Melting Phase Equilibria from 4 to 7 GPa in the System CaO-MgO-Al2O3-SiO2-CO2, the Persistence of the “Ledge” in Carbonated Basalt with Excess Silica, and the Most Likely Limits on the Depths of Termination of Carbon Cycle at Subduction Zones

Abstract

Abstract Melting phase relations involving model carbonated basalt with excess silica were studied in experiments over the pressure range of 4–7 GPa in the system CaO-MgO-Al2O3-SiO2-CO2 to determine if there is a sharp decrease in the melting temperatures along the transition from carbon dioxide vapor (vapor) to dolomite. The phase assemblages of clinopyroxene + garnet + coesite + vapor + carbon dioxide-bearing silicate liquid (silicate liquid) and clinopyroxene + garnet + coesite + dolomite + carbonate liquid, exist over 4–5 and 5.8–7 GPa, respectively. These two distinct phase assemblages form the two, vapor + silicate liquid and dolomite + carbonate liquid-bearing divariant surfaces. The dissolved carbon dioxide and the molar calcium number [Ca# 100*(Ca/Ca + Mg)] of the silicate and carbonate liquids are approximately 4–8 wt% and between 50–55 and 35–40 wt% and 69–71, respectively. The compositions of phases vary little, implying minimal topography along the two surfaces, and the temperatures rise linearly along the silicate liquid-bearing divariant surface over 4–5 GPa. Between 5.2 and 5.6 GPa, however, the temperatures decrease precipitously by ~200–250°C and, along with this steep decline, the liquid changes from silicate to carbonate, with the rest of the phase assemblage of clinopyroxene + garnet + coesite + vapor, persisting. Hence, and this is important to emphasize, this liquid, coexisting with vapor, is carbonate in composition in the absence of dolomite. Isobaric invariance, at 5.4 GPa/1250°C, 5.6 GPa/1150°C, and 5.8 GPa/1100°C, consists of the six-phase assemblage of clinopyroxene + garnet + coesite + vapor + dolomite + carbonate liquid. Melting phase relations are thus univariant, and correspond to that of a solidus ‘ledge’, i.e. with a negative Clapeyron slope, in this part of the composition space. The melting reaction along the ledge is clinopyroxene + vapor = garnet + coesite + dolomite + carbonate liquid, and the ledge separates the two divariant surfaces. The Ca# of the coexisting carbonate liquid and dolomite here are opposite to those of the carbonate liquid and dolomite on the calcite-magnesite join at similar pressures as in this study. This is most likely a consequence of the combined effects of (a) observations from experiments and theory that the fusion curve of calcite starts to diverge from that of magnesite toward lower temperatures at pressures in excess of ~5 GPa, and (b) the pressure, where ultrabasic silicate–carbonate (~2.5–3 GPa) and excess-silica carbonate-basalt (>4 GPa, as inhere) systems undergo carbonation. These, in turn, cause the liquid and dolomite in experiments here to become more calcic and more magnesian than observed in experiments on the calcite-magnesite join. The solidus ledge, here, has a profound effect because the most plausible modern-day model ocean crust subduction zone geotherms in Earth will, in all likelihood, intersect it and cause fusion of dolomite, thereby, in effect, liberating all carbon from what once was a carbonate-basalt mixture. Thereafter, little exists to suggest that there is anything ‘deep’ to the carbon cycle, through recycling, with most of it likely confined to less than ~200 km in Earth.