ATLAS deepens the search for long-lived particles with Run 3 data

4 August 2026 | By

Long-lived particles (LLPs) are among the most compelling targets in the search for physics beyond the Standard Model. Unlike most particles produced in proton–proton collisions at the Large Hadron Collider (LHC), which decay almost instantaneously, LLPs can travel a measurable distance before decaying. This delay can leave a striking signature in the ATLAS experiment: “displaced” tracks away from the interaction point (see event display). While the Standard Model contains a few LLPs – bb-hadrons, for example, travel measurable distances before decaying and muons travel through the entire ATLAS detector before decaying – many theories extending the Standard Model predict additional LLPs. These new particles could help explain outstanding mysteries in physics, such as the nature of dark matter.

Displaced ATLAS
Figure 1: Distribution of the predicted background yields and observed data in the far signal region (i.e. where the displaced vertex is more than 4 mm from the beam line) as a function of the reduced mass of the displaced vertex. The vertex’s mass is scaled by how tightly clustered its associated tracks are (see publication for definition). The data points (black) are consistent with the background prediction. An example benchmark new-physics scenario – supersymmetric Higgs-boson (“Higgsino”) pair production whose decay produces the displaced vertex and displaced muon signature this search targets – is overlaid for illustration. (Image: ATLAS Collaboration/CERN)

To search for these hypothetical particles, researchers typically look for the experimental signatures left behind when an LLP decays, which often occur at some distance from the interaction point. When several displaced tracks originate from the same location, they can be reconstructed as a displaced vertex. In addition, some LLP models predict decays that produce displaced muons. The ATLAS Collaboration has released a new search for massive, long-lived particles at a collision energy of 13.6 TeV, targeting events containing at least one displaced vertex and one displaced muon. This is the first ATLAS search for LLPs to use LHC Run-3 data collected between 2022 and 2024, corresponding to an integrated luminosity of 164 fb-1.

The analysis benefits from several important improvements in tracking, vertex reconstruction and real-time event-selection tools (“triggers”). In 2022, the ATLAS Collaboration introduced a dedicated trigger for displaced muons, capable of identifying muons with transverse momentum as low as 20 GeV, thus extending the experiment's sensitivity to LLPs with masses at the electroweak scale. Improvements in displaced tracking and vertex reconstruction further enhanced sensitivity to LLPs with mean decay lengths ranging from one millimetre – comparable to the typical decay length of a b-hadron – to tens of centimetres, covering a wide range of possible new-physics scenarios.


Leveraging Run-3 ATLAS detector capabilities, the ATLAS Collaboration has set competitive model-independent limits on events with a displaced vertex and displaced muon, and world-leading limits on multiple benchmark models of R-parity-violating supersymmetry.


Displaced ATLAS
Figure 2: Observed and expected limits at 95% confidence level for supersymmetry top-quark (“stop”) pair production, whose decay yields a displaced vertex and displaced muon signature, as a function of stop mass and proper lifetime. Previous limits from ATLAS and CMS are shown for comparison. (Image: ATLAS Collaboration/CERN)

To estimate the number of background events that could mimic the LLP signal, researchers developed a fully data-driven approach that accounted for unconventional sources, including cosmic rays, fake tracks (i.e. tracks reconstructed from unrelated hits that do not correspond to a real particle) and decays of b-hadrons. Their method used discriminating variables such as muon displacement and displaced-vertex mass to distinguish these backgrounds from the signal directly in collision data, as simulations do not accurately model these processes.

No significant excess above the expected background was observed (see Figure 1) and the results were interpreted in two complementary ways. In a model-independent approach, researchers set limits as low as 0.018 fb on the visible cross-section of events containing at least one displaced vertex and one displaced muon. In a model-dependent approach, the team set world-leading limits on several benchmark models of R-parity-violating supersymmetry (see Figure 2), improving previous limits on the production cross section by up to two orders of magnitude for some benchmark scenarios.

Future analyses will continue to explore displaced signatures using the full Run 2 and Run 3 datasets, further developing these techniques to enhance sensitivity to LLPs at the High-Luminosity LHC and beyond.


About the banner image: Event display of a candidate signal event featuring a displaced vertex (blue circle) reconstructed from four displaced tracks, located 47 mm from the primary vertex (pink circle) in the transverse plane and with an invariant mass of 32 GeV. The event also contains a displaced muon candidate (red line on the left, orange line on the right) with a transverse impact parameter (the shortest distance in the transverse plane between the particle's track and the primary interaction vertex) of approximately 3 mm. Primary tracks originating from the collision point are shown in yellow, while displaced tracks with large impact parameters are shown in cyan – all with a transverse momentum of at least 2 GeV. (Image: ATLAS Collaboration/CERN)

Learn more