ATLAS closes in on rare Higgs decays

7 July 2025 | By

The ATLAS Collaboration finds evidence of Higgs-boson decays to muons and improves sensitivity to Higgs-boson decays to a Z boson and a photon.

Since the discovery of the Higgs boson in 2012, physicists have made major strides in exploring its properties. Does that mean the subject is done and dusted? Far from it! In new results presented at the 2025 European Physical Society Conference on High Energy Physics (EPS-HEP), the ATLAS Collaboration narrowed in on two exceptionally rare Higgs-boson decays using data collected in Run 3 of the Large Hadron Collider (LHC). These studies offer deep insights into how closely the Higgs boson’s behaviour aligns with the Standard Model.

The first process under study was the Higgs-boson decay into a pair of muons (H→μμ). Despite its scarceness – occurring in just 1 out of every 5000 Higgs decays – this process provides the best opportunity to study the Higgs interaction with second-generation fermions and shed light on the origin of mass across different generations. The second investigated process was the Higgs-boson decay into a Z boson and a photon (H→Zγ), where the Z boson subsequently decays into electron or muon pairs. This rare decay is especially intriguing, as it proceeds via an intermediate "loop" of virtual particles. If new particles contribute to this loop, the process could offer hints of physics beyond the Standard Model.

Looking for needles in a haystack

Identifying these rare decays is quite the challenge. For H→μμ, researchers looked for a small excess of events clustering near a muon-pair mass of 125 GeV (the mass of the Higgs boson). This signal can be easily hidden behind the thousands of muon pairs produced through other processes (“background”).

The H→Zγ decay is even harder to isolate, as the chances of spotting its signal are complicated by the fact that the Z boson only decays into detectable leptons about 6% of the time. Compounding the challenge are the operation conditions of LHC Run 3, which features more overlapping collisions, making it easier for particle jets to mimic real photons.

To boost the sensitivity of their searches, ATLAS physicists combined the first three years of Run-3 data (165 fb-1, collected between 2022-2024) with the full Run-2 dataset (140 fb-1, from 2015-2018). They also developed a sophisticated method to better model background processes, categorised recorded events by the specific Higgs-production modes, and made further improvements to their event-selection techniques in order to maximize the likelihood of spotting genuine signals. Figure 1 shows the resulting muon-pair mass distribution obtained with the data collected in 2022 to 2024 and combined over all the categories. Figure 2 displays the observed distribution of the Zγ system in the combined data sample from Run 2 and Run 3.

Physics,ATLAS
Figure 1: Dimuon invariant mass spectrum observed in Run-3 data, for all combined analysis categories. The background and signal probability density functions (pdf) are obtained from the combined fit of all categories to the Run-3 data, corresponding to a signal strength of μ = 1.6 ±0.6. The lower panel shows the fitted signal pdf, normalized to the signal best-fit value, and the difference between the observed data and the background model. Error bars represent the data’s statistical uncertainties. (Image: ATLAS Collaboration/CERN)
Physics,ATLAS
Figure 2: Weighted m_Zγ distributions of data events across all categories of the combined Run 2 and Run 3 dataset. The black points represent the data, with statistical uncertainties shown as vertical error bars. Each event is weighted by ln(1 + S/B), where S and B are the signal and background yields extracted from fits to the data in the mass window 120 ≤ m_Zγ ≤ 130 GeV. The red curve corresponds to the result of a simultaneous signal-plus-background fit across all categories, while the blue curve shows the background-only component of the fit. (Image: ATLAS Collaboration/CERN)

Finding evidence and enhancing sensitivity

In the previous search for H→μμ using the full Run-2 dataset, the ATLAS Collaboration saw its first hint of this process at the level of 2 standard deviations. The comparable CMS result reached an observed (expected) significance of 3 (2.5) standard deviations. Now, with the combined Run-2 and Run-3 datasets, the ATLAS Collaboration has found evidence for H→μμ with an observed (expected) significance over the background-only hypothesis of 3.4 (2.5) standard deviations. This means that the chance that the result is a statistical fluctuation is less than one in 3000!

As for the H→Zγ process, a previous ATLAS and CMS combined analysis used Run-2 data to find evidence of this decay mode. It reported an observed (expected) excess over the background-only hypothesis of 3.4 (1.6) standard deviations. The latest ATLAS result, combining Run-2 and Run-3 data, reported an observed (expected) excess over the background-only hypothesis of 2.5 (1.9) standard deviations. This outcome provides the most stringent expected sensitivity to date for measuring the decay probability (“branching fraction”) of H→Zγ.

These achievements were made possible by the large, excellent dataset provided by the LHC, the outstanding efficiency and performance of the ATLAS experiment, and the use of novel analysis techniques. With more data on the horizon, the journey of exploration continues!


About the event displays: Left: Candidate Higgs boson decaying to two muons (H→μμ). The event features two muons (red tracks with their associated muon chambers shown as blue boxes with green measurement bars) with invariant mass 125.324 GeV, consistent with H → μμ, and two forward jets (yellow cones). Charged-particle trajectories in the inner detector are shown as orange lines, and the yellow/orange and green/cyan boxes represent the energy deposited in the hadronic and electromagnetic calorimeters, respectively. Right: Candidate Higgs boson decaying to a photon and a Z boson, with the Z subsequently decaying to an electron-positron (H → Zγ → eeγ). The photon is depicted by a transverse purple cone, the electron-positron by green lines, consistent with H → Zγ, and two forward jets (the yellow cones). Charged-particle trajectories in the inner detector are depicted as orange lines. The yellow/orange and green/cyan boxes represent the energy deposited in the hadronic and electromagnetic calorimeters, respectively. (Image: ATLAS Collaboration/CERN)

Learn more