Experimental determination of H2O loss from melt inclusions during laboratory heating: Evidence from Raman spectroscopy

Experiments have been conducted to quantify H2O loss from melt inclusions during laboratory heating using Raman spectroscopy. Quartz-hosted melt inclusions from the early-erupted plinian stage of the Bishop Tuff were heated to 800 °C and 1 kbar for 4 to 1512 h (63 days). Previous studies had shown that unheated melt inclusions from this unit of the Bishop Tuff contain 4.8–6.5 wt.% H2O. Many Bishop Tuff melt inclusions fluoresce under visible (514 nm) Raman excitation. To minimize fluorescence interference, a method was developed to analyze silicate glasses and melt inclusions using an UV (244 nm) excitation source. The Raman microprobe was calibrated using silicate glass standards with known H2O contents, and the results indicate that the UV excitation data produce a calibration line with a different slope compared to that produced using visible excitation. The inclusions show insignificant H2O loss when heated for less than 12 h, while approximately 75% of the original H2O was lost after 1512 h. The rate of H2O loss decreases after a few hundred hours, suggesting either a change in H2O speciation or in the mechanism of H2O loss. Our results suggest that most silicic melt inclusions maintain their original H2O concentration if they are not heated for more than about 12 h during laboratory studies.