Results from the Full Analysis of 1998-1999 CDMS Data Runs

1. Results from the Full Analysis of 1998-1999 CDMS Data Runs Richard Schnee Case Western Reserve University

2. CDMS Collaboration • Santa Clara University • B.A. Young • Stanford University • L. Baudis, P.L. Brink, B. Cabrera, • C. Chang, T. Saab • University of California, Berkeley • M.S. Armel, S.R. Golwala, V. Mandic, P. Meunier, M. Perillo Isaac, W. Rau, B. Sadoulet, A.L. Spadafora • University of California, Santa Barbara • D.A. Bauer, R. Bunker, • D.O. Caldwell, C. Maloney, • H. Nelson, J. Sander, S. Yellin • University of Colorado at Denver • M. E. Huber • Brown University • R.J. Gaitskell, J.-P. Thompson • Case Western Reserve University • D.S. Akerib, A. Bolozdynya, D. Driscoll,S. Kamat, T.A. Perera, R.W. Schnee, G.Wang • Fermi National Accelerator Laboratory • M.B. Crisler, R. Dixon, D. Holmgren • Lawrence Berkeley National Laboratory • R.J. McDonald, R.R. Ross, A. Smith • National Institute of Standards and Technology • J. Martinis • Princeton University • T. Shutt

3. CDMS Background Discrimination • Ionization Yield (ionization energy per unit recoil energy) depends strongly on type of recoil • Most background sources (photons, electrons, alphas) produce electron recoils • WIMPs (and neutrons) produce nuclear recoils 1334 Photons (external source) 233 Electrons (tagged contamination) • Particles (electrons) that interact in surface “dead layer” of detector result in reduced ionization yield 616 Neutrons (external source) Ionization Threshold • Detectors provide near-perfect event-by-event discrimination against otherwise dominant electron-recoil backgrounds , very good (>95%) against surface electron-recoil backgrounds bulk electron-recoil backgrounds

4. 10.6 m earth Active Muon Veto Pb Shield .. Fridge Inner Pb shield Polyethylene Detectors Current CDMS Site: Stanford • Shielded, low-background environment • Shallow (17 mwe rock) • Hadronic cosmic-ray flux reduced by >1000x • Muons reduced by ~5x • Active muon veto • >99.9% efficient • Reject ~100 neutrons per kg-day produced by muons within shield • Expect neutron background ~2 / kg / day produced outside shield; measure using • Two materials (Si more sensitive to neutrons, Ge more sensitive to WIMPs) • Multiple-detector neutron scatters m m n Copper n n

5. CDMS Results • Results of 1998-1999 runs announced at UCLA DM2000 • Published in Phys. Rev. Letters v.84, #25, pp.5702-6 (19 June, 2000) • Thesis at http://cosmology.berkeley.edu/preprints/cdms/Dissertations/Sunil • 1999 : 4x165 g Germanium BLIP 10.6 kg-days after cuts • 13 single-scatter nuclear recoil events observed (WIMPs or neutrons) • 4 multiple-scatter nuclear recoil events observed (neutrons) • 1998 : 100 g Silicon ZIP 1.6 kg-days after cuts • 4 nuclear recoil events observed (mostly neutrons) • Most sensitive upper limits on WIMP-nucleon cross-section • Improved analysis of these runs recently completed • Relaxation of fiducial volume cut 15.8 kg-days after cuts • Better quantitative estimates of systematic errors • More conservative treatment of data from 1998 Silicon ZIP • Long paper on results will be submitted soon (PRD) • Description of analysis, cuts, and calculation of efficiencies

6. Region of Shared Events Inner Ionization Electrode Outer Ionization Electrode Increasing the 1999 Run BLIP Fiducial Volume • Inner ionization electrode shielded from background events. Top View • Events near inner-outer gap have ionization energy shared between the two electrodes • Internal multiple scatters also appear as shared events • Including “shared” events increases exposure • by ~40% for WIMPs • by ~60% for neutrons. • The less restrictive cut yields our “ultimate results” for the data set.

7. all single-scatters nuclear recoil candidates 1999 Run Ge BLIP Muon-Anticoincident Data Set Shared-Electrode 4.4kg-days forWIMPs 10nuclear-recoil candidates > 10 keV Inner-Electrode 11.9kg-days forWIMPs 13nuclear-recoil candidates > 10 keV NR Band (-3s,+1.28s) 90% eff. NR Band (-3s,+1.28s) 90% efficient

8. Inner-electrode B3 B4 B5 Shared-electrode B6 low-yield hit in outer electrode Neutron Multiple Scatters nuclear-recoil candidate in both detectors nuclear-recoil candidate in one detector  B4 / B5  B5 / B6  B4 / B6 • Require that at least one hit be in fiducial volume • Observe 4 neutron multiple scatters in 10-100 keV multiple events • Calibration indicates negligible contamination by electron multiples • ≈1 with one misidentified • <0.05 with both misidentified photons surface electrons Ionization Yield [keV/keV] neutrons Ionization Yield [keV/keV]

9. bulk events NR candidates 1998 Run Si ZIP Data Set Early-design Si ZIP measured external neutron background • Not WIMPs: Si cross-section too low (~6x lower rate per kg than Ge) • Misidentified electrons? • Calibration predicts <0.26 events in 20-100 keV range at 90% CL, but we cannot rule out systematic error due to fact that conditions of calibration and low-background data-taking were different • Using conservative assumptions about a calibration taken under same conditions as low-background data predicts contamination of 2.2 events in NR band (<7.3 events at 90% CL) • Use this very conservative estimate (7.3 events) in calculating limits mostly neutrons

10. 13 17 4 1.7 7.7 10 0.4 + Data w/ 68% confidence interval X Prediction based on Ge M, Si S 4 2.1 Predictions based on most likely 4.6 23 25 Consistency of Neutron Hypothesis Inner-electrode Shared-electrode x Total • 4 Ge multiples and 4 Si singles imply expected background of 29 neutron singles in Ge, with large statistical uncertainty • Most likely neutron background predicts fewer inner-electrode Ge multiples than seen (1.7 vs. 4). Overall, data in good agreement. • Likelihood ratio test: expect worse agreement 30% of the time

11. CDMS Upper Limits 90% CL upper limits assuming standard halo, A2 scaling • Most constraining upper limit of any experiment for WIMPs with 10-70 GeV mass • EDELWEISS better above 70 GeV • Rules out DAMA NaI/1-4 most likely point (x) at >99.9% CL (for standard WIMPs, halo) • Rules out DAMA NaI/0-4 most likely point (circle) at >99% CL (for standard…) • Compatible with less likely points in DAMA 3s allowed regions DAMA limit X marks DAMA NaI/1-4 most likely point EDELWEISS limit Expected CDMS sensitivity DAMA NaI/1-4 3sregion

12. Best simultaneous fit to CDMS and DAMA predicts too little annual modulation in DAMA, too many events in CDMS (even for small neutron background) Compatibility of CDMS and DAMA • Likelihood ratio test • asymptotic approximations • “standard” halo • standard WIMP interactions • CDMS results incompatible with DAMA model-independent annual-modulation data (left) at > 99.99% CL predicted WIMP modulation DAMA annual modulation data predicted WIMP spectrum with n background CDMS data n background (1.1 multiples)

13. Best simultaneous fit to CDMS and DAMA predicts too little annual modulation in DAMA, too many events in CDMS (even for NO neutron background) Compatibility of CDMS and DAMA • Likelihood ratio test • asymptotic approximations • “standard” halo • standard WIMP interactions • CDMS results incompatible with DAMA model-independent annual-modulation data (left) at > 99.8% CL, even under assumption that none of the CDMS events are neutrons predicted WIMP modulation DAMA annual modulation data predicted WIMP spectrum CDMS data

14. Additional Comments on Upper Limits • Limits from inner-electrode data are near expected sensitivity (dots) EDELWEISS EDELWEISS Inner electrode Inner & shared electrodes

15. Additional Comments on Upper Limits • Limits from inner-electrode data are near expected sensitivity (dots) • New limits from inner-electrode data are worse than old CDMS limits (light blue) due to more conservative treatment of Si data EDELWEISS EDELWEISS Old CDMS Inner electrode Inner & shared electrodes

16. Additional Comments on Upper Limits • Results slightly better if Si data is ignored (dashed red curves) EDELWEISS EDELWEISS Ignoring Si Ignoring Si QI QIS Old CDMS Inner electrode Inner & shared electrodes

17. Additional Comments on Upper Limits • Results slightly better if Si data is ignored (dashed red curves) • Even without neutron subtraction (blue dash-dot curves), better limits than any other experiment for low-mass WIMPs (10-45 GeV) (See Yellin, soon-to-be astro-ph for method) CDMS no neutrons CDMS no neutrons EDELWEISS EDELWEISS Ignoring Si Ignoring Si Old CDMS Inner electrode Inner & shared electrodes

18. Conclusions • No significant change from original conclusions • New data from relaxing the fiducial volume cut are consistent with our earlier results • Ultra-conservative treatment of Si data has small impact on results • Two years of scrutiny of the data resulted in only small changes • See long paper to be submitted very soon for (extensive) details • Background is dominated by neutrons at shallow site at Stanford • Best upper limits of any experiment for WIMPs with 10-70 GeV mass • Results incompatible with signal claimed by DAMA at high confidence level • If signal is from scalar-coupled WIMP in a standard dark matter halo • We are focusing on CDMS II • First complete tower of 6 ZIPs now running at Stanford Underground Facility with internal neutron shield (factor >2) • To be transported in Soudan in Summer 2002.