Oxide ion speciation in potassium silicate glasses: New limits from 17O NMR

► 17O NMR can directly detect “free” oxide ions (FO) in some silicate glasses.
► FO can also be estimated indirectly, from bridging and non-bridging oxygen contents.
► In two potassium silicate glasses, both methods suggest < 1% FO.
► 29Si NMR spectra are also consistent with low FO contents.
► Predictions of low FO contents from thermodynamic considerations are supported.

In silicate glasses and liquids, “free” oxide ions (O2 − ions not bonded to any network-forming cations such as Si) are required by stoichiometry for ratios of O/Si > 4, for example in “sub-orthosilicate” compositions (< 33.3% SiO2 in MO-SiO2 or M2O-SiO2 binaries). Measurements of oxide ion activities and other thermodynamic considerations suggest however, that at higher silica contents the concentration of such ions should rapidly become very low, particularly in systems with highly basic modifier oxides such as Na2O or K2O. Recent 17O NMR studies of alkaline earth ortho- and sub-orthosilicate glasses have indeed directly shown the presence of a few percent of “free” oxide ions, but did not detect this species at a silica content of 44 mol%. In contrast, recent O 1s XPS analyses of bridging oxygen concentrations in Na- and K-silicate glasses have suggested as much as about 6% “free” oxide even at about 67% SiO2 [1,2]. In K-silicates, theoretical calculations presented here, as well as previous work on Ca and Mg silicates, suggest that this species should be readily and directly measurable by 17O. However, we show here that “free” oxide ions are not detectable in glasses with 34 and 40 mol% K2O, and therefore are not likely to be present above the 0.1 to 1% level. Measured ratios of bridging to non-bridging oxygens are within errors of those expected from analyzed compositions and an assumption of negligible “free” oxide, and 29Si MAS spectra can be accurately fitted with constraints based on this same assumption.