Journal of Petrology Advance Access originally published online on January 22, 2008
Journal of Petrology 2008 49(4):797-821; doi:10.1093/petrology/egn002
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Experimental Melting of Carbonated Peridotite at 6–10 GPa
1Institut Für Geowissenschaften, J. W. Goethe-Universität, Altenhöferallee 1, D-60438 Frankfurt AM Main, Germany
2Vernadsky Institute of Geochemistry and Analytical Chemistry, Russian Academy of Sciences, UL. Kosygina 19, Moscow, 119991 Russia
3Institute for Geology of Ore Deposits, Petrography, Mineralogy and Geochemistry, Russian Academy of Sciences, Staromonetny 35, Moscow, 119017 Russia
RECEIVED FEBRUARY 12, 2007; ACCEPTED JANUARY 3, 2008
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL METHODSEXPERIMENTAL RESULTSDISCUSSIONREFERENCES |
|---|
Partial melting of magnesite-bearing peridotites was studied at 6–10 GPa and 1300–1700°C. Experiments were performed in a multianvil apparatus using natural mineral mixes as starting material placed into olivine containers and sealed in Pt capsules. Partial melts originated within the peridotite layer, migrated outside the olivine container and formed pools of quenched melts along the wall of the Pt capsule. This allowed the analysis of even small melt fractions. Iron loss was not a problem, because the platinum near the olivine container became saturated in Fe as a result of the reaction Fe2SiO4Ol = FeFe–Pt alloy + FeSiO3Opx + O2. This reaction led to a gradual increase in oxygen fugacity within the capsules as expressed, for example, in high Fe3+ in garnet. Carbonatitic to kimberlite-like melts were obtained that coexist with olivine + orthopyroxene + garnet ± clinopyroxene ± magnesite depending on P–T conditions. Kinetic experiments and a comparison of the chemistry of phases occasionally grown within the melt pools with those in the residual peridotite allowed us to conclude that the melts had approached equilibrium with peridotite. Melts in equilibrium with a magnesite-bearing garnet lherzolite are rich in CaO (20–25 wt %) at all pressures and show rather low MgO and SiO2 contents (20 and 10 wt %, respectively). Melts in equilibrium with a magnesite-bearing garnet harzburgite are richer in SiO2 and MgO. The contents of these oxides increase with temperature, whereas the CaO content becomes lower. Melts from magnesite-free experiments are richer in SiO2, but remain silicocarbonatitic. Partitioning of trace elements between melt and garnet was studied in several experiments at 6 and 10 GPa. The melts are very rich in incompatible elements, including large ion lithophile elements (LILE), Nb, Ta and light rare earth elements. Relative to the residual peridotite, the melts show no significant depletion in high field strength elements over LILE. We conclude from the major and trace element characteristics of our experimental melts that primitive kimberlites cannot be a direct product of single-stage melting of an asthenospheric mantle. They rather must be derived from a previously depleted and re-enriched mantle peridotite.
KEY WORDS: multianvil; carbonatite melt; peridotite; kimberlite; element partitioning
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL METHODSEXPERIMENTAL RESULTSDISCUSSIONREFERENCES |
|---|
Although carbon is a trace component in the Earth's mantle, it plays a major role in the genesis of alkaline magmas and in mantle metasomatism. The presence of carbonates in mantle lithologies is apparent through the occurrence of calcite and magnesite in mantle xenoliths and as inclusions in diamond (e.g. Smith, 1987
; Wang et al., 1996; Leost et al., 2003). Inclusions of carbonatitic melt in cuboid diamond crystals were reported by Zedgenizov et al. (2004). Some mineralogical and geochemical features of the mantle and of mantle-derived magmas can be explained by the migration of small portions of carbonate-rich melts (Yaxley et al., 1991, 1998; Ionov et al., 1993, 1996; Rudnick et al., 1993). Experiments in simple and complex systems (e.g. Brey & Green, 1975; Wyllie & Huang, 1976; Eggler, 1978) showed that the presence of carbonate phases depresses the solidus temperatures of mantle peridotites by more than 200°C at 3 GPa and shifts partial melts to highly undersaturated compositions. In experiments up to 8 GPa Dalton & Presnall (1998a, 1998b) and Gudfinnsson & Presnall (2005) observed a gradual change in the composition of partial melt in the CaO–MgO–Al2O3–SiO2–CO2 system from carbonatitic to kimberlitic with increasing pressures and temperatures. The solidus of carbonate-bearing natural peridotites was determined by Falloon & Green (1989, 1990) up to 3·5 GPa and by Dasgupta & Hirschmann (2006) from 3 to 10 GPa. Dasgupta & Hirschmann (2007) determined the composition of carbonatitic melts near the solidus of natural peridotite with CO2 at 6·6 GPa and 1245°C. The composition of near-solidus melt in CO2- and H2O-bearing mantle peridotite was shown to be dolomitic by Wallace & Green (1988) and Thibault et al. (1992). Because carbonatitic melts are highly mobile in mantle peridotite at melt fractions as low as 0·01–1· 0% (Hunter & McKenzie, 1989; Watson et al., 1990) they may be potent agents of mantle metasomatism (Green & Wallace, 1988).
There is very limited experimental evidence on the composition of melts from natural, anhydrous carbonated peridotite. Hirose (1997) studied partial melting of CO2-bearing natural peridotite at 3 GPa using the diamond trap method. The resulting melts were carbonatitic at 1350°C and melilititic at 1400–1450°C. Dasgupta et al. (2007) explored partial melting of natural lherzolite compositions with 1·0 and 2·5 wt % CO2 at 3·0 GPa. The melts evolved with increasing temperature from carbonatitic (<10 wt % CO2) near the solidus to carbonated silicate compositions. One reason for the paucity of relevant experimental data is the difficulty of analysing small fractions of interstitial melt, which forms coarse-grained quench aggregates. To overcome this problem, low-degree melts can be segregated into various traps including diamond or vitreous carbon sphere aggregates (Hirose & Kushiro, 1993; Baker & Stolper, 1994; Wasylenki et al., 2003) and microcracks in graphite capsules (Laporte et al., 2004). However, these methods are not very practicable within the diamond stability field. Hirschmann & Dasgupta (2007) proposed a modified sandwich method to determine the equilibrium composition of low-degree partial melts, and Dasgupta & Hischmann (2007) reported its successful application to the incipient melting of carbonated peridotite.
In this study we used a new method that was serendipitously discovered during the experimental investigation of mineral equilibria in ‘dry’ natural peridotites at pressures up to 10 GPa (Brey et al., 2008). These experiments were carried out in a multianvil apparatus using cylindrical containers from natural San Carlos olivine sealed within Pt capsules. Although the peridotitic starting mixtures within the olivine capsules were nominally devoid of carbon, small amounts of quenched CO2-rich melt were found between the olivine container and the Pt wall. The source of carbon remained unclear. It may stem from material adsorbed on the platinum and olivine capsule, fluid inclusions in olivine, carbon used for welding, etc. The melt originated within the peridotite layer and migrated into the interstices between the olivine capsule and the Pt. While the melt was still migrating the Pt wall became saturated in Fe as a result of olivine–Pt interaction, which strongly reduced Fe loss from the melt. The melt pools between olivine and Pt were up to 50 µm thick and could be readily analysed with an electron microprobe using a defocused beam (Fig. 1a).
|
In this study we used the advantages of such behaviour of carbonate-rich melt to investigate melts coexisting with carbonated peridotites at pressures of 6–10 GPa. Melts and minerals were analysed with an electron microprobe and by laser ablation inductively coupled plasma mass spectrometry (LA ICP MS). Our experimental strategy was different from that of Dasgupta & Hischmann (2007
) and Dasgupta et al. (2007) in that we did not simulate partial melting of a certain peridotite composition but determined melt composition in equilibrium with a range of polymineralic peridotite assemblages at or near magnesite saturation. The bulk composition of the peridotite was not strictly fixed because of interaction with the olivine container and Pt capsule. However, many compositional parameters of the melts are buffered in polymineralic assemblages and the obtained results therefore have a bearing on partial melting of carbonated peridotite. |
EXPERIMENTAL METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL METHODSEXPERIMENTAL RESULTSDISCUSSIONREFERENCES |
|---|
Starting materials
Starting material SC1 was the same as used by Brey et al. (1990
, 2008). It consists of remixed olivine, orthopyroxene, clinopyroxene, and spinel originally separated from spinel lherzolite sample SC1 (Jagoutz et al., 1979) and ground to <20 µm. SC1 is a proxy for a fertile mantle composition and is similar to MixKLB-1 studied by Dasgupta & Hirschmann (2007) and Dasgupta et al. (2007). The compositions of the minerals are listed in Table 1 together with the original bulk composition and that of the mineral mix. The latter was determined by electron microprobe analysis (EMPA) from a glass bead prepared from 20 mg SC1 blended with 20 mg high-purity SiO2, which was added to avoid the formation of quench crystals. Trace elements were analysed in this glass by LA ICP MS.
|
Three experiments were carried out with mixture CL1, a composition estimated to be a melt in equilibrium with magnesite-bearing harzburgite at 6 GPa and 1400°C (Table 1). This mixture was prepared from pure chemical reagents [SiO2, TiO2, Al2O3, Fe2O3, Fe, Ca3(PO4)2, NiO, MnO, Na2CO3, and K2CO3] and natural magnesite and calcite.
For all other experiments SC1 was mixed with 10% of either high-purity graphite or natural magnesite or calcite (Icelandic spar) and 0·5 mg of the mixture was loaded into a container machined from San Carlos olivine. Its mg-number [molar MgO/(MgO + FeO)] ranged between 0·89 and 0·92. The container was placed within a Pt capsule with 1·6 mm outer diameter, 0·1 mm wall thickness and 1·6 mm in height. It was sealed by electric welding and pressed to a cylinder to minimize the deformation of the cell assembly during experiments.
Multianvil apparatus
Experiments were carried out in a Walker-type multianvil apparatus with WC cubes with 8 mm truncation edges. The pressure assembly consisted of a 95% MgO + 5% Cr2O3 octahedron, a zirconia sleeve, a rhenium foil heater and MgO inserts. Temperature was measured with a W–Re thermocouple inserted through an Al2O3 ceramic tube. The free space was filled with Ciatronics cement. Pressure was calibrated at room temperature and at high temperatures by S. Buhre and H. Steinberg (two unpublished Ph.D. theses, 2005) using the Bi phase transitions at room temperature, the 
–β Mg2SiO4 transition at 1200°C and 13·6 GPa (Morishima et al., 1994), the coesite–stishovite transition at 1200°C and 9·2 GPa (Yagi & Akimoto, 1976) and the CaGeO3 garnet–perovskite transition at 1109°C and 6·1 GPa (Susaki et al., 1985). The experimental technique has been described in more detail by Brey et al. (2008).
Analytical techniques
After the experiment the capsule was mounted in epoxy, cut lengthwise and oil polished for analysis with a Jeol Superprobe 8900 electron microprobe equipped with five wavelength spectrometers. An accelerating voltage of 15 kV and a beam current of 20 nA were used; counting times varied between 20 and 40 s on peak and background. Minerals were analysed with a focused beam, and quenched melts with a defocused beam from 5 to 30 µm in diameter. Standards were natural silicates and pure oxides and metals; the ZAF algorithm was used for matrix correction.
Melts and large garnet crystals were analysed for trace elements by LA ICP MS using a 30 µm laser beam. Residual bulk peridotite material within the olivine capsule was analysed in three samples using a 100 µm laser beam. The concentrations were normalized to the Ca content measured with the electron microprobe as an internal standard. NIST glasses were used as standards. The accuracy of analyses is about 10%. The concentrations of some minor elements (Ni, Cr, Mn) were analysed in experimental products by both electron microprobe and LA ICP MS. The data are consistent within 10–15%.
|
EXPERIMENTAL RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL METHODSEXPERIMENTAL RESULTSDISCUSSIONREFERENCES |
|---|
Experimental conditions and run products are shown in Table 2. The analyses of phases are given in Tables 3 and 4.
|
|
|
Phase relations
The run products are in general independent of the nature of the added carbon species, regardless of whether it is the unknown source of carbon in the precursor runs, graphite, magnesite or calcite. Figure 1a is a longitudinal section of a capsule after the experiment. It shows the residual peridotite zone in the middle of the olivine capsule and the melt pool on the side walls and towards the upper part of the capsule. The residual peridotite zone was always free of visible melt (Fig. 1b). The recrystallized olivine capsule was also usually devoid of melt. Only in a few cases were thin veinlets of quenched melt found within the olivine capsule (Fig. 1c). Melt always quenched to an aggregate of silicate and carbonate crystals, up to 10 µm in length (Fig. 1c). A thin (5–30 µm) discontinuous orthopyroxene ±magnesite ± garnet ± clinopyroxene layer was observed in all experiments between the Pt capsule and olivine (Fig. 1d). Pyroxene or garnet crystals occasionally grew in the melt pool (Fig. 1e). The residual peridotite material was always completely recrystallized, and no relics of starting minerals (e.g. spinel, or Al-rich pyroxene) were found even in subsolidus experiments. The newly formed garnets, olivines, and orthopyroxenes formed 5–15 µm euhedral to subhedral crystals (Fig. 1b). Larger (up to 50 µm across) poikilitic orthopyroxenes were occasionally found in the upper part of the residual peridotite zone (Fig. 1f). Clinopyroxene and magnesite were generally interstitial or formed subhedral grains between the other minerals. Modal vertical zoning was sometimes observed in the peridotite material with magnesite preferentially occurring in the lower part and orthopyroxene in the upper part. No significant variations in mineral compositions were related to this modal zoning.
In an attempt to understand the processes occurring during the experiments we performed a time series study with the SC1 + MgCO3 mixture at 6 GPa and 1500°C with run durations varying from 0·5 to 24 h. In the shortest experiment (0·5 h) all carbonates were melted or reduced to graphite (identified by Micro Raman spectroscopy) and clinopyroxene grains survived in the peridotite material. Mineral compositions were strongly heterogeneous, and relics of initial orthopyroxene with 
5 wt % Al2O3 were found. However, all spinel had disappeared and garnet had grown, which suggests rapid recrystallization. A small amount of melt occurred as a thin (1–3 µm) discontinuous layer between the olivine container and the upper lid of the Pt capsule (Fig. 2a). Melt channels in the olivine container are readily recognized in back-scattered electron (BSE) images as streaks of high-Mg olivine (Fig. 2a). The products of the 2 h experiment differ from those of the 0·5 h run in that no relics of Al-rich orthopyroxene and clinopyroxene crystals were found, the amount of graphite was significantly lower and larger pools of quenched melt were observed near the Pt wall (Fig. 2b). Abundant carbonate-rich melt was present in the products of the 5, 10 and 24 h runs. The peridotite zone contained magnesite and no graphite was found. The peridotite minerals are much more homogeneous than in the shorter experiments.
|
The phase assemblage in the residual peridotite was generally dependent on the P–T conditions but also varied with the type of the starting mixtures. Experiments without the addition of a carbon-bearing phase yielded olivine, orthopyroxene and garnet plus clinopyroxene and magnesite at the lower temperatures. Experiments with graphite added produced the assemblage olivine–orthopyroxene–garnet–magnesite at 6 GPa and temperatures between 1400 and 1600°C. In the experiments with SC1 + MgCO3 clinopyroxene was observed only under subsolidus conditions (1300°C, 6 GPa). Clinopyroxene-bearing assemblages with magnesite were produced in the experiments with SC1 + CaCO3.
Melt composition
The composition of melt varies depending on the P–T conditions and the mineral assemblage in the peridotite zone. The melts show slightly differing composition–temperature trends for experiments with magnesite–clinopyroxene-bearing, magnesite-bearing and clinopyroxene-free and magnesite- and clinopyroxene-free assemblages. The main features of these variations are illustrated in Fig. 3. The compositions are compared with the near-solidus melt of CO2-bearing peridotite determined by Dasgupta & Hirschmann (2007) at 6·6 GPa and 1245°C.
|
The SiO2 content of melts in the lowest temperature experiments at 6–7 GPa is 7–11 wt % both in clinopyroxene-bearing and clinopyroxene-absent runs. With increasing temperature SiO2 rises to 20–25 wt % at 1600°C, 6 GPa and 1700°C, 10 GPa. This trend is consistent with the composition of near-solidus melt reported by Dasgupta & Hirschmann (2007
) (4 wt % SiO2 at 1245°C). The total of the analyses increases with increasing temperature from about 60 to 75 wt %, indicating a decrease in CO2 contents in the melt. MgO rises with increasing temperature from about 20 to 30 wt %. The lowest MgO and highest CaO contents were observed in melts coexisting with clinopyroxene and magnesite. The CaO content of clinopyroxene-undersaturated melts is lower and shows no clear temperature dependence. The molecular ratio CaO/(CaO + MgO + FeO) decreases from about 0·45 to 0·25 with increasing temperature in clinopyroxene-bearing runs and is lower than 0·15 in clinopyroxene-free ones. The SiO2 content of melts coexisting with magnesite-free assemblages is slightly higher than in the presence of magnesite. The concentration of FeO(tot) increases with increasing temperature whereas the pressure effect is ambiguous. This is related to the influence of different degrees of melt oxidation (see below). Al2O3 in melt tends to increase with increasing temperature and decrease with increasing pressure from about 1 wt % at 6 GPa to less than 0·2 wt % at 10 GPa. The concentration of incompatible elements in the melts (Na, K, Ti, and P) decreases by a factor of approximately three between 1400 and 1600°C at 6 GPa. This implies a three-fold increase in the degree of partial melting.
Variations in trace elements are in accord with those of the major elements. Incompatible elements decrease with increasing temperature and decrease with pressure. The compatible elements Ni and Cr increase with increasing temperature. Moderately compatible and moderately incompatible elements (Sc, V, Mn, Y) show no regular variations with experimental conditions. The chondrite-normalized trace element distribution patterns of melts are shown in Fig. 4. There are pronounced positive deviations of La, Ta, Zr and Hf, which are related to previously unrecognized and unintentional contamination of the starting material in our laboratory (Fig. 4). All other elements show smooth regular distribution patterns compatible with the derivation of melts from a fertile peridotite.
|
Composition of minerals
In general, carbonate-rich melt was in direct contact with olivine and Pt capsule only. The analysis of Fe in Pt was only semiquantitative but indicates several wt % Fe in Pt near the contact with the olivine capsule. The mg-number of olivine in contact with melt slightly decreases with increasing temperature from 0·92–0·96 at 1400°C to about 0·92 at 1700°C, which is contrary to the expected trend related to an increase in the degree of partial melting. This implies that the mg-number of olivine near melt is controlled by the reaction between olivine and Pt. In contrast, the mg-numbers of low-Ca pyroxenes (presumably orthopyroxene at 6 GPa and high-P clinopyroxene at 10 GPa) in the residual peridotite zone increase with temperature in agreement with an increasing degree of partial melting. The concentration of other elements in the low-Ca pyroxenes varies regularly with pressure and temperature; that is, in equilibrium with clinopyroxene and garnet, there is an increase in Ca with increasing temperature and a decrease in Al with increasing pressure. In a few experiments orthopyroxene was also found in the melt pools outside the olivine capsule. Its composition is similar to that of pyroxene from the residual peridotite (Table 3).
The CaO content of garnet decreases with increasing temperature in agreement with experimental data for natural systems (Brey et al., 1986; Brenker & Brey, 1997). In the 6 GPa experiments Cr2O3 in garnet is practically constant when in equilibrium with two pyroxenes and is much lower in the 10 GPa experiment. TiO2 decreases with increasing temperature. The total FeO tends to be higher in higher pressure experiments. SiO2 appears to increase slightly with temperature at 6 GPa but the magnitude of a majorite component cannot be established. In the lower pressure experiments Al2O3 is generally close to 20 wt % but goes below 15 wt % in garnets with high total Fe. Stoichiometry calculations (12 O and eight cations) indicate Fe3+ amounts sufficient to compensate for the lower Al2O3. The calculated Fe3+/(Fe3+ + Fe2+) increases with temperature from 0·1 at 1400°C to 0·3 at 1600°C (Fig. 5). The compositions of garnets from the 10 GPa experiments are significantly different from those of lower pressure experiments: they are enriched in Si (up to 3·3 a.f.u.), and Fe oxidation is generally very high (i.e. Fe3+ > Fe2+). Both are highest in a garnet grown in the melt pool in run M-139 (10 GPa, 1600°C). Its calculated majorite fraction is 30%; it contains only 7 wt % Al2O3 and 14 wt % FeO(tot), which suggests 0·67 Fe3+ per a.f.u. This amount cannot be fully accommodated in the andradite end-member and a significant (10%) khoharite–skiagite (Mg3Fe3+2Si3O12–Fe2+3Fe3+2Si3O12) component must be assumed. Trace elements were analysed in three garnets that were grown within the melt pools (Fig. 4). They reflect the unintended enrichment in La, Ta, Zr and Hf of the starting material. Otherwise, the garnets are light rare earth element (LREE) depleted and heavy REE (HREE) enriched and high in Sc.
|
Clinopyroxenes do not show much compositional variation. The only significant difference is the much lower Al2O3 content of the clinopyroxenes in one 10 GPa experiment (M-139, 1600°C) compared with those from the 6 GPa experiments. The clinopyroxenes are always stoichiometric and their calculated Fe3+ content was negligible, even in the 10 GPa experimental products where very oxidized garnet was analysed.
Magnesite is an (Fe, Ca, Mg)CO3 solid solution with minor amounts of Mn and Ni.
Trace elements were measured in three residual peridotites from runs M-138 (6 GPa, 1400°C), M-140 (10 GPa, 1600°C) and M-142 (10 GPa, 1700°C) (Fig. 4). The distribution patterns are qualitatively similar to those of garnets, with depletion in the strongly incompatible trace elements and moderate enrichment in compatible trace elements. The three samples showed very similar distribution patterns, except for generally higher levels of all trace elements in the 10 GPa samples.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL METHODSEXPERIMENTAL RESULTSDISCUSSIONREFERENCES |
|---|
Mechanisms of melt formation and migration in the experiments
The present experiments suggest that the solidus of carbonated, H2O-free peridotite is slightly below 1300°C at 6 GPa and 1500°C at 10 GPa. This is consistent with the estimates of Falloon & Green (1989
) and Dasgupta & Hirschmann (2006) based on experiments with natural fertile peridotites (Fig. 6). Further evaluation of our results relies on the understanding of processes responsible for the development of the observed relationships between the melt, peridotite mixture and olivine container. The following interpretation is mainly based on the results of time series (kinetic) experiments at 6 GPa and 1500°C, which have been described above (Tables 2 and 3; Fig. 2).
|
The starting material in the kinetic experiments consisted of Fe2+–Mg silicates, Mg–Al spinel and magnesium carbonate, and was almost free of Fe3+. Oxygen fugacity was not externally buffered and depended on processes occurring within the capsule. The beginning of melting is accompanied by the formation of graphite, which is present in the products of the two shortest experiments (0·5 and 2 h). The appearance of graphite can be attributed to the fact that, in contrast to the starting minerals, the melt phase contains appreciable Fe3+; that is, the incipient melting is a redox reaction, which can be schematically written as
|
| (1) |
). During this stage the oxygen fugacity must be slightly below the enstatite-magnesite-olivine-graphite (diamond) [EMOG(D)] buffer (Eggler & Baker, 1982). Simultaneously with the melting in the peridotite zone, oxygen is liberated owing to the ferrous iron disproportionation reaction
|
| (2) |
|
| (3) |
The oxidation process can be followed by the distribution of Fe and Mg between olivine and melt. KD = (Fe/Mg)Ol/[Fe(tot)/Mg]Liq varies between 0·2 and 0·45 and is negatively correlated with the degree of Fe oxidation in garnet (Fig. 7). The highest KD values (0·40–0·45) are observed in the experiments in which garnet contains negligible amounts of Fe3+. Such values are in agreement with the distribution of Fe and Mg between olivine and carbonatitic melt at low oxygen fugacities; for example, in experiments with graphite capsules (e.g. Dalton & Wood, 1993; Hirose, 1997; Girnis et al., 2005). The negative correlation between KD and Fe3+ in garnet (Fig. 7) reflects an increase in Fe3+ in the melt. It can be supposed that KD* = (Fe2+/Mg)Ol/(Fe2+/Mg)Liq and DFe3+ = Fe2O3Grt/Fe2O3Liq are approximately constant over the range of our P–T conditions and melt compositions. The compositions of garnet and melts from our experiments suggest that KD* 0·43 and DFe3+ 1 (Fig. 7). Then, the calculated molar Fe3+/Fe2+ ratio in the melt varies from about 0·1 in the most reduced samples to 0·5 in melts coexisting with the andradite-rich garnets at 10 GPa and 1600°C. The equilibrium between Fe3+ and Fe2+ in melt can be described by the reaction
|
| (4) |
|
If the activity coefficients of FeO and FeO1·5 in the melt remain constant, a five-fold change in Fe3+/Fe2+ is equivalent to a change in fO2 by a factor of 54 = 625; that is, log fO2 increases by approximately 2·8 between the most reduced [close to the EMOG(D) buffer] and the most oxidized experiments at 10 GPa. This is probably the maximum estimate, because the high Fe3+ in garnets from 10 GPa experiments could be in part due to the pressure effect that was inferred by Rohrbach et al. (2007
), who observed a significant ferric iron content in garnet and clinopyroxene in equilibrium with metallic iron at pressures higher than 6 GPa. Thus, our experiments show that slight variations in oxygen fugacity near the EMOG(D) buffer may change phase assemblages in carbon-bearing mantle peridotite and affect its solidus temperature. In addition, the formation of small amounts of graphite at the beginning of melting suggests that the nucleation and growth of diamond in natural carbonated peridotite do not necessarily require externally controlled reduction but may be a consequence of closed-system heating or decompression.
Hunter & McKenzie (1989) and Watson et al. (1990) determined the dihedral angle between olivine and carbonate melt to be 25–30°. This means that carbonate melt wets olivine grains and forms a network of interconnected channels in mantle peridotite. McKenzie (1985) showed that carbonatite melt can be mobile in the mantle at melt fractions below 0·1%. This conclusion was experimentally supported by Hammouda & Laporte (2000), who demonstrated extremely fast dunite impregnation by carbonatite melt. Our experiments provide evidence for high mobility of both carbonatitic and kimberlitic melts in peridotite to pressures up to at least 10 GPa.
Melt pools mostly occurred in the upper parts of the capsules, but the presence of a thin melt film along the whole olivine–Pt interface is indicated by the crystallization of magnesite and, occasionally, clinopyroxene and garnet there. The reproducibility of the outwards melt migration suggests a steady pressure or temperature gradient within the capsule. To determine the main driving force for melt migration, a special experiment was performed at 6 GPa and 1500°C with a thin Re foil between the olivine container and the Pt capsule (run M-165). This minor change in sample configuration had a dramatic effect on melt behaviour (Fig. 2c and d). Neither newly formed minerals nor quenched melt were found outside the olivine container, which is in sharp contrast to even the shortest kinetic experiment (Fig. 2a) conducted under the same P–T conditions. The olivine capsule was homogeneous, and its mg-number remained exactly the same as before the experiment (0·909 ± 0·001 and 0·9089 ± 0·0006, respectively). In contrast to all other experiments, tiny pockets of interstitial melt were observed in the peridotite material (Fig. 2d). Thus, it is obvious that (1) temperature gradient had no effect on melt behaviour and (2) olivine–Pt interaction is crucial for melt segregation and migration. These inferences provide a ready explanation for the observed relations. Reaction (2) has a significant negative volume effect of about –12 cm3/mol. Therefore, iron loss from olivine to Pt maintains a stable negative pressure gradient and efficiently ‘sucks’ melt from peridotite.
Equilibrium between melt and residual peridotite
Reaction rates in carbonate-bearing systems with melts are rather high (e.g. Thibault et al., 1992; Lee et al., 1994), and the duration of our experiments was longer than in most experimental studies on peridotite melting. However, the question of equilibrium deserves special attention, because the melt rapidly escapes from the peridotite and there is no apparent connection between the melt and residual peridotite already after the first few hours. Nonetheless, several lines of evidence indicate that the melt outside the olivine container is close to equilibrium with the central peridotite zone. The first is the results of the aforementioned time series experiments that were aimed at elucidating the mechanism of melt formation and extraction. The compositions of melts from these experiments are shown against run duration in Fig. 8. The melt of the 2 h experiments is almost purely carbonate (<3 wt % SiO2) and very low in Al2O3 (0·1 wt %) and FeO (mg-number of 0·98). With increasing run duration the concentrations of SiO2, FeO and Al2O3 increase regularly and that of CaO decreases. The concentrations of all elements are identical within the analytical errors in melts from the 10 and 24 h experiments. These results show that melt–peridotite interaction continued even after the major volume of melt was squeezed outside the olivine capsule, but the melt composition changed significantly only within the first few hours. This observation is in line with the experimental results of Watson (1991), who detected very rapid Fe diffusivity through a network of interconnected channels of carbonate melt in a dunitic matrix.
|
The second argument comes from the analysis of pyroxene, garnet or magnesite that grew in several experiments within the melt near the Pt wall (Fig. 1e). The concentrations in these phases of elements that are incompatible in olivine (e.g. Cr, Ca, Na) are almost identical to those in the mineral phases of the residual peridotite (Fig. 9). There are, however, significant differences in their mg-numbers, which can be attributed to a continued increase in oxidation state.
|
Three crystallization experiments were performed using the estimated composition of melt in equilibrium with magnesite-bearing garnet harzburgite from the experiment at 6 GPa and 1400°C (mixture CL1 in Table 1). If the compositions obtained in partial melting experiments are close to equilibrium with residual peridotite under experimental P–T conditions, the liquidus temperature of CL1 must be close to 1400°C at 6 GPa, and the compositions of near-liquidus phases must be similar to those observed in the partial melting experiments. The crystallization experiments were also performed in olivine capsules at 1350, 1400 and 1450°C, 6 GPa (Tables 2 and 3). At 1350°C a tiny melt pool, too small for microprobe analysis, was found near the Pt capsule. The central zone consisted of magnesite, clinopyroxene and garnet. At 1400°C a large melt pool formed outside the olivine capsule and the central zone contained magnesite, clinopyroxene and garnet. At 1450°C the carbonatitic material was completely molten and the melt had assembled outside the olivine capsule and in the upper part of the Pt capsule, where large clinopyroxene crystals were also formed. These results show that the liquidus temperature of CL1 is close to the expected value (1400°C), although the composition of melt changed toward clinopyroxene saturation owing to interaction with the olivine capsule. The compositions of garnet and clinopyroxene from these experiments are similar to those from partial melting experiments (Table 3). Melt from the 1400°C experiment with the CL1 mixture (run M-150) coexists with olivine (capsule), clinopyroxene, garnet (residual phases) and orthopyroxene (newly formed phase near the Pt wall). Its composition appears to be similar to that of melt from run M-138, in which the same phase assemblages was observed in the residual peridotite under the same P–T conditions (Fig. 3). Considerable differences were observed only for those elements whose abundances are controlled mainly by the composition of starting mixture (e.g. Ti, Na, K). It should be noted that the interpretation of the melting experiments is more complex than expected and they cannot be considered as true ‘reversals’. This is related primarily to the larger amount of carbonate–silicate melt interacting with the olivine container and Pt capsule, which changed phase relations and compositions. Additional sources of ambiguity are the relatively high uncertainties in the composition of melt (Fig. 3) and uncontrolled oxygen fugacity.
Summing up the above, we conclude that the melts in our experiments were close to equilibrium with peridotite minerals with respect to the distribution of most components. The Fe–Mg relationships and oxidation state of minerals and melts were subject to changes because of the reactions between melt, olivine and Pt and cannot be rigorously evaluated. More accurate estimates of partial melt compositions from carbonated peridotite under strictly controlled bulk peridotite composition and redox conditions can be obtained by iterative methods utilizing sandwich experiments (Wallace & Green, 1988; Thibault et al., 1992; Hirschmann & Dasgupta, 2007). It should be noted that the near-solidus melt composition reported by Hirschmann & Dasgupta (2007) for a CO2-bearing peridotite at 6·6 GPa is generally consistent with our results (Fig. 3). Our method is advantageous in that it requires much fewer experiments and is especially helpful for reconnaissance investigations on mantle melting. It would be very efficient in combination with the iterative sandwich technique, but this is a project on its own and subject for future work.
Minor and trace element partitioning between minerals and melt
The partitioning of Ni between olivine and melt 
decreases with increasing temperature from more than 10 at 1300°C to 3–4 at 1700°C (Fig. 10). This is in the same range as obtained by Hart & Davis (1978) for CO2-free ultramafic and mafic melts.
|
Figure 11 shows the partitioning behaviour of a number of elements between orthopyroxene and melt. As for olivine,
decreases with temperature. 
shows no clear temperature dependence but decreases with pressure from around 1·8 at 6 GPa to unity at 10 GPa. Aluminium tends to change from incompatible to compatible with increasing temperature but pressure seems to counter this effect. 
shows a negative correlation with temperature, with values of 0·25 at 1300°C and 0·05–0·10 at 1700°C. Both Ca and Ti are incompatible, with 
0·5.
|
The partition coefficients
and 
average about 0·25 and 0·9, respectively, independent of temperature and pressure (Fig. 12). Ti and Cr are temperature dependent, with 
decreasing from 0·8 to 0·3 and 
from 40 to 15 as temperature increases from 1300 to 1700°C.
|
For magnesite–melt pairs
increases slightly with temperature from 0·07 to 0·12, and 
decreases from about eight to less than unity as temperature increases from 1300 to 1700°C (Fig. 13).
|
Trace element partitioning between garnet and melt was studied by LA ICP MS in three samples from runs at 6 GPa, 1500 and 1600°C and 10 GPa, 1700°C (Table 4). The partition coefficients are shown in Fig. 14 together with previous results for carbonatitic to kimberlitic melts (Sweeney et al., 1992
, 1995; Adam & Green, 2001; Girnis et al., 2006) and for volatile-free systems. It appears from Fig. 14 that the DGrt/Liq values are in general lowest for the 6 GPa, 1600°C run, slightly higher for 6 GPa, 1500°C and highest for the 10 GPa run. The partition coefficients from all three experiments are higher than those from our previous experiments in a kimberlitic system at 3–4 GPa (Girnis et al., 2006). This points to a pressure effect whereby the garnets from the 10 GPa experiment differ from the lower pressure garnets in having a high majorite and nominally a khoharite component. The present results also lie near the values reported by Sweeney et al. (1992, 1995) and Adam & Green (2001). Figure 14 also shows the range of DGrt/Liq values reported for volatile-free silicate systems. There is an overlap in values for the strongly and moderately incompatible elements including the LREE, but our DGrt/Liq values for the HREE are significantly lower. Such a behaviour for CO2-bearing systems was already pointed out by us previously (Girnis et al., 2006). The resulting lower LREE/HREE values can be potentially used to fingerprint the effect of carbonate melt metasomatism in the mantle. Vanadium behaves as an incompatible element in our experiments in contrast to all other data from the literature. This may be due to a high oxidation state of vanadium (V4+ and V5+) in our experiments (compare Canil, 1999).
|
Near-solidus liquids and primary kimberlites
Our experiments produced a series of CO2-rich melts in equilibrium with different mantle assemblages. In almost all experiments melt coexisted with olivine, orthopyroxene and garnet and ± magnesite and (or) clinopyroxene (magnesite-bearing or magnesite-free harzburgite or lherzolite). The composition of near-solidus melt is a function of both P–T conditions and peridotite phase assemblage (Figs 3 and 6).
Melts coexisting with the five-phase magnesite-bearing garnet lherzolite assemblage span a narrow compositional range (Fig. 3). They are generally low in SiO2 (10 wt %) and very high in CaO (20–25 wt %). Their MgO content is close to 20 wt % and not sensitive to pressure and temperature. Except for a higher SiO2 content, these melts are similar to liquids obtained under similar conditions near the solidus of carbonated lherzolite in the CaO–MgO–Al2O3–SiO2–CO2 system by Dalton & Presnall (1998b). The abundances of SiO2 and MgO in melts coexisting with olivine, orthopyroxene, garnet and magnesite are more variable and show distinct temperature trends (Fig. 3) with SiO2, increasing from 10 wt % at 1300–1400°C to 20 wt % at 1600–1700°C. Because of the overall high CO2 contents in the experiments there are only minor differences between the melt compositions with or without magnesite. The most important difference is the SiO2 content, which is higher in the experiments without magnesite at comparable P–T conditions. However, the concentration of SiO2 never surpasses 25 wt %.
Our derived melt compositions can now be used to test the hypothesis that some primary kimberlites were directly derived from the asthenospheric mantle (Smith et al., 1985; Ringwood et al., 1992; Tachibana et al., 2006). To this end we compared the results of our experiments with the recent compilation of probable primary kimberlite compositions by Becker & le Roex (2006). Such a comparison can be made only on a volatile-free basis, because the original concentrations of volatiles in kimberlite magmas are poorly constrained. The compositions were recalculated accordingly and are compared in Fig. 15.
|
It is clear that group II kimberlites are too rich in SiO2 to be produced by melting of a carbonated, anhydrous peridotite source. Group I kimberlites are closer to the field of the experimental melts, but further chemical features make their origin via a single-stage partial melting process in a carbon-bearing primitive mantle improbable. Figure 15 shows that melts derived from magnesite-bearing lherzolite have higher CaO and lower MgO contents compared with kimberlites. Only the high-pressure (10 GPa) melts from a magnesite-bearing garnet harzburgite are similar to group I kimberlites in MgO and SiO2 but they are extremely depleted in Al2O3 (<0·2 wt %) and thus unlike any kimberlite composition. This means that neither magnesite-bearing lherzolite nor magnesite-bearing harzburgite can be a source of kimberlite melt.
At higher degrees of partial melting with magnesite eliminated, the 6 GPa melts approach the field of group I kimberlites in terms of MgO–Al2O3–SiO2–CaO relationships. However, the source cannot be primitive (asthenospheric mantle) or depleted, because the concentrations of incompatible elements (Ti, K, P) in kimberlites are much higher than in the experimental melts. The source of kimberlite must be harzburgitic (low-Ca) and re-enriched in incompatible elements. Such kimberlites cannot have separated from their residue at pressures much higher than 6 GPa (Al2O3 in the melts becomes very low). This deduction applies for carbonated, anhydrous harzburgitic sources. The effect of water and other volatiles is not considered here. It will be subject of future experimental work.
Trace element characteristics of carbonate-rich melts in the mantle
A comparison of the trace element characteristics of the experimentally produced melts and of the residual peridotites can be used to derive empirical bulk distribution coefficients (Figs 4 and 16). They can provide more insight into the possible modifications of mantle peridotite by carbonate-rich melts. Trace elements were determined in residual peridotite in three runs at 6 GPa, 1400°C (M-138), 10 GPa, 1600°C (M-140) and 10 GPa, 1700°C (M-142) (Fig. 4). They represent a magnesite-bearing garnet lherzolite, a magnesite-bearing garnet harzburgite and a magnesite-free garnet harzburgite residue, respectively.
|
Figure 16 shows concentration ratios of melt/residual peridotite for these three experiments, which may be regarded as bulk partition coefficients. They may also be seen as the composition of incipient melts (i.e. for zero degree of partial melting) normalized to the source composition. The ratios then exhibit the maximum possible fractionation between elements with different degrees of incompatibility and provide insight into the chemistry of metasomatizing carbonatitic melt. It should be noted that such an empirical melt/peridotite ratio is not a true liquid/solid partition coefficient, because it ignores the possible presence of tiny amounts of melt in the interstices between the residual peridotite minerals and grain boundary enrichment of incompatible elements. The highly incompatible elements would be affected most in this way and their DLiquid/Solid lowered. However, these factors operate also in natural systems and we consider the observed relations as proxies for natural processes. For a carbonated primitive mantle, the carbonatitic melts are strongly enriched in Ba, U, Th, Nb. Ta, LREE and Sr. The enrichment in other incompatible elements is only minor or negligible (Y, HREE). Zr and Hf even display a slightly negative anomaly.
A comparison with the trace elements of average kimberlites (Becker & le Roex, 2006) shows many qualitative similarities, indicating that the source of kimberlite melts must not be too different from the primitive mantle. However, some differences are important and preclude the derivation of kimberlites directly from the asthenospheric mantle: (1) the REE patterns of the kimberlites are more strongly fractionated than those of the experiment melts (i.e. Ce/Yb in the kimberlites is 90–110 compared with 50–80 in the experimental melts); (2) some other element ratios, such as Sr/Ti (4·5–10 and 1·5–3·0, respectively) and Y/Zr (0·130 and 4·0–0·5, respectively), are also different. Generally, the kimberlites show a higher degree of fractionation from highly to less incompatible elements; they must be derived from a (re)enriched source that is more refractory in its major elements than the primitive mantle (see above).
Such an enrichment of mantle magma sources in incompatible elements is commonly explained by the migration of small portions of near-solidus mantle melts, which may be a hydrous silicate melt, hydrous fluid or a carbonatitic melt. Geochemical indicators were proposed to distinguish between these agents. In particular, strong REE enrichment combined with Zr and Ti depletion is regarded as evidence for carbonate metasomatism (Yaxley et al., 1991; Hauri et al., 1993; Ionov et al., 1993; Rudnick et al., 1993). This is supported by the 3 GPa experiments of Blundy & Dalton (2000), who showed that carbonate melts are characterized by elevated incompatible trace element concentrations, strongly fractionated REE patterns and depletion in Ti and Zr.
Our results show that migration of high-pressure carbonate–silicate melts cannot produce a stronger fractionation between HFSE and LILE compared with silicate melts (Figs 14 and 16) and Zr and Hf are also not significantly fractionated. The latter was considered by Dupuy et al. (1992) as an indicator of carbonate mantle metasomatism. The difference of our findings and the literature data may be explained by a dependence of HFSE and HREE partition coefficients on temperature and melt composition. Our experimental melts are transitional between carbonatites and kimberlites, and HFSE and HREE are probably more soluble in such more silicic melts compared with lower temperature ‘true’ carbonatite melts. Distinguishing between the chemically different metasomatic agents thus becomes more difficult. A possible indicator of deep carbonate metasomatism may be a relatively low fractionation amongst the REE combined with very high contents of incompatible elements. With the trace element characteristics of our experimental melts, repeated or continuous overprinting as in a chromatographic column is necessary to provide a suitable source composition for melts with strongly fractionated REE patterns.
|
ACKNOWLEDGEMENTS |
|---|
We are grateful to G. Yaxley and an anonymous reviewer for careful reading of our manuscript and insightful comments. The help of Heidi Höfer and Alan Woodland with analytical facilities is gratefully acknowledged. The work was supported by the Deutsche Forschungsgemeinschaft and the Russian Academy of Sciences.
*Corresponding author. Telephone: +49 (069) 798 40124. Fax: +49 (069) 798 40121. E-mail: brey{at}em.uni-frankfurt.de
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL METHODSEXPERIMENTAL RESULTSDISCUSSIONREFERENCES |
|---|
Adam J, Green T. Experimentally determined partition coefficients for minor and trace elements in peridotite minerals and carbonatitic melt, and their relevance to natural carbonatites. European Journal of Mineralogy (2001) 13:815–827.
Baker M, Stolper EM. Determining the composition of high-pressure mantle melts using diamond aggregates. Geochimica et Cosmochimica Acta (1994) 58:2811–2827.[CrossRef][Web of Science]
Barth MG, Foley SF, Horn I. Partial melting in Archean subduction zones: Constraints from experimentally determined trace element partition coefficients between eclogitic minerals and tonalitic melts under upper mantle conditions. Precambrian Research (2002) 113:323–340.[CrossRef][Web of Science]
Becker M, le Roex AP. Geochemistry of South African on- and off-craton, group I and group II kimberlites: Petrogenesis and source region evolution. Journal of Petrology (2006) 47:673–703.
Bennett SL, Blundy J, Elliott T. The effect of sodium and titanium on crystal–melt partitioning of trace elements. Geochimica et Cosmochimica Acta (2004) 68:2335–2347.[CrossRef][Web of Science]
Blundy J, Dalton J. Experimental comparison of trace element partitioning between clinopyroxene and melt in carbonate and silicate systems, and implications for mantle metasomatism. Contributions to Mineralogy and Petrology (2000) 139:356–371.[CrossRef][Web of Science]
Bowen NL, Schairer JF. The system FeO–SiO2. American Journal of Science, Series 5 (1932) 24:177–211.
Brenker FE, Brey GP. Reconstruction of the exhumation path of the Alpe Arami garnet-peridotite body from depths exceeding 160 km. Journal of Metamorphic Geology (1997) 15:581–592.[CrossRef][Web of Science]
Brey GP, Green DH. The role of CO2 in the genesis of olivine melilitite. Contributions to Mineralogy and Petrology (1975) 49:93–103.[CrossRef][Web of Science]
Brey G, Brice WR, Ellis DJ, Green DH, Harris KL, Ryabchikov ID. Pyroxene–carbonate reactions in the upper mantle. Earth and Planetary Science Letters (1983) 62:63–74.[CrossRef][Web of Science]
Brey GP, Nickel KG, Kogarko LN. Garnet–pyroxene equilibria in the system CaO–MgO–Al2O3–SiO2 (CMAS): prospect for simplified (‘T-independent’) lherzolite barometry and an eclogite-barometer. Contributions to Mineralogy and Petrology (1986) 92:448–455.[CrossRef][Web of Science]
Brey GP, Koehler T, Nickel KG. Geothermobarometry in four-phase lherzolites I. Experimental results from 10 to 60 kbar. Journal of Petrology (1990) 31:1313–1352.
Brey GP, Bulatov VK, Girnis AV. Geobarometry for peridotites: Experiments in simple and natural systems from 6 to 10 GPa. Journal of Petrology (2008) doi:10.1093/petrology/egm067.
Canil D. Vanadium partitioning between orthopyroxene, spinel and silicate melt and the redox states of mantle source regions for primary magmas. Geochimica et Cosmochimica Acta (1999) 63:557–572.[CrossRef][Web of Science]
Dalton JA, Presnall DC. The continuum of primary carbonatitic–kimberlitic melt compositions in equilibrium with lherzolite: data from the system CaO–MgO–Al2O3–SiO2–CO2 at 6 GPa. Journal of Petrology (1998a) 39:1953–1964.[CrossRef][Web of Science]
Dalton JA, Presnall DC. Carbonatitic melts along the solidus of model lherzolite in the system CaO–MgO–Al2O3–SiO2–CO2 from 3 to 7 GPa. Contributions to Mineralogy and Petrology (1998b) 131:123–135.[CrossRef][Web of Science]
Dalton JA, Wood BJ. The compositions of primary carbonate melts and their evolution through wallrock reaction in the mantle. Earth and Planetary Science Letters (1993) 119:511–525.[CrossRef][Web of Science]
Dasgupta R, Hirschmann MM. Melting in the Earth's deep upper mantle caused by carbon dioxide. Nature (2006) 440:659–662.[CrossRef][Medline]
Dasgupta R, Hirschmann MM. A modified iterative sandwich method for determination of near-solidus partial melt compositions. II. Application to determination of near-solidus melt compositions of carbonated peridotite. Contributions to Mineralogy and Petrology (2007) 154:647–661.[CrossRef][Web of Science]
Dasgupta R, Hirschmann MM, Smith ND. Partial melting experiments of peridotite + CO2 at 3 GPa and genesis of alkalic ocean island basalts. Journal of Petrology (2007) 48:2093–2124.
Draper DS, Xirouchakis D, Agee CB. Trace element partitioning between garnet and chondritic melt from 5 to 9 GPa: implications for the onset of the majorite transition in the Martian mantle. Physics of the Earth and Planetary Interiors (2003) 139:149–169.[CrossRef][Web of Science]
Dupuy C, Liotard JM, Dostal J. Zr/Hf fractionation in intraplate basaltic rocks: carbonate metasomatism in the mantle source. Geochimica et Cosmochimica Acta (1992) 56:2417–2423.[CrossRef][Web of Science]
Eggler DH. The effect of CO2 upon partial melting of peridotite in the system Na2O–CaO–Al2O3–MgO–SiO2–CO2 to 35 kb, with an analysis of melting in a peridotite–H2O–CO2 system. American Journal of Science (1978) 278:305–343.
Eggler DH, Baker DR. Reduced volatiles in the system C–O–H: implications to mantle melting, fluid formation and diamond genesis. In: High Pressure Research in Geophysics—Akimoto S, Manghnani MH, eds. (1982) Tokyo: Centre for Academic Publishing. 237–250.
Falloon TJ, Green DH. The solidus of carbonated, fertile peridotite. Earth and Planetary Science Letters (1989) 94:364–370.[CrossRef][Web of Science]
Falloon TJ, Green DH. Solidus of carbonated fertile peridotite under fluid-saturated conditions. Geology (1990) 18:195–199.
Girnis AV, Bulatov VK, Brey GP. Transition from kimberlite to carbonatite melt under mantle parameters: an experimental study. Petrology (2005) 13:1–15.[Web of Science]
Girnis AV, Bulatov VK, Lahaye Y, Brey GP. Partitioning of trace elements between carbonate–silicate melts and mantle minerals: experiment and petrological consequences. Petrology (2006) 14:492–514.[CrossRef][Web of Science]
Green DH, Wallace ME. Mantle metasomatism by ephemeral carbonatite melts. Nature (1988) 336:459–462.[CrossRef]
Green TH, Blundy JD, Adam J, Yaxley GM. SIMS determination of trace element partition coefficients between garnet, clinopyroxene and hydrous basaltic liquids at 2–7·5 GPa and 1080–1200°C. Lithos (2000) 53:165–187.[CrossRef][Web of Science]
Gudfinnsson GH, Presnall DC. Continuous gradations among primary carbonatitic, kimberlitic, melilititic, basaltic, picritic, and komatiitic melts in equilibrium with garnet lherzolite at 3–8 GPa. Journal of Petrology (2005) 46:1645–1659.
Hammouda T, Laporte D. Ultrafast mantle impregnation by carbonatite melts. Geology (2000) 28:283–285.
Hart SR, Davis KE. Nickel partitioning between olivine and silicate melt. Earth and Planetary Science Letters (1978) 40:203–219.[CrossRef][Web of Science]
Hauri EH, Shimizu N, Dieu JJ, Hart SR. Evidence for hotspot-related carbonatite metasomatism in the oceanic upper mantle. Nature (1993) 365:221–227.[CrossRef]
Hirose K. Partial melt compositions of carbonated peridotite at 3 GPa and role of CO2 in alkali-basalt magma generation. Geophysical Research Letters (1997) 24:2837–2840.[CrossRef][Web of Science]
Hirose K, Kushiro I. Partial melting of dry peridotites at high pressures: determination of compositions of melts segregated from peridotites using aggregates of diamonds. Earth and Planetary Science Letters (1993) 114:477–489.[CrossRef][Web of Science]
Hirschmann MM. Mantle solidus: experimental constraints and the effects of peridotite composition. Geochemistry, Geophysics, Geosystems (2000) 1. paper number 2000GC000070.
Hirschmann MM, Dasgupta R. A modified iterative sandwich method for determination of near-solidus partial melt compositions. I. Theoretical considerations. Contributions to Mineralogy and Petrology (2007) 154:635–645.[CrossRef][Web of Science]
Hunter RH, McKenzie D. The equilibrium geometry of carbonate melts in rocks of mantle composition. Earth and Planetary Science Letters (1989) 92:347–356.[CrossRef][Web of Science]
Ionov DA, Dupuy C, O’Reilly SY, Kopylova MG, Genshaft YuS. Carbonated peridotite xenoliths from Spitsbergen: implications for trace element signature of mantle carbonate metasomatism. Earth and Planetary Science Letters (1993) 119:283–297.[CrossRef][Web of Science]
Ionov DA, O’Reilly SY, Genshaft YuS, Kopylova MG. Carbonate-bearing mantle peridotite xenoliths from Spitsbergen: phase relationships, mineral compositions and trace-element residence. Contributions to Mineralogy and Petrology (1996) 125:375–392.[CrossRef][Web of Science]
Jagoutz E, Palme H, Baddenhausen H, Blum K, Cendales M, Dreibus G, Spettel B, Lorenz V, Waenke H. The abundances of major, minor and trace elements in the Earth's mantle as derived from primitive ultramafic nodules. In: Proceedings of the Lunar and Planetary Science Conference, 2—Merrill RB, Bogard DD, Hoerz F, McKay DS, Robertson PC, eds. (1979) New York: Pergamon Press. 2031–2050.
Laporte D, Toplis MJ, Seyler M, Devidal J-L. A new experimental technique for extracting liquids from peridotite at very low degrees of melting: application to partial melting of depleted peridotite. Contributions to Mineralogy and Petrology (2004) 146:463–484.[CrossRef][Web of Science]
Lee W-J, Wyllie PJ, Rossman GR. CO2-rich glass, round calcite crystals, and no liquid immiscibility in the system CaO–SiO2–CO2 at 2·5 GPa. American Mineralogist (1994) 79:1135–1144.[Abstract]
Leost I, Stachel T, Brey GP, Harris JW, Ryabchikov ID. Diamond formation and source carbonation: mineral associations in diamonds from Namibia. Contributions to Mineralogy and Petrology (2003) 145:15–24.[Web of Science]
McKenzie D. The extraction of magma from the crust and mantle. Earth and Planetary Science Letters (1985) 74:81–91.[CrossRef][Web of Science]
Morishima H, Kato T, Suto M, Ohtani E, Urakawa S, Utsumi W, Shimomura O, Kikegawa T. The phase boundary between - and β-Mg2SiO4 determined by in situ X-ray observation. Science (1994) 265:1202–1203.
Newton RC, Sharp WE. Stability of forsterite + CO2 and its bearing on the role of CO2 in the mantle. Earth and Planetary Science Letters (1975) 26:239–244.[CrossRef][Web of Science]
Nicholls IA, Harris KL. Experimental rare earth element partition coefficients for garnet, clinopyroxene and amphibole coexisting with andesitic and basaltic liquids. Geochimica et Cosmochimica Acta (1980) 44:287–308.[CrossRef][Web of Science]
Palme H, O’Neill HStC. Cosmochemical estimates of mantle composition. In: Treatise on Geochemistry, 2—Holland HD, Turekian KK, Davis AM, eds. (2003) Amsterdam: Elsevier. 1–38.
Ringwood AE, Kesson SE, Hibberson W, Ware N. Origin of kimberlites and related magmas. Earth and Planetary Science Letters (1992) 113:521–538.[CrossRef][Web of Science]
Rohrbach A, Ballhaus C, Golla-Schindler U, Ulmer P, Kamenetsky VS, Kuzmin DV. Metal saturation in the upper mantle. Nature (2007) 449:456–458.[CrossRef][Medline]
Rudnick R, McDonough WF, Chappell BC. Carbonatite metasomatism in the northern Tanzanian mantle. Earth and Planetary Science Letters (1993) 114:463–475.[CrossRef][Web of Science]
Smith CB, Gurney JJ, Skinner EMW, Clement CR, Ebrahim N. Geochemical character of Southern African kimberlites: a new approach based on isotopic constraints. Transactions of the Geological Society of South Africa (1985) 88:267–280.
Smith D. Genesis of carbonate in pyrope from ultramafic diatremes on the Colorado Plateau, southwestern United States. Contributions to Mineralogy and Petrology (1987) 97:389–396.[CrossRef][Web of Science]
Susaki J, Akaogi M, Akimoto S, Shimomura O. Garnet–perovskite transformation in CaGeO3: in-situ X-ray measurements using synchrotron radiation. Geophysical Research Letters (1985) 12:729–732.[Web of Science]
Sweeney RJ, Green DH, Sie SH. Trace and minor element partitioning between garnet and amphibole and carbonatitic melt. Earth and Planetary Science Letters (1992) 113:1–14.[CrossRef][Web of Science]
Sweeney RJ, Prozesky V, Przybylowicz W. Selected trace and minor element partitioning between peridotite minerals and carbonatite melts at 18–46 kb pressure. Geochimica et Cosmochimica Acta (1995) 59:3671–3683.[CrossRef][Web of Science]
Tachibana Y, Kaneoka I, Gaffney A, Upton B. Ocean-island basalt-like source of kimberlite magmas from West Greenland revealed by high 3He/4He ratios. Geology (2006) 34:273–276.
Thibault Y, Edgar AD, Lloyd FE. Experimental investigation of melts from a carbonated phlogopite lherzolite: implications for metasomatism in the continental lithospheric mantle. American Mineralogist (1992) 77:784–794.[Abstract]
Van Westrenen W, Blundy J, Wood B. Crystal-chemical controls on trace element partitioning between garnet and anhydrous silicate melt. American Mineralogist (1999) 84:838–847.[Abstract]
Van Westrenen W, Blundy JD, Wood BJ. Effect of Fe2+ on garnet–melt trace element partitioning: experiments in FCMAS and quantification of crystal-chemical controls in natural systems. Lithos (2000) 53:189–201.[CrossRef][Web of Science]
Wallace ME, Green DH. An experimental determination of primary carbonatite magma composition. Nature (1988) 335:343–346.[CrossRef]
Wang A, Pasteris JD, Meyer HOA, Dele-Duboi ML. Magnesite bearing inclusion assemblage in natural diamond. Earth and Planetary Science Letters (1996) 141:293–306.[CrossRef][Web of Science]
Wasylenki LE, Baker MB, Kent AJR, Stolper EM. Near-solidus melting of the shallow upper mantle: partial melting experiments on depleted peridotite. Journal of Petrology (2003) 44:1163–1191.
Watson EB. Diffusion in fluid-bearing and slightly-melted rocks: experimental and numerical approaches illustrated by iron transport in dunite. Contributions to Mineralogy and Petrology (1991) 107:417–434.[CrossRef][Web of Science]
Watson EB, Brenan JM, Baker DR. Distribution of fluids in the continental lithospheric mantle. In: The Continental Lithospheric Mantle—Menzies MA, ed. (1990) Oxford: Clarendon Press. 111–125.
Wyllie PJ, Huang WL. Carbonation and melting reactions in the system CaO–MgO–SiO2–CO2 at mantle pressures with geophysical and petrological applications. Contributions to Mineralogy and Petrology (1976) 54:79–107.[CrossRef][Web of Science]
Yagi T, Akimoto S. Direct determination of coesite–stishovite transition by in-situ X-ray measurements. Tectonophysics (1976) 35:259–270.[CrossRef][Web of Science]
Yaxley GM, Crawford AJ, Green DH. Evidence for carbonatite metasomatism in spinel peridotite xenoliths from Western Victoria, Australia. Earth and Planetary Science Letters (1991) 107:305–317.[CrossRef][Web of Science]
Yaxley GM, Green DH, Kamenetsky V. Carbonatite metasomatism in the Southeastern Australian lithosphere. Journal of Petrology (1998) 39:1917–1930.[CrossRef][Web of Science]
Zedgenizov DA, Kagi H, Shatsky VS, Sobolev NV. Carbonatitic melts in cuboid diamonds from Udachnaya kimberlite pipe (Yakutia): evidence from vibrational spectroscopy. Mineralogical Magazine (2004) 68:61–73.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||