A Raman spectroscopic tool to estimate chemical composition of natural volcanic glasses

A correlation between Raman spectra of silicate glasses and their chemical composition is investigated using a collection of 31 natural multicomponent silicate glasses. The sample suite comprises the largest database of Raman spectra collected on natural volcanic materials and spans subalkaline to Na-rich and Krich alkaline compositions. Raman spectra were acquired using a Nd solid state green laser having an excitation wavelength of 532 nm. The model was verified against an independent database of 8 additional samples (i.e. not used for calibration). Ratios of Raman peaks (R, Rn) retrieved from spectra are shown to have a strong covariance with concentrations of six oxides (SiO2, TiO2, Al2O3, FeOT, MgO and CaO) across the compositional range of the sample suite. The Raman ratios are also strongly correlated to pseudostructural parameters (e.g., NBO/T, SM) calculated from oxide concentrations of SiO2, TiO2, Al2O3, FeOT, MgO, CaO, Na2O and K2O. The Raman ratios are relatively insensitive to variations in Na2O and K2O contents and, as a consequence, their concentrations can only be estimated if additional independent constraints on chemical content are available. This work constitutes the first generalized model for retrieving chemical compositions of natural glasses from corresponding Raman spectra. The model provides a rapid, robust and inexpensive way to retrieve compositions of volcanic glasses in both laboratory and field environments and thus represents a powerful new tool for earth and planetary, archaeological and glass sciences. A similar strategy can be applied to silicate melts and glasses used in industrial activities.