This thesis presents the results of an investigation on the impact of continuous circuit and solution parameters on coprecipitation of arsenic with ferric iron from acidic sulphate solution. The techniques employed were selected, or developed, to emulate industrial practice. The concept behind the work was to better understand the link between the process of precipitation and the stability of arsenic in the resultant coprecipitates. This was performed by examining the role of circuit design and co-ions on both arsenic removal during coprecipitation and arsenic retention during ageing. The parameters/factors investigated included continuous versus batch operation, number of stages (pH profile), recycling and Ca2+, Ni2+ and Al3+ as co-ions. Arsenic removal was found to be greatly improved by continuous rather than batch coprecipitation. In addition, the presence of calcium (introduced as slaked lime) was found to be instrumental in the removal and retention of arsenic. Arsenic retention during ageing (up to 300 days) at various temperatures (3, 22, 40 and 70°C) was observed to reach an "equilibrium" that was strongly affected by the continuous circuit design, as well as the co-ions present during coprecipitation. Evidence is presented of the partitioning of arsenic within the coprecipitates in two principal phases, namely ferric arsenate (FeAsO4•xH2O) and arsenic adsorbed ferrihydrite. Continuous circuit design parameters, such as staging, that resulted in enhanced stability appear to yield coprecipitates with higher ferric arsenate content. Analysis of the kinetic and "equilibrium" arsenic retention data yielded activation energy (~ 60 kJ/mol) and enthalpy (~ -38.5 kJ/mol) values that suggest a reaction controlled exothermic dissolution mechanism. / Ce mémoire de maîtrise présente les résultats d'une étude portée sur la co-précipitation de l'arsenic et des ions ferriques en solution aqueuse sulfatée. L'utilisation d'un procédé continu et les paramètres de la solution ont été considérés comme cibles d'étude. Les techniques employées ont été sélectionnées ou développées afin de simuler les méthodes opératoires industrielles. L'idée directrice de ce travail était d'aboutir à une meilleure compréhension sur les liens entre les mécanismes de précipitation et la stabilité de l'arsenic dans les co-précipités formés. Aussi, le rôle du design du procédé ainsi que celui des co-ions sur l'extraction, respectivement la stabilisation de l'arsenic ont fait l'objet d'un examen approfondi à court terme soit au stade de co-precipitation, respectivement à long terme durant le vieillissement de la solution. Les paramètres/facteurs étudiés incluaient le caractère continu du procédé (en comparaison avec un procédé discontinu), le nombre d'étapes (le profile pH), le recyclage des produits solides et la nature (Ca2+, Ni2+ et Al3+) des co-ions introduits dans le système. Cette étude a montré que l'utilisation d'un procédé continu (en comparaison avec un procédé discontinu) améliorait considérablement l'extraction de l'arsenic à partir de solutions aqueuses sulfatées. Aussi, la présence de calcium (introduit sous forme de chaux hydratée) s'est avérée profitable à l'extraction de l'arsenic à court et à long terme. Il a été observé que la stabilisation de l'arsenic dans le cas des expériences de vieillissement (d'une durée de 300 jours) performées à différentes températures (3, 22, 40 et 70°C) atteignait un état d'équilibre fortement influencé par le design du procédé continu ainsi que par les co-ions présents lors de la co-précipitation. Il a été mis en évidence que lors de la phase de co-précipitation, l'arsenic était réparti sous forme de de

Strain and temperature gradients are developed through the cross-section during the hot rolling of steel rods. These gradients are of particular interest because of their potential effects on the homogeneity of the resulting microstructure during rolling as well as in the final product. Strain and temperature can strongly influence recrystallization and transformation behaviour. / A three-dimensional, temperature-dependent viscoplastic finite element analysis was used to predict the strain distribution developed for the hot rolling of steel square billets using the software package ABAQUS/EXPLICIT$ sp{ rm TM}.$ Constitutive parameters were determined from previous experimental data generated at McGill University and at the Metals Technology Laboratory (MTL), CANMET. / McGill-CANMET's Property Prediction Model was used to determine the microstructure during rolling and the final properties at room temperature. / Results from the mechanical and microstructural models were validated with experimental data from literature and from rolling tests performed at MTL-CANMET. / With these two models, several different hot rolling scenarios were generated to examine how the two models would predict their results by varying a single parameter.

The formation of iron-bearing intermetallics in the 413 type of aluminum alloys was investigated comprehensively. Both synthetic and commercial 413 alloys were studied with iron concentrations in the range of 0.4-1.2 wt. % and manganese up to 0.5 wt.%. The effects of cooling rate during solidification and of melt chemistry on the morphology of iron intermetallic phases were determined. Image analysis was used to quantify the intermetallic size, volume fraction, and number, as a function of both melt chemistry and cooling rate. The total volume fraction of intermetallic compounds in these alloys was related to cooling rate by an exponential equation. / The kinetics of both dissolution of intermetallics on melting, and of re-formation on cooling of the liquid were investigated by means of quenching experiments. Quantitative evaluation of intermetallic size and number revealed that the change in volume fraction of intermetallics in the liquid state is controlled by nucleation. / The effect of settling time and the rate of gravity segregation of intermetallic compounds in a stagnant liquid metal were investigated. It was found that, in the absence of convection, settling obeys Stokes' law with the terminal velocity reached at very short times and very close to the melt surface. / Strontium was used to modify or eliminate the iron-intermetallics. (Abstract shortened by UMI.)

A comprehensive characterization of silver and bismuth refining crusts produced during the pyrometallurgical refining of lead was undertaken. Parameters such as intermetallic crystal size and composition, and degree of metal entrainment were determined. Silver crust was found to be composed mostly of the $ epsilon$-phase (AgZn$ sb3$) with an average particle size of 50 $ mu$m. Liquid lead entrainment of 80 volume percent was measured. The intermetallic phase in bismuth crust was composed entirely of CaMg$ rm sb2Bi sb2$ crystals with an average diameter of 50 $ mu$m. Lead entrainment in bismuth crust was approximately 90 volume percent. The high degree of metal entrainment observed in both silver and bismuth crusts was not adequately explained through a capillary model of liquid metal entrainment in a static bed of solid particles. The models used to predict swelling in colloidal particle systems (such as clay soils) were found to better approximate the high degree of liquid entrainment. / Experiments were performed in which particle beds of $ gamma$-phase ($ rm Ag sb5Zn sb8$) intermetallic compound were infiltrated with liquid lead under controlled conditions and examined under the optical microscope. Image analysis revealed pronounced swelling of fine particle beds relative to coarse. The effect was most noticeable for average particle diameters less than 100 $ mu$m where the solids volume fraction decreased from 50 to 30 vol.%. Data showed samples LO exhibit a constant increase in inter-particle spacing of approximately 50 $ mu$m. The increase in spacing explained the degree of swelling at low particle diameters and was also used to develop an expression to predict height of capillary rise. / Based on an osmotic pressure analogy in colloidal particle systems, a swelling pressure term was defined to explain the increase in inter-particle spacing. The swelling pressure was defined as equal and opposite to the capillary pressure tending to force particles to random packing. It was concluded that swelling was related to mass transfer at the solid-liquid interface of the intermetallic crystals giving a reduced static pressure due to a localized minimization of Gibbs energy. The static pressure change (swelling pressure) was attributed to overlap of mass transfer boundary layers where mid-point static liquid metal pressure is less than the pressure in the bulk liquid. For the system studied, the mass transfer boundary layer was estimated to be approximately 100 $ mu$m in thickness. / Further experiments, using the sessile drop technique, were used to support the study findings by showing rapid spreading of liquid lead on Ag-Zn intermetallic substrates. The rapid spreading was attributed to mass transfer at the solid-liquid interface giving a reduced interfacial energy, $ gamma sb{ rm s1}$ due to a localized reduction in Gibbs energy. / The analysis of experimental results led to the development of two new industrial processes in the lead industry aimed at reducing lead loss to silver and bismuth crusts. Tests employing continuous agitation during bismuth removal from lead promoted the growth of large $ rm CaMg sb2Bi sb2$ crystals. The larger crystals minimized the effect of mass transfer boundary layer overlap and reduced lead entrainment from 90 to 80 volume percent. The addition of a low melting point KCl-ZnCl$ sb2$ salt during silver crust formation was found to remove up to 90% of entrained lead by the salt preferentially wetting Ag-Zn intermetallic particles. This was found to eliminate surface forces which retained liquid in the system of solid particles.

Microflotation, laboratory batch, and continuous in-plant minicell tests were conducted to determine if copper ions not only activate sphalerite but also depress pyrite when the minerals are together. / Microflotation was performed on mineral samples over a pH range of 3.5--11. It was found that pyrite recovery increased with addition of copper ions when alone, but decreased by more than 25% (absolute) when sphalerite was also present for the pH range 6--9.5. / Bench flotation tests were performed on Louvicourt Mine ore and Red Dog Mine ore. The results for Red Dog showed that upon addition of copper, pyrite recovery decreased significantly. Pyrite recovery decreased from ca. 70% (no copper) to ca. 10% (4.5 g Cu/kg Zn) at a sphalerite recovery of ca. 70%. / Continuous in-plant minicell tests at Louvicourt Mine also showed that pyrite recovery decreased with addition of copper and further indicated that with excess copper (20 g Cu/kg Zn) pyrite recovery can increase. / The mechanism responsible for the depression of pyrite seems to be competition first for copper then xanthate that favours sphalerite over pyrite. To take advantage of this competition it may be necessary to control the sequence and rate of copper, collector, and lime addition.

The reaction of HCl gas with MgO particles suspended in a molten salt containing magnesium chloride was studied. Chlorination experiments were carried out in a batch stirred tank reactor. The determination of the MgCl2 concentration as a function of chlorination time indicated that the chlorination of MgO to MgCl2 occured through an intermediate step. The intermediate species was assumed to be MgOHCl forming according to the reaction: MgOs+HCl g→MgOHCl d / Alkalimetric titration and thermal decomposition by injection of dry nitrogen were used to confirm the existence of MgOHCl in the molten salt during chlorination. Infrared spectrometry was also tentatively used to identify MgOHCl but, without success due to the hygroscopic nature of the material. / Conventional mass transfer theory was used to develop a model for the reaction of solid MgO particles with HCl dissolved in the molten salt to form MgOHCl. The model assumed shrinking particle behaviour for the MgO and was confirmed by the data from the chlorination experiments. / Moreover, experiments conducted at different HCl flow rates showed that the reaction was controlled by mass transfer of HCl dissolved in the molten salt across the liquid film surrounding the MgO particle to the surface of the particle. The following rate law was obtained CMgOCi MgO1/3=1-K nt the constant Kn was found to vary between 0.04 and 0.08 s-1 when the gas flow rate was varied from 2.5 to 7.5 SLPM at constant temperature and agitation speed of 500°C and 500 RPM, respectively.

Activation of sphalerite by lead ions in the presence of ethyl xanthate was investigated by microflotation, zeta potential measurements, scanning electron microscopy (SEM), X-ray photoelectron spectroscopy (XPS) and extraction by ethylenediaminetetraacetate (EDTA). The source of lead, either from solution or contact with galena particles, proved immaterial. From EDTA extraction data flotation response was determined as a function of surface concentration of Pb ([Pb]surf, mg/cm2). A critical [Pb]surf ≃ 0.5 x 10-4 mg/cm2 causing activation was estimated. By combining with a Pb ion production model from the literature, the possibility of Pb activation for an ore of given Pb grade was explored. Ores with as little as 0.1% Pb are candidates for accidental activation. / A mechanism of lead interaction is proposed: at weakly acidic to mildly alkaline conditions Pb2+ and PbOH+ adsorb on sphalerite and react with xanthate to form PbX2 and Pb(OH)X; at high pH, hydrophilic Pb(OH)2(s) precipitates depress flotation. / Various candidate deactivators were compared using microflotation. The reagents were diethylenetriamine (DETA), sodium bicarbonate (NaHCO3), silica sol (SS), sodium phosphate (Na3PO4·12H 2O) and sodium polyphosphate (NaPO3)n. The latter had the strongest effect and was selected for detailed study. The mechanism of polyphosphate action was investigated by SEM and XPS. The results show that polyphosphate acted to remove Pb ("clean") from the sphalerite by forming soluble complexes. / The polyphosphate (PP) to lead (Pb) ratio in the complexes was determined from conductometric titration of lead nitrate vs. polyphosphate. It was found that initially a precipitate formed with PP/Pb of 1/3. The precipitate dissolved in excess polyphosphate, resulting in a variety of complexes with PP/Pb of 1/2, 1/1, 3/2, 2/1, and 3/1. Knowing the amount of Pb to be removed and taking the lowest PP/Pb ratio, 1/2, the quantity of polyphosphate required to solubilize the Pb can be calculated.

The susceptibility of steels to transverse cracking during continuous casting depends, in part, on the hot ductility of the steel. In the past, hot ductility has been measured in the laboratory by tensile testing specimens reheated to a high temperature (preferably melted), and then cooled down to the desired test temperature before isothermally testing to fracture. More recently, isothermal tensile testing has been performed after imposing different thermal histories prior to reaching the test temperature. The most relevant thermal history is that of the solidifying strand surface. Such thermal histories have been found to be quite complex, and invariably involve rapid cooling and heating cycles. This will therefore lead to high thermal gradients, which, in turn, can generate strains in the surface of the solidifying strand. Thus, the purpose of this thesis was to consider the effect of the thermomechanical history on the hot ductility of steel. / A Nb-Ti microalloyed steel was chosen to be studied, since microalloying often increases susceptibility to transverse cracking. After in-situ melting and solidification, tensile test specimens were subjected to the thermal history typical of a continuously cast billet. Different amounts of prior deformations in various schedules were imposed on the specimens at selected stages during the thermal history, before tensile testing to fracture at the time and temperature corresponding to the unbending stage of the billet casting. It was found that the hot ductility varied from 1% to 98%, depending on the stage in the thermal history at which prior deformation was executed. Microstructural studies showed remarkable changes in the precipitation pattern occurred due to the employed prior deformation. Similar hot ductility measurements executed on a low carbon steel, that was free from microalloy elements, exhibited an effect of prior hot deformation which was quite different to the Nb-Ti microalloyed steel. In particular, the effect on the hot ductility was not as dramatic. The effects were also postulated to be due to precipitation.

In this work, the kinetic and equilibrium profiles of each individual reaction step involved in the production of hematite at 200°C via oxydrolysis of ferrous sulphate in concentrated zinc sulphate media were established. Crystallization of ferrous sulphate was found to play a crucial role in the overall process due to its relatively low solubility and fast crystallization kinetics at elevated temperatures. In fact, the overall kinetics of the oxydrolysis process were found to be limited by the re-dissolution of ferrous sulphate. Pre-crystallization of ferrous sulphate prior to oxidation was found to result in enhanced overall kinetics, cutting down the required retention time from 3 to 2 hours. Enhanced kinetics were also achieved by performing oxydrolysis in two-temperature stages: a low temperature (T = 180°C) first stage with retention time >20 min and an elevated temperature (T = 200°C) second stage with retention time >100 min. / The typical composition of the hematite material produced in this work was 64.3% Fe, 1.3% S (as SO4), 0.6% Zn, and 4.6% H2O. The sulphur content was found to be predominantly (0.6--0.8%) due to formation of sodium jarosite with the remaining (0.3--0.5%) attributed to SO4 chemisorption, and (to less extent) basic ferric sulphate formation. In the absence of zinc sulphate, the majority of sulphate contamination was due to basic ferric sulphate formation. Hematite was found to form via a predominantly homogeneous nucleation mechanism with sub-micron crystallites clustered together as aggregates of 5--10 mum size and around 7 m 2/g specific surface area. In contrast, hematite produced by direct hydrolysis of ferric sulphate possessed one order of magnitude higher specific surface area. As for the industrial hematite product, its composition was found to be 52.6% Fe, 4.6% S, 1.0% Zn, and 8.8% H2O. The sulphur contamination in the industrial product was mainly due to co-precipitation of jarosite and basic ferric sulphate compounds. Hydrothermal trans-formation of the industrial hematite product at elevated temperatures (>200°C) and retention time of >60 min with solids loading as high as 16 wt.% proved to be effective in reducing the sulphur content to less than 1%.

Through the application of solution thermodynamic principles, brazing alloys based on the Ni/Cr ratio of AWS BNi-5 (Ni-18Cr-19Si (atom%); Ni/Cr = 3.46) were designed for brazing SiC to itself and Nb. / The optimum alloy composition for SiC-SiC joints was computed to be Ni-14.5Cr-35Si (atom%). Brazing experiments were conducted to assess the effect of changing the Si content of the brazing alloy away from the calculated optimum on the joint microstructures and strengths. For alloys containing less than 37 atom% Si excessive reaction was observed, resulting in the formation of a porous reaction zone at the brazing alloy/SiC interface and poor joint strengths, the thickness of the porous zone decreasing and the joint strength increasing as the brazing alloy composition approached 37 atom% Si. The most favourable joint microstructures and highest joint strengths ($ approx$100MPa) were observed for the 37 atom% Si brazing alloy, in excellent agreement with the thermodynamic model. For alloys containing greater than 37 atom% Si, the microstructures exhibited localized debonding of the brazing alloys from the SiC and poorer joint strengths, also in good agreement with the thermodynamic model. / For the SiC-Nb brazing study, the thermodynamic modelling procedure determined that a series of Ni-Cr-Si-Nb alloys were suitable for forming bonds. Brazing experiments were performed using two of the "optimum" brazing alloy compositions (i.e. Ni-11.6Cr-29Si-19Nb and Ni-11.2Cr-20Si-30Nb (atom%)) and two alloys of lower Si and Nb contents in order to assess the predictions of the thermodynamic model through evaluation of the joint microstructures. All of the joints exhibited porous reaction zones, with the thickness of the reaction zone decreasing as the Si and Nb content of the brazing alloys approached the computed optima. However, due to cracking and solidification porosity in the brazing alloy, these joints were found to be unsuitable for engineering applications. / It was concluded that the solution thermodynamic method developed in this thesis was very useful for selecting Ni-based (Ni/Cr = 3.46) brazing alloy compositions suitable for forming SiC-SiC-Nb bonds.