Search for low-mass dark matter WIMPs with 12 ton-day exposure of DarkSide-50

We report on the search for dark matter weakly interacting massive particles (WIMPs) in the mass range below $10\mathrm{GeV}/{\mathrm{c}}^2$, from the analysis of the entire dataset acquired with a low-radioactivity argon target by the DarkSide-50 experiment at Laboratori Nazionali del Gran Sasso. The new analysis benefits from more accurate calibration of the detector response, improved background model, and better determination of systematic uncertainties, allowing us to accurately model the background rate and spectra down to $0.06{\mathrm{keV}}_{er}$. A 90% C.L. exclusion limit for the spin-independent cross section of $3\mathrm{GeV}/{\mathrm{c}}^2$ mass WIMP on nucleons is set at $6\times {10}^{-43}{\mathrm{cm}}^2$, about a factor 10 better than the previous DarkSide-50 limit. This analysis extends the exclusion region for spin-independent dark matter interactions below the current experimental constraints in the [1.2, 3.6] $\mathrm{GeV}/{\mathrm{c}}^2$ WIMP mass range.

Figure 1

Comparison and ratio between LAr ionization responses to nuclear (NR) and electronic (ER) recoils, as a function of the deposited energy. The two energy scales were measured using DarkSide-50 data and datasets from the ARIS [17] and SCENE [18] experiments, and are reported in Ref. [16].

Figure 2

$N_e$ spectra at different steps of the data selection, after rejection of events outside the fiducial volume and with multiple interactions.

Figure 3

$N_e$ spectra after selection cuts requiring a time coincidence ($\mathrm{\Delta }\mathrm{T}$) with the preceding event higher than 20 ms (shaded blue), between [0.8, 2] ms (blue dots) and between [2, 20] ms (red dots). The spectra are normalized to livetime of the sample with $\mathrm{\Delta }\mathrm{T}>20\mathrm{ms}$, whose spectrum is statistically compatible with the one with $\mathrm{\Delta }\mathrm{T}$ in [2, 20] ms above 4 $N_e$, where the contamination from spurious correlated electrons becomes negligible.

Figure 4

${}^{39}\mathrm{Ar}$ (orange) and ${}^{85}\mathrm{Kr}$ (blue) beta decay spectra in $N_e$ and associated systematics from atomic exchange and screening effects (shaded area) and ionization response (dashed line). The systematic error propagated from the Q-value uncertainty is too small to be illustrated in this plot.

Figure 5

Breakdown of PMTs (top) and cryostat (bottom) radioactive contributions. The stacked spectra in the energy region of interest and in the fiducial volume are scaled by the measured activity.

Figure 6

Background spectra from PMTs and cryostat and associated error bands from the detector response (shaded area) and from the Monte Carlo statistics (dotted lines).

Figure 7

Background model and uncertainty (red line and shaded area) from the data fit in the [4, 170] $N_e$ range, and the individual contributions from the internal (${}^{39}\mathrm{Ar}$ and ${}^{85}\mathrm{Kr}$) and external components (cryostat and PMTs). An excess of events with respect to the background model is observed below 4 $N_e$ (blue shaded area). The residuals, defined as the difference between the observed and expected events, normalized to the expected ones, are compared below to the model uncertainty from the fit.

Figure 8

Pulls from the background-only fit (black points) are normally distributed, as highlighted by the Gaussian fit (red line). Shaded blue histogram corresponds to prefit distribution.

Figure 9

Postfit nuisance parameters compared to the nominal values (left) and correlation matrix (right) from the background-only fit. Error bars are normalized to the prefit size of each of the nuisance parameter penalty terms.

Figure 10

Data and background model compared to expected WIMP spectra, assuming binomial quenching fluctuations (solid lines) and WIMP-nucleon scattering cross section equal to $2\times {10}^{-41}{\mathrm{cm}}^2$. The systematic error associated with WIMP spectra is due to uncertainty on the NR ionization response. For reference, WIMP spectra assuming no quenching fluctuation (dashed lines) are also shown.

Figure 11

90% upper limits on spin independent WIMP-nucleon cross sections from DarkSide-50 in the range above $1.2\mathrm{GeV}/{\mathrm{c}}^2$. Both nonquenching (NQ, solid red line) and quenching (QF, dashed red line) fluctuations models are considered. Also shown are the expected limits (green dotted lines) with the $\pm 1\text{−}\sigma$ (green shaded area) and $\pm 2\text{−}\sigma$ (yellow shaded area) bands.

Figure 12

DarkSide-50 limits with and without quenching fluctuations are labeled QF and NQ, respectively. These limits are compared to the 90% C.L. exclusion limits and claimed discovery from Refs. [6, 9, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45] and to the neutrino floor for LAr experiments [46].