⚛️ high-energy experiments

Search for compressed electroweak SUSY production in final states with two $\tau$ -leptons in $\sqrt{s} = 13$ and $13.6 $TeV$ pp$ collisions with the ATLAS detector

Using 140 fb⁻¹ of 13 TeV and 53 fb⁻¹ of 13.6 TeV proton-proton collision data, the ATLAS collaboration searched for compressed electroweak SUSY scenarios involving $\tau$ -sleptons, charginos, and neutralinos in final states with two $\tau$ -leptons, setting mass exclusion limits up to 200 GeV for electroweakinos while observing no significant deviations from the Standard Model that would allow for the exclusion of direct $\tau$ -slepton production when combining all channels.

Original authors: ATLAS Collaboration

Published 2026-09-30

📖 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 we can see and measure, yet much of what makes up the cosmos remains hidden. Scientists have long suspected that the visible world is only a small fraction of reality, with a vast, invisible sector of particles waiting to be discovered. This hidden sector is predicted by a theory called supersymmetry, which suggests that every known particle has a heavier, unseen partner. If these partners exist, they could explain why the universe has the mass it does and might even account for the mysterious dark matter that holds galaxies together. The challenge is that these partners, if they exist, are likely very heavy or hide in ways that make them incredibly difficult to spot. They might decay into other particles so quickly that they vanish before we can see them, or they might be so similar in mass to their decay products that the energy difference is too small to notice.

A researcher using the ATLAS detector at the Large Hadron Collider has taken a fresh look at this problem, specifically hunting for a type of hidden particle known as a tau-slepton. These are the supersymmetric partners of the tau lepton, a heavy cousin of the electron. The researcher focused on a tricky scenario where the tau-slepton is only slightly heavier than the lightest stable particle in the supersymmetric family. In this "compressed" situation, the energy released when the tau-slepton decays is very low, making the resulting particles move slowly and blend in with the background noise of ordinary particle collisions. To find them, the researcher looked for events where a high-energy jet of particles, created by the collision, gives a sudden push to the entire system, making the hidden particles easier to spot. They analyzed data from 140 fb⁻¹ and 53 fb⁻¹ of proton-proton collisions, combining records from two different periods of operation to maximize their chances of seeing something new.

The search involved looking for specific signatures in the debris of these collisions, particularly events containing tau leptons that decay into other particles. The researcher split their investigation into several different approaches to cover all possible ways these particles could appear. Some searches looked for collisions with no extra jets, while others focused on events where a single high-energy jet was present to provide that necessary boost. They used advanced computer algorithms to sift through the data, distinguishing between the rare signals of new physics and the overwhelming flood of ordinary events. The analysis was rigorous, comparing the observed data against detailed simulations of what should happen if only known particles existed. The researcher also paid close attention to the "fake" signals that occur when ordinary jets are mistaken for tau leptons, using data-driven methods to ensure these errors were accounted for.

After examining the vast amount of data, the results showed that the universe, at least in the energy ranges explored, behaves generally as the Standard Model of physics predicts, although several deviations with local significances of 2.1–2.4σ were observed. These deviations weaken the observed exclusion limits. While there were a few small fluctuations in the data that stood out slightly from the average, none were large enough to be considered a discovery. These small deviations were consistent with random statistical variations. Consequently, the researcher was able to set strict limits on where these hidden particles could be hiding in specific channels, though no exclusion limits were set for direct tau-slepton production when statistically combining all considered channels. In the ISR channel alone, they ruled out the existence of tau-sleptons with masses up to 115 gigaelectronvolts if the mass difference between the slepton and the lightest partner was 30 gigaelectronvolts. Similarly, they excluded charginos, another type of supersymmetric particle, with masses up to 175 gigaelectronvolts under the same conditions. For scenarios involving pairs of charginos or mixtures with neutralinos, the limits extended to masses of 200 gigaelectronvolts.

The study did not find the new particles it was looking for, but it significantly narrowed the search space. By pushing the boundaries of what is known, the researcher has eliminated a wide range of possibilities for how these particles might exist. The results are particularly important because they cover a region of mass and energy that was previously difficult to explore due to the subtle nature of the signals. The absence of a discovery does not mean the theory is wrong, but it does mean that if these particles exist, they must be heavier or behave in ways that are even more elusive than the models tested in this study. The work demonstrates the power of combining large datasets with sophisticated analysis techniques to probe the deepest layers of reality, bringing us closer to understanding the full composition of the universe.

Technical Summary: Search for Compressed Electroweak SUSY Production in Final States with Two $\tau$ -Leptons

Problem and Motivation
This paper presents a search for the production of supersymmetric (SUSY) particles in the electroweak sector, specifically targeting scenarios with "compressed mass spectra." In these scenarios, the mass difference ( $\Delta m$ ) between the produced SUSY particles (such as $\tau$ -sleptons $\tilde{\tau}$ , charginos $\tilde{\chi}^{\pm}_1$ , and neutralinos $\tilde{\chi}^0_2$ ) and the Lightest Supersymmetric Particle (LSP, $\tilde{\chi}^0_1$ ) is small. Such small mass splittings result in soft decay products that are difficult to reconstruct and trigger on, leading to reduced sensitivity in previous searches. The analysis focuses on final states containing at least one hadronically decaying $\tau$ -lepton, motivated by dark matter phenomenology where light $\tau$ -sleptons can satisfy relic density constraints. The study aims to probe mass splittings $\Delta m < 50$ GeV, a region largely unexplored due to challenges in soft $\tau$ reconstruction and large Standard Model (SM) backgrounds.

Methodology
The analysis utilizes proton-proton collision data collected by the ATLAS detector at $\sqrt{s}=13$ TeV (2015–2018, 140 fb $^{-1}$ ) and $\sqrt{s}=13.6$ TeV (2022–2023, 53 fb $^{-1}$ ). The search is divided into four distinct channels targeting different production mechanisms:

Direct $\tilde{\tau}$ production (Stau NoISR): Targets $\tilde{\tau}\tilde{\tau}$ pairs without initial-state radiation (ISR), relying on single-lepton triggers.
Direct $\tilde{\tau}$ production with ISR (Stau ISR): Targets $\tilde{\tau}\tilde{\tau}$ pairs produced with an ISR jet to boost the system, utilizing missing transverse momentum ( $E^{\text{miss}}_T$ ) triggers.
Intermediate $\tilde{\tau}$ channels: Targets chargino pair ( $\tilde{\chi}^+_1 \tilde{\chi}^-_1$ ) and chargino-neutralino ( $\tilde{\chi}^{\pm}_1 \tilde{\chi}^0_2$ ) production, where decays proceed via intermediate $\tilde{\tau}$ or $\tilde{\nu}_{\tau}$ states.

Analysis Strategy:

Event Reconstruction: The analysis employs harmonized reconstruction algorithms for Run 2 and Run 3, including dedicated low- $p_T$ calibrations for $\tau$ -leptons. Hadronic $\tau$ candidates are identified using Recurrent Neural Networks (RNNs) to suppress jet fakes.
Signal Discrimination:
- For Direct $\tilde{\tau}$ searches, a Multi-Class Deep Neural Network (DNN) is trained to distinguish signal from six background classes (Top, $Z(\tau\tau)$ , Diboson, Higgs, Fake- $\tau$ , and Signal).
- For Intermediate $\tilde{\tau}$ searches, Binary Boosted Decision Trees (BDT) are used to separate signal from background, leveraging more pronounced kinematic differences.
- A Cut-and-Count analysis is also performed for the Stau ISR channel as a cross-check and to provide re-interpretable results.
Background Estimation:
- True $\tau$ backgrounds (Top, $Z(\tau\tau)$ , Diboson) are modeled using Monte Carlo (MC) simulations, normalized to data in dedicated Control Regions (CRs).
- Fake $\tau$ backgrounds (jets misidentified as $\tau$ s) are estimated using a data-driven "fake factor" method, validated in Validation Regions (VRs).
Statistical Analysis: A binned likelihood fit is performed using the pyhf framework. Fits are conducted for background-only (to validate modeling) and signal-plus-background (to set exclusion limits) hypotheses. Run 2 and Run 3 data are combined with separate normalization factors to account for potential setup differences.

Key Contributions

Extended Sensitivity: This is the first ATLAS search with sensitivity to compressed prompt $\tau$ -slepton and electroweakino scenarios with $\Delta m < 50$ GeV.
ISR Utilization: The inclusion of ISR jet topologies significantly enhances sensitivity to compressed spectra by providing a Lorentz boost to the SUSY system, making soft decay products more detectable.
Advanced Techniques: The application of machine learning (DNN and BDT) specifically tuned for soft $\tau$ kinematics allows for better signal-background separation than traditional cut-and-count methods in this kinematic regime.
Dual Approach: The combination of ML-based analyses with a dedicated cut-and-count cross-check provides robustness and interpretability.

Results

Observation: The observed data are generally consistent with Standard Model predictions. However, several local deviations are observed with significances ranging from $2.1\sigma$ to $2.4\sigma$ (e.g., a $2.3\sigma$ excess in the Run 2 Stau NoISR 1-jet SR and a $2.4\sigma$ deficit in the Run 3 Intermediate $\tilde{\chi}^{\pm}_1 \tilde{\chi}^0_2$ LH SR). These fluctuations weaken the observed exclusion limits compared to expectations.
Direct $\tilde{\tau}$ Production:
- When combining Stau ISR and Stau NoISR channels, no exclusion limits are set at 95% Confidence Level (CL) due to the observed excesses.
- In the Stau ISR channel alone, $\tau$ -sleptons with masses up to 115 GeV are excluded for a mass splitting $\Delta m(\tilde{\tau}, \tilde{\chi}^0_1) = 30$ GeV.
Intermediate $\tilde{\tau}$ (Electroweakino) Production:
- For pure chargino pair production ( $\tilde{\chi}^+_1 \tilde{\chi}^-_1$ ), chargino masses up to 175 GeV are excluded for $\Delta m(\tilde{\chi}^{\pm}_1, \tilde{\chi}^0_1) = 30$ GeV.
- For mass-degenerate chargino pair and chargino-neutralino production, chargino masses up to 200 GeV are excluded for the same mass splitting.
- These limits significantly extend previous results in the compressed region.
Model-Independent Limits: Upper limits on the visible cross-section for new physics in the signal regions are set, ranging from 0.02 to 0.25 fb.

Significance
The paper claims that these results represent the most stringent limits to date for compressed electroweak SUSY scenarios involving $\tau$ -leptons. By utilizing the full Run 2 dataset and the initial Run 3 data, along with improved trigger strategies and machine learning techniques, the analysis successfully probes the challenging region of small mass splittings ( $\Delta m < 50$ GeV). The work demonstrates that with dedicated calibration and advanced analysis methods, sensitivity to compressed spectra can be achieved even with significant SM backgrounds, effectively closing the gap left by previous searches focused on non-compressed final states.

More like this