Searches for strong production of supersymmetric particles with the ATLAS detector

This paper presents the latest ATLAS results from LHC collisions at 13 and 13.6 TeV on searches for the strong production of supersymmetric particles, specifically targeting gluinos and squarks (including stops) across various decay modes to address naturalness and explore beyond-minimal scenarios.

The Gist

Imagine the universe as a giant, high-speed racetrack called the Large Hadron Collider (LHC). Inside this track, scientists smash tiny particles together at nearly the speed of light to see what happens. The ATLAS detector is like a massive, ultra-high-speed camera trying to capture every detail of these collisions.

The paper you're reading is a report from a team of scientists (the ATLAS Collaboration) who are looking for "ghosts" in the machine. These ghosts are theoretical particles called Supersymmetric particles (or "sparticles").

The Big Idea: The Shadow World

According to our current best map of the universe (the Standard Model), every known particle has a "shadow twin" that we haven't found yet.

• If you have a heavy quark (a building block of matter), its shadow twin is a squark.
• If you have a gluon (the glue holding atoms together), its shadow twin is a gluino.

The scientists believe these shadow twins might solve big mysteries, like why the universe has "dark matter" (the invisible stuff holding galaxies together). The theory suggests that if these twins exist, the lightest one is stable and could be the dark matter we are looking for.

The Hunt: Four Different Searches

The paper describes four specific "hunts" the scientists went on, using data from collisions at two different energy levels (like driving the racetrack at 13 and 13.6 on the speedometer). They were looking for specific combinations of particles that would appear if these shadow twins were created and then immediately fell apart.

Here is a simple breakdown of the four searches:

1. The "Heavy Top" Hunt (Search 1)

• The Target: They looked for pairs of "stop squarks" (the shadow twin of the top quark, the heaviest known particle).
• The Scenario: Imagine two heavy boxes (stop squarks) crashing into each other and breaking open. Inside, they expect to find a pair of top quarks and two invisible "ghosts" (the dark matter candidates).
• The Trick: They looked for two different ways the boxes could break:
  • The "Resolved" way: The pieces fly apart slowly enough to be seen clearly as separate jets of energy.
  • The "Boosted" way: The pieces fly apart so fast they smash together into a single, giant blob of energy.
• The Result: They didn't find the boxes. They set a rule: "If these stop squarks exist, they must be heavier than 1,230 GeV." (Think of this as saying, "If the ghost exists, it must be heavier than a blue whale.")

2. The "Charm Switch" Hunt (Search 2)

• The Target: They looked for stop squarks that might turn into "charm quarks" (a lighter cousin of the top quark) instead of top quarks. This is a bit like looking for a shape-shifter.
• The Scenario: They looked for a specific signature: one heavy jet (from a top quark) and one charm jet, with no visible electrons or muons, just missing energy.
• The Result: No shape-shifters found. They ruled out stop squarks up to 800 GeV in most cases, and up to 600 GeV if the particles were very close in weight (a "compressed" scenario).

3. The "Double Charm" Hunt (Search 3)

• The Target: They looked for pairs of stop squarks or charm squarks that both turn into charm quarks.
• The Scenario: This is like looking for a pair of twins that both turn into the same lighter sibling. They looked for two charm jets and missing energy.
• The Result: Still no ghosts. They pushed the limit even further, saying these particles must be heavier than about 900 GeV if they exist.

4. The "Gluino & Squark" Hunt (Search 4)

• The Target: This was the biggest net, looking for gluinos (glue shadows) and other squarks that decay into "tau leptons" (heavy cousins of electrons).
• The Strategy: They used two different detective tools:
  • Cut-and-Count: A traditional method of setting strict rules (e.g., "Only count events with energy over X").
  • Machine Learning: An AI brain trained to spot subtle patterns that humans might miss, sorting events into "signal" or "background noise."
• The Result: The AI and the traditional method both agreed: No gluinos or squarks found. They set the strictest limits yet, saying gluinos must be heavier than 2.25 TeV (over 2,000 times the mass of a proton) and squarks heavier than 1.7 TeV.

The Bottom Line

The paper is essentially a "Wanted" poster that says: "We looked everywhere, we used our best cameras and smartest AI, but we didn't find any of these supersymmetric particles."

Because they didn't find them, they didn't discover new physics in this specific run. Instead, they drew a line in the sand. They told the theoretical physicists: "If these particles exist, they are heavier than we thought. You need to update your maps to look for heavier ghosts."

In short: The hunt continues, but the "easy" ghosts (the light ones) have been ruled out.

Technical Summary

Technical Summary: Searches for Strong Production of Supersymmetric Particles with the ATLAS Detector

Problem and Motivation
Despite the precision of the Standard Model (SM), unresolved issues such as the hierarchy problem, the lack of fundamental interaction unification, and the absence of a dark matter candidate persist. Supersymmetry (SUSY) offers a theoretical framework addressing these issues by positing a superpartner for every SM particle. In the Minimal Supersymmetric Standard Model (MSSM), the conservation of R-parity ensures the stability of the Lightest Supersymmetric Particle (LSP), typically the lightest neutralino ($\tilde{\chi}^0_1$), making it a viable dark matter candidate.

Naturalness arguments suggest that supersymmetric partners of gluons (gluinos, $\tilde{g}$) and third-generation quarks (squarks, specifically the stop $\tilde{t}_1$) should be light enough to be produced at the Large Hadron Collider (LHC). While classical MSSM scenarios face increasingly stringent mass bounds, variations involving non-minimal particle content and flavor structures remain of significant interest. This paper presents a compilation of recent searches by the ATLAS Collaboration targeting the strong production of these particles.

Methodology and Data Sets
The analyses utilize proton-proton collision data collected by the ATLAS detector at center-of-mass energies of $\sqrt{s} = 13$ TeV (Run 2, 2015–2018, $\sim$139–140 fb$^{-1}$) and $\sqrt{s} = 13.6$ TeV (partial Run 3, 2022–2023, 51.8 fb$^{-1}$). The searches focus on final states characterized by jets, leptons (including hadronically decaying $\tau$-leptons), and significant missing transverse energy ($E^{\text{miss}}_T$), which signals the presence of stable LSPs.

Four distinct search strategies are detailed:

1. Stop Pair Production ($\tilde{t}_1\tilde{t}_1 \to t\bar{t}\tilde{\chi}^0_1\tilde{\chi}^0_1$):

   • Decay Modes: Covers both two-body decays (large mass splitting, $\Delta m > m_t$) and three-body decays (intermediate mass splitting, $m_W < \Delta m < m_t$).
   • Final State: One isolated lepton, jets, and $E^{\text{miss}}_T$.
   • Strategy: Events are categorized into "resolved" (low-$p_T$ tops, $E^{\text{miss}}_T > 230$ GeV) and "boosted" (high-$p_T$ tops, large-R jets) regions. Signal regions (SRs) are defined by b-jet multiplicity and top-tagging. A neural network (NN) is employed to enhance signal/background discrimination. Backgrounds ($t\bar{t}$, single-top, $W$+jets) are estimated using control regions (CRs) and kinematic variables like transverse mass ($m_T$) and lepton charge.

2. Stop Pair Production with Flavor Violation ($\tilde{t}_1\tilde{t}_1 \to tc\tilde{\chi}^0_1\tilde{\chi}^0_1$):

   • Model: Investigates non-minimal flavor violation (MFV) where $\tilde{t}_1$ decays equally to $t\tilde{\chi}^0_1$ or $c\tilde{\chi}^0_1$.
   • Final State: One b-tagged jet, one c-tagged jet, no leptons, and large $E^{\text{miss}}_T$.
   • Strategy: Utilizes a specialized c-tagging algorithm (based on DL1r) to identify charm jets. The phase space is divided into "bulk," "intermediate" (boosted/resolved), and "compressed" regions. In compressed scenarios, initial state radiation (ISR) jets and NN training are used to enhance sensitivity.

3. Stop and Charm-Squark Production ($\tilde{t}_1\tilde{t}_1, \tilde{c}_1\tilde{c}_1 \to cc\tilde{\chi}^0_1\tilde{\chi}^0_1$):

   • Model: Targets minimal flavor violation where charm-squarks ($\tilde{c}_1$) may be lighter than other squarks.
   • Final State: Two c-tagged jets, no leptons, and $E^{\text{miss}}_T$.
   • Strategy: Investigates high-mass ($m \gtrsim 600$ GeV) and compressed ($\Delta m < 175$ GeV) regions. The compressed region relies on high-momentum ISR jets and recursive jigsaw reconstruction (RJR), while the leading jet is required not to be c-tagged to suppress background.

4. Gluino and Squark Pair Production ($\tilde{g}\tilde{g}, \tilde{q}\tilde{q}$):

   • Decay Modes: Involves decays to $\tau$-sleptons ($\tilde{\tau}$) and $\tau$-sneutrinos ($\tilde{\nu}_\tau$).
   • Final States: Three channels based on $\tau$ and lepton content: 1$\tau$0lep, 1$\tau$1lep, and 2$\tau$.
   • Strategy: Two approaches are compared: a traditional "cut-and-count" method using $E^{\text{miss}}_T$ variables and a machine-learning approach using Boosted Decision Trees (BDTs) for multiclass classification against backgrounds ($W/Z$+jets, diboson, top, fake-$\tau$).

Key Results
No significant excess of events over the expected SM background was observed in any of the analyzed channels. Consequently, 95% confidence level (CL) exclusion limits were established:

• Stop Pairs ($t\bar{t} + E^{\text{miss}}_T$): The single-lepton analysis improves sensitivity in small mass-splitting regions. Combined with zero-lepton results, top-squark masses are excluded up to 1230 GeV (for massless $\tilde{\chi}^0_1$) and neutralino masses up to 570 GeV.
• Flavor-Violating Stop Pairs ($tc + E^{\text{miss}}_T$): Top-squark masses are excluded up to 800 GeV (massless $\tilde{\chi}^0_1$) and 600 GeV in the compressed scenario. This result is also reinterpreted to exclude new mediators coupling dark matter to SM quarks up to 1.2 TeV.
• Stop/Charm-Squark Pairs ($cc + E^{\text{miss}}_T$): Top/charm-squark masses are excluded up to approximately 900 GeV (massless $\tilde{\chi}^0_1$), representing an improvement of roughly 100 GeV over previous ATLAS limits.
• Gluino and Squark Pairs:
  • Gluino masses are excluded up to 2.25 TeV (with neutralinos up to 1.35 TeV for $m_{\tilde{g}} \approx 2$ TeV).
  • Squark masses are excluded up to 1.7 TeV (with neutralinos up to 0.85 TeV).
  • The machine-learning approach demonstrates superior sensitivity for higher gluino/squark masses and larger neutralino masses, while the cut-and-count method performs better in the compressed phase-space region due to higher event yields and dedicated selections.

Significance and Claims
The paper claims to present the latest results from the ATLAS experiment, extending the search for strongly produced supersymmetric particles into new decay modes and kinematic regimes. The significance lies in:

1. Broadening the Search Scope: Moving beyond classical MSSM scenarios to include non-minimal flavor violation and compressed spectra.
2. Technological Advancement: The successful implementation and optimization of c-tagging algorithms and the application of machine learning (NNs and BDTs) to improve signal-background discrimination in complex final states.
3. Updated Constraints: Providing the most stringent exclusion limits to date for specific SUSY scenarios, particularly in the compressed mass regions and for flavor-violating decays, thereby narrowing the parameter space for viable supersymmetric models.

The authors conclude that while no evidence for SUSY was found, these results significantly refine the experimental boundaries for strong production of gluinos, stops, and charm-squarks at the LHC.