Mortality of ultra-thin LGADs and PiN diodes from high energy deposition

This study investigates the mortality mechanisms of ultra-thin Low Gain Avalanche Diodes and PiN diodes under high-energy particle irradiation by pre-irradiating devices and exposing them to proton and heavy ion beams, identifying distinct electrical and mechanical damage signatures to better understand and mitigate permanent radiation damage like Single Event Burnout in future high-radiation detector applications.

The Gist

Imagine you are building a fleet of incredibly sensitive, high-speed cameras to capture the fastest events in the universe. These cameras are made of silicon sensors called LGADs (Low Gain Avalanche Diodes). They are so fast they can time a particle collision with the precision of a stopwatch measuring a blink of an eye.

However, these cameras are going to be placed in a very dangerous neighborhood: a particle collider (like the Large Hadron Collider). This neighborhood is filled with radiation, which is like a constant hailstorm of tiny, invisible bullets.

The Problem: The "Burnout" Risk

Usually, these sensors are tough. But if you push them too hard—specifically, if you turn up the voltage to make them work faster after they've been hit by radiation—they can suffer a catastrophic failure called Single Event Burnout (SEB).

Think of SEB like a short circuit in a house. If you turn the voltage up too high, a single stray particle (a "bullet") can hit a weak spot, causing a massive surge of electricity. This surge creates so much heat in a tiny spot that it melts the silicon, leaving a permanent crater. The camera is now dead.

Scientists knew that if the electric field inside the sensor gets stronger than 12 Volts per micrometer, this burnout becomes a real risk. But they didn't know exactly how different types of radiation caused this. Was it just the number of particles? Or did the type of particle matter?

The Experiment: The "Target Practice"

To find out, a team of scientists from Brookhaven National Laboratory and other institutions decided to play a game of target practice with these sensors.

1. The "Pre-Workout": First, they took 72 sensors (some thin, some thick) and blasted them with neutrons in a nuclear reactor. This was like giving them a "workout" to simulate years of radiation damage. This made the sensors weaker and forced them to need higher voltages to work, bringing them right to the edge of the danger zone.
2. The "Ammo": They then took these tired sensors to a giant particle accelerator (the Tandem van de Graaff) and shot them with different types of "bullets":
   • Protons: Light, fast bullets (like ping-pong balls).
   • Heavy Ions: Heavy, slow bullets made of Carbon, Oxygen, Iron, and even Gold (like bowling balls).
3. The Goal: They wanted to see which "bullets" could punch a hole in the sensor and what the damage looked like.

The Results: Three Ways to Die

After the shooting stopped, they inspected the sensors and found three distinct ways the sensors could be destroyed:

• Category 1: The Classic "Burnout" (SEB)

  • What happened: The sensor was hit by a particle while the voltage was high. A massive spark occurred, and a tiny, circular crater (about the width of a human hair) melted into the sensor.
  • The Lesson: This confirmed that if the electric field is too strong (above 12 V/µm), any particle can cause a meltdown. It doesn't matter if it's a light proton or a heavy gold ion; if the voltage is high enough, the sensor explodes.
  • Analogy: It's like over-inflating a balloon. If you blow it too hard, a single speck of dust can pop it, leaving a hole.

• Category 2: The "Self-Destruct" (No Beam Needed)

  • What happened: Some sensors broke even when the beam was turned off. They just had the voltage turned up too high for too long, and the current got too hot.
  • The Lesson: You don't even need a particle to kill these sensors; just turning the voltage up too high is enough to melt them, usually near the edges (the "guard ring").
  • Analogy: This is like leaving a lightbulb on so long that the filament burns out, even if no one touched the switch.

• Category 3: The "Heavy Ion Hammer"

  • What happened: When hit by the heaviest ions (like Gold), the sensors didn't always make a neat crater. Instead, the current just slowly crept up and up until the sensor died.
  • The Lesson: Heavy, slow particles damage the internal structure of the silicon in a different, more gradual way than the fast, light particles.
  • Analogy: Imagine hitting a glass window with a pebble (Category 1) vs. hitting it with a sledgehammer (Category 3). The pebble makes a clean hole; the sledgehammer shatters the whole frame.

The Big Takeaway

The most important discovery is that both the fancy LGAD sensors and the simpler PiN diodes suffered the same fate. This means the "gain layer" (the special part that makes LGADs fast) isn't the weak link; the whole sensor is vulnerable to burnout if the voltage gets too high.

Why does this matter?
Future particle colliders will be incredibly radioactive. Engineers need to know exactly how much voltage they can safely apply to their sensors without melting them. This study tells them: "Keep the electric field below 12 V/µm, or you risk turning your expensive, high-tech cameras into melted silicon with craters in them."

By understanding these "death modes," scientists can design better sensors that survive the harsh environment of the future, ensuring we can keep taking pictures of the universe's most mysterious events.

Technical Summary

1. Problem Statement

Low Gain Avalanche Diodes (LGADs) are critical for high-resolution timing in High Energy Physics (e.g., the High Luminosity Large Hadron Collider). However, their operation in high-radiation environments poses significant reliability risks.

• The Core Issue: While cumulative radiation damage (displacement damage) is well-studied, Single Event Burnout (SEB) remains a poorly characterized failure mode. SEB is a catastrophic, irreversible breakdown triggered by a single ionizing particle.
• The Mechanism: SEB is believed to be caused by thermal runaway resulting from high charge density deposition. Previous studies using Minimum Ionizing Particles (MIPs) suggested that SEB occurs when the internal electric field exceeds 12 V/µm.
• The Gap: In high-radiation environments, sensors suffer gain suppression, requiring operators to increase bias voltages to compensate. This pushes sensors closer to the critical 12 V/µm field. It was unknown whether high-energy particles (heavy ions) with high stopping power (energy loss per unit length) could induce SEB at lower fields or cause different damage phenomenology compared to MIPs.

2. Methodology

The study employed a systematic approach to test the mortality of silicon sensors under extreme ionization densities.

• Devices Under Test (DUT):
  • Types: Low Gain Avalanche Diodes (LGADs) and PiN diodes (control devices without gain layers).
  • Thicknesses: Active layers of 20, 30, and 50 µm (with 300 µm substrates).
  • Total Count: 72 sensors.
• Pre-Irradiation:
  • To simulate the high-radiation environment of future colliders, all sensors were pre-irradiated at the Rhode Island Nuclear Science Center (RINSC).
  • Fluence: Up to $1.5 \times 10^{15} \text{ neq/cm}^2$ using neutrons.
  • Purpose: This step increased the breakdown voltage of the sensors, allowing them to be biased up to the expected SEB threshold (approx. 12 V/µm) without immediate failure.
  • Annealing: Sensors were baked at 60°C for 80 minutes to stabilize them and remove transient annealing effects.
• Beam Exposure:
  • Facility: Brookhaven National Laboratory (BNL) Tandem van de Graaff accelerator.
  • Beam Species: A range of particles with increasing stopping power:
    • Protons (28 MeV)
    • Carbon (83.5 MeV)
    • Oxygen (89.9 MeV)
    • Iron (121 MeV)
    • Gold (216 MeV and 332 MeV)
  • Procedure: Sensors were biased in a vacuum chamber. Voltage was ramped in steps. If a current spike occurred (indicating failure), the voltage was automatically reduced. Sensors were monitored for current spikes, steady leakage increases, or breakdown.
• Post-Beam Analysis:
  • Electrical: Re-biasing to check for permanent leakage current increases.
  • Visual/Metrology: 3D laser scanning confocal microscopy (Keyence VK-X 1100) to inspect for physical damage (craters).

3. Key Contributions

• Systematic Stopping Power Study: This is the first systematic investigation of SEB in LGADs and PiN diodes across a wide spectrum of particle stopping powers, moving beyond MIP-only studies.
• Differentiation of Failure Modes: The authors defined three distinct categories of sensor mortality based on electrical signatures and physical damage:
  1. Category 1 (SEB Candidates): Current spikes during beam exposure at fixed voltage, followed by catastrophic failure and crater formation.
  2. Category 2 (High Current Damage): Damage occurring without beam exposure, triggered by high bias voltage alone, often near the guard ring.
  3. Category 3 (Beam/Electrical Effects): Steady increase in leakage current upon beam injection, leading to trip-off, without the classic SEB spike.
• Validation of Critical Field: The study experimentally confirms the 12 V/µm critical electric field threshold for SEB in pre-irradiated sensors, regardless of whether the trigger is a proton or a heavy ion.

4. Key Results

• SEB Threshold Confirmation:
  • For sensors with active thicknesses of 30–50 µm, SEB occurred at an average electric field of 12–12.5 V/µm.
  • For thinner (20 µm) sensors, the threshold was slightly higher (14.25–14.5 V/µm), consistent with previous MIP studies.
  • The likelihood of SEB did not strongly depend on the precise pre-irradiation dose within the tested range.
• Damage Morphology:
  • Craters: Category 1 failures resulted in circular craters (~30 µm diameter, ~8 µm depth) on the sensor surface.
  • Location: Craters appeared randomly across the active area with no clear dependence on sensor type (LGAD vs. PiN), thickness, or beam type.
  • Guard Ring Damage: Category 2 failures (high current without beam) often produced craters specifically near the Guard Ring (GR).
• Independence from Gain Layer: Both LGADs and PiN diodes exhibited all three failure categories. This proves that the SEB mechanism and high-current damage do not rely on the presence of the gain layer; they are intrinsic to the high-field operation of the silicon substrate.
• Heavy Ion Effects: While heavy ions (Fe, Au) caused SEB consistent with the 12 V/µm rule, they also induced "Category 3" damage (steady leakage increase) more frequently, suggesting unique damage mechanisms at very high stopping powers.

5. Significance

• Detector Safety: The findings provide critical constraints for the safe operation of timing detectors in future high-luminosity colliders. Operators must ensure the internal electric field remains below 12 V/µm to avoid catastrophic burnout, even when compensating for radiation-induced gain loss.
• Design Implications: Since both LGADs and PiN diodes are vulnerable, sensor designs must prioritize robustness against high electric fields and thermal runaway, not just gain layer optimization.
• Reliability Modeling: The identification of distinct damage categories (SEB vs. Guard Ring breakdown vs. Heavy Ion leakage) allows for more accurate reliability modeling and failure prediction in radiation-hardened electronics.
• Universal Mechanism: The confirmation that SEB is driven by the electric field strength rather than the specific particle type (MIP vs. Heavy Ion) simplifies the safety margins required for future experiments.