CN / EN
NEWS CENTER
NEWS CENTER
The influence of buffer layer thickness on defect rate in 8-inch 4H SiC homoepitaxial growth
Release time:2024.12.16 Number of views:16

introduction

Due to its excellent physical properties, wide bandgap semiconductor material silicon carbide (SiC) is an ideal candidate material for power electronic devices. Compared with silicon, it can achieve faster switching speed, lower leakage current, and higher breakdown voltage.

The formation of buffer layer (BL) is a key step in chemical vapor deposition (CVD) process, which aims to suppress substrate defects (such as dislocations, particles or stains) to improve the crystal quality of the grown epitaxial layer. In fact, many defects can cause serious damage to power devices as they increase leakage current, alter the current voltage relationship, and reduce device reliability. In addition to other parameters such as doping or growth rate [8], optimizing the BL thickness can help improve the crystal quality in SiC epitaxy.

In order to optimize the epitaxial growth conditions, we studied the effect of increasing the thickness of the buffer layer on morphological and crystal defects. This study was conducted on n-type doped 8-inch 4H SiC substrates from two different suppliers.

Experimental methods

Growth: An n-type doped SiC epitaxial layer is grown on a 200mm 4H SiC substrate with a 4 ° offset (112 ̅ 0) plane (0001) Si plane from two different suppliers (represented here as A and B). The growth experiment was conducted in a commercial low-pressure hot wall monolithic CVD reactor based on trichlorosilane (SiHCl3) - ethylene (C2H4) chemistry, using nitrogen based dopants and hydrogen gas as carrier gases. The epitaxial step is carried out at a temperature of approximately 1650 ° C, with a C/Si ratio of about 1. The doping concentration of the buffer layer is about 1018 at/cm3, while the doping concentration of the epitaxial layer is about 1016 at/cm3. All substrates have ultra-low microtubule density and have undergone epitaxial polishing on the Si surface and fresh cleaning before growth.

Two types of BL thicknesses were studied: thin BL samples were grown using standard thicknesses for commercial purposes, while thick BL samples were grown by doubling the BL deposition time (thereby increasing the BL thickness) while keeping all other growth parameters constant. The experiment ensured that each thin BL and thick BL chip pair could be compared, coming from closely related substrates with comparable crystal quality and grown in continuous CVD operation.

Characterization: Defect characterization mainly uses KLA Tencor's Candela ® Tools are used. This tool is based on scattering light methods with different laser incidence angles and equipped with photoluminescence channels, which can monitor crystal defects and morphological defects with high spatial resolution and sensitivity. Figure 1 shows an example of typical epitaxial defects detected by Candela, with their feature sizes reported in parentheses.

picture

Figure 1: Typical Example of Defects after Epitaxy and Their Characteristic Sizes

By using KLA Altair ® The bright field optical microscope of the tool further monitors chip defects. This method is only sensitive to morphological defects and can verify the distinction between crystal defects and morphological defects based on Candela detection.

Control measurements of thickness and doping concentration distribution were conducted using FT-IR and Hg probe capacitance voltage methods, respectively (data not shown). No correlation was observed between thickness or doping distribution and BL thickness or observed defect patterns.

result

Bright Field Optical Microscope (KLA Altair) ®): For example, the left figure in Figure 2 shows the distribution of Altair defects in two sample pairs (thin BL and thick BL) from the same supplier, grown on closely related substrates with comparable crystal quality during continuous CVD operation. The right figure in Figure 2 shows the trend of morphology defect grain percentage for thin and thick BL epitaxy from two suppliers. In all cases, the chip defect rate of thick BL samples is higher than that of thin BL samples.

picture

Figure 2: Distribution of morphological defects and percentage of defective grains

Scattering light method (KLA Candela) ®): Figure 3 illustrates the Candela detection results of two thin BL/thick BL chip pairs from two substrate suppliers (top: supplier A; bottom: supplier B). For each pair, the image displays the defect distribution maps of two chips, as well as the relative defect counts for each defect type. The chart in Figure 4 reports the average defect counts observed in thin BL (orange) and thick BL (blue) chip pairs from supplier A (5 sample pairs), sorted by defect type.

picture

Figure 3: Example of Candela detection results

Defect types are further divided into morphological defects (detected by scattered light) and crystal defects (detected by photoluminescence). Although the statistical data is limited (5 chip pairs), the results seem to indicate an increase in surface morphology defects compared to the standard value in the case of thick BL. This trend was observed in all defect types of all chip pairs, except for the surface triangle (see Figure 4), where the average thickness of the two BL types was comparable. The observed trend for crystal defects is variable.

picture

Figure 4: Average Defect Count of Supplier A Samples

conclusion

In this study, in order to optimize the epitaxial growth conditions on an n-type doped 200mm 4H SiC substrate, we investigated the effect of increasing the thickness of the buffer layer on the crystal quality of the drift layer, including morphological and crystal defects. This study was conducted on substrates from two different suppliers (A and B).

This study is part of the ongoing large-scale evaluation campaign on 200mm epitaxy. Although it is a preliminary result, our findings indicate that the crystal quality is related to the thickness of the buffer layer, which was observed in both suppliers in the repeated experiments. In particular, the data shows that compared to the standard thickness, there is an increase in morphological defects in thicker BL cases, as can be seen from optical microscopy (Altair) and scattered light (Candela) measurements. Due to better crystal quality (which also means less defect variation), this trend is more pronounced in supplier A.

This article is sourced from the internet and the copyright belongs to the author. If there is any infringement, please contact the editor to delete it