LAUSR

Abstract Sulfur is an important volatile element involved in magmatic systems. Its quantification in silicate glasses relies on state-of-the-art techniques such as electronprobe microanalyses (EPMA) or X-ray absorption spectroscopy but is often complicated by the fact that S dissolved in silicate glasses can adopt several oxidation states (S6+ for sulfates or S2− for sulfides). In the present work, we use micro-Raman spectroscopy on a series of silicate glasses to quantify the S content. The database is constituted by 47 silicate glasses of various compositions (natural and synthetic) with S content ranging from 1179 to 13 180 ppm. Most of the investigated glasses have been synthesized at high pressure and high temperature and under fully oxidizing conditions. The obtained Raman spectra are consistent with these fO2 conditions and only S6+ is present and shows a characteristic peak located at ~1000 cm−1 corresponding to the symmetric stretch of the sulfate molecular group (ν1 SO42− $\begin{array}{} \text{SO}_4^{2-} \end{array} $). The intensity of the ν1 SO42− $\begin{array}{} \text{SO}_4^{2-} \end{array} $ peak is linearly correlated to the parts per million of S6+ determined by EPMA. Using subsequent deconvolution of the Raman spectra, we established an equation using the ratio between the areas of the ν1 SO42− $\begin{array}{} \text{SO}_4^{2-} \end{array} $ peak and the silicate network species (Qn) in the high-frequency region: ppm S6+=34371ASO42−AQn±609. $$\begin{array}{} \text{ppm S}^{6+}=34371\frac{A\text{SO}_4^{2-}}{A\text Q^{\text n}}\pm609. \end{array} $$ We tested our calibration on several silicate glasses equilibrated under moderately reducing conditions (QFM+0.8 ≤fO2≤ QFM+1.4) in which S is dissolved as both SO42− $\begin{array}{} \text{SO}_4^{2-} \end{array} $ and S2−. We also analyzed several olivine-hosted melt inclusions collected from Etna for which the fO2 and S speciation are unknown. For these samples, the S content estimated by the Raman calibration is systematically lower than the total S measured by EPMA. We combined both methods to estimate the S2− content not accounted for by Raman and derive the S speciation and fO2 conditions. The derived fO2 is consistent with the imposed fO2 for synthesized glasses and with current assumed fO2 conditions for basaltic melt inclusions from Etna.