Are There Sterile Neutrinos at the eV Scale?

New predictions for the antineutrino flux from nuclear reactors suggest that reactor experiments may have measured a deficit in this flux, which can be interpreted in terms of oscillations between the known active neutrinos and new sterile states. We perform a reanalysis of global short-baseline neutrino oscillation data in a framework with one or two sterile neutrinos. While one sterile neutrino is still not sufficient to reconcile the signals suggested by reactor experiments and by the LSND and MiniBooNE experiments with null results from other searches, we find that, with the new reactor flux prediction, the global fit improves considerably when two sterile neutrinos are introduced.

Figure 1

Comparison of sterile neutrino models to reactor data: energy spectra from Bugey3 and the rate measurement of Bugey4 (inset). The data points show the ratio of the observed and predicted event numbers where the prediction is based on the new reactor antineutrino fluxes [5] and does not include oscillations. Bugey3 error bars are statistical only, whereas the error on the Bugey4 rate is dominated by systematics. The green dot-dashed curve shows the prediction for the no oscillation hypothesis, the blue dashed and thin light red solid curves correspond to the $3+1$ and $3+2$ best-fit points for SBL reactor data (Table ), and the thick dark red solid curve corresponds to the $3+2$ best-fit point of global SBL data (Table ).

Figure 2

${\chi }^2$ from global SBL data (upper panel) and from SBL reactor data alone (lower panel) for the $3+1$ (blue) and $3+2$ (red) scenarios. Dashed (solid) curves were computed using the old [6] (new [5]) reactor ${\overline{\nu }}_e$ flux prediction. All undisplayed parameters are minimized over. The total number of data points is 137 (84) for the global (reactor) analysis.

Figure 3

Global constraints on sterile neutrinos in the $3+1$ model. We show the allowed regions at 90% and 99% C.L. from a combined analysis of the LSND [3] and MiniBooNE antineutrino [4] signals (filled regions), as well as the constraints from the null results of KARMEN [20], NOMAD [21] and MiniBooNE neutrino [19] appearance searches (blue dot-dashed contour). The limit from disappearance experiments includes data from CDHS [22], atmospheric neutrinos, and from the SBL reactor experiments. We compare results obtained using the new ${\overline{\nu }}_e$ flux prediction [5] (solid) to those obtained using the previous prediction [6] (dashed). The region to the right of the curves is excluded at 99% C.L.

Figure 4

Predicted spectra for MiniBooNE data and the transition probability for LSND (inset). Solid histograms refer to the $3+2$ global best-fit point (Table ), dashed histograms correspond to the best fit of appearance data only (LSND, MiniBooNE $\nu /\overline{\nu }$, KARMEN, NOMAD). For MiniBooNE we fit only data above 475 MeV.

Figure 5

The globally preferred regions for the neutrino mass-squared differences $\Delta m_{41}^2$ and $\Delta m_{51}^2$ in the $3+2$ (upper left) and $1+3+1$ (lower right) scenarios.