High energy window for probing dark matter with cosmic-ray antideuterium and antihelium^*

The existence of dark matter (DM) is supported by various astronomical observations at different scales, but the nature of DM particles is still unknown. As an important probe in indirect DM detection, the antiparticles in cosmic rays (CR) may shed light on the properties of DM. In recent years, a number of experiments have shown an unexpected structure in the CR positron data [1-4], which could be related to DM annihilation or decay [5-8]. Unlike CR positrons, the CR antiproton flux data from PAMELA [9], BESS-polar II [10] and AMS-02 [11] do not show significant discrepancies with the secondary production of antiprotons, and these null results can be used to place stringent constraints on the DM annihilation cross-sections [12-15].

CR heavy anti-nuclei such as antideuterium () and antihelium-3 () are postulated to be important probes for DM [16-18]. The and in CR can be generated as secondary productions by collisions between primary CR particles and interstellar gas, or they can be produced by DM annihilation or decay. However, the secondary and are boosted to high kinetic energies because of the high production threshold in -collisions (17 for and 31 for , where is the proton mass), and thus the signal from DM can be distinguished in low energy regions (below GeV per nucleon). Although the fluxes of anti-nuclei decrease rapidly with the increase of the atomic mass number A, the high signal-to-background ratio at low energies and experiments with high sensitivities (such as AMS-02 [19, 20] and GAPS [21]) make it possible to distinguish the contributions from DM interactions. Furthermore, an advantage for considering and is that their productions are highly correlated with CR antiprotons, so the uncertainties of the and fluxes can be greatly reduced by the CR data [22].

In the literature (for a recent review, see Ref. [23]), the analysis of the DM-produced and are focused on the low kinetic energy regions, which we refer to as the low energy window. In our previous analysis [22], we studied the prospects of detecting DM through low energy antihelium. We systematically analysed the uncertainties from propagation models, DM density profiles and MC generators, and reduced the uncertainties by constraining the DM annihilation cross-sections with the AMS-02 data. However, the low energy window suffers from the uncertainties of solar activity (solar modulations). In this work, we investigate the possibility of probing DM with high energy CR and particles. In high energy regions (typically above 100 GeV per nucleon), the flux of primary CR particles drops quickly (the flux of CR proton is proportional to ), which leads to a suppression of the high energy secondary CR particles. As a result, the fluxes of anti-nuclei produced by DM annihilation may exceed the secondary background and open a high energy window for probing DM.

We use the Monte Carlo (MC) event generators PYTHIA 8.2 [24, 25], EPOS-LHC [26, 27] and DPMJET-III [28] to fit the coalescence momenta for anti-nuclei with experiments including ALEPH [29], CERN ISR [30] and ALICE [31], and generate the energy spectra of anti-nuclei. The propagation of CR particles is calculated using the GALPROP code. We use the AMS-02 [11] and HAWC data [32] and the HESS galactic center (GC) -ray data [33, 34] to constrain the DM annihilation cross-sections. We find that the conclusion depends on the choice of DM density profiles. For a large DM mass ( TeV) with the relatively flat "Cored" type DM profile, the high energy window exists, while for a typical steep DM profile like the "Cuspy" type, the high energy window closes.

This paper is organized as follows. In Section II, we briefly review the coalescence model and determine the coalescence momenta for anti-nuclei by fitting the ALEPH, ALICE and CERN-ISR data. In Section III, we review the theory of CR propagation. In Section IV, we constrain the DM annihilation cross-section by using the data from AMS-02 and HAWC and -ray data from HESS. The fluxes of and for direct DM annihilation and annihilation through mediator channels are presented in Section V. The conclusions are summarized in Section VI.

The formation of anti-nuclei can be described by the coalescence model [35-37], which uses a single parameter, the coalescence momentum , to quantify the probability of anti-nucleons merging into an anti-nucleus . The basic idea of this model is that anti-nucleons combine into an anti-nucleus if the relative four-momenta of a proper set of nucleons is less than the coalescence momentum. For example, the coalescence criterion for is written as:

where and are the four-momenta of the antiproton and antineutron respectively, and is the coalescence momentum of . If we assume that the momentum distributions of the and in one collision event are uncorrelated and isotropic, the spectrum of can be derived by the phase-space analysis:

where are the Lorentz factors, and .

For , we adopt the same coalescence criterion as in our previous analysis [22]. We compose a triangle using the norms of the three relative four-momenta and , where are the four-momenta of the three anti-nucleons respectively. Then, making a circle with minimal diameter to envelop the triangle, if the diameter of this circle is smaller than , an is generated. If the triangle is acute, the minimal circle is just the circumcircle of this triangle, and the coalescence criterion can be expressed as follows:

Otherwise, the minimal diameter is equal to the longest side of the triangle, and the criterion can be simply written as . See Ref. [22] for more details.

We use the MC generators PYTHIA 8.2, EPOS-LHC and DPMJET-III to simulate the hadronization after the DM annihilations and collisions, then use the coalescence model to produce anti-nuclei from the final state or . The spatial distance between each pair of anti-nucleons also needs to be considered, because the anti-nucleons should be close enough to undergo nuclear reactions then merge into anti-nuclei. We set all the particles with lifetime to be stable, to ensure that every pair of anti-nucleons is located within a short enough distance [17], where 2 fm is approximately the size of the nucleus.

The values of the coalescence momenta should be determined by the experimental data, which are often released in the form of coalescence parameters . The definition of the coalescence parameter is expressed by the formula:

where is the mass number of the nucleus, and Z and N are the proton number and the neutron number respectively. Under the assumption that the momentum distributions of the and are uncorrelated and isotropic, the relation is expected by comparing Eq. (2) and Eq. (4). However, the jet structure and the correlation between and play important roles in the formation of anti-nuclei. So we derive the coalescence momenta by using the MC generators to fit the experimental data, and thus the effect of jet structures and correlations are included.

To derive the coalescence momentum of in collisions, we follow the procedures described in Ref. [22] to fit the coalescence parameter data from the ALICE-7 TeV, 2.76 TeV, 900 GeV [31] and ISR-53 GeV [30] experiments. We use MC generators to simulate these experiments, and record the momenta information of , that are possible to form a . We select the and according to a sufficiently large coalescence momentum (600 MeV for , 1 GeV for ), and then calculate the spectra of for different values that are smaller than . The values are calculated by using Eq. (4), and we perform a analysis to find the value of . The best-fit values are shown in Fig. 1. We find for all these fits, which means the fitting results are in good agreement with the experimental data. As shown by the figure, the -dependences are reproduced well by the coalescence model and the MC generators.

Zoom In Zoom Out Reset image size

The fitting results for are listed in Table 1. The values in parentheses are the results given by the ALICE group, which are only available for the ALICE 7 TeV and ISR-53 GeV experiments with the PYTHIA 8.2 and EPOS-LHC generators. We can see that for ALICE 7 TeV, our best-fit values are in good agreement with those from the ALICE group, but for ISR-53 GeV, our results are larger. This is partly because the ALICE group has only generated the energy spectra of for six values [31], and used the isotropic approximation to interpolate the spectra for other values, while we generate the spectra for every integer number MeV and do not need the approximation. Moreover, in fitting the ISR-53 GeV experiment, the ALICE group has manually rescaled the spectra of generated by MC to better reproduce the experimental data, while for self-consistency we do not make this correction.

In Fig. 2, we show the values for different collision experiments with various MC generators. As can be seen, does not show a clear relation with the values of the experiments. Since the fitting results for different center-of-mass energies are similar, we assume that the value does not vary with the of the experiment. By fitting these , we get MeV, MeV, and MeV for collisions.

Zoom In Zoom Out Reset image size

By using PYTHIA to fit the data from the ALEPH experiment [29], we find MeV. Considering the similarity between the dynamics of the decay and DM annihilation, we set to be the coalescence momentum for the DM annihilation process .

For , we adopt the value obtained in our previous work [22], which determining by fitting the ALICE TeV -collision data. The results are listed in Table 2. Note that particles can be produced from two channels: direct formation from the coalescence of , or through the -decay of an antitriton (). The direct formation channel is suppressed by the Coulomb repulsion between the two antiprotons, so some previous works only considered the antitriton channel. However, our calculation shows that the direct formation channel is not negligible. The coalescence momentum of is only slightly smaller than that of , and the Gamow factor [38], which indicates that the Coulomb repulsion may not be significant. From these facts, it is reasonable to ignore the Coulomb repulsion in the MC simulation, and the contributions from both channels are included in this work. Our MC calculation shows that about 30% of is produced through the direct formation channel. Due to the lack of experiment data, we set to be the coalescence momentum for the DM annihilation process . It is known that the coalescence momentum varies for different processes and center-of-mass energies [39], and the relation makes a crucial factor for the uncertainties of the final fluxes. In some previous analyses, the value of was estimated using various approaches, for example by using the binding energy relation between and , or assuming the ratio [17], and some works just set [18]. The results from these approaches are different. It is worth mentioning that our value is relatively small compared to previous works, which leads to conservative DM contributions.

For the primary anti-particles originating from DM annihilations, the injection spectra are calculated using PYTHIA 8.2. We simulate the annihilation of Majorana DM particles by a positron-electron annihilation process , where is a fictitious scaler singlet and f is a standard model final state. We set and switch off all initial-state-radiations in PYTHIA 8.2, to mimic the dynamics of DM annihilation. Three kinds of final state are considered: (q stands for u or d quark), and . In the EPOS-LHC and DPMJET-III generators, only hadrons can be set as the initial states, which does not resemble the properties of DM annihilations.

For the secondary anti-particles produced in -collisions, EPOS-LHC and DPMJET-III are used to generate the energy spectra. The default parameters in PYTHIA 8.2 (the Monash tune [40]) are focused to reproduce the experimental results at high center-of-mass energies (like ATLAS at TeV [41]), but are not optimized for energy regions around a few tens of GeV, which give the dominating contributions for secondary CR anti-particles. To evaluate the performance of MC generators at low center-of-mass energies, we compare the differential invariant cross-section obtained by MC generators and the NA49 data at GeV [42]. This comparison shows that EPOS-LHC has the best performance, and DPMJET-III is also in relatively good agreement with experiment, while the production cross-section of given by PYTHIA 8.2 is larger than the NA49 data by roughly a factor of two. In this paper, we will draw our conclusion based on the results from EPOS-LHC, and the difference between EPOS-LHC and DPMJET-III can be used as a rough estimation of the uncertainties from different MC generators.

The propagation of charged CR particles is assumed to be random diffusions in a cylindrical diffusion halo with radius kpc and half-height kpc. The diffusion equation is written as [43, 44]:

where is the number density in phase spaces at the particle momentum p and position , and is the source term. is the spatial diffusion coefficient, which is parameterized as , where is the rigidity of the CR particle with electric charge , is the spectral power index, which takes two different values when R is below (above) a reference rigidity , is a constant normalization coefficient, and is the velocity of CR particles. quantifies the velocity of the galactic wind convection. The diffusive re-acceleration is described as diffusions in the momentum space, which is described by the parameter , where is the Alfvén velocity that characterizes the propagation of weak disturbances in a magnetic field. is the momentum loss rate, and and are the timescales of particle fragmentation and radioactive decay respectively. For boundary conditions, we assume that the number densities of CR particles vanish at the boundary of the halo: . The steady-state diffusion condition is achieved by setting . We numerically solve the diffusion equation Eq. (5) by using the GALPROP v54 code [45-49]. The primary CR nucleus injection spectra are assumed to have a broken power law behavior , with the injection index for the nucleus rigidity below (above) a reference value . The spatial distribution of the interstellar gas and the primary sources of CR nuclei are taken from Ref. [45].

The injection of CR particles is described by the source term in the diffusion equation. For the primary CR antiparticles originating from the annihilation of Majorana DM particles, the source term is given by:

where is the energy density of DM, is the thermally-averaged DM annihilation cross-section, and is the energy spectrum of discussed in the previous section. The spatial distribution of DM is described by DM profiles. In this work, we consider four commonly-used DM profiles to represent the uncertainties: the Navarfro-Frenk-White ("NFW") profile [50], the "Isothermal" profile [51], the "Moore" profile [52, 53] and the "Einasto" profile [54].

For the secondary produced in collisions between primary CR and interstellar gas, the source term can be written as follows:

where is the number density of CR components (proton, helium or antiproton) per unit momentum, is the number density of interstellar gases (hydrogen or helium), and is the inelastic cross-section for the process , which is provided by the MC generators. is the energy spectrum of in the collisions, with the momentum of incident CR particles. For , we include the contributions from the collisions of , , , , and . For the secondary and , since the experimental data are only available in -collisions, we consider the contribution from collisions between CR protons and interstellar hydrogen, which dominates the secondary background of and . The tertiary contributions of and are not included, as they are only important at low kinetic energy regions below 1 GeV [55], which are not relevant to our conclusions.

The fragmentation timescale in Eq. (5) is inversely proportional to the inelastic interaction rate between the nucleus and the interstellar gas, which is estimated as [17, 18]

where and are the number densities of interstellar hydrogen and helium, respectively, is the geometrical factor of helium, v is the velocity of relative to interstellar gases, and is the total inelastic cross-section of the collisions between and the interstellar gas. The number density ratio in the interstellar gas is taken to be 0.11 [45], which is the default value in GALPROP.

Since experimental data for the inelastic cross sections and are currently not available, we assume the relation , which is guaranteed by CP invariance. For an incident nucleus with atomic mass number A, charge number Z and kinetic energy T, the total inelastic cross-section for collisions is parameterized by the following formula [46]:

where . For example, by substituting and , one obtains the cross-section .

Finally, when anti-nuclei propagate into the heliosphere, the spectra of charged CR particles are distorted by the magnetic fields of the solar system and solar wind. The effects of solar modulation are quantified by the force-field approximation [56]:

where is the flux of CR particles, which is related to the density function f by . "TOA" denotes the value at the top of the atmosphere of the earth, "IS" denotes the value at the boundary between interstellar space and the heliosphere, and m is the mass of the nucleus. is related to as . In this work, we set the value of the Fisk potential at MV.

A. Constraints from AMS-02 and HAWC data

The experimental CR data shows good agreement with the scenario of secondary productions, and thus it is expected to place stringent constraints on the DM annihilation cross-sections. Since the production of anti-nuclei is strongly correlated with the antiproton, these constraints can greatly reduce the uncertainties of the maximal flux of and originating from DM. In 2016, the AMS-02 group [11] released the currently most accurate ratio data in the rigidity range from 1 to 450 GV. Recently, the HAWC group [32] published the upper limit of the ratio in very high energy regions, obtained by using observations of the moon shadow. In this paper, we use these two sets of ratio data to constrain the upper limit of the DM annihilation cross-sections.

To quantify the uncertainties from CR propagation, we consider three different propagation models, the "MIN", "MED" and "MAX" models [57]. The parameters of these models are obtained by making a global fit to the AMS-02 proton flux and B/C ratio data using the GALPROP-v54 code, and the names of these models represent the typically minimal, median and maximal antiproton fluxes due to the uncertainties of propagation. The parameters of these three models are listed in Table 3. In our calculations, we adopt the default normalization scheme in GALPROP, which normalizes the primary nuclei source term to reproduce the AMS-02 proton flux at the reference kinetic energy GeV.

Table 3. Values of the main parameters in the "MIN", "MED" and "MAX" models, derived from fitting to the AMS-02 and proton data based on the GALPROP code [57]. The parameter is in units of .

Model	/kpc	/kpc		/GV		/(km/s)	/GV
MIN	20	1.8	3.53	4.0	0.3/0.3	42.7	10.0	1.75/2.44
MED	20	3.2	6.50	4.0	0.29/0.29	44.8	10.0	1.79/2.45
MAX	20	6.0	10.6	4.0	0.29/0.29	43.4	10.0	1.81/2.46

The 95% CL upper limits of DM annihilation cross-sections are derived by performing a frequentist analysis, with defined as:

$$\chi^2 = \sum\limits_{i}\frac{(f^{\mathrm{th}}_i-f^{\mathrm{exp}}_i)^2}{\sigma^2_i}, \tag{ 11 }$$

where is the central value of the experimental ratio, is the data error, is the theoretical prediction of , and i denotes the th data point. For the 95% CL upper limits of given by the HAWC experiment, we set , and the value of the upper limit corresponds to . For a specific DM mass, we first calculate the minimal value of , and then the 95% CL upper limits on DM annihilation cross-sections corresponding to for one parameter. See Ref. [13] for more details. The upper limits for different annihilation channels are shown in Fig. 3, with the production cross-section of secondary generated by EPOS-LHC. We can see that the differences between the upper limits for various propagation models and DM profiles can reach one to two orders of magnitude. As shown in previous works [17, 58], the final flux uncertainties from propagation models and DM profiles can be larger than one order of magnitude. However, if we use these cross-section upper limits to constrain the maximal flux of and , the final uncertainties from the propagation model and the DM profile can be reduced to merely 30% [22]. Comparisons of the maximal fluxes in different propagation models and DM profiles are shown in Fig. 4. It can be seen that the uncertainties are small.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

B. Constraints from HESS galactic center -ray data

The galactic center (GC) is a promising place for detecting DM interactions, because of the expected high DM density. The 10-year HESS -ray data [34] focus on a small area around the galactic center, and constrains the DM annihilation cross-sections well for "Cuspy" type DM profiles, which have a large gradient at the inner galactic halo. These include the "Einasto" profile and the "NFW" profile. For DM mass TeV, the galactic center -ray can place more stringent upper limits than the data. However, for "Cored" type DM profiles, which are flat near the galactic center, such as the "Isothermal" profile, the constraints from the -ray data are relatively weak.

The latest galactic center -ray analysis, with 254 hours' exposure, was published in 2016 by HESS [34]. The upper limits are calculated for several DM profiles and DM annihilation channels, but the -ray flux data have not been publicly released. Since the relative statistical error of a data point is inversely proportional to the square root of the number of counts in the bins, we can approximately estimate the 254-hour results by rescaling previous HESS -ray data [33], which were published in 2011 and for which exposure time was 112 hours. To obtain the upper limits for other DM profiles and channels, we perform a analysis to the 112-hour -ray flux data and estimate the 254-hour results by rescaling the data errors by a factor . The flux residual is defined as , where and are the experimental -ray flux data from the source region and from the background region respectively, and the error of is provided in Ref. [33]. is the theoretical value of the flux residual, and the differential flux is calculated by the following formula:

$$F^{\mathrm{th}} = \frac{\mathrm{d}\Phi_{\gamma}}{\Omega\mathrm{d}E_{\gamma}} = \frac{\langle\sigma v\rangle}{8\pi m_{\chi}^2\Omega}\frac{\mathrm{d} N_{\gamma}}{\mathrm{d}E_{\gamma}} \int_{\Omega}\int_{\mathrm{l.o.s}}\rho^2(r(s,\theta))\mathrm{d}s \mathrm{d}\Omega, \tag{ 12 }$$

where is the energy spectrum of photons produced in one DM annihilation, and is the total spherical angle of the source or background regions. We adopt the reflected background technique described in Ref. [33] to determine the source region and background regions, and calculate the flux residual to perform the analysis. We randomly choose 540 pointing positions near the GC, with the maximal distance between the pointing position and the GC being . We first calculate the minimal value, and then the 95% CL upper limit corresponds to . The results are shown in Fig. 5. We can see that there are large gaps between the upper limits for different DM profiles. As expected, the constraints are stringent for DM profiles that are cuspy at the GC, but for a cored one like the Isothermal profile, the limits are rather weak. For various DM profiles, the constraints from data are are similar, because the DM densities in the diffusion halo are similar in different DM profiles, except for the GC region. However, for -ray data, the GC region provides the most stringent constraints for heavy DM particles, which leads to the large gaps between DM profiles.

Zoom In Zoom Out Reset image size

We compare the upper limits from the data and from the -ray data, as presented in Fig. 6. We use the "MED" propagation model as the benchmark model, and the "Isothermal" and "Einasto" profiles represent the typical "Cored" and "Cuspy" profiles respectively. The "Isothermal" profile is parameterized as follows:

Zoom In Zoom Out Reset image size

$$\rho_{\mathrm{DM}} = \rho_{\odot}\frac{1+(r_{\odot}/r_{\mathrm{Iso}})^2}{1+(r/r_{\odot})^2}\; , \tag{ 13 }$$

where is the local DM energy density, . The "Einasto" profile can be written as:

$$\rho_{\mathrm{DM}} = \rho_{\odot}\exp\left[-\left(\frac{2}{\alpha}\right)\left(\frac{r^{\alpha}- r^{\alpha}_{\odot}}{r^{\alpha}_{\mathrm{Ein}}}\right)\right]\; , \tag{ 14 }$$

where and kpc. As we can see, in all annihilation channels, for the "Isothermal" profile, the upper limits from data are much more stringent than those from the -ray data, while for the "Einasto" profile, the -ray data give stricter limits than for DM mass larger than 2 TeV.

It is worth mentioning that the Fermi-LAT experiment [59] also collected a large amount of -ray data near the GC, and the relatively large region of interest can reduce the large gaps between different DM profiles. However, the energy range of Fermi-LAT observations (from 20 MeV to more than 300 GeV) is much lower than that of HESS, and thus provides a weaker limitation on large DM mass. For example, the analysis in Ref. [60] shows that for a steep profile like "NFW", HESS gives stronger constraints than Fermi-LAT at TeV. For a flat profile like "Isothermal", the Fermi-LAT limits derived in Ref. [61] are slightly weaker than the constraints at TeV. From these facts, to make the most conservative conclusion, we use the limits to calculate the maximal and fluxes in the "Isothermal" profile, and use the HESS -ray limits for the "Einasto" profile in the following sections.

A. Direct DM annihilation

With the DM annihilation cross-section upper limits to hand, we can derive the maximal and fluxes for different annihilation channels, propagation models, and DM profiles. As shown in Fig. 4, by constraining the DM annihilation cross-section with the data, the uncertainties from the propagation models are small. We thus present our results for the "MED" propagation model, and other models do not affect our conclusions. For illustrative purposes, the particle mass of the heavy DM is set to be and 50 TeV, which is smaller than the unitarity bound for a self-conjugate DM [62]. However, it is worth mentioning that our analysis is largely model-independent, and we do not assume that DM has thermal origins.

Currently, the AMS-02 detector has the strongest detection ability for both and . For the flux, the AMS-02 detection sensitivity is given in Ref. [39], while the sensitivity for has only been released in terms of the ratio in Ref. [20]. To study the detection prospects of AMS-02, we present our results in terms of fluxes and ratios.

The results for the "Isothermal" profile with and 50 TeV are presented in Fig. 7, with the DM annihilation cross-section constrained by the AMS-02 and HAWC data. The top three figures show the results for fluxes for different annihilation channels, while the bottom three figures present the ratio results. The blue shaded areas represent the prospective AMS-02 detection sensitivity after 18 years of data collection, and the error bands show the uncertainties from coalescence momenta. Note that for , the error bands for the secondary background are thinner than the line width. We can see that for , the DM contributions exceed the secondary backgrounds in the energy region GeV. For and channels and TeV, the excess can be as large as one order of magnitude. Similarly, for , the excess exists at a kinetic energy of around 800 GeV per nucleon, and the primary fluxes originating from DM annihilation can be 20 times larger than the secondary background with TeV. Despite the fluxes of these anti-nuclei being small at high kinetic energies, and far below the AMS-02 sensitivities, these excesses can be promising windows for future detections.

Zoom In Zoom Out Reset image size

The results for the "Einasto" profile are shown in Fig. 8, with the DM annihilation cross-sections constrained by the HESS 10-year GC -ray data. For , the DM contributions are below the secondary background in all energy regions and annihilation channels, and thus the high energy window closes. However, for , the conclusion depends on the choice of MC generators. The DM contributions are lower than the secondary background given by EPOS-LHC, but can still exceed the DPMJET-III background.

Zoom In Zoom Out Reset image size

B. DM annihilation through mediators

We also consider the process , where two DM particles first annihilate into a couple of mediators, and then the mediators decay into standard model final states. If the mass of the mediator is much smaller than that of the DM particle, the mediator will be highly boosted. The and produced in this process are thus expected to assemble in high energy regions, which provides a high signal-to-background ratio for the high energy window.

By following the steps described in Sec. IV, we obtain the 95% CL upper limit of cross-sections for the annihilation process with mediators. The results for mediator mass GeV are presented in Fig. 9. Similar to the results for direct annihilations, for the "Isothermal" profile, the most stringent constraints are from data, while for the "Einasto" profile, the -ray limitations are stricter. Again, we use the AMS-02 and HAWC data to constrain the "Isothermal" profile, and the "Einasto" profile is restricted by the HESS GC -ray data.

Zoom In Zoom Out Reset image size

The results for the "Isothermal" profile and "Einasto" profile with GeV are presented in Fig. 10 and Fig. 11 respectively. As expected, the DM contributions are boosted to high energy regions, and we get similar conclusions as in the direct annihilation channels. For the "Isothermal" profile with TeV, the high energy window opens for both and in all decay channels, especially for , where the excesses can reach two orders of magnitude in the and channels. For TeV, the DM contributions for and are comparable to the secondary backgrounds.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

However, as shown in Fig. 11, for the "Einasto" profile, the excesses in high energy regions disappear for for all DM masses and mediator decay channels. For , the contributions from DM with TeV can be larger than the background calculated by using DPMJET-III. For EPOS-LHC, however, the only excess appears in the decay channel with TeV.

For other mediator masses, we reach the same conclusion. Taking the channel as an example, we calculate the fluxes and ratios for mediator mass , 200 and 600 GeV, and present the results for the "Isothermal" profile in Fig. 12. It can be seen that with the growth of the mediator mass, the DM contributions in low energy regions increase significantly. However, in high energy regions, in which we are interested, the variations of the results are small.

Zoom In Zoom Out Reset image size

In summary, we have explored the possibility of probing DM by high energy CR anti-nuclei. We used the MC generators PYTHIA 8.2, EPOS-LHC and DPMJET-III and the coalescence model to calculate the spectra of anti-nuclei. The coalescence momenta of and were derived by fitting the data from ALICE, ALEPH and CERN ISR experiments. The propagation of charged CR particles was calculated by the GALPROP-v54 code, with the inelastic interaction cross-sections between the primary CR and interstellar gases given by MC generators. We used the HESS GC -ray data to constrain the DM annihilation cross-sections for the DM profiles with large gradient at GC, while the flat profiles were limited by the AMS-02 and HAWC data.

Our results showed that, for a "Cored" type DM density profile like the "Isothermal" profile, the high energy window opened for both and in all channels. However, for a "Cuspy" type profile like the "Einasto" profile, the contributions from DM annihilations were below the secondary background in both DM direct annihilation and annihilation through mediator decay channels. For , the conclusion depended on the choice of MC generators. The flux originating from DM annihilations could exceed the secondary background for DPMJET-III, while the excess disappeared for EPOS-LHC.

Signals in high energy regions can effectively avoid the uncertainties from the solar activities. Although the fluxes of and in high energy regions are far below the sensitivity of current experiments like AMS-02 and GAPS, the high energy window could be a promising probe of DM for next-generation experiments. We believe that with the fast development of detector technologies, DM detection through the high energy window will eventually be possible.