⚛️ high-energy experiments

Search for low-mass resonances decaying into two jets and produced in association with a photon or a jet at $\sqrt{s}=13$ TeV with the ATLAS detector

Using the full Run 2 dataset from the ATLAS experiment at $\sqrt{s}=13$ TeV, this study searches for low-mass dijet resonances produced in association with a photon or a jet, finding no evidence of new physics and setting new upper limits on production cross-sections for masses between 200 and 650 GeV.

Original authors: ATLAS Collaboration

Published 2026-09-07

📖 5 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 visible world works, from the glow of a lightbulb to the fusion powering the sun. Yet, the map is incomplete. It cannot explain why the universe is filled with invisible dark matter, nor does it account for the imbalance between matter and antimatter that allowed our existence. To fill these gaps, scientists propose the existence of new, heavier particles that interact with the known ones but remain hidden in plain sight. One leading candidate is a hypothetical particle called a Z-prime boson. Think of it as a heavy, invisible cousin to the photon, capable of decaying into pairs of ordinary particles that fly apart at incredible speeds. Finding such a particle would be a direct window into the physics that lies beyond our current understanding.

To hunt for these elusive particles, researchers at the Large Hadron Collider smash protons together at nearly the speed of light, creating a shower of debris that detectors record in exquisite detail. In a recent study, the ATLAS collaboration, a team of thousands of scientists working with one of the world's largest detectors, turned their attention to a specific type of collision. They looked for a scenario where a new, heavy particle is created alongside a high-energy photon or a jet of particles. This strategy is crucial because it allows scientists to search for lighter versions of these new particles that might otherwise be missed. By focusing on events where a new particle recoils against a visible photon or jet, the team could lower the energy threshold of their search, effectively widening the net to catch particles with masses between 200 and 650 billion electron volts.

The researchers analyzed a massive dataset collected over four years, representing 140 inverse femtobarns of proton-proton collisions. They sorted these collisions into four distinct categories based on what was produced alongside the potential new particle. Two categories looked for a photon paired with two jets of particles, while the other two looked for a jet paired with two other jets. Within the jet-pair categories, they further divided the search into two groups: one that looked at all jets regardless of their internal composition, and another that specifically looked for jets containing heavy beauty quarks. This distinction was important because some theoretical models suggest that new particles might prefer to decay into these heavier quarks, and isolating them helps reduce the overwhelming background noise of ordinary particle collisions.

The core of the search involved reconstructing the mass of the two jets in each event. If a new particle had been created and decayed into those jets, the data would show a distinct spike, or a localized excess, at a specific mass value rising above the smooth, predictable curve of background events. The team used sophisticated statistical tools to compare the actual data against the expected background, which was modeled using a smooth mathematical function derived from the data itself. They scanned the entire mass range from 200 to 650 billion electron volts, looking for any sign of a bump that could not be explained by chance or known physics.

The result was a clear absence of any such signal. In every channel they examined, the data followed the smooth curve predicted by the Standard Model with remarkable precision. There were no spikes, no unexpected clusters of events, and no evidence of a new particle decaying into two jets in association with a photon or a jet. The most significant fluctuations they observed were small and consistent with random statistical variations, far too weak to claim a discovery. Consequently, the team did not find a new particle, but they did achieve something equally valuable: they set strict upper limits on how often such a particle could exist. They determined that if these Z-prime particles do exist, they must be produced less frequently than a certain threshold, effectively ruling out a wide range of theoretical models that predicted stronger signals.

This study extends the reach of previous searches, pushing the constraints on these hypothetical particles into a mass range that was previously difficult to probe with high precision. By combining the full dataset from the second run of the Large Hadron Collider with refined analysis techniques, including better methods for identifying heavy quarks and improved strategies for handling the complex combinatorics of multi-jet events, the team has tightened the noose around these theoretical possibilities. The findings confirm that within the mass range of 200 to 650 billion electron volts, no new resonances are hiding in the data. While the search for the missing pieces of the universe's puzzle continues, this work ensures that the search will have to look elsewhere or at higher energies, refining the path forward for the next generation of discovery.

Technical Summary: Search for Low-Mass Resonances Decaying into Two Jets in Association with a Photon or a Jet at $\sqrt{s}=13$ TeV

Problem and Motivation
The Standard Model (SM) of particle physics, while successful, fails to explain phenomena such as dark matter (DM) and the matter-antimatter asymmetry in the universe. A prominent class of Beyond the Standard Model (BSM) theories proposes the existence of a $Z'$ boson, a spin-1 vector mediator that couples to quarks but is "leptophobic" (does not couple to SM leptons), potentially serving as a portal to dark matter. Direct searches for $Z'$ bosons decaying into charged lepton pairs have set strong constraints, but these can be evaded if the $Z'$ is leptophobic. Consequently, searches for hadronically decaying $Z'$ resonances are essential.

While high-mass $Z'$ searches ( $>1$ TeV) are well-established using dijet invariant mass spectra, sensitivity to lower masses ($200-650$ GeV) is limited by trigger bandwidth constraints on single- and multijet triggers. To access this low-mass regime without trigger bias, this analysis employs a strategy utilizing high transverse momentum ( $p_T$ ) initial-state radiation (ISR). By requiring the production of a high- $p_T$ photon or jet that recoils against the $Z'$ , the analysis can trigger on the ISR object, allowing the reconstruction of lower-mass dijet resonances that would otherwise be inaccessible.

Methodology
The analysis utilizes the full Run 2 dataset of proton-proton collisions at $\sqrt{s}=13$ TeV collected by the ATLAS detector, corresponding to an integrated luminosity of $140 \text{ fb}^{-1}$ . The search targets four distinct channels based on the ISR particle type and the flavor composition of the $Z'$ decay products:

$\gamma jj$ : ISR is a photon; jets are inclusive.
$\gamma bb$ : ISR is a photon; both jets are $b$ -tagged.
$jjj$: ISR is a jet; jets are inclusive.
$jjb$: ISR is a jet; the two jets forming the resonance candidate are $b$ -tagged.

Object Reconstruction and Selection:
Jets are reconstructed using the anti- $k_t$ algorithm with $R=0.4$ and calibrated using particle flow objects. Photons are reconstructed from electromagnetic calorimeter deposits with tight identification and isolation criteria. $b$ -tagging is performed using the DL1r multivariate algorithm at a 77% efficiency working point.

Photon Channels: Events require a single photon with $p_T > 150$ GeV (fully efficient for the trigger) and two leading jets. A kinematic cut on the rapidity asymmetry $y^* = |y_1 - y_2|/2 < 0.825$ is applied to suppress background.
Trijet Channels: Events require at least three jets, with the leading jet $p_T > 475$ GeV. A combinatorial challenge arises in identifying which two jets correspond to the resonance decay. To ensure a smooth background mass spectrum, the analysis selects the jet pair with the minimum azimuthal separation ( $|\Delta \phi|$ ), provided they are not the two leading jets. This method was found to yield the smoothest low-mass spectrum in multijet simulations.

Background Estimation and Statistical Analysis:
The dominant backgrounds (multijet and single-photon production) are estimated using a data-driven approach. The dijet invariant mass ( $m_{jj}$ ) distribution is fitted with a parametric function: $f_B(x) = p_1(1-x)^{p_2}x^{p_3 + p_4 \ln(x) + p_5 \ln^2(x) + p_6 \ln^3(x)}$ , where $x = m_{jj}/\sqrt{s}$ . A 5-parameter fit is used, validated through spurious signal tests and signal injection tests to ensure the fit function does not mimic a signal or fail to extract one.
The search employs a binned profile likelihood fit to the $m_{jj}$ distribution. Signal templates are modeled either as Gaussian distributions (for generic narrow resonances) or using simulated $Z'$ samples (for the specific benchmark model). Systematic uncertainties, including luminosity, jet energy scale, photon identification, $b$ -tagging, and parton distribution functions (PDFs), are incorporated as nuisance parameters.

Key Contributions

Extension of Mass Reach: This analysis extends the search for dijet resonances to the mass range of 200 to 650 GeV, a region previously limited by trigger thresholds.
Inclusion of Trijet Channels: Unlike previous similar searches that focused primarily on photon-associated production, this work includes channels where the ISR is a jet ($jjj$ and $jjb$), significantly increasing the cross-section sensitivity despite the added combinatorial complexity.
Flavor-Specific Searches: The inclusion of $b$ -tagged channels ( $\gamma bb$ and $jjb$) allows for targeted searches for models where the $Z'$ couples preferentially to heavy quarks, leveraging the strong background suppression provided by $b$ -tagging.
Improved Combinatorial Strategy: The paper introduces and validates a specific jet-pairing algorithm based on minimum $|\Delta \phi|$ for the trijet channels, ensuring the background modeling remains robust and smooth in the low-mass region.

Results
No significant excess of events was observed relative to the Standard Model prediction in any of the four channels. The observed $m_{jj}$ distributions are well-described by the background-only fit, with global $p$ -values ranging from 0.10 to 0.42. The BumpHunter algorithm identified localized excesses with local significances of at most $1.9\sigma$ (at 348 GeV in the $\gamma jj$ channel), which are consistent with statistical fluctuations.

Upper limits were set at the 95% confidence level (CL) on the production cross-section times acceptance times branching ratio ( $\sigma \times A \times BR$ ) for:

$Z'$ Axial-Vector Mediator: Limits on the coupling to quarks ( $g_q$ ) were derived. The $jjj$ channel provided the strongest constraints, excluding $g_q$ values down to approximately 0.06–0.1 depending on the mass.
Generic Narrow Resonances: Limits were set for Gaussian-shaped signals with widths ranging from 5% to 15% of the resonance mass.

A combination of the $\gamma jj$ and $jjj$ channels was performed to further constrain $g_q$ , but the improvement over the $jjj$ channel alone was found to be negligible due to the lower sensitivity of the photon channel in this mass range.

Significance and Claims
The paper claims that this search improves the constraints on the $Z'$ -quark coupling ( $g_q$ ) by up to 50% compared to previous results using similar techniques, particularly in the low-mass region. By utilizing the full Run 2 dataset and incorporating trijet topologies, the analysis effectively probes the parameter space for leptophobic $Z'$ bosons and generic dijet resonances that were previously inaccessible due to trigger limitations. The results provide stringent constraints on simplified dark matter models where the $Z'$ acts as a mediator, specifically ruling out certain coupling strengths for masses between 200 and 650 GeV. The paper concludes that the data are consistent with the Standard Model, and no evidence for new physics in the form of low-mass dijet resonances associated with ISR was found.

More like this