Cryogenic operation of neutron-irradiated silicon photomultiplier arrays up to 1e14 neq/cm^2

This paper presents a comprehensive characterization of custom silicon photomultiplier arrays from FBK and Hamamatsu, demonstrating their operational viability and performance metrics after neutron irradiation up to 1e14 neq/cm² across a temperature range from room temperature down to 100 K to support the LHCb Upgrade 2 SciFi Tracker.

The Gist

The Story of the "Super-Sensitive Microphones" in a Nuclear Storm

Imagine you are trying to listen to a single cricket chirping in the middle of a hurricane. That is essentially what scientists are trying to do with the Large Hadron Collider (LHC). They need to detect tiny flashes of light (photons) created by subatomic particles, but the environment is incredibly hostile.

The paper you shared is about a special type of microphone called a Silicon Photomultiplier (SiPM). These are tiny sensors that act like super-sensitive ears for light. The researchers wanted to see if these "ears" could survive a nuclear storm and still hear that single cricket, even after being battered by radiation for years.

Here is the breakdown of their experiment, explained simply:

1. The Problem: The Nuclear Hurricane

The LHC is a giant machine that smashes particles together. This creates a lot of radiation, specifically neutrons, which are like invisible, high-speed bullets.

• The Goal: The LHC is getting an upgrade (Upgrade 2) to smash particles even harder. This means the "bullets" (neutrons) will be much more frequent and powerful.
• The Fear: These bullets damage the silicon sensors. When a sensor gets hit, it starts making "static noise" (called Dark Count Rate or DCR). If the static gets too loud, it drowns out the real signal (the cricket chirp).
• The Location: The sensors are located in a spot where they will be hit by up to 30 trillion neutrons per square centimeter over the machine's life. That is a lot of damage.

2. The Solution: The "Deep Freeze"

The scientists had a clever idea: What if we freeze the sensors?
Just as a hot summer day makes people restless and noisy, heat makes atoms in the sensor jitter and create false signals. If you put the sensors in a deep freeze (down to -173°C or 100 Kelvin, using liquid nitrogen), the atoms calm down.

• The Analogy: Imagine a crowded, hot dance floor where everyone is bumping into each other (creating noise). If you turn the AC down to freezing, everyone slows down and stands still. The "noise" disappears, and you can hear the quietest sound.

3. The Experiment: Stress-Testing the Sensors

The researchers took two different brands of these sensors (made by FBK and Hamamatsu) and subjected them to a "gauntlet":

• The Bombardment: They shot neutrons at the sensors, increasing the dose until it was 100 times higher than what the current LHC sensors usually see. Some were hit with up to 100 trillion neutrons per square centimeter.
• The Freeze: They tested them at room temperature, then cooled them down step-by-step to -173°C.
• The Variables: They changed the voltage (how hard they pushed the sensors) and looked at different pixel sizes (the size of the individual "ears").

4. The Results: What They Found

A. The "Static" Vanished (Mostly)
When they cooled the sensors to -173°C, the background noise (DCR) dropped by a massive amount—a million times quieter.

• The Good News: Even after being hit by a huge amount of radiation, the frozen sensors were still quiet enough to hear the "cricket." This means the LHC Upgrade 2 can use these sensors, provided they keep them frozen.

B. The "Bullets" Left Scars
While freezing helped, the radiation still left some permanent damage.

• The "Tunneling" Effect: At very high radiation levels, the noise didn't just drop when cooled. It stayed high.
• The Analogy: Imagine the sensor is a dam holding back water. The radiation punched holes in the dam. Cooling the water makes it less turbulent, but if the holes are big enough, water still leaks through due to pressure (electric field), not just heat. This is called tunneling. The researchers found that at extreme radiation levels, this "leakage" becomes the main problem, and cooling can't fix it completely.

C. Brand Differences

• Hamamatsu vs. FBK: One brand (Hamamatsu) generally stayed quieter than the other (FBK) under the same stress. It's like one brand of microphone having better soundproofing than the other.
• Pixel Size: Smaller "ears" (pixels) handled high voltage better, while larger ones were better at certain gain settings.

D. The "Healing" Process (Annealing)
After the sensors were hit, the scientists warmed them up to see if they could "heal."

• Room Temp Healing: Warming them to 30°C (86°F) helped a bit, like letting a bruise rest.
• Hot Healing: Warming them to 135°C (275°F) helped a lot, but only if the sensors weren't being pushed too hard (low voltage). It's like baking a cake to fix a dent in the metal, but you have to be careful not to melt the frosting (the plastic parts of the detector).

5. The Big Picture: Why This Matters

This paper is a green light for the future of the LHC.

• Before: Scientists were worried that the radiation would make the sensors too noisy to use for the next big upgrade.
• Now: They know that if they freeze the sensors, they can survive the radiation storm and still do their job perfectly.
• The Catch: They need to be careful about how much voltage they apply and which brand of sensor they choose. Also, they might need to "bake" the sensors occasionally to heal the deepest damage.

In a nutshell: By putting these sensitive light detectors in a deep freezer, scientists can protect them from the nuclear storm of the Large Hadron Collider, allowing them to keep listening to the universe's smallest whispers for years to come.

Technical Summary

1. Problem Statement

The LHCb Upgrade 2 (scheduled for Long Shutdown 4) aims to increase the instantaneous luminosity of the Large Hadron Collider, resulting in a significantly harsher radiation environment. The Scintillating Fibre (SciFi) tracker, which relies on Silicon Photomultipliers (SiPMs) for readout, faces a projected neutron fluence of 3×10¹² neq/cm² (compared to 6×10¹¹ neq/cm² for Upgrade 1).

The primary challenge is radiation-induced damage in SiPMs, which leads to a dramatic increase in the Dark Count Rate (DCR). High DCR degrades signal-to-noise ratios, potentially rendering single-photon detection impossible. While cryogenic cooling is a known mitigation strategy for reducing thermal noise, its effectiveness against radiation-induced defects at the extreme fluences expected in Upgrade 2 requires comprehensive validation. This study investigates whether cryogenic operation can extend the operational lifetime of SiPMs in this high-radiation environment.

2. Methodology

The authors conducted a systematic characterization of custom SiPM arrays from two leading manufacturers: Fondazione Bruno Kessler (FBK) and Hamamatsu Photonics (HPK).

• Device Specifications:
  • FBK: Based on NUV-HD-MT (Near-Ultraviolet High-Density Metal-in-Trench) technology, specifically the Ultra-Low-Field (ULF) variant optimized for cryogenics. Pixel sizes of 31 µm and 42 µm.
  • HPK: Production code S17104 (2024), featuring Through-Silicon Via (TSV) packaging. Pixel sizes of 42 µm, 50 µm, and 60 µm (historical reference).
  • All arrays contained 64 channels with nominal dimensions of 0.25 mm × 1.65 mm.
• Irradiation:
  • Performed at the Jožef Stefan Institute (Slovenia) using fast neutrons.
  • Fluences: 3×10¹¹, 1×10¹², 3×10¹², 1×10¹³, 3×10¹³, and up to 1×10¹⁴ neq/cm² (HPK only).
  • Annealing: All devices underwent a standard "beneficial annealing" step of 2 weeks at 30°C post-irradiation to simulate operational conditions.
• Experimental Setup:
  • Measurements were conducted in a closed-cycle helium cryostat.
  • Temperature Range: 100 K to 300 K (Room Temperature).
  • Parameters Measured: Breakdown voltage ($V_{bd}$), Gain, Dark Count Rate (DCR), Optical Crosstalk, and Afterpulsing.
  • Light Source: 450 nm laser for $V_{bd}$ extraction and uniform illumination.
  • Data Acquisition: 128-channel multiplexer, 2 GHz amplifier, and 4 GHz oscilloscope.

3. Key Contributions

• High-Fluence Cryogenic Characterization: This is one of the most extensive studies to date, testing SiPMs up to 1×10¹⁴ neq/cm² at cryogenic temperatures (100 K), far exceeding the expected fluence for LHCb Upgrade 2.
• Technology Comparison: A direct, side-by-side comparison of FBK and HPK technologies under identical irradiation and thermal conditions, revealing distinct performance differences based on microcell design.
• Mechanism Identification: Detailed analysis of the transition from thermally activated noise (Shockley-Read-Hall) to field-assisted mechanisms (trap-assisted tunneling) as fluence and overvoltage increase.
• Annealing Optimization: Investigation of high-temperature annealing (up to 135°C) to determine if further DCR reduction is possible beyond standard room-temperature recovery.

4. Key Results

A. Breakdown Voltage ($V_{bd}$)

• $V_{bd}$ decreases linearly with temperature, consistent with previous studies.
• Radiation Impact: Up to 1×10¹³ neq/cm², $V_{bd}$ remained largely unchanged. At the highest fluences (3×10¹³ and 1×10¹⁴ neq/cm²), $V_{bd}$ increased by 1–1.5 V.
• Implication: Neutron damage primarily affects generation-recombination centers rather than the space-charge distribution in the avalanche region, though extreme fluence alters the effective doping profile.

B. Dark Count Rate (DCR) and Temperature Dependence

• Cryogenic Benefit: Cooling from 300 K to 100 K reduced DCR by six orders of magnitude.
• Low Fluence (< 3×10¹² neq/cm²): DCR follows the Non-Ionising Energy Loss (NIEL) scaling hypothesis (proportional to fluence) and is dominated by thermally activated Shockley-Read-Hall (SRH) generation.
• High Fluence (> 3×10¹² neq/cm²):
  • At high overvoltages, DCR deviates from NIEL scaling, increasing faster than proportional.
  • Mechanism Shift: The dominant noise mechanism shifts from thermal generation to trap-assisted tunneling and field-enhanced defect emission. This component is temperature-independent and cannot be mitigated by cooling alone.
  • Pixel Size: Smaller pixels generally performed better at the same overvoltage, while larger pixels performed better at the same gain.

C. Manufacturer Comparison (FBK vs. HPK)

• HPK Superiority: Across all fluences and operating conditions, HPK devices exhibited significantly lower DCR than FBK devices.
• Cause: HPK devices likely have a thicker p-n junction, resulting in a lower electric field at the same overvoltage, which suppresses trap-assisted tunneling. FBK devices have a higher geometrical fill factor (better Photon Detection Efficiency) but suffer from higher electric-field-induced noise.

D. Annealing Effects

• Standard Annealing (30°C): Reduced DCR by a factor of ~3.3 (consistent with LHCb Upgrade 1 experience).
• High-Temperature Annealing:
  • 80°C: Negligible additional improvement.
  • 135°C: Produced a further significant reduction in DCR (factor of ~3) but only at low overvoltages.
  • Limitation: High-temperature annealing did not affect the DCR component associated with trap-assisted tunneling (dominant at high fluence/overvoltage), confirming these are distinct defect mechanisms.

5. Significance and Conclusion

• Feasibility for LHCb Upgrade 2: The study confirms that operating SiPMs at 100 K is a highly effective strategy for the LHCb Upgrade 2. Even at the expected fluence of 3×10¹² neq/cm², cryogenic cooling suppresses radiation-induced noise to levels comparable to non-irradiated devices at room temperature, enabling single-photon detection.
• Operational Limits: For fluences exceeding 3×10¹² neq/cm², the effectiveness of cooling diminishes due to the onset of tunneling-based noise, which is independent of temperature.
• Technology Selection: While FBK offers higher Photon Detection Efficiency (PDE), HPK devices demonstrate superior radiation hardness (lower DCR) under high-fluence, high-overvoltage conditions due to their microcell design.
• Future Outlook: The results provide a robust foundation for the design of the LHCb Upgrade 2 SciFi tracker. Future work will focus on optimizing forward-bias annealing to selectively heat active regions without damaging the surrounding plastic scintillating fibers, and characterizing timing resolution and correlated noise at cryogenic temperatures.

In summary, the paper validates that cryogenic operation combined with careful device selection (favoring HPK for noise performance) allows SiPMs to survive the extreme radiation environment of the High-Luminosity LHC, extending their utility well beyond current limits.