Quantitative characterization of the interface roughness of (GaIn)As quantum wells by high resolution STEM

Micron 79 (2015) 1–7

Contents lists available at ScienceDirect

Micron journal homepage: www.elsevier.com/locate/micron

Quantitative characterization of the interface roughness of (GaIn)As quantum wells by high resolution STEM H. Han ∗ , A. Beyer, K. Jandieri, K.I. Gries, L. Duschek, W. Stolz, K. Volz Philipps-Universität Marburg, Faculty of Physics and Materials Science Center, Hans Meerwein Str., 35032 Marburg, Germany

a r t i c l e

i n f o

Article history: Received 12 June 2015 Received in revised form 15 July 2015 Accepted 16 July 2015 Available online 26 July 2015 Keywords: HAADF STEM Quantitative evaluation Composition map Interface roughness III/V semiconductors

a b s t r a c t The physical properties of semiconductor quantum wells (QW), like (GaIn)As/GaAs, are significantly influenced by the interface morphology. In the present work, high angle annular dark field imaging in (scanning) transmission electron microscopy ((S)TEM), in combination with contrast simulation, is used to address this question at atomic resolution. The (GaIn)As QWs were grown with metal organic vapor phase epitaxy on GaAs (001) substrates under different, precisely controlled conditions. In order to be able to compare different samples, a carefully applied method to gain reliable results from high resolution STEM micrographs was used. The thickness gradient of the TEM samples, caused by sample preparation, was compensated by the intensity of group V atomic columns, where no alloying takes place. After that, the In concentration map was plotted for the investigated regions based on the intensity of the group III atomic columns. The composition maps show that the Indium distribution across the quantum well is not homogeneous. The growth temperature of the QW can greatly influence the composition fluctuation and the interface morphology, with higher growth temperature resulting in larger composition fluctuations in the QWs and slightly wider interfaces, i.e. larger In-segregation. Growth interruptions are shown to significantly homogenize the elemental depth profile especially along the (GaIn)As/GaAs interface and hence have a positive effect on interface smoothness. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction Ternary (GaIn)As nano-structure materials are widely applied in modern optoelectronic device structures (Pletschen et al., 1993; Volz et al., 2008). Their performances are strongly influenced by the quality of the interface, which can be affected by the corresponding growth conditions, like growth temperature or growth interruptions applied at internal interfaces. In order to optimize the growth parameters and physical properties, characterization techniques that can provide sufficient structure and compositional information of the nano-sized devices are needed. The well-developed analytical transmission electron microscopy (TEM) techniques including high resolution TEM (Thoma and Cerva, 1991), energy dispersive X-ray spectroscopy (EDS) (Kothleitner et al., 2014), electron energy loss spectroscopy (EELS) (Leifer et al., 2000; Kimoto et al., 2007), as well as the high resolution X-ray characterization (Metzger et al., 2005; Volz et al., 2009) made it already possible to determine the composition in quantum wells (QW) at nanometer scale. However, to characterize and understand the nano-sized devices, information at atomic scale, especially across internal interfaces, is needed.

∗ Corresponding author. http://dx.doi.org/10.1016/j.micron.2015.07.003 0968-4328/© 2015 Elsevier Ltd. All rights reserved.

The high angle annular dark field (HAADF) technique in scanning TEM (STEM) provides Z-contrast imaging and the corresponding resolution can be down to sub-angstrom. The collected intensity of HAADF imaging scales with the average atomic number and specimen thickness, which means that quantitative information can be derived, if the measurement is accompanied by contrast simulation. Recently this technique is heavily researched, including both in experiment and in simulation, because of the easy interpretation of the image and the possibility to quantitatively analyze chemical compositions. With the introduction of spherical aberration (Cs)-correctors, the technique has already promoted the investigating limit from nanometer scale (Grieb et al., 2012) to atomic scale (Van Aert et al., 2009; Rosenauer et al., 2011; Martinez et al., 2014). Through comparison with STEM simulations, quantitative information like TEM specimen thickness and chemical composition can be obtained. With such atomic-scale chemical information, it is possible to characterize internal interface between two materials and quantify the influences of different growth parameters on interface composition and morphology. It has already been shown (Rosenauer et al., 2011; Grieb et al., 2012; Martinez et al., 2014), how to derive the chemical composition map at different length scales. Here we address especially, how to characterize the interfacial morphology based on the composition

2

H. Han et al. / Micron 79 (2015) 1–7

Table 1 Growth parameters for the different investigated (Ga1-x Inx )As QWs (x = 28%).

Temperature [◦ C] GIa @BIb [s] GI@TIc [s] a b c

Table 2 Settings used for frozen phonon simulation.

Sample 1

Sample 2

Sample 3

Acceleration voltage HT

200 kV

625 120 40

525 120 120

525 0 0

Defocus C1 Spherical aberration C3 Spherical aberration C5 Objective aperture angle HAADF collection angle Specimen thickness Zone axis No. of phonon configurations Pixel size of simulated image per unit cell

−2 nm 0.00232 mm 5 mm 24 mrad 30–300 mrad Up to 49.74 nm [010] 15 50

Growth interruption. Bottom interface (GaAs-to-(GaIn)As). Top interface ((GaIn)As-to-GaAs).

maps and how different growth conditions can influence interface morphology. In the current study, (GaIn)As/GaAs has been chosen as a model system, in order to investigate the internal interfaces between GaAs and (GaIn)As QWs (GaAs-to-(GaIn)As interface) as well as (GaIn)As and GaAs ((GaIn)As-to-GaAs interface). In addition, the influences of metal organic vapor phase epitaxy (MOVPE) growth conditions on interface composition and morphology is addressed. Thereto, quantitative evaluation of high resolution HAADF STEM images is used. To realize this aim, the thickness of each TEM sample is firstly determined based on high resolution imaging and corresponding STEM simulation. Before further quantitative chemical evaluation, thickness gradient compensation is done. Finally the In composition map is plotted at atomic resolution. Since the In concentration at the interface can reflect the interface morphology, different interfaces in correlation to their growth conditions are investigated. 2. Experiments The (GaIn)As multi QWs were grown on (001) GaAs substrate by MOVPE at different temperatures to investigate the influence on the GaAs-to-(GaIn)As interface as well as on the (GaIn)As-toGaAs interface and on the (GaIn)As QW itself. In addition, growth interruptions at all interfaces were applied and the growth interruption time was varied. All QWs have a width of approximately 9 nm, separated by 70 nm wide GaAs barriers. The nominal concentration of In in the investigated (GaIn)As QWs, determined from high-resolution X-ray measurements (Volz et al., 2009), is 28%. The relevant growth details of the samples are summarized in Table 1. The preparation of TEM sample in [010] zone axis orientation, consists of conventional mechanical grinding, polishing, and further thinning by dimpling grinder (Fischione model 200). To obtain thin electron transparent regions and remove most of the degraded surface caused by mechanical polishing, the samples were finally polished by argon ions at low angle of 4◦ at a voltage changing from 5 keV to 1.5 keV using Ar-ion milling in a precision ion polishing system (PIPS) (Gatan model 691). After sample preparation, the electron transparent region had a wedged shape. Directly before STEM measurement the samples were cleaned in a plasma cleaner (Fischione model 1020) for two minutes. The HAADF STEM images were obtained at room temperature using a double Cs-corrected JEOL JEM 2200FS at an acceleration voltage of 200 kV. The electron probe was formed with an aperture size of 40 ␮m and a corresponding convergence semi-angle of ˛ = 24 mrad. The images were recorded with a dwell time of 40 ␮s and large images with 1024 × 1024 pixels were recorded. A nominal camera length of 4 cm was used. The inner angle of the EM-24590YPDFI YAP dark field image detector from JEOL was determined by measuring the size of its shadow on a CCD camera. At the used camera length, the corresponding inner detection angle was approximately 71 mrad while the outer angle is four times of the inner angle with a maximum of 170 mrad, defined by geometric limitations of the microscope. To normalize the HR STEM images, the impinging beam current was determined from a beam image

(He and Li, 2014), recorded on the charge coupled device (CCD) camera. The used Orius SC2002 has a conversion factor of 0.1087 electron/count provided by the manufacturer. The determined current was around 25pA. After background subtraction, the counts of STEM images were also converted into current by dividing by the gain value of the detection system at fixed brightness and contrast settings. Complementary HAADF contrast simulation was carried out using StemSim program (Rosenauer and Schowalter, 2007) with frozen phonon approach taking into consideration the microscope settings. The frozen phonon approximation is based on multi slice method (Kirkland, 2010), which divides the specimen into thin slices and then calculates the interactions between the electron probe and each slice. During the calculation it also considers the influences of atomic thermal vibration on the image intensity through averaging the simulated results from different atomic configurations. For each configuration, the position of each atom is changed randomly with atomic displacements determined from thermal vibration. The main parameters used during the simulation are summarized in Table 2 and correspond to our experimental setup. The simulation was carried out with detection angles between 30 mrad and 300 mrad as listed in Table 2. After simulation, the detection angle corresponding to experimental settings (71–170 mrad), determined by the used camera length, was extracted for the quantitative analysis. The values of the mean intensity of one unit cell were recorded as a function of the specimen thickness as shown in Fig. 1. One can see that the intensity increases with specimen thickness. By comparing the experimental data with the calibration curve, the specimen thickness t can be determined. 3. Image evaluation For later quantitative evaluation of the micrographs, the average background intensity was determined by taking an image of the specimen hole without illumination using the same STEM settings

Fig. 1. Normalized mean intensity I of GaAs in dependence on the TEM sample thickness t.

H. Han et al. / Micron 79 (2015) 1–7

3

Fig. 2. An exemplary experimental HAADF STEM image of (GaIn)As/GaAs in [010] zone axis with the overlay of the schematic illustration of one Ga(In)As unit cell (Thick TEM sample). The blue dots indicate the group III atomic columns, whereas the red dots denote the group V atomic columns. Group III atomic columns are brighter at (GaIn)As QW and darker at GaAs barrier compared to their neighboring group V atomic columns. The line profile of the average intensity of the whole image is shown overlaid to the image.

as during the actual measurement. To finally be able to compare experimental images with simulated images, the images have to be normalized to the impinging electron beam after background intensity subtraction (Lebeau and Stemmer, 2008). The normalization was carried out according to the following equation Inorm =

Iexp − Ibkg Ibeam − Ibkg

(1)

where Inorm is the normalized intensity and can be directly compared with simulated results. Iexp is the measured image intensity in the experiment. Ibkg is the mean intensity of the background. Ibeam is the current of the impinging beam. As an example of the investigated micrographs a view in [010] zone axis with an overlay of the crystal structure of sample 2 is shown in Fig. 2. From the image, every atomic column can be clearly distinguished in the high resolution HAADF STEM images. On one hand, due to the heavier indium, the intensity from (GaIn) columns is higher than that from Ga ones. On the other hand, because of intermixing at the interface, the corresponding (GaIn)As-to-GaAs interface, as shown in the image, is not abrupt. It is worth noting that the distribution of doping atoms in the direction of the electron beam can severely influence the measured HAADF-intensity (Voyles et al., 2003). Considering the facts that in our study the In concentration is rather high (28%) and that there are no nanoclusters, the influence of the position In atoms on the collected intensity is negligible. Therefore we assume that it is reasonable to use the intensity for further quantitative evaluation. To reveal the interface, careful quantitative evaluation has been performed to determine the chemical composition information of every atomic column. Then the In composition change from the QW to the barrier and vice versa can reveal the shape of the interface. To realize this, a procedure has been used for all images, which is described in the following. From the original HAADF image (Fig. 3a) first a background subtraction and normalization to the impinging beam is applied as describe above (Fig. 3b). After that, every atomic column was identified in the images using the Peak Pairs Analysis (PPA) software (Galindo et al., 2009) and the group III and group V sublattices were separated. The intensity of each atomic column was integrated in a circle with a radius of 1/3rd of the next nearest neighbor. Asin our samples there is no chemical composition change from the QW to the barrier on the group V sublattice, the

Fig. 3. Evaluation process of the experimental images: (a) the original HAADF STEM image, (b) the analyzed image after background subtraction and normalized to the impinging beam, (c) Indium composition x map of the group III sublattice. In (b) the grey scale indicates the normalized intensity Inorm . In (c) the color bar means the In composition x in every atomic column.

4

H. Han et al. / Micron 79 (2015) 1–7

intensity change of the group V columns can be used to compensate the thickness gradient, since the wedge-shaped structure of the sample due to preparation would disturb further quantitative evaluation. The shape is fitted using a two-dimensional function to compensate the thickness gradient. This procedure can be used, as for the sample thicknesses, electron probe size and chosen crystal zone axis investigated here, there is no “cross talk” (Allen et al., 2010) between group III and group V atomic columns, which would make this evaluation impossible. After thickness compensation, the specimen has a homogeneous thickness and the minor thickness fluctuation is caused by the surface roughness of the TEM specimen. However, the sample thickness was determined based on the mean intensity of the whole unit cell of GaAs instead of the intensity of As columns using the theoretical dependence shown in Fig. 1, which has been verified by EELS measurements (Beyer et al., 2015). The separated intensity of As columns is used to check if there is cross scattering between neighboring atomic columns and to obtain the thickness gradient information used for correction. In contrast to the group V sublattice, for group III sublattice the In concentration changes from (GaIn)As QW to GaAs barrier. To determine the relative In concentration at every atomic column, two reference regions for both Ga (>300 atomic columns) and GaIn atomic columns (∼ 1200 atomic columns) were firstly chosen and the intensity of both Ga and GaIn atomic columns are measured and defined as IGa and IGaIn respectively. The reference regions should contain as much information of atomic columns as possible in the images, to make the statistical data reliable. Assuming that the investigated region has no thickness change after thickness gradient correction, the intensity of every atomic column is only determined by the average atomic number as there is no cross-talk. The In concentration x of every atomic column can be determined according to the following expression:



x=

0.28 Iexp − IGa



IGaIn − IGa

(2)

Here Iexp is the experimental intensity of every atomic column after thickness compensation. Finally the In composition map was plotted as shown in Fig. 3c. Here, Eq. (2) is an approximation for the indium composition determination. Although Van Aert et al. (2009) assumed the linear change of scattered intensity with changing chemical composition, Rosenauer et al. (2009) and Martinez et al. (2014) suggested a non-linear relation to describe it. However, in the current case the quantitatively characterized features of the interface (width, roughness) are not influenced by this linear description, considering that the In concentration is theoretically 0 in the barrier and for the whole QW it is on average 28%, as it is known from HR-XRD measurements. From the composition map the interface can be easily recognized and determined. 4. Results and discussion The HAADF STEM images of all three QWs examined are shown in Fig. 4 and the corresponding growth parameters are listed in Table 1. For the quantitative interpretation of the image, it is important to carefully choose the investigated regions. For very thick specimen regions the intensity of group V columns is influenced by the so-called cross talk from group III columns, hence also the data of the group III columns would not be reliable. In addition, strain relaxation (Peizel et al., 2000) can also especially influence the intensities at the interface. Exemplarily for unusable data, the intensity maps as shown in Fig. 5a andb show too thick TEM specimen (∼65 nm), where cross talk between group III and group V atomic columns took place, which makes the quantitative evaluation impossible. At thin regions the above influences are small

Fig. 4. HAADF STEM images of all the investigated samples, the growth direction is marked on the right side. The In-containing QWs can be seen from their bright contrast with respect to the GaAs barriers.

enough to be neglected as shown in Fig. 5c and 5d (sample thickness 39 nm, the expected gradient is observed). For carefully chosen thin regions, the In composition at the interface can reveal the shape of the interfaces because the barrier layer and the QW have different indium concentration. Comparison between different samples is possible, if the TEM specimen thickness of the compared samples is almost identical. Otherwise, differences could arise due to different plastic relaxation of the thin TEM specimens (Dunstan, 1997; Grieb et al., 2013). In addition, projection effects of different interface geometries in beam directions can hamper the evaluation if the TEM sample thickness is not identical. For example, structured interfaces can look similar to interdiffused ones, if the TEM samples thickness is only large enough. Also, since our comparison is mainly based on statistics, to make the statistical data from collected atomic columns between different images comparable and reasonable, the chosen region should also have the same field of view besides the same specimen thickness. After quantitative evaluation, the In composition map of the different investigated QWs with different growth parameters at the chosen regions are calculated and shown in Fig. 6. The average TEM specimen thickness is indicated above every map. It is obvious that the In composition at every atomic column in the QW is not homogeneous. The In concentration at some atomic columns can be almost twice as high as the nominal value, pointing towards compositional clustering of the In. In addition, the QW grown at higher temperature (sample1) seems to have higher fluctuations of the In content than the other QWs. This QW also has the roughest interfaces exhibiting different morphologies than the other QWs. For the QWs grown at 525 ◦ C, generally the bottom interface appears smoother than the top interface. To investigate the QW as well as its interfaces with the GaAs barriers quantitatively, the integrated line scan along [001] direction is analyzed for all three In composition maps. The profiles of the average In concentration is shown in Fig. 7. Every point in the profile indicates the average In composition of the atomic planes perpendicular to [001] direction. It can be seen that on average the composition of all three samples is identical. The standard deviation of the In composition of the plateau regions is shown in Fig. 8 together with the standard deviation of the GaAs (as reference for the experimental influence on the intensity). As the compositions are normalized to their respective value, the standard deviations are directly comparable. Now it can be also seen using statistical

H. Han et al. / Micron 79 (2015) 1–7

5

Fig. 5. Normalized intensity (Inorm ) map of both thick ((a) group III and (b) group V atomic columns) and thin specimen regions ((c) group III and (d) group V atomic columns) at atomic resolution. Due to the wedge shape the intensity decreases from the left side to the right side. The higher intensity in the middle of the maps indicates either the existence of (GaIn)As QWs for (a) and (c) or cross talk for (b). It is obvious that no cross talk happens in thin specimen regions as shown in (d).

data that higher growth temperatures lead to larger composition fluctuations, clearly pointing towards the advantage of low temperature growth. The interface regions are marked in Fig. 7 as 10% and 90% of the nominal In composition of 28%, respectively. The width of different interfaces is determined and depicted in Fig. 9, to gain information on the interface morphology and In segregation. All bottom GaAsto-(GaIn)As interfaces consist of 6 group III atomic layers, whereas, the top (GaIn)As-to-GaAs interfaces are significantly wider (13, 12, 10 group III atomic layers for sample1, sample2 and sample3, respectively). To determine the influences of strain relaxation on the interface, simulation of the GaAs-to-(GaIn)As interface with a TEM sample thickness of 40 nm (in agreement with the experimental images) was also carried out. The result shows that the interface for this geometry is 7 group III atomic layers wide, when taking into consideration only the strain relaxation and no In-atom diffusion. Obviously, in the current case the bottom GaAs-to-(GaIn)As interfaces are quite abrupt (no intermixing) while the top interfaces are much wider (intermixing) than the simulated one under the “smearing” effect from plastic relaxation. This is probably caused by the fact that during sample growth there were still some In atoms segregated on the growth surface (Volz et al., 2009). This In is used up during the growth of the first couple of monolayers of the

GaAs barriers. Also, an island-like structure on the (GaIn)As surface might exist which is then overgrown with GaAs upon barrier layer growth. At higher temperature (625 ◦ C) the (GaIn)As-to-GaAs interface is slightly wider than that at lower temperature. Hence, there is larger segregation at higher temperatures compared to lower temperatures. In addition to this information on the In depth profile, the corresponding standard deviations of the In compositions of every atomic column with [010] orientation in the atomic plane perpendicular to [001] direction are depicted in Fig. 10 for all investigated samples. This standard deviation is mainly determined by the distribution and size of islands and/or In-segregation in the interface. From a first view on the In composition maps (Fig. 6) the three samples exhibit similar features at the interfaces at the investigated scale. Low temperature growth only results in slightly more abrupt top (GaIn)As-to-GaAs interfaces. The standard deviation of the composition along the interface is, however, a more quantitative measure of the interface roughness. The smaller it is, the more homogeneous the In distribution, hence the lower the interface roughness. High temperature growth at 625 ◦ C results in larger In composition fluctuation at both interfaces, also indicating a more interdiffused bottom GaAs-to-(GaIn)As interface for high temperature growth compared to low temperature growth. The growth interruptions at 525 ◦ C have, however, the largest influence

6

H. Han et al. / Micron 79 (2015) 1–7

Fig. 8. Standard deviation of In composition x of the plateau region for the investigated QWs grown at different temperature as well as the barrier.

Fig. 9. No. of group III layers forming both the bottom and the top interfaces for all three samples. The interface is defined as the region, where the In composition x is between 10% and 90% of its mean value in the QW.

Fig. 6. In composition x map of (a) sample 1, (b) sample 2 and (c) sample 3. The thickness of the investigated TEM specimen is indicated above each map. All three maps have similar number of atomic columns.

among all the parameters varied in this study, especially on the top (GaIn)As-to-GaAs interface, as verified in Fig. 10. The standard deviation of the In composition fluctuation for the sample, where the growth was interrupted for 120 s on the (GaIn)As-to-GaAs

Fig. 10. Standard deviation of In composition x at both bottom and top interfaces of all three samples.

interface, is significantly lower than for all the other investigated samples pointing towards the importance of growth interruptions to smoothen internal interfaces. It might well be that during the growth interruption surface-segregated In desorbs from the surface or that an island-structure on the surface smoothens during this process. 5. Conclusions

Fig. 7. Average In composition x of every atomic plane perpendicular to [001] direction. The results for all three investigated samples are given.

The present study applies quantitative evaluation of high resolution HAADF STEM images to investigate influences of growth parameters, like growth temperature and growth interruption, on the chemical composition and on the interface roughness of (GaIn)As/GaAs as a model system. After compensation of the TEM

H. Han et al. / Micron 79 (2015) 1–7

specimen thickness gradient, quantitative In composition maps have been derived for the investigated regions. The In composition map clearly reveals the different homogeneities of the samples and different morphologies of the interfaces of the samples grown under different conditions. Using a statistical analysis, it can be quantitatively stated that the growth temperature of the (GaIn)As has a large influence on homogeneity of In-distribution in the QW and the composition fluctuation at the interface. Growth at 625 ◦ C results in large composition fluctuation in the QW as well as large composition fluctuation at the (GaIn)As-to-GaAs interface. Applying growth interruptions has a major influence on the (GaIn)As-to-GaAs interface, which significantly homogenizes during this process. Acknowledgement We gratefully acknowledge the financial support of the Deutsche Forschungsgemeinschaft (DFG) in the frame work of SFB 1083 (‘Structure and dynamics of internal interface’). References Allen, L.J., DAlfonso, A.J., Findlay, S.D., Lebeau, J.M., Lugg, N.R., Stemmer, S., 2010. Elemental mapping in scanning transmission electron microscopy. J. Phys. Conf. 241, 012061. Beyer, A., Straubinger, R., Belz, J., Volz, K., 2015. Local sample thickness determination via scanning transmission electron microscopy defocus series. J. Microsc., http:// dx.doi.org/10.1111/jmi.12284 Dunstan, D.J., 1997. Strain and strain relaxation in semiconductors. J. Mat. Sci.: Mater. Electron. 8, 337–375. Galindo, P., Pizarro, J., Molina1, S., Ishizuka, K., 2009. High resolution peak measurement and strain mapping using peak pairs analysis. Microsc. Anal., 23–25. Grieb, T., Müller, K., Fritz, R., Grillo, V., Schowalter, M., Volz, K., Rosenauer, A., 2013. Quantitative chemical evaluation of dilute GaNAs using ADF STEM: Avoiding surface strain induced artifacts. Ultramicroscopy 129, 1–9. Grieb, T., Müller, K., Fritz, R., Schowalter, M., Neugebohrn, N., Knaub, N., Volz, K., Rosenauer, A., 2012. Determination of the chemical composition of GaNAs using STEM HAADF imaging and STEM strain state analysis. Ultramicroscopy 117, 15–23. He, D.S., Li, Z.Y., 2014. A practical approach to quantify the ADF detector in STEM. J. Phys.: Conf. Ser. 522, 012017. Kirkland, E.J., 2010. Advanced Computing in Electron Microscopy, second ed. Springer, New York.

7

Kimoto, K., Asaka, T., Nagai, T., Saito, M., Matsui, Y., Ishizuka, K., 2007. Elementselective imaging of atomic columns in a crystal using STEM and EELS. Nature 450, 702–704. Kothleitner, G., Neish, M.J., Lugg, N.R., Findlay, S.D., Grogger, W., Hofer, F., Allen, L.J., 2014. Quantitative Elemental mapping at atomic resolution using X-ray spectroscopy. Phys. Rev. Lett. 112, 085501. Leifer, K., Buffat, P.A., Stadelmann, P.A., Kapon, E., 2000. Theoretical and experimental limits of the analysis of III/V semiconductors using EELS. Micro 31, 411–427. Lebeau, M.J., Stemmer, S., 2008. Experimental quantification of annular dark-field images in scanning transmission electron microscopy. Ultramicroscopy 108, 1653–1658. Metzger, T.H., Schülli, T.U., Schmidbauer, M., 2005. X-ray characterization of selforganized semiconductor nanostructures. C. R. Phys. 6, 47–59. Martinez, G.T., Rosenauer, A., De Backer, A., Verbeeck, J., Van Aert, S., 2014. Quantitative composition determination at the atomic level using model-based high-angle annular dark field scanning transmission electron microscopy. Ultramicroscopy 137, 12–19. Peizel, R.I., Zepeda-Ruiz, L.A., Weinberg, W.H., Maroudas, D., 2000. Effects of buffer layer thickness and film compositional grading on strain relaxation kinetics in InAs/GaAs(111)A heteroepitaxy. Surf. Sci. 463, 634–640. Pletschen, W., Bachem, K.H., Tasker, P.J., Winkler, K., 1993. AlGaInP/GaInAs/GaAs MODEFT devices: candidates for optoelectronic integrated circuits. Mater. Sci. Eng. B 21, 304–306. Rosenauer, A., Schowalter, M., 2007. STEMSIM-a new software tool for simulation of STEM HAADF Z-contrast imaging. In: Microscopy of Semiconducting Materials. Springer, Netherlands, pp. 170–172. Rosenauer, A., Gries, K., Müller, K., Pretorius, A., Schowalter, M., Avramescu, A., Engl, K., Lutgen, S., 2009. Measurement of specimen thickness and composition in Alx Ga1−x N/GaN using high-angle annular dark field images. Ultramicroscopy 109, 1171–1182. Rosenauer, A., Mehrtens, T., Müller, K., Gries, K., Schowalter, M., Satyam, V.P., Bley, S., Tessarek, C., Hommel, D., Sebald, K., Seyfried, M., Gutowski, J., Avramescu, A., Engl, K., Lutgen, S., 2011. Composition mapping in InGaN by scanning transmission electron microscopy. Ultramicroscopy 111, 1316–1327. Thoma, S., Cerva, H., 1991. New methods for qualitative and quantitative analysis of the GaAs/AlGaAs interface by high-resolution electron microscopy. Ultramicroscopy 38, 265–289. Van Aert, S., Verbeeck, J., Erni, R., Bals, S., Luysberg, M., Van Dyck, D., Van Tendeloo, G., 2009. Quantitative atomic resolution mapping using high-angle annular dark field scanning transmission electron microscopy. Ultramicroscopy 109, 1236–1244. Volz, K., Stolz, W., Teubert, J., Klar, P.J., Heimbrodt, W., Dimroth, F., Baur, C., Bett, A.W., 2008. Doping, electrical properties and solar cell application of GaInNAs. In: Erol, A. (Ed.), Dilute III–V Nitride Semiconductors and Material Systems. Springer, Berlin/Heidelberg, pp. 269–404. Volz, K., Koch, J., Höhnsdorf, F., Kunert, B., Stolz, W., 2009. MOVPE growth of dilute III/V semiconductors using all liquid metalorganic precursors. J. Cryst. Growth 311, 2418–2426. Voyles, P.M., Grazul, J.L., Muller, D.A., 2003. Imaging individual atoms inside crystals with ADF-STEM. Ultramicroscopy 96, 251–273.