J'étudie en détail la solution d'un modèle simplifié d'électrons fortement corrélés, à savoir le modèle de Hubbard dimérisé. Ce modèle est la réalisation la plus simple d'un problème de cluster DMFT. Je fournis une description détaillée des solutions dans une région de coexistence où l'on trouve deux états (méta) stables des équations DMFT, l'un métallique et l'autre isolant. De plus, je décris en détail comment ces états disparaissent à leurs lignes critiques respectives. Je clarifie le rôle clé joué par la corrélation intra-dimère, qui agit ici en complément des corrélations de Coulomb.Je passe en revue la question importante du passage continue entre unisolant Mott et un isolant Peierls où je caractérise une variété de régimes physiques. Dans un subtil changement de la structure électronique, lesbandes de Hubbard évoluent des bandes purement incohérentes (Mott) à desbandes purement cohérentes (Peierls) à travers un état inattendu au caractère mixte. Je trouve une température d'appariement singulet T* en-dessous de laquelle les électrons localisés à chaque site atomique peuvent se lier dans un singulet et minimiser leur entropie. Ceci constitue un nouveau paradigme d'un isolant de Mott paramagnétique.Enfin, je discute la pertinence de mes résultats pour l'interprétation de différentes études expérimentales sur VO₂. Je présente plusieurs arguments qui me permettent d'avancer la conclusion que la phase métallique, à vie longue (métastable) induite dans les expériences pompe-sonde, et l'état métallique métastable M₁, thermiquement activé dans des nano-domaines, sont identiques. De plus, ils peuvent tous être qualitativement décrits, dans le cadre de notre modèle, par un métal corrélé dimérisé. / We study in detail the solution of a basic strongly correlated model,namely, the dimer Hubbard model. This model is the simplest realization ofa cluster DMFT problem.We provide a detailed description of the solutions in the ``coexistentregion'' where two (meta)stable states of the DMFT equations are found, onea metal and the other an insulator. Moreover, we describe in detail howthese states break down at their respective critical lines. We clarify thekey role played by the intra-dimer correlation, which here acts in additionto the onsite Coulomb correlations.We review the important issue of the Mott-Peierls insulator crossoverwhere we characterize a variety of physical regimes. In a subtle change inthe electronic structure the Hubbard bands evolve from purely incoherent(Mott) to purely coherent (Peierls) through a state with unexpected mixedcharacter. We find a singlet pairing temperature T* below which thelocalized electrons at each atomic site can bind into a singlet and quenchtheir entropy, this uncovers a new paradigm of a para-magnetic Mottinsulator.Finally, we discuss the relevance of our results for the interpretation ofvarious experimental studies in VO₂. We present a variety of argumentsthat allow us to advance the conclusion that the long-lived (meta-stable)metallic phase, induced in pump-probe experiments, and the thermallyactivated M₁ meta-stable metallic state in nano-domains are the same.In fact, they may all be qualitatively described by the dimerizedcorrelated metal state of our model.

Cette thèse se concentre sur des aspects multiorbitales des systèmes fermioniques fortement corrélés. En particulier, sur l'existence d'une différentiation orbitale dans laquelle la coexistence de caractère itinérant et localisé peut être associée à différentes orbitales. Cette problématique est examinée dans le contexte des atomes froids et des matériaux, offrant un pont entre les deux communautés.Dans la première partie de la thèse, nous donnons un aperçu du problème des corrélations fortes dans les matériaux, et nous introduisons le concept de 'transition de Mott sélective en orbitales'. Nous fournissons également les principaux outils pour comprendre comment les matériaux peuvent être simulés avec des atomes froids, et nous présentons des résultats importants liés à la transition métal-isolant de Mott. Les aspects techniques, basées sur la théorie du champ moyen dynamique sont également discutés, et la solution de deux principaux modèles de systèmes fermioniques fortement corrélés, à savoir le modèle d'Hubbard (HM) et le modèle de Falicov-Kimball (FKM), sont passés en revue.Ensuite, nous étudions en détail la physique de deux espèces fermioniques en interaction forte avec des masses différentes dans un réseau optique. Nous établissons les différentes phases (avec et sans ordre à longue portée) en termes de la force des interactions (U), du rapport des masses et de la température (T), et aussi nous discutons les variables thermodynamiques, qui sont pertinentes pour les expériences d'atomes froids. Nous montrons que dans la phase métallique (U inférieure à une valeur critique) et avec un certain degré de différence de masses, un 'crossover' apparaît entre un état métallique du type de liquide de Fermi à basse T, et un état avec différentiation orbital à haute T, où les fermions lourds se localisent tandis que les fermions légers restent itinérant. Par conséquent, nous proposons ce modèle minimal pour étudier la physique des systèmes qui présentent une différentiation orbitale avec des expériences d'atomes froids.Basé sur les propriétés du modèle étudié, nous proposons la 'chromatographie entropique' comme une nouvelle méthode pour refroidir des atomes fermioniques dans les réseaux optiques. Nous discutons son efficacité et ses limites, et fournissons quelques idées afin de les surmonter.Dans la dernière partie de la thèse, nous généralisons le modèle précédent aux matériaux corrélés à plusieurs bandes qui permet d'afficher la différentiation orbitale. Nous montrons que l'état de Mott sélectif en orbital peut être stable sous les distorsions du réseau, modélisées par une hybridation locale entre les orbitales. Cependant, l'état de Mott est caractérisé par un pseudo-gap, où les fluctuations de charge sont brusquement réduites, mais l'état reste compressible. En relation au modèle précédent, nous discutons le 'crossover' entre l'état métallique et l'état sélectif induit par des effets température, nous comparons nos résultats avec les expériences de photoémission, et nous prédisons ce qui se passerait dans les matériaux qui présentent une hybridation locale entre les bandes. / This thesis focuses on multiorbital aspects of strongly correlated fermionic systems. In particular, it focuses on the existence of orbital differentiation in which coexistence of itinerant and localized character can be associated to different orbitals. This subject is discussed in the context of cold atoms and materials, providing a bridge between both communities.In the first part of the thesis, we give an insight into the problem of strong correlations in materials, and we introduce the concept of 'orbital-selective Mott transition'. We also provide the main tools to understand how materials can be simulated with cold atoms experiments, and we present important related results in the context of the metal-Mott insulator transition. The technical aspects, based on dynamical mean-field theory are also discussed, and the solution of two key models of strongly correlated fermionic systems, i.e., the Hubbard model (HM) and the Falicov-Kimball model (FKM), are reviewed.Then we study in detail the physics of two interacting fermionic species with different masses in an optical lattice. We establish the different phases (with and without long-range order) in terms of the interactions strength (U), mass ratio and temperature (T), and also discuss the thermodynamic variables, which are relevant in cold atoms experiments. We show that in the metallic phase (U below a critical value) and for some degree of mass imbalance, a crossover appears between a Fermi-liquid metallic state at low T, and an 'orbital-selective' state at higher T, where the heavy fermions effectively localize while the light species remain itinerant. Hence, we propose this minimal model for addressing orbital-selective physics with cold atoms experiments.Based on the properties of the studied model, we propose the 'entropic chromatography' as a new method for cooling fermionic atoms in optical lattices. We discuss its efficiency and limitations, and provide some ideas in order to overcome them.In the last part of the thesis we generalize the previous model to a model relevant for multiband correlated materials that can display orbital differentiation. We show that the orbital-selective Mott state can be stable under lattice distortions modeled by local hybridization between the orbitals. However, the Mott state is characterized by a pseudogap, where charge fluctuations abruptly reduce, but the state remains compressible. In connection with the previous model, we discuss the temperature-induced orbital-selective crossover in this problem, we compare our results with photoemission experiments, and predict what would happen in materials that display local hybridization between the bands.

Mn5Ge3 suscite de l'intérêt pour des applications dans le domaine de l'électronique de spin car il a une température de Curie élevée (≈300 K) et il peut croître épitaxialement sur des substrats Ge(111) permettant ainsi d'injecter directement dans le semi-conducteur Ge un courant polarisé en spin.Nous avons étudié par photoémission résolue en angle et en spin (ARPES, SARPES), utilisant le rayonnement synchrotron, des films minces de Mn5Ge3(001), obtenus par croissance sur la surface reconstruites Ge(111)-c(2x8).Les résultats ARPES, obtenus dans les plans GALM et GAHK, sont en accord avec des simulations faites sur la base de calculs de structure de bandes faisant appel à la théorie de la fonctionnelle de la densité.Les mesures SARPES faites en plusieurs points du plan GALM sont aussi bien reproduites par ces simulations.D'une façon globale, nos résultats apportent une validation remarquable de la description des propriétés électroniques de Mn5Ge3 par le modèle de bandes. Seule l'intensité spectrale au niveau de Fermi n'est pas bien expliquée par la simulation. Cette différence est attribuée à la nature tridimensionnelle de l'échantillon et à des effets de corrélation. / Mn5Ge3 attracts strong interest for spintronics applications because it has a high Curie-temperature (≈300 K) and it can be grown epitaxially on Ge(111) substrates, permitting direct injection of a spin-polarized current into the Ge semiconductor.Mn5Ge3(001) thin films grown on Ge(111)-c(2x8) reconstructed surface were studied by angle- and spin- resolved photoemission (ARPES, SARPES) using synchrotron radiation. ARPES results, obtained in the GALM and GAHK planes, are in agreement with simulations done with the help of band structure calculations based on the density functional theory, taking into account lifetime broadening and broadening caused by correlation effects.SARPES measurements done at several k-points of the GALM plane are also well acounted for by these simulations.Overall our results provide a remarkable validation of the band structure model for a proper description of the electronic properties of Mn5Ge3. Only the spectral intensity at the Fermi level is not well explained by the simulation. This departure is attributed to the 3D nature of the sample and to correlation effects.

Nesta dissertação estudaremos sistemas de bósons ultrafrios armadilhados em uma rede ótica quadrada bidimensional sem levar em consideração o confinamento harmônico. A dinâmica desses sistemas é bem descrita pelo modelo de Bose-Hubbard, que prevê uma transição de fase quântica de um superfluido para um isolante de Mott a temperaturas baixas, e pode ser induzida variando a profundidade do potencial da rede ótica. Apresentaremos o diagrama de fases dessa transição construído a partir de uma aproximação de campo médio e também com um cálculo numérico usando um algoritmo de Monte Carlo Quântico, denominado algoritmo Worm. Encontramos o ponto crítico para o primeiro lobo de Mott em ambos os casos, concordando com trabalhos anteriores. / This work study the two-dimensional ultracold bosonic atoms loaded in a square optical lattice, without harmonic confinement. The dynamics of this system is described by the Bose-Hubbard model, which predicts a quantum phase transition from a superfluid to a Mott-insulator at low temperatures that can be induced by varying the depth of the optical potential. We present here the phase diagram of this transition built from a mean field approach and from a numerical calculation using a Quantum Monte Carlo algorithm, namely the Worm algorithm. We found the critical transition point for the first Mott lobe in both cases, in agreement with the standard literature.

The localized/delocalized duality of 5f electrons plays an important role in understanding the complex physics of actinides. Band-structure calculations based on the ad hoc assumption that 5f electrons are simultaneously localized and delocalized explained the observed dHvA experiments very well. This ad hoc assumption also gives the correct equilibrium volume for delta-Pu. Experimentally, the duality of 5f electrons is observed by inelastic neutron scattering experiments, or by soft X-ray angle-resolved photoelectron spectroscopy. It is worth recalling that the origin of partial localization in the 3d and 5f systems is quite different. In compounds with 3d electrons, the large crystalline electric field set up by the surrounding environment of transition metal ions plays a major role. On the other hand, in 5f systems, the Hund's rule correlations play the key role whilst the crystalline electric field is less important. In this thesis we have studied the effect of intra-atomic correlations on anisotropies in hopping matrix elements of different 5f orbitals. For that purpose, we used the effective model that includes on-site interactions that are responsible for Hund's rules and effective hopping terms that result from the hybridization of different 5f orbitals with the environment. Two different approximations, namely, rotationally invariant slave-boson mean-field (RISBMF) and infinite time-evolving block decimation (iTEBD), have been used to investigate the ground-state properties of the Hamiltonian. We have demonstrated that Hund's rule correlations enhance strongly anisotropies in hopping matrix elements. For a certain range of 5f bandwidth parameters this effect may result in a complete suppression of hopping processes for some of 5f orbitals, i.e., the system is in a partially localized phase. Within the RISBMF method, we calculated the ground-state properties and the phase diagram of the system. The suppression of hopping processes in some of 5f orbitals due to Hund's rule correlations can be seen through orbital-dependent quasiparticle weights. In a mean-field theory, a quasiparticle weight of zero for an orbital means a complete suppression of hopping processes in this orbital. Thus, quasiparticle weights and occupation numbers were used to classify partially localized phases. In the calculated phase diagram we obtain four partially localized phases that can be separated into two different sets. In the first set electrons in two orbitals are localized. In the second, electrons in one orbital are localized. The difference between the two sets is not simply the number of localized orbitals but the mechanism for the partial localization. For the first set, the Hund's rule mechanism applies: only those 5f electrons that enable the remaining ones to form a Hund's rule state will delocalize. This mechanism requires to have at least two localized orbitals, therefore it is definitely not applicable to those phases with only one localized orbital. For the second set, a situation similar to a single-band Mott-Hubbard transition applies. The direct on-site Coulomb interaction between jz and -jz electrons plays the key role for understanding the partial localization transition. In order to assess the validity of the RISBMF results we have used the iTEBD method to calculate the ground-state properties of a 1D system. Qualitatively, the two approaches agree with each other. However, we found an area where the RISBMF yields an artificial ground-state. Note that the mean-field method is worst for a 1D system. Therefore one shoud not judge from it the quality of the RISBMF method for the more general case.

Nesta dissertação estudaremos sistemas de bósons ultrafrios armadilhados em uma rede ótica quadrada bidimensional sem levar em consideração o confinamento harmônico. A dinâmica desses sistemas é bem descrita pelo modelo de Bose-Hubbard, que prevê uma transição de fase quântica de um superfluido para um isolante de Mott a temperaturas baixas, e pode ser induzida variando a profundidade do potencial da rede ótica. Apresentaremos o diagrama de fases dessa transição construído a partir de uma aproximação de campo médio e também com um cálculo numérico usando um algoritmo de Monte Carlo Quântico, denominado algoritmo Worm. Encontramos o ponto crítico para o primeiro lobo de Mott em ambos os casos, concordando com trabalhos anteriores. / This work study the two-dimensional ultracold bosonic atoms loaded in a square optical lattice, without harmonic confinement. The dynamics of this system is described by the Bose-Hubbard model, which predicts a quantum phase transition from a superfluid to a Mott-insulator at low temperatures that can be induced by varying the depth of the optical potential. We present here the phase diagram of this transition built from a mean field approach and from a numerical calculation using a Quantum Monte Carlo algorithm, namely the Worm algorithm. We found the critical transition point for the first Mott lobe in both cases, in agreement with the standard literature.

The vicinity of continuous quantum phase transitions displays unique properties such as scaling behavior and incoherent excitation spectra which are not found in any stable phase of matter. This fascinating quantum critical regime is crucial for progress on key problems of modern condensed matter physics. The three research projects of this thesis challenge and refine our understanding of quantum criticality in different ways. Part I concerns unexpected quantum critical behavior near the Mott transition. The bandwidth-controlled Mott transition in the half-filled one-band Hubbard model is one of the most paradigmatic phenomena of strongly correlated physics. Within the approximation of dynamical mean-field theory (DMFT) this metal-insulator transition is of first order at low temperatures, with the transition line ending at a critical temperature. Surprisingly, numerical calculations with DMFT and experiments in organic salts consistently found quantum critical scaling of the resistivity above the critical temperature. The aim of this project is to explain this unexpected scaling in the absence of a quantum critical point in the phase diagram. To this end, we perform extensive DMFT simulations with the numerical renormalization group as a state-of-the-art impurity solver. We find that the quantum critical scaling can be traced back to the metastable insulator at the boundary of the coexistence region at T = 0 which exhibits previously unknown scale-invariance on the frequency axis. In Part II we study how magnetic quantum criticality is affected by the coupling to additional non-critical degrees of freedom. Considering typical electronic energy scales the study of quantum critical phenomena in magnets requires very low temperatures in the sub-100mK range. In this regime additional effects which are typically neglected in the theoretical modeling may become important. Here we focus on one particular example, which is the hyperfine coupling to nuclear spins. We investigate the fate of the quantum critical behavior at lowest temperatures and determine crossover scales below which a purely electronic description is no longer sufficient. Explicit calculations for paradigmatic models on the level of mean-field theory plus Gaussian fluctuations reveal that the quantum phase transition can be shifted or smeared in the presence of nuclear spins. More exotic effects of nuclear spins, e.g. in spin liquids, are discussed on a qualitative level. Part III is devoted to the discussion of domain-driven phase transitions in easy-axis ferromagnets.This work is motivated by an experimental study of LiHoF4, a dipolar easy-axis ferromagnet that displays a well-studied quantum phase transition from a ferromagnetic to a paramagnetic phase as function of a transverse field. Measurements of the ac susceptibility found a well-defined phase transition even in tilted fields where the Ising symmetry is explicitly broken and Landau theory of the microscopic order parameter predicts a crossover. We are able to explain and model the transition in tilted fields by the inclusion of domain effects, i.e., by taking into account the spontaneous breaking of translational symmetry by mesoscale pattern formation in the ferromagnetic phase. The modeling of stray-field energies as effective antiferromagnetic couplings between magnetization components in different domains is in excellent quantitative agreement with the experimental results.:1 Phases and their transitions . . . . . . . . . . . . . . . . . . . . 4 1.1 Thermal and quantum phase transitions . . . . . . . . . . . . . . . . . . . . 4 1.2 Theoretical description of phase transitions . . . . . . . . . . . . . . . . . . 8 1.3 Project overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 I Mott quantum criticality in the one-band Hubbard model . . . . . . . . . . .15 2 Introduction to the Mott transition . . . . . . . . . . . . . . . . . . . . 16 2.1 Metal-insulator transitions and the Hubbard model . . . . . . . . . . . . . . 16 2.2 A local perspective: the idea of dynamical mean-field theory . . . . . . . . . 19 2.3 Quantum critical scaling near the Mott transition . . . . . . . . . . . . . . . 21 3 Dynamical mean-field theory (DMFT) . . . . . . . . . . . . . . . . . . . . 25 3.1 Single-impurity Anderson model . . . . . . . . . . . . . . . . . . . . . . . . 25 3.2 Theoretical foundations of DMFT . . . . . . . . . . . . . . . . . . . . . . . 28 3.3 Wilson's numerical renormalization group . . . . . . . . . . . . . . . . . . . 32 3.4 Implementation and choice of parameters . . . . . . . . . . . . . . . . . . . 36 4 Power-law spectra and quantum critical scaling . . . . . . . . . . . . . . . . . . 38 4.1 Scale-invariant solutions of DMFT . . . . . . . . . . . . . . . . . . . . . . . 38 4.2 Spectral power laws at T=0 in the metastable insulator . . . . . . . . . . . 40 4.3 Finite-temperature crossovers in the spectral function . . . . . . . . . . . . 47 4.4 Resistivity scaling driven by spectral power laws . . . . . . . . . . . . . . . 50 4.5 Scaling analysis of the dynamic susceptibility . . . . . . . . . . . . . . . . . 58 4.6 Ideas and obstacles towards an analytical understanding . . . . . . . . . . . 62 4.7 Conclusions and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 II Limits on magnetic quantum criticality from nuclear spins . . . . . . . . . . . . .65 5 Stability of magnetic transitions to hyperfine coupling . . . . . . . . . . . . . . . .66 5.1 Nuclear spins near quantum criticality . . . . . . . . . . . . . . . . . . . . . 66 5.2 Introduction to nuclear spins and hyperfine coupling . . . . . . . . . . . . . 67 5.3 Magnetic phases in the presence of nuclear spins . . . . . . . . . . . . . . . 69 5.4 Two scenarios for magnetic quantum criticality plus nuclear spins . . . . . . 70 6 Paradigmatic models for magnetic quantum phase transitions . . . . . . . . . 73 6.1 Transverse-field Ising model . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 6.2 Coupled-dimer model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 6.3 Frustrated spin models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 7 Crossover scales introduced by nuclear spins . . . . . . . . . . . . . . . . . . .83 7.1 Shifted transitions: transverse-field Ising magnets . . . . . . . . . . . . . . . 83 7.2 Smeared transitions: coupled-dimer magnets . . . . . . . . . . . . . . . . . . 90 7.3 Additional transitions due to nuclear spins . . . . . . . . . . . . . . . . . . . 98 7.4 Exotic magnetic quantum phase transitions plus nuclear spins . . . . . . . . 101 7.5 Conclusions and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 III Domain-driven phase transitions in easy-axis ferromagnets . . . . . . . . 105 8 Easy-axis ferromagnet LiHoF4 . . . . . . . . . . . . . . . . . . . . 106 8.1 Easy-axis ferromagnets in tilted fields . . . . . . . . . . . . . . . . . . . . . 106 8.2 LiHoF4 and its phase transitions . . . . . . . . . . . . . . . . . . . . . . . . 109 9 Modeling of microscopic degrees of freedom in LiHoF4 . . . . . . . . . . . . 112 9.1 Landau theory in tilted fields . . . . . . . . . . . . . . . . . . . . . . . . . . 112 9.2 Crystal field effects and microscopic Hamiltonian . . . . . . . . . . . . . . . 113 9.3 Crossovers in the microscopic model . . . . . . . . . . . . . . . . . . . . . . 118 10 Modeling of mesoscopic degrees of freedom in LiHoF4 . . . . . . . . . . . . . . .123 10.1 Domains in ferromagnets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 10.2 Modeling of domain effects as effective interactions . . . . . . . . . . . . . . 127 10.3 Combined mean-field Hamiltonian and domain optimization . . . . . . . . . 130 10.4 Nature of the phase transition in tilted fields . . . . . . . . . . . . . . . . . 132 10.5 Domain-driven phase transition at T = 0 . . . . . . . . . . . . . . . . . . . . 135 10.6 Domain-driven phase transition at finite temperatures . . . . . . . . . . . . 141 10.7 Comparison with experimental results . . . . . . . . . . . . . . . . . . . . . 146 10.8 Conclusions and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 IV Summary & Outlook . . . . . . . . . . . . . . . . . . . . 151 V Appendices . . . . . . . . . . . . . . . . . . . . 155 A Part I: NRG level spectra . . . . . . . . . . . . . . . . . . . . 156 B Part I: Analytical properties of scale-invariant DMFT solutions . . . . . . . . . . .159 B.1 Kondo perturbation theory as an impurity solver . . . . . . . . . . . . . . . 159 B.2 Analytical properties of a power-law self-energy . . . . . . . . . . . . . . . . 166 C Part I: Scaling analysis of the resistivity . . . . . . . . . . . . . . . . . . 168 D Part II: Solution of the transverse-field Ising model with nuclear spins . . . . . . 172 D.1 Holstein-Primakoff representation of the electronic and nuclear spins . . . . 172 D.2 Determination of the classical reference state . . . . . . . . . . . . . . . . . 174 D.3 Excitation spectrum of the coupled nuclear-electronic model . . . . . . . . . 175 D.4 Magnetization, susceptibility, and heat capacity . . . . . . . . . . . . . . . . 177 E Part II: Solution of the coupled-dimer model with nuclear spins . . . . . . . . . . . 181 E.1 Bond-operator description of the electronic spins . . . . . . . . . . . . . . . 181 E.2 Determination to the electronic ground state . . . . . . . . . . . . . . . . . 185 E.3 Holstein-Primakoff representation of the nuclear spins . . . . . . . . . . . . 188 E.4 Excitation spectrum of the coupled nuclear-electronic model . . . . . . . . . 189 E.5 Staggered magnetization and susceptibility . . . . . . . . . . . . . . . . . . 192 F Part III: Calculation of domain-induced effective interactions . . . . . . . . . . . . . 198 Bibliography . . . . . . . . . . . . . . . . . . . . 203

A nickel alloy with high chrome and molybdenum content was found to form a highly resistive and passive oxide layer. The donor density and mobility of ions in the oxide layer has been determined as a function of the electrical potential when alkaline water layers are on the alloy surface in order to account for the relative inertness of the nickel alloy in corrosive environments.

A unique feature of the 5d-iridates is that the spin-orbit interaction (SOI) and Coulomb interactions U are of comparable strength and therefore compete vigorously. The relative strength of these interactions stabilizes new exotic ground states that provide a fertile ground for studying new physics. SOI is proportional to Z^4 (Z is the atomic number), and it is now recognized that strong SOI can drive novel narrow-gap insulating states in heavy transition metal oxides such as iridates. Indeed, strong SOI necessarily introduces strong lattice degrees of freedom that become critical to new physics in the iridates. This dissertation thoroughly examines a wide array of newly observed novel phenomena induced by adjusting the relative strengths of U and SOI interactions via slight chemical doping and application of hydrostatic pressure in the layered iridates, particularly, BaIrO3 and Sr2IrO4.

Die Untersuchung stark korrelierter Elektronensysteme anhand des zweidimensionalen Hubbard-Modells bildet das zentrale Thema dieser Arbeit. Wir analysieren das Schicksal des Mott-Isolators bei Dotierung als auch bei Reduzierung der Wechselwirkungsstärke. Die numerische Auswertung erfolgt mit Hilfe von Quanten-Cluster-Approximationen, die eine thermodynamisch konsistente Beschreibung der Grundzustandseigenschaften garantieren. Der hier verwendete Rahmen der Selbstenergiefunktional-Theorie bietet eine große Flexibilität bei der Konstruktion von Cluster-Näherungen. Eine detaillierte Analyse gibt Aufschluss über die Qualität und das Konvergenzverhalten unterschiedlicher Cluster-Näherungen innerhalb der Selbstenergiefunktional-Theorie. Wir verwenden für diese Untersuchungen das eindimensionale Hubbard-Modell und vergleichen unsere Resultate mit der exakten Lösung. In zwei Dimensionen finden wir als Grundzustand des Teilchen-Loch-symmetrischen Modells bei Halbfüllung einen antiferromagnetischen Isolator unabhängig von der Wechselwirkungsstärke. Die Berücksichtigung kurzreichweitiger räumlicher Korrelationen durch unsere Cluster-Näherung führt, im Vergleich mit der dynamischen Mean-Field-Theorie, zu einer deutlichen Verbesserung des antiferromagnetischen Ordnungsparameters. Darüberhinaus beobachten wir in der paramagnetischen Phase einen Metall-Isolator-Übergang als Funktion der Wechselwirkungsstärke, der sich qualitativ vom reinen Mean-Field-Szenario unterscheidet. Ausgehend vom antiferromagnetischen Mott-Isolator zeigt sich ein füllungsgetriebener Metall-Isolator-Übergang in eine paramagnetische metallische Phase. Abhängig von der verwendeten Cluster-Approximation tritt dabei zunächst eine antiferromagnetische metallische Phase auf. Neben langreichweitiger antiferromagnetischer Ordnung haben wir in unseren Rechnungen auch Supraleitung berücksichtigt. Das Verhalten des supraleitenden Ordnungsparameters als Funktion der Dotierung ist dabei in guter Übereinstimmung sowohl mit anderen numerischen Verfahren als auch mit experimentellen Ergebnissen. / The central goal of this thesis is the examination of strongly correlated electron systems on the basis of the two-dimensional Hubbard model. We analyze how the properties of the Mott insulator change upon doping and with interaction strength. The numerical evaluation is done using quantum cluster approximations, which allow for a thermodynamically consistent description of the ground state properties. The framework of self-energy-functional theory offers great flexibility for the construction of cluster approximations. A detailed analysis sheds light on the quality and the convergence properties of different cluster approximations within the self-energy-functional theory. We use the one-dimensional Hubbard model for these examinations and compare our results with the exact solution. In two dimensions the ground state of the particle-hole symmetric model at half-filling is an antiferromagnetic insulator, independent of the interaction strength. The inclusion of short-range spatial correlations by our cluster approach leads to a considerable im\-prove\-ment of the antiferromagnetic order parameter as compared to dynamical mean-field theory. In the paramagnetic phase we furthermore observe a metal-insulator transition as a function of the interaction strength, which qualitatively differs from the pure mean-field scenario. Starting from the antiferromagnetic Mott insulator a filling-controlled metal-insulator transition in a paramagnetic metallic phase can be observed. Depending on the cluster approximation used an antiferromagnetic metallic phase may occur at first. In addition to long-range antiferromagnetic order, we also considered superconductivity in our calculations. The superconducting order parameter as a function of doping is in good agreement with other numerical methods, as well as with experimental results.