Hematite-magnetite Buffer Research Articles

Dome-related gold mineralization in the Pani volcanic complex, North Sulawesi, Indonesia; a study of geologic relations, fluid inclusions, and chlorite compositions

Gold mineralization at the Pani prospect in north Sulawesi, Indonesia, is related to a Miocene or younger rhyodacitic volcanic center. The center, which is 3.5 km in diameter, consists of porphyritic lavas, intrusions, breccias, and pyroclastics. It overlies and partly intrudes horn-blende and hornblende-biotite granodiorite and Eocene (?) basaltic volcanics.Mineralization is spatially related to the Baganite dome, which is centered on Gunung Baganite. Low-grade (0.8 ppm) disseminated gold mineralization, associated with pyrite and minor base metals, is hosted by the dome. Silver mineralization (acanthite) occurs in minor quartz-hematite veins in overlying silicified pyroclastics. The most important gold mineralization occurs on the flank of the dome at the Pani ridge, less than 1 km from Gunung Baganite. It is associated with minor base metal sulfides, quartz, and adularia encrustations, found in vuggy fractures and centimeter-wide breccia zones, in weakly silicified rhyodacites, and in adjacent wall rocks. Quartz veins on the Pani ridge are virtually absent. The assemblage chloritealbite-anatase-pyrite + or - quartz-adularia is ubiquitous in flow-banded and massive rhyodacites in the prospect area, whereas carbonates occur within the enclosing pyroclastics. Strong silicification occurs in the upper part of the Baganite dome and extends up to 130 m into the overlying volcanics. Adularia veining increases with depth as does secondary K feldspar in the groundmass. The Mg number of chlorite decreases from 60 in the pyroclastics above the Baganite dome to 25 in the rhyodacite at a 300-m depth, and there is significant intrasample variation.Temperatures of chlorite formation in the Baganite dome, calculated from electron microprobe analyses, vary from about 125 degrees to 275 degrees C and are in good agreement with the homogenization temperatures of 150 degrees to 300 degrees C from secondary fluid inclusions in quartz phenocrysts. The salinities of the majority of the fluid inclusions range from near zero up to about 4 equiv wt percent NaCl, but some salinities are as high as 40 equiv wt percent. Vapor-rich inclusions also occur. The redox states of the fluids associated with the Baganite dome, calculated from chlorite compositions, varied significantly, from about -2 to -8 log f (sub O 2 ) units below the hematite-magnetite buffer. Hydrogen sulfide levels in the fluids decreased from a maximum of 30 ppm around 250 degrees C, with decreasing temperature and oxidation state, and pH values were close to neutrality.Gold was apparently transported at 1-ppb levels at the highest temperature in the most oxidized fluids, of around 4 equiv wt percent NaCl salinity, which probably rose through the root zones and the lower contacts of the Baganite dome. Gold precipitated as the fluids cooled by boiling (indicated by vapor-rich inclusions, adularia, and Mg-rich chlorites) and also by mixing with low-salinity, reduced fluids inferred to have circulated around the flanks of the dome. The latter mechanism was probably the most efficient and suggests that there may be potential on the margins of volcanic domes for large-tonnage, low-grade gold deposits.

Fluids in Hercynian Au veins from the French Variscan belt

AbstractFluids, together with alteration and ore mineral assemblages, were studied in representative hydrothermal gold-bearing quartz veins from the western part of the Variscan belt in France (La Bellière, Montagne Noire district, Villeranges-Le Châtelet district, and Limousin province). Petrographic studies of the relationships between ores, fluid inclusions, microfracturing and quartz textures show that chronological and genetic relationships between gold deposition and fluid trapping may be very complex and difficult to establish for veins which show multi-stage fracturing and shearing. Systematic studies of secondary fluid inclusions in microcracks and recrystallized zones of the early quartz veins indicate two contrasting physical-chemical conditions: 1 relatively high temperature (250–400°C) and pressure (>1 kbar) event with CO2-CH4-H2S-N2 (±H2O-NaCl)-rich fluids related to the early sulphide deposition; 2 lower temperature (150–250°C) and pressure with aqueous fluids related to the late native-gold-sulphide (or sulphosalt) assemblage, which constitutes the economic ores in some deposits.In deposits where gold occurs predominantly in a combined state within arsenopyrite and pyrite (Châtelet and Villeranges), primary fluid inclusions in authigenic quartz combs cogenetic with arsenopyrite are almost purely aqueous (H2O-NaCl) and have a low salinity (1–4 wt. % NaCl). P-T conditions (150–250°C), nearly hydrostatic pressures) are similar to those of the second stage in the multi-stage quartz veins.Consideration of chemical equilibria in the C-O-H-N-S system using microthermometric and Raman spectrometric analysis for the fluids, together with data obtained from mineralogical studies, show that during gold deposition, fO2 was below hematite-magnetite buffer at Villeranges and around the Ni-NiO buffer at La Bellière and Montagne Noire. fS2 calculations based on H2S analyses are in good agreement with mineral assemblage estimates and close to that fixed by the pyrite-pyrrhotite boundary at high temperature. Ore fluid pH was significantly lower than in the host rocks as shown by the complete alteration of the host rocks into a quartz-K-mica assemblage. The data illustrate that during the late Hercynian, fluid circulation evolved from high P-T conditions, in some cases linked to late magma intrusions, towards conditions typical of later hydrothermal systems of the geothermal type.

Comparison of Granite, Tuff, and Basalt as Geologic Media for Long-Term Storage of High-Level Nuclear Waste

ABSTRACTSystematic differences in pH, cation/proton ion activity ratios, and redox have been observed between solutions produced in rock-water hydrothermal experiments with tuff, granite, and basalt. Stable pH values in tuff-water experiments may be as much as 1.5 pH units more acidic than basalt-water experiments at the same temperature and ionic strength. Redox (log fO2) values in 300°C tuff experiments are 4–7 orders of magnitude more oxidizing than basalt experiments and ca. 4 log units more oxidizing than the magnetite-hematite buffer. Such fluid differences could significantly affect the performance of a high-level nuclear waste repository and should be considered in repository design and siting.

Cordierite-olivine symplectites in Fe-Al-rich hornfels from the Nogo-Hakusan area, Gifu Prefecture, central Japan.

Fe-Al-rich hornfels containing cordierite, olivine, orthopyroxene and spinel is found in the contact aureole of the Nogo-Hakusan granodiorite. Spinel is hercynitic and rich in V2O3 (up to 2.2 wt.%). Symplectites composed of olivine, cordierite and small amount of spinel are present around orthopyroxene. Phase relationships of cordierite, olivine, orthopyroxene and spinel in the system Mg-Fe2+-Al-Si-O-H have been experimentally determined. The experimental results and petrographical features suggest that the symplectites were formed by the following retrogressive reaction: 10 orthopyroxene+2 spinel=5 olivine+cordierite. The maximum temperature of the metamorphism in the Nogo-Hakusan area was about 800°C at Ptotal=Pfluid=2-3 kb. The estimated fo2 during the retrogressive metamorphism was relatively low, at least below the hematite-magnetite buffer curve.

Experimental determination of the ZnFe 2 O 4 -ZnAl 2 O 4 miscibility gap with application to franklinite-gahnite exsolution intergrowths from the Sterling Hill zinc deposit, New Jersey

The Sterling Hill zinc deposit is a regionally metamorphosed stratiform oxide-silicate deposit enclosed in the Precambrian Franklin marble. Spinels, which consist of oriented exsolution intergrowths of gahnite (G) (ideally ZnAl 2 O 4 ) and franklinite (F) (ideally ZnFe 2 O 4 ), were collected at four locations in the Sterling Hill mine. The bulk chemical composition of these spinels and the composition of the phases which constitute the intergrowths were determined by broad beam and point analysis, respectively, with the electron microprobe. These data show that the original homogeneous high-temperature spinels were as aluminous as F 80 G 20 .The miscibility gap in the system ZnFe 2 O 4 -ZnAl 2 O 4 was determined experimentally by hydrothermal methods. Homogeneous synthetic spinels on the join ZnFe 2 O 4 -ZnAl 2 O 4 were prepared by dry reaction of the oxides at 1,380 degrees + or - 50 degrees C. Equimolar mechanical mixtures of two synthetic homogeneous spinels with the compositions F 90 G 10 and F 10 G 90 were sealed with water in separate platinum capsules. These, in turn, were sealed in gold capsules and subjected to P (sub H 2 O) = 1 kbar between 700 degrees and 900 degrees C for a maximum period of 72 and 11 days, respectively. The oxygen fugacity was controlled by the hematite-magnetite buffer (hematite packed between the gold and platinum capsules). Equilibrium was approached from two directions, namely, by exsolution of synthetic homogeneous spinels and by reaction of the spinels in the mechanical mixtures. An excellent fit with the experimental data was obtained by solution modeling (asymmetric Margules model) which gives a miscibility gap with a calculated consolute temperature of 957.4 degrees C at X gahnite = 0.525.Integration of the electron microprobe analyses of the natural intergrowths with the experimentally determined miscibility gap indicates that the minimum peak temperature attained during regional metamorphism at the Sterling Hill mine was 760 degrees C.

The role of migrating oil field brines in the formation of sediment-hosted Cu-rich deposits

Oil field brines commonly contain ppm quantities of Zn, Pb, and H 2 S but negligible quantities of Cu. Theoretical analysis of the water-rock interactions expected during migration through evaporitic and red-bed sequences suggests that these reducing oil field brines can become transformed into far more oxidized fluids capable of transporting Cu, Pb, Zn, and sulfate and of forming sediment-hosted Cu-rich deposits. Calculation of the chemical mass transfer accompanying reaction of a typical reducing brine with anhydrite, hematite, and quartz, followed by the assemblage quartz, K-feldspar, muscovite, hematite, chlorite, and cuprite, reveals that the aqueous phase becomes greatly undersaturated with respect to sphalerite, galena, chalcopyrite, bornite, pyrite, and pyrrhotite as it achieves log f (sub O 2 ) values well above the magnetite-hematite buffer consistent with saturation with respect to anhydrite, hematite, and chalcocite and ore-forming concentrations of copper and sulfate. Consequently, evaporites are not only possible sources of salinity and (oxidized) sulfur for evolving basinal brines but may be crucial to the development of the f (sub O 2 ) conditions required to transport Cu in ore-forming concentrations. Reaction of the oxidized brine with organic matter (approximated by graphite) and diagenetic pyrite results in precipitation of the zonation sequence chalcocite, bornite, chalcopyrite, and pyrite with variable amounts of hematite, covellite, sphalerite, and galena, depending on the relative rates of reaction of the graphite and pyrite.

Strata-bound cupriferous sulfide mineralization associated with continental rhyolitic volcanic rocks, northern Chile; I, The Jardin copper-silver deposit

The Jardin arsenian Cu-Ag deposit of the Copiapo mining district, northern Chile, is in many respects representative of the volcanic-associated, disseminated, cupriferous manto mineralization characteristic of this segment of the central Andean orogen. Containing ca. 1 to 2 million metric tons of ore with a grade of 1.75 percent Cu and 140 g/metric tons Ag, the crudely stratiform deposit is hosted by the upper, brecciated and unwelded part of a approximately 75-m-thick, continental rhyolitic ignimbrite and by an overlying approximately 2-m succession of tuffaceous lacustrine sedimentary rocks. The ignimbrite represents the local basal unit of the Paleocene Hornitos Formation. The reduced, ore-hosting, sedimentary horizons were deposited in an ephemeral saline pond and comprise coarse and fine breccias, tuffaceous sandstones and silt-stones, and calcareous-carbonaceous shales containing thin coaly seams and accumulations of plant debris.The greater part of the mineralization occurs as disseminations, discontinuous veinlets, thin concordant lenses and, most strikingly, cylindrical bodies, 1 to 8 mm in diameter, oriented perpendicular to bedding. Chalcocite, bornite, and tennantite (ca. 1-4.3% Ag) are the major ore minerals and are associated with minor sphalerite, digenite, covellite, chalcopyrite, wit-tichenite, native silver, acanthite, stromeyerite, mckinstryite, jalpaite, and the ruby silvers. Pyrite and marcasite are subordinate constituents of the ore and widely display evidence of replacement by the Cu and Ag minerals. Mineralization in both the ash flow and sediments is associated with moderate argillic and carbonate alteration. The hanging wall of the orebody lies within a red fanglomerate, encroachment of which terminated the accumulation of the lacustrine strata.The sulfidic pipes commonly display concentric mineral zonation suggestive of the successive, in part open-space, precipitation of pyrite and the Cu-Ag minerals. Textural evidence suggests that at least the introduction of pyrite into these bodies occurred prior to the deposition of the immediately overlying strata. The pipes, both individually and in aggregate, do not resemble rhizoliths or dewatering structures. More probable origins are as Scolithus sp. lebensspuren or as gregarious hydrothermal conduits directly reflecting the upward penetration of ore-forming fluids. Sulfide-free pipes are extremely rare, and we favor the latter interpretation.The Jardin mineralization displays features characteristic of several ore deposit types, particularly of the red-bed and epithermal clans. Although the temporal relationships of Cu-Ag mineral deposition and lacustrine sedimentation are uncertain and direct evidence of the conditions of mineralization unavailable, our preferred genetic model involves the convective flow of metal-bearing ground waters through the permeable upper zone of the ignimbrite into the mantling lacustrine strata, during both the fumarolic stage of the tuff and the deposition of the sediments. If this be the case, the emplacement of the disseminated mineralization may have taken place in less than a century; the temporal constraints would be reduced if the Cu-Ag minerals formed largely through replacement of early diagenetic pyrite. The hydrothermal fluids are inferred to have been Cl rich, low temperature, neutral to weakly alkaline, and with an f (sub O 2 ) exceeding the hematite-magnetite buffer; they probably derived their metals and arsenic from nearby, weathered, Upper Cretaceous base and precious metal deposits exposed by the sub-Hornitos Formation erosion surface. Ore deposition occurred in response to reduction, neutralization and moderate cooling on contact with the roof zone of the ignimbrite and the carbon- and sulfur-rich sediments; there is no evidence for sulfide replacement of organic detritus.The Jardin manto may perhaps best be interpreted as an unusual, Ag-rich, red-bed copper deposit nucleated in part by an ash flow cooling on the floor of an intermontane basin. However, the Paleocene epoch in the Copiapo district saw the development of numerous Ag-Cu-As epithermal vein systems, and the deposit could also represent an analogous center in which ore deposition was controlled by a sequence of permeable reduced sediments rather than by faults.

A model for the evolution of hematite carbonatite, based on whole-rock major and trace element data from the Fen complex, southeast Norway

REE-rich hematite carbonatites are found in numerous carbonatite complexes worldwide, commonly associated with ankerite-bearing carbonatites. The mineralogy of the hematite carbonatites contradicts a primary magmatic origin. Hematite-calcite-dolomite carbonatite in the Fen complex (southeast Norway) was formed by postmagmatic oxidation and partial dissolution of ankeritic ferrocarbonatite. In this process, major components such as CaO, MgO and CO 2, and trace elements such as the light rare earths, were removed from the rock, while more insoluble components (Fe 2O 3, Th, Y and the middle rare earths) were retained in the solid residue. It is possible to model the alteration of the rocks in terms of a two-stage recrystallization and leaching process, corresponding to two distinct stages of the reequilibration of the carbonatite mineralogy. The first step took place at oxygen fugacities below the hematite-magnetite buffer (ferrous iron in part leached from the system), the second at more oxidizing conditions (Fe immobile). The most strongly altered carbonatite has lost more than 70% of its original volume during reequilibration. Ferrous iron and REE leached from the ferrocarbonatite during early reequilibration precipitated as hydrothermal magnetite-pyrite ores with accessory REE minerals. Reequilibration, leaching and redeposition processes like those recognized at Fen may be important factors controlling the mineralogy and geochemistry of REE-enriched carbonatites in general.

Metamorphic evolution of aluminous granulites from Labwor Hills, Uganda

Sapphirine-cordierite-quartz and spinel-cordierite-quartz form relic assemblages of probable Archaean age in Fe-rich aluminous metapelites from Labwor Hills, Uganda, and reflect an unusually high temperature metamorphism (∼1,000° C) at pressures in the vicinity of 7–9 kbars and a(O2) near the magnetite-hematite buffer. Subsequent reaction textures include the replacement of spinel and cordierite by sillimanite and hypersthene and formation of sapphirine-hypersthene-K-feldspar-quartz symplectites which are interpreted as pseudomorphs after osumilite. A petrogenetic grid appropriate to these assemblages suggests these reaction textures may be due to cooling at constant or increasing pressure and constant a(O2), or decreasing a(O2) at constant temperature and pressure. The former interpretation is supported by the coexistence of ilmenohematite and magnetite during the development of the reaction textures, and by the comparatively low Al2O3-contents of secondary hypersthene. This pressure-temperature path implies that: (1) metamorphism occurred at deep levels within normal thickness crust, probably less than 40–45 km thick, due to an extreme thermal perturbation induced either by emplacement of mantle-derived magmas or by thinning of the subcontinental lithosphere in an extensional tectonic regime, (2) the excavation and surface exposure of the granulites is due to a subsequent, postgranulite facies metamorphism, crustal thickening most probably involving their incorporation into an allochthonous upper crustal thrust sheet during the formation of the Mozambique foldbelt.

Physics and chemistry of the hydrothermal system at the Panguna porphyry copper deposit, Bougainville, Papua New Guinea

Various hydrothermal processes have been suggested as important in the formation of porphyry coppers, e.g., orthomagmatic evolution of salt-rich liquid, condensation of salt-rich liquid from magmatic vapor, convection of ground water driven by magmatic heat, and boiling of ground water. A fluid inclusion study based on detailed two-dimensional sampling indicates that all of these processes may have contributed to the evolution of the Panguna deposit, but that the copper was deposited mainly by salt-rich liquid expelled directly from the magma. Significant salt-rich liquid may also have condensed from a vapor plume.A scheme for cooling the boiling liquids (initially unsaturated in NaCl at magmatic conditions) through supersaturation to unsaturation near 400 degrees C (by mixing with an externally derived liquid) is suggested to explain the systematic variation in homogenization properties of the salt-rich liquids. For a pressure near 300 bars determined at the outer boundary of the ore zone, the sequence of predicted phase changes based on the NaCl-H 2 O system corresponds with observed changes. Variations in the KCl and NaCl content of salt-rich liquid could be explained by the precipitation of halite, but that is incompatible with boiling. Salt-rich liquid was pumped into the ore zone by the hydrostatic-lithostatic pressure difference, then descended through the ore zone because of its density relative to ground water. Phenomena in the top of the system were dominated by the interaction of vapor and ground water.An oxygen fugacity near that of the hematite-magnetite buffer, and the predominance of SO 2 among vapor sulfur species, have been deduced from mineralogy and limits on partial pressures. The formation of mixed metal halides enhances the volatility of Fe, but vapor transport of Fe is insignificant. Zn and Mo undergo significant vapor transport. Cu may be transported in liquid in some cases and vapor in others. The alteration of wall rock to biotite generates HCl, which is probably removed by boiling. The sulfate in anhydrite may be a decomposition product of SO 2 , but a second mechanism is implied below 500 degrees C, where sulfide deposition becomes greater relative to sulfate. The high oxidation state of magmatic fluids during copper mineralization was due to the loss of H 2 from the magma in those early evolved volatiles that formed an amphibole-bearing assemblage.

An investigation into the environment of formation of the volcanic-hosted Mt. Lyell copper deposits using geology, mineralogy, stable isotopes, and a six-component chlorite solid solution model

The Mt. Lyell copper deposits occur within 10 km 3 of hydrothermally altered, calc-alkaline felsic pyroclastics and lavas that form the overturned, steeply west-dipping limb of a north-south-trending anticline within the Mt. Read Volcanics.Exhalative pyritic massive sulfide deposits (e.g., The Blow) and siliceous barite-bornite-chalcopyrite mineralization (e.g., North Lyell) occur at or near the top of the volcanic sequence. The bulk of the mineralization occurs as disseminated pyrite-chalcopyrite lenses within the volcanics (e.g., Prince Lyell). The disseminated deposits are lenticular, broadly concordant with the stratigraphy and subparallel to the cleavage developed during the Tabberabberan orogeny (Middle Devonian). They are associated with quartz-sericite and quartz-sericite-chlorite rocks, the latter commonly containing minor siderite-hematite-magnetite, barite, sphalerite, apatite, and monazite and more rarely inclusions of bornite-chalcopyrite and chalco-pyrite-pyrrhotite in the pyrite.The host volcanics may have developed in a subaerial environment, but the exhalative sulfide deposits require water depths approaching 1 km. North-south- and east-west-trending faults, which were active during the Tabberabberan orogeny and probably controlled sedimentation patterns in the Early Ordovician and the Late Cambrian, may have controlled a shift to a partial or total(?) submarine environment at the time of mineralization and acted as conduits for solution movement. The disseminated deposits are thought to have formed from solutions percolating laterally along permeable horizons within the volcanic sequence underlying the exhalative massive sulfide deposits. The deposits at the top of the sequence were probably partly eroded during the Late Cambrian and/or Early Ordovician, and the entire sequence was metamorphosed to lower greenschist grade during the Devonian.Determination of the physicochemical conditions of formation of the disseminated pyrite-chalcopyrite mineralization has been attempted using the composition of chlorite coexisting with quartz, pyrite, and chalcopyrite, by means of a six-thermodynamic-component solid solution model for chlorite, and assuming ideal mixing for cations on energetically equivalent sites. The model is calibrated on the composition of the chlorite from the OH vein, Creede, Colorado, and estimates of the oxidation state and temperature during vein formation. Log Sigma SO 4 /Sigma H 2 S versus pH and log Sigma SO 4 /Sigma H 2 S versus T diagrams have been constructed to illustrate the variation in chlorite stoichiometry as a function of temperature and solution conditions.The temperatures of 260 degrees to 290 degrees C calculated from chlorite compositions at Prince Lyell, assuming quartz-iron oxide-chlorite equilibrium and the f (sub O 2 ) conditions above the hematite-magnetite buffer, probably reflect metamorphic reequilibration. Assuming quartz-chlorite-pyrite equilibrium and that each chlorite is a closed system for four of the six components, it is possible to estimate ore deposition temperatures of 270 degrees to 310 degrees C, initial log f (sub S 2 ) values of about -10 to -11.5, and initial Sigma S values between 0.0005 and 0.003--the latter estimate involving additional assumptions about the concentration of the major components in the ore fluid and the pH which was taken as 1 unit below neutrality. At 300 degrees C and a Sigma S value of about 0.001, a pH of between 4 and 4.25 and a log Sigma SO 4 /Sigma H 2 S value between 0 and -2 are required to maintain about 10 ppm Fe and 1 ppm Cu in a 1-molal chloride solution in equilibrium with pyrite and chalcopyrite.Bornite-bearing ores in the North Lyell area may have formed from acid, oxidized ground waters that derived their copper from leaching of preexisting chalcopyrite.The variation in delta 34 S py values from Prince Lyell (10 to 5.2ppm) to Cape Horn (+6.4 to -0.4ppm) to Crown Lyell-North Lyell (+0.8 to -5.3ppm) is consistent with a progressive increase in the oxidation conditions at which these deposits formed. The delta 34 S (sub py-ccp) values vary from +1.8 to -1.5 per mil and can be reproduced by mass transfer modeling of possible reaction paths in a feldspar-ore solution system at constant temperature and varying initial oxidation conditions. The calculations suggest an initial value of delta 34 S (sub Sigma S) of 7 per mil for the ore fluid(s).

Stability of the assemblage orthopyroxene-sillimanite-quartz in the system MgO-FeO-Fe2O3-Al2O3-SiO2-H2O

The stability of coexisting orthopyroxene, sillimanite and quartz and the composition of orthopyroxene in this assemblage has been determined in the system MgO-FeO-Fe2O3-Al2O3-SiO2-H2O as a function of pressure, mainly at 1,000° C, and at oxygen fugacities defined mostly by the hematite-magnetite buffer. The upper stability of the assemblage is terminated at 17 kbars, 1,000° C, by the reaction opx+Al-silicate →gar+qz, proceeding toward lower pressures with increasing Fe/(Fe+Mg) ratio in the system. The lower stability is controlled by the reaction opx+sill+qz→ cord, which occurs at 11 kbars in the iron-free system but is lowered to 9 kbars with increasing Fe/(Fe+Mg). Spinel solid solutions are stabilized, besides quartz, up to 14 kbars in favour of garnet in the iron-rich part of the system (Fe/(Fe+Mg)≧0.30). Ferric-ferrous ratios in orthopyroxene are increasing with increasing ferro-magnesian ratio. At least part of the generally observed increase in Al content with Fe2+ in orthopyroxene is not due to an increased solubility of the MgAlAlSiO6 component but rather of a MgFe3+AlSiO6 component. The data permit an estimate of oxygen fugacity from the composition of orthopyroxene in coexistence with sillimanite and quartz.

Partitioning of iron and magnesium between crystals and partial melts in peridotite upper mantle

The iron-magnesium distribution coefficient, $$K'_D = (X_{\Sigma {\text{FeO}}} /X_{{\text{MgO}}} )^{{\text{olivine}}} (X_{{\text{MgO}}} /X_{\Sigma {\text{FeO}}} )^{{\text{liquid}}} ,$$ has frequently been used as a means of testing whether experimental and natural silicate liquids could have been in equilibrium with olivine of mantle composition. It is shown here that this K′ D decreases with increasing oxygen fugacity (xxx) for a hydrous partial melt in equilibrium with a natural spinel peridotite assemblage under pressure and temperature conditions corresponding to those of the upper mantle (from 0.52 at the xxx of the iron-wustite buffer to 0.04 at the xxx of the magnetite-hematite buffer). K′ D also increases with increasing pressure, with decreasing temperature, and probably with increasing Mg/(Mg+∑ Fe) of the parental peridotite, suggesting that $$K_D = (X_{{\text{FeO}}} /X_{{\text{MgO}}} )^{{\text{olivine}}} (X_{{\text{MgO}}} /X_{{\text{FeO}}} )^{{\text{liquid}}}$$ also increases with increasing pressure and decreasing temperature. Thus, unless these four variables (P, T, xxx, silicate composition) are known for a natural magma, K′ D and probably K D are variables, and the Mg/(Mg+∑ Fe) of such a magma cannot be correlated to that of the parent. The K D determined at 1 atm pressure by Roeder and Emslie has frequently been used to test whether the Mg/(Mg+∑ Fe) ratios of experimentally formed liquids at high pressure in equilibrium with olivine of known Fo content represent the equilibrium Mg/(Mg+Fe2+) of this liquid, assuming that ∑Fe=Fe2+ and that K′ D does not vary with P, T, and composition of the system. Published data demonstrate that the oxygen fugacities of the experimental designs employed by different laboratories vary between those of the magnetite-hematite and magnetite-wustite buffers (6 orders of magnitude), resulting in K′ D between 0.04 and 0.31 at 1050° C and 15 kbar, for example. Thus, published arguments as to whether the quenched liquids represent equilibrium compositions based on iron-magnesium partitioning are inadequate. The effects of P, T, xxx, and the composition of the starting material must also be considered.