Water dissolution in albite melts:: Constraints from ab initio NMR calculations

Abstract

Hartree-Fock and B3LYP NMR calculations were performed at the 6–311+G(2df,p) level on cluster models representing albite glasses using B3LYP/6 to 31G* optimized geometries. Calculation results on several well-known crystalline materials, such as low albite and KHSi 2O 5 , were used to check the accuracy of the calculation methods. Calculated 29Si-NMR results on clusters that model protonation of Al-O-Si linkages and the replacement of Na + by H + indicate a major increase in Si-O(H) bond length and a 5 ppm difference in δ iso for 29Si compared to that for anhydrous albite glass. The calculated δ iso of 27Al in such linkages agrees with the experimental data, but shows an increase in C q that cannot be fully diminished by H-bonding to additional water molecules. This protonation model is consistent with both experimental 17O NMR data and the major peak of 1H-NMR spectra. It cannot readily explain the existence of the small peak in the experimental 1H spectra around 1.5 ppm. Production of the depolymerized units Al [Q 3]-O-H upon the dissolution of water is not consistent with 27Al, 1H, or 17O NMR experimental results. Production of Si [Q 3]-O-H is consistent with all of the experimental 17O and 1H-NMR data; such units can produce both the major peak at 3.5 ppm and the small peak at 1.5 ppm in 1H spectra, either with or without hydrogen bonding. This species, however, cannot produce the main features of 29Si spectra. It is concluded that although neither protonation nor the production of Si [Q 3]-O-H alone is consistent with the available experimental data, the combination of these two processes is consistent with available experimental NMR data.