Influence of melt composition on Fe, Mn, and Ni partitioning between carbonate-silicate melts and mantle minerals: experiments and applications to the genesis of kimberlites and inclusions in diamonds

Girnis A.V.a, Bulatov V.K.b, Brey G.P.c

a Institute of Geology of Ore Deposits, Petrography, Mineralogy and Geochemistry, Russian Academy of Sciences, Staromonetny 35, Moscow, 119017 Russia

b Vernadsky Institute of Geochemistry and Analytical Chemistry, Russian Academy of Sciences, ul. Kosygina 19, Moscow, 119991 Russia

cInstutut fr Geowissenschaften, J.-W. Goethe Universitt, Altenhferallee 1, D-60438 Frankfurt am Main, Germany


It is commonly believed that the composition of kimberlites and xenoliths in them provide insight into the composition of deep mantle zones. This suggestion is based on the concept of kimberlite formation by low-degree melting of volatile-bearing peridotites in deep lithospheric or asthenospheric mantle zones. In this connection, the high mg numbers of kimberlites and kimberlite-borne mantle xenoliths is regarded as evidence for the strongly depleted character of kimberlite mantle sources (e.g. Becker, LeRoex, 2006; Bernstein et al., 2007). Quantitative interpretation of these data requires the knowledge of element partitioning between primary kimberlite melts and peridotitic minerals. The composition of primary melts formed in a carbonated peridotite under conditions of kimberlite magma genesis is strongly dependent on temperature and ranges from carbonatitic (~5 wt % SiO2) to carbonate-silicate kimberlite-like compositions (30 wt % SiO2). It was shown that the partition coefficients of trace elements are variable within this composition range (Girnis et al., 2006). We explored experimentally the behavior of some siderophile elements during melting of carbonated peridotite at 6-10 GPa and 1200-1700C. The experiments were carried out in a multianvil apparatus using the method of pressure-induced melt segregation (Brey et al., 2008). The starting materials were prepared from natural mantle minerals (SC-1 sample used by Brey et al., 2008) with addition of carbonate, carbonate + water, and carbonate + calcium fluoride. In all cases, CO2 was the major volatile component. Experimental products were analyzed using a JEOL Superprobe 8900 electron microprobe with five wavelength spectrometers. Fluorine was analyzed with the electron microprobe, but water and CO2 were only approximately estimated on the basis of analytical totals and mass balance considerations.

The obtained results showed that the melt composition is strongly variable depending on temperature and the bulk composition of starting peridotite. The lowest temperature melts are dominated by carbonates and contain about 5 wt % SiO2. With increasing temperature, the SiO2 content increases gradually to about 30 wt % at 1600-1700C at 6-10 GPa. Simultaneously, the Ca/Mg ratio of the melt decreases appreciably and the melt approaches the field of kimberlite compositions. The partition coefficients of Fe, Mn, and Ni between melts and minerals are also strongly variable, which is illustrated in Figs. 1 and 2. It is known that the olivine-melt KdFe-Mg is close to 0.3 within a wide range of thermodynamic conditions and melt composition. Our results show that KdFe-Mg is close to or even lower than 0.3 in high-temperature kimberlite-like melts and increases to more than 0.5 in carbonate-dominated melts (Fig. 1). The increase is especially significant in carbonate-fluoride melts containing more than 10 wt % F, where the mg-number of olivine is very close to that of the coexisting melt (i.e. KdFe-Mg approaches 1). The similar effect is observed for Mn partitioning (Fig. 2), i.e. Mn tends to partition into olivine as melt becomes poorer in SiO2. The effect is especially pronounced for F-bearing systems. The partition of Ni is known to be strongly temperature-dependent. Variations in olivine-melt DNi are consistent with the temperature dependence obtained for volatile-free system. Within the analytical errors, no significant effect of CO2 on Ni partitioning was established. Nonetheless, there is also a distinct increase in DNi in F-bearing system. The same tendencies were observed for Fe and Mn partitioning between garnet and melt (Ni in garnet was too low for reliable microprobe analysis).

Fig. 1. Fe-Mg distribution coefficient between olivine and melt obtained in partial melting experiments with natural CO2-free anhydrous peridotites (crosses, data from the literature), carbonate-bearing peridotites with an without additional H2O (filled circles, our data), and carbonate-bearing peridotite with F (triangles, our data).


Fig. 2. Mn partitioning between olivine and partial melts from experiments with mantle peridotites. Symbols are the same as in Fig. 1.


Thus, the obtained results show that the partitioning of Fe, Mn, and Ni between mantle melts and residual minerals is strongly dependent on thermodynamic conditions and the bulk composition of peridotite. In particular, the presence of volatile components significantly affects the behavior of these elements. A transition from CO2-free systems to carbonate-silicate compositions is associated with considerable changes in KdFe-Mg and DMn values. The addition of F significantly increases this effect and makes Fe, Mn, and Ni more compatible in silicate minerals. This dependence must be taken into account during modeling of mantle magmatism, especially at low degrees of melting. The high olivine-melt KdFe-Mg values obtained for carbonate-rich melts suggest that kimberlite melts with high Mg/Fe ratios could be derived from a less depleted source than it was supposed on the basis of KdFe-Mg values for CO2-free silicate liquids. In particular, the presence of very highly magnesian olivines (mg > 0.95) as inclusions in diamond (e.g. Creighton et al., 2008) does not necessarily implies the existence of an ultradepleted source (more depleted than the residue after komatiite extraction), but can result from local effects related to redox reactions in volatile-rich melts. If a carbonate-rich melt is formed in equilibrium with mantle peridotite containing olivine with mg = 0.90, its mg-number will be high (~0.83) owing to the high KdFe-Mg values for carbonate-dominated melts. Then, carbonate components may be lost owing to decarbonation and/or redox reactions (including diamond crystallization). In such a case, the mg value of melt will remain unchanged, but much more magnesian olivine will crystallize from it owing to an increase in KdFe-Mg value (mg = 0.94 for KdFe-Mg = 0.3).



Becker, M., Le Roex, A.P. (2006). Geochemistry of South African on- and off-craton, group I and group II kimberlites: Petrogenesis and source region evolution. Journal of Petrology 47, 673703.

Bernstein S., Kelemen P).B., Hanghoj K. (2007). Consistent olivine Mg# in cratonic mantle reflects Archean mantle melting to the exhaustion of orthopyroxene. Geology 35, 459-462.

Brey, G.P., Bulatov, V.K., Girnis, A.V., Lahaye, Y. (2008). Experimental melting of carbonated peridotite at 610 GPa. Journal of Petrology 49, 797821.

Creighton S., Stachel T., McLean H. (2008) Diamondiferous peridotitic microxenoliths from the Diavik Diamond Mine, NT. Contributions to Mineralogy and Petrology 155, 541-554.

Girnis A. V., Bulatov V. K., Lahaye, Y., Brey G. P. (2006). Partitioning of trace elements between carbonate-silicate melts and mantle minerals: experiment and petrological consequences. Petrology 14, 492-514.

" "