Probing electroweak pair production of heavy neutral leptons with displaced vertices at the LHC

We study the sensitivity of displaced vertex searches at the LHC to heavy neutral leptons (also known as sterile neutrinos) that are produced in pairs with an electroweak-size cross section. We work within the context of a supersymmetric model in which the sterile neutrino is produced along with Standard Model particles in higgsino decays. By making use of model-independent reconstruction efficiencies provided by the ATLAS collaboration in their search for displaced vertices with multiple jets, we obtain constraints on this model from $139$ fb$^{-1}$ of data collected by ATLAS during the LHC Run2, and assess the discovery reach of Run3 and of the high-luminosity LHC (HL-LHC). Depending on the higgsino mass parameter, sterile neutrino masses between $20~\mathrm{GeV}$ and $230~\mathrm{GeV}$ and active-sterile neutrino mixings in the range $4 \times 10^{-14} \lesssim V^2_N \lesssim 3 \times 10^{-10}$ can be excluded. At the HL-LHC, discovery-level significances could be reached for sterile neutrinos masses up to $295~\mathrm{GeV}$ and values of $V^2_N$ down to $3 \times 10^{-14}$. Finally, moving away from the supersymmetric scenario, we study to which extent these results can be generalized to a broader class of models in which the sterile neutrinos are produced in the decays of heavier particles that are themselves pair-produced with an electroweak-size cross section.

💡 Research Summary

This paper investigates the sensitivity of displaced‑vertex (DV) searches at the LHC to heavy neutral leptons (HNLs), also known as sterile neutrinos, that are produced in pairs via an electroweak‑size cross‑section. In the standard HNL paradigm, production proceeds through off‑shell W or Z bosons (pp → W* → ℓ N or pp → Z* → ν N) and is suppressed by the square of the active‑sterile mixing |V_{Nα}|², limiting reach to relatively large mixings (≈10⁻⁸–10⁻⁹). The authors consider an alternative scenario where the sterile neutrino N is generated from the decay of a heavier beyond‑Standard‑Model particle ψ (ψ → N + SM). If ψ is pair‑produced through electroweak interactions (σ ∼ O(10 fb)–O(1 pb)), the production rate of N is essentially independent of the tiny mixing angle, while the decay of N remains mixing‑suppressed, giving rise to macroscopic decay lengths.

A concrete realization is provided by a supersymmetric model with R‑parity violation. In this setup the only light superpartners are the higgsino‑like electroweakinos (˜χ⁰₁, ˜χ⁰₂, ˜χ⁺₁) with a common mass μ ≳ M_Z. All gauginos are heavy (M₁, M₂ ≫ μ). The higgsinos decay almost exclusively (≈100 % branching ratio) into a sterile neutrino and a SM gauge boson: ˜χ⁺₁ → W⁺ N and ˜χ⁰₁,₂ → Z N. Consequently, pp collisions produce pairs of sterile neutrinos together with two gauge bosons (e.g. pp → ˜χ⁰₁ ˜χ⁰₂ → Z Z N N). The sterile neutrino mass m_N is taken as a free parameter in the range 20–300 GeV, while the active‑sterile mixing angles V_{Nα} are dictated by the seesaw‑like relation V_N ∼ √(m_ν/m_N) ≈ 10⁻⁶–10⁻⁷, leading to displaced decays into ℓℓν, ℓ qq′, ν qq, or ν ν ν final states.

To assess experimental sensitivity, the authors recast the ATLAS multijet + DV search (Run 2, 139 fb⁻¹) which provides model‑independent reconstruction efficiencies ε(r, p_T, N_{trk}) for displaced vertices. They generate signal events with MadGraph5_aMC@NLO, shower them with Pythia 8, and simulate detector response using Delphes, applying the ATLAS efficiency maps to obtain realistic selection efficiencies. Backgrounds are assumed negligible, allowing a simple Poisson 95 % CL limit (zero observed events). Scanning over μ, m_N, and the total mixing V_N², they find that for higgsino masses μ ≲ 200 GeV the current data exclude V_N² in the range 4 × 10⁻¹⁴ – 3 × 10⁻¹⁰ for sterile neutrino masses between 20 GeV and 230 GeV. Projecting to Run 3 (300 fb⁻¹, √s = 13.6 TeV) and the High‑Luminosity LHC (3 ab⁻¹, √s = 14 TeV) under the same background‑free assumption, the discovery reach extends to m_N ≈ 295 GeV and V_N² down to ≈3 × 10⁻¹⁴.

A theoretical parameter z (complex angle) controls the size of the mixing: the “minimal mixing” case (z = π/2) yields V_min ≈ m_2/m_N, while the “maximal real mixing” case (z = 0) gives V_max ≈ m_3/m_N. For complex z, V_N can be exponentially larger (∝ e^{|Im z|}), but cannot be smaller than V_min. The analysis therefore provides both a lower bound (model‑independent) and an upper bound (model‑dependent) on the mixing that can be probed.

Finally, the authors argue that the conclusions are not limited to the specific SUSY construction. Any BSM particle ψ that is pair‑produced via electroweak interactions and decays almost exclusively to N + SM will lead to the same phenomenology. Examples include heavy W′/Z′ bosons, vector‑like leptons, or scalar doublets. The key ingredients are (i) an electroweak‑size production cross‑section for ψ, (ii) a near‑100 % branching ratio ψ → N + SM, and (iii) a displaced decay of N governed by tiny active‑sterile mixing. By leveraging the ATLAS model‑independent DV efficiencies, the study offers a broadly applicable framework for future LLP searches targeting sterile neutrinos produced in electroweak pair production.