Non-Metallic Inclusion Changes in Si-Mn Killed Steels Papadopoli ToneStephano 2018 Steel must be deoxidized during refining to avoid the formation of gas porosity during solidification, and many alloys. In the case of Si-Mn killed steels, Si and Mn are added to form low melting point inclusions that can prevent nozzle clogging and that are vitreous upon solidification, so they can deform upon mechanical processing. Depending on the activity of SiO2 of the slags used during secondary refining, the steel-slag reactions can further deoxidize the steel and reduce oxides from the slag and refractories such as MgO, Al2O3 and CaO. These reduced elements can increase the melting point and hardness of the inclusions. Some of these impurities can enter the steelmaking process from the ferroalloys themselves because of their manufacturing process.<br>In this work, the process variables that influence the thermodynamics and kinetics of the non-metallic inclusion formation, transformation and removal are reviewed. It is found that the gas stirring rate is found to have a significant impact on the kinetics of steel-slag reactions and inclusion generation and removal. It is also found that basic slags will cause pick-up of Al, Mg and Ca and promote the formation of undesired inclusions in the steel melt. During solidification, inclusions can precipitate at different chemical compositions depending on the remaining dissolved oxygen content in the melt. Ferroalloys used for deoxidizing the steel such as FeSi typically contains Ca that will impact steel cleanliness and can be used to the advantage of the process, i.e. the treatment of alumina inclusions.<br>A one-parameter kinetic model for steel-slag-inclusion reactions is developed using FactSage. The potential effect of three different slag chemistries on the trajectory of inclusion modification are tested using the model for 1.25%Mn-0.25%Si steel. It is seen that Al, Mg and Ca pick-up are expected for all tested systems. The SiO2 activity, directly influenced by slag binary basicity, controls the reduction of the oxides. For high basicity, alumina-containing slags, the inclusions will transform from liquid Mn-silicates to alumina to spinels to Ca, Mg-aluminates to liquid slag. For low basicity, low alumina slags, the inclusions will remain liquid. For CaO-SiO2-MgO-CaF2 slags, Mn-silicates are predicted to become 2CaO.SiO2 and MgO.<br>Inclusions in lollipop samples of degassed Si-killed steels with CaO-SiO2-MgO-CaF2 slags are analyzed using ASPEX automated feature analysis and inclusion extraction. The inclusions found to be MgO and CaO-rich after degassing. After degassing, CaO-rich inclusions are found both containing and not containing fluorine, indicating Ca pick-up from slag and slag emulsification. After trimming, primary inclusions are mostly Ca-silicates and do not contain fluorine - their origin is discussed. It is observed that Mg content in steel increases during a heat while Ca content decreases.<br>Induction furnace experiments are done with MgO and CaO crucibles for different slag compositions. The one-parameter kinetic model predicts the inclusion composition over time. It is found that more basic slags in the CaO-Al2O3-MgO-SiO2 will drive more Al and Mg pick-up than less basic slags. The addition of 5%MnO and 10ÊF2 to CaO-Al2O3-MgO-SiO2 slags did not significantly change the steel-slag reaction kinetics and non-metallic inclusion composition. It is found that CaO-SiO2-MgO-CaF2 are strongly desulfurizing and can modify Mn-silicate inclusions to forsterite and MgO. There is no observable CaO precipitation when the melt is reoxidized after equilibration. One experiment with CaO-SiO2 -CaF2 slags in a CaO-3%ZrO2 crucible yielded CaO-containing inclusions, with estimated [Ca] before reoxidation of 0.3 ppm. It is found that Ca solubility in the steel is significantly lower than predicted by FactSage. Experiments using FeSi containing 2Ê to deoxidize the steel and treat Al-killed and Si-killed steels showed that the Ca yield is higher than the typically observed for CaSi2 powder addition in the same experimental setup. When [O] is higher (for Si-killed steels), the Ca yield is higher, and inclusions reach equilibrium faster than for Al-killed steels which have very low [O]. The inclusion-inclusion reactions in Ca-treated Al-killed steel took several minutes more than expected from kinetic models. It was found that the mass transfer coefficient for the configuration is controlled by convection in the melt, and there should not be significant differences when using 600g or 1000g melts for the experiments.<br>Confocal Laser Scanning Microscopy was used to observe the reoxidation behavior of a steel sample taken from a Si-killed degassed heat, where [Ca] and [Mg] was expected to be high. The steel sample was oxidized using ungettered argon. The gas phase mass transfer coefficient for a CSLM sample was measured by evaporation of Mn. The oxidation experiments could precipitate Mg-rich inclusions but there was no observable increase in the CaO content of these inclusions. <br>