Improved Measurements of the β-Decay Response of Liquid Xenon with the LUX Detector Jon Balajthy UC Davis, Department of Physics September 12, 2019 1
LUX Detector ● 4850’ level (4300 m w.e.) of the Sanford Underground Research Facility in Lead, SD. ● Water tank for neutron shield/ muon veto ● 2-Phase (liquid/gas) xenon TPC ○ Sensitive to light (S1) and charge (S2) ○ 3-D position reconstruction ● Total/ fiducial mass of 370 kg/ 118 kg ● Combined science runs: Excluded scattering cross sections below 1.1×10 −46 ○ cm 2 for a 50 GeV WIMP Exposure of 3.35×10 4 kg days ○ ○ ~1100 fiducial counts 2
Combined Energy Model Charge and light work functions are averaged N r will be randomly distributed with mean value, into a single W = 13.7 ±0.2 eV/quanta. N i ·P R , and standard deviation, σ R . ER energy can equivalently be written: or, in terms of measurables: Some fraction, r , of N i will be converted to N * Which can be rewritten: 3 α is empirically constant for ER
ER Yields Energy and field dependence of the recombination process is characterized by the yields: Important note- Ly and Qy are assumed to be complementary: 4
ER Yields in Post-WS 14 C and 3 H Data Begin with S1 and S2 measurements in a collection of energy and field bins for both 14 C and 3 H. ● Fields: 43, 53, 65, 80, 98, 120, 147, 180, 220, 269, 329, 403, 491 V/cm ○ The drift field in Run04 was highly non-uniform (arXiv:1709.00095) ● Energies: 1-145 keVee The 14 C spectrum overlapped with 131m Xe above this range ○ Procedure: 1. Model detector resolution in post-WS 2. Develop model of yields and recombination fluctuations ( σ R ) 3. Numerically de-smear the reconstructed energy, S1 , and S2 spectra 4. Measure N γ /N e in each energy/field bin (can then be used to calculate LY , QY , and P R ) 5
Post-WS Doke Plot ● Doke plot measured from Kr-83 + Xe-131 at a collection of drift times ● Only two lines, but smearing due to E-field gives very good measurement of g1, g2 6
Post-WS Doke Plot Errors dominated by systematic variation Two-point Doke plots in each drift bin gives a measurement of how g1 and g2 change as a function of position Introduces subdominant- dominant errors on final Ly, Qy measurements 7
S2 Tails in Post-WS LUX Thought to be caused by emission of electron trains Modeled by adding area to the LibNEST S2 output ○ Area drawn from an exponential distribution with mean = b*S2 ○ Added only to a certain fraction, R, of MC events ○ b and R found by matching 37 Ar and 131m Xe MC to data 8
S2 Tails in Post-WS LUX The addition of the S2 to libNEST significantly improves the agreement with both 3 H and 14 C beta spectra 9
Recombination Fluctuations 1/2 for a binomial process. σ R should be proportional to N i σ R has been instead observed to be roughly proportional to ~ N i In LUX post-WS carbon-14 data, we add a Gaussian adjustment to the N i dependence Where: 10
Comparison to LUX WS2013 The adjusted σ R model better reproduces the substructure of the LUX WS2013 measurements When N e ≈ N γ , the new model reproduces the WS2013 result arXiv:1512.03133 11
De-Smearing a Beta Spectrum ● Continuous spectra will be smeared by The average E true in the 80 keV bin is finite detector resolution. 79.53 ±0.009 keV ● Similar effect is found in 2-D N e , N γ space ● Can be accounted for analytically by integration ● In Post-WS, this is unwieldy due to S2 tails. Instead we desmear numerically ○ Calculate ratio of true to measured E rec in MC & apply correction to data ○ Calculate ratio of true to measured N e , N γ in MC, and apply correction to data 12
Results- Snapshot Results are reported in terms of photon-electron ratio Very low statistical error (shown in black) Systematic error is taken as quadrature sum of: ● g1/g2 uncertainty ● Absolute magnitude of smearing correction 13
Post-WS Qy Compared to WS2013 We construct an empirical model to help compare LUX Post-WS to previous measurements. All the datasets considered agree with the model to about 1-σ. The QY results are consistent with WS2013 measurements at 180 and 100 V/cm, except... arXiv:1512.03133 arXiv:1709.00800 arXiv:1610.02076 arXiv:1705.08958 14
Post-WS Qy Compared to WS2013 … Run03 tritium has distinct kink near endpoint This is due to mistake in WS2013 de-smearing. Recombination fluctuations are double counted- they are corrected for in both reconstructed energy and individual quanta. Wrong 15
Beta vs EC Also… 127 Xe EC L-shell is consistent within systematic error bars, but just barely As per Dylan Temples’ talk on Monday- this discrepancy appears to be physical 16
Beta vs Photo-absorption 131m Xe and 83m Kr lines also show discrepancy with beta yields ● Photoabsorption events give higher light yield 83m Kr is composite, and has even more intricacies ● 17
Compared to World Data 18
Compared to World Data Good agreement with 3 H data from Xenon100 19
Compared to World Data Systematically below the Compton scattering data from neriX, Shanghai, and E. Dahl Could point to physical difference in Compton/ beta yields? 20
Compared to World Data Systematically below 37 Ar EC data from PIXeY. 21
Summary ● Developed model for S2 tails in LUX Post-WS data ● Added Gaussian adjustment to σ R model ● Numerically de-smeared S1 and S2 spectra ● Measured LY, QY, and P R as functions of energy and electric field ○ Energy range from 1-145 keV ○ Field range from 43-491 V/cm ● Compared results to existing data Very consistent with LUX WS2013 3 H data ○ Consistent with WS2013 127 Xe K- and M-shell ○ Hints of discrepancy with 127 Xe L-shell ○ Agrees with 3 H data from Xenon100 ○ ○ Disagrees with all available Compton data 22
Extra Slides 23
Energy Deposition in LXe An energetic particle interacting with a xenon nucleus will create: ● Excited xenon atoms (Xe*) Electron-ion pairs (e - , Xe + ) ● ● Heat (not detected) The excitons and ions form dimers with ground state + ). atoms (Xe 2 *, Xe 2 After the initial partitioning, there is a recombination period (~50 ns) where: e - +Xe 2 + → Xe**+Xe → Xe*+Xe + heat The net result of this is that an ion is converted to an exciton. 24
ER Yields For ER, P R , LY , and QY can all be written in terms of the N γ /N e ratio only: 25
Post-Run04 Injection Campaign Following the primary WIMP-search run (WS2014-16), a series of NR and ER calibrations were performed. ● D-D 83 Kr ● 131m Xe ● 3 H ● 14 C ● 222 Rn ● 37 Ar ● 26
Beta Model from Post-Run04 14 C and 3 H Data Fully empirical model of QY and σ R ● QY is fit to the sum of two asymmetric sigmoids Non-LUX datasets used to constrain model: Low energy: Boulton et al., 2017 High energy: Doke et al., 2002 ● The m 1 and m 7 parameters are 0-field: Baudis et al., 2013 allowed to vary with field Aprile et al., 2012 ● σ R is the Gaussian adjusted High-field: Akimov et al., 2014 linear model presented in slide 6 Compton data from neriX is in tension on the 1-2 sigma level 27
Results- 14 C 28
Results- 3 H 29
Recommend
More recommend