Estimating accuracy of 17O NMR measurements in oxide glasses: Constraints and evidence from crystalline and glassy calcium and barium silicates

The use of 17O nuclear magnetic resonance (NMR) spectroscopy to measure site populations in silicate and aluminosilicate glasses has provided insights and challenges to conventional models of glass structure. In order to better understand the level of accuracy and precision achievable, we have synthesized crystalline barium metasilicate (BaSiO3), barium orthosilicate (Ba2SiO4), tricalcium silicate (Ca3SiO5), a barium silicate glass ((BaO)0.45(SiO2)0.55), and a calcium silicate glass ((CaO)0.56(SiO2)0.44), and report 17O NMR spectra for all of these. After correcting the observed intensities for quadrupolar effects, we measure an NBO content of 66.7% ± 0.6% for the BaSiO3, compared to the known value of 66.7%. Applying the same techniques for the glasses gives an NBO content of 58.8% ± 0.8% (vs. the expected 55.5% ± 1.4% from stoichiometry) for the barium silicate and 76.9% ± 1.2% (vs. 78.6% ± 1.4%) for the calcium silicate. Within our uncertainties, we find no evidence for deviation from conventional models of glass structure for the glasses studied here. We also see no NMR signal (detection limit of about 0.5%) at the expected position for “free” oxide ions (bonded only to Ca2 +), as newly constrained by our data for crystalline Ca3SiO5, which contains this species.