New collective coordinates that describe the deformation of atomic nuclei in terms of solitary waves (rotons) that are solutions of an underlying nonlinear liquid drop dynamics are introduced. The rotons, which are highly localized surface excitations, are shown to be intelligent states of an underlying angular momentum algebra, and furthermore, to adiabatically decouple from the rotational motion of the nucleus. An expansion of the roton profile in terms of surface multipoles shows that although the quadrupole mode (&lambda= 2) dominates, there are significant contributions from other even multipoles with &lambda= 4, 6, 8, etc. The picture of nuclear deformation that emerges is that of a smooth transition from vibrational motion to rotations, signaled by the increasing amplitude of the surface oscillations. The proposed model incorporates two features: Shell effects, with only particles (holes) above (below) the core participating in the dynamics. Collective modes, governed through nonlinear terms in the interaction governing the system. The theory offers a different interpretation of the Bohr-Mottelson picture of collective motion, one that integrates in a natural way the concept of shell closures.

We present results from investigations of mass transfer instability in close binary star systems. By unstable mass transfer we mean the exchange of material where the response of the binary to the initial Roche lobe overflow causes the donor to loose even more material. Our work is guided by approximate arguments that dictate the stability boundaries for binary star systems. To proceed further one must explicitly treat extended mass and velocity distributions that are both nitially, and through their subsequent evolution in time, self-consistent. In this dissertation, we present the first three-dimensional, fully self-consistent treatment of mass transfer in close binary systems. To perform these calculations we have developed and tested a set of tools including a Self-Consistent Field code for generating polytropic binaries executing synchronous rotation upon circular orbits and a parallel, gravitational hydrodynamics code for evolving the binaries in time. We describe, in detail, these tools and their application to the evolution of binary star systems. We present extended simulations of two detached binaries that have been used to examine the accuracy of our computational techniques in addition to the simulations of interacting binaries.

The purpose of this dissertation is to investigate the nanoscale hardness of gallium arsenide thin films and the elastic-plastic behavior of gallium arsenide under an indenter. These investigations were carried out using molecular dynamics (MD) simulations. The simulations are based on interatomic potentials that accurately reproduce many properties of bulk GaAs. The MD simulations performed required scalable and efficient algorithms for implementation on large parallel computers. Nanoindentation simulations were performed using an ideal indenter that was held rigid during the simulation. To reduce the transient effects due to loading, the traversal of the indenter was interrupted periodically to allow the substrate to relax. Load-displacement curves were calculated and Vickers hardness and Youngs modulus were computed from the curves. The damage caused by the indenter was characterized in three ways. The material deposited on the surface was compared to bulk amorphous GaAs and found to be structually similar, indicating that the material underwent solid-state amorphization under the indenter. Analysis of energetic atoms beneath the surface suggested the presence of dislocation loops. A centrosymmetry method was applied to characterize these defects. It was found that the method used did not perform adequately in the presence of amorphized material. Pressure distributions were calculated and atomic configurations were plotted to determine if subsurface microcracking due to the indentation was present. No indication of microcracking or pore formation was found. Adhesion between the tip and substrate was also studied. The effect of the tip-surface attraction was studied for a modified Vickers indenter with a small flat surface instead of an atomically sharp tip. For indentations less than the yield point in GaAs, the bond formation between the tip and the surface led to nonelastic deformation of the surface layer, while the layers undeneath the surface behaved in a purely elastic fashion. Through a series of small indenter traversals, the yield point of GaAs was determined to be 0.6 µN.

Two Varian linear accelerators were used to test the modified sector-integration method. This method can predict the electron insert factor for arbitrary inserts at different source to surface distances (SSD). The effective source distances (SSD_eff) and insert factors (IF) for several accelerators are compared. The relationship of IF and SSD_eff for machine type, energy, cone size, insert size and SSD is presented. The results were fed into the electron output determination module of an existing monitor unit calculator (MUCalc) which uses the modified sector integration method. Prediction and measurements were compared for various clinical inserts at several SSDs.

Advances in computer technologies allow calculations in ever larger model spaces. To keep our understanding growing along with this growth in computational power, we consider a novel approach to the nuclear shell model. The one-dimensional harmonic oscillator in a box is used to introduce the concept of an oblique-basis shell-model theory. By implementing the Lanczos method for diagonalization of large matrices, and the Cholesky algorithm for solving generalized eigenvalue problems, the method is applied to nuclei. The mixed-symmetry basis combines traditional spherical shell-model states with SU(3) collective configurations. We test the validity of this mixed-symmetry scheme on ²⁴Mg and ⁴⁴Ti. Results for ²⁴Mg, obtained using the Wilthental USD intersection in a space that spans less than 10% of the full-space, reproduce the binding energy within 2% as well as an accurate reproduction of the low-energy spectrum and the structure of the states -- 90% overlap with the exact eigenstates. In contrast, for an m-scheme calculation, one needs about 60% of the full space to obtain compatible results. Calculations for ⁴⁴Ti support the mixed-mode scheme although the pure SU(3) calculations with few irreps are not as good as the standard m-scheme calculations. The strong breaking of the SU(3) symmetry results in relatively small enhancements within the combined basis. However, an oblique-basis calculation in 50% of the full pf-shell space is as good as a usual m-scheme calculation in 80% of the space. Results for the lower pf-shell nuclei ^44-48Ti and ⁴⁸Cr, using the Kuo-Brown-3 interaction, show that SU(3) symmetry breaking in this region is driven by the single-particle spin-orbit splitting. In our study we observe some interesting coherent structures, such as coherent mixing of basis states, quasi-perturbative behavior in the toy model, and enhanced B(E2) strengths close to the SU(3) limit even though SU(3) appears to be rather badly broken. The results suggest that a mixed-mode shell-model theory may be useful in situations where competing degrees of freedom dominate the dynamics, and full-space calculations are not feasible.

An in vivo dosimetry system that uses p-type semiconductor diodes with buildup caps was characterized for clinical use. The dose per pulse dependence was investigated. This was done by altering the SSD, field size and wedge for photons, and by altering SSD and cone size for electrons. The off-axis correction and effect of changing repetition rate were also investigated. A model was made to fit the measured diode correction factors.

Systematic constancy and accuracy of a treatment planning system (TPS) are crucial for the entire radiation treatment planning process (TPP). The Quality Assurance (QA) of individual components does not necessarily lead to satisfying performance of the whole process due to the possible errors introduced by the data transfer process between components and other fluctuations. However, most of current QA for TPS is confined to the treatment planning computers. In this study, a time efficient and integrated CT-TPS QA procedure is presented, which starts at the beginning of the TPS input --- Computer Tomograhpy (CT). The whole QA procedure is based on the concept of simulating a real patient treatment. Following the CT scan of a head phantom with geometrical objects, a set of reference treatment plans for each accelerator, with all energy beams included, were established. Whenever TPS QA is necessary, the same procedure is repeated and a QA plan is produced. Through the comparison of QA plan with the reference plan, major systematic errors can be found easily and quickly. This method was also applied to VariSeed and PLATO Brachytherapy treatment planning systems. Moreover, if any error is detected in the system, TPS is broken into several parts and individual tests are also set up.

When delivering the total dose via a sequence of small fields shaped by a multileaf collimator, it is important to consider leaf positional reproducibility. A small error in the leaf position can result in large dose errors to the entire field. This is true for both dynamic multileaf collimation and step and shoot delivery. The goal of this research project is to design a method of quality assurance that is easily reproducible, sensitive to small changes in leaf position, and requires minimal time on the part of the medical physicist to carry out. This paper describes a system of measurements performed with a two-dimensional diode array that can be used in conjunction with a leaf edge function determined from radiographic film to quickly and easily test the reproducibility of the multileaf collimator position with acceptable sensitivity.

The sophistication of radiation therapy delivery techniques at Mary Bird Perkins Cancer Center (MBPCC) creates the need for an advanced dosimetric system that can quantify and verify the dose distributions in three-dimensions. Current dosimetric systems perform this dose analysis in only one or two dimensions. This paper evaluates the application of BANG-3 polymer gel dosimetry to quantify the 3-D dose distribution of Intensity-Modulated Radiation Therapy (IMRT) using a "step and shoot" approach. The gel was irradiated by 10 MV photons at a dose rate of 400 MU/min. Relaxation rate maps were computed from proton density and T₂-weighted magnetic resonance images acquired with a GE Horizon 1.5T scanner; scans were performed 5 days and 2 months post-irradiation. The dose distribution within the gel was compared to the dose distribution calculated by the Pinnacle³ planning system. Three techniques were used for analysis: image subtraction, dose-volume analysis and contour analysis. Also, a dose correction factor was used to attempt to correct for excess dose delivered to the gel as the gels were erroneously placed in the treatment room two days prior to irradiation. Corrected 5-day post-irradiation dose maps show reasonable agreement with the Pinnacle³ plan. The absolute measurement error was +/-50 cGy; however, the relative errors were large compared to the total dose of 2 Gy delivered to the gel. Delivering a larger total dose should reduce the relative error to a reasonable magnitude. Exposure to light and other environmental factors caused substantial additional polymerization with time. The results of this project indicate that polymer gel dosimetry could be a useful routine 3D dosimetric technique at MBPCC. However, utilizing a commercial scanning service may simplify use of the gels.

Oxidation of a flat aluminum (111) surface and the reactive wetting of the aluminum (Al) droplet on a flat alumina (α-Al₂O₃) surface are investigated by using parallel molecular-dynamics simulations with dynamic charge transfer among atoms on a microscopic length scale. The interatomic potential, based on the formalism of Streitz and Mintmire, allows atoms to vary their charges dynamically between anions and cations, when atoms move and their local environment is altered. We investigate the oxidation thickness as a function of time and the oxygen density which is 10-40 times that of the normal state (1 atm and 300 K). Stable amorphous oxide scales form around 51 Å at 4.42 ns, 2.862 ns, and 2.524 ns, respectively, and molecular oxygen density 10-40 times the normal state. We also study structural correlations in the resulting final oxide scale. The structure of final oxide scales depend on depth, where density of aluminum (Al) and oxygen (O) atoms change. Reactive wetting of aluminum nanodroplet on alumina surface is also studied using parallel MD. We study heat transfer, diffusion within droplet, and the structure of the inter-metallic phases in the liquid-solid interface. Oxygen (O) atoms diffuse into the spherical aluminum (Al) droplet and form an interface between the flat solid substrate and the Al droplet. This diffusion of oxygen atoms may be the main source of adhesion between the Al drop and the flat α-Al₂O₃ substrate. The temperature in the flat α-Al₂O₃ bulk substrate rises from 0K to 200 K at the end of the simulation, 8.5 ps, but the temperature becomes much higher at the reactive interface. We have examined which oxygen atoms from the substrate participate in the wetting and the formation of a solder joint at the Al/α-Al₂O₃ interface.