JEŘÁBEK, F. Analýza dislokací v GaN pomocí mikroskopických metod [online]. Brno: Vysoké učení technické v Brně. Fakulta strojního inženýrství. 2025.

Posudek vedoucího

Kolíbal, Miroslav

František Jeřábek has focused on defect properties in GaN, a topic which is related to the Chips JU project running at BUT. The student set himself an ambitious goal – acquire a large number of datasets from different techniques, cross-correlate them and draw conclusions based on a robust database of experiments. I have to point out that he did not have experience with either technique he used; ha has to learn both ECCI and cAFM measurements from the scratch, and even more: ECCI was available at Thermo Fisher Scientific for a very limited amount of time (few days in summary) and the cAFM microscopes are quite sophisticated and non-standard (Nano Scan and Litescope). Nevertheless, the student have quickly learned to perform ECCI on different tool even in the limited time given, and he also dedicated his time to operate and use Litescope (although these measurements are not included in the thesis). He had to overcome many experimental hurdles, and sometimes with the help of PhD students in the group he managed to do so. This does not belittle his independence! Apart from experimental work, he worked intensively on a correlation methodology, which has proven extremely efficient and promising for future use within the project and beyond. František was always helpful to younger students, he has widespread skills and is able to get quite deep into problems of different nature. I had only minor comments to the thesis during writing. The thesis nicely represents his diploma work, although he omitted many more activities he had performed and results he has received. Grade A is well-deserved here.

Posudek oponenta

Chisholm, Claire

The thesis topic is well-defined and currently industry- and academic-relevant as the EU strives to increase understanding and superiority in the area of wide-bandgap semiconductors. Chapters 1 and 2 include a good amount of the relevant background research and is used effectively to highlight the current state of the art and where advances can be made. Chapter 3 has a very nice physics-based explanation of electron channeling contrast imaging, demonstrating a good physical understanding of the method. CAFM is defined concisely and the advantages of current-controlled CAFM are introduced. Chapter 4 includes a very thorough and useful report of the sample cleaning required for CAFM measurements. Such reports represent a huge amount of effort and utility that is critical for experimental repeatability, but are rarely included in journal articles. It is clear from Chapter 4 that extreme care was made to develop a repeatable workflow – a sign of a high-quality experimental design process. Chapter 5 outlines a very impressive data analysis and correlation process. The high impact of this dissertatation is largely due to this workflow. There are three very distinct topics: ECCI, CAFM, and data correlation and analysis, each with their own challenges. Even one of these alone would be an appropriate masters topic. It is the opinion of this reviewer that this body of work would require only a slightly deeper dive on some topics and a few more experiments to become a passable PhD dissertation. As a master’s thesis it goes above and beyond what should be required. The primary criticism is that there is a lack of understanding of crystallography, but given the breadth and achievement of the work, this should be overlooked. There are some corrections and inclusions, however, that should be made: -Threading dislocation should be defined. The primary assumption made in the ECCI analysis is that the line direction of the dislocations is nominally parallel to the growth direction. And it is important to make this assumption clear. -, , and type should have their Burgers vectors defined in crystallographic terms – perhaps including a figure -Edge, mixed, and screw are used too broadly (in particular, the use of “pure” is over-reaching). Without including a mention of the line direction, these terms lose their relation to their Burgers vector. It would be better to either (1) explicitly define what is meant by edge, screw, mixed, (2) always include the line direction, ex. ‘edge-type threading dislocation’ rather than simply ‘edge’, or (3) refer to them by their direction, ex. ‘-type’ instead of ‘edge’. -Chapter 3 is missing a discussion on why threading dislocations have the butterfly contrast (and a reference to WJ Tunstall, et.al. Phil Mag, 9:97 (1964)). This is important for a few reasons: (1) the surface relaxation that happens means that g dot b will never be 0 (g dot b invisibility is also not explained and is lacking a reference), and (2) when you do your ECCI image manipulation it is important to know whether this manipulation of the black-white contrast is affecting the validity of your analysis and this thought exercise should start with and understanding of Tunstall. However, this is perhaps beyond the scope of a master’s thesis. -The ECCI experimental results have no mention of the diffraction vector, g. The vector and the direction in the ECCI images are both critical parameters to report. If g ={11-20}, then I would ask why you expect type dislocations to be invisible in some orientations, and if g={10-10} then I would ask why you did not use g={11-20}. -It is stated more than once (ex. p.16) that the black-white contrast direction for type threading dislocations is “arbitrary”. This is incorrect. Because the Burgers vector includes and components the resulting ECCI contrast will include effects from both, and in 1 of 12 expected directions. -The Fan, et.al. paper (reference [15]) is unfortunately somewhat misleading and the use of it for Fig. 3.3 has lead to the propagation of the original paper’s error. In this figure, g is shown at a multi-beam condition – this means there is more than one set of planes diffracting. In the positions shown, two or perhaps even three sets of planes are diffracting. In this case one cannot determine a single diffraction vector and therefore the ECCI contrast should not be used to determine Burgers vector. It would be better to use the Naresh-Kumar PRL 2012 paper (ref [14]) for this figure, as they were more careful to use a two-beam condition.

eVSKP id 166070