EDELWEISS: Latest Results and Outlook

Low-mass Dark Matter search with EDELWEISS: latest results and outlook

Valentin Kozlov (for the EDELWEISS collaboration)

Institute for Nuclear Physics, Karlsruhe Institute of Technology

− EDELWEISS-III experiment − Low-mass analyses − 3H intrinsic background CERN, Sept. 12-16 − Outlook: 2017 and beyond

KIT – University of the State of Baden-Württemberg and National Large-scale Research Center of the Helmholtz Association www.kit.edu Dark Matter evidences Gravitational lensing

ApJ 523, p.136 (1999) Bullet cluster, Hubble Space Telescope NASA/CXC/CfA/M.Markevitch et al.

But what is Dark Matter? - A new particle or set of particles? e.g. • Axion, AxionLikeParticle (ALP)? 10-6 – 10-2 eV • Sterile neutrino? ~keV CMB • Mirror dark world (dark photon mixing with γ)? and ΛCDM • WIMP? GeV - TeV • self-interacting dark matter? 26.8% A&A 2016, • …? 4.9% arXiv: 1502.01582 - Primordial black holes? 68.3% - Modified Newtonian Dynamics (MOND)? - A new force? 2/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics DM search strategies: indirect, direct, production “The search for Cold Dark Matter candidates addresses one of the most fundamental problems in particle physics and cosmology”

Annihilation (indirect det.) χ (WIMP) (WIMP) χ t Production

SM SM t t LHC: ATLAS, CMS Search for missing AMS02, PAMELA, FERMI, Scattering (direct detection) transverse energy in H.E.S.S, IceCube.... e.g. monojets

e.g. EDELWEISS, CDMS, CRESST, Xenon, LUX ... 3/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Direct (WIMP) DM search: state of the art

EDELWEISS III MLHD & BDT DAMIC CDMS-Si

DAMA/LIBRA

CDMSLite-2 CRESST(2015)

CRESST(2012)

SCDMS-LT

ν floor

Neutrino background J. Billard et al., Phys. Rev. D (2014)

4/27 This13.9.2016 plot: V.Y. KozlovErecoil | Low- massnucleus DM search

EDELWEISS III MLHD & BDT DAMIC CDMS-Si

cryogenic crystals DAMA/LIBRA CDMSLite-2 CRESST(2015)

CRESST(2012)

2-phaseSCDMS liquid-LT noble gases

ν floor

Neutrino background J. Billard et al., Phys. Rev. D (2014)

5/27 This13.9.2016 plot: V.Y. KozlovErecoil | Low- massnucleus DM search

CEA / Irfu / Iramis (Saclay) CSNSM (Orsay) Institut Neel (Grenoble) IPNL (Lyon) LPN (Marcoussis)

KIT (Karlsruhe) Collaboration

JINR (Dubna)

University of Oxford University of Sheffield EDELWEISS 2016 @ Karlsruhe (De)

13 km

1 - y 1 - sr 2 -

LSM 5 µ/m2/day

Muon intensity,Muon m 4800 mwe (deepest in Europe)

6/27 13.9.2016Depth, metersV.Y. Kozlov water | Low equivalent-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Edelweiss-III: experimental set-up Cryogenic installation (18 mK) : Polyethylene Reversed geometry cryostat shield (ext) Pb shield Can host up to 40kg of detectors Shielding : Clean room + deradonized air: 10Bq/m3  30mBq/m3

Active muon veto (>98% coverage) Ge-detectors Polyethylene PE shield internal + external (50cm) shield (int) Lead shield 20 cm (18cm + 2cm ancient lead) (Many) others : Remotely controlled sources for Muon Veto calibrations + regenerations Radon detector down to few mBq/m3 thermal neutron monitoring (3He det.) Upgrades for EDW-III: New cryogenics (microphonics), Additional PE shield, new Cu thermal screens, Kapton cables and connectors (α-n reduction), ALL cold electronics at 100K, digitization at 300K, New event- 7based/27 13.9.2016 scalable V.Y.IPE Kozlov-DAQ, | Low -mass… DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Edelweiss-III: looking inside

Kapton cables Bolo plates

New PE (1K)

Roman Pb

8/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Edelweiss-III: FID Ge-bolometers . Simultaneous measurement • Heat @18 mK with 2 x Ge/NTD thermometer Electrodes NTD • Ionization @ few V/cm

. Evt by evt identification by ratio Q=Eioniz/Erecoil • Q=1 for electron recoil Bolometer • Q≈0.3 for nuclear recoil support 800g -6  γ-rejection factor <5.8×10 (x5 better w.r.t. EDW-II) h=40mm . Vetoing surface events (ID electrodes) ∅ 70mm surface-rejection factor 4×10-5 (>15 keV)

) Neutron calibration NTD C1 +4 V V1 -1.5 V recoil / E

ionization Electron recoils Ionization yield (E yield Ionization Nuclear recoils NTD C2 -4 V V2 +1.5 V

9/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Recoil energy, keV PLB 681 (2009) 305; JLTP (2014) 176:870 Edelweiss-III: FID Ge-bolometers . Simultaneous measurement • Heat @18 mK with 2 x Ge/NTD thermometer Electrodes NTD • Ionization @ few V/cm

133 γ Ge-FID800 ( Ba 412000 ): ) NTD C1 +4 V V1 -1.5 V recoil / E

ionization

Electron recoils Ionization yield (E yield Ionization Nuclear recoils NTD C2 -4 V V2 +1.5 V

10/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Recoil energy, keV PLB 681 (2009) 305; JLTP (2014) 176:870 Edelweiss-III: FID Ge-bolometers . Simultaneous measurement • Heat @18 mK with 2 x Ge/NTD thermometer Electrodes NTD • Ionization @ few V/cm

) NTD C1 +4 V V1 -1.5 V recoil / E

ionization

1 event, RoI, ER>15keV Ionization yield (E yield Ionization NTD C2 -4 V V2 +1.5 V

11/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Recoil energy, keV PLB 681 (2009) 305; JLTP (2014) 176:870 Edelweiss-III: Strategies for light DM searches

. Low thresh analysis: Improve exposure and extend background ID to low energy . High Voltage: Amplification of the signal to reduce the effective threshold (Neganov-Luke effect: V = 100V  boost x35)

. Optimized sensors: Improve energy resolutions to reduce thresholds

 σheat=100eV, σion=100eV

12/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Edelweiss-III: Low-mass search data set

exposure w/o deadtime correction

. 36 detectors produced & installed in June 2014 blinded ROI for high mass WIMPs . WIMP data-taking since January 2015 July 2014 – April 2015 with 24 FID800 (cabled)  More than 14 kg of Fiducial mass in Ge

. 2015-2016: 2 R&D Runs

13/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Low threshold analysis: BDT . Use 8 detectors with lowest trigger thresholds:

4 @ 1 keVee and 4 @ 1.5 keVee (thanks to new electronics)  582 kg-d fiducial exposure 4σ ionization . Define rough ROI (WIMP box):

1 < Eheat < 12 keVee No fiducial cut 0 < Eionization < 8 keVee σ Eveto < 5 . Use Boosted Decision Trees: − Combine 6 variables (4 ionizations, combined heat, time-related variable) for Signal/Background discrimination − Model WIMP signal and backg.’s within this ROI to train BDT cuts − Choose BDT cuts of each detector to optimize their combined sensitivity . Background models are data driven: − Use regions without signal (sideband, depends on background) to build the bckg model − Use calibrations as crosscheck 14/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN JCAP05 (2016) 019; Institute arXiv:1603.05120 for Nuclear Physics

Low threshold analysis: BDT

Surface + Fiducial γ, β: Internal radioactivity + cosmogenic lines (68,71Ge, 65Zn, 68Ga) + 3H 4σ ionization

Surface betas, Pb-recoils: γ betas and 206Pb nuclei from 210Pb No fiducial cut Fiducial decay chain (Rn contamination) Surf. γ Neutrons: n’s . Radiogenic origin only . Rate is estimated from coincident Nuclear Recoils scaled by the Surf. β single-to-multiple ratio derived from Geant4 simulations Pb Heat-Only events: . Dominant background at low energies Heat-Only . Estimated from Eion<0 sideband data (no WIMP expected) . Potentially associated with cracks of individual detector's holders, time variated!  Intensive R&D ongoing to find the solution 15/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN JCAP05 (2016) 019; Institute arXiv:1603.05120 for Nuclear Physics Low threshold analysis: BDT JCAP05 (2016) 019; arXiv:1603.05120

MΧ = 5GeV MΧ = 20GeV

Dominant backgrounds: 5 GeV: Heat Only; 20 GeV: radiogenic neutrons

16/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Low threshold analysis: 2D Likelihood L. Hehn et al. arXiv:1607.03367 . Using same 8 det.’s and live time periods as for BDT: Only Quality cuts all Cuts . Stricter Quality cuts . Independent fiducial cut . 2 observables:

0 < Eionization < 15 keVee

Emin < Eheat < 15 keVee where

0.9 keV < Emin < 1.5 keV (indiv. 80%-eff. thresh.)  496 kg.d fiducial exposure . Use 2D profile likelihood:

N for each  n n exp detector: L(σ χ ,µ | mχ ) = ∏Ptot (Eheat ,Eion ) × ∏Gauss(µi | µi ,σ i ) × Poisson(N |υ) n=1 i   8 Each detector has its own for all: Lcomb (σ χ ,µ1..µ8 | mχ ) = ∏ Lj (σ χ ,µ j | mχ ) background, signal PDF, nuiss. j=1 parameters and constraint terms.

If the fit for mχ results in MLE for σχ compatible with 0  exclusion limit. 17/27 Hypotesis 13.9.2016 testV.Y. based Kozlov | Lowon-mass profile DM search likelihood with EDELWEISS test | TeVPA statistics 2016, CERN. Institute for Nuclear Physics Low threshold analysis: 2D Likelihood

heat L. Hehn et al. arXiv:1607.03367

ionization

. Good agreement model vs. data . No significant excess

18/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Low threshold analysis: BDT+2D Likelihood

90%CL limit for JCAP05 (2016) 019; arXiv:1603.05120 SI WIMP-nucleon: EDELWEISS III BDT

BDT (blind analysis): 1.15x10-38 cm2 at 4 GeV/c2 to EDELWEISS II DAMA/LIBRA 6.0x10-44 cm2 at 30 GeV/c2. (up to x40 better than EDW-II)

2D Likelihood: CoGeNT-2012 CoGeNT-2013 1.6x10-39 cm2 at 4 GeV/c2 to CDMSLite2 6.9x10-44 cm2 at 30 GeV/c2. (thanks to higher signal efficiency and background subtraction)

Cross-check of background EDELWEISS III models in two independent 2D Likelihood pipelines L. Hehn et al. arXiv:1607.03367 Above 15 GeV no difference BDT vs. 2D Likelihood

19/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Low threshold analysis: closer look @ER band S. Scorza et al. arXiv:1607.04560 . Thanks to its impressively low gamma background (<0.1 DRU) and ionization energy resolution (200 eV RMS)

EDELWEISS-III is the first Ge exp. to observe the intrinsic tritium beta background convincingly (Qβ = 18.6 keV, T1/2 = 12.32 y) (activation by cosmogenic n’s):

P = 82 ± 21 nuclei/kg.d . Tritium ß-decay is one of the major background component for next generation Ge experiments . Also measured rates of 49V, 55Fe and 65Zn . Results compared to model predictions and ACTIVIA code calculations

20/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics R&D: high-voltage (NL amplification) 133Ba: 356keV Neganov-Luke (NL) amplification:

Charge propagation ∆ Resulting V NL-phonons ‚Prompt‘ phonons

First data in Modane with 133Ba:  up to 100V working  NL boost 35

 lower threshold to ~60 eVee  ,easily’ upgradable setup,

thanks to KIT HV-card

 Sensitivity to low mass WIMPs (~1GeV/c2) BUT: no electron/nuclear recoil discrimination 21/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics R&D: optimized sensors

J. Billard et al., J. Low Temp. Phys. (2016)

Detailed model of heat signals that fits very well observed pulses: − Sensitivity to ballistic phonons and presence of a parasitic heat capacity − Optimized sensors, to be tested this year, should reach 100 eV baseline resolution

HEMT R&D for ionization signal also ongoing in collaboration with SuperCDMS to reach 100 eVee baseline resolution

22/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics EDELWEISS strategy 2017++

. common cryogenic infrastructure with SuperCDMS in SNOLAB (Canada) . SuperCDMS  2019

SuperCDMS@SNOLAB

23/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics EDELWEISS strategy 2017++

. common cryogenic infrastructure with SuperCDMS in SNOLAB (Canada) . SuperCDMS  2019

SuperCDMS@SNOLAB

10cm Pb 1cm archeol Pb  3 tons acrylic

. CUTE project (Queen´s, Kingston)  2017 3.6 m 24/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics EDELWEISS strategy 2017++

∅ 101.6 mm 33 mm thick . EDW  EURECA  SCDMS@SNOLAB . compatible interface with tower design

3D model @KIT

SCDMS

∅ 70 mm 40mm thick

EDW

EDW current casing SuperCDMS tower tower designSuperCDMS EURECA detectorsin SuperSIMcode EURECA (SCDMS) Hexagonal

25/27 13.9.2016 V.Y. Kozlov | Lowcup-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Conclusions and outlook

EDW-III results 2016 . until 2018: Low-mass program in LSM (350 kg.d)

R&D on NL 100V EDW-III goals 2017 HV, HEMT, sensors, heat-only 350kg-days (100V, σheat=100eV, σion=100eV, EDW bckg.) . after 2018: going beyond EDW detectors (35 kg x 1000 days) 8V in SNOLab

(0.1 x Compton, no n-bgd, no heat-only)

26/27 13.9.2016 V.Y. Kozlov | LowRecent-mass EDELWEISS DM search results with EDELWEISS & outlook | DPG | TeVPA 2016, 2016, Darmstadt CERN Institute for Nuclear Physics Conclusions and outlook

EDW-III results 2016 . until 2018: Low-mass program CRESST-3 phase 1 in LSM (350 kg.d)

R&D on NL 100V EDW-III goals 2017 HV, HEMT, sensors, heat-only CRESST-3 phase 2 350kg-days (100V, σheat=100eV, σion=100eV, EDW bckg.) . after 2018: going beyond EDW detectors (35 kg x 1000 days) 8V in SNOLab

(0.1 x Compton, no n-bgd, no heat-only)

SuperCDMS@SNOLAB

27/27 13.9.2016 V.Y. Kozlov | LowRecent-mass EDELWEISS DM search results with EDELWEISS & outlook | DPG | TeVPA 2016, 2016, Darmstadt CERN Institute for Nuclear Physics Backup slides

BACKUP SLIDES

28/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Edelweiss-III: surface-rejection

FID800 105

<63.5keV 46.5keV

NTD Ionization yield Ionization

FID800: 4×10-5 (90% CL, >15 keV) 1 event, RoI, E >15keV ID200 : 6×10-5 (90% CL, >20 keV) R

-46 2 If no other background: σSI ~ 4 × 10 cm

Recoil energy (keV) Phys Lett B 681 (2009) 305 J Low Temp Phys (2014) 176:870

29/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Low threshold analysis, BDT: results

. one BDT per WIMP mass . conservative limit: w/o background subtraction . No statistically significant excess

 Complemetary confirmation of latest LUX and SuperCDMS results

JCAP05 (2016) 019; arXiv:1603.05120 DAMIC 90% CL limit achieved for SI WIMP-nucleon: DAMA CRESST-2015 1.15x10-38 cm2 at 4 GeV/c2 to 6.0x10-44 cm2 at 30 GeV/c2. CDMSlite Up to x40 better than EDW-II Cross check with a 2D profile likelihood analysis

30/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics Low threshold analysis: 2D Likelihood

Component FID824 Combined L. Hehn et al. arXiv:1607.03367 Heat-only 5386 ± 804 44122 ± 1356 heat Cosmo- 176 ± 14 4358 ± 77 genic γ‘s Compton γ‘s 41 ± 6 554 ± 26

Tritium β‘s 43 ± 14 624 ± 77

Surface β‘s 8.5 ± 2.4 21.0 ± 3.6

Surface 206Pb 6.2 ± 0.8 35.5 ± 1.6 recoils Neutrons 0.19 ± 0.09 1.60 ± 0.72 ionization

All 5661 ± 805 49655 ± 1361 backgrounds Observed N 5685 50715

 . Good agreement model vs. data . No significant excess

31/27 13.9.2016 V.Y. Kozlov | Low-mass DM search with EDELWEISS | TeVPA 2016, CERN Institute for Nuclear Physics