Characterisation of Crystalline Defects in 4H Silicon Carbide using DLTS and TSC

This paper characterizes intrinsic and growth-related electrically active defects, specifically identifying the $Z_{1/2}$ and Nitrogen-related defects, in state-of-the-art n-type 4H Silicon Carbide diodes using Deep-Level Transient Spectroscopy (DLTS) and Thermally Stimulated Currents (TSC) to support the development of radiation-hard sensors for future hadron collider experiments.

The Gist

Imagine you are building a super-strong, high-tech camera for a future particle collider. This camera needs to take pictures in an environment so full of radiation that it would melt or break a standard silicon camera almost instantly. Scientists are looking for a new material to build this camera, and they've chosen Silicon Carbide (SiC)—specifically a type called 4H-SiC. Think of SiC as the "titanium" of the semiconductor world: it's incredibly tough and handles heat and radiation much better than regular silicon.

However, before you can trust this new material, you have to check its quality. Even the best materials have tiny imperfections inside them, like dust in a diamond or a scratch on a lens. In the world of electronics, these imperfections are called defects. If there are too many defects, the camera won't work right.

This paper is essentially a "quality control report" for a brand-new, un-irradiated SiC diode (a basic electronic component). The scientists wanted to find out: What kind of "dust" and "scratches" are already hiding inside this material before we even start using it?

The Two Detective Tools

To find these invisible defects, the scientists used two different "flashlights" or detective techniques:

1. TSC (Thermally Stimulated Currents): Imagine the diode is a cold room full of people (electrons) hiding in dark corners (defects). The scientists slowly heat up the room. As it gets warmer, the people get restless and start running out of the corners. The scientists measure the "crowd surge" as it happens. By watching when the people run out, they can guess how deep the corners were.
2. DLTS (Deep-Level Transient Spectroscopy): This is a more precise version of the same idea. Instead of just heating the room, they give the electrons a little "shock" (a voltage pulse) to make them jump out of their hiding spots, and then they listen very carefully to how long it takes for the room to settle down again.

What They Found

The scientists found about a dozen different types of "hiding spots" (defects) inside the material. Since the material hadn't been hit by radiation yet, they knew these defects were either:

• Intrinsic: Natural imperfections that happen just because the crystal structure isn't perfect (like a missing brick in a wall).
• Growth-related: Mistakes made while the material was being grown in a lab.
• Impurities: Unwanted guests, like a speck of dirt, that got mixed in during production.

Two specific "guests" were identified:

• The $Z_{1/2}$ Defect: This is a famous troublemaker in the SiC world. It's known as a "lifetime killer," meaning it stops electrons from doing their job efficiently. The scientists confirmed it was there.
• A Nitrogen Defect: Nitrogen is used to "dope" (tune) the material, but sometimes it sits in the wrong spot, creating a glitch.

The "Heating Rate" Problem

Here is the tricky part of the story. The scientists tried to use both TSC and DLTS to measure these defects, but the results didn't always match perfectly.

Think of it like trying to measure the speed of a car.

• DLTS is like using a high-speed camera with a laser radar. It's very precise.
• TSC is like trying to guess the speed by watching the car blur past a window.

The paper explains that the TSC method they used was a bit "blurry." To get a perfect TSC measurement, you need to heat the material at many different speeds (from very slow to very fast). However, their equipment had limits:

• If they heated it too fast, the heat didn't spread evenly through the material (like trying to toast a thick steak on one side only), causing a distorted picture.
• If they heated it too slow, the signal was so weak it got lost in the electronic "static" (noise).

Because of this, the TSC numbers for the energy levels of the defects were a bit fuzzy. The scientists used a computer simulation to prove that both methods were actually looking at the same defects, just with different levels of clarity.

The Verdict

The paper concludes that DLTS is the superior tool for this job. Its measurements are much sharper and more reliable.

• The Good News: They successfully mapped out the "fingerprint" of the defects in this high-quality SiC material. They found the $Z_{1/2}$ defect and a Nitrogen-related defect.
• The Next Step: This is just the "before" picture. The scientists plan to shoot the material with protons, neutrons, and gamma rays (radiation) in the future to see how the defects change. This will help them understand if SiC is truly tough enough to survive the extreme conditions of future particle colliders.

In short: The scientists took a close look at a new, tough material, found some natural imperfections using two different methods, and decided that one method (DLTS) gave them the clearest, most trustworthy map of the territory.

Technical Summary

Technical Summary: Characterisation of Crystalline Defects in 4H Silicon Carbide using DLTS and TSC

Problem Statement
Future hadron collider experiments require sensing materials capable of withstanding significantly higher radiation fluences than current Silicon (Si) detectors can endure. While efforts to improve Si radiation hardness continue, wide-bandgap materials like Silicon Carbide (SiC) offer a promising alternative due to their inherently low leakage currents, even after irradiation. Despite recent advancements in the production of high-quality 4H-SiC monocrystals, a comprehensive understanding of the electrically active defects present in state-of-the-art, unirradiated material is necessary to accurately simulate detector performance and distinguish intrinsic/growth-related defects from those induced by radiation.

Methodology
The study investigates an n-type epitaxial 4H-SiC diode (active layer: 50 µm, substrate: 350 µm) produced by IMB-CNM. The device was characterized using two defect spectroscopy techniques:

1. Thermally Stimulated Currents (TSC): Measurements were performed in a temperature range of 20 K to 350 K (limited by the cryostat). The study varied filling temperatures ($T_{fill}$) between 20 K and 250 K under both forward bias (majority/minority carriers) and 0 V (majority carriers) conditions. Heating rates were varied between 5 and 11 K/min to extract energy levels via Arrhenius plots. Delayed heating measurements were also conducted for the $Z_{1/2}$ defect to refine energy level determination.
2. Deep-Level Transient Spectroscopy (DLTS): Measurements utilized a reverse bias of -10 V and pulse voltages for carrier injection. Capacitance transients were recorded using three rate windows (20 ms, 200 ms, 2 s) and analyzed with 23 correlator functions.

To reconcile discrepancies between the two methods, a simulation framework (pyTSC) was employed to generate TSC spectra based on defect parameters extracted from the DLTS data.

Key Results

• Defect Identification: The TSC measurements identified 11 distinct defects in the 20–350 K range. The DLTS measurements identified six primary defects (labeled 1–6) in the same temperature range.
• Specific Defects:
  • The $Z_{1/2}$ defect (labeled E241K in TSC, defect 6 in DLTS) was identified as the main lifetime killer in SiC. DLTS determined its energy level at 0.694 eV, while the delayed heating TSC method yielded 0.725 ± 0.036 eV. A strong temperature dependence for its capture cross-section was observed at low temperatures.
  • A defect associated with Nitrogen doping (Defect 1 in DLTS, ~0.104 eV) was identified, likely corresponding to a Nitrogen substitute on a Carbon cubic lattice site (k-site). Its concentration ($7.85 \cdot 10^{13}$ cm$^{-3}$) is consistent with, though slightly lower than, the manufacturer's specified doping concentration.
• Parameter Precision: The study found significant discrepancies in extracted energy levels and concentrations between TSC and DLTS. The authors attribute this to the limited range of heating rates achievable in TSC (factor of ~2), which introduces statistical fluctuations and temperature gradients, making the Arrhenius extraction less precise.
• Simulation Validation: Simulations using DLTS-derived parameters successfully reproduced the peak positions and widths observed in TSC spectra, confirming that both methods detected the same underlying defects, despite amplitude deviations.

Significance and Claims
The paper presents the first set of measured defect parameters for intrinsic, doping-related, and growth-related defects in state-of-the-art unirradiated 4H-SiC material. The authors conclude that while both TSC and DLTS detected the same defects, DLTS provides significantly higher precision and reliability for determining energy levels, concentrations, and capture cross-sections. Consequently, the authors assert that the DLTS-derived parameters should be regarded as the final results for this material.

The study establishes a baseline of pre-irradiation defect parameters, which is essential for future comparisons. The authors note that subsequent studies will investigate how these defects change after exposure to protons, neutrons, and gammas to determine if the observed defects are point-like or clustered and to further assess the radiation hardness of 4H-SiC relative to Silicon.