JOURNAL OF PETROLOGY

VOLUME 38

NUMBER 12

PAGES 17411752

1997

Constraints on Magma Degassing beneath the Far Southeast Porphyry CuAu Deposit, Philippines
H. SHINOHARA1 AND J. W. HEDENQUIST2
1 2

MINERAL AND FUEL RESOURCES DEPARTMENT, GEOLOGICAL SURVEY OF JAPAN, 1-1-3 HIGASHI, TSUKUBA 305, JAPAN INSTITUTE OF GEOLOGICAL AND NUCLEAR SCIENCES, P.O. BOX 31312, LOWER HUTT, NEW ZEALAND

RECEIVED JANUARY 1997; ACCEPTED AUGUST 1997

K-silicate alteration in the Far Southeast (FSE) porphyry CuAu deposit formed at ~14 Ma, concentric to dikes of quartz diorite porphyry. At ~2 km paleodepth the hydrothermal system consisted of magmatic hypersaline liquid and vapor at 550C and 50 MPa (lithostatic pressure). Advanced argillic alteration formed at the same time over the deposit at 1 km paleodepth from acidic condensates of the vapor. At [13 Ma, K-silicate alteration was overprinted by 350C magmatic liquid (~5 wt % NaCl equiv.) at hydrostatic pressure. Sericite alteration and much of the CuAu mineralization formed at this later stage. Evolution of the magmatic uid composition was simulated with a magma-chamber crystallization model. Homogeneous crystallization during early stage convection is assumed, whereas at 50 vol. % crystals the chamber becomes stagnant and crystallizes from rim to core over a narrow crystallization interval. The model calculation, based on a magma chamber with 2 km thickness at 6 km depth (150 MPa) and [800C (saturated melt composition at 30 vol. % crystals, 5 wt % H2O, 02 wt % Cl and D of 40), can reproduce the chemical and isotopic compositions of the early and late magmatic uids. The most critical factor controlling the compositional evolution of the model hydrothermal system is the transition from convective to stagnant magma-chamber crystallization. There is also a sharp decrease in the rate of uid exsolution associated with this transition, which can account for the thermal collapse of the FSE porphyry system, from K-silicate to sericite alteration.

INTRODUCTION
Magmas become saturated with volatile components as they ascend in the crust, and either erupt or crystallize at depth. Saturated magmas that remain at depth exsolve an aqueous uid that forms a hydrothermal system on separation from the magma chamber (Burnham, 1979). Although part of the exsolved uid (a low-density vapor) may discharge to the surface from passively degassing volcanoes (Hedenquist, 1995), part will remain at depth, and this dense, residual magmatic liquid may be associated with hydrothermal mineralization in the porphyry ore environment (Hedenquist & Lowenstern, 1994). In this paper we use evidence from a porphyry ore deposit in the Philippines to determine the composition of the uid that exsolved from the deeper parent magma. The evolving composition of the magmatic uid at the depth of the ore body can then be used to place constraints on the nature of the evolved magma chamber far below the porphyry ore body. This approach diers from that of other studies where a portion of the parent magma chamber is available to study directly (e.g. at Yerington, Nevada; Dilles, 1987). The Far Southeast (FSE) porphyry CuAu deposit and overlying Lepanto epithermal CuAu ore body are located in northern Luzon. These deposits formed at the same time, from coupled hydrothermal processes, in a system dominated by magmatic uid (Arribas et al., 1995; Hedenquist et al., 1997). We use the isotopic composition of contemporaneous K-silicate and advanced argillic alteration, formed from hypersaline liquid and vapor,

KEY WORDS:

magma; degassing; evolution; porphyry-copper; isotopes

Corresponding author. Present address: Mineral and Fuel Resources Department, Geological Survey of Japan, 1-1-3 Higashi, Tsukuba 305, Japan

Oxford University Press 1997

JOURNAL OF PETROLOGY

VOLUME 38

NUMBER 12

DECEMBER 1997

respectively, plus uid inclusion data, to estimate the composition of the initial aqueous uid that exsolved from the parent magma at depth. Similar data on the subsequent stage of mineralization, dominated by a hydrothermal uid of lower salinity but also magmatic in origin, constrain the evolution of the crystallizing magma chamber, despite it lying several kilometers beneath the ore deposit.

THE FSE DEPOSIT Geology

The geology of the FSELepanto deposit is dominated by three main lithologic groups (Fig. 1): (1) a volcanic to epiclastic basement consisting of several units (including the Lepanto metavolcanic and Balili volcaniclastic rocks) of Late Cretaceous to middle Miocene age; (2) a Pliocene dacitic pyroclastic and porphyry unit (Imbanguila hornblende dacite) which pre-dates alteration and CuAu mineralization; (3) an unaltered (post-mineralization) Pleistocene dacitic porphyry of owdome complexes (Bato hornblendebiotite dacite). Near the FSE deposit two volcanic vents are lled with Imbanguila dacite porphyry and pyroclastic rocks (Fig. 1). These vents are clearly shown by the structural contours of the base of the Imbanguila unit (Garcia, 1991), and are probably the source of the Imbanguila pyroclastic and porphyry rocks. The Imbanguila dacite that hosts the Lepanto ore body erupted at 2002 Ma, and the deposit is capped by unaltered (post-mineralization) Bato dacite, dated at ~10 Ma (Arribas et al., 1995). FSE alteration and ore are centered on dikes of melanocratic quartz diorite porphyry that cut the epiclastic basement rocks (Fig. 1). Minimum phenocryst content of these altered dikes is ~30 vol. %, consisting of andesine, quartz, hornblende and biotite in order of decreasing n & Cinco, 1989). Individual inabundance (Concepcio trusive bodies are typically 50150 m wide, and are elongated up to 5:1 in a northwesterly direction, parallel n & Cinco, 1989). to the district-wide faulting (Concepcio The intrusions extend up to about 300 m elevation, with intrusive contacts nearly vertical at sea level. An intermineral series of altered leucocratic quartz diorite dikes intrude to 500700 m elevation, cut the rst set of intrusions, and contain lower grades of porphyry mineralization. A core of early K-silicate alteration (biotite magnetiteK feldspar) is associated with veins of vitreous, anhedral quartz (biotite dated at 141005 Ma, n=6; Arribas et al., 1995). This is overlain by a shallow, blanketlike zone of contemporaneous (142008 Ma, n=5) to later advanced argillic (quartzalunite) alteration. The K-silicate alteration concentrically envelopes the dikes, and is overprinted by an assemblage of chlorite

hematitesericite. The chlorite alteration is cut by veins of euhedral quartz accompanied by anhydrite chloritesericitepyritechalcopyritebornite, with halos of sericite alteration (dated at 130007 Ma, n=10). A hydrothermal breccia, post-K-silicate alteration in timing, cut the center of the porphyry and the Imbanguila dacite (Fig. 1) at the same time as formation of the later veins with sericite halos. Porphyry ore is concentric to the Ksilicate and sericitic alteration, with much of the grade associated with the sericitic stage. The Lepanto epithermal ore is supradjacent to porphyry mineralization, hosted by the quartzalunite alteration (Fig. 1).

Summary of constraints
Magma rose from a chamber at depth to form the earlymineral quartz diorite dikes after ~18 Ma. The parent magma, well below our depth of observation, exsolved an aqueous uid on saturation at some stage in its crystallization, and this aqueous uid ascended to the porphyry environment to form K-silicate alteration at ~14 Ma, including alteration of the early-mineral dikes. We do not know the water content of the melt at saturation, nor the temperature, pressure or depth of exsolution. We assume that saturation was timed to the emplacement of the early-mineral dikes, as many studies nd that porphyry dikes intruded with the same timing as hydrothermal alteration (e.g. Gustafson & Hunt, 1975). Thus, the crystal content of the magma at the inception of saturation is taken as 30 vol. %, based on the minimum crystal content of the early-mineral quartz diorite dikes. At the maximum depth of our observation (100 m below sea level), the maximum trapping temperature of inclusions of the early aqueous uid was 550C. This uid consisted of hypersaline liquid (5055 wt % NaCl equiv.) and a coexisting low-salinity, gas-rich vapor (Hedenquist et al., 1997). The pressure of this uid was 4050 MPa, indicating a paleodepth of 162 km in a lithostatic system (i.e. paleosurface near 15001900 m elevation). A single-phase supercritical liquid at 550C intersects its solvus in the NaClH2O system at ~70 MPa (Sourirajan & Kennedy, 1962), i.e. at ~3 km paleodepth. The depth of immiscibility may have been greater if the uid was gas rich, as the 550C solvus extends to a higher pressure. The hypersaline liquid formed the K-silicate alteration, and the hydrothermal biotite indicates that this liquid had a depleted-D isotopic composition (Fig. 2), averaging 45 D (Table 1). The vapor separated from the two-phase uid and condensed into meteoric water at shallower depths to form the acidic water that produced the contemporaneous advanced argillic alteration. Based on isotopic analysis of alunite, the end-member vapor (Fig. 2) had a D composition of 20 to 25 (Table 1). The isotopic compositions of the hypersaline liquid

1742

SHINOHARA AND HEDENQUIST

MAGMA DEGASSING BENEATH A PORPHYRY Cu DEPOSIT

Fig. 1. Schematic NWSE longitudinal section through the Far Southeast (FSE) porphyry CuAu and Lepanto epithermal CuAu deposits, n & Cinco, 1989; Garcia, showing the geological units and extent of porphyry and high-suldation epithermal mineralization (from Concepcio 1991). The upper limit of the hypersaline liquid, determined from uid inclusion study, is shown by the heavy dashed line (Hedenquist et al., 1997).

Table 1: Comparison of constraints from FSELepanto deposits (Hedenquist et al., 1997) and the results of model calculation
Constraints from FSELepanto deposits
Stage Age (Ma) Mineral Temp. (C) Conditions Paleodepth (m) D () NaCl (wt %) 142008 Alunite <350 Hydrostatic 1300 20 to 25 Vapor Early 141005 Biotite 450550 Lithostatic 2000 45 50 Late 130007 Illite 350 Hydrostatic 2000 43 5 75 Residual 143021 Hornblende <700

Results of the model calculation

Stage Duration D () NaCl (wt %) Temp. (C) Vapor D () NaCl (wt %) 25 07 Early 2 kyr 27 83 550 Liquid 45 50 Late 003 Myr 43 62 Residual (Hornblende) 57

D values are calculated with a vaporliquid fractionation factor of +20 for the uid separated at 550C and 50 MPa.

1743

JOURNAL OF PETROLOGY

VOLUME 38

NUMBER 12

DECEMBER 1997

Fig. 2. Calculated uid compositions for waters in equilibrium with hydrothermal minerals at temperatures determined from uid inclusion and sulfur isotope geothermometry, and the estimated composition of local paleometeoric water (Hedenquist et al., 1997). The end-member magmatic compositions (Table 1) are for uid phases with 18O >56. The compositional elds were determined for water related to Ksilicate alteration (biotite, n=6), advanced argillic alteration (alunite, n=8; samples from SE to NW extent of the Lepanto deposit), and sericitic alteration (illite, n=12; samples from the center of the deposit, labeled SE, to the northwest margin, labeled NW). Boxes show the ranges of compositions of water dissolved in silicic melts (Taylor, 1992) and discharged from high-temperature volcanic fumaroles (Giggenbach, 1992).

and vapor are consistent with those expected for water exsolved from felsic magmas (Taylor, 1992), including that of separated vapor discharging from volcanoes (Giggenbach, 1992). Hence, the hydrothermal system at 14 Ma was dominated by magmatic uid. The +20 D dierence between the vapor and hypersaline liquid, estimated from mineral data, is similar to that predicted from extrapolation of experimental data for fractionation between high-temperature vapor and hypersaline liquid (Horita et al., 1995). Thus, FSE isotope data are consistent with the vapor and hypersaline liquid having coexisted, as also indicated by the contemporaneity of alunite and biotite. However, we do not know the vapor:liquid ratio, so we cannot estimate the composition of the bulk uid exsolved from the magma at depth. Subsequent to the 14 Ma coupled K-silicate and advanced argillic alteration event, the porphyry was cut by the hydrothermal breccia, typical of those associated with build-up of pressure from a crystallizing magma (Burnham, 1979). Sericite had already begun to form at this time. The isotopic composition of sericite (illite) in the ore body center (Fig. 2) indicates formation from liquid with a magmatic signature (i.e. 18O >5 and an average D value of 43; Table 1). This liquid, with a salinity of ~5 wt % NaCl equiv., boiled at a temperature of 350C near

sea level. This is consistent with hydrostatic pressures at [1600 m depth, i.e. similar to the paleodepth during Ksilicate alteration at 14 Ma, indicating that there was little erosion between these two alteration events.

THE MODEL
Shinohara & Kazahaya (1995) modeled the physical processes involved in the release of hydrothermal uids from a crystallizing magma chamber. We extend this model to evaluate the chemical and isotopic evolution of the uid exsolved from such a crystallizing magma, and apply the results to interpreting the hydrothermal evolution of the FSE porphyry deposit. As a stagnant magma chamber cools by conductive heat loss to the surrounding crust, the magma chamber will crystallize from rim to core with a narrow, inwardmoving zone of crystallization ( Jaeger, 1968; Brandeis & Jaupart, 1987). Fluid that exsolves from such a stagnant magma will have a constant chemical and isotopic composition (Shinohara & Kazahaya, 1995). In contrast, if there is a temperature gradient across the magma chamber, convection may occur (Sparks, 1990), although there is as yet no agreement on this topic (Marsh, 1990). If such convection occurs and mixes the chamber well, crystallization will occur homogeneously,

1744

SHINOHARA AND HEDENQUIST

MAGMA DEGASSING BENEATH A PORPHYRY Cu DEPOSIT

and exsolution may proceed in an open-system manner. As convective crystallization of a saturated magma chamber proceeds, the chemical and isotopic composition of the exsolving uid will vary (Shinohara & Kazahaya, 1995). The pattern of compositional change will be a function of several variables, particularly pressure (Shinohara, 1994; Cline, 1995). However, homogeneous crystallization of a well-mixed magma chamber is only possible during the early stages of magma chamber crystallization, because magma viscosity increases with increasing crystal content to the point where movement in a magma chamber slows and eventually stops, probably at 2555 vol. % crystals, a value that depends on the rheological properties of the magma (Marsh, 1981). Therefore, if a magma chamber is convecting in its early stages of crystallization, there must eventually be a transition to stagnant magma-chamber crystallization. This transition will be associated with a marked change in the nature of the uid being exsolved (Shinohara & Kazahaya, 1995), a change that may be recorded by alteration minerals in an overlying magmatichydrothermal (i.e. porphyry) system. It is this change that we examine by modeling data from the FSE porphyry system, in an attempt to constrain, remotely and in a preliminary fashion, the conditions of magmachamber crystallization.

Model conditions
The chemical and isotopic variations of uid discharging from a crystallizing chamber are calculated under the following conditions. A variety of assumptions have been made, such as melt composition and pressure of crystallization. These assumptions are based in part on typical compositions for silicic magmas deduced from melt inclusion studies (Lowenstern, 1995). The volume of the magma chamber is that required to supply copper to a porphyry deposit the size of the FSE (see below), and the thickness chosen controls the period of crystallization. Pressure is a major control on the model results, and was chosen to replicate the observed uid evolution. As we demonstrate below, the model is relatively insensitive to some of the geological variables, such as the crystal content of the magma at saturation, and the crystal content at which convection ceases, as long as saturation occurs before stagnation. A at-topped aphyric magma chamber 2 km thick and 50 km2 in area (100 km3) is assumed to be emplaced to a depth of 6 km (150 MPa) and 1100C (its liquidus temperature, based on that of granodiorite; Naney, 1983). The magma convects initially, and cools by conductive loss of heat to the crust (initially at ambient temperature) from the top and bottom of the chamber only, ignoring for simplicity any side-wall crystallization. The silicic

magma crystallizes and reaches saturation at 30 vol. % crystals (the minimum crystal content of early-mineral dikes at the FSE deposit) and 800900C. Magma is assumed to convect until an abrupt transition to stagnant conditions occurs at a crystal content of 50 vol. %. The composition of the initial aphyric magma (=melt) is assumed to be 35 wt % H2O, 014 wt % Cl, and D of 40, with the water and chlorine contents increasing to 50 and 02 wt %, respectively, at 30 vol. % crystals. As water solubility is also about 50 wt % at 150 MPa and 800C (Silver et al., 1990), magma with 30 vol. % crystals should be just saturated with water at its top at 6 km depth. The partition coecient of Cl, i.e. the ratio of the molal concentration of Cl in the aqueous phase to that in the melt, is ~27 at the crystallization condition (Shinohara et al., 1989). Although the chamber has a thickness of 2 km, the pressure of the crystallizing magma is assumed to be 150 MPa throughout the chamber for simplicity in this initial model. The deuterium fractionation factor between water in an aqueous uid and a melt containing 5 wt % dissolved water is ~15 (Taylor, 1991). Hornblende is assumed to form throughout the crystallization history, xing a maximum of 10% of the water present in the saturated melt (therefore 05 wt % water in the nal pluton), whereas Cl xation is assumed to be negligible. Deuterium fractionation between an aqueous uid and hornblende is 15 at 800C (i.e. there is no fractionation between melt and hornblende; Suzuoki & Epstein, 1976). The complete hornblende crystals are assumed to remain in isotopic equilibrium with the surrounding magma throughout crystallization.

RESULTS OF CALCULATION Fluid discharge rate

The thickness of the subsolidus portion of a stagnant magma chamber (x, in m) increases in proportion to the square root of time (t, in seconds), x(t)=at 1/2 (1)

where a is a constant that depends on crystal growth rate, nucleation rate, Stefan number, and thermal diffusivity. The constant is estimated to be 8104 for a silicic magma chamber in the crust ( Jaeger, 1968; Brandeis & Jaupart, 1987). It will take 151012 s (~50 000 years) for complete crystallization of a stagnant chamber with a 1 km half-thickness (Fig. 3a). The maximum rate of crystallization during convection is assumed to be twice that of the stagnant chamber (Marsh, 1989), and the rate of crystallization once convection ceases is taken to be equal to that in a stagnant chamber (Fig. 3a). Intrusion of an aphyric magma to 6 km depth into a host rock at ambient temperature will

1745

JOURNAL OF PETROLOGY

VOLUME 38

NUMBER 12

DECEMBER 1997

Fig. 3. Variation of the average crystallized fraction (a) and uid discharge rate (b) of the crystallizing magma chamber. (a) The variation for the case without convection (dashed curve) is calculated with Eq. (1), where the average crystallized fraction is equal to the thickness of the subsolidus magma (x) normalized by the half-thickness of the chamber (1000 m). Convection increases the crystallization rate by a factor of two (continuous curve). The convection stops at a crystal fraction of ~050; the subsequent rate of crystallization will be similar to that in a stagnant chamber. (b) The rate of uid discharge is calculated by multiplying the crystallization rate by the area of the chamber (50 km2) and water content (45 wt %, excluding water in hornblende). The curve is the calculated result for a chamber that convects while the crystallized fraction is 050. The uid discharge rate from the stagnant chamber is smaller by a factor of ve compared with the average for the convection stage. The hatchured and shaded areas show the duration of pre- and post-saturation convection (1000 and 2000 years), respectively.

crystallize during convection to 30 vol % crystals in 1 kyr, at which point the melt becomes saturated. The transition from convecting to stagnant magma conditions is set here at 50 vol. % crystallization, which will be achieved 2 kyr after saturation; the stagnant magma chamber will completely crystallize in a further 25 kyr. Thus the time required from melt saturation (at 30 vol. % crystals) to complete crystallization of the initially convecting chamber is 27 kyr (Fig. 3a), or 28 kyr from the beginning of crystallization. Once the melt becomes saturated with water at 30 vol. % crystals, the rate of uid discharge from the magma chamber can be approximated by the rate of uid exsolution from crystallizing magma. This is obtained by multiplying the crystallization rate by the volatile content of the magma (Shinohara & Kazahaya, 1995; Fig. 3b). In the very early stage of crystallization, the uid discharge rate from the modeled chamber ranges from 2000 to 4000 ton/day (7105 to 15106 ton/year). This is typical of the upper half of the range of hightemperature (>700C) vapor discharge from passively degassing volcanoes (<105 to >5106 ton/year; Hedenquist, 1995). After convection ceases, the uid discharge rate quickly decreases to an average of about 500 ton/ day (2105 ton/year, equal to the lower rate of lowtemperature fumarolic discharge from passively degassing volcanoes).

Compositional evolution
Open-system uid exsolution and discharge occur during convection because the bubbles formed in the chamber can be removed rapidly from the homogeneously mixed chamber (Shinohara & Kazahaya, 1995). In contrast, in a stagnant magma chamber, crystallization and uid exsolution only occur over a narrow zone of crystallization (<5 vol. % of the whole chamber). The uid that discharges from throughout the relatively narrow crystallization interval in the stagnant chamber is a mixture, originating from magma that has a crystal content ranging from nearly 100 vol. % crystals (on the outer margin of the interval) to the volume of crystals present when convection ceased. Therefore, the bulk uid exsolved from the crystallization interval will have a composition equal to that of the volatiles dissolved in the melt at the time stagnant conditions develop (excluding water which is incorporated into hornblende). Thus, the composition of the uid exsolved from a stagnant chamber does not vary over the lifetime of crystallization of the whole chamber (Shinohara & Kazahaya, 1995). Most previous modeling of the variation of chemical and isotopic compositions of the discharging uid has assumed open-system conditions (Taylor, 1991; Shinohara, 1994; Cline, 1995). Such open-system discharge, however, is limited to the early stage of

1746

SHINOHARA AND HEDENQUIST

MAGMA DEGASSING BENEATH A PORPHYRY Cu DEPOSIT

Fig. 4. Variation of NaCl (a) and D (b) content in the exsolving uid (continuous curves) resulting from the crystallization of an early convecting and late stagnant magma chamber. Initial conditions at saturation (030 crystallized fraction) for the calculations are 5 wt % H2O, 02 wt % Cl and 40 D. Open-system degassing is expected during the convection stage (030050 crystallized fraction), whereas complete, closed-system degassing occurs from the narrow crystallization interval in the later stage; variation under open-system conditions is shown by the light curve. Hornblende (dashdot curve) is assumed to incorporate 10% of the water at saturation (05 wt %) with a D 15 less than the exsolved uid.

crystallization, during magma convection, or possibly during eruption. The variation of the NaCl concentration in the exsolving uid depends on the relative exsolving tendencies of Cl and water (Shinohara, 1994). As Cl tends to exsolve more readily than water at 150 MPa, the NaCl concentration decreases with time under open-system degassing of the convecting chamber. The NaCl concentration of the exsolving uid decreases from 86 wt % at the beginning of saturation (30 vol. % crystals) to 79 wt % at 50 vol. % crystallization; the average concentration is 83 wt % (Fig. 4a). After convection ceases at 50 vol. % crystallization, the NaCl concentration will remain constant at 62 wt % (Fig. 4a). During convection, the D values of the discharging water decrease from 25 to 30 (bulk 274), then become constant at 43 on transition from the convecting to stagnant state (Fig. 4b). The D value of igneous hornblende, which is equal to that of the melt, also decreases with time in the early stage. However, if hornblende continually reequilibrates with the exsolving uid by diusion through to the later stage (Zaluski et al., 1994), the hornblende will nally have a model composition of D 58.

DISCUSSION Fluid evolution of the FSE system

The early uid in the FSE hydrothermal system was a mixture of immiscible vapor and hypersaline liquid. The

model isotopic composition of this vapor and liquid can be determined by a mass balance calculation, assuming a simple NaClH2O uid (Table 1). NaCl concentrations of the 550C, 50 MPa vapor and liquid are 07 and 50 wt %, respectively (Sourirajan & Kennedy, 1962). The deuterium fractionation factor between vapor and liquid is taken to be 20, estimated by extrapolation of 4 molal NaCl experimental data (Horita et al., 1995). The uid exsolved during initial crystallization (from 30 to 50 vol. % crystals, taking ~2000 years in a convecting magma chamber) has a bulk composition of 83 wt % NaCl and 27 D. Cooling of this uid to 550C during ascent along permeable channelways (an apophyse extending upward from the parent magma chamber and splitting into individual dikes) results in the solvus being reached at [3 km depth (70 MPa). Further ascent to 2 km depth (50 MPa) forms hypersaline liquid of 50 wt % NaCl and 45 D, and vapor of 07 wt % NaCl and 25 D. A mass balance based on the salinity indicates that the vapor accounts for 916 wt % of the bulk water exsolved from the magma, whereas the remaining 84 wt % comprises the hypersaline liquid, i.e. there is 11 times more vapor, by mass, than liquid in the early hydrothermal system. These model results of the exsolved bulk uid, present as liquid and vapor phases near the apices of the dikes, match well the isotopic compositions of the aqueous phases that formed the K-silicate and advanced argillic alteration, respectively, plus the salinity of the associated inclusion uids (Table 1).

1747

JOURNAL OF PETROLOGY

VOLUME 38

NUMBER 12

DECEMBER 1997

Once the saturated magma chamber contains >50 vol. % crystals, convection ceases, and the remaining crystallization (from 50 to 100 vol. %) will take the balance of the ~25 000 years under stagnant conditions. During this period, an evolved uid of constant composition exsolves from the magma with a composition 62 wt % NaCl and 43 D, as exsolution now occurs from only the relatively narrow rind that crystallizes from the margin inward (Candela, 1991; Shinohara & Kazahaya, 1995). Ascent and cooling of this uid to 350C also matches well the salinity and isotopic composition of water associated with the later-stage sericite that overprinted the early K-silicate alteration (Table 1). In addition to the model results tting the observed compositional variation, the lifetime of the hydrothermal system, 100 kyr, is of the same order of magnitude as the ~27 kyr duration of saturated crystallization for the model magma chamber. Doubling the half-thickness of the parent magma chamber to 2 km will increase the total crystallization period after saturation by a factor of four, i.e. ~108 kyr. But why does the later magmatic uid, that associated with sericite alteration, not separate into hypersaline liquid and vapor during ascent? In the early porphyry system, inception of immiscibility and formation of vapor and hypersaline liquid are deduced to occur at about 3 km depth, followed by vapor removal near the top of the zone of K-silicate alteration. During saturated magma convection (~2000 years), there is a huge advective ux of volatiles (Fig. 3b) and heat that generates a plume of magmatic uid (Fig. 5a). Once the crystal proportion increases to about 50 vol. %, magma convection ceases and the mass (Fig. 3b) and heat ux decrease markedly [i.e. 09(50/70)=643% of the total water is exsolved over ~25 kyr, vs 09(20/70)=257% being exsolved in only the initial 2000 years of convection that follows saturation]. In this situation, the average mass (and therefore heat) ux will decrease by a factor of ve during the period of stagnant crystallization, resulting in a downward collapse of the isotherms, along the dikefocused discharge (Fig. 5b). Thus, the uid exsolved during stagnant crystallization never intersects its solvus during ascent, and hypersaline liquid cannot form. A similar argument was used by Fournier (1987) to predict under what conditions a magmatic intrusion would cause a dilute meteoric water to exceed its critical point. In summary, the transition from a convecting to a stagnant magma chamber is required to match the observed compositional evolution of the FSE hydrothermal system. Simple uid exsolution from a crystallizing stagnant magma chamber will result in a constant composition of magmatic uid that does not match the FSE observations. Equally convincing, when coupled with the evidence for a compositional change, is the sharp decrease in advective uid (and hence heat) ux at the convective

to stagnant transition, which can account qualitatively for the temperature decrease, from 550 to 350C, recorded in veins associated with the change from the early K-silicate to later sericite stage of alteration. Igneous hornblende collected from a relatively fresh dacite breccia at surface (possibly inter-mineral with an age of 143021 Ma) has a D value of 75, similar to hornblende from pre- and post-mineral volcanic rocks in the district (ranging from 70 to 80). This D value is lighter than that calculated here (58), probably because this rock erupted to the surface. During eruption, shallow, open-system degassing (Taylor, 1991) typically results in D-depleted hornblende. The hornblende crystals may also have been aected by deuteric alteration (Nabelek et al., 1983).

Porphyry constraints on magma degassing beneath the FSE system

The model of crystallization of a 2 km thick magma at 6 km depth (150 MPa) can reproduce the chemical and isotopic evolution of the magmatic uid in the FSE porphyry system. The parameters used for the calculation (i.e. magma chamber thickness, depth and pressure, timing of saturation and crystallization history, and concentrations of water, Cl and D) are not constrained directly from eld evidence, as the source magma chamber beneath the deposit is undocumented, unlike that of the Yerington deposit (Dilles, 1987). However, our model is relatively insensitive to several variables, such as timing of saturation and crystallization history. For example, saturation of an aphyric melt at 0 vol. % crystals, and stagnation at 25 vol. % crystals, results in the model results changing by <1 D and a few tenths of a percent NaCl. The initial concentrations of water, Cl and D are those typical for silicic magmas (Lowenstern, 1995). Thus, as a rst approximation, the model results provide some constraints on the conditions of magma crystallization and degassing. The FSE and Lepanto ore bodies have a combined copper content of >36106 tonnes (Garcia, 1991). Assuming that the quartz diorite magma contained a minimum of 50 mg/kg Cu (Cline, 1995; Dilles & Proett, 1995), the minimum volume of such a source magma (density 2000 kg/m3) would be ~36 km3 if there were complete extraction of copper from the magma and 100% transport and deposition eciency in the ore bodies. Therefore, the actual size of the source magma chamber is likely to be larger, of the order of 100 km3, similar to that observed at Yerington (Dilles & Proett, 1995), and predicted by Cline (1995). The calculated duration of magmatic uid discharge from a 2 km thick, 100 km3 chamber is 27 kyr, of the same order of magnitude as the 100 kyr lifetime of

1748

SHINOHARA AND HEDENQUIST

MAGMA DEGASSING BENEATH A PORPHYRY Cu DEPOSIT

Fig. 5. Schematic representation of the parent magma chamber and hydrothermal system associated with the FSE porphyry deposit, during (a) K-silicate and (b) sericite alteration, at ~14 Ma and [13 Ma, respectively. The high-temperature magmatic plume in the early stage is related to the high discharge rate from the convecting magma chamber. The collapse of isotherms in the later stage reects the factor of ve decrease in the rate of uid exsolution from the stagnant magma chamber (Fig. 3b). The shaded area near sea level (at the apex of the dikes) shows the extent of the K-silicate and sericite alteration and CuAu mineralization in the FSE porphyry system (lower limit is a guess). The lithostatic to hydrostatic pressure transition occurs at a temperature <400C (Fournier, 1991), i.e. (a) at a shallow depth over the FSE body during the early high-ux stage (14 Ma), and (b) at a greater depth subsequently, owing to collapse of the isotherms (13 Ma).

the FSE system. If the chamber thickness is only 1 km, the calculated duration of crystallization is only 6000 years, inconsistent with the lifetime of the system. In contrast, crystallization of a 5 km thick chamber requires a longer duration (~120 kyr) than the period of activity. Therefore, the eective thickness (height or width) of the source magma chamber beneath the FSE system is likely to be a few kilometers, similar to that observed for the causative Luhr Hill Granite pluton beneath the porphyry deposits of the Yerington district (Dilles & Proett, 1995). A combination of the crystallization pressure and initial water content of the magma controls the timing of water saturation during crystallization. When the initial water content of the magma is far below its saturation level (<050 times the solubility), saturation and uid exsolution will not occur during the convective stage of a crystallizing magma (Shinohara & Kazahaya, 1995). The evidence for evolution in the FSE magmatic uid composition argues for saturation during open-system conditions of convective crystallization. Thus, at the initiation of crystallization the initial water concentration in the source

magma was much closer than 05 times water saturation. The solubility of water at 300 MPa (~12 km depth) is 75 wt %. As this exceeds the range of normal water contents in silicic melts (e.g. Lowenstern, 1995), it is unlikely that the crystallization pressure was appreciably greater than 150 MPa (6 km depth). Again, this result is similar to the conclusion by Dilles & Proett (1995), who had geological evidence for the Luhr Hill Granite being emplaced at 59 km depth, with the melt containing >4 wt % water at ~10 vol. % crystallization (Dilles, 1987; Dilles & Proett, 1995). The partition coecient of chlorine has a strong dependence on pressure, which results in a pressure-dependent pattern of Cl variation during open-system crystallization (Shinohara et al., 1989; Cline & Bodnar, 1991). As Cl tends to exsolve more readily than water at high pressure, the Cl concentration of a uid exsolving during the early stage of crystallization will start out high, and then will decrease rapidly with crystallization. For example, 90% of the Cl dissolved in a water-saturated melt at 300 MPa would exsolve during the rst 25% of

1749

JOURNAL OF PETROLOGY

VOLUME 38

NUMBER 12

DECEMBER 1997

of uid discharge, resulting in a very low salinity of lateexsolving uid. Such a high crystallization pressure is not consistent with the observation of a 5 wt % NaCl uid during the later, sericite stage at the FSE system. Likewise, if crystallization had occurred at a pressure <150 MPa, the evolution of the bulk salinity would be from low to high, opposite to what we observe. Furthermore, the critical pressure in the system NaCl H2O is ~150 MPa at 800C (Knight & Bodnar, 1989). Thus, crystallization at a pressure less than 150 MPa will result in the direct exsolution of a mixture of vapor and liquid (Shinohara, 1994). Because there is a large density (and thus viscosity) dierence between the vapor and hypersaline liquid phases, we argue that the very dense liquid will not ascend through the same channel as the vapor over a long distance. The early hypersaline (50 wt %) and later, lower-salinity (5 wt %) liquids both ascended to similar depths. Therefore, it is likely that the early magmatic uid ascended as a single phase (a supercritical uid; Bodnar, 1995) for much of the passage from the crystallizing chamber toward the depth of the deposit, having rapidly cooled to about 550C once the uid separated from the magma. This argument suggests that the crystallization pressure of the magma was probably slightly greater than 150 MPa. For a magma chamber ~2 km thick and 100 km3 in volume, its upper surface area will be ~50 km2. As the site of ore deposition is a relatively small area (<1 km2 for the FSE porphyry deposit; Garcia, 1991), 46 km above the crystallizing magma (as modeled here), there must have been an eective mechanism to focus uid ow from the large area of the crystallizing magma chamber. The porphyry deposit is associated with quartz diorite porphyry dikes which are extremely small in volume compared with the magma chamber at depth. The uid that exsolved from the deep magma chamber probably ascended along a deep apophyse and associated shallow dikes and fractures, consistent with the concentric pattern of alteration and mineralization centered on the dikes (Fig. 1).

of early-mineral dikes. Nevertheless, we believe that our results are qualitatively valid, as the model is relatively insensitive to several of the variables (e.g. degree of crystallization at saturation, and the timing of the transition from a convecting to a stagnant magma chamber). The FSE hydrothermal data require an early convecting magma, followed by stagnant crystallization, to account for the observed compositional evolution. There was also a sharp decrease in the rate of uid exsolution related to this transition (by a factor of ve), accounting for the thermal collapse from K-silicate to sericite alteration. We do not see any other mechanism as simple as this, involving stagnant crystallization only, to account for the FSE evolution. Finally, Hedenquist et al. (1997) have argued that much of the copper was deposited in the FSE system during the later stage of sericite alteration. If true, this suggests that much of the copper in the deposit was derived from the parent melt relatively late in its crystallization history, during the period of stagnant, 50100 vol. % crystallization (cf. Cline, 1995). Investigations of volatile (and metal) concentrations in melt inclusions of various intrusions associated with formation of a porphyry deposit would be useful to provide some direct constraints on the parameters applied to our model. In particular, conducting such a study on a porphyry deposit with good exposure of multiple dike events (from pre- and early-mineral to late- and postmineral) may allow the volatile evolution of the causative intrusion(s) to be identied. These results may then be integrated with studies of the metal content of fresh igneous rocks (e.g. Dilles & Proett, 1995), and/or the uid evolution recorded by porphyry hydrothermal systems, to arrive at a better understanding of how magma evolution relates to mineralization. Such modeling also has the potential to be extended to the timing of magmatic exsolution of metals, and their subsequent transport and deposition, using the results of recent experimental data on metal partitioning between melt, and aqueous vapor and liquid (Candela & Piccoli, 1995; Williams et al., 1995).

Implications
Although the results of the model are consistent with observations at the FSE, the magma-chamber crystallization path inferred here is oversimplied, and is only one possible end-member; the crystallization model itself requires further revision (Shinohara & Kazahaya, 1995). Furthermore, another shortcoming of our modeling comes from the fact that the parameters used are not constrained by direct observations of the parent magma chamber, as it is not exposed, unlike that at Yerington. Our only constraint on the timing of magmachamber saturation is the minimum phenocryst content

ACKNOWLEDGEMENTS
We thank Phil Candela and Marco Einaudi for discussions on this topic, and Antonio Arribas and Jim Reynolds for helping to provide the FSE data that have constrained our modeling. Review comments on the paper by Phil Candela, John Dilles, and Dick Sillitoe have been useful in clarifying our modeling and presentation, although we alone remain responsible for the conclusions presented.

1750

SHINOHARA AND HEDENQUIST

MAGMA DEGASSING BENEATH A PORPHYRY Cu DEPOSIT

REFERENCES
n, Arribas, A., Jr, Hedenquist, J. W., Itaya, T., Okada, T., Concepcio R. A. & Garcia, J. S., Jr, 1995. Contemporaneous formation of adjacent porphyry and epithermal CuAu deposits over 300 ka in northern Luzon, Philippines. Geology 23, 337340. Bodnar, R. J., 1995. Fluid-inclusion evidence for a magmatic source of metals in porphyry copper deposits. In: Thompson, J. F. H. (ed.) Magmas, Fluids, and Ore Deposits. Mineralogical Association of Canada Short Course 23, 139152. Brandeis, G. & Jaupart, C., 1987. The kinetics of nucleation and crystal growth and scaling laws for magmatic crystallization. Contributions to Mineralogy and Petrology 96, 2434. Burnham, C. W., 1979. Magmas and hydrothermal uids. In: Barnes, H. L. (ed.) Geochemistry of Hydrothermal Ore Deposits. New York: John Wiley, pp. 71136. Candela, P. A., 1991. Physics of aqueous phase evolution in plutonic environments. American Mineralogist 76, 10811091. Candela, P. A. & Piccoli, P. M., 1995. Model ore-metal partitioning from melts into vapor and vapor/brine mixtures. In: Thompson, J. F. H. (ed.) Magmas, Fluids and Ore Deposits. Mineralogical Association of Canada, Short Course 23, 101127. Cline, J. S., 1995. Genesis of porphyry copper deposits: the behavior of water, chloride, and copper in crystallizing melts. In: Pierce, F. W. & Bolm, J. G. (eds) Porphyry Copper Deposits of the American Cordillera. Arizona Geological Society Digest 20, 6982. Cline, J. S. & Bodnar, R. J., 1991. Can economic porphyry copper mineralization be generated by a typical calc-alkaline melt? Journal of Geophysical Research 96, 81138126. n, R. A. & Cinco, J. C., Jr, 1989. Geology of the LepantoFar Concepcio Southeast gold-rich copper deposit (abstract). In: International Geological Congress, Washington, DC, Proceedings, 1, Washington DC: IGC, pp. 319320, and preprint, 46 pp. Dilles, J. H., 1987. Petrology of the Yerington batholith, Nevada: evidence for evolution of porphyry copper ore uids. Economic Geology 82, 17501789. Dilles, J. H. & Proett, J. M., 1995. Metallogenesis of the Yerington batholith. In: Pierce, F. W. & Bolm, J. G. (eds) Porphyry Copper Deposits of the American Cordillera. Arizona Geological Society Digest 20, 306315. Fournier, R. O., 1987. Conceptual models of brine evolution in magmatichydrothermal systems. In: Volcanism in Hawaii. US Geological Survey Professional Paper 1350, 14871506. Fournier, R. O., 1991. Transition from hydrostatic to greater than hydrostatic uid pressure in presently active hydrothermal systems in crystalline rocks. Geophysical Research Letters 18, 955958. Garcia, J. S., Jr, 1991. Geology and mineralization characteristics of the Mankayan mineral district, Philippines. In: Matsuhisa, Y., Aoki, M. & Hedenquist, J. W. (eds) High-temperature Acid Fluids and Associated Alteration and Mineralization. Geological Survey of Japan Report 277, 2130. Giggenbach, W. F., 1992. Isotopic shifts in waters from geothermal and volcanic systems along convergent plate boundaries and their origin. Earth and Planetary Science Letters 113, 495510. Gustafson, L. B. & Hunt, J. P., 1975. The porphyry copper deposit at El Salvador, Chile. Economic Geology 70, 857912. Hedenquist, J. W., 1995. The ascent of magmatic uid: discharge versus mineralization. In: Thompson, J. F. H. (ed.) Magmas, Fluids and Ore Deposits. Mineralogical Association of Canada Short Course 23, 263289. Hedenquist, J. W. & Lowenstern, J. B., 1994. The role of magmas in the formation of hydrothermal ore deposits. Nature 370, 519527. Hedenquist, J. W., Arribas, A., Jr & Reynolds, T. J., 1997. Evolution of an intrusion-centered hydrothermal system: Far SoutheastLepanto

porphyry and epithermal CuAu deposits, Philippines. Economic Geology (in press). Horita, J., Cole, D. R. & Weslowski, D. J., 1995. The activity composition relationship of oxygen and hydrogen isotopes in aqueous salt solutions: III. Vaporliquid water equilibration of NaCl solutions to 350C. Geochimica et Cosmochimica Acta 59, 11391151. Jaeger, J. C., 1968. Cooling and solidication of igneous rocks. In: Hess, H. H. & Poldervaart, A. (eds) Basalts: The Poldervaart Treatise on Rocks of Basaltic Composition, Vol. 2. New York: John Wiley, pp. 503556. Knight, C. L. & Bodnar, R. J., 1989. Synthetic uid inclusions: IX. Critical PVTX properties of NaClH2O solutions. Geochimica et Cosmochimica Acta 53, 38. Lowenstern, J. B., 1995. Applications of silicatemelt inclusions to the study of magmatic volatiles. In: Thompson, J. F. H. (ed.) Magmas, Fluids and Ore Deposits. Mineralogical Association of Canada Short Course 23, 7199. Marsh, B. D., 1981. On crystallinity, probability of occurrence, and rheology of lava and magma. Contributions to Mineralogy and Petrology 78, 8598. Marsh, B. D., 1989. Magma chambers. Annual Review of Earth and Planetary Sciences 17, 439474. Marsh, B. D., 1990. Crystal capture, sorting, and retention in convecting magma: Reply. Geological Society of America Bulletin 102, 849850. Nabelek, P. I., ONeil, J. R. & Papike, J. J., 1983. Vapor phase exsolution as a controlling factor in hydrogen isotope variation in granitic rocks: the Notch Peak granitic stock, Utah. Earth and Planetary Sciences Letters 66, 137150. Naney, M. T., 1983. Phase equilibria of rock-forming ferromagnesian silicates in granitic systems. American Journal of Science 283, 993 1033. Shinohara, H., 1994. Exsolution of immiscible vapor and liquid phases from a crystallizing silicate melt: implications for chlorine and metal transport. Geochimica et Cosmochimica Acta 58, 52155221. Shinohara, H. & Kazahaya, K., 1995. Degassing processes related to magma-chamber crystallization. In: Thompson, J. F. H. (ed.) Magmas, Fluids and Ore Deposits. Mineralogical Association of Canada Short Course 23, 4770. Shinohara, H., Iiyama, J. T. & Matsuo, S., 1989. Partition of chlorine compounds between silicate melt and hydrothermal solutions: I. Partition of NaClKCl. Geochimica et Cosmochimica Acta 53, 26172630. Silver, L. A., Ihinger, P. D. & Stolper, E., 1990. The inuence of bulk composition on the speciation of water in silicate glasses. Contributions to Mineralogy and Petrology 104, 142162. Sourirajan, S. & Kennedy, G. C., 1962. The system H2ONaCl at elevated temperatures and pressures. American Journal of Science 260, 115141. Sparks, R. S. J., 1990. Crystal capture, sorting, and retention in convecting magma: Discussion. Geological Society of America Bulletin 102, 847848. Suzuoki, T. & Epstein, S., 1976. Hydrogen isotope fractionation between OH-bearing minerals and water. Geochimica et Cosmochimica Acta 40, 12291240. Taylor, B. E., 1991. Degassing of obsidian dome rhyolite, Inyo volcanic chain, California. In: Taylor, H. P., ONeil, J. R. & Kaplan, I. R. (eds) Stable Isotope Geochemistry: a Tribute to Samuel Epstein. Geochemical Society, Special Publication 3, 339353. Taylor, B. E., 1992. Degassing of H2O from rhyolite magma during eruption and shallow intrusion, and the isotopic composition of magmatic water in hydrothermal systems, In: Hedenquist, J. W. (ed.) Magmatic Contributions to Hydrothermal Systems. Geological Survey of Japan, Report 279, 190194.

1751

JOURNAL OF PETROLOGY

VOLUME 38

NUMBER 12

DECEMBER 1997

Williams, T. J., Candela, P. A. & Piccoli, P. M., 1995. The partitioning of copper between silicate melts and two-phase aqueous uids: an experimental investigation at 1 kbar, 800C and 05 kbar, 850C. Contributions to Mineralogy and Petrology 121, 388399.

Zaluski, G., Nesbitt, B.E. & Muehlenbachs, K., 1994. Hydrothermal alteration and stable isotope systematics of the Babine porphyry Cu deposits, British Columbia: implications for uid evolution of porphyry systems. Economic Geology 89, 15181541.

1752