This paper describes a three-step process to remediate surface and sub-surface defects on chemo-mechanically polished SiC surfaces. In this process, a CF4-based inductively coupled plasma with reactive ion etch was used to remove material to a depth, which is unaffected by surface and subsurface polishing damage. This produced a planarized but carbon-rich fluorinated surface. This surface was then exposed to a 2 min rapid thermal oxidation in air at 1000 °C to oxidize and volatilize the excess carbon and fluorinated species, respectively. The resulting surface oxide was then stripped using a dilute hydrofluoric acid in water solution. This process, referred to as plasma assisted remediation, reproducibly yielded planarized, stoichiometric surfaces with low levels of carbon and oxygen contamination suitable for subsequent device fabrication. In the supporting studies described here, 4H- and 6H-SiC(0001) surfaces were remediated and characterized by x-ray photoelectron spectroscopy and atomic force microscopy at each stage of the process. Experimental studies under ion-rich and radical-dominant conditions are also reported which provide greater insight into the underlying chemistry and physics of the process.

The success of the silicon-based semiconductor industry is based upon several key factors. One of these is the ability to produce large area, defect free, single crystal silicon wafers qualified for device manufacture.1 In contrast, SiC, despite its many advantages as a wide bandgap semiconductor, has not developed as rapidly.2 This is in part because SiC was much more difficult to produce with the required bulk and surface quality. Nonetheless, steady progress over the past 50 years has led to high-quality single crystal wafers with diameters up to 200 mm. One of the keys to this progress has been the development of the chemo-mechanical polishing (CMP) process.3–5 

Although CMP, as it has evolved in recent years, provides a major improvement in the surface quality of SiC substrates, there is extensive literature dealing with its limitations. Sasaki et al.,6 for example, reported the presence of small scratches on otherwise optically flat wafers. These scratches were accompanied by subsurface defects (SSDs) localized along the scratches. Although they further reported that H2 etching was effective in removing the SSDs, it led to step bunching and pitting of the surface. These detrimental effects are well known and documented in the SiC literature.7,8 Sako et al.9 made similar observations, and in subsequent studies,10 they provided more detailed understanding of the underlying CMP induced defects. Specifically, these included high-density basal-plane dislocation loops, Shockley-type stacking faults, and chevron- or Y-shaped defects. While reported dimensions varied, a 125 nm wide scratch, for example, was observed to produce a defect field on the order of 800 nm wide and up to 100 nm deep.

The studies reported here focus on remediation of defects on the SiC surface and near surface layers. They were motivated by several factors. First, the CMP generated defects described above have been shown by Sameshima et al.11 and others12 to have a negative impact on metal–oxide–semiconductor (MOS) device reliability and performance. Moreover, CMP has also been shown to produce defects in homoepitaxial films grown on 4H- and 6H-SiC.9,13,14 Consequently, cost effective and scalable surface remediation methods are required to advance SiC device technology. Second, as small users buying wafers from a variety of vendors, our experience is that all vendors and/or CMP processes are not equivalent. Wafers purported to have CMP surfaces can range from high-quality, optically perfect surfaces exhibiting well-defined step and terrace structures to low-quality surfaces, which have highly scratched regions. Thus, methods are required to remediate these surfaces if they are to be used effectively.

The approach to surface remediation explored in this paper utilizes a three-step scalable process. Specifically, these are CF4-based inductively coupled plasma reactive ion etching (ICP-RIE) followed by rapid thermal oxidation (RTO) and then an oxide strip using dilute hydrofluoric acid in water solution. This process, referred to as plasma assisted remediation (PAR), is described in detail in Sec. II along with the overall experimental approach and methods used in these studies. In Sec. III, the results are presented and discussed. To augment these studies, we describe the results from a set of supporting experiments using a different, but well-controlled plasma-based treatment of the SiC surfaces. When combined with PAR results, the findings are used to provide insight into the surface chemistry and physics associated with the process, assess the effectiveness of the process, and suggest potential additional steps that can be taken to remediate the SiC surface at wafer scale.

The SiC samples used in this study were purchased as 50 mm diameter on axis (±0.5°), n-doped, CMP 4H-, and 6H-SiC (0001) wafers from a variety of vendors. These were diced into 1 × 1 cm2 samples for subsequent experiments using a Discotech Model DAD3240 wafer dicer. It is worth noting that most wafers were in the high-quality range. That is, there was no evidence of scratches when examined optically, and they exhibited step and terrace structure when examined by atomic force microscopy (AFM). Occasionally, however, a full wafer or a select area of some wafer exhibited a high density of scratches. These defective areas comprised the subset of samples used in this study.

The PAR process investigated here involved three steps. First, the SiC substrates were plasma etched using a CF4-based ICP-RIE process. Next, the etched substrates underwent RTO for 2 min at 1000 °C. Finally, the resulting oxide was stripped using a dilute HF/H2O solution.

Prior to ICP-RIE, each sample was degreased using a mixture of acetone and methanol. This was followed by blow drying in ultrahigh purity N2. The samples were then adhered to a silicon carrier wafer using boron nitride-based COOL-GREASE (CGR7016 from AI Technology). The sample and carrier wafer were then load-locked into a Trion Technologies Minilock-Phantom III ICP-RIE system. The system utilized an ICP for plasma generation and a biasable stage for ion flux and energy control at the surface to enhance the RIE process. The samples were etched in a CF4 plasma produced using 600 W of RF power (13.56 MHz) to drive the ICP, while 300 W of RF power (13.56 MHz) was applied to the stage. The chamber pressure was maintained at 25 m Torr with a 20 sccm CF4 flow. These conditions produced a nominal SiC etch rate previously reported as 200 nm/min.15 Samples were etched for 12 min, which corresponded to a depth of 2.4 μm. This etch time/depth was chosen so that the resulting surface was well below the depth of any CMP related surface scratches or SSDs observed on the original surface.

The RTO operation was performed by loading the substrate into an open ended 0.5-inch OD quartz tube. A type K thermocouple housed in a quartz thermocouple shield located adjacent to the substrate was used to monitor the substrate temperature. The quartz tube was rapidly inserted into a furnace operating at a steady temperature of 1000 °C, the nominal oxidation temperature. The “anneal clock” was started when the sample temperature reached 940 °C. After 2 min elapsed time, the sample reached a temperature of 1000 °C, and the tube was rapidly withdrawn from the furnace. It was then allowed to cool to ambient temperature in the open tube. The initial heating rate for all RTOs was approximately 5 °C/s.15 

After the RTO, the resulting oxide was stripped by a 10 min wet etch in a solution of 10% HF in H2O. Following this etch, the samples were rinsed in DI water for approximately one minute. The sample was then dipped in electronics grade isopropanol and then dried with house N2.

At each step of the process, all samples were characterized using tapping mode atomic force microscopy (AFM) to establish the surface morphology and x-ray photoelectron spectroscopy (XPS) to determine the stoichiometry and chemical state of each surface species. Tapping mode AFM was conducted with an Agilent (formerly Molecular Imaging) PicoPlus AFM/STM with a PicoScan 3000 controller. Commercial silicon probes with spring constant k = 8.4–57 N/m and a resonant frequency between 200 and 400 kHz in high amplitude mode were used. The XPS measurements were conducted with a Physical Electronics PHI 5700 VersaProbe with a monochromated Al-Kα (hυ = 1486.6 eV) photon source with an Omni focus V lens. XPS spectra were taken in “bulk analysis mode” with a nominal photoelectron takeoff angle of 45°.

The ICP-RIE process involves both directional, energetic ions, which give rise to physical etching, and radicals, which give rise to chemical etching. To delineate the roles of these species, several samples were etched using the large area plasma processing system (LAPPS) developed at the Naval Research Laboratory (NRL). The details of the LAPPS system have been discussed elsewhere16–19 and will not be repeated here. Importantly, when LAPPS is combined with an ICP source, the system provides the ability to operate under ion-rich and radical-dominant conditions17 and also to tightly control the energy of the ions19 at substrate surfaces. In these supporting experiments, SF6 rather than CF4 was used as the source gas. This resulted in SFx+ ions and F radicals as opposed to CFx+ ions and F radicals in the ICP-RIE process. Moreover, the ion kinetic energy was set for 10 eV as compared with several hundred eV in the ICP-RIE. Despite the differences and in some cases because of them, these studies provided insight into the different roles of ions and fluorine containing radicals in the etching process, the nature of the defects, and possible methods of remediating these defects and improving the PAR process.

Figures 1(a)1(d) show typical 5 × 5 μm2 AFM images of the SiC surface at each stage of the PAR process. Figure 1(e) shows the surface after a second cycle of PAR. The AFM image shown in Fig. 1(a) is typical of high scratch density regions observed on the poor-quality CMP surfaces of interest here. These observations are consistent with Sasaki et al.6 and Sako et al.10 who reported localized surface and sub-surface damage, which was not uniform across the wafer surface. The origin of these scratches in the CMP process has been described elsewhere and need not be repeated here.6,10,20–22 For the samples studied here, the observed scratches were nominally on the order of 10 nm deep and 5 nm wide. The overall RMS roughness of the surfaces was typically in the range of 1.5–2.0 nm. As demonstrated by multiple studies, these surface defects cause a cascade of sub-surface defects, which cannot be detected by AFM.6,9,10,23

FIG. 1.

Typical 5 × 5 μm2 AFM images and overall RMS surface roughness values for (a) a defective as-received 6H-SiC (0001) surface, (b) the plasma etched surface, (c) the surface after RTO, (d) the surface after HF etch, and (e) the surface after a second cycle of PAR.

FIG. 1.

Typical 5 × 5 μm2 AFM images and overall RMS surface roughness values for (a) a defective as-received 6H-SiC (0001) surface, (b) the plasma etched surface, (c) the surface after RTO, (d) the surface after HF etch, and (e) the surface after a second cycle of PAR.

Close modal

Figure 1(b) shows a representative AFM image of the surface after CF4-based ICP-RIE. As is typical, the AFM image reveals a nominal 58% reduction in the RMS surface roughness. It should be noted that the surface at this point is representative of material several micrometers below the depth of the original scratches, yet the darker areas in the image (corresponding to greater depths) reveal patterns strikingly similar to those seen on the as-received surfaces. Since the etched surface is several microns below the original surface, these features cannot be the original scratches. Rather, they are most likely the result of templating the original scratch pattern due to the directionality of the ICP-RIE process. Most importantly, since they are formed by etching and not polishing, they will not have the SSDs associated with the original scratches left by CMP.

Figure 1(c) shows the plasma etched surface after RTO. As described later, the RTO not only desorbs volatile halogenated surface species but also forms a thin SiO2 film on the surface. The RMS roughness is of the order of 21% greater than the plasma etched surface, while the morphology appears to be quite similar. This slight roughening is consistent with the formation of a conformal oxide layer on the etched surface.

Figure 1(d) shows the RTOed surface after the HF etch to remove the oxide layer. Here, it is seen that the templated scratch pattern is still observable; however, RMS roughness is on the order of 18% less than that of the plasma etched surface and 65% less than the as-received surface.

Figure 1(e) shows the surface after a second complete cycle of PAR. The templated scratch pattern is further reduced. This is reflected in a further 14% reduction in RMS roughness compared to the first PAR cycle and a 70% reduction compared to the initial surface. The 0.500 nm RMS roughness of this surface corresponds to a variation of less than two Si–C bilayers.24,25 This is significantly less than the step heights produced by step bunching on H2 etched surfaces.6–10 Consequently, after two cycles of PAR, the surface can be characterized as planar with small hillocks across the surface.

To gain a better understanding of the surface chemistry associated with the PAR process XPS was used to monitor each step. Figures 2(a)2(d) show XPS survey spectra corresponding to the AFM images shown in Figs. 1(a)1(d). As seen by the atomic concentrations shown in Fig. 2(a), the as-received surface is carbon rich and has a significant level of oxygen contamination. After ICP-RIE, Fig. 2(b) shows an even greater level of carbon enrichment, a significant reduction in oxygen contamination, and a high level of fluorine contamination. Further details of these surfaces will be discussed based on the high-resolution C1s and Si2s peaks.

FIG. 2.

XPS survey spectra corresponding to the AFM images shown in Figs. 1(a)1(d). The atomic concentrations shown here are based on integrated peak intensities and known sensitivity factors.

FIG. 2.

XPS survey spectra corresponding to the AFM images shown in Figs. 1(a)1(d). The atomic concentrations shown here are based on integrated peak intensities and known sensitivity factors.

Close modal

Figure 2(c) reveals that RTO leads to a significant reduction in carbon and a substantial increase in the oxygen concentration as expected due to the formation of a silicon oxide film. The stoichiometry of this oxide will be discussed later based on the high-resolution spectra of the Si2s peak; however, based on the attenuation of the silicon peaks, the thickness of this film can be estimated to be of the order of 3.5 nm.26,27 The fluorine peak is no longer observed.

Figure 2(d) shows that following the HF etch, the oxide film is largely removed with only a low level of oxygen contamination remaining. This mechanism of SiO2 etching by HF been previously reported28,29 and need not be more fully discussed here. As seen here, the C/Si ratio shows that the HF etched surface is essentially stoichiometric. In our experience the composition of this surface produced by PAR is comparable with the best well-polished CMP and HF etched CMP surfaces. Thus, in both composition and morphology, the PAR surface represents a significant improvement over the CMP-damaged surface. For completeness, it should be noted that the survey spectrum for the surface after the second PAR cycle is identical with that for the first cycle.

Further insight into the surface chemistry and stoichiometry of the processed surfaces can be obtained by examining the high-resolution C1s XPS spectra. Figure 3(a) shows the C1s spectrum for the as-received surface. This spectrum can be resolved into three peaks. The leading peak at 282.5 eV is assigned to the carbon associated with the SiC substrate.30 The peak at 284.8 eV has been extensively reported in the literature31,32 and is thought to be associated with sp2 carbon in the form of graphitic clusters on the surface. We believe the high density of scratches exposes numerous Si- and C-atoms with dangling bonds, and this facilitates the oxidation of the surface and the formation of graphitic clusters. The peak at 286.6 eV is associated with oxygen-based defects. Specifically, it may be due to hydroxide (C-OH) or epoxide (C–O–C) defects formed on the graphite clusters33,34 or silicon oxycarbide species formed on the SiC surface.33,35

FIG. 3.

High-resolution C1s XPS spectra corresponding to the AFM images shown in Figs. 1(a)1(d).

FIG. 3.

High-resolution C1s XPS spectra corresponding to the AFM images shown in Figs. 1(a)1(d).

Close modal

The C1s spectrum for the plasma etched surface is shown in Fig. 3(b). It will be recalled from the survey spectrum in Fig. 2(b) that the C/Si ratio for this surface was 1.8. As before, the peak at 282.5 eV is associated with the carbon in the SiC substrate. The peak at 283.7 eV, as referenced later, has a binding energy lower that any fluorinated carbon species, and it might be supposed that it is due to carbon being implanted below or deposited at the surface when several hundred eV CFx+ ions impact the surface and dissociate. However, prior studies36 show that both CF4- and Cl2-based ICP-RIE produce a comparable peak. Since the Cl2-based plasma eliminates the possibility of carbon deposition, this suggests that the observed C-enrichment is due primarily to selective halogen-based etching of Si atoms rather than carbon deposition. This agrees with reports showing that CF4 plasmas not only etch Si-based materials37,38 but also preferentially etch Si from SiC and SiCOH.39–41 More detailed studies of the C1s peak at 283.7 eV15,42 show that it is associated with graphene tethered to Si atoms in the topmost layer of the SiC substrate with the precise binding energy being dependent on the extent of tethering. Specifically, reduced tethering corresponds in an increase in binding energy. Although implantation does not appear to contribute significantly to the observed carbon enrichment, it is reasonable to believe that energy deposition by the ions disorders the surface and near surface layers increasing the number of defects and dangling bonds. This, in turn, should increase the effectiveness of the radical etch as well as the extent of tethering.

The C1s peak at 286.4 eV, as before, may be associated with the hydroxide, epoxide, or carboxide defect species. However, for the etched surface, oxygen is present only at low levels (3.0%), while fluorine is present at high levels (21.2%). Thus, it is more likely that these and other higher binding energy carbon species represent fluorinated defects. Specifically, based on the work of several groups investigating fluorinated carbon surfaces,43–45 the peak at 286.4 eV is likely due to graphitic carbon bound to a CF2, which is denoted as C-CF2 defect. Here, the bold-underline indicates the atom from which the photoelectron originates. The peaks at 288.1 and 290.6 eV may be associated with the carbon in CF-CF3, and CF2, respectively. Given the stoichiometry of these species, this accounts for more than twice the 21.2% fluorine observed in the survey spectrum for the plasma etched surface. This to some extent may be accounted for by the way these species are distributed in the near surface layers of the substrate. The critical point is that the plasma etched SiC surface consists of a tethered graphene film covered with a high concentration of fluorinated carbon defects.

Figure 3(c) shows the SiC surface after RTO. It is evident that all the fluorinated carbon species have been desorbed or oxidized and volatilized from the surface. Aside from the substrate C1s peak at 282.5 eV, there is only a trace of graphitic carbon present at 284.6 eV (1.4%). Oxygen-based carbon defects (e.g., epoxides, hydroxides, carbonyls, etc.) are either absent or present at levels below the detection limit which is ∼0.2%,46 and as discussed below, essentially all the oxygen is associated with silicon and the oxide film. These observations are consistent with the understanding that oxygen readily etches carbon-based species by conversion to, and subsequently desorption of CO2.47–49 

After the oxide is stripped with an HF acid etch, the C1s spectrum shown in Fig. 3(d) reveals that only a low level of surface contaminant remains as indicated by the peak at 283.9 eV. This peak may be associated with tethered graphene clusters as discussed above. Most likely, however, it is silicon oxycarbide (SiOxCy) since its position is in excellent agreement with Socha et al.33 and Onneby and Pantano.35 

Surface remediation methods like sacrificial thermal oxidation (STO) employ longer oxidation times than the RTO step, and they have been reported to form not only SiOxCy but also graphite clusters. These result from carbon that is trapped under the thicker oxide film.50–52 Due to the rapid nature of the oxidation step in the PAR process, this problem appears to have been avoided. Moreover, the level of carbon contamination is much lower than that of the as-received surface.

To complete the picture of the surface chemistry and stoichiometry of the processed surfaces, Figs. 4(a)4(d) show the high-resolution Si2s XPS peaks corresponding to the AFM images shown in Figs. 1(a)1(d). Figure 4(a) represents the as-received surface. The main peak at 151.8 eV is due to silicon in the substrate.53 Comparing the C1s and Si2s [Figs. 3(b) and 4(b)] substrate components, it may be seen that the C/Si ratio for the as-received surface is of the order of 0.8. This slightly silicon rich condition is generally the case. This may be because silicon atoms are the outermost species for the (0001) crystal plane. The extent to which the as-received surface appears to be carbon rich in the survey spectrum [Fig. 2(a)] is due to graphitic and other carbon defects on the surface. The low intensity peak at 153.6 eV is due to silicon suboxide or oxycarbide.54 

FIG. 4.

High-resolution Si2s XPS spectra corresponding to the AFM images shown in Figs. 1(a)1(d).

FIG. 4.

High-resolution Si2s XPS spectra corresponding to the AFM images shown in Figs. 1(a)1(d).

Close modal

Figure 4(b) shows the Si2s peak for the plasma etched surface. Again, the peak at 151.7 eV is due to silicon bound as SiC in the substrate. By comparing this Si2s component with the corresponding C1s substrate peak [Fig. 3(b)], it may be seen that the C/Si ratio for the substrate species is again of the order of 0.8. As noted previously, the low intensity peak at 153.6 eV is due to silicon suboxide or oxycarbide.54 The fact that no fluorinated silicon species are present is consistent with the idea that that these species are preferentially volatilized compared to the fluorinated carbon species.

Figure 4(c) shows the Si2s peak after RTO. As expected, the spectrum is dominated by a large oxide peak at 154.8 eV. Based on the work of Gross et al.,55 this peak can be identified as SiO2; however, based on the 45.3% oxygen concentration shown in the survey spectrum [Fig. 2(c)], the O/Si ratio is closer to 1.6/1 than 2/1. Thus, at least on average, the oxide layer is slightly non-stoichiometric. Interestingly, the peak associated with silicon in the substrate has been shifted to higher binding energy by about 0.3 eV. The reason for this is not clear. For this surface, the C/Si ratio for the substrate species is of the order of 1.0.

Figure 4(d) shows the Si2s peak after using HF to strip the oxide layer. Again, the peaks present represent the substrate and suboxide or carboxide species. Comparing the substrate components in Figs. 3(d) and 4(d), it can be seen that the C/Si ratio for the substrate species is approximately 1.0.

Overall, these results indicate that the SiC surfaces prepared by PAR are essentially stoichiometric with very low levels of suboxide or oxycarbide contamination. Moreover, the results indicate that these surfaces are essentially planar and scratch free. The results presented here deal exclusively with 6H-SiC(0001). Not surprisingly however, comparable results were seen for defective 4H-SiC.

It is important to note that ICP-RIE process used in PAR utilizes physical etching due to ions and chemical etching due to radicals. Physical etching component associated with the ions gives ICP-RIE its directional and high etch rate characteristics. It is the directionality of the ionic etch species, for example, that gives rise to the templating previously discussed. Chemical etching due to radicals is much slower42 and isotropic.

The ion component is needed in the PAR process to rapidly etch below the depth of surface scratches and the SSDs associated with them. While ultimately leaving a planarized, stoichiometric SiC surface with very low levels of oxygen and carbon contaminants, there is undoubtedly some ion-induced damage at the substrate surface. These defects most likely include but are not limited to lattice vacancies as well as impurity and self-interstitials. To some extent, this damage may be remediated by the RTO process. However, to gain at least cursory insight into the nature of this damage as well as the basic process mechanism and a possible means of remediating the damage, several well-defined experiments were conducted using the LAPPS operating in an ion-rich and radical-dominant mode.

As noted earlier, LAPPS can be operated under well-controlled conditions, particularly when combined with a remote ICP source, which is used to partially dissociate the Ar/SF6 gas mixture prior to entering the reactor. Figure 5 shows a plot of ion and radical species concentrations as a function of RF power (13.56 MHz) applied to the ICP source. Over the range of RF power shown here, the ion concentration SFx+ remains relatively constant at a nominal value of ≈8 × 1011 cm−3, while the F radical concentration increases from 2.5 × 1012 to 8 × 1013 cm−3. It is important to note the ion-to-radical density ratio is much higher in LAPPS (ICP = 0 W) than would be expected in a conventional discharge. When the ICP is powered at 300 W, the ion-to-radical density ratio becomes similar to a conventional discharge. In the present work, samples were exposed for 5 min under the limiting cases identified as A (ion-rich) and B (radical-dominant) in Fig. 5. The ion energy in both cases was ≈10 eV.

FIG. 5.

Ion and radical species concentrations in an Ar/SF6 gas mixture as a function of ICP power in the hybrid-LAPPS. (a) and (b) represent the operational conditions used in the present work. Reproduced with permission from D. R. Boris and S. G. Walton, J. Vac. Sci. Technol. A 36, 060601 (2018). Copyright 2018 AVS: Sci. Technol. Mater., Interfaces Process.

FIG. 5.

Ion and radical species concentrations in an Ar/SF6 gas mixture as a function of ICP power in the hybrid-LAPPS. (a) and (b) represent the operational conditions used in the present work. Reproduced with permission from D. R. Boris and S. G. Walton, J. Vac. Sci. Technol. A 36, 060601 (2018). Copyright 2018 AVS: Sci. Technol. Mater., Interfaces Process.

Close modal

Figures 6(a) and 6(b) show the high-resolution spectra for SiC samples after being treated under ion-rich and radical-dominant conditions, respectively. Before discussing these spectra in detail, it is important to emphasize that SF6 is used as the reactive gas for the LAPPS exposures, rather than the fluorocarbon background gas used during the ICP-RIE process discussed earlier. Thus, carbon cannot be deposited from the plasma; however, sulfur and sulfur containing species deposited by neutral or charged SFx species are possible. Given that, it is significant that neither sulfur nor sulfur containing species were observed for surfaces treated under ion-rich and radical-dominant conditions.

FIG. 6.

High-resolution C1s spectra for samples etched under (a) ion-rich and (b) radical-dominant etch conditions.

FIG. 6.

High-resolution C1s spectra for samples etched under (a) ion-rich and (b) radical-dominant etch conditions.

Close modal

In Fig. 6(a), the peak at 282.5 eV is associated with the SiC substrate. The peak at 284.0 eV is in the range associated with tethered graphene, although it is not as intense as in the case of the CF4-based ICP-RIE exposures. The slightly higher binding energy for this peak (0.2 eV) suggests that the tethering is less extensive than that observed for the ICP-RIE process. As discussed earlier, prior work15,42 suggests that tethering is in part associated with ion-induced disorder or damage of the surface and in the near surface layers. In this case, reduced tethering (i.e., an up-shifted binding energy) is expected for the LAPPS ion-rich exposure, which utilized 10 eV ions compared with an ICP-RIE exposure, which utilized ions with energies of several hundred eV.

Based on the work of Dementjev et al.,43 the remaining peaks in Fig. 6(a) can be assigned to halogen related species. The peak at 285.5 eV can be assigned to C-CF, while the peaks at 287.0 and 288.6 eV can be assigned to C-CF2 and CF-CF3, respectively. As noted previously, the bold-underline indicates the carbon atom from which the photoelectron originates, which in these cases is a carbon atom on the SiC surface or in the very near surface layers. The peaks at 290.6 and 292.7 eV are in reasonably good agreement with CF-CF2 and CF-CF3, respectively. It is important to note that the origin of the carbon for these species is the SiC matrix. Thus, species C-CF, C-CF2 and CF-CF3, CF-CF2, and CF-CF3, whether on the surface or in the very near surface layers, represent substrate carbon atoms with a lower energy threshold for liberation.

The peaks at 285.5 and 287.0 eV in Fig. 6(a) may also be, respectively, associated with C-OH and C=O species.15 However, the survey spectrum for this sample is essentially identical to that for the samples produced using the ICP-RIE process shown in Fig. 2(b), where the oxygen level is low (≈5%) while the fluorine level is high (≈25%). Consequently, the contribution from the oxygen-based defects is expected to be quite small in this case. The results of Fig. 6(a) are consistent with the highly decorated surfaces created when using LAPPS to functionalize graphene.18 

As noted above, the species identified as -CF, -CF2, and -CF3 are incipient volatile species formed by interactions of charged and neutral the SFx and F species with carbon atoms of the SiC. Since the SFx+ ions have a kinetic energy of 10 eV, which is just above the Si–C bond energy (6 eV),56 it is possible that these species are formed both on the surface and in the very near surface layers. The latter might be the result of either ion penetration or lattice disorder produced by ion energy accommodation. Regardless, the delivery of additional fluorine or an elevation of ion flux and/or energy could ultimately lead to the desorption of these CxFy species.

When the ICP is added to produce a radical-dominant plasma, the chemical state of the surface and near surface layers is distinctly different. The C1s spectrum in Fig. 6(b) shows only peaks associated with the substrate, tethered graphene, and C-CF. The higher binding energy fluorinated defect species formed by the ion-rich treatment are not observed. This is an interesting result that points to the interplay between ions and radicals in the formation of higher binding energy species. As noted above, these higher binding energy fluorinated species could be volatilized by an increase in fluorine or an increase in the ion flux and/or energy. The latter can be eliminated since the ion density (Fig. 5) and energy (surface bias) did not vary between process conditions. Accordingly, we assume the increase in fluorine was sufficient to drive the surface evolution.

While the systems are significantly different, it is fair to consider the importance of reactive radicals or neutral species in minimizing subsurface defect formation associated with a flux of energetic ions, particularly when the ion energy is high. That is, it is reasonable to speculate that a radical-dominant etch at low incident ion energy (≈10 eV in this work) could more completely remove the ICP-RIE ion-induced damage. This, in turn, would provide a more defect free surface for STO and, ultimately, result in more defect free surface after the HF etch.

The motivation for this work was the remediation of highly scratched regions of commercial CMP SIC wafers. The results indicate that PAR removes the material associated with the scratches and underlying SSDs, and it produces an essentially planar, stoichiometric surface with very low levels of suboxide or oxycarbide contamination. The results also suggest that the use of a radical-dominant etch at low ion energies after the ICP-RIE etch may provide an even more defect free surface.

Given these results, it is interesting to speculate on the potential of PAR to replace or complement the CMP process. This, of course, is a subject for future investigation. From a manufacturing point of view, however, PAR would be attractive since it uses instrumentation and processes already widely employed in the semiconductor industry. Moreover, the PAR process is readily scalable to SiC wafer diameter. A further area of future research would be the comparison of homoepitaxial films and devices fabricated on PAR remediated surfaces with those prepared on CMP surfaces. A closely related area of exploration deals with the extent to which radical-dominant etch at low ion energies can be employed to remediate ion-induced damage.

A three-step plasma-enhanced surface oxidation (PAR) process has been used to remediate high scratch density areas and associated SSDs on defective SiC(0001) surfaces. Both AFM and XPS were used to characterize the effects of each step on surface morphology and composition. The first step, CF4-based ICP-RIE, removes SiC to a depth of several microns. This removes the SiC layers containing the surface scratches and SSDs. The second step utilizes RTO to desorb fluorinated carbon surface species and oxidize sp2-carbon species formed during the plasma etch. Due to the relatively low temperature and brief oxidation time, only a thin oxide is formed, which avoids the problem of graphite cluster formation frequently observed in conventional STO processes. The third step, an HF/H2O etch, removes the oxide layer. The resulting surface is essentially planarized with an RMS roughness on the order of several Si–C bilayers. Moreover, the surface is essentially stoichiometric with only low levels of silicon suboxide or oxycarbide present. Mechanistic studies using ion-rich and radical-dominant etch conditions at low ion energies in a non-carbon containing background gas suggest the ion-rich conditions lead to a carbon-rich film with a variety of fluorinated defects on the surface and in the near surface layers. The radical-dominated process limits the production of defects, suggesting fluorine radicals can serve to remediate ion-driven defects at the surface.

M. A. Mathews was supported in part by the Instrumentation Seed Program for Innovative Research (InSPIRe) grant provided by WVU Shared Research Facilities (SRF). D.R.B. and S.G.W. are supported by the Naval Research Laboratory base program. The authors acknowledge the use of WVU Shared Research Facilities, and we thank Mr. Harley Hart, WVU-SRF Operations Manager, and Dr. Qiang Wang, WVU-SRF Manager, Materials Characterization Facilities, for expert guidance and many helpful discussions throughout the course of this work.

The authors have no conflicts to disclose.

M. A. Mathews Jr.: Conceptualization (equal); Investigation (equal); Writing – original draft (equal). A. R. Graves: Conceptualization (equal); Investigation (equal); Writing – review & editing (equal). D. R. Boris: Conceptualization (equal); Investigation (equal); Writing – review & editing (equal). S. G. Walton: Conceptualization (equal); Data curation (equal); Investigation (equal); Methodology (equal); Supervision (equal); Writing – review & editing (equal). C. D. Stinespring: Conceptualization (equal); Data curation (equal); Methodology (equal); Supervision (equal); Writing – review & editing (equal).

The data that support the findings of this study are available within the article.

1.
G.
Fisher
, “
Silicon crystal growth and wafer technologies
,”
Proc. IEEE
100
,
1454
1474
(
2012
).
2.
P. G.
Neudeck
,
Silicon Carbide Technology
(
The VLSI Handbook
,
2006
), p.
20061800
.
3.
Y.
Wang
and
L. C.
Zhang
, “
A review on the CMP of SiC and sapphire wafers
,”
Adv. Mat. Res.
126–128
,
429
434
(
2010
).
4.
L.
Zhou
,
V.
Audurier
,
P.
Pirouz
, and
J. A.
Powell
, “
Chemomechanical polishing of silicon carbide
,”
J. Electrochem. Soc.
144
,
L161
(
1997
).
5.
Y.
Zhou
,
G.
Pan
,
X.
Shi
,
L.
Xu
,
C.
Zou
,
H.
Gong
, and
G.
Luo
, “
XPS, UV–vis spectroscopy and AFM studies on removal mechanisms of Si-face SiC wafer chemical mechanical polishing (CMP)
,”
Appl. Surf. Sci.
316
,
643
648
(
2014
).
6.
M.
Sasaki
,
H.
Matsuhata
,
K.
Tamura
,
H.
Sako
,
K.
Kojima
,
H.
Yamaguchi
, and
M.
Kitabatake
, “
Synchotron X-ray topography analysis of local damage occurring during polishing of 4H-SiC wafers
,”
Jpn. J. Appl. Phys.
54
,
91301 (7)
(
2015
).
7.
M.
Kumagawa
,
H.
Kuwabara
, and
S.
Yamada
, “
Hydrogen etching of silicon carbide
,”
Jpn. J. Appl. Phys.
8
,
421
428
(
1969
).
8.
S.
Dogan
,
D.
Johnstone
,
F.
Yun
,
S.
Sabuktagin
,
J.
Leach
,
A. A.
Baski
,
H.
Morkoc
,
G.
Li
, and
B.
Ganguly
, “
The effect of hydrogen etching on 6H-SiC studied by temperature-dependent current-voltage and atomic force microscopy
,”
Appl. Phys. Lett.
85
,
1547
1549
(
2004
).
9.
H.
Sako
,
T.
Yamashita
,
K.
Tamura
,
M.
Sasaki
,
M.
Nagaya
,
T.
Kido
,
K.
Kawata
,
T.
Kato
,
K.
Kojima
,
S.
Tsukimoto
,
H.
Matsuhata
, and
M.
Kitabatake
, “
Microstructural analysis of damaged layer introduced during chemo-mechanical polishing
,”
Mater. Sci. Forum
778–780
,
370
373
(
2014
).
10.
H.
Sako
,
H.
Matsuhata
,
M.
Sasaki
,
M.
Nagaya
,
T.
Kido
,
K.
Kawata
,
T.
Kato
,
J.
Senzaki
,
M.
Kitabatake
, and
H.
Okumura
, “
Micro-structural analysis of local damage introduced in subsurface regions of 4H-SiC wafers during chemo-mechanical polishing
,”
J. Appl. Phys.
119
,
135702
(
2016
).
11.
J.
Sameshima
,
O.
Ishiyama
,
A.
Shimozato
,
K.
Tamura
,
H.
Oshima
,
T.
Yamashita
,
T.
Tanaka
,
N.
Sugiyama
,
H.
Sako
,
J.
Senzaki
,
H.
Matsuhata
, and
M.
Kitabatake
, “
Relation between defects on 4H-SiC epitaxial surface and gate oxide reliability
,”
Mater. Sci. Forum
745
,
740
742
(
2013
).
12.
M.
Kitabatake
,
J.
Sameshima
,
O.
Ishiyama
,
K.
Tamura
,
H.
Oshima
,
N.
Sugiyama
,
T.
Yamashita
,
T.
Tanaka
, and
H.
Matsuhata
, “
The integrated evaluation platform for SiC wafers and epitaxial films
,”
Mater. Sci. Forum
740–742
,
451
(
2013
).
13.
M.
Hasegawa
,
K.
Ohira
,
N.
Kaneoka
,
T.
Ogata
,
K.
Onuki
,
K.
Kobayashi
,
T.
Osanai
,
K.
Masumoto
, and
J.
Senzaki
, “
4H-SiC epi-ready substrate qualification by using mirror electron microscope inspection system
,”
Mater. Sci. Forum
1004
,
369
375
(
2020
).
14.
M.
Kitabatake
,
J.
Sameshima
,
O.
Ishiyama
,
K.
Tamura
,
H.
Oshima
,
N.
Sugiyama
,
T.
Yamashita
,
T.
Tanaka
,
J.
Senzaki
, and
H.
Matsuhata
, “
The integrated evaluation platform for SiC wafers and epitaxial films
,”
Mater. Sci. Forum
740–742
,
451
454
(
2013
).
15.
A. R.
Graves
,
S.
Chaudhari
, and
C. D.
Stinespring
, “
Two-step synthesis of few layer graphene using plasma etching and atmospheric pressure rapid thermal annealing
,”
Diam. Relat. Mater.
101
,
107568
(
2019
).
16.
S. G.
Walton
,
D. R.
Boris
,
S. G.
Rosenberg
,
H.
Miyazoe
,
E. A.
Joseph
, and
S. U.
Engelmann
, “
Etching with electron beam-generated plasmas: Selectivity versus ion energy in silicon-based films
,”
J. Vac. Sci. Technol. A
39
,
033002
(
2021
).
17.
D. R.
Boris
and
S. G.
Walton
, “
Precise control of ion and radical production using electron beam generated plasmas
,”
J. Vac. Sci. Technol. A
36
,
060601
(
2018
).
18.
S. G.
Walton
,
S. C.
Hernandez
,
D. R.
Boris
,
T. B.
Petrova
, and
G. M.
Petrov
, “
Electron beam generated plasmas for the processing of graphene
,”
J. Phys. D: Appl. Phys.
50
,
354001
(
2017
).
19.
S. G.
Walton
,
D. R.
Boris
,
S. C.
Hernandez
,
E. H.
Lock
,
T. B.
Petrova
,
G. M.
Petrov
, and
R. F.
Fernsler
, “
Electron beam generated plasmas for ultra low Te processing
,”
ECS J. Solid State Sci. Technol.
4
,
N5033
(
2015
).
20.
P.
Pirouz
,
J. L.
Demenet
, and
M. H.
Hong
, “
On transition temperatures in the plasticity and fracture of semiconductors
,”
Philos. Mag. A
81
,
1207
1227
(
2001
).
21.
P.
Pirouz
,
M.
Zhang
,
J. L.
Demenet
, and
H. M.
Hobgood
, “
Yield and fracture properties of the wide band-gap semiconductor 4H-SiC
,”
J. Appl. Phys.
93
,
3279
3290
(
2003
).
22.
K.
Maeda
,
K.
Suzuki
,
S.
Fujita
,
M.
Ichihara
, and
S.
Hyodo
, “
Defects in plastically deformed 6H-SiC single crystals studied by transmission electron microscopy
,”
Philos. Mag. A
57
,
573
592
(
1988
).
23.
J. R.
Grim
,
M.
Benamara
,
M.
Skowronski
,
W. J.
Everson
, and
V. D.
Heydemann
, “
Transmission electron microscopy analysis of mechanical polishing-related damage in silicon carbide wafers
,”
Semicond. Sci. Technol.
21
,
1709
1713
(
2006
).
24.
P.
Lauffer
,
K. V.
Emtsev
,
R.
Graupner
,
T.
Seyller
, and
L.
Ley
, “
Atomic and electronic structure of few-layer graphene on SiC(0001) studied with scanning tunneling microscopy and spectroscopy
,”
Phys. Rev. B
77
,
155426
(
2008
).
25.
I.
Razado-Colambo
,
J.
Avila
,
D.
Vignaud
,
S.
Godey
,
X.
Wallart
,
D. P.
Woodruff
, and
M. C.
Asensio
, “
Structural determination of bilayer graphene on SiC(0001) using synchrotron radiation photoelectron diffraction
,”
Sci. Rep.
8
,
10190
(
2018
).
26.
Z. H.
Lu
,
J. P.
McCaffrey
,
B.
Brar
,
G. D.
Wilk
,
R. M.
Wallace
,
L. C.
Feldman
, and
S. P.
Tay
, “
SiO2 film thickness metrology by x-ray photoelectron spectroscopy
,”
Appl. Phys. Lett.
71
,
2764
2766
(
1997
).
27.
D. S.
Jensen
,
S. S.
Kanyal
,
N.
Madaan
,
M. A.
Vail
,
A. E.
Dadson
,
M. H.
Engelhard
, and
M. R.
Linford
, “
Silicon (100)/SiO2 by XPS
,”
Surf. Sci. Spectra
20
,
36
42
(
2013
).
28.
S.
Verhaverbeke
,
I.
Teerlinck
,
C.
Vinckier
,
G.
Stevens
,
R.
Cartuyvels
, and
M. M.
Heyns
, “
The etching mechanism of SiO2 in hydrofluoric acid
,”
J. Electrochem. Soc.
141
,
2852
(
1994
).
29.
T.
Hoshino
and
Y.
Nishioka
, “
Etching process of SiO2 by HF molecules
,”
J. Chem. Phys.
111
,
2109
2114
(
1999
).
30.
C. D.
Stinespring
and
J. C.
Wormhoudt
, “
Surface studies relevant to silicon carbide chemical vapor deposition
,”
J. Appl. Phys.
65
,
1733
1742
(
1989
).
31.
J. F.
Moulder
,
W. F.
Stickle
,
P. E.
Sobol
, and
K. D.
Bomben
,
Handbook of x-ray photoelectron spectroscopy
(
Perkin-Elmer Corporation—Physical Electronics Division
,
1992
).
32.
S.
Contarini
,
S. P.
Howlett
,
C.
Rizzo
, and
B. A.
De Angelis
, “
XPS study on the dispersion of carbon additives in silicon carbide powders
,”
Appl. Surf. Sci.
51
,
177
183
(
1991
).
33.
R. P.
Socha
,
K.
Laajalehto
, and
P.
Nowak
, “
Oxidation of the silicon carbide surface in watts’ plating bath
,”
Surf. Interface Anal.
34
,
413
417
(
2002
).
34.
S. W.
King
,
R. J.
Nemanich
, and
R. F.
Davis
, “
Dry ex situ cleaning processes for (0001)Si 6H-SiC surfaces
,”
J. Electrochem. Soc.
146
,
2648
2651
(
1999
).
35.
C.
Onneby
and
C. G.
Pantano
, “
Silicon oxycarbide formation on SiC surfaces and at the SiC/SiO2 interface
,”
J. Vac. Sci. Technol. A
15
,
1597
1602
(
1997
).
36.
S.
Raghavan
,
T. J.
Denig
,
T. C.
Nelson
, and
C. D.
Stinespring
, “
Novel surface chemical synthesis route for large area graphene-on-insulator films
,”
J. Vac. Sci. Technol. B
30
,
030605
(
2012
).
37.
S. A.
Rosli
,
A. A.
Aziz
, and
H. A.
Hamid
, “
Highly chemical reactive ion etching of silicon in CF4 containing plasmas
,” in
2006 IEEE International Conference on Semiconductor Electronics
(
IEEE
,
2006
), pp.
856
860
.
38.
L. M.
Loewenstein
, “
Selective etching of silicon nitride using remote plasmas of CF4 and SF6
,”
J. Vac. Sci. Technol. A
7
,
686
690
(
1989
).
39.
A. N.
Andreev
,
M. M.
Anikin
,
A. L.
Syrkin
, and
V. E.
Chelnokov
, “
Effect of various processing methods on the state of a 6H-SiC(000-1) surface
,”
Semiconductors
28
,
377
380
(
1994
).
40.
J.
Comeaux
,
W.
Wirth
,
J.
Courville
,
N.-W.
Baek
,
D.
Jung
, and
S.
Jang
, “
Etching characteristics of low-k SiCOH thin films under fluorocarbon plasmas
,”
Vacuum
202
,
111165
(
2022
).
41.
J. W.
Palmour
,
R. F.
Davis
,
T. M.
Wallett
, and
K. B.
Bhasin
, “
Dry etching of β-SiC in CF4 and CF4 + O2 mixtures
,”
J. Vac. Sci. Technol. A
4
,
590
593
(
1986
).
42.
M. A.
Mathews
, Jr.
,
A. R.
Graves
, and
C. D.
Stinespring
, “
Characterization of graphene films formed using radical plasma species
,”
Diam. Relat. Mater.
139
,
110416
(
2023
).
43.
A.
Dementjev
,
A. V.
Eletskii
,
K. I.
Maslakov
,
E. G.
Rakov
,
V. F.
Sukhoverhov
, and
A. V.
Naumkin
, “
Fluorination of carbon nanostructures and their comparative investigation by XPS and XAES spectroscopy
,”
Fuller. Nanotub. Carbon Nanostruct.
14
,
287
296
(
2006
).
44.
Y.
Ma
,
H.
Yang
,
J.
Guo
,
C.
Sathe
,
A.
Agui
, and
J.
Nordgren
, “
Structural and electronic properties of low dielectric constant fluorinated amorphous carbon films
,”
Appl. Phys. Lett.
72
,
3353
3355
(
1998
).
45.
H. J.
Ko
,
K. M.
Lee
,
H. J.
Lee
, and
C. K.
Choi
, “
Structural and electrical studies on fluorinated amorphous carbon films as intermetal dielectric layer in ULSI devices
,”
Thin Solid Films
506–507
,
82
86
(
2006
).
46.
A. G.
Shard
, “
Detection limits in XPS for more than 6000 binary systems using Al and Mg Kα X-rays
,”
Surf. Interface Anal.
46
,
175
185
(
2014
).
47.
Z.
Shpilman
,
I.
Gouzman
,
E.
Grossman
,
L.
Shen
,
T. K.
Minton
,
J. T.
Paci
,
G. C.
Schatz
,
R.
Akhvlediani
, and
A.
Hoffman
, “
Oxidation and etching of CVD diamond by thermal and hyperthermal atomic oxygen
,”
J. Phys. Chem. C
114
,
18996
19003
(
2010
).
48.
M.
Bagge-Hansen
,
R. A.
Outlaw
,
M. Y.
Zhu
,
H. J.
Chen
, and
D. M.
Manos
, “
Hyperthermal atomic hydrogen and oxygen etching of vertically oriented graphene sheets
,”
J. Vac. Sci. Technol. B
27
,
2413
2419
(
2009
).
49.
U. A.
Shroder
,
E.
Granas
,
T.
Gerber
,
M. A.
Arman
,
A. J.
Marinez-Galera
,
K.
Schulte
,
J. N.
Andersen
,
J.
Knudsen
, and
T.
Michely
, “
Etching of graphene on Ir(111) with molecular oxygen
,”
Carbon
96
,
320
331
(
2016
).
50.
W.
Kern
, “
The evolution of silicon wafer cleaning technology
,”
J. Electrochem. Soc.
137
,
1887
(
1990
).
51.
G.
Maeda
,
I.
Takahashi
,
H.
Kondo
,
J.
Ryuta
, and
T.
Shingyouji
, “
Adsorption and desorption of contaminant metals on Si wafer surface in SC1 solution
,”
Mater. Res. Soc. Symp. Proc.
386
,
195
200
(
1995
).
52.
A.
Koh
,
A.
Kestle
,
C.
Wright
,
S. P.
Wilks
,
P. A.
Mawby
, and
W. R.
Bowen
, “
Comparative surface studies on wet and dry sacrificial thermal oxidation on silicon carbide
,”
Appl. Surf. Sci.
174
,
210
216
(
2001
).
53.
L.
Muehlhoff
,
W. J.
Choyke
,
M. J.
Bozack
, and
J. T.
Yates
, Jr
., “
Comparative electron spectroscopic studies of surface segregation on SiC(0001) and SiC(0001-)
,”
J. Appl. Phys.
60
,
2842
2853
(
1986
).
54.
T.
Bekkay
,
E.
Sacher
, and
A.
Yelon
, “
Surface reaction during the argon ion sputter cleaning of surface oxidized crystalline silicon (111)
,”
Surf. Sci.
217
,
L377
L381
(
1989
).
55.
T.
Gross
,
M.
Ramm
,
H.
Sonntag
,
W.
Unger
,
H. M.
Weijers
, and
E. H.
Adem
, “
An XPS analysis of different SiO2 modifications employing a C1s as well as an Au 4f7/2 static charge reference
,”
Surf. Interface Anal.
18
,
59
64
(
1992
).
56.
G.
Ecke
,
R.
Kosiba
,
V.
Kharlamov
,
Y.
Trushin
, and
J.
Pezoldt
, “
The estimation of sputtering yields for SiC and Si
,”
Nucl. Instrum. Methods Phys. Res., Sect. B
196
,
39
50
(
2002
).