Electron neutrino appearance in the NOvA experiment

NOvA is an off-axis, two-detector experiment studying neutrino oscillations with the νµ beam from Fermilab. This paper describes the νe appearance analysis, including data-driven constraints from Near Detector measurements. The data set corresponds to an exposure equivalent to 6.05 × 1020 protons-on-target in the Far Detector. We observed 33 νe candidates with a predicted background of 8.2 ± 0.8 (syst.), for a significance of appearance higher than 8σ. Preliminary results for the allowed values of δCP and θ23 in both hierarchies are presented.


Introduction
NOvA is a long-baseline accelerator neutrino experiment using the NuMI muon neutrino beam. The NuMI beam is produced by directing 120 GeV proton spills onto a graphite target [1]. The resulting hadrons are focused by two magnetic horns, and decay into neutrinos and other particles inside a long pipe. The neutrinos are observed by two detectors, Near (ND) and Far (FD), located 1 and 810 km away from the target and 14 mrad off the beam axis. The beam spectrum is narrowly peaked around 2 GeV, optimal to observe ν μ → ν e transitions. This allows NOvA to draw conclusions about the neutrino mass hierarchy, the θ 23 octant and δ CP . In Sec.2, we explain data-driven methods that improve the prediction from the simulation. In Sec.3, we compare the final prediction with the FD data and highlight some results. Further details about the experiment and ν e event classification (CVN), are discussed elsewhere in these Proceedings.

Far
Detector prediction using data-driven constraints 2.1. Signal prediction using ND ν μ data The prediction of the ν μ → ν e signal in the FD is constrained using the observed ν μ spectrum in the ND. Discrepancies between ν μ data and simulation are interpreted as an inexact modeling of the underlying true energy spectrum [2]. The ν e spectrum in the FD is adjusted accordingly. The exact distribution of the predicted ν e signal will further depend on the oscillation parameters. The signal expectation varies between 11 and 28 total events for fixed sin 2 θ 23 = 0.5, corresponding to (IH, δ CP = π/2) and (NH, δ CP = 3π/2) respectively.

Beam background prediction using ND ν μ and ν e data
Three types of beam-related backgrounds are estimated using the ND: neutral currents (NC), ν μ charged currents (CC), and the intrinsic ν e component in the NuMI beam (beam ν e CC). Since each one propagates differently to the FD, we use a combination of data-driven techniques to correct their relative proportions from the simulation. Muon neutrinos that contribute to the 2 GeV peak mainly result from the decay π + → ν μ +μ + . A few anti-muons that subsequently decay as μ + →ν μ + e + + ν e give rise to the intrinsic ν e component. At higher energies, the majority of ν μ and ν e originate in kaon decays. We use ν μ events selected in the ND to constrain the pion and kaon yields, and consequently the beam ν e component. Figure 1(a) compares the spectra of contained ν μ events in data and simulation; most events have a pion ancestor. After subtracting the background, the differences between data and the simulated sample are translated into weights as function of pion forward (tp z ) and transverse (tp T ) momenta, and then applied to the ν e CC from pions. Similarly, uncontained ν μ events with energies above 4.5 GeV predominantly have kaon ancestors, as seen in Figure  1(c). Any data/MC discrepancy is used to correct the kaon yield. The result of both corrections is a 2% decrease for ν e CC from pions, and a 17% increase for ν e CC from kaons. The ν μ CC and NC components are corrected using the number of Michel Electrons (ME) in data and MC (Figure 2). On average, ν μ CC interactions have one more ME than NC or beam ν e , resulting from the decay of the muon. Combining the beam ν e estimation above and a fit to the number of ME in ND data, all three components are constrained. On average, the beam ν e component is scaled up by 4%, NC up by 10% and ν μ CC up by 17%. The corrections obtained with the ND data are translated to FD background expectations using Far/Near ratios. Unlike the signal prediction, these have small variations with the oscillation parameters. An additional background component, ν τ CC, is read directly from the simulation. The expected background counts in the FD are 3.7 NC, 3.1 beam ν e , 0.7 ν μ CC and 0.1 ν τ CC.

Cosmic background prediction using FD data
The NOvA FD is on the surface and thus susceptible to cosmogenic background. A rejection of 1 part in 10 8 cosmic ray interactions is achieved using the time structure of the NuMI beam, event topologies, and particle identification; analysis cuts are optimized for higher signal efficiency, and tuned using an independent sample. Using FD ν e candidates outside of the beam time window, we estimate a total of 0.53 cosmic events that could coincide with the appearance signal.

Systematic uncertainties
The two-detector technique described mitigates the impact of many sources of systematic uncertainty. Residual effects are assessed via variations in the simulation. These can be classified as: normalization, flux, calibration, cross section, and detector response. The overall effect in the FD event count is 5% for signal and 10% for background. In the binned fit, systematic uncertainties are included as nuisance parameters.

Results
33 electron neutrino candidates were observed. With an expected background of 8.2±0.8 events, the significance of ν e appearance is greater than 8σ. A comparison between the FD data and the final prediction is presented in Figure 3; the reconstructed energy spectra are split in three ranges of the event classifier (CVN). A combination of the NOvA ν e measurement with global θ 13 and Δm 2 32 constraints gives best fit values for NH, δ CP = 1.59π, and sin 2 θ 23 = 0.45. The preference for these parameters has low statistical significance: several values are compatible with the data, both in NH and IH. The fit was also run using NOvA's ν μ disappearance results, in the form of a constraint for sin 2 θ 23 and Δm 2 32 . This is a preliminary combination; a joint fit including correlations of systematic uncertainties and Feldman-Cousin corrections is in progress. The resulting two-dimensional contours using Gaussian limits are presented in Figure 4. The global best fit occurs at NH, δ CP = 1.49π, and sin 2 θ 23 = 0.40. Both octants and hierarchies are allowed at the 1σ level. In inverted hierarchy, there is some rejection of the lower octant for all values of δ CP ; the region around δ CP = π/2 is excluded at the 3σ level.  . Allowed values of δ CP and sin 2 θ 23 from the preliminary combination of ν e appearance and ν μ disappearance data, for both hierarchies.