Theory of Interfaces and Surfaces of Wide-Gap Nitrides
- PDF / 407,901 Bytes
- 6 Pages / 414.72 x 648 pts Page_size
- 89 Downloads / 156 Views
configuration was chosen as the atomic reference for nitrogen. The alloys were modelled using the virtual crystal approximation. We studied 4+4 superlattices in the wurtzite structure, taking the in-plane lattice parameter to be that of GaN (3.20 A). Previous work on GaN/A1N interfaces [7] demonstrated that effects of strain on the interface are significant. Consequently, the large mismatch between GaN and InN lead us to expect that the effects of strain in alloys, even with low In composition, will be large. The calculation of the polarizations in the epilayers followed Refs. [9, 7]. In surface calculations, the slabs contained four bilayers of GaN, three of which were relaxed. Calculations with thicker slabs indicate that a sufficiently thick slab was employed. Additionally, for calculations involving the Ga-terminated surface, pseudo-hydrogens of charge 0.75 passivated the Nterminated face, while for the study of the N-terminated face, pseudo-hydrogens of charge 1.25 passivated the Ga-terminated surface [10]. POLARIZATION FIELD EFFECTS IN In 0 .2 Ga0.sN/Inl_:Ga•N MULTIPLE QUANTUM WELLS The wide-gap nitrides display pyroelectric and piezoelectric behavior because of their low symmetry, wurtzite structure [7]. Thus in the active region of a nitride-based MQW a net electric field perpendicular to the plane of the well will be present. If large enough, this field will cause a spatial separation of the electron and hole wavefunctions in the well, a consequent decrease in the wavefunction overlap and ultimately a reduction in the interband recombination rate (the so-called Stark effect). In order to study this phenomenon, we chose to model our theoretical device as a periodically repeated MQW confinement heterostructure, consisting of In 0.2Ga 0 .8 N well layers (active region) and In 1 -xGa.N barrier layers, grown on a GaN substrate (GaN in-plane lattice constant). Values of x=1.0, 0.95, and 0.9 were considered. We find that all of these interfaces are of type I and that there are no electrically active interface states in the gap. In a polar material, the macroscopic electric field, F, is related to the macroscopic polarization, P, via the dielectric constant of the material: F = -P/loe. The macroscopic polarization and electric field present in a single epilayer of a MQW are shown in Table I (the field present in pure GaN is given for reference). The values chosen for the dielectric constants were c(GaN)=9 and E(InN)=15 [11]. A Vegard's law approach was used for the dielectric constants of the alloys. We observe that the size of the field spontaneously present in nitride MQW's is roughly one order of magnitude larger than the typical fields used in experiments
GaN In0.05 Ga 0 .95 N In 0 .1 Ga 0 .9 N
SIn0. 2 Ga 0 .8 N
[12]. P (10-2 C/m 2 )
F (kV/cm)
0.45 0.58 0.81
564 704 952
1.16
1019
Table 1: Calculated polarization (P) and absolute value of the macroscopic electric field (F) in the different strained epilayers.
In order to understand the effect of the field on the recombination rate, we modelled the MQ
Data Loading...
