Shedding light with jets from the dark side
15 May 2025 | By
Imagine a jet of particles that defies the rules of the Standard Model. Unlike most jets produced in high-energy collisions, its energy seems to suddenly vanish or it doesn't originate at the collision point — instead, blooming suddenly deeper in the detector. These jets aren't just detector glitches – they could be signs of elusive dark matter.
Researchers at the ATLAS Experiment are searching for a hidden “dark sector”, a realm of new sets of particles and forces that interact very weakly with the known matter, potentially producing elusive and puzzling signatures. One of the most compelling theoretical models they’re exploring is dark QCD – the shadowy counterpart to the strong force, capable of binding dark quarks into exotic matter. If it exists, it could give rise to dark jets: sprays of dark-sector particles that partially decay into visible particles, while some components escape detection.
In two new results, ATLAS physicists hunted for dark-sector physics using these distinctive jet signatures. Their first search targeted semi-visible jets, where part of the jet’s momentum is carried away by invisible dark particles, leaving behind a partial signature. Their second search studied emerging jets, where the jet seems to emerge progressively mid-flight – produced when dark particles decay into visible particles after traveling some distance in the detector (i.e they appear “displaced” from the interaction point). Emerging jets have a striking radial profile, with little activity near the collision point followed by a blossoming of tracks at larger radii. A sketch of the jet topologies is shown in Figure 1. Due to the double showering process involving both dark and Standard Model QCD, jet can generally assume wider, more fragmented structures and, depending on the scenario, can include invisible or displaced components.
Researchers at the ATLAS Experiment are searching for a hidden “dark sector”, a realm of new sets of particles and forces that interact very weakly with the known matter.
Both searches deployed a two-pronged strategy. First, researchers trained powerful neural networks on charged particle track information to recognise these subtle signatures predicted by dark-sector models. Second, they applied a model-agnostic strategy – broadening the discovery potential across a wider range of new-physics scenarios.
In the semi-visible jets analysis, researchers looked for jets paired with missing transverse momentum – an apparent momentum imbalance in the plane perpendicular to the beam, which is a hallmark of invisible dark particles escaping the detector. A central variable is Rinv,which indicates the fraction of dark particles that go undetected, representing possible dark matter candidates. As Rinv increases, jets appear faint or incomplete, and the momentum imbalance in the event often aligns with the jet direction, making these events harder to distinguish from background. To identify potential signals, researchers selected collision events with two semi-visible jets and searched for a resonant “bump” in the transverse mass distribution, a mass-related quantity defined by both the jets and the invisible component of the event. Such a signal could be a heavy new boson (Z′) mediating interactions between Standard Model particles and dark quarks (see Figure 2a). Using the full Run 2 dataset (collected between 2015 and 2018), the results set strong limits on the mass of the Z’ mediator (see Figure 2b), excluding Z′ masses from 2000 to 3200 GeV for Rinv values between 0.2 and 0.37, at 95% confidence level.
Two new ATLAS results target the distinctive signatures of dark-sector particles and mark the beginning of a new chapter in dark-sector exploration for Run 3 and beyond.
For the emerging jets study, the team focused on displaced decays from dark-sector cascades. One key observable adopted was the prompt-track-fraction, which measures how many tracks start near the collision point (see Figure 3a). For the Run 3 data-taking campaign, ATLAS physicists even introduced a dedicated event-selection system (or trigger) to catch these unusual signatures in real time. The analysis also used specialised jet reconstruction techniques and algorithms to spot displaced tracks and secondary vertices, to better distinguish emerging jets from those produced by known particles. The results were interpreted in the context of different kinds of mediator particles connecting the Standard Model to the dark sector. Using Run 3 data collected in 2022 and 2023, exclusion limits were set on mediator mass and the proper decay length of the long-lived dark hadrons (see Figure 3b). Mediator masses are excluded between 600 GeV and 2550 GeV for quark and dark quark couplings of 0.01 and 0.1, respectively, assuming dark pion decay lengths between 5 mm and 50 mm.
Together, these results mark a big step forward in ATLAS' exploration for the dark sector. The semi-visible jet study delivers ATLAS’ first s-channel dark-sector search, while the emerging jet analysis opens entirely new windows into long-lived dark phenomena. By combining model-specific neural networks with model-agnostic approaches, these results offer a powerful discovery framework – one that hunts for predicted signals while remaining sensitive to the unexpected. With advanced reconstruction algorithms, specialised triggers and state-of-the-art analysis techniques, these results mark the beginning of a new chapter in dark sector exploration for Run 3 and beyond.
