All-flavor based searches for solar dark matter with the IceCube Neutrino Observatory

Abstract

Dark matter particles may be trapped in large celestial bodies as the sun and by self-annihilation can produce a detectable neutrino flux on Earth. Well shielded volumes of natural media are used to register neutrinos by means of their Cherenkov signatures after they undergo charged or neutral-current interactions. One of these detectors, the IceCube neutrino observatory, is located in the clear glacial ice beneath the geographic South Pole, comprising a volume of one cubic kilometer which is monitored by 5160 photomultiplier modules. While traditional studies with the IceCube detector have concentrated on the good angular resolution of muon neutrino events and neglected other flavors of active neutrinos, this work attempts to achieve better sensitivities through an all-flavor based approach which increases the expected signal rate in the detector by a factor of two. The worse directional resolution of cascade-shaped events is improved by computationally intensive reconstructions and a newly developed uncertainty estimator enables the classification of individual events according to their reconstruction quality. Machine learning is applied at the final step of a multi-level event selection which aims at extracting the signal from the abundant background of atmospheric muons and neutrinos. Sensitivity limits on the annihilation rate are then obtained by means of a likelihood analysis using energy and directional information including the angular uncertainty. These sensitivities are finally interpreted as spin-dependent cross-section bounds within the supersymmetric framework of the pMSSM for which 100 billion possible models were scanned. Compared to present track-based searches with IceCube, the sensitivity for low dark matter masses could be improved by up to one order of magnitude.