Degassed versus pristine: Evaluating melt inclusions with a new ATR-FPA-FTIR calibration and water imaging method in rhyolitic melts

The concentration of pre-eruptive volatiles is a key parameter that controls the eruptive behaviour of volcanoes. Among these, water is the most abundant volatile and has a major influence on magma chamber dynamics and a primary role in driving eruptions. While accurately quantifying the pre-eruptive dissolved water concentration of magmas is a crucial step in understanding eruptive styles, we still encounter problems in this regard. One of the main challenges is in discriminating pristine melt inclusions from those that lost water due to syneruptive and post-emplacement processes. To this end, we have calibrated a Fourier Transform Infrared (FTIR) setup with a Germanium crystal objective for Attenuated Total Reflectance analyses (ATR) coupled to a Focal-Plane Array detector (FPA), which allows the measurement of single-sided polished rhyolitic melt inclusions. The samples do not require any other special preparation routine, making it a versatile method (i.e. it can be applied to samples prepared for microprobe analysis). The setup generates high-resolution (4 μm) water distribution images that allow the quantification of dissolved water with an uncertainty of ±0.35 wt%. Water diffusion inside the melt inclusions can be identified on the spectral images, allowing for a qualitative analysis that, in most cases, can easily distinguish between pristine and degassed inclusions. We apply this setup to eight units (four explosive and four effusive) from the Nisyros-Yali volcanic center, in the South Aegean Sea, and compare the results with previously published hygrometry data. The results show that even in explosive units, which are generally considered unaffected by gas loss, up to 75% of the melt inclusions lost water. Because of this, the measured water contents show a spread over a range of ~2 wt%. Dissolved water contents in pristine melt inclusions match those from previous hygrometry calculations, indicating ~4 to 5 wt% H2O. In effusive units, all the melt inclusions are degassed, yielding water contents that vary from 0 to 3.6 wt%, as compared to ~5.6 wt% H2O given by mineral-melt hygrometry. The new setup and calibration provide a fast and cheap way for measuring water contents in calc-alkaline rhyolitic melt inclusions. Furthermore, our results indicate the importance of determining the pristine versus degassed nature of inclusions before any interpretations are made.