5.5 Future directions
6.1.3 Oxidation and characterization
99
Figure 6.25: Reflection TEM images of corrugated pillars after etching but before oxidation; the caps are the e-beam deposited alumina mask. The light and dark fringes are indications of the single crystal nature of the silicon.
contrast has been used to pick out portions of the polycrystaline alumina in the mask. The final frame shows a different diffraction condition which attempts to pick out the single crystal silicon dots within the oxidized sheath. One dot is seen near the top of the pillar and the dotted box is blown up in the inset. The RTEM method, while useful for non-destructive TEM measurements, often had difficulty maintaining a precise enough focus to pick out quantum dots, even with the diffraction aperture in place. Furthermore an exceptional amount of beam current was required to see the dots with simple density mapping. Figure 6.27 shows a sequence of photographs that show the result of imaging a quantum dot with R-TEM. A wide angle view of a set of pillars is shown in the top frame. A close up on square box on the left-most pillar shows the large, (too large to photoluminesce due to lack of size dependent band modification) football-shaped, quantum dot. The final frame shows the result of such a high beam current being focused down to such a small radius;
showing the oxidized pillar has melted and begun fall over. In this condition it was not possible to see the quantum dot anymore.
101
Figure 6.26: Three images of the same corrugated pillar. (a) Light-field density mapping. (b) Dark-field with diffraction contrast to highlight the mask. (c) Dark-field with diffraction contrast to highlight the small quantum dot in the topmost bead. The inset shows the diffraction from the silicon inside the amorphous silica.
Figure 6.27: The consequences of imaging oxidized pillars with high beam current in reflection mode. The pillar is imaged with a lower maginification in the top image. The middle image shows the quantum dot present in one of the corrugations. The bottom image shows the melted pillar due to electron imparted heating.
103
To obtain better images of the quantum dots within the silicon dioxide it was necessary to cleave the pillars off of the substrate and onto a copper TEM grid. Since this method was destructive to the sample it was only done after the testing of a given sample was completely finished. Figure 6.28 shows the improvement in image quality with pillars sitting on the copper grid. In this image it is possible to see the remnant, acorn shaped silicon in the head of the pillar and it is possible to make out the much smaller quantum dot in the first bead. Note that this sample is an oxidized example of the pillars in frame (b) of figure 6.23.
Figure 6.28: Image of the head of the pillars etched in frame (b) of figure 6.23. Note the acorn shape of silicon in the head of the pillar and the football shaped quantum dot in the first bead.
Even with this method it was difficult to make out the oxidized quantum dots. We atempted to make direct measurements of the size of the embedded quantum dots, however as explained later, this size estimation was to inaccurate and the actual size estimation was conducted with another method. In figure 6.29 frame (a) shows the image as taken from the TEM. It is barely possible to make out the three quantum dots in the three oxidized beads; depending on the medium in which this work is read the contrast may be too low to see them. Frame (b) consists of the same image, made translucent, with post-processed contrast/brightness tweaking, laid over the original image in frame (a). Dotted circles are added as guides for the eye. With this enhancement it is easier to see the general position of the quantum dots, although their size remains ambiguous. Figure 6.30 shows a similar TEM and processed image with dot locations circled with a white dashed line. Both of these figures highlight the difficulty in measuring the actual quantum dot size through 50-100 nm
of thermal oxide, using density mapping alone.
105
Figure 6.29: (a) TEM image of a corrugated pillar on a copper grid. The quantum dots are present but are only weakly visible. (b) Enhanced image of the quantum dots with their general locations circled.
Figure 6.30: (a) TEM image of a corrugated pillar on a copper grid. The quantum dots are present but are only weakly visible. (b) Enhanced image of the quantum dots with their general locations circled.
107
In order to increase the visibility of the oxide encapsulated dots a dark-field, diffraction contrast method was used. The diffraction aperture was used to preferentially pick out electrons diffracted from the single crystal silicon inside the pillar showing up as bright white sections, while the randomly scattered electrons from the amorphous silicon dioxide show up only as a darker gray pattern. An example of this is seen in figure 6.31. In both the images in that figure the bright white ball visible in the center of the oxide is the acorn shaped remnant of the head of the pillar from the sample seen etched in frame (a) of figure 6.23.
Figure 6.32 shows a pillar under two different diffraction conditions. In frame (a) only the electrons passing through the silicon chunk in the head are seen. The fact that the structures retain their single crystal behavior even after etching and oxidation is shown by the diffraction rings appearing on the silicon in the image. Frame (b) shows a second diffraction image where the electrons scattered through the first dot are collected by the detector. Uniform illumination from the first dot is an indication that the silicon remains a single crystal even after etching and oxidations;
illumination at a different diffraction condition than the silicon in the head implies that although the silicon remained crystalline perhaps the oxidation related strain deformed the lattice.
The images in figure 6.33 show the quantum dots within the pillar by utilizing the diffraction aperture. It is, however, not advisable to use such an image to estimate the size of these dots as their extent is roughly the precision of the focus. If 10 images are taken there could be a variance as large as 1 nm; with the dots being 2-4 nm in diameter this is far too large of a variance to use to size the dots. This technique does, however, provide a useful method to show (after optical testing) that quantum dots were in fact present in the pillars
Figure 6.31: (a), (b) TEM images of oxidized pillars from the sample seen in frame (a) of figure 6.23. The dot seen is actually the remnant of the silicon chunk in the head; with a final diameter of slightly less than 20 nm.
109
Figure 6.32: A pillar under two different diffraction conditions: (a) Only the electrons passing through the silicon chunk in the head are seen. Note the diffraction rings present in the silicon chunk at the top. (b) The same pillar under a different diffraction condition which highlights the first quantum dot.
Figure 6.33: TEM image of the same pillar under two different diffraction conditions: (a) Condition which only shows the bottom two dots. (b) Condition where all three dots are visible with better contrast to the amorphous silica.
111
Figure 6.34: Photoluminescence spectra of three samples of etched and oxidized quantum dots of with different initial corrugation diameters. The colorized frame of the SEM images correspond to the measured, plotted spectra. The pre-oxidation size is 30 nm, 37 nm, and 45 nm for the (a) black, (b) blue and (c) green samples, respectively. Note that the larger the pre-oxidation size of the corrugated pillars the longer the peak emission wavelength. Scale bars are 200 nm in each frame.