Master of Science / Department of Chemical Engineering / James H. Edgar / Silicon oxidation has been the cornerstone of the semiconductor industry for many years, so understanding and being able to predict the oxidation process is paramount. The most popular model to date is the Deal-Grove model for the thermal oxidation of planar silicon surfaces. The Deal-Grove model owes its popularity to the overall simplicity in which it was derived and the accuracy in which it predicts the oxidation of planar silicon geometries. Due to this popularity and accuracy it is desirable to extend the Deal-Grove model beyond flat surfaces to other geometries such as cylinders and spheres. Extending the Deal-Grove model to these types of geometries would allow the prediction of the oxidation of silicon nano-wires and silicon nanocrystals. Being able to predict the oxidation is attractive due to the recent progress of integration of silicon nano-wires and silicon nano-crystals into microelectronic devices. Prediction of the oxidation of silicon cylinders (nano-wires) and spheres (nano-crystals) by simply utilizing the established planar Deal-Grovel model results in highly exaggerated oxide thicknesses compared with empirical data. This exaggeration for small silicon cylinders and spheres is due to the effects of the reduction in the available surface area for oxidation along with the stress induced due to the volumetric expansion and viscous flow of the oxide on non-planar surfaces. These stress effects retard the oxidation rate in non-planar silicon geometries with respect to flat surfaces. This reduction in the oxidation rate reduction is caused by the normal compressive stress which is normal to the SiO[subscript]2/Si interface due to the volumetric expansion during oxidation. This compressive stress reduces the reaction rate constant at the SiO[subscript]2/Si interface and thus retards the overall oxidation rate for silicon cylinders and spheres with respect to planar silicon. The focus of this paper will be to contrast cylindrical and spherical versions of the Deal-Grove model to the well established planar version. Surface area and stress effects will also be explored as they help explain the reduction in the oxidation rate for non-planar silicon geometries.

This M.A.Sc. work focuses on the development of a sensor to operate at industrially relevant conditions, directly in autoclaves, to generate accurate online free acidity estimates. The main goal is to obtain accurate (within 5% from titrated) free acid estimates at leach temperature while compensating for various metals in solution at high solid loadings. The electrodeless conductivity sensor and calibration technique that were developed in previous work have been improved upon for operation in slurry systems. The sensor operates at temperature on the basis of induction currents and is suitable for very concentrated solutions containing solid dispersions. Three systems have been investigated: limonitic feeds, limonite-saprolite blends, and a converter slag feed. Up to 40 wt.% solids were tested with investigation carried out from 25°C to 260°C. High quality estimates of acidity were produced for all systems studied, with the average difference between measured and titrated acidities being within 5%.

Hemp, one of the fast-growing plants with strong, durable fibers is being used for industrial applications, including papermaking, as a biodegradable reinforcing material for plastics as well as in construction. A series of tests has been conducted grinding hemp with the SM-220 Szego Mill equipped with three helically grooved, 30 cm long rollers with 8*8 or 4*4 mm grooves. A range of flow rates and rotational speeds has been covered in both dry and wet operation to determine the best processing conditions for various potential applications of refined hemp fibers. Particle size distributions and power consumption have been measured. Fiber length and diameter measurements allow an evaluation of fiber aspect ratio. For dry grinding, these are relatively low, typically in the 6-10 range. Wet grinding generally favors higher aspect ratios due to reduced fiber cutting and greater clearing action in the mill. Measured values are in the 60-80 range

Novel plastics that are biodegradable and made from renewable natural resources are currently being researched as alternatives to traditional petroleum-based plastics. One such plastic, thermoplastic starch (TPS) is produced from starch processed at high temperatures in the presence of plasticizers, such as water and glycerol. However, because of its hydrophilic nature, TPS exhibits poor mechanical properties when exposed to environmental conditions, such as rain or humidity. The overall objective of this thesis was to produce a thermoplastic starch based material with low water absorption that may be used to replace petroleum-based plastics. Three different methods for reducing water absorption were investigated, including the following: extrusion of starch with hydrophobic polymers, starch modifying chemicals, and citric acid/sorbitol as plasticizers. It was found that all methods reduced the water absorption of TPS.

Hedge fund performance is modeled from publically available data using feed-forward neural networks trained using a resilient backpropagation algorithm. The neural network’s performance is then compared with linear regression models. Additionally, a stepwise factor regression approach is introduced to reduce the number of inputs supplied to the models in order to increase precision. Three main conclusions are drawn: (1) neural networks effectively model hedge fund returns, illustrating the strong non-linear relationships between the economic risk factors and hedge fund performance, (2) while the group of 25risk factors we draw variables from are used to explain hedge fund performance, the best model performance is achieved using different subsets of the 25 risk factors, and, (3) out-of-sample model performance degrades across the time during the recent (and still on-going) financial crisis compared to less volatile time periods, indicating the models’ inability to predict severely volatile economic scenarios such as economic crises.

This M.A.Sc. work focuses on the development of a sensor to operate at industrially relevant conditions, directly in autoclaves, to generate accurate online free acidity estimates. The main goal is to obtain accurate (within 5% from titrated) free acid estimates at leach temperature while compensating for various metals in solution at high solid loadings. The electrodeless conductivity sensor and calibration technique that were developed in previous work have been improved upon for operation in slurry systems. The sensor operates at temperature on the basis of induction currents and is suitable for very concentrated solutions containing solid dispersions. Three systems have been investigated: limonitic feeds, limonite-saprolite blends, and a converter slag feed. Up to 40 wt.% solids were tested with investigation carried out from 25°C to 260°C. High quality estimates of acidity were produced for all systems studied, with the average difference between measured and titrated acidities being within 5%.

Hemp, one of the fast-growing plants with strong, durable fibers is being used for industrial applications, including papermaking, as a biodegradable reinforcing material for plastics as well as in construction. A series of tests has been conducted grinding hemp with the SM-220 Szego Mill equipped with three helically grooved, 30 cm long rollers with 8*8 or 4*4 mm grooves. A range of flow rates and rotational speeds has been covered in both dry and wet operation to determine the best processing conditions for various potential applications of refined hemp fibers. Particle size distributions and power consumption have been measured. Fiber length and diameter measurements allow an evaluation of fiber aspect ratio. For dry grinding, these are relatively low, typically in the 6-10 range. Wet grinding generally favors higher aspect ratios due to reduced fiber cutting and greater clearing action in the mill. Measured values are in the 60-80 range

Novel plastics that are biodegradable and made from renewable natural resources are currently being researched as alternatives to traditional petroleum-based plastics. One such plastic, thermoplastic starch (TPS) is produced from starch processed at high temperatures in the presence of plasticizers, such as water and glycerol. However, because of its hydrophilic nature, TPS exhibits poor mechanical properties when exposed to environmental conditions, such as rain or humidity. The overall objective of this thesis was to produce a thermoplastic starch based material with low water absorption that may be used to replace petroleum-based plastics. Three different methods for reducing water absorption were investigated, including the following: extrusion of starch with hydrophobic polymers, starch modifying chemicals, and citric acid/sorbitol as plasticizers. It was found that all methods reduced the water absorption of TPS.

Hedge fund performance is modeled from publically available data using feed-forward neural networks trained using a resilient backpropagation algorithm. The neural network’s performance is then compared with linear regression models. Additionally, a stepwise factor regression approach is introduced to reduce the number of inputs supplied to the models in order to increase precision. Three main conclusions are drawn: (1) neural networks effectively model hedge fund returns, illustrating the strong non-linear relationships between the economic risk factors and hedge fund performance, (2) while the group of 25risk factors we draw variables from are used to explain hedge fund performance, the best model performance is achieved using different subsets of the 25 risk factors, and, (3) out-of-sample model performance degrades across the time during the recent (and still on-going) financial crisis compared to less volatile time periods, indicating the models’ inability to predict severely volatile economic scenarios such as economic crises.

Wood service life is significantly prolonged by the use of preservatives. Unfortunately, preservative treated wood is still susceptible to weathering degradation (UV and moisture) and subsequent leaching of preserving components. These negative impacts of weathering can be reduced by the application of a coating; however, the effectiveness of the coating depends on its characteristics, especially its compatibility with preservative treated wood. In this project, the performance characteristics of semitransparent deck stains were evaluated for untreated wood samples and for samples treated with chromated copper arsenate (CCA), alkaline copper quat (ACQ) and copper azole (CA) over three years of natural and three months of accelerated weathering conditions. The parameters measured were water permeability, colour change, general appearance, checking of wood, and the coating‟s ability to reduce preservative leaching. Coatings were characterized in terms of glass transition temperature (Tg), solid content, viscosity, density, contact angle, surface tension, and film thickness. Also penetration depth of a polyurethane (PU) coating was examined using Time-of-Flight Secondary Ion Mass Spectroscopy (ToF-SIMS). iii All the stains evaluated effectively reduced cumulative leaching of preservative components by about 60% on average. An analysis of the preservative gradients and residual soluble components in the ACQ-samples after weathering indicated that preferential leaching of monoethanolamine (Mea) is most likely responsible for the reduced amounts of available copper in coated treated wood samples. Also, a two-week screening test was able to provide accurate predictions of the long-term leaching performance of different coatings. There was a significant interaction effect between coatings and preservatives: solvent-based coatings showed better water repellency for CCA and untreated wood, but there was no significant difference in water repellent effectiveness between water-based and solvent-based coatings for ACQ or CA-treated wood. Overall, preservative treatments greatly enhanced coating performance. Image analysis of the samples subjected to 3 years weathering showed that coatings reduced surface checking by 30-40%. Partial least squares regression (PLS-R) modeling was used to correlate measured coating properties with their weathering performance characteristics. The modeling results showed that coatings with low Tg and high viscosity effectively reduce the leaching of preservative components and improved water repellency and visual ratings