Search for Lepton Flavor Violating Signals at the Future Electron-Proton Colliders

TL;DR

This work investigates lepton flavor violation in Z boson decays to $e$ and $\mu$ using a model‑independent EFT framework, focusing on vector, axial‑vector, tensor, and axial‑tensor couplings. The authors assess the discovery potential at future electron–proton colliders (LHeC and FCC‑eh) via the $\mu+j$ final state, employing a multivariate XGBoost analysis to separate signal from Standard Model backgrounds. They project 95% CL upper bounds on BR$(Z\to e\mu)$ that improve upon the current ATLAS limit, with tensor‑type couplings achieving the strongest sensitivity, potentially reaching BR values as low as $\mathcal{O}(10^{-8})$–$\mathcal{O}(10^{-9})$ at FCC‑eh. The results highlight the complementarity of future $e$‑p colliders to hadron and lepton facilities in probing LFV Z decays, while acknowledging that systematic uncertainties are not exhaustively treated and the estimates are indicative.

Abstract

The search for lepton flavor violation (LFV) is a powerful probe to look for new physics beyond the Standard Model. We explored the possibility of searches for LFV $Z$ boson couplings to electron and muon pairs at the upcoming electron-proton colliders, namely the Large Hadron Electron Collider (LHeC) and the Future Circular lepton-hadron Collider (FCC-eh). We employed the study via a single muon plus an associated jet channel to search for the LFV signal. We used a multivariate technique to obtain an improved signal-background analysis. By using the condition on nonobservation of any significant deviation of the signal over the expected background, we provide an upper limit on the LFV $Z$ boson coupling and corresponding branching ratio (BR). We find that an upper limit of up to $1.13\times 10^{-7}$ and $4.64 \times 10^{-8}$ can be set on BR($Z\to e μ$) at 95\% confidence level (C.L.) with one year run of LHeC and FCC-eh, respectively, if the LFV coupling is governed by vector or axial-vector coupling. For tensor or axial-tensor coupling, these limits can be improved up to $2.34\times 10^{-8}$ and $5.02\times 10^{-9}$ for LHeC and FCC-eh machines, respectively, at 95\% C.L. The projected numbers improve significantly over the existing limit of $2.62\times 10^{-7}$ set by ATLAS.

Figures (9)

• Figure 1: Feynman diagram for the signal process $e^- p \to \mu^- j$.
• Figure 2: Representative Feynman diagrams for (a) the photoproduction process and (b) other background processes with $V = W^\pm, \gamma, Z$.
• Figure 3: Normalized distributions of ${p_T}_{\mu_1}$ for both signal and backgrounds for (a) LHeC1 and (b) FCC-eh1. Normalized distribution of ${p_T}_{j_1}$ for both signal and backgrounds for (c) LHeC1 and (d) FCC-eh1. For vector-like coupling, we use $g_v=3.68\times 10^{-4}$ and for tensor-like coupling $g_t=3.00\times 10^{-6}$ GeV$^{-1}$.
• Figure 4: Normalized distribution of $\slashed{E}_T$ for both signal and backgrounds for (a) LHeC1 and (b) FCC-eh1. Normalized distributions are shown for $\Delta R_{\mu_1 j_{1}}$ for both signal and backgrounds for (c) LHeC1 and (d) FCC-eh1. The couplings have been taken to be the same as Fig. \ref{['fig:distribution1']}.
• Figure 5: Normalized distribution of $(E-p_z)_{\mu_1}$ for both signal and backgrounds for (a) LHeC1 and (b) FCC-eh1. The same coupling values as Fig. \ref{['fig:distribution1']} have been used.