ 16O  16F + e- 10-5 νe+ e  νe + e- 10-7 0 10 20 30 40 50 Electron energy [MeV] Courtesy K. Bays, UC Irvine

5. SK Main Backgrounds O μ spallation products from cosmic m’s e X γ X e atm. nm→ stealth m±→e± relic n’s • oxgen spallation products from cosmic m’s (~600/day) • atmospheric n’s • CC ne’s • sub-Cherenkov m production: nm→stealth m→e • radioactivity • solar n’s • reactor n’s • spallation limits the energy threshold & cuts to reduce it causes greatest signal loss • sub-Cherenkov threshold muons from atmospheric neutrinos are irreducible without neutron tagging Michael Smy, UC Irvine

6. energy resolution Spallation Products 11Be 16N 15C 8Li 8B 12C half life in s 9Li 9C 8He 12B 12N 13B 12Be 14B 13O 11Li Energy in MeV Courtesy K. Bays, UC Irvine

7. Spallation Products

8. Tagging Spallation Events μentry point μ track DlLongitudinal maximum light emission DlTransverse Relic Candidate form time diff. Dt between muon and relic candidate reconstruct muon track calculate residual charge ResQ: total light minus charge expect. from length find distance of closest approach DlTransverse of muon to relic candidate use arrival time of each hit to calculate emission point along track: DlLongitudinal is difference of point of max. light emission and relic candidate projection peak light emission QPeak K. Bays, UC Irvine

9. Example of a dE/dx Plot QPeak= sum of charge in window p.e.’s spallation expected here distance along muon track (50 cm bins) Courtesy K. Bays, UC Irvine

10. A Simple Example ofSpallation Removal entry point muon peak of dE/dx LLONG relic n candidate LTRAN Spallation • LLONG (cm) • LLONG (cm) • LTRAN (cm) • LTRAN (cm) Courtesy K. Bays, UC Irvine

11. three-variable likelihood cut for successful single track fits Dt DlTransverse ResQ two-variable likelihood cut for unsuccessful single track fits Dt ResQ 150ms cut on Dt 18 < E < 34: 36 % signal inefficiency for each muon type (single m, m bundle, stopping m): four-variable likelihood cut if single, well-fit track Dt DlTransverse DlLongitudinal QPeak three-variable likelihood otherwise Dt DlTransverse Qtotal 18 < E < 24 MeV: 18.5 % ineff. 16 < E < 18 MeV: 22.5 % ineff. Previous and Improved Spallation Tag Previous Improved Michael Smy, UC Irvine

12. Removal of Spallation • Deadtime 18% (Improved from 36%) • Increase in Exposure of 28% • Further Tuning may be possible… single muons (dt < 10 s) stopping muons 12 LTRAN (cm) dt (seconds) black – before likelihood cut, red – after likelihood cut Courtesy K. Bays, UC Irvine

13. energy resolution for an event of energy: 16 MeV Solar ν Events 18 MeV pp 7Be pep 8B hep 16 18 e recoil energy (total) (MeV) Solar 8B and hep neutrino are a SRN background (hep at 18 MeV, and both at 16 MeV, because of energy resolution) Cut criteria is optimized using 8B/hep MC improved cut is energy dependent, tuned in 1 MeV bins Courtesy K. Bays, UC Irvine

14. Solar n cut solar n candidates 16-17 MeV ε = 72.5% Inefficiency: Previous: 7% for 18-34MeV Improved: 4.5% for 18-19MeV, 0% above 19MeV 17-18 MeV ε = 86.5% 18-19 MeV ε = 95.5% Courtesy K. Bays, UC Irvine

15. External Backgrounds remove events with small effective wall to killradioactive decays originating outside the detector but reconstructing within the fiducial volume of SK Inner detector wall effective wall reconstructed event vertex reconstructed event direction Effective Wall in cm Signal Inefficiency: Previous: 7% Improved: 2.5% Energy in MeV Courtesy K. Bays, UC Irvine

16. Efficiency Improvement efficiency increase > 18 MeV: (# events new/previous) unnormalized relic (Ando) unnormalized stealth m Michels new! new cuts more efficient AND at least as effective > 34MeV, efficiency increase < 1.0 due to new background reducing cuts (new pion cut especially) 18 – 34 MeV, large efficiency increase due mostly to new spallation and solar cuts. 16-18 MeV region is now usable as well! Energy [MeV] Courtesy K. Bays, UC Irvine

17. Future of this Search • New cuts improve efficiency, re-analyzing now • More planned improvements: • Fiducial volume enlargement • Finalizing event selection • Combine SK-I, SK-II and SK-III data. • Extract new combined limit. • Hope to publish result within 1 year. • Future phase: neutron tagging with Gd. Courtesy K. Bays, UC Irvine

18. Possibilities of ne tagging ne could be identified by delayed coincidence. Inverse beta decay Possibility 1 νe+ p  e+ + n n+p→d + g n 2.2MeV g-ray g ne p p DT = ~ 200 msec Number of hit PMT is about 6 in SK-IV Gd Possibility 2 e+ g n+Gd →~8MeV g DT = ~30 msec Add 0.2% Gd2(SO4)3 in water Positron and gamma ray vertices are within ~50cm. (ref. Vagins and Beacom) GADZOOKS! Courtesy Iida, ICRR

19. Gadzooks! Measured Gd n capture Spectrum in SK • dissolve Gd salts into SK water to detect gamma by neutron capture • need to investigate • water transparency • water recirculation • material effects • test tank for Gadzooks! is now being constructed!! Astrophys. J. 697, 730-734 (2009) Michael Smy, UC Irvine

20. GdCl3 Source in Super-Kamiokande • measure Gd n capture gamma cascades: • Spectrum • Vertex Resolution • Capture Time Michael Smy, UC Irvine

21. Possibility of SRN detection Relic model: S.Ando, K.Sato, and T.Totani, Astropart.Phys.18, 307(2003) with NNN05 flux revision If invisible muon background can be reduced by neutron tagging SK10 years (e=67%) Assuming 67% detection efficiency. With 10 yrs SK data, Signal: 33, B.G. 27 (Evis =10-30 MeV) Assuming invisible muon B.G. can be reduced by a factor of 5 by neutron tagging. Courtesy Iida, ICRR

22. Gd-sized impurities only (NF reject) Gadolinium Water “Band-Pass” Filter Gd plus smaller impurities (UF product) pure water plus Gd from tank Ultrafilter Nanofilter impurities smaller than Gd (NF product) impurities bigger than Gd (UF reject) De-Ioniziation/ Reverse Osmosis pure water (DI/RO product) impurities to drain (DI/RO reject) M. Vagins, ICMU

23. Filtration/Transparency Studies in Irvine Simple Filter DI IDEAL Extra DI IDEAL DI “band pass” System pure H2O System usual style water filtration system “band pass” water system DI DI DI Michael Smy, UC Irvine

24. Measuring Water Transparency • idea based on a IMB device • measure light intensity continuously as a function of light travel distance • vertical pipe for quick & easy change of distance • pipe is necessarily short (< height of lab) • look for changes when GdCl3 / Gd2(SO4)3 is introduced • plastic pipe and tank (no metal effects) • use integrating spheres and a focal lens to stabilize intensity measurements of Si photodiodes • use laser pointers (small, cheap & good beam quality) Michael Smy, UC Irvine

25. Experimental Setup Adjustable Mirrors Beam Splitter & Steerer Integrating Sphere & Photodiode Pulsed Laser Pointers Michael Smy, UC Irvine

26. Reject 0.2% Gd(NO3)3: UV (337nm) Pure Water Measurement Gd(NO3)3 Measurement linear scale linear scale 125.8±5.9m 94.87±0.46cm log scale Michael Smy, UC Irvine

27. Endorse GdCl3 Solution 0.8% Solution: 4xGadzooks! Concentration 66.8±0.9m 405nm 69.8±3.2m 478nm 21.08±0.51m 532nm 360nm 337nm 6.422±0.014m 595nm 650nm 2.864±0.004m 33.00±0.23m 27.74±0.26m Michael Smy, UC Irvine

28. Gadolinium Compound Selection • GdCl3 is considered too corrosive for stainless steel tank/PMT support structure • Gd(NO3)3 is opaque in the UV • Gd2(SO4)3 is not as corrosive and (from spectro-photometer measurements) should have good water transparency • However, it dissolves not nearly as fast: must first solve selective water filtration Michael Smy, UC Irvine

29. Gd2(SO4)3 Filtering Progress • took data with ultrafilter and two types of nanofilters • basic principle is sound • UF passed ~100% of Gd2(SO4)3 • NF rejected ~100% of Gd2(SO4)3 • actually use try multiple stages of NF; clean up product with DI & RO units • so far, cannot reproduce transparency even without Gd; need to tune the bandpass; check for impurities from additional components • when filtration is working, measure resulting water transparency of Gd2(SO4)3 solution M. Vagins, ICMU

30. EGADS Evaluating Gadolinium’s Action on Detector Systems Make 100 ton class test tank and demonstrate the GADZOOKS! Idea. PMTs Figure by A.Kibayashi 0.2%Gd water in 100 ton class water tank Transparency measurement Water system Courtesy A. Kibayshi, Okayama University

31. Current status of Gadzooks! • Excavation has started • Test tank is currently designed • Construction will start soon • Material compatibility test • Study selective water filtration at Irvine • transparency measurement at Irvine • test a large-scale water system and measure the water transparency performance with EGADS soon Michael Smy, UC Irvine

32. Supernova Relic Neutrino Requirementof DUSEL Water Detector K. Bays, UC Irvine

33. Requirements • sufficient depth to avoid being overwhelmed by spallation background • need above about six photo-electrons/MeV for sufficient energy resolution (and threshold for Gd n capture events) • need low PMT dark noise (same reason): cooling of the PMT environment • good radiopurity Michael Smy, UC Irvine

34. Expected Threshold for DUSEL Detector at 4850ft Level • assume spallation background is dominant issue • assume spallation spectrum scales with muon rate when varying depth • ignore correlation between spallation energy and lifetime • keep signal/background ratio to the same level as SK Michael Smy, UC Irvine

35. I Relic Spectrum Increase in relic rate in water compared to present SK analysis as energy threshold changes Courtesy K. Bays, UC Irvine

36. II Muon Intensity as Function of Depth Courtesy K. Bays, UC Irvine

37. III Spallation Spectrum at SK • The unnormalized spallation spectrum from SK data can be parameterized by a simple formula: • The increase in spallation as the energy threshold is lowered can be calculated by: En (MeV) Courtesy K. Bays, UC Irvine

38. III Spallation Spectrum With Gd: Guess Spallation Rate with n • shorter livetime, less products, less energy • …but what are production rates? • what if spallation list is not complete? • if reduced by ~1 order of magnitude: shift spectrum by 2.6MeV (ln(10)MeV/0.894) Michael Smy, UC Irvine

39. Energy Threshold Results Depth (m.w.e.) w/o Gd with Gd Energy Threshold (MeV) • Some particular values: 4050 (4850 ft) = 15.5/12 MeV 2930 (3500 ft) = 17.5/15 MeV 2700 m.w.e. (SK) = 18/15.5 MeV 1680 (2000 ft) = 20.5/18 MeV 250 (300 ft) = 25/22.5 MeV • since SK will lower the threshold, a DUSEL detector should be able to employ the same techniques, so this is very conservative Courtesy K. Bays, UC Irvine

40. Conclusions • SK is improving the sensitivity of the SN relic search • spallation tagging is critical for this • together with data update, sensitivity should improve by up to a factor of two • SK will lower the energy threshold of the search to 16 MeV • SK investigates introduction of Gd salt to detect anti-neutrinos via delayed coincidence using n capture on Gd: • water filtration system is currently designed in Irvine • large-scale test and material effects are studied soon in a especially built test tank next to SK • with Gd, SK should see SN relics within ten years • DUSEL detector has excellent prospects to measure and study the SN relic signal • must have sufficient photocathode coverage • must have cool enough PMT environment • must have radiopurity • DUSEL depth is sufficient with or without Gd