Porphyry and epithermal ore deposits, which are the products of magmatic hydrothermal fluids, are intimately associated with volcanoes in continental and island arcs above subduction zones, but the exact nature of this relationship has remained enigmatic. Although metal deposition is usually thought to occur during the waning stages of volcanism, numerous ore deposits have been demonstrated to be synvolcanic. Here we show how the formation of these deposits is tied to volcanic cycles. We relate the chemical variations in vapors from Merapi volcano, Indonesia, to different stages of its eruptive cycle. The chemical compositions of volcanic vapors from subduction zone volcanoes are then compared globally to those of fluid inclusions from porphyry-epithermal deposits. These data show that adiabatic decompression is the principal control on mineralization. The data also suggest that volcanic and subvolcanic magmatic-hydrothermal systems are under lithostatic pressure during quiescence but decompress rapidly during injections of mafic magma and explosive eruptions. During quiescence, the magma evolves through fractional crystallization and devolatilization, gradually becoming oxidized and enriched in gold and other incompatible metals. Upon the injection of sulfur-rich mafic magmas, subvolcanic intrusions brecciate the overlying rocks, the systems are depressurized, the volcanoes erupt explosively, supercritical fluids unmix into vapor and brine, and base metal sulfides precipitate.
Magmatic-hydrothermal ore deposits form over periods ranging from thousands to millions of years (Sillitoe, 2010), a generally shorter lifespan than those of arc volcanoes, which are thought to usually remain active for millions of years (Newhall et al., 2000). Although the economic mineralization is usually considered to postdate volcanism (Williams-Jones and Heinrich, 2005), numerous ore deposits have been shown to be genetically associated with volcanism (Hattori and Keith, 2001) or even interrupted by episodes of renewed magmatic activity (Maughan et al., 2002). It is interesting that the injection of sulfur-rich mafic magmas into shallow porphyry stocks appears to trigger explosive eruptions (Eichelberger, 1980) and generate ore deposits (Hattori and Keith, 2001). In this paper, we report data on the distribution of metals, sulfur, and chlorine in volcanic vapors collected at Merapi volcano (Indonesia) during a phase of quiescent degassing in A.D. 2004 and after a large explosive eruption in 2006, and compare them to data reported for volcanic vapors globally and for fluid inclusions from porphyry deposits. These data and comparisons provide the first convincing evidence that injections of mafic magma (which commonly coincide with explosive volcanic eruptions) are followed by decompression of the magmatic hydrothermal system, inducing fluid phase separation, rapid cooling, and the deposition of porphyry and epithermal ores. In so doing they demonstrate the important links between arc volcanism and porphyry-epithermal ore formation.
Volcanic vapors were sampled by inserting silica tubes into fumaroles and connecting the tubes to ice-trap condensers. The liquid that collected in the ice trap was analyzed by inductively coupled plasma–mass spectrometry (ICP-MS), ICP–optical emission spectrometry, instrumental neutron activation analysis, and ion chromatography. Sublimates from the vapors were collected by inserting silica tubes into fumaroles immediately after the condensates were sampled and waiting several days until sufficient solid had accumulated on the inner walls of the tubes. Temperatures were measured along the tube centerlines using a thermocouple. Aliquots of sublimates were mounted on polished sections for mineral identification via electron probe microanalysis and the remainder was analyzed using X-ray diffraction.
The concentrations of ligands (S-Cl) and metals (Au-Cu-Zn-Pb-Mo-Y-U) in condensed vapor were normalized to Na to compensate for variations in density and thus the water fugacity, fH2O, of the vapor (Fig. 1A). The vapors were enriched in S-Cl-Pb-Cu-Zn in A.D. 2006 samples (explosive) and enriched in Mo-Y-Au-U in 2004 (quiescent). In 2006, S and Cl concentrations in the vapor ranged from 0.7 wt% to 1.0 wt% and 0.5 to 0.7 wt%, respectively, whereas they varied from 260 to 520 ppm and 890 to 3030 ppm, respectively, in 2004 (Table 1). Copper and Zn concentrations ranged from 0.16 to 0.28 ppm and 0.15 to 0.42 ppm, respectively, in 2006, and only 0.003 ppm and 0.06 to 0.15 ppm, respectively, in 2004. By contrast, concentrations of Mo only reached 0.6 ppb in 2006 compared to 4–20 ppb in 2004. The concentrations of Y were 0.7–4 ppb in 2004 and 0.4–0.5 ppb in 2006, those of Au were 0.2–0.6 ppb in 2004 and 0.02 ppb in 2006, and those of U were 0.09–0.34 ppb in 2004 and 0.06–0.08 ppb in 2006.
The concentrations of the rare earth elements (REEs) were first normalized to Na and subsequently to chondrite (McDonough and Sun, 1995) to allow comparison with subduction zone fluids (Fig. 1B). The doubly normalized REE profiles all show light REE enrichment typical of subduction zone fluids. The vapors from 2004 had higher total REE contents compared to those from 2006. The concentrations of large ion lithophile elements (LILEs) and high field strength elements (HFSEs) were first normalized to Na and subsequently to mid-oceanic ridge basalt (McDonough and Sun, 1995) to evaluate fluid behavior (Fig. 1C). All samples display gradual enrichment of progressively more incompatible elements, except Ti, because it was compatible in titanomagnetite (Camus et al., 2000; Nadeau et al., 2013). All incompatible elements are enriched in 2004 samples relative to those for 2006, except for Zr and Hf in one sample.
Sublimate Mineralogy and Redox State
A variety of minerals precipitated from the vapors during cooling along the silica tubes. In 2004, the mineral assemblage consisted of chlorides, sulfates, and hydrated sulfates, oxides, and native elements, whereas the minerals comprised chlorides, sulfides, arsenosulfides, and native elements in 2006 (Fig. 2; Table DR1 in the GSA Data Repository1).
Iron precipitated in ilmenite and hematite in 2004 and in pyrite and smythite in 2006; Pb precipitated as anglesite in 2004 and as galena and tsugaruite (Pb4As2S7) in 2006; and Bi and Zn-Cd precipitated as bismuthinite and wurtzite-greenockite in 2006. The magmatic-hydrothermal system was thus relatively oxidized in 2004 and reduced in 2006.
Magmatic Arc Fluids
In order to expand our database beyond that of Merapi, and to compare volcanic vapors to ore fluids, we compiled published compositional data for 282 samples of vapor condensates from 19 subduction zone stratovolcanoes (Table DR2) and 292 fluid inclusion assemblages from 7 major porphyry Cu ± Mo ± Au deposits (Table DR3; Fig. 3). These data are illustrated in violin plots showing median (black dot) and interquartile range (black line) values as well as the total data distribution (gray envelope) (Fig. 3; see the Data Repository). The volcanic vapors ranged in temperature from 85 °C to 1020 °C, averaging 445 ± 216 °C (1σ), whereas the fluid inclusions were trapped at temperatures from 220 °C to 771 °C, or an average of 434 ± 130 °C (1σ). On average, base metal concentrations in fluid inclusions from porphyry deposits are more than three orders of magnitude higher than in volcanic vapors. Given their similar temperatures, and experimental studies of metal solubility in aqueous fluids, the higher metal content of the fluid inclusions is attributed to their higher fluid density and resulting higher capacity to hydrate metal complexes. At constant density, metal solubility increases with increasing temperature due to a corresponding increase in the partial pressure of H2O, which promotes the formation of metal complexes with high hydration numbers (e.g., Migdisov et al., 2014). Increasing temperature, however, destabilizes metal complexes with high hydration numbers, and a maximum solubility is reached when this effect outweighs that of the increasing partial pressure of H2O. In volcanic emissions, the Pb/Na ratio reaches a maximum in fluids with temperatures of 400–600 °C, Zn/Na is greatest in fluids with temperatures of 400–800 °C, and Cu/Na and Au/Na have maxima in fluids with temperatures of 200–400 °C. The use of Y as a proxy for the REEs suggests that REE/Na ratios are greatest in vapors with temperatures of 400–600 °C (Fig. 3A).
The Pb/Na, Zn/Na, and Mo/Na ratios are much lower in fluid inclusions than in volcanic vapors (Fig. 3B). This probably results from the preferential incorporation of Na in alteration minerals precipitating from the fluid between the porphyry and the fumaroles. By contrast, Cu/Na is higher in supercritical and vapor fluid inclusions and lower in brines and volcanic vapors. However, the higher content of Cu in vapor inclusions has been interpreted to result from postentrapment diffusion of Cu toward the vapor inclusions (Lerchbaumer and Audétat, 2012). The Cu/Na ratio is thus higher in supercritical fluids than in brines, vapors, and volcanic vapors, implying that Cu-bearing minerals undergo large-scale precipitation during the unmixing of supercritical fluids into brine and vapor.
Physical State of the Volatile Phase
An important difference between porphyry fluids and volcanic vapors is their physical state. Magmatic volatiles exsolved at pressures and temperatures greater than those of the critical point, i.e., porphyry fluids at depth, are in a supercritical state. When supercritical fluids ascend, they may reach the critical point and unmix into liquid and vapor. Upon unmixing, Cl preferentially partitions into the liquid forming brine (Bodnar et al., 1985), whereas S partitions preferentially into the vapor that is ultimately emitted as a volcanic gas (Simon and Ripley, 2011; Nadeau et al., 2013).
Pressure has a major impact on the physical state of magmatic hydrothermal fluids because decreasing pressure may lead to phase separation. It is possible to relate the pressure at which supercritical fluids unmix into brine and vapor to a critical depth, by assuming lithostatic and hydrostatic pressure gradients. According to experimental studies in the H2O-NaCl system, for a magmatic fluid at 1000 °C and 5–10 wt% NaCleq (Zajacz et al., 2008), the critical depth is 5 km under a lithostatic pressure gradient and 15 km under a hydrostatic pressure gradient (Driesner and Heinrich, 2007). The addition of CO2 and sulfur species to the system will increase this critical depth, but phase relationships in the C-O-H-S–NaCl system are not sufficiently known to quantify this effect (Webster and Mandeville, 2007).
Variations in Sulfur and Chlorine
Porphyry-type deposits have greater contents of sulfur than of the base and precious metals for which they are exploited. At Merapi, volcanic vapors were highly enriched in S following the explosive eruption of 2006 compared to volcanic vapors that were quiescently degassed in 2004 (Table 1). The total concentration of S in vapor condensates was 6730 and 10,580 ppm (in samples 1 and 2, respectively) in 2006, and 260 and 520 ppm, respectively, in 2004; the higher concentration of 2006 was attributed to the injection of sulfide melt–saturated mafic magma in that year (Nadeau et al., 2010). Chlorine was present in significant concentrations in volcanic vapors released from Merapi in both 2004 and 2006, but its concentration varied much less than that of sulfur between periods of quiescence and eruption.
During periods of quiescent degassing, magma becomes oxidized due to the preferential incorporation of Fe2+ in minerals (Kelley and Cottrell, 2012) and fluids (Bell and Simon, 2011) and preferential degassing of H2 relative to O2 during H2O disproportionation (Mathez, 1984). Although primitive arc magmas are usually considered to be oxidized [oxygen fugacity, fO2, ≥ FMQ (fayalite-magnetite-quartz) + 2; Mungall, 2002], the deep mafic magmas feeding Merapi were sufficiently reduced, with fO2 at or near FMQ (Jugo, 2009), to exsolve an immiscible sulfide melt (Nadeau et al., 2010) and an aqueous fluid that ultimately precipitated sulfides (Fig. 2B). Such immiscible sulfide melt inclusions also have been observed at other arc volcanoes and in porphyry Cu deposits (Larocque et al., 2000), suggesting that deep mafic feeders commonly may be less oxidized than shallower porphyry stocks.
Variations in the degree of oxidation of magmas and fluids have an important impact on metal transport characteristics. Highly oxidized magmas lack sulfides that could scavenge chalcophile and siderophile metals from magmatic-hydrothermal systems. In contrast, S-rich reduced magmas may exsolve chalcophile element–rich immiscible sulfide melts that, in turn, can be subsequently dissolved in magmatic hydrothermal fluids (Nadeau et al., 2010; Mungall et al., 2015). High oxidation states of hydrothermal fluids lead to low H2S/SO2 ratios, inhibiting the solubility of metals that are dissolved dominantly as bisulfide complexes (Zezin et al., 2011) but enhancing the capacity of the melt to transport sulfur as SO42– (Jugo, 2009).
Metallogenesis of Porphyry-Type Deposits
We propose that porphyry-epithermal systems evolve by cycling through periods of volcanic quiescent degassing dominated by fractional crystallization and volatile exsolution, and periods of explosive activity triggered by injection of mafic magmas (Fig. 4). According to this hypothesis, quiescently devolatilizing magmas evolve under lithostatic pressure gradients to relatively shallow depths (the critical depth is 5 km in the H2O-NaCl system; Driesner and Heinrich, 2007). Incompatible metals such as Au, Mo, REEs, HFSEs, and LILEs are gradually enriched in the magmas and fluids and deposited in these relatively oxidizing environments. Injections of less oxidized mafic magmas open up the systems and abruptly decrease the confining pressures to values closer to the hydrostatic gradient, thereby displacing the critical depth to deeper levels (15 km in the H2O-NaCl system; Driesner and Heinrich, 2007). Regional tectonic earthquakes, such as that near Merapi during the eruption of 2006, also may enhance the pressure regime transition by opening the systems. Although experimental studies in the H2O-NaCl system suggest a critical depth of 15 km under a hydrostatic pressure gradient, a study of rock rheology suggests that a hydrostatic regime would be difficult to maintain at depths greater than ∼5–10 km (Sibson, 1992). Therefore the pressure regime at this stage may be between lithostatic and hydrostatic. Upon depressurization, sulfur partitions preferentially into the vapor, which rises and expands, whereas chlorine remains in the brine that ponds beneath the volcano.
After the injection of mafic magma, Cu bonds with S to precipitate as a sulfide mineral under the combined effects of decompression and expansion of the hydrated metal complexes. The precipitation of Zn and Pb, which are also enriched in the hydrothermal system after rock failure and decompression, is not as abrupt as that of Cu (Fig. 3B) as these metals form stronger bonds with Cl. Their precipitation in sulfide minerals is thus delayed relative to Cu. After explosive eruptions, which typically last several months and contribute to fracturing and brecciating rocks at progressively greater depths, the magmas evolve by fractional crystallization, gradually becoming more oxidized and enriched in incompatible elements. The hydrothermal system seals itself by precipitation of minerals along fractures and in breccias under the weight of the rock column, with pressure returning to lithostatic levels. The ability of the fluid to dissolve metals gradually increases until the next explosive eruption (or rock failure at depth), typically a few years later.
CONCLUSIONS AND IMPLICATIONS
Although economic mineralization is usually thought to postdate volcanism (Williams-Jones and Heinrich, 2005), numerous ore deposits have been shown to be genetically associated with volcanism (Hattori and Keith, 2001). In this paper we have related variations in the composition of volcanic vapors from Merapi volcano to different stages of its eruptive cycle and compared these compositions to those of volcanic vapors from other subduction zone volcanoes and porphyry-hosted fluid inclusions. These data provide the first convincing evidence that injections of mafic magma (that commonly coincide with explosive volcanic eruptions) are followed by decompression of the magmatic hydrothermal system, inducing fluid phase separation, rapid cooling, and the deposition of porphyry and epithermal ores. In so doing they demonstrate that decompression is the principal control on mineralization, and that pressure in the systems transforms abruptly from lithostatic during quiescence to hydrostatic during and immediately after an explosive eruption (rock failure at depth). During quiescence, magmatic-hydrothermal systems evolve through crystallization and devolatilization, gradually becoming oxidized and enriched in gold and other incompatible metals. Upon injection of mafic magmas, they are depressurized, supercritical fluids unmix into vapor and brine, temperature drops, base metal sulfides precipitate at depth, and volcanoes erupt explosively.
This research was made possible by Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery grants to Williams-Jones and Stix, and an NSERC Ph.D. scholarship to Nadeau. We thank F. Gaillard and two anonymous reviewers for their constructive comments.