Melt inclusion vapour bubbles: the hidden reservoir for major and volatile elements

Co-auteur étrangerInternational audienceolivine-hosted melt inclusions (Mis) provide samples of magmatic liquids and their dissolved volatiles from deep within the plumbing system. Inevitable post-entrapment modifications can lead to significant compositional changes in the glass and/or any contained bubbles. Reheating is a common technique to reverse MI crystallisation; however, its effect on volatile contents has been assumed to be minor. We test this assumption using crystallised and glassy basaltic MIs, combined with Raman spectroscopy and 3D imaging, to investigate the changes in fluid and solid phases in the bubbles before and after reheating. Before reheating , the bubble contains CO 2 gas and anhydrite (caSo 4) crystallites. The rapid diffusion of major and volatile elements from the melt during reheating creates new phases within the bubble: So 2 , gypsum, Fe-sulphides. Vapour bubbles hosted in naturally glassy MIs similarly contain a plethora of solid phases (carbonates, sulphates, and sulphides) that account for up to 84% of the total MI sulphur, 80% of CO 2 , and 14% of FeO. In both reheated and naturally glassy MIs, bubbles sequester major and volatile elements that are components of the total magmatic budget and represent a "loss" from the glass. Analyses of the glass alone significantly underestimates the original magma composition and storage parameters. Olivine-hosted melt inclusions (MIs) provide insight into the nature of the magma mantle source, storage conditions , and pre-eruptive volatile contents 1. Following entrapment, MIs undergo compositional modifications due to growth of the host olivine along the MI walls, and to crystallisation of daughter minerals from the glass due to slow ascent rates, and/or cooling 1,2. Another modification is the nucleation of a vapour bubble in response to decompression during cooling and post-entrapment crystallisation 1,2 , further reducing the solubility of volatiles in the glass. Vapour, or shrinkage, bubbles produced by differential thermal contraction between the melt (glass) and the host crystal are considered to be inherent to the MI 1,2. However, pre-existing bubbles that formed externally in a vapour-saturated system may also become trapped inside MIs. Vapour bubbles may also form during MI leakage and decrepitation of the host crystal 3,4. Discriminating between various bubble types depends upon the size of the bubble relative to the total inclusion. Since the volumetric proportions of vapour bubbles depend on the cooling rate, volatile content and melt composition, cooling-related shrinkage and melt-saturated bubbles normally comprise 0.2 to 10 vol% of the inclusion 3,4. Bubbles with greater volumetric proportions are not considered inherent to the MI 3,4. Due to the strong pressure-dependency of CO 2 solubility, the contraction of the melt and the decrease of the internal pressure in response to cooling and post-entrapment crystallisation first leads to rapid CO 2 saturation of the melt, and consequently, the transfer of CO 2 gas into the bubble 5-7 , so that analyses of the glass yield erroneously low magmatic CO 2 concentrations. This poses significant problems as the CO 2 content is commonly used to infer the pressure of crystallisation and MI entrapment, as well as magmatic storage depths. By only considering the glass, these values are grossly underestimated. Many of the post-entrapment modifications that occur within olivine-hosted MIs can be corrected using well-constrained exchange coefficients and recently established methods to quantify the amount of volatiles, particularly CO 2 , sequestered by the bubble. These approaches include the use of trace element proxies, such as CO 2 / Nb, to determine the pre-eruptive CO 2 content of the undegassed melt 8 , reheating the MI to resorb the bubbl