The reaction between carbon and iron oxide-containing slag is crucial to efficient electric arc furnace steelmaking. The reaction occurs via gaseous intermediates, and the rate of gas generation by carbon gasification is limited by the chemical reactions at the slag-gas and carbon-gas interfaces. The aim of the present study was to obtain an understanding of the gasification rate limiting factors and slag foaming behaviour that could be readily applied to industrial electric arc furnace situations. The rate of carbon gasification was measured in experimental simulations of an electric arc furnace heat with slags containing between 21.6 and 48.2 wt% 'FeO'. It was found that rate control was dominated by the carbon-gas chemical reaction. A model was developed which describes the carbon gasification rate, amount of residual carbon in the slag, gas composition, slag-gas interfacial area and bubble diameter during carbon injection into slag. The model predicts rate control by the carbon-gas chemical reaction, in agreement with experimental observations. The slag foaming behaviour was investigated, and it was found that the foaming index is a useful parameter in quantifying foam height only if void fraction is constant with respect to gas flow rate. The average bubble size was observed to be an important factor in determining foam stability, with smaller bubble size resulting in greater foam height. / Thesis / Master of Applied Science (MASc)

<p> Substantial development of silicon-based technology is required to continue to improve product reliability and production yield of silicon-based IC devices. Defects play a key role in process technology and research is required into their properties and interactions with host and impurity atoms. Cavities formed by ion implantation of helium into a silicon substrate are known to be efficient gettering sites for mobile interstitials and metallic impurities. In addition, the existence of a buried void layer drastically reduces the unintentional parasitic transistor gain in power devices by introducing mid-gap energy levels in the substrate. Utilizing slow positron annihilation spectroscopy (PAS), void formation by implanted He+ at a dose of lxl016/cm2 and energy of 60 keV subsequently subjected to various annealing conditions, i.e. different temperatures, times, ramp rate and ambients, is reported. Quantitative measurement of cavity sizes shows that the annealing temperature largely influences helium out-diffusion from the implanted region. Consequently, different void evolution processes associated with specific annealing temperatures are found. Furthermore, larger voids are formed in oxygen ambient leading us to suggest that the supersaturation of interstitials enhances bubble migration and coalescence during the stage when a large fraction of He atoms remains in the cavities. </p> <p> In recent years, SiGe pseudomorphic alloys have become attractive for heterostructure devices due to their higher mobility, lower noise and lower power consumption, as compared with traditional Si devices. More importantly, SiGe is highly compatible with Si processing technologies. With the continuous improvement of SiGe technology, it has been coupled with complementary metal oxide semiconductor (CMOS) technology and has the potential to replace III-V compound semiconductor devices in the near future. However, many unknowns remain in this material system. The study of point defect injection during various thermal treatments is important in itself, and would also allow additional tools for the study of dopant diffusion in the material under different conditions. With preexistent voids in the buried substrate, we performed furnace wet oxidation on a Si0.98Geo.08 sample at 900 °C for half an hour. A small shrinkage of the voids, as compared to complete annihilation in the case of inert annealing, signifies vacancy injection during the oxidation process. Possible defect generation mechanisms and difference in growth rate enhancement in dry and wet oxidation are discussed. Based on a literature review of Si and SiGe oxidation, we suggest that stress relaxation and the Si replacement mechanism are responsible for the catalytic effect of the oxide growth and the change of point defect generation. </p> / Thesis / Master of Applied Science (MASc)

Crop production, soil system, water quality, and economic impacts of four corn silage production systems were compared through a field study including 16 plots (4 replications of each treatment). Systems included a rye cover crop and application of liquid dairy manure in the spring and fall. The four management systems were: 1) traditional, 2) double-crop, 3) roll-down, and 4) undercut. In the fourth system, manure was applied below the soil surface during the undercutting process. In all other systems, manure was surface-applied. In the third system, the rye crop was flattened with a heavy roller after manure application. Simulated rainfall was applied within 48 h of manure application. Measured constituent concentrations in runoff were compared with water quality criteria. Costs and returns of all systems were compared. The undercut system reduced loadings of all nutrients, but increased total suspended solids (TSS) concentration as compared with all other systems. The mean volume of runoff from the undercut system was less than half that from any other system, which influenced all constituent loadings. Mean TSS concentration in runoff from the undercut system was over three times the mean of any other system. The roll-down system had no significant effect on water quality as compared to the traditional system. The undercut system was reasonably effective in keeping phosphate phosphorus levels below the criterion set for bathing water. None of the systems generally exceeded nitrate nitrogen concentration criteria. However, total phosphorus, orthophosphate, fecal coliform and e. coli criteria for drinking, bathing, shellfish harvest, and aesthetics were regularly exceeded by all of the systems. There were no differences among the treatments in effects on bacterial concentrations. The double-crop system produced significantly higher net returns than all other systems only if the value of the rye crop was $92.31/Mg or more. There were no significant differences in net returns of the traditional, roll-down, or undercut systems. / Ph. D.

Generally, the model developed in this investigation predicted quite well the values of the parameters measured experimentally. In certain cases, deviations were observed, but these were due to known factors which have been described in the previous section. This model represents the first link in a chain that is just beginning. The end use of the program will be in DOC feed-forward control applications for injection molding cycles. Several improvements can be made on this model. First, a more detailed analysis should yield a better relationship between the volumetric flow rate and the pressure at a given point. This is necessary because the volumetric flow rate should be smoothly decreased in order to keep the runner entrance pressure at the constant preset injection pressure. In this investigation, a power law relationship was used. This exponential relationship proved to decrease the volumetric flow rate faster than it should have. This caused some oscillations in the predicted temperature and pressure values. It was observed from the experimental data that a linear relationship could do the job, and that the volumetric flow rate should be decreased at each time interval as the pressure increased, instead of waiting until the preset injection pressure was reached as it was done in this investigation. Second, a non-linear regression fit should be performed on the experimental data available for heat capacity and thermal conductivity. The regression equations could be used in the model. This will reduce the consistently high temperatures predicted by the model. Third, the model should be extended to take into account the heat transferred to the wall of the channels. In this investigation a constant wall temperature was assumed in order to simplify the initial development of the model. At this point the finite difference grid used in the model could be extended to include the metal containing the cooling channels where the temperature is known. The inclusion of this feature could be accomplished with a small programming effort. Finally better packing and cooling models could be developed. The cooling model proved to be the poorest of the proposed models. An extra effort to develop a better cooling model was not considered to be necessary, because this investigation was more concerned with the filling and packing stages, which are the critical stages for gate design purposes. / Ph. D.

Background: Glaucoma is a chronic ocular disease of both dogs and humans that results in blindness and ocular discomfort. Most commonly, end-stage glaucomatous eyes in dogs are enucleated to provide comfort. This intervention requires significant financial investment, general anesthesia, and has a psychological impact on some owners. Retrobulbar neurolytic injections are used in humans to provide immediate and long-acting pain relief, while simultaneously preserving the globe. Objectives: To determine the safety and efficacy of retrobulbar neurolytic ethanol injection in canine eyes with end-stage glaucoma. Animals: 16 client-owned dogs (19 eyes) diagnosed with end-stage glaucoma. Methods: All eyes underwent an ophthalmic examination, including Schirmer Tear Testing (STT), intraocular pressure (IOP) measurement, corneal touch threshold (CTT), anterior and posterior segment examination, and fluorescein staining. Subjects were sedated and administered a retrobulbar block with bupivacaine, followed by retrobulbar injection of ethanol or saline solution. At specified time points after the procedure, clients assessed their pet's comfort and side effects of the injections via survey. Subjects returned for enucleation. Owner perceived comfort after the enucleation was assessed at identical post-procedure time points and compared to that achieved with retrobulbar ethanol injection or control solution. Overall client satisfaction with each procedure, as well as the effects of retrobulbar ethanol injection on STT, IOP, CTT, and histological changes in retrobulbar tissues were investigated. Results: Retrobulbar neurolytic ethanol injections did not signficiantly improve owner perceived comfort compared to control group treatment or provide more comfort than enucleation. Retrobulbar ethanol injections did not signficantly lower IOP, but did significantly elevate CTT. There was a trend towards lower STT in eyes receiving retrobulbar ethanol injections. Retrobulbar ethanol injections were safe, well tolerated, and no differences in client satisfaction with particpation in the study were noted in either injection group. Histologically, globes in the treatment group displayed significantly greater inflammation and fibrosis; retrobublar tissue samples were not significantly different between control and treatment groups with regard to inflammation or fibrosis. Conclusions and Clinical Importance: Retrobulbar neurolytic ethanol injections were not determined to be an effective globe-sparing alternative treatment to provide analgesia for end-stage canine glaucoma. Enucleation remains an effective way to provide comfort to dogs with elevated IOP. / Master of Science / The glaucomas represent a diverse group of blinding and painful diseases associated with elevated intraocular pressure (IOP). Despite advances in the medical and surgical treatment of glaucoma, the long-term prognosis in dogs remains dismal for IOP control, comfort, and globe retention. Blindness and pain are common long-term outcomes, necessitating surgical salvage procedures aimed at restoring patient comfort, including enucleation (eye removal), intrascleral prosthesis, or intravitreal chemical cyclodestruction. The most commonly performed, effective, and predictable of these options is enucleation, but this requires general anesthesia, a considerable financial investment, risks post-surgical complications, and has a negative psychological impact on some owners. Retrobulbar neurolytic injections with absolute ethanol have been performed in humans with blind, painful eyes since the early 1900s. Immediate and long lasting pain relief can be achieved from 2 weeks to 2 years after a single injection. The purpose of this study is to determine the safety and efficacy of retrobulbar ethanol injections as a globe-sparing therapeutic option for end-stage glaucoma in dogs. Nineteen dogs presenting to the VTH ophthalmology service with end-stage glaucoma were enrolled in a prospective, randomized, double-masked clinical trial. Subjects were sedated and administered a retrobulbar injection of ethanol (n=9) or control saline solution (n=10). At specified time points after the procedure, clients assessed their pet’s comfort and side effects of the injections via survey. Three weeks later, subjects returned for enucleation and the level of comfort after the enucleation was assessed at identical post-procedure time points and compared to that achieved with retrobulbar ethanol injection or control solution. Retrobulbar neurolytic ethanol injections did not significantly improve comfort compared to control group sham treatment or provide more comfort than enucleation. Retrobulbar ethanol injections did not significantly lower IOP, but did significantly elevate corneal touch threshold in treated patients. There was a trend towards lower tear production in eyes receiving retrobulbar ethanol injections. Retrobulbar ethanol injections were safe, well tolerated, and no differences in client satisfaction with participation in the study were noted in either injection group. Further investigation is warranted to determine the optimal volume of retrobulbar ethanol to provide analgesia for patients with end-stage glaucoma as well as to determine the duration of clinical effect of these injections.

A newly designed liquid fuel (kerosene) aeroramp injector/plasma igniter was tested in cold flow using the Virginia Tech supersonic wind tunnel at Mach 2.4. The liquid fuel (kerosene) injector is flush wall mounted and consists of a 2 hole aeroramp array of impinging jets that are oriented in a manner to improve mixing and atomization of the liquid jets. The two jets are angled downstream at 40 degrees and have a toe-in angle of 60 degrees. The plasma torch used nitrogen and air as feedstocks and was placed downstream of the injector as an ignition aid. First, schlieren and shadowgraph photographs were taken of the injector flow to study the behavior of the jets, shape of the plume, and penetration of the liquid jet. The liquid fuel aeroramp was found to have better penetration than a single, round jet at 40 degrees. However, the liquid fuel aeroramp does not penetrate as well as an upstream/downstream impinging jet in a plane aligned with the flow. Next, the Sauter mean droplet diameter distribution was measured downstream of the injector. The droplet diameter was found to vary from 21 to 37 microns and the atomization of the injector does not appear to improve beyond 90 effective jet diameters from the liquid fuel aeroramp. These results were then used to decide on an initial location for the plasma torch. The combined liquid injector/plasma torch system was tested in an unheated (300 K) Mach 2.4 flow with a total pressure of 345 kPa. The liquid fuel (kerosene) volumetric flow rate was varied from 0.66 lpm to 1.22 lpm for the combined liquid injector/plasma torch system. During this testing the plasma torch was operated from 1000 to 5000 watts with 25 slpm of nitrogen and air as feedstocks. The interaction between the spray plume and the plasma torch was observed with direct photographs, videos, and photographs through an OH filter. It is difficult to say that any combustion is present from these photographs. Of course, it would be surprising if much combustion did occur under these cold-flow, low-pressure conditions. Differences between the interaction of the spray plume and the plasma torch with nitrogen and air as feedstocks were documented. According to the OH wavelength filtered photographs the liquid fuel flow rate does appear to have an effect on the height and width of the bright plume. As the liquid fuel flow rate increases the bright plume increases in height by 30% and increases in width slightly (2%). While, a decrease in liquid fuel flow rate resulted in an increase in height by 9% and an increase in width by 10%. Thus, as the liquid fuel flow rate varies the width and height of the bright plume appear to always increase. This can be explained by noticing that the shape of the bright plume changes as the liquid fuel flow rate varies and perhaps anode erosion during testing also plays a part in this variation of the bright plume. From the OH wavelength filtered photographs it was also shown that the bright plume appears to decrease in width by 9% and increase in height by 22% when the plasma torch is set at a lower power setting. When air is used as the torch feedstock, instead of nitrogen, the penetration of the bright plume can increase by as much as 19% in width and 17% in height. It was also found that the height and width of the bright plume decreased slightly (2%) as the fuel flow rate increased when using air as the torch feedstock. Testing in a hot-flow facility is planned. / Master of Science

An experimental investigation of a flush-wall, diamond-shaped injector was conducted in the Virginia Tech supersonic wind tunnel. The diamond injector was elongated in the streamwise direction and is aimed downstream angled up at 60° from the wall. Test conditions involved sonic injection of helium heated to approximately 313 K into a nominal Mach 4.0 crossstream airflow. These conditions are typical of a scramjet engine for a Mach 10 flight, and heated helium was used to safely simulate hydrogen fuel. The injector was tested at two different injectant conditions. First, it was investigated at a baseline mass flow rate of 3.4 g/s corresponding to an effective radius of 3.54 mm and a jet-to-freestream momentum flux ratio of 1.04. Second, a lower mass flow rate of 1.5 g/s corresponding to an effective ratio of 2.35 mm and a jet-to-freestream momentum flux ratio of 0.49 was studied. The diamond injector was tested both aligned with the freestream and at a 15&#176; yaw angle for the baseline mass flow rate and aligned with the freestream at the lower mass flow rate. For comparison, round injectors angled up at 30° from the wall were also examined at both flow rates. A smaller round injector was used at the lower mass flow rate such that the jet-to-freestream momentum flux ratio was 1.75 for both cases. A concentration sampling probe and gas analyzer were used to determine the local helium concentration, while Pitot, cone-static and total temperature probes were used to determine the flow properties. The results of the investigation can be summarized as follows. For the baseline case, the aligned diamond injector penetrated 44% higher into the crossflow than did the round injector. The addition of yaw angle increased the crossflow penetration to 53% higher than the round injector. The aligned diamond injector produced a 34% wider jet than the round injector, while the addition of yaw angle somewhat reduced this widening effect to 26% wider than the round injector. The aligned and yawed diamond injectors exhibited 10% and 15% lower mixing efficiency than the round injector, respectively. The total pressure loss parameter of the aligned diamond was 22% lower than the round injector, while the addition of yaw angle improved the total pressure loss parameter to 34% lower than the round injector. For the lower mass flow (and momentum flux ratio) case, the diamond injector demonstrated 52% higher penetration and a 39% wider plume than the round injector. The mixing efficiency was nearly identical between the two injectors with just a 4% lower mixing efficiency for the diamond injector. The total pressure loss parameter of the diamond injector was 32% lower than round injector. These results confirm the conclusions of earlier, lower free stream Mach number and higher molecular weight injectant, studies that a slender diamond injector provides significant benefits for crossflow penetration and lower total pressure losses. / Master of Science

The integration of large language models (LLMs) with third party applications has allowed for LLMs to retrieve information from up-to-date or specialized resources. Although this integration offers numerous advantages, it also introduces the risk of indirect prompt injection attacks. In such scenarios, an attacker embeds malicious instructions within the retrieved third party data, which when processed by the LLM, can generate harmful and untruthful outputs for an unsuspecting user. Although previous works have explored how these attacks manifest, there is no benchmarking framework to evaluate indirect prompt injection attacks and defenses at scale, limiting progress in this area. To address this gap, we introduce InjectBench, a framework that empowers the community to create and evaluate custom indirect prompt injection attack samples. Our study demonstrate that InjectBench has the capabilities to produce high quality attack samples that align with specific attack goals, and that our LLM evaluation method aligns with human judgement. Using InjectBench, we investigate the effects of different components of an attack sample on four LLM backends, and subsequently use this newly created dataset to do preliminary testing on defenses against indirect prompt injections. Experiment results suggest that while more capable models are susceptible to attacks, they are better equipped at utilizing defense strategies. To summarize, our work helps the research community to systematically evaluate features of attack samples and defenses by introducing a dataset creation and evaluation framework. / Master of Science / Large language models (LLMs), such as ChatGPT, are now able to retrieve up-to-date information from online resources like Google Flights or Wikipedia. This ultimately allows the LLM to utilize current information to generate truthful, helpful and accurate responses. Despite the numerous advantages, it also exposes a user to a new vector of attacks known as indirect prompt injections. In this attack, an attacker will write a instruction onto an online resource that an LLM will process when retrieved from the online resource. The primary aim of the attacker is to instruct the LLM to say something it is not supposed to, and thus may manifest as a blatant lie or misinformation given to the user. Prior works have studied and showcased the harmfulness of this attack, however not many works have tried to understand which LLMs are more vulnerable to indirect prompt injection attacks and how we may defend from them. We believe that this is mainly due to the non-availability of a benchmarking dataset which allows us to test LLMs and new defenses. To address this gap, we introduce InjectBench, a methodology that allows the automated creation of these benchmarking datasets, and the evaluation of LLMs and defenses. We show that InjectBench can produce a high quality dataset that we can customize to specific attack goals, and that our evaluation process is accurate and agrees with human judgement. Using the benchmarking dataset created from InjectBench, we evaluate four LLMs and investigate defenses for indirect prompt injection attacks.

I present a detailed study of the magnetic propertie of electron-beam evaporated Pdo.4Nio.6 alloy thin films by means of ferromagnetic resonance measurements on extended films of varying thickness and anisotropic magnetoresistance measurements lithographically patterned high aspect-ratio ferromagnetic electrodes, respectively. The results reveal that the direction of the magnetization with respect to the film plane strongly depends on the electrode lateral dimensions, transitioning from in-plane magnetization for extended films to out of the plane magnetization for electrode width below 2-3 microns, reaching ~58 degrees for electrode widths of about 100nm (nanowires). This behavior arises from a competition between the film demagnetizing vector, which leads to in-plane magnetization for extended films , and an intrinsic uniaxial anisotropy, which overcomes the magnetostatic energy for laterally constrained films, pulling the magnetization off plane.

Improvements to heritage gas turbine engines will be needed in the coming years as the demand made on these systems increase. While industry continually presses for higher performance of both military and civilian aero engines, the government simultaneously raises the bar for emissions standards in the commercial sector to support public health. The next generation of aerospace gas turbine engines will be defined by their ability to operate at high power conditions while maintaining efficiency. This challenge is compounded by airlines' proposition of a return to supersonic flight- an operating regime characterized by higher total temperatures, and thus more NOx production. Lean Direct Injection (LDI) is a combustion scheme that was proposed by NASA, and inherently addresses the needs of both the private sector and the military. LDI is a liquid fueled combustor that promotes rapid mixing of fuel and air at the entrance of the combustor. Despite the benefits of LDI, it has never been implemented, nor has any other lean burning scheme been implemented in an aircraft due to the system level complications of such technology. This dissertation focuses on the dynamics of thermoacoustic instability and lean blowout (LBO), two of the major complications that industry will face when they attempt to incorporate LDI in a production engine. The present dissertation investigates these dynamics from a fundamental and applications standpoint. Fundamental insights on thermoacoustic instabilities are developed by investigating droplet dynamics in a self-excited flow field, and significant oscillations in droplet diameters are discerned. PDPA measurement will be taken to identify coupling of the fuel spray with the instability, and a phase locking algorithm will be used to develop a new spray parameter than is more indicative of combustion heat release that the standard Sauter mean diameter. Next, while varying the swirl number and the venturi geometry of the combustor, the evolution of the flow field will be characterized. An in-house innovation called the Direct Rotation Swirler (DRS) is built for this purpose. The DRS uses an active geometry to provide continuously variable swirl number modulation. The effects of these changes on lean blow out, pressure drop and NOx emissions will then be experimentally determined. Venturis were rapidly manufactured using a ii casting procedure that was developed to make venturi geometries from a commercially available ceramic at very low cost. / Doctor of Philosophy / Improvements to heritage gas turbine engines will be needed in the coming years as the demand made on these systems increase. While industry continually presses for higher performance of both military and civilian aero engines, the government simultaneously raises the bar for emissions standards in the commercial sector to support public health. The next generation of aerospace gas turbine engines will be defined by their ability to operate at high power conditions while maintaining efficiency. This challenge in compounded by airlines' proposition of a return to supersonic flight- an operating regime characterized by higher total temperatures, and thus more NOx production. Lean Direct Injection (LDI) is a combustion scheme that was proposed by NASA, and inherently addresses the needs of both the private sector and the military. LDI is a liquid fueled combustor that promotes rapid mixing of fuel and air at the entrance of the combustor. Rapid mixing yields a clean, even flame and eliminates the fuel-rich primary zone which is the heart of NOx production. Despite the benefits of LDI, it has never been implemented, nor has any other lean burning scheme been implemented in an aircraft due to the system level complications of such technology. This dissertation focuses on two of the major complications that industry will face when they attempt to incorporate LDI in a production engine. Drastically reducing the local hot spots in the primary zone is fundamentally necessary to improve pattern factor and emissions, but this change is directly at odds with two dynamic phenomenon that plague combustors. These effects are thermoacoustic instabilities, and lean blow out. Thermoacoustic instabilities are a major concern in any type of combustor and instabilities are more common and more intense in lean engines which is a significant safety risk to aircraft. A thermoacoustic instability occurs when pressure waves in an engine grow to high amplitudes. Small pressure waves are normal in any combustion process, but when the acoustic waves couple with the heat release, the waves can grow uncontrollably. The amplitudes can reach magnitudes capable of damaging the combustor or significantly reducing its cyclic life. Due to the high iv standard of safety in the aerospace industry, lean combustion will not be implemented until engines can be designed to avoid instabilities throughout the entire flight envelope. Lean blow out occurs when the fuel to air ratio of the engine becomes too low to sustain a flame. Lean blow out is a transient phenomenon that is dependent on local flame speeds, local chemical time scales and turbulence parameters. Typically, lean blow out is combated by designing a rich flame anchoring region that burns with plenty of excess fuel so that even if the fuel flow rate is reduced, a core region is still within its flammability regions. The present dissertation investigates these dynamics from a fundamental and applications standpoint. Fundamental insights on thermoacoustic instabilities are developed by investigating droplet dynamics in a self-excited flow field, and significant oscillations in droplet diameters are discerned. PDPA measurement will be taken to identify coupling of the fuel spray with the instability, and a phase locking algorithm will be used to develop a new spray parameter than is more indicative of combustion heat release that the standard Sauter mean diameter. Next, while varying the swirl number and the venturi geometry of the combustor, the evolution of the flow field will be characterized. An in-house innovation called the Direct Rotation Swirler (DRS) is built for this purpose. The DRS uses an active geometry to provide continuously variable swirl number modulation. The effects of these changes on lean blow out, pressure drop and NOx emissions will then be experimentally determined. Venturis were rapidly manufactured using a casting procedure that was developed to make venturi geometries from a commercially available ceramic at very low cost.