TCS 气流和预蚀刻对 4H-SiC 同质外延生长的影响

In this study, the epitaxial growth of 6-inch n-type 4° off-axis Si-face substrates using a horizontal hot-wall LPCVD system was investigated. The study explored the epitaxial growth under different source gas flow rates, growth pressures, and pre-etching times, with particular emphasis on their effects on epitaxial growth rate, epitaxial layer thickness uniformity, doping concentration and uniformity, and epitaxial layer surface roughness. The observation was made that the increase in source gas flow rate led to variations in dopant concentration due to different transport models between nitrogen gas and source gas. Additionally, with the increase in etching time, overetching phenomena occurred, resulting in changes in both dopant concentration and uniformity. Furthermore, the relationships between these three factors and their corresponding indicators were explained by combining the CVD growth process with the laminar flow model. These observed patterns are beneficial for further optimizing growth conditions in industrial settings, ultimately enhancing the quality of the growth process.

Introduction

Currently, most epitaxial growth processes are predominantly homoepitaxial, and the matured SiC device structures in the industry are primarily based on 4H-SiC. Consequently, the optimization of epitaxial techniques and growth processes in the industry is primarily focused on homoepitaxial growth of 4H-SiC.Previous research has shown that under conditions of significant fluctuations in source gas flow rates during epitaxial growth, especially when the magnitude of the flow rate variation is nearly equal to the magnitude of the source gas flow rate itself, there is a gradual transition of the epitaxial growth from 4H to 3C polytype as the source gas flow rate increases.5 Under specific source gas flow rate conditions, both 3C and 4H polytypes may simultaneously appear on the same epitaxial wafer. In order to determine the influence of source gas flow rates on homoepitaxial growth of 4H-SiC during homoepitaxial growth, and to provide better references for controlling the growth parameters of homoepitaxial 4H-SiC processes, it is necessary to vary experimental conditions. Specifically, under lower source gas flow rates, reducing the magnitude of uctuations in source gas flow rates and conducting experiments on the effects of different source gas flow rates on epitaxial growth under the premise of homoepitaxial growth. Additionally, chamber pressure also affects the flow rate of the source gas, the thickness of the boundary layer, and the stability of the gas flow, thereby influencing the kinetics of epitaxial growth. Therefore, it is necessary to study the relationship between chamber pressure and the quality of gas-phase epitaxial growth.

The quality of epitaxial layers depends not only on various process parameters of the chemical vapor deposition process but also significantly on the quality of the substrate.There is ample evidence from numerous studies indicating that many defects in epitaxial layers are caused by defects on the substrate surface or extensions of substrate defects.Common methods for eliminating defects in epitaxial layers caused by substrate surface defects involve treating the substrate surface to improve the quality of the epitaxial layer. For example, common substrate surface defects such as carbon inclusions and scratches can be eliminated through surface treatment. Pre-growth in situ etching with H2 is one such surface treatment process.At a certain temperature, using H2 for in situ etching can eliminate surface scratches, improve substrate surface smoothness, and enhance substrate surface quality. Importantly, through chemical reactions, a layer of SiC is removed from the surface while forming steps. The formal growth stage then involves step flow growth, where C atoms and Si atoms decomposed from the source gas molecules will adsorb and nucleate at the steps, ultimately completing the homogeneous epitaxial growth of SiC. Therefore, in situ etching is an important process for SiC homogeneous epitaxial growth.

Experimental 

To investigate the influence of source gas flow rate on the homogeneous epitaxial growth of 4H-SiC during the epitaxial growth process and to provide better references for regulating the growth parameters of 4H-SiC homogeneous epitaxy, experiments were conducted under the premise of homogeneous epitaxy, varying the source gas flow rates to observe their effects on epitaxial growth. The experimental groups were labelled as Group A, and the experimental procedure was as follows: at an epitaxial growth temperature of 1570 °C in a horizontal hot-wall CVD system, 6-inch 4H-SiC n-type substrates were selected, and epitaxial layer growth experiments were conducted on the (0001) surface of these substrates. Trimethylsilane (TCS) diluted with H2 and ethylene (C2H4) were used as the Si and C sources, respectively. During the formal epitaxial layer growth, the C/Si ratio was controlled at 0.72, and nitrogen (N2) was used as the n-type doping source. The flow rates of different source gases and doping gases were controlled, with reduced variations compared to previous experiments, to avoid the formation of the 3C crystal structure. During flow rate control, the flow rates of each step of the source gases and doping gases were varied in the same proportion. After determining the flow rates for one group of epitaxy, the flow rates for the other two groups of source gases and doping gases were set at 0.9 times and 1.1 times those of the experimental group, respectively. Samples with flow rates set at 0.9 times, 1.0 times, and 1.1 times were labelled as A1, A2, and A3, respectively.

Results and discussion

Source gas flow rate In Group A experiments, a graph was plotted with the TCS flow rate of the formal epitaxial layer as the horizontal axis. The relationship obtained is depicted in Fig. 1.

Fig. 1 Growth rate and thickness uniformity under different source gas flow rates.

However, the doping concentration appeared to deviate from expectations. Despite maintaining a fixed ratio of N2 to TCS flow rates at 0.61, as shown in Fig. 2, the doping concentration decreased in the final epitaxial wafers. This indicates a different scenario compared to the case where mass transport rate limits the source gas. In the epitaxial growth process of SiC, the limitation on the decomposition of N2 and its incorporation as doping atoms into SiC is governed by the chemical reaction rate. In other words, within the boundary layer, the limitation on the rate of N atoms entering the epitaxial layer arises from the decomposition rate of N2 within the boundary layer and the mechanism of N2 entering the epitaxial layer. Since the consumption rate is relatively slow, the concentration of N2 within the boundary layer does not differ much from that in the chamber environment. Without a concentration gradient, the effect of gas flow rate on doping is relatively small. The internal bond energy of N2 itself is high, and even at a high temperature of 1570 °C, the number of N atoms generated by decomposition is very limited.Therefore, simply increasing the flow rate of N2 does not correspondingly increase the number of doping atoms entering the substrate and thus the doping concentration. Additionally, due to the acceleration of the growth rate, the doping concentration will decrease accordingly. This is only part of the reason for the decrease in doping concentration. In the homogeneous epitaxial growth process of SiC, the mechanism of N atom doping competes for lattice positions with C atoms. If the concentration of C elements in the gas increases, the competitive ability of N atoms with C atoms decreases, resulting in a decrease in doping concentration. With an increase in the concentration of source gas, the concentration of C atoms within the boundary layer increases. However, N atoms are limited by the decomposition rate and do not increase correspondingly in concentration, resulting in a decrease in doping concentration.

Fig. 2 Doping concentration and doping uniformity under different source gas flow rates.

From the relationship between source gas flow rate and substrate surface roughness, as shown in Fig. 3, it can be observed that variations in source gas flow rate within this small range did not induce changes in surface roughness. The phenomena observed in previous experiments, such as step clustering on the smooth central region of 4H-SiC, which led to changes in surface roughness, did not occur here. In other words, as long as the growth process remains within the range of step-flow growth, the roughness of the epitaxial layer will not undergo significant changes.

Fig. 3 Surface roughness of epitaxial layers under different source gas flow rates.

Fig. 6 depicts AFM test images of the epitaxial wafer surfaces under different growth pressures. Here, Rq represents the surface roughness of the samples, measured in nanometers (nm). The relationship between epitaxial growth pressure and surface roughness is illustrated in Fig. 7. It can be observed that as the epitaxial growth pressure increases, the surface roughness exhibits a downward trend.

Fig. 6 AFM images of sample surface under different growth pressures. The shading of colors in the figure represents the surface height at that location.

The relationship between pre-etching time, growth rate, and epitaxial uniformity is illustrated in Fig. 8. From this graph, it can be observed that the growth rate does not vary significantly with pre-etching time. Compared to the growth rate of 21.3 mm h without pre-etching, when the pre-etching time is 12 minutes, the growth rate only decreases to 19.26 mm h^-1 , a change of less than 5%. However, there is an overall discernible trend where the growth rate slightly decreases with longer pre-etching times. Several factors may contribute to this trend. When the pre-etching time is short, the substrate surface roughness remains relatively high, and the substrate steps are not fully exposed. Numerous defects still exist on the substrate surface, leading to epitaxial growth that tends to nucleate at these defect sites, resulting in epitaxial layers of potentially different crystallographic orientations from the substrate and non-strict step-flow growth. Under the influence of various growth factors, the epitaxial growth rate is relatively high. As the growth time increases, the substrate steps are fully exposed, and the surface roughness becomes smoother and more regular, providing better conditions for homoepitaxial growth. However, with further prolongation of the pre-etching time, over-etching of the substrate surface may occur, resulting in wider and less regular steps. Such step morphology is not conducive to step- flow epitaxial growth and may even lead to two-dimensional nucleation phenomena. Consequently, the growth rate further decreases, and the quality of the epitaxial layer may degrade.

Fig. 8 Growth rate and thickness uniformity of epitaxial growth under different pre-etching times.

Conclusions

In summary, we conducted SiC epitaxial experiments on 6-inch n-type substrates with a 4° off-axis orientation. By controlling process parameters such as source gas flow rate, chamber pressure, and pre-etching time, we investigated the relationship between these parameters and the quality of the epitaxial layer. Our focus in evaluating epitaxial quality included parameters such as epitaxial growth rate, thickness uniformity, doping concentration, uniformity of doping concentration, and surface roughness of the epitaxial layer.

In scenarios where the variation in source gas flow rate is minimal, distinct epitaxial outcomes were observed under the same growth system and process conditions. Under minor changes in source gas flow rate, the epitaxial growth mode remained unchanged, preserving homogeneous epitaxy. The epitaxial growth rate exhibited a linear increase with the source gas flow rate, while the uniformity of epitaxial layer thickness improved with increasing flow rate. However, no significant correlation was observed between the surface roughness of the epitaxial layer and the source gas flow rate. Conversely, the doping concentration decreased with increasing source gas ow  rate, although the uniformity of doping concentration did not exhibit noticeable variations.

The pre-etching time has minimal impact on the epitaxial growth rate, but it generally exhibits a slight decrease as the pre-etching time prolongs. Moreover, an increase in pre-etching time leads to a reduction in substrate surface roughness and an improvement in the uniformity of epitaxial layer thickness. However, with further increase in pre-etching time leading to over-etching, the thickness uniformity deteriorates. Similarly, the doping concentration and its uniformity show a similar trend with increasing pre-etching time.