Desulfurization using CaO-Al2O3 based flux in secondary refining processes: a critical review

Efficient desulfurization is one of the key objectives in secondary steelmaking processes in the scrap-based electric arc furnace (EAF) route. Desulfurization is primarily governed by slag-steel interactions, which depend strongly on both thermodynamics and kinetics. This study reviews recent research on CaO-Al2O3 based refining slags used in secondary refining processes to meet sulfur removal targets. Thermodynamic principles of desulfurization, including the slag-steel reaction mechanisms, sulfide capacity, and sulfur partition ratio, are first introduced. The effects of steel composition and slag chemistry on desulfurization are systematically summarized. Active oxygen in steel plays a crucial role in desulfurization, which is controlled by the Al content in bulk steel and by Fe-O equilibria at the slag-metal interface. Even a small amount of FeO (e.g., 1 wt.%) can generate approximately 30 ppm O in steel at the interface. Therefore, complete deoxidation is essential if deep desulfurization (<20 ppm S) is required. Optimizing slag composition, particularly ensuring sufficient CaO to maintain unit activity in the liquid phase, along with a low melting temperature and good fluidity, is beneficial for achieving efficient sulfur removal. In addition, the kinetics of desulfurization, with emphasis on diffusion coefficients and mass-transfer behavior in molten steel and slag, as well as process simulation methods, are reviewed. CFD coupled with thermodynamic modeling provides one of the most reliable approaches. Finally, industrial practices for secondary refining operations are discussed as well. Vacuum degassing units can reduce sulfur levels to approximately 10 ppm due to the intense stirring at the slag-metal interface. One of the major challenges that remains is measuring high-temperature physicochemical properties of slag and improving the accuracy of existing slag properties prediction models, artificial intelligence and big data based predictive models shows great potential for estimating physicochemical properties of slag, thereby enabling deeper understanding of desulfurization in the future. A unified desulfurization-related evaluation index that incorporates both steel composition and slag characteristics is still lacking and warrants further investigation.