Development of Segmented 4H-SiC LGADs

This paper presents the design, fabrication, and initial characterization of the first segmented 4H-SiC Low-Gain Avalanche Detectors (LGADs), which utilize internal gain and various isolation strategies to achieve clear charge separation in strip and pixel configurations for harsh environment particle detection.

The Gist

The Super-Tough, Super-Fast Detective: A New Kind of Sensor

Imagine you are trying to catch a speeding bullet (a subatomic particle) in a room that is on fire, freezing cold, and being bombarded by radiation. Standard silicon sensors, which are the "eyes" of most particle detectors, would melt, freeze, or go blind in such a harsh environment.

Enter 4H-SiC (Silicon Carbide). Think of this material as the "titanium" of the semiconductor world. It is incredibly tough, handles heat like a champion, and doesn't mind radiation. However, it has a catch: it's a bit shy. When a particle hits it, it doesn't scream as loudly as silicon does. It generates a very tiny signal, making it hard to hear the "bullet" over the background noise.

To fix this, scientists added a "megaphone" inside the material, creating a device called an LGAD (Low-Gain Avalanche Detector). This megaphone amplifies the tiny signal so it can be heard clearly.

The Big Challenge: The "Crowded Room" Problem

For years, scientists could only build these megaphone sensors as one giant, solid block (a single pad). But to track particles accurately, you need to know exactly where they hit. This requires cutting the sensor into tiny strips or pixels, like a grid of individual microphones.

Here is the problem: When you cut the sensor, you have to stop the "megaphone" effect at the edges of each strip. If the amplification bleeds over into the next strip, you get a muddled signal. In silicon sensors, scientists solved this by building tiny "soundproof walls" (isolation trenches) between the strips.

This paper reports the first time anyone has successfully built these "soundproof walls" inside the tough Silicon Carbide material.

How They Built It: The "Garden Fence" Analogy

The team created a new batch of sensors (called "Lot 4") with two main shapes:

1. Strips: Long, thin lines (like a picket fence) with a spacing of 80 micrometers.
2. Pixels: Tiny squares (like a grid of tiles) with spacings of 55 and 110 micrometers.

To keep the signals from mixing, they tried two different strategies, similar to how you might separate neighbors in a garden:

• Strategy A: The "Empty Space" Fence (Geometric Separation). They simply left a small gap of empty space between the active parts of the sensor. No physical wall, just a gap.
• Strategy B: The "Oxide Trench" Fence. They dug a tiny trench between the strips and filled it with an insulating material (oxide), like filling a ditch with concrete to stop water from flowing between gardens.

The Results: What Worked and What Didn't

The team tested these sensors with electricity and a special laser that acts like a "flashlight" to see how charge moves inside.

1. The "Gap" Rule (The Most Important Discovery)
They found a critical rule for building these sensors: You must leave a gap.

• If they tried to put the strips right next to each other (zero gap), the sensors would short-circuit and break at very low voltages. It was like trying to build a wall with no space between the bricks; the electricity would arc over the top.
• Once they added a small gap (about 1 micrometer), the sensors became stable and could handle high voltages. The "gap" acts as a buffer zone to stop the electricity from crowding and breaking the sensor.

2. The "Trench" Reality
The "Oxide Trench" strategy worked, but with a caveat. The trenches they dug were deep, but not deep enough to completely stop the electrical connection underneath. It was like digging a shallow ditch to stop a flood; the water still seeped through the bottom. However, they still managed to separate the signals well enough to prove the concept works.

3. The "Laser Test" (TPA-TCT)
Using a high-powered laser at a facility called ELI ERIC, they scanned the sensors to see if the "megaphone" effect stayed inside its own strip.

• The Result: Success! When the laser hit the left strip, only the left strip screamed. When it hit the right strip, only the right strip screamed.
• The "cross-talk" (hearing the neighbor's signal) was minimal. This proved that the segmentation works: the sensor can now tell exactly which strip a particle hit, even while amplifying the signal.

The Bottom Line

This paper is a "proof of concept." The researchers have successfully taken the complex idea of "segmented, amplified sensors" and built it for the first time in the tough, heat-resistant world of Silicon Carbide.

They proved that:

1. You can cut these sensors into strips and pixels.
2. You can add a "megaphone" (gain) to make the signal loud.
3. You can build "walls" (gaps and trenches) to keep the signals separate.

This is a major step toward creating detectors that can survive inside nuclear reactors, space satellites, or future particle colliders, where standard silicon sensors would simply give up. The paper does not claim these are ready for commercial use yet; it simply says, "We built the first prototype, and it works."

Technical Summary

Technical Summary: Development of Segmented 4H-SiC LGADs

Problem Statement
The demand for radiation-hard, fast-timing detectors in high-energy physics and nuclear applications has driven the exploration of wide-bandgap semiconductors, specifically 4H-silicon carbide (4H-SiC). While 4H-SiC offers superior thermal stability, high critical electric fields, and resistance to bulk damage compared to silicon, it suffers from a significant limitation: its wide bandgap (3.26 eV) results in low charge generation (57 electron-hole pairs per micrometer) for minimum ionizing particles. Furthermore, available epitaxial layers are often thin (<100 µm), yielding small signals unsuitable for precision tracking or timing.

To address this, the Low-Gain Avalanche Detector (LGAD) concept, which utilizes a highly doped gain layer for internal signal amplification, has been adapted for 4H-SiC. While single-pad 4H-SiC LGADs have demonstrated internal gain, the transition to segmented geometries (strips and pixels) presents a new challenge: defining inter-channel isolation without compromising the gain layer or introducing excessive inactive regions. Previous work in silicon has shown that terminating the gain implant at pad boundaries creates "no-gain" regions, reducing the effective fill factor. No segmented 4H-SiC LGADs incorporating internal gain had been fabricated or characterized prior to this work.

Methodology
This work presents the design, fabrication, and initial characterization of the first segmented 4H-SiC LGADs, produced by the CAPADS collaboration (CTU Prague, FZU CAS, and onsemi). The devices were fabricated using an ion-implantation-based approach on commercial 6-inch n-type 4H-SiC wafers with epitaxial layers ranging from 30 to 100 µm.

• Device Geometry: The study covers two primary geometries: strip detectors with an 80 µm pitch and pixel arrays with 55 µm and 110 µm pitches.
• Isolation Strategies: Two distinct inter-channel isolation methods were investigated:
  1. Geometric Separation: Relying on the lateral spacing between active implants without physical trenches.
  2. Oxide-Filled Trenches: Utilizing deep, oxide-filled trenches to physically separate channels.
• Design Parameters: Variants were produced with different "spacing" (lateral distance from the p+ implant edge to the isolation boundary) and "gap" (distance from the p+ edge to the LGAD gain implant edge) parameters.
• Simulation: TCAD simulations were employed to model electric field profiles and charge multiplication factors, predicting the impact of spacing and trench dimensions on gain and charge sharing.
• Characterization:
  • Electrical: Reverse bias IV and CV measurements were performed to assess leakage currents, breakdown voltages, and depletion profiles.
  • TPA-TCT: Two-Photon Absorption Transient Current Technique measurements were conducted at the ELI ERIC facility using a focused femtosecond laser to map charge collection and inter-channel isolation with high spatial resolution.

Key Contributions and Results

1. First Fabrication of Segmented 4H-SiC LGADs: This paper reports the first successful fabrication and characterization of 4H-SiC LGADs with multi-channel segmentation, covering both strip and pixel geometries.
2. Validation of Isolation Strategies:
   • Geometric Separation: TCAD simulations indicated that geometric separation effectively suppresses avalanche multiplication at channel boundaries but creates a "no-gain" region whose size is governed by the gap parameter.
   • Trench Isolation: Simulations and initial data suggest that while trenches improve isolation, insufficient trench depth or continuous gain implants underneath the trench can lead to charge sharing.
3. Electrical Characterization Findings:
   • Breakdown Behavior: A critical correlation was observed between the "gap" parameter and breakdown voltage. Devices with a gap of 0 µm exhibited premature breakdown below 100 V due to electric field crowding at the inter-channel interface. Conversely, devices with a gap ≥1 µm remained stable up to at least 700 V. This suggests that retracting the gain implant from the channel boundary is essential for stable operation.
   • Leakage Current: LGAD devices showed leakage currents (30 nA) approximately two orders of magnitude higher than reference PN diodes (0.1 nA), consistent with gain-layer multiplication of thermally generated carriers.
4. TPA-TCT Confirmation:
   • Measurements confirmed clear charge separation between adjacent strips with internal gain.
   • The LGAD signal amplitude was significantly higher than that of reference PN devices, consistent with an internal gain factor of approximately 6–10.
   • The transition between channels occurred sharply within a few micrometers of the geometric mid-plane, validating the effective electrical isolation and the simulated fill factor.

Significance and Outlook
The paper claims that these results demonstrate the successful transfer of the segmentation concept from silicon LGADs to 4H-SiC. The work establishes that internal charge multiplication and multi-channel segmentation can be combined in a single 4H-SiC device.

The authors note that while the initial results are promising, the gain layer implantation profiles in the current batch (Lot 4) require further optimization to match design targets. Future work will involve:

• Combining spatially resolved TPA-TCT scans with Minimum Ionizing Particle (MIP) data from CERN SPS test beams.
• Systematic comparison of trench-isolated versus geometrically separated designs.
• Optimization of implant parameters for subsequent wafer production.
• Bump-bonding the 55 µm pitch pixel devices to Timepix4 readout chips to create fully integrated 4H-SiC pixel detector modules.

The study concludes that segmented 4H-SiC LGADs represent a viable path toward radiation-hard, high-granularity detectors for harsh environments, provided that isolation strategies are carefully tuned to prevent premature breakdown and minimize inactive regions.