Magnesium isotopes and zircon geochemistry verify the entrainment of garnet increasing the maficity of S-type granites

Abstract

It is controversial how S-type granites get higher maficity (MgO + FeOt) than initial melts derived from metasedimentary rocks. Different models, including source-controlled processes (residual/peritectic mineral assemblage entrainment), mixing of magmas derived from different source rocks, and fractional crystallization, have been proposed. However, whole-rock major-trace element and radiogenic isotope data generally provide ambiguous constraints to these models, and the entrained minerals can be later modified by host magmas to have re-equilibrated compositions or replaced by new mineral phases. We address this issue by means of whole-rock Mg isotope and zircon trace element and O isotope analyses on the ∼250 Ma Jiuzhou S-type granite pluton from South China. δ26Mg values of the granites exhibit a negative relationship with increasing maficity, varying from −0.14‰ at 3.00 wt% to −0.43‰ at 8.72 wt%. Zircon yields high and variable δ18O values of 10.0–15.2‰, suggesting that the dominant source rocks are metasedimentary rocks with heterogeneous compositions. No relationship between zircon δ18O value versus whole-rock δ26Mg value or maficity is observed, indicating that magma mixing has played a negligible role in the origin of the granites. Meanwhile, neither Hf nor Ti concentration in zircon shows correlation with whole-rock maficity, precluding fractional crystallization as the dominant mechanism for the compositional variation of the granites. Instead, zircon from granite samples with different maficities seems to have grown from magmas with different compositions but rather similar initial magma temperatures of ca. 810–850 °C as suggested by the upper limit values of the Ti-in-zircon thermometer. Consequently, the negative relationship between δ26Mg value and maficity is best explained as melt entraining a solid assemblage from the melting source, with garnet as the main ferromagnesian phase because this phase is enriched in lighter Mg isotopes than other ferromagnesian phases. This conclusion is supported by the negative relationship between δ26Mg and (Yb/Dy)N values, since garnet is also enriched in heavy rare earth elements (HREE). Furthermore, phase equilibrium modeling using the sample with the lowest δ26Mg value and highest maficity indicates that garnet is always the dominant ferromagnesian phase (exceeding 50% to nearly 100% among the ferromagnesian phases) in the solid assemblage, which equilibrates with melt across a variety of P–T–H2O initial magma conditions (6–8 kbar/800–900 °C/1–5 wt% H2O) at the melting sources. Simple mixing calculations also confirm that the mixing between melt and a solid assemblage consisting mainly of garnet can account for the compositional variation of Mg isotopes with maficity. This study highlights the advantage of integrated geochemical analyses, particularly Mg isotopes, in placing important constraints on the petrogenetic processes that produce high-maficity S-type granites.