Sensitivity of the Cherenkov Telescope Array to the detection of a dark matter signal in comparison to direct detection and collider experiments

Imaging atmospheric Cherenkov telescopes (IACTs) that are sensitive to potential $\gamma$-ray signals from dark matter (DM) annihilation above $\sim 50\mathrm{GeV}$ will soon be superseded by the Cherenkov Telescope Array (CTA). CTA will have a point source sensitivity an order of magnitude better than currently operating IACTs and will cover a broad energy range between 20 GeV and 300 TeV. Using effective field theory and simplified models to calculate $\gamma$-ray spectra resulting from DM annihilation, we compare the prospects to constrain such models with CTA observations of the Galactic center with current and near-future measurements at the Large Hadron Collider (LHC) and direct detection experiments. For DM annihilations via vector or pseudoscalar couplings, CTA observations will be able to probe DM models out of reach of the LHC, and, if DM is coupled to standard fermions by a pseudoscalar particle, beyond the limits of current direct detection experiments.

Figure 1

J factor as a function of angular distance from the GC for the two profiles considered in this work (derived from a fit to the rotation curve of the Milky Way [45]) with those parameters in the text (Sec. 2a): NFW (solid blue line) and Einasto (solid orange). For comparison, we also show as a dashed blue line the J factor given by the Einasto profile used in Refs. [68, 69, 73]. For each value of $\theta$ the integral over the solid angle in Eq. (1) is evaluated over an azimuthal angle interval of 0.1°. Regions of the GC halo that will be used in our analysis are also depicted in the figure as striped areas delimited by vertical lines (see legend). These regions depend on the observational strategy adopted in each case; see Sec. 4a for full details.

Figure 2

Average photon spectrum per DM annihilation into quarks. Photon spectra are from pppc4dmid [79, 80] and branching ratios are determined using Eqs. (8)–(11).

Figure 3

Branching ratios for the $S$ and $V$ models for $g_q=g_{\chi }=1$ ($S$) and $g_q=0.25$, $g_{\chi }=1$ ($V$), with $M_{\mathrm{med}}=1\mathrm{TeV}$.

Figure 4

Simplified model pppc4dmid spectra for $M_{\mathrm{med}}=1\mathrm{TeV}$.

Figure 5

Adopted pointing schemes. From left to right the panels show the expected count rate for the different model components included here (DM, CR, and GDE). For the DM, we show the expected rate for the ${\mathcal{O}}_V$ operator and $m_{\chi }=1\mathrm{TeV}$ with $⟨\sigma v⟩=3\times {10}^{-26}{\mathrm{cm}}^3{\mathrm{s}}^{-1}$. Top: Pointing schemes for the morphological analysis adopted for the NFW profile. Bottom: True on/off pointings for the Einasto profile.

Figure 6

Expected count rates for the different source components for the NFW profile and morphological pointing strategy. The DM component is shown for the ${\mathcal{O}}_V$ operator and $m_{\chi }=1\mathrm{TeV}$ with $⟨\sigma v⟩=3\times {10}^{-26}{\mathrm{cm}}^3{\mathrm{s}}^{-1}$. Top: Expected count rate as a function of energy for the innermost RoI. The energy above which the GDE is extrapolated is marked by a blue arrow. Bottom: Count rate for each RoI integrated above 30 GeV.

Figure 7

Upper limits on the cross section for 100% annihilation of Majorana DM into $b\overline{b}$ for our fiducial analysis (1% systematic uncertainty, including GDE) for the two assumed DM density profiles (blue solid line, NFW; orange solid line, Einasto). For comparison we also show our limits for 0% systematic uncertainty for the NFW profile with and without GDE (blue dotted and dashed line, respectively) and for the Einasto profile of the Aquarius simulation (light-blue dashed line). The blue dashed-dotted line shows the limits without GDE but 1% systematic uncertainty. We compare our limits to previous results for the same observation time but not including systematic uncertainties and GDE [68, 73, 95]. The annihilation cross section resulting in the right relic density is shown as a gray dashed line using the result from Ref. [98] (and an extrapolation thereof to 100 TeV). The NLO approximation of pppc4dmid breaks down for $m_{\chi }>10\mathrm{TeV}$, indicated by the gray shaded region.

Figure 8

Limits on the annihilation cross section for the different EFT operators and DM density profiles. The NLO approximation of pppc4dmid breaks down for $m_{\chi }>10\mathrm{TeV}$, indicated by the gray shaded region.

Figure 9

Lower limits on EFT scale $M_{\star }$ (red shaded regions) and comparison to LHC (purple shaded regions) and DD limits (green lines) for each tested EFT operator. Values of $M_{\star }$ and $m_{\chi }$ within the gray band result in the DM relic abundance as measured with Planck and the adopted range of values for $x_F$ and $g_{*}$. The EFT approximation is valid above the black dotted lines.

Figure 10

Examples for excluded annihilation cross sections for the different simplified models. Limits are shown for both considered DM density profiles (blue and orange lines) and for different mediator masses $M_{\mathrm{med}}=0.1$, 0.3, 1, and 3.2 TeV. Theoretical cross sections are also shown in black and are upscaled by a factor of $10^6$ and 10 for the scalar and axial-vector DM, respectively. The NLO approximation of pppc4dmid breaks down for $m_{\chi }>10\mathrm{TeV}$, indicated by the gray shaded region.

Figure 11

Projected limits and excluded values of $m_{\chi }$ vs $M_{\mathrm{med}}$ for the different DM models. CTA observations will exclude combinations of $m_{\chi }$ and $M_{\mathrm{med}}$ indicated by the orange (NFW and Einasto DM profiles) and blue (NFW only) squares. Only the parameter space to the right of the green and purple solid lines is allowed from current DD and dijet limits, respectively (dashed lines show projections of future searches). Purple regions show LHC monojet limits and projections. The gray lines show the parameter values that yield the correct relic abundance. No points are excluded in the scalar Dirac DM case for $g_q=g_{\chi }=1$ and none of the excluded parameters of the axial-vector model obey perturbative unitarity; thus these cases are not shown.