3.1. Properties of the SiC Powder Sources
The size distribution and additional powder properties are presented in Figure 1
and Table 1
. The particle size distribution of the SiC powder from SIKA shows a narrow size distribution, with a span of only 1.2. The mean diameter was 324 µm with a d90 and d10 of 160 µm and 555 µm, respectively. The LPD value was 1.6 g/cm3
. The packing density of the powder along with high mean particle diameter are important factors to increasing efficiency and stability in the crystal growth process.
The size distribution of the SIKA High Purity powder was developed to optimize the crystal growth process. The high packing density enables higher yield in PVT reactors. The large particle size and narrow distribution enables stable growth. Rounded grains would also improve stability during sublimation. The low free carbon and total oxygen contents are a result of the powder processing and are low compared with typical SiC products.
The SEM image of the powder in Figure 2
a shows the grains were sharp, as is typical for SiC produced by the Acheson process. The optical image in Figure 2
b shows the color of the powder is light green. Pure SiC is colorless, so the green tint is most likely due to nitrogen impurity. The nitrogen is introduced in the crystal during synthesis as the Acheson process is done in air. The image in Figure 2
c depicts a piece of SiC ingot before processing, which clearly shows the range of colors that forms in the crude from white to dark green, corresponding to the diffusion of impurities through the furnace during synthesis. Sorting the material by color prior to processing helped increase the purity of the final product.
presents a typical trace element analysis for the SIKA High Purity SiC powder. The total impurity is less than 17 ppm with metallic impurity below 14 ppm. The Acheson process and the raw materials used limit the purity that can realistically be achieved in a high capacity industrial process and therefore cannot compare to powders produced by CVD or other similar methods reaching 6N purity and higher.
3.2. Sublimation Behavior of the SiC Powder
Two SiC boules with a single crystalline diameter of 75 mm were grown using the SIKA and FAU SiC powder source for comparison. The average growth temperature and ambient inert gas pressure were set to 2050 °C and 20 mbar, respectively. Nitrogen gas was added in order to perform n-type doping. Simultaneously to the growth process, in situ 3D computed tomography X-ray visualization [12
] was applied to study the sublimation behavior of the novel SIKA SiC powder source. Figure 3
illustrates, for both growth runs, the evolution of the SiC source material and SiC crystal growth process. In the case of the SIKA powder, in order to investigate the tendency of incorporation of carbon dust-like particles inside the growing SiC boule, no carbon dust shield was placed between source and seed. As a consequence, the source powder tends to rise up towards the crystal growth with progressing growth time. This trend is typical for all kinds of SiC source materials and not specific to the application of the SIKA powder. It can be suppressed in an optimized growth environment by marginal adaptions of the design of the hot zone. The formation of a needle-like structure on the top of the source material is observed, which is a typical behavior related to the concept of the applied growth setup (see, e.g., [9
]). A significant difference between the SIKA and FAU powder is related to the difference in packaging density of 1.80 g/cm3
(SIKA) versus 1.15 g/cm3
(FAU). In the case of the much lower value for the FAU powder, a much faster shrinkage of the core of the SiC source material is observed, which causes a larger change of the growth conditions related to the temperature field. In the case of the SIKA source material, however, a more smooth and continuous consumption of the SiC powder is observed. Although the SIKA powder undergoes larger morphological changes in its top area because of the missing carbon dust shield, the grown crystal exhibits a much flatter growth interface which is beneficial to reduce thermo-elastic stress. The top area of the SIKA powder even tends to adapt with a slightly concave surface to the slightly convex crystal growth interface.
Optical analysis was carried out on a series of wafers cut from the SiC boule grown with the SIKA powder in order to investigate the occurrence of unwanted carbon dust particles from the SiC source. Within the resolution of the optical microscope of ca. 1 µm, no evidence for an incorporation of carbon particles from the SIKA powder into the grown SiC boule was found. In addition, KOH defect etching did not show an increase in the micropipe density that would be an indication of the presence of carbon particles even below the optical resolution. In both cases, the best areas of the SiC crystals grown by the FAU and SIKA powders exhibited a micropipe density below 5 cm−3. Areas with higher micropipe density were related to defective areas in the starting SiC seed wafer and are not related to the growth process itself.
In order to investigate the polytype stability during the application of the SIKA powder in the PVT system, Raman spectroscopy was carried out subsequent to the growth process on one wafer cut from the boule (Figure 4
a). Solely a Raman signal related to the 4H-SiC polytype is observed, which is a proof for the stable PVT growth conditions using the SIKA powder. Compared to the reference crystal grown with the FAU SiC powder (Figure 4
b), the position of the FTO line at ca. 776 cm−1
indicates comparable low stress values in the two crystals. The slightly smaller line width of the FTO line in the case of the FAU SiC source is related to a slightly higher SiC seed quality, but not to the applied SiC powder properties.
As an important conclusion, the morphology of the SIKA SiC powder does not release carbon dust particles at the sublimation interface that would interfere with the crystal growth process, causing the incorporation of macro-defects into the SiC boule. In addition, birefringence measurements did not indicate an increase of macro-defects in the SiC boule while using the SIKA instead of the FAU powder.
Another two SiC boules, however, with a single crystalline diameter of 100 mm were grown using the SIKA and FAU SiC powder source. The average growth temperature and ambient inert gas pressure were set to 2100 °C and 5 mbar, respectively. Nitrogen gas was added in order to perform n-type doping. Simultaneously to the growth process, in situ 2D X-ray imaging [9
] was applied to study sublimation behavior in a setup that matches an industrial crystallization tool. Figure 5
shows an optical image of the top of the grown SiC boule which exhibits a predominantly mirror-like surface.
depicts for both growth runs, i.e., with SIKA and with FAU SiC powder, the evolution of the SiC source material and SiC crystal growth process. Similar to the case of the 75 mm SiC growth runs presented in Figure 3
, the SIKA powder also exhibits a smoother sublimation-recrystallization behavior in the 100 mm SiC growth run than the FAU powder. As an example, in Figure 6
, the SIKA SiC source exhibits one dense SiC powder block even after 77 and 119 h of growth. In the case of the FAU SiC source, however, the source separates in different SiC areas which are most pronounced after 45 and 67 h of growth (Figure 6
). This observation is again to a large extent related to the difference in packing density of 1.73 g/cm3
(SIKA) vs. 1.44 g/cm3
(FAU). The crystal growth interface appears with a slightly more convex shape in the case of the SIKA powder compared to the FAU powder. As a high density in the powder area supports the formation of a temperature field with very low radial thermal gradients, the hot zone was adapted to introduce a stronger thermal gradient prior to the growth run with the SIKA powder. In the evolution of the growth run the curvature of the crystal is maintained steady over the growth period.
Based on the evolution of the growth process shown in Figure 5
, the SIKA powder source material exhibits the properties for long term crystal growth runs. There is no visible gap between the crucible wall and the dense disk of the SIKA powder even after 119 h crystal growth while the FAU powder develops such a gap after only 45 h growth time. This enables much more constant growth conditions, e.g., the temperature field for long growth runs. In addition, very little morphology change in the images between 77 h and 119 h can be observed for the SIKA powder, indicating a very steady sublimation behavior for extended growth times. As a result, in the PVT setup used for the crystals as shown in Figure 6
, the processing time using the SIKA source material could be prolonged to achieve a crystal height of 40 mm to 45 mm in the center.