Dark Neutrino Portal to Explain MiniBooNE excess

We present a novel framework that provides an explanation to the long-standing excess of electron-like events in the MiniBooNE experiment at Fermilab. We suggest a new dark sector containing a dark neutrino and a dark gauge boson, both with masses between a few tens and a few hundreds of MeV. Dark neutrinos are produced via neutrino-nucleus scattering, followed by their decay to the dark gauge boson, which in turn gives rise to electron-like events. This mechanism provides an excellent fit to MiniBooNE energy spectra and angular distributions.

Introduction.-Neutrinos have been connected to anomalies in experimental data since their commencement in the realm of Physics. From the problems with beta decays in the dawn of the XX th century, that culminated with the proposal and subsequent discovery of the first of these remarkable particles, to the solar and atmospheric neutrino puzzles, that revealed the phenomenon of neutrino oscillations driven by masses and mixings, the neutrino road has been full of surprises. Some, however, like the 17-keV neutrino [1] or the superluminal neutrinos [2] turned out to be mere bumps on the road as they were resolved by explanations unrelated to new physics. As it happens, one never knows which small clouds hovering on the horizon of Physics will eventually vanish and which will instead ignite a revolution.
Even today some peculiar data anomalies remain unsolved. On one hand, there is an apparent deficit of ν e in short-baseline reactor experiments [3] and of ν e in radioactive-source experiments [4], both amounting to a 2.5-3σ discrepancy that many believe may be connected to unknown nuclear physics. On the other hand, the LSND [5] and MiniBooNE neutrino experiments [6][7][8][9] have reported an excess of ν e and ν e charge-current quasi-elastic (CCQE) events in their data. All these conundrums have been offered a number of exotic inter-pretations in the literature [10][11][12][13][14], typically invoking eV sterile neutrinos in schemes easily in tension with other neutrino data [15][16][17].
Recently, after 15 years of running, MiniBooNE updated their analysis revealing that the excess of electronlike events in the experiment [18], consistently observed in the neutrino and antineutrino modes, is now a 4.8σ effect. That makes the MiniBooNE result the most statistically relevant anomaly in the neutrino sector. The origin of such excess is unclear -it could be the presence of new physics, or a large background mismodeling. In this Letter we propose a phenomenological solution to understand the MiniBooNE data [19].
Framework.-We introduce a dark sector composed by a new vector boson, Z D , coupling directly solely to a dark neutrino, ν D , which mixes with the standard ones as where ν i and ν α are the neutrinos mass and flavor eigenstates, respectively. The new vector boson will, in general, communicate with the Standard Model (SM) sector via either mass mixing or kinetic mixing. The relevant part of the dark Lagrangian is where m Z D is the mass of Z D and g D is the coupling in the dark sector, e is the electromagnetic coupling, g/c W is the Z coupling in the SM, while and parametrize * E-mail:bertuzzo@if.usp.br † E-mail:sudip.jana@okstate.edu ‡ E-mail:pmachado@fnal.gov § E-mail:zukanov@if.usp.br the kinetic and mass mixings, respectively. The electromagnetic and Z currents are denoted by J em µ and J Z µ . For simplicity, we assume the mass mixing between the Z and the Z D boson to be negligible. We resort to kinetic mixing between B µν and B µν [20], the SM hypercharge and the dark field strengths, as a way to achieve a naturally small coupling between the Z D and the electromagnetic current J em µ . We will take m N D > m Z D , so the dark neutrino can decay as N D → Z D + ν i , and m Z D < 2 m µ so arXiv:1807.09877v1 [hep-ph] 25 Jul 2018 the Z D can only decay to electrons and light neutrinos.
The dark neutrino decay width into Z D + ν s is simply while the Z D decay width into e + e − and light neutrinos are, respectively, and We observe that as long as We want both N D and Z D to decay promptly. Taking the typical energy E N D , E Z D ∼ 1 GeV, and as- MeV would guarantee prompt decay for both particles. We will see shortly that m N D and m Z D between a few tens to a few hundred of MeV is exactly what is needed to explain the experimental data.
Analysis and results.-The MiniBooNE experiment is a pure mineral oil (CH 2 ) detector located at the Booster Neutrino Beam line at Fermilab. The Cherenkov and scintillation light emitted by charged particles traversing the detector are used for particle identification and neutrino energy reconstruction, assuming the kinematics of CCQE scattering. MiniBooNE has observed an excess of 381 ± 85.2 (79.3 ± 28.6) electron-like events over the estimated background in neutrino (antineutrino) beam configuration in the energy range 200 < E rec ν /MeV < 1250 corresponding to 12.84 × 10 20 (11.27 × 10 20 ) protons on target [18].
Our proposal to explain MiniBooNE's low energy excess from the production and decay of a dark neutrino relies on the fact that MiniBooNE cannot distinguish a collimated e + e − pair from a single electron. Muon neutrinos produced in the beam would up-scatter on the mineral oil to dark neutrinos, which will subsequently lead to Z D → e + e − as shown schematically in Fig. 1. If N D is light enough, this up-scattering in CH 2 can be coherent, enhancing the cross section. To take that into account, we estimate the up-scattering cross section to be where F (E r ) is the nuclear form factor [21] for Carbon, while σ coh C and σ p are the elastic scattering cross sections on Carbon and protons, which can be easily calculated. For Carbon, F (E r ) is sizable up to proton recoil energies of few MeV.
To obtain the spectrum of events, a simplified model was implemented in FeynRules [22] in which Carbon and protons were taken to be an elementary fermion and events were generated in MadGraph5 [23]. Since Mini-BooNE would interpret Z D → e + e − decays as electronlike events, the reconstructed neutrino energy would be incorrectly inferred by the approximate CCQE formula (see e.g. Ref. [24]) where m p is the proton mass, and E Z D and θ Z D are the dark Z D boson energy and its direction relative to the beam line. The fit to MiniBooNE data was then performed using the χ 2 function from the collaboration official data release [18], which includes the ν µ andν µ disappearance data, re-weighting the Montecarlo events by the ratio of our cross section to the standard CCQE one, and taking into account the wrong sign contamination from Ref. [25]. Note that the official covariance matrix includes spectral data in electron-like and muonlike events for both neutrino and antineutrino modes. In Fig. 2 we can see the electron-like event distributions, including all of the backgrounds, as reported by MiniBooNE. We clearly see the event excess reflected in all of them. The neutrino (antineutrino) mode data as a function of E rec  an approximated systematic uncertainty from the background estimated from Table I of Ref. [18]. On the bottom panel we show the cos θ distribution of the electronlike candidates for the neutrino data, as well as the distribution for cos θ Z D for the benchmark point (blue line). The cos θ distribution of the electron-like candidates in the antineutrino data is similar and not shown here and our model is able to describe it comparably well. We remark that our model prediction is in extremely good agreement with the experimental data. In particular, our fit to the data is better than the fit under the electron-Volt sterile neutrino oscillation hypothesis [18] if one considers the constraints from other oscillation experiments. We find a best fit with χ 2 bf /dof = 31.2/36, while the background only hypothesis yields χ 2 bg /dof = 63.8/38, corresponding to a 5.4σ preference for our model.
In Fig. 3 we see the region in the plane |U µ4 | 2 versus m N D consistent with MiniBooNE data at 1σ to 5σ CL, for the exemplifying hypothesis m Z D = m N D /5, α Z D = 0.25 and α 2 = 3×10 −9 . Other values of these pa-rameters can also provide good agreement with the data. We also show the combined non-oscillation bounds from meson decays, muon decay Michel spectrum and lepton universality compiled in Refs. [26,27], which exclude the region above the red line. The dashed gray lines represent γcτ = 1 cm for N D and Z D with 1 GeV of energy, as a reference. The ship hull shape region can be divided in two parts: a high mixing region at |U µ4 | 2 ∼ 10 −3 − 10 −6 , corresponding to m N D 300 MeV, and a low mixing region for |U µ4 | 2 10 −7 and m N D 200 MeV. The latter seems to be favored by spectral data. As a side remark, we have checked that the typical opening angle θ e + e − of the e + e − pair satisfy cos θ e + e − = 0.99, ensuring that MiniBooNE will identify these events as electron-like.
The MicroBooNE experiment at Fermilab [28] is currently investigating the low energy excess of electron-like events observed by MiniBooNE. They can distinguish electrons from photon conversions into a e + e − pair by their different ionization rate at the beginning of their trajectory in the liquid argon detector. So by analyzing the energy deposited along the track as a function of the range (dE/dX) they hope to distinguish a photon from a single electron. Our model predicts a dE/dX distribution similar to photons but with a prompt Z D decay to a collimated e + e − pair. In addition our framework allows for the possibility of the experimental observation of the We also have inquired into the possible effects of N D and Z D on oscillation experiments. While low energy sources, such as the sun or nuclear reactors, do not have enough energy to produce these particles, they could be, in principle, produced in higher energy oscillation experiments. Typically ν µ and ν µ beams in accelerator neutrino experiments have an insurmountable O(1%) contamination of ν e + ν e , and atmospheric neutrinos have a large ν e and ν e component. While Cherenkov detectors, like Super-Kamiokande, cannot distinguish between electrons and photons, detectors like MINOS, NOνA or T2K would have a hard time to see any signal over their neutral current contamination. That is particularly relevant at lower energies where one would expect the signal of new physics to lay.
In a different note, we do not foresee any issues with cosmological data, as the particles in the dark sector decay too fast to affect Big Bang Nucleosynthesis, and the ν − ν self-interactions are too small to change neutrino free streaming. Supernova cooling would not constrain the model, as the Z D is trapped due to the large kinetic mixing.
Finally, one may wonder if the phenomenological approach we propose here can arise in a UV-complete anomaly free model. We have checked that such realization is possible as follows. A gauge U (1) D symmetry, under which the only charged fermions are the dark neutrinos, protects neutrino masses from the standard Higgs mechanism. An enlarged scalar sector is called Region of our model in the |Uµ4| 2 versus mN D plane satisfying MiniBooNE data at 1σ to 5σ CL, for the hypothesis mZ D = mN D /5, αZ D = 0.25 and α 2 = 3 × 10 −9 . The region above the red curve is excluded at 99% CL by meson decays, the muon decay Michel spectrum and lepton universality [26,27].
upon to ensure non-zero neutrino masses, naturally leading to ν − N D mixing, as well as the mass of the dark gauge boson. In this realization, both kinetic and mass mixing are unavoidable, but typically small. The model naturally connects neutrino masses with the new interaction. We will explore the rich phenomenology of this model in detail elsewhere.
Conclusion.-We have shown that the low energy excess observed by MiniBooNE can by explained by a light dark sector to which neutrinos are a portal. The framework is elegant and no tuning is need to fit the excess. We find an excellent agreement with spectral and angular data distributions, in both neutrino and antineutrino modes. This solution is consistent with all current experimental data and can be probed by Liquid Argon detectors in the near future.