Effect of variable CO2on eclogite-derived andesite and lherzolite reaction at 3 GPa-Implications for mantle source characteristics of alkalic ocean island basalts

We have performed reaction experiments between 1, 4, and 5 wt % CO2-bearing MORB-eclogite (recycled oceanic crust)-derived low-degree andesitic partial melt and fertile peridotite at 1375°C, 3 GPa for infiltrating melt fractions of 25% and 33% by weight. We observe that the reacted melts are alkalic with degree of alkalinity or Si undersaturation increasing with increasing CO2 content in reacting melt. Consequently, an andesite evolves through basanite to nephelinite owing to greater drawdown of SiO2 from melt and enhanced precipitation of orthopyroxene in residue. We have developed an empirical model to predict reacted melt composition as a function of reacting andesite fraction and source CO2 concentration. Using our model, we have quantified the mutual proportions of equilibrated melt from andesite-peridotite (+ CO2) hybridization and subsequent peridotite (± CO2)-derived melt required to produce the major element composition of various ocean island basalts. Our model can thus be applied to characterize the source of ocean islands from primary alkalic lava composition. Accordingly, we determined that average HIMU source requires 24 wt % of MORB-eclogite-derived melt relative to peridotite containing 2 wt % CO2 and subsequent contribution of 45% of volatile-free peridotite partial melt. We demonstrate that mantle hybridization by eclogite melt-peridotite (± CO2) reaction in the system can produce high MgO (>15 wt %) basaltic melts at mantle potential temperature (TP) of 1350°C. Therefore, currently used thermometers to estimate TP using MgO content of primary alkalic melts need to be revised, with corrections for melt-rock reaction in a heterogeneous mantle as well as presence of CO2.