DAMIC DARK MATTER IN CCD

Transcription

1 1 DAMIC DARK MATTER IN CCD ROMAIN GAIOR (LPNHE PARIS) DARK MATTER DAY (APC 2016/12/01)

2 MOTIVATION! A. Aguilar-Arevalo et al. (DAMIC Collaboration) Phys. Rev. D 94, DAMIC (2016) Super CDMS Si HV DAMIC 1kg Ge HV WIMP mass [GeV/c 2 ] 2

3 THE DAMIC COLLABORATION FERMILAB U Chicago U Michigan SNOLAB UNAM (Mexico) FIUNA (Paraguay) UFRJ (Brasil) U. Nacional del Sur Centro Atomico Bariloche U. Zurich LPNHE (Paris 6/7) 11 institutions 8 countries 39 collaborators 3

4 EXPERIMENTAL METHOD 4

5 DETECTION PRINCIPLE coherent elastic scattering CCD y DM Si Si Si Si Si Si Si Si Si pixel Si nuclear recoil electron holes z x 3.77eV / e-h pair (T = 130K) (Si band gap = 1.2eV) Si + Light mass target: dr/de 1/m A Low noise ~2e- = 7.5 ev >low E threshold (~ 0.06 kevee) 5

6 CCD TECHNIQUE PROS AND CONS Pros low E threshold spatial resolution 3D reconstruction energy resolution Cons 1 detection signal (ionisation) timing resolution ~ hours no directionality info compact and cheap detector 6

7 3D RECONSTRUCTION charge diffusion σ along z axis x The DAMIC setup at SNOLAB was devoted to background studies throughout the years , with 7 more than ten installations involving changes to the exmuon track pixel x z y diffusion lim ted σhits y [pix] >6 Ionization [kevee] x [pix] Z σ xy 3D reconstruction surface event tagging FIG. 4. A Minimum Ionizing Particle (MIP) observed in cosmic ray background data acquired on the surface. Only pixels whose values 55 Fe are X above raysthe noise in the image are colored. Darker colors represent pixels with more collected charge. The large area of di usion on the top left corner of the image is where the MIP crosses the back of the CCD. Conversely, the narrow end on the bottom right corner is where the MIP crosses the front of the device. The reconstructed track is shown by the long-dashed line. The short-dashed lines show the 3 band of the charge distribution according to the bestfit di usion model. V. DATA SETS AND IMAGE PROCESSING

8 ENERGY LINEARITY & RESOLUTION 4 55 Fe 3 K 2 Si Esc K K Al Esc K E [kev] Calibration data to X-ray lines Energy resolution from X-ray lines Reconstructed energy / kev 1 C O Al-K 55 Fe Ca-K Si-K 241 Am 1 Energy / kev Var(E) / kev Fano = eV noise 1 Energy / kev = p EF E resolution at 5.9 kev: : 54eVee 8

9 DAMIC CCD developed at LBNL (Microsystem lab) originally for DECam Thick CCD: mm mm 2.9g (5.8g)/ CCD 8 (16) MegaPixels pixel size: 15 x 15 μm High resistivity: -20 kω.cm (low donor density >fully depleted at 40V) low dark current ( -3 e- /pix /day at 120K) 9

10 RADIOGENIC BACKGROUND no effective discrimination nuclear vs electronic recoil potential bkg from β and γ decay chain E = 5.39 MeV E = 6.75 MeV E = 8.66 MeV RUNID= 490, EXTID= 6, cluster_id= 1345 RUNID= 491, EXTID= 6, cluster_id= Th Po Po Δt = 17.8 d Δt < 5.5 h (a) Triple a sequence RUNID= 716, EXTID= 4, cluster_id= (b) a b coincidence unique spatial and energy resolution observe decay chain from a single isotope 238 U and 232 Th decay chain 32 Si chain Figure 7. a) Three a particles detected in different CCD images at the same x-y position. Their energies and the time separation between images are consistent with a sequence from a 232 Th decay chain. b) A peculiar cluster found in a single image, consistent with a plasma a and a b track originating from the same CCD position. This may happen for a radioactive decay sequence occurring within the 8-hour exposure time of an image. same exposure, shown in Figure 7(b). As the energy of the a cannot be measured due to digitizer saturation, this particular decay sequence cannot be identified. 4. Limits on 32 Si and 2 Pb contamination from b decay sequences We have performed a search for decay sequences of two b tracks to identify radioactive contamination from 32 Si and 2 Pb and their daughters, X-ray? whose b spectra extend to the lowest energies and could represent a significant background in the region of interest for the WIMP search. These isotopes do not emit a or penetrating g radiation, n, WIMP? and their decay rates are Diffusion significant for extremely low atomic abundances due to their 0 y half-lives, making conventional screening methods ineffective in determining their presence at the low levels enecessary forlimited a WIMP search. 32 Si is produced by cosmic ray spallation of argon in the atmosphere, and then transported to the Earth s surface, mainly by rain and snow. Detector-grade silicon is obtained through a chemical process starting from natural silica. Therefore, the α 32 Si content of a silicon detector should be close to its natural abundance in the raw silica. Spectral measurements of radioactive background in silicon detectors suggest a rate of 32 Si at the level of hundreds of decays per kg day [15]. μ 32 Si leads to the following decay sequence: particle identification 32 Si! 32 P + b with t 1/2 = 150y, Q value = 227keV (4.1) 32 P! 32 S + b with t 1/2 = 14d, Q value = 1.71MeV (4.2)

11 READ OUT NOISE exposure blank (taken after exposure) Entries per bin Image Blank Gaussian fit mean = = Pixel Value [e ] noise limited by read out improved by CDS (Correlated Double Sampling) limited to 2e- with the current electronics 11

12 STATUS 12

13 alibration 2014 reanalysis 2015 campaign new results (installed Dec12) DAMIC AT SNOLAB Summary InBACK UP searches. addition, neutrons are emitted in (, n)- and s reactions from natural radioactivity (called radiogenic neutrons). Th lower energies of around a few MeV. Dark matter experiments are typically placed at underground la to minimise the number of produced muon-induced neutrons. The dee the experiment, the lower the muon flux. Figure 3 shows the muon fl depth for di erent laboratories hosting dark matter experiments. Th Muon flux [cm-2 s-1] Snolab damic0 matter -6 WIPP/LSBB Kamioka Soudan Y2L Boulby LNGS -7-8 LSM SURF -9 DA M IC SNOLAB Jin-Ping Depth [km w. e.] Installed at Snolab: 2km of norite overburden! 6000m water Figure 3. Muon flux as function of depth in kilometres water for various underground laboratories hosting dark matter exper equivalent depth is calculated using the parametrisation curve (thin line) f / 34 2 km down a mine (6000m water equivalent) 19 muon rate < 0.27 m-2 d-1 (1μ /m2 every 3 days!) 13

14 DAMIC DETECTOR 14

15 & CALIBRATION k(e) / k(5.9 kev ee ) X-rays Optical photons 1 1 Ionization signal [kev ee ] Electronic recoil: linear response down to 40 evee (e- recoil with X-ray and LED at low E) resolution of 54 evee at 5.9keVee Ionization energy / kev ee 1 Ionization efficiency in silicon 9 UChicago 124 Sb Be Antonella (systematic) Antonella Gerbier et al (1990) Lindhard, k=0.15 Lindhard, k=0.05 (Fano factor of 0.133) Nuclear recoil: fast neutron source (2-20 kevnr) photoneutron (0.7-2 kevnr) (Phys. Rev. D 94, ) -1 Deviation from Lindhard model (at 1 Recoil energy / kev r low E) 15

16 DAMIC BACKGROUND SPECTRUM v1 pkg outages v2 pkg outages v3 pkg Pb shield upgrade v4 pkg Jan13 Jun13 Sep13 Jan14 Aug14 Dec14 16

17 ANALYSIS STEPS Entries per bin Blanks (noise) Simulated ionization events Data exposures (1 1) Fit to tail of noise we perform a fit to each cluster and record the LL LL 1. data selection (E < kevee, noisy pixel) σxy [pix] Surface (sim) E [kevee] All data candidates Entries FIG. 7. Lateral spread ( xy ) versus measured energy (E) of the clusters that pass the selection criteria outlin Black (red) markers correspond to candidates in the 1 1 (1 0) data set. Gray markers show the simulated energy deposits near the front and back surfaces of the device. The plot on the right is the projection on the identified clusters. The horizontal dashed lines represent the fiducial selection described in Sec. VII, while the v line shows the upper bound of the WIMP search energy range. VII. REJECTION OF SURFACE EVENTS 2. find hits with LL clustering algo. (comparison bkg vs bkg+signal) 3. exclusion of surface events 4. fit of the candidate spectrum The selection criteria presented in Sec. VI were implemented to distinguish events due to ionization by particle interactions from electronic noise. High-energy photons that Compton scatter in the bulk of the device produce background ionization events with a uniform spatial distribution because the scattering length is always much greater than the thickness of the CCD. Hence, ionization events from Compton scattering are only distinguishable from WIMP interactions through their energy spectrum. Nuclear recoils from WIMP interactions would produce a for events uniformly distributed in the CCD The rejection factor for surface backgroun cial region was estimated by simulating ev front and back surfaces of the CCD. The gr FIG. 7 show the xy versus energy for one ulations, where the interactions were simul <15 µm from the front and back surfaces of the 1 1 data. The rejection factor is >95 electrons with energy depositions >1.5 kev ternal photons with incident energies rejection factor decreases for higher energ 85% at 6.5 kev ee due to their longer absor

18 RESULTS 2 WIMP-nucleon cross-section [cm ] This work 0.6 kg d CRESST II kg d CDMSLite - 70 kg d LUX - 14 ton d DAMA/Na CDMS-II Si kg d WIMP Mass [GeV c ] compatible background hypothesis (Compton scatt.) sensitivity at low mass WIMP ( m χ < GeV/c 2 ) exclusion of a part of CDMSII signal with same target (Si) 18

19 STATUS OF OPERATION April 2016: installation of 6 new CCDs (8 total) replaced copper box and modules replacement of parts of the shielding with ancient lead (Roman lead from Modane) cleaning and etching Issues appeared on 2 CCDs Tests/fix at Fermilab since then due to mechanical stress CCDs (~60g) to be installed in January

20 FUTURE PLANS 20

21 DAMIC FORESEEN SENSITIVITY DAMIC2016 (0.1e- read out noise) target mass to kg scale detector threshold down to ~8 evee (~ 0.3e - ) background ~ 0.01 d.r.u. 21

22 INCREASE MASS current mass: 5.8g /CCD (DAMIC0=>18CCDs) goal: increase CCD mass 3X (DAMIC00=>~50CCDs) ~1mm with same fabrication process ~ few mm with new fabrication process (dev. at U. Chicago) larger format :4k x 4x > 6k x 6k kg scale DAMIC is feasible with current technology in a short time 22

23 & & LOWER THE ENERGY THRESHOLD & BACKGROUND read out noise goal: <0.3e- (w.r.t. 2 now) first amplifier optimisation skipper CCD Digital filtering radio background goal: 0.01 d.r.u (w.r.t. 5 now) Use electroformed copper already one module in test eventually limited by 32 Si background 23

24 DAMIC FORESEEN SENSITIVITY DAMIC2016 (0.1e- read out noise) target mass to kg scale detector threshold down to ~8 evee (~ 0.3e - ) background ~ 0.01 d.r.u. 24

25 CONCLUSION CCD is an efficient DM detector for low mass WIMP stable operation very good energy & spatial resolution After a phase of development / bkg reduction DAMIC has released competitive limits Currently upgrading to DAMIC0 Development for DAMIC1KG: electronics to reduce readout noise CCD fabrication to increase the mass 25

26 THANKS FOR YOUR ATTENTION 26

27 CCD 6 cm CCD Wire bonds Clocks, Bias, and Signal cable Copper frame

28 SIGNAL HYPOTHESIS 28

29 CCD Cu box

30 Stability

31 image example

32 UNDERSTANDING BACKGROUND particle identification decay chain X-ray? n, WIMP? e Diffusion limited E = 5.39 MeV E = 6.75 MeV E = 8.66 MeV RUNID= 490, EXTID= 6, cluster_id= 1345 RUNID= 491, EXTID= 6, cluster_id= Th Po Po Δt = 17.8 d Δt < 5.5 h (a) Triple a sequence RUNID= 716, EXTID= 4, cluster_id= (b) a b coincidence α μ Figure 7. a) Three a particles detected in different CCD images at the same x-y position. Their energies and the time separation between images are consistent with a sequence from a 232 Th decay chain. b) A peculiar cluster found in a single image, consistent with a plasma a and a b track originating from the same CCD position. This may happen for a radioactive decay sequence occurring within the 8-hour exposure time of an image. same exposure, shown in Figure 7(b). As the energy of the a cannot be measured due to digitizer saturation, this particular decay sequence cannot be identified. 4. Limits on 32 Si and 2 Pb contamination from b decay sequences Radiogenic background identification (2015 JINST P08014) Th and U contamination 32 Si bkg estimation We have performed a search for decay sequences of two b tracks to identify radioactive contamination from 32 Si and 2 Pb and their daughters, whose b spectra extend to the lowest energies and could represent a significant background in the region of interest for the WIMP search. These isotopes do not emit a or penetrating g radiation, and their decay rates are significant for extremely low atomic abundances due to their 0 y half-lives, making conventional screening methods ineffective in determining their presence at the low levels necessary for a WIMP search. 32 Si is produced by cosmic ray spallation of argon in the atmosphere, and then transported to the Earth s surface, mainly by rain and snow. Detector-grade silicon is obtained through a chemical process starting from natural silica. Therefore, the 32 Si content of a silicon detector should be close to its natural abundance in the raw silica. Spectral measurements of radioactive background in silicon detectors suggest a rate of 32 Si at the level of hundreds of decays per kg day [15]. 32 Si leads to the following decay sequence: 32

33 DM candidate spectrum 3 ee 2.5 Events per 0 ev E [kev ] ee

34 DAMIC FORESEEN SENSITIVITY 2 WIMP-nucleon cross-section / cm WIMP 90% exclusion limits CDMSLite(2015) DAMIC(2016) 0.6 kg-d DAMIC0(2017) 30 kg-d DAMIC1K 300 kg-d 0.01 dru, 0.5 e - noise target mass to kg scale LUX(2015) DAMA/Na(2009) CRESST(2015) CDMSII-Si(2013) 1-2 WIMP Mass / GeV c detector threshold down to ~8 evee (~ 0.3e - ) background ~ 0.01 d.r.u. 34

35 COMPARISON OF EXPECTED SENSITIVITY DAMIC0 DAMIC2016 DAMIC00 35

36 Electron recoil

37 Axion like particle DAMIC0(2017) DFSZ Axion models CoGeNT(2008) XENON0(2014) Red giant KSVZ DAMIC1K (also millicharged particles)