🔬 optics

Confocal Subsurface Backscattering Microscopy for Optical Identification of Nanoscale Threading Dislocations in SiC Substrates

This paper introduces Confocal Subsurface Backscattering Microscopy (CSBM), a nondestructive optical technique that leverages confocal filtering and strain-induced photoelastic scattering to achieve high-contrast, high-resolution imaging and classification of nanoscale threading dislocations in SiC substrates, overcoming the limitations of existing surface profiling and photoluminescence methods.

Original authors: Russel Cruz Sevilla, Hsiu-Ming Hsu, Irwan Saleh Kurniawan, Ruth Jeane Soebroto, Hsiu-Ying Huang, Jia-Ren Wu, Chia-Shain Chuang, Chao-Ming Fu, Sheng-Hsiung Chang, Wen-Chung Li, Chi-Tsu Yuan

Published 2026-06-04

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Russel Cruz Sevilla, Hsiu-Ming Hsu, Irwan Saleh Kurniawan, Ruth Jeane Soebroto, Hsiu-Ying Huang, Jia-Ren Wu, Chia-Shain Chuang, Chao-Ming Fu, Sheng-Hsiung Chang, Wen-Chung Li, Chi-Tsu Yuan

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to find a tiny, invisible crack inside a block of clear glass. If you shine a bright flashlight directly at the glass, the light bounces off the surface so brightly that you can't see anything inside. If you try to look for a crack that is only a few nanometers wide (thinner than a strand of DNA), it's like trying to spot a single grain of sand in a desert using a magnifying glass; the grain is just too small to scatter enough light to be seen.

This is the problem scientists face with Silicon Carbide (SiC), a super-hard material used to make powerful electronics. Inside these materials, there are tiny defects called Threading Dislocations (TDs). Think of these as microscopic "twists" or "kinks" running through the crystal structure. These kinks are bad news because they can cause electronic devices to fail or leak electricity.

The problem is that these kinks are:

Too small to be seen by standard cameras.
Buried deep inside the material.
Hard to spot in certain types of SiC (like high-purity ones) because the material itself is "noisy" or blocks the usual light signals used to find defects.

The New Solution: "Confocal Subsurface Backscattering Microscopy" (CSBM)

The authors of this paper developed a new way to see these invisible kinks without breaking or damaging the material. They call their method CSBM. Here is how it works, using simple analogies:

1. The "Blind Spot" Trick (Confocal Filtering)

Imagine you are in a dark room with a very bright spotlight shining on a mirror in front of you. The glare is so strong you can't see anything else in the room.

Old Method: You try to look at the mirror, but the glare blinds you.
CSBM Method: Instead of looking at the surface, the scientists move their "eyes" (the microscope lens) slightly behind the surface, into the glass.
The Result: Because the lens is focused inside the glass, the bright glare from the surface becomes blurry and out of focus. The microscope has a special "pinhole" that blocks this blurry glare. Suddenly, the background goes dark (like turning off the spotlight), and the room becomes a "dark field." Now, even tiny things inside the glass can be seen because they stand out against the dark background.

2. The "Stress-Induced Glow" (Photoelastic Effect)

Even with the glare gone, a tiny kink is still too small to scatter light on its own. It's like a tiny pebble in a dark ocean; it doesn't reflect much light.

The Secret: The paper explains that these kinks aren't just empty holes; they are places where the crystal is stressed or stretched, like a rubber band that has been twisted.
The Analogy: Imagine the crystal is a calm lake. A twisted rubber band (the dislocation) is submerged in it. Even if the rubber band is small, the twist it creates in the water ripples out far beyond the rubber band itself.
The Science: This "twist" changes how light travels through that specific area (changing the refractive index). When the laser hits this stressed area, it scatters light much more strongly than the tiny kink itself would. It's as if the stress creates a "halo" or a "glow" around the defect that is much bigger and brighter than the defect itself.

3. Identifying the "Fingerprint"

The scientists found that different types of kinks create different patterns of scattered light, almost like a fingerprint.

Screw Dislocations (TSDs): These create a symmetrical, star-shaped pattern of light, like a snowflake.
Mixed Dislocations (TMDs): These create a more messy, asymmetrical pattern, like a splatter of paint.
By looking at the shape of the light pattern, the microscope can tell exactly what kind of defect it is looking at.

Why This Matters

No Damage: You don't have to cut the material or etch it away to see the defects. It's a non-destructive scan.
Works in "Noisy" Materials: It works even in high-quality, pure SiC where other methods fail because the material usually blocks the light signals.
Simple Hardware: They didn't need to build a giant, expensive machine. They just took a standard microscope, adjusted the focus to look under the surface, and used a special laser.

Summary

The paper presents a clever optical trick: Move the focus inside the material to block the surface glare, and use the "stress halo" around tiny defects to make them glow. This allows engineers to see and identify microscopic flaws in SiC wafers that were previously invisible, helping to make better and more reliable electronics.

Technical Summary: Confocal Subsurface Backscattering Microscopy for Optical Identification of Nanoscale Threading Dislocations in SiC Substrates

Problem Statement
Threading dislocations (TDs) are prevalent extended defects in Silicon Carbide (SiC) substrates that facilitate reverse leakage and degrade the performance of power devices. Despite their critical impact on device reliability, the nondestructive detection of individual nanoscale TDs remains a significant challenge. Commercial surface-profiling tools and Photoluminescence (PL) dark-contrast imaging often fail to resolve these defects. Surface-profiling methods miss TDs because they lack optically resolvable surface signatures, while PL imaging is ineffective in highly n-doped or high-purity semi-insulating (HPSI) SiC substrates where background dopants or compensating defects uniformly quench band-edge PL. Furthermore, conventional laser scattering techniques struggle in the Rayleigh regime, where the scattering cross-section of nanoscale defects scales as $a^6$ , causing signals to fall below the noise floor, especially when buried within a host matrix with weak refractive-index contrast. While Synchrotron X-ray Topography (SXRT) offers a solution, its requirement for specialized beamlines and complex alignment limits its utility for routine industrial in-line inspection.

Methodology: Confocal Subsurface Backscattering Microscopy (CSBM)
To address these limitations, the authors developed Confocal Subsurface Backscattering Microscopy (CSBM), a purely scattering-based optical technique that requires no polarization analysis or phase-contrast optics. The method relies on two synergistic mechanisms:

Confocal-Filtering-Induced Quasi-Dark-Field Configuration: By shifting the focal plane from the surface to an optimized subsurface depth, the system suppresses specular surface reflection. The confocal pinhole efficiently rejects out-of-focus background scattering, effectively converting a bright-field setup into a dark-field mode.
Strain-Induced Photoelastic Scattering: The technique exploits the photoelastic effect, where the long-range elastic strain field surrounding a TD perturbs the local permittivity. This creates spatially varying refractive-index gradients around the dislocation line, acting as an extended, anisotropic scattering center that enhances backscattered light.

The experimental setup utilizes a conventional confocal microscope (MT100, PicoQuant) with a pulsed 375 nm laser and a high-numerical-aperture objective (NA = 0.9). To validate the technique, the authors employed defect-selective etching (using molten KOH at 510°C) on HPSI SiC substrates to create partially revealed TDs. These etch pits served as surface markers to correlate specific dislocation types (identified by pit morphology) with their subsurface scattering signatures.

Key Results

High-Contrast Visualization: CSBM successfully visualized individual nanoscale TDs in HPSI SiC substrates, achieving a Contrast-to-Background Ratio (CBR) of 206. The spatial resolution was estimated to be on the order of a few micrometers (FWHM).
Distinct Scattering Signatures: The study demonstrated that different TD types exhibit unique photoelastic backscattering patterns:
- Threading Screw Dislocations (TSDs): Produced a pronounced, symmetric, star-shaped scattering pattern with a bright central core. This pattern arises from the azimuthal shear strain field creating a rotationally symmetric refractive-index perturbation.
- Threading Mixed Dislocations (TMDs): Exhibited reduced total scattering intensity, a less pronounced central spot, and a more extended, asymmetric spatial redistribution of light. This reflects the mixed screw–edge character, where the edge component introduces anisotropic strain fields that redirect light into off-axis directions.
3D Reconstruction: The authors performed depth-resolved 2D-stack mapping (from the surface to ~65.5 µm depth) and reconstructed 3D volumes. This allowed for the direct visualization of the spatial continuity between surface etch pits and the underlying dislocation lines, confirming the correlation between surface morphology and subsurface strain signatures.
Sensitivity Analysis: Theoretical modeling suggests that while the core of a TD is too small for direct geometric scattering, the strain-assisted photoelastic mechanism increases the effective scattering cross-section to a detectable level ( $10^{-15} \sim 10^{-17}$ cm²), yielding photon count rates ( $10^4 \sim 10^6$ counts/s) well above detector noise levels.

Significance and Claims
The paper claims to establish a simple, robust, and effective optical approach for the direct identification of nanoscale TDs. Unlike previous methods, CSBM is:

Nondestructive: It does not require destructive etching for the final inspection (etching was only used here for validation).
Surface-Imperfection Tolerant: By imaging subsurface features, it is less susceptible to surface particles and roughness that often plague SXRT and surface scattering methods.
Hardware-Efficient: It achieves high sensitivity without modifying existing microscope hardware or adding complex optical components like polarizers.
Industrial Viable: The technique offers a practical route toward high-throughput, in-line inspection for wide-bandgap semiconductor manufacturing, specifically addressing the detection gap in highly defective or doped SiC substrates where PL imaging fails.

The authors conclude that CSBM provides a scalable solution for distinguishing dislocation types based on their characteristic photoelastic scattering patterns, offering a significant advancement over current commercial and research-grade defect inspection tools.