AlGaAs Core Multishell

Jun 26, 2013 - Juan Arturo Alanis , Dhruv Saxena , Sudha Mokkapati , Nian Jiang , Kun Peng , Xiaoyan Tang , Lan Fu , Hark Hoe Tan , Chennupati Jagadis...
5 downloads 9 Views 3MB Size
Download PDF
Letter pubs.acs.org/NanoLett

Polarity-Driven 3‑Fold Symmetry of GaAs/AlGaAs Core Multishell Nanowires Changlin Zheng,† Jennifer Wong-Leung,‡,§ Qiang Gao,‡ Hark Hoe Tan,‡ Chennupati Jagadish,‡ and Joanne Etheridge*,†,∥ †

Monash Centre for Electron Microscopy, Building 81, Monash University, Victoria, 3800, Australia Department of Electronic Materials Engineering, Research School of Physics and Engineering, The Australian National University, Canberra, ACT 0200, Australia § Centre for Advanced Microscopy, The Australian National University, Canberra, ACT 0200, Australia ∥ Department of Materials Engineering, Monash University, Victoria 3800, Australia ‡

ABSTRACT: AlGaAs/GaAs quantum well heterostructures based on core-multishell nanowires exhibit excellent optical properties which are acutely sensitive to structure and morphology. We characterize these heterostructures and observe them to have 3-fold symmetry about the nanowire axis. Using aberration-corrected annular dark field scanning transmission electron microscopy (ADF-STEM), we measure directly the polarity of the crystal structure and correlate this with the shape and facet orientation of the GaAs core, quantum wells and cap, and the width of radial Al-rich bands. We discuss how the underlying polarity of the crystal structure drives the growth of these heterostructures with a 3-fold symmetry resulting in a nonuniform GaAs quantum well tube and AlGaAs shell. These observations suggest that the AlGaAs growth rate is faster along the ⟨112⟩ B compared to the ⟨112⟩ A directions and/or that there is a polarity driven surface reconstruction generating AlGaAs growth fronts inclined to the {110} planes. In contrast, the observations suggest that the opposite is true for the GaAs growth, with the preferred surface reconstruction plane being parallel to {110} and an apparent faster growth rate along the ⟨112⟩ A. This two-dimensional layer growth of the nanowire multishells strongly depends on the surface energies and surface reconstruction of the facets which are related to the crystal polarity and lead to the 3-fold symmetry observed here. KEYWORDS: Polarity, AlGaAs/GaAs, nanowire, ADF-STEM, growth

G

in determining the optical properties, it is critical to understand the growth mechanisms that determine the structure. This is the goal of this paper. AlGaAs/GaAs heterostructures have been studied extensively in the last 30 years, and there are many reports of aluminum inhomogeneity in planar heterostructures6−8 and quantum wires.9 In addition, key studies have shown morphology variation and faceting due to variable growth rates on different atomic planes10−18 making the growth of AlGaAs/GaAs quantum well heterostructures a challenge in both metal organic chemical vapor deposition (MOCVD) and molecular beam epitaxy (MBE) growth. An important goal in the growth of core−shell nanowires is to achieve conformal shell growth around the core. This process is complicated by existing pregrowth conditions, such as facets on the nanowire core in addition to variable growth parameters such as temperature, V/III ratio, and growth rate.

aAs/AlGaAs is an interesting material system for the study of semiconductor heterostructures with minimal strain issues. The small lattice mismatch between AlAs and GaAs allows great flexibility in tuning the band gap energy and the mobility of charge carriers by varying the aluminum compositions without enhanced strain.1,2 This has generated wide applications in electronic and optoelectronic devices, including terahertz lasers and high speed electronics.3,4 Recently, we synthesized a unique AlGaAs/GaAs quantum well heterostructure based on core−multishell nanowires which showed excellent optical properties, including extremely high quantum efficiency and intense emission for extremely low submicrowatt excitation powers.5 These asymmetric quantum wells with nonparallel interfacial planes were grown concentrically about the nanowire axis with a tube-like geometry, thereby imposing boundary conditions on the surface and interface that significantly enhance the quantum size effect. The quantum well “tubes” have a pseudohexagonal shape, with the electrons and holes being confined in their ground states to onedimensional filaments at the corners of the pseudohexagonal tube.5 Given the importance of the structure and morphology © 2013 American Chemical Society

Received: May 8, 2013 Revised: June 26, 2013 Published: June 26, 2013 3742

dx.doi.org/10.1021/nl401680k | Nano Lett. 2013, 13, 3742−3748

Nano Letters

Letter

Figure 1. (a) Schematic of the TEM cross-section of the nanowire prepared by microtoming. (b−e) ADF-STEM images of an AlGaAs/GaAs nanowire cross-section viewed along the ⟨111⟩ direction ([111] or [1̅1̅1̅] incident zone axis cannot be distinguished); (b) low magnification overview (inset: Fourier transform spectrum from a corresponding atomic resolution image); (c) atomic resolution image of the GaAs quantum well and the AlGaAs barrier layers. (d) Higher magnification image of the top region of b and (e) higher magnification image of the left region of b. (f) Schematic showing the tilt direction from ⟨111⟩ toward ⟨110⟩ (around the blue arrow).

nonparallel interfacial planes, (b) aluminum rich bands along the radial ⟨112⟩ directions which alternate in thickness, three thick and three thin bands, and (c) two different sizes of facets on {112} planes. Given the localization of ground states to the quantum well (QW) corners and their sensitivity to QW width and radius,5 it is important to understand the mechanisms driving the growth of these “3-fold” features. The key questions we address in this paper are what is the polarity of the crystal with respect to these geometrical features and what, if any, role does it play in their growth? In most previous studies, the polarity of the nanowires has not been determined directly but has been assumed to be inherited from the substrate, the polarity of which is usually known. This approach has several limitations. First, it can only be used if the TEM specimen is examined along the length of the nanowire (i.e. in plan view) where the growth direction can be identified from the location of the Au eutectic. However, plan view specimens do not enable the characterization of key structures, such as the QW shell, so it is not possible to correlate polarity with these radial structures. Second, substrate polarity is not necessarily a reliable predictor of nanowire polarity. During growth, the generation of twin defects or kinks on {111} surfaces can change the polarity by a 180 degree rotation.18,23 Finally, this approach cannot easily be applied to nanowires grown on nonpolar substrates like Si, where the relationship between substrate and polarity is more tenuous. If we are to be certain of the polarity and of its relationship to the important QW heterostructures, we must apply a method that enables the direct and unambiguous measurement of polarity at high spatial resolution, simultaneously with the characterization of the QW heterostructures. Polarity can be determined at low spatial resolution from methods such as chemical or thermal etching, X-ray diffraction, and ion channeling and with unit cell resolution from the asymmetric features in convergent beam electron diffraction (CBED) patterns which arise due to strong dynamical

Details about both the crystal morphology and structure of the heterostructures cannot be accurately determined from thermodynamic simulation. Previous investigations18 suggest that the growth of zincblende structure compound semiconductor nanowires may be affected by polarity, which originates from the noncentrosymmetric atomic structure and hence electrostatic potential distributions within the unit cell. The crystal surfaces, {hkl} and {hkl}, perpendicular to the polarity axis are terminated with a different type of atom (metallic or metalloid elements) and have different chemical and electronic characteristics.19 Typically, the As terminated {111} surfaces are known as (111)B or (1̅1̅1̅) surfaces, and the Ga terminated {111} surfaces are known as (111)A or (111) surfaces. In our earlier work,5 we investigated cross-sectional samples of the GaAs/AlGaAs multishell heterostructure nanowires by scanning transmission electron microscopy (STEM). The GaAs core showed a hexagonal cross-section with {110} facets whereas several geometrical features in the radial shell growth exhibit a 3-fold rotational symmetry about the nanowire growth axis ⟨111⟩ (including Al-rich bands along the ⟨112⟩ direction).5 3-fold rotational symmetries are also evident in three other recent papers on GaAs/AlGaAs nanowires, although the 3-fold symmetry was not explicitly commented on. Tambe et al.20 observed anisotropy in the shape of the shell of MOCVD grown nanowires, however, the image resolution was insufficient to identify the nature of the anisotropy. Very recently in MBE grown nanowires, Heiss et al.21 reported quantum dots located in three of six radial Al-rich bands and in Figure 2 of Rudolph et al.,22 it is just evident that the radial Alrich bands alternate in thickness resulting in a 3-fold symmetry (although the authors report a 6-fold symmetry). Here we investigate the structures with 3-fold symmetry in more detail and determine their relationship to the underlying crystal polarity which we measure. These “3-fold” features include (a) the slight wedge-shape of the GaAs quantum wells due to 3743

dx.doi.org/10.1021/nl401680k | Nano Lett. 2013, 13, 3742−3748

Nano Letters

Letter

Figure 2. EDX analysis of the cross-sectional AlGaAs/GaAs nanowire. (a) Two-dimensional EDX elemental mapping of As (blue), Ga (yellow), and Al (red) together with the synchronously recorded ADF-STEM signal. (b) EDX line profile across the AlGaAs barrier layer, the GaAs quantum well, and the GaAs cap layer. (c) EDX line profile across a thin dark band confirms it is rich in Al.

scattering.24 However, CBED can be particularly timeconsuming when it is applied to multiple compound semiconductor nanowires with varying morphologies and where both the composition and the thickness of the sample are unknown. High-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) has previously been used to identify polarity in bulk and thin film specimens25,26 and more recently in plan view nanowires.27,28 Here we use aberration corrected ADF-STEM to image directly the polarity of AlGaAs/GaAs nanowire cross sections at the atomic scale and correlate this directly with the local morphology. The intensity in atomic resolution ADF-STEM images is very sensitive to the atomic number of the atom being imaged, making it possible to distinguish the lighter Ga column from the heavier As column in electron microscopes with sufficient resolution.28−30 Experimental Methods. GaAs core nanowires were grown on (111)B GaAs substrates by MOCVD using 50 nm Au nanoparticles as the catalyst. A two-temperature growth method31 with a nucleation temperature of 450 °C and growth temperature of 375 °C was used, and this growth process has been shown to result in a defect-free zinc blende GaAs nanowires.31 Then, a 90 nm thick AlGaAs shell was grown with a ratio of Al to Ga in the vapor phase of 60:40 at a higher temperature of 650 °C followed by a thin GaAs quantum well sandwiched with a second AlGaAs barrier layer of the same composition. The whole structure was terminated with a thin GaAs cap layer to prevent oxidation of the AlGaAs barrier. Cross-sectional TEM samples were prepared by embedding the nanowires in resin and microtoming into thin slices perpendicular to the growth direction (see Figure 1a). The approximate thickness of the slice, as determined by convergent beam electron diffraction (CBED), is ∼50 nm. The STEM investigations were performed on an FEI Company Titan3 80− 300 field emission gun transmission electron microscope fitted with aberration correctors (CEOS GmbH) on both the probe and image forming lens systems. The microscope was operated at 300 kV and a Fischione Instruments 3000 annular dark field detector was used to record ADF-STEM images. A focused electron beam, ∼1 Å fwhm (full width at half maximum) in diameter, was raster-scanned over the sample and high-angle scattered electrons are synchronously collected and integrated by the annular dark field detector located in the far field diffraction plane. The intensity of the resulting ADF-STEM image can be described phenomenologically as being propor-

tional to Zn where Z is the atomic number of the constituent atoms and n depends strongly on the experimental conditions32 but is typically between one and two. Hence the polarity can be determined from the asymmetry in the image intensity profile across the Ga−As “dumbbells”. To acquire atomic resolution, Z-sensitive STEM images, the probe semi angle was set to 18 mrad, and the effective collection angle of the ADF detector was 46−200 mrad. An EDAX 30 mm2 retractable Si(Li) X-ray detector was utilized to collect the EDX spectrum and spectrum imaging in STEM mode. To enhance the X-ray collection efficiency, the nanowire was tilted ∼16 degrees toward the EDX detector. Results. The microtomed TEM cross-section generates a thin slice with orientation as illustrated schematically in Figure 1a. A low-magnification ADF STEM image (Figure 1b) recorded along the zone axis [1̅1̅1̅] or [1 1 1] which is parallel or antiparallel, respectively, to the axial growth direction shows a typical overview of the AlGaAs/GaAs nanowire. From this overview, it is evident the cross-section of the nanowire is bounded by six main facets together with another six small facets in between them. By comparing the Fourier transform spectrum (inset in Figure 1b) of a high-resolution image with the low magnification image in Figure 1b, it can be determined that the six main facets are approximately parallel to the {110} planes while the six small facets are approximately parallel to the {112} planes. The GaAs core facets lie parallel to the {110} planes. The overview of the morphology shows an approximately hexagonal cross-section which contains a central core and two other concentric layers, which all have higher image intensity. Given the dependence of ADF-STEM image intensity on atomic number, these three high intensity regions can be identified as the GaAs core (∼50 nm), the thin GaAs quantum well (∼4−8 nm) and a thick GaAs cap layer (∼7−13 nm), respectively. An atomic resolution image of the GaAs quantum well is shown in Figure 1c. The remaining relatively lower intensity regions are identified as the AlGaAs barrier layers, since the atomic number of Al is smaller than Ga. This assignment was confirmed from energy dispersive X-ray (EDX) maps, as discussed below. In addition, within the AlGaAs layers, there are six radial bands with lower intensity, consistent with a higher Al content, located along the diagonal of the hexagonal structure (perpendicular to the {112} facets). This was also confirmed by EDX maps. These Al-rich bands separate the nanowire into six subsections. 3744

dx.doi.org/10.1021/nl401680k | Nano Lett. 2013, 13, 3742−3748

Nano Letters

Letter

sizes: “large” and “small”. (The small facets are so small they give the impression that the corner is “rounded”. See an enlarged view in Figure 1d.) It is plausible that the features showing a 3-fold symmetry are linked to the polarity of the {112} planes which are composed of {112}A and {112}B planes, both having a 3-fold symmetry.18 However, the relationship between the polarity and these features is not clear. Neither the possible incident beam directions [111] or [1̅1̅1̅] (two opposite polar directions), nor the polarity of the six {112} facets can be determined from this orientation. As explained in the schematic in Figure 1f, the Ga atoms and As atoms are projected along the same atomic columns. In order to determine the polarity, the crystal must be tilted to different zone axes. To distinguish the polarity direction, the nanowire was tilted 35.5 degrees anticlockwise from the original position to the ⟨110⟩ direction. The tilting axis is indicated by the blue arrow in Figure 1f. The corresponding ADF-STEM image after tilting is shown in Figure 3a. Only two GaAs QWs and two Al-rich bands perpendicular to the tilting axis are visible in this geometry. A closer view of the bottom left section of the nanowire is presented in Figure 3b. The corresponding Fourier transform spectrum (inset) confirms that the nanowire is oriented along the ⟨110⟩ incident direction. The enlarged atomic resolution image of the GaAs QW and AlGaAs barrier is displayed in Figure 3c, where the dumbbell structure in the [110] zone axis can now be resolved (the distance between the (Al,Ga) and the As atomic columns in the dumbbell structure is 1.4 Å). The whole cross-sectional sample after tilting is illustrated schematically in Figure 3d. A portion of the AlGaAs layer is further enlarged and displayed in Figure 3e. An intensity line profile along the path from S to E is plotted in Figure 3g. As revealed in the line profile, the individual columns in the dumbbell structure are resolved, and there is an intensity difference between the pairs of group III and group V atom columns. Since the atomic number of Al and Ga are both smaller than As, the lower and higher intensity peaks can be assigned to the Ga(Al) atomic columns and the As atomic columns, respectively. The corresponding atomic arrangement is illustrated schematically in the structure model of Figure 3f. By combining the results from these two different incident zone axes, and considering their relative orientations, we can determine uniquely that the incident zone axis in Figure 1 is [111] rather than [1̅1̅1̅], (otherwise the positions of the (Ga, Al) columns and the As columns in the dumbbell structure would be swapped in the second incident zone axis). Similarly, the polarity of the six {112} facets in Figure 1 can also be determined uniquely, and we find the three small “rounded” corner facets of the nanowires correspond to {112}A planes and the large {112} corner facets corresponds to {112}B planes. Figure 4a shows a schematic summarizing the assignment of polarity and absolute facet orientation. These results establish the relationship between the three types of features with 3-fold symmetry and the polarity of the sample. The difference between (112)A and (112)B as determined from this analysis is illustrated in Figure 4b. Discussion. Our results clearly show that polarity plays an important role in the asymmetry of radial growth of GaAs/ AlGaAs nanowires. We discuss here the mechanisms by which polarity determines the geometry of the QW, the geometry of the facets and the Al-rich radial bands and their corresponding 3-fold symmetry.

The chemical composition of the nanowires was characterized by EDX analysis. Figure 2a shows a two-dimensional EDX elemental map (extracted from EDX spectrum taken at each position as the electron probe was scanned across the specimen) together with the simultaneously recorded ADFSTEM image. The elemental distributions are consistent with the STEM results described above. A more quantitative analysis of the EDX spectrum, using the GaAs core to calibrate the Kfactors, shows that the average Al concentration in the AlGaAs barrier layers is 0.5 ± 0.1. Figure 2b shows a one-dimensional EDX line profile across the nanowire along the white arrow superimposed on a micrograph showing the GaAs quantum well, the AlGaAs barrier layer, and the GaAs cap. As expected, the Al atoms only replace the Ga atoms in the AlGaAs barrier. The Al signal does not drop completely to zero across the GaAs quantum well. This is due to some experimental limitations, including the low resolution sampling points (1.9 nm between two neighboring data points) and the larger probe size required to increase the beam current (in this Schottky emitter TEM with low angle EDX detector), and hence resulting in a lower spatial resolution of the EDX spectrum. An EDX line profile across the low intensity radial bands in the ADF STEM image is shown in Figure 2c and confirms that it has a higher Al concentration than any other part of the cross-section. Several structural features exhibit a 3-fold rotational symmetry about the nanowire axis. Consider first the wedgeshape of the GaAs quantum well (and similarly of the GaAs capping layer). The GaAs quantum wells are roughly parallel to the {110} planes but all show a narrowing in their thickness from one corner of the hexagon to another. This results in two types of corners, a “thick” corner of about 8 nm width and a “thin” corner of about 4 nm width (see the dotted circles in Figure 1d). These corners have been identified as the emission centers from microphotoluminescence (PL) spectra.5 The two interfacial planes of the GaAs QWs are not parallel to each other. The inner AlGaAs/GaAs interface (closer to the nanowire core) is ∼1° off the {110} plane, while the outer GaAs/AlGaAs interface (closer to the cap layer) is exactly parallel to the {110} plane, the thickness of the quantum wells narrows from one corner to the next (Figure 1e), giving a slightly wedged shape. The GaAs cap layer also shows a similar slight wedge shape with the inner AlGaAs/GaAs interface ∼1° off the {110} plane while the outside surface is exactly parallel to the {110} plane. From the overview (Figure 1b), the wedge shapes of the six GaAs QW segments are arranged with a 3-fold symmetry. The second geometric feature with a 3-fold symmetry about the nanowire axis is the Al-rich bands in the AlGaAs layers. The six radial Al-rich bands can be divided into two groups according to their thickness, the thin Al-rich bands and the thick Al-rich bands (as indicated by the arrows in Figure 1d). The thin Al-rich bands only intersect with the thick corners of the GaAs quantum well, while the thick bands only intersect with the thin corners of the QW. The net direction of the Alrich bands is along the ⟨112⟩ direction but close inspection of atomic resolution images suggest the bands comprise a zigzag of short sections parallel to ⟨110⟩ directions, with the zigzag following the ⟨112⟩ direction. The third feature with a 3-fold symmetry is the “corner” of the GaAs cap layers. These GaAs layers are terminated with major facets parallel to the {110} planes, as observed for the GaAs cores. These major facets meet at the corner “facets” which are parallel to {112} planes and come in two alternating 3745

dx.doi.org/10.1021/nl401680k | Nano Lett. 2013, 13, 3742−3748

Nano Letters

Letter

mechanism of the shells is quite different. When the growth temperature is increased to 650 °C, the decomposition rate of the precursors is significantly improved, the growth species will deposit directly on the side walls via the two-dimensional (2D) layer growth mode. This results in fewer adatoms diffusing into the Au particle, and axial growth is then significantly reduced. In this case, the final morphology of the nanowires after the shell growth mainly depends on the relative growth rates of the different facets of the sidewalls. The general rule is that the slowest growing surface will appear as the largest developed facets.33 Zinc blende GaAs nanowires typically have {112} sidewall facets18 which are somehow favored in the VLS growth and related to the energy implications associated with the triple phase boundary. The {110} facets of the GaAs cores in the present core-multishell nanowires may be attributed to the fact that the {112} planes have higher surface energies compared to {110}. The ramping up of the temperature for the twodimensional growth must enhance surface diffusion such that the cores are now faceted on the surfaces with the lowest surface energies which are the {110} planes. According to Chadi et al.,34 the {112} planes are expected to have surface energies of at least three times the surface energy of {110} planes, in agreement with the larger {112}B facets of the cap layer observed here. For planar structures,12 {110} planes have also been reported to be the slowest growth planes in agreement with what is observed in the current nanowires. The radial Al-rich bands are intriguing and warrant some discussion. First, the inhomogeneities of Al composition in AlGaAs layers have been reported for planar structures. Petroff et al.6 reported the alloy clustering in AlGaAs layer which was a surprising result, given that thermodynamic calculations for this system do not show a miscibility gap. They reported a surface orientation dependent phase diagram for the AlGaAs system due to difference in the cathodoluminescence emission between (100) and (110) oriented growths. Impurity pinning and surface effects were believed to be the predominant factors. Asom et al.7 proposed that the Al−As bonds are much stronger than Ga−As and that regions of high concentrations of Al−As are nucleating sites for further clustering and hence enhanced growth. Kim et al.9 reported the presence of Ga-rich wells in AlGaAs buffer layers deposited on substrates patterned to exhibit a sawtooth profile on (111) planes. The radial AlGaAs growth on the GaAs core represents a unique opportunity to study the effect of different planes under the same growth conditions, temperature and V−III ratio. Our results show two features with regards to the AlGaAs shell, (1) thicker Al bands at the {112}B facets compared to the {112}A facets and (2) a slightly wedge-shaped AlGaAs shell. The first feature can be attributed to the fact that the {112}B facet on the GaAs core is larger compared to the {112}A facets. The larger {112}B facets compared to {112}A facets have been reported for GaAs35,36 and attributed to a higher growth rate on the {112}A facet compared to the {112}B facet in GaAs. On the other hand, the wedge shape of the AlGaAs shell would seem to suggest that the radial growth rate of the {112}B facets for the AlGaAs shell has a slightly faster growth rate than the {112}A facets, thus making the GaAs core/AlGaAs interface between these two corners about one degree from being parallel to the {110} plane (although we note that the GaAs quantum well and GaAs cap have approximately the same wedge angle, even though the growth time is different). This is initially surprising, however, there is evidence for reversal of the growth rate of these two surfaces in the case of GaP.36

Figure 3. (a−c) ADF-STEM images of AlGaAs/GaAs nanowires viewed along the ⟨110⟩ direction. (a) Overview of the nanowire crosssection after tilting to the ⟨110⟩ zone axis; (b) an enlarged part of the bottom left section of (a) with the corresponding Fourier transform (inset); (c) atomic resolution image of the GaAs quantum well and the AlGaAs barrier layers. (d) Schematic of the nanowire after tilting down the ⟨110⟩ zone axis with atomic structure model inserted. (e) An enlarged part of c showing the AlGaAs barrier layer (close to bottom left corner). The dumbbell structure and its asymmetry can be clearly resolved. (f) Corresponding atomic structure model. (g) Intensity line profile taken from “S to E” of e. The Ga(Al) site and As site can be distinguished from the relative intensity of the adjacent dumbbell peaks.

The core and shells were grown at different temperatures and have very different growth mechanisms. For the core growth, the growth temperature is set at 375 °C, and the decomposition rate of the precursors is low. Growth occurs predominantly through the vapor−liquid−solid (VLS) mechanism with the Au nanoparticle acting as the catalyst. The growth species either directly impinge into the Au nanoparticle or are adsorbed on the sidewall and further diffuse to the Au nanoparticle. The low supersaturation vapor minimizes the direct deposition of the growth species on the sidewalls (the vapor−solid growth) via two-dimensional nucleation. Thus the nanowire growth in the axial direction is much faster than that in the radial direction. Furthermore, the size of the GaAs core is determined by the size of the Au nanoparticle. On the other hand, the growth 3746

dx.doi.org/10.1021/nl401680k | Nano Lett. 2013, 13, 3742−3748

Nano Letters

Letter

Figure 4. Schematic summarizing the facet polarity of the AlGaAs/GaAs nanowire relative to the morphology of the nanowire core, shell, and cap and the Al-rich radial bands. (b) Atomic structure model of zinc-blende GaAs showing the atomic structure at the polar planes.

During the growth of the GaAs quantum well, the step bunched {110} growth fronts merge closer to the {110} planes, and this results in the wedged shape of the quantum well structure. The relative faceting of the {112}B facets in the final structure of this quantum tube arises again as a result of the surface energy of {112}B being more favorable than the {112}A due to the relative surface energies involved. Also a faster radial growth rate is expected along the {112}A for GaAs compared to the {112}B, thus creating a GaAs QW thickness at the {112}A corner compared to the {112}B corner. Our results are also consistent with the exact {110} planes being extremely favorable as the termination plane for the GaAs surfaces such that the {110} surface is reconstructed as the largest facet during GaAs radial growth. In summary, we have observed that the GaAs quantum well shape, the Al-rich radial bands, and the size of the {112} facets all exhibit 3-fold symmetry about the nanowire axis in AlGaAs/ GaAs core−multishell nanowires. We have determined the relationship between the crystal polarity and morphology of the nanowire heterostructures and proposed mechanisms by which polarity drives the growth of these morphologies. One interpretation of our observations is that there is a higher growth rate along the ⟨112⟩B direction compared to ⟨112⟩A direction in the AlGaAs layer and the opposite for the GaAs radial growth. Alternatively or in addition, they suggest a preference for a polarity driven inclined reconstruction of the pseudo-{110} surface in AlGaAs and a perfect reconstruction of the {110} surface in GaAs. The growth of the GaAs quantum well and the overlayer along the radial direction occurs via a two-dimensional layer growth mechanism, unlike the axial growth of nanowires via vapor−liquid−solid (VLS) (the growth of the nanowire core in our sample) or vapor−solid (VS) growth. Thus the growth rate along different facets is influenced by (1) the surface energies and surface reconstruction of the facets and (2) the growth directions which are related to the polarity of the crystal. All of our findings indicate that the growth of the nanowire shells is driven by the crystal polarity leading to a heterostructure morphology with 3-fold symmetry.

Furthermore, the bonding is the same for the Al atoms whether it is incorporated on {110}, {112}A, or {112}B surface growths. In reality, the growth rate is mediated by several growth parameters such as V/III ratio and growth temperature, and such parameters will affect the surface reconstruction and the adatom sticking coefficients which are much more complex.36 One interpretation is that the introduction of Al under the conditions used for the AlGaAs layer changes the growth rates on the different facets with the fastest to slowest growth planes occurring in the following order: {112}B, {112}A, and then much slower on the {110} plane. The Al incorporation is the highest along these {112} directions which have high growth rate, and the thin and thick Al bands are the footprints of preexisting {112}A and {112}B planes which are very narrow polar surfaces compared to the {110} surface, thus explaining the relatively sharp corner on the low-magnification image representation. The reversal of the growth rate between AlGaAs and GaAs has also been observed in the growth of planar structures.37 Indeed, as proposed by Asom et al.,7 the higher Al content is consistent with faster growth rates. To accommodate for the difference between growth rates on {112} B and {112}A, the {110} planes, a slower-growing plane between adjacent {112}A and {112}B planes, will adopt a step bunched aspect during the AlGaAs layer growth, as has been observed for planar growths on vicinal (110) substrates.16 Petroff et al.6 observed that AlGaAs grown on vicinal (110) GaAs substrates were rough when the miscut with a tilt of ±2 degrees about the [110 ] axis created ledges with Ga bonds, in contrast to ledges with As bonds. Clearly, AlGaAs growth exhibits a preference for a given polarity of the ledge even in these planar (110) growths. Moreover, we know that the {112} planes are not atomically straight facets but have microfacets corresponding to {111}B and {001} for {112}B facets.18 Microfacets provide a collection of step edges where enhanced growth is more likely to occur. It is clear from our observations that polarity is the driving force for the asymmetry observed in the AlGaAs layer growth, either due to a higher growth rate for AlGaAs along the ⟨112⟩B direction or to a preferred “pseudo{110}” AlGaAs plane reconstructed with an inclination associated with a preferred ledge polarity. In planar growth on (001) surfaces, the streakiness of AlAs on (001) planes has previously been observed,7 and perhaps such inhomogeneities on the (001) planes are reflected in the zigzag structure observed in these Al bands in atomic resolution imaging along the (112) growth front. However, this is hard to identify from the inclined nature of the (001) surface to the geometry of the cross-sectional sample studied.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel.: +61-3-9905 1836. Fax: +61-3-9905 3600. Author Contributions

Q.G., H.H.T., and C.J. grew the nanowires, and C.L.Z., J.W.-L., and J.E. analyzed the structure and polarity with electron 3747

dx.doi.org/10.1021/nl401680k | Nano Lett. 2013, 13, 3742−3748

Nano Letters

Letter

(22) Rudolph, D.; Funk, S.; Doblinger, M.; Morkotter, S.; Hertenberger, S.; Schweickert, L.; Becker, J.; Matich, S.; Bichler, M.; Spirkoska, D.; Zardo, I.; Finley, J. J.; Abstreiter, G.; Koblmuller, G. Nano Lett. 2013, 13 (4), 1522−1527. (23) Uccelli, E.; Arbiol, J.; Magen, C.; Krogstrup, P.; Russo-Averchi, E.; Heiss, M.; Mugny, G.; Morier-Genoud, F.; Nygard, J.; Morante, J. R.; Fontcuberta i Morral, A. Nano Lett. 2011, 11 (9), 3827−3832. (24) Goodman, P. Space-group determination by convergent-beam electron diffraction. In International Tables for Crystallography Vol. B: Reciprocal Space; International Union of Crystallography: Chester, Englad, 2006; pp 285−306. (25) Pennycook, S. J.; Jesson, D. E.; McGibbon, A. J.; Nellist, P. D. J. Electron. Microsc. 1996, 45 (1), 36−43. (26) Mkhoyan, K.; Batson, P.; Cha, J.; Schaff, W.; Silcox, J. Science 2006, 312 (5778), 1354. (27) Den Hertog, M.; Elouneg-Jamroz, M.; Bellet-Amalric, E.; Bounouar, S.; Bougerol, C.; Andre, R.; Genuist, Y.; Poizat, J. P.; Kheng, K.; Tatarenko, S. 17th International Conference on Microscopy of Semiconducting Materials, Cambridge, U.K., Apr 4−7, 2011; p 326. (28) de la Mata, M.; Magen, C.; Gazquez, J.; Utama, M. I. B.; Heiss, M.; Lopatin, S.; Furtmayr, F.; Fernandez-Rojas, C. J.; Peng, B.; Morante, J. R.; Rurali, R.; Eickhoff, M.; Fontcuberta i Morral, A.; Xiong, Q. H.; Arbiol, J. Nano Lett. 2012, 12 (5), 2579−2586. (29) Nellist, P. D.; Pennycook, S. J. Ultramicroscopy 1999, 78 (1−4), 111−124. (30) Pennycook, S. J.; Chisholm, M. F.; Lupini, A. R.; Varela, M.; van Benthem, K.; Borisevich, A. Y.; Oxley, M. P.; Luo, W.; Pantelides, S. T. Adv. Imag. Elect. Phys. 2008, 153, 327−384. (31) Joyce, H. J.; Gao, Q.; Tan, H. H.; Jagadish, C.; Kim, Y.; Zhang, X.; Guo, Y. N.; Zou, J. Nano Lett. 2007, 7 (4), 921−926. (32) Dwyer, C.; Maunders, C.; Zheng, C. L.; Weyland, M.; Tiemeijer, P. C.; Etheridge, J. Appl. Phys. Lett. 2012, 100 (19), 191915. (33) Berkovitchyellin, Z. J. Am. Chem. Soc. 1985, 107 (26), 8239− 8253. (34) Chadi, D. J. J. Vac. Sci. Technol., B 1985, 3 (4), 1167−1169. (35) Joyce, H. J.; Wong-Leung, J.; Gao, Q.; Tan, H. H.; Jagadish, C. Nano Lett. 2010, 10 (3), 908−915. (36) Verheijen, M. A.; Algra, R. E.; Borgstrom, M. T.; Immink, G.; Sourty, E.; van Enckevort, W. J. P.; Vlieg, E.; Bakkers, E. P. A. M. Nano Lett. 2007, 7 (10), 3051−3055. (37) Biasiol, G.; Gustafsson, A.; Leifer, K.; Kapon, E. Phys. Rev. B 2002, 65 (20), 205306 (1−15).

microscopy. All authors contributed to the discussion and preparation of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank Yanan Guo and Jin Zou of the University of Queensland for help with the preparation of the nanowire cross sections. The electron microscopy was conducted at the Monash Centre for Electron Microscopy using the FEI Titan3 80-300 S/TEM funded by the ARC Grant LE0454166. The Australian Research Council is acknowledged for the financial support, and the authors also acknowledge the use of the Australian National Fabrication Facility at ACT node for this work.

■

REFERENCES

(1) Alferov, Z. I. Semiconductors 1998, 32 (1), 1−14. (2) Schaefer, W.; Wegener, M. Semiconductor optics and transport phenomena; Springer: Berlin, 2002; p xi. (3) Kohler, R.; Tredicucci, A.; Beltram, F.; Beere, H. E.; Linfield, E. H.; Davies, A. G.; Ritchie, D. A.; Iotti, R. C.; Rossi, F. Nature 2002, 417 (6885), 156−159. (4) Harder, C. S.; Vanzeghbroeck, B. J.; Kesler, M. P.; Meier, H. P.; Vettiger, P.; Webb, D. J.; Wolf, P. IBM J. Res. Dev. 1990, 34 (4), 568− 584. (5) Fickenscher, M.; Shi, T.; Jackson, H. E.; Smith, L. M.; YarrisonRice, J. M.; Zheng, C. L.; Miller, P.; Etheridge, J.; Wong, B. M.; Gao, Q.; Deshpande, S.; Tan, H. H.; Jagadish, C. Nano Lett. 2013, 13 (3), 1016−1022. (6) Petroff, P. M.; Cho, A. Y.; Reinhart, F. K.; Gossard, A. C.; Wiegmann, W. Phys. Rev. Lett. 1982, 48 (3), 170−173. (7) Asom, M. T.; Geva, M.; Leibenguth, R. E.; Chu, S. N. G. Appl. Phys. Lett. 1991, 59 (8), 976−978. (8) Sato, M.; Maehashi, K.; Asahi, H.; Hasegawa, S.; Nakashima, H. Superlattice Microstruct. 1990, 7 (4), 279−282. (9) Kim, Y.; Yuan, S.; Leon, R.; Jagadish, C.; Gal, M.; Johnston, M. B.; Phillips, M. R.; Kalceff, M. A. S.; Zou, J.; Cockayne, D. J. H. J. Appl. Phys. 1996, 80 (9), 5014−5020. (10) Jacobi, K.; Platen, J.; Setzer, C. Phys. Status Solidi B 2000, 218 (2), 329−364. (11) Notzel, R.; Daweritz, L.; Ploog, K. Phys. Rev. B 1992, 46 (8), 4736−4743. (12) Demeester, P.; Vandaele, P.; Baets, R. J. Appl. Phys. 1988, 63 (7), 2284−2290. (13) Hersee, S. D.; Barbier, E.; Blondeau, R. J. Cryst. Growth 1986, 77 (1−3), 310−320. (14) Hasegawa, S.; Sato, M.; Maehashi, K.; Asahi, H.; Nakashima, H. J. Cryst. Growth 1991, 111 (1−4), 371−375. (15) Hasegawa, S.; Kimura, K.; Sato, M.; Maehashi, K.; Nakashima, H. Surf. Sci. 1992, 267 (1−3), 5−7. (16) Hasegawa, S.; Arakawa, K.; Oooka, H.; Nakashima, H. Appl. Surf. Sci. 2000, 162, 430−434. (17) Krishnamurthy, M.; Wassermeier, M.; Williams, D. R. M.; Petroff, P. M. Appl. Phys. Lett. 1993, 62 (16), 1922−1924. (18) Zou, J.; Paladugu, M.; Wang, H.; Auchterlonie, G. J.; Guo, Y. N.; Kim, Y.; Gao, Q.; Joyce, H. J.; Tan, H. H.; Jagadish, C. Small 2007, 3 (3), 389−393. (19) Iwanaga, H.; Fujii, M.; Takeuchi, S. Phase Transit. 1998, 66 (1− 4), 147−165. (20) Tambe, M. J.; Allard, L. F.; Gradecak, S. J. Phys. Conf. Ser. 2010, 209, 012033. (21) Heiss, M.; Fontana, Y.; Gustafsson, A.; Wust, G.; Magen, C.; O’Regan, D. D.; Luo, J. W.; Ketterer, B.; Conesa-Boj, S.; Kuhlmann, A. V.; Houel, J.; Russo-Averchi, E.; Morante, J. R.; Cantoni, M.; Marzari, N.; Arbiol, J.; Zunger, A.; Warburton, R. J.; Fontcuberta i Morral, A. Nat. Mater. 2013, 12 (5), 439−444. 3748

dx.doi.org/10.1021/nl401680k | Nano Lett. 2013, 13, 3742−3748