Suppression of partial dislocation glide motion during contraction of stacking faults in SiC epitaxial layers by hydrogen ion implantation

Silicon carbide (SiC), which has superior physical properties for power device applications such as a wide bandgap, high breakdown field, and high thermal conductivity, is a promising semiconductor material for high-power and high-temperature devices. ^1–3) Different from silicon (Si), which can be produced dislocation-free by the Cz crystal growth technique, ⁴⁾ SiC wafers contain many dislocations, which influence the performance, reliability, and yield of power devices. ^5–9) Although the dislocation density in SiC wafers continues to improve day by day, ^10–16) technologies for SiC power devices "with dislocations" are required. One of the critical issues that hinder the widespread use of SiC power devices is the bipolar degradation phenomenon. ^17–21) SiC bipolar devices such as pin diodes and insulated gate bipolar transistors face problems with defect-related degradation, in which the forward voltage is degraded due to the spontaneous expansion of single Shockley-type stacking faults (SFs) in the epitaxial layer during forward bias application. The SFs expand from basal plane dislocations (BPDs) to accommodate the glides of partial dislocations. ^19,22) The driving force of SF expansion was understood to be the electronic energy gain from carrier trapping at SFs. ^23–27) Localized energy states are formed at SFs trap carriers, and as a result, electronic energy gain (Δ) is generated. In this case, the SF energy (γ) is lowered by the energy gain compared to the original SF energy without carrier trapping (γ₀), which is expressed by the following equation:

$$\begin{eqnarray}&&\gamma ={\gamma }_{0}-{\rm{\Delta }}\end{eqnarray} \tag{ 1 }$$

When Δ exceeds γ₀, the SF energy with carrier trapping γ takes on a negative value, and spontaneous SF expansion takes place with the negative SF energy as the driving force. Therefore, many researchers have tried to control the carrier concentration during device operation to suppress the generation of the driving force for SF expansion. ^26,28–31) Another anomalous aspect of the spontaneous expansion of SFs is a drastic decrease in the critical resolved shear stress (CRSS) for the gliding of partial dislocations by recombination at the dislocations, which is called "recombination-enhanced dislocation glide motion." ^19,32,33) It is surprising that the partial dislocations glide even near room temperature with a very small driving force under a high current density, considering that CRSS for dislocation glide in SiC without forward bias application is several gigapascals. ^29,34) Thus, another way to suppress SF expansion is by controlling the mobility of the partial dislocations. To hinder the dislocation glide motion, pinning of dislocations by impurity elements is a common approach. ^35,36) Recently, we have reported that proton irradiation, which is commonly used in semiconductor processing, can suppress SF expansion in the SiC epitaxial layer driven by ultraviolet (UV) illumination and forward bias application. ^21,37) Although we carefully separated the effect of change in carrier lifetimes on the driving force of SF expansion and the effect of the pinning of dislocations by proton implantation, the relationship between dislocation mobility and proton implantation has remained unclear. In the present study, we investigated the contraction behavior of SFs subjected to proton implantation and revealed the relationship between dislocation mobility and proton implantation.

An n-type 150 mm 4H-SiC wafer with an epitaxial layer thickness of 12 μm and a donor doping concentration of 0.8 × 10¹⁶ cm⁻³ was used as the specimen. This epitaxial wafer contained much more BPDs than typical SiC epitaxial wafers and is described in Ref. 38. Firstly, the specific positions of the propagating BPDs were confirmed by X-ray topography (XRT) observations. Then, UV light was illuminated at the position of the BPDs at 374 K for 10 h to expand the SFs. A UV LED with a wavelength of 365 nm was focused to a diameter of 3 mm, and the illumination intensity was adjusted to 10 W cm⁻². After UV illumination, we observed the specimens by XRT to confirm the SF expansion. Next, specimens were implanted at room temperature with 0.6 MeV protons, which results in a maximum hydrogen density at about 5 μm beneath the surface. The detailed hydrogen density profile is shown in Ref. 37. After the proton implantation, the specimens were heated to 873 K in a nitrogen atmosphere for 10 min. High-temperature annealing has been reported to contract the SFs due to an increase in the mobility of partial dislocation glides. ^39,40) For the comparison, we conducted the same annealing process on the specimen without proton implantation after UV illumination. XRT observation was conducted in grazing incidence reflection geometry using a monochromatic X-ray beam (λ = 0.15 nm) with a g vector of −1–128 or 11–28 at BL8S2 in the Aichi Synchrotron Radiation Center. The details of the conditions are described in Ref. 41. In this geometry, SFs never exhibit a contrast, and the surrounding partial dislocations appear dark or bright depending on the direction of the Burgers vector. ⁴²⁾ For high-resolution XRT observation, we used our developed nuclear emulsion plates. ⁴³⁾

Figure 1 shows the XRT image taken before and after UV illumination with the same field of view. Large circular contrasts of threading screw dislocations, small point contrasts of threading edge dislocations, and linear contrasts of BPDs are observed in the XRT image taken before UV illumination [Fig. 1(a)]. After UV illumination for 10 h [Fig. 1(b)], a line contrast corresponding to a partial dislocation appeared and the surrounding area corresponding to the SF expanded from the position of a BPD before UV illumination, as reported in Refs. 37, 39, 40.

Zoom In Zoom Out Reset image size

Figure 2 shows XRT images taken before and after annealing to observe the contraction of SFs. Without proton implantation, SFs completely contracted to BPDs as a result of the annealing, as shown in Figs. 2(a) and 2(b). On the other hand, it is evident that the contraction of SFs in the epitaxial layer subjected to proton implantation was hindered. The contraction of SFs in the epitaxial layer subjected to proton implantation with a fluence of 1 × 10¹⁴ cm⁻² was incomplete, and the SF was still observed at the downstream part of the step flow as shown in Figs. 2(d) and 2(f). It seems that the more the proton fluence increases, the more difficult it becomes for the partial dislocations to move. The SFs in the epitaxial layer implanted with a fluence of 1 × 10¹⁶ cm⁻² were almost unchanged at the downstream part of the step flow, as shown in Fig. 2(h). The incomplete contraction at the downstream part of the step flow indicates that suppression of the movement of the partial dislocation took place at the shallower side of the epitaxial layers subjected to proton implantation, which is consistent with the maximum proton implantation depth of 5 μm. As the fluence of proton implantation increased, the contraction of SFs was evidently suppressed.

Zoom In Zoom Out Reset image size

Note that the driving force for the contraction of SFs during annealing is never influenced by proton implantation, as is the case for the expansion of SFs by UV illumination. ³⁷⁾ Reduction of carrier lifetime has also no influence on the contraction behavior of SFs. Therefore, the current results clearly indicate that proton implantation evidently suppressed the glide motion of partial dislocations. Proton implantation of SiC epitaxial layers introduces point defects such as proton impurities, vacancies, interstitials, and passivation of the cores of dislocations, which all lead to the suppression of dislocation glide motion. The interaction between dislocation and point defects is a possible mechanism for the suppression of dislocation glide motion by proton irradiation. So far, immobilization of partial dislocations in 4H-SiC by the interaction of point defects has been reported. ^36,44) Although the detailed mechanism is still unclear, including the types of the point defects that suppress the dislocation motion and how the suppression of the dislocation motion takes place, the present results suggest that the dislocation glide motion is independently controlled and bipolar degradation in SiC power devices with SF expansion to accommodate the movement of partial dislocations can possibly be suppressed by proton implantation, which is compatible with semiconductor processes.

In summary, we have revealed that proton implantation can suppress the partial dislocation glide motion through investigation of the contraction behavior of SFs in SiC epitaxial layers subjected to proton implantation. The present results support the conclusion that proton implantation suppresses the SF expansion in bipolar device operation without significant reduction in carrier lifetime in the epitaxial layer, as we have previously reported in Refs. 21 and 37. The present results imply that the problem of bipolar degradation in SiC power devices can be solved by proton implantation, which is compatible with semiconductor processing.

Acknowledgments