<p>This thesis contributes significantly to the advanced applications of scattering parameter sensitivity analysis including the design optimization of high-frequency printed structures and in microwave imaging. In both applications, the methods exploit the computational efficiency of the self-adjoint sensitivity analysis (SASA) approach where only one EM simulation suffices to obtain both the responses and their gradients with respect to the optimizable variables.</p> <p>An<em> S</em>-parameter self-adjoint sensitivity formula for multiport planar structures using the method of moments (MoM) current solution is proposed. It can be easily implemented with existing MoM solvers. The shape perturbation which is required in computing the system-matrix derivatives are accommodated by changing the material properties of the local mesh elements. The use of a pre-determined library system matrix further accelerates the design optimization because the writing/reading of the system matrix to/from the disk is avoided. The design optimization of a planar ultra-wide band (UWB) antenna and a double stub tuner are presented as validation examples.</p> <p>In the application of the sensitivity-based imaging, the SASA approach allows for real-time image reconstruction once the field distribution of the reference object (RO) is known. Here, the RO includes the known background medium of the object under test (OUT) and the known antennas. The field distribution can be obtained using simulation or measurement.</p> <p>The spatial resolution is an important measure of the performance of an imaging technique. It represents the smallest detail that can be detected by a given imaging method. The resolution of the sensitivity-based imaging approach has not been studied before. In this thesis, the resolution limits are systematically studied with planar raster scanning and circular array data acquisition. In addition, the method’s robustness to noise is studied. A guideline is presented for an acceptable signal-to-noise ratio (SNR) versus the spatial and frequency sampling rates in designing a data-acquisition system for the method.</p> <p>This thesis validates the sensitivity-based imaging with measured data of human tissue phantoms for the first time. The differences in dielectric properties of the targets are qualitatively reflected in the reconstructed image. A preliminary study of imaging with inexact background information of the OUT is also presented.</p> / Doctor of Philosophy (PhD)

The frequency demands of radiating systems are moving into the terahertz band with potential applications that include sensing, imaging, and extremely broadband communication. One commonly used method for generating and detecting terahertz waves is to excite a voltage-biased photoconductive antenna with an extremely short laser pulse. The pulsed laser generates charge carriers in a photoconductive substrate which are swept onto the metallic antenna traces to produce an electric current that radiates or detects a terahertz band signal. Therefore, analysis of a photoconductive antenna requires simultaneous solutions of both semiconductor physics equations (including drift-diffusion and continuity relations) and Maxwell’s equations. A multi-physics analysis scheme based on the Discontinuous-Galerkin Finite-Element Time-Domain (DGFETD) is presented that couples the semiconductor drift-diffusion equations with the electromagnetic Maxwell’s equations. A simple port model is discussed that efficiently couples the two equation sets. Various photoconductive antennas were fabricated using TiAu metallization on a GaAs substrate and the fabrication process is detailed. Computed emission intensities are compared with measured data. Optimized antenna designs based on the analysis are presented for a variety of antenna configurations.

This dissertation is about efficient computation of the electromagnetic fields with the locally corrected Nyström (LCN) method as a point-based boundary element method (BEM). The concept of surface integral equations is discussed and the electric field integral equation (EFIE) is derived from the Maxwell’s equations. Due to its point-based nature, the LCN discretization of the EFIE has some advantages over discretizing the EFIE by the method-of-moments (MoM) which is an element-based BEM. On the other hand, due to maturity of the MoM, a large body of work is available to resolve the numerical issues arising in MoM while there has been less work related to the relatively new LCN. To combine the benefits of the LCN method and the classical Rao-Wilton-Glisson MoM, equivalence between these BEMs are established and their exact relationships are derived. Both the vector-potential EFIE and the mixed-potential EFIE are covered. Various aspects of achieving HO convergence to the correct answer using high-order (HO) LCN method are discussed. In particular, the patch size limitation, predicting the optimal degrees of freedom, and the effect of dynamic range in the solution are discussed both analytically and numerically to provide concrete motivations towards HO LCN. The benefits of an HO BEM can not be realized unless an HO geometry representation is used in conjunction with the BEM. Non-uniform rational b-spline (NURBS) surfaces are the most widely adopted HO geometry modelling technique in various disciplines due to their many advantages. However, a typical mesh created out of NURBS surfaces contain both triangular and quadrilateral elements while formulating LCN based on Gaussian quadrature rules on triangular elements have limitations. As a result, the LCN community has mostly adopted LCN based on curvilinear quadrilateral modelling of the geometry. A new class of Newton-Cotes quadrature rules for triangles is proposed to facilitate incorporating NURBS surfaces into the HO LCN. / May 2016

Modeling of the electromagnetic, structural, thermal, or acoustic response of the human body to various external and internal stimuli is limited by the availability of anatomically accurate and numerically efficient computational models. The models currently approved for use are generally of proprietary or fixed format, preventing new model construction or customization. 1. This dissertation develops a new Visible Human Project - Female (VHP-F) computational phantom, constructed via segmentation of anatomical cryosection images taken in the axial plane of the human body. Its unique property is superior resolution on human head. In its current form, the VHP-F model contains 33 separate objects describing a variety of human tissues within the head and torso. Each obejct is a non-intersecting 2-manifold model composed of contiguous surface triangular elements making the VHP-F model compatible with major commercial and academic numerical simulators employing the Finite Element Method (FEM), Boundary Element Method (BEM), Finite Volume Method (FVM), and Finite-Difference Time-Domain (FDTD) Method. 2. This dissertation develops a new workflow used to construct the VHP-F model that may be utilized to build accessible custom models from any medical image data source. The workflow is customizable and flexible, enabling the creation of standard and parametrically varying models facilitating research on impacts associated with fluctuation of body characteristics (for example, skin thickness) and dynamic processes such as fluid pulsation. 3. This dissertation identifies, enables, and quantifies three new specific computational bioelectromagnetic problems, each of which is solved with the help of the developed VHP-F model: I. Transcranial Direct Current Stimulation (tDCS) of human brain motor cortex with extracephalic versus cephalic electrodes; II. RF channel characterization within cerebral cortex with novel small on-body directional antennas; III. Body Area Network (BAN) characterization and RF localization within the human body using the FDTD method and small antenna models with coincident phase centers. Each of those problems has been (or will be) the subject of a separate dedicated MS thesis.

In traditional Permanent Magnet Machines, such as electric motors and generators, power is transmitted by magnetic flux passing through an air gap, which has a very low magnetic permeability, limiting performance. However, reducing the air gap through traditional means carries risks in manufacturing, with tight tolerances and associated costs, and reliability, with thermal and dynamic effects requiring adequate clearance. Using a magnetically permeable, high dielectric strength material has the potential to improve magnetic performance, while at the same time offering performance advantages in heat transfer. Ferrofluids were studied as a method for improved permeability in the rotor / stator gap with a combined experimental and computational approach. Results show promise for the ferrofluid technique. An off-the-shelf motor system showed improved performance with ferrofluids vs. fluids of equivalent viscosity, and improved performance vs. an air gap at low RPM. New generator designs showed design dependent performance gains, although some potential for negative performance effects. A proof of concept generator was built and tested, with increased voltage vs. RPM predicted through virtual prototyping, and validated through experimentation, showing ~10% improvement on voltage vs. RPM at the <600 RPM range. More repeatable engineering tests demonstrated a ~30% increase in the voltage / RPM relationship for designs with an isolated stator chamber and a large stator - rotor gap. However, the effects were negative for a similar system with a small stator-rotor gap due to leakage flux effects. New contributions to the body of knowledge in this area include: • Application of the ferrofluid technique to axial flux designs. • Development of a virtual prototype, including variations in the fluid viscosity due to ferrohydrodynamic effects. • Consideration of negative effects of ferrofluid immersion, such as shear losses and increases in leakage flux. • Optimization of the design to eliminate increased viscous losses. The improved design has been designed, built, and tested, featuring isolation of the ferrofluid from the rotating region. This offers all of the performance gain of improved magnetic permeability, while minimizing the offsetting losses from increased shear effects.

Small antennas have attracted significant attention due to their prolific use in consumer electronics. Such antennas are highly desirable in the healthcare industry for imaging and implants. However, most small antennas are not highly directive and are detuned when in the presence of a dielectric. The human body can be compared to a series of lossy dielectric media. A novel antenna design, the orthogonal coil, is proposed to counter both of these shortcomings. As loop antennas radiate primarily in the magnetic field, their far field pattern is less influenced by nearby lossy dielectrics. By exciting two orthogonal coil antennas in quadrature, their beams in the H-plane constructively add in one direction and cancel in the other. The result is a small, yet directive antenna, when placed near a dielectric interface. In addition to present a review of the current literature relating to small antennas and dipoles near lossy interfaces, the far field of the orthogonal coil antenna is derived. The directivity is then plotted for various conditions to observe the effect of changing dielectric constants, separation from the interface, etc. Numeric simulations were performed using both Finite Difference Time Domain (FDTD) in MATLAB and Finite Element Method (FEM) in Ansys HFSS using a anatomically accurate high-fidelity head mesh that was generated from the Visible Human Project® data. The following problem has been addressed: find the best radio-frequency path through the brain for a given receiver position - on the top of the sinus cavity. Two parameters: transmitter position and radiating frequency should be optimized simultaneously such that (i) the propagation path through the brain is the longest; and (ii) the received power is maximized. To solve this problem, we have performed a systematic and comprehensive study of the electromagnetic fields excited in the head by the aforementioned orthogonal dipoles. Similar analyses were performed using pulses to detect Alzheimer’s disease, and on the femur to detect osteoporosis.

Examination of quantum critical points of itinerant electron systems will aid with understanding of d-electron magnetism that exhibits both local and itinerant characteristics in different families of compounds. Doping-induced quantum phase transition of the itinerant ferromagnet Sc 3.1 In that is composed of non-magnetic elements is the focus of our work. Polycrystalline samples of (Sc 1-x Lu x ) 3.1 In with 0≤ x≤ 0.08 were prepared by arcmelting and then annealing for an extended period of time. Susceptibility measurements were performed in an applied magnetic field H = 0.1 T for temperatures T = 1.85 K to 300 K. Linearity of Arrott plots in low-field region was significantly improved by implementing the non-mean-field Arrott-Noakes technique where plotting M 1/β vs. ( H/M ) 1/γ is used to determine both the Curie temperature and composition. Modified Arrott plot approach was used in order to determine the new critical exponents β, γ and δ that better describe this compound. The Curie temperature of the Sc 3.1 In compound was found to be T C = 4.4 K and the critical composition x c = 0.02. This work was supported by NSF DMR 0847681.

Modeling of radio wave propagation is indispensable for the design and analysis of wireless communication systems. The use of the Finite-Difference Time-Domain (FDTD) method for wireless channel modeling has gained significant popularity due its ability to extract wideband responses from a single simulation. FDTD-based techniques, despite providing accurate channel characterizations, have often employed point sources in their studies, mainly due to the large amounts of resources required for modeling fine geometrical details or features inherent in antennas into a discrete spatial domain. The underlying influences of the antenna on wave propagation have thus been disregarded. This work presents a possible approach for the efficient space-time analysis of antennas by deducing FDTD-compatible macromodels that completely encapsulate the electromagnetic behaviour of antennas and then incorporating them into a standard FDTD formulation for modeling their interactions with a general environment.

Modeling of radio wave propagation is indispensable for the design and analysis of wireless communication systems. The use of the Finite-Difference Time-Domain (FDTD) method for wireless channel modeling has gained significant popularity due its ability to extract wideband responses from a single simulation. FDTD-based techniques, despite providing accurate channel characterizations, have often employed point sources in their studies, mainly due to the large amounts of resources required for modeling fine geometrical details or features inherent in antennas into a discrete spatial domain. The underlying influences of the antenna on wave propagation have thus been disregarded. This work presents a possible approach for the efficient space-time analysis of antennas by deducing FDTD-compatible macromodels that completely encapsulate the electromagnetic behaviour of antennas and then incorporating them into a standard FDTD formulation for modeling their interactions with a general environment.

The discontinuous Galerkin method (DGM) is applied to a number of problems in computational electromagnetics. This is achieved by obtaining numerical solutions to Maxwell's equations using the DGM. The aim of these simulations is to highlight the strengths of the method while showing its resilience in handling problems other schemes may not be able to accurately model. Although no method will ever be the best choice for every problem in electromagnetics, the discontinuous Galerkin method is able to accurately approximate any problem, although the computational costs can make the scheme impractical for some. Like other time domain schemes, the DGM becomes inefficient on large domains where the solution contains small wavelengths. We demonstrate that all of the different types of boundary conditions in electromagnetic wave propagation can be implemented into the DGM. Reflection and transmission boundaries fit easily into the framework, whereas perfect absorption requires a more advanced technique known as the perfectly matched layer. We begin by simulating mirrors with several different geometries, and analyze how the DGM method performs, and how it offers a more complete evaluation of the behavior in this problem than some other methods. Since Maxwell's equations describe the macroscopic features of electromagnetics, our simulations are able to capture the wave features of electromagnetics, such as interference and diffraction. We demonstrate this by accurately modelling Young's double slit experiment, a classic experiment which features well understood interference and diffraction phenomena. We also extend the basic electromagnetic wave propagation simulations to include situations where the waves travel into new media. The formulation of the DGM for Maxwell's equations allows the numerical solutions to accurately resolve the features at the interface of two media as predicted by the Fresnel coefficients. This allows the DGM to model lenses and other sources of refraction. We predict that the DGM will become an increasingly valuable method for computational electromagnetics because of its wide range of applicability as well as the lack of undesirable features in the numerical solutions. Furthermore, the only limiting factor for applying DGM, its computational cost, will become less influential as computing power continues to increase, allowing us to apply the DGM to an increasing set of applications.