Low phosphorus content steels are essential for steel applications where high ductility is required, such as thin sheets, deep drawn, pipelines and automobile exteriors. In the past, phosphorus control was not considered a big challenge in steel production in the US because iron ores with low phosphorus contents were readily available and considered cheap. However, in the last decade, the iron ore price has risen by roughly 400% and lower cost iron sources generally have higher phosphorus content. In integrated steel plants, phosphorus removal usually takes place during the oxygen steelmaking process (OSM) but in Japan a intermediate step for hot metal dephosphorization is commonly used. There are various types of OSM furnaces but the most widely used remains the top-blown Basic Oxygen Furnace (BOF). The BOF slag can be recycled to a sinter plant or directly to the blast furnace, ultimately increasing the phosphorus input in the process. In order to meet new demands for phosphorus control, it is necessary to improve our understanding on the thermodynamics and kinetics of the phosphorus partitioning reaction between slag and metal melts during steelmaking. Therefore, the present work has been divided in three strongly correlated sections: phosphorus equilibrium between metal and slag; analysis of plant data; and observations of the reaction kinetics. Phosphorus equilibrium between liquid metal and slag has been extensively studied since the 1940's. It is well known that CaO and FeO are the main slag constituents that help promote dephosphorization. On the other hand, dephosphorization decreases with temperature due to the endothermic nature of the reaction. Many correlations have been developed to predict the phosphorus partition ratio as a function of metal and slag composition as well as temperature. Nevertheless, there are still disagreements in the laboratory data and the equilibrium phosphorus partition can be predicted with an uncertainty of a factor of up to 5. The first part of the present work focuses on generating more reliable equilibrium data for BOF-type slags by approaching equilibrium from both sides of the reaction. The experimental results were combined with two other sets of data from different authors to produce a new correlation that includes the effect of SiO2 on the phosphorus partition coefficient, LP . Although the quantification of phosphorus equilibrium is extremely important, most industrial furnaces do not operate at equilibrium, usually due to liquid slag formation, kinetics and time constraints. Thus, it is important to know how close to equilibrium different furnaces operate in order to suggest optimal slag compositions to promote dephosphorization. The present work analyzed four large sets of data containing the chemical compositions of both slag and metal phase as well as the tapping temperature of each heat. Each set of data corresponded to different furnaces: one AOD (Argon Oxygen Decarburization), two top-blown BOFs and one Q-BOP or OBM. It was found that the bulk slag composition can greatly \mask" the data due to solid phases coexisting with the liquid slag. The author used the software package FactSage to estimate the amount of solids in the slag and liquid slag composition. It was found that the AOD is the reactor closest to equilibrium, followed by the Q-BOP (OBM) and the two top-blown BOFs. It was noted that the stirring conditions and slag composition are two key variables to enable optimum phosphorus removal. Also, over saturating the slag with CaO and MgO does not seem to benefit the process to any extent. Lastly, interesting observations on the behavior of small metal droplets reacting with slag are presented and discussed. It was found that dynamic interfacial phenomena at the metal-slag interface is likely to play a significant role in the kinetic behavior of the system, due to the exchange of surface active elements, such as oxygen, which dramatically lowers the interfacial tension and cause spontaneous emulsification. Although this phenomenon has been studied, actual quantification of changes in interfacial area remain a challenge. The author developed an experimental method to enable better quantification of spontaneous emulsification and two sets of experiments were carried. One with an Fe containing 0.2 wt.% P and another in a P-free system where pure iron was oxidized. It was found that phosphorus did not play a role in spontaneous emulsification and it was rapidly removed before the onset of dynamic interfacial phenomena. Emulsificaion was maybe caused by de-oxidation of the metal after phosphorus removal took place and the metal became super saturated with oxygen by an unknown reason. The estimated surface area rapidly increases by over an order of magnitude during the beginning and intermediate periods of the reaction. The metal drop breaks into hundreds of small droplets, effectively emulsifying the metal into the slag. With time, the surface area decreases and the metal droplets coalesce. Similar results were observed for an Fe droplet being oxidized. Spontaneous emulsification takes place regardless of the direction of oxygen transfer and the changes in surface area are similar for both cases. The last chapter describes the industrial relevance of the present work, summarizes the findings, revisits the hypotheses and presents potential future work where further research is encouraged.

A literary study has been made to find parameters affecting dephosphorization of stainless steel. Ways to lower phosphorus content without major loss of chromium in order to decrease the production costs of Sandvik AB’s stainless steel. The study was conducted with respect to oxidizing dephosphorization and parameters affecting dephosphorization are carbon and chromium content, temperature and slag properties. It was revealed that higher carbon content and basicity is beneficial to the dephosphorization process. However, the choice of temperature, chromium content and flux were dependant on which way dephosphorization was approached. A method of refining chromium in slag using an electric arc furnace (EAF) was also discovered. This made it possible to extract 97% of all chromium in slag. The conclusions are that in order for Sandvik to successfully remove phosphorus, scrap metal with higher carbon content should be used, together with a basic CaO slag with constituents of e g Li2O3 and CaF2. A deslagging step can be added to the argon oxygen decarburizer (AOD) process in order to remove the phosphorus bound to the slag, before decarburization.

Purifying silicon to a certain degree is needed in the photovoltaic industry. In the metallurgical route, phosphorus is removed from molten silicon in an induction vacuum refining furnace.This study presents the simulation of the silicon dephosphorization process that takes into account a neglected factor thus far, namely the residual oxygen amount.A thermodynamic assessment is performed for the Si - P system, and the phosphorus solubility in silicon is determined. Resulting activity coefficients are incorporated in an algorithm created to simulate this phenomenon. By also using other literature data, this program predicts the distillation time of a n-type silicon sample as well as its weight loss as a function of the oxygen pressure in the enclosure.The experimental validation of those results is intended with an electromagnetic levitation device. It enables to melt a silicon sample without any contaminant crucible and under a controlled atmosphere. / La purification du silicium à un certain degré est requis pour l'industrie photovoltaïque. Par voie métallurgique, le phosphore est éliminé du silicium fondu dans un four à induction sous vide.Cette étude présente la simulation du processus de déphosphoration du silicium en prenant en compte un facteur jusqu'alors négligé, à savoir la présence d'oxygène résiduel.Un travail d'optimisation thermodynamique du système Si - P est effectué, où la solubilité du phosphore dans le silicium est notamment déterminée. Les coefficients d'activité qui découlent de celle-ci sont incorporés dans un algorithme créé pour simuler ce phénomène. En s'appuyant également sur des données issues de la littérature, ce programme prédit le temps de distillation d'un échantillon de silicium de type n ainsi que sa perte de masse en fonction de la pression d'oxygène dans l'enceinte.La validation expérimentale de ces résultats est envisagée en utilisant un dispositif de lévitation électromagnétique. Il permet de faire fondre un échantillon de silicium sans creuset contaminant et avec une atmosphère contrôlée.

[pt] Um modelo de descarburação e formação de escórias foi desenvolvido e aplicado ao processo de fabricação de aço em forno elétrico a arco de 120 ton de capacidade, com carga de ferro gusa e de sucata ferrosa. O carregamento de carbono foi significativamente variado para testar a consistência do modelo, considerando a cinética de oxidação do carbono, oxidação do fósforo e de redução do óxido de ferro. Gusa e coque foram empregados como fontes mais relevantes de carbono, resultando na entrada de 15 a 35 kg carbono/ton. As taxas de fusão do gusa e da sucata governam a disponibilidade dos elementos mais relevantes tais como carbono, fósforo e silício em solução, portanto, afetam as taxas de descarburação e de formação de escórias. A principal fonte de fósforo na carga ferrosa é o gusa. Desta forma, a evolução do teor de fósforo na fase metal mostrou-se importante para as estimativas das taxas de fusão do gusa, uma vez que o fósforo pode ser empregado como traçador adicional ao carbono. Modelos cinéticos envolvendo as reações do fósforo e silício operam simultaneamente com os modelos cinéticos referentes às reações do carbono e do ferro. Integrações numéricas associadas a um algoritmo de gradientes reduzidos generalizado foi empregado para o sistema não linear com restrições, de forma a determinar a maioria dos parâmetros cinéticos do modelo. A taxa de fusão global da carga de sucata foi maior do que a taxa de fusão aparente do gusa. Supõe-se que, o gusa apesar de ter relações geométricas desfavoráveis à transferência de calor em relação à sucata, poderia fundir mais rapidamente influenciado pelo seu baixo ponto de fusão. Entretanto, devido à formação de camada solidificada a partir da massa líquida na qual é imerso, é provável que mesmo fundido posteriormente, ocorra um processo de encapsulamento temporário, conferindo-lhe uma taxa aparente de fusão mais baixa. A constante cinética da reação de descarburação quando o teor de carbono é inferior ao carbono crítico de 0.19 por cento em massa e pelo menos 60 por cento da carga ferrosa estão fundidas, foi estimada em 0.74 min-1, taxas comparáveis às obtidas em aciaria a oxigênio. A principal fonte de oxigênio para oxidação do ferro é disponibilizada por lanças supersônicas. Estima-se que 20 por cento do oxigênio injetado via lanças sejam consumidos para a formação de óxido de ferro. Entretanto, cerca de 31 por cento e 26 por cento do oxigênio oriundo de injetores de póscombustão podem contribuir na formação de óxido de ferro ou são captados pelo sistema de exaustão de gases, respectivamente. Os resultados indicam que em torno de 15-30 por cento do carbono injetado podem não reagir no forno, sendo removidos com a escória. Adicionalmente ao estado de não-equilíbrio no sistema Fe-C-O observado, a dispersão nas estimativas de carbono solúvel na fase metal também pode ter sido influenciada pela intensidade de penetração da injeção de coque. O algoritmo proposto se constitui num promissor simulador de práticas que visam otimizar o rendimento metálico do ferro, a partir da dependência da cinética de redução do óxido de ferro com sua atividade química na escória. / [en] A decarburization and slag formation model was developed and applied to a steelmaking process based on scrap and pig iron mixes melted in a conventional AC electric arc furnace (EAF) with 120 ton capacity. The amount of carbon input was varied significantly in order to evaluate the model consistency regarding mainly the kinetics of carbon oxidation, phosphorus oxidation and iron oxide formation and reduction. Pig iron and coke were used as sources of carbon, resulting in variation of total carbon input in the range of 15 to 35 kg carbon/ton. The pig iron and scrap melting rates determine the availability of the most relevant elements such as carbon, phosphorus and silicon in solution in Fe-C melts, and therefore, affecting the decarburization as well the slag formation rates. The pig iron is the main source of phosphorus in the ferrous charge. Hence, the evolution of the phosphorus content in the metal phase is important to predict the pig iron melting rate, since phosphorus can be used as a tracer element in addition to carbon. Kinetic models regarding phosphorus and silicon were applied simultaneously to kinetic models of carbon and iron reactions. A numerical integration method supported a generalized reduced gradient algorithm for non-linear and constrained system (GRG) was applied to determine most of the kinetic model parameters. The scrap melting rates were found to be higher than pig iron apparent melting rates. This is expected that, even though the heat transfer issues related to significant differences in the area to volume ratio compared to scrap, pig iron may melt faster influenced by its low melting point. However, a solidified shell maybe created from the hot heel where pig iron is immersed, even when further melting occur, Fe-rich carbon melts could be encapsulated temporarily and present lower apparent melting rate. The decarburization rate parameter, when at least 60 percent of the charge is melted, was estimated as 0.74 min-1, when carbon content is lower than the critical carbon 0.19 percent wt, which is similar to the rate range observed in oxygen steelmaking facilities. Around 31 percent and 26 percent of the oxygen input through post combustion injectors were addressed to iron oxidation and to the off-gas system, respectively. The main source of oxygen taking part of iron oxidation is available from supersonic lances. Approximately 20 percent of the oxygen input through lancing are consumed to form iron oxide. The results also indicate about 15-30 percent of the injected carbon may not react and leave EAF during slagoff. In addition to the observed non-equilibrium state in Fe-C-O system, the dispersive behavior of the prediction of soluble carbon content in the metal phase could also be influenced by the intensity of penetration coke. The model framework is a promising tool to work preliminarily in what-if process scenario builder as a static model for iron yield optimization, regarding the kinetics of iron oxide reduction reaction and the proposed dependence on its chemical activity in the slag phase.