Characterization of CMOS SPADs for future RICH Detectors

This paper presents experimental characterization results, including dark count rate measurements under neutron irradiation and cryogenic cooling, of CMOS SPADs developed in 55 nm and 110 nm technologies for future high-granularity, radiation-hard RICH detectors in LHCb, ALICE, and Belle II experiments.

The Gist

Imagine you are trying to take a photograph of a very fast, very chaotic party. The guests (particles) are moving so quickly that you need a camera with incredibly fast shutter speeds and the ability to see a single spark of light in a dark room. This is the job of RICH detectors in giant particle physics experiments like those at the LHC.

However, there's a problem: the party is getting rowdier. The "background noise" (radiation) is getting so intense that it's like trying to take a photo in a room filled with flashing strobe lights and people throwing confetti. The current cameras (photodetectors) are getting damaged by this chaos and becoming too "noisy" to see the real signal.

Here is a simple breakdown of what this paper is about, using some everyday analogies:

1. The Problem: The "Broken Camera" in a Storm

The scientists are planning upgrades for massive particle detectors. These detectors need to count individual photons (particles of light) with extreme precision. But, the environment is harsh. It's like driving a car through a hailstorm of tiny, invisible bullets (neutrons).

Over time, these "bullets" hit the camera's sensors (called SPADs), damaging them. A damaged sensor starts firing randomly, creating "noise" (false signals) that drowns out the real data. If the noise gets too loud, the camera becomes useless.

2. The Solution: Two Tricks to Save the Camera

The researchers (the spadRICH project) are testing two main tricks to keep the camera working in this storm:

• Trick A: The "Deep Freeze" (Cryogenic Cooling)
  Imagine your computer gets hot and starts glitching. If you put it in a freezer, it might calm down and work better.
  The scientists found that if they cool these sensors down to the temperature of liquid nitrogen (colder than a winter night in Antarctica, around -160°C), the "glitches" caused by radiation stop happening. The cold makes the sensors so quiet that even if they are slightly damaged, they can still hear the whisper of a single photon.

• Trick B: Building Stronger Sensors (Radiation Hardening)
  Instead of just freezing the camera, they are also trying to build the sensors out of tougher materials. They tested sensors made with two different manufacturing "recipes" (55 nm and 110 nm technologies).
  Think of it like testing two different types of armor. They found that the sensors made with the "55 nm" recipe were slightly tougher against the hailstorm than the "110 nm" ones, especially when you account for the size of the sensor.

3. The Experiment: The "Stress Test"

The researchers took these sensors and subjected them to a simulated hailstorm in a lab.

• The Shot: They blasted the sensors with neutrons (the hail) up to levels expected in the real experiments.
• The Result: At room temperature, the sensors went crazy. The noise increased by 10,000 times (4 orders of magnitude). It was like turning a whisper into a rock concert.
• The Fix: When they put the damaged sensors in the "freezer" (liquid nitrogen), the noise dropped back down to a manageable level. The sensors recovered most of their ability to see clearly.

4. The "Pixel" Puzzle

They also tested a whole grid of these sensors (like a camera sensor with thousands of tiny pixels).

• Some pixels got hit harder than others (just like some people in a hailstorm get hit more than others).
• They tried "baking" the sensors (heating them up) to see if that would fix the damage (a process called annealing). It helped a little bit, but the "Deep Freeze" was the real hero.

The Bottom Line

This paper is basically a report card on how well these tiny, high-tech light sensors can survive a nuclear-level hailstorm.

The takeaway: If we want to upgrade our particle physics experiments to see deeper into the universe, we can't just use standard cameras. We need to build super-tough sensors and then freeze them solid to keep them quiet enough to hear the universe's faintest whispers.

By combining better engineering (stronger sensors) with extreme cold, the scientists believe they can build detectors that will survive the rest of the experiment's life, even in the most chaotic environments.

Technical Summary

1. Problem Statement

Future upgrades to major particle physics experiments (LHCb, ALICE, and Belle II) require Ring Imaging Cherenkov (RICH) detectors capable of operating under significantly increased beam interaction densities. The photodetectors for these upgraded systems must meet three critical criteria:

• High Granularity: To resolve dense particle tracks.
• Single-Photon Sensitivity: To detect faint Cherenkov light.
• Excellent Timing: For precise event reconstruction.

However, these detectors will be exposed to extreme background radiation, estimated at $10^{13}$ 1-MeV neutron equivalent/cm² over the experiment's lifetime. Standard Silicon Photomultipliers (SiPMs) suffer from severe performance degradation (specifically increased Dark Count Rate or DCR) under such neutron fluence. To maintain functionality, SiPMs may need to operate at cryogenic temperatures (near liquid nitrogen), but their radiation hardness at these temperatures remains a critical challenge.

2. Methodology

The spadRICH project is developing a radiation-hard, cryogenic-compatible digital SiPM. To validate this approach, the authors conducted experimental characterizations on existing Single-Photon Avalanche Diode (SPAD) samples before and after irradiation.

• Device Samples:
  • Technologies: SPADs fabricated in 55 nm BCD (Bipolar-CMOS-DMOS) and 110 nm CMOS Image Sensor (CIS) technologies.
  • Variations: Samples differed in junction design (P+/N, P/N, N/P) and active diameters (ranging from 5 µm to 12 µm).
  • Arrays: A 144×32 SPAD array (110 nm CIS, 5 µm active diameter, 50 µm pitch) was also tested to gather statistical data on pixel-to-pixel variations.
• Experimental Conditions:
  • Temperature: Characterization was performed at Room Temperature (RT) and cooled to Liquid Nitrogen (LN) temperatures (~−160 °C).
  • Irradiation: Samples were exposed to neutron fluences up to $10^{12}$ 1-MeV neq/cm².
  • Annealing: One chip (Chip 2) underwent high-temperature annealing (100°C to 160°C in 20°C steps) to study defect recovery.
• Measurement Setup:
  • DCR was measured using a scaler for high rates (>100 cps) and a custom Time-to-Digital Converter (TDC) for low rates (<100 cps).
  • For the array, binary frames were accumulated at 32 kHz in passive quenching mode to generate DCR maps.

3. Key Contributions

• Cryogenic Radiation Hardness Validation: The study provides empirical evidence that cooling SPADs to cryogenic temperatures significantly recovers performance lost due to neutron irradiation.
• Technology Comparison: It offers a comparative analysis of 55 nm BCD vs. 110 nm CIS technologies regarding radiation tolerance and active area scaling.
• Array-Level Statistics: Unlike previous studies focusing on single devices, this work characterizes a large SPAD array, providing statistical distributions of DCR degradation and recovery across hundreds of pixels.
• Annealing Effects: The paper documents the impact of thermal annealing on irradiated SPADs, showing a partial recovery of DCR performance.

4. Key Results

• DCR Degradation at Room Temperature:
  • At RT, neutron irradiation of $10^{12}$ neq/cm² caused the DCR to increase by approximately 4 orders of magnitude.
  • At this level, damaged SPADs could reach noise levels of 10 kcps/µm², rendering them unusable for single-photon detection.
• Cryogenic Recovery:
  • Cooling the irradiated SPADs to −160 °C recovered most of the DCR performance, bringing it back to near pre-irradiation levels.
  • A few specific SPADs showed residual high noise, suggesting they suffered "catastrophic" damage that cooling could not mitigate.
• Technology and Size Dependence:
  • 55 nm BCD devices performed slightly better than 110 nm CIS devices when normalized by diameter.
  • Consistent with expectations, larger active area SPADs were more susceptible to neutron damage (higher DCR increase per unit area) compared to smaller ones.
• Array Statistics (Chip 1 & 2):
  • Chip 1 (irradiated to $10^{11}$ neq/cm²) and Chip 2 (irradiated to $10^{12}$ neq/cm²) showed median DCR increases of 10× and 1500×, respectively, at RT.
  • Annealing: Chip 2, after high-temperature annealing, showed a 3× reduction in DCR compared to its post-irradiation state, though it did not fully return to pre-irradiation levels.
  • Scatter plots confirmed that while the median DCR increased, the distribution of pixel noise widened significantly after irradiation.

5. Significance

This work is pivotal for the design of next-generation RICH detectors for the High-Luminosity LHC and Belle II upgrades. The findings confirm that cryogenic operation is a viable strategy to mitigate neutron-induced noise in SPADs, allowing them to function effectively even after receiving fluences up to $10^{12}$ neq/cm².

The data guides the spadRICH project in optimizing the design of radiation-hard digital SiPMs. By combining radiation-hard transistor design, integrated mitigation electronics, and microlens arrays (to reduce sensitive volume), future detectors can achieve the necessary performance in harsh radiation environments, provided they are operated at cryogenic temperatures. This validates the feasibility of using CMOS SPADs in the most demanding particle physics experiments.