Characterisation of silicon photomultipliers in a dilution refrigerator down to 9.4 mK towards a cryogenic cosmic-ray muon veto system

This paper reports the successful characterization of an FBK NUV-HD-cryo silicon photomultiplier operated at 9.4 mK within a dilution refrigerator, demonstrating its viability for detecting cosmic-ray muons in low-background dark matter experiments like QUEST-DMC.

The Gist

Imagine you are trying to listen to a single, tiny whisper in a room that is supposed to be perfectly silent. This is the challenge faced by scientists looking for Dark Matter—the invisible stuff that makes up most of the universe. To hear this "whisper," they need detectors that are incredibly sensitive and operate in a state of absolute zero, colder than outer space.

However, there is a problem: Cosmic rays (high-energy particles from space, like muons) constantly rain down on Earth. When they hit the detector, they create a loud "bang" that drowns out the dark matter whisper. To solve this, scientists need a "bouncer" system—a veto system—that can spot these cosmic rays and tell the main detector, "Ignore this signal, it's just a cosmic ray!"

This paper is about testing a new type of "bouncer" sensor called a Silicon Photomultiplier (SiPM) inside a machine so cold it's almost at absolute zero.

Here is the story of their experiment, broken down into simple concepts:

1. The Setting: The Deep Freeze

The scientists used a Dilution Refrigerator. Think of this as a super-fridge that can cool things down to 9.4 millikelvin. To put that in perspective:

• Room temperature is about 300 Kelvin.
• Liquid nitrogen is about 77 Kelvin.
• This fridge is 30,000 times colder than liquid nitrogen. It is so cold that atoms barely move.

The goal was to see if a silicon sensor could survive and work in this deep freeze, right next to the main experiment.

2. The Hero: The SiPM Sensor

The sensor they tested is called a NUV-HD-cryo SiPM.

• What is it? Imagine a tiny, high-tech solar panel made of millions of microscopic "buckets" (pixels). When a single photon (a particle of light) hits a bucket, it triggers a tiny electrical avalanche, like a row of dominoes falling.
• Why use it? Traditional sensors (like old-school vacuum tubes) are bulky and need high voltage. This SiPM is small, cheap, and works with low voltage.
• The Challenge: Usually, these sensors are "noisy" because heat makes electrons jump around randomly (thermal noise). But in the deep freeze, heat is gone, so the noise should disappear, making the sensor incredibly quiet and sensitive.

3. The Experiment: Putting the Sensor in the Fridge

The team built a custom copper box (like a tiny, insulated lunchbox) and put the sensor inside. They connected it to wires that went out of the fridge to a computer.

They wanted to answer three big questions:

1. Did it survive? (Did the extreme cold break it?)
2. Is it quiet? (Does it generate false signals?)
3. Can it see light? (Can it detect the flash from a cosmic ray hitting a scintillator?)

4. The Findings: The Good, The Bad, and The Weird

✅ The Good News: It Works!

• Survival: The sensor didn't break. It worked perfectly at 9.4 mK.
• Silence: The "Dark Count Rate" (random noise) was incredibly low. It was so quiet that it was basically as silent as it is at 77 Kelvin (liquid nitrogen temps). This is great news because it means the sensor won't trigger false alarms.
• Sensitivity: It could detect single photons of light, just as expected.

⚠️ The Bad News: The "Echo" Effect (Afterpulsing)

Here is where things got interesting. While the sensor was quiet, it developed a strange habit called Afterpulsing.

• The Analogy: Imagine you clap your hands once. In a normal room, you hear one clap. But in this deep freeze, the clap seems to trigger a series of faint, delayed echoes that keep going for a long time.
• What happened: When a photon hit the sensor, it didn't just stop. It trapped some electrons in the silicon. As the temperature dropped, these electrons got "stuck" in deep traps and were released very slowly, triggering fake signals long after the original light hit.
• The Result: Instead of one clean signal, the sensor saw a "train" of signals that could last up to a millisecond. This is much longer than expected.

🌟 The Proof of Concept: Catching the "Bang"

To test if this sensor could actually catch cosmic rays, they attached a piece of plastic scintillator (a material that glows when hit by a particle) to the sensor.

• When they simulated a high-energy event (like a cosmic ray hitting the plastic), the sensor saw a massive burst of light.
• Even with the "echo" problem, the sensor could clearly distinguish between a tiny whisper (noise) and a loud shout (a cosmic ray).

5. The Conclusion: What Does This Mean?

The scientists concluded that yes, this sensor can be used as a cosmic-ray bouncer for the QUEST-DMC dark matter experiment.

• The Verdict: The sensor is quiet enough and sensitive enough to do the job.
• The Catch: The "echo" (afterpulsing) is stronger in the deep freeze than at higher temperatures.
• The Fix: Since cosmic rays create a huge flash of light, the "echoes" won't hide the signal. However, the echoes might make the system slightly slower to reset. The team will need to tweak the design (like choosing the right type of plastic scintillator) to make sure the light flash is fast and sharp, so the echoes don't cause confusion.

In a nutshell: The team successfully tested a high-tech light sensor in the coldest environment on Earth. It survived, it's very quiet, and it can spot cosmic rays. While it has a quirky habit of "echoing" in the cold, it's still a promising tool for protecting our search for the universe's biggest mystery: Dark Matter.

Technical Summary

1. Problem Statement

The QUEST-DMC collaboration is developing a direct dark matter search experiment using a superfluid helium-3 ($^3$He) target. To achieve world-leading sensitivity, the experiment requires an ultra-low background environment. Since the experiment operates at sea level (unlike underground facilities), cosmic-ray muons are a primary source of background interference.

To mitigate this, a cryogenic cosmic-ray muon veto system is being designed. The proposed design involves a scintillator volume surrounding the detector, coupled directly to a Silicon Photomultiplier (SiPM) mounted inside the dilution refrigerator at the millikelvin (mK) stage.

• The Challenge: Traditional photon detectors (like PMTs) are unsuitable due to size, magnetic field sensitivity, and high voltage requirements. While SiPMs are promising, their viability at ultra-low temperatures (sub-Kelvin) had not been fully demonstrated, particularly regarding their noise characteristics (dark count rate, crosstalk, afterpulsing) and thermal load at ~10 mK.
• The Goal: To characterize the performance of a specific SiPM model (FBK NUV-HD-cryo) at 9.4 mK to determine if it can function as a reliable photon sensor for the muon veto system without disrupting the cryostat's cooling power.

2. Methodology

The authors operated a single FBK NUV-HD-cryo SiPM (12 $\times$ 8 mm$^2$, 30 $\mu$m cell pitch) inside a cryogen-free dilution refrigerator at Royal Holloway, University of London.

• Experimental Setup:
  • Temperature: The SiPM was mounted on a custom PCB inside a copper box, thermally coupled to the mixing chamber plate. The base temperature was measured at $T = 9.4 \pm 0.2$ mK using Current Sensing Noise Thermometry (CSNT).
  • Thermal Load: The power dissipation of the SiPM was estimated to be $\sim$20 pW (including wiring), resulting in a negligible temperature difference ($\Delta T \approx 203$ nK) between the sensor and the mixing chamber.
  • Readout: Signals were amplified by a custom board (130 V/V gain) located at room temperature and read out by a 23 GHz bandwidth oscilloscope.
  • Data Acquisition: Approximately 300,000 waveforms were acquired per bias voltage (32 V, 33 V, 34 V) in dark conditions. A "FastFrame" memory scheme was used to minimize dead time.
  • Analysis: A pulse-finding algorithm was applied to identify single photon events, calculate gain, and analyze noise correlations (time-over-threshold, pulse amplitude, and inter-pulse timing).

3. Key Contributions

• First Operation at 9.4 mK: This is the first report of a SiPM operating successfully inside a dilution refrigerator at temperatures below 10 mK.
• Comprehensive Noise Characterization: The study provides the first detailed breakdown of SiPM noise parameters (Dark Count Rate, Crosstalk, Afterpulsing) specifically at mK temperatures, comparing them against 77 K benchmarks.
• Proof-of-Concept Veto: The authors demonstrated the detection of high-energy events using a scintillator coupled to the SiPM inside the cryostat, validating the concept for a muon veto system.
• Breakdown Voltage Determination: Due to thermal effects during I-V sweeps, the authors developed an alternative method (extrapolating Single Photoelectron amplitude to zero) to accurately determine the breakdown voltage at mK temperatures.

4. Key Results

A. Device Performance at 9.4 mK

• Survival: The SiPM survived the cool-down from room temperature to 9.4 mK without damage.
• Breakdown Voltage ($V_{bd}$): Determined to be $25.70 \pm 0.04$ V. This is lower than the 77 K value ($\sim$27.1 V), consistent with the expected temperature dependence of SiPMs.
• Gain: The gain ($G$) ranged from $7 \times 10^5$ to $1.2 \times 10^6$ for overvoltages ($\Delta V$) between 6.3 V and 8.3 V. This is approximately 3.7 times lower than the gain reported at 77 K for the same overvoltage, likely due to the "freeze-out" of free carriers at ultra-low temperatures.

B. Noise Characteristics

• Dark Count Rate (DCR): Measured at $4.05 \pm 0.22$ mHz/mm$^2$ at 32 V. This is consistent with the <5 mHz/mm$^2$ reported at 77 K, suggesting DCR is roughly constant or weakly dependent between 77 K and 9.4 mK.
• Direct Crosstalk (DiCT): The probability was measured at 15.8% (at $\Delta V = 6.3$ V), showing a 49% increase as $\Delta V$ increased from 6.3 V to 8.3 V. No strong temperature dependence was observed compared to 77 K data.
• Correlated Delayed Avalanches (CDA) / Afterpulsing (AP):
  • Significant Increase: The CDA probability was measured at 42.3% (within a 10 $\mu$s window) at $\Delta V = 6.3$ V. This is a substantial increase compared to the 12% reported at 77 K.
  • Long Trains: The study observed "self-sustaining AP trains" lasting up to 1 ms. This is attributed to the suppression of thermal emission from deep-level traps at ultra-low temperatures, causing charge carriers to be released over much longer timescales.

C. Scintillator Coupling (Proof-of-Concept)

• A prototype scintillator coupled to the SiPM via wavelength-shifting fibers was tested.
• The system successfully detected high-light events (consistent with cosmic-ray muons and environmental gammas), producing a broad charge distribution distinct from the intrinsic noise background.
• The long AP trains did not prevent the detection of high-energy muon signals, though they introduce fluctuations in the integrated charge.

5. Significance and Implications

• Feasibility Confirmed: The results confirm that FBK NUV-HD-cryo SiPMs can operate at the mK stage of a dilution refrigerator without exceeding the thermal budget, making them viable for the QUEST-DMC muon veto system.
• Design Constraints:
  • The Dark Count Rate and Crosstalk are low enough to meet the veto system's requirements.
  • The increased Afterpulsing at mK temperatures is the primary challenge. While it does not prevent muon tagging, it creates "high-charge tails" in the signal distribution. This could increase the rate of accidental vetoes (false positives) from environmental gamma rays if the trigger threshold is set too low.
• Future R&D: To mitigate the impact of long AP trains, the next stage of development must focus on:
  1. Selecting scintillator materials with fast emission times to ensure the primary signal is collected before the AP trains dominate.
  2. Optimizing the geometry and light collection to maximize the signal-to-noise ratio, allowing for a higher trigger threshold that rejects background noise while maintaining >90% muon veto efficiency.

In conclusion, this work provides the critical experimental validation required to deploy SiPM-based muon vetoes in ultra-low temperature dark matter experiments, while highlighting specific noise mechanisms that must be managed in the final system design.