The composition of the bulk silicate Earth (BSE) is the product of planetary accretion, core differentiation and Moon formation. By establishing the composition of the BSE, one can determine the composition of the bulk Earth and by subtraction, calculate the core’s composition. The BSE represents the bulk Earth minus the core, which in today’s terms equals the modern mantle, the continental crust, and the hydrosphere-atmosphere systems. The modern mantle can be framed in terms of two compositionally distinct components, an enriched and a depleted mantle, with the latter as the MORB (mid-ocean ridge basalt) source and the former as the OIB (ocean island basalt) source.

The spectrum of geochemical compositions of Oceanic Island Basalts (OIBs) and their systematic differences from Mid-Ocean Ridge Basalts (MORBs) reveal that the Earth’s mantle is chemically and isotopically heterogeneous. Two main processes, both related to plate tectonics, contribute to the creation of mantle heterogeneities: (1) partial melting generates melts enriched in incompatible elements and leaves a depleted residual rock; and (2) subduction of the oceanic lithosphere injects heterogeneous material at depth, in particular, altered oceanic crust and continental/oceanic sediments. Moreover, delamination and foundering of metasomatized subcontinental lithospheric mantle might have been important in the early Earth history, when plate tectonics did not operate as today. The fate of the subducted plate is still a matter of debate; presumably some of it is stirred by convection and some may segregate at the base of the mantle, in particular the oceanic crust, which is compositionally denser than the pyrolitic mantle. The view of the lower mantle as a “graveyard” of subducted crust prevailed for decades and was supported by the Hofmann and White ( 1982 ) observation that the geochemical fingerprint of most OIB reveals the presence of ancient recycled crust. However, recent geochemical data on short-lived systems ( e.g. 182 Hf→ 182 W has a half-life of 8.9 My) showed that some hotspots, namely Hawaii, Samoa, Iceland and Galápagos, have a negative µ 182 W anomaly. This discovery prompted a change in our view of the deep mantle because anomalies in short-lived systems require additional processes, which include, but are not limited to, the preservation of ‘pockets’ of melt from a primordial magma ocean, and/or chemical reactions between the metallic core and the silicate mantle. Exchanges at the core-mantle boundary would cause a negative µ 182 W anomaly, and might also add 3 He to mantle material later entrained by plumes. It is now clear that some plumes probe the deepest mantle and are highly heterogeneous, as revealed by isotope ratios from long-lived radiogenic systems, noble gases and short-lived isotope systems. Here I will focus on the dynamics of plumes carrying compositional and rheological heterogeneities. This contribution attempts to be pedagogic and multi-disciplinary, spanning from seismology to geochemistry and geodynamics.

The present contribution synthesizes the main petrographic, mineralogical and chemical features of mantle xenoliths uplifted by Phanerozoic lavas. The collections of mantle xenoliths consist predominantly of peridotites but minor pyroxenites are commonly associated. Two main petrogenetic processes are responsible for the features of mantle xenoliths: partial melting and circulation of melts/fluids and associated metasomatic and magmatic processes. Partial melting processes lead to the formation of residual pieces of upper mantle while two main types of mantle metasomatism could be recognized such as LILE enrichment, the first referring to asthenosphere upwelling settings (essentially mantle plumes, rifting zones and asthenosphere window zones) and the second to mantle wedge settings. The AUZ (asthenospheric upwelling zones) metasomatism is essentially related to the migration of more or less CO 2 -rich alkaline silicate melts and associated fluids while the MWZ (mantle wedge zones) metasomatism is associated with the activity of hydrated liquids (fluids) commonly SiO 2 -rich.

Batch melting, fractional melting, continuous melting and two-porosity melting models have been used widely in geochemical studies of trace element fractionation during mantle melting. These simple melting models were developed for melting an homogeneous mantle source. Here we revisit and further develop these melting models in the context of decompression melting of a two-lithology mantle. Each lithology has its own source composition and melting parameters. During decompression melting, melt and solid flow vertically in the melting column. Part of the melt produced in one lithology is transferred to the other lithology at a prescribed rate. We use a set of conservation equations to solve for melt and solid mass fluxes, extent of melting and concentrations of a trace element in interstitial melt and aggregated melt in each lithology and mixed-column melt between the two lithologies. We uncover conditions under which batch melting, fractional melting, continuous melting and two-porosity melting models are realized during decompression melting through four case studies. We show that porosity in the continuous melting model varies along the melting column during decompression melting, contrary to what was assumed in its original development. We unify the batch melting, fractional melting, continuous melting and two-porosity melting models through a two-lithology melting model for decompression melting in a two-lithology mantle column. We discuss basic features of the two-lithology melting model through worked examples. We show that it is possible to produce partial and well-mixed melts with a range of REE patterns, from LREE depleted to LREE enriched, similar to those observed in mid-ocean ridge basalts by decompression melting of a two-lithology mantle.

The footprints of mafic melts travelling from the depths to the surface are abundant in the mantle section of ophiolites. They constitute an important source of information about the melt migration mechanisms and related petrological processes in the shallowest part of the mantle beneath former oceanic spreading centres. In the field, these so-called ‘melt migration structures’ attract attention when they consist of mineral assemblages contrasting with that of their host peridotite. They therefore record a particular moment in the migration history: when the melt becomes out of equilibrium with the peridotite and causes a reaction impacting its modal composition, and/or when a temperature drop initiates the crystallization of the melt. The existence of cryptic effects of migration revealed by geochemical data shows that melts do not always leave a trail visible in the field. Although incomplete and patchy, the melt migration structures preserved in ophiolites are witnesses of processes that do actually occur in nature, which constitutes an invaluable support to the interpretation of geophysical data and inescapable constraints for numerical simulations and models of chemical geodynamics. Here we show how field observations and related petrological and geochemical studies allow us to propose answers to fundamental questions such as these: At which temperature is porous flow superseded by dyking? What are the factors governing melt trajectories? What is the nature of the ‘universal solvent’ initiating infiltration melting and making channelized porous flow the most common mode of transport of magmas through a peridotite matrix regardless the tectonic setting? A fundamental message delivered by ophiolites is that the shallow mantle behaves as a particularly efficient reactive filter between the depths and the surface of the Earth. Unexpectedly, the reactions occurring there are enhanced by the hybridization between mafic melts and a hydrous component, whatever its origin ( i.e. magmatic vs. hydrothermal). This hybridization triggers out of equilibrium reactions, leading to the formation of exotic lithologies, including metallic ores, and impacting the global geochemical cycle of a whole range of chemical elements.

In this chapter, we evaluate how the incorporation of H 2 O as a thermodynamic component influences phase relations in a peridotite composition. This component – present either in the form of hydrous minerals, aqueous fluids and hydrous melts, or as a structurally-bonded trace element at defect sites of nominally anhydrous minerals (NAMs) – may influence upper-mantle rheology in diverse ways. By presenting various natural cases, we identify key incorporation mechanisms and assess their role in the microstructural evolution of ultramafic rocks at different depths in the Earth’s interior. These data suggest that the influence of either aqueous fluids or hydrous melts on rheology out-matches that of NAMs or stable hydrous phases across much of the lithospheric mantle. Consequently, future research is expected to shift towards a better understanding of the transient conditions in the lithosphere that control the availability and transport of aqueous fluids and hydrous melts. These transient conditions are likely to play a more dominant role than the sole ability of hydrous defects in NAMs – a role that is currently less well-constrained experimentally – in controlling the ductile deformation of the upper mantle.

Temperature-dependent equilibrium partitioning of elements between different mineral (or melt/glass) phases forms the basis of geothermometry. In natural rock systems it is necessary to determine whether equilibrium partitioning of a given element was obtained between two phases before calculating temperatures using the tool. With the improvement of spatial resolution of analytical tools and our understanding of solidstate kinetics it has become clear that compositional heterogeneities on different scales exist in mantle rocks because of incomplete equilibration, and a kinetic evaluation is necessary before application of geothermometers. This work summarizes the kinetic situations that may arise and provides some guidelines and criteria for testing whether partitioning equilibrium was obtained. A suite of dunites and harzburgites from an ophiolite suite in the Himalaya (Xigaze, Tibet) is used to illustrate the application of some of these concepts. It is shown that when compositions used for geothermometry are chosen bearing these kinetic considerations in mind, a systematic pattern of freezing temperatures is obtained from the geothermometers. These data provide insights into the cooling histories of these rocks with complex, multistage (e.g. melt percolation) histories. Some potential pitfalls for geospeedometry are also illustrated along the way.

Full article available in PDF version.