Hydrodynamic modeling of magmatic–hydrothermal activity at submarine arc volcanoes, with implications for ore formation

Earth and Planetary Science Letters 404 (2014) 307–318

Contents lists available at ScienceDirect

Earth and Planetary Science Letters www.elsevier.com/locate/epsl

Hydrodynamic modeling of magmatic–hydrothermal activity at submarine arc volcanoes, with implications for ore formation Gillian Gruen a,b,∗ , Philipp Weis a , Thomas Driesner a , Christoph A. Heinrich a,c , Cornel E.J. de Ronde d a

Institute of Geochemistry and Petrology, Department of Earth Sciences, ETH Zurich, 8092 Zurich, Switzerland focusTerra – ETH Zurich, 8092 Zurich, Switzerland c Faculty of Mathematics and Natural Sciences, University of Zurich, 8001 Zurich, Switzerland d GNS Science, P.O. Box 31-312, Lower Hutt 5010, New Zealand b

a r t i c l e

i n f o

Article history: Received 30 December 2013 Received in revised form 25 July 2014 Accepted 29 July 2014 Available online xxxx Editor: T.M. Harrison Keywords: hydrothermal multi-phase numerical simulation fluid ore

*

a b s t r a c t Subduction-related magmas have higher volatile contents than mid-ocean ridge basalts, which affects the dynamics of associated submarine hydrothermal systems. Interaction of saline magmatic fluids with convecting seawater may enhance ore metal deposition near the seafloor, making active submarine arcs a preferred modern analogue for understanding ancient massive sulfide deposits. We have constructed a quantitative hydrological model for sub-seafloor fluid flow based on observations at Brothers volcano, southern Kermadec arc, New Zealand. Numerical simulations of multi-phase hydrosaline fluid flow were performed on a two-dimensional cross-section cutting through the NW Caldera and the Upper Cone sites, two regions of active venting at the Brothers volcanic edifice, with the former hosting sulfide mineralization. Our aim is to explore the flow paths of saline magmatic fluids released from a crystallizing magma body at depth and their interaction with seawater circulating through the crust. The model includes a 3 × 2 km2 sized magma chamber emplaced at ∼2.5 km beneath the seafloor connected to the permeable cone via a ∼200 m wide feeder dike. During the simulation, a magmatic fluid was temporarily injected from the top of the cooling magma chamber into the overlying convection system, assuming hydrostatic conditions and a static permeability distribution. The simulations predict a succession of hydrologic regimes in the subsurface of Brothers volcano, which can explain some of the present-day hydrothermal observations. We find that sub-seafloor phase separation, inferred from observed vent fluid salinities, and the temperatures of venting at Brothers volcano can only be achieved by input of a saline magmatic fluid at depth, consistent with chemical and isotopic data. In general, our simulations show that the transport of heat, water, and salt from magmatic and seawater sources is partly decoupled. Expulsion of magmatic heat and volatiles occurs within the first few hundred years of magma emplacement in the form of rapidly rising low-salinity vapor-rich fluids. About 95% of the magmatically derived salt is temporarily trapped in the crust, either as dense brine or as precipitated halite. This retained salt can only be expelled by later convection of seawater during the waning period of the hydrothermal system (i.e., “brine mining”). While the abundant mineralization of the NW Caldera vent field at Brothers could not be classified as an economic ore deposit, our model has important implications for submarine metal enrichment and the origin of distinct ore types known from exposed systems on land. Sulfide-complexed metals (notably Au) will preferentially ascend during early vapor-dominated fluid expulsion, potentially forming gold ± copper rich vein and replacement deposits in near-seafloor zones of submarine volcanoes. Dense magmatic brine will initially accumulate chloride-complexed base metals (such as Cu, Fe, Pb and Zn) at depth before they are mobilized by seawater convection. The resulting mixed brines can become negatively buoyant when they reach the seafloor and may flow laterally towards depressions, potentially forming layers of base metal sulphides with distinct zonation of metals. © 2014 Published by Elsevier B.V.

Corresponding author at: focusTerra – ETH Zurich, 8092 Zurich, Switzerland. E-mail address: [email protected] (G. Gruen).

http://dx.doi.org/10.1016/j.epsl.2014.07.041 0012-821X/© 2014 Published by Elsevier B.V.

308

G. Gruen et al. / Earth and Planetary Science Letters 404 (2014) 307–318

1. Introduction Submarine hydrothermal systems form a variety of economically important mineral deposits, commonly known as volcanogenic massive sulfide (VMS) deposits (Stanton, 1958; Solomon and Walshe, 1979; Scott and Binns, 1995; Petersen et al., 2000; Scott, 2001; Franklin et al., 2005; Hannington et al., 2011). While hydrothermal activity at mid-ocean ridges (MOR) is widespread and most significant in terms of global mass and heat flux (e.g., Stein and Stein, 1994; German et al., 2003; Fisher and Wheat, 2010; Cathles, 2011; Hannington, 2011), hydrothermal systems related to bimodal or felsic magmatism in submarine arcs are considered to be better modern analogues for these fossil ore-forming systems and are being actively explored today (e.g., Hannington, 2014). Magmatism at MOR systems provides a heat source for hydrothermal circulation, while crystallization of arc magmas releases significant amounts of saline magmatic fluids that strongly interact with the convecting seawater (Yang and Scott, 1996, 2005; Zaw et al., 1996; Solomon et al., 2004a; de Ronde et al., 2014a). For example, the water content of dacitic and rhyolitic melts generated in subduction zone settings ranges from 1 to 6 wt.% (Wallace, 2005; Wysoczanski et al., 2012). In addition, subductionrelated magmas are chemically more diverse than MOR basalts and contain higher concentrations of precious metals and volatiles, in keeping with the variable metal inventory of large VMS deposits found today on land (e.g., Binns et al., 1993; Scott and Binns, 1995; Yang and Scott, 1996, 2005; Petersen et al., 2000; de Ronde et al., 2005). The typically shallower water depths at submarine arc volcanoes compared to MOR settings (e.g., de Ronde et al., 2003b) and the correspondingly lower fluid pressures favor fluid separation into a high-density, high-salinity brine and a low-density, lowsalinity vapor (e.g., Bischoff and Rosenbauer, 1987; Fournier, 1987; Foustoukos and Seyfried, 2007; Coumou et al., 2009). Temperature and pressure of phase separation control the partitioning of metal complexing agents such as sulfur or chlorine and, therefore, the preference of dissolved ore metals to enter either the vapor or the brine phase (Heinrich et al., 1999; Pokrovski et al., 2008). Hot, buoyant fluids are preferentially focused into topographic highs on the seafloor (Gruen, 2011; Bani-Hassan et al., 2012), reaching shallower depths before interacting with seawater (e.g., Baker et al., 2003). By contrast, dense brines may sink into depressions on the seafloor, as proposed to explain large, sheet-like massive sulfide ore bodies seen in ancient deposits exploited on land (Solomon et al., 2004a, 2004b). Numerous studies based on research voyages have highlighted the variable characteristics of submarine hydrothermal activity and seafloor metallogenesis along convergent settings (e.g., Wright et al., 1998; de Ronde et al., 2001, 2003b, 2007; Massoth et al., 2007). However, numerical models investigating the sub-seafloor hydrology of modern arc-related hydrothermal systems are largely lacking in the literature. Here, we use the geological and hydrological constraints obtained from studies on Brothers volcano of the southern Kermadec intraoceanic arc (34.86◦ S/179.06◦ E; see Fig. 1) to investigate processes of fluid flow and phase separation in submarine arc volcanoes, using numerical simulations with realistic properties for saline, high-temperature fluids. To obtain a better understanding of the governing parameters for submarine hydrothermal fluid flow, we compare our model results with specific field observations at Brothers. We have not attempted to quantitatively reproduce every detail of seafloor hydrothermal activity at Brothers volcano, as these will mainly be governed by (observationally unconstrained) local variations of lithologies, faults, and other geological features. In this study, we focus on the physical hydrology of the present volcanic structure, including a volcanic cone emplaced into an earlier caldera, but do not consider the

complex thermal pre-history of the submarine volcano. Even in a simplified single-stage model, a broad diversity of hydrological regimes develops in our simulations. However, they do allow identification of some first-order processes and principles that control sub-seafloor hydrology and the potential of forming ore deposits associated with submarine arc magmatism. 2. Magmatic–hydrothermal evolution and hydrological constraints at Brothers volcano Hydrothermal activity at Brothers volcano has been observed on the northwestern inner walls of the caldera (NW Caldera site), along the western caldera wall (W Caldera site), on the NE slope of the larger, main cone (Upper Cone site) and at the summit of the smaller, more degraded cone (Lower Cone site) (Fig. 1A; de Ronde et al., 2005, 2011; Baker et al., 2012; Caratori Tontini et al., 2012). In addition, an older, inactive hydrothermal field was found on the inner slopes of the southeastern caldera wall (SE Caldera site; de Ronde et al., 2005; Caratori Tontini et al., 2012). At least two major magmatic stages can be distinguished at Brothers volcano (Embley et al., 2012); the first stage includes caldera formation and then collapse of the initial volcanic edifice, with the initial hydrothermal system considered to include magmatic fluids exsolved from an underlying magma chamber (de Ronde et al., 2011). Negative anomalies in total magnetic intensity indicate that extensive, pervasive hydrothermal alteration has taken place especially along the northwestern caldera wall, but also along the southeastern and western caldera walls (Caratori Tontini et al., 2012). Dating of hydrothermal barite from the NW Caldera site using the 226 Ra/Ba method confirms that hydrothermal activity and associated massive sulfide formation dates back to at least 1200 yrs (de Ronde et al., 2005, 2011). Fluid flow at that time was probably controlled by near-vertical, discontinuous caldera ring faults (Embley et al., 2012), with vent fields migrating from early locations close to the caldera rim to later discharge near the base of the caldera walls (de Ronde et al., 2005). Sulfur isotope analyses of sulfate–sulfide mineral pairs and fluid inclusions indicate formation and trapping temperatures between 240 and 305 ◦ C, consistent with measured vent fluid temperatures of up to 302 ◦ C (de Ronde et al., 2005, 2011). The second magmatic stage includes at least two episodes of volcanic eruption that post-date caldera collapse, the first forming the smaller, more degraded Lower Cone followed by the younger, larger, less degraded Upper Cone (Fig. 1A; see also Fig. 20 in de Ronde et al., 2005; Embley et al., 2012). Present venting at the Cone sites is diffuse, of low temperature (46–68 ◦ C, with one vent up to 122 ◦ C), with low metal contents in the fluid. Nearseawater salinities of venting fluids indicate high permeability for the cones and sub-seafloor mixing of hydrothermal fluids with seawater (de Ronde et al., 2011). High gas contents as well as isotope (helium, hydrogen, oxygen, and sulfur) signatures and pH analyses, however, suggest a strong magmatic signature, especially for Cone vent fluids (de Ronde et al., 2011). Harmonic tremor measurements confirm a reservoir of gas-laden hydrothermal fluids underlying the cones (Dziak et al., 2008). At the NW Caldera site, presentday venting is focused and characterized by relatively high temperatures (265–302 ◦ C). Salinity variations in the calculated endmember vent fluids indicate that sub-seafloor phase separation is occurring at the NW Caldera site (de Ronde et al., 2011). Brothers volcano is one of four volcanoes along the Kermadec arc known to host recently deposited sulfide mineralization (de Ronde et al., 2011, 2014b; Leybourne et al., 2012a, 2012b). In general, two types of massive sulfide samples have been obtained from the NW Caldera site: (1) Cu-rich chimneys, related to high vent temperatures (274–302 ◦ C) that contain minor Mo, Bi, Co, Se, Sn, and Au mineralization, and (2) Zn-rich

G. Gruen et al. / Earth and Planetary Science Letters 404 (2014) 307–318

309

Fig. 1. Brothers volcano and modeling domain. A. Three-dimensional view looking south towards Brothers volcano. The composite volcanic edifice, 8 km by 13 km in lateral dimensions, consists of a collapse caldera of 3–3.5 km rim diameter encircling two volcanic cones (Wright and Gamble, 1999; de Ronde et al., 2005; Embley et al., 2012). The base of the volcanic edifice is at a water depth of ∼2200 m, with the deepest part of the caldera floor at ∼1850 m. The caldera walls locally rise up to 1350 m, and the two cones to 1330 m and 1220 m, respectively. The cross-section ‘X–Y’ cuts through the Upper (main) Cone and the NW Caldera vent sites and is used for the geometric models (thick white line). B. Two-dimensional unstructured mesh, showing the dimensions, the boundary conditions, and the different regions of the models.

chimneys, widespread within the vent field and which contain minor Cd, Hg, Sb, Ag, and As mineralization (de Ronde et al., 2003a, 2003b, 2005, 2011; Berkenbosch et al., 2012). By contrast, at the two main Cone sites, no massive sulfide mineralization has been found. Rather, large amounts of native sulfur, including chimneys up to 6–7 m high as well as Fe-oxide crusts, are deposited around vent orifices (de Ronde et al., 2005, 2011). In an earlier study, Gruen et al. (2012) presented two-dimensional fluid flow models using geological constraints from Brothers volcano and pure water as a proxy for salty fluids. The results showed that, for single-phase magmatic–hydrothermal circulation, the observed venting today along the caldera walls can be best explained by the presence of caldera ring faults, whereas homogeneous permeability would lead to dominant venting at the caldera floor. Venting temperatures largely depend on the degree of interaction between hydrothermal fluids and recharge by cold seawater,

which in turn depends on the permeability structure. Gruen et al. (2012) recognized that a magmatic fluid source strongly affected the dynamics and thermal structure of the hydrothermal system. They also showed that a pure water model could not satisfy the evidence for phase separation of the hydrothermal fluids, derived from a range in Cl concentrations (i.e., salinities) both greater than and less than that of seawater. In this study, we report on simulations of the Brothers volcano magmatic–hydrothermal system using the full NaCl–H2 O phase diagram and realistic properties for saline fluids (Driesner, 2007; Driesner and Heinrich, 2007). We explore a physical hydrology scenario that would evolve from the current conceptual interpretation of geological, geochemical, and geophysical observations summarized by de Ronde et al. (2011), focusing on the post-caldera cone stage of the magmatic–hydrothermal evolution (Fig. 2A). Additional

310

G. Gruen et al. / Earth and Planetary Science Letters 404 (2014) 307–318

Fig. 2. Geologic constraints and model configuration. A. Conceptual model of the present-day magmatic–hydrothermal system at Brothers volcano (modified after de Ronde et al., 2005). Vertically exaggerated. After emplacement of a high permeability cone (indicated by darker brown), venting at the Cone site(s) is diffuse and of relatively low temperatures, suggesting mixing of magmatic fluids (red arrows) with seawater (blue arrows). High-temperature venting at the NW Caldera site suggests sub-seafloor phase separation but shows weaker magmatic input (shown by lighter gray). A hydrothermal fluid reservoir underlies the cone, whereas the magmatic body driving hydrothermal activity at the NW Caldera wall has been partially crystallized (i.e., evolved hydrothermal fluids, shown as yellow arrows; cf. de Ronde et al., 2011). B. Initial and transient conditions for the central part of the modeling domain, with magma quantities, magmatic heat and fluid production scaled from an assumed 3D circular geometry to a 2D cross-section. Geothermal here means a temperature gradient of ∼20 ◦ C.

simulations exploring the earlier magmatic stage with venting at the NW caldera site are presented in Gruen (2011). 3. Method and model configuration We have computed the transient flow of saline fluids in the sub-seafloor with a continuum, porous medium approach. The governing equations of energy, fluid mass, and salt (NaCl) mass conservation are solved with a Control Volume Finite Element method implemented in the CSMP++ software platform (CVFEM; Weis et al., 2012, 2014), which can accommodate compressible flow of variably-miscible, multi-phase fluids. Properties of the saline fluid are calculated from the NaCl–H2 O model described by Driesner and Heinrich (2007) and Driesner (2007). The simulation method has been described in detail by Weis et al. (2014). Their study presented a series of benchmarks, comparing results with those from two other well-established codes, namely HYDROTHERM (Kipp et al., 2008) and TOUGH2 (Pruess, 2004), where the ranges of applicability overlap. The mass conservation equation for the combined H2 O and NaCl fluid components considers the possible phases liquid l, vapor v, and solid halite h, and is formulated as

∂( S l ρl + S v ρ v + S h ρh ) = −∇ · (vl ρl ) − ∇ · (v v ρ v ) + Q f , (1) ∂t where φ is the porosity, ρi is the density and S i the saturation of φ

phase i, and Q f is the source term of fluid mass (H2 O + NaCl). The Darcy velocity vector vi of fluid phase i is computed according to the formulation

vi = −k

kri

μi

(∇ p − ρi g),

i = {l, v }

(2)

with k as the bulk rock permeability, kri as the relative permeability and μi the dynamic viscosity of phase i, p as the fluid pressure, and g as the gravitational acceleration vector (9.81 m s−2 ). Solid halite is considered to be immobile, and its precipitation will reduce pore space, which is reflected by the applied linear relative permeability model krl + kr v = 1 − S h . The residual saturation has

been set to 0.3(1 − S h ) for the liquid phase and to 0.0 for the vapor phase. Since the system is advection-dominated, NaCl conservation is formulated without the negligible NaCl diffusion contribution as

φ

∂( S l ρl Xl + S v ρ v X v + S h ρh X h ) ∂t = −∇ · (vl ρl Xl ) − ∇ · (v v ρ v X v ) + Q NaCl ,

(3)

where X i is the NaCl mass fraction (i.e., salinity) of phase i and Q NaCl is a source term for NaCl mass. For energy conservation, we assume local thermal equilibrium between rock and fluid. As porosity within the domain is generally low, heat conduction is approximated by assuming conduction through a rock matrix of constant thermal conductivity only. The conservation equation is then formulated as

(1 − φ)

∂ Hr ∂T = (1 − φ)ρr c pr = ∇ · (K ∇ T ) + Q c , ∂t ∂t

(4)

with the enthalpy of the rock H r , the rock’s heat capacity c pr and thermal conductivity K , and a heat source term Q c . Advective energy transport by the fluid is calculated as

φ

∂H f = −∇ · (vl ρl hl ) − ∇ · (v v ρ v h v ) + Q a , ∂t

(5)

with H f as the enthalpy of the bulk fluid, h i as the specific enthalpy of the phase indicated, and Q a describing an energy added or subtraced by a fluid source or sink, respectively. Thermal equilibrium is computed by an iterative procedure, in which temperature is varied at constant pressure and bulk salinity until thermal equilibrium is reached (Weis et al., 2014). Using available geological, geochemical, and geophysical data, we set up a two-dimensional (2D) geometry to explore the current observation-based conceptual model of recent hydrothermal activity at Brothers volcano. Our model represents a cross-section through the uppermost ∼5 km of oceanic crust (Fig. 1B), cutting the Brothers volcanic edifice through two of its active hydrothermal systems: the NW Caldera and the Upper Cone sites

G. Gruen et al. / Earth and Planetary Science Letters 404 (2014) 307–318

(section ‘X–Y’ in Fig. 1A). The bathymetry is derived from a SIMRAD EM300 multi-beam system used to swath map the volcano in detail (Embley et al., 2012). Information about the presence and the location of the underlying magmatic source is obtained from harmonic tremor and seismic recordings at Brothers volcano (Dziak et al., 2008). Based on these observations, we placed the top of a magmatic body at a crustal depth of 2.5 km underneath the caldera floor (de Ronde et al., 2011). We used dimensions for the elliptical magma chamber of 3 × 2 km2 , estimated from experimentally and numerically determined relations between the collapse caldera morphology and the magmatic body responsible for collapse (Acocella, 2007, and references therein; Gruen et al., 2012). A postulated feeder dike of ∼200 m width connects the magma chamber with the post-collapse, higher permeability main cone. This dike represents the central, high-permeability part of an inferred upflow zone of ∼300 m width (de Ronde et al., 2011). To discretize the model domain, we used an unstructured mesh with a nominal thickness of 1 m in the third dimension, composed of ∼9000 triangular elements with side lengths between 100 m and 500 m, or ∼4600 nodes (Fig. 1B). The top boundary represents the seafloor where temperature, pressure, and salinity are fixed at 10 ◦ C, hydrostatic values from the load of the ocean water, and seawater salinity (i.e., 3.2 wt.% NaCl), respectively. The fully fluid-saturated model domain is initialized with seawater salinity, a geothermal gradient of ∼20 ◦ C km−1 for old oceanic crust (which relates to a basal heat flux of Q c = 0.045 W m−2 applied to the bottom boundary), and the corresponding hydrostatic pressure. The host rock has a porosity of φ = 0.1, a homogeneous permeability of k = 10−14 m2 (a value typically assumed for hydrothermal convection in oceanic crust, e.g., Jupp and Schultz, 2004), a thermal conductivity of K = 2 W m−1 ◦ C−1 , a heat capacity of the rock of c pr = 880 J kg−1 ◦ C−1 , and a density of the solid rock matrix of ρr = 2700 kg m−3 . The values correspond to the conditions used in comparable hydrothermal simulation projects (e.g., Hayba and Ingebritsen, 1997). For the intrusive units, that is, the magma chamber and dike, we use similar relationships as in Hayba and Ingebritsen (1997). For example, we initially assign higher fluid pressures with near lithostatic values of 1 kbar, porosities of 0.05, and individual temperatures and permeabilities (Fig. 2B). Permeability is temperaturedependent to represent the transition from a hot, ductile magmatic body impermeable to fluid flow, to a cold, permeable and brittle rock (Hayba and Ingebritsen, 1997; Fournier, 1999; Weis et al., 2012). As the magmatic rocks cool, their permeabilities continuously increase to the value of the surrounding rock (from 10−22 m2 at 500 ◦ C, to 10−17 m2 at 400 ◦ C, and finally 10−13 m2 in the dike and 10−14 m2 in the magma chamber, respectively, at 350 ◦ C). The release of latent heat of crystallization during the cooling of the magma chamber is taken into account by initially assigning a doubled heat capacity, which is gradually reduced to the constant value at an assumed solidus temperature of 700 ◦ C (Hayba and Ingebritsen, 1997; Weis et al., 2012). After 300 yrs of simulated time, when the dike has essentially solidified and cooled to temperatures below the brittle–ductile transition, a magmatic fluid source is applied to the uppermost edge of the residual dike (Fig. 2B), representing a pulse of magmatic fluid exsolving from the crystallizing magma chamber. Hot (750 ◦ C), saline (10 wt.% NaCl), and pressurized (1 kbar) fluid is supplied during a period of one hundred years at a rate of Q = 160 kg s−1 , which approximately corresponds to a volatile flux observed at Satsumo Iwojima, a subaerial arc volcano similar to Brothers volcano (Hedenquist et al., 1994). By assuming a circular, horizontal injection location, and scaling the source rate from 3D to 2D, we acknowledge the radial character of magmatic fluid expulsion and can capture the governing hydrological processes in our 2D section. Assuming a circular cross section of the dike with

311

a radius R = 100 m results in a source rate of Q f = 1.02 kg s−1 ( Q f = 2Q /(π R )). The source terms for the conservation of salt mass are calculated accordingly as Q NaCl = 0.1Q f and Q a = Q f h f , where h f is the specific enthalpy of a fluid with 10 wt.% NaCl at a temperature of 750 ◦ C and a fluid pressure of 1 kbar. In order to reach these source rates in our simulations, fluid expulsion has to be faster than the crystallization rate of the magma chamber. A number of lines of evidence for transient fluid pulses are indicated at Brothers volcano, including variations in 3 He/4 He and CO2 /3 He values in the hydrothermal fluids expelled at different sites, the occurrence in some chimneys of mineral assemblages related to high sulfidation environments, and the covariance of δ 34 S values and Au content within discrete bands of sulfides within individual chimneys (de Ronde et al., 2005, 2011). Similar fluid pulses have also been recognized in porphyry-mineralizing magma chambers exposed on land (e.g., Yerington, USA: Dilles, 1987). Starting with a dike solidification time of 300 yrs, we computed four different scenarios: one without a fluid source and three others with fluid sources but with different permeabilities for the cone (i.e., 10−13 m2 , 10−12 m2 , 10−11 m2 ). 4. Simulation results 4.1. Evolution of the magmatic–hydrothermal system The simulation starts with a 900 ◦ C hot magma chamber and a 750 ◦ C hot feeder dike, which are both initially impermeable (10−22 m2 ) due to their high temperatures. Within a few hundred years (Fig. 3A), the circulation of seawater cools the dike below its solidus. In response, the dike’s permeability increases, creating a 200-m-wide zone maintaining upflow of a single-phase fluid with liquid-like density (Fig. 3A) and temperatures of ∼290 ◦ C. Abundant seawater percolation through the highly permeable cone lowers temperatures of the venting fluids in the uppermost part to ∼220 ◦ C. The hydrologic regime changes significantly with the release of magmatic fluids at the top of the residual dike at about 2 km depth, after 300 simulated years (see right-hand side illustration in Fig. 2B). Fluid pressure increases near the injection point and the added heat and salinity lead to two-phase liquid and vapor coexistence (i.e., boiling) throughout the entire upflow zone (Fig. 3B). The magmatic fluids quickly become dominant over seawater (Fig. 3B). The continuous increase in temperature and bulk salinity induces the separation of a volatile vapor phase (which rises and selectively vents via the seafloor within a few years) from an increasingly saline brine (up to 80 wt.% NaCl) that is dense, and thus remains in the deeper parts of the system, retaining large amounts of salt. Further boiling and over-saturation of salt forms pockets of precipitated halite laterally around the dike, in the zone with highest bulk salinity. When the magmatic fluid pulse terminates after 100 yrs, fluid pressure drops rapidly and induces another major change in the hydrologic regime. That is, the entire high-temperature (>600 ◦ C) upflow zone changes its state from a liquid + vapor or liquid + halite coexistence to a state of vapor + halite, leading to an even lower, near-vaporstatic pressure gradient beneath the load of the seawater column (Fig. 3C). While the vapor phase is highly mobile and escapes upward, this phase change leaves the immobile halite behind, filling most of the rock pore space and therefore reducing permeability. Continuing the simulation after termination of magmatic fluid supply, we see that convection of seawater starts to dominate the hydrology again and the precipitated halite is gradually redissolved. The resulting hot, but moderately salty brine (<20 wt.% NaCl) generated in the lower part of the system is slowly “washed

312

G. Gruen et al. / Earth and Planetary Science Letters 404 (2014) 307–318

G. Gruen et al. / Earth and Planetary Science Letters 404 (2014) 307–318

313

Fig. 3. Simulation results before (A), during (B and C), and after (D) the release of magmatic fluid. Center snapshots show the distribution of temperature and pressure. Right-hand side snapshots show the bulk salinity of the fluid (color background; gray is seawater salinity) and the fraction of magmatic fluid versus seawater (grayscale contour lines; 0 = pure seawater, 1 = magmatic fluid only), as well as areas of multiple phases (inside red line) and of precipitated halite (inside green line). Left-hand plots show temperature, pressure, and salinity along a vertical line (indicated in the center snapshot pictures), as well as the present phase states of the fluid along this line (L = liquid; V = vapor; H = precipitated halite).

Fig. 4. Temporal evolution of temperature, salinity, and magmatic fraction ∼100 m below the cone summit. Also shown are the venting temperature and salinity measurements made in 2005 (data from de Ronde et al., 2011). Note a change in scale at 200 yrs (linear to log scale) and 2000 yrs (log to log scale).

out” as a seawater-dominated fluid containing an excess of magmatically derived salt (Fig. 3D). 4.2. Temporal variations of vent fluids The evolution of temperature and salinity with time, at a location approximately 100 m below the seafloor within the main cone, is shown in Fig. 4. The curves show the physico-chemical conditions of the ascending hydrothermal fluid before it mixes with ambient seawater at the ocean floor. Temperatures of the single-phase, liquid-like fluids close to the seafloor vary between 210 and 270 ◦ C during the first 300 yrs, due to the cooling of the dike. Salinities of the fluids vary only slightly (i.e., ±0.4 wt.% NaCl), indicating that minor phase separation occurs at depth. With the onset of magmatic fluid input after 300 yrs, the temperature increases up to a nearly constant value of ∼340 ◦ C, which leads to a magmatically dominated two-phase region extending over the entire upflow zone and venting of a low-salinity vapor phase together with a liquid phase of varying salinity (Fig. 4). After shut-down of the magmatic fluid source 100 yrs later, the temperature decreases by several tens of ◦ C, leading to venting of a single-phase, liquid-like fluid with increased salinity due to the dissolution of previously stored salt of magmatic origin. Without an applied fluid source (thin red line in Fig. 4), the variations in temperature decrease to ±20 ◦ C after ∼200 yrs before the temperature drops below 200 ◦ C after ∼550 yrs. Phase separation in the upflow zone does not take place throughout the entire simulation, as the physical state of the fluid is below its critical point at all times. Measured temperatures at the Cone vent sites are lower (by mostly <70 ◦ C) than the simulated values, despite a clear magmatic signal in the vent fluids. This suggests that the permeability of the cone today may be significantly higher than the 10−13 m2 assumed for our initial simulation, thereby allowing more inflow of cold seawater. We thus performed the magmatic fluid supply simulations with higher permeabilities for the cone (e.g., Fig. 5). Venting temperatures were seen to decrease with increasing permeability, yet even with a permeability of 10−11 m2 temperatures

were still higher than the observed ones, indicating either that locally even higher permeabilities in the cone may influence temperatures and salinities, or that magmatic degassing rates at the Cone sites may be smaller than the simulated ones. 4.3. Transport and transient storage of magmatically vs. seawater-derived salt In our simulations we tagged the salt added by magma degassing as a passive tracer that is transported with the hydrothermal fluids. During phase separation, the fraction of magmatically derived salt is preserved and is equal for coexisting brine, vapor, and halite phases alike. The mass of magmatically derived salt integrated over the entire modeling domain divided by the total salt mass injected during the expulsion phase shows how much salt of magmatic origin has been retained in the crust at any given time of the simulation (Fig. 6). For example, at the end of the expulsion period about 95% of the magmatically derived salt has remained in the crust. During the waning period, this transiently stored salt of magmatic origin is effectively “mined” by convecting seawater. In the simulation with an intermediate permeability for the cone (i.e., 10−12 m2 ), “brine mining” occurs within three major flushing periods that are separated by two periods of minor venting of magmatically derived salt. Between the first and the second flushing period, brines from shallower levels rise through the volcanic edifice where over a period of ∼100 yrs they initially flow laterally within the cone and then subparallel to the slope of the cone, towards the caldera floor (Fig. 5C). During this period, hydrothermal convection in the area of the cone happens almost entirely within the crust with no significant inflow from, or outflow to, the ocean (see “plateau” between 550 and 650 yrs in Fig. 6). After the second flushing period, high-temperature brines from deeper levels (see high bulk salinities at the bottom of Fig. 5D) are slowly diluted by convecting seawater and become buoyant. This second phase, with minor venting of magmatically derived salt (see “plateau” between 1200 and 1400 yrs in Fig. 6), is not related to lateral flow, but rather to a delay due to the time needed for the fluids to ascent through the crust from deeper levels. These

314

G. Gruen et al. / Earth and Planetary Science Letters 404 (2014) 307–318

Fig. 5. Close-up of circulation within, and underneath, the high-permeability (10−12 m2 ) cone region (see excerpt in Fig. 4). Indicated are temperature (color background), bulk salinity (lines), and pore velocity of the liquid phase (arrows). A. Shortly after injection of a magmatic fluid, precipitation of halite underneath the high-permeability cone more-or-less seals the entire pore space, while entrainment of seawater through the cone flanks enables efficient cooling and dilution close to the seafloor. B. The salt that has been stored in the deep part of the system is brought up towards the seafloor as the precipitated halite is slowly re-dissolved. This leads to a relatively late increase in salinity in the venting fluid (cf. Fig. 4). C. At around 650 yrs, the remaining heavy brine (∼27 wt.% NaCl) flows downwards and laterally within the high-permeability cone. D. Finally, at around 750 yrs, seawater recharge through the cone is again dominant, with the remaining salt being continuously removed from the system.

brines eventually vent to the ocean during the third flushing period. The mass balance depicted in Fig. 6 enables us to determine the amount of magmatically derived salt that; (1) has been lost to the ocean by hydrothermal venting on the seafloor (white area), (2) approaches the seafloor but may not vent completely due to the bulk salinity of the fluid being up to ∼20 wt.% NaCl (light-gray area), (3) may be “recycled”, i.e., involved in hydrothermal circulation within the crust (intermediate-gray area), or (4) is retained in the crust since the bulk salinity of >40 wt.% NaCl inhibits the rise of fluid (dark-gray area). Approximately 15% of the total salt of magmatic origin is being recycled at ∼650 yrs of simulated time. As the fraction of magmatically derived salt at this time remains almost constant (see “plateau” in the curve in Fig. 6), the portion of the salt that would be able to vent (given the fluid’s bulk salinity) also remains in the crust. This suggests that moderately saline fluids are accumulating more salt along their upflow path and then flow back downwards to lower levels within the volcanic edifice, before they are able to vent. Thus, up to 25% of the total magmatically derived salt could be involved in the convection pattern shown in Fig. 5C. The average salt flux within the cone for the period of lateral flow is about 6.9 × 106 kg yr−1 . Given the total mass of injected salt of about 3.6 × 108 kg NaCl, the fraction of salt mass available for recycling (i.e., 25% of 3.6 × 108 kg = 0.9 × 108 kg) has a predicted turnover time of about 13 yrs. As the period of lateral flow lasts significantly longer, the magmatically derived salt can be recycled several times within this convection pattern. 5. Discussion Our simulations show that different hydrological regimes can transiently occur beneath submarine arc volcanic systems. These regimes can lead to the formation of low-salinity, bouyant vapor, dense brines, and the precipitation of salt. Physical separation of the phases during fluid flow decouples the transport of heat, water, and salt from magmatic and seawater sources in a systematic way, leading to distinct thermal and geochemical characteristics of hydrothermal venting on the seafloor. The physical separation of major fluid components has profound implications for selective enrichment of ore-forming minor elements and the formation of distinct ore deposit types. 5.1. Hydrologic evolution at Brothers volcano The physical process model obtained from our simulations can explain many specific observations at Brothers volcano. Some discrepancies between predicted and observed locations and in the time sequence result from modeling the younger, post-caldera collapse cone stage alone. In reality, at least two magmatic stages are indicated by field observations (i.e., the caldera stage and the post-collapse cone stage) that temporally overlap in the real geological system. Coupled magmatic degassing at the Cone and the NW Caldera sites is indicated by correlated time variations of 3 He/4 He and CO2 /3 He ratios at both sites between 1999 and 2005

G. Gruen et al. / Earth and Planetary Science Letters 404 (2014) 307–318

315

5.2. Implications for ore formation

Fig. 6. “Brine mining” at submarine arc volcanoes, with data from simulations at Brothers volcano. The graph shows the fraction of magmatically derived salt retained in the crust after a short period of magmatic fluid release. The hatched area marks the fraction of magmatically derived salt that is located below the injection point 3 km below the cone summit and thus is not involved in hydrothermal circulation. The grey shaded areas distinguish bulk salinities of the fluids containing salt of magmatic origin: fluids with salinities above 40 wt.% NaCl are too heavy to rise, whereas fluids with bulk salinities up to 20 wt.% NaCl may be expelled on the seafloor (cf. Fig. 4). The time period around 650 yrs is characterized by almost no venting of magmatically derived salt to the ocean due to lateral flow of brines within the cone.

(de Ronde et al., 2011). The inferred entrainment of recent magmatic gas in a previously started hydrothermal convection system, however, is not captured by our model. For the Cone sites at Brothers volcano, the development of a magmatic fluid-dominated two-phase upflow zone is consistent with the tremor data of Dziak et al. (2008) and the model proposed by de Ronde et al. (2011) of a vertical fluid plume beneath the cone. The predicted mixing of the magmatic fluid with seawater in the permeable cone corroborates the interpretation of vent fluid chemistry (de Ronde et al., 2011). The inferred presence of a two-phase fluid underneath the Cone sites suggests that the hydrothermal system is currently in the phase of active degassing from the magma chamber. This implies that magmatically derived salt currently accumulates in a heavy, potentially halite-saturated brine phase at depth. Our numerical simulations predict that this salt may be remobilized and vented in the future evolution of the hydrothermal system. The physical separation of fluid phases is controlled by contrasting densities and viscosities. Low-salinity fluids (i.e., less than seawater values) are dominated by vapor and therefore rapidly rise and vent on the seafloor, leading to the rapid expulsion of magmatic heat and volatiles in the early periods of the evolution of the hydrothermal system, following magma emplacement. By contrast, the bulk of the salt added by magmatic fluid is temporarily trapped within the crust and can only be slowly expelled by later convection of seawater as it is heated by the residual heat of the magmatic intrusion, after most of the magmatic water and other magma-derived volatiles have escaped. One remarkable consequence is that, although the simulated period of magmatic fluid supply only lasted 100 yrs, vent fluids can have a significant component of magmatically derived salt for a much longer period of up to 2000 yrs. Due to the negative buoyancy of the cooling brine, it may flow laterally and downwards within highly permeable volcanoclastics immediately beneath the seafloor; a process that requires an optimal balance between mass flux, thermal conditions and fluid composition. The extended stage of “brine mining” or “washing-out” of magmatically derived salt may therefore occur without conspicuous venting of brine at the seafloor.

Physical separation and recycling of vapor, brine, and salt can lead to fractionation of ore metals. Sulfur species and sulfidecomplexed metals can become enriched in the volatile vapor phase (Seo et al., 2009), whereas metal-chloride complexes contained in a deep-seated, hypersaline liquid may be intermittently stored at depth, and may even transiently precipitate in halite-saturated pore space. The physical separation of magmatic brine and vapor and their interaction with convecting seawater may therefore lead to selective (and potentially more efficient) enrichment of ore metals at, or below, the ocean floor. Rocks underlying magmatic vapor-dominated vent sites are expected to be enriched in magmatic volatile components, creating fluid-chemical conditions with high sulfur activity ( f S 2 ) characterized by co-precipitation of sulfate + sulfide minerals with native sulfur, and by extensive acid silicate (e.g., argillic) alteration resulting from the disproportionation of SO2 to H2 S and H2 SO4 (Giggenbach, 1997; Sillitoe, 2010). Such high-sulfidation assemblages with polymorphs of silica, clay minerals, natroalunite, native sulfur and pyrite are seen in abundance in samples from the Cone sites and more rarely at the NW Caldera site (de Ronde et al., 2005). Trace metals partitioning into sulfur-rich magmatic vapor will become enriched, including B, As, Sb, Hg, Au, and possibly Cu. Further evidence for the contribution of magmatic volatiles to subseafloor mineralization of the exposed Cu- and Au-rich massive sulfide deposit at the NW Caldera site are given by the occurrence of mineral phases such as bornite + pyrite, tellurides, and natroalunite in active and inactive chimneys from this site and that of the now extinct SE Caldera site, formed during an earlier highsulfidation phase when volcanic gas dominated (de Ronde et al., 2005). Direct expulsion of significant amounts of magmatic brine onto the seafloor is physically unlikely according to our model results. It would require near-lithostatic overpressures reaching up to the seafloor, which is unlikely in the extensional regime of submarine arcs, in contrast to the compressional regime favoring porphyry ore formation on land (Sillitoe, 2010; Weis et al., 2012). Nevertheless, fluids with salinities greater than seawater concentration have been observed to actively vent on the seafloor (e.g., von Damm, 1995; de Ronde et al., 2011), including at Brothers volcano. Our simulations show that the bulk of the salt derived from the magmatic source ultimately will get expelled into the ocean, by late-magmatic convection. Seawater enriched in magmatically derived salt and chloride-complexed metals can be transported to shallower levels in the crust, and such a mixed brine may reach salinities up to 20% NaCl in the future evolution of the cone. With the likely elevated permeability in the cone, increased hydrological decoupling of water source, heat, and salt transport is expected to lead to lateral flow of this mixed brine, shifting the future points of fluid expulsion to the flanks, the base, or the shallow subsurface of the cone. A possible indication for the emergence of this future brine stage could be the presence of Fe-oxide crusts covering the northeastern slope of the Upper Cone site, which have been correlated with Fe-rich plumes over this site (de Ronde et al., 2005; Figs. 5E and 5F in de Ronde et al., 2011). 5.3. Relation to submarine ore deposits exposed and mined on land The inferred mineralization derived from magmatic vapors at Brothers volcano described in Section 5.2 may serve as an analogue for a variety of ancient Au–Cu-rich massive sulfide deposits exposed on land, including the Miocene gold- and barite-rich ores of Wetar Island (Scotney et al., 2005), the Cambrian Mount Lyell Cu–Au ore system in Tasmania (Corbett, 2001), and possibly the

316

G. Gruen et al. / Earth and Planetary Science Letters 404 (2014) 307–318

Proterozoic Boliden Cu–Au–As deposit (Weihed et al., 1996). All of these deposits have high gold grades together with elevated concentrations of As and other volatile metalloids, and they contain mineral associations of sulfides + sulfates with indications of acid leaching from Al-rich alteration assemblages. All preserve geological features indicating a transitional environment between massive sulfide exhalation and epithermal, or intrusion-hosted, ore deposition, in a shallow sub-seafloor setting beneath volcanic edifices of felsic to intermediate composition. Measured salinities and homogenization temperatures of fluid inclusions preserved in several ancient VMS deposits on land (e.g., Zaw et al., 1996; Ioannou and Spooner, 2007) indicate that the mineralizing fluids may have been negatively buoyant, i.e., denser than ambient seawater (Solomon et al., 2002; Hannington, 2014). Such negatively buoyant brines have been invoked to explain the sheet-like aspect ratio of some of the largest VMS deposits such as Rosebery in Western Tasmania and the giant massive sulfide ore bodies in the Iberian Pyrite Belt (e.g., Sato, 1972; Potter and Brown, 1977; Turner and Campbell, 1987; Solomon et al., 2004b), as a result of transient containment of ore fluids in brine pools (Solomon et al., 2002; Tornos et al., 2008). Our model results indicate that such a scenario is hydrologically possible as a natural consequence of magmatic fluid dynamics, without any external (e.g., evaporite-derived) salt. Brine pools have never been observed in active submarine arc systems to date, and Roseberytype orebodies are indeed rare features in the geological record. More commonly, brines may flow laterally through the porous subsurface, giving rise to lower-grade mineralization within volcanic breccias. Our observation that late hydrothermal brines are dominantly of seawater origin, but owe their excess salinity to magmatically derived salts, adds an interesting geochemical implication for the metal budget of submarine hydrothermal systems in arc settings. The magmatically derived salts that are transiently stored as brine, or solid halite, will also be enriched in chloride-complexed ore metals such as Cu, Fe, Zn, and Pb (e.g., Hemley et al., 1992). Plausible metal ratios (e.g., Cu/Na, Zn/Na, Fe/Na) can be inferred from microanalyses of fluid inclusions in exposed magmatic– hydrothermal systems on land (e.g., Audétat et al., 2008) and may be used to estimate the quantity of metals that are transiently stored with magmatically derived salt in submarine hydrothermal systems. Let us assume that the magmatic fluid initially containing 10 wt.% NaCl additionally carries 500 ppm Cu, 1000 ppm Zn, and 1000 ppm Pb (Audétat et al., 2008), and that these metals are predominantly chloride-complexed (Hemley et al., 1992; Yardley, 2005). We derived that 25% of the total magmatically derived salt, containing such high metal concentrations, could be involved in a lateral convection cell (cf. results discussed in Section 4.3). Cooling along a temperature gradient from initially >300 ◦ C to below <100 ◦ C (cf. Fig. 5C) causes a decrease in solubility of these metals by orders of magnitude, provided that there is enough sulfide available (Hemley et al., 1992; Yardley, 2005). If the contained metals were completely precipitated during a future episode of lateral flow within the volcaniclastics of the cone, a VMS deposit with 6.3 × 107 kg of Cu (160 kg s−1 × 500 ppm Cu × 100 yrs × 0.25), and 12.6 × 107 kg of both Pb and Zn could form. This fits remarkably well with the average metal endowment of economic VMS deposits mined on land (Hannington, 2014). The temperature structure of the flow field, combined with the different solubilities of the metals, would result in metal zonation of a Cu-rich zone near the upflow region and a Pb–Zn zone further up and away from the feeder, as observed in many VMS deposits.

6. Conclusions We conducted numerical simulations of saline hydrothermal fluid flow based on geological and hydrological constraints from a modern submarine arc volcanic system. The results show that different hydrological regimes develop successively, controlled by the state of magmatic activity, i.e., if pressurized magmatic fluids are being released at depth. If magmatic fluid release is minor or absent, a seawaterdominated convecting system will develop around a cooling magmatic body. In our simulations we did not observe phase separation during such a stage, which does not exclude that phase separation in general would be possible in response to favorable configurations of magma chamber size and depth, similar to conditions at relatively shallow MORs. If magmatic fluid release is prevalent, hydrothermal upflow zones and venting will be dominated by magmatic fluids, as is observed at the Cone vent sites at Brothers volcano. Considerable variations during such a magmatic-dominant stage are possible depending on fluid temperature, pressure, and salinity as different regions of the H2 O–NaCl phase diagram will be encountered. For example, the upflow part of the system may be in the liquid + vapor region, leading to a separation of upflowing low-salinity vapor from highly saline brines remaining at depth, or it may be in the vapor + salt region, leading to massive precipitation of salt in the subsurface. Under certain conditions, this temporarily trapped salt may be recycled within the crust, but its discharge to the seafloor always requires seawater-dominated hydrothermal circulation after cessation of magmatic fluid injection. The physical processes at depth may be masked by the shallow permeability structure of a volcano, as exemplified by the characteristics of venting at the current Cone sites at Brothers. While a magmatic signature can be seen in the vent fluids, measured temperatures are significantly lower than the values simulated for the stage of magmatic fluid injection. A more profound understanding of the sub-seafloor structure and permeability is required to further study their control on the hydrological regime. The mechanism of lateral flow near the seafloor, which we observed in one of our simulations, may describe a general process that naturally evolves from a highly dynamic hydrothermal system in any submarine arc environment involving magmatic fluid input. The combination of our hydrological scenario with an adequate sulfide source can lead to efficient precipitation of the entire metal load of recirculating brine within the uppermost part of the crust, where it is more likely to be preserved than in black smoker chimneys exposed to open seawater. Understanding specific ore systems will require better knowledge of geological structures in the sub-seafloor and fluid-compositional data, as essential input to constrain numerical models for the transient evolution of thermohaline fluid flow. Acknowledgements This study was supported by an ETH Independent Investigators Research Award (ETHIIRA, project Nr. 0-20489-08 at ETH Zurich) funding the PhD project of Gillian Gruen, and by public research funding from the Government of New Zealand to Cornel de Ronde. Constructive discussions and reviews by an anonymous reviewer are gratefully acknowledged. References Acocella, V., 2007. Understanding caldera structure and development: an overview of analogue models compared to natural calderas. Earth-Sci. Rev. 85 (3–4), 125–160. Audétat, A., Pettke, T., Heinrich, C.A., Bodnar, R.J., 2008. The composition of magmatic–hydrothermal fluids in barren and mineralized intrusions. Econ. Geol. 103 (5), 877–908.

G. Gruen et al. / Earth and Planetary Science Letters 404 (2014) 307–318

Baker, E.T., Feely, R.A., de Ronde, C.E.J., Massoth, G.J., Wright, I.C., 2003. Submarine hydrothermal venting on the southern Kermadec volcanic arc front (offshore New Zealand): location and extent of particle plume signatures. In: Larter, R.D., Leat, P.T. (Eds.), Intra-Oceanic Subduction Systems: Tectonic and Magmatic Processes: Geological Society Special Publication, vol. 219. The Geological Society, London, pp. 141–162. Baker, E.T., Walker, S.L., Embley, R.W., de Ronde, C.E.J., 2012. High-resolution hydrothermal mapping of Brothers caldera, Kermadec Arc. Econ. Geol. 107 (8), 1583–1593. Bani-Hassan, N., Iyer, K., Rüpke, L.H., Borgia, A., 2012. Controls of bathymetric relief on hydrothermal fluid flow at mid-ocean ridges. Geochem. Geophys. Geosyst. 13, Q05002. Berkenbosch, H.A., de Ronde, C.E.J., Gemmell, J.B., McNeill, A.W., Goemann, K., 2012. Mineralogy and formation of black smoker chimneys from Brothers submarine volcano. Kermadec Arc. Economic Geology 107 (8), 1613–1633. Binns, R.A., Scott, S.D., Bogdanov, Y.A., Lisitzin, A.P., Gordeev, V.V., Gurvich, E.G., Finlayson, E.J., Boyd, T., Dotter, L.E., Wheller, G.E., Muravyev, K.G., 1993. Hydrothermal oxide and gold-rich sulfate deposits of Franklin Seamount, western Woodlark Basin, Papua-New-Guinea. Economic Geology and the Bulletin of the Society of Economic Geologists 88 (8), 2122–2153. Bischoff, J.L., Rosenbauer, R.J., 1987. Phase separation in seafloor geothermal systems: an experimental study of the effects on metal transport. Am. J. Sci. 287 (10), 953–978. Caratori Tontini, F., Davy, B., de Ronde, C.E.J., Embley, R.W., Leybourne, M., Tivey, M.A., 2012. Crustal magnetization of Brothers volcano, New Zealand, measured by autonomous underwater vehicles: geophysical expression of a submarine hydrothermal system. Econ. Geol. 107 (8), 1571–1581. Cathles, L.M., 2011. What processes at mid-ocean ridges tell us about volcanogenic massive sulfide deposits. Miner. Depos. 46 (5–6), 639–657. Corbett, K.D., 2001. New mapping and interpretations of the Mount Lyell mining district, Tasmania: a large hybrid Cu–Au system with an exhalative Pb–Zn top. Econ. Geol. 96 (5), 1089–1122. Coumou, D., Driesner, T., Weis, P., Heinrich, C.A., 2009. Phase separation, brine formation, and salinity variation at Black Smoker hydrothermal systems. J. Geophys. Res., Solid Earth 114, B03212. de Ronde, C.E.J., Baker, E.T., Massoth, G.J., Lupton, J.E., Wright, I.C., Feely, R.A., Greene, R.R., 2001. Intra-oceanic subduction-related hydrothermal venting, Kermadec volcanic arc, New Zealand. Earth Planet. Sci. Lett. 193 (3–4), 359–369. de Ronde, C.E.J., Faure, K., Bray, C.J., Chappell, D.A., Wright, I.C., 2003a. Hydrothermal fluids associated with seafloor mineralization at two southern Kermadec arc volcanoes, offshore New Zealand. Miner. Depos. 38 (2), 217–233. de Ronde, C.E.J., Massoth, G.J., Baker, E.T., Lupton, J.E., 2003b. Submarine hydrothermal venting related to volcanic arcs. In: Simmons, S.F., Graham, I.J. (Eds.), Volcanic, Geothermal, and Ore-Forming Fluids: Rulers and Witnesses of Processes Within the Earth, Special Publication, vol. 10. Geochemical Society and Society of Economic Geologists, pp. 91–110. de Ronde, C.E.J., Hannington, M.D., Stoffers, P., Wright, I.C., Ditchburn, R.G., Reyes, A.G., Baker, E.T., Massoth, G.J., Lupton, J.E., Walker, S.L., Greene, R.R., Soong, C.W.R., Ishibashi, J., Lebon, G.T., Bray, C.J., Resing, J.A., 2005. Evolution of a submarine magmatic–hydrothermal system: Brothers volcano, southern Kermadec arc, New Zealand. Econ. Geol. 100 (6), 1,097–1,133. de Ronde, C.E.J., Baker, E.T., Massoth, G.J., Lupton, J.E., Wright, I.C., Sparks, R.J., Bannister, S.C., Reyners, M.E., Walker, S.L., Greene, R.R., Ishibashi, J., Faure, K., Resing, J.A., Lebon, G.T., 2007. Submarine hydrothermal activity along the mid-Kermadec arc, New Zealand: large-scale effects on venting. Geochem. Geophys. Geosyst. 8, Q07007. de Ronde, C.E.J., Massoth, G.J., Butterfield, D.A., Christenson, B.W., Ishibashi, J., Ditchburn, R.G., Hannington, M.D., Brathwaite, R.L., Lupton, J.E., Kamenetsky, V.S., Graham, I.J., Zellmer, G.F., Dziak, R.P., Embley, R.W., Dekov, V.M., Munnik, F., Lahr, J., Evans, L.J., Takai, K., 2011. Submarine hydrothermal activity and gold-rich mineralization at Brothers volcano, Kermadec arc, New Zealand. Miner. Depos. 46 (5–6), 541–584. de Ronde, C.E.J., Chadwick, W.W.J., Ditchburn, R.G., Embley, R.W., Tunnicliffe, V., Baker, E.T., Walker, S.L., Ferrini, V.L., Merle, S.M., 2014a. Molten sulfur lakes of intraoceanic arc volcanoes. In: Rouwet, D., Christenson, B., Tassi, F., Vandelbroulemuck, J. (Eds.), Volcanic Lakes. Springer-Verlag. In press. de Ronde, C.E.J., Walker, S.L., Ditchburn, R.G., Caratori Tontini, F., Hannington, M.D., Merle, S.G., Timm, C., Handler, M.R., Wysoczanski, R.J., Dekov, V.M., Kamenov, G.D., Baker, E.T., Embley, R.W., Lupton, J.E., 2014b. The anatomy of a buried submarine hydrothermal system, Clark volcano, Kermadec arc, New Zealand. Econ. Geol. 109. In press. Dilles, J.H., 1987. Petrology of the Yerington Batholith, Nevada – evidence for evolution of porphyry copper ore fluids. Econ. Geol. 82 (7), 1750–1789. Driesner, T., 2007. The system H2 O–NaCl. Part II: Correlations for molar volume, enthalpy, and isobaric heat capacity from 0 to 1000 ◦ C, 1 to 5000 bar, and 0 to 1 XNaCl . Geochim. Cosmochim. Acta 71 (20), 4,902–4,919. Driesner, T., Heinrich, C.A., 2007. The system H2 O–NaCl. Part I: Correlation formulae for phase relations in temperature–pressure-composition space from 0 to 1000 ◦ C, 0 to 5000 bar, and 0 to 1 XNaCl . Geochim. Cosmochim. Acta 71 (20), 4,880–4,901.

317

Dziak, R.P., Haxel, J.H., Matsumoto, H., Lau, T.K., Merle, S.G., de Ronde, C.E.J., Embley, R.W., Mellinger, D.K., 2008. Observations of regional seismicity and local harmonic tremor at Brothers volcano, south Kermadec arc, using an ocean bottom hydrophone array. J. Geophys. Res., Solid Earth 113 (B8), B08S04. Embley, R.W., de Ronde, C.E.J., Merle, S.G., Davy, B., Tontini, F.C., 2012. Detailed morphology and structure of an active submarine arc caldera: Brothers volcano, Kermadec arc. Econ. Geol. 107 (8), 1557–1570. Fisher, A.T., Wheat, C.G., 2010. Seamounts as conduits for massive fluid, heat, and solute fluxes on ridge flanks. Oceanography 23 (1), 74–87. Fournier, R.O., 1987. Conceptual models of brine evolution in magmatic– hydrothermal systems. In: Decker, R.W., Wright, T.L., Stauffer, P.H. (Eds.), Volcanism in Hawaii 2. Hawaiian Volcano Observatory, US Geological Survey, pp. 1487–1506. Fournier, R.O., 1999. Hydrothermal processes related to movement of fluid from plastic into brittle rock in the magmatic–epithermal environment. Econ. Geol. 94 (8), 1193–1211. Foustoukos, D.I., Seyfried, W.E., 2007. Fluid phase separation processes in submarine hydrothermal systems. In: Liebscher, A., Heinrich, C.A. (Eds.), Fluid–Fluid Interactions: Reviews in Mineralogy and Geochemistry, vol. 65. Mineralogical Society of America, Geochemical Society, pp. 213–239. Franklin, J.M., Gibson, H.L., Jonasson, I.R., Galley, A.G., 2005. Volcanogenic massive sulfide deposits. In: Hedenquist, J.W., Thompson, J.F.H., Goldfarb, R.J., Richards, J.P. (Eds.), One Hundredth Anniversary Volume. Society of Economic Geologists, Littleton, CO, pp. 523–560. German, C.R., Von Damm, K.L., Heinrich, D.H., Karl, K.T., 2003. Hydrothermal Processes, Treatise on Geochemistry. Pergamon, Oxford, pp. 181–222. Giggenbach, W.F., 1997. The origin and evolution of fluids in magmatic– hydrothermal systems. In: Barnes, H.L. (Ed.), Geochemistry of Hydrothermal Ore Deposits. Wiley-Interscience, New York, pp. 737–796. Gruen, G., 2011. Subsurface hydrology of submarine arc volcanoes, with application to the active hydrothermal system at Brothers volcano, Kermadec arc, New Zealand. Unpublished PhD thesis. ETH, Zurich. Gruen, G., Weis, P., Driesner, T., de Ronde, C.E.J., Heinrich, C.A., 2012. Fluid-flow patterns at Brothers volcano, southern Kermadec arc: insights from geologically constrained numerical simulations. Econ. Geol. 107 (8), 1595–1611. Hannington, M., 2011. Comments on “What processes at mid-ocean ridges tell us about volcanogenic massive sulfide deposits” by L.M. Cathles. Miner. Depos. 46 (5–6), 659–663. Hannington, M.D., 2014. Volcanogenic massive sulfide deposits. In: Holland, H.D., Turekian, K.K. (Eds.), Treatise on Geochemistry. 2nd edition. In: Geochemistry of Ore Deposits, vol. 13. Elsevier-Pergamon, Oxford, pp. 463–488. Hannington, M., Jamieson, J., Monecke, T., Petersen, S., Beaulieu, S., 2011. The abundance of seafloor massive sulfide deposits. Geology 39 (12), 1155–1158. Hayba, D.O., Ingebritsen, S.E., 1997. Multiphase groundwater flow near cooling plutons. J. Geophys. Res., Solid Earth 102 (B6), 12,235–12,252. Hedenquist, J.W., Aoki, M., Shinohara, H., 1994. Flux of volatiles and ore-forming metals from the magmatic–hydrothermal system of Satsuma Iwojima volcano. Geology 22 (7), 585–588. Heinrich, C.A., Günther, D., Audétat, A., Ulrich, T., Frischknecht, R., 1999. Metal fractionation between magmatic brine and vapor, determined by microanalysis of fluid inclusions. Geology 27 (8), 755–758. Hemley, J.J., Cygan, G.L., Fein, J.B., Robinson, G.R., Dangelo, W.M., 1992. Hydrothermal ore-forming processes in the light of studies in rock-buffered systems. 1. Iron–copper–zinc–lead sulfide solubility relations. Econ. Geol. 87 (1), 1–22. Ioannou, S.E., Spooner, E.T.C., 2007. Fluid temperature and salinity characteristics of the Matagami volcanogenic massive sulfide district, Quebec. Econ. Geol. 102 (4), 691–715. Jupp, T.E., Schultz, A., 2004. Physical balances in subseafloor hydrothermal convection cells. J. Geophys. Res., Solid Earth 109 (B5), B05101. Kipp, K.L., Hsieh, P.A., Charlton, S.R., 2008. Guide to the revised ground–water flow and heat transport simulator: HYDROTHERM – version 3. 160 p. Leybourne, M.I., de Ronde, C.E.J., Wysoczanski, R.J., Walker, S.L., Timm, C., Gibson, H.L., Layton-Matthews, D., Baker, E.T., Clark, M.R., Tontini, F.C., Faure, K., Lupton, J.E., Fornari, D.J., Soule, S.A., Massoth, G.J., 2012a. Geology, hydrothermal activity, and sea-floor massive sulfide mineralization at the Rumble II west mafic caldera. Econ. Geol. 107 (8), 1649–1668. Leybourne, M.I., Schwarz-Schampera, U., de Ronde, C.E.J., Baker, E.T., Faure, K., Walker, S.L., Butterfield, D.A., Resing, J.A., Lupton, J.E., Hannington, M.D., Gibson, H.L., Massoth, G.J., Embley, R.W., Chadwick, W.W., Clark, M.R., Timm, C., Graham, I.J., Wright, I.C., 2012b. Submarine magmatic–hydrothermal systems at the Monowai volcanic center, Kermadec arc. Econ. Geol. 107 (8), 1669–1694. Massoth, G., Baker, E., Worthington, T., Lupton, J., de Ronde, C., Arculus, R., Walker, S., Nakamura, K., Ishibashi, J., Stoffers, P., Resing, J., Greene, R., Lebon, G., 2007. Multiple hydrothermal sources along the south Tonga arc and Valu Fa Ridge. Geochem. Geophys. Geosyst. 8. Petersen, S., Herzig, P.M., Hannington, M.D., 2000. Third dimension of a presently forming VMS deposit: TAG hydrothermal mound, Mid-Atlantic Ridge, 26◦ N. Miner. Depos. 35 (2–3), 233–259. Pokrovski, G.S., Borisova, A.Y., Harrichoury, J.C., 2008. The effect of sulfur on vapor–liquid fractionation of metals in hydrothermal systems. Earth Planet. Sci. Lett. 266 (3–4), 345–362.

318

G. Gruen et al. / Earth and Planetary Science Letters 404 (2014) 307–318

Potter, R.W., Brown, D.L., 1977. The volumetric properties of aqueous sodium chloride solutions from 0 ◦ C to 500 ◦ C at pressures up to 2000 bars based on regression of data in the literature. U.S. Geol. Surv. Bull. 1421C, 1–36. Pruess, K., 2004. The TOUGH codes – a family of simulation tools for multiphase flow and transport processes in permeable media. Vadose Zone J. 3 (3), 738–746. Sato, T., 1972. Behaviour of ore-forming solutions in seawater. Mining Geol. 22, 31–42. Scotney, P.M., Roberts, S., Herrington, R.J., Boyce, A.J., Burgess, R., 2005. The development of volcanic hosted massive sulfide and barite–gold orebodies on Wetar Island, Indonesia. Miner. Depos. 40 (1), 76–99. Scott, S.D., 2001. Deep ocean mining. Geosci. Can. 28 (2), 87–96. Scott, S.D., Binns, R.A., 1995. Hydrothermal processes and contrasting styles of mineralization in the western Woodlark and eastern Manus basins of the western Pacific. In: Parson, L.M., Walker, C.L., Dixon, D.R. (Eds.), Hydrothermal Vents and Processes, Geological Society Special Publication, vol. 87. The Geological Society, London, pp. 191–205. Seo, J.H., Guillong, M., Heinrich, C.A., 2009. The role of sulfur in the formation of magmatic–hydrothermal copper–gold deposits. Earth Planet. Sci. Lett. 282 (1–4), 323–328. Sillitoe, R.H., 2010. Porphyry copper systems. Econ. Geol. 105 (1), 3–41. Solomon, M., Walshe, J.L., 1979. Formation of massive sulfide deposits on the sea floor. Econ. Geol. 74 (4), 797–813. Solomon, M., Tornos, F., Gaspar, O.C., 2002. Explanation for many of the unusual features of the massive sulfide deposits of the Iberian pyrite belt. Geology 30 (1), 87–90. Solomon, M., Gemmell, J.B., Zaw, K., 2004a. Nature and origin of the fluids responsible for forming the Hellyer Zn–Pb–Cu, volcanic-hosted massive sulphide deposit, Tasmania, using fluid inclusions, and stable and radiogenic isotopes. Ore Geol. Rev. 25 (1–2), 89–124. Solomon, M., Tornos, F., Large, R.R., Badham, J.N.P., Both, R.A., Zaw, K., 2004b. Zn–Pb– Cu volcanic-hosted massive sulphide deposits: criteria for distinguishing brine pool-type from black smoker-type sulphide deposition. Ore Geol. Rev. 25 (3–4), 259–283. Stanton, R.L., 1958. Abundances of copper, zinc, and lead in some sulfide deposits. J. Geol. 66 (5), 484–502. Stein, C.A., Stein, S., 1994. Constraints on hydrothermal heat flux through the oceanic lithosphere from global heat flow. J. Geophys. Res., Solid Earth 99 (B2), 3081–3095. Tornos, F., Solomon, M., Conde, C., Spiro, B.F., 2008. Formation of the Tharsis massive sulfide deposit, Iberian pyrite belt: geological, lithogeochemical, and stable isotope evidence for deposition in a brine pool. Econ. Geol. 103 (1), 185–214.

Turner, J.S., Campbell, I.H., 1987. Temperature, density and buoyancy fluxes in “black smoker” plumes, and the criterion for buoyancy reversal. Earth Planet. Sci. Lett. 86 (1), 85–92. von Damm, K.L., 1995. Controls on chemistry and temporal variability of seafloor hydrothermal fluids. In: Humphris, S.E., Zierenberg, R.A., Mullineaux, L.S., Thomson, R.E. (Eds.), Seafloor Hydrothermal Systems: Physical, Chemical, Biological, and Geological Interactions. In: Geophys. Monogr., vol. 91. American Geophysical Union, pp. 223–247. Wallace, P.J., 2005. Volatiles in subduction zone magmas: concentrations and fluxes based on melt inclusion and volcanic gas data. J. Volcanol. Geotherm. Res. 140 (1–3), 217–240. Weihed, J.B., Bergstrom, U., Billstrom, K., Weihed, P., 1996. Geology, tectonic setting, and origin of the paleoproterozoic Boliden Au–Cu–As deposit, Skellefte district, northern Sweden. Econ. Geol. 91 (6), 1073–1097. Weis, P., Driesner, T., Heinrich, C.A., 2012. Porphyry–copper ore shells form at stable pressure–temperature fronts within dynamic fluid plumes. Science 338 (6114), 1613–1616. Weis, P., Driesner, T., Coumou, D., Geiger, S., 2014. Hydrothermal, multiphase convection of H2 O–NaCl fluids from ambient to magmatic temperatures: a new numerical scheme and benchmarks for code comparison. Geofluids. Wright, I.C., Gamble, J.A., 1999. Southern Kermadec submarine caldera arc volcanoes (SW Pacific): caldera formation by effusive and pyroclastic eruption. Mar. Geol. 161 (2–4), 207–227. Wright, I.C., de Ronde, C.E.J., Faure, K., Gamble, J.A., 1998. Discovery of hydrothermal sulfide mineralization from southern Kermadec arc volcanoes (SW Pacific). Earth Planet. Sci. Lett. 164 (1–2), 335–343. Wysoczanski, R.J., Handler, M.R., Schipper, C.I., Leybourne, M.I., Creech, J., Rotella, M.D., Nichols, A.R.L., Wilson, C.J.N., Stewart, R.B., 2012. The tectonomagmatic source of ore metals and volatile elements in the southern Kermadec arc. Econ. Geol. 107 (8), 1539–1556. Yang, K.H., Scott, S.D., 1996. Possible contribution of a metal-rich magmatic fluid to a sea-floor hydrothermal system. Nature 383 (6599), 420–423. Yang, K.H., Scott, S.D., 2005. Vigorous exsolution of volatiles in the magma chamber beneath a hydrothermal system on the modern sea floor of the eastern Manus back-arc basin, western Pacific: evidence from melt inclusions. Econ. Geol. 100 (6), 1,085–1,096. Yardley, B.W.D., 2005. 100th anniversary special paper: metal concentrations in crustal fluids and their relationship to ore formation. Econ. Geol. 100 (4), 613–632. Zaw, K., Gemmell, J.B., Large, R.R., Mernagh, T.P., Ryan, C.G., 1996. Evolution and source of ore fluids in the stringer system, Hellyer VHMS deposit, Tasmania, Australia: evidence from fluid inclusion microthermometry and geochemistry. Ore Geol. Rev. 10 (3–6), 251–278.