Oxygen isotopes and the origin of high- 87Sr/ 86Sr andesites

Abstract

Andesites from the Peruvian Andes and the Banda arc of Indonesia are characterized by unusually high and variable 87Sr/ 86Sr ratios. The Banda arc samples, including two cordierite-bearing lavas from Ambon, show a clear positive correlation between 87Sr/ 86Sr and δ 18O. The andesitic rocks have δ 18O values that range from 5.6 to 9.2‰. Over that range in δ 18O, 87Sr/ 86Sr increases from 0.7044 to 0.7095. The cordierite-bearing lavas have δ 18O values of approximately 15‰ and 87Sr/ 86Sr ratios of approximately 0.717. The similarity between δ 18O values and 87Sr/ 86Sr ratios in total rocks and separated plagioclase phenocrysts of the Banda arc samples indicates that the measured isotope ratios are primary and have not been affected by secondary, low-temperature post-eruptive alteration. The observed variation between O and Sr isotopic ratios can be modeled by two-component mixing in which one component is of mantle isotopic composition. As the crust beneath the Banda arc is probably oceanic, contamination of the manle component may have resulted from the subduction of either continentally-derived sediments or continental crust. Mixing calculations indicate that the contaminant could have an isotopic composition similar to that observed in the cordierite-bearing lavas. The Andean samples, despite petrographic evidence of freshness, exhibit whole-rock δ 18O values significantly higher than those of corresponding plagioclase phenocryst separates, indicating extensive low-temperature post-eruptive alteration. The plagioclase mineral separates show a range of δ 18O values between 6.9 and 7.9‰. The 87Sr/ 86Sr ratios of these same samples are, in most instances, not significantly different from those measured for the whole rock, thus signifying that the phenocrysts and groundmass were in isotopic equilibrium at the time of eruption. Unlike the lavas of the Banda arc, the Andean lavas show no strong positive correlation between 87Sr/ 86Sr ratios and δ 18O values, but instead lower 87Sr/ 86Sr ratios appear to be associated with higher δ 18O values. The δ 18O and 87Sr/ 86Sr values of the Peruvian samples are both slightly higher than those of “normal” island arc volcanics. The small proportions of contaminant implied by the O isotope results seem to preclude continental crustal contamination as a primary cause of high 87Sr/ 86Sr ratios. The most plausible process that can explain both O and Sr isotope results is one in which sediments of continental origin are partially melted in the subduction zone. These melts rise into overlying mantle material and subsequently participate in the formation of calc-alkaline magmas. If the involvement of a sialic component in the genesis of andesitic magma occurs in the subduction zone, melting of that sialic material signifies temperatures of at least 750–800°C at the top of the subducted lithospheric slab at depths of approximately 150 km. The fact that contamination has apparently occurred in the Banda arc samples without producing any simple widespread correlations between Sr and O isotopic compositions on the one hand and major or trace element abundances on the other, shows that isotopic correlations, possibly including pseudo-isochrons, can be produced by mixing without producing trace element mixing correlations. Because O versus Sr isotope correlations are little affected by processes of partial melting of differentiation, they provide a direct means of testing whether Sr isotopic variations in volcanic rocks are of mantle origin or are due instead to mixing with sialic material.