Comparing readout strategies to directly detect dark matter

Over the past decades, several ideas and technologies have been developed to directly detect weakly interacting massive particles (WIMP) from the galactic halo. All these detection strategies share the common goal of discriminating a WIMP signal from the residual backgrounds. By directly detecting WIMPs, one can measure some or all of the observables associated to each nuclear recoil candidates, such as their energy and direction. In this study, we compare and examine the discovery potentials of each readout strategies from counting only (bubble chambers) to directional detectors (Time Projection Chambers) with 1d-, 2d-, and 3d-sensitivity. Using a profile likelihood analysis, we show that, in the case of a large and irreducible background contamination characterized by an energy distribution similar to the expected WIMP signal, directional information can improve the sensitivity of the experiment by several orders of magnitude. We also found that 1d directional detection is only less effective than a full 3d directional sensitivity by about a factor of 3, or 10 if we assume no sense recognition, still improving by a factor of 2 or more if only the energy of the events is being measured.

Figure 1

Left: Evolution of the direction pointing toward the constellation Cygnus over a sidereal day in the detector’s coordinates. Middle and right: Angular distributions of WIMP induced nuclear recoils at 50 keV in the detector frame along the $\theta$ (middle) and $\varphi$ (right) angles over a sidereal day. We assumed a Xe-based experiment located in Modane and a WIMP mass of $50\mathrm{GeV}/{\mathrm{c}}^2$. The black dashed line correspond to an isotropic distribution.

Figure 2

Number of WIMP events required to get a $3\sigma$ detection in 90% of the experiment as a function of the signal purity $\lambda$ for all different types of readout: counting only (black dotted line), Energy only (black dashed line), $\text{Energy}+1\mathrm{d}$, 2d and 3d directionality (red, blue, and green solid lines, respectively). We considered a Xe-based experiment located in Modane with a recoil energy range from 5 keV to 100 keV and a $50\mathrm{GeV}/{\mathrm{c}}^2$ WIMP mass. Left (right) panel assumes an isotropic background with an exponential energy distribution with a slope of 100 (10) keV. Small fluctuations in the results are due to the finite size of the Monte Carlo simulations of 1,000 samples.

Figure 3

Evolution of the discovery limit for a SI interaction for a $50\mathrm{GeV}/{\mathrm{c}}^2$ WIMP mass as a function of the number of expected background events (exposure) for all different types of readout: counting only (black dotted line), counting only including the effect of annual modulation (black short-dotted line), Energy only (black dashed line), $\text{Energy}+1\mathrm{d}$, 2d and 3d directionality (red, blue, and green solid lines, respectively). We considered a Xe-based experiment located in Modane with a recoil energy range from 5 keV to 100 keV, a background rate of 100 events/ton/year, a systematic uncertainty of ${\sigma }_{R_b}=20\%$, and an exponential slope of 10 keV. Small fluctuations in the results are due to the finite size of the Monte Carlo simulations of 1000 samples.

Figure 4

Number of WIMP events required to get a $3\sigma$ detection in 90% of the experiment as a function of the detector latitude ${\lambda }_{\text{lab}}$ for all different directional readouts: $\text{Energy}+1\mathrm{d}$, 2d and 3d directionality (red, blue, and green solid lines respectively). We considered a Xe-based experiment with a recoil energy range from 5 keV to 100 keV, a $50\mathrm{GeV}/{\mathrm{c}}^2$ WIMP mass, a signal purity of $\lambda =0.4$, and an exponential energy distribution for the isotropic background of 10 keV. Small fluctuations in the results are due to the finite size of the Monte Carlo simulations of 1000 samples.

Figure 5

Number of WIMP events required to get a $3\sigma$ detection in 90% of the experiment as a function of the Head-Tail energy threshold for the three directional readouts: 1d (red), 2d (blue), and 3d (green). Also shown for comparison is the case of reading out only the energy of the events (black dashed line). We considered a Xe-based experiment located in Modane with a recoil energy range from 5 keV to 100 keV, a $50\mathrm{GeV}/{\mathrm{c}}^2$ WIMP mass, a signal purity of $\lambda =0.4$, and an exponential energy distribution for the isotropic background of 10 keV. Small fluctuations in the results are due to the finite size of the Monte Carlo simulations of 1,000 samples.