Magmatic–hydrothermal evolution and devolatilization beneath Merapi volcano, Indonesia

•Melt inclusions were analyzed for major and volatile elements in clinopyroxene, amphibole and plagioclase phenocrysts.•Barometry was conducted on inclusions and amphibole, suggesting the presence of magma reaching 26 km beneath the summit.•Volcanic gases were sampled and analyzed in 2006, and compiled from the literature.•A mass balance model for devolatilization was built by comparing the volatiles exsolved from the melt to volcanic gases.•Magmatic volatiles are first supercritical but later unmix into Cl-rich brine and CO2-rich vapor, 5 km beneath the summit.

At Merapi volcano, Indonesia, explosive eruptions lasting several months result from injections of reduced mafic magmas and are interspersed by periods of lava dome growth and collapse, quiescent degassing, assimilation of wall rock and fractional crystallization lasting a few years. Silicate melt inclusions and volcanic gases were analyzed to study processes of magmatic volatile exsolution and volcanic degassing during periods of quiescent degassing at Merapi. Volcanic gases were sampled at Merapi in 1994 during a phase of quiescent degassing and in 2006 at the end of a dome-growth and collapse explosive eruption. Silicate melt inclusions were collected and analyzed from lavas and scorias for their major element and volatile components. Solubility relationships between H2O and CO2 in melt inclusions demonstrate that melt was being trapped beneath Merapi to depths reaching 19 km. Host amphibole geobarometry indicates crystallization pressures reaching up to 26 km. Hence there is evidence of deep pooling and crystallization of magma beneath Merapi at appreciable depths. Using a mass balance model of magmatic volatile exsolution, we demonstrate that the magma degassing from 19 km to the surface approaches closed-system behavior. Compared to the bulk magmatic volatiles analyzed in melt inclusions, volcanic gases had similar amounts of CO2, were enriched in H2O and S and depleted in Cl and F. We thus propose that the magmatic volatile phase was initially exsolved from the magma as a supercritical fluid and that it subsequently exsolved into a H2O–Cl–F-rich brine and CO2–S-rich vapor. According to the H2O–NaCl model, brine-vapor exsolution occurred at ~ 5 km in depth for a 5 wt.% NaCl, 900 °C to 1000 °C fluid, although the presence of H2S, SO2 and CO2 may cause the supercritical fluid to unmix at greater depths. Magma at shallower depths exsolved a brine and a vapor directly from the melt. The volcanic gases at Merapi thus represent the vapor fraction of a magmatic volatile phase, as brine ponded below the surface. We further suggest that carbon dioxide liberated by the magmatic assimilation of limestone, as proposed by other studies, fluxed the melt thereby promoting CO2 enrichment of the melt and H2O degassing. Finally, we propose that Merapi volcanic gases are enriched in sulfur compared to the modeled volatile phase composition because deep, reduced mafic magma supplies sulfur to the shallow magmatic plumbing system.