Anhydrous melting of peridotite at 0–15 Kb pressure and the genesis of tholeiitic basalts

The anhydrous melting behaviour of two synthetic peridotite compositions has been studied experimentally at temperatures ranging from near the solidus to about 200° C above the solidus within the pressure range 0–15 kb. The peridotite compositions studied are equivalent to Hawaiian pyrolite and a more depleted spinel lherzolite (Tinaquillo peridotite) and in both cases the experimental studies used peridotite –40% olivine compositions. Equilibrium melting results in progressive elimination of phases with increasing temperature. Four main melting fields are recognized; from the solidus these are: olivine (ol)+orthopyroxene (opx)+clinopyroxene (cpx)+Al-rich phase (plagioclase at low pressure, spinel at moderate pressure, garnet at high pressure)+liquid (L); ol+opx+cpx+Cr-spinel+L; ol+opx+Cr-spinel +L: ol±Cr-spinel+L. Microprobe analyses of the residual phases show progressive changes to more refractory compositions with increasing proportion of coexisting melt i.e. increasing Mg/(Mg+Fe) and Cr/(Cr+Al) ratios, decreasing Al₂O₃, CaO in pyroxene.The degree of melting, established by modal analysis, increases rapidly immediately above the solidus (up to 10% melting occurs within 25°–30° C of the solidus), and then increases in roughly linear form with increasing temperature.Equilibrium melt compositions have been calculated by mass balance using the compositions and proportions of residual phases to overcome the problems of iron loss and quench modification of the glass. Compositions from the melting of pyrolite within the spinel peridotite field (i.e. 15 kb) range from alkali olivine basalt (<15% melting) through olivine tholeiite (20–30% melting) and picrite to komatiite (40–60% melting). Melting in the plagioclase peridotite field produces magnesian quartz tholeiite and olivine-poor tholeiite and, at higher degrees of melting (30–40%), basaltic or pyroxenitic komatiite. Melts from Tinaquillo lherzolite are more silica saturated than those from pyrolite for similar degrees of partial melting, and range from olivine tholeiite through tholeiitic picrite to komatiite for melting in the spinel peridotite field.The equilibrium melts are compared with inferred primary magma compositions and integrated with previous melting studies on basalts. The data obtained here and complementary basalt melting studies do not support models of formation of oceanic crust in which the parental magmas of common mid-ocean ridge basalts (MORB) are attributed to segregation from source peridotite at shallow depths ( 25 km) to leave residual harzburgite. Liquids segregating from peridotite at these depths are more silica-rich than common MORB.     

Manganese partitioning during hydrous melting of peridotite

Manganese contents and the iron/manganese ratio of igneous rocks have been used as a method of probing the heterogeneity in the Earth’s mantle during melting of peridotite and pyroxenite lithologies. Most previous work has assumed that changes in these parameters require differences in either source lithology or composition based on experiments indicating that manganese is slightly incompatible during melting and that the iron/manganese ratio is fixed by the presence of olivine. However, the presence of volatiles in the mantle drives melting at lower temperatures and with different compositions than in volatile-free systems, and thus the partitioning of Fe and Mn may in fact vary. We have produced silicate liquids in equilibrium with a peridotite assemblage under hydrous conditions at 3 GPa that show that Mn can also be unexpectedly compatible in garnet at 1375 °C and that Mn partitioning between solids and liquids can be strongly affected by temperature and liquid composition. The compatibility of Mn in garnet provides a mechanism for large variations of Mn contents and the Fe/Mn ratio in silicate melts that solely involves melting of mantle peridotite with only small compositional changes. Correlations between Mn variations and other indices indicative of melting in the presence of garnet may provide a means of more completely understanding the role of garnet at high pressures in peridotite melting.     

The influence of water on melting of mantle peridotite

This experimental study examines the effects of variable concentrations of dissolved H₂O on the compositions of silicate melts and their coexisting mineral assemblage of olivine + orthopyroxene ± clinopyroxene ± spinel ± garnet. Experiments were performed at pressures of 1.2 to 2.0 GPa and temperatures of 1100 to 1345 °C, with up to ∼12 wt% H₂O dissolved in the liquid. The effects of increasing the concentration of dissolved H₂O on the major element compositions of melts in equilibrium with a spinel lherzolite mineral assemblage are to decrease the concentrations of SiO₂, FeO, MgO, and CaO. The concentration of Al₂O₃ is unaffected. The lower SiO₂ contents of the hydrous melts result from an increase in the activity coefficient for SiO₂ with increasing dissolved H₂O. The lower concentrations of FeO and MgO result from the lower temperatures at which H₂O-bearing melts coexist with mantle minerals as compared to anhydrous melts. These compositional changes produce an elevated SiO₂/(MgO + FeO) ratio in hydrous peridotite partial melts, making them relatively SiO₂ rich when compared to anhydrous melts on a volatile-free basis. Hydrous peridotite melting reactions are affected primarily by the lowered mantle solidus. Temperature-induced compositional variations in coexisting pyroxenes lower the proportion of clinopyroxene entering the melt relative to orthopyroxene. Isobaric batch melting calculations indicate that fluid-undersaturated peridotite melting is characterized by significantly lower melt productivity than anhydrous peridotite melting, and that the peridotite melting process in subduction zones is strongly influenced by the composition of the H₂O-rich component introduced into the mantle wedge from the subducted slab. Received: 7 April 1997 / Accepted: 9 January 1998     

Melting equilibria in multicomponent systems and liquidus/solidus convergence in mantle peridotite

The melting of undepleted mantle peridotite proceeds through a temperature interval which decreases with increasing pressure. If liquidus and solidus actually meet in the range 100–150 Kb, as suggested by Herzberg (1983), peridotite must transform there directly to a melt of its own composition. Thermodynamic analysis shows that such a liquidus/solidus meeting would be very unlikely in a system as chemically complex as mantle peridotite and would require that unanticipated phase equilibrium relations suppress all incongruent melting behavior. But Takahashi and Scarfe's (1985) preliminary experiments suggest that the upper mantle itself may indeed have a special composition with respect to phase equilibrium relations between liquids and solids at very high pressure. If so, mantle peridotite composition cannot be generated as a crystal accumulate or melting residue, because these two popular theories of origin are difficult to reconcile with a supposed eutecticlike composition. If upper mantle peridotite were itself a solidified liquid composition produced either as a partial melt or, more likely, as a crystallization residue of some more primitive melt composition representative of the whole mantle, an approach of liquidus to solidus might be expected at high pressure although the phase relations of Herzberg (1983) and Herzberg and O'Hara (1985) remain implausible.     

Petrogenesis of the garnet peridotite and garnet-free peridotite of the Zhimafang ultramafic body in the Sulu ultrahigh-pressure metamorphic belt, eastern China

The CCSD‐PP1 drillhole penetrated a 110‐m‐thick sequence of the Zhimafang ultramafic body in the Sulu ultrahigh‐pressure (UHP) metamorphic belt, east China. The sequence consists of interlayered garnet‐bearing (Grt) and garnet‐free (GF) peridotite. Eleven layers of Grt‐peridotite, ranging from 1.2 to 9.5 m in thickness, have an aggregate thickness of 54.49 m, whereas eight layers of GF‐peridotite, ranging from 2.2 to 14.2 m in thickness, have a total thickness of 57.53 m. The boundaries between the two rock types are gradational. The Grt‐peridotites have slightly higher contents of Al₂O₃, CaO and SiO₂, and lower Mg^#s (0.90–0.92) than the GF‐peridotites (Mg^#s 0.91–0.93). Both contain low TiO₂ (<0.05 wt%) and have higher modal abundances of enstatite (average 10 vol.%) than diopside (1–5 vol.%), typical of depleted‐type upper mantle. The diopside in these rocks has high and relatively uniform Mg^# members (0.93–0.95), but highly variable Al₂O₃ (0.2–2.3 wt%), Na₂O (0.5–2.5 wt%) and Cr₂O₃ (0.38–2.09 wt%). Enstatite (En_92?93) contains very low Al₂O₃ (0–0.3 wt%). Both porphyroblastic and equigranular garnet are present. The equigranular varieties are zoned, from core to rim in Cr₂O₃ (3.4–4.2 wt%), MgO (18.4–17.5 wt%) and Al₂O₃ (21.1–20.1 wt%). Titania is very low in all the garnet, mostly <0.05 wt%. Chromite or chromium (Cr)‐spinel occur both in the Grt‐ and GF‐peridotite, and are characterized by high contents of Cr₂O₃ (49–58 wt%) and FeO (24–43 wt%), similar to that in iron‐rich Alpine‐type peridotites. Based on the bulk‐rock MgO–FeO compositions, the Zhimafang Grt‐peridotite probably underwent 20–30% partial melting, whereas the GF‐peridotite may have undergone as much as 35–40% partial melting, suggesting that the two rock types owe their differences to different degrees of partial melting rather than to pressure differences during metamorphism.     

The beginnings of hydrous mantle wedge melting

This study presents new phase equilibrium data on primitive mantle peridotite (0.33 wt% Na₂O, 0.03 wt% K₂O) in the presence of excess H₂O (14.5 wt% H₂O) from 740 to 1,200°C at 3.2–6 GPa. Based on textural and chemical evidence, we find that the H₂O-saturated peridotite solidus remains isothermal between 800 and 820°C at 3–6 GPa. We identify both quenched solute from the H₂O-rich fluid phase and quenched silicate melt in supersolidus experiments. Chlorite is stable on and above the H₂O-saturated solidus from 2 to 3.6 GPa, and chlorite peridotite melting experiments (containing ~6 wt% chlorite) show that melting occurs at the chlorite-out boundary over this pressure range, which is within 20°C of the H₂O-saturated melting curve. Chlorite can therefore provide sufficient H₂O upon breakdown to trigger dehydration melting in the mantle wedge or perpetuate ongoing H₂O-saturated melting. Constraints from recent geodynamic models of hot subduction zones like Cascadia suggest that significantly more H₂O is fluxed from the subducting slab near 100 km depth than can be bound in a layer of chloritized peridotite ~ 1 km thick at the base of the mantle wedge. Therefore, the dehydration of serpentinized mantle in the subducted lithosphere supplies free H₂O to trigger melting at the H₂O-saturated solidus in the lowermost mantle wedge. Alternatively, in cool subduction zones like the Northern Marianas, a layer of chloritized peridotite up to 1.5 km thick could contain all the H₂O fluxed from the slab every million years near 100 km depth, which suggests that the dominant form of melting below arcs in cool subduction zones is chlorite dehydration melting. Slab P–T paths from recent geodynamic models also allow for melts of subducted sediment, oceanic crust, and/or sediment diapirs to interact with hydrous mantle melts within the mantle wedge at intermediate to hot subduction zones.     

Constraints on clinopyroxene/melt partitioning of REE, Rb, Sr, Ti, Cr, Zr, and Nb during mantle melting: First insights from direct peridotite melting experiments at 1.0 GPa

Earlier piston-cylinder experiments in our laboratory produced a collection of mantle melting run products at 1.0 GPa that have now been analyzed by ion probe for selected REE, Ti, Cr, Rb, Sr, Y, Zr, and Nb. Natural starting materials were used and experiments were run in graphite-lined Pt capsules with the melt separated from the residual minerals into a layer of vitreous carbon spheres (VCS) to circumvent quench modification. The glass phase in 18 run products, representing melt percentages of ∼2-20 wt%, yielded excellent data that were inverted to yield the first estimates ever of clinopyroxene/melt distribution coefficients, Ds, derived from direct peridotite partial melting experiments. Uncertainties were estimated with a Monte Carlo method.For the REE and Y, these Ds were then compared to Ds calculated with the widely-used model of Wood and Blundy (1997) and the two sets overlap at the ±2σ level in 123 of 128 cases (∼96%). This indicates to us that: 1) the experiments analyzed here are well equilibrated with respect to major and trace element distributions, thus supporting the efficacy of the VCS technique and its variation involving diamond (e.g., Baker and Stolper, “Determining the composition of high-pressure mantle melts using diamond aggregates” [1994], Geochim. Cosmochim. Acta58, 2811-2827); 2) the model of Wood and Blundy (1997), calibrated largely on the basis of large melt fraction, inverse- or sandwich-type experiments, describes REE and Y partitioning during peridotite melting well, even very near the solidus; and it suggests that the cpx/melt Ds derived here for other elements, not modeled by the Wood and Blundy formulation, are probably also correct for peridotite melting to within their ±2σ uncertainties. D^sp/liq and D^cpx/liq values for Cr calculated directly from electron microprobe data decrease by about a factor of five with increasing temperature and melt percentage.The degree to which our experiments appear to have equilibrated seems at odds with recent measurements of the diffusivities of REE in diopside which suggest that relatively small percentages of our starting mineral grains should have equilibrated diffusively. Instead, we suggest that equilibration occurs much more rapidly through the processes of recrystallization and grain coarsening, accomplished through dissolution and reprecipitation. This suggestion is supported by the observation that our final grain sizes are typically 5-10 times larger than the ∼10 μm starting sizes, indicating that substantial mass transfer occurred in our experiments, probably mediated by the melt phase in which diffusion is faster.     

The Ronda high temperature peridotite: Geochemistry and petrogenesis

The Ronda peridotite in southern Spain is a large (~300 km²) exposure of upper mantle which provides direct information about mantle processes on a scale much larger than that provided by mantle xenoliths in basalt. Ronda peridotites range from harzburgite to lherzolite, and vary considerably in major element content, e.g., Al₂O₃ from 0.9 to 4.8%, and trace element abundances, e.g., Sr, Zr and La abundances vary by factors of 20 to 40. These compositional variations are systematic and correlate with (pyroxene + garnet)/olivine ratios and olivine compositions. The data are consistent with formation of residual peridotites by variable degrees of melting (~0 to 30%) of a compositionally homogeneous peridotite. None of the peridotites have geochemical characteristics of residues formed by extensive (?5%) fractional melting and the data can be explained by equilibrium (batch) melting, possibly with incomplete melt segregation in some samples. Based on compositional differences between Ronda peridotites, the segregated melts were picritic (12–22% MgO) with relative rare earth element abundances similar to mid-ocean ridge basalt (MORB). Prior to the melting event the Ronda peridotite body was a suitable source for MORB. The compositional characteristics of Ronda peridotites are consistent with diapiric rise of a fertile mantle peridotite with relatively small degrees of melting near the diapir-wall rock interface yielding residues of garnet iherzolite, and larger degrees of melting in the diapir interior yielding residues of garnet-free peridotite. Subsequently these residual rocks were recrystallized at sub-solidus conditions (Obata, 1980), and emplaced in the crust by thrusting (Lundeen, 1978).     

Mineral/melt partitioning of trace elements during hydrous peridotite partial melting

This experimental study examines the mineral/melt partitioning of incompatible trace elements among high-Ca clinopyroxene, garnet, and hydrous silicate melt at upper mantle pressure and temperature conditions. Experiments were performed at pressures of 1.2 and 1.6 GPa and temperatures of 1,185 to 1,370 °C. Experimentally produced silicate melts contain up to 6.3 wt% dissolved H ₂O, and are saturated with an upper mantle peridotite mineral assemblage of olivine+orthopyroxene+clinopyroxene+spinel or garnet. Clinopyroxene/melt and garnet/melt partition coefficients were measured for Li, B, K, Sr, Y, Zr, Nb, and select rare earth elements by secondary ion mass spectrometry. A comparison of our experimental results for trivalent cations (REEs and Y) with the results from calculations carried out using the Wood-Blundy partitioning model indicates that H ₂O dissolved in the silicate melt has a discernible effect on trace element partitioning. Experiments carried out at 1.2 GPa, 1,315 °C and 1.6 GPa, 1,370 °C produced clinopyroxene containing 15.0 and 13.9 wt% CaO, respectively, coexisting with silicate melts containing ~1–2 wt% H ₂O. Partition coefficients measured in these experiments are consistent with the Wood-Blundy model. However, partition coefficients determined in an experiment carried out at 1.2 GPa and 1,185 °C, which produced clinopyroxene containing 19.3 wt% CaO coexisting with a high-H ₂O (6.26±0.10 wt%) silicate melt, are significantly smaller than predicted by the Wood-Blundy model. Accounting for the depolymerized structure of the H ₂O-rich melt eliminates the mismatch between experimental result and model prediction. Therefore, the increased Ca ²⁺ content of clinopyroxene at low-temperature, hydrous conditions does not enhance compatibility to the extent indicated by results from anhydrous experiments, and models used to predict mineral/melt partition coefficients during hydrous peridotite partial melting in the sub-arc mantle must take into account the effects of H ₂O on the structure of silicate melts.     

The influence of melt structure on trace element partitioning near the peridotite solidus

This experimental study examines the mineral/melt partitioning of Na, Ti, La, Sm, Ho, and Lu among high-Ca clinopyroxene, plagioclase, and silicate melts analogous to varying degrees of peridotite partial melting. Experiments performed at a pressure of 1.5 GPa and temperatures of 1,285 to 1,345 °C produced silicate melts saturated with high-Ca clinopyroxene, plagioclase and/or spinel, and, in one case, orthopyroxene and garnet. Partition coefficients measured in experiments in which clinopyroxene coexists with basaltic melt containing ~18 to 19 wt% Al₂O₃ and up to ~3 wt% Na₂O are consistent with those determined experimentally in a majority of the previous studies, with values of ~0.05 for the light rare earths and of ~0.70 for the heavy rare earths. The magnitudes of clinopyroxene/melt partition coefficients for the rare earth elements correlate with pyroxene composition in these experiments, and relative compatibilities are consistent with the effects of lattice strain energy. Clinopyroxene/melt partition coefficients measured in experiments in which the melt contains ~20 wt% Al₂O₃ and ~4 to 8 wt% Na₂O are unusually large (e.g., values for Lu of up to 1.33±0.05) and are not consistent with the dependence on pyroxene composition found in previous studies. The magnitudes of the partition coefficients measured in these experiments correlate with the degree of polymerization of the melt, rather than with crystal composition, indicating a significant melt structural influence on trace element partitioning. The ratio of non-bridging oxygens to tetrahedrally coordinated cations (NBO/T) in the melt provides a measure of this effect; melt structure has a significant influence on trace element compatibility only for values of NBO/T less than ~0.49. This result suggests that when ascending peridotite intersects the solidus at relatively low pressures (~1.5 GPa or less), the compatibility of trace elements in the residual solid varies significantly during the initial stages of partial melting in response to the changing liquid composition. It is unlikely that this effect is important at higher pressures due to the increased compatibility of SiO₂, Na₂O, and Al₂O₃ in the residual peridotite, and correspondingly larger values of NBO/T for small degree partial melts.Editorial responsibility: T.L. Grove     

Deep solid-state equilibration and deep melting of plagioclase-free spinel peridotite from the slow-spreading Mid-Atlantic Ridge, ODP Leg 153

A petrological investigation of abyssal, plagioclase-free spinel peridotite drilled during ODP cruise 153 in the North Atlantic revealed that the peridotite represent refractory, partial residual mantle material that experienced depletion of incompatible trace elements during upper mantle melting. The degree of partial melting as estimated from spinel compositions was c. 12%. Fractionated middle and heavy rare earth elements imply polybaric melting, with c. 1–4% initial melting in the garnet peridotite stability field and subsequent partial melting of ~7–10% in the spinel peridotite stability field. Geothermobarometric investigations revealed that the solid-state equilibration of the spinel peridotite occurred at some 1,100–1,150°C and c. 20–23 kbar, corresponding to an equilibration depth of c. 70?±?5 km and an unusually low thermal gradient of some 11–17°C/km. A thermal re-equilibration of the peridotite occurred at ~850–1,000°C at similar depths. Naturally, the initial mantle melting in the garnet-peridotite stability field must have commenced at depths greater than 70?±?5 km. It is likely that the residual peridotite rose rapidly through the lithospheric cap towards the ridge axis. The exhumation of the abyssal peridotite occurred, at least in parts, via extensional detachment faulting. Given the shallow to moderate dip angles of the fault surfaces, the exhumation of the peridotite from its equilibration depth would imply an overall ridge-normal horizontal displacement of c. 50–160 km if tectonic stretching and detachment faulting were the sole exhumation mechanism.     

Geochemical evidence for melting of carbonated peridotite on Santa Maria Island, Azores

The islands of the Azores archipelago emerge from an oceanic plateau built on lithosphere increasing in age with distance from the Mid-Atlantic Ridge from 10 to 45 Ma. Here, we present the first comprehensive major and trace element and Sr–Nd–Pb isotope data from Santa Maria, the easternmost island of the archipelago, along with published data from the other Azores islands situated much closer to the Mid-Atlantic Ridge axis. We can show that the distinctively more variable and more enriched trace element ratios at Santa Maria combined with a relatively small range in Sr–Nd–Pb isotope ratios are the result of low degrees of partial melting of a common Azores mantle plume source underneath thicker lithosphere. This implies that melt extraction processes and melting dynamics may be able to better preserve the trace element mantle source variability underneath thicker lithosphere. These conclusions may apply widely for oceanic melts erupted on relatively thick lithosphere. In addition, lower Ti/Sm and K/La ratios and SiO₂ contents of Santa Maria lavas imply melting of a carbonated peridotite source. Mixing of variable portions of deep small-degree carbonated peridotite melts and shallow volatile-free garnet peridotite could explain the geochemical variability underneath Santa Maria in agreement with the volatile-rich nature of the Azores mantle source. However, Santa Maria is the Azores island where the CO₂-rich nature of the mantle source is more evident, reflecting a combination of a smaller extent of partial melting and the positioning at the edge of the tilted Azores mantle plume.     

Constraints on the origin of mantle-derived low Ca garnets

Current hypotheses for the source rock of low Ca garnets hosted in mantle-derived diamonds and xenoliths range from residues of komatiite generation, to subducted serpentinite, to subducted mid-ocean ridge (MORB) harzburgite. Experiments designed to test these hypotheses were undertaken. The stability and compositional variation of garnets at pressures above 4 GPa through the melting interval of hydrous peridotite, in the subsolidus of depleted harzburgite and peridotite compositions, and along the liquidus of aluminium-undepleted and aluminium-depleted komatiites were examined, and compared with petrological data for natural low Ca garnets. Partitioning of Cr between garnet and ultramafic liquid along the liquidus of komatiites and within the melting interval of peridotite, indicates that garnets in mantle residues after single stage Archean ultramafic liquid removal would contain 2 to 4 wt% Cr₂O₃. Thus, the more Cr-poor population of mantle-derived low Ca garnets, with Cr₂O₃ less than 4 wt%, could have originated by such a process. Experimental results for other compositions indicate that average cratonic peridotite or its hydrated equivalent is typically too Cr-poor to be the protolith from which low Ca garnets containing greater than 4 wt% Cr₂O₃ could have crystallized in the upper mantle. Experiments on a spinel harzburgite composition indicate that an extremely Cr-rich protolith (Cr/Cr+Al>0.3) is required to crystallize spinel and Cr-rich low Ca garnets, at pressures deduced for the ultramafic inclusion suite in diamonds (5 to 7 GPa). Natural examples of such Cr-rich protoliths are represented in some ophiolite harzburgites. All the experimental data taken together require that low Ca garnets with greater than 4 wt% Cr₂O₃ originated from residues that underwent multiple melt extraction. Whether such multi-stage events formed protoliths for low Ca garnets at shallow (i.e. MORB source region) or deep (i.e. komatiite source region) levels in the Precambrian mantle is not completely resolvable. The former environment can better account for the abundance of spinel in many diamonds hosting low Ca garnets, but the latter scenario best explains the presence of low Ca garnets in harzburgite xenoliths with cratonic bulk compositions well removed from typical MORB residues.     

Dihedral angle measurements and infiltration property of SiO<Subscript>2</Subscript>-rich melts in mantle peridotite assemblages

Petrological and experimental studies demonstrated that nepheline-normative, SiO₂-rich melts can be present in the upper mantle at pressures 1.5 GPa. To evaluate the role of such melts in mantle processes and magma genesis, we carried out two types of experiments: (1) melt distribution experiments to characterize the grain-scale distribution of a small fraction of typical SiO₂-rich mantle melt (SRMM) in polycrystalline olivine (Ol) at 1,180°C, 1.2 GPa; and (2) an infiltration experiment to test the ability of SRMM to impregnate and metasomatise neighbouring non-molten mantle rocks. The median dihedral angles at Ol-Ol-SRMM contacts are equal to 50°, implying that melt should be interconnected at all melt fractions. Complications arise, however, in the investigated system because Ol–liquid interfacial energy is anisotropic, and we estimate that the connectivity threshold in the SRMM–Ol system is 0.3 vol%. Regarding the very low volume fraction of SRMM in peridotites, we conclude that these melts either occur as isolated pockets or form a network of grain edge channels with a low degree of connectivity due to a large number of dry grain edges. Even in the case where an interconnected network exists, their large viscosities should prohibit the extraction of SRMM from peridotite sources. The infiltration experiment also points to a very reduced mobility of SRMM in the upper mantle. In this experiment, a slice of synthetic dunite was immersed into a magma reservoir made of 60 wt% SRMM+40 wt% Ol, and subjected to 1,180°C-1.2 GPa for 113 h: despite this long duration, the SiO₂-rich liquid was unable to infiltrate measurably the dunite. Our experiments do not support the hypothesis that SRMM represent agents of mantle metasomatism.     

Partial Melting Experiments of Peridotite + CO2 at 3 GPa and Genesis of Alkalic Ocean Island Basalts

We document compositions of minerals and melts from 3 GPa partialmelting experiments on two carbonate-bearing natural lherzolitebulk compositions (PERC: MixKLB-1 + 2·5 wt% CO₂; PERC3:MixKLB-1 + 1 wt% CO₂) and discuss the compositions of partialmelts in relation to the genesis of alkalic to highly alkalicocean island basalts (OIB). Near-solidus (PERC: 1075–1105°C;PERC3: 1050°C) carbonatitic partial melts with <10 wt%SiO₂ and 40 wt% CO₂ evolve continuously to carbonated silicatemelts with >25 wt% SiO₂ and <25 wt% CO₂ between 1325 and1350°C in the presence of residual olivine, orthopyroxene,clinopyroxene, and garnet. The first appearance of CO₂-bearingsilicate melt at 3 GPa is 150°C cooler than the solidusof CO₂-free peridotite. The compositions of carbonated silicatepartial melts between 1350 and 1600°C vary in the rangeof 28–46 wt% SiO₂, 1·6–0·5 wt% TiO₂,12–10 wt% FeO*, and 19–29 wt% MgO for PERC, and42–48 wt% SiO₂, 1·9–0·5 wt% TiO₂,10·5–8·4 wt% FeO*, and 15–26 wt% MgOfor PERC3. The CaO/Al₂O₃ weight ratio of silicate melts rangesfrom 2·7 to 1·1 for PERC and from 1·7 to1·0 for PERC3. The SiO₂ contents of carbonated silicatemelts in equilibrium with residual peridotite diminish significantlywith increasing dissolved CO₂ in the melt, whereas the CaO contentsincrease markedly. Equilibrium constants for Fe*–Mg exchangebetween carbonated silicate liquid and olivine span a rangesimilar to those for CO₂-free liquids at 3 GPa, but diminishslightly with increasing dissolved CO₂ in the melt. The carbonatedsilicate partial melts of PERC3 at <20% melting and partialmelts of PERC at 15–33% melting have SiO₂ and Al₂O₃ contents,and CaO/Al₂O₃ values, similar to those of melilititic to basaniticalkali OIB, but compared with the natural lavas they are moreenriched in CaO and they lack the strong enrichments in TiO₂characteristic of highly alkalic OIB. If a primitive mantlesource is assumed, the TiO₂ contents of alkalic OIB, combinedwith bulk peridotite/melt partition coefficients of TiO₂ determinedin this study and in volatile-free studies of peridotite partialmelting, can be used to estimate that melilitites, nephelinites,and basanites from oceanic islands are produced from 0–6%partial melting. The SiO₂ and CaO contents of such small-degreepartial melts of peridotite with small amounts of total CO₂can be estimated from the SiO₂–CO₂ and CaO–CO₂ correlationsobserved in our higher-degree partial melting experiments. Thesesuggest that many compositional features of highly alkalic OIBmay be produced by 1–5% partial melting of a fertile peridotitesource with 0·1–0·25 wt% CO₂. Owing to verydeep solidi of carbonated mantle lithologies, generation ofcarbonated silicate melts in OIB source regions probably happensby reaction between peridotite and/or eclogite and migratingcarbonatitic melts produced at greater depths. KEY WORDS: alkali basalts; carbonated peridotite; experimental petrology; ocean island basalts; partial melting     

Review of melting experiments on carbonated eclogite and peridotite: insights into mantle metasomatism

Experimental studies on the partial melting of eclogite and peridotite provide important clues on mantle metasomatism. Here, we review results from some of the recent experiments and show that melting of carbonated eclogite and peridotite can produce carbonatitic to carbonated silicate melt, in which carbonates melt preferentially before Ti oxides and silicates. Low-degree melting results in carbonatitic melt coexisting with Ti oxides and silicates. This process also leads to the fractionation between some high-field strength elements (Nb, Ta, Zr, Hf, and HREE) and highly incompatible elements (U and Th) in the melt. When Ti oxides are nearly exhausted in eclogite, extremely high TiO₂ contents (e.g. 19 wt.%) are present in the melt with marked concentration of Nb and Ta. These results help to explain the features of carbonatitic metasomatism and the Nb–Ta spike in oceanic island basalts as identified in experimental studies. These studies also explain the reducing conditions that stabilize diamond in the deep mantle (>150 km) as well as the occurrence of diamond at different depths reported in various studies. Melting in such a reduced mantle can happen through redox reaction between diamond, pyroxene, and olivine, in which the initial liquid is a carbonated silicate melt. However, the theoretical oxygen fugacity (fO₂) in the asthenosphere is much lower than that predicted by the reaction and requires elevated fO₂, which can be caused by the addition of relatively oxidized materials from the lower mantle, deep asthenospheric material, and various recycled components. A combination of these processes generates locally oxidized domains in the deep mantle.     

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

We performed modified iterative sandwich experiments (MISE) to determine the composition of carbonatitic melt generated near the solidus of natural, fertile peridotite + CO₂ at 1,200–1,245°C and 6.6 GPa. Six iterations were performed with natural peridotite (MixKLB-1: Mg# = 89.7) and ∼10 wt% added carbonate to achieve the equilibrium carbonatite composition. Compositions of melts and coexisting minerals converged to a constant composition after the fourth iteration, with the silicate mineral compositions matching those expected at the solidus of carbonated peridotite at 6.6 GPa and 1,230°C, as determined from a sub-solidus experiment with MixKLB-1 peridotite. Partial melts expected from a carbonated lherzolite at a melt fraction of 0.01–0.05% at 6.6 GPa have the composition of sodic iron-bearing dolomitic carbonatite, with molar Ca/(Ca + Mg) of 0.413 ± 0.001, Ca# [100 × molar Ca/(Ca + Mg + Fe*)] of 37.1 ± 0.1, and Mg# of 83.7 ± 0.6. SiO₂, TiO₂ and Al₂O₃ concentrations are 4.1 ± 0.1, 1.0 ± 0.1, and 0.30 ± 0.02 wt%, whereas the Na₂O concentration is 4.0 ± 0.2 wt%. Comparison of our results with other iterative sandwich experiments at lower pressures indicate that near-solidus carbonatite derived from mantle lherzolite become less calcic with increasing pressure. Thus carbonatitic melt percolating through the deep mantle must dissolve cpx from surrounding peridotite and precipitate opx. Significant FeO* and Na₂O concentrations in near solidus carbonatitic partial melt likely account for the ∼150°C lower solidus temperature of natural carbonated peridotite compared to the solidus of synthetic peridotite in the system CMAS + CO₂. The experiments demonstrate that the MISE method can determine the composition of partial melts at very low melt fraction after a small number of iterations.     

Mantle melting beneath Gakkel Ridge (Arctic Ocean): abyssal peridotite spinel compositions

The ultraslow spreading Gakkel Ridge represents one of the most extreme spreading environments on the Earth. Full spreading rates there of 0.6–1.3 cm/year and Na_8.0 in basalts of 3.3 imply an extremely low degree of mantle partial melting. For this reason, the complementary degree of melting registered by abyssal peridotite melting residues is highly interesting. In a single sample of serpentinized peridotite from Gakkel Ridge, we found spinels which, though locally altered, have otherwise unzoned and thus primary compositions in the cores of the grains. These reflect a somewhat higher degree of melting of the uppermost oceanic mantle than indicated by basalt compositions. Cr/(Cr+Al) ratios of these grains lie between 0.23 and 0.24, which is significantly higher than spinels from peridotites collected along the faster spreading Mid-Atlantic and Southwest Indian Ridges. Crustal thickness at Gakkel Ridge can be calculated from the peridotite spinel compositions, and is thicker than the crustal thickness of less than 4 km estimated from gravity data, or predicted from global correlations between spreading rate and seismically determined crustal thickness. The reason for this unexpected result may be local heterogeneity due to enhanced melt focussing at an ultraslow spreading ridge.     

Phase relations of a natural MARID composition and implications for MARID genesis,lithospheric melting and mantle metasomatism

Melting experiments were performed on a natural mica-amphibole-rutile-ilmenite-clinopyroxene (MARID) sample from the Kaapvaal mantle lithosphere (AJE137) at 20 to 35 kbar and 800 to 1450°C. A solidus was determined at 1260°C and 30 kbar above which phlogopite, clinopyroxene and olivine were stable with an alkali-rich silicate melt. Olivine is the only crystallizing phase just below the liquidus of the AJE137 bulk composition and K-richterite was only stable in the subsolidus region ( 1100°C at 30 kbar). These results are consistent with previous studies in more simple systems. In experiments with 10 wt% added water the solidus was depressed by ca. 300°C and K-richterite was stabilized above this solidus. MARIDs represent a potential lowtemperature component in the lithospheric mantle beneath the Kaapvaal Craton of southern Africa. The addition of > 10 wt% water (with less than a 120°C rise of temperature above the geotherm) to this mantle region would create conditions for the melting of this component. This may then be incorporated in any continental flood basalt parent magma that traverse this lithospheric mantle. The derivation of MARIDs from a silicate melt of their bulk composition, even if water saturated, is considered unlikely as such small degree melts could not sustain the elevated liquidus temperatures required (> 1200°C at 30 kbar) in a cold (< 800°C at 30 kbar) mantle lithosphere. MARID xenoliths may be produced by the interaction of an alkali-rich fluid with a peridotite or as the residue to a group II kimberlitic parent magma that has undergone fractionation of olivine and the exsolution of a carbonatite component.     

An experimental study of the grain-scale processes of peridotite melting: implications for major and trace element distribution during equilibrium and disequilibrium melting

The grain-scale processes of peridotite melting were examined at 1,340°C and 1.5 GPa using reaction couples formed by juxtaposing pre-synthesized clinopyroxenite against pre-synthesized orthopyroxenite or harzburgite in graphite and platinum-lined molybdenum capsules. Reaction between the clinopyroxene and orthopyroxene-rich aggregates produces a melt-enriched, orthopyroxene-free, olivine + clinopyroxene reactive boundary layer. Major and trace element abundance in clinopyroxene vary systematically across the reactive boundary layer with compositional trends similar to the published clinopyroxene core-to-rim compositional variations in the bulk lherzolite partial melting studies conducted at similar P–T conditions. The growth of the reactive boundary layer takes place at the expense of the orthopyroxenite or harzburgite and is consistent with grain-scale processes that involve dissolution, precipitation, reprecipitation, and diffusive exchange between the interstitial melt and surrounding crystals. An important consequence of dissolution–reprecipitation during crystal-melt interaction is the dramatic decrease in diffusive reequilibration time between coexisting minerals and melt. This effect is especially important for high charged, slow diffusing cations during peridotite melting and melt-rock reaction. Apparent clinopyroxene-melt partition coefficients for REE, Sr, Y, Ti, and Zr, measured from reprecipitated clinopyroxene and coexisting melt in the reactive boundary layer, approach their equilibrium values reported in the literature. Disequilibrium melting models based on volume diffusion in solid limited mechanism are likely to significantly underestimate the rates at which major and trace elements in residual minerals reequilibrate with their surrounding melt. Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorized users.