Characterization of Passive CMOS Strip Detectors After Proton Irradiation

This paper demonstrates the feasibility of using stitched passive CMOS strip detectors fabricated in a commercial 150 nm foundry for high-energy particle tracking by showing that 24 GeV proton irradiation causes no adverse effects from the stitching process.

The Gist

Imagine you are trying to build a giant, ultra-sensitive net to catch tiny, invisible particles zipping through space at nearly the speed of light. These particles are the "ghosts" of the universe, and to study them, scientists need to catch them in a web of silicon sensors.

For years, the best way to make these nets was to use a single, massive mold (called a "reticle") that could cover an entire silicon wafer at once. It was like using a giant cookie cutter to stamp out cookies. But this method is expensive and limited to a few specialized factories.

The Big Idea: The Patchwork Quilt
The scientists in this paper asked a bold question: What if we could build these giant nets using smaller, cheaper molds, stitching them together like a patchwork quilt?

Think of it like making a long tablecloth. Instead of needing one giant piece of fabric (which is hard to weave), you sew together several smaller squares. The challenge? If you sew them together poorly, you might get a weak spot or a bump where the seams meet. In the world of particle detectors, a "seam" is called a stitch. If the stitch is bad, the detector might fail to catch the particles right at the join.

The Experiment: Building the "Quilt"
The team went to a commercial chip factory (LFoundry) and used a standard, modern technology (150 nanometers) to build these detectors. They took two small square molds and stitched them together to create strips that were either 2.1 cm or 4.1 cm long.

They made two different "patterns" for these strips:

1. The Regular Design: A standard, tried-and-true layout.
2. The "Low Dose" Design: A more complex layout with extra features, like a tiny capacitor (a battery for holding charge) built right into the strip.

The Stress Test: The Proton Shower
To see if these detectors were tough enough for the real world, they didn't just look at them; they threw a hurricane at them. They took the detectors to CERN (the home of the Large Hadron Collider) and blasted them with a high-energy beam of protons.

Imagine this as a hailstorm of tiny, super-fast bullets hitting the silicon. This simulates the harsh radiation environment of space or a particle accelerator. They hit the detectors with two different levels of "hail": a moderate storm and a massive, super-storm.

The Results: The Quilt Holds Up!
After the bombardment, they checked the detectors to see if the "stitches" had caused any problems. Here is what they found:

• No Weak Seams: The most important discovery was that the stitches didn't matter. The detectors worked perfectly across the seams. It's as if the patchwork quilt was so well-made that you couldn't tell where one square ended and the next began. The particles were caught just as well in the middle of the strip as they were at the join.
• The "Low Dose" Surprise: The two different designs behaved differently under the hailstorm.
  • In the moderate storm, the "Low Dose" design actually performed better than the regular one, catching almost as many particles as a brand-new detector.
  • However, in the super-storm, the "Low Dose" design got a bit tired and started catching fewer particles, while the "Regular" design stayed steady.
  • Analogy: Think of the "Low Dose" design as a sports car with a fancy engine. It's fast and efficient in normal traffic (moderate radiation), but if you push it too hard on a rough road (heavy radiation), it struggles. The "Regular" design is like a sturdy pickup truck—it's not as fancy, but it keeps chugging along no matter how rough the road gets.

Why This Matters
This paper is a huge win for the future of particle physics and even cancer treatment.

1. Cheaper and Faster: Because they proved that "stitching" works, scientists can now use any commercial chip factory to make these detectors. They don't need to rely on a few expensive, specialized labs anymore. It's like being able to buy custom-made suits at a regular department store instead of a high-end tailor.
2. Bigger Nets: This opens the door to building much larger detectors to cover more area, which is essential for future experiments like the Future Circular Collider.
3. The Next Step: Since the "passive" (simple) strips work so well, the team is now planning to build "active" strips. Imagine putting the brain (electronics) inside the net itself, rather than having to wire it up later. This would create "Monolithic Active Strip Sensors," making the whole system even more efficient.

In a Nutshell
The scientists proved that you can build high-tech particle detectors by stitching together small pieces of silicon, just like a quilt. Even after being blasted by a proton storm, the seams held strong, and the detectors worked perfectly. This means we can now mass-produce these detectors in regular factories, making the future of particle physics brighter, bigger, and more affordable.

Technical Summary

1. Problem Statement

Silicon strip detectors are essential for the outer tracking layers of future high-energy particle physics experiments (e.g., FCC, EIC, CEPC) and medical applications like hadron therapy. While strip detectors offer advantages in power consumption and cost over pixel sensors for large-area coverage, they face a manufacturing challenge in commercial CMOS foundries.

• The Challenge: Standard strip sensors are typically fabricated using a single mask covering the entire wafer. However, commercial foundries often use smaller reticles (e.g., 1 cm²). To create long strips (2–4 cm), multiple reticles must be "stitched" together.
• The Concern: There is a lack of empirical evidence proving that the stitching process (joining multiple reticles) does not degrade the electrical performance or charge collection efficiency of the sensors, particularly after exposure to high levels of radiation.

2. Methodology

The study involved the fabrication, irradiation, and comprehensive electrical characterization of passive CMOS strip detectors.

• Fabrication:

  • Foundry & Process: Produced at LFoundry using a 150 nm node technology on a 150 µm thick p-type Float Zone (FZ) silicon wafer (resistivity 3–5 kΩcm).
  • Stitching Strategy: Two different reticle masks were used to create two strip lengths: 2.1 cm (stitching 3 reticles) and 4.1 cm (stitching 5 reticles).
  • Design Variations: Two distinct strip geometries were tested on the 4.1 cm sensors:
    1. Regular Design: Similar to the ATLAS Phase 2 upgrade design.
    2. Low Dose Design: Incorporates specific CMOS features, including MIM capacitors and n-well implants with two different widths (30 µm and 55 µm).
  • Optimization: Previous batches suffered from early breakdown; subsequent batches underwent improved backside processing, achieving stable currents up to 300 V and full depletion at ~35 V.

• Irradiation:

  • Facility: Irradiated at CERN's IRRAD facility.
  • Beam: 24 GeV protons.
  • Fluences: Two levels were tested: $5 \times 10^{14} \, n_{eq}/cm^2$ and $1 \times 10^{15} \, n_{eq}/cm^2$.
  • Conditions: Irradiation was performed at room temperature. Sensors were tilted to ensure the beam covered the full stitched area.

• Characterization:

  • Temperature: Measurements conducted at -20 °C.
  • Annealing: Sensors underwent a beneficial annealing step (80 min at 60 °C) prior to testing.
  • Tests Performed:
    • I-V (Current-Voltage): Measured leakage current and breakdown voltage.
    • C-V (Capacitance-Voltage): Determined full depletion voltage ($V_{FD}$) and effective doping concentration ($N_{eff}$).
    • Charge Collection: Measured using a $^{90}Sr$ beta source and an ALiBaVa readout system to quantify collected charge (in electrons).

3. Key Contributions

• Feasibility Demonstration: This work provides the first experimental proof that passive CMOS strip detectors fabricated using stitched reticles in a commercial foundry (150 nm node) function correctly for high-energy particle detection.
• Radiation Hardness Validation: The study characterizes the radiation tolerance of these specific CMOS designs under proton irradiation, a critical requirement for future collider experiments.
• Design Comparison: It offers a comparative analysis between a "Regular" design and a "Low Dose" design, revealing how specific CMOS implant geometries (n-well widths) influence radiation response.

4. Results

• Stitching Impact: No negative effects from the stitching process were observed. The sensors performed consistently across the stitched boundaries, validating the manufacturing approach.
• Leakage Current & Damage Rate:
  • Leakage current increased with fluence, as expected.
  • The Low Dose design exhibited a higher current-related damage rate ($\alpha$) compared to the Regular design, attributed to higher leakage currents.
• Depletion Voltage ($V_{FD}$):
  • Unirradiated: ~30 V (Low Dose) and ~35 V (Regular).
  • After irradiation ($1 \times 10^{15} \, n_{eq}/cm^2$): $V_{FD}$ increased to ~80 V for Low Dose and ~70 V for Regular designs.
• Effective Doping ($N_{eff}$):
  • An inversion in behavior was observed. Unirradiated Low Dose sensors had lower $N_{eff}$ than Regular sensors. However, at higher fluences, the Low Dose design showed higher effective doping, suggesting a dependence of radiation damage on the specific strip implant geometry.
• Charge Collection:
  • At $5 \times 10^{14} \, n_{eq}/cm^2$: Low Dose designs maintained excellent charge collection (~11,000 electrons), comparable to unirradiated values, while the Regular design showed a noticeable decrease.
  • At $1 \times 10^{15} \, n_{eq}/cm^2$: The trend reversed. The Regular design maintained stable charge collection, while the Low Dose design showed a decrease.
  • Conclusion: The performance differences are attributed to the strip design geometry rather than bulk damage, as the bulk material was identical.

5. Significance

• Commercial Viability: The results confirm that commercial CMOS foundries can produce high-performance strip detectors using stitching techniques. This significantly expands the pool of available manufacturing facilities beyond specialized semiconductor labs.
• Future Production: The success of passive stitched sensors paves the way for large-scale production of strip sensors for future experiments (e.g., FCC).
• Path to Monolithic Sensors: This success is a critical stepping stone toward developing Monolithic Active Strip Sensors (MASS). By proving that stitching does not degrade performance, the authors suggest that future iterations can embed active electronics directly into the silicon strip sensors, eliminating the need for complex and costly hybridization steps.