ATLAS measures rare Higgs boson interaction with tau leptons

15 December 2023 | By

Over a decade since its discovery, particle physicists still have much to learn about the Higgs boson. A unique feature of the Higgs boson, according to the Standard Model, is its intrinsic link to the masses of fundamental particles. Using the full LHC Run 2 dataset (collected from 2015 - 2018), ATLAS physicists are revealing a more nuanced portrait of the Higgs boson in hopes of understanding how closely it aligns with Standard Model predictions.

In a new result, the ATLAS Collaboration reports the first evidence of a Higgs boson produced in association with a leptonically-decaying W or Z boson and decaying into a pair of tau leptons. This interaction is valuable for studying the relationship between the Higgs boson and mass, as the Higgs boson is expected to couple more strongly to particles with larger masses. Consequently, studies of the Higgs boson’s interaction with the heaviest lepton (the tau lepton) are a major target of LHC researchers.

ATLAS researchers find evidence for a rare Higgs boson process, which accounts for less than 0.02% of all Higgs bosons produced at the LHC.

Figure 1: The fitted values of the Higgs boson signal strength for a Higgs boson with 125 GeV mass are plotted for the WH and ZH processes separately and in combination. Their combination shows that the measurement agrees with the expected behavior of the Standard Model Higgs boson within uncertainties. (Image: ATLAS Collaboration/CERN)

Previous ATLAS measurements of this process were only able to report on the approximately 70% of events in which the W or Z boson decayed into hadrons. This new result marks the first time ATLAS has measured the cross section of this interaction in which the W or Z boson decays into leptons.

This rare process accounts for less than 0.02% of all Higgs bosons produced at the LHC – so how were researchers able to spot it? While the enormous size of the LHC Run 2 dataset was vital, so too was the use of new machine learning techniques. Researchers introduced a neural network-based strategy to their analysis, enhancing the selection efficiency of signal vs. background events. The neural network was trained to classify simulation samples of Higgs boson and background diboson events (WZ or ZZ). To validate the neural network strategy, the analysis was also done using variables related to the Higgs boson mass to identify background events.

The neural network analysis strategy yielded a measured cross section of 8.5 ± 2.6 fb, which agrees with the Standard Model prediction of 6.59 ± 0.03 fb within one standard deviation. This corresponds to a measured signal strength, i.e. the ratio of measured to predicted cross-sections, of μVH = 1.28 ± 0.39. The statistical significance of the measurement is 4.2σ (see Figure 1). By contrast, the mass-based analysis strategy yielded μVH = 1.40 ± 0.49, corresponding to a significance of 3.5σ. Figures 2a) and 2b) show distributions of the measured ATLAS data compared with predicted amounts of signal and background for both the neural network and mass-based analyses.

Figure 2: The data (black points) are compared with distributions of the expected signal (in red) and the estimated backgrounds (in blue, green, and purple) for the WH(τhadτhad) sub-category. The same events are plotted with respect to (a) the neural network score and (b) a variable related to the Higgs boson mass. The hatched band indicates the total post-fit uncertainty of the total predicted yields. (Image: ATLAS Collaboration/CERN)

This new measurement exemplifies how Higgs boson research has shifted into a new phase at the LHC experiments. In the coming decade, physicists will use the Higgs boson not only to report initial discoveries, but also for increasingly precise measurements – enabled by the LHC’s ever-growing datasets, and better reconstruction and analysis strategies that take full advantage of the latest developments in artificial intelligence.

Learn more