Results from the LUX Experiment Sally Shaw DMUK Meeting UCL, 18th - - PowerPoint PPT Presentation

Results from the LUX Experiment

Sally Shaw
 DMUK Meeting UCL, 18th January 2016

1

Large Underground Xenon Detector

2

3

The LUX Collaboration

The Black Hills

4

Sanford Lab, South Dakota

5

WIMP-nucleon scattering:

• Spin Independent: scalar,

coherent across nucleus, σ∝A2

• Spin Dependent: axial vector,

needs unpaired nucleon

Direct Detection of WIMPs

6

Need a medium that produces something detectable after a nuclear recoil, and if possible a way to discriminate between signal (DM) and background (𝛿,e-,n) Need a low background environment, well shielded from cosmic rays and with minimal radioactivity

particle/nuclear physics astrophysics detector exposure

LUX TPC

7

250 kg liquid xenon, 122 PMTs, 4850 ft underground

Signal and Background

8

• Electron recoil (ER) tracks:


cylindrical, distance between ions much shorter than e- thermalisation length → S2/S1 higher


99.8% electron recoil (ER) nuclear recoil (NR) discrimination

Nuclear recoil (NR) tracks:
 dense, high recombination, low extraction → S2/S1 lower

Calibrating LUX - Electron Recoils (ER)

32.2 keV 9.4 keV

83Kr 83mKr 83Rb

T1/2 = 1.83 h E = 32.2 keV T1/2 = 154.4 ns
 E = 9.4 keV T1/2 = 86.2 d J = 5/2-

• J = 1/2-
• J = 7/2+
• J = 9/2+
• Tritium (CH3T) - ER band, Ly, Qy

83mKr → position reconstruction, S1 & S2 corrections, field mapping


𝛿-rays → intrinsic Xe isotopes → photon and electron detection efficiencies (g1 & g2, via Doke Plot)

CH3T injected via gas system distributes uniformly within LXe

Emax = 18.6 keV
 <E> = 5.9 KeV

performed every 3 months rubidium source left to decay, then build-up of krypton injected into LUX performed weekly

9

arXiv:1512.03133

Calibrating LUX - Nuclear Recoils (NR)

Neutrons pass through an air-filled acrylic conduit, collimated to 1°

10

Deuterium-deuterium fusion neutrons:
 mono-energetic at 2.45 MeV 
 fired horizontally into the TPC

Beam aligned 15.5 cm below the liquid level within the LUX active region (maximise double scatters)

• Cuts applied ensure 95% of

neutrons are within 4% of 2.45 MeV and are well constructed

• Validated extensively with

Monte Carlo

arXiv:1608.05381

11

Calibrating LUX - Nuclear Recoils (NR)

Charge yield (Qy) measured down to 0.7 keVnr Light yield (Ly) measured down to 1.1 keVnr Lowest ever energies for liquid xenon!

NR band characterised with much better statistics than previously Results confirmed validity of past NR calibrations (AmBe,

252Cf)

arXiv:1608.05381

LUX Timeline

12

Improvements due to:

• Low energy threshold 3 keV → 1.1 keV
• additional 10 days of data
• improved background model
• improved data processing algorithms

→ improved efficiency

2016 (Feb): spin-dependent limits arXiv:1602.03489 arXiv:1310.8214 arXiv:1512.03506 2006 2008 2012 2013 (April) 2013 (Nov) 2014 (Jan) 2014 (Sept) 2015 (Dec) 2016 (May) 2016 (July) 2016 (Sep) 2016 (Oct)

collaboration founded LUX funded LUX moved underground 1st science run starts 85-day limit published grid conditioning 2nd science run starts 2nd science run ends reanalysis limit published (95-day) 332-day limit published LUX decommissioned 427-day limit published

LUX Timeline

13

e-! e-!

Effects of grid conditioning between 1st and 2nd science run:

• Extraction field increased fro .. kV/cm to .. kV/cm
• Electron extraction efficiency increased from

~49% to >70%

• Time and drift dependent field distortions
• These found to be consistent with a build up of

charge on PTFE wall 2006 2008 2012 2013 (April) 2013 (Nov) 2014 (Jan) 2014 (Sept) 2015 (Dec) 2016 (May) 2016 (July) 2016 (Sep) 2016 (Oct)

collaboration founded LUX funded LUX moved underground 1st science run starts 85-day limit published grid conditioning 2nd science run starts 2nd science run ends reanalysis limit published (95-day) 332-day limit published LUX decommissioned 427-day limit published

Solution - 16 “detectors”…

14

4 bins in time, 4 bins in z Treat each bin as a separate detector, with its own calibration and model!

NEST models tuned to data by adjusting E-field and Fano factor

15

3 1.7 9 2.9 15 4 21 5.2 27 6.3 33 keVnr 7.5 8.7 9.8 keVee

S1 (phd) log10[S2 (phd)] 10 20 30 40 50 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4

16

3 1.7 9 2.9 15 4 21 5.2 27 6.3 33 keVnr 7.5 8.7 9.8 keVee

S1 (phd) log10[S2 (phd)] 10 20 30 40 50 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4

Salting the Data

17

Traditional blinding can mask rare backgrounds and pathologies

• > rarely seen in dark matter community
• Instead, we “salted” the data

Fake NR events created from uncorrelated S1s and S2s in Tritium (ER) data Collaboration-wide Turing test done to ensure no one could identify the salt Key to unsalting only known by two people

18

3 1.7 9 2.9 15 4 21 5.2 27 6.3 33 keVnr 7.5 8.7 9.8 keVee

S1 (phd) log10[S2 (phd)] 10 20 30 40 50 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4

19

3 1.7 9 2.9 15 4 21 5.2 27 6.3 33 keVnr 7.5 8.7 9.8 keVee

S1 (phd) log10[S2 (phd)] 10 20 30 40 50 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4

A B C

3 Bad Events

20

80% of light in single top PMT

• consistent with energy

deposited outside of TPC, and light leaked through gap near edge of PMT array

• Concentrated under a few top PMTs
• time structure consistent with

gas scintillation emission

• event came < 1s after high rate
• Do not correspond to interactions within the TPC:


develop loose post-unsalting cuts to target these pathologies cuts were checked for a flat and high signal acceptance on calibration data

21

3 1.7 9 2.9 15 4 21 5.2 27 6.3 33 keVnr 7.5 8.7 9.8 keVee

S1 (phd) log10[S2 (phd)] 10 20 30 40 50 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4

A B C

22

3 1.7 9 2.9 15 4 21 5.2 27 6.3 33 keVnr 7.5 8.7 9.8 keVee

S1 (phd) log10[S2 (phd)] 10 20 30 40 50 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4

Run 4 Limit

23

2006 2008 2012 2013 (April) 2013 (Nov) 2014 (Jan) 2014 (Sept) 2015 (Dec) 2016 (May) 2016 (July) 2016 (Sep) 2016 (Oct)

collaboration founded LUX funded LUX moved underground 1st science run starts 83-day limit published grid conditioning 2nd science run starts 2nd science run ends reanalysis limit published (93-day) 332-day limit published LUX decommissioned 427-day limit published

arXiv:1608.07648

End-of-LUX Calibrations / R&D

24

• D-D Calibrations for low energy measurements & LZ R&D:
• short pulse (S2-only)
• D2O reflector (1-4 keV, 8B ν region)

• Injection campaigns
• Xe-131m (CH3T replacement for LZ)
• C-14 (new energy range),
• Rn-220 (skin calibration for LZ)
• Ar-37 (low E K,L,M peaks, signal physics),

25

2006 2008 2012 2013 (April) 2013 (Nov) 2014 (Jan) 2014 (Sept) 2015 (Dec) 2016 (May) 2016 (July) 2016 (Sep) 2016 (Oct)

collaboration founded LUX funded LUX moved underground 1st science run starts 85-day limit published grid conditioning 2nd science run starts 2nd science run ends reanalysis limit published (95-day) 332-day limit published LUX decommissioned 427-day limit published

Combining the Science Runs

26

1 2 Livedays 95 332 Spatial 
 Co-ordinates 2 (r,z) 3 (r, ɸ, z) Corrections x, y, e- lifetime x, y Backgrounds Xe-127, 𝛿 sources, β sources, B-8 ν,wall 𝛿 sources, β sources, B-8 ν,wall Fiducial mass (kg) 145.4 105.4, 107.2, 99.2 98.4

Treat the 1st as the 17th detector/exposure segment, and treat nuisance parameters in PLR as independent, except for Lindhard k factor (measured with DD data)

FV no longer a simple cylinder - calculate using acceptance of uniform 83mKr events and multiply by total LXe mass for each time bin Decayed away by the start of the 2nd run (half life ~36 days)

Combined Run 3 + Run 4

27

arXiv:1608.07648 2006 2008 2012 2013 (April) 2013 (Nov) 2014 (Jan) 2014 (Sept) 2015 (Dec) 2016 (May) 2016 (July) 2016 (Sep) 2016 (Oct)

collaboration founded LUX funded LUX moved underground 1st science run starts 85-day limit published grid conditioning 2nd science run starts 2nd science run ends reanalysis limit published (95-day) 332-day limit published LUX decommissioned 427-day limit published

28

Lots more interesting analyses and papers still to come from LUX → axions, spin-dependent run 4 limits, 83mKr, DD, 134Xe 0νββ decay, position reconstruction, double e- capture, pulse shape discrimination, S2-only, inelastic DM, modulations…


2006 2008 2012 2013 (April) 2013 (Nov) 2014 (Jan) 2014 (Sept) 2015 (Dec) 2016 (May) 2016 (July) 2016 (Sep) 2016 (Oct)

collaboration founded LUX funded LUX moved underground 1st science run starts 83-day limit published grid conditioning 2nd science run starts 2nd science run ends reanalysis limit published (93-day) 332-day limit published LUX decommissioned 427-day limit published

It’s not over yet!

Back-up

29

• Scintillates in the VUV (178 nm)
• Very transparent to its own light
• Massive (A =131) - enhances spin-independent

scattering

• Isotopes with unpaired nuclei (129Xe - 29.5%,

131Xe - 23.7%) for spin-dependent sensitivity

• Self-shielding is extremely effective at reducing

background

Xenon

30

Triplet 27ns Singlet 3ns

recombination

Recoil Excitation Ionisation Xe* Xe+ Xe2+ + e- Xe** + Xe

+Xe +Xe

Xe2 2Xe 2Xe

Kr-83m Calibrations

Krypton calibrations done weekly throughout operation of LUX. 83mKr generated with a Rubidium source.

• Low energy → no PMT saturation

Mixes homogeneously with LXe → can use to check electric field models 
 
 Used for accurate measurements of:

• Electron livetime
• Detector leveling
• S1 xyz light collection
• S2 xy light collection

WIMP search resumes after ~8 hours (~4 half lives). Total energy ~41.6 keV is outside of WIMP search region of interest, events clearly identifiable by two S1s

83Kr 83mKr 83Rb

T1/2 = 1.83 h E = 32.2 keV T1/2 = 154.4 ns
 E = 9.4 keV T1/2 = 86.2 d J = 5/2-

• J = 1/2-
• J = 7/2+
• J = 9/2+
• 32.2 keV

9.4 keV

31

Tritium Calibration - ER Band

32

S1 vs log10(S2/S1): 
 ER/NR discrimination space

Quantisation at low S1 is due to digital spike counting of photons

— ER mean


• - ER 10% and 90% contours

— NR mean
 — energy contours

• - S2 threshold of 165 detected photons

Discrimination: 99.81 ± 0.02 (stat) ± 0.1 (sys)% arXiv:1512.03133

Tritium Ly and Qy

33

data at 180 V/cm data at 105 V/cm NEST at 180 V/cm NEST at 105 V/cm bands are 1σ

Ly relative to 32.1 keV 83mKr decay

Aprile et al (zero field)
 Baudis et al (zero field) Baudis et al (450 V/m)

Tritium Efficiency

34

LUX used to use AmBe and 252Cf neutrons for nuclear recoil band calibrations (original run 3 result)

Previous NR Calibrations

35

10-4 10-3 10-2 10-1 1 2 3 4 5 6 7 8 9 10 probability energy [MeV] Spectra for spontaneous Cf252 fission 1 neutron emitted 2 neutron emitted 3 neutron emitted 4 neutron emitted 5 neutron emitted 6 neutron emitted 7 neutron emitted 8 neutron emitted

log10(S2b/S1) xyz corrected

DD Calibration

36

37

Qy determination: Calculate ne using S2 size, extraction efficiency and single electron size Bin in keVnr (0.7-24.2 keVnr) and fit signal model to each bin Unbinned extended maximum likelihood (resolution effects as nuisance parameter)

• Uncertainties on Qy:
• Position reconstruction (worsens as S2s get smaller)
• Neutron beam entry point
• Neutron beam energy spectrum
• g2 measurement
• Eddington bias
• Statistical fluctuations in recombination

DD Calibration - Ly and Qy

Ly determination: Model S1 and S2 distributions of single scatters using NEST with measured Qy Bin data in S2s, compare S1 distribution to model
 Ly measurement used S2s of 50 - 900 phd (900+ used for normalisatioin), corresponds to ~0-20 keV Fit each bins ith maximum-likelihood optimisation of the simulated S1 spectrum, extract nᵧ

• Uncertainties on Ly:
• Qy energy scale
• S1 corrections
• g1 measurement

38

effect of 2 phe emission

Run 3 Improvements

Run 3 threshold: 
 3 keV Reanalysis threshold: 1.1 keV

better background model → upper S1 threshold increased from 30 photoelectrons to 50

Other Improvements

39

• More accurate estimation of detected

photons:

• Accounted for probability of 2 photoelectron

emission per photon at the photocathode in a PMT (VUV effect)

• Digital “spike counting” of individual photons
• Improvements to data processing:
• Pulse finding and classification algorithms

improved

• Position reconstruction improved
• Better cuts to remove topologies such as “gas

events”

• More livetime - further 10 days:
• Better understanding of krypton calibration

data (low rates can still be used for WIMP search as is high energy)

• Improved background model allowing

a larger fiducial volume (118kg → 147kg)

• Total livetime 1.4⨉104 kg·days

Reanalysis data

Run 4 Background Modelling

40

Post-unsalting Cuts

41

A B C A B C

Based on applying a maximum amount of light in one PMT and signal promptness cut (S1 signals are very prompt)