Antiproton Bounds on Dark Matter Annihilation from a Combined Analysis Using the DRAGON2 Code

TL;DR

This study uses the DRAGON2 code to analyze AMS-02 antiproton data, constraining dark matter annihilation signals and setting limits on WIMP properties, while accounting for astrophysical and nuclear uncertainties.

Contribution

It introduces a comprehensive Bayesian analysis combining multiple cosmic-ray measurements with updated nuclear cross-sections to improve constraints on dark matter annihilation signals.

Findings

No significant dark matter signal detected.

Constraints exclude WIMP annihilation below 200 GeV at thermal cross-section.

Possible residual excess around 70 GeV with low significance.

Abstract

Early studies of the AMS-02 antiproton ratio identified a possible excess over the expected astrophysical background that could be fit by the annihilation of a weakly interacting massive particle (WIMP). However, recent efforts have shown that uncertainties in cosmic-ray propagation, the antiproton production cross-section, and correlated systematic uncertainties in the AMS-02 data, may combine to decrease or eliminate the significance of this feature. We produce an advanced analysis using the DRAGON2 code which, for the first time, simultaneously fits the antiproton ratio along with multiple secondary cosmic-ray flux measurements to constrain astrophysical and nuclear uncertainties. Compared to previous work, our analysis benefits from a combination of: (1) recently released AMS-02 antiproton data, (2) updated nuclear fragmentation cross-section fits, (3) a rigorous Bayesian parameter space scan that constrains cosmic-ray propagation parameters. We find no statistically significant preference for a dark matter signal and set strong constraints on WIMP annihilation to $b\bar{b}$, ruling out annihilation at the thermal cross-section for dark matter masses below $\sim200$~GeV. We do find a positive residual that is consistent with previous work, and can be explained by a $\sim70$~GeV WIMP annihilating below the thermal cross-section. However, our default analysis finds this excess to have a local significance of only 2.8$\sigma$, which is decreased to 1.8$\sigma$ when the look-elsewhere effect is taken into account.