Abstract
Basal plane dislocations (BPDs) have been a concern for SiC high-voltage bipolar devices for many years as they source Shockley-type stacking faults in the presence of an electron-hole plasma and reduce minority carrier lifetimes [1]. There has been much focus on the reduction of BPD densities including pre-growth treatments [2], growth interrupts [3], and proton implantation [4]. It has been demonstrated that with sufficient carrier injection, BPDs from the highly doped buffer layer or substrate can expand and propagate into the drift layer [5], causing device failure. Therefore, a technique is needed to prevent BPD faulting during high carrier injection for high peak, pulsed power applications. In this paper, we use an in-situ H2 etch process to mitigate BPD faulting under high-power UV exposure at 12 kW/cm2, which is equivalent to ~1x1019 cm-3 carrier injection in the substrate. Intentionally doped (ID) n-type (~ 5x1015 cm-3) epitaxial layers were grown on 4° off-axis substrates in a horizontal hot-wall chemical vapor deposition reactor using the standard chemistry of silane (2% in H2) and propane. Growth was conducted at 1620 °C, 100 mbar, and a C/Si of 1.55. A 6 µm buffer layer (n-type ~ 2x1018 cm-3) was grown prior to the ID drift layer. A H2 etch was performed at different stages of the growth to determine its impact on BPD faulting. The etch was conducted at 1665 °C and 70 mbar for 50 min, either before the buffer layer (BL), after the BL, or both before and after the BL. A fourth sample was grown with a H2 etch prior to the BL and a growth interrupt after the BL. The growth interrupt consisted of reducing the temperature to 1000 °C in H2 and immediately ramping back to the growth temperature, after which the ID drift layer was grown. A control sample was grown without any etches or growth interrupt. Ultraviolet photoluminescence (UVPL) imaging was used to identify the BPD faulting behavior after epitaxial growth, and before and after UV stressing. When exposing the control sample to 300 W/cm2, the BPDs present in the BL faulted after 420 s of UV exposure and continued to propagate throughout the drift layer. For the sample with a H2 etch between the BL and ID drift layer, only the pre-existing BPDs in the drift layer faulted at 300 W/cm2, and no BPDs from the BL faulted after > 1240 s. Similar results were found for the sample grown with a H2 etch before and after the BL. When these two H2 etched samples (etched before and after the BL, and etched after the BL) were stressed at an increased UV exposure of 1 kW/cm2, again, no faulting of BL or substrate BPDs were observed after 1900 s. With 10X higher power density (12 kW/cm2), BPD faulting from the substrate occurred for both of the H2 etched samples. However, for the sample with just a single H2 etch between the BL and drift layer, the density of additional BPD faulting was ~ 200 BPDs/cm2, whereas when the H2 etch was performed before and after the BL, the density of newly faulted BPDs was only ~ 50 BPD/cm2. For the sample grown with a H2 etch before and a growth interrupt after the BL, BPDs did not expand at 12 kW/cm2. Carrier concentration was simulated as a function of epitaxial layer depth for the various UV laser powers. At the highest power density, the injected carrier concentration is higher than the doping of all the layers, which should lead to recombination induced BPD faulting in the substrate, BL and ID epilayer. We propose that H2 from the H2 etch and growth interrupt pin the BPDs and prevent them from expanding. This level of suppression of BPD expansion, to our knowledge, has not been demonstrated before at these power densities.[1] J.P. Bergman, et al., Mater. Sci. Forum Vol. 353-356, 299 (2001).[2] N.A. Mahadik et.al., Mater Sci Forum 858, 233 (2016).[3] R. E. Stahlbush, et al., Appl. Phys. Lett. 94, 041916 (2009).[4] M. Kato, et al., Sci. Rep., 12, 18790 (2022).[5] N.A. Mahadik, et. al., Appl. Phys. Lett., 100, 042102 (2012).
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have