Latest results from the DEAP-3600 dark matter search at SNOLAB On - - PowerPoint PPT Presentation
Latest results from the DEAP-3600 dark matter search at SNOLAB On behalf of the DEAP-3600 collaboration DMUK Meeting, Kings College London, 11th April 2019, R. Ajaj et al. Ashlea Kemp, https://arxiv.org/pdf/1902.04048.pdf Royal Holloway,
DMUK Meeting, Kings College London, 11th April 2019, Ashlea Kemp, Royal Holloway, University of London.
1
https://arxiv.org/pdf/1902.04048.pdf
On behalf of the DEAP-3600 collaboration
2
3
holding 3279 kg of target LAr,
recoiling Ar nuclei after interaction with dark matter particle (WIMP),
➡ 128 nm VUV scintillation photons (ɣ) wavelength shifted to
420 nm by TPB layer for PMT detection (~75% coverage).
4
holding 3279 kg of target LAr,
recoiling Ar nuclei after interaction with dark matter particle (WIMP),
➡ 128 nm VUV scintillation photons (ɣ) wavelength shifted to
420 nm by TPB layer for PMT detection (~75% coverage).
5
γ γ γ γ γ γ γ γ γ γ γ
holding 3279 kg of target LAr,
recoiling Ar nuclei after interaction with dark matter particle (WIMP),
➡ 128 nm VUV scintillation photons (ɣ) wavelength shifted to
420 nm by TPB layer for PMT detection (~75% coverage).
Prompt light Prompt + Late light = PSD
➡ Pulse-shape discrimination (PSD).
6
http://darkmatter.ethz.ch/ Published PSD paper from DEAP-1: Astroparticle Physics 85 (2016) 1-23.
world,
7
https://phys.org/news/2018-05-world-sensitive-dark.html
from cosmic rays,
produce neutrons…
➡ Neutrons mimic WIMPs!
8
* Data used for first dark matter search: Phys. Rev.
9
t
Initial design phase begins
2006
First injection with purified argon
May 2016 Aug 2016
First physics data collected*
Nov 2016 - Dec 2017
Continue to collect physics data (open)**
Jan 2018 - Present
Collect further physics data with blinding scheme (20%
** Data used for second dark matter search: https:// arxiv.org/pdf/1902.04048.pdf (2019), “Design and Construction of the DEAP-3600 Dark Matter Detector”: Astropart. Phys. 108 (2019) 1-23.
10
80+ researchers
11
a range of time,
➡ More photons emitted, hence detected, later from electronic recoils
ERs (β/ɣ) compared to nuclear recoils NRs (neutrons, WIMPs),
12
ER + NR pile-
Data from AmBe (neutron emitter) run.
Fprompt = ∑60ns
t=−28ns PE(t)
∑10μs
t=−28ns PE(t)
➡ For ER-type event for which q PE reconstructed, the probability of observing given Fprompt value, f,
is described by convolution of Gamma function with the Gaussian smearing term.
13
FER( f, q) = Γ( f; ¯ f, b) * Gauss( f; σ)
Mean Fprompt, ¯ f(q) Shape parameter, b(q)
Width of Gaussian response, σ(q)
➡ Trigger efficiency based on prompt PE, ➡“Lose” events, data and model diverge.
matter WIMP search.
14
which we expect 90% (green) and 50% (purple) of NRs,
probability of 2.8+1.3-0.6 x 10-7 at low energy threshold for WIMP search.
‘Leakage’ probability: probability of ER type event being detected above given Fprompt value at the low energy WIMP threshold (15.6 keVee),
In energy range of WIMP search region [15.6 - 32.8 keVee], average PSD leakage probability of 4.1+2.1-1.0 x 10-9 at 90% NR acceptance.
used:
PMT hits),
information of early pulses used to determine position).
Ar39 β decays in data distributed uniformly across LAr volume (see back- up slides).
15
➡ Apply radial cut of 630 mm to remove Cherenkov, external neutron and surface ⍺ events that reconstruct
with R > 630 mm,
➡ Position resolution of 35 mm for events near radius of 630 mm in WIMP PE search region, ➡ Leakage probability at 630 mm radial cut from surface ⍺ decays into ROI is ~ 10-5
‘Leakage’ probability of simulated ⍺ decays in WIMP PE range vs contained LAr mass as determined by events within given reconstructed radius.
WIMP ROI driven by signal and background models, 2D WIMP ROI drawn in PE, Fprompt, Designed such that background expectation < 1.
16
LAr: Ar39 β decays, ⍺ decays from Rn222/ Rn220, Acrylic Vessel (AV) surface: Po210 ⍺ decays.
17
External: Cosmogenic- induced neutrons produced inside water tank/ rock.
PMTs & other detector components: Radiogenic neutrons, ɣ/β produced in glass…
➡ Cherenkov light
produced in light guide acrylic.
Neck: Po210 ⍺ decays through LAr ‘film’ on surface of acrylic flowguides, originating from long-lived Pb210 (Rn222).
18
LAr: Ar39 β decays, ⍺ decays from Rn222/ Rn220, Acrylic Vessel (AV) surface: Po210 ⍺ decays.
19
The three decay scenarios for surface ⍺ in DEAP-3600 (decays in the LAr, the TPB layer or the AV acrylic) Image taken from the PhD thesis of P . Giampa.
AV surface Po210 ⍺ decays with degraded energies arise from long-lived Pb210 nuclei residual on AV surface.
LAr ⍺ decays from Rn222/ Rn220 chain, Produce high energy events which reside outside of the WIMP PE search region.
20 Lowest bulk Rn contamination of any noble liquid experiment Rn222 LAr: (0.153 ± 0.005) µBq/kg
AV surface Po210 ⍺ decays with degraded energies arise from long-lived Pb210 nuclei residual on AV surface.
LAr ⍺ decays from Rn222/ Rn220 chain, Produce high energy events which reside outside of the WIMP PE search region. Predict < 0.08 events in ROI from surface ⍺
21
External: Cosmogenic- induced neutrons produced inside water tank/ rock.
LAr: Ar39 β decays, ⍺ decays from Rn222/ Rn220, Acrylic Vessel (AV) surface: Po210 ⍺ decays.
PMTs & other detector components: Radiogenic neutrons, ɣ/β produced in glass…
➡ Cherenkov light
produced in light guide acrylic.
22
Cosmogenic neutrons: Veto 0.1 s before a Muon Veto PMT is fired and Veto 1 s after a Muon Veto PMT is fired to mitigate cosmogenic neutron backgrounds. Radiogenic neutrons: Estimate radiogenic neutron background expectation using data-driven neutron capture coincidence tagging analysis,
➡ After scattering in LAr, neutrons thermalise and capture, to produce ɣ ray.
Courtesy of Andrew Erlandson.
Cherenkov: Physics data with U232 calibration source (2.6 MeV ɣ’s) deployed collected,
➡ Develop cut flow to remove Cherenkov.
Background expectations: Cosmogenic n: < 0.11 ROI events, Radiogenic n: 0.10+0.10-0.09 ROI events, Cherenkov: < 0.14 ROI events.
LAr: Ar39 β decays, ⍺ decays from Rn222/ Rn220, Acrylic Vessel (AV) surface: Po210 ⍺ decays.
23
External: Cosmogenic- induced neutrons produced inside water tank/ rock.
Neck: Po210 ⍺ decays through LAr ‘film’ on surface of acrylic flowguides, originating from long-lived Pb210 (Rn222).
PMTs & other detector components: Radiogenic neutrons, ɣ/β produced in glass…
➡ Cherenkov light
produced in light guide acrylic.
background rate is Po210 ⍺ decays occurring on surfaces of acrylic flowguides in the neck,
➡ Thin layer of LAr on surfaces needed to produce
detectable scintillation photons & high Fprompt events,
shadowing effects,
24
show that events can be up to 5000 PE,
correlation between PE and reconstructed Z position.
25
Simulation ‘Sample’ region in data fit to obtain rate from each surface component.
Data
Image taken from paper:
‘Sample’ region: control regions, motivated by MC ‘Background’ region: sideband regions, chosen outside PE range of neck alpha background and cover same fiducial volume as control regions.
reconstruction algorithms (see Slide 16),
➡ Algorithm using PE + timing information is more sensitive to early arriving photons, thus reconstructing
these events higher along the Z-axis,
➡ Difference in reconstructed Z positions can offer level of separation between neck alphas and WIMP
events.
26
Predict 0.49+0.27-0.26 events in ROI from neck alphas.
Difference between reconstructed Z position from algorithm using PE and timing information and reconstructed Z position from algorithm using only PE information.
in GAr PMTs,
than expected from WIMPs.
➡ Expected background contribution < 1 event.
27
cuts are the most severe effect on WIMP acceptance.
For WIMPs inside the ROI, the average detection efficiency inside the fiducial volume after background rejection cuts is 35.4%.
Spatial distribution for events in data surviving all cuts expect from reconstructed radius (R < 630 mm).
28
29
30
cm2 for a 100 GeV WIMP at 90% C.L,
31
x 10-45 cm2 for a 100 GeV WIMP at 90% C.L,
32
➡ Will collect data until end of 2020,
factor to WIMP sensitivity,
which plans to improve WIMP sensitivity down to neutrino floor.
33
34
➡ Simulation propagates detector timing response and PMT effects (such as after-pulsing) into output
Fprompt distribution,
➡ Shape and Gaussian width parameters governed by same fit parameters as ERs.
35
¯ f
* Reference here: H. Cao et al. (SCENE Collaboration), Phys. Rev. D 91, 092007 (2015).
¯ f
FNR( f, q) = Γ(1 − f; 1 − ¯ f, b) * Gauss( f; σ)
Mean Fprompt for NRs, ¯ f(q) Shape parameter, b(q)
Width of Gaussian response, σ(q)
36
Fprompt distribution from AmBe data from calibration source (NRs) along with simulations of AmBe source, simulations of single-scatter neutrons, the ER PSD model and their sum.
GEANT4/RAT simulations used to build model, For radiogenic neutrons, components normalised to material assay values.
hits),
position).
decays in data distributed uniformly across LAr volume, split into two ‘pseudo-events’,
reconstructed distances between pseudo-events, as function of average pseudo-event PE and
37
➡ Apply radial cut of 630 mm to remove Cherenkov, external neutron and surface ⍺ events that reconstruct
with R > 630 mm,
➡ Position resolution of 35mm for events near radius of 630 mm in WIMP PE search region, ➡ Leakage probability at 630 mm radial cut from surface ⍺ decays into ROI is ~ 10-5
38
Courtesy of Rob Stainforth [Lake Louise Conference, February 2019].
Region A Observe: 1 Predict: 0.46+0.13-0.18 Region B Observe: 4 Predict: 1.25+0.26-0.42 Large variations in neck alpha light yield and reconstructed position required to predict event rates consistent with observation in Region B, No significant effect on WIMP exclusion presented. Future analyses: include additional background sources above upper PE bound of WIMP ROI.