Partial reactive crystallization of variable CO2-bearing siliceous MORB-eclogite-derived melt in fertile peridotite and genesis of alkalic basalts with signatures of crustal recycling

The presence of heterogeneity in the form of recycled altered oceanic crust (MORB-eclogite) has been proposed in the source of HIMU ocean island basalts (OIBs) [1]. Partial melts of recycled oceanic crust, however, are siliceous and Mg-poor and thus do not resemble the major element compositions of alkalic OIBs that are silica-poor and Mg-rich. In an upwelling heterogenous mantle, MORB-eclogite undergoes melting deeper than volatile-free peridotite, hence, andesitic partial melt derived from eclogite will react with subsolidus peridotite. We have examined the effect of such a melt-rock reaction under volatile-free conditions at 1375 °C, 3 GPa by varying the melt-rock ratio from 8 to 50 wt.% [2]. We concluded that the reacted melts reproduce certain major element characteristics of oceanic basanites, but not nephelinites. Also, the melt-rock reaction produces olivine and garnet-bearing websteritic residue. Because presence of CO₂ has been invoked in the source of many HIMU ocean islands, the effect of CO₂ on such a melt-rock reaction needs to be evaluated. Accordingly, we performed reaction experiments on mixtures of 25% and 33% CO₂-bearing andesitic partial melt and peridotite at 1375 °C, 3 GPa by varying the dissolved CO₂ content of the reacting melts from 1 to 5 wt.% (bulk CO₂ from 0.25 to 1.6 wt.%) [3, this study]. Owing to melt-rock reaction, with increasing CO₂ in the bulk mixture, (a) modes of olivine and cpx decrease while melt, opx and garnet increase, (b) reacted melts evolve to greater degree of Si-undersaturation (from andesite through basanite to nephelinite), (c) enhanced crystallization of garnet take place with higher CO₂ in the melt, reducing alumina content of the reacted melts, and (d) CaO and MgO content of the reacted melts increase, without affecting FeO* and Na₂O contents (indicating greater propensity of Ca²⁺ and Mg²⁺ over Fe²⁺ and Na⁺ to enter silicate melt as carbonate). For a given melt-MgO, the CO₂-bearing reacted melts are a better match for alkalic basalts in terms of SiO₂, Al₂O₃, CaO and CaO/Al₂O₃ than the CO₂-free ones [3]. Using the experimental data, we have further developed an empirical model to predict mineral modes in residue and reacted melt compositions for olivine-opx saturated lithologies as a function of melt:rock ratio and bulk CO₂ content. For example, in case of 5 wt.% eclogite melt infiltrating in fertile peridotite, with bulk CO₂ from 0 to 2 wt.%, the derivative melts show an increase in CaO and MgO from 11 to 16 wt.%, 15 to 24 wt.%, respectively and decrease in SiO₂ and Al₂O₃ from 45 to 39 wt.% and 14 to 5 wt.%, respectively. From this model, we have created a major element composition space of MORB-eclogite-derived reactive melt mass vs. bulk CO₂ and we predict that primary HIMU-type magmas require <5 to 10 wt.% of MORB-eclogite melt input and up to 0.8 wt.% bulk CO₂ in their source. Our model also allows determining the residual lithology at the source of alkalic basalts, produced owing to eclogite melt-peridotite reaction with or without CO₂. [1] Jackson & Dasgupta (2008) EPSL 276, 175-186. [2] Mallik & Dasgupta (2012) EPSL 329-330, 97-108. [3] Mallik & Dasgupta (in press) JPetrol.