⚛️ high-energy experiments

Search for high-mass dilepton resonances in $pp$ collisions at $\sqrt{s}=13.6$ TeV combined with 13 TeV results using the ATLAS detector

The ATLAS collaboration presents a search for high-mass spin-1 dilepton resonances using a combined dataset of 13 TeV and 13.6 TeV proton-proton collisions, finding no significant deviations from the Standard Model and setting the most stringent limits to date on various $Z'$ boson models, with mass exclusion limits reaching up to 5.5 TeV.

Original authors: ATLAS Collaboration

Published 2026-07-31

📖 3 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

Imagine the universe as a giant, invisible ocean of energy. For decades, scientists have been trying to map this ocean, looking for islands that don't appear on their maps. These "islands" are new particles—tiny, heavy building blocks of matter that might explain why gravity is so weak or why the universe is made of stuff at all. To find them, physicists build massive machines called particle accelerators, which are like super-powered slingshots. They smash tiny particles together at speeds close to the speed of light, creating a chaotic explosion of energy. In that flash, the energy can briefly turn into heavy, new particles that usually don't exist in our everyday world. By studying the debris from these crashes, scientists hope to spot a "ghost" particle that hints at a deeper, hidden layer of reality. This paper is about one of the most famous ghost-hunting teams, the ATLAS collaboration, using the world's biggest slingshot, the Large Hadron Collider, to scan the debris for a specific type of ghost: a heavy messenger particle that might carry a new kind of force.

The ATLAS team recently decided to take a closer look at a very specific type of crash debris: pairs of electrons or pairs of muons (which are like heavy cousins of electrons). They call these "dileptons." If a new, heavy particle existed, it would likely decay into these pairs, creating a sudden, sharp spike in the number of events at a specific mass—like a sudden, loud note in a quiet song. The scientists wanted to see if they could hear that note.

To do this, they combined two huge piles of data. First, they looked at 165 units of data (called femtobarns, or fb⁻¹) collected between 2022 and 2024, when the collider was running at a record-breaking energy of 13.6 TeV. Then, they mixed this with 140 fb⁻¹ of data from 2015 to 2018, when the machine was running at 13 TeV. It's like listening to a recording of a quiet room from last year and combining it with a louder recording from this year to see if you can finally hear a whisper that was previously drowned out. They used a clever mathematical trick to draw a smooth curve through the "background noise" of ordinary particle collisions. If a new particle existed, the real data would jump up above this smooth curve, creating a distinct bump.

The result? The curve stayed smooth. The data followed the background noise perfectly, with no sudden spikes or bumps. The team looked for these signals across a massive range of masses, from 0.5 TeV all the way up to 7.1 TeV. They found nothing. No new heavy messenger particles were hiding in the data.

Because they didn't find the particles, the team didn't just throw up their hands; they used the silence to set strict boundaries. They calculated that if these specific types of new particles did exist, they would have to be heavier than certain limits. For the most popular theoretical models, they ruled out any such particle weighing less than 5.5 TeV. For other specific models, the limits were 5.1 TeV and 4.8 TeV. In other words, if these particles are out there, they are hiding deeper in the "heavy" zone than ever before. The authors state that these are the most stringent limits to date, meaning they have pushed the search further than anyone has before, but the mystery of whether these particles exist remains unsolved. They are still out there, perhaps, but if they are, they are heavier and harder to find than the scientists had hoped.

Technical Summary: Search for High-Mass Dilepton Resonances in $pp$ Collisions at $\sqrt{s}=13.6$ TeV Combined with 13 TeV Results

Problem and Motivation
This paper presents a search for new heavy resonances decaying into high-mass dilepton pairs ($ee$ and $\mu\mu$ ) using proton-proton collision data recorded by the ATLAS detector at the Large Hadron Collider (LHC). While the Standard Model (SM) has been extensively validated, theoretical extensions addressing questions such as the unification of forces (Grand Unified Theories, GUTs) or the hierarchy problem (extra dimensions) predict the existence of new spin-1 vector bosons, denoted as $Z'$ . These particles would manifest as resonant local excesses over the smoothly falling SM background in the dilepton invariant mass spectrum. Previous searches by ATLAS and CMS at $\sqrt{s}=13$ TeV have set stringent limits, but the increased center-of-mass energy of Run 3 ( $\sqrt{s}=13.6$ TeV) and the accumulation of additional integrated luminosity offer enhanced sensitivity to higher mass scales and increased production cross-sections.

Methodology
The analysis utilizes two distinct datasets:

Run 3 Data: 165 fb $^{-1}$ of data recorded between 2022 and 2024 at $\sqrt{s}=13.6$ TeV.
Run 2 Data: 140 fb $^{-1}$ of data recorded between 2015 and 2018 at $\sqrt{s}=13$ TeV (rescaled from the preliminary 139 fb $^{-1}$ used in previous publications).

Event Selection and Reconstruction:

Channels: The search targets same-flavor dilepton final states ($ee$ and $\mu\mu$ ).
Triggers: Events are selected using unprescaled triggers requiring two electrons ( $p_T > 24$ GeV) or single muons ( $p_T > 24$ GeV and $50$ GeV).
Object Quality: Electrons and muons are required to have $p_T > 30$ GeV and $|\eta| < 2.5$ (excluding the transition region for electrons). Strict identification and isolation criteria are applied to suppress background from misidentified jets and non-prompt leptons.
Kinematics: The dilepton invariant mass ( $m_{\ell\ell}$ ) is required to be $> 130$ GeV to exclude the $Z$ boson peak.

Signal and Background Modelling:

Background: The dominant background is the irreducible Drell–Yan (DY) process. Other contributions (top-quark pair, single top, diboson) are estimated via Monte Carlo (MC) simulation. Non-prompt and fake backgrounds (e.g., from multijets) in the electron channel are estimated using a data-driven matrix method.
Fit Function: The background shape is modeled using a functional form fitted to the observed data: $f(x) = f_{BW,Z}(m) \times (1-x^\alpha)^{p_1} x^{-\sum p_i (\ln x)^{i-2}}$ , where $x = m/\sqrt{s}$ . The parameters $\alpha$ and the number of free parameters $N$ are optimized using spurious signal tests and F-tests on Asimov datasets to ensure the function does not mimic a signal. For Run 3, both channels use $\alpha=1$ and $N=5$ .
Signal: Signal templates for generic spin-1 resonances are constructed by reweighting NLO DY MC samples with NNLO QCD corrections. The signal shape is a convolution of a Breit–Wigner distribution (with varying widths) and the detector resolution.
Systematic Uncertainties: Major uncertainties include background modelling (spurious signal), signal acceptance (reconstruction, isolation, trigger), and theoretical uncertainties (PDFs, $\alpha_s$ , scale variations). Experimental uncertainties are treated as uncorrelated between Run 2 and Run 3, while signal modelling uncertainties are correlated.

Statistical Interpretation
A binned profile likelihood fit is performed on the $m_{\ell\ell}$ spectra (0.5–7.1 TeV) to test the background-only hypothesis against signal-plus-background hypotheses. The significance of any local excess is quantified using the $p$ -value, with global significance calculated to account for the "look-elsewhere" effect. In the absence of a significant global excess, 95% confidence level (CL) upper limits are set on the fiducial cross-section times branching ratio ( $\sigma_{fid} \times \mathcal{B}$ ) using the modified frequentist $CL_s$ method.

Key Results

Observation: No significant global deviation from the SM background expectation was observed in the Run 3 data alone or in the combined Run 2 and Run 3 dataset.
- The largest local excess in the Run 3 data was $2.8\sigma$ at 0.8 TeV in the $ee$ channel and $3.2\sigma$ at 1.3 TeV in the $\mu\mu$ channel. However, the global significances for these fluctuations were consistent with zero (or low, e.g., $< 2\sigma$ ).
- The highest mass events observed were 3.13 TeV ($ee$) and 3.96 TeV ( $\mu\mu$ ).
Exclusion Limits:
- Model-Independent: Upper limits on $\sigma_{fid} \times \mathcal{B}$ were set for various resonance widths (0% to 10%).
- Benchmark Models: Mass limits were extracted for specific $Z'$ $Z^{'}$ models:
  - $Z'_{SSM}$ (Sequential Standard Model, $\Gamma/m \approx 3\%$ ): Observed limit of 5.5 TeV (Expected: 5.5 TeV) in the combined channel.
  - $Z'_{\chi}$ ( $E_6$ model, $\Gamma/m \approx 1.2\%$ ): Observed limit of 5.1 TeV (Expected: 5.1 TeV).
  - $Z'_{\psi}$ ( $E_6$ model, $\Gamma/m \approx 0.5\%$ ): Observed limit of 4.8 TeV (Expected: 4.9 TeV).
- The combination of Run 2 and Run 3 data resulted in mass limits that are approximately 25% to 50% more stringent than the Run 2 results alone.

Significance and Claims
The authors state that these results represent the most stringent limits to date on high-mass dilepton resonances for the considered $Z'$ models. By combining the increased center-of-mass energy of Run 3 (which increases the production cross-section for heavy resonances) with a dataset more than double the size of the previous Run 2 analysis, the search extends the exclusion reach significantly. The paper emphasizes that the results are derived using a model-independent fitting strategy for the background, ensuring robustness against specific signal assumptions, while providing specific interpretations for well-motivated theoretical benchmarks. No evidence for new physics was found, thereby constraining the parameter space for extended gauge symmetry models and extra-dimensional scenarios.

More like this