La circulation de matière et la signature sismique des zones de subduction sont fortement dépendantes despropriétés élastiques de deux grands types de minéraux hydratés : les serpentines (antigorite, lizardite etchrysotile) provenant de l'hydratation des minéraux du manteau et le glaucophane, minéral marqueur des schistesbleus et éclogites, faciès métamorphiques caractéristiques des zones de subduction. La détection de ces phasesest parfois difficile. Il est nécessaire de connaître leurs propriétés élastiques afin de mieux comprendre lesimages sismiques acquises. L'objectif de cette thèse est de mesurer et de calculer les propriétés élastiques deminéraux hydratés par la spectroscopie Brillouin (dans le cas de l'antigorite et du glaucophane) et par les calculsab initio (pour la lizardite). Caractériser au mieux ces propriétés permet de les relier à la sismicité dans les zonesde subduction et notamment aux zones de faibles vitesses détectées par diverses méthodes sismiques. Nosmesures ont été réalisées à l'ambiante pour l'antigorite et le glaucophane et à haute pression pour l'antigorite encellule à enclumes de diamants. Les mesures élastiques ont ensuite été couplées à des mesures d'orientationspréférentielles par Electron Back-Scattered Diffraction. Ceci a permis de quantifier l'anisotropie élevée(AVP=37% et AVS=50%) dans la roche totale ainsi que les vitesses sismiques faibles des ondes P et S. Nousavons pu ainsi relier le décalage des ondes S aux retards observés par sismique dans la zone de subduction deRyukyu (Japon). Les observations sismiques montrent que la serpentine est présente dans les zones où lasismicité est faible et apparaît non seulement comme un minéral essentiel des zones de subduction mais en pluscomme un "lubrifiant" permettant aux couches de glisser les unes sur les autres sans engendrer de séismes. Pourle glaucophane, les schistes bleus présentent une anisotropie plus élevée que les éclogites à glaucophane maisces roches sont toutefois difficiles à détecter avec la profondeur, du fait de leurs vitesses élevées comparables àcelles du manteau environnant. Nos calculs par méthode ab initio portent sur un analogue de l'antigorite, lalizardite pour laquelle nous avons établi les constantes élastiques à diverses pressions et en présence de fer ounon. Nous avons mis en évidence une anomalie élastique vers 5 GPa pour la lizardite et 7 GPa pour l'antigorite,que nous avons confirmée ensuite par des mesures de spectroscopies Brillouin et Raman à haute pression. / The flow of material and the seismic signature of subduction zones are highly dependent on the elastic propertiesof two major types of hydrated minerals: serpentines (antigorite, lizardite and chrysotile) produced by thehydration of mantle minerals and glaucophane, a marker of blueschists and eclogites, which are metamorphicfacies characteristic of subduction zones. Detection of these phases is sometime difficult. It is important to knowtheir elastic properties in order to better understand the seismic images. The goal of this work is to measure andcalculate the elastic properties of hydrated minerals by Brillouin spectroscopy (for antigorite and glaucophane)and ab initio calculations (for lizardite). The precise knowledge of such properties allows linking them to theseismicity in the subduction zones including areas of low velocities detected by various seismic methods. Ourmeasurements were performed at room conditions for antigorite and glaucophane and at high-pressure forantigorite in a diamond anvil cell. Elastic measurements were then coupled with measurements of latticepreferredorientations by Electron Back-Scattered Diffraction. This allowed quantifying the high anisotropy(AVP= 37% and AVS= 50%) in the whole rock and the low seismic velocities of P and S waves. We were able tolink it to the shear wave splitting observed by seismology in the Ryukyu arc (Japan). Seismic observations showthat serpentine is present in areas of low seismicity; it appears to be not only an essential mineral of thesubduction zones but also a “lubricant” allowing sliding layers to slip over each other without leading toearthquakes. For glaucophane, blueschists exhibit a higher anisotropy than glaucophane eclogites, but theserocks are difficult to detect at higher depth, because of their high velocities comparable to those of thesurrounding mantle. We performed ab initio calculations for a similar serpentine, the lizardite, for which weestablished the elastic constants at various pressures and in the presence of iron or not. We highlighted ananomaly around 5 GPa for lizardite and 7 GPa for antigorite, which was later confirmed by Brillouin and Ramanspectroscopies at high pressure.

Nous avons réalisé des calculs de structure électronique ab initio, basés sur la théorie de lafonctionnelle de la densité (DFT), pour étudier les propriétés des dislocations vis h111i dansles métaux de transition cubiques centrés (V, Nb, Ta, Mo, W et Fe). Dans tous ces éléments,le coeur facile non-dégénéré est la configuration d’énergie minimale et la configuration de coeurdissociée a une énergie très élevée, comparable ou plus élevée que celle du coeur difficile, encontradiction avec les prédictions des potentiels interatomiques. Nous avons mis en évidence destendances de groupe marquées sur l’énergie de coeur de la dislocation facile, reliées à la positiondu niveau de Fermi par rapport au minimum du pseudo-gap de la densité d’états électroniques.Notre travail fait aussi apparaitre un comportement atypique du fer, avec une énergie relativedu coeur difficile basse, proche de celle du point col entre deux coeurs faciles, conduisant à unpotentiel de Peierls plat autour de la configuration difficile, contrairement aux autres éléments.A partir de ces calculs DFT, nous avons construit le paysage énergétique à deux dimensionsdans le plan {111} (potentiel de Peierls) et nous avons étudié plusieurs propriétés relativesau glissement des dislocations, et en particulier l’énergie de formation de la paire de décrochementset la dépendance de la contrainte de Peierls en fonction de l’orientation cristalline.Nous proposons une modification simple de la loi de Schmid, qui prend en compte la trajectoirenon rectiligne de la dislocation et qui permet d’expliquer qualitativement pourquoi l’asymétriemaclage/antimaclage est moins marquée dans Fe que dans les autres métaux cubiques centrés. / We performed electronic structure ab initio calculations based on density functional theory(DFT) to study the h111i screw dislocation properties in body-centered cubic transition metals(V, Nb, Ta, Mo, W and Fe). In all investigated elements, the nondegenerate easy coreis the minimum energy configuration and the split core configuration has a high energy nearor above that of the hard core, contrary to interatomic potential predictions. A strong groupdependence of the core energy of the easy dislocation is also evidenced, related to the positionof the Fermi level with respect to the minimum of the pseudogap of the electronic density ofstates. Our work also reveals an atypical behavior in Fe, with a low relative energy at the hardcore position, close to that of the saddle configuration between easy cores, resulting in a flatPeierls potential around the hard core configuration, at variance with other elements. Fromthese DFT calculations, the two-dimensional energetic landscape in the {111} plane (Peierlspotential) is constructed and we investigated several properties of dislocation glide and in particular,the kink-pair formation enthalpy, as well as the dependence of the Peierls stress oncrystal orientation. We proposed a simple modification to the Schmid law that takes accountof the non-straight trajectory of the dislocation and that qualitatively explains why the twinning/antitwinning asymmetry is less pronounced in Fe than in other body-centered cubic metals.

Desde sua descoberta por S. Iijima em 1991, os nanotubos de carbono têm sido considerados um dos materiais nanoestruturados mais promissores para o desenvolvimento de novos dispositivos eletrônicos em escala nanoscópica. Devido _a sua alta razão entre a área superficial e o volume, esse material se destaca para aplicações como sensores de gás. No presente trabalho, estudamos através de simulações computacionais, a possibilidade de nanotubos de carbono com defeitos de nitrogênio (os chamados nanotubos CNx), poderem ser usados como sensores de moléculas gasosas. Na primeira parte do trabalho foram realizados cálculos de estrutura eletrônica baseados na Teoria do Funcional da Densidade (DFT) para diferentes sistemas formados pelo nanotubo e pela molécula. Através de cálculos de energia de ligação, foi possível identificar quais gases poderiam ou não serem adsorvidos à superfície do nanotubo. Dentre as moléculas investigadas, o monóxido de carbono e a amônia mostraram ser as mais facilmente adsorvidas ao nanotubo. Na segunda parte, foram realizados cálculos das propriedades de transporte utilizando o formalismo das funções de Green fora do equilíbrio (NEGF) recursivo. Foi possível concluir que os nanotubos estudados poderiam ser usados para detectar o monóxido de carbono e a amônia. Todavia, em relação à seletividade, os resultados indicaram que não parece possível distinguir essas duas moléculas, caso o sistema fosse inserido em um ambiente contendo uma mistura desses gases. Ainda, foram feitas simulações de nanotubos contendo defeitos aleatoriamente distribuídos, de forma a levar em conta os fatores de desordem característicos de sistemas mais realistas. / Since their discovery by S. Iijima in 1991, carbon nanotubes have been considered as one of the most promising nanostructured materials for the development of new nanoscopic electronic devices. Due to its high surface area to volume ratio, this material stands out as a candidate for possible gas sensoring applications. In this thesis, we have studied, by means of computational simulations, the possibility of using carbon nanotubes containing nitrogen defects (the so-called CNx nanotubes) as gas sensors. In the first part, we have performed electronic structure calculations based on Density Functional Theory (DFT) of several systems to address the possible binding of different molecules to the nanotube surface. Our results indicate that, among the molecules which were investigated, carbon monoxide and ammonia adsorb more easily to the nanotube surface. In the second part of this thesis, we have performed calculations of the transport properties by means of non-equilibrium Green\'s function formalism (NEGF). The results have shown that the nitrogen-defect carbon nanotubes could be used to detect, mainly carbon monoxide and ammonia molecules. On the other hand, when dealing with the selectivity of this system, it seems to be not possible to distinguish these gases, in the case of inserting the system in a environment containing a mixture of these molecules. Finally, we have simulated carbon nanotubes with defects randomly distributed along its length, in order to take into account disordering factors usually found in more realistic nanosensors.

Este trabalho é dividido em duas partes. Na primeira parte são estudadas moléculas isoladas, na segunda, moléculas em meio solvente. Em sua primeira parte, métodos de mecânica quântica de alto nível de correlação são utilizados para caracterizar o estado fundamental da molécula de CaC. Propriedades elétricas e óticas desta molécula, como também da molécula de MgH, são obtidas usando-se o método de campo finito. O estudo dessas moléculas é de interesse para a Astrofísica. A segunda parte deste trabalho foca o estudo teórico do solvatocromismo da molécula de 4-[(4-dimethylamino)phenyl)imino]-2, 5-cyclohexadien-1-one, ou DIA (Dimethylaminoindoaniline). Esta molécula apresenta também uma grande variação da sua primeira hiperpolarizabilidade medida em solução. Sabe-se que ela existe em solução como uma combinação de duas formas extremas: a forma com separação de cargas eno e a forma neutra ceto. Após otimizarmos teoricamente as estruturas dessas duas formas, realizamos uma simulação Monte Carlo em um número de moléculas de solventes polares e apoiares. A análise da estrutura do solvente, em torno de cada soluto, é realizada usando-se funções de distribuição radial de pares. É feita, também, a análise das pontes de hidrogênio para cada estrutura eno ou ceto. Estruturas supermoleculares, envolvendo o soluto com as moléculas mais próximas e, em separado. com as moléculas ligadas por pontes de hidrogênio, são usadas como entrada para cálculos semi-empíricos de mecânica quântica da primeira transição eletrônica de absorção e da primeira hiperpolarizabilidade. Apenas superestruturas estatisticamente descorrelacionadas foram usadas no cálculo. Os resultados são analisados e comparados com resultados experimentais em solução e com resultados teóricos em fase gasosa. Todas as interpretações sugerem que a estrutura eno é responsável pelos espectros experimentais de absorção UV-visível, tanto em solvente polares quanto em apoiares. A estrutura eno tem também uma primeira hiperpolarizabilidade calculada que concorda adequadamente com aquelas observadas em alguns solventes. Para clorofórmio a primeira hiperpolarizabilidade do DIA é dez vezes maior que em outros solventes. / This work is divided in two parts. In the first part isolated molecules are studied, in the second part the interest is in molecules in solvent media. In the first part of this work, high level quantum mechanical correlation methods are employed to characterise the ground state of the CaC molecule. Electrical and optical properlies of this molecule and also of the MgH molecule are obtained using the finite field method. The study of these molecules is of interest to Astrophysics. The second part of this work focuses on the theoretical study of the solvatochromism of the 4-[(4-dimelhylamino)phenyl)imino]-2, 5-cyclohexadien-1-one molecule, or DIA (Dimethylaminoindoaniline). This molecule also presents a large variation of its measured first hyperpolarizability in solution. It is known that in solution this molecule exists in a combination of two extreme forms, the chargeseparated eno form and the neutral kelo form. After these two structures are theoretically optimised, a Monte Carlo simulation is performed in a number of polar and non-polar solvent molecules. Analysis of1 the solvent structure around each solute is performed using radial distribution function. Hydrogen bonding analysis for each eno or keto structure is also carried out. Supermolecular structures involving the solute and the nearest solvent molecules and, separately, the hydrogen bonded molecules, are used as input to quantum mechanical semi-empirical calculations of the first electronic absorption transition and the first hyperpolarizability. Only statistically non-correlated superstructures were used in the calculation. The results are analysed and compared to experimental results in solution and theoretical results in gas-phase. All interpretations suggest that the eno structure is responsible for the experimental UV-Visible absorption spectrum, both in polar and non-polar solvents. The eno structure has also a calculated first hyperpolarizability that agrees well with that observed for some solvents. For chloroform the first hyperpolarizability of DIA is ten times greater than in other solvents.

Este documento de Tese é o resultado de um trabalho de pesquisa voltado à análise cristalográfica de novas formas sólidas cristalinas derivadas de fármacos antirretrovirais, diante do contexto da engenharia de cristais para o desenho das novas formas sólidas, e principalmente diante da ótica da análise das densidades de carga, o que permitiu um entendimento mais acurado da estrutura eletrônica molecular desta classe de compostos. Compostos farmacêuticos antirretrovirais do tipo inibidores nucleosídeos da transcriptase reversa (INTRs), são de grande importância, uma vez que são amplamente usados na terapêutica antirretroviral, principalmente contra o vírus HIV. Nesse contexto, são conhecidos alguns problemas associados na manufatura destes fármacos, principalmente aos processos de extração e purificação dos fármacos Lamivudina (3TC) e Emtricitabina (FTC). Diante desta problemática, a engenharia de cristais fornece uma solução, mediante o planejamento racional de formas sólidas (sais, cocristais, solvatos, polimorfos, etc.) que apresentam maior estabilidade e facilitem principalmente o processo de purificação em grande escala. Daí surge a importância de estudar a estrutura molecular das diferentes formas sólidas derivadas destes fármacos, sendo uma das principais técnicas para este estudo a difração de raios X em monocristais (DRXM). Neste trabalho um total de nove novas formas sólidas foram avaliadas e reportadas, com uma discussão detalhada das conformações moleculares e supramoleculares. Entretanto, é realizada uma análise das densidades de carga mediante métodos experimentais, uma vez que foram conduzidos experimentos de DRXM em alta resolução, em virtude da boa qualidade dos cristais que algumas das formas sólidas apresentaram. Desta maneira foi possível propor modelos de densidade de carga experimentais construídos mediante o formalismo de Hansen & Coppens, utilizando refinamento por mínimos quadrados baseados nos dados de difração em alta resolução. Por último, com o intuito de ter um estudo mais completo e detalhado da estrutura eletrônica, foram realizados cálculos teóricos de primeiros princípios em condições gasosas e periódicas de contorno. Desta forma, é apresentada uma sinergia entre os resultados obtidos pelas análises das distribuições de densidade de cargas de algumas formas sólidas, com os resultados gerais da engenharia de cristais e, portanto, concluir e extrapolar alguns aspectos importantes, principalmente no que se refere às energias associadas com as interações intermoleculares. A sinergia dos estudos de engenharia de cristais e de densidade de carga, é um tipo de pesquisa pouco publicada dentro da área da cristalografia de pequenas moléculas. / This Thesis is the result of the research proposal aimed to the crystallographic analysis of new crystalline solid forms derived from antiretroviral drugs, in the context of the crystal engineering for the design of the new solid forms, mainly since the viewpoint of the charge density analysis, which allowed an accurate comprehension of the molecular electronic structure of this kind of compounds. Antiretroviral drug compounds of nucleoside analog reverse-transcriptase inhibitors (NRTI) type, are of great importance once they are large used in the antiretroviral therapy, mainly against the HIV. In this context, some problems are known regard to the manufacture process of these drugs, mainly in the extraction and purification procedures of the lamivudine (3TC) and emtricitabine (FTC) drugs. On this issue, the crystal engineering provides an answer, through the rational planning of solid forms (salts, cocrystals, solvates, polymorphs, etc.) that exhibit an increased stability and facilitate mainly the large-scale purification process. Hence is important to study the molecular structure of the diverse solid forms derived from these drugs, mainly through the single-crystal X-ray diffraction (SCXD) experiments. In this research a total of nine new solid forms were assessed and reported, along with a detailed discussion of the molecular and supramolecular conformations. Meantime, it was carried out an analysis of the experimental charge density, once it was performed high-resolution SCXD experiments, since some of the solid forms showed good quality single crystals. In this way, it was possible to propose models of experimental charge density through the Hansen & Coppens formalism, using least-square refinement against high-resolution X-ray diffraction data. Finally, with the aim to have a more complete and detailed study of the electronic structure, it was also carried out first principles theoretical calculations in gas-phase and periodic boundary conditions. Thus, it is shown a synergy between the results obtained by the analysis of the charge density distributions of some solid forms and the crystal engineering results and, therefore, to conclude and to extrapolate some important aspects, mainly involved with the intermolecular interaction energies. The synergy of the crystal engineering and charge density studies is a kind of research little published, within the small molecule crystallography area.

Este projeto tem dois objetivos. Primeiramente estudou-se a aplicabilidade dos splines tricúbicos para a construção de superfícies de energia potencial globais. Um dos obstáculos que este método tem de superar e a escolha de um sistema de coordenadas apropriado, que minimize a influência de pontos não físicos. Para isto, propôs-se o uso do sistema de coordenadas de Pekeris, nunca usado para este fim. Este procedimento foi realizado para três sistemas químicos bem descritos na literatura, [Cl,H2], [F,H,D] e [H,O,Cl], cujas superfícies de energia potencial e propriedades das reações foram usadas como referência. Com base nestes modelos, aplicamos o método proposto variando-se a quantidade e a disposição dos nós das interpolações, a fim de verificar sua influência na qualidade das superfícies interpoladas. Os resultados mostram que as superfícies construídas por este método reproduzem muito bem os cálculos de dinâmica química, tanto por métodos quânticos quanto por métodos clássicos. Para isto, os nós da interpolação devem cobrir as regiões mais importantes da superfície de energia potencial e os valores mais baixos das coordenadas de Pekeris devem ser priorizados. O segundo objetivo consiste na aplicação deste procedimento na construção da superfície de energia potencial [H,S,F]. Com esta superfície, diversas características deste sistema foram analisadas, tais como geometrias dos pontos estacionários, energias relativas e frequências vibracionais. Os valores obtidos estão de acordo com os dados descritos na literatura. A superfície construída também foi usada para a realização de cálculos de dinâmica para a reação F+HS → S+FH. Observamos a existência de dois tipos de mecanismos, um com a formação de um intermediário de longa duração e outro com a abstração direta do átomo de hidrogênio. / This project has two goals. First, we studied the applicability of the tricubic splines to construct global potential energy surfaces. One of the diculties this approach has to overcome is the choice of an appropriate coordinate system that minimises the in uence of non-physical points. For such, we proposed the use of the Pekeris coordinate system, never employed for this purpose. This procedure was carried out for three well described systems, [Cl,H2], [F,H,D] and [H,O,Cl], whose potential energy surfaces and reaction properties were taken as references. Based on these models, we applied the proposed method varying the amount and arrangement of the interpolation knots, to verify their influence on the quality of the interpolated surfaces. The results showed that surfaces constructed by this approach reproduce very well the chemical dynamics calculations, both for the quantum as well as for the classical methods, provided that the interpolation knots cover the most important regions of the potential energy surfaces, and the lower values of the Pekeris coordinates are prioritised. The second goal was the application of this procedure to the construction of the [H,S,F] potential energy surface. With this surface, several characteristics of this system were analysed, such as the geometry of the stationary points, relative energies and vibrational frequencies. The values obtained are in agreement with the data described in the literature. The constructed surface was also used for quantum dynamics calculations on the reaction F + HS → S + FH. We observed two kinds of mechanisms, one of them with the formation of a long-living intermediate and the other with the direct abstraction of the hydrogen atom.

Neste trabalho de mestrado apresentamos um estudo sobre a manifestação do efeito Rashba em isolantes topológicos na ausência de simetria de inversão estrutural. Os cálculos das propriedades atomísticas, energéticas e as estruturas eletrônicas são abordados através de métodos de primeiros princípios baseados na teoria do funcional da densidade. E seus resultados foram utilizados para o desenvolvimento de hamiltoniana efetiva baseado no modelo de Zhang. Realizamos o estudo de dois sistemas: 1) Bi$_2$Se$_3$ com átomos de Sn depositados na superfície: Este sistema pode ser entendido através da manifestação do efeito Rashba sobre um isolante topológico dada a quebra de simetria de inversão estrutural. Para um sítio de deposição específico, os átomos de Sn causam uma reconstrução da superfície e um terceiro cone de Dirac é observado na estrutura eletrônica. Este terceiro cone é não localizado na superfície e pode ser entendido como a manifestação do efeito Rashba. 2) PbBiI: Reportado aqui como um novo isolante topológico 2D com efeito Rashba. Descobrimos este sistema por um estudo sistemático sobre uma família de materiais formados por átomos tipo IV, V, e VII, cuja estrutura cristalina é hexagonal e não centrossimétrica. Mostramos que o PbBiI possui: i) Estabilidade mecânica, ii) Spin-splitting Rashba de 60 meV, iii) um gap de energia não trivial de 0.14 eV, iv) retroespalhamento proibido entre os estados de borda e v) retroespalhamento proibido entre os estados do bulk no entorno do nível de Fermi. Estas propriedades fazem do PbBiI um candidato para construção de dispositivos de spintrônica que atenua a perda de energia. / In this work, were studied the Rashba effect in topological insulators without structural inversion symmetry. We performed a first principles study based on density functional theory to calculate the atomistic properties, formation energy and electronic structure. These results were used to development a effective Hamiltonian based on Zhang model. They were studied two systems: 1) Bi$_2$Se$_3$ with Sn atoms deposited on the surface: This system can be seen as the Rashba effect manifestation on a topological insulator due to the structural inversion symmetry breaking. For a specific deposition site, the Sn atoms cause a reconstruction of the surface and display a third Dirac cone in the electronic structure. This third cone is not located on the surface and can be understood as the giant Rashba effect manifestation. 2) We propose a new non-centrosymmetric honeycomb-lattice QSH insulator family formed by the IV, V, and VII elements. The system formed by Bi, Pb and I atoms is reported here as a new 2D topological insulator with Rashba effect. We show that this system has: i) Mechanical stability, ii) spin-splitting Rashba of 60 meV, iii) nontrivial energy gap of 0.14 eV, iv) backscattering forbidden for both edge and bulk conductivity channels in the nanoribbon band structure. These properties make PbBiI a good candidate to construct spintronic devices with less energy loss.

Les piézoélectriques à haute performance sont des composants clés pour les dispositifs agiles. Il a été démontré récemment que les solutions solides (Ba,Ca)(Ti,Zr)O3 (BCTZ) présentent des propriétés électromécaniques prometteuses. Cependant, les mécanismes microscopiques conduisant à de telles caractéristiques restent à éclaircir, et les investigations théoriques de BCTZ demeurent très limitées à ce jour. En conséquence, cette thèse propose d’étudier les propriétés de différentes compositions de solutions solides (Ba,Ca)TiO3-Ba(Ti,Zr)O3 au moyen de calculs de premiers principes, en mettant l’accent sur la dynamiquedu réseau et sur la compétition entre différentes phases ferroélectriques. Nous nous intéressons d’abord aux quatre composés parents BaTiO3, CaTiO3, BaZrO3 et CaZrO3, afin de comparer leurs propriétés et leurs différentes tendances à la ferroélectricité. Ensuite, le coeur de notre étude est une caractérisation systématique des systèmes binaires (Ba,Ca)TiO3 et Ba(Ti,Zr)O3 en utilisant à la fois l’approximation du cristal virtuel (VCA) et des calculs directs sur supercellules.Lorsqu’on passe continument de BaTiO3 à CaTiO3 dans (Ba,Ca)TiO3, la caractéristique principale est une transformation progressive de la ferroélectricité de type B en type A en raison d’effets stériques, lesquels déterminent en grande partie le comportement du système. En particulier, pour les petites concentrations en calcium, nous avons mis en évidence que laferroélectricité guidée par le site B est globalement affaiblie, conduisant à la disparition de la barrière d’énergie entre différents états polaires et à une polarisation quasi-isotrope. Une amélioration considérable de la réponse piézoélectrique résulte de ces caractéristiques. En passant de BaTiO3 à BaZrO3 dans Ba(Ti,Zr)O3, en revanche, le comportement est dominé parles mouvements coopératifs Zr-Ti et l’électrostatique locale. En particulier, la phase R3m est stabilisée significativement pour les faibles concentrations en zirconium. Sous l’effet d’une augmentation de la concentration en zirconium, le système montre une tendance à la réduction de la distorsion polaire; néanmoins, la ferroélectricité peut être préservée localement dans les régionsriches en titane. Grâce à un modèle électrostatique basé sur un super-réseau BaZrO3/mBaTiO3, nous avons également découvert une activation polaire inattendue pour Zr, en fonction d’un ordre atomique spécifique. Un facteur microscopique expliquant la réponse piézoélectrique exaltée dans BCTZ, pour de faibles concentrations en Ca et Zr, peut donc résider dans l’interaction entre la ferroélectricité affaiblie induite par Ti et la ferroélectricité émergente induite par Ca, interaction produisant une anisotropie minimale pour la polarisation. En outre, notre étude comparative révèle que la physique microscopique spécifique de ces solutions solides limite sévèrement l’applicabilité de l’approximation du cristal virtuel (VCA) à ces systèmes. / High-performance piezoelectrics are key components of various smart devicesand, recently, it has been discovered that (Ba,Ca)(Ti,Zr)O3 (BCTZ) solid solutions show appealingelectromechanical properties. Nevertheless, the microscopic mechanisms leading to suchfeatures are still unclear and theoretical investigations of BCTZ remain very limited. Accordingly,this thesis analyzes the properties of various compositions of (Ba,Ca)TiO3-Ba(Ti,Zr)O3solid solutions by means of first-principles calculations, with a focus on the lattice dynamicsand the competition between different ferroelectric phases. We first analyze the four parentcompounds BaTiO3, CaTiO3, BaZrO3 and CaZrO3 in order to compare their properties andtheir different tendency towards ferroelectricity. Then, the core of our study is a systematiccharacterization of the binary systems (Ba,Ca)TiO3 and Ba(Ti,Zr)O3 within both the virtualcrystal approximation (VCA) and direct supercell calculations. When going from BaTiO3 toCaTiO3 in (Ba,Ca)TiO3, the main feature is a gradual transformation from B-type to A-typeferroelectricity due to steric effects that largely determine the behavior of the system. In particular,for low Ca-concentration we found out an overall weakened B-driven ferroelectricitythat produces the vanishing of the energy barrier between different polar states and resultsin a quasi-isotropic polarization. A sizable enhancement of the piezoelectric response resultsfrom these features. When going from BaTiO3 to BaZrO3 in Ba(Ti,Zr)O3, in contrast, thebehavior is dominated by cooperative Zr-Ti motions and the local electrostatics. In particular,low Zr-concentration produces the further stabilization of the R3m-phase. Then, the systemshows the tendency to globally reduce the polar distortion with increasing Zr-concentration.Nevertheless, ferroelectricity can be locally preserved in Ti-rich regions. We also found out anunexpected polar activation of Zr as a function of specific atomic ordering explained via a basicelectrostatic model based on BaZrO3/mBaTiO3 superlattice. A microscopic factor behind theenhanced piezoelectric response in BCTZ, at low concentration of Ca and Zr, can thus be theinterplay between weakened Ti-driven and emerging Ca-driven ferroelectricity, which producesminimal anisotropy for the polarization. In addition, our comparative study reveals that thespecific microscopic physics of these solid solutions sets severe limits to the applicability of thevirtual crystal approximation (VCA) for these systems.

En chimie quantique, le principe variationnel est largement utilisé pour calculer la limite supérieure de l'énergie exacte d'un système atomique ou moléculaire. Des méthodes pour calculer la valeur limite inférieure de l'énergie existent mais sont bien moins connues. Une méthode précise pour calculer une telle limite inférieure permettrait de fournir une barre d'erreur théorique pour toute méthode de chimie quantique. Nous avons appliqué des méthodes de type variance pour calculer différentes énergies limites inférieures de l'atome d'hydrogène en utilisant des fonctions de base gaussiennes. L'énergie limite supérieure se trouve être toujours plus précise que ces différentes limites inférieures, i.e. plus proche de l'énergie exacte. L'importance de points singuliers sur l'évaluation de valeurs moyennes d'opérateurs quantiques a également été soulignée.Nous avons étudié les réactions d'adsorption d'un atome de fluor et d'un ion fluorure sur de petits agrégats de lithium Li$_n$ (n=2-15), à l'aide de méthodes de chimie quantique précises. Pour le plus petit système, nous avons montré que la formation de complexes stables Li$_2$F et Li$_2$F$^-$ se produit par un transfert d'électrons sans barrière et à longue portée, de Li$_2$ vers F pour le système neutre et l'inverse pour le système anionique. De telles réactions pourraient être rapides à très basse température. De plus, les complexes formés présentent des caractéristiques uniques de "longue liaison". Pour les systèmes plus gros Li$_n$F/Li$_n$F$^-$ ($n\geqslant4$), nous avons montré que les énergies d'adsorption peuvent être aussi grandes que 6~eV selon le site d'adsorption et que plus d'un état électronique est impliqué dans le processus d'adsorption. Les complexes formés présentent des propriétés intéressantes de "super alcalins" et pourraient servir d'unités de base dans la synthèse de composés à transfert de charge avec des propriétés ajustables. / In quantum chemistry, the variational principle is widely used to calculate an upper bound to the true energy of an atomic or molecular system. Methods for calculating the lower bound value to the energy exist but are much less known. An accurate method to calculate such a lower bound would allow to provide a theoretical error bar for any quantum chemistry method. We have applied variance-like methods to calculate different lower bound energies of a hydrogen atom using Gaussian basis functions. The upper bound energy is found to be always more accurate than the lower bound energies, i.e. closer to the exact energy. The importance of singular points on mean value evaluation of quantum operators has also been brought to attention.The adsorption reactions of atomic fluorine (F) and fluoride (F$^-$) on small lithium clusters Li$_n$ (n=2-15) have been investigated using accurate quantum chemistry ab initio methods. For the smallest system, we have shown that the formation of the stable Li$_2$F and Li$_2$F$^-$ complexes proceeds via a barrierless long-range electron transfer, from the Li$_2$ to F for the neutral and conversely from F$^-$ to Li$_2$ for the anionic system. Such reactions could be fast at very low temperature. Furthermore, the formed complexes show unique long bond characteristics. For the bigger Li$_n$F/Li$_n$F$^-$ systems ($n\geqslant 4$), we have shown that the adsorption energies can be as large as 6~eV depending on the adsorption site and that more than one electronic state is implied in the adsorption process. The formed complexes show interesting "superalkali" properties and could serve as building blocks in the synthesis of charge-transfer compounds with tunable properties.

<p>The main focus of this thesis is the investigation of the dissociation dynamics of aryl halides using femtosecond pump-probe spectroscopy. In the monohalogenated aryl halides, iodo-, bromo- and chlorobenzene, the rate of dissociation following excitation at 266 nm in the gas phase increased with increasing mass of the halogen atom. This process was assigned to predissociation of the initially excited singlet (π, π*) state via a repulsive triplet (n, σ*) state due to spin-orbit interaction. In addition to the predissociative mechanism, a direct dissociation channel was observed in iodobenzene. The rate of the predissociation in bromobenzene was found to be faster in the condensed phase than in the gas phase, which can be explained by solvent-induced symmetry perturbations. <i>Ab initio</i> calculations of the potential energy surfaces of the ground state and several low lying excited states in bromobenzene have been performed in order to verify the suggested mechanism. Substituting one of the hydrogen atoms in bromobenzene affected the predissociation rate significantly. In o-, m- and p-dibromobenzene the predissociation rate increased with decreasing distance between the bromine atoms in accordance with an increased spin-orbit interaction introduced by the bromine substituent. The fastest predissociation rate was observed in 1,3,5-tribromobenzene. With chlorine and fluorine substitution, inductive and conjugative effects were found to be of importance. In the o- and m-isomers of the dihalogenated aryl halides, an additional faster dissociation channel was observed. Guided by <i>ab initio</i> calculations of the potential energy surfaces in the dibromobenzene isomers, we ascribed the fast dissociation pathway to predissociation of an initially excited triplet state. Upon methyl group substitution in bromobenzene, the decreased lifetime of the initially excited state was attributed to an incresaed density of coupled states.</p><p>Another system which has been studied in the condensed phase is diiodomethane. Using Car-Parrinello molecular dynamics simulations we observed a prompt dissociation and subsequent recombination to the isomer, iso-diiodomethane, in acetonitrile solution.</p><p>Vibrational wavepacket dynamics in the C (¹Σ⁺) state of NaK were studied using a direct ionization probing scheme. A simple analytical expression for the pump-probe signal was developed in order to see what factors that govern direct ionization of the vibrational wavepacket. Our experimental data was consistent with a photoionization transition dipole moment that varies with internuclear distance.</p>