Expected Sensitivity of the Light Dark Matter eXperiment to Long-Lived Dark Photons and Axion-Like Particles

This paper presents a detailed simulation-based evaluation demonstrating that the Light Dark Matter eXperiment (LDMX) can achieve competitive sensitivity to long-lived, visibly decaying dark photons and axion-like particles, thereby complementing its primary search for invisible dark matter and significantly broadening its exploration of the sub-GeV dark sector.

The Gist

Imagine the universe is a giant, bustling city. For decades, scientists have been trying to find the "ghosts" of this city—particles that make up Dark Matter. These ghosts are everywhere, but they are invisible; they don't shine, they don't bounce light off them, and they barely interact with anything.

The Light Dark Matter eXperiment (LDMX) is a high-tech detective agency built to catch these ghosts. But this paper isn't about catching the ghosts directly. Instead, it's about catching the clues they leave behind, or perhaps, catching the "messengers" they send out.

Here is the story of the paper, broken down into simple concepts and everyday analogies.

1. The Setup: A High-Speed Bullet and a Thin Wall

Imagine firing a super-fast bullet (an electron) at a very thin sheet of metal (a tungsten target).

• The Goal: Usually, LDMX is looking for the bullet to hit the metal and disappear completely, taking its energy with it into a hidden dimension. This is the "Missing Momentum" search.
• The Twist: This paper asks: "What if the bullet hits the metal, creates a secret messenger particle, and that messenger travels a bit before popping open to reveal itself?"

These "messengers" are called Long-Lived Particles (LLPs). They are like a spy who gets on a train, travels for a while, and then jumps off at a specific station to meet a contact.

2. The Two Types of Messengers

The paper focuses on two specific types of spies:

1. The Dark Photon ($A'$): Think of this as a "shadow twin" of the regular photon (light). It's a particle that usually hides but can occasionally turn into a pair of electrons (an electron and a positron).
2. The Axion-Like Particle (ALP): Think of this as a "wobbly string" of energy that can also decay into an electron pair.

Both of these particles are created in the target, fly through the detector, and then—POP!—they decay into a pair of electrons inside a specific room called the Hadronic Calorimeter (HCal).

3. The Detective Work: How LDMX Spots Them

The LDMX detector is like a multi-layered security checkpoint.

• Layer 1 (The Tracker): Watches the bullet. If the bullet slows down significantly, it means something was created.
• Layer 2 (The ECal): A high-tech energy meter. It checks if the bullet lost energy but didn't disappear.
• Layer 3 (The HCal): This is the "trap room." It's a giant box made of steel and plastic scintillators (materials that glow when hit by particles).

The Signature:
If a Dark Photon or ALP is created, it flies through the first two layers unnoticed. Then, deep inside the HCal, it decays. This creates a shower of energy that looks like a tiny, controlled explosion of light.

4. The Challenge: Distinguishing the Signal from the Noise

The problem is that the detector is noisy. Sometimes, regular particles (like neutrons or muons) bounce around and create energy in the HCal that looks like our spy. This is the "Background Noise."

• The Analogy: Imagine you are trying to hear a specific ringtone in a crowded, noisy stadium.
  • The Signal: A clean, sharp ringtone (the electron pair from the dark particle).
  • The Noise: The roar of the crowd, people clapping, and random shouting (background particles).

To solve this, the scientists built a Smart Filter (a BDT - Boosted Decision Tree).

• Think of the BDT as a super-smart bouncer at a club.
• It looks at the shape of the energy explosion.
  • Signal: A neat, round, electromagnetic shower (like a perfect snowball).
  • Noise: A messy, jagged, hadronic shower (like a pile of rubble).
• The Bouncer checks 12 different features (how wide the snowball is, how far it is from the center, etc.) and decides: "Is this our spy, or just a random fan?"

5. The Results: A Clean Sweep

The paper runs a massive computer simulation (like a video game) to see how well this plan works.

• The Good News: The Smart Filter is incredibly good. It successfully rejects almost 100% of the background noise. In their simulation, they found zero fake events that looked like the real thing.
• The Efficiency: They can catch about 30% of the real spies. This is a great success rate for such a rare event.

6. Why This Matters

The paper concludes that LDMX isn't just a one-trick pony.

• Primary Mission: Catching invisible dark matter (the bullet disappearing).
• Secondary Mission (This Paper): Catching the visible messengers (the bullet creating a spy that pops open).

By doing both, LDMX becomes a "Swiss Army Knife" for dark matter research. Even if the dark matter is too heavy or too light for the primary search, this secondary search might catch it.

The Bottom Line

This paper is a "proof of concept." It says: "We built a sophisticated computer model of our detector. We simulated billions of collisions. We found that our detector is sensitive enough to spot these long-lived, visible particles, and our smart filters can tell the difference between a real discovery and a false alarm."

If the LDMX experiment runs as planned, it could be the first to discover these "shadow twins" or "wobbly strings," opening a new window into the hidden 95% of our universe.

Technical Summary

1. Problem Statement and Motivation

The Light Dark Matter eXperiment (LDMX) is primarily designed to search for sub-GeV dark matter (DM) via a "missing momentum" signature, where a DM particle escapes detection, carrying away energy and momentum. However, the unique design of LDMX—a fully instrumented beam dump with high-granularity calorimetry—also makes it ideal for searching for Long-Lived Particles (LLPs) that decay visibly within the detector.

While the primary analysis focuses on invisible decays, this paper addresses the complementary search for visibly decaying LLPs, specifically:

• Dark Photons ($A'$): Produced via kinetic mixing with the Standard Model photon, decaying into $e^+e^-$ pairs.
• Axion-Like Particles (ALPs): Produced via couplings to electrons ($g_{ae}$), also decaying into $e^+e^-$ pairs.

The challenge lies in distinguishing these signal events (an electromagnetic shower in the Hadronic Calorimeter) from Standard Model backgrounds, primarily photo-nuclear (PN) reactions and muon photoproduction, which can mimic large energy depositions in the calorimeter.

2. Methodology

The authors performed a comprehensive simulation study using the LDMX software framework (ldmx-sw) and Geant4 for particle propagation and detector response.

Detector Configuration

• Beam: 8 GeV electron beam from SLAC's LCLS-II.
• Target: Thin tungsten target (0.1 $X_0$).
• Tracking: Silicon-strip trackers upstream (tagger) and downstream (recoil) to measure charged particles ($p \ge 50$ MeV).
• Calorimetry:
  • ECal (Electromagnetic Calorimeter): High-granularity Si-W sampling calorimeter. Used for triggering and vetoing backgrounds via shower shape analysis.
  • HCal (Hadronic Calorimeter): Steel-scintillator sampling calorimeter located ~87 cm to 587 cm downstream. This is the primary detection volume for LLP decays.

Simulation and Event Selection

1. Signal Generation: MadGraph/MadEvent was used to simulate $A'$ and ALP production and decay ($e^- Z \to e^- A'/a \to e^- e^+ e^-$) for masses of 5, 10, 50, and 100 MeV.
2. Background Generation: Simulated PN reactions and muon photoproduction in the target and ECal. Samples correspond to $10^{14}$ to $10^{15}$ Electrons on Target (EoT).
3. Selection Strategy:
   • Trigger: Require low energy deposition in the first 20 layers of the ECal (< 3160 MeV) to ensure the incident electron lost significant energy.
   • Recoil Tracker: Require exactly one track with momentum < 2400 MeV.
   • ECal Veto: Use a Missing-Momentum BDT to reject events with additional photon-induced showers in the ECal.
   • HCal Requirements:
     • Require > 4840 MeV energy deposition (near beam energy).
     • Require the shower to be fully contained within the HCal (no activity in the first layer).
   • Visibles BDT: A Boosted Decision Tree (XGBoost) trained to distinguish electromagnetic showers (signal) from hadronic showers (background).
     • Key Features: Shower shape (x, y, z standard deviation), hit multiplicity, and crucially, the distance of hits from the projected photon trajectory (calculated using the recoil electron momentum).

3. Key Contributions

• First Detailed Simulation: This is the first comprehensive evaluation of LDMX's sensitivity to visibly decaying LLPs using realistic detector efficiencies and background levels.
• Novel Analysis Strategy: The development of a specific "Visibles BDT" that leverages the HCal's ability to distinguish electromagnetic vs. hadronic showers, a capability not utilized in the primary missing-momentum search.
• Background Rejection: Demonstrated that the combination of ECal vetoes and the HCal Visibles BDT can reduce Standard Model backgrounds to effectively zero at the pilot run scale ($10^{14}$ EoT).
• Systematic Uncertainty Quantification: Rigorous assessment of uncertainties arising from shower shape modeling (Geant4 physics lists), theoretical cross-sections, and detector resolution (tracker, HCal, ECal).

4. Results

• Background Rejection:
  • At $10^{14}$ EoT, the analysis yields zero background events after all selection cuts (including the Visibles BDT).
  • The most difficult backgrounds (neutral kaons decaying to $\pi^0$) are successfully vetoed by the BDT.
• Signal Efficiency:
  • The overall signal efficiency ranges from 27% to 38%, depending on the LLP mass and model.
  • Efficiency is relatively constant for decay distances > 90 cm (inside the HCal).
  • Heavier masses (100 MeV) show slightly lower efficiency (~27%) due to broader electromagnetic showers resembling hadronic backgrounds more closely.
• Projected Sensitivity (90% CL):
  • ALP Search: For $10^{16}$ EoT, LDMX offers unique sensitivity in the mass range ~20–56 MeV for electron couplings ($g_{ae}$), surpassing current and upcoming experiments.
  • Dark Photon Search: The experiment achieves competitive sensitivity to kinetic mixing ($\epsilon$) for masses up to ~100 MeV, covering regions not yet excluded by other experiments.
  • The sensitivity curves are presented for scenarios with 0.5, 5, and 50 expected background events to account for statistical extrapolation uncertainties.

5. Significance

• Complementarity: This work establishes that LDMX is not limited to missing-momentum searches. It can simultaneously probe invisible dark matter and visible long-lived particles, significantly broadening the physics reach of the experiment.
• Unique Physics Reach: The proposed analysis targets a specific mass-coupling parameter space for ALPs and Dark Photons that is currently inaccessible to other fixed-target or collider experiments.
• Robustness: The analysis demonstrates that LDMX can achieve a background-free search even with large datasets ($10^{16}$ EoT) by utilizing advanced machine learning (BDT) and the unique geometry of the HCal.
• Future Potential: The paper outlines paths for further improvement, such as using Graph Neural Networks (e.g., ParticleNet) for better ECal vetoes and searching for "dark tridents" in the recoil tracker to extend sensitivity to shorter-lived particles.

In conclusion, this paper validates the LDMX detector design as a powerful, multi-purpose facility capable of exploring the sub-GeV dark sector through both invisible and visible decay channels, offering world-leading sensitivity to long-lived dark sector particles.