Kimberlite petrogenesis: Insights from clinopyroxene-melt partitioning experiments at 6 GPa in the CaO-MgO-Al2O3 -SiO2-CO2 system

Shantanu Keshav, Alexandre Corgne, Gudmundur H. Gudfinnsson, Michael Bizimis, William F. McDonough, Yingwei Fei

Research output: Contribution to journalArticlepeer-review

56 Citations (Scopus)


In this experimental study, we examine the mineral-melt partitioning of major and trace elements between clinopyroxene and CO2-rich kimberlitic melts at a pressure of 6 GPa and temperatures of 1410°C and 1430°C. The melts produced contain ∼ 28 wt% dissolved CO2, and are saturated with olivine and clinopyroxene. To assess the effects of temperature, crystal and melt compositions on trace element partitioning, experiments were performed in the model CaO-MgO-Al2O3-SiO2-CO2 system. Our results reveal that all the elements studied, except Al, Mg, Si, and Ga, are incompatible in clinopyroxene. Partition coefficients show a considerable range in magnitude, from ∼ 10-3 for DU and DBa to ∼ 2.5 for DSi. The two experimental runs show similar overall partitioning patterns with the D values being lower at 1430°C. Rare earth elements display a wide range of partition coefficients, DLa (0.012-0.026) being approximately one order of magnitude lower than DLu (0.18-0.23). Partition coefficients for the 2+ and 3+ cations entering the M2-site exhibit a near-parabolic dependence on radius of the incorporated cations as predicted from the lattice strain model. This underlines the contribution made by the crystal structure toward controlling the distribution of trace elements. Using data obtained in this study combined with that in the published literature, we also discuss the effects that other important parameters, namely, melt composition, pressure, and temperature, could have on partitioning. Our partition coefficients have been used to model the generation of the Group I (GI) kimberlites from South Africa. The numerical modeling shows that kimberlitic melts can be produced by ∼0.5% melting of a MORB-type depleted source that has been enriched by small-degree melts originating from a similar depleted source. This result suggests that the source of GI kimberlites may be located at the lithosphere-asthenosphere transition. Percolation of small degree melts from the asthenosphere would essentially create a metasomatic horizon near the bottom of the non-convecting sublithospheric mantle. Accumulation of such small degree melts together with the presence of volatiles and conductive heating would trigger melting of the ambient mantle and subsequently lead to eruption of kimberlitic melts. Additionally, our model shows that the GI source can be generated by metasomatism of a 2 Ga old MORB source ca. 1 Ga ago. Assuming that MORB-type mantle is the most depleted source of magmas on earth, then this is the oldest age at which the GI source could have existed. However, this age most likely reflects the average age of a series of metasomatic events than that of a single event.

Original languageEnglish
Pages (from-to)2829-2845
Number of pages17
JournalGeochimica et Cosmochimica Acta
Issue number11
Publication statusPublished - 2005 Jun 1


Dive into the research topics of 'Kimberlite petrogenesis: Insights from clinopyroxene-melt partitioning experiments at 6 GPa in the CaO-MgO-Al2O3 -SiO2-CO2 system'. Together they form a unique fingerprint.

Cite this