V33F-01 INVITED
Sulfur and Copper in Magmatic Arcs: Sources and Linkages
Economic metal mineralization, principally porphyry and related hydrothermal deposits, occurs in the shallow crust of volcanic arcs, in association with felsic rocks. However, the processes that lead to the formation of giant deposits take place at much deeper levels in the subarc mantle. Sulfur is a critical component because it is the most anomalous element in deposits in terms of enrichment factor. Sulfur isotopes and other information from giant deposits indicate that S was introduced to the host felsic rocks by a primitive mafic melt. The mafic melt was oxidized, allowing the transport of a large quantity of S from the mantle to the shallow crust. An oxidized melt also allows the transfer of chalcophile metals from the mantle, thus avoiding the metals being scavenged by dense sulfide liquid. Formation of an oxidized primitive melt requires an oxidized subarc mantle because of the minimal change in fO2 during partial melting. A substantial range in the fO2 of subarc mantle is most likely caused by the nature of subducted sediments, because the compositions of oceanic lithosphere have been comparable over 2 Ga. The subarc mantle is depleted in chalcophile elements because they are mildly to highly incompatible during partial melting, as shown by inverse correlations between Cu and compatible elements in residual peridotites. However, Cu concentrations in primitive arc magmas have a wide range, from very low to high values, > 300 ppm, in contrast to the MORB values of about 70 ppm. Furthermore, there are broad spatial variations, with high concentrations in some arcs such as Izu. The wide variation in both Cu contents and the ratios of Cu to immobile elements in primitive arc magmas suggest that Cu is introduced to the subarc mantle by aqueous fluids from slabs. In a given subduction zone, the Cu concentration in primitive arc magmas does not correlate with Ni, Mg, alkalis, alkali-earths, and other fluid-mobile or chalcophile elements, such as Pb. Rather, Cu concentration correlates positively with V, which is a fluid-immobile lithophile element. The evidence suggests that Cu and possibly other chalcophile elements may not be hosted by sulfides in the magma source region in subarc mantle. Vanadium valence is sensitive to fO2 and oxidized species are not compatible with mantle minerals. Therefore, the positive correlation between Cu and V confirms an oxidized mantle source for the generation of fertile Cu-rich arc magmas that can lead to mineralized systems at shallow depths.
V33F-02 INVITED
Controls on the Solubility of Sulfur in Anhydrous and Hydrous Silicate Melts Saturated with Sulfide Liquid, Pyrrhotite or Anhydrite: Empirical Models and Applications
The solubility of S in silicate melts depends on P, T, composition and oxidation state. Available experimental results indicate that under reducing conditions S is mainly dissolved as S2- whereas under oxidizing conditions S is mainly dissolved as S6+ in silicate melts. Under reducing conditions excess S in Fe-bearing silicate melts occurs as sulfide liquid at high T and as pyrrhotite at low T. In Ca-bearing systems under oxidizing conditions excess S is present as anhydrite. At sulfide saturation the composition of silicate melt is controlled by fO2/fS2 and the volume of sulfide liquid. Such interdependency allows the construction of an empirical equation that relates S solubility to P, T and silicate melt composition using available experimental results of sulfide-saturated systems. This empirical equation can be used as a yardstick to measure deviation of S solubility in anhydrite-saturated systems and hydrous systems saturated with pyrrhotite or anhydrite. A comparison between the predicted values from the empirical equation and the results of available experiments reveals that S solubility in anhydrite- saturated, anhydrous silicate melts is one order of magnitude higher than that in sulfide-saturated silicate melts. Such a large difference is also present in hydrous systems. The solubility of S in hydrous melts saturated with pyrrhotite may be as much as 3 times the predicted values. In contrast, the solubility of S in hydrous melts saturated with anhydrite is one order of magnitude higher than the predicted value. The experiments of Luhr (1990) indicate that the positive deviation of S solubility in H2O-saturated dacitic magma is positively correlated with CaO and H2O concentrations when anhydrite is present and negatively correlated with H2O concentration when pyrrhotite is present. The empirical equation, when considered in conjunction with the effects of oxidation and hydration on S solubility, can be used to evaluate the consequences of magma differentiation. The fate of chalcophile elements during magma degassing can be evaluated based on available experimental results. Regardless of whether or not magma is saturated with sulfide or anhydrite, degassing is an effective mechanism to transfer S from magma to a vapor because of the high vapor/magma D value for S. However, mass transfer of chalcophile elements from magma to a vapor during degassing is more complicated. If magma is not saturated with sulfide liquid, degassing can rapidly transfer chalcophile elements such as Os, Pd, Au and Cu from magma to a vapor because these elements partition more strongly into vapor than magma. Degassing is unlikely to be an effective mechanism to transfer chalcophile elements from sulfide liquid to a vapor because these elements favor sulfide liquid over a vapor. In pyrrhotite-saturated systems Cu and Au may be transferred from magma to a vapor if the vapor/pyrrhotite ratio is high. The transfer of chalcophile elements from magma to a vapor is possible in anhydrite-saturated systems because anhydrite does not concentrate PGE, Au and Cu from coexisting magma.
V33F-03
Higher Melt Fractions are Generated in Metamorphosed Hydrothermally Altered Rocks: Significance to Ore Genesis
Felsic magma-related ore deposits are Earth's largest source of Cu and Mo, and a major source of Au. Despite this, there have been few field-based studies of processes that affect metal distribution in magma source regions. Based on geochemical studies at ore deposits, it has been suggested that in magmatic arcs, magmas and metal may be derived from subducted oceanic slab, overlying metasomatized mantle wedge, or from the lower crust during anatexis. Intrusion-related Au systems form distally to arcs and are typically associated with reduced granitic magmas thought to be largely derived from partial melting of crustal rocks, casting into question the absolute need for oceanic slab or mantle involvement. Even within arcs, evolution of ore producing magmas may involve mixing between mantle- and crust-derived magmas. It is therefore crucial to understand processes that influence the metal content of these crust-derived magmas. We have compared melting of alteration zones in metamorphosed gold deposits with that of unaltered rocks of the same protolith using the computer program THERMOCALC to examine the relative contributions of metal-rich and metal-poor source regions to crust-derived magmas. Potassium enrichment, caused by pre-metamorphic hydrothermal alteration, is found to stabilize a higher proportion of muscovite and biotite at high metamorphic grades than is possible in unaltered rocks. Because these micas contain water, they control the melt fraction generated through dehydration melting. Therefore, in the vicinity of metamorphosed gold mineralization, the greatest volume of felsic melt is generated in the more metal- and sulfur-rich rocks. Sulfides and Au dissolve, and are physically incorporated into the resulting felsic melt, which thereby becomes metal- and S-enriched. Since the presence of melt significantly weakens rocks, strain preferentially partitions into these melt-rich mineralized zones promoting melt segregation there. As strain increases, high-melt domains become shear zones that expand and link up with other incipient shear zones, forming low pressure channelways for migrating melts. In this way, mineralized zones become part of a regional network of magma pathways through the crust, capable of continuously enriching migrating magmas in metal.
V33F-04
Simple Models of Melting and Trace Element Transport in Subduction Zones
Melting at subduction zones has been recognized as a combination of dry decompression melting and wet melting. This study focuses on 2D models in which hydrous fluids released by slab dehydration reactions rise into the wedge by porous flow to provoke melting following the numerical models of Cagnioncle et al. [2007]. In these models, melting is parameterized according to MELTS [Hirschmann et al., 1999] and therefore depends on pressure, temperature, water fraction and degree of depletion. Depletion is in turn influenced by solid flow which dictates the rate at which fertile material is brought into the melting region. Solid flow thereby plays an important role in determining both melt production rates and the extent of melting. Transitions from an oceanic-oceanic to an oceanic-continental subduction are good locations to investigate the influence of solid flow on melting because they significantly alter solid flow: indeed, the increase in crustal lid thicknesses associated with the presence of a continent may reduce solid flow relative to the amount of water released by the slab and influence the relative contributions of flux and decompression melting. Trace elements concentrations in arc lavas provide a good indication of melting behavior in the wedge and have long been used to help understand the respective contributions from the slab, sediments and the mantle. However, in the past, models used to interpret trace element patterns have focused on melting processes (batch melting, fractional melting, or hybrid models of these two end members) and have not taken into account the fluid dynamics of melt transport. This study therefore also investigates the influence of melting and melt migration on stable trace element geochemistry in an open system. We vary the lid thickness in our temperature-dependent viscosity thermal models and examine the variations both in the degree of melting and trace element concentrations. Two end-member trace element transport behaviors are studied: equilibrium transport in which the melt completely equilibrates with the solid as it percolates from its source and which is therefore equivalent to batch melting; and disequilibrium transport in which melt is locally in equilibrium with the matrix and is thereby similar to fractional melting. We concentrate on the contribution from the solid source to trace element patterns and compare the resulting trace element concentrations of the aggregated melt to those of subduction zones which transition from an oceanic to a continental setting (e.g. Aleutians/Alaska Peninsula, Tonga).
V33F-05
The effect of assimilation and fractional crystallization on U-series disequilibria in arc lavas
U-series disequilibria in young arc lavas provide critical temporal constraint on magmatism in arc settings. Most young arc lavas have (231Pa/235U) and (226Ra/230Th) greater than unity. Although the majority of young arc lavas have 238U excess over 230Th, a significant amount of samples have (238U/230Th) < 1. Models involving recent addition of U-Ra rich fluids to the mantle wedge can explain the 238U and 226Ra excesses over 230Th, but it cannot explain 231Pa excess or 230Th excess. In-growth melting models can produce 231Pa excess and 230Th excess, but it cannot account for the overall positive correlation between 226Ra excess and Ba/Th or Sr/Th, which is interpreted as a typical feature of fluid addition. There are still heated debates on generation of U-series disequilibria in arc lavas and time-scale of magmatisms in convergent margins. Most arc lavas have experienced significant assimilation and fractional crystallization (AFC) during magma evolution. Although AFC processes can fundamentally change trace element and isotopic features of magma, its effect on U-series disequilibria has not been rigorously investigated despite its importance. It is not quite clear yet whether the correlations of U-series data with major-trace element compositions of arc lavas (e.g., (226Ra/230Th) vs. Sr/Th or Ba/Th) reflect the geochemical features of the fluid-metasomatised mantle wedge or result from magma differentiation processes. In this study, we use a numerical model to show that time-dependent AFC processes can significantly change U-series data and other geochemical features of arc lavas via fractional crystallization, decay of short-lived nuclides, and dilution by old crustal assimilant. The positive correlation between 226Ra excess and Ba/Th or Sr/Th can be produced by fractionation of plagioclase and amphibole with contemporaneous decay of 226Ra. Assimilation of old crustal materials may dilute the primary U-series disequilibria originating from melting of the metasomatised mantle. Our model can also reconcile the discrepancy of the time scales of arc magmatism indicated by U-Th-Pa-Ra and 10Be/Be systematics. Because correlations between parameters sensitive to fluid addition and U-series data could simply be produced by the AFC processes in a magma chamber within several thousands years, recent fluid addition and ultra-fast upwelling rates of magma may not be required for generation and preservation of 226Ra excess in young arc lavas. U-series disequilibria observed in arc settings might reflect in-growth melting in the mantle wedge plus magma differentiation processes in magma chamber at crustal depths.
V33F-06
Sulfate Saturated Hydrous Magmas Associated with Hydrothermal Gold Ores
Hydrothermal ore deposits associated with arc magmatism represent important sulfur anomalies. During degassing of magmatic systems the volatile may transport metals and sulfur and produce deposits. The ultimate origin of the magma-derived sulfur is still uncertain. The Yanacocha high-sulfidation epithermal Au deposit, Peru, is hosted by a Miocene volcanic succession (ca. 16 to 8 Ma). Magmatic rocks are highly oxidized >NNO+2 and show a range of composition from andesite to dacite. Two populations of amphibole occur in the Yanacocha dacitic ignimbrite deposits (~7 and 12 wt% Al2O3). Low Al amphiboles crystallized at ~ 1.5-2 kbar and 800°C (Plag-Hb thermobarometry) in equilibrium with plagioclase and pyroxene. High Al amphiboles only contain inclusions of anhydrite associated with apatite (up to 1.2 wt% SO3), and have a higher Cr2O3 content (up to 1000 ppm). We estimate these amphiboles form near the magma's liquidus at P(H2O)> 3kbar and 950 to 1000°C of a basaltic, basaltic andesite ascending magma. Low Al amphibole presents an REE pattern with negative anomalies in Sr, Ti and Eu, characteristic of plagioclase and titanite fractionation in the magma. High Al amphiboles are less enriched in REE and have no Sr, Ti, or Eu anomaly. Rare crystals of high Al amphibole display a low Al rim marked by higher REE contents compared to the core and a negative Eu anomaly. Magmatic sulfate occurrences have been discovered through the 8 m.y. volcanic sequence. Rounded anhydrite crystals are found included within clinopyroxene and both high and low Al amphibole. The rare high Al amphiboles (from the sample RC6) contain up to ~10 vol.%, ~5-80 micrometer-long anhydrite as irregularly shaped (amoeboid) blebs that do not show crystallographic forms and do not follow host cleavages. Extremely rare sulfide inclusions are found in plagioclase (Brennecka, 2006). The major and trace element contents of Yanacocha magmatic anhydrite have been analyzed by electron microprobe and LA-ICP-MS. Yanacocha anhydrite, hosted by amphiboles, are enriched in FeO (up to 0.6 wt%) and present positive anomalies in Eu and SrO (up to 8000ppm in anhydrite blebs hosted by high Al amphibole of the sample RC6). Anhydrite hosted by clinopyroxene (CPx) and low Al amphibole present higher Ce2O3 content (up to 2000ppm in CPx). In comparison, hydrothermal anhydrite analyzed from El Salvador, Butte and Ajo ore deposits contain less SrO (~ 2000 ppm) and no FeO. Pinatubo anhydrite phenocrysts and inclusions from the 1991 Pinatubo dacite yield low FeO contents, except anhydrite included in amphibole. These data suggest FeO in anhydrite is a product of subsolidus diffusion from the host. The breakdown of abundant anhydrite crystals "stored" in the magma may source of SO2-rich hydrothermal fluids that produced the sulfur enrichment (>500 M Tonnes) observed the Yanacocha hydrothermal gold deposits. The two populations of amphibole are evidence of magma mixing in the Yanacocha magmatic rocks. A sulfate-saturated oxidized dacitic magma chamber resided at about 4 to 8 km depth and 800°C was periodically underplated or fed by hydrous sulfate-rich oxidized basaltic-andesite magma. The shape of the irregular anhydrite blebs suggest that these inclusions could have been trapped as an immiscible sulfate- phosphate rich melt, despite the fact that anhydrite normally has a liquidus temperature of 1450°C and the host amphiboles crystallized at no more than 1050°C based on experiments on andesites and dacites.
V33F-07
Ancient crust in the world's youngest giant porphyry Cu-Au deposit, Ok Tedi, Papua New Guinea
The Ok Tedi Intrusive Complex in Papua New Guinea is dominantly composed of shoshonitic monzonite to monzodiorite. It is associated with one of the world's largest Au-rich porphyry Cu deposits (PCDs). The first zircon U-Pb zircon crystallization age determination for the Ok Tedi intrusions obtained using SHRIMP analysis are presented here. The mean concordant age of the mineralized monzonite porphyry is 1.12 plusminus 0.05 Ma, making Ok Tedi the world's youngest giant PCD and implying that it was exhumed at a rate of 3-5 km/Myr. This is consistent with Pliocene exhumation rates for Papua New Guinea. Many of the analyzed zircon grains contained inherited Proterozoic cores (ranging ~1.7-2.1 Ga, 206Pb/207Pb age), suggesting that magma evolution involved assimilation and/or partial melting of an ancient fragment of the North Australian Craton, possibly in the lower crust. Recycling of older cratonic continental crust has also been documented for the giant deposits of Bingham (USA) and Chuquicamata (Chile), suggesting that crustal contamination is a common process in the genesis of magma that forms PCDs, either by assimilation or by mixing of lower-to-middle crustal material with hydrous mantle derived melts. It is discussed whether this should be considered vital for the genesis of this deposit type.