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.

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.

Metal-Oxide-Silicon Field-Effect-Transistors with a layer of electrically isolated Si nanocrystals (NCs) embedded in the gate oxide are known to improve conventional floating gate flash memories. Data retention, program and erase speeds as well as the memory operation voltages can be substantially improved due to the discrete charge storage in the isolated Si NCs. Using ion beam synthesis, Si NCs can be fabricated along with standard CMOS processing. The optimization of the location and size of ion beam synthesized Si NCs requires a deeper understanding of the mechanisms involved, which determine (i) the built-up of Si supersaturation by high-fluence ion implantation and (ii) NC formation by phase separation. For that aim, process simulations have been conducted that address both aspects on a fundamental level and, on the other hand, are able to avoid tedious experiments. The built-up of a Si supersaturation by high-fluence ion implantation were studied using dynamic binary collision calculations with TRIDYN and have lead to a prediction of Si excess depth profiles in thin gate oxides of a remarkable quality. These simulations include in a natural manner high fluence implantation effects as target erosion by sputtering, target swelling and ion beam mixing. The second stage of ion beam synthesis is modeled with the help of a tailored kinetic Monte Carlo code that combines a detailed kinetic description of phase separation on atomic level with the required degree of abstraction that is necessary to span the timescales involved. Large ensembles of Si NCs were simulated reaching the late stages of NC formation and dissolution at simulation sizes that allowed a direct comparison with experimental studies, e.g. with electron energy loss resolved TEM investigations. These comparisons reveal a nice degree of agreement, e.g. in terms of predicted and observed precipitate morphologies for different ion fluences. However, they also point clearly onto impact of additional external influences as, e.g., the oxidation of implanted Si by absorbed humidity, which was identified with the help of these process simulations. Moreover, these simulations are utilized as a general tool to identify optimum processing regimes for a tailored Si NC formation for NC memories. It is shown that key properties for NC memories as the tunneling distance from the transistor channel to the Si NCs, the NC morphology, size and density can be adjusted accurately despite of the involved degree of self-organization. Furthermore, possible lateral electron tunneling between neighboring Si NCs is evaluated on the basis of the performed kinetic Monte Carlo simulations.

Metal-Oxide-Silicon Field-Effect-Transistors with a layer of electrically isolated Si nanocrystals (NCs) embedded in the gate oxide are known to improve conventional floating gate flash memories. Data retention, program and erase speeds as well as the memory operation voltages can be substantially improved due to the discrete charge storage in the isolated Si NCs. Using ion beam synthesis, Si NCs can be fabricated along with standard CMOS processing. The optimization of the location and size of ion beam synthesized Si NCs requires a deeper understanding of the mechanisms involved, which determine (i) the built-up of Si supersaturation by high-fluence ion implantation and (ii) NC formation by phase separation. For that aim, process simulations have been conducted that address both aspects on a fundamental level and, on the other hand, are able to avoid tedious experiments. The built-up of a Si supersaturation by high-fluence ion implantation were studied using dynamic binary collision calculations with TRIDYN and have lead to a prediction of Si excess depth profiles in thin gate oxides of a remarkable quality. These simulations include in a natural manner high fluence implantation effects as target erosion by sputtering, target swelling and ion beam mixing. The second stage of ion beam synthesis is modeled with the help of a tailored kinetic Monte Carlo code that combines a detailed kinetic description of phase separation on atomic level with the required degree of abstraction that is necessary to span the timescales involved. Large ensembles of Si NCs were simulated reaching the late stages of NC formation and dissolution at simulation sizes that allowed a direct comparison with experimental studies, e.g. with electron energy loss resolved TEM investigations. These comparisons reveal a nice degree of agreement, e.g. in terms of predicted and observed precipitate morphologies for different ion fluences. However, they also point clearly onto impact of additional external influences as, e.g., the oxidation of implanted Si by absorbed humidity, which was identified with the help of these process simulations. Moreover, these simulations are utilized as a general tool to identify optimum processing regimes for a tailored Si NC formation for NC memories. It is shown that key properties for NC memories as the tunneling distance from the transistor channel to the Si NCs, the NC morphology, size and density can be adjusted accurately despite of the involved degree of self-organization. Furthermore, possible lateral electron tunneling between neighboring Si NCs is evaluated on the basis of the performed kinetic Monte Carlo simulations.