Compound Semiconductor Inspection Service (CSIS)
Analytical Testing Service
MONSTR Sense Technologies’ Compound Semiconductor Inspection Service (CSIS) is the first of its kind. Based on our proprietary, multi-modal ultrafast imaging microscope technology, CSIS provides sensitive and specific defect detection and classification in wide bandgap semiconductors, including within substrate wafers and epitaxial layers, as an Analytical Testing Service. Our Ultrafast Imaging technique is relatively background free, which produces high contrast detections of Killer Defects, in whole wafers (overview scans), at high-resolution and in volumetric images of localized defects. CSIS defect inspection reports are designed to support wafer growth recipe research and development, as well as failure analysis evaluations.
CSIS is proving its value to semiconductor manufacturers and is available now, as MONSTR Sense is working to productize defect inspection capability into a commercially mature form factor for operation within semiconductor manufacturing environments.
We currently offer three service packages:
CSIS Preferred:
- Resonant linear reflectance (RLR) defect inspection analysis across an entire wafer (up to 6″) with survey scan resolution (4 um)
- Four-wave mixing (FWM) defect inspection analysis across an entire wafer (up to 6″) with survey scan resolution (4 um)
- High-resolution (<1 um) scan of 50 customer-selected areas
- Defect Inspection Report with location and classification of Defect Classes (KLARF File)
CSIS Premium:
- Resonant linear reflectance (RLR) defect inspection analysis across an entire wafer (up to 6″) with survey scan resolution (4 um)
- Four-wave mixing (FWM) defect inspection analysis across an entire wafer (up to 6″) with survey scan resolution (4 um)
- High-resolution (<1 um) scan of 100 customer-selected areas
- Volumetric scan of 10 customer-selected areas
- Defect Inspection Report with location and classification of Defect Classes (KLARF File)
CSIS Custom:
- Engage with our Ultrafast Imaging & Defect Inspection Experts to develop a Customized Defect Inspection Project that addresses your unique needs.
| Materials | SiC, GaN, GaAs | |
| Wafer size | 2″, 4″, 6″ | |
| Spatial resolution (lateral) | 1-4 microns | |
| Spatial resolution (axial) | down to 4 microns | |
| Measured channels | Resonant linear reflection (RLR), Four-wave mixing (FWM) | |
| Wavelength range (for selectivity) | 350 nm-500 nm, 690 nm-1000 nm |
Here’s why we think you should try our nonlinear imaging to inspect your SiC and GaN wafers:
- Nonlinear imaging reveals more than morphology: we measure defects that impact electronic structure. There are many things lighting up in a scattering image: Killer defects as much as non-killer defects, particulates, and dust. Our technique is resonant – if the defect does not alter the bandstructure, we can tell.
- Surface, subsurface, and substrate defects. Scattering is only sensitive to surface defects. A photoluminescence excitation beam can only penetrate so deep into the material. Our excitation beam penetrates deep and is sensitive to surface, subsurface, and even substrate defects.
- Nonlinear imaging is coherent, and therefore it is fast. Our nonlinear imaging is a coherent technique. This means we can go to larger spots without compromising signal collection. It also means we can go to sub-microsecond pixel dwell times, which enables high throughput (<10 mins) whole wafer scanning.
Nonlinear optical measurements solve some problems of white-light imaging and photoluminescence. Nonlinear measurements, particularly four-wave mixing, are sensitive to fundamental defects in the crystalline structure of the semiconductor that impact the carriers. Therefore, defects, even subsurface defects, will be visible in the four-wave-mixing image if the defect has a appreciable impact on the electrical properties of the semiconductor.
Compound semiconductors are the backbone of high-power electronics, high-frequency electronics, and opto-electronics. As global demand for compound semiconductors grows, improving yield rates of materials such as gallium nitride (GaN) and silicon carbide (SiC) is more important than ever. By developing better optical inspection tools, we seek to help manufacturers identify yield-killing defects in epitaxy layers long before the material is used in device production.
MONSTR Sense Technologies’ current line of products is already being used in research labs worldwide to improve the development of GaN devices, two-dimensional materials such as graphene and transition metal dichalcogenides, perovskites, colloidal quantum dots, gallium arsenide (GaAs)-based devices, and quantum materials such as vacancy centers.
Our ultrafast nonlinear imaging inspection technique is rather versatile – we work with tunable excitation lasers that currently allow us to study materials with bandgaps over the broad ranges 2.5-3.5 eV (wavelengths 350-500 nm) and 1.25-1.75 eV (690-1000 nm). This makes the technique perfectly suitable for, but not limited to, most compound semiconductor materials of interest, including SiC, GaN, and GaAs.
Silicon Carbide
Silicon Carbide (SiC) is the current focus for many high-power electronics manufacturers. The crucial purpose of inspection tools is to distinguish between killer and non-killer defects. Because our nonlinear technique is highly sensitive to small changes in the bandstructure, the killer defects that will affect device performance show up more strongly by either glowing or darkening.
As part of the CSIS defect analysis, we typically image 6″ SiC wafers with a 4 um spot size and classify defects across the entire wafer in a survey scan. Upon request, we also collect high-resolution scans with sub-micron resolution in specific areas. Examples of high-resolution scans of some defects of interest in SiC are shown below, including examples of high-resolution volumetric scans.
Surface triangles, at least four types of distinguishable stacking faults, carrots, micropipes and partials in SiC epilayers alter the bandstructure. Depending on how the bandstructure is changed, defects can either glow or darken in the nonlinear (NL) image. Surface triangles glows in a nonlinear image through the entire volume of the triangle. Stacking faults glow more weakly and have a thickness that is less than our axial resolution. Carrot defects and micropipes excited using the same laser wavelength, darken near the center of the nonlinear image, though carrots have a bright edge.
Nonlinear imaging is not only sensitive to surface defects. Even substrate defects, such as grain boundaries, show up in the nonlinear image. Unlike photoluminescence (PL), our technique does not rely on above-gap excitation, therefore, we are more sensitive to deep defects in the epilayer and substrate if the focus is set accordingly.
The signal from our nonlinear imaging modality is emitted from the microscope focus, which provides <4 micron axial resolution in the SiC with our 0.85 NA objective. Since we also excite with light that transmits the semiconductor (photon energy is somewhat less than the bandgap), MONSTR Sense nonlinear imaging can measure deep within the wafer. Below we show volumetric images of SiC surface triangle defects in the epitaxy layer, in which we measure the volume of 3C of polytype inclusion. We have also measured various stacking faults and shown that they exist in the basal plane of SiC and have a thickness that is less than the axial resolution of our measurement. We offer volumetric scan down through entire 1000 micron substrates with high axial resolution.
Gallium Arsenide
Gallium Arsenide (GaAs) has made it from the lab to the marketplace with applications including integrated circuits, LEDs, and solar cells. Defect inspection of GaAs and GaAs quantum wells remains at the forefront of the industry. However, correlating defects with their impact on device performance and measuring subsurface defects is challenging.
Full wafer scans of a multiple quantum well sample highlight some features unique to these samples. For one, thickness variations in the substrate lead to Fabry-Perot fringes in the wafer image. Secondly, many of the defects in the wafer manifest themselves through bright emission properties. The provided wafer excerpt has a diameter of 2″, though wafers of up to 6″ can be scanned with sub-10 um resolution. Moreover, we offer the option to provide high-resolution images of certain defect areas, as illustrated below.
Scratches, as well as crystallographic defects, show up strongly in the FWM. Additionally, defects that partially or fully emit brightly can also be found with our nonlinear inspection method. Those defects are crucial because conventional brightfield or darkfield microscopy is often not sensitive to them. However, they clearly alter the bandstructure of the material, affecting the electrical properties and hence often acting as a killer defect. Our nonlinear imaging is also sensitive to subsurface defects that do not show up in any surface-sensitive scattering technique.
Gallium Nitride
Surface heterogeneities (triangular pyramids) are readily visible in the Mg-doped, P-type GaN Templates grown on (0001) sapphire substrates by low-pressure metal organic chemical vapor phase deposition (MOCVD). Surface heterogeneity is likely due to stacking fault formation and three-dimensional growth). It is reported that the density and size of these structures increase with the amount of magnesium incorporated into the GaN samples.
Compared to SiC, bulk GaN has a long decay time, evidenced in the figure below by a signal lasting well past 50 ps. Defects remain distinguishable by signal enhancement and changes in the decay time. Because GaN is a direct gap semiconductor, the bandstructure can interact well with light without the introduction of phonons. Beyond defect characterization, four-wave mixing has proven powerful for the measurement of excitons in direct gap semiconductors. The direct gap in the bandstructure allows for excitons, and it is part of the reason GaN measurements with nonlinear imaging and photoluminescence are so different from SiC.
