III-nitride compounds have attracted considerable interest in the last decade due to their wide application in solid-state-lighting devices, which are used widely in the blue/green and ultraviolet spectral regions with higher quantum efficiency [1-3]. AlGaN alloys are important for optoelectronic devices such as light-emitting diodes (LEDs), laser diodes (LDs) and photo-detectors in the UV spectral region between the wavelength range from 200 nm to 365 nm, which can be tuned according to Al content in the AlGaN alloy system . Owing to difficulty in fabricating bulk GaN and AlN, nearly all AlGaN films are heteroepitaxially grown on lattice-mismatched substrates, such as sapphire and SiC substrates, which results in threading dislocations, and residual strains occur greatly in AlGaN epifilms. For reducing the threading dislocation density, a thin AlN nucleation layer or GaN buffer layer is utilized to improve the crystal quality [5-7]. D H Wang
Growth of an HT-AlN nucleation layer was achieved using a cold-wall showerhead low-pressure metal-organic chemical vapor deposition (LP-MOCVD) system. Hydrogen was used as carrier gas, triethylaluminum (TEAl), triethylgallium (TEGa) and ammonia (NH3) were used as Al, Ga and N sources, respectively. The pressure of the LP-MOCVD growth system was about 5.2×10−3 Torr, and the calculated ratio of V/III (NH3/TEAl) is 7058. As shown in Fig. 1, a 210-nm-thick HT-AlN nucleation layer was first deposited on (0001) sapphire substrate (2 inch) at 1100℃ denoted as Sample A. Sample B has the same structure except that a 12-nm-thick LT-AlN nucleation layer was grown at 660℃ between the sapphire substrate (2 inch) and a 210-nm-thick HT-AlN nucleation layer. In order to study and analyze the influence of the LT-AlN nucleation layer on AlGaN epifilms, we have grown an AlGaN with 1200-nm-thickness on Sample A and Sample B at 1030℃. The as-grown samples were characterized by high resolution X-ray diffraction (HRXRD), atomic force microscopy (AFM), scanning electron microscope (SEM) and Raman scattering. For the purpose to reveal and obtain a comprehensive knowledge of the threading dislocation density of AlGaN epifilms grown on the two AlN nucleation layers, we measured X-ray rocking curves (XRCs) for both symmetry and asymmetry diffraction planes by HRXRD. The HRXRD was performed using Bruker D8-discover system equipped with Ge (220) monochromator and channel-cut analyzer, delivering a pure Cu-K
Figure 3 shows the 10×10 μm2 AFM images of the two HT-AlN nucleation layers. We can see the differences in surface morphology between Sample A and Sample B. The micrograph in Fig. 3(a) and (b) indicates an improvement of quality of HT-AlN nucleation layers. These V-shaped pits are similar to those observed in c-plane GaN epifilm. It was found that the number of V-shaped pit defects on the surface of Sample B is less than on Sample A when a thin film LT-AlN nucleation layer is grown on sapphire substrates. Figs. 3(c) and (d) show the 3D surface morphology of the two HT-AlN nucleation layer samples. It is seen that Sample B shows clearly a lower quantity of dislocations on the surface than Sample A does. In addition, the root mean square (RMS) value is 4.09 nm for Sample A and 0.47 nm for Sample B, which indicates that Sample B has a smoother surface morphology than Sample A does. Furthermore, we have carried out SEM analysis in Figs. 3(e) and (f). As can be shown, there appear irregular convex disks on the surface of Sample A and regular convex disks on the surface of Sample B. From a quantitative point of view, the number of irregular convex disks for Sample A is more than Sample B has. We have marked the size of convex disks, and found the size of the convex disks on Sample A is a little smaller than those on Sample B. Consequently, be that as it may, the higher crystal quality of HT-AlN nucleation layers can be obtained by insertion of the LT-AlN nucleation layer between the sapphire substrate and the HT-AlN nucleation layer.
[FIG. 3.] AFM and SEM images of 10×10 μm2 surface morphologies for Sample A and Sample B. (a) Surface morphology of sample A, (b) Surface morphology of sample B, (c) 3D Surface morphology of sample A (RMS=4.09 nm), (d) 3D Surface morphology of sample B (RMS=0.47 nm), (e) SEM image of sample A, (f) SEM image of sample B.
Raman scattering was performed to examine residual strain of the two AlN nucleation layer samples recorded in the z(xx)-z backscattering configuration with 514-nm wavelength at room temperature. As is well known, the lattice and thermal mismatch between the substrate and the epifilm, characteristic for the heteroepitaxial growth of nitrides on a foreign substrate, results in the presence of strain under a linear elasticity biaxial system [11, 12]. Phonon frequency shifts are often employed as a tool for strain assessment in mismatched semiconductor heterostructures. The Raman scattering spectra of Sample A and Sample B are shown in Fig. 4 for a comparison. As can be shown in Fig. 4, the Raman peaks located at 651.0 cm−1 and 645.90 cm−1 are E2 (high) phonon modes for Sample A and Sample B, respectively. It was well known that the red-shift of AlN E2 (high) phonon mode corresponds to tensile strain in the HT-AlN nucleation layer, which is caused by differences in the thermal expansion coefficient and lattice mismatch between the AlN epilayer and sapphire substrate [5, 9]. The theoretical value of E2 (high) for AlN is 657.4cm−1 of stress-free [13, 14], so there existed red-shift of AlN E2 (high) mode of 6.4 cm−1 and 11.5 cm-1 relative to the theoretical value of 657.4 cm−1 for Sample A and Sample B, respectively, which implies that Sample A and Sample B are under tensile strain, and the tensile strain existing in Sample B is larger than that in Sample A. As we know, the strain along the growth direction is found to be tensile as elasticity theory predicts in the case of biaxial stress [15-18]. By comparison to the number of V-shaped pits, more tensile strain can be relieved in the condition of more V-shaped pits on the surface of Sample B. Furthermore, it is widely accepted that the larger tensile strain existing in-plane is apt to produce high crystalline quality epifilm. Thus, the conclusion can be drawn that inserting the LT-AlN nucleation layer can improve crystal quality of the HT-AlN nucleation layer.
In order to study further the influence on the crystal quality for AlGaN epifilms by inserting the LT-AlN nucleation layer, we grew an AlGaN with 1200-nm-thickness on Sample A and Sample B at 1030℃, as denoted as Sample A’ and Sample B’ respectively, as shown in Fig. 5. In Ref., we have carried out the symmetrical (002) and asymmetrical (102) XRD-
FWHMs and the calculated threading dislocation densities of AlGaN epifilms on Sample A’ and Sample B’
Ref.  studied the size of AlN island under different growth temperature modes, as can be seen from Fig. 6, the size of AlN islands grown under low-temperature mode is smaller than the size of those grown under intermediate temperature and high temperature , which coincides with AFM images results.
It was well known that the relationship between saturated vapor pressure and radius of grain in a nucleation layer is shown as the Kelvin formula in Eq. (1):
In this paper, we have focused on the influence on AlN nucleation layer and AlGaN layer caused by inserting a LT-AlN nucleation layer between the c-plane sapphire substrate and HT-AlN nucleation layers. HRXRD, AFM, SEM and Raman scattering have been employed to characterize crystal quality, surface morphology and residual strain of HT-AlN nucleation layer and AlGaN epilayer. Results indicated that crystal quality of HT-AlN and AlGaN layer can be improved greatly, and the threading dislocation density existing in AlGaN epifilms has been reduced when LT-AlN nucleation layer was inserted, which is consistent with the surface morphology of the HT-AlN nucleation layer and AlGaN epifilm probed by AFM and SEM.