Up-scattering production of a sterile fermion at DUNE: complementarity with spallation source and direct detection experiments

The paper addresses whether MeV-scale sterile fermions can be produced via neutrino up-scattering on electrons or nuclei and how different experiments can probe this scenario. Motivated by neutrino oscillation evidence and theories predicting sterile states, the authors consider generalized non-derivative interactions between neutrinos and electrons/quarks mediated by light bosons, enabling inelastic transitions ν → χ (a sterile fermion). They focus on experimental complementarity: DUNE Near Detector (DUNE-ND) using neutrino–electron elastic scattering with on-/off-axis configurations and beam options; dark matter direct detection experiments (XENONnT, LZ) via solar-neutrino–electron scattering; and COHERENT via coherent elastic neutrino–nucleus scattering (CEνNS). The goal is to map sensitivities across mediator and sterile-fermion masses, including potential signatures from sterile-fermion decays, and to show how different neutrino sources and kinematics probe distinct parameter regions.

The study builds on extensive work on heavy neutral leptons and light new states probed at oscillation, CEνNS, and dark matter detectors. Prior analyses of up-scattering into sterile fermions have typically considered limited mediator types (e.g., scalar only) or integrated-out (effective) interactions, and often used earlier COHERENT datasets. Recent efforts combined CEνNS and ν–e scattering with light mediators but usually restricted to vector/scalar couplings. The authors extend these by including all Lorentz-invariant non-derivative interactions (S, P, V, A, T) with explicit light-mediator propagators. They also compare to the dipole portal scenario, previously studied at DUNE, COHERENT, and DM detectors, noting their framework generalizes beyond electromagnetic-type interactions. The work incorporates updated datasets (COHERENT CsI-2021; XENONnT; LZ) and projects DUNE-ND sensitivities, emphasizing complementarity across facilities and energy regimes.

Theory and cross sections: The authors introduce a simplified low-energy Lagrangian enabling inelastic νℓ + f → χ + f transitions (f = e, u, d) through generalized interactions: scalar, pseudoscalar, vector, axial-vector, and tensor. They work in a light-mediator regime, replacing effective couplings with propagator factors (m_a^2 + |q|^2) in denominators, and derive differential cross sections for ν–e elastic scattering (EνES) and CEνNS including detector-specific momentum transfers. For ν–e scattering, atomic binding is modeled via an energy-dependent effective electron charge Z_eff(Te). For CEνNS, they map quark-level to nuclear couplings (C_a), include nuclear form factors (Klein–Nystrand) and spin structure functions (longitudinal/transverse) for axial and tensor cases using shell-model inputs. Appendices detail hadronic inputs and spin responses. DUNE-ND simulation: They simulate ν–e events in the E_e θ_e^2 plane to suppress dominant backgrounds (CCQE, misidentified π^0). They use LBNF fluxes for on-axis and multiple off-axis (PRISM) positions up to 30 m, considering both CP-optimized and τ-optimized beam configurations. Kinematic thresholds and relations determine integration limits. Event spectra are computed with angular resolution smearing; energy resolution effects are checked and found negligible. Sensitivities are extracted via a binned likelihood (Poisson χ^2) simultaneously fitting on-/off-axis, neutrino/antineutrino modes over 7 years, with nuisance parameters for flux (5%) and background (10%) normalizations and backgrounds modeled from previous studies. XENONnT and LZ analysis: For solar-neutrino–induced ν–e events, they compute predicted event rates using detector efficiencies and resolutions, accounting for pp and 7Be flux components and oscillation probabilities. For LZ, they use a Poisson likelihood across 51 energy bins with nuisance parameters for backgrounds and flux normalizations; for XENONnT, a Gaussian χ^2 with modeled background spectrum and flux uncertainties is used. Detector response functions are taken from the respective collaborations’ publications. COHERENT CsI-2021 analysis: They compute CEνNS and ν–e event rates for prompt and delayed SNS neutrinos, fold in detector efficiency, energy resolution, quenching factor and light yield to obtain PE spectra, and include timing information by distributing events over time bins using provided time profiles and efficiencies. A comprehensive Poisson likelihood fits energy-time binned data including CEνNS, ν–e, and backgrounds (beam-related neutrons, neutrino-induced neutrons, steady-state), with multiple nuisance parameters for flux/efficiency (11%), backgrounds (25–35%, 2.1%), quenching (3.8%), nuclear form factor (5%), timing, and CEνNS efficiency variations. Sterile-fermion decays: For vector-mediated interactions, they estimate order-of-magnitude decay sensitivities from χ decays inside detectors, considering two-body χ → V ν and three-body χ → ν ℓ+ℓ− (off-shell mediator) modes. Decay widths are taken from prior work; mediator dilepton decays are included where kinematically allowed. They approximate forward kinematics, prompt mediator decays, and background-free sensitivity (N_decay ≥ 2.3 per year), using detector lengths for decay probabilities and simple efficiency assumptions. DM DD experiments are not expected to see χ-decay electron signals due to kinematics; COHERENT and DUNE-ND can in certain mass ranges.

- DUNE-ND projected sensitivities (7 years, combined on-/off-axis, ν and anti-ν): On-axis dominates; off-axis reduces both signal and background with minor impact on overall reach. For m_χ of order few to tens of MeV, the smallest couplings reachable are approximately g_S/P ~ 1e-4, g_V/A ~ 2e-5, and g_T ~ 1e-5. Tensor interactions are most strongly constrained; scalar/pseudoscalar are weakest due to kinematic suppression at leading order relative to V/A and T. - Beam configuration: The τ-optimized beam slightly improves sensitivity at larger mediator and sterile masses; CP-optimized can be better at low masses (notably for S/P). Overall differences are modest; conclusions on complementarity remain unchanged. - Complementarity across experiments: XENONnT and LZ provide the strongest limits at very small mediator and sterile-fermion masses due to low thresholds and recoil-energy scaling of cross sections. COHERENT CsI-2021 becomes more constraining for m_χ ≳ 300 keV up to several MeV–tens of MeV, especially where CEνNS contributes (scalar) or ν–e events dominate for spin-dependent cases. For mediator/sterile masses beyond about a few MeV (∼2 MeV and up), DUNE-ND substantially improves sensitivities and covers regions consistent with cosmology, particularly for pseudoscalar, axial-vector, and tensor interactions where CEνNS is spin-suppressed. - Kinematic mass reach: DUNE-ND can probe χ masses up to ≈200 MeV (from scattering/decay analyses), while COHERENT sensitivity drops around ≈52 MeV, set by the SNS neutrino endpoint. Many low-mass regions probed by DM DD and some COHERENT regions may face cosmological constraints (e.g., BBN), highlighted as a cautionary vertical line in figures. - Sterile-fermion decays: Decay-based searches at DUNE-ND and COHERENT can provide additional sensitivity. DUNE-ND covers m_χ up to ∼200 MeV; COHERENT up to ∼52 MeV. DM DD experiments do not expect extra electron signals from χ decays due to m_χ < 2 m_e in their kinematic range.

The results demonstrate that different experimental setups with distinct neutrino sources and energy spectra probe complementary regions of the ν → χ up-scattering parameter space. Low-threshold DM detectors using solar neutrinos are optimal at very low mediator and sterile masses. CEνNS at spallation sources extends reach to higher m_χ, with scalar interactions benefitting most from coherent enhancement; for spin-dependent interactions, ν–e contributions dominate. DUNE-ND, with GeV-scale fluxes and high statistics, is most powerful at higher mediator and sterile masses, extending coverage into regions not excluded by typical cosmological bounds. The on-axis DUNE-ND configuration yields the best sensitivity; off-axis operation provides cross-checks and background mitigation with limited sensitivity gains. Beam configuration choices (τ-optimized) can modestly extend mass reach. Decay signatures can further enhance coverage at DUNE and COHERENT, consistent with kinematic limits from the neutrino sources. Overall, the combined landscape provides robust and complementary probes of generalized interactions leading to sterile-fermion production.

The study presents a comprehensive, model-agnostic framework for sterile fermion production via generalized non-derivative interactions and quantifies sensitivities across DUNE-ND, COHERENT, XENONnT, and LZ. DUNE-ND can reach couplings down to ~1e-4 (S/P), 2e-5 (V/A), 1e-5 (T) for m_χ in the few–tens of MeV range after 7 years, with on-axis operation providing the strongest reach; τ-optimized beams slightly extend the mass reach. Current XENONnT/LZ data dominate constraints at very low masses, COHERENT constrains intermediate masses (especially for scalar interactions), and DUNE-ND will lead at higher masses up to ~200 MeV. Decay channels of χ can provide additional sensitivity at DUNE and COHERENT but are irrelevant for current DM DD kinematics. The work highlights strong complementarity across facilities and motivates further theoretical development of UV-complete models and more detailed experimental studies, including refined background modeling and decay-signature analyses.

- Cosmological and astrophysical constraints (e.g., BBN) potentially exclude parts of the low-mass parameter space, but a full treatment is model-dependent and not performed. - Nuclear physics inputs (form factors, spin structure functions) carry uncertainties (typically ~5–10%), incorporated via nuisance parameters but still a source of systematic error. - Simplified phenomenological framework without a specific UV completion; assumes flavor-universal couplings and light mediators with explicit propagators. - DUNE-ND simulation uses approximate background models and assumes certain thresholds; off-axis background shape variations are noted but treated as small. - Decay sensitivity estimates are order-of-magnitude: assume forward decays, prompt mediator decay, background-free requirement (N ≥ 2.3/yr), simplified efficiencies, and neglect of mesonic channels at DUNE except in marginal regions. - COHERENT analysis focuses on CsI-2021 dataset; the LAr dataset is not included due to lower statistics at present.