No description available.

Ces dix dernières années ont vu émerger des avancées technologiques majeures, nous permettant capturer une image instantanée des processus physique. L'amélioration systématique de la résolution temporelle de ces instants, grâce aux lasers (de différente sorte) aux impulsions ultracourtes, a joué un rôle important dans l'exploration des transitions de phases photo-induites dans différents matériaux, et leur potentiel applicatif. Néanmoins, ce progrès technologique incontestable a poussé à ses limites notre capacité de décrire les phénomènes hors-équilibre très complexes qui pilotent les transitions. Ils sont intrinsèquement multi-échelles dans le temps et dans l'espace, s'étalant de la femtoseconde aux plusieurs jours, et de la dimension atomique jusqu'à celle d'un cristal macroscopique. Les expériences résolues en temps permettent de séparer temporellement différents dégrées de liberté et les phénomènes pilotés par ceux-ci, au lieu d'observer seulement leur moyenne statistique. La première étape (processus photo-induit) de cette séquence temporelle est liée à l'absorption d'un photon, la deuxième (élastique) est pilotée par la dilation du volume macroscopique du matériau, et la troisième étape (thermique) est due aux effets de chaleur. Cette approche séquentielle offre de nouvelles possibilités pour mieux comprendre comment impacter les matériaux de façon contrôlée et efficace. Les lasers opérant dans le moyen infrarouge (mid IR) permettent de suivre le déroulement d'une transition de phase par le changement de vibrations des molécules/liaisons ciblées. Cette spécificité au site moléculaire combinée avec la résolution en temps ultracourte devrait ouvrir une nouvelle fenêtre d'observation des phénomènes qui échappaient aux études scientifique. Ce travail de thèse a commencé exactement dans cet esprit. L'effort majeur a été dédié à l'application de la spectroscopie mid IR ultrarapide aux matériaux présentant une conversion de l'état de spin, [Fe(3-MeO-SalEen)]2PF6 en particulier. La principale difficulté de ce travail consistait à décrypter le contenu spectral des molécules hors-équilibre. Nous avons découvert que l'approche utilisé dans les spectroscopies résolues en temps de plus haute énergie (UV/VIS) ne suffit pas pour étudier la problématique posé dans le cadre de cette thèse. Une nouvelle approche a été pensée pour modéliser les spectres résolus en temps, et celle-ci consistait à séparer la réponse spectrale en deux contributions : le transfert de poids spectral, et un décalage spectral. J'ai pu démontrer que ces deux contributions suivent sensiblement le changement d'état de spin, et la pression (dilatation du volume). L'analyse de données basée sur ce modèle, corrobore les résultats obtenus jusqu'alors avec d'autres techniques. Sur l'échelle de temps ultracourts, plus difficile à modéliser, nous avons pu néanmoins résoudre très clairement le refroidissement vibrationnelle (VC) de l'état électronique haut spin -chaud. A ma connaissance, ce phénomène dans un système solide présentant crossover de spin n'a jamais été observé directement. / The past few decades have seen great advancements in technology that allow us to capture the picture of a physical process, as the adage “seeing is believing” implies how people understand the world. The increasing temporal resolution of lasers played an important role in the study of materials, among which materials exhibiting photo-induced phase transition are of great importance thanks to their potential for future applications. However, as we proceed further and further in the investigation of the mechanism of phase transition, we found ourselves confronted with the very complex nature of phase transition dynamics. It is intrinsically multi-scale in time and space, from femtosecond to days and from atomic dimensions to macroscopic distances. Time resolved experiments disentangle different degrees of freedom and different phenomena in a step-like manner, rather than providing a statistical average. The first step is photo induced due to absorption of photons, the second step (elastic step) is pressure induced due to volume dilation, and the third step is temperature induced due to dissipation of heat. This step-like approach offers an opportunity to understand the mechanism, so that we could effectively impact the materials and possibly control phase transition. Mid IR lasers have a unique advantage of monitoring phase transition through vibrational modes on specific molecular sites. Implementing ultrafast mid IR spectroscopy in phase transition materials should be therefore very insightful in discovering new phenomena and revealing hidden mechanism. This PhD project, focusing on mid IR technique, started exactly in this context. The main effort is dedicated to the application of ultrafast mid IR spectroscopy to the spin crossover solids, [Fe(3-MeO-SalEen)]2PF6. The major challenge in this work was to comprehend the shape of transient mid IR spectra. We found out that this is conceptually different from the experiences accumulated in UV/VIS spectroscopy. A suitable model had to be developed, separating the transient IR spectra into two contributions: spectral weight transfer and spectral shift. I demonstrated that these two components are sensitive to the spin change and pressure effect (volume dilation), respectively. Data analysis based on the new model shows consistency with previously published results. On the ultrafast timescale, more difficult to fit our model to, vibrational cooling (VC) of electronically hot HS state has been very well resolved. To the best of my knowledge, direct observation of VC in solid state SCO compound by IR spectroscopy has not been reported earlier.

Echo state networks represent a special type of recurrent neural networks. Recent papers stated that the echo state networks maximize their computational performance on the transition between order and chaos, the so-called edge of chaos. This work confirms this statement in a comprehensive set of experiments. Afterwards, the best performing echo state network is compared to a network evolved via neuroevolution. The evolved network outperforms the best echo state network, however, the evolution consumes significant computational resources. By combining the best of both worlds, the simplicity of echo state networks and the performance of evolved networks, a new model called locally connected echo state networks is proposed. The results of this thesis may have an impact on future designs of echo state networks and efficiency of their implementation. Furthermore, the findings may improve the understanding of biological brain tissue. 1

No description available.

Philosophiae Doctor - PhD / In this thesis we report on the metal-mediated-thermally induced changes of the structural and optical properties of hydrogenated amorphous silicon deposited by hot-wire CVD, where aluminium and nickel were used to induce crystallization. The metal-coated amorphous silicon was subjected to a thermal annealing regime of between 150 and 520°C. The structural measurements, obtained by Raman spectroscopy, show partial crystallization occurring at 350 °C. At the higher annealing temperatures of 450°C and 520°C complete crystallization occurs. Reflection and transmission measurements in the UV-visible range were then used to extract the optical properties. By adopting the effective medium approximation a single optical model could be constructed that could successfully model material that was in different structural phases, irrespective of metal contamination. Changes in the absorption of the material in various stages of transition were confirmed with a directly measured absorption technique, and the modelled absorption closely followed the same trends This study forms part of the larger overall solar cell research project, of which the primary aim is to eventually develop a silicon solar panel that optimises the characteristics for best performance. / South Africa

Le sujet de cette thèse est principalement ciblé sur l’étude du transport ionique dans les sulfates d’alcalins de la famille LIMSO₄, où M=Na, K, Rb, Cs. Une attention particulière est portée sur l’étude du système LiNaSO₄, le plus intéressant en termes de dynamique ionique, par RMN en fonction de la température. Dans le cadre de cette étude, nous avons effectué des mesures de temps de relaxation et de largeurs de raie pour toute la série de composés. Des mesures de coefficients d’autodiffusion du ₇li et du ²³Na, ont été menées depuis l’ambiante jusqu’à la température de fusion. Pour la première fois, nous avons pu mesurer la cinétique de transition de phase dans LiNaSO₄, à partir d’une nouvelle méthode basée sur la différence de temps de relaxation dans les deux phases de part et d’autre d e la transition, mais sans mesurer forcément le T₁. Cette technique élaborée dans le cadre de ce travail permet de mesurer l’évolution au cours du temps du volume de la phase qui apparait pour des vitesses de refroidissement contrôlées. Nous avons aussi mené une étude par RMN des réorientations des groupements sulfates dans la phase basse température. L’influence des réorientations des SO ₄² sur les couplages quadripolaires au noyau 7li, a été étudiée par un modèle de réorientation par sauts, qui n’avait encore jamais été utilisé pour les sulfates. La méthode proposée est une méthode à « bas couts » car elle permet d’atteindre l’information sur la dynamique des groupements sulfates sans enrichir l’échantillon en ¹⁷O ou de mener des expériences très longues pour le ³³S, ou sans passer par les mesures de temps de relaxation. Afin d’analyser le rétrécissement par le mouvement (motional narrowing) en fonction de la température dans les solides avec deux sous réseaux cationiques diffusant comme dans le cas du LiNaSO₄ , nous avons élaboré un modèle permettant de fitter l’évolution observée à deux marches de la largeur de raie RMN avec la température. La fonction analytique obtenue a été étendue au cas de distributions de temps de corrélation. / The main goal of a present research is a detailed study of ionic transfer in double sulfates belonging to the LIMSO₄ family, where M = Na, K, Rb, Cs. The most attention has been paid to LiNaSO₄ as to the most interesting (in terms of the ion dynamics) compound among other double sulfates. We have carried out magnetic relaxation measurements and line width analysis for all compounds under consideration. Moreover, PGF NMR measurements of ₇li and ²³Na self-diffusion coefficients in LiNaSO4 have been carried out. For the first time, we have measured the phase transition kinetics in LiNaSO₄. For this purpose, we developed a new technique, which is based on the difference of spin-lattice relaxation times in the two phases, but which does not involve the direct measurement of T₁. Elaborated technique allows measuring time evolution of the volume of the appearing phase at controlled cooling rates. We have carried out NMR study of the sulfate ion reorientations in the low-temperature modification of LiNaSO₄. The influence of the SO ₄² reorientational jumps on the quadrupolar interactions of 7Li nuclei was investigated b y a j ump reorientational model, which has not previously been app lied to sulfates. The proposed method is a “low-cost” technique, since it does not require an ¹⁷O enriched sample and dispenses with time-consuming ³³S NMR. Other advantage of a given method is a possibility to probe reorientational motions without NMR relaxation measurements. To analyze motional narrowing in solids with two diffusing spin sublattices (such case occurs, e.g., in LiNaSO₄) we deduced a formula, which can be used for fitting of the two-step temperature dependencies of the NMR line width. The obtained function has been al so ex tended to the case, when a distribution of correlation times takes place. The advantage of this approach is that even in the case of distribution of correlation times, the fitting function could be expressed in the analytical form.

Abstract In this work, evolution of structural properties of anhydrous and hydrated alkali halide clusters are studied using synchrotron radiation based photoelectron spectroscopy. Alkali metal core level spectra of small anhydrous RbCl, RbBr, CsCl and CsBr clusters indicate a NaCl structure. For larger CsBr clusters a structural phase transition to CsCl structure is likely the case. Alkali halide core level spectra of mixed RbBr-water clusters indicate that at dilute concentration the salt is dissolved by the water cluster but ion pairing increases with concentration. Modeling of gas phase cluster formation and electronic structure calculations of core level chemical shifts are used to interpret the experimental spectra.

Este trabalho visa a estudar o modelo de percolação independente, de Bernoulli, em grafos, tendo como objetivo principal obter condições que garantam a ocorrência de transição de fase. Iniciamos apresentando as definições e algumas técnicas fundamentais para o modelo de percolação (de elos ou de sítios) em um grafo infinito, conectado e localmente finito. Demonstramos então dois resultados essenciais: os fatos do parâmetro crítico não depender da escolha do vértice e da existência de um aglomerado infinito ter probabilidade 0 ou 1. Também obtemos um limitante inferior para o parâmetro crítico quando o grafo é de grau limitado. Para finalizar esta parte introdutória, analisamos a percolação em grafos particulares, a saber, a rede hipercúbica Z^d (para a qual mostramos a existência de transição de fase em dimensão d >= 2 e a unicidade do aglomerado infinito na fase supercrítica) e alguns tipos de árvores (para as quais apresentamos os parâmetros críticos). Na parte mais importante da dissertação, tendo como base os trabalhos de Benjamini e Schramm, de Häggström, Schonmann e Steif e de Lyons e Peres, introduzimos os conceitos de transitividade, amenabilidade e amenabilidade forte para um grafo. Fazemos uma detalhada discussão destas definições: provamos que a constante de Cheeger ancorada não depende do vértice em que é ancorada, estudamos relações entre os conceitos (amenabilidade e amenabilidade forte são noções distintas, bem como condições necessárias e suficientes para ambas) e calculamos a constante de Cheeger e a constante de Cheeger ancorada para alguns grafos. Finalmente, utilizando a técnica de crescimento do aglomerado, apresentamos para a probabilidade crítica um limitante superior que depende da constante ancorada. Isto nos permite concluir que ocorre transição de fase para qualquer grafo infinito, conectado, fracamente não-amenável (de constante de Cheeger ancorada positiva) e de grau limitado. / This work intends to study independent Bernoulli percolation model on graphs; the main purpose is obtaining conditions for phase transition. We begin presenting the definitions and some basic techniques for bond percolation and site percolation models on infinite, connected, locally finite graphs. We prove two essential results: the critical parameter is independent of the choice of the vertex and the probability that there exists an infinite cluster takes the values 0 and 1 only. We also obtain a lower bound for critical parameter when the graph is of bounded degree. To finish this preliminary part, we analyze percolation on particular graphs, namely the d-dimensional cubic lattice Z^d (for which we prove that there exists phase transition in dimension d >= 2 and the uniqueness of the infinite cluster in supercritical phase) and some trees (for which we present the critical parameters). In the most important part of this essay, founded in the works of Benjamini and Schramm, Häggström, Schonmann and Steif and Lyons and Peres, we introduce the concepts of transitivity, amenability and strong amenability. We discuss in detail these definitions: we prove that anchored Cheeger constant does not depend on the choice of the vertex, we study some relations (amenability and strong amenability are distinct notions, and necessary and sufficient conditions for both) and we obtain Cheeger constant and anchored Cheeger constant for some graphs. Finally, using the growing cluster technique, we present for the critical probability an upper bound that depends on the anchored constant. This permits us to conclude that there exists phase transition on infinite, connected, weakly non-amenable graphs of bounded degree.

Information on the sensitivity of explosives is highly valuable, and the short time scales in which chemical reactions occur in explosives, along with the ability of microstructure to have significant effects on sensitivity, often make this information difficult and expensive to acquire and interpret. Significant changes in impact and shock sensitivity are expected as a result of inducing structural damage in an explosive sample, and thermally damaged HMX-based samples can incur a solid-solid phase transition from beta to delta with non-extreme thermal inputs. Changes in sensitivity due to this phase transition, as well as the simultaneously induced damage, and their relative influence on sensitivity, are of interest to determine experimentally. <div><br></div><div>Drop-weight impact tests are a commonly used measure of explosive impact sensitivity. Often, simply the L50 of a given material is reported and compared with that of other materials to give a sense of its impact sensitivity. The practice of reporting the impact sensitivity as a single number, the L50, is likely inadequate. It is important to additionally report a measure of the spread of the distribution of reaction probabilities in order to assess the hazard of reaction in situations that may induce a stimulus level well below the L50 of a material. Additionally, multiple distribution forms have been suggested previously for fitting of binary sensitivity data; these distributions typically deviate from each other most near the tails (low and high stimulus levels). The consequences of choosing one distribution form over another in the analysis of explosive drop-weight impact results is explored.<br></div><div><br></div><div>Changes in impact sensitivity due to phase change have received some previous exploration, though the phase change influence is generally conflated with the induced damage upon said phase transition; however, sensitivity changes in the shock regime due to beta to delta-phase change have received little attention. Work is shown which includes methods to isolate variables of HMX phase transition and damage typically incurred upon said phase transition.<br></div>

Železo-rhodium (FeRh) je látka procházející magnetickou fázovou přeměnou prvního druhu z antiferomagnetické (AF) do feromagnetické (FM) fáze, ke které dochází při zahřátí materiálu nad teplotu fázové přeměny nebo působením dostatečně velkého magnetického pole. Tato fázová přeměna je mimo jiné provázena výraznou změnou entropie, magnetizace a elektrického odporu, přičemž její tvar a poloha teploty přeměny je silně závislá na stechiometrii krystalu, na příměsích, tlaku a v případě tenkých vrstev na napjatosti vrstvy způsobené substrátem. Tato práce se zaměřuje na studium magnetotransportních vlastností drátů připravených z tenkých FeRh vrstev rostlých na substrátech indukujících různou napjatost vrstvy. Jedním z hlavních jevů studovaných v této práci je anizotropní magnetorezistance (AMR) projevující se změnou odporu pro různé natočení magnetických momentů v látce vůči směru elektrického proudu. AMR byla studována jak ve FM fázi, tak i v AF fázi FeRh. Byla změřena hodnota AMR ve vysokoteplotní FM fázi a objeveno neočekávané chování AMR ve zbytkové FM fázi v nízkoteplotním stavu. Dále byla pozorována výrazná závislost AMR na orientaci měřených segmentů vůči krystalografickým směrům FeRh.