Leakage current of high-fluence neutron-irradiated 8" silicon sensors for the CMS Endcap Calorimeter Upgrade

This paper presents electrical characterization results, specifically focusing on leakage current behavior and temperature dependence, for 8-inch silicon sensors irradiated with high-fluence neutrons up to $1.4 \times 10^{16}~n_{eq.}/cm^{2}$ to support the development of the CMS High-Granularity Calorimeter for the HL-LHC upgrade.

The Gist

Imagine the Large Hadron Collider (LHC) as the world's most powerful particle accelerator, a giant circular racetrack where tiny subatomic particles zoom around at near-light speed and crash into each other. In 2030, this racetrack is getting a massive upgrade called the High-Luminosity LHC. Think of it as turning a busy highway into a super-highway: instead of just a few cars passing by, you'll have a massive, non-stop flood of traffic.

This flood of particles is great for discovering new physics, but it's terrible for the detectors watching the crashes. They are like security cameras that are about to get blinded by a thousand times more flashbulbs than they were designed for.

This paper is about testing a specific type of "camera sensor" (silicon sensors) that will be used in the CMS experiment to survive this extreme environment. Here is the breakdown of their findings, explained simply:

1. The Challenge: The "Radiation Rain"

The sensors are going to be exposed to a level of radiation so intense it's like standing in a hurricane of invisible, high-energy neutrons.

• The Analogy: Imagine you have a delicate glass window. Now, imagine someone starts throwing millions of tiny, invisible pebbles at it every second. Eventually, the glass gets pitted and cloudy. In the world of electronics, this "cloudiness" causes the sensor to leak electricity, like a bucket with holes in it. The more radiation it gets, the more holes appear, and the more electricity leaks out.

2. The Test: The "Stress Test"

The researchers took these new silicon sensors and subjected them to a "stress test" at the Rhode Island Nuclear Science Center (RINSC). They blasted them with neutrons to simulate the worst-case scenario the sensors will face in 10 years of operation.

• The Twist: They didn't just test the full, perfect sensors. They also tested "partial" sensors—sensors cut into weird shapes (like hexagons) to fit into the tight corners of the detector.
• The Question: Do these weirdly shaped sensors, which have internal cut lines and extra wiring inside them, behave differently? Do they leak more electricity because of their shape?
• The Result: No. The sensors were surprisingly robust. Whether they were full squares or partial hexagons, they leaked electricity at the same smooth, predictable rate. The internal wiring didn't cause any "short circuits" or unexpected leaks.

3. The Temperature Problem: The "Hot Stove"

One of the biggest enemies of these sensors is heat. When the sensors get hot, the "leakage" gets much worse.

• The Analogy: Think of the sensor like a sponge. If the sponge is cold, it holds its shape. If you put it on a hot stove, it starts to melt and drip.
• The Issue: During the testing, some sensors were left in the reactor for too long without enough cooling (dry ice). They got so hot that they started to "reverse anneal."
• What is Reverse Annealing? Usually, if you heat something up and then cool it down, it settles. But in these damaged sensors, heating them up too much actually made the damage worse and permanent. It was like baking a cake too long; instead of fixing the batter, you turned it into a rock.
• The Fix: The researchers realized that if they split the long irradiation sessions into two shorter sessions with a break in between (to refill the dry ice), the sensors stayed cool enough to avoid this "melting." This simple trick saved the sensors from exponential leakage.

4. The Verdict: Will They Survive?

The team calculated exactly how much electricity these sensors would leak when they are finally installed in the CMS detector, operating at a chilly -35°C.

• Good News: At the planned operating temperature, the sensors are within the safety limits. They won't leak so much electricity that the readout chips get fried.
• Bad News: If the cooling system fails and the sensors warm up even a little bit (to -30°C), the total leakage would exceed the safety limit.
• The Takeaway: The sensors are tough enough to handle the radiation, BUT the cooling system is the hero of the story. If the cooling fails, the sensors fail.

5. The "Damage Factor"

The researchers calculated a number called the "damage factor" (alpha). This is like a "leakage score" that tells you how much damage a specific amount of radiation does to the silicon.

• They found that their sensors had a slightly higher leakage score than some other sensors tested in different labs.
• Why? It's likely due to slight differences in how the radiation was measured or the specific conditions of the reactor. However, the difference was small enough that they could combine all their data to get a reliable "average" score for the future detector.

Summary in a Nutshell

The scientists built new, super-tough silicon sensors for the CMS detector to survive the upcoming "High-Luminosity" era of the LHC. They tested them under extreme radiation and found:

1. Shape doesn't matter: Weirdly shaped sensors work just as well as full ones.
2. Heat is the enemy: Keeping them cool is critical; if they get too hot during the process, they get permanently damaged.
3. Cooling is key: The sensors will work perfectly in the real detector, as long as the cooling system keeps them cold.

The paper essentially gives the green light to use these sensors, provided we keep the air conditioning running!

Technical Summary

Here is a detailed technical summary of the paper "Leakage current of high-fluence neutron-irradiated 8" silicon sensors for the CMS Endcap Calorimeter Upgrade."

1. Problem Statement

The High-Luminosity Large Hadron Collider (HL-LHC), scheduled to begin operations in 2030, will increase the integrated luminosity by a factor of nearly 10 compared to previous runs. This environment subjects the CMS detector to extreme radiation levels, particularly in the endcap regions.

• Radiation Challenge: The High-Granularity Calorimeter (HGCAL) will operate in regions with neutron fluences up to $1 \times 10^{16} , \text{neq/cm}^2$ and doses up to 1.5 MGy.
• Sensor Degradation: High radiation fluence induces bulk damage in silicon sensors, leading to increased leakage current. Excessive leakage current causes electronic noise, power dissipation issues, and potential sensor failure.
• Specific Concerns: The HGCAL uses 8" p-type silicon wafers cut into hexagonal shapes. Some sensors are "partial" (cut from multi-geometry wafers to fit detector borders), introducing internal dicing lines and high-voltage (HV) protection structures within the active area. There was a need to verify if these internal structures or the specific manufacturing processes (epitaxial vs. float-zone) caused anomalous leakage current behavior or instabilities under extreme fluence.
• Annealing Control: High-fluence irradiations require long exposure times in the reactor, leading to elevated temperatures and "in-reactor annealing." Uncontrolled annealing can push sensors into a "reverse annealing" regime, potentially causing charge multiplication and exponential leakage current growth.

2. Methodology

The study analyzed Version 2 prototypes of the CMS Endcap (CE) silicon sensors, irradiated at the Rhode Island Nuclear Science Center (RINSC).

• Sensor Specifications:
  • Types: Full sensors and "Partial" sensors (LD Five and HD Bottom cuts) derived from multi-geometry wafers.
  • Thicknesses: 300 $\mu$m, 200 $\mu$m (Float-Zone process), and 120 $\mu$m (Epitaxial process).
  • Fluence Range: Target fluences between $6.5 \times 10^{14}$and$1.3 \times 10^{16} , \text{neq/cm}^2$ (30% above the expected End-of-Life fluence).
• Irradiation Procedure Adaptations:
  • To mitigate in-reactor annealing, high-fluence rounds were split into two parts with dry ice refilled between sessions.
  • New puck materials (PEEK, Aluminum) were tested for thermal conductivity.
  • Pre-cooling of the aluminum transport cylinder was investigated but deemed less effective than splitting rounds.
• Characterization:
  • Sensors were measured using the ARRAY system at CERN on a cold chuck at $-40^\circ\text{C}$.
  • Leakage Current Profiles: Measured cell-by-cell to map spatial distributions.
  • IV Curves: Analyzed voltage dependence for full and partial sensors, comparing different cell geometries (standard, edge, calibration).
  • Temperature Dependence: Measurements taken between $-40^\circ\text{C}$ and $-36^\circ\text{C}$ to extract activation energy.
  • Fluence Assessment: Compared three methods (target fluence, dosimetry sensors, and irradiation time). The study concluded that irradiation time (based on reactor power curves) provided the most consistent results across campaigns.

3. Key Contributions

• Validation of Partial Sensors: Demonstrated that partial sensors with internal HV rings and dicing lines exhibit leakage current behavior identical to full sensors. No instabilities were found near internal HV protection structures.
• Mitigation of Reverse Annealing: Proved that splitting high-fluence irradiation rounds effectively prevents sensors from entering the reverse annealing regime, thereby avoiding the exponential rise in leakage current associated with charge multiplication.
• Refined Damage Factor ($\alpha$): Calculated the current-related damage factor for CE sensors, providing a benchmark for future detector simulations and cooling requirements.
• Fluence Assessment Protocol: Established irradiation time as the most reliable metric for fluence estimation in this specific reactor setup, correcting discrepancies found in previous dosimetry-based methods.

4. Key Results

• Leakage Current Profiles:
  • Profiles are smooth and continuous across the sensor surface, well-modeled by a 2D elliptical Gaussian function.
  • Leakage current variations are attributed to fluence gradients in the reactor rather than thermal annealing gradients or internal sensor structures.
  • Different cell geometries (standard, edge, calibration) show stable IV curves with deviations $<8\%$ after volume normalization.
• Exponential Current Rise:
  • Sensors subjected to prolonged annealing (e.g., Round 4, equivalent to ~43,000 min at $60^\circ\text{C}$) exhibited an exponential increase in leakage current at high bias voltages.
  • This behavior is attributed to charge multiplication in the reverse annealing regime.
  • Split irradiation rounds (e.g., Rounds 12, 14) successfully prevented this effect, maintaining stable diode-like IV characteristics.
• Damage Factor ($\alpha$):
  • The combined $\alpha$ value at $-20^\circ\text{C}$ is $8.2 \pm 0.2 \pm 1.6 \times 10^{-19} , \text{A/cm}$.
  • This value is systematically higher (~18-21%) than values reported for similar p-type sensors at other facilities (e.g., JSI), likely due to facility-specific neutron flux differences and fluence estimation uncertainties.
• Operational Limits:
  • At the planned operating temperature of $-35^\circ\text{C}$, both per-cell and total sensor leakage currents remain within specifications ($<50 \, \mu\text{A}$/cell and $<10 \, \text{mA}$/sensor) even at $1.3 \times 10^{16} , \text{neq/cm}^2$.
  • At $-30^\circ\text{C}$, the total sensor current exceeds the 10 mA limit, highlighting the critical need for efficient cooling.
• Activation Energy:
  • Extracted activation energies ($E_A$) range from 0.58 to 0.63 eV.
  • This corresponds to a bandgap energy of ~1.21 eV, confirming that leakage current is dominated by bulk recombination centers (Shockley-Read-Hall theory) rather than surface defects.

5. Significance

• Detector Design Confidence: The results validate the use of partial sensors and internal HV structures for the HGCAL, ensuring that border regions of the calorimeter will perform reliably without leakage current hotspots.
• Operational Strategy: The study confirms that controlling in-reactor annealing via split irradiation rounds is essential for accurate radiation hardness testing and prevents misleading "charge multiplication" artifacts.
• Cooling Requirements: The data underscores the necessity of maintaining sensor temperatures at or below $-35^\circ\text{C}$ to ensure the total leakage current stays within the power supply and readout chip limits at the end of the HL-LHC lifetime.
• Future Irradiations: Recommendations include deploying more temperature sensors (RTDs) on both sides of the sensor puck and ensuring consistent sensor orientation to better monitor annealing and fluence profiles in future high-fluence campaigns.

In conclusion, the CMS HGCAL silicon sensors have demonstrated robust radiation hardness up to fluences exceeding the HL-LHC design requirements, provided that thermal management during irradiation and operation is strictly controlled.