No description available.

No description available.

La structure et la spectroscopie de molécules d'intérêt astrophysique ont été étudiées par des méthodes de calcul électronique ab initio hautement corrélées. Notre démarche a tout d'abord été d'étudier des systèmes allant de deux à quatre atomes, afin de voir l'influence du niveau de calcul électronique utilisé lors de la détermination de données comparables aux résultats expérimentaux. Ainsi, l'étude de la spectroscopie des états électroniques de la molécule AlCl a été entreprise. Les résultats obtenus pour ce système reproduisent parfaitement les mesures expérimentales existantes tout en introduisant la compréhension des phénomènes observés. Nous avons ensuite prédit un ensemble de données pour les ions AlCl+ et AlCl2+, en utilisant la même méthodologie. Nous avons également entrepris des calculs de durée de vie de prédissociation pour les molécules SH et SH+. Nos courbes d'énergie potentielle hautement corrélée nous ont permis de reproduire avec une grande précision les durées de vie de prédissociation des niveaux rovibrationnels de SH. Par la suite, nous avons étudié une molécule d'intérêt astrophysique, détectée dans l'atmosphère de Titan, N2H+. Nous avons généré des surfaces d'énergie potentielle multidimensionnelles pour les états fondamentaux des molécules N2H+, N2H2+ et N2HAr+. Ces surfaces ont été utilisées pour le calcul des niveaux rovibrationnels de ces ions et de leurs espèces deutérées. Un excellent accord avec les donné es expérimentales disponibles a été obtenu, et un ensemble de constantes spectroscopiques a été prédit. Enfin, nous nous sommes intéressés à une molécule plus grande, Si(C2H3)4. Cette étude nous a permis d'établir une méthodologie pour le calcul de la structure et des états électroniques excités de molécules de taille moyenne. Nous avons également effectué des calculs sur l'ion Si(C2H3)4+, afin d'interpréter sa fragmentation mesurée lors de l'expérience d'impact d'électron réalisée à l'Université de Bratislava durant cette thèse / Highly correlated ab initio methodologies were used to investigate the structure and the spectroscopy of small and medium sized molecules at different levels of theory. Our results are compared to the available experimental data. For the AlCl molecule, our computed spectroscopic constants were in good agreement with the experimental ones, allowing us to predict a set of predictive data for the less known AlCl+ and AlCl2+ ions. After that, we treated the predissociation lifetimes of the A state of SH and SH+. Our highly correlated calculations allowed us to reproduce the lifetimes measured experimentally. Thereafter, multidimensional potential energy surfaces of the ground state of N2H+, N2H2+ and N2HAr+ were generated. These surfaces were used for the calculations of rovibrational levels of these ions and their respective deuterated species. An overall excellent agreement was obtained comparing to the available experimental measurements. Finall y, we investigated the Si(C2H3)4 molecule in order to establish a methodology for the treatment of the structure and electronic excited states of medium sized molecules. We also studied the reactivity of the Si(C2H3)4+ ion for a tentative assignment of the mass spectra obtained in the University of Bratislava by an electron impact experiment

Les molécules se trouvent très rarement isolées, ceci implique qu'une modélisation de leur environnement doit être faite lors du calcul de propriétés physiques ou chimiques. Il est possible de considérer l'environnement par plusieurs méthodes de chimie théorique. Le modèle du continuum polarisable est un exemple dont les premières applications ont maintenant plus de 30 ans. Ce modèle permet de reproduire l'influence d'un solvant mais n'est pas capable de représenter des milieux fortement anisotropes tels que les macro-molécules. Afin de représenter de tels environnements, des méthodes couplant la mécanique quantique, pour le traitement de la partie d'intérêt chimique ou physique, et la mécanique moléculaire pour la représentation de l'environnement, ont été développées. Cette thèse est consacrée à l'étude de complexes de ruthénium en interaction avec l'ADN. Leurs spectres d'émission présentent des particularités trés intéressantes dues à cette interaction. Nous montrons que les propriétés photophysiques calculées doivent prendre en compte l'environnement. En particulier, nous avons utilisé une méthode permettant de modéliser la réponse électronique de l'environnement lors de transitions électroniques verticales. Les états triplets de ces complexes intercalés entre deux paires de bases de l'ADN sont également étudiés. En effet, les propriétés d'émission sont liées à la nature de ces derniers et il est important de modéliser de façon correcte le double-brin pour comprendre les mécanismes mis en jeu. Nous avons ainsi donné une interprétation physique à l'effet light-switch / Molecules are rarely isolated and a modelisation of their environment must be carried out when computing their physical or chimical properties. Quantum chemistry offers various ways to take into account this environment. For instance, polarizable continuum model is available for more than 30 years. This model is able to reproduce the influence of a solvent upon a solute but while the environment is becoming less isotropic, serious limitations are found for the model. In order to represent such environments, methods coupling quantum mechanics, for the treatment of the physically or chemically interesting part, and molecular mechanics for the environment have been developped. This thesis is dedicated to the study of ruthenium complexes in interaction with DNA. Moreover, their emission spectra are strongly modified by this interaction. We show that the photophysical properties calculated must take into account the environment. Eventually, we used a methodology able to include effects linked to the electronic response of the surroundings when computing vertical transitions. Triplets of these complexes intercalated between 2 DNA base pairs are also studied. Indeed, emission properties are linked to the nature of these and it is necessary to modelize correctly the double-strand to better understand mecanisms involved. The light-switch effect is then elucidated

Peroxidase de rábano (\"horseradish peroxidase - HRP), atuando como uma oxidase frente a substratos apropriados, catalisa a formação de espécies eletronicamente excitadas triplete. Os produtos obtidos são os esperados da clivagem de um intermediário 1,2-dioxetânico hipotético. Espécies tripletes geradas enzimaticamente são capazes de transferir energia a vários aceptores incluindo macromoléculas tais como: fitocromo, RNA, DNA e proteínas, além de organelas como cloroplastos. Acompanhamos a oxidação aeróbica de aldeídos alifáticos lineares (C2-C6) catalisada pela HRP; esta reação gera o aldeído inferior (Cn-1) no estado excitado triplete e ácido fórmico. Dois aspectos principais são abordados neste trabalho: (i) análise dos resultados em base ao comprimento da cadeia carbônica do substrato (C2 → C6) e (ii) a procura de emissões sensitizadas em estruturas biológicas (cloroplastos e microsomas). - Oxidação aeróbica dos aldeídos alifáticos lineares C2-C6 catalisada pela HRP. Corantes xantênicos e clorofila a solubilizada em micelas foram utilizados para monitorar estados eletronicamente excitadas geradas pela oxidação aeróbica dos aldeídos alifáticos lineares catalisada pela HRP. Quando eosina é o aceptor fluorescente, máxima emissão ocorre com butanal. Este resultado é discutido em conecçao com o fato de que uma série de efeitos biológicos de ácidos alifáticos lineares são máximos com o ácido butírico. A possível paticipação de estados excitados no processo que desencadeia as atividades biológicas deste ácido é sugerida. - TransferêncIa de energia para cloroplastos. Aldeídos alifáticos lineares (C2-C6) promovem emissão em cloroplastos na região de fluorescência de clorofila. Esta habilidade deve provavelmente constituir um caso de \"luminescência escura\" (\"dark luminescence\") . Se HRP estiver presente o aldeído (Cn) é oxidado ao homólogo inferior (Cn-1) no estado triplete, o qual sensitiza direta e/ou indiretamente a fluorescência de clorofila. Clorofila excitada por este processo é incapaz de reduzir um aceptor de Hill. - Quimioluminescência de microsomas expostos à oxidação aeróbica de aldeídos lineares catalisada pela HRP. Microsomas expostos ao sistema propanal/HRP/O2 desenvolvem luminescência de baixa intensidade. Este processo emissivo é distinto daquele originário durante a peroxidação de lipídeos uma vez que: (i ) a emissão se situa próxima a 560 nm e não na região espectral do vermelho (como esperado para a emissão bimol de oxigênio singlete) e (ii) não há formação de malonaldeído. Acetaldeído triplete parecer ser a espécie responsável pela indução deste processo através da excitação de algum componente microsomal, possivelmente uma flavoproteína. / Horseradish peroxidase (HRP), acting as anoxidase upon appropriate substrates, promotes the formation of electronically excite triplet species. The products obtained are those which would be expected from the cleavage of a hypothetical 1,2 dioxetane intermediate. Enzyme-generated triplet species are able to transfer energy to several acceptors, including macromolecules such as phytochrome, RNA, DNA and proteins, as well as to organeles such as chloroplasts. In the present study, we investigated the HRP-catalyzed aerobic oxidation of linear aliphatic aldehydes (C2 to C6), a reaction which generates the next lower aldehyde in the triplet state and formic acid. Two main aspects were emphasized in this work: (i ) the dependence of the results of the length of the carbonic chain of the substrate (C2 to C6) and (ii) the possibility of inducing sensitized emission from biological structures (chloroplasts and microsomes). HRP-catalyzed aerobic oxidation of aliphatic aldehydes. Xanthene dyes and micelle-solubilized chlorophyll-a were used to monitor the electronically excited species generated by the peroxidase-catalyzed aerobic oxidation of C2-C6 linear aldehydes. Maximal emission occurs with butanal as substrate and eosine as the fluorescent acceptor. This result is discussed in connection with the fact that the multiple biological effects of short chain aliphatic acids are maximal for butyric acid. Our observations strengthen the case for a possible role of excited state formation in the biological activity butyrate. Energy transfer to chloroplasts Linear aldehydes trigger red emission from chloroplasts. If horseradish peroxidase is also present, the aldehyde is oxidized to the next lower homolog in the triplet state, which in turn sensitizes (directly or indirectly) chlorophyll fluorescence. This phenomenon probably is a case of \"dark luminescence\"; the chlorophyll excited by this process is unable to reduce a Hill acceptor. MIicrosomal luminescence elicited by enzymatic systems that generate triplet species. Microsomes exposed to the propanal/HRP/02 system develop a weak luminescence. The underlying process is distinct from that occuring during lipid peroxidation because the emission intensity peaks at around 560 nm rather than in the red (as would be expected for bimol singlet oxygen emission) and no malonaldehyde is formed. Triplet acetaldehyde appears to be responsible for initiating the process, which in turn leads to excitation of a component of microsomes, possibly a flavoprotein.

Neste trabalho, utilizamos o desenvolvimento recente do método DFT/LDA-1/2 para cálculos de estados excitados em materiais. Começamos com um resumo da teoria do funcional da densidade (DFT) e incluímos uma introdução ao método LDA-1/2 para cálculos de excitações em sólidos. Na compilação dos resultados esperamos ter demonstrado a utilidade do LDA-1/2 para cálculos de alinhamentos de bandas em junções semicondutor/semicondutor e semicondutor/isolante. A aplicação do método envolve o conhecimento da química básica dos sistemas. Para tanto, escolhemos sistemas importantes para diversas aplicações, e cujos modelos de simulação estão o limite ou fora do alcance de metodologias que envolvem alto custo computacional, mas que foram bem caracterizados experimentalmente. Concentramos nossas ações no estudo da capacidade preditiva do LDA-1/2 para alinhamentos de bandas, os chamados band offsets, particularmente importantes para a micro e optoeletrônica. Quando não foi possível compararmos nossos resultados com o experimento, procuramos a comparação com métodos estado-da-arte como GW. Bons resultados foram obtidos para band gaps e band offsets de interfaces A1As/GaAs, Si/SiO2, A1N/GaN e CdSe/CdTe, que representam os diferentes tipos de jun_c~oes poss__veis, com (e.g. A1As/GaAs, A1N/GaN) e sem (e.g. Si/SiO2, CdSe/CdTe) ^anions omuns, com (e.g. A1As/GaAs) e sem (e.g. CdSe/CdTe, Si/SiO2) casamento de parâmetros de rede e diferentes tipos de alinhamentos (\"straddling\", e.g. A1As/GaAs ou \"staggered\"e.g. CdSe/CdTe). Analisamos de maneira sistemática o comportamento do entorno do bandgap ao longo da interface, verificando plano a plano atômico o comportamento das bordas de valência e condução com LDA-1/2 em comparação com o LDA, ou comparando diferentes modelos dentro do LDA-1/2, como o caso do CdSe/CdTe e do Si/SiO2. Para o caso A1As/GaAs, aproveitamos o casamento de parâmetros de rede dos semicondutores constituintes e tentamos um modelo de interface de ligas A1xGa1-x As/GaAs para estudar a variação de valência, condução e bandgap em função da composição x. No AlN/GaN, estudamos também os offsets com as contribuições dos orbitais separadamente. Em todos os casos o LDA-1/2 levou-nos a resultados interessantes com modelos simples. A exploração de novas fronteiras de aplicação do método fez-se necessária com a diminuição da dimensionalidade dos sistemas, de 3D (bulk ) para 2D (interfaces) e depois para 1D, ou seja, _os quânticos (\"nanofios\"). Nosso material de estudo para os foi o ZnO que, além da motivação oriunda de conhecidas aplicações em optoeletrônica, apresenta desafios para simulações bulk com qualquer método, e que foi abordado com certo sucesso usando o LDA-1/2 anteriormente, sendo que para fios quânticos encontramos resultados interessantes em geometrias triangulares que facilitaram os modelos. Calculamos o bandgap ZnO bulk e de nanofios passivados e não passivados com hidrogênios usando LDA e LDA-1/2 sem polarização de spin. As estruturas de bandas e o bandgap como função do diâmetro do ano_o foram calculados e ajustes com funções de decaimento foram feitos para comparação, por extrapolação, dos bandgaps com valores experimentais. Foi possível comparar nossos resultados de fios com o bulk, e predizer uma faixa de variaação de bandgaps que os experimentais podem encontrar para nanofios triangulares de ZnO. Também foi feita análise de energias de confinamento em fios quânticos de ZnO, comparando o LDA com LDA-1/2. Finalmente, mostramos os resultados de uma oportunidade de aplicação do método a um material com defeitos, recentemente descoberto e promissor, e com enorme mercado potencial em fotocatálise, o Ti1-O4N. Nosso trabalho envolveu a aplicação do LDA-1/2 a um problema muito desafiador, e.g. a geração de energia limpa, especificamente a separação da molécula de água para produção de hidrogênio. O desafio maior vem da dificuldade de predição de bandgaps teoricamente, em particular para sistemas grandes como é o caso de modelos atomísticos com defeitos, devido aos altos custos computacionais envolvidos. Tais dificuldades forçam os pesquisadores a usarem parâmetros ajustáveis ou métodos semi-empíricos, ou modelos simplificados demais para descrever precisamente resultados experimentais. Isto dificulta o estudo dos sistemas fotocatalíticos potencialmente eficientes e que não foram ainda caracterizados ou otimizados. O LDA-1/2 é aqui validado para esta classe de materiais, abrindo assim a oportunidade para estudar sistemas mais realísticos e complexos para cálculos ainda mais precisos, particularmente para geração de energia limpa. Em particular, modelamos o TiO2 na estrutura rutile com nitrogênio substitucional, cuja estrutura eletrônica é ainda debatida. Foi a primeira aplicação do LDA-1/2 a sistemas com algum tipo de defeito, com ótimos resultados para o novo sistema Ti1- _O4N com vacâncias de Ti. / In this work, we used the recent development of DFT/LDA-1/2 method for calculations of excited states in materials. We begin with a summary of the density functional theory (DFT) and included an introduction to the method LDA-1/2 for calculations of excitations in solids. In compiling the results we hope to have demonstrated the usefulness of the LDA-1/2 for calculating alignments of bands at junctions semiconductor / semiconductor and semiconductor / insulator. The method involves the knowledge of basic chemical systems. To do this we chose systems important for several applications, and simulation models which are the limit or beyond the reach of methodologies involving high computational cost, but have been well characterized experimentally. We focus our actions in the study of the predictive capability of the LDA-1/2 for alignments of bands, the band called offsets, particularly important for micro and optoelectronics. When it was not possible to compare our results with experiment, we compared the methods with state of the art as GW. Good results were obtained for band gaps and band offsets of interfaces A1As/GaAs, Si/SiO2, A1N/GaN and CdSe / CdTe, which represent the different types of jun_c poss__veis-tions, with (eg A1As/GaAs, A1N/GaN) and without (eg Si/SiO2, CdSe / CdTe) ^ omuns anions with (eg A1As/GaAs) and without (eg CdSe / CdTe, Si/SiO2) matching network parameters and different types of alignments (\"straddling\" eg A1As/GaAs or \"staggered\" eg CdSe / CdTe). Systematically analyze the behavior of the environment along the interface bandgap, plane by plane scanning behavior of the edges atomic valence and conduction with LDA-half in comparison with LDA, or comparing templates within the LDA-1 / 2, as the case of CdSe / CdTe and Si/SiO2. For the case A1As/GaAs, we take the marriage of network parameters of semiconductor components and try an interface model alloys A1xGa1-x As / GaAs to study the variation of valence, conduction and bandgap as a function of composition x. In the AlN / GaN, we also studied the offsets with the contributions of the orbitals separately. In all cases the LDA-half led us to interesting results from simple models. The exploration of new frontiers of the method was necessary to decrease the dimensionality of the systems, the 3D (bulk) for 2D (interfaces) and then to 1D, ie, quantum _os (\"nanowires\"). Our study material for the ZnO was that, apart from the motivation coming from known applications in optoelectronics, presents challenges for bulk simulations with any method, and that was addressed with some success using the LDA-half earlier, and for wireless find interesting results in quantum triangular geometries that facilitated models. We calculate the bandgap and bulk ZnO nanowires passivated and not passivated with hydrogen using LDA and LDA-1/2 without spin polarization. The bandgap structures and strips as a function of the diameter of ano_o adjustments are calculated and decay functions for comparison were made by extrapolation of the bandgaps with experimental values. It was possible to compare our results with the bulk of wires, and predict a range of bandgaps that variaação can find experimental triangular ZnO nanowires. It was also made analysis of energy confinement in ZnO quantum wires, comparing LDA with LDA-1/2. Finally, we show the results of an opportunity to apply the method to a material with defects, newly discovered and promising, and with huge market potential in photocatalysis, the Ti1-O4N. Our work involved the application of LDA-1/2 to a very challenging problem, eg the generation of clean energy, specifically the separation of the water molecule for hydrogen production. The main challenge has been the difficulty of predicting bandgaps theoretically, in particular for large systems such as the model atomistic defects because of the high computational costs involved. These difficulties force the researchers to use adjustable parameters or semi-empirical methods, or other simplified models to accurately describe experimental results. This complicates the study of potentially efficient photocatalytic systems which have not yet been characterized or optimized. The LDA-1/2 is here validated for this class of materials, thus opening the opportunity to study more realistic and complex systems for more accurate calculations, particularly for clean energy generation. In particular, we modeled the structure of TiO2 in the rutile with substitutional nitrogen, whose electronic structure is still debated. It was the first application of the LDA-1/2 systems with some kind of defect, with excellent results for the new system Ti1-_O4N with Ti vacancies.

Submitted by Jorge Silva (jorgelmsilva@ufrrj.br) on 2017-02-15T17:02:45Z No. of bitstreams: 1 2009 - William Pereira.pdf: 3794972 bytes, checksum: 64fbf9758bbf44329031caf3b369b79a (MD5) / Made available in DSpace on 2017-02-15T17:02:45Z (GMT). No. of bitstreams: 1 2009 - William Pereira.pdf: 3794972 bytes, checksum: 64fbf9758bbf44329031caf3b369b79a (MD5) Previous issue date: 2009-10-22 / CAPES / The present study has two specific phases: Early, was studied the isathine irradiation with the presence of diisopropil-phosphoril-dimethylidrazone (diisopfh), in chloroform, at room temperature and 300nm. The structure of photo-product was elucidated by GC/MS, IR, NMR 1H, 13C and 31P, as well as hety-cosy, homo-cosy and other techniques. The proposal mechanism involve a bi-radical from triplet excited state of isathine that react with C=N of the diisopfh and has a ring expansion. The successive pass would be thermal and with propanone elimination, as thermal rearrangement on phosphoric derivates described in the literature. In a second phase was studied, by Laser Flash Photolysis, the kinetic effect of the six aromatic derived of phosphoril-hidrazone (H, p-Cl, p-NO2, p-CN, p-NMe2 e p-CO2H) on the triplet excited state of four aromatic ketones: Xanthone, Tioxanthone, Benzophenone and Benzyl. The observed process to quencher, in all cases, was energy transference, because the rate constants are in the order of 1010M-1s-1. In this form was possible to estimate the energy of triplet excited state of these six phosphoril-hidrazone derivates. / O presente estudo divide-se em duas etapas bem distintas: Inicialmente, estudou-se a irradia??o de isatina na presen?a de diisopropil-dimetilfosforilidrazona (diisopfh) em clorof?rmio, a temperatura ambiente e comprimento de onda de 300nm. H? forma??o de um produto majorit?rio cujo peso molecular corresponde ? soma dos reagentes, diminu?do de 58g/mol. A estrutura deste produto foi elucidada por CG/EM, IV e RMN de 1H, 13C e 31P, al?m de t?cnicas como homocosy, hetycosy e outros. Prop?e-se um mecanismo via bi-radical formado a partir do estado excitado triplete da isatina com a liga??o C=N do composto diisopfh e conseq?ente expans?o do anel isat?nico; levando ? forma??o de um prov?vel produto fotoqu?mico prim?rio resultante de cicloadi??o. Numa segunda etapa, por processo t?rmico de elimina??o de propanona, similar a processo descrito na literatura por Olah, levaria a forma??o do fotoproduto detectado (derivado fosforil-benzodiazep?nico). Numa segunda etapa, estudou-se, por Fot?lise por Pulso de laser, o efeito cin?tico da presen?a de seis derivados arom?ticos de fosforilidrazona (H, p-Cl, p-NO2, p-CN, p-N (CH3)2 e p-CO2H) sobre o estado excitado triplete de quatro cetonas arom?ticas: tioxantona, xantona, benzofenona e benzil, todos em solu??o de acetonitrila. As constantes de velocidades obtidas (da ordem de difus?o da acetonitrila 1,9x1010 M-1s-1) indicam que o processo de supress?o de estado excitado se faz por transfer?ncia de energia. Desta forma foi poss?vel estimar a energia de estado excitado triplete destes derivados arom?ticos de fosforilidrazona.

Un très grand nombre de représentations a été proposé pour modéliser la liaison chimique, mais les structures de Lewis en particulier sont largement utilisées par la communauté des chimistes expérimentateurs. Les méthodes théoriques se développant sur des structures chimiques claires de type Lewis sont essentiellement utilisées pour la description des états fondamentaux. Par ailleurs, la majorité des chimistes théoriciens utilise des orbitales moléculaires pour décrire les état excités, et manque ainsi de lisibilité. Les états excités sont difficiles à prédire, il convient donc d’utiliser un langage simple pour aboutir à une compréhension commune de ces états. Nous proposons dans cette thèse deux méthodes afin d'accéder aux états excités décrits sur des structures facilement lisibles. D’abord, une méthode de projection permettant de développer un état excité en structure chimique claire de type VB. Cette méthode est rapide car elle ne diagonalise pas d’Hamiltonien VB et elle calcule un taux de confiance servant de garde-fou pour juger la fiabilité de la description de l’état excité. Ensuite, une méthode itérative utilisant un Hamiltonien Super-IC optimise des orbitales VB pour un état excité. Cette méthode couplée à la méthode de projection ouvre un passage vers une compréhension simple des états excités. / A large number of chemical representations has been proposed to model the chemical bond, but in particular Lewis structures are widely used by the experimenters community. The theoretical methods on developing the clear chemical Lewis structures are mainly used for the description of the ground states. Moreover, the majority of theoretical chemists uses molecular orbitals to describe the excited states, and thus lacks clarity. The excited states are difficult to predict, it is appropriate to use simple language to reach a common understanding of these states. We propose in this thesis two methods to access the excited states described on easily readable structures. First, a projection method developing an excited state in clear chemical structure type VB. This method is fast because it does not VB Hamiltonian diagonalizes and calculates a trust factor for a safeguard to judge the description of the excited state. Then an iterative method using a Super-CI Hamiltonian to optimize VB orbitals to an excited state. This method coupled with the projection method opens a way to a simple understanding of the excited states.

La réponse linéaire de la théorie de la fonctionnelle de la densité dépendante du temps est étudiée dans le cadre du formalisme d'ondelettes du code BigDFT, qui permet d'exprimer les fonctions d'onde électroniques sur une grille de simulation dans l'espace réel. L'objectif est de déterminer un spectre d'excitations de référence pour un système et un potentiel d'échange-corrélation donnés.Il apparaît que seule une partie du spectre, concernant les transitions entre orbitales liées, peut être facilement amenée à convergence par rapport aux paramètres d'entrée de BigDFT, que sont l'extension de la grille de simulation et le nombre d'orbitales du continuum qui sont considérées pour le calcul des spectres. L'énergie de la dernière orbitale inoccupée utilisée dans les calculs se révèle d'ailleurs être un paramètre plus important que ce nombre d'orbitales inoccupées. La justification vient de l'étude de la complétude des bases formées par les orbitales de l'état fondamental du système. Tout ceci permet de porter un regard neuf sur les résultats obtenus avec le formalisme à base gaussienne, tel qu'implémenté dans le code NWChem.En ce qui concerne la convergence du spectre de plus haute énergie, concernant des transitions entre orbitales occupées et orbitales inoccupées du continuum, l'espoir d'une convergence se heurte au problème du tassement du continuum. Il faut alors songer à une manière différente de capter l'information contenue dans ce continuum.Le formalisme des états résonants, dont les fondements ont été posés lors de la première moitié du XXème siècle, est une piste très encourageante pour cela. Une étude préliminaire dans le cas du puits de potentiel carré à une dimension est donc présentée. La première étape a consisté en la détermination de ces états résonants, dont les énergies et fonctions d'onde sont complexes. Une normalisation a notamment pu leur être attribuée. Il est ensuite montré, sous certaines conditions, que la base formée par les états propres de ce potentiel, dont une partie est constituée par les états du continuum, peut être efficacement remplacée par une base discrète et complète faite d'états résonants. Des applications numériques montreront qu'ils peuvent être avantageusement utilisés pour définir la fonction de Green ou encore calculer la propagation temporelle d'un paquet d'onde. / The linear response on the time-dependent density functional theory is studied in the wavelets formalism used in the BigDFT code, that allows the representations of electronic wave-functions on a simulation grid in real space. The goal of this study is to determine a reference excitation spectrum for a given system and exchange-correlation potential.It appears that only one part of the spectrum can be easily brought to convergence with respect to the input parameters of BigDFT, which are the simulation grid extension and the number of unoccupied continuum orbitals considered in the spectrum calculation. The energy of the last unoccupied orbital used actually proves to be more important as a parameter than this number of unoccupied orbitals. This is justified by the study of the completeness of the basis sets made of the ground state orbitals of the system. This gives another point of view regarding spectrum obtained by using the Gausian basis sets formalism, as the one implemented in the code NWChem.As to the convergence of the spectrum at higher energy, concerning transitions between occupied orbitals and unoccupied orbitals of the continuum, the hope for a convergence faces the problem of the continuum collapse. One therefore has to think of another way of retrieving the data contained in this continuum.The resonant states formalism, whose foundations were laid in the first half of the 20th century, is very encouraging in this regard. A preliminary study in the case of the one-dimension square well potential is therefore presented. The first step consisted in the determination of these resonant states, whose energies and wavefunctions are complex valued in general. Their normalization was also clearly defined. It is then shown, under certain conditions, that the basis set formed by the eigenstates of this potential, including the continuum states, can be efficiently replaced by a discrete and complete basis set made of resonant states. Numerical applications also show that these states can also be advantageously used to define the Green's function or even compute the time propagation of a wavepacket.

No description available.