The current–voltage (I-V) characteristics of the PiN diodes in regions with and without proton implantation before the pulsed-current stress are shown in Fig. 1. PiN diodes with proton implantation show rectifying properties similar to those without proton implantation. We plotted frequencies of voltages at a forward current density of 2.5 A/cm2 (corresponding to 100 mA) to statistically evaluate the effects of proton implantation as shown in Fig. 2. The curves fitted by the normal distribution are also indicated by the dotted lines. As illustrated by the peaks of the curves, the on-state resistance slightly increased with proton doses of 1014 and 1016 cm− 2, whereas the PiN diodes with proton doses of 1012 cm− 2 showed almost the same performance as those without proton implantation. We also performed proton implantation after PiN diode fabrication, and the diodes did not exhibit uniform EL, as shown in Fig. S1, due to the damage caused by the proton implantation, as reported in previous studies.37–39 Therefore, annealing at 1600°C after Al ion implantation which is an essential process for device fabrication to activate Al acceptor recovered the damages induced by proton implantation, resulting in similar I-V characteristics between the PiN diodes with and without proton implantation.
EL images of the PiN diodes at a current density of 25 A/cm2 after electrical stress are shown in Fig. 3. Before the pulsed-current stress is applied, no dark region is observed for any of the diodes, as shown in Fig. S2. However, as shown in Fig. 3(a), after applying electrical stress, several bar-shaped dark regions with bright edges are observed in the PiN diode without proton implantation. Such bar-shaped dark regions in EL images were observed for 1SSFs expanded from BPDs in the substrates.28,29 In contrast, a few extended stacking faults were observed in the proton-implanted PiN diodes, as shown in Fig. 3(b)–(d). Using X-ray topography, we confirmed the presence of PDs that could be moved from the BPDs in the substrate at the periphery of the contact in the PiN diode without proton implantation. Therefore, the dark regions in the EL images correspond to the expanded 1SSFs from the BPDs in the substrate. The EL images of the other stressed PiN diodes are shown in Figs. S2–S5 and videos with and without expansion of the dark region (Time changes in EL images for the PiN diodes without proton implantation and with implantation of 1014 cm− 2) are also shown in the supplementary information.
We calculated the density of the expanded 1SSFs by counting the dark regions with a bright edge in the three PiN diodes for each condition, as shown in Fig. 5. The expanded 1SSF densities decreased with increasing proton doses, and even at a dose of 1012 cm− 2, the density of the expanded 1SSF was significantly lower than that of PiN diodes without proton implantation.
The reduction in carrier lifetime also influences the suppression of expansion, and proton implantation reduces carrier lifetime.32,36 We observed carrier lifetimes in a 60 µm-thick epitaxial layer with 1014 cm− 2 proton implantation. From the initial carrier lifetime, although implantation reduced the value to ~ 1 %, subsequent annealing recovered it to ~ 5 %, as shown in Fig. S6. Therefore, the reduced carrier lifetime owing to proton implantation was recovered by high temperature annealing and carrier-lifetime reduction should not be the major factor for the suppression of 1SSF expansion in proton-implanted PiN diodes.
Although no hydrogen was detected by SIMS after annealing at 1600°C, as reported in a previous study,43 we observed the effects of proton implantation on the suppression of 1SSF expansion, as shown in Figs. 3 and 4. Therefore, we consider that PDs were pinned by hydrogen atoms that had a density lower than the detection limit of SIMS (2 × 1016 cm− 3) or point defects introduced by implantation. It should be noted that we did not confirm an increase in the on-resistance due to the expanded 1SSF after the pulsed-current stress. This is possibly due to imperfect ohmic contacts fabricated using our process, which will be solved in the near future.
In summary, we developed a suppression method for the expansion of BPDs to 1SSFs in 4H-SiC PiN diodes using proton implantation before device fabrication. The deterioration of the I-V characteristics by proton implantation was not significant, particularly at a proton dose of 1012 cm− 2; however, the effect of the suppression of 1SSF expansion was significant. Although we fabricated PiN diodes in this study, proton implantation can also be applied to the fabrication of 4H-SiC devices. Additional device-fabrication costs during proton implantation should be considered, but they will be similar to the Al-ion implantation costs, which is an essential process for the fabrication of 4H-SiC power devices. Therefore, proton implantation prior to device processing is a potential method for fabricating bipolar degradation-free 4H-SiC power devices.