Direct Dark Matter Searches with LUX Cláudio Silva on behalf of the LUX Collaboration Rencontres de Moriond, La Thuile, Italia, 20th March 2017 Very High Energy Phenomena in the Universe Snow in Summer - Location of the Sanford Underground Laboratory (and LUX)
Outline 2 The LUX detector - two phase Liquid/Gas xenon time projection chamber o Direct dark matter detection o How LUX detector works The LUX calibrations Krypton ( 83m Kr); Neutron recoils: D-D Electron recoils: 3 H 332 live-days second science run (2014-2016) results o LUX backgrounds o Main dark matter search analysis o Spin independent o Spin dependent (NEW!) Image: LUX inside the water tank (September 2012)
Dark Matter Evidence 3 Rotation curves of galaxies Bullet-cluster: Cosmic Microwave DM not Background Planck NGC-3198 Galaxies Surveys Gravitational lensing Bullet Cluster Background figure: X-ray: NASA/CXC/M.Markevitch et al. Optical: NASA/STScI; Magellan/U.Arizona/D.Clowe et al. Lensing Map: NASA/STScI; ESO WFI; Magellan/U.Arizona/D.Clowe et al. Dark Matter 5.44 ± 0.14 Ordinary Matter
Cold Dark Matter Particle Candidates 4 owimps (weakly interactive massive particles): o Stable and neutral in most scenarios; o Solves the Gauge Hierarchy Problem o WIMP Miracle (density from freeze out) o Physics beyond the standard model: o Super-symmetry - neutralino oaxions (solution to the strong CP problem - Peccei-Quinn solution) oand many, many others XENON-0 o superwimps, light gravitinos, branons, Sterile Neutrinos, Kaluza- Klein bosons
Dark Matter Detection 5 LUX is a Direct Detection exp. o Galactic WIMPs scatter with nucleus of the target o Spin independent (M) - proport. to A 2 o Spin dependent interaction (Σ and Σ ) o Other effective field interactions o Axions are detected through axioelectric effect Challenges o Isothermal model: expect recoil < kev requiring detectors with a very low threshold. o Challenge backgrounds o Go underground, passive and active veto, careful selection of materials with very low background, discriminate nuclear recoils from electron recoils Why Xenon as a Target Nucleus Good for most EFT interactions o High atomic mass (A=131) o Odd-neutron isotopes ( 129 Xe, 131 Xe) enhance spin-dependent sensitivity studies High density (2.9 g/cm 3 ) High light yield No intrinsic backgrounds
Liq. Xenon Time Projecting Chamber 6 Energy depositions produce light and charge o Prompt scintillation (S1) o Proportional scintillation (S2): Measurement of the electrons extracted from the liquid to the gas 3D Position Reconstruction o Depth obtained from the time difference between S1 and S2 - called here drift time o XY reconstructed from the S2 light pattern Ratio of charge to light (S2/S1) is a discriminator against backgrounds (>99%): o Nuclear Recoil (NR): WIMPs and neutrons interact with nuclei - short, dense tracks o Electronic Recoil (ER): axions, γs and e- interact with the electrons - longer, less dense tracks TPCs are scalable with improvement of performance
The LUX Experiment 370 kg Liquid Xenon Detector (59 cm height, 49 cm diameter) o 250 kg in the active region (with field) Construction materials chosen for low radioactivity (Ti, Cu, PTFE) ~47 cm 59 cm (cath.-surface) 49 cm 122 ultra lowbackground PMTs (61 on top, 61 on bottom) observe both S1 and S2 7 Active region defined by PTFE reflectors (high reflectivity >97%) - high light collection)
Typical LUX Pulses S1 - Prompt scintillation Sharp rise with a exponential decay o Pulse FWHM: ~0 ns ~60-90% of light in bottom PMTs o Ratio depends on the depth of the event Threshold of 2 detected photons 1.5 μs All PMTs Signal sum S2 - Electroluminescence Near-gaussian pulse shape o pulse width depends on the depth (z) ~57/43% (top/bot.) light in PMTs ~25 phd per extracted electron Threshold of 200 phd (WIMP-Search) Phd stands for detected photons 8 5.5 μs
LUX AT SURF 9 (Sanford Underground Research Facility) Sanford Underground Research Facility Lead, South Dakota, USA. Former Home of the Homestake Solar Neutrino Experiment 1970-1994 1478 m deep (4300 m.w.e.) μ flux reduced x -7 compared to sea level) Raymond Davis (Nobelpriset i fysik 2002)
Timeline 2006-2016 11 Two main scientific runs First Science Run (FSR): 2013/04-2013/09, 95 live-days Second Science Run (SSR): 2014/09-2016/09, 332 live-days 2008: LUX funded (DOE+NSF) 2013 (Apr): First Science Run (FSR) starts 2014 (Jan): 2016 (May): Grid SSR finished - conditioning 332 live-days 2014 (Sep): Second Science Run (SSR) Starts 2016 (Sep): LUX decommission 2006: LUX collab. formed 2012 (Jul): LUX moves underground WIMP nucleon cross section (cm 2 ) 2013 (Nov): First Science Run results reported 44 45 40 42 44 1 2 3 m WIMP (GeV/c 2 ) 6 8 12 PRL, 112, 091303 2014 WIMP nucleon cross section (zb 2015 (Dec): 3-month run reanalysis posted 3 2 1 0 1 8 B CDMSlite 2 SuperCDMS 2014 PandaX 2015 DarkSide 50 2015 XENON0 2012 LUX 2014 This Result 1 2 3 m WIMP (GeV/c 2 ) 42 43 44 45 PRL, 116, 161301 2016 PRL, 116, 161302 2016 WIMP nucleon cross section (cm 2016 (Jul): SSR results released 332 live-days PRL, 118, 021303, 2017
Krypton Calibrations 12 83m Kr injected in the gas system and decaying uniformly inside the detector o 83m Kr emits two gamma rays E γ,1 = 32.2 kev (T 1/2 = 1.83 h) and E γ,2 = 9.4 kev (T 1/2 = 154 ns) o 1 to 2 times a week 83m Kr used to o Develop S1 and S2 position corrections: both S1 and S2 pulses depend on the location of the event due to geometrical light collection and electronegative impurities. o Map variations of the electric field in the detector o Develop and test the position reconstruction: krypton data is used to get the light response functions (LRFs)of the PMTs. These functions are found by iteratively fitting the distribution of S2 signal for each PMT. Krypton data 83m Kr (Drift Time 4-8!s) Second Science Run The large difference between the drift field (180 V/cm) and the extraction field (2.8 kv/cm in liquid) causes the the drift field lines to be compressed as they pass through the gate plane; any electrons leaving the drift volume appear only in narrow strips between each pair of gate wires.
ER Calibrations 13 Tritium is an ideal source for determination of the detector s electron recoil band and low energy threshold o E(max) - 18.6 kev, <E> - 5.9 kev o β decay with T (1/2) = 12.6 a - Long Lifetime Tritiated methane was injected in the system and removed by the getter. ER calibrations performed every three months Phys. Rev. D 93, 072009 (2016) Light Yield (ph/kev) Charge Yield (e-/kev) gas Charge Light
NR Calibrations 14 Deuterium-Deuterium neutron Generator installed outside LUX water tank The 2.45 MeV emitted neutrons are collimated to the level of ~1 degree Two analysis are performed o Double-scatters - ionization yield Q y (0.7 to 74 kev nr ) o Single-scatters - scintillation yield L y and NR band calibration (1.1 to 74 kev nr ) NR calibrations every three months and at different depths. Ionization scintillation Efficiency https://arxiv.org/abs/1608.05381
4 3.8 3.6 LUX 2014/2016 Detector s Response 15 log [S2 (phd)] 3.4 3.2 3 2.8 2.6 2.4 ER Band Mean NR Band Mean ER Band 80% contours NR Band 80% contours 2.2 0 20 30 40 50 S1 (phd) phd stands for detected photons
4 3.8 3.6 LUX 2014/2016 Detector s Response 2.9 4 5.2 6.3 7.5 8.7 9.8 kevee 16 log [S2 (phd)] 3.4 3.2 3 2.8 2.6 2.4 1.7 15 21 27 33 kevnr Energy contours (constant number of quanta) 3 9 2.2 0 20 30 40 50 S1 (phd)
log [S2 (phd)] 4 3.8 3.6 3.4 3.2 3 2.8 2.6 2.4 1.7 2.9 3 4 5.2 6.3 9 7.5 LUX Salting 8.7 15 9.8 kevee A common approach is to blind oneself to events in the signal regions but it often blinds us to rare backgrounds and pathologies 27 21 Instead of traditional blinding, we employ a technique where fake signal events ( salt ) are injected into data stream. NOT SIM!! 33 kevnr 2.2 0 20 30 40 50 S1 (phd) 17
S2 Quality Cuts and Efficiency 18 After single-scatter event selection (1 S1 + 1 S2), fiducializing, and cutting out periods of detector instability, few pathologies remain. These pathologies are targeted applying cuts to the S2: o Position reconstruction χ 2 cut analysis the PMT hit pattern. It removes single-electron pile-up, additional multiple scatters (x,y separated) and PMT afterpulsing. o S2 waveform cuts to remove misclassified gas events and merged S2s from multiple scatters (z separated). Efficiencies calculated with calibration data! Position Reconstruction χ2 Cut S2 Gaussian Fit Width S2 Gaussian Sigma Cut
Estimation of Backgrounds 19 Background source Expected number below NR median External Gamma Rays 1.51±0.19 Internal Betas 1.20±0.06 238 U late chain wall back. 8.7±3.5 Accidental S1-S2 0.34±0. Solar 8 B neutrinos 0.15±0.02 Bulk volume, but leakage at all energies Low-energy, but confined to the edge of our fiducial volume In the bulk volume, lowenergy, in the NR band These figures are figure of merit only. In our analysis we use a likelihood analysis. o + ~ 0.3 single scatter neutrons, e.g. from (α, n), not included in PLR
LUX Likelihood Analysis 20 A profile-likelihood test (PRL) was implemented to compare the models with the observed data 5 un-binned PLR dimensions o z/drift time, r, ϕ, S1 and log (S2) 1 binned PLR dimension: o Event date Detector s response (S1,S2) modeled with NEST (Noble Element Simulation Technique) with input from our situ calibration data o See M. Szydagis 2013 JINST 8 C003 Data in the upper-half of the ER band were compared to the model (plot at right) to assess goodness of fit. Good agreement with background-only model, p-value >0.6 for each projection. Events Events Events Events Events 30 25 20 15 5 0 5 15 20 25 30 35 40 45 50 S1 c (phd) 0 80 60 40 20 0 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4 log(s2 ) c 0 80 60 40 20 0 0 2 4 6 8 12 14 16 18 20 r (cm) 140 120 0 80 60 40 20 0 250 200 150 0 50 0 50 0 150 200 250 300 drift time (µs) 3 2 1 0 1 2 3 phi (radian) Observed data Background only model
4 3.8 3.6 LUX 2014/2016 Detector s Response 9.8 kevee 8.7 7.5 6.3 5.2 4 2.9 21 log [S2 (phd)] 3.4 3.2 3 2.8 2.6 1.7 21 27 33 kevnr 2.4 15 3 9 2.2 0 20 30 40 50 S1 (phd)
log [S2 (phd)] 3.8 3.6 3.4 3.2 2.8 2.6 2.4 WIMP-search data from 332 live days 4 3 1.7 2.9 3 4 5.2 6.3 9 7.5 8.7 9.8 kevee 21 15 Black Dots: Second Science Run Data (Salt Included) Open circles: High radius events 27 33 kevnr 2.2 0 20 30 40 50 S1 (phd) 22
log [S2 (phd)] 4 3.8 3.6 3.4 3.2 3 2.8 2.6 2.4 WIMP-search data - Salt Identified 9.8 kevee 8.7 7.5 6.3 5.2 4 2.9 1.7 27 21 15 3 9 33 kevnr Salt - blue dots 2.2 0 20 30 40 50 S1 (phd) 23
log [S2 (phd)] 3.8 3.6 3.4 3.2 2.8 2.6 WIMP-search data - Pathological Events 4 3 1.7 2.9 4 5.2 A 6.3 7.5 8.7 red dots: 2 populations of rare pathological events were identified contributing 3 sub- 15 NR-band events 9.8 kevee A B 21 top PMT array 27 33 kevnr 2.4 C 3 9 2.2 phe samples ( ns) 0 20 30 40 50 S1 (phd) C phe / sample phe / sample phe / sample B 24
Post-Unsalting Quality Cuts 25 Two additional cuts on the S1 pulse were implemented. Flat signal acceptance of 98.5% when both cuts are applied to the DD and Tritium data Density map: CH 3T calibration data x : WS data passing S1 cuts : WS data cut by S1 Max. PMT Area cut x : WS data cut by S1 Prompt Fraction cut Density map: CH 3T calibration data x : WS data passing S1 cuts : WS data cut by S1 Max. PMT Area cut x : WS data cut by S1 Prompt Fraction cut Removes events with S1 that has gasevent-like time structure Removes events with S1 light overly concentrated in a single PMT
log [S2 (phd)] 3.8 3.6 3.4 3.2 2.8 2.6 2.4 WIMP-search data from 332 live days 4 3 1.7 2.9 4 3 5.2 6.3 9 7.5 8.7 15 9.8 kevee 27 33 kevnr 21 Pathological events removed p-value = 40% consistent with background-only hypothesis (includes also information on r, z, φ) 2.2 0 20 30 40 50 S1 (phd) 26
WIMP-nucleon SI Exclusion - SSR 0.22 zeptobarns (at 50 GeV/c 2 ) Brazil bands show our 1 and 2 σ regions We observed an improvement of a factor of four for high WIMP masses compared with the results from the first science run (PRL, 116, 161301 (2016)).
SI Exclusion - FSR+SSR 28 0.11 zeptobarns (at 50 GeV/c 2 ) WIMP nucleon cross section [zb] 3 2 1 0 1 2 8 B Both LUX Runs Combined SUSY-cMSSM (1 σ) DarkSide 50 2015 XENON0 2016 PandaX II 2016 LUX WS2013 LUX WS2014 16 LUX WS2013+WS2014 16 1 2 3 4 5 WIMP Mass [GeV/c 2 ] 42 43 44 45 46 o http://journals.aps.org/prl/abstract/.13/physrevlett.118.021303 LUX now excludes significant portions of the 1-sigma regions for WIMPs favored by certain supersymmetric models. WIMP nucleon cross section [cm 2 ]
Spin-Dependent Neutrons (FSR+SSR) 29 1.6-41 cm 2 (at 35 GeV/c 2 ) WIMP neutron cross section [pb] 1 2 3 4 5 6 g q = g c = 0.25 ATLAS CMS g q = 0.25,g c = 1 gq = gc = 1.45 CMS XENON0 LUX WS2013 PandaX-II LUX WS2013+WS2014 16 Gray regions: MSSM-15 https://arxiv.org/abs/1405.0622 LUX Preliminary 1 2 3 4 5 WIMP Mass [GeV/c 2 ] 37 38 39 40 41 42 WIMP neutron cross section [cm 2 ] Both runs combined We observed an improvement of a factor of six compared with the results from the first science run (PRL, 116, 161302 (2016)).
Spin-Dependent Protons (FSR+SSR) 30 5-40 cm 2 (at 35 GeV/c 2 ) WIMP proton cross section [pb] 0 1 2 3 4 5 PICO-2L PICO-60 DAMA SuperK (t t) PandaX-II IceCube (b b) LUX WS2013 LUX WS2013+WS2014 16 IceCube (W + W ) IceCube (t t) LUX Preliminary 1 2 3 4 5 WIMP Mass [GeV/c 2 ] 36 37 38 39 40 41 WIMP proton cross section [cm 2 ] We observed an improvement of a factor of six compared with the results from the first science run - (PRL, 116, 161302 (2016))
The LUX-ZEPLIN Experiment 31 Turning on by 2020 with 1,000 initial live-days plan In the same location of LUX tons total, 7 tons active, ~5.6 ton fiducial Unique triple veto system LXe heat exchanger tower Instrumentation conduits Existing water tank Gd-loaded liquid scintillator 120 outer detector PMTs (Main Detector) 2-phase XeTPC 494 (131) TPC (Xe skin) PMTs n tubes
The LUX-ZEPLIN Experiment -3-4 -5-6 LZ 90%CL Median (Baseline) LZ 90%CL Median (Goal) CMSSM (1σ) CMSSM (2σ) Zeplin-III (2012) XENON0 (2016) PandaX (2016) LUX WS2013+WS2014-16 -39-40 -41-42 SI ) [pb] (σ p log -7-8 -9 - -43-44 -45-46 ] 2 [cm SI σ p -11-47 -12-13 -14 ν-n coherent scattering 1 event 3σ significance 00 tonne-years WIMP mass [GeV/c 2 2 ] 3-48 -49-50 (LZ 5.6 Tonnes, 00 live days)
Conclusions 33 LUX has since 2013 the world-leading result in the dark-matter research. LUX had significant improvements in the calibration of xenon detectors - essential to improve detector s sensitivity. The LUX s 332 live-day search, cutting into un-probed parameter space. Excluding SI WIMPs down to 0.22 zeptobarns (2.2x -46 cm 2 ) at 50 GeV/c 2. When both runs are combined SI WIMPs are excluded down to 0.11 zeptobarns at 50 GeV/c 2. Spin-Dependent results show a cross section of 1.6-41 cm 2 (at 35 GeV/c 2 ) for neutrons (most sensitive constraint to date) and a cross section 5-40 cm 2 (at 35 GeV/c 2 ) for protons. Results available on: o http://journals.aps.org/prl/abstract/.13/physrevlett.118.021303 More analysis forthcoming o Axion searches/alp, effective field theory, neutrino less double beta decay, additional calibrations etc. Onwards and downwards: LUX-ZEPLIN (LZ) experiment under construction, 7 tonne active mass (2020).
Thanks! Obrigado!
Backup Slides LUX in the Water Tank 2012
Grid Conditioning 36 Results from the first science run featured a 48.9% electron extraction efficiency. During the first half 2014 the voltage of the grids was raised for an extended period of time until significant current is drawn. The main objective was to burn any dust or asperities present in the grids. After the grid conditioning the electron extraction efficiency increased to >70%. but upon refilling we observed a large radial component in the drift field. Moreover the effect of the radial field is time dependent increasing along the run. Krypton data reconstructed positions 250 kg still in here Measured position of wall -2 kv e - -4 kv z [cm] -6 kv -8 kv - kv r [cm]
Modeling the Electric Field 37 A Fully 3-D model is constructed in the COMSOL Multiphysics FEM software to compute the electric field in the active region of LUX The observed radial field is consistent with a build up of negative charge (0 to - μc/m 2 ) on the PTFE walls. Charges are added to the walls to produce the radial field that best produces the observed distribution of 83m Kr decays. Charges in the PTFE Wall x variation in the field amplitude
Dealing with the Fields 38 Gray density: CH 3 T calibration (ER) Orange density: DD calibration (NR) Solid lines: NEST model, ER, NR band mean Dashed lines: NEST model, -90 percentile.
First Science Run Reanalysis 39 4 3.5 6.6 7.8 9.0 5.4 4.2 2.9.2 kevee [S2(phd)] 3 1.7 33 kevnr log 2.5 3 9 15 21 27 2 0 5 15 20 25 30 35 40 45 50 S1 (phd)
S2/S1 Position Corrections 40 Size of the S1 depends on the location of the event (due to geometrical light collection), and S2 (due electronegative impurities) On the FSR the correction factors for both S1 and S2 were obtained by flat fielding a mono-energetic source 83m Kr. However, a spatially varying E-field ALSO affects S1 and S2 sizes, but differently for every particle type and energy. Anode Gate Cathode P M S1 larger for events lower in the detector T S2 larger for events higher in the detector P M T
S2/S1 Position Corrections 073303-4 Manalaysay et al. 41 Our strategy is: o Disentangle position effects from field effects; o Apply a correction to account for position effects only. 83m Kr has two decays close in time. The ratio of the first-tosecond S1 pulse area depends on field alone. This allows us to measure the component of variation due to applied field alone. p.e./sample S(E)/S(0), Q(E)/Q 0 60 40 20 120 0 0 500 00 1500 2000 1 0.8 0.6 0.4 0.2 57 Co t [ns] 83m Kr (41.5 kev, Charge) 83m Kr (9.4 kev) 83m Kr (32 kev) 0 0 500 00 1500 2000 Applied Field [V/cm]
Wall-surface backgrounds 42 238 U late chain plate-out on PTFE surfaces survives as 2 Pb and its daughters (mainly 2 Bi and 2 Po). Betas and 206 Pb recoils travel negligible distance, but they can be reconstructed some distance from the wall as a result of position resolution (especially for small S2s). These sources can be used to define the position of the wall in measured coordinates, for the 4 data bins and any combination of drift-time and ϕ. The boundary of the fiducial volume is defined at 3 cm from the observed wall in S2 space and for a drift time between 50 and 300 μs. Fiducial volume Measured wall position Events Events Events Events Data Wall Model 00 0 S2 [200, 500] phd S1 [ 55, 500] phd 1 00 0 S2 [500, 900] phd S1 [ 55, 500] phd 1 00 0 S2 [900, 1300] phd S1 [ 55, 500] phd 1 00 0 S2 [1300, 1800] phd S1 [ 55, 500] phd 1 00 0 S2 [1800, 3000] phd S1 [ 55, 500] phd 1 5 4 3 2 1 0 1 2 3 4 5 Distance to the wall (cm) Events Fiducial boundary Into the detector Out the detector
Wall Surface Model How to predict the number of wall events leaking into fiducial volume? P wall = N (S1,S2,φ,z,DB) P r r (S2,φ,z,DB) raw,d wall ) Number of Events Leakage depends on the number of events, N, and how they distribute around the wall, P r. To get N we used our WIMP search data, but with the radial position of the events larger than the radial position of the wall - events leaking towards the wall. To get P r our strategy is: o Describe the events as function of the ratio between the distance to the wall and the uncertainty called pulls g g = d Wall o Use the data with S1>50 phd to get systematics. r Radial Model Fiducial boundary ) c (S2 log 4 3.5 3 2.5 2 Events Events Events Events Data Wall Model 00 0 S2 [200, 500] phd S1 [ 55, 500] phd 1 00 0 S2 [500, 900] phd S1 [ 55, 500] phd 1 00 0 S2 [900, 1300] phd S1 [ 55, 500] phd 1 00 0 S2 [1300, 1800] phd S1 [ 55, 500] phd 1 00 0 S2 [1800, 3000] phd S1 [ 55, 500] phd 1 5 4 3 2 1 0 1 2 3 4 5 Distance to the wall (cm) Events Into the detector Wall PDF Proj log(s2) vs S1 20 30 40 50 S1 c in detected photons Wall PDFS Rel Prob -5-6 -7 drift time [us] 50 0 150 200 250 Out the detector Wall PDF Proj drift time vs r 300 0 5 15 20 r [cm] Rel Prob -1-2 -3-4 -5-6 -7-8 -9 -
LUX Proposal VS Main Result 44 Blue: DOE/NSF LUX Proposal, Projected (2007) Black: Observed 332-liveday limit Brazil band: 1- and 2-sigma 332-LD expected sensitivities
The Water Shield 45 Water Tank: ø = 8 m, h = 6 m (300 tonnes) Cherenkov based active shielding o Dimensions: ø = 8 m, h = 6 m (300 tonnes). o Muon active veto: 20 PMTs Ø. Ultra-low Background o γ suppression: x -9 o Neutron sup. (E n > MeV ~ -3 and E n < MeV > -9 ).
Krypton Removal 46 85 Kr - beta decay intrinsic background in liquid Xe o 85 Kr: 0.687 MeV β, yr half-life o Research grade Xenon: ~0 ppb Kr => 4-5 reduction needed August 2012 - January 2013: Kr removal at Case Western Reserve University o Chromatographic separation system o Kr lighter & less polarisable than Xe. Kr bonds weaker, travels faster through charcoal and pure xenon is left behind. Kr concentration reduced from 130 ppb to 4 ppt, (factor of 30000) o 1 ppt achievable (useful for next-generation detectors) He charcoal Xe Kr