710 influence of temperature, residence time, and gas flow rate on reduction kinetics, phase evolution, and the formation of recoverable iron- and titanium-bearing phases. After reduction, the phases present were identified by X-ray diffraction (XRD). The experimentally observed phase assemblages and recoveries were compared against the model predictions. This validation step was crucial to confirm that the thermodynamic models captured the essential chemistry of the real system, and to calibrate any differences (for instance, due to kinetic effects or minor components not in the model). XRF analyses of the BR sample were utilised for setting up the calculations. Major oxide components (Al₂O₃, SiO₂, Fe₂O₃/FeO, TiO₂, CaO, Na₂O, etc.). The pseudosection has been calculated using Perple_X_7.1.13 (Connolly, 2009) in the system NCFMASHTO. We used the internally consistent thermodynamic database of Holland and Powell (1998, revised version of 2002) and the following solid solution models: Ppv(stx8), O(stx8), FeSi(BCC), Opx(stx8), Neph(FB), Cpx(stx8), Pl(stx8), Kf, Gt(stx8), Crd(HGP), Sapp(HP), Sp(stx8), Mt(W), Ilm(WPH), Wus, Fper(H), Cor(H), ZrRu. The CORK equation of state for H2O-CO2 fluids (Holland & Powell, 1998) was used, although fluids have been considered pure H2O. The pressure was fixed at 1 bar for simulating atmospheric and furnace conditions (a reasonable approximation for open reactors). Oxygen fugacity (Log fO₂) was adjusted to represent different redox conditions (-29 to -5). PyWerami, a visualization and analysis tool for the Perple_X software package, was used to extract and display the stability fields and modal abundances of individual mineral phases from the calculated thermodynamic datasets. The thermodynamic modelling was performed using a publicly available dataset originally developed for metamorphic and geological systems. While this dataset provides a robust framework for evaluating phase stability trends, it is not specifically optimized for highly alkaline industrial environments, hydrogen-rich atmospheres, or synthetic phases commonly encountered in cementitious and metallurgical systems, such as larnite or gehlenite. Therefore, the results are interpreted as indicative stability trends rather than absolute quantitative predictions. The use of this dataset was selected based on accessibility and reproducibility considerations, as comprehensive industrial thermodynamic databases (e.g., full FactSage cement or metallurgical databases) require specialized licenses and were beyond the scope of this study. 3. RESULTS The primary feedstock in this study was bauxite residue from Alumina d.o.o. (Zvornik, BosniaHerzegovina). The chemical composition was determined by ICP-OS, and the mineral composition by XRD (Table 1). The materials are dry solids, finely grained (<100 µm), making them suitable for thermal processing and subsequent leaching. The BR sample displays a (normalized) composition of ~50% FeO, 10-20% Al₂O₃, 3% TiO₂, along with significant amounts of Na₂O, CaO, SiO₂ and minor rare metals (Table 1). X-ray diffraction (XRD) analysis of the starting material revealed a complex and heterogeneous mineralogical composition typical of bauxite residue (Table 1). The preliminary study identified predominant crystalline phases included hydrogarnet, hematite, magnetite, ilmenite, gibbsite, perovskite, cancrinite, and coesite. The previous research from Stopic et al., 2024 identifies hematite and goethite (iron phases), cancrinite and hydrogarnet (Na-Ca-Al silicates), gibbsite/boehmite (alumina phases), and anatase/ilmenite/perovskite as Ti-bearing phases. Most of the primary phases were identified as well in our study, as displayed in Table 1. Stopic et al., 2024 discusses that a portion of Ti in the raw BR is already present as perovskite (CaTiO₃), owing to the lime additions during Bayer processing, and as ilmenite (FeTiO₃). This affirmation is consistent with our results (Table 1). Following hydrogen reduction, significant mineralogical transformations were observed, with the formation of metallic α-iron as a major reduction product, alongside silicate and titanate phases such as perovskite, and nepheline. Due to the complexity and multiphase nature of the BR, no Rietveld refinement was possible, and the analyses should be interpreted as semi-quantitative (from a non-dimensional point of view), which is appropriate for identifying phase evolution and reaction pathways under the investigated reduction
RkJQdWJsaXNoZXIy MTM0Mzk2