The relationship of alkaline melts to hydrous pyroxenites in the mantle

Mantle peridotite has long been assumed to be the main source lithology for mafic magmas found at the Earth’s surface. However, even extremely low degree melting (<<1 wt.% melt fraction) of peridotite at 3.0 GPa generates melt with K2O content < 1.0 wt.%, far lower than the high K2O contents of some mafic magmas in intraplate and collisional settings. A metasomatised mantle formed through interaction with fluids or melts that is enriched in trace and major element compositions has been invoked as the source for many alkaline magmas. Modal metasomatism introduces various phases into the lithospheric mantle. The very common phases are clinopyroxene, orthopyroxene, phlogopite, amphibole, apatite, garnet, and carbonate, as evidenced by the natural xenoliths and exposed mantle massifs. These minerals commonly occur as hydrous pyroxenitic veins (phlogopite- and/or amphibole- bearing pyroxenite ± carbonate and others), as evidenced by natural xenoliths and exposed mantle massifs. However, we have little knowledge of the melt compositions derived from these hydrous pyroxenitic assemblages.

In this thesis, the melting reactions, conditions, and melt compositions of two hydrous pyroxenite rocks were investigated in high-pressure experiments: (1) a phlogopite websterite consisting of 33.3 wt.% phlogopite, 33.3 wt.% orthopyroxene and 33.3 wt.% clinopyroxene at 1.5 to 4.5 GPa, 1050 to 1500 ℃; and (2) a garnet phlogopite websterite consisting of 20 wt.% phlogopite, 35 wt.% orthopyroxene and 35 wt.% clinopyroxene at 3.0 to 4.0 GPa, and 1100 to 1500 ℃. In the 1.5 GPa experiments, olivine is a peritectic phase, resulting in SiO2-saturated melts, whereas at pressures ≥ 3.0 GPa, orthopyroxene replace olivine as the peritectic phase, resulting in SiO2-undersaturated melts. The melt compositions show remarkable resemblance to commonly occurring ultrapotassic alkaline magmas that display transitional geochemical characteristics between those typical of orogenic and intraplate settings, suggesting that phlogopite websterite is probably a common source lithology for these ultrapotassic rocks.

Melts of the garnet phlogopite websterite particularly resemble the leucitites of New South Wales, eastern Australia, an area typical for ultrapotassic rocks of the transitional group, for which the source had remained unclear. The age-progressive leucitite occurrences constitute the longest continental hotspot track on Earth. Eastern Australia has irregular lithospheric thickness, but the leucitites regularly occur in areas with lithospheric thickness between 120 and 140 km, which may indicate some interesting processes associated with hotspot–continental lithosphere interaction. In this thesis, a model of ponded melt interaction invokes garnet phlogopite websterite in the areas with lithospheric thickness between 120 and 140 km, with garnet phlogopite websterite remelted by the passage of the plume.

I also investigated the early Miocene melilitites of the northeastern Tibetan Plateau, concluding that the melilitite melts are probably derived from a phlogopite-bearing carbonated source in the lithosphere. This agrees with enriched Sr–Nd–Pb–Os isotopes and the occurrence of phlogopite and calcite in the xenoliths. Geothermobarometer of silica-undersaturated melts and clinopyroxene–liquid pairs reveal that the primitive melilititic magma was generated at ~4.3 GPa (~142 km) and 1330–1464 ℃. This is consistent with the deepest xenoliths (~3.9 GPa) entrained in the melilitites and with experimental constraints for melilitite melt origin (≥ 3.5 GPa). The results reveal that the lithosphere was around 30 km thicker at the time of eruption than the lithospheric thickness today (~110 km), revealed by S-wave receiver functions. We interpret this as petrological confirmation of lithosphere erosion by asthenospheric flow that has occurred since the early Miocene.