Constitution of the deep transition zone and lower mantle shown by diamonds and their inclusions

Diamonds from the Sao Luiz alluvial deposit, Brazil, have been found to occlude syngenetic inclusions whose associations are evidence for formation in the mantle transition zone and lower mantle (Wilding, 1990; Harte and Harris, 1994). These diamonds represent the most extensive sample of deep mantle available to date, and have been subjected to detailed study. Five principal associations are identified. One association involves orange garnet inclusions (with diopside and pyrrhotite) which contain a significant pyroxene component in solid-solution (majorite component) indicating formation in the transition zone (Akaogi and Akimoto, 1977). Trends of major element composition against Si content are most consistent with formation within the deepest regions of the transition zone where equilibrium with perovskite structured CaSiO3 (CaSi-Pvk) is envisaged (Irifune and Ringwood, 1987). The remaining associations all involve MgO - FeO (fPer) and are also believed to have been in equilibrium with CaSiO3 composition inclusions. The association of fPer and (Mg, Fe)Si03 (LM I) is understood to have formed at pressures of >24 GPa (Yagi et al., 1978), within the lower mantle, where (Mg, Fe)SiO3 adopts a perovskite structure (MgSi-Pvk) at pressures above the breakdown of (Mg, Fe)2SiO4 ringwoodite. Indeed, all the broadly pyroxene composition phases recovered in association with fPer are envisaged to have formed with perovskite structures. The LM I association also includes grains of broadly pyrope-almandine composition with high Fe3+ content (Fe3+/EFe=-0.7) and very low Ca (<0.15 wt% CaO) and depleted rare earth element (REE) concentrations consistent with equilibrium with REE-phyllic CaSi-Pvk. This new mineral is shown to adopt a tetragonal I :F -2d structure and is referred to provisionally as 'TAPP' (tetragonal almandine-pyrope phase). Given the propensity for MgSi-Pvk to adopt the entirety of the likely lower mantle A12O3 budget within its structure at depths over 820 km (e. g. Kesson et al., 1995), and the stability of an A1203-involving association at depths of 720-820km (Irifune et al., 1996), TAPP is believed to form in aluminous bulk compositions in the depth region, 670-720km. A deeper association of fPer, aluminous and Fe 3'-rich MgSi-Pvk and A1203 (ruby) from Sao Luiz diamonds forms a third (LM H) association. The remaining two associations have characteristics indicative of formation in the deepest regions of the transition zone. An association (LM III) of low Ca-garnet with a small majoritic component, a previously unrecorded C2/c structured Al-Ca-Na-Few-rich magnesium silicate (with 11,5 and 6 wt% A1203, CaO and Na2O respectively) and fPer is reported. Trace element compositions of this garnet are found to be transitional between majoritic garnet (Harte, 1992) and TAPP. The final association, found in a single diamond involves a (Mg, Fe)2SiO4 composition inclusion, fPer and TAPP (UM/LM association), and is suggestive of formation within the range 460-720km depending on bulk Constitution of the deep transition zone and lower mantle shown by diamonds and their inclusions 111 composition (Jeanloz and Thompson, 1983). Also identified from Sao Luiz is the first recorded sapphire inclusion in diamond. Change in cell parameters on release of two fPer inclusions (one from Guinea, West Africa) have been measured and interpreted on the basis of expected mantle geotherms and physical properties of compressibility and expansivity. Depths of formation of -300km are inferred which, on correction due to the fractured and plastically deformed nature of the diamond hosts, extend to within the lower mantle. The very low Fe 3+ content of fPer and the large Fe 3+ content of aluminous MgSi-Pvk inclusions additionally support formation at high pressure (McCammon et al., 1995 and McCammon, 1997). Furthermore, the presence of significant quantities of magnesioferrite as inclusions in many fPer inclusions is consistent with the high Fe3+ content of associated phases and indicates relatively oxidised conditions of formation. Partitioning of Fe, Ni and Mg between fPer and MgSi-Pvk is indicative of high temperature (>2000K) within the lower mantle which suggests a steep thermal gradient at 670km and hence a thermal boundary layer between the upper and lower mantle. This observation, in addition to indications from associations of a compositional distinction between upper mantle and lower mantle, supports separate regimes of mantle convection. The diamonds themselves show cathodoluminescencep atterns indicative of a complex interplay of growth and resorption. Transition zone stones show a range in nitrogen content from <15 to 311ppm, and are highly aggregated indicating a long, high temperature history. Lower mantle stones are even more deficient in nitrogen (mostly Type II diamond), and show a very tight clustering of 513C composition around -5%o. Given ranges of up to 9%o within single stones, precipitation under fluctuating conditions within a homogeneous reservoir is concluded. Values for 815N of -6 and -5.2%o have been obtained for an upper / lower mantle boundary sourced stone. Thermoelastic modelling is applied to a variety of deep mantle phases and it is concluded that, with a thermal boundary between upper and lower mantle, there exists a narrow depth region just below 670km where many phases, (particularly diamond) are gravitationally stabilised. Diamond moving within the circulatory system of the lower mantle will, therefore, tend to pond in this region. Exhumation from the deep mantle is believed to have been relatively swift due to the lack of: reequilibration of composite grains; complete exsolution of majoritic garnet; and recombination of magnesioferrite with Per. A regime of transportation by upwelling mantle plume is envisaged. The dominance within thin cratonic areas amongst world-wide locations of deep mantle diamonds is also discussed. This observation is interpreted in terms of thin cratonic areas being suitably reduced to stabilise diamond at shallow depths, unlike in oceanic settings where diamond bums to form C02- Additionally, the crust in thin cratonic regions is not suitable for formation of lithospheric diamond and so the deep population of stones is not outnumbered by shallow sourced diamonds.