introduce
Silicon carbide (SiC) is one of the most attractive and promising wide bandgap semiconductor materials, with excellent physical properties and huge potential for electronic applications. At present, 150mm diameter SiC wafers have been commercialized, and due to the expansion of the market, a lot of effort and resources have been invested, and the dislocation density of the wafer has been reduced to 5000/cm2. The most successful method for growing large SiC crystals with high quality is the seed crystal sublimation process, which inevitably leads to inhomogeneous nucleation and thermal stress leading to the defects of the growing SiC wafer.
Especially in the initial stage of growth, many dislocations are created near the interface between the seed crystal and the growing crystal. Nucleation of dislocations at the beginning of growth is due to strain caused by thermal stress and lattice mismatch caused by the difference in nitrogen concentration between seed crystal crystals and growing crystals. Recently, H. Suo et al. reported an increase in penetration dislocations (TDs) of SiC crystals grown under different temperature conditions. Crystals grown at high temperatures have a significant increase in TD at the initial stage of growth compared to crystals grown at low temperatures. The increase in TDs is due to lattice mismatches due to thermal stress and/or differences in nitrogen concentration between seed and growing crystals.
Especially for the lattice mismatch caused by the difference in nitrogen concentration, the lattice constant of 4H-SiC at different nitrogen concentrations and the lattice mismatch caused by doping were calculated. S. Sadaki et al. reported lattice shrinkage at high temperatures for diazo-doped and a calculated lattice mismatch (Δd/d) of 1.7x10-4 between lightly doped epitaxial layer (6x1014/cm3) and heavily doped substrate (2x1019/cm3) at 1100°C. Therefore, due to the difference in doping concentration, the mismatch around the growth temperature of 2200~2500 °C is much larger than that of room temperature. However, the formation of defects due to lattice mismatch due to the difference in nitrogen concentration between seed crystals and growing crystals at the beginning of growth has not been studied.
experiment
The purpose of this experiment was to analyze the effects of nitrogen concentration difference and thermal stress on dislocation nucleation in crystals at the beginning of growth. 4H-SiC single crystals with a diameter of 4 inches were grown on 4-degree off-axis seed crystals by PVT method. The seed crystal and silicon carbide source material are placed separately in sealed graphite crucibles with insulation. Crystal growth is performed at the same growth temperature and pressure. Controlling the crucible bottom temperature and measuring the variation of crucible top temperature within ±5°C can eliminate the influence of thermal stress differences between growing crystals. To investigate the relationship between nitrogen concentration in the initial phase to dislocation nuclei and lattice mismatch, crystals are grown in different proportions of nitrogen flow and/or shape during the heating and decompression phases, as shown in Table 1. Ar gas and N2 gas are introduced into the crystal growth chamber, fixing the total flow.
Table 1 Nitrogen ratio of samples during heating and decompression phases.
The grown SiC ingots were sliced and polished to prepare x-section samples < 11-20 >. Secondary ion mass spectrometry (SIMS) was used to perform depth profile analysis of sample nitrogen concentration. SIMS measurements are performed using Cs+ Gun with an impact energy of 15 keV and a current of 50 nA. The analysis area was 60um (Φ) and the detected ions were 30Si and 30Si14N. In PAL (Pohang Accelerator Lab), synchronized X-ray topographic images were obtained using a white beam source with reflection patterns to study the generation of penetrating dislocations in the initial stages of growth. After corrosion at 500°C KOH for 15~25 min, the density of etching pit (EPD) was determined.
Results and discussion
We minimized temperature differences during operation, which helped to study the effect of nitrogen concentration on dislocation nucleation and rule out the effects of thermal stress. The amount of nitrogen added in SiC crystals is affected by the growth rate and growth temperature, the growth temperature changes in the range of ±5°C, and the growth rate changes in the range of ~0.05 um/hr. At lower growth temperatures and lower growth rates, the greater the amount of nitrogen incorporated in the crystals.
The N2 atomic concentration (atoms/cm3) measured with SIMS is shown in Table 2. For each sample, the N2 atomic concentration was measured at several locations, including: seed crystal, upper seed crystal 100um, 300um, and 1mm positions. The N2 atomic concentration of the growing crystal (1 mm above the seed crystal) is highly dependent on the proportion of N2 gas introduced, as shown in Figure 1. Except for the samples (d), the N2 atomic concentration increased significantly in all samples at the beginning of growth, and the atomic concentration decreased monotonically with growth until it grew more than 300um from the seed crystal. Samples (d) were grown at an extremely low nitrogen flow ratio of 0.0375.
Table 2 Relationship between nitrogen atomic concentration and crystal position in the sample
Fig. 1 The relationship between the concentration of N2 atoms in the growth crystal and the proportion of N2 gas entering the crystal growth chamber to the total gas flow.
We studied the X-section samples grown at different nitrogen flows in Table 1. The optical image of the sample (a-1, b-1, c-1, d-1, e-1) and X-ray topographic image (a-2, b-2, c-2, d-2, e-2) are shown in Figure 2. These optical and X-ray topographic images were taken from the center of the X-profile sample. From the nitrogen atomic concentration measured in Table 2 and the optical image in Figure 2, we can assume that the initial stage of growth is from seed crystal to 300~400um, corresponding to the decompression phase. The average growth rate in the initial stage is assumed to be about 30um/hr.
Figure 2 shows. Optical images (A-1, B-1, C-1, D-1, E-1) and X-ray topographic images (A-2, B-2, C-2, D-2, D-2, E-2) of X-profile samples grown at different nitrogen flows in Table 1.
Sample (a) is grown at a ratio of nitrogen to total gas flow of 0.5, and we can see a very dark interface between seed and growth crystals in the optical image (a-1). Sample (a) had a nitrogen atom concentration of 1.6E+19 atoms/cm3 at a distance of 1 mm from the seed crystal, and a correlation resistivity of 0.012 Ω-cm. In the X-ray topography (A-2) of diazo-doped sample (A), dislocation-free thick bands were observed in crystals grown to ~200um, after which a large number of penetrating dislocations began to appear. In sample (b), we found that the penetration dislocation is generated at the interface. The number of newly generated penetrating dislocations is 50% greater than the number of penetrating dislocations transferred from seed crystals. Sample (b) was grown with a nitrogen ratio of 0.15, and the difference between the nitrogen atom concentration at 100um from the seed crystal was 1.2E+ 19 atoms/cm3. Samples (c) ~ (d) show that only a few new penetrating dislocations are observed at the interface. Compared with samples (a) and (b), the difference in nitrogen atomic concentration between seed crystals and growing crystals at 100um away from seed crystals was less than 4.0E+18 atoms/cm3 for samples grown at a lower nitrogen ratio. In particular, the interface of sample (d) is very thin, as shown in Figure 2 (d-1). The concentration of nitrogen atoms in the growing crystal at 100um from the seed crystal is not higher than that of the seed crystal. The interfacial thickness of sample (e) is similar to that of other samples (a, b, c), but we found from the optical image (e-1) that the nitrogen concentration in the initial stage is very low, and the results of measuring nitrogen concentration are shown in Table 2. However, in regrown silicon carbide, short, curved dislocations and base dislocations exist at very high densities. The cause is unknown, but in sample (e), the existing penetrating dislocation has been transformed into a base dislocation.
Figure 3 X-ray topography of the X-profile sample edge locations corresponding to nitrogen flows (b) and (c) in Table 1 (b-3, e-3). The crystallographic orientation of the X-ray topography image is the same as in Figure 2(f).
Figure 3 shows the X-ray topography of samples (b) and (e) in Table 1. These X-ray topographic images are taken from the edges of the X-profile sample. From the X-ray topography of the edges, it can be seen that the dislocation formation results from the center to the interface are similar, but the dislocation density of the edge in Figure 2 is greater than the center dislocation density. The dislocation density of the selected sample was studied by molten KOH etching. These samples were grown using similar seed crystal qualities, and the difference in defect density evolved from seed crystals was negligible. The dislocation densities of the selected samples (b), (c), and (e) are shown in Table 3, and the associated defect images are shown in Figure 4. The sampling position is 1 mm away from the seed crystal.
Table 3 Dislocation density of selected samples after KOH etching.
The measurement of dislocation density after KOH etching is consistent with the observation of X-ray morphology. The penetrating edge dislocation density (TED) of sample (b) is almost 3~5 times that of sample (c) and sample (e), and the base surface dislocation density (BPD) tends to be consistent with the penetrating edge dislocation density, as shown in Table 3.
Figure 4 Optical image of KOH etched sample at different nitrogen flows in Table 1.
Another factor to consider is that a mismatch of nitrogen at the growth front can change the surface morphology. D.D. Avrov et al. reported that nitrogen actively prevents three-dimensional growth by reducing the height of microsteps on the crystalline surface, and reports that the surface catalytic mechanism of nitrogen makes it grow layer by layer. However, in nitrogen-heavily doped 4H-SiC crystals, higher tiny steps are generated, transforming the penetrating dislocation into a base dislocation. By overlaying higher microsteps over the penetrating dislocation, the growth direction of the dislocation is changed to 90° and the growth direction of the dislocation is transformed into the basal dislocation.
Figure 5 (a) X-ray topography of sample (b) in Table 1. The orientation of the X-ray slices is <1-100 >. The crystallographic direction of X-ray topography images is (b).
We observed thick bands without dislocations in diazo-doped sample (a) and extensive threading dislocations at the interface of sample (b). For closer inspection, we obtained an X-ray topography image of G=101-4, as shown in Figure 5. In the X-ray topography of the initial growth area (b), a penetrating dislocation sliding towards the basal surface was found.
conclusion
We investigated the generation of penetration dislocations at the beginning of the growth of 4H-SiC single crystals by PVT method. In order to elucidate the effect of nitrogen concentration difference between seed crystal and growth crystal on dislocation nucleation at the early stage of 4H-SiC single crystal growth, crystals were formed under the same growth conditions under different nitrogen flow and/or morphology to eliminate the effect of thermal stress differences between samples. The N atomic concentration of the growing crystal is highly dependent on the proportion of nitrogen introduced, and the N atomic concentration increases sharply at the beginning of growth. The creation of new penetrating dislocations at the interface depends largely on the difference in nitrogen atomic concentration between seed and growth crystals. In samples with low N atomic concentration differences, penetration dislocations are rarely generated, while in samples with high N atomic concentration differences, penetration dislocations nucleation occurs at the interface. In the initial stage of the PVT growth process, the generation of penetration dislocations caused by lattice mismatch caused by the difference in nitrogen concentration between seed crystals and growing crystals was studied, and the appropriate nitrogen flow and distribution in the heating and decompression phases were determined.