Searching for top-philic heavy resonances in boosted four-top final states

TL;DR

The paper tackles the search for heavy top-philic resonances through boosted four-top final states at the LHC. It delivers the first complete NLO QCD predictions for $pp\to t\bar{t}X\to t\bar{t}t\bar{t}$ with $X=S_1,S_8$, including off-shell effects, by implementing two simplified models and constructing UFO libraries, using a complex-mass scheme and tailored loop handling to ensure pole cancellation. A validated detector-level framework with state-of-the-art top-tagging reconstructs all boosted top quarks, enabling a shape-based bump hunt in the resonance mass distribution and dedicated signal regions for octet and singlet mediators. The results show HL-LHC sensitivity reaching $M_{S_8}\approx 2-2.5$ TeV (colour octet) and $M_{S_1}\approx 1-1.5$ TeV (colour singlet) for couplings in the $0.1-1$ range, underscoring the importance of NLO modelling and boosted-object techniques for multi-top BSM searches.

Abstract

New heavy resonances with sizeable couplings to top quarks can be probed through searches for beyond-the-Standard-Model effects in four-top production at the LHC. In this work, we present the first next-to-leading-order QCD predictions for the full on-shell and off-shell production of four-top events via new electroweak singlet states, along with dedicated analysis strategies based on the reconstruction and tagging of all final-state top quarks. We develop a detector-level simulation incorporating recent advances in top-tagging and boosted object reconstruction. Moreover, we demonstrate that searches at LHC Run 3 and high-luminosity phase in the zero-lepton, one-lepton and same-sign di-lepton channels can improve the sensitivity to the new physics cross sections by up to two orders of magnitude. In particular, colour-octet resonances with masses up to 2-2.5 TeV and colour-singlet states with masses up to 1-1.5 TeV are within reach for coupling values in the 0.1-1 range.

Figures (11)

• Figure 1: Representative Feynman diagrams contributing to four-top production in a simplified model with an additional scalar octet field $S_8$. When the scalar particles are on-shell, the associated production of the scalar octet with a top-antitop pair (left) is proportional to the square of the Yukawa coupling from the second Lagrangian of eq. \ref{['eq:lags']}, while the QCD-driven pair production of two scalar octets (right) depends only on the octet mass $M_{S_8}$ after assuming that $\mathrm{Br}(S_8 \to t\bar{t}) = 1$.
• Figure 2: Comparison of LO and NLO cross sections as functions of the BSM resonance mass. Results are shown for the scalar octet ( left) and singlet ( right) cases, assuming couplings to top quarks of 0.5 ( top) and 3 ( bottom) . The lower panels of the figures display the associated $K$-factors defined as the ratio of the NLO cross section to the LO prediction computed with NLO PDFs.
• Figure 3: Minimum $\Delta R$ separation between (parton-level) top quarks, computed for events that pass at least one signal region selection after reconstruction. Events are categorised according to the number of leptons produced in the top quark decays.
• Figure 4: LO (dashed orange) and NLO (solid green) distribution of the largest reconstructed resonance mass for the colour-octet benchmark points BP1 (left) and BP2 (right), for the SR1 signal region with conservative top-tagging performance. The predictions are normalised to 1, the hatched bands represent the NLO scale variation and statistical uncertainties and the last bin includes the overflow.
• Figure 5: LO (dashed orange) and NLO (solid green) distributions of the reconstructed resonance mass for the colour-singlet benchmark points BP3 (left) and BP4 (right), for the SR1 signal region with conservative top-tagging performance. The predicted differential cross sections $\mathrm{d}\sigma$ are normalised to 1, the hatched bands represent the NLO scale variation and statistical uncertainties and the last bin includes the overflow.