## research papers

## X-ray diffraction strain analysis of a single axial InAs_{1–x}P_{x} nanowire segment

**Mario Keplinger,**

^{a}^{*}Bernhard Mandl,^{a,}^{b}Dominik Kriegner,^{a}Václav Holý,^{c}Lars Samuelsson,^{b}Günther Bauer,^{a}Knut Deppert^{b}and Julian Stangl^{a}^{a}Solid State and Semiconductor Physics, Johannes Kepler University Linz, A-4040 Linz, Austria, ^{b}Solid State Physics, Lund University, S-22 100 Lund, Sweden, and ^{c}Faculty of Mathematics and Physics, Charles University, Ke Karlovu 5, 121 16 Prague, Czech Republic^{*}Correspondence e-mail: kepimario@gmx.at

The spatial strain distribution in and around a single axial InAs_{1–x}P_{x} hetero-segment in an InAs nanowire was analyzed using nano-focused X-ray diffraction. In connection with finite-element-method simulations a detailed quantitative picture of the nanowire's inhomogeneous strain state was achieved. This allows for a detailed understanding of how the variation of the nanowire's and hetero-segment's dimensions affect the strain in its core region and in the region close to the nanowire's side facets. Moreover, ensemble-averaging high-resolution diffraction experiments were used to determine statistical information on the distribution of wurtzite and zinc-blende crystal in the nanowires.

Keywords: nanowire; nano-focus; strain analysis; hetero-structure; finite-element simulation.

### 1. Introduction

For the fabrication of future electronic and opto-electronic devices, investigations of new device designs made from precisely tailored semiconductor materials are crucial. Nanowires (NWs), for instance, offer a great flexibility of altering their material's electronic and optoelectronic properties in beneficial ways. Being one-dimensional in the quantum mechanical sense renders their physical properties different from three-dimensional bulk. Moreover, by including axial hetero-segments of different materials, the formation of zero-dimensional quantum dots or tunnel barriers becomes possible. Elastic relaxation due to the small diameter allows for larger *et al.*, 2012; Borgström *et al.*, 2011; Wallentin *et al.*, 2010; Roddaro *et al.*, 2011; Reimer *et al.*, 2012; Romeo *et al.*, 2012). Inhomogeneous strain fields are intimately connected to such heterostructures, which can alter the band structure of semiconductor materials dramatically. This can be exploited, for example, to enhance electron mobility (Chu *et al.*, 2009; Hrauda *et al.*, 2011) or to alter electro-optical properties (Greil *et al.*, 2012). Control on the details of strain distributions in nanostructures is thus an important task in device fabrication, which in turn requires precise analysis tools to monitor those strain fields. In this work we present a detailed strain characterization of InAs NWs containing a short axial InAs_{1–x}P_{x} segment. Both ensemble-average experiments as well as the analysis of a single NW using nano-focused synchrotron radiation have been carried out. The NWs are not only strained due to the mismatch of the hetero-segment; strain is also introduced from a stacking of wurtzite (WZ) and zinc-blende (ZB) since the wurtzite polytype has a slightly different spacing than zinc-blende (Kriegner *et al.*, 2011*a*,*b*). High-resolution X-ray diffraction (HRXRD) measurements on ensembles of the as-grown heterostructured NWs were used to obtain the average WZ/ZB polytype segment length and the wurtzite content. With nano-focused X-ray diffraction (XRD) we precisely monitored the chemical composition and dimension of the investigated InAs_{1–x}P_{x} segment. The NW's chemical composition and dimension define the strain distribution in the region inside and around the hetero-segment. Nano-focused XRD allows for illuminating a single InAs_{1–x}P_{x} segment of one NW and to record the diffracted intensity of the segment and its vicinity, from which its three-dimensional strain distribution is found. The WZ/ZB needs to be considered in the analysis to yield correct results. With the presented method a precise knowledge of the spatial strain variation is gained, a prerequisite for the fabrication of novel strain engineered NW devices.

### 2. Sample fabrication

The NW fabrication followed in principle the self-seeded In particle-assisted growth scheme reported by Mandl *et al.* (2006) and Mandl *et al.* (2010). InP (111)B substrates were covered with a ∼13 Å thin SiO_{x} layer (*x* ≃ 1). This deposition was carried out using thermal and the layer thickness was controlled by an oscillating quartz system mounted in the deposition chamber. During all growth steps the temperature was fixed to 853 K, and H_{2} was used as with a flow rate of 13000 ml min^{−1}. By using a special switching sequence of gas precursors in an Aixtron 200/4 metal-organic vapour phase reactor, an InAs_{1–x}P_{x} segment was embedded in an InAs NW. First the InAs core NW was grown for 2 min with a trimethylindium and arsine precursor molar fraction of 1.1 × 10^{−5} and 3.8 × 10^{−4}. Then InAs_{1–x}P_{x} growth was initiated for 10 s with a phosphine and arsine precursor molar fraction of 6.2 × 10^{−3} and 1.9 × 10^{−4}. Finally, a second InAs segment was grown under the same conditions as the first one. Fig. 1(*a*) shows an illustration of the fabricated structure, where additional radial shells, due to a finite radial growth rate typical of the self-seeded particle-assisted mechanism, are indicated. However, these shells are extremely thin compared with the axial segment lengths. From a detailed growth study (to be published elsewhere) we found a rather wide distribution of shell thicknesses within the NW ensemble from only a few nanometres up to 20 nm.

The coordinate system used later on is also depicted in the inset of Fig. 1(*a*): the NW growth direction, the cubic [111] direction, is depicted as the *z*-direction. The in-plane directions *x* and *y* in the substrate surface are along the crystallographic [] and [] directions, respectively.

### 3. Ensemble characterization experiments

For a first characterization of the NW shape and size, scanning electron microscope (SEM) images were recorded in a Jeol 6400F system, shown in Fig. 1(*b*). An average NW diameter of 180 ± 40 nm and an average NW length of 3.8 ± 0.7 µm were found. This means that the NWs on the sample exhibit a certain variation of their geometric properties within the as-grown ensemble. The hetero-segment is not visible in the SEM micrographs, and also the WZ/ZB stacking properties cannot be obtained.

For the NW ensemble's X-ray characterization, intensity distributions around the symmetric (111) and asymmetric () InAs/InAs_{1–x}P_{x} Bragg reflections^{1} were recorded at the HRXRD beamline BW2 at HASYLAB in Hamburg, Germany, using an X-ray energy of 9.5 keV. The recorded reciprocal-space map around the (111) InAs/InAs_{1–x}P_{x} signal is shown in Fig. 2(*a*), and Fig. 2(*b*) shows the intensity distribution around the hexagonal () peak. Note that no information on the InAs_{1–x}P_{x} segment and the thin shells is gained from the ensemble HRXRD experiments, since their scattering volume is too small compared with the entire NW volume to contribute significantly to the diffraction signal. However, the fraction of WZ material in the NWs and the average WZ segment length can be deduced from this experiment.

The (111) Bragg reflection is allowed for both ZB and WZ structure, for which it corresponds to the (000.2) hexagonal Bragg reflection. Therefore, it is sensitive to the *d*-spacing distribution in the [111] direction of the whole NWs, but almost insensitive to the WZ/ZB stacking sequence. The only influence of the stacking is indirect *via* the slightly different parameters for ZB and WZ. This leads to shifts of the NW's Bragg peak position as a function of the overall WZ/ZB ratio and a significantly smaller broadening in the *Q*_{z} direction as for the () reflection discussed below. The position of the (111) NW Bragg reflection is therefore a way to determine the *total fraction* of WZ segments in the NW. The average NW ensemble parameter and its variation due to the WZ/ZB was calculated using finite-element-method (FEM) simulations and was then compared with the (111) NW Bragg reflection's position and shape in the *Q*_{z} direction, which resulted in a WZ content of around 70%. Detailed explanation of the FEM simulations, which have been performed using a commercial software package, will be given in the next section. The *Q*_{z} positions of pure zinc-blende InAs and WZ/ZB mixture are indicated in Fig. 2(*a*). Moreover, a cut through the NW signal is plotted which shows an asymmetry of the (111) NW Bragg reflection in the *Q*_{x} direction. This is also seen in the intensity profile of the () Bragg reflection, see Fig. 2(*b*), which stems from an asymmetric tilt distribution of the NWs on the sample. Such NW tilting and bending have been discussed by Keplinger *et al.* (2010), and possible reasons are defects at the substrate interface or an inhomogeneous thickness of the grown shell in heterostructured NWs. From the evaluation of the (111) Bragg peak position as well as its FWHM in the [] direction we calculated a tilt distribution of ±0.3° around an offset tilt angle of ±0.1°.

For the () Bragg reflection shown in Fig. 2(*b*), the parts of the sample having WZ provide the major amount of scattered intensity, except for a small, almost constant, contribution from the crystal truncation rod of the cubic (224) reflection, which is already far away at *Q*_{z} = 4.8 Å^{−1}. Since we are not dependent on the absolute intensity values in the data evaluation but only the peak shape, this crystal truncation rod contribution to the intensity of the () Bragg reflection can be neglected. From a reference measurement (to be published elsewhere) we confirmed that the WZ phase occurs only within NWs on this sample, and not from a possible two-dimensional layer, as well as rather bulky crystallites on the sample visible in Fig. 1(*b*), which are purely ZB. This finding confirms earlier studies, *e.g.* by Eymery *et al.* (2007) and Kriegner *et al.* (2011*a*), who also find WZ phase exclusively in NWs. In the former case, truncation rod analysis in grazing-incidence geometry is used to disentangle planar overgrowth in NW material.

For a single NW with alternating WZ/ZB segments the stacking gives rise to a special speckle pattern of hexagonal Bragg peaks (Favre-Nicolin *et al.*, 2009; Chamard *et al.*, 2009), which is nothing else than the Fourier transformation of the particular WZ distribution function along this particular NW. The average length of the WZ segments defines the speckle pattern's envelope function's full width at half-maximum (FWHM) in the *Q*_{z} direction. When performing ensemble measurements as in this case, *i.e.* illuminating a large number of NWs, the recorded intensity distribution is an incoherent superposition of intensities originating from various NWs. This `smears out' the speckle pattern and eventually its envelope defines the shape of the Bragg reflection, as observed in Fig. 2(*b*).

To simulate the shape function of the () Bragg reflection in the *Q*_{z} direction, a Monte Carlo approach was used: arbitrary stacks of WZ and ZB segments with randomly varied lengths were assembled using computer-generated pseudo random numbers. This results in WZ and ZB segment lengths distributed according to a Gamma function with a certain mean length and variation. Then, the ZB and WZ segments were assigned with a scattering density according to = ± (Δρ_{WZ/ZB}/2). The mean scattering density was set to zero to avoid very sharp and intense features at zero frequency in Fourier space, since the simulations should just describe the shape of the scattered intensity distribution. From such WZ/ZB distribution functions simulated diffraction patterns of the hexagonal () Bragg reflection were then calculated using a discrete Fourier transformation. By calculating the arithmetic average of many statistically varied simulated diffraction patterns the real situation of many incoherently scattering wires within the illuminated volume was simulated. To verify the statistical quality of this Monte Carlo approach, *i.e.* ensuring that the resulting signal's properties only depend on the average length as well as the length variation of the simulated WZ segments, the relative root-mean-square deviation was calculated for each Monte Carlo step:

where is the *j*th simulated intensity distribution and *M* is the total number of simulations. With *M* = 150 simulated individual ZB/WZ stacks a max[] far below 1% was obtained. Quantitative results were then found by varying the Gamma distribution function parameters, *i.e.* the average WZ segment length and its variation, until a good correspondence with the measured intensity pattern was obtained. Fig. 2(*b*) shows the measured two-dimensional intensity distribution of the () InAs/InAs_{1–x}P_{x} Bragg reflection along with the simulated shape function (green) compared with the measured intensity distribution (blue) at the position of the white dashed line. The resulting wurtzite segment length distribution parameters are: a mean length of 19 nm with a σ of 15 nm, see Fig. 2(*c*).

### 4. Investigation of a single InAs_{1–x}P_{x} segment

To gain further insight into the investigated structures and in particular into the short InAs_{1–x}P_{x} hetero-segment, we performed nano-focused X-ray diffraction experiments at beamline ID01 at the ESRF in Grenoble, France, at an X-ray energy of 10.2 keV. By using a Fresnel zone plate the X-ray beam was focused down to a FWHM of 200 nm × 350 nm. With such a nano-focused beam, only a small part of a single NW is illuminated. To align the InAs_{1–x}P_{x} hetero-segment of a single wire into the focal spot of the X-ray beam, first a rough pre-alignment of the sample with an optical microscope, mounted on the goniometer, was performed. However, in this way the position of the X-ray spot on the sample can only be determined within a precision of a few micrometres. A finer alignment followed using scanning diffraction microscopy, as described by Diaz *et al.* (2009): by setting the detector and sample angle being sensitive to the NW signal, one records the intensity while moving the sample laterally. This gives a real-space map of the NW positions with a resolution according to the focused X-ray beam size. After finding and selecting one NW with this method and moving the X-ray spot along the NW to find the hetero-segment position, the scattered intensity was recorded using a two-dimensional MAXIPIX detector by rocking the sample around the (333) InAs/InAs_{1–x}P_{x} Bragg reflection. This results in the measurement of a three-dimensional reciprocal-space map (RSM). After every sample rocking step of 0.01° a realignment of the InAs_{1–x}P_{x} segment into the centre of the beam, due to a minimal sample drift with respect to the X-ray beam, was necessary, which was carried out using two lateral sample movement scans. Fig. 3 shows the measured intensity distribution as a three-dimensional contour plot and additionally as two-dimensional contour plots showing slices through the point of maximum intensity along all three principle axes. In the *Q*_{x}/*Q*_{z} and *Q*_{y}/*Q*_{z} slices as well as in the three-dimensional illustration of the data the intensity maximum around *Q*_{z} = 5.375 Å^{−1} is attributed to the InAs part of the illuminated real-space volume, whereas the broader signal around *Q*_{z} = 5.41 Å^{−1} comes from the InAs_{1–x}P_{x} segment. Moreover, in the *Q*_{x}/*Q*_{y} slice facet streaks with thickness oscillations from the NW side facets are observed. The InP (333) substrate peak, which would occur at *Q*_{z} = 5.563 Å^{−1}, is outside the plotted *Q* range. The (333) Bragg reflection is sensitive to the *d*-spacing distribution only along the [111] direction. However, due to the known symmetric shape of the NWs, the other strain components become linked to the strain along the [111] direction, so that performing finite-element calculations and simulations of the resulting intensity distribution around the (333) reflection actually allow conclusions to be drawn also on the strain components in the *x* and *y* directions. To deduce the spatial strain distribution, FEM calculations and kinematical scattering simulations were used to reproduce the measured RSMs. The NW was simulated as a three-dimensional object with the outer dimensions taken from SEM measurements. The phosphorous content in the InAs_{1–x}P_{x} segment was then introduced by applying initial strain according to the mismatch of InAs and InAs_{1–x}P_{x} in the heterostructure part of the simulated NW. Additionally, arbitrarily distributed wurtzite and zinc-blende segments were mimicked by randomly distributing wurtzite segments, with the same dimensions and distribution function as obtained from the simulations of the ensemble measurements. As an initial strain state of these wurtzite segments, anisotropic strain according to the parameter deviations reported by Kriegner *et al.* (2011*a*) for WZ InAs and by Kriegner *et al.* (2011*b*) for WZ InP was used. The parameters as well as the elastic constants of cubic InAs and InP were taken from IOFFE (2014). Elastic constants were transformed according to Hirth & Lothe (1992) and Martin (1972) for ZB and WZ respectively. For the parameter variation of InAs_{1–x}P_{x} according to the P content, Vegard's law was used. With the simulated displacements field from FEM^{2} the diffusely scattered intensity was calculated using kinematical scattering theory.

### 5. Discussion

The parameter space of the simulations was spanned by the InAs_{1–x}P_{x} segment thickness, its phosphorous concentration, the thickness of the InAs and InAs_{1–x}P_{x} shells, the outer diameter of the NW, and finally the WZ/ZB segment structure. Also the illumination of only a part of the NW by the focused beam needs to be taken into account. First, to narrow down the parameter space, a range of reasonable parameter values was mapped in a series of simulations, where the agreement between measurement and simulation was judged `guided by the eye'. After the relevant range of parameter space was found, the accuracy of the simulations was additionally judged by calculating the sum of squared residuals for every simulation. The whole procedure has been iterated several times to find the optimum values for all parameters.

The outer wire diameter was initially chosen according to SEM micrographs and slightly varied around this value. The simulated thickness fringes depend very sensitively on the NW diameter, so that it could be fixed to 175 nm. An initial guess of the P content was taken from the position of the diffusely scattered intensity from the InAs_{1–x}P_{x} segment. The variation of the phosphorous content strongly influences the *Q*_{z} position of the InAs_{1–x}P_{x} signal. An agreement between simulation and measurement was found within a variation of ±0.5% of the phosphorous content from the optimum value of 20%.

The remaining NW dimensions can be quantified with less precision. The InAs_{1–x}P_{x} segment length, for example, can already be varied by ±4 nm around the optimal value of 43 nm, and still an acceptable agreement between simulation and measurement is found. Similarly, the thicknesses of the inner InAs_{1–x}P_{x} and the outer InAs shells have a more subtle influence on the simulations, as is illustrated in Fig. 4. Although the shells are very thin, neglecting them does significantly change the strain distribution, and therefore it is important to take them into account. Fig. 4(*a*) shows a simulation performed without shells, where the agreement of simulation and measurement is poor at higher *Q*_{z} values: above 5.425 Å^{−1} the intensity values of the simulation are higher compared with the measurement. On the other hand, simulating thicker shells results as well in a poorer agreement between simulation and measurement. This is illustrated in Fig. 4(*c*), where the regions with very poor agreement are again indicated. The optimum inner- and outer-shell thicknesses were found to be 1 nm and 2.5 nm, respectively. The strain values from the FEM simulation with these parameters are illustrated in Fig. 5, and the resulting simulated intensity pattern compared with the measurement is shown in Fig. 4(*b*).

In summary, the phosphorous content and the outer NW diameter influence the simulated signal stronger than the thickness of the InAs_{1–x}P_{x} segment and the shell thicknesses. The results show a large axial/radial growth ratio producing only very thin shells.

To illustrate the influence of different shell parameters and also the WZ/ZB segment distribution on the strain variation inside the NW, Fig. 6 shows a comparison of different FEM simulations with different shell dimensions. This results in a change of the strain values close to the circumference of the NW. One can clearly see that not only is the InAs_{1–x}P_{x} segment strained but also the region around it as well as the thin shells. The highest strain values perpendicular to the NW growth direction (of the order of 0.6%) are located in the NW core region of the hetero-segment, but also above and below the segment in the InAs region. While the hetero-segment exhibits tensile strain, the regions above and below are compressively strained, see Fig.6(*b*). Owing to relaxation at the side walls, the strain values close to the NW border decrease compared with the NW centre. The shells influence the strain distribution particularly in these regions, and strains are underestimated in the simulation neglecting the shells. Already the introduction of a thin core-shell structure increases the strain values in the region near the NW side walls, leaving the strain in the core region virtually unaltered. An increase of the shell thicknesses is enhancing this effect, see Fig. 6(*c*). Thus, by changing the structural parameters one can achieve a tailored strain distribution. Moreover, the different WZ/ZB constants locally introduce strain changes of the order of 0.2 to 0.3%, as can be seen in Fig. 6. This cannot be neglected compared with the strain values due to the hetero-segment. Simulations of the scattered intensity distribution from FEM results with different random distributions of WZ and ZB segments give very similar results, therefore we are not able to determine the particular segment distribution in the illuminated NW segment. However, not taking into account the strain caused by the results in an incorrect overall peak position. For a tight control of the strain distribution, it will thus be extremely beneficial to control the WZ/ZB during growth (Bolinsson *et al.*, 2011; Spirkoska *et al.*, 2009).

To account for the finite size of the focused beam a simple approach was used, where the Fourier coefficients of the crystal polarizability in the simulated NW's growth direction were multiplied by an illumination function with minimum and maximum values of 0 and 1, respectively. This approach is reflecting the real physical situation of the nano-focused scattering experiment, where the focused beam and its transverse coherence length are at the same order of magnitude due to a partial illumination of the Fresnel zone plate during the experiment. Since we did not determine the exact shape and phase of the focused beam in this study, three simple illumination functions were compared: a box function, a sinc^{2} function and a constant illumination of the whole simulated NW as depicted in Fig. 7(*b*). Fig. 7(*a*) illustrates a comparison of the resulting RSMs from simulations with different illumination functions. The results for the sinc^{2} and Box functions are rather similar, *i.e.* we are not sensitive to details in the illumination function. However, by neglecting the finite beam size and calculating the scattered intensity distribution from the whole simulated NW (corresponding to a plane incident wave illuminating the whole simulated NW), very poor agreement between measurement and simulation was achieved. This can be seen around *Q*_{z} = 5.375 Å^{−1} where the features from the measurement have not been reproduced, *e.g.* the diagonal intensity stripes in the measurement are not reflected in the simulation. A more precise approach for the estimation of the focused beam's properties was used by Mastropietro *et al.* (2013), where the beam size and shape was calculated from the estimated undulator source size and the utilized FZP, and Schropp *et al.* (2010) have analyzed the beam properties using ptychography experiments, resulting in a detailed picture of the focused beam's shape and coherence properties. Taking into account those beam profiles does not, however, improve our simulation results any more.

### 6. Conclusion

By using X-ray diffraction with a nano-focused beam, we have performed a detailed analysis of the strain distribution in and around a single axial InAs_{1–x}P_{x} hetero-segment in an InAs NW. Quantitative results were deduced using simulations of the scattered intensity distribution, resulting in a complete picture of the investigated structure. The hetero-segment's chemical composition, the geometrical parameters and also the thicknesses of the thin radial shells were deduced for a single as-grown nanowire. Additional ensemble-averaged experiments were used to evaluate the WZ/ZB ratio of the fabricated nanowires as well as the average crystallite's segment length. The strain distribution was found to be rather inhomogeneous due to the combination of axial and radial growth in the self-seeded particle-assisted growth mode. The precise measurement of the strain distribution in NW heterostructures provides the knowledge for the fabrication of NW devices with dedicated physical properties.

### Footnotes

^{1}We use Miller and Miller–Bravais point index notation to distinguish cubic and hexagonal crystal respectively.

^{2}In the construction of the FEM mesh, the maximum length of the longest mesh elements edge was set to 3 nm in the shell region, whereas in the rest of the simulated structure it was chosen coarser at 6 nm. To reduce computational cost the model contained only one-quarter of the NW; the result of the full NW was then reconstructed using symmetry operations according to the wire geometry and crystal structure.

### Acknowledgements

This work was supported by the European Commission (EC) (NODE cont. no. 015783, CALIPSO cont. no. 312284, AMON-RA cont. no. 214814), the FWF Vienna (SFB IR-On F2507-N08, P23706-N19), the Swedish Foundation for Strategic Research (SSF), the Swedish Research Council (VR), the Knut and Alice Wallenberg Foundation, and the Czech Science Foundation (project 14-08124S). We would like to thank Dmitri Novikov for assistance with the synchrotron measurements at DESY in Hamburg, and the staff of beamline ID01 at the ESRF Grenoble for assistance during the nano-focus experiment.

### References

Bolinsson, J., Caroff, P., Mandl, B. & Dick, K. A. (2011). *Nanotechnology*, **22**, 265606. Web of Science CrossRef PubMed Google Scholar

Borgström, M. T., Wallentin, J., Heurlin, M., Fält, S., Wickert, P., Leene, J., Magnusson, M. H., Deppert, K. & Samuelson, L. (2011). *Selected Topics Quantum Electron. IEEE*, **17**, 1050–1061. Google Scholar

Chamard, V., Diaz, A., Stangl, J. & Labat, S. (2009). *J. Strain Anal.* **44**, 533–542. Web of Science CrossRef Google Scholar

Chu, M., Sun, Y., Aghoram, U. & Thompson, S. E. (2009). *Annu. Rev. Mater. Res.* **39**, 203–229. Web of Science CrossRef CAS Google Scholar

Diaz, A., Mocuta, C., Stangl, J., Mandl, B., David, C., Vila-Comamala, J., Chamard, V., Metzger, T. H. & Bauer, G. (2009). *Phys. Rev. B*, **79**, 125324. Web of Science CrossRef Google Scholar

Eymery, J., Rieutord, F., Favre-Nicolin, V., Robach, O., Niquet, Y.-M., Fröberg, L., Mårtensson, T. & Samuelson, L. (2007). *Nano Lett.* **7**, 2596–2601. Web of Science CrossRef PubMed CAS Google Scholar

Favre-Nicolin, V., Eymery, J., Koester, R. & Gentile, P. (2009). *Phys. Rev. B*, **79**, 4546–4553. Google Scholar

Ganjipour, B., Wallentin, J., Borgström, M. T., Samuelson, L. & Thelander, C. (2012). *ACS Nano*, **6**, 3109–3113. Web of Science CrossRef CAS PubMed Google Scholar

Greil, J., Lugstein, A., Zeiner, C., Strasser, G. & Bertagnolli, E. (2012). *Nano Lett.* **12**, 6230–6234. Web of Science CrossRef CAS PubMed Google Scholar

Hirth, J. P. & Lothe, J. (1992). *Theory of Dislocations.* Malabar: Krieger. Google Scholar

Hrauda, N., Zhang, J., Wintersberger, E., Etzelstorfer, T., Mandl, B., Stangl, J., Carbone, D., Holý, V., Jovanovic, V., Biasotto, C., Naver, L. K., Moers, J., Grützmacher, D. & Bauer, G. (2011). *Nano Lett.* **1**, 2875. Web of Science CrossRef Google Scholar

IOFFE (2014). *Semiconductors on NSM*, http://www.ioffe.ru/SVA/NSM/Semicond/index.html. Google Scholar

Keplinger, M., Kriegner, D., Stangl, J., Mårtensson, T., Mandl, B., Wintersberger, E. & Bauer, G. (2010). *Nucl. Instrum. Methods Phys. Res. B*, **268**, 316–319. Web of Science CrossRef CAS Google Scholar

Kriegner, D., Panse, C., Mandl, B., Dick, K. A., Keplinger, M., Persson, J. M., Caroff, P., Ercolani, D., Sorba, L., Bechstedt, F., Stangl, J. & Bauer, G. (2011*a*). *Nano Lett.* **11**, 1483–1489. Web of Science CrossRef CAS PubMed Google Scholar

Kriegner, D., Wintersberger, E., Kawaguchi, K., Wallentin, J., Borgström, M. T. & Stangl, J. (2011*b*). *Nanotechnology*, **22**, 425704. Web of Science CrossRef PubMed Google Scholar

Mandl, B., Stangl, J., Hilner, E., Zakharov, A. A., Hillerich, K., Dey, A. W., Samuelson, L., Bauer, G., Deppert, K. & Mikkelsen, A. (2010). *Nano Lett.* **10**, 4443–4449. Web of Science CrossRef CAS PubMed Google Scholar

Mandl, B., Stangl, J., Mårtensson, T., Mikkelsen, A., Eriksson, J., Karlsson, L., Bauer, G., Samuelson, L. & Seifert, W. (2006). *Nano Lett.* **6**, 1817–1821. Web of Science CrossRef PubMed CAS Google Scholar

Martin, R. M. (1972). *Phys. Rev. B*, **6**, 4546–4553. CrossRef CAS Web of Science Google Scholar

Mastropietro, F., Eymery, J., Carbone, G., Baudot, S., Andrieu, F. & Favre-Nicolin, V. (2013). *Phys. Rev. Lett.* **111**, 215502. Web of Science CrossRef PubMed Google Scholar

Reimer, M. E., Bulgarini, G., Akopian, N., Hocevar, M., Bavinck, M. B., Verheijen, M. A., Bakkers, E. P. A. M., Kouwenhoven, L. P. & Zwiller, V. (2012). *Nanotechnology*, **3**, 737. Google Scholar

Roddaro, S., Pescaglini, A., Ercolani, D., Sorba, L. & Beltram, F. (2011). *Nano Lett.*, **11**, 1695–1699. Web of Science CrossRef CAS PubMed Google Scholar

Romeo, L., Roddaro, S., Pitanti, A., Ercolani, D., Sorba, L. & Beltram, F. (2012). *Nano Lett.* **12**, 4490–4494. Web of Science CrossRef CAS PubMed Google Scholar

Schropp, A., Boye, P., Feldkamp, J. M., Hoppe, R., Patommel, J., Samberg, D., Stephan, S., Giewekemeyer, K., Wilke, R. N., Salditt, T., Gulden, J., Mancuso, A. P., Vartanyants, I. A., Weckert, E., Schoder, S., Burghammer, M. & Schroer, C. G. (2010). *Appl. Phys. Lett.* **96**, 091102. Web of Science CrossRef Google Scholar

Spirkoska, D., Arbiol, J., Gustafsson, A., Conesa-Boj, S., Glas, F., Zardo, I., Heigoldt, M., Gass, M. H., Bleloch, A. L., Estrade, S., Kaniber, M., Rossler, J., Peiro, F., Morante, J. R., Abstreiter, G., Samuelson, L. & Fontcuberta i Morral, A. (2009). *Phys. Rev. B*, **80**, 245325. Web of Science CrossRef Google Scholar

Wallentin, J., Persson, J. M., Wagner, J. B., Samuelson, L., Deppert, K. & Borgström, M. T. (2010). *Nano Lett.* **10**, 974–979. Web of Science CrossRef CAS PubMed Google Scholar

This is an open-access article distributed under the terms of the Creative Commons Attribution (CC-BY) Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are cited.