Mean inner potential variation with strain in III-nitrides studied by off-axis electron holography
Résumé
Background incl. aims
Wurtzite III-nitride semiconductors are well known for their intense spontaneous and piezoelectric polarization fields. A deeper understanding of these properties is relevant for applications in various fields such as optoelectronic or power electronic devices. Off-axis electron holography is a powerful tool to measure directly and with a nanometer resolution the total electrostatic potential in these materials. However, to be able to study the polarization fields in III-nitrides, we need to remove the contribution from mean inner potential (MIP), which corresponds to the volume average of the Coulomb potential in the specimen. In the literature, MIP is traditionally derived for bulk materials within density functional theory (DFT).1,2 Evaluating the MIP using electron holography, we noticed disparities with values from the literature. One explanation that we propose is that the calculation for bulk material does not reflect the reality of heterostructures that are used in devices, with for instance the importance of strain. We propose here a deep analysis of the MIP evaluation for III-nitrides, comparing electron holography results with DFT calculation of the MIP that include the variation of strain. Methods
MIP was computed through ab-initio simulations performed using GPAW code with plane wave basis-set of 300 eV cutoff energy, PBE functional, and 8x8x1 k-point sampling. The calculations were performed in a slab containing 11 atomic layers terminated in nonpolar m-{10-10} facets, centered in a 6.2-nmlong vacuum region with periodic boundary conditions. The MIP was obtained by averaging the Coulomb potential in the center 7 atomic layers, to prevent surface effects. The effect of strain has been modelled considering a biaxial strain configuration for layers grown along the <0001> direction. From the experimental point of view, we study here 4 different samples. The first structure consists in a 6-period non-polar AlN/GaN (13nm/16nm) on 60 nm of AlN deposited on a (1-100) SiC substrate. The second sample is a 10period polar AlN/GaN (20nm/20nm) superlattice grown by plasma-assisted molecular beam epitaxy (PAMBE) on 100 nm of GaN deposited on 1-μm-thick AlN-on-sapphire templates.3 The third and fourth samples consist in 5-period GaN/InGaN multi-quantum wells (MQWs) grown on a 2 µm-thick GaN layer that is itself grown on a 220-nm thick AlN buffer layer on patterned Si(111) templates using low-pressure metal organic vapor phase epitaxy (MOVPE).
Origine | Fichiers éditeurs autorisés sur une archive ouverte |
---|---|
Licence |