Growth of GaN without Yellow Luminescence
- PDF / 322,608 Bytes
- 5 Pages / 414.72 x 648 pts Page_size
- 75 Downloads / 238 Views
native defects are the natural candidates for the origin of the electronic levels participating in the YL. Contrary to conventional beliefs, Neugebauer and Van de Walle showed that in n-type GaN, the gallium vacancy is the most energetically favorable native defect [13]. The above studies show important characteristics of the YL. However, a correlation between the growth conditions and the YL would be more helpful to relate these results with each other. In our previous studies [14,15,16], we noticed that the YL becomes barely detectable for Si- or Mg-doped GaN. For undoped samples, GaN grown on SiC substrates shows the weakest YL. In this paper, we report a systematical photoluminescence study of GaN samples grown under different growth conditions. With the correlation between photoluminescence and growth conditions, the results shown here strongly suggest that: (i) the gallium vacancy and related complexes are responsible for the YL emission; (ii) These defects are generated due to the large lattice and thermal mismatch between epilayer and substrates. The closer to the epilayer-substrate interface, the higher the density of defects. EXPERIMENTS The growth conditions are the same as those we have reported [14,15,16]. Briefly, all epitaxial layers were grown by MOCVD. Trimethylgallium, trimethylaluminum, ammonia, germane and bis-cyclopentadienylmagnesium were used as starting materials for Ga, Al, N, Ge and Mg elements, respectively. The carrier gas was H 2. The effect of lattice mismatch was studied by using two different substrates: sapphire (00.1) and 6H-SiC (00.1). These substrates were always put side by side on a 4.5"-diameter SiC-coated graphite susceptor. The susceptor is rotated by gas-foil rotation technology and heated by a RF-coil. Thus excellent uniformity can be achieved across the susceptor. For other studies, only sapphire (00.1) was used as the substrate. A thin AIN buffer layer was deposited before the GaN layer with a thickness of 0.5pm to 4pm. Photoluminescence spectra were measured with a He-Cd laser as the excitation source (25nm, =5mW), a grating monochromator and a standard synchronous detection setup. The power density on the sample is approximately 50W/cm-2 after focusing. At this power density, we measured that the dependence of the integrated peak intensities of YL and bandedge emission are both linear. All photoluminescence experiments reported in this paper are carried out at room temperature. Two series of experiments were performed to study the correlation between the growth conditions and YL. One is the doping of GaN with Ge and Mg; the other is the growth of GaN with different TMGa flow. RESULTS AND DISCUSSION The photoluminescence of undoped GaN, Ge-doped GaN and Mg-doped GaN are shown in Fig. 1 (a), (b) and (c) respectively. Clearly shown is the absence of YL emission from both GaN:Ge and GaN:Mg on (00.1) A120 3. Such emission was not observed from doped layers on other substrates either. Ge (group IV) acts as a donor and Mg ( group II) as an acceptor in GaN by substituting to
Data Loading...
