Low-mass dark matter search results from full exposure of the PandaX-I experiment

We report the results of a weakly interacting massive particle (WIMP) dark matter search using the full 80.1 live-day exposure of the first stage of the PandaX experiment (PandaX-I) located in the China Jin-Ping Underground Laboratory. The PandaX-I detector has been optimized for detecting low-mass WIMPs, achieving a photon detection efficiency of 9.6%. With a fiducial liquid xenon target mass of 54.0 kg, no significant excess events were found above the expected background. A profile likelihood ratio analysis confirms our earlier finding that the PandaX-I data disfavor all positive low-mass WIMP signals reported in the literature under standard assumptions. A stringent bound on a low-mass WIMP is set at a WIMP mass below $10\mathrm{GeV}/{\mathrm{c}}^2$, demonstrating that liquid xenon detectors can be competitive for low-mass WIMP searches.

Figure 1

The PE distribution of the smallest S2 signals (corrected for horizontal nonuniformity), summed over all top and bottom PMTs. The single-electron gain is determined by fitting two constrained Gaussians, shown as the dashed green lines.

Figure 2

Distribution in uniformity-corrected S1 and S2 for the deexcitation gamma peaks in the neutron calibration runs. The antidiagonal lines are the anticorrelation fit at the two energies. The dashed vertical and horizontal lines indicate the mean NR energy in mixture with the gamma energies at 40 and 80 keV.

Figure 3

The $L_y$ (blue) and $C_y$ (red) (in units of quanta per ${\mathrm{keV}}_{\mathrm{ee}}$) extracted based on PDE and EEE obtained at different energies overlaid with corresponding curves predicted by NEST-0.98. The reconstructed energy spectra for the deexcitation peaks and metastable xenon isotopes are also overlaid with the $y$ axis scaled for visual clarity with fitted energy resolutions indicated in the figure.

Figure 4

The band of ${\log }_{10}(\mathrm{S}2/\mathrm{S}1)$ vs S1 for the NR calibration data without (a) and with (b) the “X” cut. See text for definition of the cut. The 54.0-kg fiducial cut is applied. The solid blue line in panel (b) is the median of the pure NR band in MC, and the dots are the Gaussian mean obtained from the data for $\mathrm{S}1>15$ (where the detection efficiency is flat so that the data and MC comparison is straightforward). The dashed magenta lines in both panels are the 300 PE cut on S2, below which no dark matter candidate is considered. The gray dashed lines are the equal-energy lines with NR energy indicated in the panels.

Figure 5

Comparison of the distribution of ${\log }_{10}(\mathrm{S}2/\mathrm{S}1)$ in the NR data and tuned MC in six slices of S1 as indicated by the panel titles: data (blue), tuned “neutron-X” events in the MC (magenta), and the sum of pure NR and tuned “neutron-X” in MC (red). In each slice of S1, the value of ${\log }_{10}(\mathrm{S}2/\mathrm{S}1)$ is shifted relative to the median value in that slice. The efficiency in Fig. 6 has been applied to the MC to compare with the data.

Figure 6

The nuclear recoil efficiency in (S1,S2) determined using the method discussed in the text.

Figure 7

(a) The band of ${\log }_{10}(\mathrm{S}2/\mathrm{S}1)$ vs S1 for the ${}^{60}\mathrm{Co}$ ER calibration runs with the median and $\pm 2\sigma$ of the band indicated as the solid and dashed blue lines, respectively. The median of the NR band is indicated as the solid red line, below which the 12 leaked events are plotted as green markers. The dashed magenta line is the 300 PE cut on S2. The gray dashed lines are the equal-energy lines with ER energy indicated in the figures. (b) The ER efficiency in S1 obtained by taking the ratio between the data and MC (histogram), and the red curve is a fit to the efficiency.

Figure 8

The measured isolated S1 (a) and S2 (b) charge spectra, and the band of ${\log }_{10}(\mathrm{S}2/\mathrm{S}1)$ vs S1 (c) for the statistically obtained accidental background, with the sharp cutoffs at the top and bottom corresponding to the 10 000 and 300 PE cut. The median of the NR band is indicated as the red line.

Figure 9

(a) The vertex distribution of events in the TPC during dark matter search data taking. The blue dashed box indicates the fiducial volume, and the red dashed box indicates the entire active volume within the TPC, confined by the cathode, gate grid, and the PTFE wall. The events below the NR median are indicated by the green markers. (b) Projection in $r^2$, with data (TM method for position reconstruction) in blue, data (CoG method) in green, and the MC energy deposition CoG in red, (c) Projection in the drift time, with data (TM) in blue and the MC energy deposition CoG in red.

Figure 10

The band of ${\log }_{10}(\mathrm{S}2/\mathrm{S}1)$ vs S1 for the dark matter search data. The ER band is indicated by the blue solid line (median) and the dashed line ($\pm 2\sigma$). The median of the NR band is indicated as the solid red line. The dashed magenta line is the 300 PE cut on S2. The green stars represent events below the NR median. The gray dashed lines are the equal-energy lines with NR energy indicated in the figures.

Figure 11

The overall detection efficiency for WIMPs at different masses (red) and those obtained with PDE and EEE set at $+1\sigma$ (solid green) and $-1\sigma$ (dashed green).

Figure 12

The comparison of the mean energy thresholds, translated from the cuts on S1 and S2, for different experiments under the same energy model (NEST v0.98). The solid curves (red: PandaX; violet: XENON100; blue: LUX) represent the mean values for S1 and S2 obtained from NEST with slanted ticks (along the equal-energy vector) indicating the corresponding mean NR energy in divisions of ${\mathrm{keV}}_{nr}$. The curve for PandaX based on NEXT v1.0 is drawn as the green dashed curve. The selection thresholds in S1 and S2 are indicated in the figure as the solid circles. The antidiagonal dashed lines (red: PandaX; violet: XENON100; blue: LUX) are the equal-energy lines projected from the corresponding threshold points for different experiments.

Figure 13

The 90% C.L. upper limit for the spin-independent isoscalar WIMP-nucleon cross section for the PandaX-I experiment (red curves). Recent world results are plotted for comparison: XENON100 225 day results [11] (black solid), LUX first results [12] (blue), SuperCDMS results [15] (orange solid), DarkSide results [14] (magenta solid), CRESST-II 2014 limits [8] (brown dashed), and CDEX 2014 limits [17] (solid violet). The claimed WIMP signals are shown as closed contours: CoGeNT 2014 results [6] (cyan solid), CDMS-II-Si results [9] (gold dashed), DAMA/LIBRA $3\sigma$ contours [38] (green solid), and CRESST-II 2012 results [7] (brown solid).

Figure 14

PandaX-I WIMP search limit from the data (red line) overlaid with the $\pm 1\sigma$ sensitivity band obtained from toy MC (yellow) as well as the alternative upper limits using either $+1\sigma$ or $-1\sigma$ values for the PDE and EEE, but with the same NEST-0.98 model. For comparison, a few world leading limits for the low-mass WIMP are plotted: LUX first results [12] (blue), SuperCDMS results [15] (orange), and CRESST-II 2014 limits [8] (brown dashed).