The Earth is believed to have formed with a relatively homogeneous composition, which fractionated into the dominantly silicate crust and mantle and the metallic core. A principle objective in geochemistry is to understand how these separate reservoirs came into existence and how elements were divided among these regions and their subdivisions. Fractionation of the Earth occurred during periods of melting where either crystals and liquids or immiscible liquids separated to form distinct domains. By reproducing melting events in the laboratory and measuring element partitioning between the separating phases, the conditions at which reservoirs formed and the likely compositions of the phases during formation can be determined. Such measurements are particularly important for reservoirs such as the Earth's core, which is inaccessible to direct chemical analysis.
Highly siderophile elements (HSE), such as Pt, Rh, Ir, Re, and Os, are so-called because they favour existence in the metallic state over wide ranges of conditions. During core separation these elements should have been stripped from the silicate mantle and partitioned into separating bodies of metallic liquid. A number of experimental studies have been designed to test the effectiveness of HSE extraction from the mantle. If the present day HSE concentrations in the mantle are higher than predicted from such experiments, this would support a more complex accretion model, where material was added back to the mantle after core separation had mainly finished. This theory of a secondary meteoritic influx to the silicate portion of the Earth is often termed the "late veneer" model.
High temperatures and slow convection in the Earth's mantle mean that element partitioning between coexisting phases can, in general, be handled with a thermodynamic approach. Laboratory measurements of thermodynamic quantities, such as molar volume and chemical activity, can be used to extrapolate element partitioning data to wider, inaccessible ranges of temperature, pressure and composition. There are, however, many partitioning situations where equilibrium may not be attained. During core formation, for example, segregating bodies of metal may be large enough and moving rapidly enough for them not to fully equilibrate with the surrounding mantle. Similarly when reactions are confined to the solid state there is evidence from natural samples that disequilibrium can be sustained to high grades of metamorphism, if a fluid phase is absent. In these situations experimental measurements of element diffusion rates can give crucial information on the likely chemical compositions of resulting reservoirs and the phases therein.