⚛️ high-energy experiments

Search for ttbar resonances in final states with exactly one or two leptons using 140 fb $^{-1}$ of pp collision data at $\sqrt{s}=13$ TeV with the ATLAS experiment

The ATLAS experiment analyzed 140 fb $^{-1}$ of 13 TeV proton-proton collision data in final states with one or two leptons to search for heavy spin-1 and spin-2 resonances decaying into top-antitop quark pairs, finding no significant deviation from the Standard Model and setting exclusion limits on hypothetical $Z'$ bosons, Kaluza-Klein gravitons, and Kaluza-Klein gluons.

Original authors: ATLAS Collaboration

Published 2026-10-30

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: ATLAS Collaboration

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The universe is built on a foundation of particles and forces that physicists have spent a century mapping out. This map, known as the Standard Model, successfully explains how the known building blocks of matter interact. However, it is incomplete. It leaves out gravity, the force that holds planets in orbit, and it offers no explanation for why some particles are incredibly heavy while others are nearly weightless. To fill these gaps, scientists have proposed that there are heavier, undiscovered particles lurking just beyond our current reach. These hypothetical particles would be massive, far heavier than anything we have ever created, and they might decay into pairs of top quarks, the heaviest known elementary particles. Finding such a heavy particle would be a monumental discovery, signaling that the laws of nature are richer and more complex than our current theories suggest.

To hunt for these elusive giants, the ATLAS experiment at the Large Hadron Collider in Switzerland smashes protons together at nearly the speed of light. When these protons collide, they release enough energy to briefly create heavy particles that do not exist in nature under normal conditions. If a new, heavy particle is born in such a collision, it would almost instantly fall apart into a pair of top quarks. The challenge is that top quarks are unstable and decay immediately into other particles, creating a chaotic spray of debris that is difficult to distinguish from the billions of ordinary collisions happening every second. The researchers must sift through this noise, looking for a specific signature: a pair of top quarks that appear to have come from a single, massive source rather than from the random jumble of standard collisions.

A team of scientists using the ATLAS detector has now performed the most comprehensive search to date for these heavy resonances. They analyzed a massive dataset containing 140 inverse femtobarns of collision data, collected over several years at an energy level of 13 tera-electronvolts. This represents the full collection of data from the second major run of the collider. The team focused on two specific ways the top quarks could decay: one where a single electron or muon is produced alongside other particles, and another where two such particles are produced. By combining these two approaches, they were able to look for signals across a vast range of masses, from 400 gigaelectronvolts up to 5 teraelectronvolts. They were searching for particles that could be either very narrow, appearing as a sharp spike in the data, or very broad, appearing as a wide, gentle hill.

The researchers built a detailed computer model of what the universe should look like if only the known Standard Model particles were present. They then compared this model against the actual data recorded by the detector. They looked for any deviation, any place where the real data showed more events than the model predicted, which would indicate the presence of a new particle. After carefully accounting for every known source of error and uncertainty, from the way the detector measures energy to the theoretical calculations of how particles behave, they found no such deviation. The data matched the Standard Model prediction perfectly across the entire mass spectrum they examined.

Because no new particles were found, the team did not report a discovery, but rather a set of strict boundaries. They established that if certain types of heavy particles exist, they cannot be lighter than specific mass limits. For example, they ruled out the existence of a specific type of heavy particle, known as a Z-prime boson, with a mass up to 4.2 teraelectronvolts if it is narrow, or up to 3.65 teraelectronvolts if it is slightly wider. They also excluded a different type of heavy particle, a Kaluza-Klein gluon, with masses up to 4.1 teraelectronvolts. For another candidate, a Kaluza-Klein graviton, they excluded masses up to 1.3 teraelectronvolts. These limits are the most stringent ever set for these specific types of particles in the decay channels they studied.

The search covered a wide variety of theoretical possibilities, including models that suggest extra dimensions of space or new forces that interact strongly with the top quark. While the results do not confirm the existence of these new particles, they provide a crucial guide for the future. By telling scientists exactly where these particles are not, the study narrows the field of possibilities. It forces theorists to refine their ideas, pushing the potential mass of these new particles even higher or suggesting that they might interact in ways that are even more difficult to detect. The work stands as a testament to the power of precision measurement, proving that even in the absence of a discovery, knowing what is not there is a vital step toward understanding what is.

Technical Summary: Search for $t\bar{t}$ Resonances in Final States with Exactly One or Two Leptons

Problem and Motivation
This paper presents a search for heavy resonances decaying into top-antitop ( $t\bar{t}$ ) quark pairs, a process predicted by various extensions of the Standard Model (SM) that aim to address open questions such as the hierarchy between the electroweak and Planck scales. The analysis targets spin-1 and spin-2 resonances, specifically $Z'$ bosons (spin-1, colour singlet), Kaluza–Klein (KK) gravitons (spin-2, colour singlet), and KK gluons (spin-1, colour octet). While previous searches by ATLAS and CMS have covered fully-hadronic and single-lepton final states, and CMS has included dilepton states, this work represents the first ATLAS search for these specific resonances in the dilepton channel. Furthermore, it utilizes the full Run 2 dataset (140 fb $^{-1}$ ), improving upon previous 36 fb $^{-1}$ analyses through enhanced reconstruction techniques and the inclusion of validation regions.

Methodology
The analysis utilizes proton-proton collision data collected by the ATLAS detector at $\sqrt{s} = 13$ TeV. The search focuses on two final states:

1-lepton channel: One top quark decays semi-leptonically ( $t \to bW \to b\ell\nu$ ) and the other hadronically ( $t \to bW \to bqq'$ ).
2-lepton channel: Both top quarks decay leptonically ( $t \to bW \to b\ell\nu$ ).

Events are selected requiring exactly one or two prompt electrons or muons, large missing transverse momentum ( $E_T^{miss}$ ), and at least two jets, with at least one identified as a $b$ -jet.

Event Categorization:
- In the 1-lepton channel, events are split into "merged" and "resolved" topologies based on the Lorentz boost of the hadronically decaying top quark. For high masses ( $m_{t\bar{t}} > 2$ TeV), decay products merge into a single large-radius jet, identified using a deep neural network (DNN) top-tagger. Resolved topologies require at least four small-radius jets. These categories are further subdivided based on the number of $b$ -tags (1b or 2b) and angular variables ( $\Delta\eta$ ) to define orthogonal signal and validation regions.
- In the 2-lepton channel, events are categorized into five regions based on the azimuthal angle difference between the two leptons ( $\Delta\phi_{\ell\ell}$ ) to optimize sensitivity to spin-1 resonances. A validation region is defined using the pseudorapidity difference ( $\Delta\eta_{\ell\ell}$ ).
Background Estimation:
- The dominant background, SM $t\bar{t}$ production, is modeled using Monte Carlo (MC) simulation (Powheg Box v2 + Pythia 8) at NLO accuracy, corrected to NNLO QCD + NLO EW differential predictions.
- $W/Z$ +jets and single-top backgrounds are modeled with MC (Sherpa, Powheg) and corrected using data-driven scale factors (e.g., charge asymmetry for $W$ +jets).
- The "fakes" background (non-prompt leptons or jets misidentified as leptons) in the 1-lepton channel is estimated using a data-driven matrix method.
- In the 2-lepton channel, the $Z$ +jets background shape is corrected using a reweighting procedure derived from a $Z$ -enriched control region.
Statistical Analysis:
A binned profile-likelihood fit is performed simultaneously across all signal regions (3 in 1-lepton, 5 in 2-lepton) using the reconstructed $t\bar{t}$ invariant mass ( $m_{t\bar{t}}$ ) for the 1-lepton channel and the invariant mass of the two leptons and two $b$ -jets ( $m_{\ell\ell bb}$ ) for the 2-lepton channel. Systematic uncertainties (experimental and theoretical) are treated as nuisance parameters.

Key Contributions

First ATLAS Dilepton Search: This is the first time the dilepton channel has been included in an ATLAS search for $t\bar{t}$ resonances, complementing the fully-hadronic and single-lepton searches.
Full Run 2 Dataset: The analysis utilizes the complete 140 fb $^{-1}$ dataset from 2015–2018, significantly increasing statistical power compared to previous 36 fb $^{-1}$ analyses.
Advanced Reconstruction: The use of a DNN-based top-tagger for merged topologies and improved $t\bar{t}$ reconstruction in resolved topologies enhances sensitivity across the mass range.
Comprehensive Background Validation: The inclusion of dedicated validation regions allows for robust validation of the SM $t\bar{t}$ background modeling prior to the final fit.

Results

Observation: No significant deviation from the Standard Model prediction was observed across the entire probed invariant mass spectrum ($0.4$ TeV to $5.0$ TeV). The data is consistent with the background-only hypothesis, with a goodness-of-fit probability of 95%.
Exclusion Limits: Upper limits at the 95% confidence level (CL) were set on the production cross-section times branching ratio ( $\sigma \times \mathcal{B}$ $σ \times B$ ) for the benchmark models:
- $Z'_{TC2}$ (Spin-1, narrow width): Excluded masses up to 4.2 TeV (3% width) and 3.65 TeV (1.2% width). For $m_{t\bar{t}} < 1.5$ TeV, these are the most stringent limits to date.
- $G_{KK}$ (Spin-2, narrow width): Excluded masses up to 1.3 TeV. This represents the strongest limit to date in the $t\bar{t}$ decay channel, though diboson searches provide stronger mass exclusions.
- $g_{KK}$ (Spin-1, broad width, $\Gamma/m = 30\%$ ): Excluded masses up to 4.1 TeV, setting the strongest limits to date for this benchmark model.

Significance
The paper claims that these results enhance existing exclusion limits across a broad range of models predicting heavy resonances decaying into top quark pairs. While the fully-hadronic ATLAS search remains more sensitive for $Z'_{TC2}$ masses above 4 TeV, the current analysis provides complementary sensitivity in alternative decay channels and establishes the most stringent constraints to date for $g_{KK}$ and $G_{KK}$ signals in the $t\bar{t}$ channel. The inclusion of the dilepton channel and the full Run 2 dataset marks a significant step in the comprehensive search for new physics in the top-quark sector.

More like this