Investigation of Inhomogeneity in Single Crystal SiC Wafers Using C-Scan Acoustic Scanning Microscopy

In this work, C-Scan Acoustic Scanning Microscopy (ASM) is used to map the defects of three SiC samples. The acoustic images indicate that numerous defects with different shapes and area sexist in the wafers. Some of the defects have areas of more than 100,000 μm 2 . The number of defects ranges from 1 to 50 defects/wafer. Defect mapping is essential for defect repairing or avoidance. This work shows that ASM can locate the precise positions of the crys-tallographic defects, which enables defects repair and yield enhancement.


Introduction
Silicon Carbide material has become one of the critical materials in the microelectronics industry. SiC has many properties that are hard to find in a single semiconducting material. It is a wide band gap material that can be used to build blue, green, or white optoelectronic devices [1] [2] [3]. Additionally, SiC is widely used to build power electronic devices, such as Bipolar Junction Transistors (BJT) and Field Effect Transistors (FETs) for high voltage applications [4] [5] [6] [7] [8]. It is also used to build FETs for digital, analog, and microwave applications [9] [10] [11].
SiC crystals are grown at temperatures up to more than 1600˚C [12]. Crystal growth at these high temperatures can cause many intrinsic defects including missing atoms, displaced atoms, missing stacks of atoms, micro-pipes, screw dislocations, line defects, and point defects [13] [14]. These defects can result in device failure, performance degradation [15], low reliability and sensitivity to self-heating [10].
Several techniques have been used to study defects in semiconductor substrates, including SiC wafers. Among these are photomicrography and scanning electron microscopy (SEM). These techniques are effective in identify surface defects. Bulk defects can be identified using techniques such as Optical Coherence Tomography (OCT) that can detect defects on the surface and in the bulk [16]. However, OTC requires the transmission of the light into the material, which may not be possible for less transparent materials. Raman spectroscopy is another technique that can provide information about the lattice structural and electrical properties of the material [17]. But, its mapping capability remains a problem. Selective etching is used to reveal the exact location of the defects [18]. However, this technique is destructive. White-beam synchrotron topography has been also used to study defects in SiC [19] [20]. This technique is costly with safety concerns. Deep Level Transient Spectroscopy (DLTS) is another technique to be used to study defects in SiC, as it was used in GaAs and other semiconductor materials [21] [22]. This technique is less costly, but it provides information about the energy level of the defects not their locations. As can be seen from above, the main problem of these techniques is that they suffer from one or more shortcomings. They may be expensive, destructive, unable to map defects, or cannot locate defects residing inside the material.
Acoustic Scanning Microscopy (ASM) is a viable alternative to the above techniques. It is a very cost-effective technique that can provide scanning for the crystal defects to determine their exact locations, shapes, and areas. Knowing the X-Y location allows for a targeted repair or avoidance of defects. Knowing the Z-location can help in determining the type and value of energy required for the repair. ASM is also used in imaging the samples to determine voids, cracks, and inhomogeneity inside the materials. One of the advantages of ASM is that it can be used for all types of materials such as, semiconductor, crystalline, amorphous, polycrystalline, metallic, insulator, nonmetallic, biological, or organic material.
In 1959, Dunn and Fry demonstrated the first experiment of acoustic microscopy [23]. Acoustic microscopy has been proposed for tissue characterization by Kessler and Yuhas [24]. The C-Scam ASM was developed by Kessler and others in the 1980's [25]. ASM can be performed using reflected or transmitted signals. In this study, the reflected echo microscopy is used. A detailed information about acoustic microscopy can be found in the book authored by Briggs and Kolosove [26].
Acoustic scanning microscopes use acoustic frequencies ranging from 5 -500 MHz. The frequency of high speed microscopecan reach up to 2 GHz. Low frequency waves can penetrate deeper in the sample, but has lower imaging resolution. On the other hand, high frequency waves penetration is shallower, but their resolution is higher, reaching the micron level. One of the advantages of ASM is that the samples do not need special preparation for scanning, as long as it can withstand water or other acoustically conductive liquid. Acoustic conductive medium is needed for interfacing the sample with the signal. Air is a poor conduc-

Experimental Results and Analysis
In this study, KSI-V8 [27], high-speed scanning acoustic microscope, is used to image and map the defects of three SiC wafers. The wafers are purchased from Cree Inc. This acoustic microscope can perform a nondestructive testing, imaging, and characterization at frequencies up to 500 MHz. Therefore, this microscope is capable of a resolution in the micrometer range. The maximum scan field is 400 × 400 mm 2 and the maximum magnification is 625x.
The test was performed on three 3-inch samples. These samples are shown in    The defects of Sample-1 are mapped, as shown in Figure 3. This sample is a semi-insulating crystal. The mapping shows that there are 37 defects in the 3-inch area. These defects are assumed to be intrinsic vacancies, deviancies, and other defects created during the material growth. Figure 4 shows the shape of the defects of Sample-1. Table 1 shows the area and the X-Y position of each defect. In all mapping, upper left corner of the graph is considered the origin of the X-Y axes. The defects are named as P1R1 to P1R37. The shapes and areas of the defects vary. The defects take different shapes, as shown in Figure 4. The size of the areas ranges from 0.0007 to 0.1238 mm 2 for defect P1R3 and P1R31. Figure 5 shows the defect mapping of Sample-2 and Figure 6 shows the shapes of the defects. Table 2 shows the area and the X-Ycoordinates for each defect. The wafer has 50 defects with defect areas ranging from 0.0007 to 0.1017 mm 2 . This sample is 6H, N-type SiC. The larger number of defects could be intrinsic defects, similar to Sample-1, with other defects due to doping. The density of defects for Samples-1 & 2 are high that renders the wafers unsuitable for electronic fabrication. For these wafers to be used, recrystallization to remove the defects needs to be performed. Crystal Structure Theory and Applications      Figure 7 shows the mapping of the defects for Sample-3, N-doped 4H SiC.
For this sample there is only one defect. This shows that this sample has little intrinsic or doping defects. Figure 8, shows the shape of the defect. As indicated in Table 3, the area of the defect is 0.1045 mm 2 . The difference between the samples is, more likely a result of the place of the sample in the ingot, although the type of the crystal, 6H vs. 4H, could be also a factor.

Conclusions
To the best of my knowledge, this is one of the few studies of defects in SiC substrates using ASM, if not the only one. The study suggests that defects can take any shape and cover an area greater that 0.1 mm 2 or a circular area of more than 300 μm diameter. Such defects render the wafer unusable for device or integrated circuit manufacturing, without some sort of repair.
This study can open the door for developing new processes and instruments.
First, the study shows that ASM technique is a cost effective technique to evaluate     location and the area, and B scan determines the defects Z-locations and thicknesses. With this information, the wave length capable of penetration the material to that depth can be determined, and the energy needed to repair the specific defect size can be estimated. In fact, these proposed methods and instruments can be adopted for other substrates and materials.

Conflicts of Interest
The author declares no conflicts of interest regarding the publication of this paper.