Search for pair-produced vector-like $T$-quarks decaying into $Ht$ final states in the lepton-plus-jets channel in $pp$ collisions at $\sqrt{s}$=13 TeV with the ATLAS detector

Using 139 fb$^{-1}$ of 13 TeV proton-proton collision data from the ATLAS detector, this study searches for pair-produced vector-like $T$-quarks decaying into Higgs-top final states in the lepton-plus-jets channel, finding no significant excess over Standard Model predictions and setting 95% confidence level lower mass limits of 1.40 to 1.66 TeV depending on the $T$-quark representation and branching fraction.

The Gist

The Big Picture: Hunting for "Heavy Twins"

Imagine the universe is a giant, high-speed racetrack (the Large Hadron Collider, or LHC). Physicists at CERN are like race officials who smash protons together at nearly the speed of light to see what happens. Usually, these crashes produce a predictable set of particles, like standard cars and motorcycles.

However, the Standard Model (our current rulebook for physics) has a few gaps. One big question is: Why is the Higgs boson (the particle that gives other particles mass) so light? To fix this "fine-tuning" problem, some theories suggest there are "heavy twins" of the top quark (the heaviest known particle). These are called Vector-Like T-quarks.

This paper is a report from the ATLAS experiment team saying: "We looked very hard for these heavy twins, but we didn't find them. However, we can now say with high confidence that if they do exist, they must be heavier than we previously thought."

The Strategy: The "Heavyweight Boxing Match"

Since these T-quarks are so heavy, they are hard to make. When they are made, they don't last long; they immediately break apart (decay) into other particles.

The team decided to look for a specific scenario:

1. The Pair: They are looking for two T-quarks being created at once (like a pair of heavyweight boxers entering the ring).
2. The Decay: At least one of them breaks down into a Higgs boson and a top quark.
3. The Trail: The Higgs boson then splits into two "bottom" quarks, and the top quark splits into a light particle (an electron or muon), a ghost-like neutrino, and another bottom quark.

The Analogy: Imagine you are trying to find a specific type of rare, heavy fruit (the T-quark) in a massive orchard. You know that when this fruit falls, it splits into a specific combination of seeds and juice. Instead of looking for the fruit itself, you are looking for the unique pile of seeds and juice it leaves behind.

The Detective Work: Sorting the Noise

The problem is that the orchard is full of regular fruit falling all the time (Standard Model background). The team had to filter out the noise to find the rare signal.

• The "Reclustering" Trick: When heavy particles decay, they move so fast that their debris (jets of particles) gets squished together. The team used a special technique called "variable-radius jets." Think of this like using a smart camera lens that automatically zooms in or out depending on how fast the object is moving, ensuring they capture the whole "debris pile" correctly, even when it's moving incredibly fast.
• The Neural Network (The AI Detective): They trained a computer brain (a neural network) to look at the shape, speed, and arrangement of these debris piles. It's like teaching a dog to sniff out a specific scent. The AI learned to distinguish between the messy, random debris of normal collisions and the clean, structured debris of a heavy T-quark decay.

The Results: "Not Found, But We Know Where They Aren't"

After analyzing 139 "inverse femtobarns" of data (which is a massive amount of collision data, equivalent to years of running the collider), the team found no evidence of these heavy T-quarks. The data matched the predictions for normal physics perfectly.

Because they didn't find them, they set a "fence" around where the T-quarks could be. They can now rule out T-quarks that are lighter than certain weights:

• If the T-quark is a "Singlet" (a specific type of particle), it must be heavier than 1.40 TeV.
• If it's a "Doublet," it must be heavier than 1.56 TeV.
• If it only decays into the Higgs and top quark (100% of the time), it must be heavier than 1.66 TeV.

The Metaphor: Imagine you are searching for a hidden treasure chest in a field. You dig up the whole field and find nothing. You can't say the chest doesn't exist, but you can say, "If the chest is there, it must be buried deeper than 10 feet, because we dug up everything above that." This paper digs deeper than anyone has before, pushing the "burial depth" limit further down.

Why This Matters

This is the most sensitive search of its kind to date. By using more data (139 fb⁻¹ vs. previous 36 fb⁻¹) and better AI tools, the ATLAS team has pushed the boundaries of our knowledge. They haven't found the "heavy twins," but by proving they aren't hiding in the lighter mass range, they are forcing physicists to rethink their theories or look for these particles at even higher energies in the future.

In short: The hunt for the heavy T-quark continues, but the search zone has been narrowed down significantly. If they are out there, they are heavier and harder to find than we hoped.

Technical Summary

Technical Summary: Search for Pair-Produced Vector-Like T-Quarks Decaying into Ht Final States

Problem and Motivation
The Standard Model (SM) of particle physics faces the hierarchy problem regarding the large disparity between the electroweak mass scale and the Planck scale. One proposed solution involves extending the SM with vector-like quarks (VLQs), which can counterbalance quadratically divergent radiative corrections to the Higgs boson mass. This paper presents a search for pair-produced up-type vector-like $T$-quarks, which are expected to couple preferentially to third-generation SM quarks. The analysis specifically targets the decay channel where at least one $T$-quark decays into a Higgs boson and a top quark ($T \to Ht$), with subsequent decays $H \to b\bar{b}$ and $t \to \ell \nu_\ell b$ (where $\ell = e, \mu$). The search is conducted in the lepton-plus-jets final state using proton-proton collision data at a center-of-mass energy of $\sqrt{s} = 13$ TeV.

Methodology
The analysis utilizes the full Run 2 dataset collected by the ATLAS detector at the Large Hadron Collider (LHC), corresponding to an integrated luminosity of 139 fb$^{-1}$.

• Event Reconstruction and Selection: The signal topology is characterized by exactly one isolated electron or muon, significant missing transverse momentum ($E_T^{\text{miss}}$), and multiple jets, including $b$-tagged jets. To handle the boosted nature of heavy resonances, the analysis employs variable-radius ($R_{\text{eff}}$) jets reconstructed via reclustering of small-$R$ jets. These are mass-tagged to identify candidates for hadronically decaying $W$, $H$, and top quarks. Leptonic top-quark candidates are reconstructed by combining a leptonic $W$-boson candidate with a $b$-tagged jet.
• Background Modeling and Reweighting: The dominant backgrounds are top-quark pair production ($t\bar{t}$), single-top, and $V$+jets processes. To address discrepancies between simulation and data, particularly in the modeling of $t\bar{t}$ and $V$+jets cross-sections at high jet multiplicities and transverse momenta, an iterative data-driven reweighting technique is applied. This corrects the normalization and shape of background distributions based on control regions enriched in $Z$+jets and $t\bar{t}$ events.
• Multivariate Analysis: A feed-forward neural network (NN) is employed to discriminate signal from background. The classifier utilizes 30 input variables, including kinematic properties (e.g., effective mass $m_{\text{eff}}$, transverse momenta of variable-$R$ jets) and topological features (e.g., angular separations, constituent multiplicities). The NN is trained on simulated signal and background samples.
• Event Categorization: Events are categorized into signal regions (SRs) and control regions (CRs) based on the NN output score ($h_{\text{NN}}$), the number of $b$-tagged jets (2, 3, or $\ge 4$), and the presence of $H$-tagged jets. Three SR categories are defined by NN score thresholds: High-NN (HNN), Mid-NN (MNN), and Low-NN (LNN). The LNN region, with lower signal purity but high background statistics, is used to constrain background normalizations in the statistical fit.
• Statistical Analysis: A binned profile-likelihood fit is performed on the effective mass ($m_{\text{eff}}$) distribution across all analysis regions. The fit includes nuisance parameters for systematic uncertainties (experimental and theoretical) and free normalization factors for specific background components ($t\bar{t}+c$ and $t\bar{t}+b$).

Key Contributions

• Improved Background Modeling: The implementation of a data-driven reweighting technique significantly improves the agreement between data and SM predictions for $t\bar{t}$ and $V$+jets backgrounds, particularly in kinematic regimes relevant to heavy resonance searches.
• Enhanced Signal Discrimination: The use of a neural network classifier, combined with explicit reconstruction of leptonic top-quark candidates and the exploitation of the back-to-back topology of pair-produced heavy particles, provides superior separation power compared to previous ATLAS searches in this channel.
• Comprehensive Scenario Testing: The search interprets results under three distinct benchmark scenarios: the SU(2) singlet, the SU(2) doublet, and a scenario where the branching fraction $B(T \to Ht)$ is set to 100%.

Results
No significant excess of events above the Standard Model prediction is observed in any of the analysis regions. Consequently, upper limits on the pair-production cross-section of vector-like $T$-quarks are set at the 95% confidence level (CL).

The observed lower mass limits for the $T$-quark are:

• 1.40 TeV for the SU(2) singlet representation.
• 1.56 TeV for the SU(2) doublet representation.
• 1.66 TeV for the scenario where $B(T \to Ht) = 100\%$.

These limits represent the most stringent constraints to date for the SU(2) doublet and the $B(T \to Ht) = 100\%$ scenarios, extending previous exclusion limits by 0.07 TeV and 0.16 TeV, respectively. For the SU(2) singlet scenario, this result improves upon the previous ATLAS limit by 0.04 TeV. The analysis is noted to be primarily limited by statistical uncertainties.

Significance
The paper claims that these results provide the most stringent exclusion limits currently available for vector-like $T$-quarks in the specified decay channels and gauge representations. By utilizing the full Run 2 dataset and advanced analysis techniques (reweighting and machine learning), the search effectively probes higher mass ranges than previous ATLAS analyses. The authors note that while the current search is statistically limited, the increased data volume expected from Run 3 and the High-Luminosity LHC offers significant potential for extending sensitivity to even higher masses. The results are made available for reinterpretation via HEPData, including event yields, systematic uncertainties, and the statistical workspace.