No description available.

Die Herstellung von verspannten Schichten mittels Heteroepitaxie gewinnt in der aktuellen Festkörperphysik zunehmend an Bedeutung, insbesondere wenn es gelingt, Schichten mit unterschiedlichen Gitterparametern in hoher kristalliner Perfektion, wie sie für die Herstellung elektronischer Bauelemente notwendig ist, aufeinander abzuscheiden. Der Einsatz verspannter Schichtsysteme erlaubt es, bestimmte Materialeigenschaften, wie die Ladungsträgerbeweglichkeiten und den Bandabstand, gezielt zu beeinflussen. Im Rahmen der vorliegenden Arbeit wurde die frühe Phase der Relaxation von dünnen verspannten metastabilen Silizium-Germanium-Mischkristallschichten auf (001)-Siliziumsubstrat untersucht. Derartige metastabile Schichten bilden bei Temperaturbehandlung sogenannte Fehlanpassungsversetzungen aus. Die Ausbildung dieser Versetzungen in makroskopischer Ausdehnung in nahezu perfekt kristallinen Materialien setzt einen Nukleationsvorgang und die Ausbreitung durch Gleiten bzw. Klettern voraus. Diese Vorgänge wurden am o.g. Materialsystem systematisch untersucht. Die Untersuchung außerordentlich geringer Versetzungsdichten erfolgte mittels in-situ und ex-situ Röntgentopographie sowie ergänzender Messungen mittels hochauflösender Röntgendiffraktometrie. Die Plane-view-Transmissionselektronenmikroskopie sowie Atomkraftmikroskopie ergänzten die Experimente auf der damit möglichen Längenskala insbesondere für die Beurteilung des Verhaltens von sich kreuzenden Versetzungen. Es wurden Versetzungsgleitgeschwindigkeiten in Abhängigkeit von verschiedenen Fremdstoffkonzentrationen und Epitaxieverfahren gemessen. Ein signifikante Einfluß dieser Verfahren auf die Gleitgeschwindigkeit konnte nicht nachgewiesen werden, jedoch veringert der Sauerstoffgehalt einiger Proben die Versetzungsgleitgeschwindigkeit. Eine Kohlenstoffdotierung von 0,1% führt bei Einbau auf Gitterplätzen zu keinem messbaren Einfluß auf die Ausbreitungsgeschwindigkeit von Fehlanpassungsversetzungen. Der Prozess der Nukleation von Versetzungen wurde durch die heterogene Nukleation an Einzeldefekten, deren Vorhandensein stark vom Epitaxieverfahren abhängt, dominiert. Dies wurde an einer Vielzahl von Proben verifiziert. Die Aktivität der einzelnen Nukleationszentren ist sehr unterschiedlich und lässt sich nur durch ein "Spektrum" von Nukleationszentren unterschiedlicher "Stärke" erklären. Starke Nukleationszentren können bis ca 100 Versetzungen, die zusammen Versetzungsbündel bilden, induzieren. Die Nukleationszentren werden sequentiell aktiviert. Es wurde gezeigt, dass Laserbeschuss schwache Zentren "aktiviert" und eine damit eine sehr gleichmäßige Relaxation von Schichten möglich wird. Durch in-situ-Beobachtungen konnte erstmals röntgentopografisch der Prozess des Versetzungsblockierens und des Quergleitens an sich kreuzenden Versetzungsbündeln über einen großen Skalenbereich verfolgt werden.Es konnte gezeigt werden, dass die Zweikristall-Röntgentopografie ein geeignetes Verfahren zur Untersuchung großer teilrelaxierter Proben mit extrem geringem Relaxationsgrad darstellt. / The deposition of metastable layers by means of heteroepitaxy is gaining more and more in importance in current solid state physics, especially if it is possible to deposit layers with rather different lattice parameters in perfect quality, necessary for the fabrication of electronic devices and semiconductor technology. The use of such layer systems enables to controlling specific material properties, for example the mobility of charge carriers and the energy gap. Within the framework of this work the early stages of the relaxation of very thin strained silicon-germanium layers deposited onto (001)silicon substrates were investigated. Such metastable layers create so-called misfit dislocations at annealing. The formation of dislocations in nearly perfect materials as silicon in macroscopic extension presumes a nucleation process and the propagation by means of gliding and/or climbing of the threading segments. These processes were investigated systematically at silicon-germanium-films on silicon substrates as a modelsystem. Very low densities of dislocations were investigated by in-situ and ex-situ X-ray topography as well as high resolution X-ray diffractometry. Plane-view transmission electron microscopy and atomic force microscopy complemented the experiments in order to study the behavior at the dislocation crossings at smaller length scale. Dislocation glide rates were measured at different concentrations of impurities (e.g. carbon) in the sample and for different growth techniques. There was no significant influence of the epitaxial method on the glide velocities but a very low amount of oxygen decreases the the velocity effectively. A low density of carbon (0,1%) wich occcupy silicon or germanium lattice sites had no significant influence on the propagation of dislocations. The process of nucleation is dominated by heterogeneous nucleation as was verified at a large number of samples and varies in detail for different epitaxial depostion methods. The activity of single nucleation centers is rather different and can interpreted with a "spectrum" of centers of different strength. A strong center can induce as many as 100 dislocations. These dislocations form bunches of dislocations as was observed with X-ray topography. The centers are sequentially activated. This was demonstrated at samples where a pretreatment with an excimer laser "activated" weak centers thereby inducing a comparatively homogeneous distribution of dislocations. By means of in-situ synchrotron radiation experiments the prozesses of propagation, blocking and crossslip could be observed for the first time in large areas of samples. The double crystal X-ray topography is a suitable method to observe these processes in comparatively large samples with a very low degree of relaxation.

本論文觀察台灣地區國語閩南口音，也就是俗稱的「台灣國語」中展現的單元音化以及擦音卸解現象，建立「台灣國語語料庫」作為本論文分析的基礎，並且從優選理論的觀點探討這些台灣國語音韻現象。 根據「台灣國語語料庫」的數據顯示，韻核後介音 /j/ 以及 /w/ 容易被語者刪除，韻核前介音 /j/ 以及 /w/ 則傾向被語者保留。輕擦音 /f/ 在台灣國語中常被卸解為 /hw/。 本文的分析認為，國語閩南口音是一種方言的變異，反映了閩南語以及國語的部份制約排序。當語者表現出明顯的閩南語口音時，反映出閩南語的部份制約排序；而語者的閩南語口音較不明顯時，則反映了國語的部份制約排序。 / This thesis investigates monophthongization and fricative reconfiguration in the Taiwanese Accent of Mandarin (TM), and establishes the TM corpus as the base of phonological analysis. The Optimality Theory approach is adopted to account for the phonological phenomenon of TM. As indicated by the TM corpus, the postnuclear glides /j/ and /w/ tend to be deleted in TM; the prenuclear glides /j/ and /w/ tend to be preserved in TM; and the fricative /f/ tend to reconfigurate as /hw/ in TM. Our analysis considers that TM is a dialectal variation which reflects the subgrammars of Taiwanese and Mandarin. When the speaker shows a clear Taiwanese accent, the partially ordered ranking of Taiwanese is reflected; and when the speaker shows no obvious Taiwanese accent, the partially ordered ranking of Mandarin is reflected.

This thesis explains how the classified design parameters of existing passenger jets can be determined. The classified design parameters are; the maximum lift coefficient for landing and take-off, the maximum aerodynamic efficiency and the specific fuel consumption. The entire concept is based on the preliminary sizing of jet powered civil aeroplanes. This preliminary sizing is explained in detail because it is the foundation of the final result. The preliminary sizing is combined using reverse engineering which is not a strict method. Therefore, only the basics are explained. By applying reverse engineering on the preliminary sizing and aiming for the classified design parameters as output, formulas are derived to calculate the maximum lift coefficients, the maximum aerodynamic efficiency and the specific fuel consumption. The goal is to calculate these parameters, using only aircraft specifications that are made public by the manufacturer. The calculations are complex with mutual relations, iterative processes and optimizations. Therefore, it is interesting to integrate everything in a tool. The tool is built in Microsoft Excel and explained in detail adding operating instructions. The program is executed for miscellaneous aeroplanes, supported with the necessary comments. Investigated aeroplanes are: Caravelle 10B (Sud-Aviation), Boeing 707-320C, BAe 146-200 (British Aerospance), A320-200 (Airbus), "The Rebel" (based on A320), Boeing SUGAR High, Boeing 747-400, Blended Wing Body VELA 2 (VELA) and Dassault Falcon 8X.

<p>Synthesized speech may be used for nefarious purposes, such as fraud, spoofing, and misinformation campaigns. We present several speech forensics methods based on deep learning to protect against such attacks. First, we use a convolutional neural network (CNN) and transformers to detect synthesized speech. Then, we investigate closed set and open set speech synthesizer attribution. We use a transformer to attribute a speech signal to its source (i.e., to identify the speech synthesizer that created it). Additionally, we show that our approach separates different known and unknown speech synthesizers in its latent space, even though it has not seen any of the unknown speech synthesizers during training. Next, we explore machine learning for an objective in the aerospace domain.</p> <p><br></p> <p>Compared to conventional ballistic vehicles and cruise vehicles, hypersonic glide vehicles (HGVs) exhibit unprecedented abilities. They travel faster than Mach 5 and maneuver to evade defense systems and hinder prediction of their final destinations. We investigate machine learning for identifying different HGVs and a conic reentry vehicle (CRV) based on their aerodynamic state estimates. We also propose a HGV flight phase prediction method. Inspired by natural language processing (NLP), we model flight phases as “words” and HGV trajectories as “sentences.” Next, we learn a “grammar” from the HGV trajectories that describes their flight phase transition patterns. Given “words” from the initial part of a HGV trajectory and the “grammar”, we predict future “words” in the “sentence” (i.e., future HGV flight phases in the trajectory). We demonstrate that this approach successfully predicts future flight phases for HGV trajectories, especially in scenarios with limited training data. We also show that it can be used in a transfer learning scenario to predict flight phases of HGV trajectories that exhibit new maneuvers and behaviors never seen before during training.</p>