| Aqueous solubility data have been obtained using a "dynamic flow" technique at temperatures in the range 400 to 600(DEGREES)C and pressures in the range 0.5 to 3.5 kbar. Pure water was used as the solvent. Quartz, albite, K-feldspar, labradorite, nepheline, leucite and mixtures of these minerals were used as starting solids. The data are of variable quality. Experiments under carefully controlled conditions yield aqueous concentrations of Si, Al, Na and K which are reproducible to (+OR-)5%.; The data were interpreted thermodynamically in terms of a simple aqueous solution model, involving mononuclear species. Values for the apparent molal Gibbs free energy of AlO(,1.5)(aq) were obtained. Predictions of corundum solubility using these values were too high by one or two orders of magnitude. This failure of the model led to the development of more complex models. Additional experimental data from a variety of sources were used to constrain the models. The data were sufficient to precisely constrain the new models only at 600(DEGREES)C and 2 kbar.; The first model investigated involved the charged hydroxyl species Al(OH)(,4)('-) and Si(OH)(,5)('-). This model could be constrained to fit the experimental data, but only by allowing the activity coefficients for charged aqueous species to drop to well below unity. This is an indication that some kind of strong interaction, or complex formation, is occurring in the solution.; The second model investigated involved the neutral polynuclear species KH(,2)AlO(,3), KH(,3)SiO(,4), NaH(,2)AlO(,3), and NaH(,3)SiO(,4). This model was found to be consistent with most of the experimental data, using a simple activity model, wherein all activity coefficients plus the activity of water were set equal to unity.; The only major failure of this model at 600(DEGREES)C and 2 kbar concerns the prediction of Al concentrations in solutions equilibrated with quartz plus other (Na,K)-Al-Si-bearing phases. In the experimental solutions, Al concentration is higher than the predicted concentrations by up to an order of magnitude.; This shortcoming may be explained by complex activity model corrections. However, these corrections cannot be accurately developed without highly accurate experimental data, and they have little theoretical basis.; Some suggestions for future experiments are given, which will help to resolve these problems. |
