Open Access
Add to BibSonomyAbstract
An ultrasmooth reaction-sintered silicon carbide surface with an rms roughness of 0.424 nm is obtained after thermal oxidation for 30 min followed by ceria slurry polishing for 30 min. By SEM-EDX analysis, we investigated the thermal oxidation behavior of RS-SiC, in which the main components are Si and SiC. As the oxidation rate is higher in the area with defects, there are no scratches or cracks on the surface after oxidation. However, a bumpy structure is formed after oxidation because the oxidation rates of Si and SiC differ. Through a theoretical analysis of thermal oxidation using the Deal-Grove model and the removal of the oxide layer by ceria slurry polishing in accordance with the Preston equation, a model for obtaining an ultrasmooth surface is proposed and the optimal processing conditions are presented.
©2013 Optical Society of America
1. Introduction
Reaction-sintered silicon carbide (RS-SiC) is an ideal mirror material for space telescope systems [1, 2] as it has the advantages of a low thermal expansion coefficient, high radiation resistance, high specific stiffness, good size stability, and a low manufacturing cost [3, 4]. Therefore, the fabrication process of RS-SiC is a focus of research in the space optics field.
However, RS-SiC is a difficult-to-machine material because of its high hardness and chemical inertness [5, 6]. Furthermore, the composition and structure of RS-SiC are not uniform [7, 8]. The fabrication process of RS-SiC [9] generates a SiC domain and a Si domain as the major components in the substrate [10]. Figures 1(a) and 1(b) show the surface and cross-sectional morphology of the RS-SiC specimen used in our study, respectively.
Fig. 1 Morphology of RS-SiC. (a) Surface morphology observed by SEM. (b) Cross-sectional morphology observed by TEM.
Up to now, many techniques have been developed for processing RS-SiC [11], but few of them can yield an ultrasmooth surface. In the case of diamond lapping [12] and plasma chemical vaporization machining (PCVM) [13], scratches and bumpy structures are observed on the processed surface, as shown in Figs. 2(a) and 2(b), respectively, because the hardnesses and chemical reactivities of Si and SiC in RS-SiC differ. The hardnesses of Si and SiC are less than that of diamond; thus, in the case of the diamond lapping of RS-SiC, scratches are inevitably introduced on the surface. Similarly, in the case of the PCVM of RS-SiC, it is considered that the difference in chemical reactivity between the reactive species and SiC/Si causes the formation of the bumpy structure on the processed surface.
Fig. 2 AFM images of the processed RS-SiC surface. (a) Mechanical material removal by diamond lapping. (b) Chemical material removal by plasma chemical vaporization machining.
If we directly use a relatively soft abrasive compared to RS-SiC, such as ceria, to polish the RS-SiC, of course no scratches or bumpy structures will be introduced. However, there also will be almost no material removal, and the soft abrasive polishing process will make no sense, as the hardness of ceria is far smaller than that of RS-SiC. Therefore, an effective way of obtaining an ultrasmooth RS-SiC surface is a combination of surface oxidation and polishing of the oxide by soft abrasive compared with SiC. We applied the combination of thermal oxidation and polishing using ceria slurry, and an ultrasmooth RS-SiC surface with an rms roughness of 0.424 nm was achieved. To clarify the removal mechanism of RS-SiC in the combined process, we conducted scanning electron microscopy/energy-dispersive X-ray spectroscopy (SEM-EDX) observation of the RS-SiC surface before oxidation, after oxidation, and after polishing.
2. Experimental setup
The initial RS-SiC specimen was prepared by diamond lapping, and the surface morphology is shown in Fig. 2(a). The thermal oxidation of RS-SiC was conducted using a lamp annealer (MILA-5000: ULVAC-RIKO Inc.). The oxidation parameters are listed in Table 1.
Table 1. Thermal oxidation parameters
After oxidation, the oxidized surface was polished with ceria slurry, and the schematic diagram of the polishing system and corresponding polishing parameters are summarized in Fig. 3. The specimen and the polishing pad are both immersed into the slurry, and the polishing pad scanned the specimen surface. The polishing parameters in Fig. 4(b), Fig. 8(a) and 8(b) are listed in Table 2, and the corresponding polishing times are 30 min, 20 min, and 40 min, respectively.
Table 2. Ceria slurry polishing parameters
3. Results and discussion
Atomic force microscopy (AFM) images of the specimen as above after thermal oxidation and after ceria slurry appropriate polishing (polishing time: 30 min) are shown in Figs. 4(a) and 4(b), respectively. The rms roughnesses of the specimen before oxidation (after diamond lapping), after oxidation, and after ceria slurry polishing for 30 min are 1.250 nm, 16.417 nm, and 0.424 nm, respectively, and the Ra roughnesses are 0.856 nm, 12.426 nm, and 0.274 nm, respectively.
Fig. 4 RS-SiC surface measured by AFM. (a) After thermal oxidation. (b) After ceria slurry polishing for 30 min.
As the research aim is to fine finish not figure the surface, we set the AFM scanning areas to 5 µm × 5 µm and try to investigate the improvement of high spatial frequency roughness (HSFR) of the RS-SiC surface. For the improvement of low spatial frequency roughness (LSFR) and mid spatial frequency roughness (MSFR) of RS-SiC surface, it’s the aim of our other research in the figuring of RS-SiC substrate. The oxidation of RS-SiC increases its volume; thus, the surface roughness of the specimen increases after oxidation. Through the removal of the oxide layer by ceria slurry polishing, an ultrasmooth surface with an rms roughness of 0.424 nm was obtained.
To elucidate the thermal oxidation process of RS-SiC, we compared the morphology of surfaces before and after oxidation, as shown in Figs. 5(a) to 5(d). The Figs. 5(a) and 5(c) show the surfaces before and after thermal oxidation at position A, and Figs. 5(b) and 5(d) show those at position B. The results of elemental analysis by SEM-EDX indicate that the dark area in the original RS-SiC surface is Si and the other area is SiC, as shown in Fig. 1(a). A comparison of the RS-SiC surface before and after thermal oxidation at the same position is conducive to obtaining a better understanding of the reason why the polishing of thermally oxidized RS-SiC enables us to obtain an ultrasmooth and scratch-free surface.
Fig. 5 SEM images of RS-SiC surface before and after thermal oxidation. (a) Position A before oxidation. (b) Position B before oxidation. (c) Position A after oxidation. (d) Position B after oxidation.
It is interesting to note that there are few scratches on the RS-SiC surfaces that are thermally oxidized, as shown in Figs. 5(c) and 5(d), although the initial surfaces, which are prepared by diamond lapping, have many scratches, as shown in Figs. 5(a) and 5(b). We hypothesize that there are two reasons for the disappearance of scratches on the RS-SiC surface. First, the area with scratches has a larger surface area than the other areas, and the oxidation of RS-SiC increases its volume; thus, the scratches are closed by the expanded oxide. Second, the area surrounding that with scratches has a higher oxidation rate because of lattice strain; thus, the rapidly expanding oxide will fill the scratches. Through the closing and filling of scratches by oxidation, scratches are eliminated on the oxidized RS-SiC surface.
As shown in Figs. 5(c) and 5(d), the oxidized surface of each crystal grain of Si or SiC is very smooth. These results indicate that the oxidation rate in each crystal grain is uniform. Taking one crystal grain in RS-SiC into consideration, regardless of it being Si or SiC, the thermal oxidation behavior exhibits no difference from the thermal oxidation of a single Si and SiC crystal; thus, the oxide surface of each crystal grain in RS-SiC is flat. However, the oxidation rate is different among different Si and SiC crystal grains because the orientation of each crystal is different. Therefore, the thermal oxidation rate is not uniform over the whole RS-SiC surface.
By comparing Figs. 5(a) and 5(c), and 5(b) and 5(d), we find that the thermal oxidation rate of Si is higher than that of SiC [14–17] as the oxide layer of the Si grains is thicker than that of the SiC grains. Because the chemical bond of Si-C in the SiC crystal is much stronger than the Si-Si bond in the Si crystal, and the crystal structure of SiC is denser than that of Si, the oxidation rate of SiC is lower than that of Si under the same thermal oxidation conditions.
SEM-EDX analysis was conducted to investigate the composition distribution of the thermally oxidized RS-SiC surface. Figure 6(a) shows a SEM image of the oxidized surface, and Figs. 6(b), 6(c), and 6(d) show the elemental distributions of Si, O, and C, respectively. The density distribution of O reflects the different oxidation rates of the crystal grains.
Fig. 6 Surface morphology and element distributions of oxidized RS-SiC surface analyzed by SEM-EDX. (a) Surface morphology. (b) Si distribution. (c) O distribution. (d) C distribution.
As the oxidation rate of Si grain in RS-SiC is faster than that of SiC, the material compositions from the surface to the interior along the cross-sectional direction are pure oxide, oxide + SiC, oxide + Si + SiC, Si + SiC, successively. For obtaining an ultrasmooth surface, the polishing depth should be kept in the pure oxide layer. Although there are still oxide layer remained on the polished surface, the thickness of the remained oxide layer is only several nanometers, so it has almost no influence to the excellent mechanical property of the RS-SiC substrate. What’s more, the RS-SiC specimen in our experiment is expected to be used in the mirror for space telescope system and it will be uniformly deposited a thin film for the reflection of signal, so the remained oxide doesn’t affect the application of the RS-SiC mirror.
4. Theoretical analysis
We used the Deal-Grove model [17–20] to analyze the thermal oxidation of RS-SiC and the Preston equation [21–23] to calculate the polishing rate of the oxide layer. To better express the oxidation and polishing processes, we proposed a schematic diagram of the thermal oxidation of RS-SiC and ceria abrasive polishing of the oxide, as shown in Fig. 7.
Fig. 7 Schematic diagram of the thermal oxidation of RS-SiC and removal of the oxide layer by abrasive polishing.
Supposing that the thermal oxidation time is t1, on the basis of the Deal-Grove model, the oxidation depth d1 in the Si grain and the oxidation depth d2 in the SiC grain are calculated using Eq. (1) and Eq. (2), respectively. Here A1, B1, A2, and B2 are constants governed by the activation energy of SiC/Si, the orientation of the SiC/Si crystal, and the thermal oxidation temperature [17, 18].
From the analysis of the thermal oxidation of RS-SiC, it is confirmed that the oxidation rate of Si is much higher than that of SiC [14–17], and the oxidation rate among different SiC/Si crystal grains in RS-SiC is nonuniform. The Vickers hardnesses (GPa) of SiC, ceria, SiO2, and Si are 24-28 [24], 5-7.5 [25], 7.6 [26], and 7-9 [27], respectively. As the hardness of a ceria particle is much less than that of SiC and slightly smaller than those of silica and Si, to obtain an ultrasmooth surface, the expected thickness to be removed should be no more than the thinnest oxidation layer among the SiC grains with different crystal orientations. The thinnest oxidation layer of SiC grains, dmax, can be obtained by comparing the oxidation rates of SiC grains with different crystal orientations, as shown in Eq. (3). As the oxidation of SiC is an expansion process, a bumpy surface is formed after oxidation, and the largest peak-to-valley distance dpv is defined as the height difference between the thickest area of the oxidized Si grain and the thinnest area of the oxidized SiC grain, as shown in Eq. (4). The datum line of the definition of dpv and dmax is the baseline of the oxide, as shown in Fig. 7. To remove these bumps, the removal depth of polishing should be larger than dpv.
The removal of the oxide layer is conducted by mechanical polishing with ceria abrasive. For the polishing time t2, the polishing depth D can be expressed by Eq. (5) in accordance with the Preston equation. Here, k, p, and v are Preston’s coefficient, the downward pressure, and the relative velocity between the polishing pad and the sample, respectively.
From the above analysis, it is concluded that the polishing depth D should be controlled to be less than dmax and more than dpv to obtain an ultrasmooth surface. Insufficiently polished (polishing time: 20 min) and over-polished (polishing time: 40 min) RS-SiC surfaces observed by AFM are shown in Figs. 8(a) and 8(b), respectively. The importance of controlling the removal depth is apparent. For better control of the thermal oxidation time and corresponding ceria polishing time, in the future we will confirm the lowest oxidation rate among differently oriented SiC/Si crystals, with the aim of determining dpv and dmax. Simultaneously, through experiments on the ceria abrasive polishing of silicon oxide, we will verify the polishing rate, with the aim of keeping the polishing depth uniform in the oxide layer.
5. Conclusions
The combination of thermal oxidation and ceria slurry polishing enabled us to obtain an ultrasmooth scratch-free RS-SiC surface with an rms roughness of 0.424 nm, and the high spatial frequency roughness (HSFR) is markedly improved. Therefore, this combination is an effective way for the fine finishing of RS-SiC.
Through SEM-EDX analysis, we clarified the thermal oxidation behavior of RS-SiC and explained the reasons why the ultrasmooth surface was obtained. There are few scratches on the RS-SiC surfaces that are thermally oxidized; the oxide surface of each crystal grain in RS-SiC is flat, but the oxidation rate is different among different Si and SiC crystal grains; the thermal oxidation rate of Si is higher than that of SiC; Although there are still oxide layer remained on the polished surface, it doesn’t affect the application of the RS-SiC mirror.
Using the Deal-Grove model and the Preston equation, we determined the control conditions that yield a better surface, which will improve the finishing process of RS-SiC. The thinnest oxidation layer of SiC grains dmax can be obtained by comparing the oxidation rates of SiC grains with different crystal orientations; the largest peak-to-valley distance dpv can be achieved by comparing the height difference between the thickest area of the oxidized Si grain and the thinnest area of the oxidized SiC grain; the polishing depth D should be controlled to be less than dmax and more than dpv to obtain an ultrasmooth RS-SiC surface.
It is expected that these achievements will promote the application of RS-SiC mirrors in space telescope systems.
Acknowledgments
This work was supported by the professors and students of the Research Center for Ultra-precision Science and Technology of Osaka University. Furthermore, the authors express their gratitude to Prof. Y. Sano in Osaka University for providing the thermal oxidation equipment.
References and links
1. Y. Dai, “Surface finishing of new type RS-SiC ceramics,” Int. J. Comput. Appl. Tech. 29(2/3/4), 145–149 (2007). [CrossRef]
2. D. A. Ersoy, M. J. McNallan, and Y. Gogotsi, “Carbon coatings produced by high temperature chlorination of silicon carbide ceramics,” Mat. Res. Innovat. 5(2), 55–62 (2001). [CrossRef]
3. S. P. Lee, Y. S. Shin, D. S. Bae, B. H. Min, J. S. Park, and A. Kohyama, “Fabrication of liquid phase sintered SiC materials and their characterization,” Fusion Eng. Des. 81(8-14), 963–967 (2006). [CrossRef]
4. A. Sayano, C. Sutoh, S. Suyama, Y. Itoh, and S. Nakagawa, “Development of a reaction-sintered silicon carbide matrix composite,” J. Nucl. Mater. 271–272, 467–471 (1999). [CrossRef]
5. S. Suyama, T. Kameda, and Y. Itoh, “Development of high-strength reaction-sintered silicon carbide,” Diamond Related Materials 12(3–7), 1201–1204 (2003). [CrossRef]
6. Y. W. Kim, M. Mitomo, H. Emoto, and J. G. Lee, “Effect of initial α‐phase content on microstructure and mechanical properties of sintered silicon carbide,” J. Am. Ceram. Soc. 81(12), 3136–3140 (1998). [CrossRef]
7. O. P. Chakrabarti, P. K. Das, and J. Mukerji, “Growth of SiC particles in reaction sintered SiC,” Mater. Chem. Phys. 67(1–3), 199–202 (2001). [CrossRef]
8. T. Grande, H. Sommerset, E. Hagen, K. Wiik, and M. A. Einarsrud, “Effect of weight loss on liquid-phase-sintered silicon carbide,” J. Am. Ceram. Soc. 80(4), 1047–1052 (1997). [CrossRef]
9. S. S. Shinozaki, J. E. Noakes, and H. Sato, “Recrystallization and phase transformation in reaction-Sintered SiC,” J. Am. Ceram. Soc. 61(5–6), 237–242 (1978). [CrossRef]
10. J. Yan, Z. Zhang, and T. Kuriyagawa, “Mechanism for material removal in diamond turning of reaction-bonded silicon carbide,” Int. J. Mach. Tools Manuf. 49(5), 366–374 (2009). [CrossRef]
11. H. Y. Tam, H. B. Cheng, and Y. W. Wang, “Removal rate and surface roughness in the lapping and polishing of RB-SiC optical components,” J. Mater. Process. Technol. 192–193, 276–280 (2007). [CrossRef]
12. H. Deng and K. Yamamura, “Smoothing of reaction sintered silicon carbide using plasma assisted polishing,” Curr. Appl. Phys. 12(3), S24–S28 (2012). [CrossRef]
13. K. Yamamura, Y. Yamamoto, and H. Deng, “Preliminary study on chemical figuring and finishing of sintered sic substrate using atmospheric pressure plasma,” Procedia CIRP 3, 335–339 (2012). [CrossRef]
14. E. A. Irene, H. Z. Massoud, and E. Tierney, “Silicon oxidation studies: Silicon orientation effects on thermal oxidation,” J. Electrochem. Soc. 133(6), 1253–1256 (1986). [CrossRef]
15. H. C. Lu, T. Gustafsson, E. P. Gusev, and E. Garfunkel, “An isotopic labeling study of the growth of thin oxide films on Si (100),” Appl. Phys. Lett. 67(12), 1742–1744 (1995). [CrossRef]
16. J. Blanc, “A revised model for the oxidation of Si by oxygen,” Appl. Phys. Lett. 33(5), 424–426 (1978). [CrossRef]
17. Y. Song, S. Dhar, L. C. Feldman, G. Chung, and J. R. Williams, “Modified Deal Grove model for the thermal oxidation of silicon carbide,” J. Appl. Phys. 95(9), 4953–4957 (2004). [CrossRef]
18. B. E. Deal and A. S. Grove, “General relationship for the thermal oxidation of silicon,” J. Appl. Phys. 36(12), 3770–3778 (1965). [CrossRef]
19. A. Fargeix and G. Ghibaudo, “Dry oxidation of silicon: A new model of growth including relaxation of stress by viscous flow,” J. Appl. Phys. 54(12), 7153–7158 (1983). [CrossRef]
20. A. Fargeix, G. Ghibaudo, and G. Kamarinos, “A revised analysis of dry oxidation of silicon,” J. Appl. Phys. 54(5), 2878–2880 (1983). [CrossRef]
21. C. W. Liu, B. T. Dai, W. T. Tseng, and C. F. Yeh, “Modeling of the wear mechanism during chemical-mechanical polishing,” J. Electrochem. Soc. 143(2), 716–721 (1996). [CrossRef]
22. Q. Luo, S. Ramarajan, and S. V. Babu, “Modification of the Preston equation for the chemical–mechanical polishing of copper,” Thin Solid Films 335(1–2), 160–167 (1998). [CrossRef]
23. F. G. Shi and B. Zhao, “Modeling of chemical-mechanical polishing with soft pads,” Appl. Phys., A Mater. Sci. Process. 67(2), 249–252 (1998). [CrossRef]
24. J. Qian, G. Voronin, T. W. Zerda, D. He, and Y. Zhao, “High pressure, high temperature sintering of diamond-SiC composites by ball milled diamond-Si mixtures,” J. Mater. Res. 17(8), 2153–2160 (2002). [CrossRef]
25. A. B. Shorey, K. M. Kwong, K. M. Johnson, and S. D. Jacobs, “Nanoindentation hardness of particles used in magnetorheological finishing (MRF),” Appl. Opt. 39(28), 5194–5204 (2000). [CrossRef] [PubMed]
26. E. Pippel, J. Woltersdorf, H. O. Olafsson, and E. O. Sveimbjornsson, “Interfaces between 4H-SiC and SiO2: microstructure, nanochemistry and near-interface traps,” J. Appl. Phys. 97(3), 034302 (2005). [CrossRef]
27. . S. Williams, B. Haberl, and J. E. Bradby, “Nanoindentation of ion implanted and deposited amorphous silicon,” Mat. Res. Soc. 843, T6.3.1/R10.3.1- T6.3.5/R10.3.5 (2005).
