Submitted by Cássia Santos (cassia.bcufg@gmail.com) on 2014-07-31T12:16:56Z No. of bitstreams: 2 license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Calculo_da_condutividade_termica_do_argonio_solido_puro_e_com_defeito_pontual.pdf: 283748 bytes, checksum: d7815053104cf5341740be05c829feff (MD5) / Made available in DSpace on 2014-07-31T12:16:57Z (GMT). No. of bitstreams: 2 license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Calculo_da_condutividade_termica_do_argonio_solido_puro_e_com_defeito_pontual.pdf: 283748 bytes, checksum: d7815053104cf5341740be05c829feff (MD5) Previous issue date: 2008-03-14 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / In this work, using the Green-Kubo method combined with Molecular Dynamic (DM), we calculate the thermal conductivity of a solid Argon "free of defects"and with point defect present, for temperatures varying from 10 up to 60 K at density 22,3 ml/mol. The obtained results are in good agreement with the available theoretical and experimental results in the limites of low and high temperatures, but with some discrepances in about 15 % for intermediate values of temperatures. The purpose to include point defects with the objective of correction of the simulational results to compare with experimental measuremments for intermediate temperatues had not the expected e?ect. However, we believe that it should be due to the fact that the density used in the simulation for the point defect is high based on the experimental estimates of point defect density in this system. Our results suggest that the Green-Kubo method combined with Molecular Dynamics is a powerful tool to calculate the thermal conductivity of solids at high temperatures. With the construction of accurate and reliable interatomic potentials to describe more complex materials, such as high temperature ceramic and minerals at extreme condiction of pressure and temperature, this method could soon become very useful to calculate thermal conductivity in materials where the access to experimental data is hard. / Neste trabalho, usando o método de Green-Kubo combinado com a Dinâmica Molecular (DM), calculamos a condutividade têrmica do Argônio sólido livre de defeitos ;e com defeitos pontuais presentes, para um intervalo de temperatura variando de 10 a 60 K e uma densidade de 22,3 ml/mol. Os resultados obtidos estão em pleno acordo com os resultados teóricos e experimentais disponíveis nos limites de baixa e alta temperatura, mas com alguma discrepância em torno de 15 % para valores intermediários de temperatura. A proposta para incluir defeitos pontuais com o objetivo de correção dos resultados da simulação para comparar com as medidas experimentais para temperaturas intermediárias não surtiu o efeito esperado, no entanto, acreditamos que isto se deve ao fato da densidade de defeitos ser alta baseado em estimativas da densidade de defeitos neste sistema. Nossos resultados sugerem que o método de Green-Kubo combinado com DM é uma ferramenta poderosa para se calcular a condutividade térmica de sólidos a altas temperaturas. Com a construção de potenciais interatômicos mais precisos e con fiáveis para descrever materiais mais complexos, como é o caso de cerâmicas a altas temperaturas e minerais em condições extrema de pressão e temperatura, esse método poderá em breve ser muito útil para calcular a condutividade térmica em materiais onde o acesso a dados experimentais é mais difícil.

La conception des dispositifs nanométriques CMOS apporte de nouveaux défis à la communauté TCAD. En effet, de nos jours, les améliorations des performances des transistors ne sont plus simplement dû à une simple diminution des dimensions des dispositifs, mais aussi à l'introduction de boosters de technologies tels que des nouvelles architectures (FDSOI, trigate), des oxydes de grille à forte permittivité, l'ingénierie de la contrainte ou de nouveaux matériaux de canal (Ge, III-V). Pour faire face à tous ces nouveaux défis technologiques, la modélisation TCAD (Technology Computer Aided Design) est un outil puissant pour guider le développement mais aussi pour réduire les coûts. Dans ce contexte, ce travail de thèse vise à améliorer la modélisation TCAD pour les technologies 28/14 et 10FDSOI, avec une attention particulière sur les impacts des contraintes mécaniques sur leurs performances. Dans un premier temps, les différents mécanismes impactant la mobilité des technologies FDSOI ont été étudiés en détail. Les modèles implémentés dans des outils de simulations avancés (NEGF, Multi subbands Monte Carlo, Kubo-Greenwood) sont étudiés, comparés et des développements du logiciel interne à STMicroelectronics (UTOXPP) sont proposés. Dans un second temps, une approche « top down » a été mis en place. Elle consiste à calibrer les modèles TCAD empiriques non pas sur des mesures mais sur des outils de simulations avancés (Kubo-Greenwood). Les modèles TCAD calibrés montrent de très bons accords avec les mesures de mobilité (split-CV) en variant la température, la polarisation du substrat et l’épaisseur de l’IL (Interfacial layer). Dans un troisième temps, les méthodes utilisées lors de cette thèse pour modéliser les contraintes induites par le procédé de fabrications sont décrites. Enfin, la dernière partie concerne la modélisation TCAD des technologies 28 et 14FDSOI. Des simulations mécaniques sont effectuées pour modéliser les profils de contraintes dans les transistors. Des solutions pour optimiser la configuration des contraintes dans le canal pour ces technologies sont proposées. / The design of nanoscale CMOS devices brings new challenges to TCAD community. Indeed, nowadays, CMOS performances improvements are not simply due to device scaling but also to the introduction of new technology “boosters” such as new transistors architectures (FDSOI, trigate), high-k dielectric gate stacks, stress engineering or new channel material (Ge, III-V). To face all these new technological challenges, Technology Computer Aided Design (TCAD) is a powerful tool to guide the development of advanced technologies but also to reduce development time and cost. In this context, this PhD work aimed at improving the modeling for 28/14 and 10FDSOI technologies, with a particular attention on mechanical strain impacts. In the first section, a summary of the main models implemented in state of the art device simulators is performed. The limitations and assumptions of these models are highlighted and developments of the in-house STMicroelectronics KG solvers are discussed. In the second section, a “top down” approach has been set-up. It has consisted in using advanced physical-based solvers as a reference for TCAD empirical models calibration. Calibrated TCAD reproduced accurately split-CV mobility measurements varying the temperature, the back bias and the Interfacial Layer (IL) thickness. The third section deals with a description of the methodologies used during this thesis to model stress induced by the process flow. Simulations are compared to nanobeam diffraction (NBD) strain measurements. The use and calibration of available TCAD models to efficiently model the impact of stress on mobility in a large range of stress (up to 2GPa) is also discussed in this section. The last part deals with TCAD modeling of advanced CMOS devices for 28/14 and 10FDSOI technology development. Mechanical simulations are performed to model the stress profile in transistors and several solutions to optimize the stress configuration in sSOI and SiGe-based devices have been presented.

In dieser Arbeit wird ein Modell für das Eu-reiche EuO formuliert. Es besteht in einer Erweiterung des Kondo-Gitter-Modells (KGM). Für das KGM existieren nur einige exakte Aussagen. In dieser Arbeit kommt eine neue hinzu, nämlich die exakte Abbildung des periodischen Anderson-Modells auf das antiferromagnetische KGM für beliebige Kopplungsstärke J. Reines EuO ist ein ferromagnetischer Halbleiter. Eu-reiches EuO zeigt einen gewaltigen Metall-Isolator-Übergang in der Nähe der Curie-Temperatur mit einem Sprung im Widerstand von bis zu 13 Größenordnungen. Das ist der größte Sprung im Widerstand, der jemals in der Natur beobachtet wurde. Wir reproduzieren diesen Sprung theoretisch mit der Kubo-Formel. Wir erzielen sehr gute Fits bereits in einer nicht vollständig selbstkonsistenten Theorie, bei der die Magnetisierung der Eu-Spins einer Brillouin-Funktion entnommen ist. In einer vollständig selbstkonsistenten Theorie bestimmen wir die Magnetisierung, die Curie-Temperatur, den spezifischen Widerstand und andere Transporteigenschaften. Wir berechnen Größen wie die elektronische Wärmeleitfähigkeit und die Thermokraft, für die weniger experimentelle Daten zum Vergleich vorhanden sind. Nichtsdestoweniger erscheinen z.B. die Rechnungen für die thermische Leitfähigkeit vertrauenswürdig, da das Wiedemann-Franz-Verhältnis mit der elektrischen Leitfähigkeit einen vernünftigen Wert liefert. Die Leitungselektronenzahl des Eu-reichen EuO kommt aus der Theorie unabhängig von der Leitfähigkeit heraus. Daher können wir aus der Leitfähigkeit und der Leitungselektronenzahl die durchschnittliche Drude-Mobilität (oder Streuzeit) berechnen. Diese Größe hat für höhere Impurity-(Sauerstoff-Leerstellen)-Konzentrationen einen Sprung in der Nähe der Curie-Temperatur von bis zu zwei Größenordnungen in Übereinstimmung mit dem Experiment. / In this thesis a model is formulated for Eu-rich EuO. It consists in an extension of the Kondo lattice model (KLM). For the KLM only a few exact statements exist. To those we add a new one, namely the exact mapping of the periodic Anderson model on the antiferromagnetic KLM for arbitrary coupling constant J. Pure EuO is a ferromagnetic semiconductor. Eu-rich EuO exhibits a huge metal--insulator transition near the Curie temperature with a jump in resistivity of up to 13 orders of magnitude. It is the biggest jump in resistivity ever observed in nature. We theoretically reproduce this jump. We achieve very good fits already within a not fully self-consistent theory where the magnetization of the Eu spins is taken from a Brillouin function. In a fully self-consistent theory we determine the magnetization, the Curie temperature, the resistivity and other transport properties. We calculate quantities like the electronic thermal conductivity and the thermopower, for which there are less experimental data to compare with. Nevertheless, e.g. the calculations for the thermal conductivity seem reliable since the Wiedemann-Franz ratio with the electrical conductivity gives a reasonable result. The conduction-electron number of Eu-rich EuO comes out of the theory independently of the conductivity. So we can calculate from the conductivity and the conduction-electron number the average Drude mobility (or scattering time). This quantitiy has a jump near the Curie temperature of up to two orders of magnitude for higher impurity (oxygen vacancy) concentrations in agreement with the experiment.

L'équation de la chaleur est un phénomène macroscopique, émergeant après une limite d’échelle diffusive (en espace et en temps) d’un système d'oscillateurs couplés. Lorsque les interactions entre oscillateurs sont linéaires, l'énergie évolue de manière balistique, et la conductivité thermique est infinie. Certaines non-linéarités doivent donc apparaître au niveau microscopique, si l’on espère observer une diffusion normale. Pour apporter de l'ergodicité, on ajoute à la dynamique déterministe une perturbation stochastique qui conserve l'énergie. En premier lieu nous étudions la dynamique Hamiltonienne d'un système d'oscillateurs linéaires, perturbé par un bruit stochastique dégénéré conservatif. Ce dernier transforme à des temps aléatoires les vitesses en leurs opposées. On montre que l'évolution macroscopique du système est caractérisée par un système parabolique non-linéaire couplé pour les deux lois de conservation du modèle. Ensuite, nous supposons que les oscillateurs évoluent en environnement aléatoire. La perturbation stochastique est très dégénérée, et on prouve que le champ de fluctuations de l'énergie à l'équilibre converge vers un processus d'Ornstein-Uhlenbeck généralisé dirigé par l’équation de la chaleur.Il est désormais connu que les systèmes unidimensionnels présentent une diffusion anormale lorsque le moment total est conservé en plus de l'énergie. Dans une troisième partie, on considère deux perturbations, l'une préservant le moment, l'autre détruisant cette conservation. En faisant décroître l'intensité de la seconde perturbation, on observe une transition de phase entre un régime de diffusion normale et un régime de superdiffusion. / The heat equation is known to be a macroscopic phenomenon, emerging after a diffusive rescaling of space and time. In linear systems of interacting oscillators, the energy ballistically disperses and the thermal conductivity is infinite. Since the Fourier law is not valid for linear interactions, non-linearities in the microscopic dynamics are needed. In order to bring ergodicity to the system, we superpose a stochastic energy conserving perturbation to the underlying deterministic dynamics.In the first part we study the Hamiltonian dynamics of linear coupled oscillators, which are perturbed by a degenerate conservative stochastic noise. The latter flips the sign of the velocities at random times. The evolution yields two conservation laws (the energy and the length of the chain), and the macroscopic behavior is given by a non-linear parabolic system.Then, we suppose the harmonic oscillators to evolve in a random environment, in addition to be stochastically perturbed. The noise is very degenerate, and we prove a macroscopic behavior that holds at equilibrium: precisely, energy fluctuations at equilibrium evolve according to an infinite dimensional Ornstein-Uhlenbeck process driven by the linearized heat equation.Finally, anomalous behaviors have been observed for one-dimensional systems which preserve momentum in addition to the energy. In the third part, we consider two different perturbations, the first one preserving the momentum, and the second one destroying that new conservation law. When the intensity of the second noise is decreasing, we observe (in a suitable time scale) a phase transition between a regime of normal diffusion and a regime of super-diffusion.

Die zuverlässige Berechnung der elektrischen Leitfähigkeit vieler Materialien aus ersten Prinzipien erfordert die Berücksichtigung der anharmonischen Gitterdynamik. Der ab initio Kubo-Greenwood (KG)-Ansatz, der die KG-Leitfähigkeitsformel und die ab initio-Molekulardynamik kombiniert, scheint vielversprechend zu sein, da er die Anharmonizität des Gitters auf natürliche Weise berücksichtigt. Seine Anwendung auf kristalline Materialien hat jedoch bisher nur wenig Beachtung gefunden. Diese Arbeit beschreibt den KG-Ansatz und stellt eine numerische Implementierung dieses Ansatzes für den harmonischen Kristall Si und den anharmonischen Kristall SnSe vor. Die Fallstudie für Si zeigt erhebliche numerische Schwierigkeiten bei den KG-Berechnungen auf. Insbesondere behindert die erforderliche dichte k-Punkt-Abtastung die Konvergenz in Superzellengröße und macht die Berechnungen nur innerhalb der (semi-)lokalen Dichtefunktionaltheorie (DFT) durchführbar. Außerdem führt die notwendige Einführung eines Verbreiterungsparameters (η) zu einer erheblichen Unsicherheit bei der Bestimmung der Leitfähigkeit. Um diese Probleme zu lösen, werden rechnerisch effiziente Strategien diskutiert, darunter: (i) der "Scherenoperator"-Ansatz zur Korrektur des DFT-Bandlückenproblems; (ii) das "Optimal-η-Schema" zur Wahl eines geeigneten Wertes von η; und (iii) die Finite-Size-Scaling-Methode zur Ableitung der Leitfähigkeit in der thermodynamischen Grenze. Es wird festgestellt, dass die KG-Berechnungen mit diesen Strategien Leitfähigkeiten in angemessener Übereinstimmung mit den Experimenten ergeben. Der Vergleich mit früheren ab initio Boltzmann-Transportberechnungen zeigt jedoch, dass das η-Problem und die Frage der Konvergenz in Superzellengröße weiter verbesserte Konzepte erfordern. Die Fallstudie für SnSe zeigt sehr ähnliche numerische Schwierigkeiten wie im Fall von Si. Es werden Einblicke in die Auswirkung der Anharmonizität auf die Konvergenz der Superzellengröße gegeben. / Reliable first-principles calculation of the electrical conductivity in many materials requires accounting for the anharmonic lattice dynamics. The ab initio Kubo-Greenwood (KG) approach, which combines the KG conductivity formula and ab initio molecular dynamics, appears to be promising because it naturally includes lattice anharmonicity. However, its application to crystalline materials has so far received very little attention. This thesis describes the KG approach and presents a numerical implementation of this approach for the harmonic crystal Si and the anharmonic crystal SnSe. The case study for Si identifies considerable numerical difficulties in the KG calculations. In particular, the dense k-point sampling required hinders supercell-size convergence and makes the calculations only feasible within (semi)local density-functional theory (DFT). Besides, the necessary introduction of a broadening parameter (η) introduces a significant uncertainty in determining the conductivity. To address these issues, computationally efficient strategies are discussed, including: (i) the "scissor operator" approach to correct the DFT band-gap problem; (ii) the "optimal-η scheme" to choose an appropriate value of η; and (iii) the finite-size scaling method to deduce the conductivity in the thermodynamic limit. It is found that with these strategies, the KG calculations yield conductivities in reasonable agreement with experiment. Yet, comparison with previous ab initio Boltzmann transport calculations shows that the η problem and the issue of supercell-size convergence still require improved concepts. The case study for SnSe shows very similar numerical difficulties as in the case of Si. Insights into the effect of anharmonicity on the supercell-size convergence are provided.

Thesis (MSc)--Stellenbosch University, 2012. / ENGLISH ABSTRACT: The experimental observation of the quantum Hall effect in a two-dimensional electron gas posed an intriguing question to theorists: Why is the quantization of conductance so precise, given the imperfections of the measured samples? The question was answered a few years later, when a connection was uncovered between the quantum Hall effect and topological quantities associated with the band structure of the material in which it is observed. The Hall conductance was revealed to be an integer topological invariant, implying its robustness to certain perturbations. The topological theory went further than explaining only the usual integer quantum Hall effect in a perpendicular magnetic field. Soon it was realized that it also applies to certain systems in which the total magnetic flux is zero. Thus it is possible to have a quantized Hall effect without Landau levels. We study a carbon nanotube in a magnetic field perpendicular to its axial direction. Recent studies suggest that the application of an electric field parallel to the magnetic field would induce a gap in the electronic spectrum of a previously metallic carbon nanotube. Despite the quasi onedimensional nature of the carbon nanotube, the gapped state supports a quantum Hall effect and is associated with a non zero topological invariant. This result is revealed when an additional magnetic field is applied parallel to the axis of the carbon nanotube. If the flux due to this magnetic field is varied by one flux quantum, exactly one electron is transported between the ends of the carbon nanotube. / AFRIKAANSE OPSOMMING: Die eksperimentele waarneming van die kwantum Hall effek in ’n twee-dimensionele elektron gas laat ’n interessante vraag aan teoretiese fisikuste: Waarom sou die kwantisasie van die geleiding so presies wees al bevat die monsters, waarop die meetings gedoen word, onsuiwerhede? Hierdie vraag word ’n paar jaar later geantwoord toe ’n konneksie tussen die kwantum Hall effek en topologiese waardes, wat verband hou met die bandstruktuur van die monster, gemaak is. Dit is aan die lig gebring dat die Hall geleiding ’n heeltallige topologiese invariante is wat die robuustheid teen sekere steurings impliseer. Die topologiese teorie verduidelik nie net die gewone kwantum Hall effek wat in ’n loodregte magneetveld waargeneem word nie. Dit is ook moontlik om ’n kwantum Hall effek waar te neem in sekere sisteme waar die totale magneetvloed nul is. Dit is dus moontlik om ’n gekwantiseerde Hall effek sonder Landau levels te hˆe. Ons bestudeer ’n koolstofnanobuis in ’n magneetveld loodreg tot die aksiale rigting. Onlangse studies dui daarop dat die toepassing van ’n elektriese veld parallel aan die magneetveld ’n gaping in die elektroniese spektrum van ’n metaliese koolstofnanobuis induseer. Ten spyte van die een-dimensionele aard van die koolstofnanobuis ondersteun die gapings-toestand steeds ’n kwantum Hall effek en hou dit verband met ’n nie-nul topologiese invariante. Hierdie resultaat word openbaar wanneer ’n bykomende magneetveld parallel tot die as van die koolstofnanobuis toegedien word. Indien die vloed as gevolg van hierdie magneetveld met een vloedkwantum verander word, word presies een elektron tussen die twee kante van die koolstofnanobuis vervoer.

The aim of this work has been to use atomistic computer simulation methods to calculate the thermal conductivity and investigate factors that will modify the behaviour when applied to three different oxide materials: MgO, SiO2 and SrTiO3. These were chosen as they represent distinct classes of materials and are substrates for thermoelectric devices, where one of the primary goals is to tailor the system to reduce the thermal conductivity. Chapter 1 introduces thermoelectric concepts, gives a background of the theory and a review of various important thermoelectric materials. In Chapter 2 an overview of the interatomic interactions is presented along with details on the implementation of these interactions in a simulation of a 3D periodic crystal. Chapter 3 outlines the importance of phonon processes in crystals and several approaches to the calculation of thermal conductivity are presented. MgO results are given in Chapter 4. Both the Green-Kubo and Boltzmann transport equation (BTE) methods of calculating thermal conductivity were used. The effect on thermal conductivity of two different grain boundary systems are then compared and finally extended to MgO nanostructures, thus identifying the role of surfaces and complex nanostructure architectures on thermal conductivity. In Chapter 5 two different materials with the formula unit SiO2 are considered. The two materials are quartz and silicalite which show interesting negative thermal expansion behaviour which may impact upon the thermal transport within the material. Chapter 6 presents results on the promising thermoelectric material STO. Once again the results from both Green-Kubo and BTE calculations are compared. Grain boundaries are also studied and the effect of inter-boundary distance and boundary type on the thermal conductivity is explored. Finally, a nanostructured STO system (assembled nanocubes) with promising thermoelectric applications is studied. Chapter 7 outlines the conclusions made from this work and suggests areas for future study.

As the foundation for a new computational implementation, we survey the calculation of the complex electrical conductivity tensor based on the Kubo-Greenwood (KG) formalism (Kubo, 1957; Greenwood, 1958), with emphasis on derivations and technical aspects pertinent to use of projector augmented wave datasets with plane wave basis sets (BIlichl, 1994). New analytical results and a full implementation of the KG approach in an open-source Fortran 90 post-processing code for use with Quantum Espresso (Giannozzi et al., 2009) are presented. Named KGEC ([K]ubo [G]reenwood [E]lectronic [C]onductivity), the code calculates the full complex conductivity tensor (not just the average trace). It supports use of either the original KG formula or the popular one approximated in terms of a Dirac delta function. It provides both Gaussian and Lorentzian representations of the Dirac delta function (though the Lorentzian is preferable on basic grounds). KGEC provides decomposition of the conductivity into intra- and inter band contributions as well as degenerate state contributions. It calculates the dc conductivity tensor directly. It is MPI parallelized over k-points, bands, and plane waves, with an option to recover the plane wave processes for their use in band parallelization as well. It is designed to provide rapid convergence with respect to k-point density. Examples of its use are given.

京都大学 / 0048 / 新制・課程博士 / 博士(理学) / 甲第21551号 / 理博第4458号 / 新制||理||1640(附属図書館) / 京都大学大学院理学研究科物理学・宇宙物理学専攻 / (主査)教授 佐々 真一, 准教授 藤 定義, 准教授 荒木 武昭 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DGAM

京都大学 / 0048 / 新制・課程博士 / 博士(理学) / 甲第18816号 / 理博第4074号 / 新制||理||1586(附属図書館) / 31767 / 京都大学大学院理学研究科化学専攻 / (主査)教授 長谷川 健, 准教授 安藤 耕司, 教授 林 重彦 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DGAM