Potassium distribution and isotope composition in the lithospheric mantle in relation to global Earth's reservoirs.

Recent analytical advances have provided means to measure potassium (K) isotopes in various terrestrial materials, but little is known about K distribution and stable isotope composition in the lithospheric mantle because of: (a) common low K abundances, (b) potential contamination and alteration, (c) diversity of mantle rocks and minerals hosting K in different tectonic settings. We report K abundances and δ41K values for well-studied whole-rock (WR) mantle xenoliths (spinel and garnet peridotites and pyroxenites) from mobile belts, a craton, a subduction zone, as well as for K-rich phases (mica, amphibole, silicate glass) and xenolith-bearing volcanic materials (67 samples). The xenolith materials show extremely broad ranges of K content (7 μg/g to 6.6 wt.%) and δ41K (-2.77‰ to 0.62‰). They contrast with the narrow δ41K range for host volcanic materials (-0.53‰ to -0.27‰) and literature data on oceanic basalts (melting products of upwelling asthenosphere: -0.43 ± 0.17‰, 2sd). Amphibole-bearing subduction zone peridotites show the highest WR δ41K values (0.40 to 0.62‰) likely inherited from fluids released into the mantle wedge from subducted oceanic crust. All other WR samples yield negative δ41K: -0.06‰ to -2.77‰ for peridotites and -0.17‰ to -0.52‰ for pyroxenites. The δ41K in K-rich mantle phases range from positive values (0.16 to 0.57‰) for phlogopite in strongly metasomatized peridotites to negative values (-0.27 to -0.94‰) for phlogopite, amphibole and glass pockets from other samples, which cannot be explained by equilibrium inter-mineral fractionation inferred from ab initio calculations. We attribute the broad δ41K variations to (a) isotope fractionation during fluid-rock interaction in the mantle, and (b) distinct sources of K-bearing fluids. Kinetic isotope fractionation during fluid percolation and diffusion is inferred for composite xenoliths (phlogopite and pyroxenite veins in peridotites) that have lower δ41K in the hosts than in the veins due to slower migration of 41K than 39K from the veins (former fluid channels) to host mantle. Overall, the K isotope fractionation in the lithospheric mantle appears to be greater than for magmatic fractionation in the crust. The average δ41K of normal off-craton continental lithospheric mantle calculated from the least altered fertile and lightly metasomatized lherzolites is -0.57 ± 0.28‰ (2sd). This value is within error (though a little lower) of estimates for both continental crust and MORB and OIB mantle sources indicating that these major silicate Earth reservoirs have similar bulk δ41K values, apparently due to low or negligible K isotopic fractionation during their formation by magmatic differentiation and melting. By contrast, K isotopes in modern and fossil subduction zones are fractionated via fluid-related equilibrium and kinetic processes.