Constraining $A\to ZH$ with $H\to t\bar t$ in the Low-Mass Region

TL;DR

Addressing the search for a heavy pseudoscalar $A$ decaying to $Z$ with $H\to t\bar t$ in the low-mass region, the authors recast differential $t\bar t Z$ measurements from ATLAS and CMS to constrain $\sigma(A\to ZH)\times\mathrm{Br}(H\to tt)$. They perform a global $\chi^2$ fit to binned differential observables and obtain 95% CL upper limits in the range $0.12$–$0.62$ pb across the $(m_A,m_H)$ plane, with a mild $\sim 2.5\sigma$ preference for a nonzero signal near $(m_A,m_H)\approx(450$–$460,290)$ GeV and a best-fit value of $\approx 0.3$ pb. Interpreting these results in a top-philic 2HDM favors a top-Yukawa rescaling $\mu_t\gtrsim 0.16$ with $\sin\alpha\approx 0$ (alignment) and shows consistency with perturbative unitarity and EW constraints. The work demonstrates the power of reinterpreting $t\bar t Z$ data to probe $A\to ZH$ in the low-mass regime and motivates future high-luminosity LHC studies and dedicated $A\to ZH$ searches to test the top-philic Higgs scenario.

Abstract

The decay $A\to ZH$ is a characteristic signal of two-Higgs-doublet models (2HDMs), where $A$ and $H$ lie primarily within the same $SU(2)_L$ multiplet, leading to a coupling of order $g_2$ to the $Z$ boson. The subsequent decay $H\to tt^{(*)}$ is particularly promising, as it gives rise to distinct final states involving multiple leptons and $b$-jets. The required splitting between $m_A$ and $m_H$ can naturally occur near the electroweak scale while being consistent with perturbative unitarity. Whereas dedicated ATLAS and CMS searches focused on the region with both top-quarks on-shell, we cover lower masses where one top quark is off-shell by recasting Standard Model $t\bar{t}Z$ measurements of ATLAS and CMS. The obtained limits on $σ(A\to ZH)\times {\rm Br} (H\to t\bar t)$ are between $0.12$ pb and $0.62$ pb. Interestingly, we observe these stringent limits despite a preference (up to $2.5σ$) for a non-zero new physics signal, most pronounced around for $m_A \approx 450-460$ GeV and $m_H\approx 290$ GeV, with a best-fit value of $σ(A \to ZH) \times {\rm Br}(H \to t\bar t) \approx 0.3$ pb. This cross section can be accommodated within a top-philic 2HDM for a top-Yukawa coupling of the second Higgs doublet of $μ_t \gtrsim 0.16$.

Figures (8)

• Figure 1: Feynman diagram depicting the process $pp\to A\to ZH$ with $H\to t\bar{t}$, leading to a $t\bar{t}Z$-like signature.
• Figure 2: Points allowed by EW precision data and perturbative unitarity for different upper limits on the tree-level scattering amplitudes: $<16\pi$ (red), $<4\pi$ (yellow) and $<2\pi$ (green). The search region from ATLAS, as well as the one covered in this work, is hatched. One can see that while only part of the ATLAS region contains viable points, our region is fully populated and closes the gap for $A\to ZH$ with $H\to t\bar{t}$ searches.
• Figure 3: 95% CL upper limit (left) and best-fit (right) value of $\sigma(A \to ZH) \times {\rm Br}(H \to t\bar{t})$ in units of pb in the $m_A$--$m_H$ plane with $m_A-m_H \geq 100$ GeV, obtained by combining the CMS and ATLAS analyses. The color bar in the left (right) plot indicates the 95% CL upper limit for the cross section (the preference for a non-zero NP signal in units of standard deviation). An extended region with $m_A< 800\,\mathrm{GeV}$ is provided in Fig. \ref{['fig:extended']} in the Appendix.
• Figure 4: Left: Regions in the $m_A$–$m_H$ plane for the top-philic 2HDM, assuming $\sin\alpha = 0$, that are excluded by our analysis for different values of the top Yukawa rescaling parameter $\mu_t$. Right: Corresponding exclusion in the aligned 2HDM realization with $\xi_u\neq 0$, i.e., both top and charm couplings.
• Figure 5: Preferred $1\sigma$ and $2\sigma$ regions within the top-philic 2HDM for the best-fit point $(m_A, m_H) = (460, 290)~\mathrm{GeV}$, where a $2.5\sigma$ preference is observed. The purple curve indicates the best-fit line for the aligned 2HDM with $\xi_u \neq 0$, $\xi_d = \xi_\ell = 0$. The regions above the red and green lines are excluded by the ATLAS and CMS searches for $A \to t\bar{t}$ATLAS:2024vxmCMS:2025dzq.