GeV excess in the Milky Way: The role of diffuse galactic gamma-ray emission templates

Several groups have analyzed the publicly available Fermi-LAT data and have reported a spatially extended $\gamma$ ray excess of around 1–3 GeV from the region surrounding the Galactic center that might originate from annihilation of dark-matter particles with a rest mass $m_{\chi }\sim 30-40\mathrm{GeV}$. In this work we examine the role of the diffuse galactic gamma-ray emission templates played in suppressing the GeV excess. For such a purpose, we adopt in total 128 background templates that were generated by Ackermann et al. [Astrophys. J. 750, 3 (2012)] in the study of the Fermi-LAT observations of the diffuse gamma-ray emission considering the effects of cosmic rays and the interstellar medium. The possible GeV excess, assumed to follow the spatial distribution of the prompt gamma rays produced in the annihilation of dark-matter particles taking a generalized Navarro-Frenk-White profile with an inner slope $\alpha =1.2$, has been analyzed in some regions of interest. The introduction of such an additional component centered at the Galactic center is found to have improved the goodness of fit to the data significantly in all background template models regardless of whether the excess spectrum is fixed or not. Our results thus suggest that the presence of a statistically significant GeV excess in the inner Galaxy is robust, though its spectrum depends on the diffuse galactic gamma-ray emission model adopted in the analysis. The possible physical origin of the GeV excess component is discussed and, in the dark-matter model, the annihilation cross section of such particles is evaluated.

Figure 1

The regions of interest are chosen to minimize the uncertainties from background modeling. What is shown here are the count maps of gamma rays in the energy range of 300 MeV–300 GeV. In the top panel (i.e., ROI I), we mask $|b|<5°$ to reduce the contamination from known underpredicted gamma-ray emission above a few GeV in the galactic plane [31] and also to avoid analyzing the complex Galactic center. We also mask $|l|>80°$ to minimize the effect of the outer Galaxy region which dominates in an all sky likelihood ratio test but will not considerably contribute to the dark-matter annihilation signal [31]. In the bottom-left panel (i.e., ROI II), the Fermi bubbles are further masked to minimize its possible degeneracy with a dark-matter signal in the low latitude region. In the bottom-right panel, we mask the regions of $|b|<10°$ and $|l|>80°$ (i.e., ROI III) to minimize the contamination of the possibly misleading GeV emission, which would be less extended, from millisecond pulsars, pulsars, SNRs, bremsstrahlung, or neutral pion decay.

Figure 2

The dark-matter-like GeV excess identified in our analysis of the Fermi bubble regions with the P6V11 galactic diffuse template. (Left panel) The SED of five bubble slices and the corresponding background templates. (Right panel) The same as the left panel except that the dark-matter template is added. Our results are consistent with that found in [28]; i.e., in both of the slices for $0°<b<10°$ and $10°<b<20°$, a dark-matter-like GeV excess component is highly preferred.

Figure 3

Log-likelihood values obtained in the separate fits for each group of the galactic diffuse gamma-ray emission template, without and with the dark-matter component (the left and right panels, respectively). The zero levels of the log-likelihood values are arbitrary but are equivalent in the same region. Therefore, for a given region, the value difference between two models represents their likelihood ratio. For each group of templates, three regions of interest defined in Fig. 1 are considered, including ROI I for the top panel, ROI II for the middle panel, and ROI III for the bottom panel.

Figure 4

TS values for the presence of an additional dark-matter-like component obtained in the fits with different galactic diffuse emission templates (see Fig. 3 for legend).

Figure 5

Log-likelihood (adopted from Fig. 3) versus TS values (adopted from Fig. 4) in three regions of interest defined in Fig. 1. The solid (dashed) lines are the log-likelihood values without (with) an additional dark-matter-like component.

Figure 6

The spectrum energy distribution of the GeV excess, averaged within 10° of the Galactic center and assuming a generalized NFW profile with an inner slope $\gamma =1.2$, for all 128 of the DGE models. Note that the fits are preformed in ROI I. The excess peaking in the energy range of 1–3 GeV is distinct in all of the fits.

Figure 7

The further increase of the 2*log(-likelihood) values of the fits to the data with an additional spatially extended component thanks to the relaxing of the SED.

Figure 8

The $⟨\sigma v⟩$ and $m_{\chi }$ obtained in the $b\overline{b}$-annihilation-channel spectral fits to the GeV excess SED obtained in different DGE models. Note that the red, blue, green, and cyan areas represent the four kinds of cosmic ray distribution models (i.e., Lorimer, OBstar, SNR, and Yusifov), respectively. The dashed line is the latest thermal relic cross section of dark-matter annihilation [51]. We also show constraints for dark matter from Milky Way halo [52] and dwarf galaxies [50].

Figure 9

The influence of sources in the 2FGL on the fit results. Lorimer’s pulsar-traced cosmic ray distribution model has been adopted and the fits are in ROI I. The left panel is extracted from the top panel of Fig. 4. The right panel presents the results when all of the sources in the 2FGL were ignored.