Ge is an attracting material for high-speed devices because of its high-carrier- mobility    . A Ge film on Si substrate has many advantages compared with bulk Ge wafer when we think about future industrial application. However, it is not easy to grow low dislocation-density Ge films onto Si because of their 4% lattice mismatch  . In order to reduce dislocation density of Ge films, post-growth annealing might be the most common method. However, it may cause interdiffusion between the Ge films and the Si substrates because the Ge-Si is an isomorphous system . In order to overcome this issue, we applied a new rapid heating technique proposed by our group and succeeded in forming Ge thin films on Si substrates   .
From the other viewpoint, it is important to study lattice strains or lattice- space distributions of semiconductors because they are strongly affected electric properties. Recently other researchers discovered that the moiré fringes between semiconductor lattices and scanning lines produced by STEM could be used to analyze distributions of lattice strains -.
In this study we applied this quite new technique to our specimen (Ge/Si) and it was evaluated that we were succeeded in forming Ge films with no strains all over the specimen.
2. Experimental Procedure
2.1. MBE Growth and Plasma Heating
We prepared the specimens using same procedure which has been published . Therefore only important part of the Ge film growth and plasma heating will be shown in this paper. Ge films with 300 nm-thick were epitaxially grown onto the Si (100) substrate by MBE. The substrate temperature was 300 C. Then 150 nm-thick SiO2 was deposited onto the Ge/Si at 300 C by plasma CVD as a barrier against W diffusion into the Ge. After that, 100 nm-thick W was deposited onto the SiO2/Ge/Si at room temperature by RF-sputtering as a heat source of the plasma heating. Then we set the specimen in the microwave-hydrogen- plasma heating chamber and heated the specimen up to 700 C. It took less than 2 seconds that the temperature reached to 700 C from the room temperature. Just after the specimen-temperature reached to 700 C, we shut down the plasm power and it took about 10 seconds that the specimen-temperature decreased to room temperature.
2.2. STEM Moiré Observation
We deposited protection layers onto the annealed specimen to avoid damages caused by focused ion beam (FIB). First, we deposited an amorphous carbon onto the W/SiO2/Ge/Si by simple vacuum evaporation. Second, Pt-Pd alloy was deposited onto the C/W/SiO2/Ge/Si by magnetron sputtering. Then the specimen was set into the FIB and W was deposited onto the Pt-Pd/C/W/SiO2/Ge/Si by ion-assisted chemical deposition in the FIB chamber. After preparing these protection layers, the specimen was fabricated by FIB (Hitachi FB-2100A). Then we checked the microstructure of the specimen using conventional TEM and STEM methods. After that, we set up the STEM condition carefully and took high-resolution STEM images and STEM moiré with the acceleration voltage of 200 kV. We mainly used field-emission type STEM (FEI, Tecnai Osiris). The STEM is not a Cs corrected microscope so it is almost impossible to take atomic-resolution STEM images but it has enough potential to take crystal-lattice- resolution STEM images. Therefore this STEM is suitable for our purpose to take STEM moiré fringes.
Figure 1 is a schematic illustration of the STEM moiré pattern. STEM moiré is a kind of fringe between the crystal-lattice spacing and the electron-beam scanning lines. In this study, we focused on the Ge and Si (111) planes and we set the scanning-line period to 632 pm.
Figure 1. An schematic illustration of the STEM moiré pattern, interaction between the STEM scanning lines and the crystal lattices.
3. Results and Discussion
Figure 2 and Figure 3 are results of overall STEM observations. The bright field STEM image (Figure 2) shows that the dislocation density of this Ge film is low even though there are 4% lattice mismatch between the Si substrate. The Si there and the high-angle-annular-dark-filed (HAADF) STEM image (Figure 3) suggests that the Ge film is compositionally uniform.
The bright field STEM image (Figure 2) shows that the dislocation density of this Ge film is low even though there are 4% lattice mismatch between the Si substrate. The Si there and the high-angle-annular-dark-filed (HAADF) STEM image (Figure 3) suggests that the Ge film is compositionally uniform. Figure 4 is a high-resolution HAADF-STEM image of the Ge part which is located just on the Si substrate. This high-resolution image suggests that the Ge lattice is not strained even on the Si substrate. However, high-resolution TEM and STEM images can not cover micrometer order wide area in general. This is one of the weak point of TEM and STEM.
In order to overcome this weak point and to analyze the lattice spacing distribution, we carried out a quite new STEM technique proposed by other researchers -. Figure 5 is a schematic illustration how to take a STEM moiré pattern. In this work, we tried to take two different STEM moiré patterns simultaneously: 1) moiré pattern between Ge (111) planes and STEM scanning lines and 2) moiré pattern between Si (111) planes and STEM scanning lines. Therefore, we set the STEM scan direction almost parallel to but not exactly parallel to (111) planes. Figures 6-8 are STEM moiré patterns taken from the annealed specimen. Figure 6 is the original data which we did not do any contrast/ brightness modification. In the case of Figure 7 and Figure 8, the image contrast was adjusted for Si area and Ge area, respectively. These data include two important information. First, the directions and spacing or the STEM moiré patterns in Ge area and Si area are drastically changed across the Ge/Si interface. This is because of the 4% lattice mismatch between Ge and Si. Second, the STEM moiré pattern in Ge area is not bending at all. This means that the Ge (111) lattice planes keep same direction and same spacing at least inside the observed are. The observed are was submicron squared, so it is assumed that the Ge film is not strained all over the specimen.
Figure 2. Low magnification STEM bright field image. There exist some dislocations but the dislocation density is not so high.
Figure 3. Low magnification STEM HAADF image. It is clear that the Ge film is compositionally uniform.
Figure 4. A high-resolution HAADF-STEM image of the Ge part.
Figure 5. A schematic illustration how to take a STEM moiré pattern. The STEM scan direction is almost parallel to (111) planes but it is not exactly parallel to them.
Figure 6. A STEM moiré patterns pattern taken from the annealed specimen. This micrograph is the original one which we did not do any contrast/brightness modification. In Figure 7 and Figure 8, modified micrographs are shown.
Figure 7. A STEM moiré patterns pattern taken from the annealed specimen. Contrast and brightness were adjusted for Si are. The original micrograph is shown in Figure 6. The other modified micrograph is shown in Figure 8.
Figure 8. A STEM moiré patterns pattern taken from the annealed specimen. Contrast and brightness were adjusted for Ge are. The original micrograph is shown in Figure 6. The other modified micrograph is shown in Figure 7.
In this study, we produced a low-dislocation-density Ge thin film onto Si substrate by MBE followed by microwave-hydrogen plasma heating. We precisely analyze the lattice-space distribution by using STEM moiré pattern and it is proved that the Ge film is not strained. This means that we are succeeded in forming a completely relaxed uniform Ge thin film on the Si, and the film has good crystallinity. It was also shown that the STEM moiré pattern is very useful to investigate the lattice-space distributions.
The authors thank Dr. Yukihito Kondo and Dr. Noriaki Endo of JEOL Ltd. for useful suggestions. We also thank Dr. Kazuo Ishizuka, Dr. Akimitsu Ishizuka, and Dr. Hirofumi Matsuhata of HREM Research Inc. for their kind comments.