Observation of High-Energy Astrophysical Neutrinos in Three Years of IceCube Data

A search for high-energy neutrinos interacting within the IceCube detector between 2010 and 2012 provided the first evidence for a high-energy neutrino flux of extraterrestrial origin. Results from an analysis using the same methods with a third year (2012–2013) of data from the complete IceCube detector are consistent with the previously reported astrophysical flux in the 100 TeV–PeV range at the level of ${10}^{-8}\mathrm{GeV}{\mathrm{cm}}^{-2}{\mathrm{s}}^{-1}{\mathrm{sr}}^{-1}$ per flavor and reject a purely atmospheric explanation for the combined three-year data at $5.7\sigma$. The data are consistent with expectations for equal fluxes of all three neutrino flavors and with isotropic arrival directions, suggesting either numerous or spatially extended sources. The three-year data set, with a live time of 988 days, contains a total of 37 neutrino candidate events with deposited energies ranging from 30 to 2000 TeV. The 2000-TeV event is the highest-energy neutrino interaction ever observed.

Figure 1

Arrival angles and deposited energies of the events. Cosmic-ray muon background would appear as low-energy track events in the southern sky (bottom). Atmospheric neutrino backgrounds would appear primarily in the northern sky (top), also at low energies and predominantly as tracks. The attenuation of high-energy neutrinos in the Earth is visible in the top right of the figure. One event, a pair of coincident unrelated cosmic-ray muons, is excluded from this plot. A tabular version of these data, including additional information such as event times, can be found in Ref. [29].

Figure 2

Deposited energies of observed events with predictions. The hashed region shows uncertainties on the sum of all backgrounds. Muons (red) are computed from simulation to overcome statistical limitations in our background measurement and scaled to match the total measured background rate. Atmospheric neutrinos and uncertainties thereon are derived from previous measurements of both the $\pi /K$ and charm components of the atmospheric ${\nu }_{\mu }$ spectrum [9]. A gap larger than the one between 400 and 1000 TeV appears in 43% of the realizations of the best-fit continuous spectrum.

Figure 3

Arrival angles of events with $E_{dep}>60\mathrm{TeV}$, as used in our fit and above the majority of the cosmic-ray muon background. The increasing opacity of the Earth to high-energy neutrinos is visible on the right side of the plot. Vetoing atmospheric neutrinos by muons from their parent air showers depresses the atmospheric neutrino background on the left. The data are described well by the expected backgrounds and a hard astrophysical isotropic neutrino flux (gray lines). Colors are as in Fig. 2. Variations of this figure with other energy thresholds are in Ref. [29].

Figure 4

Extraterrestrial neutrino flux ($\nu +\overline{\nu }$) as a function of energy. Vertical error bars indicate the $2\mathrm{\Delta }\mathcal{L}=\pm 1$ contours of the flux in each energy bin, holding all other values, including background normalizations, fixed. These provide approximate 68% confidence ranges. An increase in the charm atmospheric background to the level of the 90% CL limit from the northern hemisphere ${\nu }_{\mu }$ spectrum [9] would reduce the inferred astrophysical flux at low energies to the level shown for comparison in light gray. The best-fit power law is $E^2\varphi (E)=1.5\times {10}^{-8}(E/100\mathrm{TeV}{)}^{-0.3}\mathrm{GeV}{\mathrm{cm}}^{-2}{\mathrm{s}}^{-1}{\mathrm{sr}}^{-1}$.

Figure 5

Arrival directions of the events in Galactic coordinates. Showerlike events (median angular resolution $\sim 15°$) are marked with $+$ and those containing muon tracks ($\lesssim 1°$) with $\times$. Approximately 40% of the events (mostly tracks [13]) are expected to originate from atmospheric backgrounds. Event IDs match those in the catalog in Ref. [29] and are time ordered. The grey line denotes the equatorial plane. Colors show the test statistic (TS) for the point source clustering test at each location. No significant clustering was observed.