Sterile neutrino oscillations after first MiniBooNE results

TL;DR

The paper analyzes short-baseline neutrino oscillations including one, two, or three sterile neutrinos in light of MiniBooNE results. It systematically tests (3+1), (3+2), and (3+3) mass schemes using appearance and disappearance data, employing a parameter goodness-of-fit metric to quantify data-set compatibility. The main finding is a persistent tension between appearance and disappearance data: (3+1) remains strongly disfavoured, (3+2) can fit LSND and MB in appearance via CP violation but struggles with disappearance constraints, and (3+3) offers little improvement. The results imply that, except for discarding MB, sterile-neutrino explanations remain disfavoured, though the (3+2) scenario remains the most compatible option among the considered models and motivates further CP-violation tests with antineutrino data.

Abstract

In view of the recent results from the MiniBooNE experiment we revisit the global neutrino oscillation fit to short-baseline neutrino data by adding one or two sterile neutrinos with eV-scale masses to the three Standard Model neutrinos, and for the first time we consider also the global fit with three sterile neutrinos. Four-neutrino oscillations of the (3+1) type have been only marginally allowed before the recent MiniBooNE results, and become even more disfavored with the new data (at the level of $4σ$). In the framework of so-called (3+2) five-neutrino mass schemes we find severe tension between appearance and disappearance experiments at the level of more than $3σ$, and hence no satistfactory fit to the global data is possible in (3+2) schemes. This tension remains also when a third sterile neutrino is added, and the quality of the global fit does not improve significantly in a (3+3) scheme. It should be noted, however, that in models with more than one sterile neutrino the MiniBooNE results are in perfect agreement with the LSND appearance evidence, thanks to the possibility of CP violation available in such oscillation schemes. Furthermore, if disappearance data are not taken into account (3+2) oscillations provide an excellent fit to the full MiniBooNE spectrum including the event excess at low energies.

Figures (12)

• Figure 1: Allowed region for MB475 (solid and dashed curves) and LSND+KARMEN+NOMAD (shaded regions) at 90% and 99% CL (2 dof) in (3+1) mass schemes.
• Figure 2: Allowed regions in (3+1) schemes from no-evidence (NEV) data including MB475 (solid and dashed curves) and LSND (shaded regions) at 90% and 99% CL (2 dof).
• Figure 3: MB spectral data in bins of reconstructed CCQE neutrino energy. The histograms show the prediction at the best fit points in (3+2) mass schemes for SBL appearance data LSND, KARMEN, NOMAD, MB (left), and for the global data (right). For the solid histograms the full MB energy range has been used in the fit (MB300), whereas for the dashed histogram the two lowest energy data points have been omitted (MB475). The corresponding parameter values are given in Tab. \ref{['tab:bfp']}.
• Figure 4: The $\chi^2$ in (3+2) schemes as a function of the CP phase $\delta$ defined in Eq. \ref{['eq:5nu-def']} for appearance data from LSND, KARMEN, NOMAD, and MB (bottom), and for global data (top). Results are shown without (MB475), and including (MB300) the low energy data from MB. All other parameters have been minimised, respecting the constraint $\Delta m^2_{51} \ge \Delta m^2_{41}$.
• Figure 5: Allowed regions for SBL appearance data in (3+2) schemes at 90%, 95%, 99%, 99.73% CL (2 dof) in the plane of $\Delta m^2_{41}$ and $\Delta m^2_{51}$. All other parameters have been minimised. We use data from LSND, KARMEN, NOMAD, and MB475 (left) or MB300 (right).