Scintillation efficiency and ionization yield of liquid xenon for monoenergetic nuclear recoils down to 4 keV

Liquid xenon (LXe) is an excellent material for experiments designed to detect dark matter in the form of weakly interacting massive particles (WIMPs). A low energy detection threshold is essential for a sensitive WIMP search. The understanding of the relative scintillation efficiency (${\mathcal{L}}_{\mathrm{eff}}$) and ionization yield of low energy nuclear recoils in LXe is limited for energies below 10 keV. In this article, we present new measurements that extend the energy down to 4 keV, finding that ${\mathcal{L}}_{\mathrm{eff}}$ decreases with decreasing energy. We also measure the quenching of scintillation efficiency caused by the electric field in LXe, finding no significant field dependence.

Figure 1

Illustration of the signal production and collection in a two-phase xenon detector.

Figure 2

The setup for nuclear recoil scintillation efficiency measurement in LXe. Not drawn to scale.

Figure 3

Schematic of the dual-phase LXe detector. Spaces not drawn were filled with PTFE pieces. PMTs, LXe, and xenon gas regions drawn to scale.

Figure 4

Trigger system used in the dual-phase runs. The PMT signals are summed and integrated to select the $S2$ signals. The organic scintillator signal generates a 15-$\mu$s pulse. This pulse and the $S2$ signals in coincidence trigger the acquisition system. For the single-phase runs, the triple coincidence of the $S1$ signals and the organic scintillator signal triggers the acquisition system.

Figure 5

Scintillation light yield for 122-keV $\gamma$ rays in the LXe detector running in single phase. The left histogram was taken at 1.5 kV/cm yielding $4.8\pm 0.1$ pe/keVee and a resolution ($\sigma /E$) of $18\%$. The right histogram was taken at 0.0 kV/cm yielding $10.8\pm 0.1$ pe/keVee and a resolution of $8.8\%$.

Figure 6

$S2$ spectrum for 122-keV $\gamma$ rays in the dual-phase detector.

Figure 7

Two cuts used in the analysis to select single scatter nuclear recoil events: (a) PSD in the organic scintillator and (b) neutron ToF between the LXe and the organic scintillator. Both plots show experimental data for 56-keV${}_{\mathrm{r}}$ run.

Figure 8

$S1$ spectra in pe for (a) ${6-\mathrm{keV}}_{\mathrm{r}}$ nuclear recoils at zero electric field and (b) ${66-\mathrm{keV}}_{\mathrm{r}}$ nuclear recoils at 1.5 kV/cm, after applying the cuts. $S2$ spectra in pe for (c) ${6-\mathrm{keV}}_{\mathrm{r}}$ nuclear recoils at zero electric field and (d) ${66-\mathrm{keV}}_{\mathrm{r}}$ nuclear recoils at 1.5 kV/cm.

Figure 9

Simulated distribution of energy deposition in the LXe for 2.8-MeV neutrons that scatter and are tagged by the organic scintillator, for three different angles, assuming the same ToF cuts that were used in the experimental data. The solid line corresponds to all types of events, the blue dashed line to single elastic nuclear recoils, and the purple dotted line to multiple elastic scatters. Multiple scatter events and inelastic events are negligible compared with the single scatter events. The energy and uncertainty values indicated in the figure legends are based on the mean and variance (one sigma) of the peak for all events.

Figure 10

Detector triggering and software efficiencies for the single-(▴) and dual-phase (•) runs. The ▪ points show the ideal case when the PMTs and the software have $100\%$ efficiency for a single photoelectron.

Figure 11

$S1$ measured spectrum (data points) and Monte Carlo comparisons (solid line) for (a) ${6-\mathrm{keV}}_{\mathrm{r}}$ nuclear recoils at zero electric field and (b) ${66-\mathrm{keV}}_{\mathrm{r}}$ nuclear recoils at 1.5 kV/cm. The background tail ($S1<50$ pe) for the ${66-\mathrm{keV}}_{\mathrm{r}}$ run has been removed for the ${\chi }^2$ test. Also shown are the ${\chi }^2$ versus ${\mathcal{L}}_{\mathrm{eff}}$ histograms for (c) ${6-\mathrm{keV}}_{\mathrm{r}}$ nuclear recoil at zero electric field and (d) ${66-\mathrm{keV}}_{\mathrm{r}}$ nuclear recoils at 1.5 kV/cm. The points show the different ${\mathcal{L}}_{\mathrm{eff}}$ values tested while the curve is a fit to the points used to find the $1\sigma$ errors on the ${\mathcal{L}}_{\mathrm{eff}}$ value.

Figure 12

Scintillation efficiency for nuclear recoils relative to that of 122-keV $\gamma$ rays in LXe at zero field, comparing this work (•) to previous measurements from Arneodo (▵) [5], Akimov (□) [6], Aprile (⋆) [7], Chepel (⋄) [8], and Aprile (◯)[9]. Also shown is the theoretical model (dashed line) explained in Sec. 5, which includes the Lindhard factor, an electronic quenching resulting from biexcitonic collisions, and the effect of escaping electrons.

Figure 13

Scintillation efficiency for nuclear recoils measured in this work (•) and the theoretical model (dashed line) compared to the scintillation efficiency found from the neutron calibration data by the XENON10 (top shaded area) [12] and the ZEPLIN-III (bottom shaded area) [4] collaborations.

Figure 14

Ionization yield as a function of recoil energy. Shown are the measured values in this work at 1.00 kV/cm (▴) and 4.00 kV/cm (•), along with previously measured values at 0.10 kV/cm (◯), 0.27 kV/cm (□), 2.00 kV/cm (△), and 2.30 kV/cm (⋄) from Ref. [10], error bars omitted for clarity. Also shown are the ionization yields calculated by comparing the XENON10 nuclear recoil data and the Monte Carlo simulations [12] for single elastic recoils at 0.73 kV/cm, using two different methods (∗ and ⋆).

Figure 15

Fraction of escape electrons for nuclear recoils, ${\beta }_{\mathrm{NR}}$, as a function of recoil energy used in this work.

Figure 16

Projected spin-independent dark matter limits for a LXe detector with 30,000 kg/day exposure and 0.95 to 5.7 keVee energy window. The bottom solid line shows the limit with ${\mathcal{L}}_{\mathrm{eff}}=0.19$ (5 to 30 keV${}_{\mathrm{r}}$ window), whereas the dotted line shows the limit calculated with the measured ${\mathcal{L}}_{\mathrm{eff}}$ (8.4 to 39.0 keV${}_{\mathrm{r}}$ window). Also plotted are the XENON10 result [3] (top solid line) and the regions predicted by Refs. [31] (red shaded region) and [32] (green shaded region). Plot generated using Ref. [33].