Testing flavour symmetry with the Higgs boson

Lepton number is a quantum number that represents the difference in the number of leptons and antileptons participating in a process, while lepton flavour is a corresponding quantity that accounts for each generation of lepton (e, μ or τ) separately. Lepton number is always conserved but lepton flavour violation (LFV) is known to exist in nature, as this phenomenon has been observed in neutrino oscillations – the transition of a neutral lepton of a given flavour to one with a different flavour. This observation motivates searches for additional manifestations of LFV that may be the result of beyond-the-Standard Model (SM) physics, key among which is the search for LFV decays of the Higgs boson. 

The ATLAS collaboration has recently announced the results of searches for H → eτ and H → μτ decays based on the full Run 2 data set, which was collected at a centre-of-mass energy of 13 TeV. The unstable τ lepton decays to an electron or a muon and two neutrinos, or to one or more hadrons and one neutrino. Most of the background events in these searches arise from SM processes such as Z → ττ, the production of top–antitop and weak-boson pairs, as well as from events containing misidentified or non-prompt leptons (fake leptons). These fake leptons originate from secondary decays, for example of charged pions. Several multivariate analysis techniques were used for each final state to provide the maximum separation between signal and background events.

To ensure the robustness of the measurement, two background estimation methods were employed: a Monte Carlo (MC) template method in which the background shapes were extracted from MC and normalised to data, and a “symmetry method”, which used only the data and relied on an approximate symmetry between prompt electrons and prompt muons. Any difference between the branching fractions B(H → eτ_μ) and B(H → μτ_e), where the subscripts μ and e represent the decay modes of the τ lepton, would break this symmetry. In both cases, contributions from events containing fake leptons were estimated directly from the data.

The MC-template method enables the measurement of the branching ratios of the LFV decay modes. Searches based on the MC-template method for background estimation involve both leptonic and hadronic decays of τ leptons. A simultaneous measurement of the H → eτ and H → μτ decay modes was performed. For the H → μτ (H → eτ) search, a 2.5 (1.6) standard deviation upward fluctuation above the SM background prediction is observed. The observed (expected) upper limits on the branching fractions B(H → eτ) and B(H → μτ) at 95% confidence level are slightly below 0.2% (0.1%), which are the most stringent limits obtained by the ATLAS experiment on these quantities. The result of the simultaneous measurement of the H → eτ and H → μτ branching fractions is compatible with the SM prediction within 2.2 standard deviations (see figure 1).

The observed upper limits on the branching fractions are the most stringent limits obtained by the ATLAS experiment

The symmetry method is particularly sensitive to the difference in the two LFV decay branching ratios. For this measurement, only the fully leptonic final states were used. Special attention was paid to correctly account for asymmetries induced by the different detector response to electrons and muons, especially regarding the trigger and offline efficiency values for lepton reconstruction, identification and isolation, as well as regarding contributions from fake leptons. The measurement of the branching ratio difference indicates a small but not significant upward deviation for H → μτ compared to H → eτ. The best-fit value for the difference between B(H → μτ_e) and B(H → eτ_μ) is (0.25 ± 0.10)%. 

The expected twice-larger LHC Run 3 dataset at the higher centre-of-mass energy of 13.6 TeV will shed further light on these results.