Earth’s lithospheric mantle is dominated (usually >90% vol.%) by a nominally volatile-free mineral paragenesis with peridotitic composition. The remaining lithologies are typically hydrous and/or carbonate-bearing assemblages (the latter mostly stable at P >2 GPa), representing the products of mantle metasomatism with, or without, involvement of subduction. Despite their importance, the modal compositions of these metasomes are poorly constrained, probably containing hydrous phases (amphibole and/or phlogopite), coupled with accessory minerals, such as apatite, ilmenite, rutile and/or carbonates. These assemblages are usually considered to be the source of unusual magmas, in particular ultrapotassic compositions, especially kamafugites (K2O-rich, commonly ultracalcic, basic/ultrabasic lithologies). To evaluate if partial melting of such metasomes could effectively produce kamafugites, we performed partial melting experiments at 2.7 and 5 GPa, and 1200°C to 1550°C on clinopyroxenites variably enriched in phlogopite, olivine and accessory phases (apatite, oxides, titanite), and on a clinopyroxene glimmerite (with apatite and magnetite).

At low degrees of melting, the glasses show extremely high TiO2 (<15 wt.%), CaO (<18 wt.%) and P2O5 (<7.4 wt.%), coupled with low SiO2 (>21.6 wt.%), as accessory minerals are the principal contributors in the melting reactions. Although there are no known natural counterparts for these low-degree melt compositions, they might play a key role in re-fertilisation events in the upper mantle. At increased degrees of partial melting (∼50 to ∼90%), the experimental melts approach the compositions of silica-poor, potassic/ultrapotassic and ultracalcic rocks. Indeed, experimental-produced glasses share several geochemical similarities with natural kamafugites, partially overlapping for most of the major oxides. Clinopyroxene- and phlogopite-rich lithologies, variably enriched in olivine and accessory phases (apatite, oxides, titanite) probably occur as veins pervading the lithospheric peridotitic matrix, and their partial melting, especially at high degrees, may be a plausible explanation for the genesis of SiO2-poor, K2O- and CaO-rich compositions.

The hydrogen isotope composition of the mantle provides insight into the advection, melting and metasomatism of the mantle, and the Earth’s water cycle. Because of sampling challenges, most estimates for the hydrogen composition of the mantle have been derived from basaltic glasses rather than from direct samples of the mantle. We present the first direct measurement of D/H ratios in nominally anhydrous orthopyroxene and olivine inclusions hosted within diamonds. The diamond hosts were recovered from placer deposits in Guyana, South America. The diamond-hosted inclusions preserve a population with δD of –31 ± 59‰ (normalised to Vienna Standard Mean Ocean Water). This value is D-enriched relative to that expected for a homogeneous upper mantle and is similar to values measured for arc basalts. We introduce three explanations for the observed D-enrichment in the anhydrous silicate inclusions. The hydrogen isotopic composition of the mantle might simply be heterogeneous in space and/or time. The measured D-enrichment could instead preserve local heterogeneity produced by fluids genetically related to the crystallisation of the diamond hosts. Local and mantle-wide processes might also have operated in concert, and overprint each other. Diamond-hosted silicate inclusions preserve an underexplored record of Earth’s water cycle.

Anomalously high amounts of orthopyroxene are an unexpected feature of some depleted peridotite xenoliths from the cratonic lithosphere. This investigation presents new petrographic, geochemical and thermodynamic modelling data for silica-rich spinel harzburgites and one garnet–spinel harzburgite from the Kaapvaal Craton, South Africa, to constrain the pressure–temperature (P–T) conditions of orthopyroxene formation. At the time of sampling by the host kimberlite, the garnet–spinel harzburgite resided on the geotherm at 25 kbar, 765°C, wheras the compositionally similar garnet-free samples probably equilibrated at broadly comparable P–T conditions. The samples are subdivided into ‘bronzite type’ peridotites characterised by orthopyroxene with micrometre scale exsolutions of spinel ± clinopyroxene, and ‘normal type’ peridotites in which orthopyroxene does not contain visible exsolved phases. The presence of these exsolved phases indicates unmixing from a more Al-Cr-(Ca)-rich precursor orthopyroxene i.e. an original orthopyroxene that precipitated from, and was, in equilibrium with the Si-enriching agent at high temperature, before cooling and undergoing sub-solidus re-equilibration to form the exsolved orthopyroxene we observe now. Major and trace element compositions, geochemical interpretation and thermodynamic modelling of the samples show that the precursor orthopyroxene coexisted with melt at high pressure and temperature (~40–50 kbar, 1600–1700°C) within the garnet stability field. We consider these to be the conditions of orthopyroxene formation, and thus silica enrichment. These conditions, together with the refractory composition of the host peridotite (olivine forsterite contents > 92) suggest that silica enrichment is a secondary process potentially resulting from interaction with komatiitic melts. Our observations demonstrate that orthopyroxene enrichment in spinel-facies peridotites is not limited to low-pressure processes but can also occur via high-pressure melt–rock interaction, offering new insights into the thermal and chemical evolution of the sub-cratonic lithospheric mantle.

Silica enrichment in cratonic peridotites, expressed as high modal contents of orthopyroxene is a common yet enigmatic feature of Archean lithospheric mantle. Less widely recognised, silica depletion, expressed as high olivine content is also present in some locations. Although high-pressure melting alone cannot account for the full range of observed silica contents, open-system interaction between mantle melts and lithospheric peridotite offers a viable alternative. We present results from high-pressure (5 GPa), high-temperature (1600–1690°C) multi-anvil experiments designed to investigate the reaction between Al-depleted komatiitic melts and both fertile and moderately depleted peridotite. Using hybrid and reaction-couple experimental configurations, we track mineralogical and compositional changes across controlled thermal gradients within the experimental capsules. Modal and compositional data reveal a two-stage reaction process. Stage 1 involves high-temperature olivine precipitation from primitive Al-depleted komatiitic melts interacting with peridotite to produce olivine-rich residues with high MgO/SiO2 and low Al2O3/ SiO2. As a result of olivine removal, the coexisting melt becomes enriched in SiO2. In stage 2, the now silica-enriched komatiite melt reacts with residual olivine (±clinopyroxene and garnet) to form orthopyroxene at moderate temperature, increasing the bulk SiO2 content. Residue bulk compositions from olivine-rich zones resemble high-Mg# garnet dunites from Archean lithosphere, whereas orthopyroxene-rich zones are analogues to natural silica-rich cratonic peridotites. Coexisting melts evolve from komatiitic toward picritic compositions as the reaction progresses. We infer that reactive porous flow of komatiitic melts through depleted lithosphere can simultaneously generate silica-depleted and silica-enriched, refractory residues and diversify melt compositions, providing a process-based framework for the chemical and textural diversity of cratonic mantle.

The viscosity (η) of a geofluid dictates its roles within the Earth, such as migration in subduction zones and volcanism. High pressures and temperatures at depth influence visocity (η). The falling sphere method is effective to measure viscosity at high-pressures for geofluids that form solids at ambient temperature. A typically metallic sphere is placed atop a solid sample, which is then compressed to high-pressure and later melted by heating. This method is more challenging for geofluids that do not form solids at ambient temperature. A diamond-anvil-cell (DAC) is often used to contain such geofluids, but requires a small sample chamber and that the sphere either falls parallel to the diamond culet faces or rolls along one face. This geometry produces complicated drag effects on a sphere fall and additional frictional forces for a roll. The sphere may also adhere to chamber surfaces, preventing its fall/roll. To circumvent these issues, in this study, we quantify the viscosity of a geofluid (H2O) at pressures <2.5 GPa using the Brownian motions of suspended particles in DAC. Previous high-pressure efforts used particles of polystyrene, which are unstable at ≥300°C, or silica, but only at ambient temperatures. Such temperatures are relatively low for hydrothermal to supercritical geofluids. We tested quartz particles (∼1–2 µm diameter) with heating, as quartz does not significantly dissolve/melt until ≥600°C. Although three times denser than water, the particles remained suspended and displayed Brownian motions for long timescales at temperatures ≤200°C. The measured viscosities are relatively high due to drag from the culets and particle–particle interactions. Regardless, our measured pressure-effect on viscosity shows excellent agreement with the standard reference for water. After correcting for the drag, the η are very low (< 2 mPa s) highlighting that H2O-rich geofluids should be highly mobile at depth.