A search for decays of the Higgs boson into a $Z$ boson and a light resonance, with a mass of 0.5-3.5 GeV, is performed using the full 140 fb$^{-1}$ dataset of 13 TeV proton-proton collisions recorded by the ATLAS detector during Run~2 of the LHC. Leptonic decays of the $Z$ boson and hadronic decays of the light resonance are considered. The resonance can be interpreted as a $J/\psi$ or $\eta_c$ meson, an axion-like particle, or a light pseudoscalar in two-Higgs-doublet models. Due to its low mass, it would be produced with high boost and reconstructed as a single small-radius jet of hadrons. A neural network is used to correct the Monte Carlo simulation of the background in a data-driven way. Two additional neural networks are used to distinguish signal from background. A binned profile-likelihood fit is performed on the final-state invariant mass distribution. No significant excess of events relative to the expected background is observed, and upper limits at 95% confidence level are set on the Higgs boson's branching fraction to a $Z$ boson and a light resonance. The exclusion limit is 10% for the lower masses, and increases for higher masses. Upper limits on the effective coupling $C^\text{eff}_{ZH}/\Lambda$ of an axion-like particle to a Higgs boson and $Z$ boson are also set at 95% confidence level, and range from 0.9 to 2 TeV$^{-1}$.
The angularity, for data, background (pre- and post-reweighting) and three $H\rightarrow Za$ signal hypotheses (for $a\rightarrow q\bar{q}/gg$ inclusively). Events are required to pass the complete event selection but not the classification NN requirement. The background normalization is set equal to that of the data for events passing the preselection and being in the $m_{\ell\ell j}$ 100-180 GeV region. The signal normalization assumes the SM Higgs boson inclusive production cross-section, $\mathcal{B}(H\to Za)=100\%$, and it is scaled up by a factor of 100. The error bars (hatched regions) represent the data (MC) sample's statistical uncertainty in the histograms and the ratio plots. Vertical arrows indicate data points that fall outside the displayed $y$-axis range.
The modified energy correlation function, for data, background (pre- and post-reweighting) and three $H\rightarrow Za$ signal hypotheses (for $a\rightarrow q\bar{q}/gg$ inclusively). Events are required to pass the complete event selection but not the classification NN requirement. The background normalization is set equal to that of the data for events passing the preselection and being in the $m_{\ell\ell j}$ 100-180 GeV region. The signal normalization assumes the SM Higgs boson inclusive production cross-section, $\mathcal{B}(H\to Za)=100\%$, and it is scaled up by a factor of 100. The error bars (hatched regions) represent the data (MC) sample's statistical uncertainty in the histograms and the ratio plots. Vertical arrows indicate data points that fall outside the displayed $y$-axis range.
$Z$ boson transverse momentum, for data, background (pre- and post-reweighting) and three $H\rightarrow Za$ signal hypotheses (for $a\rightarrow q\bar{q}/gg$ inclusively). Events are required to pass the complete event selection but not the classification NN requirement. The background normalization is set equal to that of the data for events passing the preselection and being in the $m_{\ell\ell j}$ 100-180 GeV region. The signal normalization assumes the SM Higgs boson inclusive production cross-section, $\mathcal{B}(H\to Za)=100\%$, and it is scaled up by a factor of 100. The error bars (hatched regions) represent the data (MC) sample's statistical uncertainty in the histograms and the ratio plots. Vertical arrows indicate data points that fall outside the displayed $y$-axis range.
A search for Higgs boson decays into a $Z$ boson and a light resonance in two-lepton plus jet events is performed, using a $pp$ collision dataset with an integrated luminosity of 139 fb$^{-1}$ collected at $\sqrt{s}=13$ TeV by the ATLAS experiment at the CERN LHC. The resonance considered is a light boson with a mass below 4 GeV from a possible extended scalar sector, or a charmonium state. Multivariate discriminants are used for the event selection and for evaluating the mass of the light resonance. No excess of events above the expected background is found. Observed (expected) 95$\% $ confidence-level upper limits are set on the Higgs boson production cross section times branching fraction to a $Z$ boson and the signal resonance, with values in the range 17 pb to 340 pb ($16^{+6}_{-5}$ pb to $320^{+130}_{-90}$ pb) for the different light spin-0 boson mass and branching fraction hypotheses, and with values of 110 pb and 100 pb ($100^{+40}_{-30}$ pb and $100^{+40}_{-30}$ pb) for the $\eta_c$ and $J/\psi$ hypotheses, respectively.
Observed number of data events and expected number of background events in the signal region.
Efficiencies of the MLP selection, complete selection and total expected signal yields for each signal sample, assuming B$(H\to Z(Q/a))=100\%$ and $\sigma(pp\to H) = \sigma_\text{SM}(pp\to H)$. Pythia 8 branching fractions of $a$ are assumed using a $\tan\beta$ value of 1. The MLP efficiencies, total efficiencies, and expected yields are determined using MC samples, with uncertainties due to MC sample statistics, except for the expected background yield. The expected background yield and its uncertainty is calculated as described in the main text of the paper.
Expected and observed 95% CL upper limits on $\sigma(pp\to H)B(H\to Za)/$pb. These results are quoted for $B(a\to gg)=100\%$ and $B(a\to s\bar{s})=100\%$ for each signal sample. The smaller (larger) quoted ranges around the expected limits represent $\pm 1\sigma$ ($\pm 2\sigma$) fluctuations.
Forum