Objective: Combine quasi-particle methods with band-structure and molecular dynamics results to find optical accessible states of important semiconductor defects -- localized and extended.
Band offsets at heterojunction interfaces are important for designing devices for desirable opto- or microelectronic effects such as quantum confinement, high-carrier injection efficiency, and high gain. Many epitaxial growth structures have strain intentionally built into them in order to match semiconductors with different lattice constants. An example is the AlN-GaN heterojunction in which strain induced by the lattice mismatch induces piezoelectric charge and, in turn, a two-dimensional electron gas at the interface that provides both high sheet carrier density and high mobility for high frequency and power applications.
To date AlN-GaN band offsets have not been measured. Theory can fill this "gap." The structures considered are necessarily thin, but the results do not change at 0.1 eV level with an additional layer of GaN and AlN. The one free variable, the in-plane lattice constant, ranges from SiC (a typical growth substrate) to AlN and GaN, and it induces both a surface dipole (responsible for the offset) and interface monopole charge (producing an internal electric field). Our results are the first to include these important polarization effects. The most important result is the strong sensitivity of the conduction and valence band offsets to lattice constant: 0.5 eV shift in valence band offset between SiC and GaN in-plane lattice constants; the conduction band offset shift is smaller, 0.3 eV, but important.
Extended defects in silicon can dominate the device fabrication and performance. For example, boron diffusion is transiently enhanced by the Si interstitials present in the extended {311} planar defect formed by the implantation. These defects have been seen by TEM. More valuable would be a technique that could monitor their evolution. The figure shows the quasi-particle spectrum for an II interstitial defect. The II fragment has been carved out of the equilibrium structure we have found with molecular dynamics for the extended {311} defect. In contrast to the other fragment, IO, of the {311}, the II defect has a states in the gap starting 0.55 eV above the top of the valence band. We speculate that these states are related to the 0.50 eV excitation seen with Deep Level Transient Spectroscopy (DLTS) on silicon material known to have extended {311} defects. The additional supporting evidence is that the signal has logarithmic kinetics, a feature typical of an extended defect.
Interstitial cluster precursors must be formed during implantation and subsequent annealing. How they form extended defects is unknown. Currently we are using molecular dynamics methods to find stable and important metastable interstitial-Si clusters. Quasiparticle methods can identify the in-the-gap states. We know, for example, a low Si ion dose, DLTS spectra are dominated by two signatures at 0.29 and 0.48 eV. Perhaps small clusters are the source of these signatures.
|
| Comparison of quasiparticle and LDA band structures for the Si planar {311} II defect along the high symmetry [011] direction. For simplicity, only highest 4 valence and lowest 4 conduction bands are shown, along with first conduction band of bulk Si. GWA and LDA predict a direct band gap at Gamma. Quasiparticle corrections increase the size of the LDA direct gap 0.69 eV to 1.25 eV. The quasiparticle shifts show some dispersion: the shifts at K are about 0.15 eV smaller than the shifts at Gamma, producing a flatter conduction band and more localized in-the-gap state. |
Contact: John W Wilkins, voice: 614-292-5193; fax: 614-688-3871; wilkins@mps.ohio-state.edu