DISEQUILIBRIUM INTER-MINERAL MAGNESIUM ISOTOPE PARTITIONING IN GARNET-BEARING PYROXENITES

Mantle minerals are generally expected to be in isotopic equilibrium given the rapid diffusion rates of isotopes at high temperature. This is particularly the case for magnesium (Mg) due to the strong buffer capacity of mantle minerals for Mg. However, disequilibrium inter-mineral Mg isotope fractionation has recently been documented in rocks that were subjected to mantle metasomatism, such as wehrlites [1]. In an effort to further constrain the magnitude and mechanism of inter-mineral Mg isotope fractionation in the mantle, we performed high-precision Mg isotopic measurements on a suite of garnet-bearing pyroxenites from the North China Craton, which were interpreted as the direct products of melt-peridotite interaction [2]. Our data reveal broad Mg isotopic equilibrium among olivine, clinopyroxene, and orthopyroxene, with a normal-mantle like δ²⁶Mg (-0.25 ± 0.07 [3]) for them, whereas coexisting garnets give a systematically lighter δ²⁶Mg by 0.11 to 0.80‰. This observation is consistent with the fact that Mg in garnet has a higher coordination number (8 vs. 6), thus the weaker Mg-O bonds formed will preferentially take light Mg isotopes [4]. Notably, the highly variable offsets in δ²⁶Mg between garnet and olivine/pyroxene do not seem to correlate with their corresponding equilibrium temperatures and are plotting off the equilibrium fractionation lines. Therefore, it more likely reflects disequilibrium partitioning of Mg isotopes during incomplete peridotite-melt interaction that gave rise to these pyroxenites. Our results reinforce previous suggestion that significant kinetic Mg isotope fractionation could occur at mantle temperature, and highlight the role of melt migration in generating Mg isotopic disequilibrium among coexisting minerals, as well as the usefulness of Mg isotopes in tracing such processes.

[1] Xiao et al. (2013) GCA 115, 241-261.

[2] Liu et al. (2005) EPSL 234, 39-57.

[3] Teng et al. (2010) GCA 74, 4150-4166.

[4] Huang et al. (2013) EPSL 367, 61-70.