Continuous temperature measurement of molten steel in new as well as existing steelmaking processes is not always feasible with current technology. However, it is a well known fact that molten steel temperature, especially during the continuous casting operation, can have a profound effect on the quality of the product. A knowledge of the instantaneous steel temperature in the tundish can serve as a valuable tool in the quest for better quality. With the implementation of statistical process control techniques such occurrences as breakouts, tundish freeze up or the nucleation of an extensive area of columnar grains in the cast section can be reduced. / However, conventional thermocouples cannot withstand the severe corrosion around the slag line for a sustained period of time. Even though the sensing wires that comprise the hot junction can, if well protected, be used for long periods of time without serious deterioration, a suitable refractory cover or sleeve must be applied to the thermocouple prior to using it. / For the present work, two methods for continuously measuring melt temperature were developed. One of these relies on deducing melt temperature while the other is based on providing a cooled sleeve for the thermocouple to minimize slag line corrosion. The first technique involves the use of multiple thermocouples embedded in a refractory section at various displacements. When the refractory is contacted by the melt, transient heat transfer is initiated through the section. By analyzing this transient behaviour with a suitable heat transfer model, it is possible to infer the temperature of the melt. A mathematical model that adequately describes this process has been developed and tested. / An alternative approach that was developed uses heat pipe technology to prevent the corrosion of the thermocouple probe through solidification of slag onto the heat pipe which can serve as a protector of the probe. Heat pipes are devices capable of transferring large quantities of heat with very small temperature differences. Advantages of the heat pipe when used as a heat transmission device are constructional simplicity, flexibility, high heat transport capability, and no need for external pumping device. The heat pipe consists of a closed evacuated tube, porous wicking material, and working fluid. Heating one part of the external surface leads to evaporation of the working fluid and the establishment of a pressure gradient within the heat pipe. The resulting pressure difference drives the vapor from the evaporator to the condenser where the pressure and temperature are slightly lower. The effective conductance of the heat pipe can be several orders of magnitude higher than that of an equivalent solid copper bar. In the second part of this thesis, the feasibility of incorporating heat pipe technology to solidify and maintain a thin layer of slag on the heat pipe while it is in operation is presented.

The crystallization behaviour of iron intermetallics during solidification and the dissolution behaviour of iron intermetallics during heat treatment and their subsequent effect on mechanical properties have been studied. / Addition of iron decreases DAS and affects eutectic silicon particle size as nucleation and growth of silicon crystals takes place on the $ beta$-AlFeSi platelets, thus minimizing isothermal dendrite arm coarsening at the eutectic temperature. Increase of iron also increases the size and volume percent of the iron bearing intermetallics. / In the absence of manganese, the iron intermetallics crystallize in the $ beta$-phase, at all cooling rates ranging from 0.1 to 20$ sp circ$C/s when cast from a normal casting temperature of 750$ sp circ$C. In the presence of manganese, the iron intermetallic crystallizes in $ alpha$-phase at low cooling rates and in both the $ alpha$- and $ beta$-phases at high cooling rates. This reverse crystallization behaviour is explained based on the segregation effect displayed by the phase diagram. / When the melt is superheated to a high temperature (about 200 to 300 degrees above the liquidus temperature), the iron intermetallic crystallizes in the $ alpha$-phase at high cooling rates. This behaviour is attributed to the fact that $ gamma$-alumina which forms at low melt temperatures ($ le$750$ sp circ$C) acts as a nucleus for crystallization of $ beta$-phase. When the melt is superheated to a high temperature ($ ge$850$ sp circ$C), the $ gamma$-alumina transforms to $ alpha$-alumina. The $ alpha$-alumina is found to be a poor nucleus for the $ beta$-phase crystallization, and as a result the $ alpha$-phase forms. The importance of nucleation and growth undercooling for the crystallization of iron intermetallics is highlighted. / Investigation of the dissolution behaviour of the iron intermetallics on non-equilibrium heat treatment indicates that the $ beta$-phase platelets dissolve slowly through concurrent fragmentation and then dissolution at the plate tips. Addition of manganese hinders the dissolution of iron intermetallics. The amount of liquid phase formed during non-equilibrium heat treatment increases dramatically once a critical temperature is exceeded. This critical temperature is estimated to be 520 and 515$ sp circ$C for samples initially solidified at 10 and 15$ sp circ$C/s respectively. / The 0.15% Fe alloy exhibits the highest tensile strength and percent elongation compared to 1.0% Fe and 1.0% Fe + 0.5% Mn alloys under as-cast and equilibrium heat treated conditions. However, under non-equilibrium heat treatment conditions, ie., 30 degrees above the T4 equilibrium solution temperature, the strength properties of 1.0% Fe alloy exceed, or at least equal, that of the equilibrium heat treated 0.15% Fe alloy. / An attempt is made to correlate the iron intermetallics present in the microstructure and mechanical properties with the associated fracture mode in this alloy.

In this work, an artificial standard material was developed and used to examine the robustness of various different stereological correction procedures. The four correction procedures that were examined were: large-sections correction, Hill's fast approximation, Barbery's correction and PARGEN correction. / There were three steps to this work: (1) refinement of the standard material--certain modifications were made to make the standard material (developed as an M.Eng. project) more versatile and easier to use. (2) sectioning and correction of computer-generated spheres--different liberation distributions of single-capped spheres were computer-generated and sectioned. The four correction procedures were used to correct the stereological bias in the sectioning data. The corrected liberation distributions were compared with the true liberation distributions. (3) sectioning and correction of standard material particles--the standard material was used to re-create the same distributions that were computer-generated. The particles were mounted, sectioned and the sectioning data measured with an electron microscope and image-analyzer. The data were corrected using the correction procedures and the corrected and true distributions were compared. / A two-phase (glass/lead borate) standard material was successfully developed. This standard material can be made to exhibit granular, layered or simple locking. / The correction of the sectioning of the computer-generated spheres and standard material particles yielded similar observations about the different correction procedures. / The large-sections correction provided a simple, uncorrupted correction. It performed better in the sphere cases than in the standard material cases. / Hill's fast approximation performed well except in the cases of narrow liberation distributions. The sectioning data of the standard material cases appeared to support the assumption in this correction that the locked section and locked particle distributions are identical. / Barbery's correction performed better in the standard material cases than in the sphere cases. The correction had problems in situations where an incomplete beta function could not be fitted to the true liberation distribution. / The PARGEN correction was able to provide a good estimate of the true amount of free material, but it had difficulty estimating the locked particle distribution.

Heat dissipation from silicon chips is becoming critical as a result of increasing circuit density and power applied to these chips. In this regard, aluminum nitride (AlN) is the most promising candidate to be used as a substrate since it possesses high a thermal conductivity and thermal expansion coefficient close to that of silicon. In addition, metal-ceramic interfaces determine the heat dissipation through a circuit. In order to minimize the effect of interfaces, direct bonding (DB) of AlN to Cu was studied. DB process has been reported to provide a low thermal barrier and an interface which is free of thermal fatigue. / In this work process parameters of DB were optimized based on time, temperature and thickness of the Cu-foil for Cu-Al$ sb2$O$ sb3$ system in a N$ sb2$ atmosphere containing 500 ppm O$ sb2$ in a temperature range of 1065 to 1075$ sp circ$C. These conditions were then applied to the Cu-AlN system. Wettability of AlN by Cu was studied and improved through oxidation of AlN and modification of Cu by adding 1 at.% O$ sb2$. The activation energy for oxidation of AlN was found to be 94 kJ/mol. It was then shown that direct bonding of Cu to AlN can be performed without any intermediate layer. The average peel strength of AlN-Cu, A$ sb2$O$ sb3$-Cu and AlN-Al$ sb2$O$ sb3$-Cu systems were measured to be 42, 49 and 14.7 MPa, respectively.

The recrystallization behavior of three Nb-bearing HSLA steels was investigated during multipass deformation under continuous cooling conditions. The niobium concentrations of these steels varied from 0.05 to 0.09 wt%. The specimens were tested on a computerized torsion machine using a simulation schedule of 17 passes. Deformation temperatures of 1180$ sp circ$C to 700$ sp circ$C were employed, together with pass strains of 0.1 to 0.7, strain rates of 0.2 to 10 s$ sp{-1}$, and interpass times of 2 to 200 seconds. By means of mean flow stress vs. 1000/T diagrams, the effect of reheating temperature, initial grain size, and chemical composition on the T$ sb{ rm nr}$ (temperature at which recrystallization is no longer complete) was determined. Reheating temperature and microalloying elements have significant influence on the T$ sb{ rm nr}$, while initial grain size has no effect on the T$ sb{ rm nr}$ during multipass deformation. / Furthermore, the effect of deformation parameters such as the pass strain, strain rate, and interpass time on the T$ sb{ rm nr}$ during multipass deformation was investigated in this way. The T$ sb{ rm nr}$ decreases with increasing strain and also decreases slightly with increasing strain rate. There is a T$ sb{ rm nr}$ minimum at times of about 12 $ sim$ 15 seconds and both increases and decreases from this value raise this characteristic temperature. When the interpass times are short, solute atoms control the rate of recrystallization, the extent of which decreases as the time is decreased. When the interpass times are long, precipitation takes place and retards recrystallization, so that the extent of softening decreases as the time is extended instead. / The evolution of Nb(C,N) precipitation during simulated rolling was studied with the aid of carbon extraction replication and electron microscopy. Finally, by applying the additivity rule to the isothermal model of Dutta and Sellars, continuous cooling T$ sb{ rm nr}$'s were predicted from recrystallization-precipitation-temperature-time (RPTT) diagrams; these are shown to be in good agreement with the experimental observations.

The work of this thesis centres around two related themes, the first being the characterization of the through-thickness inhomogeneity found in rolled sheet at various stages in its production. The second is involved with attempting to understand how and to what extent the existing inhomogeneity affects the correlations with the anisotropies of some physical or mechanical properties. For this, the inhomogeneity of six steels were measured. / Three were specimens removed from the sheet during the early processing of conventional grain-oriented ferrosilicon steel, one sample taken after the first cold rolling stage, another taken after the intermediate anneal, while the last was taken after the following stage, i.e. after the second cold rolling. / To contrast the extensive inhomogeneity expected in the ferrosilicon steels, the remaining three steels consisted of two continuous-annealed, interstitial-free, extra-low-carbon steels and one aluminum-killed, batch-annealed, low-carbon steel, all three of which are commercial final product, deep drawing steels.

In hot strip mills, in the steel industry, reheat furnaces are utilized to reheat slabs, billets, blooms, etc., to uniform rolling temperatures prior to hot working. / During this reheating operation, hot steel surfaces react with the in-furnace oxidizing atmosphere resulting in the formation of an iron oxides layer (scale). The yield loss due to this phenomenon depends on furnace operating conditions, i.e. steel temperature, excess combustion air, steel residence time in the furnace, etc., and ranges between 1.5 to 3% of reheated steel. / In this research, the oxidation of a mild steel during reheating cycles as function of furnace atmosphere, steel residence time in the furnace and steel temperature was investigated. / It has been found that scale formation was most sensitive to oxygen levels in the furnace, steel residence time and temperature. A saving of up to 35% of steel lost to scale was achieved by reducing the excess air in the furnace from 70% to 20%. Also, longer residence time and higher temperatures of the steel in the furnace resulted in the formation of excessive amounts of scale.

ABSTRACT In the continuous casting process, the tundish not only serves as an intermediate buffer, but it also acts as a useful reactor for liquid steel refining. Modern tundishes are now designed to carry out different metallurgical operations, such as inclusion separation and flotation, alloy trimming, calcium doped inclusion modification, and thermal homogenization. To carry out such operations effectively, fluid flow inside a tundish plays an important role. It is now a proven fact that the insertion of different types of flow modifying devices can alter flow patterns within the tundish and thus affect the performance of the tundish significantly. Due to adverse operating conditions, direct experimental investigations are difficult to carry out. For that reason, physical and mathematical modeling is predominantly used to study tundish performance. Parameters like 'Residence Time Distribution' (RTD), tracer dispersion, velocity distribution, inclusion separation, etc. were mostly used to study and predict the performance of a tundish. Slag entrainment, though, is a vital problem during ladle changing that has been given less attention. In this research, it is intended to form a physical and mathematical modeling framework, to study and predict the performance of a 12 t, delta shaped, four strand, billet casting tundish. A full scale water model was studied both physically and mathematically. The phenomenon of slag entrainment occurring during a ladle changing operation was used as the key parameter to assess tundish performance. The amount of slag entering the 'Submerged Entry Nozzle' during a ladle change was measured to quantify the performance of different flow modifying arrangements. It is believed that the results of tests carried out under transient conditions can give a fairly good idea about tundish performance at steady state. To strengthen this belief, mathematical modeling of inclusion s / RÉSUMÉDans les procédés de coulées continu, les paniers répartiteurs ne servent pas seulement comme interface tampons, mais aussi comme un réacteur efficace pour raffiner l'acier. Les paniers répartiteurs modernes sont maintenant conçus pour effectuer différentes opérations métallurgiques comme la séparation des inclusions et leur flottation, l'ajustement fin de la nuance, le contrôle de la température de surfusion, la modification des inclusions par ajout de calcium et l'uniformisation de la température.Pour y arriver efficacement, l'écoulement du fluide dans le panier joue un rôle majeur. Il est maintenant prouvé que l'ajout de différents systèmes pour modifier les écoulements peut altérer les patrons d'écoulement dans le panier et en changer significativement la performance. Due aux conditions d'opérations difficiles, des investigations par expérimentation directe sont très difficiles. Pour cette raison la modélisation physique et mathématique est largement utilisée pour étudier la performance des paniers répartiteurs. Des paramètres comme la «Distribution des temps de résidence (RTD)», la dispersion d'éléments traceurs, le champ de vélocité, la séparation des inclusions, etc. sont largement utilisés pour étudier et prédire les performances d'un panier. L'entrainement du laitier, quoiqu'un problème vital lors du changement de creuset, a pourtant reçu moins d'attention. Dans cette recherche, le but est de développer une méthodologie de modélisation physique et mathématique pour étudier et prédire la performance d'un panier répartiteur de douze tonnes de forme triangulaire à quatre jets de coulée. Un modèle aquatique de grandeur nature fut étudié physiquement et mathématiquement. Le phénomène d'entrainement de laitier durant les opérations de changement de creuset fut utilise comme paramètre clé pour quantifier la performance du panier. La$

The hydrolysis of iron(III) from sulphate bearing aqueous solutions was investigated to significantly improve the neutralization of acidic mineral effluents. Both theoretical and experimental studies were undertaken. A literature review and a thermodynamic analysis of the Fe(III)-SO$ sb4$-H$ sb2$O system were performed. / The thermodynamic analysis determined that Fe(III)-sulphato complexes dominate in the acidic pH range (pH $<$ 4) while at neutral to mild alkaline pH range Fe(OH)$ sb3 sp circ sb{ rm (aq)}$ is the dominant species. It was found that a relationship exists between the precipitation pH and the sulphate content of the precipitate (17% at pH 3, 7-9% at pH 6 and 1-2% at pH 9) and it was attributed in part to the existence of the precursor Fe(III)-sulphato complexes and in part to SO$ sb4 sp{2-}$ adsorption on the Fe(III) oxyhydroxide solid phase. / Several variables were examined for their influence on the properties of the iron(III) hydrolysis products, such as: pH, neutralization rate, temperature, agitation rate, sulphate concentration and ionic strength. Tests were performed in a batch reactor and NaOH was selected as the most effective neutralizing reagent. Treatment sludges were characterized physically (e.g., solids content, settling rate, etc.), chemically and morphologically, while the treated effluent was examined for residual iron(III) content. These tests proved only partially successful as the precipitates produced were highly amorphous with low solids content (${<8 %}$) and poor settling (${<3}$ m/hr). However, exceptionally good precipitates were produced when the hydrolysis tests were performed under supersaturation control. This work led to the development of a novel approach to neutralization of effluents involving sludge recycling and staged neutralization in order to maintain a low and controlled supersaturation level. Via this novel method partially crystallized precipitates of excellent properties were produced at 50$ sp circ$C after eight recycles. (Abstract shortened by UMI.)

A quantitative texture analysis has been done on silicon steel specimens having different precipitate distribution in order to understand the texture development and the role of precipitates during secondary recrystallization. / To improve the size and distribution of precipitates, several annealing treatments were done on specimens obtained at different stages in the production of silicon steel. Tests for magnetic properties such as core loss and permeability, were done on final product specimens which had different precipitate distribution before secondary recrystallization. It is proposed that good precipitate distribution in the sheet before secondary recrystallization leads to a higher quality final product. / Grain size distribution and orientation distribution function (ODF) were obtained for the primary recrystallized specimens and for the annealed specimens, just before and just after, secondary recrystallization. From the ODF data, the probability of coincidence site lattice (CSL) boundaries were calculated up to $ Sigma$27b. The probability of CSL boundaries in specimens having good precipitate distribution was different from the probability observed in specimens having poor precipitate distribution. / An etch-pit method was used to give a more detailed, quantitative analysis of the texture development. An image analyzer was used to digitize the etch pit shapes, as well as the shape of grains. A computer program was developed to calculate the grain size of different texture components, the frequency of CSL boundaries between all grains, and the frequency of CSL boundaries between Goss grains and neighbouring grains. (Abstract shortened by UMI.)