EPSL: Mg isotopic fractionation during basalt differentiation as recorded by evolved magmas(XJ Wang)

      Magnesium isotopic fractionation during basalt differentiation is commonly thought to be negligible atcurrent levels of analytical precision. However, Mg isotopic variation of ∼0.4 (δ26Mg) can be observed in some highly evolved volcanic rocks with MgO contents of <5 wt.%, suggesting that detectable Mg isotopic fractionation may occur during late-stage basalt differentiation.

      Here we examine this possibility with a Mg isotopic study of a suite of well-characterized cogenetic alkaline volcanic rocks from St.Helena Island (South Atlantic), which vary from primitive nepheline-normative basalt to highly evolved trachyandesite with MgO contents decreasing from 15.72 wt.% to 0.81 wt.%. Our results show that the basalt samples which only experienced segregation of olivine and clinopyroxene have a narrow δ26Mgrange (−0.23 to −0.32), while the evolved rocks saturated with Fe–Ti oxides (MgO < 5 wt.%) display larger Mg isotopic variation with δ26Mg vary to higher (−0.14) or lower (−0.36) values relative to mantle value (δ26Mg = −0.25 ± 0.04). For most of the Fe–Ti oxide saturated samples, their δ26Mg values are positively correlated with TiO2 contents and Ti/Ti* ratios and negatively correlated with total alkali contents. This indicates that detectable Mg isotopic fractionation occurred in MgO-poor samples, probably through fractional crystallization of Mg-bearing Fe–Ti oxides. Further Mg isotopic analysis of Fe–Ti oxide phenocrysts (titanomagnetite) in the evolved samples reveals that titanomagnetite has remarkably higher δ26Mg value (+0.15 to +0.52) than the corresponding bulk sample and silicate minerals. Fractional crystallization of such isotopically heavy titanomagnetite is thus expected to drive the residual magma progressively enriched in light Mg isotopes. In order to reproduce the observed Mg isotopic variation in St. Helena evolved samples, quantitative modeling requires a titanomagnetite-melt Mg isotopic fractionation factor (26MgTi-Mgt−melt) of ∼0.6 , which is among our measured δ26Mg difference value (0.41–0.73) between titanomagnetite and bulk samples. When combining our results with published data for alkaline lavas from the Antipodes volcano (New Zealand), contrasting patterns of Mg isotope fractionation emerge during late-stage basalt differentiation. Quantitative modeling shows that the positive and negative δ26Mg–MgO correlations shown by evolved lavas could be related to segregation of isotopically heavy titanomagnetite and isotopically light ilmenite, respectively, depending on redox state of the magma system.

      Therefore, this study highlights that basalt differentiation involving separation of Fe–Ti oxides may induce resolvable Mg isotopic fractionation, and the δ26Mg values of compositionally evolved rocks should be used with caution in studies of petrogenesis.

Fig: Comparison between δ26Mg values of whole rock samples and some separated phenocrysts therein from St. Helena Island