Search for Higgs boson exotic decays into Lorentz-boosted light bosons in the four-$τ$ final state at $\sqrt{s}=13$ TeV with the ATLAS detector

TL;DR

This study searches for exotic Higgs decays $H\rightarrow aa\rightarrow \tau^+\tau^-\tau^+\tau^-$ in the mass range $4\,\mathrm{GeV}<m_a<15\,\mathrm{GeV}$ using $140\,\mathrm{fb}^{-1}$ of ATLAS Run 2 data at $\sqrt{s}=13$ TeV. The analysis focuses on Lorentz-boosted $a\rightarrow\tau^+\tau^-$ decays with one $\tau$ decaying hadronically and the other to a muon, employing a dedicated muon-removal technique to reconstruct overlapped di-\tau systems and two signal regions (SS and OS) based on muon charge. Backgrounds with fake $\tau_{\text{had}}$ are estimated data-driven via tight-to-loose factors measured in $Z$+jets events, while genuine $ZZ$ and related SM backgrounds are modeled by simulation. No significant excess is observed, and upper limits at 95% CL are set on $(\sigma(H)/\sigma_{\mathrm{SM}}(H))\times\mathcal{B}(H\to aa\to 4\tau)$ between 0.03 and 0.10, depending on $m_a$, providing stringent constraints on the Type-III 2HDM+S with $\tan\beta=5$. These results advance the exploration of Higgs portals to light hidden sectors and complement other exotic decay searches at the LHC.

Abstract

A search for exotic decays of the Higgs boson into a pair of low-mass scalars that subsequently decay into $τ$-leptons, $H\rightarrow aa\rightarrow τ^+τ^-τ^+τ^-$, is presented. In models with Yukawa-like couplings, the decay to $τ$-leptons is favoured for light $a$-bosons, with mass in the range of $2m_τ < m_a < 2m_{b}$. Results are presented in the range of $4\,\mathrm{GeV} < m_a < 15\,\mathrm{GeV} $ using the $140\,\mathrm{fb}^{-1}$ of proton-proton collisions at $\sqrt{s}=13$ TeV recorded with the ATLAS detector during Run 2 of the Large Hadron Collider. This search focuses on di-$τ$ pairs where one of the $τ$-leptons decays to hadrons and neutrinos, and the other decays to a muon and neutrinos. In this mass range, the $a\rightarrow τ^+τ^-$ is Lorentz-boosted and a dedicated muon removal technique is used to reconstruct the di-$τ$ pairs. No significant excess above the Standard Model background prediction is observed. Upper limits on $(σ(H)/σ_{\mathrm{SM}}(H))\times \mathcal{B}(H\rightarrow aa\rightarrow 4τ)$ at $95\%$ confidence level are provided, ranging from $0.03$ to $0.10$ depending on the $a$-boson mass.

Figures (5)

• Figure 1: Monte Carlo estimates of the $\Pgt_{\text{had}}\xspace$ reconstruction and RNN identification efficiency (\ref{['fig:RecoIDEff1prongStdTaus']} and \ref{['fig:RecoIDEff3prongStdTaus']}) before and (\ref{['fig:RecoIDEff1prongMRTaus']} and \ref{['fig:RecoIDEff3prongMRTaus']}) after the muon removal for generator-level (\ref{['fig:RecoIDEff1prongStdTaus']} and \ref{['fig:RecoIDEff1prongMRTaus']}) 1-prong and (\ref{['fig:RecoIDEff3prongStdTaus']} and \ref{['fig:RecoIDEff3prongMRTaus']}) 3-prong $\Pgt_{\text{had}}$ candidates for all working points (Very Loose, Loose, Medium, Tight) defined in Ref. ATL-PHYS-PUB-2019-033, as a function of generator-level $\Delta R(\Pgt_{\text{had}}\xspace, \mu)$. The 'Reconstruction (Reco)' markers indicate the efficiency of a $\tau_{\text{seed}}$ jet to match a generator-level $\Pgt_{\text{had}}$; The '$N_{\text{trk}}^{\text{reco}}=$1 or 3' markers show the efficiency of a generator-level $\Pgt_{\text{had}}$ reconstructed with the same number of associated charged-particle tracks as charged hadrons at generator-level. These plots are based on a mixture of $m_a=4, 6, 8, 10, 15$$\text{Ge V}$ samples, with equal weighting applied to each sample. The slight decrease in reconstruction efficiency for $0.2<\Delta R(\Pgt_{\text{had}}\xspace,\mu)<0.4$ is attributed to the $\Pgt_{\text{had}}\xspace$ generator-level matching being less than 100% for the $m_a = 10$$\text{Ge V}$ and $15$$\text{Ge V}$ signal samples, due to the geometrical criterion used.
• Figure 2: Distribution of the average mass of the two $\mu\Pgt_{\text{had}}\xspace$ candidates in each event for data and the expected background in the \ref{['fig:mvismutauSSmuLinearPerbin']} SS$\mu$ and \ref{['fig:mvismutauOSmuLinearPerbin']} OS$\mu$ validation regions. The hashed area represents the total background uncertainty, including both statistical and systematic components. Overflow events up to 15 $\text{Ge V}$ are included in the last bin. The contributions from $\Pq{}\Paq/gg\to ZZ$ and $H\to ZZ^*$ are considered, but they are not visible as they make up for less than 0.1% in every bin.
• Figure 3: Distribution of the average mass of the two $\mu\Pgt_{\text{had}}\xspace$ candidates in each event for data, the expected background and $m_a=4$, 8, and 10 $\text{Ge V}$ signals, assuming $\mathcal{B}(H\to aa\to 4\tau)=100\%$, in the \ref{['fig:BonlyAvMvisSSmuLinearPerbin']} SS$\mu$ and \ref{['fig:BonlyAvMvisOSmuLinearPerbin']} OS$\mu$ signal regions. Overflow events up to 15 $\text{Ge V}$ are included in the last bin. The hashed area represents the total background uncertainty, including both statistical and systematic components. The contributions from $\Pq{}\Paq/gg\to ZZ$ and $H\to ZZ^*$ are considered, but they are not visible as they make up for less than 0.1% in every bin.
• Figure 4: Observed (solid line) and expected (dashed line) 95% CL upper limits on $(\sigma(H)/\sigma_{\text{SM}}(H))\times\mathcal{B}(H\to aa\to 4\tau)$ as a function of $m_a$. The inner green and outer yellow shaded bands represent the $\pm 1$ and $\pm 2$ standard deviations around the expected limits, respectively. Limits between the mass points are interpolated linearly.
• Figure 5: The observed (solid line) and expected (dashed line) limits are interpreted within the framework of the Type-III 2HDM+S model with $\tan\beta=5$, which maximizes the branching ratio of $a \to \tau^+\tau^-$. These limits are expressed in terms of $(\sigma(H)/\sigma_{\text{SM}}(H))\times\mathcal{B}(H\to aa)$ as a function of $m_a$. The observed limits for the $H \to aa \to \mu\mu\mu\mu$ channel are identical to the expected limits, as no events are observed in the corresponding search. The dotted red line represents the scenario where the branching ratio of the Higgs boson exotic decay $H \to aa$ equals 100%, assuming $\sigma(H)=\sigma_{\text{SM}}(H)$. The branching ratios of $a\to\tau^+\tau^-$ are calculated based on the methodology described in Ref. Haisch:2018kqx.