Search for decays of the Higgs boson into pair-produced pseudoscalar particles decaying into $τ^+τ^-τ^+τ^-$ using $pp$ collisions at $\sqrt{s}=13$ TeV with the ATLAS detector

Using 140 fb⁻¹ of proton-proton collision data at 13 TeV collected by the ATLAS detector, this study searches for Higgs boson decays into pairs of low-mass pseudoscalars that subsequently decay into four tau leptons, finding no significant excess over Standard Model backgrounds and setting 95% confidence level upper limits on the branching fraction between 0.06 and 0.23 for pseudoscalar masses ranging from 15 to 60 GeV.

The Gist

Here is an explanation of the ATLAS paper, translated from "particle physics jargon" into everyday language with some creative analogies.

The Big Picture: Hunting for Ghosts in the Machine

Imagine the Large Hadron Collider (LHC) at CERN as the world's most powerful cosmic blender. It smashes protons together at nearly the speed of light, creating a chaotic soup of energy where new particles are born and die in a fraction of a second.

The Higgs boson is like a famous celebrity in this soup. We know it exists, but sometimes it might be doing something "secretive." The Standard Model (our current rulebook for physics) says the Higgs should decay in very specific, boring ways. But what if, occasionally, it sneaks off to a secret party and decays into something exotic?

This paper is the report from the ATLAS experiment (a giant, high-tech camera surrounding the blender) investigating one specific secret party: The Higgs turning into two invisible "ghosts," which then turn into four tau particles.

The Cast of Characters

1. The Higgs Boson ($H$): The celebrity. It's heavy and unstable.
2. The Pseudoscalar ($a$): The "ghost." This is a hypothetical, light particle that doesn't exist in our current rulebook. The theory is that the Higgs might split into a pair of these ghosts ($H \to aa$).
3. The Tau Lepton ($\tau$): The "messy party guest." When the ghosts decay, they turn into tau particles. Taus are heavy cousins of electrons, but they are very short-lived. They immediately explode into other things: either a jet of pions (a spray of particles) or a lepton (an electron or a muon) plus some invisible neutrinos.
4. The Background Noise: The universe is full of "fake" events. Just like trying to hear a whisper in a rock concert, the signal we are looking for is drowned out by billions of ordinary particle collisions (like Drell-Yan processes or top quarks).

The Strategy: The "Four-Leaf Clover" Search

The ATLAS team is looking for a very specific signature: Four tau particles coming from a single Higgs decay.

Since taus decay so fast, they don't leave a single track. They leave a "fingerprint" that looks like one of two things:

• The "Jet" ($\tau_{had}$): A small, tight spray of particles (like a firework that didn't fully explode).
• The "Lepton" ($\tau_{\ell}$): An electron or a muon (like a single, fast bullet).

The team looked for two specific "handshakes" (decay patterns) to catch the Higgs in the act:

1. The "2-and-2" Handshake: Two taus turn into jets, and two turn into electrons/muons.
2. The "3-and-1" Handshake: Three taus turn into electrons/muons, and one turns into a jet.

The Detective Work: Filtering the Noise

To find this needle in a haystack, the scientists had to build a very strict filter:

• The "No-Top" Rule: They threw away any event that looked like it came from a top quark (a very common heavy particle), because top quarks are the "loud rock band" drowning out the whisper.
• The "No-Z" Rule: They threw away events where an electron and muon pair had the exact mass of the Z boson (another common particle), because that's just background noise.
• The "Fake" Detector: Sometimes, a regular jet of particles looks exactly like a tau particle. This is called a "fake." The team used a clever statistical trick (the "Fake Factor" method) to estimate how many of their "tau" candidates were actually just regular jets pretending to be taus. They validated this by checking "control rooms" (validation regions) where they knew only fakes should exist.

The Results: The Silence is the Answer

After analyzing 140 femtobarns of data (which is like looking at 140 trillion trillion collisions), here is what they found:

• The "2-and-2" Room: They saw 0 events. The background prediction was almost 0. Perfect silence.
• The "3-and-1" Room: They saw 31 events. The background prediction was 28.
  • Analogy: Imagine you are looking for a specific type of rare bird. You expect to see 28 regular sparrows. You count 31 birds. Is that 3 extra rare birds? Or just a statistical fluke? In this case, the difference was small enough to be just random noise.

Conclusion: They found no evidence of the Higgs decaying into these exotic ghosts.

The Verdict: Setting the Limits

Even though they didn't find the ghosts, they didn't come up empty-handed. They set limits.

Think of it like a search for a lost key. If you search a whole house and don't find it, you can't say "it's not there." But you can say, "If the key was there, it would have to be hiding in a very specific, tiny spot that we didn't check."

In this paper, they calculated:

• If the Higgs does decay into these ghosts, it happens less than 6% to 23% of the time (depending on the mass of the ghost).
• This rules out a huge chunk of theoretical models that predicted this would happen much more often.

Why This Matters

This is like checking a map for a "X marks the spot."

• Before: Theories said, "Maybe the Higgs turns into these ghosts 50% of the time!"
• Now: The map says, "Nope, if it happens, it's less than 6%."

This forces physicists to rewrite their theories. If the "ghosts" exist, they must be much more elusive than we thought, or they might not exist at all in this mass range. It's a step forward in understanding the universe, even if the "ghosts" managed to stay hidden this time.

In short: The ATLAS team looked very hard for a weird, four-part decay of the Higgs boson. They didn't find it, but they successfully narrowed down the hiding spots, telling the rest of the physics world, "Keep looking, but not in this specific corner."

Technical Summary

Here is a detailed technical summary of the ATLAS paper CERN-EP-2026-026, titled "Search for decays of the Higgs boson into pair-produced pseudoscalar particles decaying into $\tau^+\tau^-\tau^+\tau^-$ using $pp$ collisions at $\sqrt{s}= 13$ TeV with the ATLAS detector."

1. Problem and Motivation

The Standard Model (SM) of particle physics, while successful, leaves several fundamental questions unanswered, such as the nature of dark matter and the hierarchy problem. Theoretical extensions, such as the Two-Higgs-Doublet Model extended by a singlet (2HDM+S) or the Next-to-Minimal Supersymmetric Standard Model (NMSSM), predict the existence of light, beyond-the-Standard-Model (BSM) pseudoscalar particles ($a$).

• The Scenario: If these pseudoscalars are lighter than half the mass of the Higgs boson ($m_a < m_H/2 \approx 62.5$ GeV), the Higgs boson can decay exotically into a pair of these particles: $H \to aa$.
• The Decay Chain: If the pseudoscalar $a$ couples preferentially to $\tau$-leptons (common in Type-III 2HDM+S models with large $\tan\beta$), the decay chain becomes $H \to aa \to 4\tau$.
• The Challenge: Previous ATLAS searches covered the low-mass range ($4 < m_a < 15$GeV) where the$\tau$-leptons are highly Lorentz-boosted and overlap. This paper addresses the **intermediate mass range ($15 < m_a < 60$GeV)**, where the$\tau$-leptons are distinct but the kinematics differ significantly from the low-mass regime, requiring a different reconstruction strategy.

2. Methodology

Data Set and Detector

• Data: The analysis utilizes the full Run-2 dataset from the Large Hadron Collider (LHC), corresponding to an integrated luminosity of 140 fb$^{-1}$ of proton-proton collisions at a center-of-mass energy of $\sqrt{s} = 13$ TeV.
• Detector: Data was recorded by the ATLAS detector, which provides near $4\pi$ solid angle coverage.

Event Selection and Signal Regions (SR)

The search targets the final state of four $\tau$-leptons. Since $\tau$-leptons decay either leptonically ($\tau \to \ell \nu \nu$, where $\ell = e, \mu$) or hadronically ($\tau_{had} \to \text{hadrons} + \nu$), the analysis defines two non-overlapping signal regions based on the decay modes:

1. $2\ell2\tau_{had}$ Region:

   • Topology: Two same-charge leptons ($e$ or $\mu$) and two same-charge hadronic $\tau$ candidates.
   • Logic: This targets the decay $H \to (a \to \tau_{\ell}\tau_{had}) + (a \to \tau_{\ell'}\tau_{had})$. The same-charge requirement suppresses Drell-Yan backgrounds while remaining consistent with the exotic decay topology.
   • Charge Constraint: The total charge of the four selected objects must be zero.

2. $3\ell1\tau_{had}$ Region:

   • Topology: Three leptons ($e$ or $\mu$) and one hadronic $\tau$ candidate.
   • Logic: This targets $H \to (a \to \tau_{\ell}\tau_{\ell'}) + (a \to \tau_{\ell''}\tau_{had})$.
   • Charge Constraint: The three leptons must have a total charge of $\pm 1$, and the total charge of all four objects must be zero.

Kinematic and Object Selection

• Lepton Selection: Electrons and muons are required to have $p_T > 20$ GeV (leading) and $15$ GeV (subleading) to ensure high trigger efficiency. Tight isolation and charge identification criteria are applied.
• $\tau_{had}$ Selection: Reconstructed using a Recurrent Neural Network (RNN) algorithm. Candidates must have $p_T > 20$ GeV and exactly 1 or 3 charged tracks.
• Background Suppression:
  • $Z$-boson Veto: To reduce the dominant Drell-Yan background, any opposite-charge lepton pair with an invariant mass $71 < m_{\ell\ell} < 111$ GeV is rejected.
  • $b$-jet Veto: Events with $b$-tagged jets are rejected to suppress top-quark backgrounds.
  • Visible Mass: The visible mass of the four-body system ($m_{vis}$) is required to be $< 110$ GeV to account for neutrino energy loss while staying below the Higgs mass.

Background Estimation

• Fake/Non-Prompt (FNP) Background: The dominant background arises from jets misidentified as leptons or $\tau$-leptons, or heavy-flavor decays producing non-prompt leptons. This is estimated using a data-driven "fake-factor" method.
  • Fake factors are measured in a $Z$+jets control region and refined in validation regions (VRs) enriched in same-sign leptons.
  • A profile-likelihood fit is performed on the VRs to constrain systematic uncertainties and refine the fake factor estimates.
• Prompt Backgrounds: Diboson processes ($WZ$, $ZZ$) are estimated using Monte Carlo (MC) simulations (Sherpa 2.2.2) and normalized to theoretical cross-sections.

Statistical Analysis

• A profile likelihood ratio test statistic is used to combine the two signal regions.
• Systematic uncertainties (luminosity, object efficiencies, energy scales, and background modeling) are included as nuisance parameters.
• Upper limits on the branching ratio are set using the $CL_s$ method.

3. Key Contributions

• First Search in this Mass Range: This is the first ATLAS search for $H \to aa \to 4\tau$ specifically targeting the mass range **$15 < m_a < 60$GeV**, complementing previous low-mass searches ($4 < m_a < 15$ GeV).
• Advanced Reconstruction Strategy: The analysis explicitly reconstructs hadronic $\tau$ decays using RNN algorithms, distinct from the boosted-object strategies used for lower masses.
• Robust Background Modeling: The implementation of a multi-step fake-factor method with validation regions and a profile-likelihood fit provides a rigorous constraint on the dominant FNP background.
• Interpretation in 2HDM+S: The results are interpreted within the context of the 2HDM+S model, specifically probing Type-III scenarios with large $\tan\beta$.

4. Results

• Observation:

  • **$2\ell2\tau_{had}$SR:** **0** observed events (Expected background:$0.12^{+0.17}_{-0.12}$).
  • **$3\ell1\tau_{had}$SR:** **31** observed events (Expected background:$28.0 \pm 4.6$).
  • The observed yields are consistent with the Standard Model background expectations. No significant excess is found.

• Limits on Branching Ratio:

  • Upper limits at the 95% Confidence Level (CL) on the branching ratio $\mathcal{B}(H \to aa \to \tau^+\tau^-\tau^+\tau^-)$ were set.
  • The limits range from 0.23 to 0.06 for $m_a$ between 15 GeV and 60 GeV.
  • Specifically, for $m_a = 45$ GeV, the observed limit is approximately 0.06 (expected: 0.04).
  • The limits are primarily limited by statistical uncertainty.

• Figure 3 (Summary): The observed limits are shown to be competitive with and complementary to previous searches in different final states (e.g., $4b$,$b\bar{b}\mu\mu$) within the 2HDM+S framework.

5. Significance

• Constraining BSM Physics: The results significantly constrain the parameter space of models predicting light pseudoscalars coupled to the Higgs sector, particularly those where the pseudoscalar decays almost exclusively to $\tau$-leptons.
• Completeness of the Search: By covering the $15-60$GeV range, this analysis closes a gap in the search for exotic Higgs decays, ensuring that the entire kinematically allowed region ($m_a < m_H/2$) is now probed by ATLAS (combining with the$4-15$ GeV search).
• Future Prospects: While no signal was found, the methodology established (RNN-based $\tau$ identification, refined fake-factor techniques) sets the stage for future searches with the higher luminosity of the High-Luminosity LHC (HL-LHC), where sensitivity to even smaller branching ratios will be achievable.

In conclusion, this paper represents a comprehensive and rigorous search for a specific class of exotic Higgs decays. The null result places stringent constraints on new physics models involving light pseudoscalars, reinforcing the Standard Model's current standing while defining the boundaries for future discoveries.