Iron isotope composition of depleted MORB

Abstract:

In terrestrial basalts, iron isotope ratios are observed to weakly fractionate as a function of olivine and pyroxene crystallization. However, a ~0.1‰ difference between chondrites and MORB had been reported (Dauphas et al. 2009, Teng et al. 2013 and ref. therein). This observation could illustrate an isotope fractionation occurring during partial melting, as a function of the Fe valence in melt versus crystals.

Here, we present high-precision Fe isotopic data measured by MC-ICP-MS on well-characterized samples from the Pacific-Antarctic Ridge (PAR, n=9) and from the Garrett Transform Fault (n=8). These samples allow exploring the Fe isotope fractionation between melt and magnetite, and the role of partial melting on Fe isotope fractionation.

Our average δ⁵⁶Fe value is +0.095±0.013‰ (95% confidence, n=17), indistinguishable from a previous estimate of +0.105±0.006‰ (95% confidence, n=43, see ref. 2). Our δ⁵⁶Fe values correlate weakly with MgO contents, and correlate positively with K/Ti ratios.

PAC1 DR10 shows the largest Ti and Fe depletion after titanomagnetite fractionation, with a δ⁵⁶Fe value of +0.076±0.036‰. This is ~0.05‰ below other samples at a given MgO. This may illustrate a significant Fe isotope fractionation between the melt and titanomagnetite, in agreement with experimental determination (Shahar et al. 2008).

GN09-02, the most incompatible-element depleted sample, has a δ⁵⁶Fe value of 0.037±0.020‰. This is the lowest high-precision δ⁵⁶Fe value recorded for a MORB worldwide. This basalt displays an incompatible-element depletion consistent with re-melting beneath the transform fault of mantle source that was depleted during a first melting event, beneath the ridge axis (Wendt et al. 1999). The Fe isotope observation could indicate that its mantle source underwent ⁵⁶Fe depletion after a first melting event. It could alternatively indicate a lower Fe isotope fractionation during re-melting, if the source was depleted of its Fe³⁺, likely producing a relatively reduced melt. These hypotheses are testable, and will be discussed in detail at the conference.