Primary Na contents of kimberlite melts from experimentally determined olivine/melt partition coefficients

Kimberlite rocks are rich in Ca, yet almost Na-free (<0.3 wt% Na2O), raising the question of whether they tap a unique Na-poor but clinopyroxene-saturated mantle source or whether Na is lost en route to the surface. To reconstruct primary Na contents of kimberlitic melts, we quantified Na partitioning between olivine and volatile-rich kimberlitic melts (Note: abstract contains math symbols that cannot be translated here; see PDF). Applying this calibration to magmatic olivine from 22 kimberlites yields reconstructed Na2Omelt of 1.9−8.6 wt% (typically ±0.25 wt%) at 1 GPa, showing that magmatic Na contents are much higher than indicated by bulk rock measurements. The lower range of these reconstructed Na2O values aligns with the predicted 3.5 ± 0.5 wt% of asthenospheric carbonated silicate melts generated from typical lherzolitic mantle at 7 GPa. The higher values correspond to the same 3.5 ± 0.5 wt% if recalculated for olivine crystallization at 2−4 GPa. Because Na2O loss through fluid exsolution during kimberlite ascent cannot exceed 0.5 wt%, the extreme Na loss leading to the observed low bulk-kimberlite values must occur upon solidification when copious amounts of Na-rich fluids are expelled. These findings suggest that kimberlitic melts do not require a peculiar mantle source, but rather carbon-bearing fertile lherzolites that melt at adiabatic conditions, placing kimberlites into a broad continuum with primitive alkaline ocean island melts.