A double partial melt zone in the mantle beneath mid-ocean ridges

For a lherzolite mantle with about 0.1 wt.-percent CO₂ or less, and a CO₂/H₂O mole ratio greater than about one, the mantle solidus curve in P-T space will have two important low-temperature regions, one centered at about 9 kbar (30 km depth) and another beginning at about 28 kbar (90 km depth). It is argued that the depth of generation of primary tholeiitic magmas beneath ridge crests is about 9 kbar, and that the geotherm changes from an adiabatic gradient at greater pressures to a strongly superadiabatic gradient at lesser pressures. Such a ridge geotherm would intersect the solidus at two separate depth intervals corresponding to the two low-temperature regions on the solidus. With increasing age and cooling of the lithosphere, the shallow partial melt zone would pinch out and the thickness of the deep partial melt zone would decrease. With increasing depth in a mature oceanic lithosphere, the rock types would consist of depleted harzburgite from directly beneath the crust to about 30 km depth, fertile spinel lherzolite from about 30 km to 50–60 km, and fertile garnet lherzolite from about 50–60 km to the top of the deep partial melt zone at about 90 km.     

Melting relationships in CaO-CO2 and MgO-CO2 to 36 kilobars with comments on CO2 in the mantle

The melting curves of CaCO₃ and MgCO₃ have been extended to pressures of 36 kb by experiments in piston-cylinder apparatus. At 30 kb, the melting temperatures of calcite and magnesite are 1610°C and 1585°C, respectively. New data for the magnesite dissociation reaction permit the location of an invariant point for the assemblage magnesite + periclase + liquid + vapor near 26 kb-1550°C. New data are also presented for the calcite-aragonite transition at 800°C, 950°C and 1100°C. At pressures above 36–50 kb, calcite and magnesite melt at temperatures lower than the solidus of dry mantle peridotite. Natural and experimental evidence suggests that carbon dioxide in the Earth's mantle could be present in a variety of forms: (a) a free vapor phase, (b) vapor dissolved in silicate magma, (c) crystalline carbonate, (d) carbonatite liquid, (e) carbon-bearing silicate analogs, or (f) carbonato-silicates (such as scapolite, spurrite, tilleyite, and related compounds).     

Discussion of recent papers on carbonated peridotite, bearing on mantle metasomatism and magmatism: response

The solidus temperature for peridotite-CO₂ drops sharply within a narrow temperature interval, with increasing pressure, to an invariant point where dolomite becomes stable at the solidus. This geometry presents a barrier to some volatile-charged magmas rising through the lithosphere, with many petrological applications. The phase relationships are based on results in the model system CaOMgOSiO₂CO₂, and peridotite-CO₂H₂O. Two published sets of experimental results for the solidus of amphibole-dolomite-peridotite are here interpreted in terms of the topology of the system peridotite-CO₂H₂O, with the conclusion that the two results yield quite different values for the pressure of the significant invariant point in peridotite-CO₂: 17 kbar or 27 kbar (equivalent to 30 km in the upper mantle). Better definition of this pressure, and clarification of the topology of the intersecting stability fields for dolomite, amphibole and phlogopite with the peridotite-vapor solidus, are essential for understanding magmatic and metasomatic processes in the lithosphere, and the petrogenesis of alkaline magmas and associated carbonatites.     

Experimental investigation on forsterite-grossularite incompatibility in presence of excess water

A mixture containing equal amounts of forsterite and grossularite by weight (Fo₅₀Gr₅₀) has been studied at temperatures between 750 and 1400°C under pressures ranging from 6 to 25 kbar in presence of excess water. The assemblages noted under low pressure (<8 kbar) are as follows: Diopside_ss+forsterite_ss+monticellite_ss+vapor and Diopside_ss+forsterite_ss+monticellite_ss+liquid+vapor. (ss denotes solid solution) Under intermediate pressures between 8 and 24 kbar following assemblages were noted in the order of increasing temperature: Diopside_ss+forsterite_ss+spinel+vapor, Diopside_ss+forsterite_ss+spinel+liquid+vapor, Diopside_ss+forsterite_ss+liquid+vapor, and Forsterite_ss+liquid+vapor. At pressures above 24 kbar the assemblages are as follows: Diopside_ss+forsterite_ss+garnet+vapor, Diopside_ss+forsterite_ss+garnet+liquid+vapor, Diopside_ss+forsterite_ss+liquid+vapor, and Forsterite_ss+liquid+vapor. Electron microprobe analyses of diopside and forsterite crystallized at 1050°C and 23 kbar, show that the former contains 6 to 6.5 wt % of Al₂O₃ as solid solution whereas the latter incorporates 1.3 wt % of monticellite in solid solution. The monticellite content of forsterite increases at low pressures at a given temperature to about 6 wt % at 1050°C and 6 kbar. The study indicates that forsteritic olivine does not coexist with pure grossularite in the studied temperature and pressure ranges, although the former is in equilibrium with pyrope-rich garnet, containing 23 mole % grossularite. The study supports the conclusion ofWerner andLuth (1973) that the solubility of monticellite in forsterite decreases with increasing pressure at a given temperature. The results of the investigation are also in agreement with the findings ofKushiro andYoder (1966), who noted that spinel peridotites found in folded belts and in alkalic basalts are produced under intermediate pressures, whereas garnet peridotite xenoliths found in kimberlite and in orogenic belts are formed at high pressures.     

Melting experiments on a high-magnesian andesite

Melting experiments were conducted on a high-magnesian bronzite olivine andesite (Teraga-Ike andesite) which is considered to be a primary andesite. The high-magnesian andesite magma is in equilibrium with both olivine and orthopyroxene at about 15.5 kbar and 1080°C under H₂O-saturated conditions and at lower pressure and higher temperature under H₂O-undersaturated conditions. This suggests that high-magnesian andesites could be generated by the partial melting of upper mantle peridotite containing a small amount of H₂O.     

Basalt-andesite-rhyolite-H2O: Crystallization intervals with excess H2O and H2O-undersaturated liquidus surfaces to 35 kolbras,with implications for magma genesis

Three rocks representing the calc-alkaline rock series gabbro-tonalite-granite or basalt-andesite-rhyolite were reacted with varying percentages of water in sealed capsules between 600 and 1300°C and pressures to 36 kbars, corresponding to depths of more than 120 km within the earth. For each rock we present complete P-T diagrams with excess water, and the water-undersaturated liquids surface projected from P-T-X_H2O space mapped with contours for constant H₂O contents and with the fields for near-liquidus minerals. All changes in liquidus and solidus slopes can be correlated with changes in mineralogy from less dense to more dense, or with expansion of crystallization fields, without appeal to changes in molar volume of H₂O in liquid and vapor phases. The results indicate that tholeiites and andesites of the calc-alkaline series with compositions similar to the rocks studied are not primary magmas from mantle peridotite at depths greater than about 50 km. Primary andesitic magmas from shallower levels would require very high water contents and we do not believe such magmas could normally reach the surface. The liquids results are consistent with the derivation of andesites with little dissolved water as primary magmas from subducted ocean crust (quartz eclogite), but multi-stage models are preferred. Temperatures required for the generation of andesites by fusion of continental crust are higher than considered reasonable. The evidence precludes the generation of primary rhyolites or granites from the mantle of subducted oceanic crust at mantle depths. Primary rhyolite or granite magmas with moderate water contents (saturated or undersaturated) can be generated in the crust at reasonable temperatures, and could reach near-surface levels before vesiculation. Water-undersaturated granite liquid with residual crustal minerals could constitute plutonic magmas of intermediate composition.     

New data on magmatic H2O contents of pantellerites,with implications for petrogenesis and eruptive dynamics at Pantelleria

Infrared spectroscopic analyses of melt inclusions in quartz phenocrysts from pantellerites erupted at Pantelleria, Italy, show that the magmas contained moderate pre-eruptive H₂O contents, ranging from 1.4 to 2.1 wt.%. Melt H₂O concentrations increase linearly with incompatible elements, demonstrating that H₂O contents were not buffered significantly during fractionation by any crystalline or vapor phase. The relatively low H₂O contents of pantellerites are consistent with an origin by partial melting of alkali gabbros rather than fractional crystallization of basalt. Preeruptive H₂O concentrations do not correlate with the volume or explosivity of pantellerite eruptions; decompression history is critical in determining the style of pantellerite (and other) eruptions.     

Phase relations in the system LiFMgF2 at elevated pressures: Implications for the proposed mixed-oxide zone of the earth's mantle

Solvi and liquidi for various LiFMgF₂ mixtures have been determined at pressures up to 40 kbar by differential-thermal-analysis in a piston-cylinder high-pressure device. The melting curves of pure LiF and MgF₂ were also studied and the initial slopes (dT_m/dP)_{P = 0} were found to be 11.2 and 8.3°C/kbar, respectively. The eutectic composition (LiF)_0.64(MgF₂)_0.36 is independent of pressure to 35 kbar and the eutectic temperature rises approximately 6.3°C per kbar. Initial slopes of 11°C/kbar and 35°C/kbar are inferred for the melting curves of MgO and SiO₂ (stishovite) respectively, on the basis of data for their structural analogue compounds. The observed solid solution of LiF in MgF₂ and other evidence suggest the possibility of solid solution in the system (Mg,Fe)OSiO₂ (stishovite) under mantle conditions which may have important consequences for the elastic properties of a “mixed-oxide” zone of the earth's mantle.     

Origin of high-magnesian andesites in the Setouchi volcanic belt,southwest Japan,II. Melting phase relations at high pressures

Melting phase relations of an augite-olivine high-magnesian andesite and an augite-olivine basalt from the Miocene Setouchi volcanic belt in southwest Japan have been studied under water-saturated, water-undersaturated and under anhydrous conditions. Both the andesite and the basalt are characterized by low FeO*/MgO ratios (0.86 and 0.76 in weight, respectively) and qualify as primary magmas derived from the upper mantle.The andesite melt coexists with olivine, orthopyroxene and clinopyroxene at 15 kbar and 1030°C under water-saturated conditions, and at 10 kbar and 1070°C under water-undersaturated conditions (7 wt.% H₂O in the melt). The basalt-melt also coexists with the above three phases at 11 kbar and 1305°C under anhydrous conditions, and at 15 kbar and 1205°C in the presence of 4 wt.% water.Present studies indicate that high-magnesian andesite magmas may be produced even under water-undersaturated conditions by partial melting of mantle peridotite. It is suggested that two types of high-magnesian andesites in the Setouchi volcanic belt (augite-olivine and bronzite-olivine andesites) were produced by different degrees of partial melting; augite-olivine andesite magmas, whose mantle residual is lherzolite, were formed by lower degrees of partial melting than bronzite-olivine andesite magmas, which coexist with harzburgite. The basalt magmas, which were often extruded in close proximity to the high-magnesian andesite magmas, are not partial melting products of a mantle peridotite which had previously melted to yield high-magnesian andesite magmas.     

Melting of enstatite from 13 to 18 GPa under hydrous conditions

Experiments on MgSiO₃ enstatite were conducted in the pressure range from 13 to 18 GPa under hydrous conditions in order to clarify the effect of water on the melting phase relations of enstatite at pressures corresponding to the Earth’s mantle transition zone. In some previous experiments [Geol. Soc. Am. Bull. 79 (1968) 1685; Phys. Earth Planet. Inter. 85 (1994) 237], incongruent melting behavior to form Mg₂SiO₄ forsterite and SiO₂ enriched liquid up to 5 GPa was observed, and congruent melting behavior at pressures up to 12 GPa was observed. Under hydrous conditions, we found that the melting reaction changes from congruent to incongruent at around 13.5 GPa. Liquid formed above 13.5 GPa is enriched in MgO component relative to MgSiO₃ because it coexists with stishovite (SiO₂). Moreover, the solidus temperature decreases drastically at around 13.5 GPa, in unison with the change in the melting reaction. The solidus temperature is about 1400 °C at 13 GPa, but approximately 900 °C at 15 GPa. Our results show that the liquidus phase changes from clinoenstatite to stishovite with increasing pressure and water content above 13.5 GPa. MgSiO₃ enstatite is one of the major constituent minerals in the Earth’s mantle, and it is expected that MgO-enriched liquid will be generated in the transition zone if water is present.     

Melting of hydrous upper mantle and possible generation of andesitic magma: An approach from synthetic systems

Phase equilibria in a portion of the system forsterite-plagioclase (An₅₀Ab₅₀ by weight)-silica-H₂O have been determined at 15 kbar pressure under H₂O-saturated conditions. The composition of the liquid pertinent to the piercing point forsterite + enstatite solid solution + amphibole + liquid + vapor is similar to that of calc-alkaline andesite. The electron microprobe analysis of the glass coexisting with the above three crystalline phases is very close to that of the piercing point determined by phase assemblage observations; however, the glass near (< 8 μm) forsterite crystals is significantly depleted in the normative forsterite component. With the addition of 10 wt.% KAlSi₃O₈, the composition of this piercing point becomes even closer to the compositions of calc-alkaline andesites. It is also shown that the liquid coexisting with forsterite and enstatite solid solution remains silica-rich (60–62 wt.%) over a wide (～ 100°C) temperature range. The present experimental studies support the view that liquids similar in composition to calc-alkaline andesites can be generated by direct partial melting of hydrous upper mantle at least at or near 15 kbar.     

Generation of komatiite magma and gravitational differentiation in the deep upper mantle

The densities of silicate liquids with basic, picritic, and ultrabasic compositions have been estimated from the melting curves of minerals at high pressures. Silicate liquids generated by partial melting of the upper mantle are denser than olivine and pyroxenes at pressures higher than 70 kbar, and garnet is the only phase which is denser than the liquid at pressures from 70 kbar to at least 170 kbar. In this pressure range, garnet and some fraction of liquid separate from ascending partially molten diapirs. It is therefore suggested that aluminium-depleted komatiite with a high Ca/OAl₂O₃ ratio may be derived from diapirs which originated in the deep upper mantle at pressures from 70 kbar to at least 140 kbar (200–400 km in depth), where selective separation of pyropic garnet occurs effectively. On the other hand, aluminium-undepleted komatiite is probably derived from diapirs originating at shallower depths (< 200 km). Enrichment of pyropic garnet is expected at depths greater than 200 km by selective separation of garnet from ascending diapirs. The 200-km discontinuity in the seismic wave velocity profile may be explained by a relatively high concentration of pyropic garnet at depths greater than 200 km.     

Experimental evidence for the role of accessory phases in magma genesis

Recent experimental studies have established petrogenetic models based on melting processes involving major phases. The possible residual character of trace-element-enriched accessory phases is not considered for temperatures well above the solidus in these models. In contrast, geochemists, applying trace element data to independently test the experimentally-based models, have concluded that residual (or fractionating) accessory phases may have an essential role in controlling the trace element (especially REE) distributions in magmas.Some recent experimental work provides data on the stability of potentially significant accessories such as sphene, rutile, apatite, zoisite and mica in basaltic compositions at elevated P and T. Sphene is stable to 1000°C with 60% melting of a hydrous tholeiite at 15 kbar. At higher pressure, rutile is the only Ti-rich accessory phase, and is present to at least 1000°C and high degrees of melting. Published REE data on sphene and rutile suggest that these phases may be important in controlling REE distribution in some magmas. For example, island are high-Mg, low-Ca-Ti tholeiites with low REE abundances and U-shaped patterns (Hickey and Frey, 1979) may reflect the role of sphene. In addition to rutile, similar close-packed Ti-rich accessory phases such as priderite, perovskite, crichtonite and loveringite may occur in mantle-derived magmas. These phases readily accommodate the REE but their possible role needs experimental confirmation.Apatite is recorded in hawaiite (1.16% P₂O_s) with 2% H₂O added at 5–6 kbar and 1050°C within 30°C of the liquidus, but at present no other experimental data are available on its high P, T stability, although thermodynamic calculations indicate that F may increase its stability markedly. Apatite is well known in high-pressure inclusions and as a phenocryst phase in rocks of the alkaline and calc-alkaline series.Ilmenite is known as a near-liquidus phase in some mafic magmas at 5–10 kbar, but its stability decreases to near-solidus at 25–30 kbar. Zoisite occurs in hydrous mafic compositions at mantle pressures, but it is confined to temperatures < 780°C. Finally, mica has a wide temperature range of stability at mantle pressures, especially in potassic magmas, and phlogopitic mica is stable to 1040°C at 20–25 kbar in a hydrous, K-rich “tholeiite” (1.6% K₂O).     

Oxygen diffusion rates in quartz exchanged with CO2

Oxygen self diffusion rates were determined in quartz samples exchanged with¹⁸O-enriched CO₂ between 745 and 900°C and various pressures, and the diffusion profiles were measured using an ion microprobe. The activation energy (Q) and preexponential factor (D₀) at P(CO₂) = P(tot) = 100 bar, for diffusion parallel to the c-axis are 159 ( ± 13) kJ/g atom and 2.10 (+0.75/ −0.55) × 10⁻⁸ cm²/s. This rate is approximately 100 times slower than that obtained from hydrothermal experiments and 100 times faster than a previous 1-bar quartz-O₂ exchange experiment. The oxygen diffusion rate measured at 0.6 bar, 888°C, and at 900°C in vacuum is in agreement with the previous 1-bar exchange experiments with¹⁸O₂. The effect of higher CO₂ pressures is small. At 900°C, the diffusion rate exchanged with CO₂ is = 2.35 × 10⁻¹⁵ cm²/s at 100 bar, 2.24 × 10⁻¹⁵ cm²/s at 3.45 kbar and 8.13 × 10⁻¹⁵ cm²/s at 7.2 kbar.There is probably a diffusing species, other than oxygen, that enhances the oxygen diffusion rate in these quartz-CO₂ systems, relative to that occurring at very low pressures or in a vacuum. The effect of this diffusing species, however, is not as strong as that associated with H₂O. Preserved oxygen isotope fractionations between coexisting minerals in a slowly cooled, high-grade metamorphic terrane will vary depending upon whether a water-rich phase was present or not. Closure temperatures will be approximately 100°C higher in rocks where no water-rich phase was present during cooling. The measured fractionations between coexisting minerals in metamorphic rocks may potentially be used as a sensor of water presence during retrogression.     

Lateral variations of H2O contents in quaternary magmas of Northeastern Japan

Volcanoes of the East Japan volcanic arc are divided into two groups on the basis of their phenocryst assemblages; volcanoes with lavas or pyroclastic rocks containing quartz phenocrysts and no hornblende phenocrysts (type A), and those with rocks containing hornblende phenocrysts and no quartz phenocrysts (type B). Type A volcanoes occur only in the narrow region along the volcanic front, whereas type B volcanoes are distributed in the area closer to the Sea of Japan.Recent experimental studies on calc-alkaline andesite-dacite under H₂O-saturated and -undersaturated conditions indicate that the liquidus temperature (maximum thermal stability limit) of quartz decreases drastically with increasing H₂O content in magma, whereas the liquidus temperatures of hornblende and biotite are relatively constant with variations in the H₂O content and bulk chemical composition of the magma.It is suggested from the lateral variation of mafic phenocryst assemblages [1] and from the above result that the temperature of the parental magmas of these volcanoes increases, and their H₂O contents decrease, towards the volcanic front in the East Japan volcanic arc.Such lateral variations in the H₂O contents of magmas under the East Japan volcanic arc are in agreement with those of other incompatible elements (K, Rb, REE, etc.). If H₂O-undersaturated partial melting of upper mantle peridotite can be represented by the univariant line (olivine, Ca-rich clinopyroxene, orthopyroxene and liquid coexist) in the system H₂OMg₂SiO₄z.sbnd;CaMgSi₂O₆z.sbnd;SiO₂, the decrease of H₂O content in the magma suggests that the melting temperature of the peridotitic mantle may gradually increase, and so the degree of partial melting may increase, towards the volcanic front. The lateral variation of other incompatible elements can also be explained by this model.     

Water and magma generation at subduction zones

The basaltic ocean crust, metasomatized and metamorphosed during and after generation at the ocean ridge, contains H₂O stored in minerals and pore fluid. Phase equilibrium data establish the conditions for dehydration, and the conditions for melting of amphibole-gabbro or amphibole-quartz-eclogite, or for quartz-eclogite or mantle peridotite if aqueous fluids are available. But there is no concensus about the temperature distribution through the subducted crust, or within the overlying mantle wedge. Therefore, a variety of magmatic models can be derived from the experimental data. According to some calculations, endothermic dehydration reactions in the depth interval 75–125 km cool the oceanic crust to such an extent that it cannot be a major source of magmas; instead, concentrated aqueous fluids released from the crust generate magmas in the overlying peridotite. However, according to most existing thermal models, if temperatures in ocean crust are cool enough to prohibit melting of amphibolite, then temperatures in the mantle above the main sources of expelled fluids are too low for hydrous melting. The ocean crust appears to be effectively dehydrated by 100–125 km depth. Dense hydrous magnesian silicates are not likely candidates for deeper H₂O transport. The extent to which H₂O can be fixed in metasomatic phlogopite in crust or mantle is a significant but undetermined factor. Experimental data on minerals and liquid compositions do not support the concept of primary magmas for andesites and associated lavas from mantle or subducted crust. Complex, multi-stage processes appear to be more likely, which is consistent with recent interpretations of geochemical data.     

Formation time of the big mantle wedge beneath eastern China and a new lithospheric thinning mechanism of the North China craton—Geodynamic effects of deep recycled carbon

High-resolution P wave tomography shows that the subducting Pacific slab is stagnant in the mantle transition zone and forms a big mantle wedge beneath eastern China. The Mg isotopic investigation of large numbers of mantle-derived volcanic rocks from eastern China has revealed that carbonates carried by the subducted slab have been recycled into the upper mantle and formed carbonated peridotite overlying the mantle transition zone, which becomes the sources of various basalts. These basalts display light Mg isotopic compositions(δ26 Mg = –0.60‰ to –0.30‰) and relatively low87 Sr/86 Sr ratios(0.70314–0.70564) with ages ranging from 106 Ma to Quaternary, suggesting that their mantle source had been hybridized by recycled magnesite with minor dolomite and their initial melting occurred at 300-360 km in depth. Therefore, the carbonate metasomatism of their mantle source should have occurred at the depth larger than 360 km, which means that the subducted slab should be stagnant in the mantle transition zone forming the big mantle wedge before 106 Ma. This timing supports the rollback model of subducting slab to form the big mantle wedge. Based on high P-T experiment results, when carbonated silicate melts produced by partial melting of carbonated peridotite was raising and reached the bottom(180–120 km in depth) of cratonic lithosphere in North China, the carbonated silicate melts should have 25–18 wt% CO2 contents, with lower Si O2 and Al2 O3 contents, and higher Ca O/Al2 O3 values, similar to those of nephelinites and basanites, and have higher εNdvalues(2 to 6). The carbonatited silicate melts migrated upward and metasomatized the overlying lithospheric mantle, resulting in carbonated peridotite in the bottom of continental lithosphere beneath eastern China. As the craton lithospheric geotherm intersects the solidus of carbonated peridotite at 130 km in depth, the carbonated peridotite in the bottom of cratonic lithosphere should be partially melted, thus its physical characters are similar to the asthenosphere and it could be easily replaced by convective mantle. The newly formed carbonated silicate melts will migrate upward and metasomatize the overlying lithospheric mantle. Similarly, such metasomatism and partial melting processes repeat, and as a result the cratonic lithosphere in North China would be thinning and the carbonated silicate partial melts will be transformed to high-Si O2 alkali basalts with lower εNdvalues(to-2). As the lithospheric thinning goes on,initial melting depth of carbonated peridotite must decrease from 130 km to close 70 km, because the craton geotherm changed to approach oceanic lithosphere geotherm along with lithospheric thinning of the North China craton. Consequently, the interaction between carbonated silicate melt and cratonic lithosphere is a possible mechanism for lithosphere thinning of the North China craton during the late Cretaceous and Cenozoic. Based on the age statistics of low δ26 Mg basalts in eastern China, the lithospheric thinning processes caused by carbonated metasomatism and partial melting in eastern China are limited in a timespan from 106 to25 Ma, but increased quickly after 25 Ma. Therefore, there are two peak times for the lithospheric thinning of the North China craton: the first peak in 135-115 Ma simultaneously with the cratonic destruction, and the second peak caused by interaction between carbonated silicate melt and lithosphere mainly after 25 Ma. The later decreased the lithospheric thickness to about70 km in the eastern part of North China craton.     

The solubility mechanisms of volatiles in silicate melts and their relations to crystal-andesite liquid equilibria

The anionic structure of magmatic liquids has been estimated at 1 atm and at pressures corresponding to those of the upper mantle. These estimates are based predominantly on spectroscopic data on binary metal oxide-silica and ternary metal oxide-silica-alumina melts. Structural information on melt compositions in aluminate-silica joins has been used to provide detailed information on the role of Al³⁺ in natural magma at atmospheric and high pressure.Regardless of pressure, andesitic melts may be described as combinations of chain, sheet, and three-dimensional network units. Nearly all Al³⁺ in the magmatic liquid resides in the three-dimensional network units. This Al³⁺ is locally charge-balanced with Na⁺, K⁺, Ca²⁺, and Mg²⁺. In the latter two cases, Al³⁺ and Si⁴⁺ are ordered, whereas for Na⁺ and K⁺, Si⁴⁺ and Al³⁺ are randomly mixed. Solution of water in natural magma results in the formation of new nonbridging oxygens in addition to OH groups attached to Si⁴⁺ and metal cations.On the basis of determined solution mechanisms of CO₂ and H₂O in silicate melts, thermodynamic properties of HO+CO₂, fluids and hydrous silicate melts and melting phase relations in peridotite-H₂O-CO₂, systems, it is found that natural andesitic magma in equilibrium with spinel Iherzolite in the upper mantle (10–20 kbar) must contain at least 5–7 wt.% H₂O. Andesitic magma with 5–7 wt.% H₂O in solution may be described as a mixture of Al-free three-dimensional units, sheets, and chains with a small proportion (less than 10%) of monomers.     

Dehydration of clastic sediments in subduction zones: Theoretical study using thermodynamic data of minerals

Pseudosections for two sediments and one basalt calculated in the system K₂O–Na₂O–CaO–MgO–FeO–Fe₂O₃–Al₂O₃–TiO₂–SiO₂–H₂O for the P–T range 10 to 35 kbar, 300 to 900°C give useful insights into the amount of H₂O released from oceanic crust in subduction zones. In cold subduction zones (20 kbar–300°C to 35 kbar–500°C) hydrous minerals storing 3 to 4 wt% H₂O are still present in metasediments at depths of 120 km. In the same environment, metabasite releases 1 wt% H₂O in the depth range 100 to 120 km, but 4.5 wt% H₂O is transported to greater depths. In hot subduction zones (300°C hotter than the cold subduction zone at 100 km depth), dehydration events of metasediments in the depth range 50 to 80 km correspond to the breakdown of chlorite and paragonite. In the calculations no further water is released at greater depths because the modal content of phengite, the only hydrous mineral phase at these depths, remains almost constant. For the same P–T path, metabasite shows continuous dehydration between 40 and 80 km releasing almost 3 wt% H₂O. At 120 km depth less than 0.4 wt% of H₂O remains. In an average modern subduction zone (～6°C/km) most dehydration of sediments occurs at depths of 70 to 100 km and that of basalts at depths of 80 to 120 km. Only 1.3 wt% H₂O in metasediments and 1.6 wt% H₂O in metabasalt has the potential to be subducted to depths greater than 120 km. The dehydration behavior of sediments concurs with the generally held idea that subduction zone fluids are most effectively transported to great depths by cold subduction. In hot subduction zones, such as those characteristic of early Earth, most H₂O carried by oceanic crust is liberated at depths less than 120 km and, thus, would not contribute to island‐arc magmatism.     

Stability of forsterite + CO2 and its bearing on the role of CO2 in the mantle

An experimental determination of the reaction MgCO₃ + MgSiO₃ = Mg₂SiO₄ + CO₂ between 20 and 40 kbars and in the range 1000–1500°C yields an average pressure effect on the equilibrium of 44 bars/°C. This result shows that the assemblage forsterite and carbon dioxide is not stable under most pressure and temperature conditions expected in the upper mantle. Hypotheses requiring the presence of free CO₂ in the low-velocity zone, CO₂ as a drive mechanism for kimberlite emplacement, or action of a free CO₂ phase in ultramafic rocks may need considerable revision.