Study of ZZ and ZH production in the bb$\tau\tau$ final state and search for high-mass spin-0 and spin-1 resonances in proton-proton collisions at $\sqrt{s}$ = 13 TeV

Technical Summary: Study of ZZ and ZH Production in the $bb\tau\tau$ Final State and Search for High-Mass Resonances

Problem and Motivation
The discovery of the Higgs boson ($H$) validated the Standard Model (SM), yet critical parameters such as the Higgs self-coupling ($\lambda$) remain unmeasured. The measurement of Higgs boson pair ($HH$) production is the primary avenue for constraining $\lambda$. The $HH \to bb\tau\tau$ decay channel offers high sensitivity but faces significant experimental challenges, including large backgrounds and complex reconstruction. To refine analysis strategies for the High-Luminosity LHC (HL-LHC) and improve the robustness of $HH \to bb\tau\tau$ searches, it is necessary to study experimentally well-established SM processes that share similar topological features. The associated production of a $Z$ boson and a $Z$ boson ($ZZ$) or a $Z$ boson and a Higgs boson ($ZH$) in the $bb\tau\tau$ final state serves as an ideal benchmark. These processes share the experimental challenges of the $HH$ search but possess significantly larger cross sections (approximately 64 times larger for $ZZ$ and 2.2 times larger for $ZH$). Furthermore, these channels provide a unique opportunity to probe physics beyond the Standard Model (BSM), specifically searching for heavy spin-0 resonances ($X$) decaying to $ZZ$ and spin-1 resonances ($Z'$) decaying to $ZH$, which could arise from extensions such as singlet models, two-Higgs-doublet models, or warped extra dimensions.

Methodology
The analysis utilizes proton-proton collision data collected by the CMS experiment at $\sqrt{s} = 13$ TeV, corresponding to an integrated luminosity of 138 fb$^{-1}$ (2016–2018). The study targets three specific final states:

1. $ZZ \to bb\tau\tau$ (denoted $ZbbZ\tau\tau$)
2. $ZH \to (Z \to bb)(H \to \tau\tau)$ (denoted $ZbbH\tau\tau$)
3. $ZH \to (Z \to \tau\tau)(H \to bb)$ (denoted $Z\tau\tau Hbb$)

Event Reconstruction and Selection:

• Tau Decay Modes: The analysis considers three $\tau\tau$ decay modes: $\tau_\mu\tau_h$, $\tau_e\tau_h$, and $\tau_h\tau_h$, covering 87.6% of possible $\tau\tau$ decays.
• Kinematic Regimes: Events are categorized into "resolved" and "boosted" regimes based on the Lorentz boost of the $\tau\tau$ and $bb$ systems.
  • Resolved: Uses AK4 jets ($R=0.4$) and standard isolation criteria.
  • Boosted: Uses AK8 jets ($R=0.8$) and dedicated reconstruction algorithms for collimated objects.
• Object Identification:
  • Jets: AK4 jets are identified using the DEEPJET algorithm (medium working point). AK8 jets are identified using the PARTICLENET algorithm (low-purity working point) to tag $Z/H \to bb$ decays.
  • Taus: Hadronic tau candidates are reconstructed using the HPS algorithm and identified using the DEEPTAU (resolved) or BOOSTEDDEEPTAU (boosted) algorithms.
  • Leptons: Electrons and muons are reconstructed with standard CMS criteria, with specific isolation adjustments for the boosted regime.
• Categorization: Events are classified into 12 mutually exclusive categories based on the $\tau\tau$ decay channel (3) and the reconstruction regime ($bb$ resolved 1b, resolved 2b, boosted $bb$-resolved $\tau\tau$, boosted $bb$-boosted $\tau\tau$).
• Mass Reconstruction: The invariant mass of the $\tau\tau$ system ($m_{\tau\tau}$) is reconstructed using the FASTMTT algorithm, which employs a collinear approximation and missing transverse momentum ($p_T^{miss}$). The $bb$ mass ($m_{bb}$) is derived from dijet invariant mass (resolved) or soft-drop groomed mass (boosted). An elliptical selection in the $(m_{bb}, m_{\tau\tau})$ plane is applied to select signal candidates.

Background Modeling and Signal Extraction:

• Backgrounds: The dominant backgrounds are top quark pair production ($t\bar{t}$) and Drell-Yan ($Z/\gamma^* \to \ell\ell, \tau\tau$) processes. QCD multijet background is estimated using data-driven ABCD methods. Other contributions (single top, $W$+jets, dibosons) are modeled using Monte Carlo (MC) simulations.
• Signal Extraction: Deep Neural Networks (DNNs) are trained to discriminate signal from background.
  • Non-resonant: Dedicated DNNs are trained for each SM process ($ZZ$, $ZbbH$, $Z\tau\tau H$).
  • Resonant: A parameterized neural network is trained for resonant searches ($X \to ZZ$, $Z' \to ZH$), using the resonance mass as an input parameter to allow for model-independent limits across a mass range.
• Statistical Analysis: A binned maximum likelihood fit is performed on the DNN output distributions across all categories and channels. Systematic uncertainties (luminosity, object identification/energy scales, theoretical cross sections, PDFs, and MC statistics) are treated as nuisance parameters.

Key Contributions and Results

• First Measurements: This work presents the first measurements of the $ZZ \to bb\tau\tau$ and $ZH \to bb\tau\tau$ processes.
• SM Measurements:
  • The signal strength ($\mu$) for $ZZ \to bb\tau\tau$ is measured as $0.10^{+0.70}_{-0.68}(\text{stat})^{+0.83}_{-0.84}(\text{syst})$.
  • The signal strength for $ZH \to bb\tau\tau$ is measured as $-1.4^{+1.2}_{-1.1}(\text{stat}) \pm 1.0(\text{syst})$.
  • Both results are consistent with Standard Model predictions.
  • Upper limits at 95% confidence level (CL) are set on the production cross sections, ranging from 300 pb to 24 fb for $ZZ$ and 0.4 pb to 12 fb for $ZH$.
• Resonant Searches:
  • Searches are conducted for heavy spin-0 resonances ($X \to ZZ$) in the mass range 200 GeV – 5 TeV and spin-1 resonances ($Z' \to ZH$) in the range 500 GeV – 6 TeV.
  • No significant deviation from SM expectations is observed.
  • Upper limits are set on $\sigma(X)\mathcal{B}(X \to ZZ)$ and $\sigma(Z')\mathcal{B}(Z' \to ZH)$ across the mass ranges.
• Sensitivity Projections: Projections for the HL-LHC suggest that evidence for $ZZ \to bb\tau\tau$ could be reached with $\sim 1$ ab$^{-1}$ and observation with $>2.5$ ab$^{-1}$, provided systematic uncertainties are reduced. The $ZH \to bb\tau\tau$ process remains more challenging, likely requiring substantial improvements in analysis methods and background modeling for observation.

Significance
The paper establishes the $ZZ/ZH \to bb\tau\tau$ processes as viable "standard candles" for the $bb\tau\tau$ final state. By demonstrating that analysis techniques developed for these well-known processes can be successfully applied, the study validates the strategies required for the future observation of non-resonant Higgs boson pair production ($HH \to bb\tau\tau$). The results provide the first experimental constraints on $ZZ$ and $ZH$ production in this specific final state and set the first limits on heavy resonances decaying to these states in the $bb\tau\tau$ channel, probing potential enhancements in couplings to third-generation fermions. The analysis confirms that while the $ZZ \to bb\tau\tau$ process offers a more accessible path to observation than $HH \to bb\tau\tau$ due to larger cross sections, it remains subject to significant background challenges that must be addressed to fully exploit the HL-LHC dataset.