Miniaturiser les composants impose des changements radicaux pour l'élaboration des dispositifs micro électroniques du futur. Dans ce cadre, les oxydes de grille MOS atteignent des épaisseurs limites qui les rendent perméables aux courants de fuite. Une solution est de remplacer le SiO2 par un matériau de permittivité plus élevée permettant l'utilisation de couches plus épaisses pour des performances comparables. Dans ce travail nous présentons une modélisation multi-échelle de la croissance par couche atomique (ALD) d'HfO2 sur Si permettant de relier la nano-structuration d'une interface au procédé d'élaboration. Nous montrons que la connaissance de processus chimiques élémentaires, via des calculs DFT, permet d'envisager une simulation procédé qui repose sur le développement d'un logiciel de type Monte Carlo Cinétique nommé "HIKAD". Au delà des mécanismes les plus évidents, adsorption, désorption, décomposition et hydrolyse des précurseurs sur la surface, nous introduirons la notion de mécanismes de densification des couches d'oxyde déposées. Ces mécanismes sont l'élément clé permettant de comprendre comment s'effectue la croissance de la couche en termes de couverture. Mais au delà de cet aspect ils nous permettent d'appréhender comment, à partir de réactions de type moléculaire le système évolue vers un matériau massif. Nous discuterons ces divers éléments à la lumière de résultats de caractérisations obtenus récemment sur le plan expérimental du dépôt d'oxydes d'hafnium.

<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>

<p>Metal ions solvated in aqueous, non-aqueous, and mixtures of solvents occur in many chemical contexts, for example in electrochemical applications and solvent separation. Solvated ions appear in high concentration in the living organisms, where their presence or absence can fundamentally alter the functions of life. In many of these cases, understanding the selective solvation and the dynamics of the ions is essential for the understanding of the processes involved.</p><p>Computer simulation provides a molecular level of detail of the solvation process usually not available from experiments. The quality of the interaction models employed in the theoretical description is of particular importance, since even rather small changes in the interaction can lead to substantial and qualitative differences.</p><p>This thesis describes the development of a sequence of increasingly refined analytical ion-solvent potentials from <i>ab initio</i> calculations for the systems Li⁺(<i>aq</i>), Na⁺(<i>aq</i>), Mg²⁺(<i>aq</i>), Al³⁺(<i>aq</i>), Li⁺(<i>MeCN</i>), Na⁺(<i>MeCN</i>), Li⁺(<i>aq, MeCN</i>), and Na⁺(<i>aq, MeCN</i>). Molecular dynamics simulations using these potentials were subsequently performed, and some key-properties computed. The reliability of the computed thermodynamical, structural and dynamical properties was scrutinized.</p>

<p>The aqueous solvation of metal ions is one of the long-standing and complex problems in chemistry, with implications for and applications in a broad range of biochemical and electrochemical systems, where water is the all-pervasive medium.</p><p>This thesis describes computer simulations of Al³⁺(<i>aq</i>), Fe²⁺(<i>aq</i>), Fe³⁺(<i>aq</i>) and Cu²⁺(<i>aq</i>). Various aspects of the solvation of these polyvalent metal ions in water are addressed, at different levels of theory, using Car-Parrinello molecular dynamics, classical molecular dynamics and quantum-mechanical cluster calculations. Polyvalent metal ions are particularly interesting because of their large influence on the solvent structure, dynamics and thermodynamics, as well as on the properties of the individual solvent molecules. Polyvalent metal ions in aqueous solution also constitute a challenging subject for computer simulations since a sophisticated interaction model is needed to incorporate the large many-body effects. </p><p>All the ion-water coordination figures in this thesis are octahedral, except in the Cu²⁺(<i>aq</i>) solution, where the ion is penta-coordinated with four equatorial neighbours in a plane and one axial neighbour located ~0.45 Å further out from the ion. The equatorial ion-water bonds have covalent character, while the axial water molecule is only electrostatically bound. For all the ions, the OD stretching frequencies of the first-shell water molecules are much more downshifted than in liquid water. In the case of Cu²⁺(<i>aq</i>), however, only the OD frequencies of the equatorial water molecules are downshifted with respect to bulk water whereas the OD frequencies of the axial water molecule are slightly upshifted. </p><p>Various limitations of the Car-Parrinello molecular dynamics simulations have been explored and compared, such as finite system-size effects and shortcomings in the electronic structure calculations. The Car-Parrinello simulations are found to give reasonable descriptions of the polyvalent metal ions in aqueous solution.</p>

<p>The development and improvement of the functionality of metal oxides in heterogeneous catalysis and other surface chemical processes can greatly benefit from an atomic-level understanding of the surface chemistry. Atomistic calculations such as quantum mechanical (QM) calculations and molecular dynamics (MD) simulations can provide highly detailed information about the atomic and electronic structure, and constitute valuable complements to experimental surface science techniques. </p><p>In this thesis, an embedded-cluster approach for quantum mechanical calculations has been developed to model the surface chemistry of metal oxides. In particular, CO adsorption on the MgO(001) and CeO₂(110) surfaces as well as O vacancy formation at the CeO₂(110) surface have been investigated. The cluster model has been thoroughly tested by comparison with electronic structure calculations for the periodic slab model.</p><p>The chemical implications of distorted surface structures arising from the surface dynamics have been investigated by combining the QM embedded-cluster calculations with force-field based MD simulations. Here QM embedded-cluster calculations were performed using surface structures sampled from the MD simulations.</p><p>This combined MD+QM embedded-cluster procedure was applied to the CO adsorption on MgO(001) at 50 K and the O vacancy formation on CeO₂(110) at 300 K. Significant thermal variations of the CO adsorption energy and the O vacancy formation energy were observed. It was found that these variations could be estimated using the force field of the MD simulation as an interaction model. With this approach, the QM results were extrapolated to higher temperature and doped systems.</p>

Theoretical studies of point defect interactions and structural stability of compounds have been performed using density functional theory. The defect-related properties, such as activation energy of diffusion, electronic and magnetic structure of selected materials have been studied. The major part of the present work is devoted to a very important material for semiconductor industry, GaAs. The formation energies of intrinsic point defects and the solution energies of 3d transitions in GaAs have been calculated from first principles. Based on the calculated energies, we analysed the site preference of defects in the crystal. The tendency of defects to form clusters has been investigated for the intrinsic defects as well as for impurities in GaAs. The magnetic moment of 3d impurities has been calculated as a function of the chemical environment. The possibility of increasing the Curie temperature in (Ga,Mn)As by co-doping it with Cr impurities has been examined on the basis of calculated total energy difference between the disordered local moment and the ferromagnetically ordered spin configurations. We found that, in order to reach the highest critical temperature, GaAs should be separately doped with either Cr or Mn impurities. Also, we have shown that diffusion barrier of interstitial Mn depends on the charge state of this impurity in (Ga, Mn)As. The formation of defect complexes between interstitial and substitutional Mn atoms, and their influence on the value of diffusion barrier for interstitial Mn, has been studied. The pair interactions energies between interstitial oxygen atoms in hcp Zr, Hf and Ti have been calculated using first principles. Based on the calculated energies, the oxygen ordering structures in IVB transition metal solid solutions have been explained. A prediction of nitrogen ordering in Hf-N solid solution has been made. The thermodynamic description of intermetallic compounds in the Zr-Sn binary system has been obtained. The conclusion has been made that Zr substitution on the Sn sites takes place in the Zr4Sn phase, which accounts for the unusual stoichiometry of this Cr3Si structure type compound. The influence of pressure on the phase stability in the Fe-Si system has been investigated. We have found instability of the hcp Fe0.9Si0.1 random alloy with respect to the decomposition onto the Si-poor hcp Fe alloy and the B2 FeSi under high pressure. The tendency of this decomposition becomes stronger with increasing the applied pressure. / QC 20100624

Point defects are an inherent part of crystalline materials and they influence important physical and chemical properties, such as diffusion, hardness, catalytic activity and phase stability. Increased understanding of point defects enables us to tailor the defect-related properties to the application at hand. Modeling and simulation have a prominent role in acquiring this knowledge. In this thesis thermodynamic and kinetic properties of point defects in metals and ceramics are studied using first-principles calculations based on density functional theory. Phenomenological models are used to translate the atomic level properties, obtained from the first-principles calculations, into functional materials properties. The next paragraph presents the particular problems under study. The formation and migration of vacancies and simple vacancy clusters in copper are investigated by calculating the energies associated with these processes. The structure, stability and electronic properties of the low-oxygen oxides of titanium, TiOx with 1/3 &lt; x &lt; 3/2, are studied and the importance of structural vacancies is demonstrated. We develop an integrated first-principles and Calphad approach to calculate phase diagrams in the titanium-carbon-nitrogen system, with particular focus on vacancy-induced ordering of the substoichiometric carbonitride phase, TiCxNy (x+y &lt; 1). The possibility of forming higher oxides of plutonium than plutonium dioxide is explored by calculating the enthalpies for nonstoichiometric defect-containing compounds and the analysis shows that such oxidation is only produced by strong oxidants. For ceria (CeO2) doped with trivalent ions from the lanthanide series we probe the connection between the choice of a dopant and the improvement of ionic conductivity by studying the oxygen-vacancy formation and migration properties. The significance of minimizing the dopant-vacancy interactions is highlighted. We investigate the redox thermodynamics of CeO2-MO2 solid solutions with M being Ti, Zr, Hf, Th, Si, Ge, Sn or Pb and show that reduction is facilitated by small solutes. The results in this thesis are relevant for the performance of solid electrolytes, which are an integral part of solid oxide fuel cells, oxygen storage materials in automotive three-way catalysts, nuclear waste materials and cutting tool materials. / QC 20100622

In this thesis I have performed theoretical studies on the reaction dynamics in few-atom molecules. In particular, I have looked at reaction processes in which highly excited resonant states are involved. When highly excited states are formed, the dynamics becomes complicated and approximations normally used in chemical reaction studies are no longer applicable.To calculate the potential energy curve for some of these states as a function of internuclear distance, a combination of structure calculations and scattering calculations have to be performed, and the reaction dynamics on the potentials has been studied using both time-independent and time-dependent methods.The processes that have been studied and which are discussed in this thesis are ion-pair formation in electron recombination with H3+, dissociative recombination and ion-pair formation of HF+, mutual neutralization in H++F- collisions and dissociative recombination of BeH+. Isotope effects in these reactions have also been investigated. Our calculated cross sections are compared with experimentally measured cross sections for these reactions.

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. Ab initio 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 ab initio 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. 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. 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.

The aqueous solvation of metal ions is one of the long-standing and complex problems in chemistry, with implications for and applications in a broad range of biochemical and electrochemical systems, where water is the all-pervasive medium. This thesis describes computer simulations of Al3+(aq), Fe2+(aq), Fe3+(aq) and Cu2+(aq). Various aspects of the solvation of these polyvalent metal ions in water are addressed, at different levels of theory, using Car-Parrinello molecular dynamics, classical molecular dynamics and quantum-mechanical cluster calculations. Polyvalent metal ions are particularly interesting because of their large influence on the solvent structure, dynamics and thermodynamics, as well as on the properties of the individual solvent molecules. Polyvalent metal ions in aqueous solution also constitute a challenging subject for computer simulations since a sophisticated interaction model is needed to incorporate the large many-body effects. All the ion-water coordination figures in this thesis are octahedral, except in the Cu2+(aq) solution, where the ion is penta-coordinated with four equatorial neighbours in a plane and one axial neighbour located ~0.45 Å further out from the ion. The equatorial ion-water bonds have covalent character, while the axial water molecule is only electrostatically bound. For all the ions, the OD stretching frequencies of the first-shell water molecules are much more downshifted than in liquid water. In the case of Cu2+(aq), however, only the OD frequencies of the equatorial water molecules are downshifted with respect to bulk water whereas the OD frequencies of the axial water molecule are slightly upshifted. Various limitations of the Car-Parrinello molecular dynamics simulations have been explored and compared, such as finite system-size effects and shortcomings in the electronic structure calculations. The Car-Parrinello simulations are found to give reasonable descriptions of the polyvalent metal ions in aqueous solution.