Flavor-physics benchmarks for tracker-based particle identification at the FCC-ee

This paper benchmarks the flavor-physics particle identification performance of the proposed CLD and IDEA detectors at the FCC-ee, demonstrating that while silicon tracker-based timing and energy-deposit measurements effectively suppress backgrounds for low-momentum hadrons, accessing drift-chamber cluster counts is essential for high-momentum light-quark jet tagging, with optimal results requiring timing resolutions of 30ps or better.

The Gist

Imagine a massive, high-speed race where tiny particles zoom around a circular track. The goal of the Future Circular Collider (FCC-ee) is to crash these particles together to see what new things pop out, helping us understand the fundamental rules of the universe.

To do this, scientists need giant "cameras" (detectors) to catch the debris. The paper by Anja Beck and Eluned Smith is essentially a design review for two different camera concepts, named CLD and IDEA.

Here is the core problem they are solving:
When particles crash, they create a chaotic spray of other particles. Some are "pions," some are "kaons," and some are "protons." To the camera, they all look like charged dots moving in a curve. But to the scientists, knowing exactly which type of particle it is (like telling a red car from a blue car) is crucial. If you mistake a red car for a blue one, your entire analysis of the race is wrong.

Usually, cameras have special "particle ID" gadgets (like a dedicated scanner) to tell them apart. But these two camera designs are trying to be minimalist and cost-effective. They don't have those special scanners. Instead, they want to see if the tracking system (the part that just follows the path of the particles) can do the job on its own.

How the "Tracking System" Tries to Guess the Identity

Since the tracker can't just "look" at the particle, it has to guess based on two clues, much like a detective trying to identify a suspect:

1. The Stopwatch (Time-of-Flight): If you know how far a particle traveled and how long it took, you know its speed. Heavy particles (like protons) move slower than light ones (like pions) if they have the same energy.
   • The Catch: The "stopwatch" needs to be incredibly precise. If the clock is off by even a tiny fraction of a second, the detective gets confused.
2. The Energy Meter (dE/dx or Cluster Counting): As a particle moves through the detector, it bumps into atoms and loses a little bit of energy.
   • CLD (The Silicon Tracker): Uses silicon sensors to measure how much energy is lost. It's like feeling the heat of a passing car.
   • IDEA (The Drift Chamber): Uses a gas-filled chamber. As particles zip through, they create "ionization clusters" (like little sparks). Counting these sparks is a very precise way to tell particles apart.

The Three "Test Drives"

The authors tested these two camera designs on three specific types of "races" (physics scenarios) to see how well they could tell the particles apart:

1. The "Sidekick" Tagging (Low Speed)

• The Scenario: Identifying a specific type of B-meson by looking at the low-speed "sidekick" particles flying next to it.
• The Result: This is easy! The particles are moving slowly, so even a mediocre stopwatch works. Both cameras did a great job here. The IDEA camera was slightly better because counting the "sparks" (clusters) in its gas chamber gave it a clear advantage.

2. The "Rare Event" Hunt (Medium Speed)

• The Scenario: Looking for very rare, weird decays that happen only once in a blue moon.
• The Result: This is tricky. The particles are moving at medium speeds where the "stopwatch" needs to be very sharp.
  • If the stopwatch is slow (low resolution), the cameras get confused.
  • However, the IDEA camera's "spark counting" was so good that it could identify the particles even without a perfect stopwatch.
  • The CLD camera needed a very fast stopwatch (30 picoseconds or better) to get the same level of accuracy. Without it, the "background noise" (mistaken identities) was too high.

3. The "Heavy Hitter" Jet (High Speed)

• The Scenario: Identifying jets of particles coming from a Higgs boson decay. These particles are moving incredibly fast.
• The Result: This is the hardest challenge. When particles move near the speed of light, the stopwatch becomes useless because they all arrive at the same time.
  • CLD: Failed to distinguish them well. The silicon sensors couldn't tell the difference between the fast-moving particles.
  • IDEA: Still performed well! Even at high speeds, the "spark counting" (cluster counting) in the drift chamber provided enough information to tell the particles apart.

The Big Takeaway

The paper concludes that you don't necessarily need a separate, expensive "particle ID machine" if you design your tracker right.

• The "Spark Counter" (IDEA): The drift chamber design that counts ionization clusters is a superstar. It works well at low, medium, and high speeds, even if the timing isn't perfect.
• The "Silicon Tracker" (CLD): It works great for slow particles, but for medium and fast particles, it needs a super-precise stopwatch (30 picoseconds or better) to do the job.

In summary: If you want to build a camera for this future collider, you can save money by skipping the dedicated particle scanner, but you must choose your tracking technology wisely. The "spark counting" method (IDEA) is the most versatile tool, while the silicon method (CLD) needs a very high-tech stopwatch to compete.

Technical Summary

Technical Summary: Flavor-physics benchmarks for tracker-based particle identification at the FCC-ee

Problem Statement
The Future Circular Collider (FCC-ee) is designed to perform precision measurements of Standard Model parameters, particularly in the flavor sector. While the collider environment offers unprecedented cleanliness and data volume, the final detector configurations remain undetermined. Current proposals, specifically the CLIC-like Detector (CLD) and the Innovative Detector for e+e− Accelerators (IDEA), do not include dedicated particle-identification (PID) subsystems (such as Ring Imaging Cherenkov detectors). Instead, they rely on tracking subsystems for PID. This study addresses the critical question of whether silicon trackers and drift chambers, utilizing time-of-flight (ToF) and energy-deposit (dE/dx or dN/dx) measurements, can provide sufficient PID performance to suppress background contamination in key flavor-physics analyses, including b-flavor tagging, rare b → s transitions, and s-jet tagging.

Methodology
The authors benchmark the PID performance of the CLD and IDEA detector concepts using fully simulated events within the key4hep framework.

• Detector Models: The CLD concept utilizes an entirely silicon-based tracking system. The IDEA concept combines a silicon pixel vertex tracker with a high-transparency drift chamber (filled with 90% He and 10% C4H10) and an outer tracker.
• Input Variables: The study utilizes three primary PID observables:
  1. Time-of-Flight (ToF): Measured between the first and last hits in the silicon layers. The study simulates various ToF resolutions (from 1 ps to 500 ps) by smearing true values with Gaussian distributions.
  2. Energy Deposit (dE/dx): In silicon sensors, calculated as the harmonic mean of hits.
  3. Cluster Counting (dN/dx): Specific to the IDEA drift chamber, representing the number of primary ionization clusters per unit distance. This is modeled using Garfield simulations and test-beam extrapolations, with varying cluster-counting efficiencies (50%, 80%, 100%).
• Algorithm: Particle identification is performed using Boosted Decision Trees (BDTs) trained with the XGBoost algorithm. Three independent BDTs are trained to classify particles as pions, kaons, or protons. The training samples consist of 600,000 isotropically generated particles (particle gun) covering momentum bins relevant to FCC-ee physics (e+e− → Z → bb and e+e− → ZH).
• Physics Cases: The PID tools are applied to three specific flavor-physics scenarios:
  1. Same-side b-flavor tagging: Identifying the flavor of $B^0_s$ mesons based on the net charge of associated kaons (low momentum regime).
  2. Rare decays: Analyzing contamination in fully visible and semivisible decays (e.g., $B^0_s \to K^+K^-\mu^+\mu^-$) in the medium momentum regime.
  3. s-jet tagging: Distinguishing strange jets from up/down jets based on the highest-momentum hadron in the jet (high momentum regime).

Key Results

• Low Momentum (Same-side tagging): For low-momentum hadrons (< 1 GeV/c), both CLD and IDEA achieve high signal efficiency with moderate background suppression. A ToF resolution of 30 ps or better at CLD yields performance comparable to IDEA's dN/dx-based classification. The addition of silicon dE/dx provides only marginal benefits for CLD in this regime.
• Medium Momentum (Rare decays): In the medium momentum range (1–10 GeV/c), where kinematic criteria alone are insufficient, PID becomes crucial.
  • CLD: Achieving a background reduction of one order of magnitude below kinematic-only levels requires a ToF resolution of 30 ps or better. Silicon dE/dx information offers negligible improvement over ToF alone in this range.
  • IDEA: The drift chamber's dN/dx information alone can reduce contamination by an order of magnitude compared to no PID. This performance is largely independent of cluster-counting efficiency. Adding ToF information (30–50 ps) further improves performance, but dN/dx is the dominant factor.
• High Momentum (s-jet tagging): For particles with momenta > 10 GeV/c, ToF and silicon dE/dx lose discrimination power.
  • CLD: Cannot effectively distinguish jet flavors at realistic ToF resolutions.
  • IDEA: The dN/dx measurement in the drift chamber provides strong background suppression and flavor distinction, even at high momenta. This performance is robust across different cluster-counting efficiencies.
• General Observations: PID quality shows only a small dependence on cluster-counting efficiency. The separation power is generally lower for kaons than for protons or pions due to the crossing of dN/dx curves in the medium momentum region.

Significance and Conclusions
The paper concludes that dedicated PID detectors may not be strictly necessary for all flavor-physics measurements at the FCC-ee, provided the tracking systems are optimized.

• Kinematic vs. PID: For fully reconstructed final states with excellent invariant-mass resolution (e.g., rare decays with muons), kinematic selections alone can suppress backgrounds to negligible levels, making dedicated PID less critical for these specific channels.
• Necessity of PID: However, for processes where kinematic criteria are insufficient—such as inclusive decays, s-jet tagging, or decays involving invisible particles—tracker-based PID is essential.
• Detector Implications: The study suggests that the IDEA concept, with its drift chamber and dN/dx capability, offers a distinct advantage for flavor physics across a wide momentum range without requiring a dedicated PID subsystem. Conversely, the CLD concept relies heavily on achieving very high ToF resolution (≤ 30 ps) to achieve similar performance in the medium momentum regime.
• Future Work: The authors note that the current study assumes ideal reconstruction and uniform magnetic fields. Future investigations should explore realistic cluster formation, occupancy effects, and the potential benefits of adding dedicated PID systems to further enhance sensitivity. The classifiers and tools developed in this study are made publicly available to facilitate future detector design studies.