Influence of hole shape/size on the growth of site-selective quantum dots
© Mayer et al.; licensee Springer. 2013
Received: 6 September 2013
Accepted: 15 November 2013
Published: 1 December 2013
Skip to main content
© Mayer et al.; licensee Springer. 2013
Received: 6 September 2013
Accepted: 15 November 2013
Published: 1 December 2013
The number of quantum dots which nucleate at a certain place has to be controllable for device integration. It was shown that the number of quantum dots per nucleation site depends on the size of the hole in the substrate, but other dimensions of the nucleation site are vague. We report on the influence of hole shape on site-selectively grown InAs quantum dots (QDs) by molecular beam epitaxy. Dry etching of the GaAs wafers was used because of its high anisotropic etching characteristic. Therefore, it was possible to verify the influence of several hole shape parameters on the subsequent QD growth independently. We show that the nucleation of these QDs depends on several properties of the hole, namely its surface area, aspect ratio of the surface area, and depth. Especially, the aspect ratio shows a big influence on the number of nucleating QDs per site. With knowledge of these dependencies, it is possible to influence the number of QDs per site and also its distribution.
Quantum dots (QDs) can be formed by growing InAs on GaAs by molecular beam epitaxy (MBE) in the Stranski-Krastanov growth mode [1–6]. The finite lattice mismatch between the two materials leads to the formation of nanometer-scaled InAs islands which, if covered with GaAs, act as QDs due to the lower bandgap of InAs [7, 8]. These QDs show unique properties which make them interesting for many applications like single photon sources [9–13].
For device fabrication, it is sometimes required to place QDs at certain locations. For example, in a microcavity, the QDs have to be placed exactly at the mode positions of photonic cavities in order to maximize coupling and therefore device performance . The positioning of QDs can be achieved by modification of the GaAs surface in the nanoscale. Electron beam lithography (EBL) [4–6, 14], local oxidation [15–17], focused ion beam , or nanomechanical stamping  can be used to fabricate small holes on the substrate surface. The deformation of surface chemical potential leads to accumulation of In adatoms in the holes which then act as preferential nucleation sites for InAs QDs. These pre-patterning techniques come with disadvantages due to surface degregation in terms of defects and impurities, which can limit the performance of the optical quality of the quantum dots. Nevertheless, it was shown that with an appropriate treatment, such as efficient sample cleaning , multistacking , or partial capping , good optical qualities can be achieved, e.g., small linewidths down to 100 μ eV for single-layer QDs  or even 43 μ eV for certain single QDs .
QD nucleation can be controlled by several methods. In prior works, we investigated the influence of hole spacing and post-growth annealing [23, 24]. It was also shown by other groups that growth parameters like flux , InAs deposition , and growth temperature [27, 28] can influence the nucleation. In this work, we focused on the effects of hole geometry and fabrication, such as hole size, shape, and depth, on the subsequent growth of site-selective QDs. Improving and adapting these parameters provide an additional control mechanism and might lead to further optimization. We used EBL combined with dry etching in our work as this is the most versatile patterning technique and therefore allows changing various pattern parameters easily. Dry etching showed superior controllability compared to the previously used wet chemical etching (WCE) [24, 29] as it is able to influence the hole shape and size much better due to a highly anisotropic etching [30, 31]. While hole size is known to influence the number of nucleating QDs  and post-growth techniques such as in situ annealing have been shown to modify these , knowledge on the influence of other hole parameters like aspect ratio or depth remained vague.
Using this etching recipe, the depth and shape of the holes can be influenced separately, and also, the shape of the hole in the resist is transferred almost 1:1 into the underlying GaAs substrate. After etching, the resist was removed by an adequate remover mainly consisted of acetone, followed by cleaning with different solvents (trichlorethylene, acetone, n-methyl-2-pyrrolidone) and dipping in a heated ultrasonic bath (isopropyl acolhol, methanol, ethanol), as also performed in prior studies . The cleaning procedure was finalized with a 35 min plasma asher treatment in oxygen atmosphere and a 10 s dip into diluted hydrochloric acid. A 12 nm thin GaAs buffer layer is deposited followed by a small annealing step for 20 s in order to reduce surface roughness created during etching. The beam equivalent pressures were ≈8×10-9 bar for As and ≈3.5×10-10 bar for Ga. The InAs QDs are grown for 24 s, which is equivalent to 1.5 ML. For all steps, the substrate temperature was held at 500°C. The influence of the hole properties, e.g., the hole shape, was then investigated by comparing the amount of QDs nucleated in the holes. Information on these properties were obtained from scanning electron microscopy (SEM) images using the image analysis tool ImageJ (NIH, Bethesda, MD, USA) . The depth of the holes was obtained from atomic force microscopy (AFM) scans.
In addition, it can be seen that the occupation increases more strongly for the 15 s etched sample. While the average number of QDs per hole seems to be lower for the 15 s sample compared to the 10 s sample for small holes, for holes larger than 120 nm, the occupation seems to be equal or even higher for the longer-etched sample. The reason for such behavior must be related to the increased depth of the holes because the increase in lateral size due to chemical etching does not lead to an expected higher occupation. Therefore, besides the lateral size, the shape of the hole influences the number of nucleating QDs.
The result of writing ellipses instead of round holes into the resist is shown in Figure 4e. The aspect ratio of the major elliptical axes is given with respect to the separation of the two exposure spots before buffer layer growth (black) and after buffer layer growth (red). As intended and shown in Figure 4, the aspect ratio increases (decreases) with increasing distance of the two exposure spots before the buffer layer growth (after the buffer layer growth).
Next, the influence of the aspect ratio on the QD nucleation was investigated. Two samples, dry etched for 10 and 15 s, are compared. With increasing distance between the two exposure spots, the final aspect ratio decreases, while the hole size increases. This effect can be seen for both samples. The differences in hole size between the two samples emerge as mentioned above. Longer-etched holes become larger due to a pullback of the resist near the holes by sputtering from the etching gases (compare Figure 1 where the resist is affected near the holes).
Furthermore, the aspect ratios of longer-etched holes are smaller. This might be explained by insufficient optimization of the etching gas parameters. The used etching gas is composed of two gases, SiCl4 and Ar. With these two selected etching gases, there is a chemical component (from the SiCl4) and a sputter component (mainly Ar). The resulting etching characteristic then depends on the gas mixture and selected powers. Chemical etching of GaAs in the direction is usually two to five times faster than in the perpendicular [0 1 1] direction, therefore increasing the effect of the separated holes.
At last, the influence of the etching depth is investigated. The etch rate depends strongly on the size of the etched structure, see Figure 3. At first, it increases very strongly with the hole area, which is due to the supply shortage of the etching gases through the small hole size. With increasing size of the etched structure, this effect becomes negligible and the etch rate converges to the etch rate of a free surface. The largest structures show about an eight times higher etching rate than the smallest investigated structures, which has to be taken into account if structures with different sizes are etched at the same time.
The 20 s etched sample has a maximum at one QD per hole of about 0.6. This means that 60% of all holes are occupied with one quantum dot. With decreasing etching depth, the maximum of the distribution is heading to a higher number of QDs per hole. Also, the distributions get broader for smaller etching depths, meaning that the average number of QDs per hole has a larger standard deviation. This behavior was seen for all investigated hole sizes and also hole spacings. This is remarkable because the size of the holes increases with increasing etching time, as seen before, which should increase the number of QDs for the longer-etched samples. The influence of depth can also be seen in Figure 6b where the number of QDs is given with respect to the volume of the holes. Since the depth and lateral size cannot be fully adjusted separately, the volume of the holes is given. It is calculated by the lateral size and depth of the holes. Despite the fact that the holes gain size, the influence of depth is dominant, and with increasing depth, fewer QDs nucleate within one nucleation site. At last, one AFM image of a 20 s etched sample is shown in Figure 6c. Two separated exposure spots with a distance of 20 nm were used in order to decrease the aspect ratio. The picture shown is an amplitude picture of this sample in order to also show the nucleated QDs inside the holes. As can be seen, there is still a small elongation of the holes with an aspect ratio of 1.16 ± 0.07 in the [0 1 1] direction and the holes are large with a diameter of 150.3±4.1 nm. Although the aspect ratio and diameter of the holes might be optimized further, the sample shows only a small number of QDs of one to two per hole. Decreasing of aspect ratio and diameter and increasing of hole depth might therefore lead to even smaller values of occupation.
The number of quantum dots which nucleate at a certain place has to be controllable for device integration. We investigated the influence of the size, aspect ratio, and depth of the nucleation site on quantum dot nucleation. The occupation increases with increasing aspect ratio, where the QDs align along a chain in the elongated direction. Increasing the distance of two separated exposure spots in the direction leads to a decrease of holes after the buffer layer growth. We showed that a smaller aspect ratio has an advantageous effect on the QD growth, which is not compensated by the worsening influence of the increased nucleation site. The depth of the nucleation site decreases the average number of QDs nucleating inside a hole and also seems to decrease its standard deviation. We showed that the nucleation of QDs can be influenced by the size, shape, and depth of the nucleation site. With in situ annealing, this should provide another possibility of influencing and optimizing the number of QDs within a nucleation site. The strong dependence of the etching rate on the structure size was also shown.
Atomic force microscopy
Electron beam lithography
Inductively coupled plasma
Molecular beam epitaxy
Methyl isobutyl ketone
Polymethyl methacrylate with metacrylic acid
Reactive ion etching
Scanning electron microscope.
We acknowledge the financial support from the Energy Research Centre Lower Saxony (EFZN), the Open Access Publishing Fund of Clausthal University of Technology, Deutsche Forschungsgemeinschaft (DFG), and the State of Baden-Württemberg through the DFG-Center for Functional Nanostructures (CFN) within subproject A2.6.