Iron isotope fractionation in highly evolved magmas results from ilmenite crystallization

Abstract

Globally, granites and rhyolites with SiO2 > 70 wt% show large Fe isotope variation (δ56Fe = −0.05 to +0.65 ‰) relative to less silicic igneous rocks in δ56Fe vs. SiO2 space. The upper bound of the data tends to show δ56Fe increase with increasing SiO2. Granitic magma differentiation can be invoked to explain magma compositional variation, including Fe isotope variation, but clearly cannot explain the highly varied δ56Fe values. The latter may result from magma differentiation of varying liquidus phases, magma mixing, assimilation and magma source compositional variation. To decipher how each of these and altogether explain the large δ56Fe variation requires rigorous studies of varying well characterized sample suites. This paper is not to solve all these issues but demonstrates clearly using three sample suites with well-defined liquid lines of descent from alkaline basalts to peralkaline rhyolites to show that the δ56Fe increases with continued magma differentiation (increasing SiO2, SiO2/MgO and decreasing MgO). The rapid δ56Fe increase for samples with SiO2 > 70 wt% results from ilmenite (vs. magnetite) fractionation. Among all the major liquidus phases, ilmenite has a distinctive affinity with light-Fe isotope, whose crystallization elevates δ56Fe in the residual melts. This result demonstrates the affinity of isotopically heavy Fe with Fe3+ and the correlation of isotopically light Fe with Fe2+ because δ56Fe values of ilmenite (TiFe2+O3) ≪ δ56Fe values of magnetite (Fe2+O∙Fe23+O3). We can conclude that ilmenite solid solution is likely the major oxide liquidus phase at the late-stage felsic melt evolution for relatively dry magmas with low fO2 such as peralkaline rhyolites we study here and mid-ocean ridge basalts. We further predict that magnetite (vs. ilmenite) solid solution may be the important liquidus phase for wet magmas with high fO2 such as volcanic arc magmas, where crystallization of magnetite with high δ56Fe will deplete the heavy Fe isotopes in the residual melts. It is thus possible that the large Fe isotope variation of global igneous rocks with SiO2 > 70 wt% may result from varying parental magma compositions (varying water content and fO2) plus bulk-rock modal mineralogy controls of granitic rocks. This work thus lays the foundation for testing this hypothesis through rigorous studies on ideal sample suites of global significance.