Many portions of energy generated in the U.S. are not used and take the form of wasted heat due to a poor heat transfer efficiency. This fact leads research communities to focus on thermoelectrics as a means for using waste heat through direct thermal to electrical energy conversion. One way to enhance thermoelectric efficiency is to reduce thermal conductivity through nanostructuring. In nanostructures, understanding energy transport across the interface of two materials is important because interfaces dominate the resistance to overall thermal transport of the system and can be described by thermal boundary conductance (TBC). Also of note, an understanding of thermal transport cannot be achieved without an understanding of transfer via atomic vibration, known as phonons. In this study, two different techniques of molecular dynamics (MD) simulation are introduced in order to improve the understanding of the phonon transport at the interface of dissimilar materials and the impact of different material properties on TBC. Non-equilibrium MD simulations are used to study relative and combined contributions of mass and bond energy difference on TBC and phonon wave-packet simulations are used to obtain a detailed description of phonon interactions at the interface. At the end of this study, a simple analytical model for the prediction of effective thermal conductivity, using knowledge of thermal boundary resistance, an inverse of TBC, and the interface geometry, is developed.

The thesis deals with a concept and creation of classifiers of defects found on continuous production lines. The first part presents an overview of methods used for image classification and a analysis of defects. The main part of the thesis consist of a description of created classifier interface and graphical user interface for classifier. The last part sums up reliability of each implemented classifer.

A precise understanding of the fundamental correlation between synthesis, microstructure and physical properties is of vital importance towards rational design of improved functional epitaxial thin films. With the presence of heterogeneous interface and associated inhomogeneous lattice strain, film microstructure becomes sensitive to subtle interfacial perturbations and hence may exhibit intriguing physical properties. Control of the epitaxial film functionality requires accurate knowledge of the actual film chemistry, interfacial defects and associated strain field. This dissertation reports in-depth microstructural characterization of the intrinsic chemical inhomogeneity in selected epitaxial thin films including superconducting Fe1+yTe1-xSex/SrTiO3(STO) heterogeneous systems, the flux-pinning defects at both of conversional YBa2Cu3O7-δ (YBCO)/substrate lateral interfaces and vertical interfaces of YBCO&BaSnO3(BSO) nanocomposite films, and the misfit dislocation core configurations of STO/MgO and MgO/STO heterostructures pair, using the state-of-the-art aberration-corrected scanning transmission electron microscopy (CS-corrected STEM) in combination with geometric phase analysis (GPA). For the first time, the local atomic arrangement of Te and Se as well as interstitial Fe(2) has been clearly revealed in superconducting Fe1+yTe1-xSex/STO epitaxial films. We found that the film growth atmosphere can greatly affect the film stoichiometry, the homogeneity of Se/Te ordering and thus the overall film superconductivity. YBCO/substrate interface mismatch and YBCO&BSO vertical interface contact have been explored through substrate selection and doping-concentration variation. We observed a diverse nature of intrinsic defects in different YBCO/substrate heterosystems; thermal stable defects capable of maintaining individual strain field have been found effective in flux-pinning. Along the vertical heterointerface of YBCO/BSO, misfit dislocations were found throughout the film thickness. It adds another dimension to the flux-pinning landscape design. Four basic misfit dislocation core configurations of a STO/MgO heterosystem have been identified, and found strongly dependent on the actual interface disordering such as substrate atomic-height steps and interdiffussion. To precisely quantify the heterointerface lattice strain, we first conducted systematic investigations on the accuracy of STEM-based GPA. Follow our protocol, 1 pm accuracy has been proven in the STEM fast-scan direction with a spatial resolution less than 1 nm. The effectiveness and reliability of this optimized GPA strain profile were demonstrated in both applications of a relaxed STO/MgO and a partially strained LaAlO3/STO heterointerfaces, respectively.

Al–O monolayers embedded in ultrathin SiO₂ were shown previously to contain Al-induced acceptor states, which capture electrons from adjacent silicon wafers and generate a negative fixed charge that enables efficient Si-surface passivation. Here, we show that this surface passivation is just in part attributed to field-effect passivation, since the electrically active interface trap density Dit itself at the Si/SiO₂ interface is reduced by the presence of the acceptor states. For sufficiently thin tunnel-SiO₂ films between the Si-surface and the Al–O monolayers, Dit is reduced by more than one order of magnitude. This is attributed to an interface defect deactivation mechanism that involves the discharge of the singly-occupied dangling bonds (Pb0 defects) into the acceptor states, so that Shockley-Read-Hall-recombination is drastically reduced. We demonstrate that the combined electronic and field-effect passivation allows for minority carrier lifetimes in excess of 1 ms on n-type Si and that additional H₂-passivation is not able to improve that lifetime significantly.

No description available.