La signature spectroscopique des ions moléculaires est fondamentale pour l'étude et la caractérisation de plasma en astrophysique et en laboratoire. Différentes techniques peuvent être utilisées pour caractériser ces plasmas parmi lesquelles la spectroscopie de photoélectrons induits par rayons X et la photo-absorption X. L'objectif de cette thèse est la simulation des spectres de photo-absorption au seuil L (2p) des ions moléculaires de silicium SiHn+ (n= 1, 2, 3) et au seuil K (1s) des ions moléculaires de l'oxygène OHn+ (n=1, 2) et du carbone CHn+ (n= 1, 2) produits lors d'une décharge plasma. Nous avons développé différents protocoles numériques permettant de calculer les spectres d'absorption aux seuils K et L en combinant des méthodes de structures électroniques et de propagation de paquet d'ondes. Les optimisations de géométrie et le calcul des seuils d'ionisation (IP) sont obtenus en utilisant la théorie de la fonctionnelle de la densité (DFT). Les effets de relaxation électronique due à la formation d'un trou en couche interne sont pris en compte au niveau SCF (convergence de la fonction d'onde électronique à N-1 électrons). Les surfaces d'énergie potentielle (PES) et les moments de transition dipolaires sont calculés à un niveau Post-HF (Interaction de configurations, CI). L'introduction explicite du couplage spin-orbite à l'aide de l'opérateur Breit-Pauli est utilisée pour l'étude du processus d'excitation au seuil L du silicium. Les spectres théoriques calculés pour les différentes molécules étudiées présentent un accord raisonnable avec les mesures expérimentales. La présence d'états électroniques métastables produits lors de la décharge plasma est discutée. / Molecular ions cover important roles in study and characterization of astrophysical and laboratory plasma. To this purposes, different spectroscopic techniques are used among which we found the X-ray photoelectron spectroscopy and the X-ray photoabsorption spectroscopy. This PhD work is focused on the calculation of X-ray photoabsorption spectra of molecular ions of silicon, carbon and oxygen XHn+ (X= Si, C, O; n= 1, 2, 3). The former is excited on the L (2p) shell while the others on the K (1s) shell produced in plasma discharged. We developed numerical protocols which permits to compute with reasonable precision the K and L-shell photoabsorption spectra combining electronic structure and nuclear wavepacket propagation methods. The optimization of the geometries and the calculation of the ionization potentials (IP) are carried out using the density functional theory (DFT). The relaxation effects due to the core hole creation are taken into account at the self-consistent field (SCF) level. The potential energy surfaces (PES) and the dipole moment transitions are computed at the post Hartree-Fock (configuration interaction, CI) level. The spin-orbit coupling effect are explicitly taken into account through the Breit-Pauli operator. The theoretical results have been compared with the experimental data and they allow the interpretation of the experimental bands.

À la surface ou à l’interface d’un matériau ferroélectrique, la polarisation peut être déstabilisée voire même annulée par le champ dépolarisation qui résulte de charges de polarisation non compensées. En l’absence de mécanismes d’écrantage extrinsèques (adsorbats, électrodes) ou intrinsèques (défauts, dopants), l’ordonnancement en domaines est le moyen le plus naturel dont un système a recourt pour rester ferroélectrique et minimiser son énergie électrostatique. Cette thèse se concentre sur l’étude de la stabilité de multiples configurations en domaines, ainsi que de la façon dont elle dépend de facteurs géométriques, chimiques et élastiques, à la surface du BaTiO3(001).Des calculs ab initio, fondés sur la théorie de la fonctionnelle de la densité, ont été menés pour éclaircir l’influence de la terminaison de surface, de la contrainte parallèle à la surface, de l’épaisseur du système et de la taille des domaines sur la polarisation de différentes phases ferroélectriques dans des couches ultraminces. L’effet de lacunes d’oxygène en surface sur la polarisation et sur la structure électronique a aussi été étudié. La microscopie d’électrons lents (LEEM) et la microscopie d’électrons photoémis (PEEM) ont été utilisées pour étudier les propriétés de la polarisation à la surface de monocristaux de BaTiO3 réduit, afin d’examiner l’influence des lacunes d’oxygène sur la polarisation ferroélectrique par une approche expérimentale, renforçant ainsi les résultats de calculs. / At a surface or interface of a ferroelectric material, the polarization can be destabilized and even suppressed by the depolarizing field which arises from uncompensated polarization charge. In the absence of external (adsorbates, electrodes) or internal (defects, dopants) screening mechanisms, domain ordering is the most natural way for a system to remain ferroelectric and minimize its electrostatic energy. This thesis focuses on the study of the stability of various possible domain configurations, and on how it depends and interplays on size, chemical and elastic factors, at the (001) surface of BaTiO3.First-principles calculations, based on density functional theory, have been performed to enlighten the influence of surface terminations, in-plane strain, system thickness and domain size in ultrathin films on the polarization in different ferroelectric phases. The effect of surface oxygen vacancies on the polarization and electronic structure has also been investigated. Low Energy electron microscopy (LEEM) and photoemission electron microscopy (PEEM) have been used to study the surface properties of ferroelectric polarization in reduced BaTiO3 single crystals, allowing to address the influence of oxygen vacancies on the ferroelectric polarization from experimental approaches, supporting the calculations results.

Les ions lanthanide sont employés avec succès pour la synthèse de molécules-aimants caractérisées par une relaxation lente du moment magnétique d’origine purement moléculaire. Ces propriétés sont principalement liées à la forte anisotropie intrinsèque de ces ions, ainsi qu’à leur moment magnétique intense. Dans le cas de complexes contenant plusieurs porteurs de spin (métaux 3d, 4f ou radicaux organiques), la proximité de ces centres peut induire entre ces derniers une interaction de couplage aussi appelée échange magnétique. La rationalisation de telles propriétés est notamment rendue possible par l’approche ab initio. / Lanthanide ions have been used successfully in the synthesis of single molecule magnets for more than a decade. This particular class of molecules shows slow relaxation of their magnetization from purely molecular origin. This property come mainly from the strong single ion anisotropy of these ions, and from their high magnetic moment. In the case of complexes with more than one spin carrier (3d, 4f metal or organic radical), coupling interactions can arise. These are called magnetic exchange. In this framework, ab initio calculations are a useful tool for magneto-structural correlations.

Les travaux présentés dans ce manuscrit sont consacrés à l'étude des molécules d'éthylène C2H4 et de méthane CH4 pour des applications planétologiques et astrophysiques. La première partie de ce travail de thèse concerne la construction de surfaces ab initio (surface d'énergie potentielle et surfaces de moment dipolaire) pour l'éthylène. Une procédure permettant la détermination précise de la géométrie d'équilibre de cette molécule a été mise en place, car elle conditionne en partie la précision sur les niveaux rotationnels. Dans la seconde partie, ces surfaces ont été utilisés pour le calcul des niveaux d'énergie vibrationnels de l'éthylène et des isotopologues 12C2D4, 13C2H4 et 13C12CH4 par calcul variationnel, ainsi que l'étude des positions et intensités de raies pour les molécules 12C2H4 et 13C12CH4. Ces nouvelles surfaces ont permis des avancées significatives pour la prédiction variationnelle ab initio des spectres de l'éthylène. L'extension de ces calculs à l'étude de ces spectres à des températures élevées, pour des applications astrophysiques, a été pour la première fois abordée. La dernière partie concerne la construction non empirique d'un modèle effectif pour le moment dipolaire du méthane à partir de surfaces ab initio. Pour cela, une technique algébrique, basée sur la méthode des Transformations de Contact, et développée au sein de notre équipe pour les molécules triatomiques, a été pour la première fois étendue au cas des molécules penta-atomiques. L'ensemble de ces résultats a été confronté aux informations présentes dans les bases de données spectroscopiques, ce qui a permis de valider nos différentes approches. / The work presented in this manuscript is devoted to the study of ethyleneC2H4 and methane CH4 for planetological and astrophysical applications. The first part of this thesis concerns the construction of ab initio surfaces (potential energy surface and dipole moment surfaces ) for ethylene molecule. A procedure for the accurate determination of the equilibrium geometry in this molecule has been established, because it determines partially the accuracy of the rotational levels. In the second part, these surfaces were used to calculate the vibrational energy levels of ethylene and isotopologues 12C2D4, 13C2H4 and 13C12CH4 by variational calculations, and the study of the line positions and line intensities for 12C2H4 and 13C12CH4 molecules. These new surfaces have led to significant advances in ab initio variational predictions of ethylene spectra. For the first time, the extension of these calculations to the study of spectra at high-temperatures, for astrophysicalapplications were addressed. The last part concerns the non-empirical constructionof effective models for the dipole moment of methane from ab initio surfaces. For thisstudy, an algebraic technique based on the Contact Transformations approach, and developed in our team for triatomic molecules has been extended to the case of penta-atomic molecules for the very first time. All of these results was confronted with informations in spectroscopic databases, which were used to validate our approaches.

Le but principal de ce travail est d'examiner la capacité des méthodes clusters couplés explicitement corrélées (CCSD(T)-F12) pour la génération des surface de potentiel (SEPs) multidimensionnelles à courte, moyenne et longue portée intermoléculaires et leur utilisation dans la détermination des spectres des systèmes van der Waals (vdW) et des données de collision sur une large gamme d'énergie de collision. Ceci a été fait par une comparaison directe des surfaces de potentiel et des résultats générés à partir de l'utilisation des méthodes clusters couplés standards avec simple, double et un traitement perturbatif des excitations triples (CCSD(T)) et explicitement corrélée CCSD(T)-F12. En effet, nous démontrons que la méthode CCSD(T)-F12 en connexion avec la base aug-cc-pVTZ est la méthode de choix pour la génération de ces surfaces de potentiel avec une très grande réduction du coût computationnel (temps CPU et occupation disque). Ceci est illustré à travers l'étude des systèmes vdW HCl/DCl-He et CO2 -CO2 qui sont importants pour l'astrophysique et les atmosphères planétaires. Pour HCl/DCl-He, nous avons comparé les résultats issus des SEPs explicitement corrélées et celles calculées avec les méthodes cluster couplés standards. Ces SEPs ont été générées le long des coordonnées intermonomères. A travers ces comparaisons, nous avons montré que l'utilisation des méthodes CCSD(T)-F12/aug-cc-pVTZ explicitement corrélées permet de dépasser la précision de l'approche CCSD(T)/CBS (prise comme référence) avec un coût computationnel très réduit. La qualité de la surface de potentiel explicitement corrélée a été vérifiée par le calcul des sections efficaces et des taux de collision d'excitation rotationnelle de HCl/DCl par He. Par la suite, Nous étendons, notre étude aux systèmes moléculaires plus larges à travers l'étude du dimère de dioxyde de carbone,(CO2 )2 . Ainsi, nous avons généré une SEP d'interaction à 4 dimensions pour le dimère de dioxyde de carbone (CO2 )2 en utilisant la méthode CCSD(T)-F12/aug-cc-pVTZ qui a été comparée avec d'autres surfaces et spécialement avec celle générée par les approches de référence CCSD(T)/CBS et SAPT. Cette comparaison a bien prouvé la performance des méthodes explicitement corrélées pour la description de la corrélation électronique. Ceci, a été validé par des calculs de dynamique qui ont permis de déterminer le second coefficient du viriel, le premier moment spectral et les niveaux vibrationnels du dimère de CO2 et leur comparaison aux résultats expérimentaux et théoriques existants. Nous recommandons donc l'utilisation de la méthode CCSD(T)-F12/aug-cc-pVTZ pour l'étude des interactions moléculaires impliquant des systèmes polyatomique-polyatomique faiblement liés / The main goal of this work is to examine the capabilities of explicitly correlated coupled cluster (CCSD(T)-F12) methods for the mapping of multidimensional potential energy surfaces (PESs) at short, medium and long intermolecular ranges and therefore their use in the determination of accurate collisional data for a wide range of collision energies after dynamical computations. This is done through close comparisons of these PESs and corresponding data to those deduced using standard coupled cluster approaches with perturbative treatment of triple excitations (CCSD(T)). Therefore, we definitely establish that CCSD(T)-F12 in connection with the aug-cc-pVTZ basis set is the method of choice for the generation of such accurate PESs with strong reduction of computational cost (CPU and disk occupancy). As a benchmark, we treated the rotational excitation induced by collision of HCl/DCl-He and CO2 -CO2 van der Waals molecular systems, which are relevant for astrophysics and planetary atmospheres, respectively. For HCl-He and DCl-He, we closely compared the results deduced from PESs calculated with explicitly correlated and standard coupled cluster methods. These PESs were generated over the intermonomer degrees of freedom. Through these comparisons, we have shown that the use of explicitly correlated CCSD(T)-F12/aug-cc-pVTZ method overcomes the accuracy of the reference approach CCSD(T) associated with the complete basis set (CBS) with a very low computational cost (2 orders of magnitude less). The quality of the explicitly correlated potential surface was verified by cross sections and collision rate calculations. Then, we extended our findings to the CO2 -CO2 larger molecular systems. Hence, we generated four-dimentional (4D) PES of the carbon dioxide dimer (CO2 )2 using the CCSD(T)-F12 method and we compared it with other surfaces specially with that generated by the CCSD(T)/CBS and SAPT approaches. Again, we established the performance of explicitly correlated method to account fully for electron correlation. This was validated after dynamical calculations, where second virial coefficient, first spectral moment and vibrational energies of the CO2 dimer were computed and compared to previous experimental and theoretical data. In total, we recommend the use of CCSD(T)/aug-cc-pVTZ method for the investigation of weak interactions occurring between polyatomic-polyatomic systems for a wide range of applications

Dans la plupart des applications concernant un fluide s'écoulant sur une surface solide, la condition de non-glissement est largement utilisée car elle est simple et produit des résultats en accord avec les expériences. Toutefois, cette condition de limite n'est plus appropriée lorsque l'écoulement considéré est à l'échelle micro ou nano-métrique. Pour modéliser cet effet à l'échelle macroscopique, les conditions aux limites de Navier ont été introduites, avec la longueur de glissement comme paramètre. Lorsque le fluide est un gaz, cette longueur est liée au coefficient d'accommodation tangentiel (TMAC) et au libre parcours moyen, selon le modèle de Maxwell. Le but de ce travail est de traiter systématiquement ce modèle par une approche multi échelle et de l'étendre en incorporant la morphologie et l'anisotropie de la surface. La thèse est composée de cinq chapitres. Après l'Introduction, les notions de base de la théorie cinétique des gaz, l'équation de Boltzmann et les solutions associées (Navier-Stokes-Fourier, Burnett, Grad, Direct Simulation Monte Carlo…) sont rappelées dans le chapitre 1. Les modèles d'interaction gaz-paroi ainsi que les modèles de glissement introduits dans le cadre le la mécanique des fluides sont aussi rappelés. Le chapitre se termine par la description de la méthode de calcul par dynamique moléculaire (MD) utilisée dans ce travail. Le chapitre 2 est dédié au développement d'une technique simple afin de simuler les écoulements induits par la pression. Le principe est de se baser sur les formules atomistiques du tenseur des contraintes (Irving Kirkwood, Méthode de Plan, Contraintes Virielles) et de modifier les conditions périodiques, tout en maintenant la différence entre l'énergie cinétique des atomes à l'entrée et à sortie du domaine de calcul. Plusieurs types de conduite sont étudiés avec cette technique. Les résultats (température, vitesses, …) sont discutés et comparés. Le chapitre 3 concerne l'étude du potentiel d'interaction gaz-paroi par la méthode ab-initio. Le code CRYSTAL 09 est utilisé pour obtenir le potentiel entre un atome d'argon (Ar) et une surface de platine (Pt) <111> en fonction de la distance. Ensuite, le potentiel atome/surface est décomposé en potentiel binaire et approché par une fonction analytique. Cette fonction est ensuite implémentée dans un code MD pour simuler les collisions gaz-paroi et déterminer le coefficient TMAC.Dans le chapitre 4, l'effet de morphologie est étudié. Le potentiel multi-corps Quantum Sutton Chen(QSC) est utilisé pour le solide Pt <100> et un des potentiels binaires étudié dans le chapitre précédent pour le couple Ar-Pt. Le potentiel QSC est nécessaire pour reproduire l'effet de surface qui affecte le résultat final. Différentes surfaces sont traitées : surface lisse, surface nanostructurée, surface aléatoire obtenue par déposition de vapeur (CVD). Le coefficient TMAC est déterminé de façon généralisée, c.à.d en fonction de l'angle du flux d'atomes incidents sur la surface. Les anisotropies de surface et le noyau de collision sont également examinés. Dans le chapitre 5, on propose un modèle de glissement anisotrope pour les fluides en fonction du tenseur d'accommodation. Le modèle est obtenu par les calculs analytiques approchés développés dans le cadre de la théorie cinétique. On a ainsi généralisé l'équation de Maxwell en montrant que le tenseur de longueur de glissement est directement lié au tenseur d'accommodation. Le modèle est en bon accord avec les résultats de la méthode MD. Concernant la simulation MD, on a développé une technique pour reproduire l'anisotropie du tenseur d'accommodation. Le mémoire de thèse se termine par une synthèse des résultats obtenus. Des perspectives pour de futures études sont proposées / In most applications concerning a fluid flowing over a solid surface, the no-slip velocity condition was widely used because it is simple and produces the results in agreement with experiments. However, this dynamical boundary condition is not appropriate when the flow under consideration is at a micro or nano length scale.In order to model this effect at the macroscopic scale, the Navier boundary conditions have been introduced, with the slip length as a parameter. When the fluid is a gas, this length is related to the tangential momentum accommodation coefficient (TMAC) and the mean free path, according to the Maxwell model. The aim of this work is to systematically address this model using a multi-scale approach and to extend it by incorporating both the morphology and the anisotropy of a surface. The thesis consists of five chapters. In Chapter 1, the basics of the kinetic theory of gases, the Boltzmann equation and related solutions (Navier-Stokes-Fourier, Burnett, Grad, Direct Simulation Monte Carlo ...) are briefly presented. The models of gas-wall interaction and slip models introduced in the fluid mechanics are also recalled. The chapter ends with a description of the computational method used for the molecular dynamics simulations performed in this work. Chapter 2 is dedicated to the development of a simple technique to simulate the pressure driven flows. The principle is to rely on the atomistic formulas of the stress tensor (Irving Kirkwood, Method of Plane, Virial Stress) and to modify the periodic conditions by maintaining the difference between the kinetic energy of the ingoing and outgoing particles of the simulation domain. Several types of channels are studied with this technique. The results (temperature, velocity ...) are discussed and compared. Chapter 3 deals with the study of the gas-wall interaction potential by the ab-initio method. The code CRYSTAL 09 is used to obtain the potential between an atom of argon (Ar) and a surface of platinum (Pt) <111> as a function of distance. Then the gas-wall potential is decomposed into binary potential and approached by an analytic function. This function is then implemented in a MD code to simulate the gas-wall collisions and determine the TMAC coefficient. In Chapter 4, the effect of morphology is studied. The multi-body Quantum Sutton Chen (QSC) potential is used for Pt <100> solid and the binary potential proposed in the previous chapter for the Ar-Pt couple is employed. The QSC potential is needed to reproduce the surface effects that affect the final results. Different surfaces are treated : smooth, nanostructured surface and, random surface obtained by Chemical vapor deposition (CVD). The TMAC is determined using a generalized approach, i.e. depending on the angle of incident flux of gas atoms on the surface. The surface anisotropy and the scattering kernel are also examined. In Chapter 5, we propose a model of anisotropic slip for fluids based on accommodation tensor. The model is obtained by the analytical approximate calculations developed in the framework of the kinetic theory. We thus generalize Maxwell's equation by showing that the slip length tensor is directly related to the accommodation tensor. The model is in good agreement with the MD results. Thanks to our MD simulations, we develop a suitable technique for reproducing the anisotropy of the accommodation tensor. The thesis ends with a conclusion section in which we suggest some perspectives for a continuation of this work

Depuis des années, le silicium est le semiconducteur principalement utilisé dansl’industrie électronique et photovoltaïque. Intensivement étudié depuis plusieursdécennies, ses propriétés sont essentiellement connues, mais de nouvelles questionsviennent se poser. En particulier, une meilleure connaissance des nombreux défauts etimpuretés ainsi que leurs propriétés et leur impact sur les performances des dispositifsà base de Si est souhaitable.Ce travail couvre un éventail de problèmes liés aux défauts ponctuels en interactionau moyen de calculs dits de premiers principes (Density Functional Theory).Une première partie est dédiée à l’impact du dopage sur la diffusivité de l’oxygèneinterstitiel. Les coefficients de diffusion obtenus en fonction de la température sonten très bon accord avec les résultats expérimentaux ce qui démontre la validité dela méthodologie appliquée. Nous avons montré que l’augmentation de la diffusivitédans le silicium dopé bore se produit par un mécanisme de transfert de charge depuisle dopant de type p.Une deuxième partie se rapporte aux différents complexes de défauts ponctuels etleur thermodynamique, leur cinétique, et leurs propriétés optiques. La formation de cescomplexes peut être induite expérimentalement par une irradiation par des électrons.Plus généralement, ils apparaissent aussi dans des environnements opérationnelsparticuliers comme le spatial. Ici, nous avons réalisé une étude expérimentale etthéorique combinée pour identifier l’impact du dopage isovalent (C, Ge) et du codopage(C-Ge, C-Sn, C-Pb) sur la production de différents complexes (VOi, CiOi,CiCs), qui sont électriquement et optiquement actifs.Enfin, une attention particulière a été portée à la paire de défaut carbone-carboneet ses propriétés. Récemment, il a été établi que le silicium fortement dopé en carboneprésente des propriétés d’émission laser. Ici nous avons cherché à étudier les formespossibles du complexe et leurs propriétés, afin de comprendre lequel est présentexpérimentalement. / Since many years, silicon is the primary semiconductor material in electronic andphotovoltaic industry. Intensively studied through decades, its properties are essentiallyknown, however new questions keep arising. We need to achieve deep insightinto the numerous possible defects and impurities properties as well as their impacton the performances of the Si based devices. This work covers a range of problemsrelated with point defects interaction of both types long range and short range bymeans of parameter free first principles calculations.The former refers to the impact of heavy doping on diffusivity of interstitialoxygen species. The obtained diffusion coefficients as a function of temperature arein a very good agreement with experimental results that demonstrates the validityof the applied methodology. We showed that the enhanced diffusivity in B-dopedsilicon occurs through a charge transfer mechanism from the p-type dopantThe latter accounts for the various point defect complexes and their thermodynamic,kinetic, and optical properties. Formation of these complexes can beinduced by electron irradiation of Czochralski silicon. This aspect is of extremeimportance for particular operational environment. Here, we performed a combinedexperimental-theoretical investigation to identify the impact of isovalent doping (C,Ge) and co-doping (C-Ge, C-Sn, C-Pb) on the production of different complexes(VOi, CiOi, CiCs, etc.), which are electrically and optically active.Finally, particular attention is addressed to the carbon-carbon defect pair and itsproperties. Recently, it was established that heavily carbon doped silicon elucidateslasing properties. Here we aimed to revisit the possible forms of the complex andtheir properties, in order to associate one of them with light emitting G-centre,observed in experiments.v

Cette these est consacree a l’etude des interactions entre les lacunes et les elements d'alliages d’un acier renforcé par une dispersion de nano-oxydes (ODS (Oxide Dispersion Strengthened)), matériau de structure envisagé pour les réacteurs nucléaires du futur. Ces travaux ont été réalisés pour un système simplifié constitué d'une matrice de fer α contenant de l'oxygène, de l'yttrium, du titane et des lacunes à partir de modélisations multi-échelles. Nous nous sommes attachés en particulier au rôle des lacunes formées en exces au cours de l’elaboration de ces aciers. La stabilite et la mobilite des amas lacunes-solutés ont été examinées à partir de calculs ab initio, d’une part, pour l’oxygene qui a ete compare au carbone et a l’azote, solutes interstitiels egalement presents dans les aciers, et d’autre part, pour le titane et l’yttrium, solutes substitutionnels. Les trois solutés interstitiels ont révélé un comportement tres analogue. L’impact de la mobilité des amas lacunes-solutes a ete etabli en utilisant un modele en dynamique d’amas parametre sur nos résultats ab initio. Il a été ainsi démontré que, en sursaturation de lacunes, la diffusion des solutés interstitiels peut etre acceleree, alors que celle des solutes substitutionnels ne l’est pas forcement. Ces conclusions se sont averees coherentes avec des observations experimentales existantes. L’ensemble de ces résultats ont ensuite été exploité pour améliorer notre compréhension des mécanismes de formation des nanoparticules. Il est apparu que la diffusion relative de l’yttrium et du titane, ainsi que le nombre de noyaux potentiels pour former des nanoparticules dépendent de la concentration en lacunes dans le système. / This PhD thesis is devoted to the study of interactions between vacancies and alloying elements in Oxide Dispersion Strengthened (ODS) steels, which are promising candidate materials for future nuclear reactors. This work is based on multiscale modeling of a simplified system composed by oxygen, yttrium and titanium atoms and vacancies in an α-iron lattice. We particularly focused on the role of vacancies which are created in excess during the fabrication of these steels. The stability and mobility of vacancy-solute clusters have been examined using ab initio calculations for oxygen, on one hand, which has been systematically compared to carbon and nitrogen, interstitial solutes frequently present in iron-based materials, and, on the other hand, for substitutional solutes : titanium and yttrium. The three interstitial solutes show very similar energetic and kinetic behaviors. The impact of small mobile vacancy-solute clusters has been verified using a cluster dynamics model based on our ab initio results. It has been thus demonstrated that with oversaturation of vacancies, diffusion of interstitial solutes may be accelerated, while substitutional solutes do not become necessarily faster. These conclusions are consistent with existing experimental observations. All these results have been then used to complete our understanding of nanoclusters formation mechanisms. It appeared that the relative mobility of yttrium and titanium, as well as the number of potential nuclei to form nanoparticles strongly depend on the total vacancy concentration in the system.

Orientador: Alex Antonelli / Tese (doutorado) - Universidade Estadual de Campinas, Instituto de Física Gleb Wataghin / Made available in DSpace on 2018-08-29T05:01:21Z (GMT). No. of bitstreams: 1 CajahuaringaMacollunco_OscarSamuel_D.pdf: 15985103 bytes, checksum: f54d17d6335170c62196c9e329f257d9 (MD5) Previous issue date: 2015 / Resumo: Apesar de serem substâncias muito diferentes, silício e gálio, na fase líquida, compartilham muitas anomalias nas suas propriedades termodinâmicas. Evidências teóricas e experimentais sugerem que esses líquidos podem sofrer a chamada transição de fase líquido-líquido (LLPT). Especula-se que este tipo de transição é acompanhada por uma transição dinâmica entre um líquido frágil e um líquido forte, com base nas hipóteses de que o surgimento de um "dip" na função auto-intermediária de espalhamento logo após o regime balístico e o aparecimento do excesso de modos vibracionais em baixas frequências após a LLPT estariam relacionados com o comportamento de líquidos fortes. Foi realizado um estudo da dinâmica desses sistemas através das funções de correlação nas vizinhanças da LLPT usando simulações clássicas. Observamos apenas no caso do silício o aparecimento do "dip" na função auto-intermediária de espalhamento. A densidade de estados vibracionais reduzida de ambos os líquidos apresenta picos em baixas frequências, sugerindo que ambos os líquidos seriam fortes. Desta forma, nos dois casos estudados, as duas hipóteses não são conclusivas. Visando um melhor entendimento, determinamos a viscosidade de cisalhamento de ambos os líquidos em um amplo intervalo de temperatura. A apresentação desses resultados no chamado gráfico de Angell, indica que, em ambos os casos, a LLPT é acompanhada de uma transição de um líquido frágil para um líquido menos frágil, o que não dá suporte às especulações de uma transição frágil-forte. Além de investigar a LLPT em gálio "bulk", foi também estudado o gálio líquido confinado em nanofendas, ou seja, um sistema quase-bidimensional. Foi observada a transição de fase de primeira ordem entre um líquido isotrópico e um líquido ordenado. Por meio de uma série de análises estruturais, tanto de ordem translacional quanto orientacional, conseguimos identificar o líquido ordenado como a fase hexática, que é prevista ocorrer em sistemas bidimensionais segundo a teoria de Kosterlitz-Thouless-Halperin-Nelson-Young (KTHNY). Como a LLPT em gálio bulk foi identificada em simulações de dinâmica clássica, realizamos um estudo visando obter evidência da LLPT em simulações de primeiros princípios, que são em princípio mais realísticas. Utilizando o chamado método-Z, que permite estimar a temperatura de fusão, determinamos que o regime super-resfriado do gálio líquido descrito por cálculos de primeiros princípios ocorre em temperaturas abaixo de 400 K. Partindo de um líquido em equilíbrio em 500 K, realizamos simulações a pressão constante igual a zero em que o líquido foi resfriado para 260 K em um intervalo de 145 ps. Não foi observada uma mudança abrupta no volume durante o resfriamento. A fim de melhorar a amostragem em nossas simulações de primeiros princípios, realizamos um estudo de dinâmica molecular a volume constante com troca de réplicas. Foram usadas um total de 10 réplicas entre 400 K e 260 K, em simulações com a duração de 100 ps. Neste caso, a pressão decresceu monotonicamente com a temperatura, o que é característico de líquidos simples que não sofrem LLPT. Cabe ressaltar que nas simulações clássicas que identificaram a LLPT em gálio, a taxa de resfriamento foi muito mais lenta, da ordem de nano-segundos, que seria inexequível para simulações de primeiros princípios / Abstract: Despite of being very different substances, silicon and gallium, in their liquid phase, share several anomalous thermodynamic properties. Theoretical and experimental evidence suggest that these liquids can undergo the so-called liquid-liquid phase transition (LLPT). There has been speculated that this transition is accompanied by a dynamic transition form a fragile liquid to a strong liquid, based on the hypothesis that the development of a dip in the self-intermediate scattering function, just after the ballistic regime, and the appearance of excess low frequency vibrational modes just after the LLPT would be related to the behavior of strong liquids. It was performed a study of the dynamics of these systems through the correlation functions in the vicinity of the LLPT using classical simulations. We observed the development of a dip in the self-intermediate scattering function only in the case o silicon. The reduced density of vibrational states of both liquids exhibits peaks in low frequencies, suggesting both liquids to be strong. Therefore, for the two cases we studied, the two hypotheses are not conclusive. In order to achieve a better understanding, we determined the shear viscosity of both liquids for a wide range of temperatures. By plotting the data in the so-called Angell plot one can see that, in both cases, the LLPT is accompanied by a transition of a fragile liquid to a less fragile liquid, which does not give support to the speculations of a fragile-to-strong transition. Aside from investigating the LLPT in bulk gallium, it was also studied liquid gallium confined in nanoslits. i.e., a quasi two-dimensional system. It was observed a first-order transition between an isotropic liquid and an ordered liquid. Through a series of structural analyses, both translational and orientational, we were able to identify the ordered liquid as the hexatic phase, which is predict to occur in two-dimensional systems according to the Kosterlitz-Thouless-Halperin-Nelson-Young (KTHNY) theory. Since the LLPT in gallium has been found in classical molecular dynamics simulations, we performed a study aimed to obtain evidence of the LLPT in first-principles simulations, which are, in principle, more realistic. Using the so-called Z-method, which allows one to estimate the melting temperature, we determined the supercooled regime of liquid gallium, described by first-principles, to occur below 400 K. Starting from an equilibrated liquid at 500 K, we performed simulations at zero pressure in which the liquid was cooled down to 260 K in an interval of 145 ps. No abrupt change in the volume was observed as the liquid was cooled. In order to improve sampling in our first-principles simulations, we performed a study using constant volume molecular dynamics with replica exchange. Ten replicas were used between 400 K and 260 K, in simulations 100 ps long. In this case, pressure decreased monotonically with temperature, which is the behavior of simple liquids, which do not exhibit the LLPT. It should be emphasized that in the classical simulations that identified the LLPT in gallium, the cooling rates were much slower, of the order of nanoseconds, which would be unfeasible for first-principles simulations / Doutorado / Física / Doutor em Ciências / 1000949 / CAPES

As proteínas são moléculas que desempenham funções essenciais para a vida. Para entender a função de uma proteína é preciso conhecer sua estrutura tridimensional. No entanto, encontrar a estrutura da proteína pode ser um processo caro e demorado, exigindo profissionais altamente qualificados. Neste sentido, métodos computacionais têm sido investigados buscando predizer a estrutura de uma proteína a partir de uma sequência de aminoácidos. Em geral, tais métodos computacionais utilizam conhecimentos de estruturas de proteínas já determinadas por métodos experimentais, para tentar predizer proteínas com estrutura desconhecida. Embora métodos computacionais como, por exemplo, o Rosetta, I-Tasser e Quark tenham apresentado sucesso em suas predições, são apenas capazes de produzir estruturas significativamente semelhantes às já determinadas experimentalmente. Com isso, por utilizarem conhecimento a priori de outras estruturas pode haver certa tendência em suas predições. Buscando elaborar um algoritmo eficiente para Predição de Estruturas de Proteínas livre de tendência foi desenvolvido um Algoritmo de Estimação de Distribuição (EDA) específico para esse problema, com modelagens full-atom e algoritmos ab initio. O fato do algoritmo proposto ser ab initio é mais interessante para aplicação envolvendo proteínas com baixa similaridade, com relação às estruturas já conhecidas. Três tipos de modelos probabilísticos foram desenvolvidos: univariado, bivariado e hierárquico. O univariado trata o aspecto de multi-modalidade de uma variável, o bivariado trata os ângulos diedrais (&Phi; &Psi;) de um mesmo aminoácido como variáveis correlacionadas. O hierárquico divide o problema em subproblemas e tenta tratá-los separadamente. Os resultados desta pesquisa mostraram que é possível obter melhores resultados quando considerado a relação bivariada (&Phi; &Psi;). O hierárquico também mostrou melhorias nos resultados obtidos, principalmente para proteínas com mais de 50 resíduos. Além disso, foi realiza uma comparação com algumas heurísticas da literatura, como: Busca Aleatória, Monte Carlo, Algoritmo Genético e Evolução Diferencial. Os resultados mostraram que mesmo uma metaheurística pouco eficiente, como a Busca Aleatória, pode encontrar a solução correta, porém utilizando muito conhecimento a priori (predição que pode ser tendenciosa). Por outro lado, o algoritmo proposto neste trabalho foi capaz de obter a estrutura da proteína esperada sem utilizar conhecimento a priori, caracterizando uma predição puramente ab initio (livre de tendência). / Proteins are molecules that perform critical roles in the living organism and they are essential for their lifes. To understand the function of a protein, its 3D structure should be known. However, to find the protein structure is an expensive and a time-consuming task, requiring highly skilled professionals. Aiming to overcome such a limitation, computational methods for Protein Structure Prediction (PSP) have been investigated, in order to predict the protein structure from its amino acid sequence. Most of computational methods require knowledge from already determined structures from experimental methods in order to predict an unknown protein. Although computational methods such as Rosetta, I-Tasser and Quark have showed success in their predictions, they are only capable to predict quite similar structures to already known proteins obtained experimentally. The use of such a prior knowledge in the predictions of Rosetta, I-Tasser and Quark may lead to biased predictions. In order to develop a computational algorithm for PSP free of bias, we developed an Estimation of Distribution Algorithm applied to PSP with full-atom and ab initio model. A computational algorithm with ab initio model is mainly interesting when dealing with proteins with low similarity with the known proteins. In this work, we developed an Estimation of Distribution Algorithm with three probabilistic models: univariate, bivariate and hierarchical. The univariate deals with multi-modality of the distribution of the data of a single variable. The bivariate treats the dihedral angles (Proteins are molecules that perform critical roles in the living organism and they are essential for their lifes. To understand the function of a protein, its 3D structure should be known. However, to find the protein structure is an expensive and a time-consuming task, requiring highly skilled professionals. Aiming to overcome such a limitation, computational methods for Protein Structure Prediction (PSP) have been investigated, in order to predict the protein structure from its amino acid sequence. Most of computational methods require knowledge from already determined structures from experimental methods in order to predict an unknown protein. Although computational methods such as Rosetta, I-Tasser and Quark have showed success in their predictions, they are only capable to predict quite similar structures to already known proteins obtained experimentally. The use of such a prior knowledge in the predictions of Rosetta, I-Tasser and Quark may lead to biased predictions. In order to develop a computational algorithm for PSP free of bias, we developed an Estimation of Distribution Algorithm applied to PSP with full-atom and ab initio model. A computational algorithm with ab initio model is mainly interesting when dealing with proteins with low similarity with the known proteins. In this work, we developed an Estimation of Distribution Algorithm with three probabilistic models: univariate, bivariate and hierarchical. The univariate deals with multi-modality of the distribution of the data of a single variable. The bivariate treats the dihedral angles (&Phi; &Psi;) within an amino acid as correlated variables. The hierarchical approach splits the original problem into subproblems and attempts to treat these problems in a separated manner. The experiments show that, indeed, it is possible to achieve better results when modeling the correlation (&Phi; &Psi;). The hierarchical model also showed that is possible to improve the quality of results, mainly for proteins above 50 residues. Besides, we compared our proposed techniques among other metaheuristics from literatures such as: Random Walk, Monte Carlo, Genetic Algorithm and Differential Evolution. The results show that even a less efficient metaheuristic such as Random Walk managed to find the correct structure, however using many prior knowledge (prediction that may be biased). On the other hand, our proposed EDA for PSP was able to find the correct structure with no prior knowledge at all, so we can call this prediction as pure ab initio (biased-free).