Redox-variability and controls in subduction zones from an iron-isotope perspective

Abstract An ongoing controversy in mantle geochemistry concerns the oxidation state of the sources of island arc lavas (IAL). Three key factors control oxidation–reduction (redox) of IAL sources: (i) metasomatism of the mantle wedge by fluids and/or melts, liberated from the underlying subducted slab; (ii) the oxidation state of the wedge prior to melting and metasomatism; and (iii) the loss of melt from IAL sources. Subsequently, magmatic differentiation by fractional crystallisation, possible crustal contamination and degassing of melts en route to and at the surface can further modify the redox states of IAL. The remote nature of sub-arc processes and the complex interplay between them render direct investigations difficult. However, a possible gauge for redox-controlled, high-temperature pre-eruptive differentiation conditions is variations in stable Fe isotope compositions (expressed here as δ 57 Fe) in erupting IAL because Fe isotopes can preserve a record of sub-surface mass transfer reactions involving the major element Fe. Here we report Fe isotope compositions of bulk IAL along the active Banda arc, Indonesia, which is well known for a prominent subducted sediment input. In conjunction with other arc rocks, δ 57 Fe in erupted Banda IAL indicates that fractional crystallisation and possibly crustal contamination primarily control their Fe isotope signatures. When corrected for fractional crystallisation and filtered for contamination, arc magmas that had variable sediment melt contributions in their sources show no resolvable co-variation of δ 57 Fe with radiogenic isotope tracers. This indicates that crustal recycling in the form of subducted sediment does not alter the Fe isotope character of arc lavas, in agreement with mass balance estimates. Primitive sources of IAL, however, are clearly isotopically lighter than those sourced beneath mid-ocean ridges, indicating either preferential Fe 3+ -depletion in the mantle wedge by prior, δ 57 Fe-heavy melt extraction, and/or addition of an isotopically-light slab-derived agent. Based on our findings and previous models of arc redox conditions, we propose a three-stage process to explain the Fe isotope composition of IAL: (i) prior melt depletion lowers Fe 3+ /ΣFe (Fe 3+ over total Fe) in the residues, leaving refractory, δ 57 Fe-light and possibly reduced mantle wedge protoliths beneath arcs. The oxygen fugacity ( f O 2 ) of these refractory protoliths may be up to −2 log 10 units reduced relative to the fayalite–magnetite–quartz synthetic oxygen buffer (ΔFMQ); (ii) oxidised, slab-derived fluids, Fe-poor but possibly rich in sulphate (i.e., S 6+ ), trigger melting of depleted protoliths with minimal effect on δ 57 Fe. Melts derived from this fluid-modified wedge source have high Fe 3+ /ΣFe, oxidised by the reduction of S 6+ , but importantly retain the light δ 57 Fe from their mantle wedge source; (iii) after melt liberation from the mantle wedge, arc magmas initially become progressively oxidised and isotopically heavier in Fe through fractional crystallisation of ferromagnesian silicates. In summary, reduction consequent to Fe 3+ -rich melt extraction and subsequent oxidation, likely by S 6+ -rich fluids, results in a “redox yo-yo” in IAL sources. Fractional crystallisation will further oxidise and elevate δ 57 Fe in erupting IAL. Iron isotope signatures in IAL record a complex magmatic history with no simple relation between δ 57 Fe and calculated f O 2 in erupted lavas. Records of higher f O 2 in subduction zones compared to MORB sources are inherited from the subduction component.