The FACT oxide database provides critically evaluated/optimized thermodynamic data for the molten oxide phase containing Al2O3-As2O3-B2O3-CaO-CoO-CrO-Cr2O3-Cu2O-FeO-Fe2O3-GeO2-K2O-MgO-MnO-Na2O-NiO-PbO-SiO2-SnO-TiO2-Ti2O3-ZnO-ZrO2-[S2--SO42--PO43--CO32--H2O-OH--F--Cl--Br--I-] and for many extensive oxide solid solutions such as:
Spinel: (Al, Co2+, Co3+, Cr2+, Cr3+, Fe2+, Fe3+, Mg, Ni2+, Zn)T [Al, Co2+, Co3+, Cr3+, Fe2+, Fe3+, Mg, Ni2+, Zn, ð ]2O O4
Pyroxene: (Ca, Fe2+, Mg)M2 (Al, Fe2+, Fe3+, Mg)M1 (Al, Fe3+, Si)T1 SiT2 O6
Melilite: (Ca, Pb)2A [Al, Fe2+, Fe3+, Mg, Zn]T1 {Al, Fe3+, Si}2T2 O7
Olivine: (Ca, Co, Fe2+, Mg, Mn, Ni, Zn)M2 [Ca, Co, Fe2+, Mg, Mn, Ni, Zn]M1 SiO4
For each particular solid solution, a special model was developed within the framework of the Compound Energy Formalism (CEF). Normally, this is straightforward for solutions such as olivine, when all end-members are neutral. When many end-members are charged, their number usually increases drastically. Although there is generally no problem in reproducing all data in simple systems, it becomes very easy to end up with an unbalanced model that gives unreasonable extrapolations of lower-order sub-systems into a multicomponent solution. That is, in lower-order sub-systems, different models can give equally good or, sometimes, even mathematically equivalent fits. However, extrapolations into multicomponent systems are different. This problem is most pronounced for solutions, such as spinel, pyroxene and melilite, where the same cations can be present on more than one sublattice.
Furthermore, the CEF is too general and too symmetric to give any indication about the proper sequence of optimization of model parameters. For example, when a polynomial model is used to describe the thermodynamic properties of a binary solution, it is reasonable to try first a regular solution model, then a sub-regular model and so on, but in the case of the CEF, it is not clear where to start. Furthermore, not all parameters are independent, because of the condition of electroneutrality. Therefore, a physically meaningful model must take into account availability of experimental data for a particular solution and must specify a set of model parameters and a sequence of their optimization, so that if only a few parameters are needed to reproduce the experimental data, the rest of them are kept equal to zero. The modeling of the solid solutions mentioned above will be briefly discussed.
The quasichemical model has been used to treat short-range ordering in molten oxide slags. The generality and flexibility of this model have been greatly increased by the latest modifications which will be briefly outlined. It has become possible to develop a particular model within the framework of the Quasichemical Formalism based on the physical nature of various types of oxide liquids such as borosilicate melts or the Al2O3-Na2O-SiO2 liquid exhibiting the “charge compensation effect”. The Quasichemical Formalism provides a seamless combination of these models into a single liquid oxide phase.