X-Ray Diffraction from Periodically Patterned GaAs Nanorods Grown onto GaAsB
- First Online:
- Cite this article as:
- Davydok, A., Biermanns, A., Pietsch, U. et al. Metall and Mat Trans A (2010) 41: 1191. doi:10.1007/s11661-009-9868-3
- 509 Views
We present a high-resolution X-ray diffraction pattern of periodic GaAs nanorod (NR) ensembles and individual GaAs NRs grown catalyst-free throughout a prepatterned amorphous SiNx mask onto GaAsB surfaces. The experiments were performed at a home laboratory using synchrotron radiation in combination with a micron-sized beam prepared by compound refractive lenses. The structural properties were probed by measuring RSMs (qx, qz) in the vicinity of GaAs(111) and (222) reflections. Besides the GaAs substrate peak, we found a second peak referring to NRs with lattice mismatch of 0.23 pct with respect to the substrate, probably caused by structural defects. The lateral periodicity of NRs was probed by qx scans, and the NR height obtained from the width of the diffraction curve along qz. Grazing-incidence in-plane diffraction revealed the appearance of small crystallites of cubic γ-Si3N4 caused by recrystallization of SiNx during NR growth. Whereas measurements at the home diffractometer provided average structure parameters, the micron-sized X-ray beam experiment was used to probe the parameters at individual NRs.
Semiconductor nanorods (NRs) grown by vapor-liquid-solid (VLS) epitaxy are of particular interest for the creation of materials with new electronic and optical properties. Exploiting quantum effects, one is able to tune the emission wavelength and to induce one-dimensional (1-D) electronic transport. In the VLS mode, NRs are grown onto  planes of a zinc-blend or diamond-type semiconductors by solution from a molten eutectic alloy formed by a metallic seed, with the diameter and position of grown NRs crucially depending on the statistical process of the metallic droplet formation. The spatial position and the diameter of the molten seeds (typically, Au) determine the position and size of the grown NRs onto the substrate. Due to the statistical character of both quantities, the NRs are nonuniform and rather randomly located at the substrate.
One route to prepare NRs of uniform diameter and defined inter-rod spacing is the use of prepatterned substrates. This can be performed via selective area–metal organic vapor phase epitaxy (SA-MOVPE), where the NRs are grown from small circular openings defined by electron-beam lithography and wet chemical etching of a thin SiNx layer. Using opening diameters in the range of a few hundred nanometers and inter-rod distances in the range of several microns, uniformly sized NRs were grown.
One key problem for the understanding of NR growth in such arrays is the mutual interaction of growing NRs with the SiNx mask, which can result in alloy formation or recrystallization of the initially amorphous SiNx.
Structure characterization of individual NRs is typically performed by high-resolution transmission electron microscopy at single NRs. In addition, X-ray diffraction has been used to obtain structural information from a statistical ensemble of nonuniform NRs.[6,7] Nowadays, the achievements in X-ray optics to define an intense coherent X-ray beam with submicron diameter allows for three-dimensional (3-D) characterization of individual nano-objects using coherent diffraction imaging (CDI).[7–9] However, strain analysis by CDI of individual nano-object is still a challenging task.
The aim of this work was to measure the average and individual properties of GaAs NRs grown in periodic arrays by catalyst-free SA-MOVPE throughout a prepatterned SiNx mask onto GaAsB surfaces.
The samples were characterized using coplanar high-resolution X-ray diffraction in a home laboratory and by coplanar X-ray diffraction at the European Synchrotron Radiation Facility (ESRF) using the microfocus setup at the ID1 beamline. The latter method can provide spatially resolved information about the shape of individual NRs within the NR pattern.
2 Sample Preparation
The structures have been grown onto B oriented GaAs substrate covered by a 15-nm-thick amorphous SiNx layer. The silicon nitride has been deposited by plasma-enhanced chemical vapor deposition at 300 °C. Within an area of 250 × 250 μm2, the SiNx layer was partially removed by electron beam lithography, defining an ordered array of circular openings with diameters of 600 nm in an electron-sensitive resist followed by wet chemical etching using NH4F:HF:H2O solution. The selective-area growth was carried out using low-pressure (50 mbar) MOVPE in an AIXTRON AIX200 reactor (Aachen, Germany). The total flow into the reactor amounted to 7 slm. Thrimethylgallium (TMGa = 3.75 mL) and arsine (AsH3 = 50 mL) were used as group-III and group-V materials. The growth temperature was set to 750 °C, providing equally hexagonally shaped NRs with uniform spacing.
3 X-ray Characterization of the Entire NR array
4 Characterization of the whole NR array using synchrotron radiation
The experiments have been performed at beamlines ID1 and ID10b of ESRF. The high intensity and the small divergence of the incidence beam motivated measurements of the whole NR array using the method of X-ray grazing incidence diffraction (GID). Here, the incident beam strikes the samples under a shallow angle αi with respect to the sample surface, and the diffracted beam is measured in a direction perpendicular to the surface normal of the sample. Subsequently, GID measures lattice planes perpendicular to the NR main axis. One of these lattice planes is (4-2-2) using zinc-blend notation. Exploiting the effect of total external reflection of the X-ray beam at the air-sample interface, the penetration depth can be extremely reduced to a few nanometers in the case where αi is smaller then the critical angle αc of total external reflection. For the wavelength λ = 0.12399 nm, αc is 0.3 deg. For the purpose of detection, we used a 1-D position sensitive detector (PSD) aligned parallel to the surface normal.
These reflections appear because of randomly oriented crystallites grown onto the amorphous SiNx layer in between the perfectly aligned NRs. In addition, the peak at q|| = 15.13 nm−1 can be indexed by the crystalline γ-Si3N4 (1-11) reflection. Unfortunately, no further peaks are visible because of the high instrumental background.
In order to clarify the question of whether peak 2 in Figure 2 could be originated from γ-Si3N4 as well, we recorded the same in-plane line scan after chemical removal of amorphous SiNx using a solution of NH4F (top curve in Figure 3). Again, one can see the (4-2-2) reflection, but no further GaAs peaks. These peaks have been removed by the etching process. However, this scan has been performed using a different experimental setup, providing much higher intensity and lower noise compared to the first experiment. Because of this, the structure of the background can be resolved, providing several peaks that can be identified with crystalline γ-Si3N4, in addition to the (1-11). They reveal a lattice parameter, which is slightly compressed by about 3 pct compared to the literature value.
5 X-ray Diffraction from Individual NRs
Nowadays, the ID1 beamline at ESRF provides the possibility of using a microfocus setup for diffraction measurements defined by a set of compound refractive lenses, focusing the incident beam to a spot size of about 500 × 1500 nm2 in the present case. Using coplanar-diffraction geometry, the beam spot onto the sample equals roughly the inter-rod spacing of sample 1, but it allows for single NR measurement in the case of sample 2. Using this setup, we recorded RSMs of individual NRs close to (111) and (222) Bragg reflection of GaAs using a beam energy of 10 keV and a 1-D position sensitive detector.
One has to note that the crossed lines and the perpendicular line in Figures 5(b) and (c) are artifacts of the experiment (monochromator and PSD streaks) For further explanation, see. Also, the additional two oblique lines on the left are artifacts of the experimental setup.
We analyzed GaAs NRs grown on GaAsB by high-resolution X-ray diffraction. Using coplanar diffraction in a home laboratory, we measured the average structure parameters of the whole NR array. The microfocus setup at ESRF was used to probe the structure parameters of individual NRs.
In both experiments, we have seen a second Bragg peak close to the GaAs one. Using microfocus, we could verify that this peak appears only at positions of GaAs NRs. Therefore, peak 2 was associated with scattering from NRs. The measured lattice expansion with respect to the substrate could be explained by inclusions of atoms with larger covalent radius compared to Ga and As. However, after SEM and EDX inspection, we can exclude any contamination of another element with concentration larger then 0.5 pct. Thus, the measured lattice mismatch must originate from the influence of structural defects such as decorated stacking faults, for example. Indeed, qz scans taken from the RSM of individual NRs revealed different lattice mismatches, which may refer to different numbers of such defects. Unfortunately, for this study, we were not able to probe individual NRs by use of asymmetric diffractions, which are necessary for detailed phase analysis.
One should note that the appearance of the lattice mismatch was very advantageous for our experiment, because low intense structural features could be identified outside the high-intense substrate diffraction.
Grazing-incidence diffraction at the whole NR ensemble revealed peaks of crystalline γ-Si3N4, which might be grown also during the growth process. Obviously, they cannot be removed by the etching process, which is sensitive to amorphous SiNx. These Si3N4 crystals are grown with random orientation with respect to the sample surface. Using this suggestion, peak 2 in Figures 2 and 5 could be indexed by the (200) Bragg reflection of γ-Si3N4 as well, but has to show nearly constant intensity and changing qx. Because the peak is sharp, this suggestion can be ruled out.
The crystallization of Si3N4 crystallites probably takes place at the growth temperature of NRs or at the recooling process after the growth. From the FWHM of peaks related to Si3N4 shown in Figure 4, the crystallite size can be estimated to be 32 nm, which is much smaller than the NR diameter. One may suggest that these crystallites grow preferentially at the interface between the growing GaAs NR and the amorphous SiNx layer to release the interface strain. The appearance of strain is visible by satellite peaks close to the substrate, which is originated by scattering from drilled holes within the SiNx mask, as shown in Figure 2(a). This explains why these satellites disappear after etching the SiNx layer (Figure 2(b)).
Finally, one has to note that our nanofocus experiment was not ideal because we could not measure an entire NR. This is caused by the limited resolution, which did not fit the size of the object, but also by the fact that we did not tune the beam spot to the center of an individual NR. Using a nanofocus diameter similar to the site of nano-objects, one can characterize the whole 3-D shape of individual NRs by CDI.[16,17] Our sample system promises revelation of the details of the different shapes and defect structures of heteroepitaxial NRs.
The authors thank Ana Diaz and T.H. Metzger, ID1 beamline at ESRF, for support.