Le formalisme des fonctions de Green hors-équilibre (NEGF pour « Non-equilibrium Green’s function) a suscité au cours des dernières décennies un engouement fort pour étudier les propriétés du transport quantique des nanostructures et des nano-dispositifs dans lesquels les interactions inélastiques, comme la diffusion des électrons-phonons, jouent un rôle significatif. L'incorporation d'interactions inélastiques dans le cadre du NEGF s’effectue généralement dans l'approximation auto-cohérente de Born (SCBA pour « Self-consistent Born approximation) qui représente une approche itérative plus exigeante en ressources numériques. Nous proposons dans ce travail de thèse une méthode efficace alternative dite LOA pour (« Lowest Order Approximation. Son principal avantage est de réduire considérablement le temps de calcul et de décrire physiquement la diffusion électron-phonon. Cette approche devrait considérablement étendre l'accessibilité de l'utilisation de codes atomistiques de transport quantique pour étudier des systèmes 3D réalistes sans faire à des ressources numériques importantes. / Non-equilibrium Green’s function (NEGF) formalism during recent decades has attracted numerous interests for studying quantum transport properties of nanostructures and nano-devices in which inelastic interactions like electron-phonon scattering have a significant impact. Incorporation of inelastic interactions in NEGF framework is usually performed within the self-consistent Born approximation (SCBA) which induces a numerically demanding iterative scheme. As an alternative technique, we propose an efficient method, the so-called Lowest Order Approximation (LOA) coupled with the Pade approximants. Its main advantage is to significantly reduce the computational time, and to describe the electron-phonon scattering physically. This approach should then considerably extend the accessibility of using atomistic quantum transport codes to study three-dimensional (3D) realistic systems without requiring numerous numerical resources.

In our pursuit of ever smaller transistors, with greater computational throughput, many questions arise about how material properties change with size, and how these properties may be modelled more accurately. Metallic nanocontacts, especially those for which magnetic properties are important, are of great interest due to their potential spintronic applications. Yet, serious challenges remain from the standpoint of theoretical and computational modelling, particularly with respect to the coupling of the spin and lattice degrees of freedom in ferromagnetic nanocontacts in emerging spintronic technologies. In this thesis, an extended method is developed, and applied for the first time, to model the interplay between magnetism and atomic structure in transition metal nanocontacts. The dynamic evolution of the model contacts emulates the experimental approaches used in scanning tunnelling microscopy and mechanically controllable break junctions, and is realised in this work by classical molecular dynamics and, for the first time, spin-lattice dynamics. The electronic structure of the model contacts is calculated via plane-wave and local-atomic orbital density functional theory, at the scalar- and vector-relativistic level of sophistication. The effects of scalar-relativistic and/or spin-orbit coupling on a number of emergent properties exhibited by transition metal nanocontacts, in experimental measurements of conductance, are elucidated by non-equilibrium Green’s Function quantum transport calculations. The impact of relativistic effects during contact formation in non-magnetic gold is quantified, and it is found that scalar-relativistic effects enhance the force of attraction between gold atoms much more than between between atoms which do not have significant relativistic effects, such as silver atoms. The role of non-collinear magnetism in the electronic transport of iron and nickel nanocontacts is clarified, and it is found that the most-likely conductance values reported for these metals, at first- and lastcontact, are determined by geometrical factors, such as the degree of covalent bonding in iron, and the preference of a certain crystallographic orientation in nickel. / Physics / Ph. D. (Physics)