High Resolution Transmission Electron

0 downloads 0 Views 330KB Size Report
Further, the float polishing process did not introduce any mechanical damages in the CaF2 crystals. The TEM images showed that the float- polished surface had ...
Materials Transactions, Vol. 47, No. 2 (2006) pp. 267 to 270 #2006 The Japan Institute of Metals

High Resolution Transmission Electron Microscopy Study of Calcium Fluoride Single Crystal (111) Surfaces Processed by Ultraprecision Machining Naoyuki Ohnishi, Shinji Yoshida* and Yoshiharu Namba Department of Mechanical Engineering, Chubu University, Kasugai 487-8501, Japan The microstructure of a mechanically finished calcium fluoride (CaF2 ) was studied by transmission electron microscopy (TEM). The (111) oriented surfaces of high-purity CaF2 single crystals were finished to a flat surface by ultra-precision mechanical polishing. A float polishing method—with 7-nm-diameter silicon dioxide powder, pure water, and a tin lap—was employed for the finishing. Prior to the float polishing, ultraprecision grinding was performed as the preliminary treatment. A cross-sectional TEM study indicated that the thickness of the subsurface damage introduced by ultraprecision grinding was relatively small when compared with that by the conventional optical polishing process. Further, the float polishing process did not introduce any mechanical damages in the CaF2 crystals. The TEM images showed that the floatpolished surface had a faceted structure consisting of (111) terrace planes and nanometer-sized steps. The size of a single terrace depends on the mismatch angle between the sample surface and the (111) plane. The high-resolution TEM observation suggested that an atomically smooth (111) surface with a bulk fluorite structure was obtained over a relatively wide area on the large terrace planes. (Received October 25, 2005; Accepted December 22, 2005; Published February 15, 2006) Keywords: microstructure analysis, transmission electron microscopy, cross-sectional observation, surface structure, calcium fluoride, fluorite, single crystal, ultraprecision machining, mechanical polishing, microlithography optics

1.

Introduction

High-purity calcium fluoride (CaF2 ) single crystals are the only candidate materials for microlithography optics using excimer lasers in the deep ultraviolet (DUV) and vacuum ultraviolet (VUV) regions. In order to increase the laserinduced damage threshold, the polished surface must be extremely smooth and without any subsurface damage caused due to finishing. However, a CaF2 crystal is difficult to polish because it is soft and has a relatively high thermal expansion coefficient, although the crystal clearly shows a cleavage along the (111) plane, which is ordinarily used for optical applications. Float polishing is one of the ultraprecision polishing techniques used for preparing high-quality flat optical surfaces.1) It has been applied to various single crystals and optical glasses to achieve an ultrasmooth surface without a deformed subsurface layer. In previous reports,2,3) we demonstrated that the smoothest surface of (111) oriented CaF2 single crystals was obtained by float polishing. This was also confirmed at an atomic scale for the first time by a study using cross-sectional transmission electron microscopy (TEM).2,4) However, the above papers primarily reported the surface morphologies of float-polished specimens with the results of surface profiler measurements. In this paper, we discuss the detailed microstructure of floatpolished CaF2 crystal (111) surfaces as well as the surfaces processed before float polishing, focusing mainly on the cross-sectional TEM observation results. 2.

Experimental Procedure High-purity DUV-grade CaF2 single crystals were sawed

*Graduate

Student, Chubu University. Present address: DISCO Corporation, Tokyo 143-8580, Japan

into several sizes with diameters in the range of 10–100 mm. The large flat area of the samples was aligned parallel to the (111) plane and precisely ground by an ultraprecision surface grinder.5) The samples were then float polished on a tin lap with 7-nm-diameter SiO2 powder in a water solution on an ultraprecision float-polishing machine.1) The polishing fluid was a mixture of pure water and SiO2 powder with a mass ratio of H2 O:SiO2 = 97:3, and its temperature variation was maintained within 0.01 K. Polishing pressure was applied only by the weight of the sample and sample holder. After performing float polishing for 80 min, following the ultraprecision grinding, a flat surface of 31.9-nm p-v and 6.03nm rms, as measured with a laser interferometer on a 90-mmdiameter area, was obtained.2) The surface roughness values of a float-polished CaF2 sample were 0.120-nm Ra, 1.25nm Ry, and 0.151-nm rms, as measured with an optical profiler over a measurement area of 250  250 mm2 . For the TEM specimen preparation, the float-polished crystal samples were sectioned with an orientation normal to the polished surface, similar to that for an ordinary crosssectional surface observation. The samples were then glued to each other by an epoxy resin, and then mechanically sliced, ground, and polished using the conventional machining processes. They were finally ion polished at room temperature with a 3-kV argon ion beam with an orientation of 12 deg to the sample surface. The TEM observations were made using two electron microscopes with accelerating voltages of 200 and 400 kV (JEOL: JEM-2010 and JEM4000EX); its theoretical point resolutions were 0.23 and 0.17 nm, respectively. For the observation, the 400-kV TEM was used with a relatively high magnification. In order to minimize the electron beam irradiation damage, the observations were made under a low-dose condition, and the images were recorded using high-sensitivity photographic films.

268

N. Ohnishi, S. Yoshida and Y. Namba

(a)

(b)

Fig. 1

Cross-sectional TEM images of CaF2 crystal surfaces processed by ultraprecision grinding (a) and float polishing (b).

Fig. 2 Cross-sectional TEM image of a CaF2 crystal surface processed by conventional optical polishing. The magnification of the image is the same as in Figs. 1(a) and (b).

3.

Results and Discussion

Figures 1(a) and (b) are cross-sectional TEM images of CaF2 single crystal surfaces obtained after the ultraprecision grinding and float polishing, respectively. The images were obtained with a relatively low magnification. The incident electron beam was parallel to the machined or polished surface. After the ultraprecision grinding, the specimen had a high surface roughness of approximately 100 nm, as observed in Fig. 1(a). Further, this image clearly shows the subsurface damage introduced by the grinding process. The damaged layer observed in Fig. 1(a) has a thickness of approximately 200–300 nm. The image contrast indicates that the damaged area consists of two layers—a severely deformed upper layer with a fine strain contrast and an elastic strain field with a broad contrast beneath the upper layer. This is a typical example of the microstructure of brittle ionic crystals with subsurface damage introduced by the machining process. After the float-polishing process following the ultraprecision grinding, the surface roughness became less than a few nanometers, as shown in Fig. 1(b). Further, the image shows that the subsurface damage observed in Fig. 1(a) was completely removed. The image does not show any diffraction contrast in the subsurface region. This suggests that the float-polishing process does not introduce any mechanical damage, such as dislocations and deformation twins, in the CaF2 crystals. Figure 2 shows the TEM image of a CaF2 crystal surface processed using an optical polishing method, which employs a conventional polishing pad and slurry; this is used for the

finishing of most commercial optical surfaces. The surface roughness observed in this image seems to be comparable with that of the float-polished surface. However, the image shows a layer with severe subsurface damage, with a thickness of approximately 100–200 nm. This type of damage was realized when the surface was scratched by relatively large polishing particles. It should be noted that the average thickness of the severe subsurface damage—the fine strain contrast region in Fig. 2—is nearly equal to that in Fig. 1(a). The thickness of the damaged layer introduced by the ultraprecision surface grinder definitely depends on the crystal orientation as well as the grinding condition.6) Nevertheless, the above TEM result reveals that the subsurface damage to the CaF2 crystal introduced by the ultraprecision grinding is basically comparable with that introduced by conventional optical polishing. Although the float-polished crystal showed a damage-free subsurface structure, the surface as seen in the TEM images was not always smooth—it depended on the specimens—in comparison with those of the float-polished ordinary optical glass material surfaces.1) Figure 3(a) shows the cross-sectional TEM image of a float-polished specimen, indicating a large surface roughness. The image was obtained with a relatively high magnification with the incident electron beam parallel to the [110] direction. The surface profile of the TEM image indicates a characteristic ‘‘faceted’’ shape. This implies a relatively large surface roughness greater than several nanometers. Figure 3(b) is a higher-magnification TEM image of a part

High Resolution Transmission Electron Microscopy Study of Calcium Fluoride Single Crystal (111) Surfaces

269

(a)

20nm (b) 0.32nm

5nm Fig. 3 Cross-sectional TEM images of a float-polished CaF2 crystal surface, indicating a characteristic faceted shape roughness. The surface terraces parallel to the (111) plane are indicated by white lines in (b).

(a) plane view e

(111) trace

(111)

ac ) tr

1 (11

(111) (111)

(11

1)

tra

ce

(111) (111) (111) (b) profile view (TEM image) Fig. 4 Schematic illustration of the faceted surface shape. The profile view (b) corresponds to the image in Fig. 3. The plane view (a) is a proposed model based on the image contrast of Fig. 3.

of Fig. 3(a). It shows an image of the (111) lattice fringes with a spacing of 0.32 nm.7) As marked with white lines, the lattice image clearly indicates that the faceted surface is composed of flat (111) terrace planes separated by steps with nanometer heights. This suggests that the roughness of the surface mainly arises from a misalignment of the macroscopic sample surface inclined from the (111) crystal plane. In fact, the average surface orientation shown in Fig. 3(a) is mismatched by around 2 deg from the (111) plane. This results in a relatively large surface roughness. Figures 4(a) and (b) are the schematic drawings of a faceted surface shape derived from the TEM image contrast of Figs. 3(a) and (b). Figure 4(b) shows the profile view corresponding to the cross-sectional TEM images, while Fig. 4(a) shows the plane-view surface, which can be proposed on the basis of the cross-sectional image contrast. According to the symmetry and crystallographic nature of the CaF2 structure, it is reasonable to assume that a (111) terrace is surrounded by a certain number of steps parallel to the trace lines of the equivalent {111} planes. This type of multifaceted shape leads to a difference in the TEM specimen

thickness and results in a discontinuous contrast variation beneath each surface terrace in the cross-sectional-view images, as observed in Figs. 3(a) and (b). The proposed structure in Fig. 4(a) is basically consistent with the results of a low-magnification plane-view observation obtained by scanning probe microscopy.2) The results of the above TEM observations indicate that it is necessary to precisely align the polishing surface with the (111) plane to obtain relatively wide surface terraces with surface steps having large intervals. This may also improve the statistically defined roughness, such as rms roughness, over an entire macroscopic polished surface. Figures 5(a) and (b) are cross-sectional TEM images of a float-polished CaF2 surface obtained from a relatively wide surface terrace. The interval between the major steps of this area exceeded several hundred nanometers. The images show a smooth surface that is exactly parallel to the (111) plane. Figure 5(b) indicates that the cross-sectional surface profile is mostly composed of an array of ‘‘platforms’’ of single (111) lattice planes. This makes the surface roughness of this area nearly equal to the (111) lattice spacing (0.32 nm). These results are consistent with the recent observations from a high-resolution dynamic force microscopy (DFM) study on the float-polished CaF2 surface.8) Figures 5(a) and (b) also show that there is no mechanically induced subsurface damage observed since there was no modification in the contrasts of these high-resolution images. (A characteristic circular contrast, as shown in Fig. 5(a), is due to the damage by electron irradiation, which did not affect the surface structure within the observation time.) Figure 6(a) is a magnified TEM image of a part of the same specimen surface shown in Figs. 5(a) and (b). Figures 6(b) and (c) show the structural model and a calculated image corresponding to the observed one, respectively. The structural model was simply derived from the fluorite-type bulk crystal structure.7) Fluorine was chosen as the terminating element, according to an atomic-scale DFM study of the cleaved CaF2 (111) clean surface.9) The calculation was performed for an imaging condition with a specimen thickness t of 6.0 nm and a defocus value  f of 40 nm, using a commercial software package (MacTem-

270

N. Ohnishi, S. Yoshida and Y. Namba

(a) 10nm (b) 0.32nm 5nm Fig. 5

Cross-sectional TEM images of the float-polished CaF2 crystal, indicating a relatively large flat surface parallel to the (111) plane.

(a)

1nm (b)

(111)

Ca

F

(c)

Fig. 6 High-resolution TEM image compared with the corresponding structural model and the calculated image. The image (a) is a magnification of a section of Fig. 5(a). The atomic arrangement of the structural model (b) is also superimposed in a part of the calculated image (c).

pas). The position of atoms in the structural model is also superimposed on a part of Fig. 6(c). The calculated image contrast of both the surface and subsurface region show a reasonably close resemblance to the observed image. Although more precise analyses are necessary to reveal the detailed atomic structure on both the surface and subsurface region, it can be concluded that the observed surface basically has an unmodified fluorite-type bulk crystal structure. This proves that an atomically smooth (111) terrace plane with no subsurface mechanical damage can be obtained on an ultraprecision machined CaF2 single crystal surface. 4.

Conclusions

The microstructure of ultraprecision machined high-purity CaF2 single crystal (111) oriented surfaces was studied by cross-sectional TEM observations. The thickness of the subsurface damage introduced by the ultraprecision grinder was relatively small as compared with that by conventional optical polishing. The float-polished CaF2 surface formed a characteristic faceted shape with (111) terrace planes and nanometer-height steps. Subsurface mechanical damages were not observed in the float-polished samples. An atomically smooth (111) surface, basically with a fluorite-type bulk structure, was obtained over a relatively wide area on the large terrace region.

Acknowledgements This work was supported by the Grants-in-Aid for Scientific Research (B) No. 14350077 of the Japan Society for the Promotion of Science, the High-Tech Research Center Establishment Project (2004), and the Nanotechnology Support Project (2004) of the Ministry of Education, Culture, Sports, Science and Technology, Japan. REFERENCES 1) Y. Namba, H. Tsuwa and R. Wada: CIRP Ann. 36 (1987) 211–214. 2) Y. Namba, N. Ohnishi, S. Yoshida, K. Harada and K. Yoshida: CIRP Ann. 53 (2004) 459–462. 3) Y. Namba, K. Harada, N. Ohnishi and K. Yoshida: Proc. 17th ASPE Annual Meeting, (The American Society for Precision Engineering, 2002), pp. 450–453. 4) N. Ohnishi, Y. Namba, S. Yoshida, T. Yoshida, Y. Kanda and K. Yoshida: Proc. 19th ASPE Annual Meeting, (The American Society for Precision Engineering, 2004), pp. 510–513. 5) Y. Namba, R. Wada, K. Unno and A. Tsuboi: CIRP Ann. 38 (1989) 331– 334. 6) Y. Namba, T. Yoshida, S. Yoshida and K. Yoshida: CIRP Ann. 54 (2005) 503–506. 7) P. Villars ed.: Pearson’s Handbook of Crystallographic Data for Intermetallic Phases-Desk Edition (ASM International, Materials Park, 1997), vol. 1, pp. 1062. 8) S. Gritschneder, Y. Namba and M. Reichling: Nanotechnology 16 (2005) 883–887. 9) A. S. Foster, C. Barth, A. L. Shluger and M. Reichling: Phys. Rev. Lett. 86 (2001) 2373–2376.