Multiple Saturation Point Research Articles

Assessing the accuracy of phase equilibrium software in reproducing the liquidus multiple saturation conditions of lunar and Martian basalt compositions

AbstractAccurate computational modeling allows the use of software as a first approach to some petrological problems that typically require experimentation, but most programs have not yet been fully tested for accuracy with lunar or Martian melt compositions. The programs pMELTS, MAGPOX, and Perple_X stand out for phase equilibrium modeling, as their calibrations include experiments of lunar compositions or have precise thermodynamic constraints for similar compositions. A set of lunar mare basalts, picritic glasses, and basaltic Martian compositions with known experimentally determined multiple saturation point (MSP) conditions were used here for phase equilibrium modeling. The accuracy of each program was tested through the determination of MSPs on the liquidus of the selected compositions. This point in pressure–temperature space can be considered as a direct proxy of the stable phases and the equilibrium conditions during partial melting of mantle sources. We identify a trend in experimental data between MSP temperature and MgO, CaO, and SiO2 concentrations, and similar trends are found in model results. However, only Perple_X is able to closely match the experimental data, despite the fact it does not accurately model ilmenite saturation for high‐Ti lunar basalts. We find that pMELTS miscalculates olivine saturation for MgO‐rich compositions and MAGPOX systematically underestimates MSP pressure and temperatures and can only be used when olivine is the liquidus phase. For modeling lunar or Martian basalt compositions, Perple_X can be used for optimal results, although no software is yet capable of bypassing the need to constrain MSP conditions through experimentation.

Experimental evidence for decompression melting of metasomatized mantle beneath Colima Graben, Mexico

Primitive subduction zone magmas provide information about the composition and thermal structure of the underlying mantle wedge. In the Colima Graben, Mexico, primitive lavas erupted from cinder cones range from high magnesium calc-alkaline basalts to high-K trachybasalts. This chemical diversity suggests that the sub-arc mantle wedge from which they derive is heterogeneous. To explore the conditions of magma generation in the wedge beneath Colima we used an inverse experimental approach to constrain multiple saturation points on the liquidus surface of a primitive high-K basanite (COM-1). Equilibrium piston-cylinder experiments were carried out between 1.0 and 2.4 GPa under hydrous (1.8–3.8 wt% H2O) and oxidizing (ƒO2 = − 0.5 to 4.3 log units relative to NNO) conditions. COM-1 + 3.8 wt% H2O is shown to be multiply-saturated with a phlogopite-bearing spinel pyroxenite assemblage (cpx + opx + phl + sp) close to its liquidus at 1.9–2.4 GPa and 1300 ºC. Experimental mapping of the liquidus surface reveals a multiple saturation point (MSP) where a lherzolitic phase assemblage of ol + cpx + opx + sp + phl coexist. The topology of the MSP indicates a peritectic of the form cpx + opx + phl + sp = liquid + ol. Four bracketing experiments define the MSP of COM-1 as 1300 ± 10 °C, 1.7 ± 0.1 GPa, ∆NNO = 3.4 ± 0.5 log units, for melt containing 3.6 ± 0.4 wt% H2O. The MSP olivine is too forsterite-rich (Fo92-94) to be in equilibrium with mantle lherzolite, but matches phenocryst core compositions in the natural basanite. Thus, experimental results indicate that COM-1 was produced by incongruent melting of an olivine-free, phlogopite-pyroxenite source that itself is the result of metasomatism of mantle wedge by slab-derived fluids. These conditions provide a valuable constraint on the thermal structure and chemical composition of the mantle wedge beneath Colima.

Melting processes and mantle sources of lavas on Mercury

The MESSENGER spacecraft provided geochemical data for surface rocks on Mercury. In this study, we use the major element composition of these lavas to constrain melting conditions and residual mantle sources on Mercury. We combine modelling and high-temperature (1320–1580 °C), low- to high-pressure (0.1 to 3 GPa) experiments on average compositions for the Northern Volcanic Plains (NVP) and the high-Mg region of the Intercrater Plains and Heavily Cratered Terrains (High-Mg IcP-HCT). Near-liquidus phase relations show that the S-free NVP and High-Mg IcP-HCT compositions are multiply saturated with forsterite and enstatite at 1450 °C – 1.3 GPa and 1570 °C – 1.7 GPa, respectively. For S-saturated melts (1.5–3 wt.% S), the multiple saturation point (MSP) is shifted to 1380 °C – 0.75 GPa for NVP and 1480 °C – 0.8 GPa for High-Mg IcP-HCT. To expand our experimental results to the range of surface compositions, we used and calibrated the pMELTS thermodynamic calculator and estimated phase equilibria of ∼5800 compositions from the Mercurian surface and determined the P–T conditions of liquid–forsterite–enstatite MSP (1300–1600 °C; 0.25–1.25 GPa). Surface basalts were produced by 10 to 50% partial melting of variably enriched lherzolitic mantle sources. The relatively low pressure of the olivine–enstatite–liquid MSP seems most consistent with decompression batch melting and melts being segregated from their residues near the base of Mercury's ancient lithosphere. The average melting degree is lower for the young NVP (0.27±0.04) than for the older IcP-HCT (0.46±0.02), indicating that melt productivity decreased with time. The mantle potential temperature required to form Mercurian lavas and the initial depth of melting also decreased from the older High-Mg IcP-HCT terrane (1650 °C and 360 km) to the younger lavas covering the NVP regions (1410 °C and 160 km). This evolution supports strong secular cooling of Mercury's mantle between 4.2 and 3.7 Ga and explains why very little magmatic activity occurred after 3.7 Ga.

Origin of the Apollo 14, 15, and 17 yellow ultramafic glasses by mixing of deep cumulate remelts

We examine the fO2-dependent melting conditions of the Apollo 14 yellow intermediate-Ti ultramafic glasses and the melting processes that formed the full suite of lunar yellow ultramafic glasses. Multiple saturation experiments indicate that the Apollo 14 yellow glass would have been in equilibrium with residual olivine and low-Ca pyroxene near 1530°C and 2.4GPa at ΔIW=+2. At ΔIW=−2, the multiple saturation point moves to greater depth and higher temperature to 1580°C and 3.0GPa. Combining the results of this study with that of Krawczynski and Grove (2012) on more Ti-rich Apollo orange and red glass indicates that the fO2-induced change in multiple saturation pressure correlates with the Fe–Ti# (molar (FeO+TiO2*)/(MgO+FeO+TiO2*), where TiO2*=all Ti calculated as Ti4+) of the liquid. Further, a decrease in the olivine Fe–Mg exchange coefficient at lower fO2 suggests that Fe2+ is complexing more efficiently with Ti3+ at the expense of Mg in the melt than it did with Ti4+ at higher fO2.Processes involving assimilation and/or fractionation and/or melt-wall rock reaction all fail to produce the within-suite compositional variability observed in the Apollo 14, 15, and 17 yellow glasses. Mixing of remelted source cumulates, combined with small amounts of olivine fractionation, are the only mechanisms that can reproduce all three trends. We present a quantitative model of the mixing process with simultaneous olivine fractionation. Remarkably, the trends can be explained by mixing melts of an ultramafic source (olivine+pigeonite or orthopyroxene), a clinopyroxene+ilmenite bearing cumulate, and KREEP. Lunar mantle overturn is the most likely process that can reconcile the observed major and trace element compositional characteristics and the experimental results. These two constraints are consistent with different models of lunar magma ocean crystallization (Snyder et al., 1992; Eklins-Tanton et al., 2011). A complex, hot thermal history is necessary to generate the magmas that mixed to form the yellow ultramafic glasses.

Petrogenesis of primitive and evolved basalts in a cooling Moon: Experimental constraints from the youngest known lunar magmas

We have conducted high-temperature experiments over a range of pressures to constrain the petrogenesis of the youngest sampled lunar magmas, which have contrasting primitive and evolved compositions. Our results indicate that at ∼3 Ga, melting still occurred within the same mantle depth range that produced crystalline mare basalts for the previous ∼1 Ga, although our data cannot support or confirm that the shallowest extents of melting moved deeper into the mantle by ∼3 Ga, as is predicted by most thermal evolution models. Furthermore, melting still occurred in regions with low abundances of heat-producing elements. Basaltic lunar meteorite NEA 003A has some of the lowest abundances of incompatible trace elements among all mare basalts and no negative Eu anomaly. Our experiments show that NEA 003A is multiply saturated with olivine and low-Ca pyroxene on its liquidus at ∼1.1 GPa (∼215 km) and ∼1330 °C. If the primitive NEA 003A liquid composition is a minimally-modified melt, the relatively low Mg# of its source region (73–75), its lack of a Eu anomaly, and its chondritic initial Nd isotopic composition indicate its source region likely escaped mixing during mantle overturn with later-stage magma ocean cumulates that formed after plagioclase saturation. This condition would require the sources of the ultramafic glasses to have experienced cumulate mixing, or for assimilation of later-stage magma ocean cumulates by the ultramafic glass parental magmas before eruption in order to account for their higher Mg#'s and deeper negative Eu anomalies. Alternatively, NEA 003A may have undergone some fractional crystallization, in which case its more primitive source region would be deeper than 215 km and may approach the depth range of the ultramafic glass source regions. Iron- and incompatible trace element-rich basaltic lunar meteorites LAP 02205, NWA 032/479, and NWA 4734 have nearly identical bulk compositions and have a multiple saturation point on their high pressure liquidus at <0.6 GPa (<105 km) and 1150–1200 °C. The P–T conditions of this multiple saturation point are inconsistent with the expectation from thermal modeling of a thick, cold lunar elastic lithosphere at 3 Ga, and likely indicate these melts underwent fractional crystallization before eruption, thus increasing their abundances of incompatible trace elements. However, these LREE-enriched basalts possess a very deep negative Eu-anomaly but do not show other chemical fingerprints of KREEP, indicating that a component of Fe-rich, low-Ti late-stage magma ocean cumulates are required in their hybridized source region.

Melting phase relations of a mica–clinopyroxenite from the Milk River area, southern Alberta, Canada

Melting experiments were conducted on a mica–clinopyroxenite xenolith brought up in a minette dyke in southern Alberta, Canada, near Milk River. Both the minettes and mica–clinopyroxenite xenoliths were studied by Buhlmann et al. (Can J Earth Sci 37:1629–1650, 2000), who hypothesized that the minettes formed by partial melting of a mantle source containing clinopyroxene + phlogopite ± olivine, at pressures ≥1.7 GPa. In liquidus experiments performed on the most primitive minette in our previous study (Funk and Luth in Contrib Mineral Petrol 164:999–1009, 2012), we found a multiple saturation point where olivine and orthopyroxene coexisted with liquid at 1.77 GPa and 1,350 °C. We argued that the minette originally formed by partial melting of clinopyroxene + phlogopite, but had re-equilibrated with a harzburgite during ascent. In the current study, we wanted to test both the source region hypothesis of Buhlmann et al. and our re-equilibration hypothesis by studying the near-solidus phase equilibria of a mica + clinopyroxene assemblage. We found the solidus for our xenolith has a steep slope in P–T space and lies at temperatures above those of a normal cratonic geotherm, implying that this mica–clinopyroxenite is stable in the cratonic mantle. Melting could occur at greater depths, where the solidus is extrapolated to cross the geotherm or must be induced either by raising the temperatures of the surrounding rocks or by introducing hydrous fluids into the source. Our melts are in equilibrium with clinopyroxene and olivine. The compositions of the liquids derived from melting this xenolith are similar to madupitic lamproites from the Leucite Hills, Wyoming, studied by Carmichael (Contrib Mineral Petrol 15:24–66, 1967) and Barton and Hamilton (Contrib Mineral Petrol 66:41–49, 1978; Contrib Mineral Petrol 69:133–142, 1979). Barton and Hamilton (Contrib Mineral Petrol 69:133–142, 1979) proposed that the madupitic lamproites may have come from a source containing mica and pyroxene. This study supports their hypothesis. The composition of the most primitive minette from southern Alberta lies between our experimental melt and a population of representative mantle orthopyroxenes. We conclude from our study that the Milk River minettes were likely derived from a source containing phlogopite, clinopyroxene and trace amounts of apatite, which formed olivine upon melting. During ascent, the melts changed composition by reacting with orthopyroxene.

Experimental petrology of the Apollo 15 group A green glasses: Melting primordial lunar mantle and magma ocean cumulate assimilation

Phase equilibrium experiments have been performed to place constraints on the melting processes in the lunar interior using the Apollo 15 group A ultramafic green glass, the most abundant pyroclastic glass composition found in the polymict regolith breccia (15426) near Hadley Rille at station 7. Experiments on two compositional end members of the 15A green glass trend yield experiments saturated with olivine and orthopyroxene on the liquidus at 1520°C and 2.1GPa. Experiments were also performed on high-Al and -Ca analog compositions and on mixtures of the green glass compositions and natural minerals (olivine, orthopyroxene and garnet) in order to obtain experimental liquids multiply saturated with olivine, orthopyroxene, clinopyroxene and garnet. Three liquid compositions in equilibrium with a garnet-lherzolite residual phase assemblage were produced at 2.35GPa, 1440 and 1420°C and at 2.6GPa, 1460°C.Observed compositional variations in the 15A glasses rule out any simple partial melting process that would involve a olivine+orthopyroxene bearing source region. The results of these experiments are used to test petrogenetic models for the origin of the 15A green glasses. Synergistic evidence from major and trace element geochemistry, geophysical data on the lunar interior structure and results from models of lunar magma ocean differentiation is used to develop a model for the origin of the 15A ultramafic glasses. We propose that a mixing process involving a primary magma, derived by melting primordial lunar mantle, and remelted late-stage magma ocean cumulates led to the origin of the 15A glass. The model uses the 15A green glass multiple saturation point at 1520°C and 2.1GPa to constrain the depth of a cumulate remelting process. Decompression partial melting of undifferentiated lunar mantle, produced by adiabatic upwelling of the deep, hot interior of the moon, produced melts that infiltrated and selectively melted late stage magma ocean cumulates that had sunk to the base of a lunar magma ocean. The compositional variations observed in the 15A glasses are consistent with this model.

An experimental study of a minette from the Milk River area, southern Alberta, Canada

Buhlmann et al. (Can J Earth Sci 37: 1629–1650, 2000) studied the minettes and xenoliths from the Milk River area of southern Alberta, Canada. Based on previous work, they hypothesized that the minettes were derived from a source containing phlogopite + clinopyroxene ± olivine, at pressures ≥1.7 GPa. To test this hypothesis, liquidus experiments were performed on a primitive minette between 1.33 and 2.21 GPa and between 1,300 and 1,400 °C to constrain the mineralogy of its source region. We found a multiple saturation point along the liquidus at 1.77 GPa and 1,350 °C, where the liquid coexists with orthopyroxene and olivine. Neither phlogopite nor clinopyroxene were found to be liquidus phases, which is inconsistent with Buhlmann et al.’s hypothesis. We suggest that our minette is not primary, but had re-equilibrated with harzburgitic mantle subsequent to formation. In such a scenario, partial melting of a veined source containing mica and clinopyroxene occurred at or near the base of the Wyoming craton (~200 km). Minimal heating or the introduction of hydrous fluids into the source would be required to induce partial melting. Rapid ascent rates, coupled with slow cooling rates, of the “primary minette magma” would preserve the high temperature observed in our experiments. At ~58 km, our “primary minette magma” likely stalled and re-equilibrated with the harzburgite surroundings.

Compressibility of molten Apollo 17 orange glass and implications for density crossovers in the lunar mantle

We performed density measurements on a synthetic equivalent of lunar Apollo 17 74,220 “orange glass”, containing 9.1 wt% TiO 2, at superliquidus conditions in the pressure range 0.5–8.5 GPa and temperature range 1723–2223 K using the sink/float technique. In the lunar pressure range, two experiments containing pure forsterite (Fo 100) spheres at 1.0 GPa and 1727 K, and at 1.3 GPa–1739 K, showed neutral buoyancies, indicating that the density of molten orange glass was equal to the density of Fo 100 at these conditions (3.09 ± 0.02 g cm −3). A third tight sink/float bracket using Fo 90 spheres corresponds to a melt density of 3.25 ± 0.02 g cm −3 at ∼2.8 GPa and ∼1838 K. Our data predict a density crossover for the molten orange glass composition with equilibrium orthopyroxene at ∼2.8 GPa, equivalent to a depth of ∼600 km in the lunar mantle, and a density of ∼3.25 g cm −3. This crossover depth is close to the orange glass multiple saturation point, representing its minimum formation depth, at the appropriate oxygen fugacity (2.8–2.9 GPa). A density crossover with equilibrium olivine is predicted to fall outside the lunar pressure range (>4.7 GPa), indicating that molten orange glass is always less dense than its equilibrium olivines in the Moon. Our data therefore suggest that that lunar liquids with orange glass composition are buoyant with respect to their source region at P < ∼2.8 GPa, enabling their initial rise to the surface without the need for additional external driving forces. Fitting the density data to a Birch–Murnaghan equation of state at 2173 K leads to an array of acceptable solutions ranging between 16.1 and 20.3 GPa for the isothermal bulk modulus K 2173 and 3.6–8 for its pressure derivative K′, with best-fit values K 2173 = 18.8 GPa and K′ = 4.4 when assuming a model 1 bar density value of 2.86 g cm −3. When assuming a slightly lower 1 bar density value of 2.84 g cm −3 we find a range for K 2173 of 14.4–18.0 and K′ 3.7–8.7, with best-fit values of 17.2 GPa and 4.5, respectively.

Flowsheet synthesis of fractional crystallization processes with cake washing

A flowsheet synthesis methodology is presented for the design of fractional crystallization processes. The methodology is based on four interconnected/nested networks. The first network is derived by identifying a set of thermodynamic states providing a basis for feasible separation alternatives. In this network, the nodes correspond to multiple saturation points, solute intermediates, and process feeds and end products. The second network type represents the variety of tasks that can be performed at each multiple saturation point. These tasks include cooling crystallization, evaporative crystallization, reactive crystallization, dissolution, and leaching. Heat integration is incorporated using the transshipment formulation to represent the heat exchanger network. The fourth network represents the filtration and cake washing alternatives. The cake wash and task networks are modeled using disjunctive programming and then converted into a mixed-integer program. The method is applied to the design of three salt separation example problems. The optimal design costs are seen to depend on design parameters such as product impurity level and liquid retention in the cake. The advantage of an integrated synthesis approach is demonstrated.

Experimental and petrological constraints on lunar differentiation from the Apollo 15 green picritic glasses

Abstract— Phase equilibrium experiments on the most magnesian Apollo 15C green picritic glass composition indicate a multiple saturation point with olivine and orthopyroxene at 1520°C and 1.3 GPa (about 260 km depth in the moon). This composition has the highest Mg# of any lunar picritic glass and the shallowest multiple saturation point. Experiments on an Apollo 15A composition indicate a multiple saturation point with olivine and orthopyroxene at 1520°C and 2.2 GPa (about 440 km depth in the moon). The importance of the distinctive compositional trends of the Apollo 15 groups A, B, and C picritic glasses merits the reanalysis of NASA slide 15426,72 with modern electron microprobe techniques. We confirm the compositional trends reported by Delano (1979, 1986) in the major element oxides SiO2, TiO2, Al2O3, Cr2O3, FeO, MnO, MgO, and CaO, and we also obtained data for the trace elements P2O5, K2O, Na2O, NiO, S, Cu, Cl, Zn, and F.Petrogenetic modeling demonstrates that the Apollo 15 A‐B‐C glass trends could not have been formed by fractional crystallization or any continuous assimilation/fractional crystallization (AFC) process. The B and C glass compositional trends could not have been formed by batch or incremental melting of an olivine + orthopyroxene source or any other homogeneous source, though the A glasses may have been formed by congruent melting over a small pressure range at depth. The B compositional trend is well modeled by starting with an intermediate A composition and assimilating a shallower, melted cumulate, and the C compositional trend is well modeled by a second assimilation event. The assimilation process envisioned is one in which heat and mass transfer were separated in space and time. In an initial intrusive event, a picritic magma crystallized and provided heat to melt magma ocean cumulates. In a later replenishment event, the picritic magma incrementally mixed with the melted cumulate (creating the compositional trends in the green glass data set), ascended to the lunar surface, and erupted as a fire fountain. A barometer created from multiple saturation points provides a depth estimate of other glasses in the A‐B‐C trend and of the depths of assimilation. This barometer demonstrates that the Apollo 15 A‐B‐C trend originated over a depth range of ˜460 km to ˜260 km within the moon.

Separation system synthesis of fractional crystallization processes with heat integration

A methodology is presented for the synthesis of fractional crystallization processes with heat integration. The methodology is based on the construction of three networks. The first network is based on the identification of feasible thermodynamic states. Using equilibrium data for a candidate set of potential operation point temperatures, a network flow model is constructed to represent the set of potential separation flowsheet structures that can result. In this network the nodes correspond to multiple saturation points, solute intermediate, process feeds and end products. The second network is used to represent the variety of tasks that can be performed at each multiple saturation point. Multiple saturation nodes can be used for different tasks depending on the characteristic of the input and output streams. These tasks include cooling crystallization, evaporative crystallization, reactive crystallization, dissolution, and leaching, This multiple task condition for each equilibrium state is modelled using disjunctive programming and then converted into a mixed integer program. Heat integration is included using a heat exchanger network which can be regarded as a transhipment problem. The method is illustrated through the design of two salt separation examples.

Origin of lunar ultramafic green glasses: constraints from phase equilibrium studies

Phase equilibria experiments on an Apollo 14B green glass composition are used to develop a petrogenetic model for the Apollo 14B green glasses. New compositional data for the Apollo 14B green glasses are presented, which replace the original data set of Delano (1986). Near-liquidus phase relations of the 14B green glass are determined from 1.3 to 2.7 GPa and 1380 to 1570°C. The liquidus is saturated with olivine to 2.4 GPa and orthopyroxene above 2.4 GPa. A multiple saturation point with olivine and orthopyroxene occurs at 1560°C and 2.4 GPa. Crystallization of the ol + opx saturation boundary involves 40 wt% ol and 60 wt% opx, and occurs over a 140°C interval before cpx appears at a reaction boundary. Crystallization/melting models are developed to reproduce the compositional variations displayed by the Apollo 14B glasses using these phase equilibrium constraints. Fractional crystallization alone cannot account for the observed variation in the major element abundance of the Apollo 14B green glasses. Assimilation is also necessary, in particular, to fit variations in TiO 2, FeO, Na 2O and Al 2O 3. The 14B glasses show evidence of simultaneous high-Ti assimilation and fractional crystallization of olivine, followed by simultaneous KREEP assimilation and olivine fractionation. The data requires the high-Ti assimilant to be at a greater depth than the KREEP assimilant, but only constrains the total assimilation/fractional crystallization process to occur between approximately 2.3 and 0.5 GPa. Assimilation of a high-Ti cumulate within this depth range is consistent with an overturn/hybrid mantle remelting model. Highlands crust assimilation is precluded for the Apollo 14B green glasses.

Separation System Synthesis for Fractional Crystallization from Solution Using a Network Flow Model

A method is presented to synthesize process flowsheets for separations of mixtures by fractional crystallization. Using equilibrium data for a candidate set of potential operating point temperatures, a network flow model is constructed to represent the set of potential separation flowsheet structures that can result. By employing specified approaches to multiple saturation point conditions, linear network constraints are obtained. Solution of the network flow model shows the optimal mass flow pattern between the candidate equilibrium states, and from this the corresponding process flowsheet is readily deduced. The method as presented is generally applicable to problems with two salts and one or more solvents, including systems forming one or more multiple salts or hydrates. Situations having multiple feeds and multicomponent products are also included. Several salt separation examples are given which demonstrate the method's application and show some of the types of coupled cycles that are obtained as sol...

Melting experiments on a synthetic olivine lamproite composition up to 8 GPa: Implication to its petrogenesis

We have determined melting phase relationships on a synthetic olivine lamproite composition with H2O contents from 6 to 12 wt % in the pressure range of 3.5 to 8 GPa and at temperatures of 1100 to 1500°C. The composition with 6 wt % H2O is multiply saturated with olivine, orthopyroxene, and garnet at 5.3 GPa and 1370°C, indicating that olivine lamproite can coexist with garnet harzburgite. The Mg # (molar 100Mg/Mg+Fe) of olivine, orthopyroxene, and garnet in experiments that closely bracket the multiple saturation point are 94.3, 93.7, and 88.9, respectively, and the garnet contains 3 wt % CaO. This residual mineral assemblage corresponds to low‐Ca garnet harzburgite. Phlogopite crystallized at a temperature 150°C below the multiple saturation point, indicating that all phlogopite in the source is consumed during partial melting prior to magma segregation. If the conditions of multiple saturation represent those of magma segregation, then the source region of the olivine lamproite exists near the base of continental lithosphere. Metasomatized, depleted harzburgite with veins containing centimeter‐sized phlogopite grains have been reported to occur in mantle xenoliths from kimberlites. We propose that olivine lamproite magma is formed by preferential melting of phlogopite veins, with subsequent equilibration of the magma with surrounding garnet harzburgite.

Anhydrous and water‐saturated melting experiments of an olivine andesite from Mt Yakushi‐Yama, northeastern Shikoku, Japan

Abstract Melting experiments have been carried out on an olivine andesite of Mt Yakushi‐Yama from the Miocene Setouchi volcanic belt in northeastern Shikoku, Japan. This andesite has been characterized by a low ratio of FeO*/Mg° (= 0.78). Phase relations have been determined within the pressure range of 2.8 to 19.3 kbar at 1000‐1300°C under anhydrous and water‐saturated conditions. At pressures less than 8.8 kbar, olivine is a liquidus phase. Orthopyroxene appears on the liquidus at 9.3 kbar under the anhydrous conditions. The multiple saturation point rises up to 17.5 kbar for water‐saturated experiments. The andesite melt coexists with olivine and orthopyroxene just below the liquidus at 8.8–9.3 kbar and 1230°C for dry conditions, and at 17.5 kbar and 1060°C under water‐saturated conditions. These experimental results indicate that the Yakushi‐Yama olivine andesite magma could coexist with a harzburgitic mantle at depths between about 30 and 60 km, and at temperatures between 1060 and 1230°C. Experimental data also suggest a possibility that a high magnesian andesite magma would be generated by a direct partial melting of the uppermost harzburgitic mantle under anhydrous conditions.

Phase relations of aluminum‐undepleted and aluminum‐depleted komatiites at pressures of 4–12 GPa

The anhydrous phase relations of natural Al‐undepleted komatiite (AUK) with 25% MgO and Al‐depleted komatiite (ADK) with 32% MgO have been investigated in the pressure range from 4 to 12 GPa, using samples from Munro Township, Ontario, and the Barberton Mountains, South Africa. The experiments were carried out with graphite or molybdenum capsules in 18‐mm octahedral pressure cells in a uniaxial split‐sphere multianvil apparatus. The thermal gradient within the stepped resistance heaters of graphite and lanthanum chromite was about 50°C/mm. For both compositions, the liquidus phase is olivine at low pressure (up to 5.6 and 9.6 GPa for the AUK and ADK, respectively) and garnet at high pressure. Pyroxene is the only liquidus phase in an intermediate pressure range for the ADK (9.6–11.5 GPa) and reaches the liquidus at a multiple saturation point with olivine and garnet at 5.6 GPa pressure for the AUK. The subsolidus assemblage consists of olivine, pyroxene, and garnet in the entire pressure range for both of the compositions. The pyroxenes have variable CaO contents, and most of them are subcalcic clinopyroxenes, but minor amounts of orthopyroxene have been observed in some of the run products. If AUK and ADK compositions represent near primary magmas, the melt separation most likely occurred at pressures corresponding to the three phase saturation point at 5.6 GPa for AUK and one of the two cosaturation points (olivine‐pyroxene at 9.6 GPa and pyroxene‐garnet at 11.6 GPa) for the ADK. The low Ca/Al ratio of the AUK requires a nearly complete melting of garnet. The high Ca/Al ratio of the ADK indicates that garnet was a residual phase at the time of melt separation, or that the mantle source experienced some premelting or synmelting garnet fractionation. The clinopyroxene liquidus interval between 9.6 and 11.6 GPa excludes the coexistence of garnet and olivine in the residue, and the following alternative petrogenetic models are suggested for the ADK: 1. The initial melt was more magnesian than the source peridotite, and the melting progressed to the stage of disappearance of the residual olivine. The melting progressed further along the pyroxene‐garnet cotectic before melt separation. 2. Garnet fractionation took place from an ascending and partially molten peridotite diapir. Further melting of the upper portion of the diapir produced a melt that was less magnesian than the source diapir. This could lead to total melt consumption of any remaining residual garnet, and further melting could proceed along the olivine‐pyroxene cotectic before final melt separation.