Ces dernières décennies, la constante évolution des besoins de stockage de données a mené à un bouleversement du paysage technologique qui s’est complètement métamorphosé et réinventé. Depuis les débuts du stockage magnétique jusqu’aux plus récents dispositifs fondés sur l’électronique dit d’état solide, la densité de bits stockés continue d’augmenter vers ce qui semble du point de vue du consommateur comme des capacités de stockage et des performances infinies. Cependant, derrière chaque transition et évolution des technologies de stockage se cachent des limitations en termes de densité et performances qui nécessitent de lourds travaux de recherche afin d’être surmontées et repoussées. Ce manuscrit s’articule autour d’une technologie émergeante prometteuse ayant pour vocation de révolutionner le paysage du stockage de données : la mémoire à pont conducteur ou Conductive Bridge Random Access Memory (CBRAM). Cette technologie est fondée sur la formation et dissolution réversible d’un chemin électriquement conducteur dans un électrolyte solide. Elle offre de nombreux avantages face aux technologies actuelles tels qu’une faible consommation électrique, de très bonnes performances d’écriture et de lecture et la capacité d’être intégré aux seins des interconnexions métalliques d’une puce afin d’augmenter la densité de stockage. Malgré tout, pour que cette technologie soit compétitive certaines limitations ont besoin d’être surmontées et particulièrement sa variabilité et sa stabilité thermique qui posent encore problème. Ce manuscrit propose une compréhension physique globale du fonctionnement de la technologie CBRAM fondée sur une étude expérimentale approfondie couplée à un modèle Monte Carlo cinétique spécialement développé. Cette compréhension fait le lien entre les propriétés physiques des matériaux composant la mémoire CBRAM et ses performances (Tension et temps d’écriture et d’effacement, rétention de donnée, endurance et variabilité). Un fort accent est mis la compréhension des limites actuelle de la technologie et comment les repousser. Grâce à une optimisation des conditions d’opérations ainsi qu’à un travail d’ingénierie des dispositifs mémoire, il est démontré dans ce manuscrit une forte amélioration de la stabilité thermique ainsi que de la variabilité des états écrits et effacés. / The constant evolution of the data storage needs over the last decades have led the technological landscape to completely change and reinvent itself. From the early stage of magnetic storage to the most recent solid state devices, the bit density keeps increasing toward what seems from a consumer point of view infinite storage capacity and performances. However, behind each storage technology transition stand density and performances limitations that required strong research work to overcome. This manuscript revolves around one of the promising emerging technology aiming to revolutionize data storage landscape: the Conductive Bridge Random Access Memory (CBRAM). This technology based on the reversible formation and dissolution of a conductive path in a solid electrolyte matrix offers great advantages in term of power consumption, performances, density and the possibility to be integrated in the back end of line. However, for this technology to be competitive some roadblocks still have to be overcome especially regarding the technology variability, reliability and thermal stability. This manuscript proposes a comprehensive understanding of the CBRAM operations based on experimental results and a specially developed Kinetic Monte Carlo model. This understanding creates bridges between the physical properties of the materials involved in the devices and the devices performances (Forming, SET and RESET time and voltage, retention, endurance, variability). A strong emphasis is placed on the current limitations of the technology previously stated and how to overcome these limitations. Improvement of the thermal stability and device reliability are demonstrated with optimized operating conditions and proper devices engineering.

A l'échelle nanométrique, les solides peuvent changer de forme par diffusion de surface, et présentent alors des propriétés de mouillage qui s'apparentent à celles des liquides. Dans cette thèse, nous nous sommes plus particulièrement intéressés au comportement de mouillage des nano-solides sur des substrats nanopatternés, comportant par exemple des piliers ou des tranchées. Sur ces substrats, les nanoparticules (ou ilots) solides peuvent être multi-stables : c'est-à-dire qu'ils peuvent présenter plus d'un état localement stable. Comme les liquides, les solides ont été observés par exemple dans des états dits de Wenzel (pénétrant dans la structure du substrat) ou de Cassie-Baxter (ne pénétrant pas). Grâce à une combinaison de simulations Monte Carlo Cinétiques et de modèles analytiques, nous avons étudié la stabilité de ces états et leur dynamique de transition. Plus particulièrement, avons mis en évidence le rôle de la diffusion de surface et de la nucléation bidimensionnelle sur la dynamique de transition. Nous avons aussi montré que les contraintes élastiques augmentent la stabilité des états de Cassie-Baxter, et mènent à de nouveaux états, avec des morphologies asymétriques ou partiellement empalées dans les nanostructures. Finalement, nous avons proposé de contrôler les transitions de mouillage à l'aide de l'électromigration induite par un faisceau d'électrons. Nos résultats ouvrent la voie vers une nouvelle direction pour les investigations expérimentales / At the nanometer scale, solids can change shape thanks to surface diffusion and therefore display wetting properties that can be likened to those of liquids. This doctoral thesis intends to study particularly the wetting behaviour of nano-solids located on nanopatterned substrates, containing for instance pillars or trenches. Upon these substrates, solid nanoparticles (or islands) can be multi-stable – that is to say they can display more than one locally-stable state. Just like liquids, solids have been observed for example in the context of the so-called Wenzel state (penetrating the very structure of the substrate) and Cassie-Baster state (no penetration). By combining Kinetic Monte Carlo simulations with analytical models, we conducted a study on the stability of these states along with their dynamics of transition. In particular, we highlighted the specific roles that surface diffusion and bidimensional nucleation play in regards to the dynamics of transition. We also demonstrated that elastic constraints increase the stability of Cassie-Baxter states and lead to new states, with either asymmetric morphologies or morphologies that are partially impaled into the nanostructures. Last but not least, we proposed to control wetting transitions using the electromigration brought on by an electron beam. Our results pave the way for a new direction in the field of experimental investigations

Molecular beam epitaxy has recently been applied to the growth and self assembly of nanostructures on crystal substrates. This highlights the importance of understanding how microscopic rules of atomic motion and assembly lead to macroscopic surface shapes. In this thesis, we present results from two computational studies of these mechanisms. We identify a kinetic mechanism responsible for the emergence of low-angle facets in recent epitaxial regrowth experiments on patterned surfaces. Kinetic Monte Carlo simulations of vicinal surfaces show that the preferred slope of the facets matches the threshold slope for the transition between step flow and growth by island nucleation. At this crossover slope, the surface step density is minimized and the adatom density is maximized, respectively. A model is developed that predicts the temperature dependence of the crossover slope and hence the facet slope. We also examine the "step bunching" instability thought to be present in step flow growth on surfaces with a downhill diffusion bias. One mechanism thought to produce the necessary bias is the inverse Ehrlich Schwoebel (ES) barrier. Using continuum, stochastic, and hybrid models of one dimensional step flow, we show that an inverse ES barrier to adatom migration is an insufficient condition to destabilize a surface against step bunching. / Science, Faculty of / Physics and Astronomy, Department of / Graduate

No description available.

This thesis addresses the metal nanocluster growth process on prepatterned substrates, the development of atomistic simulation method with respect to an acceleration of the atomistic transition states, and the continuum model of the ion-beam inducing semiconductor surface pattern formation mechanism. Experimentally, highly ordered Ag nanocluster structures have been grown on pre-patterned amorphous SiO^2 surfaces by oblique angle physical vapor deposition at room temperature. Despite the small undulation of the rippled surface, the stripe-like Ag nanoclusters are very pronounced, reproducible and well-separated. The first topic is the investigation of this growth process with a continuum theoretical approach to the surface gas condensation as well as an atomistic cluster growth model. The atomistic simulation model is a lattice-based kinetic Monte-Carlo (KMC) method using a combination of a simplified inter-atomic potential and experimental transition barriers taken from the literature. An effective transition event classification method is introduced which allows a boost factor of several thousand compared to a traditional KMC approach, thus allowing experimental time scales to be modeled. The simulation predicts a low sticking probability for the arriving atoms, millisecond order lifetimes for single Ag monomers and ≈1 nm square surface migration ranges of Ag monomers. The simulations give excellent reproduction of the experimentally observed nanocluster growth patterns. The second topic specifies the acceleration scheme utilized in the metallic cluster growth model. Concerning the atomistic movements, a classical harmonic transition state theory is considered and applied in discrete lattice cells with hierarchical transition levels. The model results in an effective reduction of KMC simulation steps by utilizing a classification scheme of transition levels for thermally activated atomistic diffusion processes. Thermally activated atomistic movements are considered as local transition events constrained in potential energy wells over certain local time periods. These processes are represented by Markov chains of multi-dimensional Boolean valued functions in three dimensional lattice space. The events inhibited by the barriers under a certain level are regarded as thermal fluctuations of the canonical ensemble and accepted freely. Consequently, the fluctuating system evolution process is implemented as a Markov chain of equivalence class objects. It is shown that the process can be characterized by the acceptance of metastable local transitions. The method is applied to a problem of Au and Ag cluster growth on a rippled surface. The simulation predicts the existence of a morphology dependent transition time limit from a local metastable to stable state for subsequent cluster growth by accretion. The third topic is the formation of ripple structures on ion bombarded semiconductor surfaces treated in the first topic as the prepatterned substrate of the metallic deposition. This intriguing phenomenon has been known since the 1960\'s and various theoretical approaches have been explored. These previous models are discussed and a new non-linear model is formulated, based on the local atomic flow and associated density change in the near surface region. Within this framework ripple structures are shown to form without the necessity to invoke surface diffusion or large sputtering as important mechanisms. The model can also be extended to the case where sputtering is important and it is shown that in this case, certain \\lq magic\' angles can occur at which the ripple patterns are most clearly defined. The results including some analytic solutions of the nonlinear equation of motions are in very good agreement with experimental observation.

<div>Accelerating the development of novel materials is one of the central goals of the Materials Genome Initiative and improving the predictive power of computational</div><div>material science methods is critical to attain this goal. Molecular dynamics (MD) is one such computational technique that has been used to study a wide range of materials since its invention in the 1950s. In this work we explore some examples of using and increasing the predictive power of MD simulations to understand materials phenomena and provide guidelines to design tailored materials. We first demonstrate the use of MD simulations as a tool to explore the design space of shape memory alloys, using simple interatomic models to identify characteristics of an integrated coherent second phase that will modify the transformation characteristics of the base shape memory alloy to our desire. Our approach provides guidelines to identify potential coherent phases that will achieve tailored transformation temperatures and hysteresis. </div><div><br></div><div>We subsequently explore ideas to enhance the length and time scales accessible via MD simulations. We first discuss the use of kinetic Monte Carlo methods in MD simulations to predict the microstructure evolution of carbon fibers. We ?find our approach to accurately predict the transverse microstructures of carbon fibers, additionally predicting the transverse modulus of these fibers, a quantity difficult to measure via experiments. Another avenue to increase length and time scales accessible via MD simulations is to explore novel implementations of algorithms involved in machine-learned interatomic models to extract performance portability. Our approach here results in significant speedups and an efficient utilization of increasingly common CPU-GPU hybrid architectures.</div><div><br></div><div>We finally explore the use of machine learning methods in molecular dynamics, specifically developing machine learning methods to discover interpretable laws directly from data. As examples, we demonstrate the discovery of integration schemes for MD simulations, and the discovery of melting laws for perovskites and single elements. Overall, this work attempts to illustrate how improving the predictive capabilities of molecular dynamics simulations and incorporating machine learning ideas can help us design novel materials, in line with the goals of the Materials Genome Initiative.</div>

Single-crystalline nanostructures often exhibit gradients of surface (and/or interface) curvature that emerge from fabrication and growth processes or from thermal fluctuations. Thus, the system-inherent capillary force can initiate morphological transformations during further processing steps or during operation at elevated temperature. Therefore and because of the ongoing miniaturization of functional structures which causes a general rise in surface-to-volume ratios, solid-state capillary phenomena will become increasingly important: On the one hand diffusion-mediated capillary processes can be of practical use in view of non-conventional nanostructure fabrication methods based on self-organization mechanisms, on the other hand they can destroy the integrity of nanostructures which can go along with the failure of functionality. Additionally, capillarity-induced shape transformations are effected and can thereby be controlled by applied fields and forces (guided or driven evolution). With these prospects and challenges at hand, formation and shape transformation of single-crystalline nanostructures due to the system-inherent capillary force in combination with external fields or forces are investigated in the frame of this dissertation by means of atomistic computer simulations. For the exploration (search, description, and prediction) of reaction pathways of nanostructure shape transformations, kinetic Monte Carlo (KMC) simulations are the method of choice. Since the employed KMC code is founded on a cellular automaton principle, the spatio-temporal development of lattice-based N-particle systems (N up to several million) can be followed for time spans of several orders of magnitude, while considering local phenomena due to atomic-scale effects like diffusion, nucleation, dissociation, or ballistic displacements. In this work, the main emphasis is put on nanostructures which have a cylindrical geometry, for example, nanowires (NWs), nanorods, nanotubes etc.

The irradiation of crystalline materials in environments such as nuclear reactors leads to the accumulation of micro and nano-scale defects with a negative impact on material properties such as strength, corrosion resistance, and dimensional stability. Point defects in the crystal lattice, the vacancy and self-interstitial, form the basis of this damage and are capable of migrating through the lattice to become part of defect clusters and sinks, or to annihilate themselves. Recently, attention has been given to HEAs for fusion and fission components, as some materials of this class have shown resilience to irradiation-induced damage. The ability to predict defect diffusion and accelerate simulations of defect behaviour in HEAs using ML techniques is consequently a subject that has gathered significant interest. The goal of this work was to produce an unsupervised neural network capable of learning the interatomic dynamics within a specific HEA system from MD data in order to create a KMC type predictor of defect diffusion paths for common point defects in crystal systems such as the vacancy and self-interstitial. Self-interstitial defect states were identified and purified from MD datasets using graph-isomorphism, and a proof-of-concept model for the HEA environment was used with several interaction setups to demonstrate the feasibility of training a GCN to predict vacancy defect transition rates in the HEA crystalline environment.

Le fer et les alliages à base de fer présentent un intérêt considérable pour la communauté de la modélisation des matériaux en raison de l’immense importance technologique de l’acier. Les alliages ferritiques à base de fer sont largement utilisés dans les industries aéronautique et nucléaire en raison de leur résistance mécanique élevée, de leur faible dilatation à haute température et de leur résistance à la corrosion. Ces propriétés sont cependant affectées par des défauts ponctuels intrinsèques et extrinsèques. Dans cette thèse, nous décrivons en détail la cinétique des défauts ponctuels dans le fer α en utilisant la technique d’activation-relaxation cinétique (ARTc), une méthode de Monte Carlo cinétique hors réseau avec construction de catalogue à la volée. Plus précisément, nous nous intéressons aux mécanismes de diffusion du carbone (C) et des amas de lacunes dans le fer α. Dans un premier temps, nous étudions l’effet de la pression sur la diffusion du carbone dans le joint de grains de fer α. Nous constatons que l’effet de la pression peut fortement modifier la stabilité et la diffusivité du carbon dans le joint de grains d’une manière qui dépend étroitement de l’environnement local et de la nature de la déformation. Ceci peut avoir un impact majeur sur l’évolution des matériaux hétérogènes, avec des variations de pression locale qui altéreraient fortement la diffusion à travers le matériau. Nous étudions également l’évolution structurale des amas de lacunes contenant de deux à huit lacunes dans le fer α. Nous décrivons en détail le paysage énergétique, la cinétique globale et les mécanismes de diffusion associés à ces défauts. Nos résultats montrent des mécanismes de diffusion complexes même pour des défauts aussi simples que de petits amas de lacunes. Enfin, dans le dernier chapitre, nous discutons une approche de gestion de petites barrières par bassin local dans ARTc. Les simulations de Monte Carlo cinétiques deviennent inefficaces dans les systèmes où le paysage énergétique est constitué de bassins avec de nombreux états reliés par des barrières énergétiques très faibles par rapport à celles nécessaires pour quitter ces bassins. Au fur et à mesure que le système évolue état par état, il est beaucoup plus susceptible d’effectuer des événements répétés (appelés oscillateurs) à l’intérieur du bassin d’énergie de piégeage que de s’échapper du bassin. De tels osccilateurs ne font pas progresser la simulation et ne fournissent que peu d’informations au-delà d’uen première évaluation de ces états. Notre algorithme de bassin local détecte, à la volée, des groupes d’états oscillants et les consolide en bassins locaux, que nous traitons avec la méthode de taux moyen d’auto-construction de bassin (bac-MRM), une approche de type équation maîtresse selon la méthode du taux moyen. / Iron and iron-based alloys are of considerable interest to the materials modelling community because of the immense technological importance of steel. Iron-based ferritic alloys are widely used in aeronautic and nuclear industries due to their high mechanical strength, low expansion at high temperatures, and corrosion resistance. These properties are affected by intrinsic and extrinsic point defects, however. In this thesis, we describe in detail the kinetics of point defects in α−iron using the kinetic activation-relaxation technique (kART), an off-lattice kinetic Monte Carlo method with on-the-fly catalog building. More specifically, we focus on the diffusion mechanisms of carbon and vacancy clusters in α−iron. First, we study the pressure effect on carbon diffusion in the grain boundary (GB) of α−iron. We find that the effect of pressure can strongly modify the C stability and diffusivity in the GB in ways that depend closely on the local environment and the nature of the deformation. This can have a major impact on the evolution of heterogeneous materials, with variations of local pressure that would strongly alter diffusion across the material. We also study the structural evolution of vacancy clusters containing two to eight vacancies in α−iron. We describe in detail the energy landscape, overall kinetics, and diffusion mechanisms associated with these defects. Our results show complex scattering mechanisms even for defects as simple as small vacancy clusters. Finally, in the last chapter, we discuss a local basin approach to managing low barrier events in the kART. Kinetic Monte Carlo simulations become inefficient in systems where the energy landscape consists of basins with numerous states connected by very low energy barriers compared to those needed to leave these basins. As the system evolves state by state, it is much more likely to perform repeated events (so-called flickers) inside the trapping energy basin than to escape the basin. Such flickers do not progress the simulation and provide little insight beyond the first identification of those states. Our local basin algorithm detects, on the fly, groups of flickering states and consolidates them into local basins, which we treat with the basin-auto-constructing Mean Rate Method (bac-MRM), a master equation-like approach based on the mean-rate method.

Gasphasenabscheidungen werden zur Produktion dünner Schichten in der Mikro- und Nanoelektronik benutzt, um eine präzise Kontrolle der Schichtdicke im Sub-Nanometer-Bereich zu erreichen. Elektronische Eigenschaften der Schichten werden dabei von strukturellen Eigenschaften determiniert, deren Bestimmung mit hohem experimentellem Aufwand verbunden ist. Die vorliegende Arbeit erweitert ein hochparalleles Modell zur atomistischen Simulation des Wachstums und der Struktur von Dünnschichten, welches Molekulardynamik (MD) und Kinetic Monte Carlo-Methoden (KMC) kombiniert, um die Beschreibung beliebiger Gasphasenabscheidungen. KMC-Methoden erlauben dabei die effiziente Betrachtung der Größenordnung ganzer Nano-Bauelemente, während MD für atomistische Genauigkeit sorgt. Erste Ergebnisse zeigen, dass das Parsivald genannte Modell Abscheidungen in Simulationsräumen mit einer Breite von 0.1 µm x 0.1 µm effizient berechnet, aber auch bis zu 1 µm x 1 µm große Räume mit 1 Milliarden Atomen beschreiben kann. Somit lassen sich innerhalb weniger Tage Schichtabscheidungen mit einer Dicke von 100 Å simulieren. Die kristallinen und amorphen Schichten zeigen glatte Oberflächen, wobei auch mehrlagige Systeme auf die jeweilige Lagenrauheit untersucht werden. Die Struktur der Schicht wird hauptsächlich durch die verwendeten molekulardynamischen Kraftfelder bestimmt, wie Untersuchungen der physikalischen Gasphasenabscheidung von Gold, Kupfer, Silizium und einem Kupfer-Nickel-Multilagensystem zeigen. Stark strukturierte Substrate führen hingegen zu Artefakten in Form von Nanoporen und Hohlräumen aufgrund der verwendeten KMC-Methode. Zur Simulation von chemischen Gasphasenabscheidungen werden die Precursor-Reaktionen von Silan mit Sauerstoff sowie die Hydroxylierung von alpha-Al2O3 mit Wasser mit reaktiven Kraftfeldern (ReaxFF) berechnet, allerdings ist weitere Arbeit notwendig, um komplette Abscheidungen auf diese Weise zu simulieren. Mit Parsivald wird somit die Erweiterung einer Software präsentiert, die Gasphasenabscheidungen auf großen Substraten effizient simulieren kann, dabei aber auf passende molekulardynamische Kraftfelder angewiesen ist.