The DAMIC (Dark Matter in CCD) Experiment at SNOLAB

Status of DAMIC

Alvaro E Chavarria The DAMIC (Dark Matter UniversityIn of Chicago for the DAMIC Collaboration

CCD) experiment at SNOLAB1 Mariangela Settimo Subatech, CNRS-IN2P3, Nantes (France)

1 DAMIC : DArk Matter In CCDDAMIC @ SNOLAB 675 Charge-Coupledm thick, 16 Mpix Device CCD, (CCD) 6 g Detection of point-like energy deposit from nuclear recoils ! induced by WIMPSWhy interactions Dark in the M bulkatter of CCDs. in CCDs ?

• Detection of point-like energy deposits from nuclear recoils induced by WIMP interactions 16 Mpix, 15 µm x15 µm, (10 keV675 Si µm ion thick, range 5.9 g 200mass A) 5) Unique spatial resolution: 3D position reconstruction and particle ID 3.7 eV to create e-h pair 1) High-resistivity (1011 donors/cm3) 2) Fully-depleted over several Sensitivity to DM masses 100s (typical CCDs few extremely pure silicon < 10 GeVµm (nuclear recoil) X-rays from ~ eVtens (electron of µm )recoil) 55Fe

R&D program (2013-2015) 3D reconstruction (x, y, z) and unique 2 The chargespatial diffuses resolution towards the CCD40g detector commissioned in 2017 pixels gates, producing a “diffusion- 2 limited” cluster

15 a muon piercing Float-zonea 675 µm Si thick DAMIC CCD single muon track

σ ≈ Z : fiducial volume definition and surface event rejection 18 Some results with the R&D detector

Result with 10 g detector mass (data: 2015-2016) Hidden photon DM Selected results PRL 118, 141803 (2017) WIMP DM Hidden photon DM PRL 118, 141803 (2017) Selected resultsPRD 994, 082006 (2016) WIMP DM Exposure 0.6 kg d

PRD 994, 082006 (2016)

Exposure 11.5 g d

radioactive bkg in the silicon bulk 2015 JINST 10 P08014 nuclear recoil calibration … and also: radioactive bkg in the silicon bulk 2015 JINST 10 P08014 • radioactive bkg in the silicon bulk, 2015 JINST 10electron P08014 recoil calibration nuclear recoil• nuclearcalibration recoil calibration, PRD 94, 082007 (2016) • electron recoil calibration, PRD 96, 042002 (2017)electron recoil calibration 3

PRD 94, 082007 (2016)

PRD 94, 082007 (2016) PRD 96, 042002 (2017)

4 PRD 96, 042002 (2017)

4 DAMIC @ SNOLAB (2000 m underground lab) DAMIC @ SNOLAB 675 !m thick, 16 Mpix CCD, 6 g

24 DAMIC @ SNOLAB (2000 m underground lab) DAMIC @ SNOLAB 675 !m thick, 16 Mpix CCD, 6 g exposure ~ 1 min On surface 4) Unprecedented low energy threshold < 0.001 e/ • Negligible noise contribution from dark current fluctuations (dark current pixel/day with CCD cooled at 120 K). Readout noise dominant contribution.

blank (taken after A readout noise of ≈ 2 e- is exposure8h • exposure) achieved by slow CCD readout (≈ 10 min / 16 Mpix image).

3.6 eV to produce 1 e-hole pair 1.2 eV band gap At Snolab + shielding

• Very long exposures (8 hours!) to minimize the n. of noise pixels SNOLAB data above the energy threshold • Lower threshold, higher WIMP recoil rate (exponential), • small mass detector competitive 17 Privitera Part B2 DAMIC-M 1280 1290 1300 1310 1320 1330 image 1330 a) 4180 c)

Low-energy Electron 1320 Unique spatial resolution4190 candidates

1310 4200 image

Particle identiﬁcation and bulk / surface background50 pixels rejection

13004210 α

b) Muon

12904220 On-going background measurements32 “Worms”Spatial straggling coincidence of beta decays ( Si) • 1280 4180 4190 4200 4210 4220 5 10 15 20 25 30 electrons 5 10 15 20 25 30 Energy measured by pixel [keV] 32 32Si (T1/2= 150 y, β ) ➔ 32P (T1/2= 14 days, β) • Cosmogenic Si • StraightFigure tracks: 5. a) DM detection minimum in a CCD; b) 3D position reconstruction; c) signatures of different ionizing particles in a CCD: a straight track (cosmic ray muon), large blob (alpha particle), “worm” (straggling electron) and small round clusters (low-energy X-ray, ionizingnuclear particles recoil, DM candidate) . On-going background measurements Search for spatially • MeV chargeexposure to air.blobs: This background rejection capability is particularly relevant for 32Si, a naturally occurring alphasisotope in silicon that is not removed by chemical purification, which is expected to be the dominant correlated beta decays. background• Cosmogenic contribution in the 32 currentSi generation of silicon detectors32Si (cf.(T 1/2 SuperCDMS= 150 [25y, ]).β DAMIC) ➔ 32 P (T1/2= 14 days, β) 32 32 Diffusion-limitedidentifies and rejects clusters: the decay sequence Si (T1/2= 150 y, β) → P (T1/2= 14 days, β) by spatial correlation, • reducing this background by at least a factor 100. Sensitivity with current low-energy X-rays, ThisSearch novel experimental for spatially technique has been demonstratedcandidate in a successful 32Si R&D - carried32P out at SNOLAB data is few Bq/kg nuclearduring recoils the last three years under PI Privitera leadership. Stringent limits on intrinsic radioactive contaminants incorrelated the silicon bulk of thebeta CCDs, incdecays.luding the first measurementin the of 32 Sinew activity indata ultra-pure set silicon using the spatial correlation method, show the power of DAMIC background identification and rejection methods [36]. TheSensitivity potential for DM withdetection currentis demonstrated in a search for low-mass WIMPs (Fig. 3.b; [35]) and for CCD spatial resolution 32 32 • hiddendata-ph otonis DMfew (Fig. Bq/kg 3.a; [34]), the latter establishing the world-best limits for ≈ eV DM masses evcandidateen Si - P • Bulk and surface radiogenicprovides backgroundswith minimal a unique exposure. handle The seven - CCD 40-g detector which will operate until 2018 at SNOLAB has achieved background rates as low as 5 dru, a level comparable to other low-mass dark matter experiments in the new data set to thelike understanding CRESST [15] and CDMSL of ite [23], and will provide further210 guidance – e.g. with a precise measurement ofDecay 32Si and tritium chain backgrounds, ofsee Section a single b.5 – for the design of PbDAMIC -M. the background• Bulk and surface radiogenic backgrounds Δt = 4 days Δt = 10 days 4 keV e 41 keV ppt e levels 4323 keV α Energy [keV] 3168 3168 3168 102

3166 3166 3166 3164 3164 sensitivity3164 to bulk ppt levels 3162 3162 3162 10 3160 3160 contamination3160 sensitivity to bulk 3158 193158 3158 y [pixel] y [pixel] y [pixel] 1 3156 3156 3156 contamination 3154 3154 3154 σ ≈ z : for ﬁducial volume deﬁnition 3152 3152 3152 10 1 5330 5332 5334 5336 5338 5340 5342 5344 5346 5348 5330 5332 5334 5336 5338 5340 5342 5344 5346 5348 5330 5332 5334 5336 5338 5340 5342 5344 5346 5348 and210 surface event rejection 210 210 x [pixel] x [pixel] x [pixel] Pb (T1/2= 22.3 y, β) ➔ Bi (T1/2=5.0 days, β) ➔ Po (T1/2= 138210 days, α) Figure210Pb 6. T he(T decay1/2= chain 22.3 of a y, single β) ➔Pb nucleus210Bi on(T the1/2 CCD=5.0 surface days,: a sequence β) ➔ of two210 betaPo and(T one1/2= alpha 138 particle days, separated α) in time by several days and detected in the same location (see text).

• Cosmogenic tritium sensitivity•b.2 CosmogenicDAMIC-M at the Laboratoire tritium Souterrain sensitivity de Modane 2015 JINST 10 P08014, PRD 994, 0820006 (2016) The Laboratoire Souterrain de Modane offers several advantages when compared to SNOLAB. Its horizontal6 Sensitivity to activation rate downSensitivityaccess to through few the to Frejustens activation tunnel of greatly tritium simplifies rate the down atoms/kg/day logistics toto perform few detector tens packaging, of tritium test and atoms/kg/day 11 assembly underground. Supply of radon-free air, crucial to minimize surface backgrounds, is already11 part of the laboratory infrastructure. While LSM is not as deep as SNOLAB, its rock overburden is sufficient to

5 Low background noise and energy threshold Comparing images (8h exposure) and “Blanks” (exposure = readout time)

1. Negligible dark current: < 0.001 e/pix/day (the lowest ever measured in a Si detector)

- - σ = 1.6 e ~ 6 eV 2. read out noise: 1.6 e @140K ‣ Energy threshold ~ 40 eVee (~ 6σ above background, 3.77 eV Data 2017 for e-h pair in Si )

Linearity demonstrated for signals <10 e-

) 1 . 0 8 e e X-rays V 1 . 0 6 e

k Optical photons 1 . 0 4 9 Linearity of the CCD response within .

5 1 . 0 2 (

k ± 5% down to 40 eVee

1 /

0 . 9 8 ) E

( 0 . 9 6 k 10−1 1 10 7 I o n i z a t i o n s i g n a l [ k e V ee]

Response to electrons

Tritium Al Energy loss in gates and SiO2 < 2 µm / 675 µm O σ ≈ 21 eV Si Current detector conﬁguration Current status • Seven CCDs‣ 7 CCDs in stable in stable data data taking; taking since 2017 40 g detector(1 CCD sandwiched in ancient lead)

One CCD ‣sandwiched40 g detector massin ancient lead • A data set (7.6 kg day) collected with full spatial resolution‣Operating (1x1 temperature binning), of optimized 140K for background‣ Exposure characterization for image : 8h and and 24h measurement(each image(32Si, acquisition 210Pb) is followed by a “blank” • A second datawhose setexposure being is the collected readout time) (so far 4.7 kg day‣) 7.6with kg bestday of energy data for thresholdbackground (1x100 binning)characterization

‣ So far, 4.6 kg day of data collected for binning: charge ofDM several search pixels (in 1x100 are addedhardware before binning) readout 3x3 1x1 some loss of spatial 8 resolution but improved signal to noise (same α-β event readout noise but more charge in a binned pixel) 5 Event clusters search strategy ‣ Pedestal and correlated noise subtraction (hot pixels among several images masked) ‣ LL ﬁt of the signal in a moving window across the image

ΔLL = ℒn - ℒs ﬂat noise Gaus signal + ﬂat noise

data Example of one event

Events blanks 102 Fit simulation Erec = 7.18 keV Data σ = 0.7 pix

−45 −40 −35 −30 −25 −20 −15 −10 −5 ∆ LL DLL cut : < 0.001 bkg events from exponential ﬁt of the “blanks” distrib 9 Surface background rejection

Surface (back)

Fiducial Bulk volume 0.3 < σ < 0.8

Surface (front)

10 Background model and acceptance

Backgrounds: on front Back • Background on front/back Back and back surfaces of surfaces of the CCD Front CCDs, in the bulk Front Bulk •BkgData model compared compared with to data Bulk background model (50/25/25 of bulk/front/back) 50/25/25 bulk/front/back (estimated from high energy data) #

MC sims • Acceptance for energy “Bulk” Acceptancedeposit in the for bulk bulk events estimated(from MC from simulations) MC simulations; threshold at Reconstructed clusters Energy50 eVee threshold : 50 eVee After ﬁducial cut

11 Energy spectrum above 2 keV ee All events 210 30 Pb EnergyFiducial spectrum region Cu above 2 keV

ee 210Pb 25 All events 30 210Pb Fiducial region Cu 20 25 210Pb Event rate / dru 20 15 Event rate / dru 15 10 10 5 5

0 0 2 2 4 4 6 6 88 10 10 12 12 14 14 Energy / keV Energy ee/ keVee ≃ 5 dru in fiducial region, consistent between CCDs a factor of ≃ 3-4 lower than our previous background level ≃ 2 dru for lead sandwiched CCD 9

Diﬀerential rate unit (dru) = 1 event /keV/kg/day 12 Low energy data (0.05 - 2 keV)

Analysis of the low energy data in progress (most sensitivity to low mass WIMPS) Low energy data • We are analyzing the data from 50 eV to 2 keV, which provide most sensitivity to low massTwo WIMPexamples. eventsSome (dataexamples + ﬁt) of candidates.

E = 0.56 keV, σ = 0.6 E = 0.14 keV, σ = 0.5 ΔLL = -780 ΔLL = -130 E = 1.6 keV E = 0.1 keV $LL = -30 $LL = -14000

• NOTE: CDMS II silicon potential signal obtained with a 7 keVnr threshold (≈ 2 keVee) We are exploring for the first time the silicon target with a much 13 lower threshold of 0.6 keVnr (≈ 0.05 keVee) Our results will be relevant for SuperCDMS Si and DAMIC-1K 10 Expected sensitivity

1035

1036 Current exposure (03/2018) 37 10 and bkg (5 keV/kg/d) CRESST (2015) 1038

39 CDMS-II Si 2 10 (2013) DAMIC (2016) DAMIC (4.6 kg d) [cm ]

n 40 10 CDMSLite (2015) DAMIC (13 kg d)

1041

1042 exposure by end 2018 DarkSide-50 (2018) 1043 Xenon experiments 1044 1 10 [GeV]

Exploring for the 1st time the CDMS signal with the silicon target and a much lower energy threshold (0.6 keVnr ~ 0.05 keVee) 14 DAMIC-M @ Laboratoire Souterrain de Modane A 0.5 kg detectorDAMIC-1K for WIMPs and at dark-sector the LSM candidates at low-masses

• Proposals• Largest submitted CCD ever for a kg-built size(6k detector x 6k atx the1mm, Laboratoire mass 20 g) DAMIC-1K at the LSM Souterrain de Modane • CCDs with skipper readout for • Proposals submitted for a kg- sub-electron• Skipper noise readout for sub-eV noise size detector at the Laboratoire Souterrain de Modane • Bkg reduction1 e- to a fraction of dru • CCDs with skipper readout for (improved design, materials, procedures) sub-electron noise • Large size Skipper CCDs (6k x 4k) σ = 0.06 e- ! for DAMIC-1K development will 1 e- 6k x 6k CCD, 1mm arrive at UW in May/June 2018. thick, 20g SENSEI R&D at Fermilab

University of Chicago, University of Washington, Pacific Northwest National Laboratory, SNOLAB, Laboratoire de Physique Nucléaire et de Hautes Energies (LPNHE), the Laboratoire de l'Accélérateur Linéaire (LAL) and the Laboratoire Souterrain de Modane/Grenoble (LSM), University of Zurich, Niels Bohr Institute, the University of - Large size Skipper CCDs (6k x 4k) Southern Denmark,σ = University 0.06 e !of Santander, Universidad Federal do Rio de Janeiro, Centro Atomico Bariloche 12 • Wafer-layout by for DAMIC-1K development will J.Holland (LNBL) arrive at UW in May/June 2018. 1st skipper CCDs (10g) at UW in summer for testing 15 SENSEISENSEI R&D R&D @ Fermilab at Fermilab

Hidden sector DM. Going beyond the WIMP paradigm is another approach now actively pursued by the international community [28]. In particular, DM particles arising from a hidden-sector (also called dark sector) appear as viable candidates that may have escaped so far detection. Hidden-sector particles have their own set of interaction forces and thus do not couple directly with (are “hidden” from) ordinary matter. Interaction with Standard Model particles may happen as illustrated in Fig. 1, where the mixing of a dark photon A’ with an ordinary photon opens a portal for weak coupling of the hidden-sector with ordinary matter [29]. In addition to acting as a mediator, the dark photon could be itself a DM particle [30]. Hidden-sector particles could then, similarly to WIMPs, play a crucial role in the thermal history of the universe and be a dominant component of DM. The Figure 1. DM scattering phenomenology of hidden-sector DM particles results in a much larger unexplored through a dark photon parameter space than that of WIMPs, in particular for DM masses m << GeV. χ mediator. This has already motivated new approaches to search for hidden particles at accelerators [28]. For direct detection, the nuclear recoil induced by these light DM particles in a detector is so feeble to become undetectable. However, energy transfer in the elastic scattering with electrons is much more efficient [31], allowing direct detection experiments to probe down to mχ ≈ MeV. Also, absorption of dark photon DM results in ionization for mA’ as low as ≈ eV. The current experimental limits on DM-electron scattering cross section as a function of mχ are shown in Fig. 2.a-b for two benchmark hidden-sector scenarios with DM scattering through a heavy or light A’. Also shown are theoretical expectations when assuming that the entire DM density observed today is accounted for by the hidden-sector DM. Notably, a large area of the parameter space is still unexplored. Current limits on dark photon DM are shown in Fig. 3.a. Experimental challenges. Independently of these theoretical motivations, it is important to recognize that current experiments have limited sensitivity to DM-electron interactions, and a light DM particle may have well escaped detection. Most of the interactions result in the production of few charges [31], requiring the detector to be able to resolve individual electrons. In addition, the detector’s dark current – from thermal excitation, or from charge released by traps in the surface and bulk of the detector material or produced by ionizing background particles – must be extremely low for a signal to be recognizable. These are very stringent and difficult to meet constraints. The best limits with the noble liquid technique have been placed by Ref. [32] with XENON10 data [33], but the results are limited by background-induced dark current and were not improved with the tenfold increase in mass of XENON100 (Figs. 2.a and 3.a). Cryogenic crystals are not sensitive to a single electron since the corresponding phonon energy (from the recombination of the electron-holeDAMIC-M pair) is equal to the sensitivityband gap (≈ eV), well belowto theDM detection searches threshold. SuperCDMS may achieve single electron sensitivity by operating in high-voltage mode [25], where the electron and hole drift under the external electric field producing additional phonons; however, the required improvements in phonon resolution and leakage current have not been demonstrated asDM-electron of today. scattering via WIMP search ultra-light hidden photon 10−35 a) b) 10−36

10−37

CRESST (2015) 10−38

−39 CDMS-II Si 2 10 (2013) DAMIC (2016) DAMIC (4.6 kg d) [cm ]

n 40 10− CDMSLite (2015) DAMIC (13 kg d) DAMIC (4.6 kg d) DAMIC (13 kg d) 10−41

10−42 DarkSide-50 DAMIC-M (1 kg y) (2018) 10−43 Xenon experiments 10−44 1 10 [GeV]

Figure 2. a) DM-electron cross section vs mχ for a heavy A’ mediator (mA’ >> keV). Shaded areas are excluded by current experiments (in gray those from accelerators and WIMP nuclear-recoil searches). Orange lines are theoretical expectations16 assuming that χ constitutes all of DM (the scalar and fermion labels refer to the possible particle nature of χ). DAMIC-M 90% C.L. sensitivity is shown as a purple line; b) same as a) for a light A’ mediator (mA’ << keV). In this case, the very weakly coupled χ does not reach thermal equilibrium and its relic abundance comes from a “freeze-in”. 2 Outlook

• DAMICDAMIC is operating collecting with high 40g quality detector data since - few 2017. dru background and 50 eVCollected threshold exposure: - with a 40 g CCD detector at SNOLAB. We will reach ≈ 15- 4.6 kg kg day d so in far, 2018. ~ 13kg day by the end of 2018 High quality data: ‣ 50 eV threshold • These‣ Low data noise will (dominated provide by essential readout) information for the next generation of ‣silicon Few dru detectors background (DAMIC-1K, SuperCDMS): - explore spectrum below 2 keVee in silicon - measureThese data cosmogenic will provide essentialand radiogenic information backgrounds for the next in generation silicon - measureof silicon detectorsCCD dark (DAMIC-M, current atSuperCDMS): the lowest temperatures ‣ spectrum below 2 keV ‣ Cosmogenic and radiogenic background in silicon • Next‣ CCD stage dark is current a kg-size at lowest DAMIC temperature detector to be installed at the LSM in France. Large-size skipper CCDs will be characterized this year. WeNext will stage continue : a kg to size profit DAMIC from detector the current at LSM, setup in France at SNOLAB in this developmentA large skipper stage CCD will be characterized this year (SNOLAB setup continuing running)

17 14 Thank you

18 BACKUP

19 Nuclear-recoil ionization efficiency in silicon

deviation from Lindhard theory observed – crucial for low-mass WIMP searches with silicon detectors 20 HiddenHidden photon photon search search ~1 week of data with 1 CCD. Absorption - -2 -1 Si bulk Leakage current 4 e mm d . of hidden- 675 µm e- photon - - arXiv:1611.03066 - - Ionization accepted by PRL dark matter. Hidden Photon

Pixel distribution consistent with white noise + uniform leakage current. 21 11 ββ coincidences 64 keV 1.2 MeV 57 days of data in 1 CCD: 210 210 210 Pb Bi Po 210Pb < 37 kg-1d-1 τ1/2 = 5 d (95% C.L.) 0.22 MeV 1.7 MeV 32 +110 -1 -1 32Si 32P 32S Si = 80 -65 kg d τ1/2 = 14 d (95% C.L.)

JINST 10 P08014 32 32 30000 25000 20000 15000 10000 5000 0

Si - P candidate 4890 4888 4886 E1 = 114.5 keV 4884

Δt = 35 days 4882 Cluster #79 4880 (xo, y o) 4878 22000 20000 18000 16000 14000 12000 10000 8000 6000 4000 2000 0 4876 Decay 4876 4874 1360 1355 1350 1345 1340 1335 1330

point 4872 4870 Cluster #51

4868 E2 = 328.0 keV

4866 22 1364 1362 1361 1360 1358 1357 1365 1363 1359 1356 1355 12 DAMIC background characterization E = 5.4 MeV E = 6.8 MeV E = 8.8 MeV RUNID= 491, EXTID= 6, cluster_id= 1388 RUNID= 345, EXTID= 6, cluster_id= 1801 RUNID= 490, EXTID= 6, cluster_id= 1345 506 5000 504 5000 5000 504 4500 502 4500 4500 504 502 4000 502 4000 500 4000 3500 3500 500 3500 500 498 3000 3000 498 3000 498 496 2500 2500 2500 496 496 494 2000 2000 2000 494 494 492 1500 1500 1500 492 492 1000 1000 490 1000 490 490 500 500 500 488 0 488 0 488 0 6050 6052 6054 6056 6058 6060 6062 6064 6066 6050 6052 60546056 6058 6060 6062 60646066 60486050605260546056 60586060606260646066 1 Δt = 17.8 d 2 Δt = 5.5 h 3 Three α at the same location!

Powerful method to measure U/Th bkg Example in the bulk – ppt limits 2015 JINST 10 P08014 of α + β

2 Bragg peak 1

23 Not seen 3 32 Flexibility in readout Pixels can be readout in “groups” and the total charge DAMIC estimated in a single measurement. Readout, charge extraction Less pixels but same noise per pixel!

RUNID= 2153, EXTID= 5, cluster_id= 1564 148 50000 146 1x10 144 3x3 40000 142 55 30000 Fe from back: Loss of x, y and140 138 Data shows z information 20000 136 1x1 clear 134 10000

1640 1660 132 1680 1700 1720 1740 0 3165 3170 3175 3180 3185 3190

improvement in 5510

energy 5515 RUNID= 2222, EXTID= 5, cluster_id= 131 5520 1x1

resolution. 5525 5530 5535

5540 α-β coincidence 5 6 7 8 5545 Energy [keV] 24 0 10000 20000 30000 40000 50000

5/6 25 Background, background, background • Lead shielding to stop environmental γ rays Spanish galleon (Chicago) Inner 2” shielding made of ancient lead to avoid bremmstrahlung γs from 210Pb β-decay (22 yrs half-life)

<0.02 Bq/kg Roman ship (Modane, France) 50 Bq/kg

• Material selection and cleaning: copper machining, “secret” recipe etching (surface bkg) Radioactive!

25 SENSEI R&D Fermilab 26