Quantification of the water content in synthetic glasses and natural melt inclusions using confocal Raman spectroscopy

Abstract

The importance of water in controlling magmatic ore-forming processes cannot be understated. To accurately constrain the water content in these systems, the application of confocal 532 nm laser Raman spectroscopy was evaluated on silicate glasses of varying bulk composition and water content with a final goal of testing the method developed on hydrous melt inclusions from natural samples. Water derivation is ultimately based on the ratio between areas of the silicate region at 700-1250 cm[superscript -1] and the O–H region at ~3600 cm[superscript -1] of the Raman spectra. Calibration of this method was carried out using hydrous synthetic glasses of rhyolitic, dacitic, and trachytic compositions with a range in water contents (2.68 to 6.59 wt% H[subscript2]O). This study identified important steps for spectral treatment in the water quantification process, including baseline correction of the spectra and application of the frequency-temperature correction. The largest source of error for this determination was identified as a combination of glass sample heterogeneity and variations in the baseline correction of the spectra. The phenomenon of fluorescence, in coloured or impure glasses, obscures the water band and makes baseline corrections difficult. This was investigated through comparison of hydrous and anhydrous synthetic glasses of the same composition to evaluate a correction protocol, and the use of near-UV excitation sources to reduce fluorescence. Finally, this method was applied to natural, quartz-hosted melt inclusions from Late Paleozoic rhyolites from Southern New Brunswick. Ultimately, the method developed enables constraining of water content to within an average ~0.85 wt% accuracy for both synthetic glasses and natural melt inclusions. This is sufficient to differentiate between degassed vs. undegassed liquids, or melts trapped at contrasting crustal depths.