Development of an RPC-based gaseous photodetector with picosecond resolution

This thesis presents the development of an improved RPC-based gaseous photodetector (GasPM) for the Belle II experiment, featuring a new algorithm to suppress photon feedback, single-electron discrimination capabilities, and the qualification of a radiation-resistant LaB$_6$ photocathode to achieve picosecond time resolution.

The Gist

The Big Picture: Catching the "Ghost" Particles

Imagine the Belle II experiment as a high-speed camera trying to take a perfect photo of a rare event: two particles colliding and creating something new. To get a clear picture, the camera needs to be in a very quiet room.

However, the room is actually a chaotic construction site. The machine that smashes the particles together (the collider) is so powerful that it creates a lot of "noise"—unwanted particles and light bouncing off the walls and pipes. This noise is like a blinding flash of light that ruins the photo.

The goal of this thesis is to build a super-fast "noise-canceling" sensor (called a GasPM) that can tell the difference between the "real" collision light and the "noise" light. It does this by measuring the exact moment a photon arrives. If it arrives even a tiny fraction of a second too late, the sensor knows it's just noise and ignores it.

The Problem: The "Echo" Effect

The sensor works like a gas-filled room with a special floor (a photocathode). When a light particle hits the floor, it kicks out an electron, which then zooms through the gas, creating a chain reaction (an avalanche) that the machine can detect.

But there's a glitch. As the electron zooms through the gas, it gets excited and emits its own tiny flash of ultraviolet light. This light bounces back and hits the floor again, kicking out another electron.

• The Analogy: Imagine you shout in a canyon. You hear your voice (the real signal), but then you hear an echo (the noise). In this detector, the echo arrives so quickly that it blends with your original shout, making it impossible to tell exactly when you started speaking. This "echo" (called photon feedback) messes up the timing, making the sensor slower and less accurate.

The Solution: A Faster Camera and a Better Filter

The author, Simone Garnero, set out to fix this timing problem. Here is what they did:

1. The Super-Fast Camera (The Digitizer)
In previous tests, the sensor was like a camera taking 10 pictures per second. It was too slow to see the difference between the shout and the echo.

• The Upgrade: The author installed a new "camera" (a digitizer) that takes 10 billion pictures per second.
• The Result: This high-speed view allowed them to see the "echo" as a separate blip on the graph, distinct from the main signal. They then wrote a computer algorithm to act like a filter, automatically ignoring those echoes so only the real signal is measured.

2. The "One-Person" Rule (Single-Electron Selection)
Sometimes, the beam sends two or more particles at once. It's like two people shouting at the same time; the sound gets louder and messier, confusing the timing.

• The Fix: The author added a special "gatekeeper" (a Multi-Pixel Photon Counter) before the main sensor. This gatekeeper checks how many people are shouting. If it sees more than one person, it discards the event. This ensures the timing data is only taken when a single "shout" (electron) is happening, giving a much cleaner measurement.

3. The "Indestructible" Floor (The LaB6 Photocathode)
The sensor's floor (the photocathode) is made of a special material. In previous tests, the "noise" from the gas (ions) acted like sandpaper, slowly wearing down the floor and ruining the sensor over time.

• The Experiment: The author tested a new type of floor made of Lanthanum Hexaboride (LaB6). This material is like a diamond floor—it's much harder and resists the "sandpaper" damage.
• The Result: They tested this new floor using cosmic rays (particles from space) instead of the big machine. They found that while the new floor is tough, it might be a bit "lazy" (less sensitive) at catching the specific type of light they need. They are still figuring out if it's sensitive enough to be used in the final upgrade.

The Outcome

The thesis didn't just find a problem; it built the tools to solve it.

• Success: They proved that with the new super-fast camera and the "echo" filter, they can distinguish single particles from multiple ones and clean up the timing signals.
• Next Steps: They have a plan to test the new "diamond floor" (LaB6) in a real beam test soon. If it works, this new sensor could be installed in the Belle II experiment to help physicists see the universe's rarest events with crystal-clear precision, free from the blinding noise of the construction site.

In short: The author built a faster, smarter sensor that can ignore its own internal echoes and filter out crowds, paving the way for a clearer view of the fundamental building blocks of our universe.

Technical Summary

Technical Summary: Development of an RPC-based Gaseous Photodetector with Picosecond Resolution

Problem Statement
The Belle II experiment at the SuperKEKB collider aims to probe the Standard Model through precision flavour physics, particularly in decays involving $b$ and $c$ quarks and $\tau$ leptons. However, the experiment's performance is severely limited by beam-induced backgrounds, specifically photons generated by beam interactions with residual gas and infrastructure. These background photons are "out-of-time" relative to collision events but currently overlap with the signal due to the electromagnetic calorimeter (ECL) timing resolution of approximately 100 ns. While the current resolution suppresses 80% of these backgrounds, the remaining 20% significantly degrades the reconstruction of rare signals, such as $\pi^0$ mass resolution.

A proposed upgrade involves installing a gaseous photodetector with $\sim$20 ps time resolution between the interaction point and the ECL to reject out-of-time photons. The candidate technology is the GasPM (Gaseous Photomultiplier), which couples a photocathode with a Resistive Plate Chamber (RPC). Previous testing revealed a significant discrepancy: while laser tests achieved a time resolution of 25 ps, a 2023 beam test with a CsI photocathode degraded to 73 ps. The primary suspected causes for this degradation were photon feedback (secondary avalanches triggered by UV photons emitted during gas de-excitation) and the inability to isolate single-electron events in a multi-particle beam environment. Additionally, the CsI photocathode showed vulnerability to ion feedback damage.

Methodology
The thesis reports on an improved beam test conducted in March 2025 at the KEK Photon Factory, alongside a cosmic-ray test, to address the performance degradation and characterize new components.

1. Experimental Setup: A 5 GeV electron beam was used. The setup included the GasPM prototype, reference Microchannel Plate Photomultipliers (MCP-PMTs), a Resistive Plate Chamber (RPC), and a Multi-Pixel Photon Counter (MPPC).
2. Configuration Improvements:
   • Gas Gap and Field: The gas gap was reduced from 200 $\mu$m to 150 $\mu$m, increasing the electric field from 140 kV/cm to 187 kV/cm to enhance gain and drift velocity.
   • Radiator: The radiator was upgraded from 2.4 mm quartz to 5.0 mm MgF$_2$ to increase UV photon yield.
   • Readout: A new high-speed digitiser (DSA-C10-8+, 10 GSPS sampling rate) was introduced to resolve closely spaced primary and feedback signals, replacing the previous 5 GSPS DRS4 for specific studies.
   • Event Selection: An MPPC was added upstream to discriminate between single-electron and multiple-electron events, a capability missing in the 2023 test.
3. Calibration: The new digitiser underwent extensive offline voltage and time calibration. Due to the device's AC-coupled input, standard DC calibration was impossible; instead, sine-wave injection and polynomial fitting were used to map ADC counts to voltage and correct for cell-to-cell timing jitter.
4. Photon Feedback Analysis: A signal-shape analysis was developed using the high-sampling data. The rising edge of the waveform was fitted with an 8th-order polynomial, and the second derivative was analyzed to detect inflection points (kinks) indicative of overlapping primary and secondary avalanches.
5. Photocathode Characterization: A cosmic-ray test was performed using a Lanthanum Hexaboride (LaB$_6$) photocathode to assess its resistance to ion feedback and estimate its Quantum Efficiency (QE) in the UV range, as LaB$_6$ is known to be more robust than CsI but has lower QE.

Key Contributions and Results

• Photon Feedback Discrimination Algorithm: The author developed a classification method based on the second derivative of the signal's rising edge. This algorithm successfully identified events affected by photon feedback, classifying approximately 53.2% of the beam test events as feedback-affected. This method is more sensitive than previous approaches and leverages the 10 GSPS sampling rate to resolve overlapping peaks that lower-frequency digitisers could not distinguish.
• Single-Electron Event Selection: By correlating the total collected ADC counts with the number of hit channels on the MPPC, a selection criterion was established (fewer than 7 hits and total ADC < 4500). This effectively isolated single-electron events, removing a source of bias in time-resolution measurements that plagued the 2023 analysis.
• Digitiser Characterization: A preliminary calibration procedure for the DSA-C10-8+ digitiser was established. While the calibration improved voltage linearity, the time resolution of the digitiser itself was measured at 9.49 ps after calibration (worsening from 6.99 ps uncalibrated), indicating that further refinement of the timing calibration or manufacturer intervention is required for optimal performance.
• LaB$_6$ Photocathode Testing: The cosmic-ray test with the LaB$_6$ photocathode revealed significant operational instability. The detector exhibited a high rate of streamer discharges (90% of events) and visible physical damage (bubbles and discoloration) to the photocathode surface.
  • Attempts to mitigate this by adjusting the gas mixture (increasing SF$_6$ quenching gas) or electric field were unsuccessful.
  • A comparison with an RPC configuration (metallic electrode) showed that the hit rates were consistent, suggesting that the LaB$_6$ photocathode was not detecting Cherenkov photons effectively.
  • The observed hit rate (7.19%) was consistent with the RPC ionization rate, implying the Quantum Efficiency of the LaB$_6$ for the relevant UV wavelengths is likely too low for the intended application, or the streamer regime prevented proportional operation.

Significance and Claims
The paper claims that the work provides a critical step forward in the development of the GasPM for the Belle II upgrade by:

1. Identifying and mitigating photon feedback: The new classification algorithm offers a pathway to recover time resolution in the presence of feedback, a major limitation observed in previous beam tests.
2. Validating single-electron selection: The demonstration of MPPC-based single-electron selection ensures that future time-resolution measurements will be unbiased by multi-particle effects.
3. Highlighting component limitations: The cosmic-ray test serves as a cautionary qualification for the LaB$_6$ photocathode. It suggests that while LaB$_6$ may be resistant to ion damage, its low quantum efficiency in the UV range and the difficulty in operating it in a stable proportional mode (without streamers) under current conditions may preclude its immediate use.

The author concludes that these results, including the feedback suppression algorithm and the single-electron selection criteria, are being prepared for presentation at the 7th International Workshop on New Photon Detectors (December 2025). The work underscores the necessity of further optimization, specifically an LED-based test to precisely measure the LaB$_6$ QE and signal attenuation adjustments to match the digitiser's dynamic range, before a definitive beam test of an improved GasPM prototype can be conducted.