Tags:

Topping it off: ATLAS completes the single-top quark picture at the LHC

26 October 2026 | By

The ATLAS Collaboration reports the first observation at the LHC of a single top quark produced via the s-channel. Previously observed through a combined result from the CDF and DØ experiments at Fermilab's Tevatron accelerator, this rare process is now observed for the first time by a single experiment, completing the experimental picture of electroweak single-top-quark production at the LHC.

The LHC is a top-quark “factory”, churning out hundreds of millions of top quarks. While most of these top quarks emerge in pairs through the strong nuclear force, they can occasionally be produced individually via the electroweak force. These elusive single top quarks offer significant opportunities for physicists to carry out precision measurements and search for physics beyond the Standard Model.

Single top quarks can be produced in one of three ways: via the t-channel, involving the exchange of a W boson; in associated production with a W boson, known as tW production; and through the s-channel, in which a quark and an antiquark annihilate through a W boson to produce a top quark and a bottom antiquark (or an antitop and a bottom quark) (see Figure 1). While the t-channel and tW production have been studied extensively, the s-channel proved particularly elusive at the LHC.

Testing,Milestones,Technology,Detectors,ATLAS — Figure 1: Leading order Feynman diagrams of single top-quark production in the t-channel, s-channel and tW-channel. (Image: ATLAS Collaboration/CERN)

Part of this challenge arises from the accelerator. The s-channel requires a quark and an antiquark to collide, giving Fermilab’s proton–antiproton Tevatron collider a natural advantage over the LHC, which collides protons with protons. Even so, the process was only observed in 2014 after the CDF and DØ Collaborations combined their full datasets.

The increase in the LHC's collision energy during recent runs – while vital for the LHC physics programme – brought further complications for s-channel top-quark production searches. While the higher energy slightly increased the s-channel production rate, the dominant backgrounds grew more rapidly. Searches typically rely on the production’s distinctive signature, with particle jets originating from bottom quarks (“b-jets”). However, top-quark pair events produce a very similar signature and occur far more often, leaving the rare s-channel events buried beneath a large background.

ATLAS observes rare s-channel single-top-quark production – a first for the LHC and the first such observation by a single experiment.

In a recent paper submitted for publication to PRL, the ATLAS Collaboration presented their search for s-channel production among the 20 billion proton–proton collision events collected during LHC Run 2. To achieve the required sensitivity, the team refined their search strategy by combining advanced machine-learning techniques, improved b-jet identification and precise theoretical modelling. In particular, they used the new GN2 algorithm, which employs a cutting-edge graph-transformer architecture to distinguish b-jets more effectively from jets originating from light and charm quarks. As a result, the analysis team selected three times more signal events compared to the previous search using the same dataset, while maintaining adequate background control.

ATLAS physicists also used a Graph Neural Network (GNN) to learn patterns in the event kinematics and assign scores reflecting how closely each event resembled a signal or background. Further gains in sensitivity came from refined theoretical modelling of the signal and backgrounds and reduced systematic uncertainties.

By comparing the distribution of GNN scores in the signal region and several background control regions (see Figure 2), the ATLAS Collaboration observed the s-channel signal with a statistical significance of 5.1 standard deviations (compared to an expected 5.2). The measured cross section is 10.8 ± 1.5 (stat.) ± 1.6 (syst.) pb, in agreement with the Standard Model prediction of 11.07 ± 0.19 pb. This is the most precise cross-section measurement of this process to date, with a relative uncertainty of about 20%. Selecting events with a GNN score above 0.8 clearly reveals the s-channel signal in other kinematic distributions (see Figure 3).

This milestone closes the final gap in the experimental picture of single-top-quark production at the LHC, completing the observation of all three production channels. Beyond this achievement, it opens new opportunities to explore the electroweak properties of the top quark with increasing precision. The larger datasets from LHC Run 3 (2022-2026) and the upcoming High-Luminosity LHC, together with continued advances in b-jet identification, machine-learning techniques, detector uncertainty treatments and theoretical modelling, will enable more precise measurements of single-top-quark production. These measurements will provide new ways to search for signs of new physics beyond the Standard Model.

About the banner image: Event display of a candidate single-top-quark s-channel event, recorded by the ATLAS detector in proton-proton collisions at a centre-of-mass energy of 13 TeV on 12 September 2015. The reconstructed muon is shown by the red line through the detector. The green and yellow bars indicate energy deposits in the liquid-argon and scintillating-tile calorimeters; from these deposits, two jets are identified and shown as yellow cones. Both jets are identified as originating from b-quarks. The missing transverse momentum is shown by the dotted white line. (Image: ATLAS Collaboration/CERN)