Role of Dopants and Impurities on Pinhole Formation; Defects Formed at InGaN/GaN And AlGaN/GaN Quantum Wells
- PDF / 2,740,117 Bytes
- 6 Pages / 414.72 x 648 pts Page_size
- 52 Downloads / 184 Views
"**** APA Optics, Inc., 2950 N.E. 84th Lane,
Blaine, MN 55449, USA
ABSTRACT Transmission electron microscopy has been applied to study defects in epitaxial doped and undoped GaN layers grown by MOCVD on sapphire and SiC substrates. Samples with InGaN/GaN and AlGaN/InGaN heterostructures have also been investigated. The results of this study show that incorporation of "foreign" atoms increases formation of nano-tubes and pinholes. The highest density of these defects was formed close to the interface with sapphire where oxygen outdiffusion might be expected, or in the subsurface area in the samples where oxygen was added deliberately. Addition of In (or Al) at QW's also leads to pinhole formation. Increased In fraction or a larger number of QW's results in a greater density of pinholes and more surface roughness. Many of the "hollow" nanotube defects are terminated during growth. INTRODUCTION GaN has been recognized as a wide-bandgap, high temperature semiconductor material that has important applications in both short-wavelength optoelectronic and high power/high frequency devices [1]. Because of lack of lattice matched substrates GaN is grown mostly on A12 0 3 or SiC with about 14% or 3.5 % lattice mismatch, respectively. This lattice, and also the thermal expansion coeficient, mismatch between the layer and the substrate causes to three dimensional growth and a high density of defects. These defects are mainly dislocations arranged in small-angle grain boundaries. However, there are other defects like inversion domains (IDs), nanotubes and pinholes, which do not appear to be caused by lattice or thermal mismatch since they are observed also in homoepitaxial layers [2,3]. These last two defects, nanotubes and pinholes, appear to have the same origin [4,5]. Both start with a V-shape cone-like feature faceted on (10"11) low energy polar planes, and both are empty inside. Their presence is not limited to a particular growth method or substrate. In plan-view micrographs both these defects have perfect or slightly elongated hexagonal shape, therefore, some defects that have been assumed to be "nano-tubes" may not be. From the study of SiC one could conclude that nanotubes are empty core screw dislocations following the Frank model [6], who suggested that a screw dislocation of large Burgers vector might have an empty core. The open-core, should represent an equilibrium between the extra surface free energy and the decrease in lattice strain energy. According to Frank's model the radius of the hole is proportional to the square of the Burgers vector. Since elastic constants in GaN are not well established the validity of the empty 375 Mat. Res. Soc. Symp. Proc. Vol. 482 © 1998 Materials Research Society
core screw dislocation model is questionable especially because Burgers vectors observed experimentally are an order of magnitude smaller than the Burgers vector required to fulfill Frank's model. In our earlier paper [5] a different possible explanation was provided for the formation of "hollow" type defects based o
Data Loading...
