Doctor of Philosophy / Department of Chemical Engineering / James H. Edgar / Gallium nitride is a promising semiconductor for fabricating field effect transistors for power electronics because of its unique physical properties of wide energy band gap, high electron saturation velocity, high breakdown field and high thermal conductivity. However, these devices are extremely sensitive to the gate leakage current which reduces the breakdown voltage and the power-added efficiency and increases the noise figures. To solve this problem, employing a gate dielectric is crucial to the fabrication of metal insulator semiconductor high electron mobility transistors (MISHEMTs), to reduce the leakage current and increase the magnitude of voltage swings possible. For this device to be successful, imperfections at the oxide-semiconductor interface must be suppressed to maintain the high electron mobility of the device. This research explored multiple high dielectric constant gate oxides (Al[subscript]2O[subscript]3, TiO[subscript]2, and Ga[subscript]2O[subscript]3), deposited on different crystalline orientations and polarities of GaN by atomic layer deposition (ALD) to form metal oxide semiconductor capacitors, including effects of pretreatment on N-polar GaN, ALD TiO[subscript]2/Al[subscript]2O[subscript]3 nano-laminate on thermal oxidized Ga-polar GaN and ALD Al[subscript]2O[subscript]3 on [Italic]c- and [Italic]m-plane GaN Surface pretreatments were shown to greatly alter the morphology of reactive N-polar GaN which is detrimental to the electrical properties. 14 nm thick ALD Al[subscript]2O[subscript]3 films were directly deposited on N-polar GaN without thermal or chemical pretreatments which yield a smooth surface (RMS=0.23 nm), low leakage current (2.09 x 10[superscript]-[superscript]8 A/cm[superscript]2) and good Al[subscript]2O[subscript]3/GaN interface quality, as indicated by the low electron trap density (2.47 x 10[superscript]10 cm[superscript]-[superscript]2eV[superscript]-[superscript]1). In the nano-laminate study, a high dielectric constant of 12.5 was achieved by integrating a TiO[subscript]2/Al[subscript]2O[subscript]3/Ga[subscript]2O[subscript]3 oxide stack layer, while maintaining a low interface trap density and low leakage current. There was a strong correlation between the surface morphology and electrical properties of the device discovered from comparing the ALD Al[subscript]2O[subscript]3 on [Italic]c- and [Italic]m-plane GaN, namely smooth surface lead to small hysteresis. These results indicate the promising potential of incorporation gate dielectric for future GaN devices.

Master of Science / Department of Chemical Engineering / Jennifer L. Anthony / A polymer has drastically different physical properties above versus below some characteristic temperature. For this reason, the precise identification of this glass transition temperature, T[subscript]g, is critical in evaluating product feasibility for a given application. The objective of this report is to review the behavior of polymers near their T[subscript]g and assess the capability of predicting T[subscript]g using theoretical and empirical models. It was determined that all polymers begin to undergo structural relaxation at various temperatures both nearly above and below T[subscript]g, and that practical assessment of a single consistent T[subscript]g is successfully performed through consideration of only immediate thermal history and thermodynamic properties. It was found that the best quantitative structure-property relationship (QSPR) models accurately predict T[subscript]g of polymers of theoretically infinite chain length with an average error of less than 20 K or about 6%, while T[subscript]g prediction for shorter polymers must be done by supplementing these T[subscript]g (∞) values with configurational entropy or molecular weight relational models. These latter models were found to be reliable only for polymers of molecular weight greater than about 2,000 g/mol and possessing a T[subscript]g (∞) of less than about 400 K.

Master of Science / Department of Chemical Engineering / Keith Hohn / Acid catalysts have been shown to be very successful in the pretreatment of cellulosic biomass to improve glucose yield and improve overall yield of ethanol. This report presents the results of a techno-economic study that looks into the use of nanoscale magnetic solid acid catalysts for glucose production. Magnetic solid acid catalysts are an improvement over using diluted acid due to eliminating acid-waste generation and corrosion hazards. Their magnetic nature also allows them to be easily separated from reaction products by an external magnetic force. After the technology is analyzed, a series of unit operations is proposed to go from the laboratory scale to the industrial plant scale. The next step was to develop material and energy balances using HYSYS process simulation software. Capital and operating costs are estimated and all the information is combined into a discounted cash flow economic model. The economic portion of the report uses a probabilistic cost assessment. It is used to quantify the range of risks in the project from swings in feedstock costs, differences in yield from catalysts, and any other significant variables. Both capital costs (initial equipment & construction investment) and operating costs (feedstock supply, chemicals, and personnell) are included with ranges of error based on databases and expert opinion. This method of evaluating investment efficiency can be helpful for predicting the cost benefits of proposed future research. The yield and percent catalyst magnetically recovered is assumed based on laboratory research to simplify the model. A 2000 metric tons of biomass per day facility was analyzed. Using the magnetic solid acid catalyst technology, the capital costs are estimated to be $160 million and this technology saves around 10% of capital costs compared to ethanol plants that uses conventional acid hydrolysis. The yield of the magnetic solid acid catalysts should be around 75% to compete with existing ethanol technologies. The metric used for this report is the discount profitability index (DPI) which is the ratio of future cash flows divided by investment. A DPI “hurdle rate” of 1.3 is used, which is similar to industry economic metrics of projects that include new process plants. The calculated DPI for the project is 1.38 DPI which is higher than using conventional cellulose treatment technologies. The recommendation is continue to study this technology’s large scale applicability before attempting any plant pilot studies.

Master of Science / Department of Chemical Engineering / J. H. Edgar / Boron phosphide, BP, is a III-V compound semiconductor with a wide band gap of 2.0 eV that is potentially useful in solid state neutron detectors because of the large thermal neutron capture cross-section of the boron-10 isotope (3840 barns). In this study, cubic BP crystals were grown by crystallizing dissolved boron and phosphorus from a nickel solvent in a sealed (previously evacuated) quartz tube. The boron - nickel solution was located at one end of the tube and held at 1150°C. Phosphorus, initially at the opposite end of the tube at a temperature of 430°C, vaporized, filling the tube to a pressure of 1–5 atmospheres. The phosphorus then dissolved into solution, producing BP. Transparent red crystals up to 4 mm in the largest dimension with mostly hexagonal shape were obtained with a cooling rate of 3°C per hour. The crystal size decreased as the cooling rate increased, and also as growth time decreased. The characterization with x-ray diffraction (XRD) and Raman spectroscopy established that the BP produced through this method were highly crystalline. The lattice constant of the crystals was 4.534 Ǻ, as measured by x-ray diffraction. Intense, sharp Raman phonon peaks were located at 800 cm[superscript]-1 and 830 cm[superscript]-1, in agreement with the values reported in the literature. The FWHM for XRD and Raman spectra were 0.275° and 4 cm[superscript]-1 which are the narrowest ever reported and demonstrates the high quality of the produced crystals. Energy dispersive x-ray spectroscopy (EDS) and scanning electron microscope (SEM) also confirmed the synthesized crystals were cubic BP crystals, with a boron to phosphorus atomic ratio of 1:1. Defect selective etching of BP at 300ºC for two minutes with molten KOH/NaOH revealed triangular and striated etch pits with low densities of defects of ~4 x 10[superscript]7 cm[superscript]-2 and 9.2 x 10[superscript]7 cm[superscript]-2 respectively. The BP crystals were n-type, and an electron mobility of ~39.8 cm[superscript]2/V*s was measured. This is favorable for application in neutron detection. Scaling to larger sizes is the next step through gradient freezing and employing a larger crucible.

Doctor of Philosophy / Chemical Engineering / Mary E. Rezac / Membrane contact reactors (MCRs) have been evaluated for the selective hydro-treating of model reactions; the partial hydrogenation of soybean oil (PHSO), and the conversion of lactic acid into commodity chemicals. Membranes were rendered catalytically active by depositing metal catalyst onto the polymer "skin" of an asymmetric membrane. Hydrogen was supplied to the support side of the membrane and permeated from the support side to the skin side, where it adsorbed directly onto the metal surface. Liquid reactant was circulated over the membrane, allowing the liquid to come into direct contact with the metal coated surface of the membrane, where the reaction occurred. Our membrane contact reactor approach replaces traditional three-phase batch slurry reactors. These traditional reactors possess inherent mass transfer limitations due to low hydrogen solubility in liquid and slow diffusion to the catalyst surface. This causes hydrogen starvation at the catalyst surface, resulting in undesirable side reactions and/or extreme operating pressures of 100 atmospheres or more. By using membrane reactors, we were able to rapidly supply hydrogen to the catalyst surface. When the PHSO is performed in a traditional slurry reactor, the aforementioned hydrogen starvation leads to a high amounts of trans-fats. Using a MCR, we were able to reduce trans-fats by over 50% for equal levels of hydrogenation. It was further demonstrated that an increase in temperature had minimal effects on trans-fat formation, while significantly increasing hydrogenation rates; allowing the system to capture higher reaction rates without adversely affecting product quality. Additionally, high temperatures favors the hydrogenation of polyenes over monoenes, leading to low amounts of saturated fats. MCRs were shown to operator at high temperatures and: (1) capture high reaction rates, (2) minimize saturated fats, and (3) minimize trans-fats. We also demonstrated lactic acid conversion into commodity chemicals using MCRs. Our results show that all MCR experiments had faster reaction rate than all of our controls, indicating that MCRs have high levels of hydrogen coverage at the catalyst. It was also demonstrated that changing reaction conditions (pressure and temperature) changed the product selectivities; giving the potential for MCRs to manipulate product selectivity.