eV-Scale Sterile Neutrino Search Using Eight Years of Atmospheric Muon Neutrino Data from the IceCube Neutrino Observatory

The results of a $3+1$ sterile neutrino search using eight years of data from the IceCube Neutrino Observatory are presented. A total of 305 735 muon neutrino events are analyzed in reconstructed energy-zenith space to test for signatures of a matter-enhanced oscillation that would occur given a sterile neutrino state with a mass-squared differences between 0.01 and $100{\mathrm{eV}}^2$. The best-fit point is found to be at ${\sin }^2(2{\theta }_{24})=0.10$ and $\mathrm{\Delta }{m}_{41}^2=4.5{\mathrm{eV}}^2$, which is consistent with the no sterile neutrino hypothesis with a $p$ value of 8.0%.

Figure 1

Muon-antineutrino oscillogram. Atmospheric ${\overline{\nu }}_{\mu }$ disappearance probability vs true energy and cosine zenith at the globally preferred sterile neutrino hypothesis of Ref. [11] [$\mathrm{\Delta }{m}_{41}^2=1.3{\mathrm{eV}}^2$, ${\sin }^2(2{\theta }_{24})=0.07$, ${\sin }^2(2{\theta }_{34})=0.0$]. Effects include a matter-enhanced resonance at TeV energies, neutrino absorption at high energy and small zenith, and vacuumlike oscillation at low energies. The matter-enhanced resonance appears only in the antineutrino flux for the case of small angles and $\mathrm{\Delta }{m}_{41}^2>0$. Vertical white lines indicate transitions between inner to outer core [$\cos ({\theta }_{\nu }^{\mathrm{true}})=-0.98$] and outer core to mantle [$\cos ({\theta }_z^{\mathrm{true}})=-0.83$].

Figure 2

Reconstructed muon energy. Data points are shown as black markers with error bars that represent the statistical error. The solid blue and red lines show the best-fit sterile neutrino hypothesis and the null (no sterile neutrino) hypothesis, respectively, with nuisance parameters set to their best-fit values in each case.

Figure 3

Best-fit signal shapes compared to data. Top: the signal shape at the best-fit point compared to the null hypothesis with the same nuisance parameters. Bottom: data compared to the null hypothesis with the nuisance parameters held at the same values.

Figure 4

Frequentist analysis result. The 90% and 99% C.L. contours, assuming Wilks’s theorem, shown as dashed and solid bold blue lines, respectively. The green and yellow band shows the region where 68% and 95% of the pseudoexperiment 99% C.L. observations lie; the dashed white line corresponds to the median. Other muon-neutrino disappearance measurements at 99% C.L. are shown in black [25, 26, 27, 28, 29, 30, 123, 124]; where results were not available at 99% C.L., methods of Ref. [11] were applied using public data releases. Finally, the star marks the analysis best-fit point location.

Figure 5

Bayesian analysis result. The logarithm of the Bayes factor [125] relative to the null hypothesis (color scale). Red indicates hypotheses preferred over the null hypothesis, while the blue indicates the null is preferred. Solid lines delineate likelihood ratios of 1 in 10 for a priori equally likely hypotheses. The best-model location is shown at the white star with a ${\log }_{10}$ (Bayes factor) minimum of $-1.03$.