5.5 kV Bipolar Diodes From High Quality CVD 411-SiC
- PDF / 1,048,518 Bytes
- 6 Pages / 414.72 x 648 pts Page_size
- 65 Downloads / 160 Views
119
Mat. Res. Soc. Symp. Proc. Vol. 512 © 1998 Materials Research Society
RESULTS AND DISCUSSION One major difficulty in growing thick layers, especially at a growth rate of 10 gm/hr, is trying to maintain a good surface morphology without 3C inclusions. Normally, the thicker the epitaxial layer and the faster the growth rate, the more pronounced some epitaxial defects such as step bunching or decorations around micropipes become. Also, disturbances such as step bunching may occur, depending on the growth conditions. Therefore, the conditions for growth are chosen in order to optimize the surface morphology, that means growing at very low C/Si ratios. We also found that the window of operation becomes more narrow as the growth rate is increased. The morphology is greatly influenced by the growth conditions and the design of the susceptor. Defects associated with the substrate such as micropipes, pits and dislocations, become amplified when thick layers are grown as demonstrated in the Figure 1. For example, this particular defect is about 210 jim long for a 30 jtm thick layer.
a Figue th1.Shos sufacemorholoy Figure 1. Shows is 215 jim. Feature layer. thick jim 30 typical
This defect has been observed by several groups [3,4] and is referred to as the "carrot" defect [4]. These defects begin at the generally substrate/epitaxial interface since their length (L) is usually the sarie and equal to (epi thickness)/L = tan 8. The tan 8° comes from. the fact that we are growing on 8 off-axis wafers. We have found that the number of carrots per unit area is independent of the micropipe density of the wafer and can be different for layers grown on wafers from the same boule. Currently, we are and tryingit. to this defect find ways to minimize or eliminate investigating 3
The n-type doping has been controlled reproducibly from high 1014 to low 1019 cm" through the introduction of nitrogen in the gas phase. Obtaining a low intrinsic background and good surface morphology can be a challenging task. This is especially true when the ultimate morphology occurs at lower C/Si ratios, where the intrinsic background doping reaches its highest n-type values. The intrinsic background doping is generally low in the hot-wall reactor as previously shown [2]. However, by increasing the growth rate we can reduce the background doping further. Hence, we can tailor our doping. For process to grow at low C/Si ratios and still obtain low intrinsic background example, at 10 jim/hr our background doping is about 3 x 1014 cm 3 n-type. Figure 2 shows a low temperature PL spectrum of an extremely pure layer. Besides the nitrogen bound exciton related lines, Ph, and Qhv , strong free exciton-related luminescence is seen. Most encouraging to see is that the layer appears to be very low doped, uncompensated n-type since the Al and B bound excitons are not detectable. Q0 line and the 176.7line, the doping concentration was From a comparison between the determined to be 7.6 x 1013 cm 3 [5]. We have compared the PL spectra from samples that were grown
