Channel-Level Calibration Methods of Silicon Photomultiplier for JUNO-TAO Central Detector

This paper presents and validates novel channel-level calibration methods for the Silicon Photomultipliers in the JUNO-TAO detector, enabling precise measurement of key parameters such as dark count rate, photon detection efficiency, gain, and optical crosstalk to achieve the experiment's goal of sub-2% energy resolution.

The Gist

Imagine you are trying to listen to a very faint whisper in a room that is supposed to be perfectly silent. But, the room is filled with thousands of tiny, hyper-sensitive microphones (called Silicon Photomultipliers, or SiPMs) that are so sensitive they can hear a single photon (a particle of light).

The goal of the JUNO-TAO experiment is to listen to "antineutrinos" (ghostly particles from a nuclear reactor) to understand the secrets of the universe. To do this, they need their microphones to be perfect. If the microphones are too noisy or misaligned, the whisper gets lost.

This paper is essentially a user manual and quality control guide for tuning these thousands of microphones so they work together perfectly. Here is the breakdown in simple terms:

1. The Problem: The "Static" and the "Echo"

Even in a dark room, these microphones aren't truly silent. They have two main problems:

• Dark Noise (The Static): Sometimes, a microphone "thinks" it heard a sound when it didn't. It's like a microphone picking up the hum of the air conditioner or a random pop. This is called Dark Count Rate (DCR).
• Crosstalk (The Echo): When one microphone fires, it accidentally sends a tiny flash of light to its neighbor, causing the neighbor to fire too.
  • Internal Crosstalk: The neighbor is in the same "room" (channel).
  • External Crosstalk: The neighbor is in a different "room" (channel).

Because the TAO detector is so sensitive and cold, there is so much "static" that the old methods of measuring these errors (like waiting for two mics to fire at the exact same time) don't work. The static is just too loud and happens too often.

2. The Solution: A New Way to Tune the Microphones

The authors developed a set of new tricks to calibrate (tune) every single microphone individually. Think of it like tuning a massive choir of 4,000 singers.

A. Measuring the "Static" (Dark Count Rate)

• The Trick: They look at the time before a real signal arrives. Any "hits" they see there are just static.
• The Catch: Some of that "static" is actually the "echo" from other microphones firing.
• The Fix: They developed a new method to turn groups of microphones off and on using a special light source (an LED). By comparing the noise when neighbors are "asleep" vs. "awake," they can subtract the echo and find the true static level.

B. Measuring the "Echo" (Crosstalk)

• The Problem: In a crowded room, it's hard to tell if a singer started singing because they heard the conductor or because they heard a neighbor.
• The New Method: They use a "Switching Game."
  1. Turn all microphones on and shine a light.
  2. Turn only one microphone (or a small group) on and shine the same light.
  3. The difference in the signal tells them exactly how much "echo" (crosstalk) is happening.
• Bonus: They can even figure out the direction of the echo. It's like knowing if the echo came from the left, right, or behind you.

C. Measuring the "Volume" (Gain) and "Timing" (Time Offset)

• Gain: How loud is the microphone? They use a known light source to see how much electrical "oomph" comes out for a single photon.
• Time Offset: Are all microphones singing in perfect rhythm? Since the cables are different lengths, some signals arrive a split-second later. They use a flash of light in the center of the room to measure exactly how late each microphone is and adjust their clocks to match.

D. Measuring "Sensitivity" (Photon Detection Efficiency)

• How good is each microphone at catching a photon? They use a special radioactive source (Germanium-68) that creates a very even, uniform light field. By counting how many photons each mic catches compared to the others, they can rank their sensitivity.

3. The Results: A Perfectly Tuned Orchestra

The authors tested these methods using a super-computer simulation (a "digital twin" of the detector). Here is what they found:

• The "Static" (DCR): Without the new fix, the error was huge (23%). With the new "switching" method, the error dropped to almost zero (0.4%).
• The "Echo" (Crosstalk): They can now measure the echo rate with less than 0.1% error and map out exactly where the echoes are coming from with high precision.
• Timing: They can sync the microphones to within 0.2 nanoseconds (that's faster than a blink of an eye).
• Volume: They can measure the sensitivity of each mic with about 2% accuracy.

4. Why Does This Matter?

If the microphones aren't calibrated, the experiment will get the wrong answer.

• Temperature: The detector is kept at -50°C (colder than a freezer) to reduce the "static." The paper shows that even if the temperature wiggles a tiny bit, the calibration holds up.
• The Big Picture: By fixing these tiny errors, the JUNO-TAO experiment can measure the energy of the antineutrinos with incredible precision (better than 2%). This will help scientists solve mysteries about why the universe is made of matter and not just antimatter.

In summary: This paper teaches us how to take a room full of 4,000 hyper-sensitive, noisy microphones, turn them on and off in specific patterns, and use math to filter out the noise and echoes. The result is a crystal-clear listening device capable of hearing the faintest whispers from the heart of a nuclear reactor.

Technical Summary

1. Problem Statement

The Taishan Antineutrino Observatory (TAO), a satellite detector for the JUNO experiment, aims to measure the reactor antineutrino energy spectrum with sub-percent precision (better than 2% at 1 MeV). To achieve this, the TAO Central Detector (CD) utilizes a large array of Silicon Photomultipliers (SiPMs) (4,024 units) operating at cryogenic temperatures (-50°C).

However, SiPMs introduce intrinsic noise and signal distortions that degrade energy resolution and vertex reconstruction accuracy. Key challenges include:

• Dark Count Rate (DCR): High thermal/tunneling noise at operating temperatures.
• Optical Crosstalk (OCT): Photons generated by avalanche events triggering neighboring pixels. This is split into Internal OCT (IOCT) (within the same channel) and External OCT (EOCT) (between different channels).
• Calibration Limitations: Standard EOCT calibration methods (e.g., time-correlation of coincident signals) fail in the TAO CD because the total DCR is extremely high (~500 MHz for the whole detector). The probability of accidental dark noise coincidences within the time window approaches 100%, making it impossible to distinguish genuine crosstalk from random noise using traditional methods.
• Parameter Sensitivity: Uncertainties in gain, time offset, PDE, and crosstalk rates directly impact the reconstructed event energy and position.

2. Methodology

The authors developed a comprehensive channel-level calibration strategy using 1 million simulated events generated by the TAO offline software framework (based on Geant4 and SNiPER). The simulation models the detector geometry, physics processes, and specific SiPM effects (Dark Noise, Gain, PDE, IOCT, EOCT, Afterpulse).

The calibration methods are categorized by the information source:

A. Calibration Based on Hit Time Information

• Dark Count Rate (DCR): Estimated by counting hits in the pre-trigger plateau (-100 ns to 0 ns). The paper identifies that raw DCR measurements are biased by EOCT from other SiPMs. A correction factor is derived using EOCT calibration results.
• Time Offset: Calibrated using an LED source placed at the center of the liquid scintillator with an external trigger. Since the light path is identical for all channels, relative time offsets are determined by fitting the hit time distribution (using a Double-Sided Crystal Ball function) for each channel relative to a reference.
• Relative Photon Detection Efficiency (PDE): Calibrated using a $^{68}$Ge source (positron annihilation source) at the center. The method uses the "zero-photoelectron" ($N_{PE=0}$) counting technique. Crucially, a statistical correction is applied to subtract the contribution of dark noise hits, which otherwise biases the PDE calculation if DCR varies between channels.

B. Calibration Based on Charge Information

• Gain: Determined by fitting the charge spectrum of single-photoelectron (PE) events with a multi-Gaussian function. The separation between adjacent peaks (1 PE, 2 PEs, etc.) defines the gain.
• Internal Optical Crosstalk (IOCT): Two methods were compared:
  1. Multiple PE Hit Analysis: Counting multi-PE hits in dark noise. This yielded a high bias (~5.7%) due to contamination from dark noise-induced multi-PEs and afterpulses.
  2. Generalized Poisson Fitting: Fitting the entire charge spectrum with a Generalized Poisson distribution. This method successfully isolates the IOCT parameter ($\lambda_{IOCT}$), reducing the bias to ~1.4% by mathematically accounting for the cascaded nature of crosstalk and dark noise.

C. Novel Method for External Optical Crosstalk (EOCT)

To overcome the high DCR limitation, the authors proposed a SiPM On-Off Switching Method:

• Mechanism: Utilizing the high-voltage system to selectively turn off groups of SiPMs (or individual tiles) while keeping an LED source active.
• EOCT Rate: By comparing the total charge ($Q_O$) when all SiPMs are ON versus the charge ($Q_C$) when only the target SiPM (or tile) is ON, the EOCT contribution is isolated ($Q_{EOCT} = Q_O - Q_C$).
• Emission Angle Distribution: A reference SiPM is used to detect EOCT photons from other SiPMs at specific angular intervals (5° steps). By rotating the "active" SiPM groups and measuring the response on the reference, the angular distribution of EOCT emission is reconstructed.
• Correction: The method accounts for non-uniform LED light fields by cross-calibrating with a uniform $^{68}$Ge source.

3. Key Contributions

1. Novel EOCT Calibration: Developed a robust method to calibrate EOCT rate and angular distribution in a high-background environment where traditional coincidence methods fail. This relies on HV switching and charge subtraction.
2. Bias Correction Frameworks:
   • Introduced a statistical correction for DCR calibration to remove EOCT contamination.
   • Developed a dark-noise subtraction algorithm for Relative PDE calibration.
   • Demonstrated that Generalized Poisson fitting is superior to simple counting for IOCT calibration in the presence of dark noise.
3. Comprehensive Uncertainty Analysis: Quantified the impact of temperature fluctuations (0.1°C) and calibration uncertainties on the final physics performance (vertex and energy reconstruction).

4. Key Results

Using 1 million simulated events, the calibration performance was evaluated with the following results:

Parameter	Calibration Bias	Standard Deviation	Notes
Relative PDE	~2.1% (max)	0.17%	Bias reduced after dark noise correction; residual bias due to optical reflection.
IOCT Rate	1.4%	3.02%	Using Generalized Poisson fitting; reduced from 5.7% bias of simple counting.
DCR	-0.4%	1.32%	After correcting for EOCT contamination (raw bias was 23.6%).
Time Offset	< 0.03 ns	0.03 ns	Limited by source positioning uncertainty (~1 cm).
Gain	< 0.1%	< 0.1%	High precision.
EOCT Rate	< 0.1%	0.1%	Achieved via On-Off switching method.
EOCT Angle	< 4%	< 1%	Measured in main angular range (5° intervals).

Impact on Physics Performance:
When these uncertainties (from both temperature fluctuation and calibration) were propagated into the event reconstruction (vertex and energy):

• Vertex Uncertainty: Increased by only 0.39 mm (from 13.08 mm to 13.12 mm).
• Energy Resolution: Degrading from 1.808% to 1.830% (a relative degradation of ~1.2%).
• Conclusion: The uncertainties are negligible compared to the detector's design goals, confirming the calibration strategy is sufficient.

5. Significance

This work provides the theoretical and algorithmic foundation for the commissioning and operation of the TAO detector.

• Feasibility: It proves that high-precision calibration is achievable even in a detector with extremely high dark noise rates, provided novel switching techniques are used.
• Physics Impact: By ensuring SiPM parameters are calibrated to the required precision, the TAO experiment can reliably measure the reactor antineutrino spectrum's fine structures. This data is critical for resolving the "5 MeV bump" anomaly and reducing model dependence in neutrino mass ordering measurements for the main JUNO experiment.
• General Applicability: The proposed On-Off switching method for EOCT calibration could be adapted for other large-scale SiPM-based detectors operating in high-noise environments.