Type of presentation: Poster

MS-9-P-3108 Precise IDB-type identification by phase shift in HRTEM imaging

Koukoula T.1, Kioseoglou J.1, Kehagias T.1, Georgakilas A.2, Komninou P.1
1Physics Department, Aristotle University of Thessaloniki, GR-54124 Thessaloniki, Greece, 2Microelectronics Research Group, IESL, FORTH, P.O. Box 1385, GR-71110 Heraklion Crete, and Department of Physics, University of Crete, P.O. Box 2208, GR-71003 Heraklion Crete, Greece
tkouk@auth.gr

Inversion domain boundaries (IDBs), bounding two neighboring domains with opposite polarity, are extended defects that may have detrimental implications in the performance of III-Nitride active layers, where the IDB* and the Holt type IDB models are observed by Transmission Electron Microscopy (TEM) methods. In particular, the high formation energy Holt IDB has been found to be electrically active by inducing electronic states in the band-gap, while the IDB* is inactive. Therefore, it is crucial to unambiguously determine the type of IDBs in order to appreciate their influence on the electronic properties of III-Nitrides heterostructures-nanostructures.
Here, we present a concise methodology, based on Geometrical Phase Analysis (GPA), in order to precisely identify the type of the observed IDBs, by inspecting the phase shift P(r) between adjacent inverse polarity domains in the phase images of High-Resolution TEM (HRTEM) micrographs. This means that for a specific type of IDB, the phase in two neighboring inverse polarity crystals would be shifted by a certain value that corresponds to the translation vector of the IDB. Depending on the diffracting conditions, P(r) is ±3π/4 and ±π/4 for the Holt and IDB* models, respectively, when the 0001 reflection is used. The corresponding values for the 0002 reflection are ±3π/2 and ±π/2. An interesting case is that the rigid body translation between the two models (±1/2[0001]), introduces an ambiguity in the identification of the specific IDB model, when the 0002 reflection is used. Following an extensive series of through focus and thickness HRTEM image simulations, it was found that the phase shift of the Holt model is strongly influenced by the imaging conditions (defocus and thickness values of the specimen). Thus, in order to irrefutably identify a Holt IDB, the 0001 reflection should be used.
The above methodology was implemented to identify the related IDB model between a GaN nanoisland (NI) and a GaN nanowire (NW), grown on Si(111) by MBE [Fig. 1(a)]. The corresponding phase images are shown in Figs. 1(b) and (c), using the 0002 g vector and the 0001 g vector of the Fast Fourier Transform (FFT), respectively, while line profiles of the phase images perpendicular to the IDBs are given in Figs. 1(e) and (f). Fig. 1(d) is the phase image using the 0002 vector of a simulated Holt IDB [inset in Fig.1 (a)]. The GaN NI is used as reference region in both cases and even though identification of the IDB type is unclear when using the 0002 g vector [Fig. 1(f)], this ambiguity is raised employing the 0001 g vector [Fig. 1(e)], which reveals the Holt type character of the IDB.


This research has been co-financed by the European Union (European Social Fund – ESF) and Greek national funds through the Operational Program "Education and Lifelong Learning" of the National Strategic Reference Framework (NSRF) - Research Funding Program: THALES, project “NanoWire”.

Fig. 1: (a) HRTEM image of an IDB between a GaN NI and a GaN NW. HRTEM image simulation of a Holt IDB is given as inset (b) & (c) Corresponding phase images using the 0002 g vector and the 0001 g vector in the FFT, respectively. (d) Phase image using the 0002 g vector of a simulated Holt IDB (e) & (f) Line profiles of the phase images.