Search for sterile neutrinos with the IceCube Neutrino Observatory

One of the open questions of particle physics is whether there are more than the three known flavors of neutrinos. The properties of these new particles would have to deviate from the known neutrinos by either being heavier than half the mass of the Z boson, or by not participating in the weak interaction. The latter of these alternatives gained relevance in the recent years from neutrino oscillation experiments that observed anomalies in their measured neutrino rates that could be explained by the introduction of one or more of these sterile neutrinos with a squared-mass difference $\Delta\mathrm{m}^2_{42}$ of roughly $1\,\mathrm{eV}^2$. In vacuum, the oscillation probabilities between the known flavors and potential sterile neutrinos have to be small, because otherwise they would have been detected already. However, when neutrinos traverse matter, the differences in the way the flavors can interact with matter induce an additional potential that can lead to resonances that can strongly increase these probabilities. Together with the large energy range, the high density of the Earth's core and mantle causes atmospheric neutrinos to be ideal to search for muon neutrino disappearance signatures of sterile neutrinos in the $1\,\mathrm{eV}^2$ range. The IceCube Neutrino Observatory at the Geographic South Pole is capable of recording several ten thousands of atmospheric muon neutrino events per year with a purity of over 99.9%. This analysis uses one year of data from an incomplete detector configuration with 59 of the final 86 strings installed. A likelihood ratio test is performed as scan in two parameters -- the sterile mixing angle $\theta_{24}$ and the squared mass difference $\Delta\mathrm{m}^2_{42}$ -- on two-dimensional histograms of neutrino events versus reconstructed energy and zenith angle to search for disappearance signatures. Systematic uncertainties in the atmospheric neutrino flux and detector responses have been taken care of, and have been parameterized and implemented into the likelihood function as nuisance parameters where applicable. The analysis does not find evidence for sterile neutrinos. The best fit lies at $\theta_{24} = 12.08^\circ$, $\Delta\mathrm{m}^2_{42} = 0.0401\,\mathrm{eV}^2$ and is compatible with the null hypothesis at a significance of $0.463\,\sigma$; it lies outside the median expected 90% C.L. sensitivity at the lower boundary of the analysis' energy range. The goodness of fit is good, and exclusion contours in the two scan parameters are given. Possible improvements to this analysis method are discussed in the outlook