Modeling complex oxides: Thermochemical behavior of nepheline-forming Na-Al-Si-B-K-Li-Ca-Mg-Fe-O and hollandite-forming Ba-Cs-Ti-Cr-Al-Fe- Ga-O systems

High concentrations of Na2O and Al2O3 in the liquid high-level radioactive waste (HLW) stored at the Hanford Site can cause nepheline (NaAlSiO4) to precipitate in a vitrified monolithic waste form upon cooling. Nepheline phase formation removes glass- former SiO2 and -modifier Al2O3 from the immobilization matrix in greater proportion to alkalis, which can reduce glass durability and consequently increase the leach rate of radionuclides into the surrounding environment. Current uncertainty in defining the HLW glass composition region prone to precipitating nepheline necessitates targeting a conservative waste loading, which raises operational costs by extending the liquid radioactive waste disposal mission and increases the required permanent repository storage capacity. An accurate thermochemical representation of HLW glass compositions is necessary to obtain a comprehensive understanding of the composition-temperature space for nepheline formation, which can facilitate the development of a phase field model of the mesoscale microstructural evolution of nepheline crystallization in HLW glass. As such an understanding of nepheline nucleation and grain growth kinetic behavior may lead to significant improvements in the production efficiency of durable HLW glass, generating thermochemical descriptions of the constituent phases is of primary importance. Thus, a database consisting of the oxides of the nepheline-forming Na2O-Al2O3- SiO2 system and HLW glass nepheline solutes B2O3, K2O, CaO, Li2O, MgO, Fe2O3, and FeO has been developed to yield a thermochemical model capable of characterizing nepheline precipitation in HLW glass at equilibrium. Due to their high molar concentrations within vitrified glass, Na2O, Al2O3, B2O3, and SiO2 were considered major oxides whereas more dilute B2O3, K2O, CaO, Li2O, MgO, Fe2O3, and FeO were treated as minor constituents. All pseudo-binary systems composed of the major as well as major- minor oxide systems were thermodynamically assessed according to the CALculation of PHAse Diagrams (CALPHAD) methodology. Additionally, all pseudo-ternary systems consisting of the major oxides were assessed due to the increased probability of interactions between these higher concentration oxides. Gibbs energies of solid solution phases and the oxide liquid were modeled using the compound energy formalism (CEF) and two- sublattice partially ionic liquid (TSPIL) model, respectively. Accuracy of the thermodynamic database was validated by comparing model calculations to HLW glass experimental data. Both annealed and canister centerline cooled (CCC) glass sample data were considered. Additionally, nepheline phase compositional data was included for comparison with database computations. Results of these comparisons indicate that the database-derived calculations agree well with HLW glass experimental data. As phase precipitation in a CCC glass sample is dependent on kinetics, however, a phase field or similar model will need to be utilized to obtain a non-equilibrium description of CCC HLW glass behavior, which in turn often require accurate Gibbs energies of phases. Hollandite has been studied as a candidate ceramic waste form for the disposal of HLW due to its inherent leach resistance and ability to immobilize alkaline-earth metals such as Cs and Ba at defined lattice sites in the crystallographic structure. The chemical and structural complexity of hollandite-type phases with a large number of potential additives and compositional ranges for high-level waste immobilization would require impractical systematic experimental exploration. Modeling the equilibrium behavior of the complex hollandite-forming oxide waste system would aid in the design and processing of hollandite waste forms by predicting their thermodynamic stability. Thus, a BaO-Cs2O- TiO2-Cr2O3-Al2O3-Fe2O3-FeO-Ga2O3 thermodynamic database was developed according to the CALPHAD methodology. The CEF was used to model solid solutions such as hollandite while the TSPIL model characterized the oxide melt. The database was validated by experimental hollandite compositional data, and an isothermal BaO-Cs2O-TiO2 pseudo- ternary diagram with added hollandite solutes was generated to extrapolate phase equilibrium behavior to regions not experimentally explored