Influence of substrate properties on the defectivity and minority carrier lifetime in 4H-SiC homoepitaxial layers

The quality of epilayers is an important factor for the yield in device production as well as for the performance and reliability of SiC devices. Important quality aspects are, e.g., the homogeneity and reproducibility of epilayer doping concentration and thickness as well as the density of extended defects like dislocations and stacking faults and the minority carrier lifetime. The minority carrier lifetime is essential with regard to bipolar device performance as it is crucial for conductivity modulation and switching losses. The minority carrier lifetime of the epilayer itself cannot be easily determined, but an effective carrier lifetime can be measured for epiwafers, which is influenced by the surface recombination, epilayer thickness and doping concentration as well as bulk lifetime of the epilayer [1]. For typical measurement conditions on low doped epilayers, the bulk lifetime is equivalent to the Shockley-Read-Hall (SRH) lifetime, which depends on point defects and extended structural defects such as dislocations and stacking faults [2]. This paper aims to identify the critical properties of the underlying substrate for the defectivity of the epitaxial layer and the effective carrier lifetime of the epiwafers. Therefore, the epitaxial growth-related parameters thickness, doping level and point defect concentration (especially Z1/2) were kept constant for all epilayers by means of multi-wafer epitaxial growth. A set of 16 substrates has been selected with a large spread of specific resistance and dislocation densities as summarized in Table I. The set consists of wafer series A, B and C cut from three different crystals, which were intentionally chosen with respect to the targeted dislocation density of the 16 substrates. All substrates are 100 mm in diameter with 0 micropipes/cm². Epitaxial growth has been done by two fully loaded runs in an AIXTRON G5 WW reactor in 8 x 150 mm configuration (loaded with 100 mm wafers) under identical growth conditions. Trichlorosilane (TCS) and propane have been used as precursors. The mean epilayer thickness is 35.1 µm with ± 1 % reproducibility and the net carrier concentration of the nitrogen doped epilayers amounts to 8.25 x 1014 cm-3 ± 12 %. Epidefects were determined by UVPL imaging with subsequent image analysis for defect classification and quantification using the Intego defect luminescence s canner [3]. It is found that the density of bright triangles, mixed triangles and bar-shaped stacking faults is correlated with dislocation densities of the underlying substrates as shown in Figure 1. We do not find any correlation between stacking fault densities and epigrowth run or reactor hardware. We will discuss the relationship between dislocations in the substrate and stacking faults in the epilayer in more detail in the paper. The effective minority carrier lifetime of the epiwafers has been measured by µ-PCD method in the Semilab WT-2500 [2]. All epiwafers show a radially symmetric lifetime profile, with lower lifetimes in the wafer center and higher lifetimes at the wafer edge as shown in Figure 2a. This finding corresponds well to the doping profile of the epilayers. Besides that, carrier lifetime can be reduced locally by 30 % by dislocations and stacking faults. The mean effective lifetimes of epiwafers are statistically analyzed in Figure 2b: the carrier lifetimes are comparable for both epigrowth runs in our study, which underlines the reproducibility of the epitaxial growth in the G5 WW reactor. Additionally, it can be clearly seen in Fig. 2b that the wafer series influence significantly the effective minority carrier lifetime of the epiwafers. We will discuss this finding in the view of the respective electrical properties of epilayers (especially point defects) and substrates (specific resistance).