Refine
Has Fulltext
- yes (5)
Is part of the Bibliography
- yes (5)
Document Type
- Doctoral Thesis (5)
Language
- English (5) (remove)
Keywords
- Simulation (5) (remove)
Institute
- Institut für Theoretische Physik und Astrophysik (5) (remove)
This thesis deals with the chaotic dynamics of nonlinear networks consisting of semiconductor lasers which have time-delayed self-feedbacks or mutual couplings. These semiconductor lasers are simulated numerically by the Lang-Kobayashi equations. The central issue is how the chaoticity of the lasers, measured by the maximal Lyapunov exponent, changes when the delay time is changed. It is analysed how this change of chaoticity with increasing delay time depends on the reflectivity of the mirror for the self-feedback or the strength of the mutal coupling, respectively. The consequences of the different types of chaos for the effect of chaos synchronization of mutually coupled semiconductor lasers are deduced and discussed. At the beginning of this thesis, the master stability formalism for the stability analysis of nonlinear networks with delay is explained. After the description of the Lang-Kobayashi equations and their linearizations as a model for the numerical simulation of semiconductor lasers with time-delayed couplings, the artificial sub-Lyapunov exponent $\lambda_{0}$ is introduced. It is explained how the sign of the sub-Lyapunov exponent can be determined by experiments. The notions of "strong chaos" and "weak chaos" are introduced and distinguished by their different scaling properties of the maximal Lyapunov exponent with the delay time. The sign of the sub-Lyapunov exponent $\lambda_{0}$ is shown to determine the occurence of strong or weak chaos. The transition sequence "weak to strong chaos and back to weak chaos" upon monotonically increasing the coupling strength $\sigma$ of a single laser's self-feedback is shown for numerical calculations of the Lang-Kobayashi equations. At the transition between strong and weak chaos, the sub-Lyapunov exponent vanishes, $\lambda_{0}=0$, resulting in a special scaling behaviour of the maximal Lyapunov exponent with the delay time. Transitions between strong and weak chaos by changing $\sigma$ can also be found for the Rössler and Lorenz dynamics. The connection between the sub-Lyapunov exponent and the time-dependent eigenvalues of the Jacobian for the internal laser dynamics is analysed. Counterintuitively, the difference between strong and weak chaos is not directly visible from the trajectory although the difference of the trajectories induces the transitions between the two types of chaos. In addition, it is shown that a linear measure like the auto-correlation function cannot unambiguously reveal the difference between strong and weak chaos either. Although the auto-correlations after one delay time are significantly higher for weak chaos than for strong chaos, it is not possible to detect a qualitative difference. If two time-scale separated self-feedbacks are present, the shorter feedback has to be taken into account for the definition of a new sub-Lyapunov exponent $\lambda_{0,s}$, which in this case determines the occurence of strong or weak chaos. If the two self-feedbacks have comparable delay times, the sub-Lyapunov exponent $\lambda_{0}$ remains the criterion for strong or weak chaos. It is shown that the sub-Lyapunov exponent scales with the square root of the effective pump current $\sqrt{p-1}$, both in its magnitude and in the position of the critical coupling strengths. For networks with several distinct sub-Lyapunov exponents, it is shown that the maximal sub-Lyapunov exponent of the network determines whether the network's maximal Lyapunov exponent scales strongly or weakly with increasing delay time. As a consequence, complete synchronization of a network is excluded for arbitrary networks which contain at least one strongly chaotic laser. Furthermore, it is demonstrated that the sub-Lyapunov exponent of a driven laser depends on the number of the incoherently superimposed inputs from unsynchronized input lasers. For networks of delay-coupled lasers operating in weak chaos, the condition $|\gamma_{2}|<\mathrm{e}^{-\lambda_{\mathrm{m}}\,\tau}$ for stable chaos synchronization is deduced using the master stability formalism. Hence, synchronization of any network depends only on the properties of a single laser with self-feedback and the eigenvalue gap of the coupling matrix. The characteristics of the master stability function for the Lang-Kobayashi dynamics is described, and consequently, the master stability function is refined to allow for precise practical prediction of synchronization. The prediction of synchronization with the master stability function is demonstrated for bidirectional and unidirectional networks. Furthermore, the master stability function is extended for two distinct delay times. Finally, symmetries and resonances for certain values of the ratio of the delay times are shown for the master stability function of the Lang-Kobyashi equations.
In this PhD thesis, the effect of strain on heteroepitaxial growth is investigated by means of Kinetic Monte Carlo simulations. In this context the lattice misfit, arising from the different lattice constants of the adsorbate and the substrate material, is of particular interest. As a consequence, this lattice misfit leads to long-range elastic strain effects having strong influence on the entire growing crystal and its resulting surface morphology. The main focus of this work is the investigation of different strain relaxation mechanisms and their controlling parameters, revealing interesting consequences on the subsequent growth. Since epitaxial growth is carried out under conditions far away from thermodynamic equilibrium, it is strongly determined by surface kinetics. At this point the relevant kinetic microscopic processes are described, followed by theoretical considerations of heteroepitaxial growth disclosing an overview over several independent methodological streams, used to model epitaxy in different time and length scales, as well as the characterization of misfit dislocations and the classification of epitaxial growth modes based on thermodynamic considerations. The epitaxial growth is performed by means of Kinetic Monte Carlo simulations which allows for the consideration of long range effects in systems with lateral extension of few hundred atoms. By using an off-lattice simulation model the particles are able to leave their predefined lattice sites, which is an indispensable condition for simulating strain relaxation mechanisms. The main idea of our used model is calculating the activation energy of all relevant thermally activated processes by using simple pair potentials and then realizing the dynamics by performing each event according to its probability by means of a rejection-free algorithm method. In addition, the crystal relaxation procedure, the grid-based particle access method, which accelerates the simulation enormously, and the efficient implementation of the algorithm are discussed. To study the influence of long range elastic strain effects, the main part of this work was realized on the two dimensional triangular lattice, which can be treated as a cross section of the real three dimensional case. Chapter 4 deals with the formation of misfit dislocations as a strain relaxation mechanism and the resulting consequences on the subsequent heteroepitaxial growth. We can distinguish between two principally different dislocation formation mechanisms, depending strongly on the sign as well as on the magnitude of the misfit, but also the surface kinetics need to be taken into account. Additionally, the dislocations affect the lattice spacings of the crystal whose observed progression is in qualitative good agreement with experimental results. Furthermore, the dislocations influence the subsequent growth of the adsorbate film, since the potential energy of an adatom is modulated by buried dislocations. A clear correlation between the lateral positions of buried dislocations and the positions of mounds grown on the surface can be observed. In chapter 5, an alternative strain relaxation mechanism is studied: the formation of three dimensional islands enables the particles to approach their preferred lattice spacing. We demonstrate that it is possible to adjust within our simulation model each of the three epitaxial growth modes: Volmer–Weber, Frank–van der Merve or layer-by-layer, and Stranski–Krastanov growth mode. Moreover, we can show that the emerging growth mode depends in principle on two parameters: on the one hand the interaction strength of adsorbate particles with each other, compared to the interaction of adsorbate with substrate particles, and on the other hand the lattice misfit between adsorbate and substrate particles. A sensible choice of these two parameters allows the realization of each growth mode within the simulations. In conclusion, the formation of nanostructures controlled by an underlying dislocation network can be applied in the concept of self-organized pattern formation as well as by the tendency to form ordered arrays of strain-induced three dimensional grown islands. In chapter 6, we extend our model to three dimensions and investigate the effect of strain on growth on bcc(100) surfaces. We introduce an anisotropic potential yielding a stable bcc lattice structure within the off-lattice representation. We can show that the strain built up in submonolayer islands is mainly released at the island edges and the lattice misfit has strong influence on the diffusion process on the plane surface as well as on the situation at island edges with eminent consequences on the appearance of submonolayer islands.
In this PhD thesis, we study the heteroepitaxial crystal growth by means of Monte Carlo simulations. Of particular interest in this work is the influence of the lattice mismatch of the adsorbates relative to the substrate on surface structures. In the framework of an off-lattice model, we consider one monolayer of adsorbate and investigate the emerging nanopatterns in equilibrium and their formation during growth. In chapter 1, a brief introduction is given, which describes the role of computer simulations in the field of the physics of condensed matter. Chapter 2 is devoted to some technical basics of experimental methods of molecular beam epitaxy and the theoretical description. Before a model for the simulation can be designed, it is necessary to make some considerations of the single processes which occur during epitaxial growth. For that purpose we look at an experimental setup and extract the main microscopic processes. Afterwards a brief overview of different theoretical concepts describing that physical procedures is given. In chapter 3, the model used in the simulations is presented. The aim is to investigate the growth of an fcc crystal in the [111] direction. In order to keep the simulation times within a feasible limit a simple pair potential, the Lennard-Jones potential, with continuous particle positions is used, which are necessary to describe effects resulting from the atomic mismatch in the crystal. Furthermore the detailed algorithm is introduced which is based on the idea to calculate the barrier of each diffusion event and to use the barriers in a rejection-free method. Chapter 4 is attended to the simulation of equilibrium. The influence of different parameters on the emerging structures in the first monolayer upon the surface, which is completely covered with two adsorbate materials, is studied. Especially the competition between binding energy and strain leads to very interesting pattern formations like islands or stripes. In chapter 5 the results of growth simulations are presented. At first, we introduce a model in order to realize off-lattice Kinetic Monte Carlo simulations. Since the costs in simulation time are enormous, some simplifications in the calculation of diffusion barriers are necessary and therefore the previous model is supplemented with some elements from the so-called ball and spring model. The next point is devoted to the calculation of energy barriers followed by the presentation of the growth simulations. Binary systems with only one sort of adsorbate are investigated as well as ternary systems with two different adsorbates. Finally, a comparison to the equilibrium simulations is drawn. Chapter 6 contains some concluding remarks and gives an outlook to possible further investigations.
In this PhD thesis, we develop models for the numerical simulation of epitaxial crystal growth, as realized, e.g., in molecular beam epitaxy (MBE). The basic idea is to use a discrete lattice gas representation of the crystal structure, and to apply kinetic Monte Carlo (KMC) simulations for the description of the growth dynamics. The main advantage of the KMC approach is the possibility to account for atomistic details and at the same time cover MBE relevant time scales in the simulation. In chapter 1, we describe the principles of MBE, pointing out relevant physical processes and the influence of experimental control parameters. We discuss various methods used in the theoretical description of epitaxial growth. Subsequently, the underlying concepts of the KMC method and the lattice gas approach are presented. Important aspects concerning the design of a lattice gas model are considered, e.g. the solid-on-solid approximation or the choice of an appropriate lattice topology. A key element of any KMC simulation is the selection of allowed events and the evaluation of Arrhenius rates for thermally activated processes. We discuss simplifying schemes that are used to approximate the corresponding energy barriers if detailed knowledge about the barriers is not available. Finally, the efficient implementation of the MC kinetics using a rejection-free algorithm is described. In chapter 2, we present a solid-on-solid lattice gas model which aims at the description of II-VI(001) semiconductor surfaces like CdTe(001). The model accounts for the zincblende structure and the relevant surface reconstructions of Cd- and Te-terminated surfaces. Particles at the surface interact via anisotropic nearest and next nearest neighbor interactions, whereas interactions in the bulk are isotropic. The anisotropic surface interactions reflect known properties of CdTe(001) like the small energy difference between the c(2x2) and (2x1) vacancy structures of Cd-terminated surfaces. A key element of the model is the presence of additional Te atoms in a weakly bound Te* state, which is motivated by experimental observations of Te coverages exceeding one monolayer at low temperatures and high Te fluxes. The true mechanism of binding excess Te to the surface is still unclear. Here, we use a mean-field approach assuming a Te* reservoir with limited occupation. In chapter 3, we perform KMC simulations of atomic layer epitaxy (ALE) of CdTe(001). We study the self-regulation of the ALE growth rate and demonstrate how the interplay of the Te* reservoir occupation with the surface kinetics results in two different regimes: at high temperatures the growth rate is limited to one half layer of CdTe per ALE cycle, whereas at low enough temperatures each cycle adds a complete layer. The temperature where the transition between the two regimes occurs depends mainly on the particle fluxes. The temperature dependence of the growth rate and the flux dependence of the transition temperature are in good qualitative agreement with experimental results. Comparing the macroscopic activation energy for Te* desorption in our model with experimental values we find semiquantitative agreement. In chapter 4, we study the formation of nanostructures with alternating stripes during submonolayer heteroepitaxy of two different adsorbate species on a given substrate. We evaluate the influence of two mechanisms: kinetic segregation due to chemically induced diffusion barriers, and strain relaxation by alternating arrangement of the adsorbate species. KMC simulations of a simple cubic lattice gas with weak inter-species binding energy show that kinetic effects are sufficient to account for stripe formation during growth. The dependence of the stripe width on control parameters is investigated. We find an Arrhenius temperature dependence, in agreement with experimental investigations of phase separation in binary or ternary material systems. Canonical MC simulations show that the observed stripes are not stable under equilibrium conditions: the adsorbate species separate into very large domains. Off-lattice simulations which account for the lattice misfit of the involved particle species show that, under equilibrium conditions, the competition between binding and strain energy results in regular stripe patterns with a well-defined width depending on both misfit and binding energies. In KMC simulations, the stripe-formation and the experimentally reported ramification of adsorbate islands are reproduced. To clarify the origin of the island ramification, we investigate an enhanced lattice gas model whose parameters are fitted to match characteristic off-lattice diffusion barriers. The simulation results show that a satisfactory explanation of experimental observations within the lattice gas framework requires a detailed incorporation of long-range elastic interactions. In the appendix we discuss supplementary topics related to the lattice gas simulations in chapter 4.
This thesis contains two major parts: The first part introduces the reader into three independent concepts of treating strongly correlated many body physics. These are, on the analytical side the SO(5)-theory (Chap.3), which poses the general frame. On the numerical side these are the Stochastic Series Expansion (SSE) (Chap.1) and the Contractor Renormalization Group (CORE) approach (Chap. 2}). The central idea of this thesis was to combine these above concepts, in order to achieve a better understanding of the high-T_c superconductors (HTSC). The results obtained by this combination can be found in the second major part of this thesis (chapters 4 and 5). The main idea of this thesis, i.e., to combine the SO(5)-theory with the capabilities of bosonic Quantum-Monte Carlo simulations and those of the CORE approach, has been proven to be a very successful Ansatz. Two different approaches, one based on symmetry and one on renormalization-group arguments, motivate an effective bosonic Hamiltonian. In a subsequent step the effective Hamiltonian has been simulated efficiently using the SSE. The results reproduce salient experiments on high-T_c superconductors. In addition, it has been shown that the model can be extended to capture also charge ordering. These results also form a profound basis for further studies, for example one could address the open question of SO(5)-symmetry restoration at a multicritical point in the extended pSO(5) model, where longer ranged interactions are included.