shows the growth parameters of two InN samples fabricated using RF-MOMBE. Each InN sample was grown on a 4-μm-thick GaN buffer layer. The InN nanorods were grown under N-rich condition, and the InN films were close to stoichiometry confirmed by TEM-EDX analysis.
a shows the XRD patterns of a high-quality GaN template and the diffraction peaks with FWHM (0002) of approximately 250 arc sec. Figure
b exhibits the typical θ-2θ XRD profiles of InN films/nanorods grown on GaN template. The XRD patterns clearly reveal several strong diffraction peaks corresponding to the (0002) of InN, (0004) of InN, (0002) of GaN, and (0004) of GaN. These results indicate that high-quality InN films/nanorods with a hexagonal structure were preferentially oriented and grown on GaN template. In addition, the lattice parameters of InN film of the a-
-axis were calculated as 3.55 and 5.71 Å, respectively. Previous literature reported that the a
-axis of InN was endured and slightly strained
]. On the contrary, our InN films had nearly relaxed on both c
- and a
-axis directions. The observed InN(0002) and (
) diffraction peaks, which are close to InN stoichiometry, suggest that the strain of InN film had nearly relaxed after growth. These values coincide with the stress-free lattice parameters of InN
XRD patterns of GaN and InN materials. (a) ω-scan XRD profile of (0002) GaN by MOCVD; (b) XRD patterns of InN films/nanorods deposited on GaN template.
It has been qualitatively determined that the X-ray rocking curve for the symmetric (0002)-reflecting plane is related to the screw and mixed dislocations; whereas the X-ray rocking curve for the asymmetric (
)-reflecting plane is directly influenced by all threading dislocations. Figure
a shows both InN(0002) and InN(
) peaks for InN films and InN nanorods, respectively. For all samples, the FWHM of InN(
) peak has a significantly greater value than that of InN(0002) peaks. InN films in particular have a minimum FWHM value of 680 and 1,200 arc sec for the symmetrical (0002) and asymmetrical (
) diffraction peaks which are due to the low-defect density, suggesting that the strain of InN relaxes considerably following its growth. To determine the epitaxial relationship between InN and the underlying GaN epilayers, we adopted the azimuthally Φ scan of the InN(10-12) diffraction peak. The in-plane orientations of the InN(10-12) are presented in Figure
b. The hexagonal structure of InN(10-12) produced six equally spaced peaks. Besides, the absence of any other random peaks suggests that the InN film grains predominantly grow in the direction of . Based on the results of XRD analysis, we concluded that InN film is monocrystalline, with a normal surface on  and in-plane orientation of [
. These results confirm in agreement with the results of selected area electron diffraction patterns of TEM. Therefore, these results indicate that high-quality InN films/nanorods with a hexagonal structure were heteroepitaxially grown on GaN template.
XRC and Phi-scan patterns in XRD of InN films and InN nanorods. (a) FWHM values of high-resolution X-ray diffraction InN(0002) ω-rocking curves of InN films/nanorods and (b) InN(1012) phi-scan of InN films/nanorods.
shows a cross section and plane-view FE-SEM images of InN films/nanorods deposited on GaN template. Figure
(a) clearly illustrates that the thickness of the film was approximately 1.7 μm with a growth rate of approximately 0.85 μm/h under this condition. Though the growth rate is still lower than that of InN grown by MOCVD, it is faster than the conventional PA-MBE with a growth rate of about 0.6 μm/h by RF-MBE
]. In addition, the surface of the film was continuous and uniform, illustrating a two-dimensional mode growth. The nanorods formed a cone-shaped columnar structure with separated InN columns, as shown in the Figure
(b). According to earlier literature
], the growth condition of N-rich regime can suppress the formation of the indium droplet; and if no indium droplet appears, we speculate that the InN nanorods were grown by means of a catalytic-free growth mechanism. The average height of the columns was 1.5 μm, and the columns were aligned in the  direction. Furthermore, it is noted that no droplet was observed at the end of any nanorod. Although no metallic particles were observed on the top, we cannot completely rule out the possibility of a VLS-like mechanism because In could desorb at high temperatures. The catalytic growth has been widely employed to grow 1D InN nanostructures via metals, like Au and Ni, which are used as the primary catalysts
]. However, the unwanted metals might limit the potential applications of 1D InN nanostructures. Moreover, their investigation on InN nanorods growth revealed a radial random orientation from the substrates as well. In our catalytic-free growth of InN nanorods, the nanorods were on c
-axis and well-aligned on the GaN/sapphire substrate.
Figure 3 SEM cross-sectional images of InN film/nanorods by RF-MOMBE in a 4-μm thick GaN template. (a) The thickness of the film was approximately 1.7 μm with a growth rate of approximately 0.85 μm/h (b) Nanorods formed a cone-shaped columnar (more ...)
is a bright-field TEM image of a cross section along the zone axes of InN
, with a corresponding selected-area diffraction (SAD) pattern of the InN film deposited on GaN template at 500°C. The thickness of the InN film was approximately 1.7 μm, clearly revealing the threading dislocations. The dislocation density is roughly estimated in the order magnitude of 2
(b) SAED pattern in the InN film and the diffraction pattern of hexagonal wurtzite structure with an incident beam direction of
can be clearly observed. No additional diffraction spots were observed in the pattern, implying that no interlayer existed between InN and GaN. Figure
(c) is a high-resolution TEM image of a cross-section of InN along the
direction. The (0002) lattice fringes reveal that the lattice constant of InN epilayer grown at 500°C was approximately 0.57 nm, similar to our previous TEM observations of InN on ZnO
]. An image of the lattice in the interfacial region between InN and GaN in
and corresponding fast-Fourier-transform patterns are shown in Figure
(d). The (0002) lattice fringes reveal a sharp, smooth interface without the formation of an interlayer, indicating that no reaction occurred between them. In addition, lattice fringes with fast-Fourier-transform patterns in the inset illustrate that InN is in epitaxy with GaN. This pattern indicated that only wurtzite InN and GaN exist at the vicinity of the interface. Therefore, the coherency of InN and GaN layers across the interface is clearly visible.
Figure 4 TEM images of the cross-section of the InN/GaN. (a) Cross-sectional TEM image of InN; (b) SAD pattern of typical InN/GaN interface; (c) image of 1attices of InN; and (d) high-resolution transmission electron microscopy of InN on GaN showing the interface (more ...)
shows the measurement of photoluminescence (PL) spectra at 13 K from InN films/nanorods deposited on GaN layer at 500°C. The PL spectra show that the fundamental band gap of the InN nanorods was located at 0.77 eV with FWHM of approximately 92 meV. The value measured was smaller than band gap of InN film at approximately 0.83 eV. According to the empirical function related to the FWHM of PL
], the carrier concentration in the InN nanorods was roughly estimated to 1.1
, which was slightly smaller than when getting from the Hall measurement. The difference may be due to the surface accumulation in the nanorods. The Hall measurements of the InN nanorods exhibited a carrier concentration of 3
and electron mobility of 253 cm2
/Vs. The reduced number of defects and low background carrier concentration were the results of the high-quality InN. However, the PL emissions in InN nanorods can be attributed to the diameter of the rods, which would increase the density of crystal defects induced by the growth of mismatched lattice. Chao et al.
characterized the PL peaks of InN nanorods as governed primarily by the quantum size effect
]. The emission peak and FWHM of photoluminescence of our InN nanorods were smaller than those reported in
]. The result implies that the surface accumulation in our InN nanorods has a minor influence.
PL spectra of epitaxial InN films/nanorods grown on GaN template at 13 K.