Assessing Near-Future Direct Dark Matter Searches with Benchmark-Free Forecasting

Forecasting the signal discrimination power of dark matter (DM) searches is commonly limited to a set of arbitrary benchmark points. We introduce new methods for benchmark-free forecasting that instead allow an exhaustive exploration and visualization of the phenomenological distinctiveness of DM models, based on standard hypothesis testing. Using this method, we reassess the signal discrimination power of future liquid xenon and argon direct DM searches. We quantify the parameter regions where various nonrelativistic effective operators, millicharged DM, and magnetic dipole DM can be discriminated, and where upper limits on the DM mass can be found. We find that including an argon target substantially improves the prospects for reconstructing the DM properties. We also show that only in a small region with DM masses in the range 20–100 GeV and DM-nucleon cross sections a factor of a few below current bounds can near-future xenon and argon detectors discriminate both the DM-nucleon interaction and the DM mass simultaneously. In all other regions only one or the other can be obtained.

Figure 1

Limit on the SI cross section vs DM mass, assuming operator ${\mathcal{O}}_1$, for a XENONnT detector. Contours show 68% confidence contours in $d=2$ dimensions. The radius of these contours in the Euclidean space is therefore $r_{\alpha }(\mathcal{M})=1.52$. The number of discriminable signals in the $\text{blue}+\text{blue}–\text{orange}$ region of the middle panel of Fig. 2 can be approximated by counting the number of closed green contours.

Figure 2

Signal-discrimination power of combinations of direct detection experiments, summarized in infometric Venn diagrams. Central-Right panel: the full area corresponds to the number of distinct detectable signals within the alternative hypothesis ${\mathcal{H}}_A$, unfolded as function of number of XENONnT signal events. The subsets indicate the fraction of signals consistent with various null hypotheses ${\mathcal{H}}_0$. Overlapping subsets correspond to signals consistent with multiple null hypotheses simultaneously. The numbers correspond to ${\nu }_{\mathcal{M},X}({\mathrm{\Omega }}_{\mathcal{M}})$ in each region. In the right panel, the overlapping ($\text{blue}+\text{orange}$) region corresponds to the model parameters between the purple-dot-dashed and orange contours in Fig. 3. The nonoverlapping (blue only) ${\mathcal{O}}_1$ region corresponds to parameters between the purple-dashed and blue contours in Fig. 3. Left panel: standard Venn diagram summed over number of signal events.

Figure 3

Discriminability of DM interactions. To the left-below each broken line, it is not possible to discriminate an ${\mathcal{O}}_1$ signal (with the indicated cross section and DM mass) from the corresponding best-fit ${\mathcal{O}}_4$, ${\mathcal{O}}_{11}$, or magnetic dipole signal. Above-right of each broken line, such a discrimination is possible with at least $2\sigma$ significance. Solid lines display 90% C.L. for XENON1T-2017 and XENONnT. All lines include DM halo uncertainties.

Figure 4

Discriminability of the DM mass. To the right of each broken line, the DM mass is unbounded from above (at $2\sigma$). Lines for other operators are mapped onto ${\sigma }_{\mathrm{SI}}$ by converting to an effective cross section and rescaled to match ${\mathcal{O}}_1$ at high masses. For ${\mathcal{O}}_1$, we also show constraints for the xenon-only case and when halo uncertainties are neglected. Solid lines display 90% CL limits for XENON1T-2017 and XENONnT.